J. An ceran.Soc,83[2]3014-2002000) urna Ceramic Composites with Multilayer Interface Coatings Theodore M. Besmann, Elizabeth R Kupp, Edgar Lara-Curzio, and Karren L. More Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6063 Silicon carbide matrix composites have been fabricated from contact with the ma cause the carbon layer is ither ceramie -grade Nicalon'Mor Hi-NicalonM fibers coated progressive oxida silica grows into the annul e left with an interface material consisting of six alternating carbon by the oxidized carbon, and the fiber becomes well bonded to the and silicon carbide lavers. Initial efforts involved the use of matrix, resulting in a material with little strain tolerance 11-15 chemical vapor infiltration to produce minicomposites (single Thin layers of carbon leave significantly more narrow channels tows of fibers). In subsequent work, forced-flow thermal- s they are oxidized than do thicker layers. The portion of the gradient chemical vapor infiltration was used to produce a carbon layer that is removed near the exposed surface is quickly single composite plate with a multilayer interface from replaced with silica growing from the Sic layers and fiber; the eramic-grade Nicalon fabric and two plates from Hi-Nicalon narrow entrance is sealed before a significant fraction of the abric, one with a single carbon layer and one with a multilayer carbon is lost down the axial length of the layer. Thus, the strongly interface. Tensile testing of the minicomposites and of speci- bonded length is relatively short and, if there is not extensive mens cut from the plates revealed typical composite cracking, the carbon interface layer is retained in much of the specimens to 950 C air for 100 h resulted in large losses in composite. Filipuzzi and co-workers, 12 and Tortorelli and co- and strengths for the as-processed samples. Exposure of tensile strength and strain tolerance regardless of the interface coat- temperatures(>800 C in 100 kPa of dr xygen) and thin carbor ing. The results demonstrate that foreed-nlow thermal-gradient layers(100 nm) single or multiple carbon layers are suscep benefit of multilayer interfaces may be in their ability to cause cracks to follow a tortuous, energy-absorbing path within the tible to oxidation that causes the loss of composite properties. interface layers. 4 ackey and co-workers have advanced that concept to produce multilayer matrices Reducing the thickness of the individual carbon layers in SiC-matrix composites has other advantages. For applications in A CONTINUING issue in the development of ceramic fiber high-radiation environments, such as nuclear fusion reactors, the ceramic matrix composites is the need for an interface presence of carbon can lead to dimensional instability, Carbon material that both controls the friction and bonding between the tends to change dimensions anisotropically during irradiation, fibers and the matrix and protects the fibers from the detrimental ausing fiber decohesion and loss of stress transfer between fibers effects of matrix infiltration and environmental attack. -'In non and the matrix, thus causing loss of composite behavior, The use oxide systems the interface materials that have yielded the best of thinner layers of carbon may reduce this effect perties are graphitic carbon and hexagonal BN. Recentl In the work of Naslain and Rebillat, composites were fabri- multilayers of these materials have been considered as interfaces cated from ceramic-grade Nicalon coated with one to four pairs of or increasing oxidation resistance C/SiC coatings that had a total thickness of 0.5 um. The work The concept of multilayer interfaces consisting of unbonded consisted of preparing both minicomposites, which consist of a layers was first described by Carpenter and Bohlen. They depos- single tow of fibers infiltrated with the fiber coating and matrix, ited multiple layers of Sic on the surface of a large-diameter Sic and larger, two-dimensionally reinforced tensile specimens. Me lament and oxide layers in the outer region, which would be in chanical properties were measured by fiber pushout and tensile contact with an oxide matrix. Work on alternating layers of carbon testing. The mechanical properties of the as-fabricated multilayer nd Sic (C/Sic layers) for small-diameter-fiber SiC-matrix mate interface composites did not appear to be different from those of rials was begun in 1990 at the University of Bordeaux. Steffier samples with a single-carbon-layer interface. Microscopic exami- was also an early developer of this concept for SiC-based fibers. nation revealed that crack deflection occurred at the interface etween the fiber and the first carbon layer in composites prepared of the multilayer interface should be better than that of a single with as-received Nicalon, regardless of the number of interface arbon layer because significantly thinner individual carbon layers layers. They determined that a fiber-surface treatment would be an be used. 4, l,2 Under oxidative conditions, thick(100 nm) required to increase the adhesion between the first carbon layer single carbon layers are removed by oxidation, and silica grows and the fiber. Cracks would then be directed into the weak from both the matrix and the fiber to fill the void volume. The multilayer interface material result is an initially weakened material in which fibers make poor More et al. studied ceramic-matrix composites with multi- layer C/SiC-interface layers prepared by Hyper-Therm, Inc.(Hun- ton Beach, CA). The work involved the characterization of 12 pairs of C/SiC coatings with the SiC film thickness increasing R. Naslain--contributing editor rom the fiber to the matrix side of the interface. The total thickness of these interfaces was almost 3 um. More et al, in luscript No. 189173. Receive closest to the fiber controlled the mechanical behavior of the composite for composites in which untreated fibers were utilized and Technology Devel 需 Stress-rupture testing(four-point bending) at 950C in air at did LC, for the u.s. Department of Energy under not show any improvement in oxidation resistance over a compos- Member, American Ceramic Society te with a single 0.25-pm carbon interface layer. 3014
Ceramic Composites with Multilayer Interface Coatings Theodore M. Besmann,* Elizabeth R. Kupp,* Edgar Lara-Curzio,* and Karren L. More Metals and Ceramics Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6063 Silicon carbide matrix composites have been fabricated from either ceramic-grade NicalonTM or Hi-NicalonTM fibers coated with an interface material consisting of six alternating carbon and silicon carbide layers. Initial efforts involved the use of chemical vapor infiltration to produce minicomposites (single tows of fibers). In subsequent work, forced-flow thermalgradient chemical vapor infiltration was used to produce a single composite plate with a multilayer interface from ceramic-grade Nicalon fabric and two plates from Hi-Nicalon fabric, one with a single carbon layer and one with a multilayer interface. Tensile testing of the minicomposites and of specimens cut from the plates revealed typical composite behavior and strengths for the as-processed samples. Exposure of tensile specimens to 950°C air for 100 h resulted in large losses in strength and strain tolerance regardless of the interface coating. The results demonstrate that forced-flow thermal-gradient chemical vapor infiltration can be used to prepare multilayer interface material. The results also verified that relatively thick (>100 nm) single or multiple carbon layers are susceptible to oxidation that causes the loss of composite properties. I. Introduction A CONTINUING issue in the development of ceramic fiber– ceramic matrix composites is the need for an interface material that both controls the friction and bonding between the fibers and the matrix and protects the fibers from the detrimental effects of matrix infiltration and environmental attack.1–3 In non oxide systems the interface materials that have yielded the best properties are graphitic carbon and hexagonal BN.1 Recently, multilayers of these materials have been considered as interfaces for improving properties or increasing oxidation resistance.4–7 The concept of multilayer interfaces consisting of unbonded layers was first described by Carpenter and Bohlen.8 They deposited multiple layers of SiC on the surface of a large-diameter SiC filament and oxide layers in the outer region, which would be in contact with an oxide matrix. Work on alternating layers of carbon and SiC (C/SiC layers) for small-diameter-fiber SiC-matrix materials was begun in 1990 at the University of Bordeaux.9 Steffier10 was also an early developer of this concept for SiC-based fibers. Naslain and his co-workers noted that the oxidation resistance of the multilayer interface should be better than that of a single carbon layer because significantly thinner individual carbon layers can be used.4,11,12 Under oxidative conditions, thick ($100 nm) single carbon layers are removed by oxidation, and silica grows from both the matrix and the fiber to fill the void volume. The result is an initially weakened material in which fibers make poor contact with the matrix because the carbon layer is absent. After progressive oxidation the silica grows into the annular space left by the oxidized carbon, and the fiber becomes well bonded to the matrix, resulting in a material with little strain tolerance.11–15 Thin layers of carbon leave significantly more narrow channels as they are oxidized than do thicker layers. The portion of the carbon layer that is removed near the exposed surface is quickly replaced with silica growing from the SiC layers and fiber; the narrow entrance is sealed before a significant fraction of the carbon is lost down the axial length of the layer. Thus, the strongly bonded length is relatively short and, if there is not extensive cracking, the carbon interface layer is retained in much of the composite. Filipuzzi and co-workers11,12 and Tortorelli and coworkers14 have shown theoretically and experimentally that high temperatures (.800°C in 100 kPa of dry oxygen) and thin carbon layers (,100 nm) allow this mechanism to operate, thus potentially protecting the integrity of the composite. An additional benefit of multilayer interfaces may be in their ability to cause cracks to follow a tortuous, energy-absorbing path within the interface layers.4 Lackey and co-workers16 have advanced that concept to produce multilayer matrices. Reducing the thickness of the individual carbon layers in SiC-matrix composites has other advantages. For applications in high-radiation environments, such as nuclear fusion reactors, the presence of carbon can lead to dimensional instability. Carbon tends to change dimensions anisotropically during irradiation, causing fiber decohesion and loss of stress transfer between fibers and the matrix, thus causing loss of composite behavior.17 The use of thinner layers of carbon may reduce this effect. In the work of Naslain4 and Rebillat,7 composites were fabricated from ceramic-grade Nicalon coated with one to four pairs of C/SiC coatings that had a total thickness of 0.5 mm. The work consisted of preparing both minicomposites, which consist of a single tow of fibers infiltrated with the fiber coating and matrix, and larger, two-dimensionally reinforced tensile specimens. Mechanical properties were measured by fiber pushout and tensile testing. The mechanical properties of the as-fabricated multilayer interface composites did not appear to be different from those of samples with a single-carbon-layer interface. Microscopic examination revealed that crack deflection occurred at the interface between the fiber and the first carbon layer in composites prepared with as-received Nicalon, regardless of the number of interface layers. They determined that a fiber-surface treatment would be required to increase the adhesion between the first carbon layer and the fiber. Cracks would then be directed into the weaker multilayer interface material. More et al.18 studied ceramic-matrix composites with multilayer C/SiC-interface layers prepared by Hyper-Therm, Inc. (Huntington Beach, CA).10 The work involved the characterization of 12 pairs of C/SiC coatings with the SiC film thickness increasing from the fiber to the matrix side of the interface. The total thickness of these interfaces was almost 3 mm. More et al., 18 in agreement with Naslain4 and Rebillat,7 concluded that the coating closest to the fiber controlled the mechanical behavior of the composite for composites in which untreated fibers were utilized. Stress-rupture testing (four-point bending) at 950°C in air at did not show any improvement in oxidation resistance over a composite with a single 0.25-mm carbon interface layer. R. Naslain—contributing editor Manuscript No. 189173. Received August 25, 1999; approved July 17, 2000. This research was sponsored by the Office of Fossil Energy, Advanced Research and Technology Development Materials Program, U.S. Department of Energy, and performed by Oak Ridge National Laboratory, which is operated by UT-Battelle, LLC, for the U.S. Department of Energy under Contract No. DE-AC05-00OR22725. *Member, American Ceramic Society. J. Am. Ceram. Soc., 83 [12] 3014–20 (2000) 3014 journal
December 2000 eramic Composites with Multilayer Interfa c Bertrand et al. tested SiC-matrix minicomposites containing Honeywell, Morristown, NJ). Before deposition, the fibers expe- lin C/SiC- interface layers(10 sets of layers each containing a 20 rienced a heating cycle to 950C for 1 h, which removed the nm C-and a 50 nm SiC-thickness layer) on untreated Hi-Nicalon protective sizing from the fiber surface. The conditions for and fibers treated with a proprietary surface modification process. depositing all of the interface layers and chemical vapor infiltra- Samples were tested in static fatigue in an oxidizing atmosphere at tion(CVi) of the matrix are given in Table I 700C under a load 10% above the proportional limit to assure matrix cracking. The results indicated a factor of ll increase in life for minicomposites with multilayer C/SiC interfaces on untreated (3) Plate Fabrication fibers(2 vs 22 h) and a 50% increase for minicomposites prepared The fabricated plates measured 152 mm on a side and were with treated fibers(100 vs 150 h)over those with equivalent total prepared using FCVI. A total of 18 layers of cloth with a fabric thickness single carbon layer interfaces. The study thus supports layer orientation of (-30%/0 /30)6 were restrained in a graphite the concept that thin(<100 nm)carbon layers are required to fixture. The protective sizing on the cloth was removed by washing allow in situ sealing of the exposed interlayer by silica growth vith acetone while in the graphite fixture. A uniform pressure(6.9 The work reported here was undertaken to study the fabrication MPa) was then applied to the surface of the preform holder to by forced-flow, thermal-gradient chemical vapor infiltration compress the cloth sufficiently(to 4.76 mm thickness)to obtain (FCvIo of multilayer interfaces in composites containing either the desired -40 vol% fiber loading. The fixture was then placed ir ceramic-grade Nicalon or Hi-Nicalon fibers within a SiC matrix. an FCVI unit capable of fabricating 300-mm-diameter, 25-mm- Only Lackey and co-workers have previously prepared layered thick components. After infiltration the parts were ground to the atrix structures by FCvi. The intent of the current effort was to desired thickness (3. 18 mm) demonstrate FCVI fabrication, to verify the previous conclusions It was necessary to identify the deposition times for each type of with regard to composite contal coating for the alternating C/SiC-interface layers. Times were multilayer C/SiC interfaces, and to determine whether the use of determined by calibration runs and from previous experience with Hi-Nicalon fibers with their improved properties would differ from carbon deposition. Control of interface coating thicknesses in an those observed in the previous studies. FCVI system is difficult because of the changing conditions Minicomposites were used in the initial work to obtai data and experience with the multilayer interface system n through the thickness of the preform. The single-layer carbon interface coating was deposited in the FCVI system at a hot- composites require significantly less labor, time, and materials surface temperature of 1100.C and at 12 kPa total pressure with (including fibers, graphite fixturing, and precursor gases)to flow rates of 150 cm/min C3 Hs and 4400 cm /min argon for 3 h process than do composite plates. In addition, no cutting or The individual carbon layers of the C/SiC-multilayer interfaces rinding steps are required to prepare samples for tensile testing. were deposited under the same conditions; however, each carbon Results of the work with minicomposites were used to develop the layer was deposited for 42 min, Each SiC layer was deposited at process for depositing multilayers on fabric preforms for preparing hot-surface temperature of 1100C and at atmospheric pressure omposite plates. Specimens from the plates were tested in tension with an MTS flow rate of 540 cm /min carried by 5000 cm/min as fabricated and after exposure to ambient air at 950 C for 100 h of hydrogen for 2 min. Carbon was deposited initially, and the coatings were alternated to make six independent carbon laye and five independent SiC layers. IL. Experimental Details After the interface layers were deposited, the matrix was (1 Materials filtrated at a hot-surface temperature of 1100oC at atmospher pressure with an MTS flow rate of 270 cm/min carried by 3000 Two types of fibers were used: ceramic-grade Nicalon and cm/min of hydrogen. The exception was sample No. CVI-1172, Hi-Nicalon(Nippon Carbon, Tokyo, Japan). The tows for which a 1200C hot-surface temperature was used. (It was minicomposites contained 500 filaments. The cloth used determined from sample No. CVI-1172 that 1200C was too high composite plates was 600 X 600 tows/m, plain-weave fabric for effective infiltration of these relatively thin samples. whose tows also each contained 500 filaments The precursor for carbon deposition was 99% purity C3H6 (propylene, Matheson, Morrow, GA)and for Sic deposition, (4 Mechanical Property Testing technical-grade CH3 SiCl, [methyltrichlorosilane(MTS), Gelest, The minicomposites were evaluated in a testing machine that nc,Tullytown, PA]. In addition, 99.997% argon and 99.999% was developed in-house. Samples were loaded in tension until hydrogen(Air Liquide, Houston, TX) were used. failure in ambient air under a constant crosshead displacement rate of I um/s. The samples were gripped by a technique (also developed in-house)in which the ends of the minicomposite (2 Incomposite Fabrication specimens were embedded in epoxy inside aluminum rivets. The The fiber tows were supported in a graphite fixture that allowed load was transferred from the machine to the specimens through the fibers to be kept relatively straight. The fixture was set the rivets, which were held by a pair of specially designed grips vertically in a resistance-heated graphite furnace Gas flowed from The grips were connected to the load train by a self-aligni the bottom to the top of the reactor, and temperatures were mechanism. The resolution of the load cell was one part in 2. The measured with an optical pyrometer (Leeds and Northrup 8627, maximum capacity at 10 V is 1. 5kN (e, the resolution is 0. 2 N) Table I. Interface Layer Deposition Conditions and Infiltration Conditions for Minicomposites Interface coatings Infiltration SiC emperature°C) 0-1000 900-1000 Gas flow(cm/min unless Ar, 250-500 noted otherwise) C3H6, 12-25 CH3 SICl,(MTS), MTS, 0.3 g/min 0.15-0.3g Pressure(kPa) 0.67 0.67 5-60 min 2-30 min 8 h
Bertrand et al.19 tested SiC-matrix minicomposites containing thin C/SiC-interface layers (10 sets of layers each containing a 20 nm C- and a 50 nm SiC-thickness layer) on untreated Hi-Nicalon and fibers treated with a proprietary surface modification process. Samples were tested in static fatigue in an oxidizing atmosphere at 700°C under a load 10% above the proportional limit to assure matrix cracking. The results indicated a factor of 11 increase in life for minicomposites with multilayer C/SiC interfaces on untreated fibers (2 vs 22 h) and a 50% increase for minicomposites prepared with treated fibers (100 vs 150 h) over those with equivalent total thickness single carbon layer interfaces. The study thus supports the concept that thin (,100 nm) carbon layers are required to allow in situ sealing of the exposed interlayer by silica growth. The work reported here was undertaken to study the fabrication by forced-flow, thermal-gradient chemical vapor infiltration (FCVI)20 of multilayer interfaces in composites containing either ceramic-grade Nicalon or Hi-Nicalon fibers within a SiC matrix. Only Lackey and co-workers16 have previously prepared layered matrix structures by FCVI. The intent of the current effort was to demonstrate FCVI fabrication, to verify the previous conclusions with regard to composites containing ceramic-grade Nicalon with multilayer C/SiC interfaces, and to determine whether the use of Hi-Nicalon fibers with their improved properties would differ from those observed in the previous studies. Minicomposites were used in the initial work to obtain some data and experience with the multilayer interface system. Minicomposites require significantly less labor, time, and materials (including fibers, graphite fixturing, and precursor gases) to process than do composite plates.21,22 In addition, no cutting or grinding steps are required to prepare samples for tensile testing. Results of the work with minicomposites were used to develop the process for depositing multilayers on fabric preforms for preparing composite plates. Specimens from the plates were tested in tension as fabricated and after exposure to ambient air at 950°C for 100 h. II. Experimental Details (1) Materials Two types of fibers were used: ceramic-grade Nicalon and Hi-Nicalon (Nippon Carbon, Tokyo, Japan). The tows for the minicomposites contained 500 filaments. The cloth used for the composite plates was 600 3 600 tows/m, plain-weave fabric whose tows also each contained 500 filaments. The precursor for carbon deposition was 99% purity C3H6 (propylene, Matheson, Morrow, GA) and for SiC deposition, technical-grade CH3SiCl3 [methyltrichlorosilane (MTS), Gelest, Inc., Tullytown, PA]. In addition, 99.997% argon and 99.999% hydrogen (Air Liquide, Houston, TX) were used. (2) Minicomposite Fabrication The fiber tows were supported in a graphite fixture that allowed the fibers to be kept relatively straight. The fixture was set vertically in a resistance-heated graphite furnace. Gas flowed from the bottom to the top of the reactor, and temperatures were measured with an optical pyrometer (Leeds and Northrup 8627, Honeywell, Morristown, NJ). Before deposition, the fibers experienced a heating cycle to 950°C for 1 h, which removed the protective sizing from the fiber surface. The conditions for depositing all of the interface layers and chemical vapor infiltration (CVI) of the matrix are given in Table I. (3) Plate Fabrication The fabricated plates measured 152 mm on a side and were prepared using FCVI. A total of 18 layers of cloth with a fabric layer orientation of (230°/0°/30°)6 were restrained in a graphite fixture. The protective sizing on the cloth was removed by washing with acetone while in the graphite fixture. A uniform pressure (6.9 MPa) was then applied to the surface of the preform holder to compress the cloth sufficiently (to 4.76 mm thickness) to obtain the desired ;40 vol% fiber loading. The fixture was then placed in an FCVI unit capable of fabricating 300-mm-diameter, 25-mmthick components. After infiltration the parts were ground to the desired thickness (3.18 mm). It was necessary to identify the deposition times for each type of coating for the alternating C/SiC-interface layers. Times were determined by calibration runs and from previous experience with carbon deposition. Control of interface coating thicknesses in an FCVI system is difficult because of the changing conditions through the thickness of the preform. The single-layer carbon interface coating was deposited in the FCVI system at a hotsurface temperature of 1100°C and at 12 kPa total pressure with flow rates of 150 cm3 /min C3H6 and 4400 cm3 /min argon for 3 h. The individual carbon layers of the C/SiC-multilayer interfaces were deposited under the same conditions; however, each carbon layer was deposited for 42 min. Each SiC layer was deposited at a hot-surface temperature of 1100°C and at atmospheric pressure with an MTS flow rate of 540 cm3 /min carried by 5000 cm3 /min of hydrogen for 2 min. Carbon was deposited initially, and the coatings were alternated to make six independent carbon layers and five independent SiC layers. After the interface layers were deposited, the matrix was infiltrated at a hot-surface temperature of 1100°C at atmospheric pressure with an MTS flow rate of 270 cm3 /min carried by 3000 cm3 /min of hydrogen. The exception was sample No. CVI-1172, for which a 1200°C hot-surface temperature was used. (It was determined from sample No. CVI-1172 that 1200°C was too high for effective infiltration of these relatively thin samples.) (4) Mechanical Property Testing The minicomposites were evaluated in a testing machine that was developed in-house. Samples were loaded in tension until failure in ambient air under a constant crosshead displacement rate of 1 mm/s. The samples were gripped by a technique (also developed in-house) in which the ends of the minicomposite specimens were embedded in epoxy inside aluminum rivets. The load was transferred from the machine to the specimens through the rivets, which were held by a pair of specially designed grips. The grips were connected to the load train by a self-aligning mechanism. The resolution of the load cell was one part in 2.16 The maximum capacity at 10 V is 1.5 kN (i.e., the resolution is 0.2 N). Table I. Interface Layer Deposition Conditions and Infiltration Conditions for Minicomposites Interface coatings Carbon SiC Infiltration SiC Temperature (°C) 900–1000 900–1000 900–1000 Gas flow (cm3 /min unless noted otherwise) Ar, 250–500 H2, 250–500 H2, 500 C3H6, 12–25 CH3SiCl3 (MTS), 0.15–0.3 g/min MTS, 0.3 g/min Pressure (kPa) 0.67 0.67 0.67 Time 5–60 min 2–30 min 8 h December 2000 Ceramic Composites with Multilayer Interface Coatings 3015
can Ceramic SocienBe Vol. 83. No. 12 oth load and crosshead displacement were recorded during the test at a constant rate of 3 Hz. Specimens were cut from each of the plates and were prepared following ASTM specification C 1275-95 for shoulder-loaded tensile specimens(the overall specimen length was 101 mm with 100-grit section of 32 mm x 6.5 mm X 3. 18 mm). 2Surfaces were in multiple overlapping passes with a 12.7-mm-diameter, diamond wheel mounted on a Harig Slicer/Grinder (Model 718, Bridgeport Corp, Birmingham, AL). All specimens were tested in tension at ambient conditions(20.C, 18% relative humidity) under a constant loading rate of 50 N/s in an electro- mechanical testing machine(Instron Model 1380, Canton, MA) equipped with a low-contact force-capacitance extensometer with 100-nm resolution. The shoulder-loaded specimens were gripped by a pair of in-house-developed grips connected to the load train through a pair of self-aligning swivels. Load was transferred from the grips to the specimens through the shoulders. Both load and strain were recorded during the test at a constant rate of 5 Hz. Oxidative exposures under zero load were conducted with the specimens resting on an alumina plate in a hot-wall, resistively Fig. 1. SEM heated box furnace at 950C in ambient air (exterior 20oC, 18% Nicalon specimen(MC-32)revealing the interface layer relative humidity ) The 950C exposure temperature was chosen to be consistent with similar exposures of materials in other ceramic composite programs. No measurements were made of possible temperature gradients within the furnace 250 (5) Microstructural Characterization 200 All the minicomposite samples were examined with a scanning electron microscope (SEM)(Hitachi S-800, NSA Hitachi Scien- tific Instruments, Mountain View, CA). SEM images of polished 21500.25 um ross sections were used for obtaining interface layer thicknesses, which were determined by visual measurement of the length between tangents applied to the boundaries in the images. Reso- lution of the measurements was estimated to be 5 nm Layers 0.3 um Some of the later samples that had thinner coatings were also etched with a NaoH-K, Fe(CN) solution to emphasize the bound- aries between layers. Fracture surfaces of samples were also examined to look for indications of fiber pullout and crack deflection. Transmission electron microscopy (TEM) was used to examine the structure of the interfacial coatings in greater detail 00511.522.53 (Hitachi HF2000, NSA Hitachi Scientific Instruments, Mountain Displacement(mm) Fig. 2. Representative tensile load-displacement curves for the ceramic- grade Nicalon minicomposites. The curves have been shifted for clarity Il. Results (I Minicomposites The interface-layer deposition times varied from 10 to 60 min to fiber strengths within a Hi-Nicalon process batch has been dentify the periods that would produce the desired interface-layer eported to be -22%, and therefore the differences fall within thickness. The initial efforts utilized ceramic-grade Nicalon(sam the error limit ple Nos. MC-22, 31, 32, and 36): an example of the deposited The crosshead displacement at peak load recorded during interlayer structure can be see in Fig. 1. The dark region between evaluation of the minicomposite specimens with multilayer inter- the fiber and the first interface layer is void space where the fiber facial coatings was consistently less than 200 um; those recorded has separated from the interface layer. Either the layer thicknesses for specimens with a single carbon-layer interface were generally for the sample with minimal deposition times, 15 min carbon and greater than 200 um. The shape of the load versus displacement 10 min SiC (MC-31), were too small to discern or there was curves is instructive with regard to the failure behavior of these insufficient material deposited to form coherent layers materials. The minicomposite that did not show any distinct layers Figure 2 contains representative tensile load-displacement MC-31)also did not exhibit strength retention after peak load curves for minicomposite specimens containing ceramic-grade Strength retention was observed only when both the interfacial Nicalon fibers. The actual tensile strengths of the minicomposite shear stress and the fiber-bond strength were low, resulting in large specimens were not determined because it was impractical to matrix crack spacing and long fiber-pullout lengths measure the cross-sectional area of the fracture plane. Neverthe The results from the evaluation of minicomposites with less, the peak loads listed in Table Il are useful measures of ceramic-grade Nicalon fibers guided the development of multilay strength. Furthermore, all specimens contain the same number of ered coatings on Hi-Nicalon fibers. For example, longer deposition fibers(500), and the strength of the minicomposite is dictated times were explored, particularly for carbon (Table ID). Figure 3 primarily by the distribution of fiber strengths and the character- contains micrographs of polished and etched cross sections of the of the fiber-matrix interface. Therefore, comparisons be- interfacial regions of two selected samples. The Hi-Nicalon sample peak loads indicate the effect of fibers and interfacial MC-85, with the longest SiC deposition time and ther gs on the composite tensile strength. The average peak load SiC layer, exhibited rougher layer surfaces than thos other for minicomposites with multilayered interfacial coating (Table l) samples because of its larger SiC grain size. The marked departure appears to be 18%25% lower than that for minicomposites with time versus thickness of sample MC-97 compared with a single layer of carbon(MC-22). However, the standard error fo samples was caused by a change in furnace configuration
Both load and crosshead displacement were recorded during the test at a constant rate of 3 Hz. Specimens were cut from each of the plates and were prepared following ASTM specification C 1275–95 for shoulder-loaded tensile specimens (the overall specimen length was 101 mm with a gauge section of 32 mm 3 6.5 mm 3 3.18 mm).23 Surfaces were ground in multiple overlapping passes with a 12.7-mm-diameter, 100-grit diamond wheel mounted on a Harig Slicer/Grinder (Model 718, Bridgeport Corp., Birmingham, AL). All specimens were tested in tension at ambient conditions (20°C, 18% relative humidity) under a constant loading rate of 50 N/s in an electromechanical testing machine (Instron Model 1380, Canton, MA) equipped with a low-contact force-capacitance extensometer with 100-nm resolution. The shoulder-loaded specimens were gripped by a pair of in-house-developed grips connected to the load train through a pair of self-aligning swivels. Load was transferred from the grips to the specimens through the shoulders. Both load and strain were recorded during the test at a constant rate of 5 Hz. Oxidative exposures under zero load were conducted with the specimens resting on an alumina plate in a hot-wall, resistively heated box furnace at 950°C in ambient air (exterior 20°C, 18% relative humidity). The 950°C exposure temperature was chosen to be consistent with similar exposures of materials in other ceramic composite programs. No measurements were made of possible temperature gradients within the furnace. (5) Microstructural Characterization All the minicomposite samples were examined with a scanning electron microscope (SEM) (Hitachi S-800, NSA Hitachi Scientific Instruments, Mountain View, CA). SEM images of polished cross sections were used for obtaining interface layer thicknesses, which were determined by visual measurement of the length between tangents applied to the boundaries in the images. Resolution of the measurements was estimated to be ;5 nm. Some of the later samples that had thinner coatings were also etched with a NaOH–K3Fe(CN)6 solution to emphasize the boundaries between layers. Fracture surfaces of samples were also examined to look for indications of fiber pullout and crack deflection. Transmission electron microscopy (TEM) was used to examine the structure of the interfacial coatings in greater detail (Hitachi HF2000, NSA Hitachi Scientific Instruments, Mountain View, CA). III. Results (1) Minicomposites The interface-layer deposition times varied from 10 to 60 min to identify the periods that would produce the desired interface-layer thickness. The initial efforts utilized ceramic-grade Nicalon (sample Nos. MC-22, 31, 32, and 36); an example of the deposited interlayer structure can be see in Fig. 1. The dark region between the fiber and the first interface layer is void space where the fiber has separated from the interface layer. Either the layer thicknesses for the sample with minimal deposition times, 15 min carbon and 10 min SiC (MC-31), were too small to discern or there was insufficient material deposited to form coherent layers. Figure 2 contains representative tensile load–displacement curves for minicomposite specimens containing ceramic-grade Nicalon fibers. The actual tensile strengths of the minicomposite specimens were not determined because it was impractical to measure the cross-sectional area of the fracture plane. Nevertheless, the peak loads listed in Table II are useful measures of strength. Furthermore, all specimens contain the same number of fibers (500), and the strength of the minicomposite is dictated primarily by the distribution of fiber strengths and the characteristics of the fiber–matrix interface. Therefore, comparisons between peak loads indicate the effect of fibers and interfacial coatings on the composite tensile strength. The average peak load for minicomposites with multilayered interfacial coating (Table II) appears to be 18%–25% lower than that for minicomposites with a single layer of carbon (MC-22). However, the standard error for fiber strengths within a Hi-Nicalon process batch has been reported24 to be ;22%, and therefore the differences fall within the error limit. The crosshead displacement at peak load recorded during evaluation of the minicomposite specimens with multilayer interfacial coatings was consistently less than 200 mm; those recorded for specimens with a single carbon-layer interface were generally greater than 200 mm. The shape of the load versus displacement curves is instructive with regard to the failure behavior of these materials. The minicomposite that did not show any distinct layers (MC-31) also did not exhibit strength retention after peak load. Strength retention was observed only when both the interfacial shear stress and the fiber-bond strength were low, resulting in large matrix crack spacing and long fiber-pullout lengths. The results from the evaluation of minicomposites with ceramic-grade Nicalon fibers guided the development of multilayered coatings on Hi-Nicalon fibers. For example, longer deposition times were explored, particularly for carbon (Table II). Figure 3 contains micrographs of polished and etched cross sections of the interfacial regions of two selected samples. The Hi-Nicalon sample MC-85, with the longest SiC deposition time and therefore thickest SiC layer, exhibited rougher layer surfaces than those of the other samples because of its larger SiC grain size. The marked departure of coating time versus thickness of sample MC-97 compared with the other samples was caused by a change in furnace configuration Fig. 1. SEM image of a polished cross section of a ceramic-grade Nicalon specimen (MC-32) revealing the interface layers. Fig. 2. Representative tensile load–displacement curves for the ceramicgrade Nicalon minicomposites. The curves have been shifted for clarity. 3016 Journal of the American Ceramic Society—Besmann et al. Vol. 83, No. 12
December 2000 Table Il. Deposition Time, Interlayer Thicknesses, and Average Peak Load for Multilayer C/SiC Minicomposite Samples Tensile testing Peak load Sample Nicalon fiber type (μm) (um) MC-22 0.25 0.25 0.2 110±6 MC-3 Ceramic grade 0.5 90±37 Ceramic grade l5/15 0.008 0.03 85±41 Ceramic grade 83±15 MC-80 0.05 MC-85 Hi-Nicalon 15/30 0.4 122±15 MC-97 Hi-Nicalol 0.5 0.083 0.1 150±14 000 MC-114 Hi-Nicalon 120 C only 0.25 187±22 iNdividual layers were not discernible 250 025um carbon only 0 0. 4 um 03m 1.5 2.5 Displacement(mm) (a) Nicalon minicomposites. The curves have been shifted for clarity crosshead displacement at the peak load than those of minicom- posites made with ceramic-grade Nicalon fibers but do not exhibit their gradual strength reduction after peak load. The load versus crosshead displacement curves are nonlinear after the matrix cracking load is exceeded, and the small discontinuities along the curves typically indicate the occurrence of matrix cracks. The peak loads and crosshead displacements were greatest for the specimens that had only a single carbon-layer interface(MC-114) (2) Composite Plates The results of infiltration and the properties of the composite plates are listed in Table Ill, The density of each sample was I um calculated from geometrical measurements and the mass of the sample with the preform containing 40 vol% fiber. The variation in filtration time was, in part, due to difficulties in performing (b) FCVI on such thin components. Experience with preforms less than 10 mm in thickness suggests that there is a tendency fo Fig. 3. SEM images of polished and etched cross sections of specimens premature sealing of the entrance surface by deposited Si of the Hi-Nicalon multilayer interface minicomposites with total carbon Interrupting the infiltration after several hours and refixturing the thicknesses of (a)0.3 um(MC-80)and(b)0.5 Hm(MC- sample to allow for a small gap between the bottom(cooled side) of the graphite holder and the sample solved this problem. The result, however, is poorer reproducibility with regard to infiltratio time that caused greater MTS depletion before the deposition of SiC on An example of the alternating carbon and Sic interface laye the fiber tow seen in the TEM image of Fig. 5. There appears to be some small Representative tensile load-displacement curves for Hi-Nicalon variability in the layers, which are 0. 1 to 0.2 um in thickness, and fiber-reinforced minicomposites are plotted in Fig. 4. The maxi- the SiC layers become increasingly rougher away from the fiber, mum load of these materials was found to be -50% higher than apparently because of the fairly large grains of the deposited Si nose of the ceramic-grade Nicalon minicomposites, probably Each layer is conformal with respect adjacent layers. Images
that caused greater MTS depletion before the deposition of SiC on the fiber tow. Representative tensile load–displacement curves for Hi-Nicalon fiber-reinforced minicomposites are plotted in Fig. 4. The maximum load of these materials was found to be ;50% higher than those of the ceramic-grade Nicalon minicomposites, probably because of the better strength retention of Hi-Nicalon fibers during high-temperature matrix deposition.25,26 The curves show greater crosshead displacement at the peak load than those of minicomposites made with ceramic-grade Nicalon fibers but do not exhibit their gradual strength reduction after peak load. The load versus crosshead displacement curves are nonlinear after the matrixcracking load is exceeded, and the small discontinuities along the curves typically indicate the occurrence of matrix cracks. The peak loads and crosshead displacements were greatest for the specimens that had only a single carbon-layer interface (MC-114). (2) Composite Plates The results of infiltration and the properties of the composite plates are listed in Table III. The density of each sample was calculated from geometrical measurements and the mass of the sample with the preform containing 40 vol% fiber. The variation in infiltration time was, in part, due to difficulties in performing FCVI on such thin components. Experience with preforms less than 10 mm in thickness suggests that there is a tendency for premature sealing of the entrance surface by deposited SiC. Interrupting the infiltration after several hours and refixturing the sample to allow for a small gap between the bottom (cooled side) of the graphite holder and the sample solved this problem. The result, however, is poorer reproducibility with regard to infiltration time. An example of the alternating carbon and SiC interface layers is seen in the TEM image of Fig. 5. There appears to be some small variability in the layers, which are 0.1 to 0.2 mm in thickness, and the SiC layers become increasingly rougher away from the fiber, apparently because of the fairly large grains of the deposited SiC. Each layer is conformal with respect to adjacent layers. Images obtained from other areas of the specimen are similar, indicating relatively good through-thickness uniformity. Table II. Deposition Time, Interlayer Thicknesses, and Average Peak Load for Multilayer C/SiC Minicomposite Samples Sample Nicalon fiber type C/SiC deposition times (min/layer) Total interlayer thickness (mm) Total carbon layer thickness (mm) Average carbon layer thickness (mm) Average SiC layer thickness (mm) Tensile testing Peak load (N) Number of samples MC-22 Ceramic grade 120 C only 0.25 0.25 0.25 110 6 6 8 MC-31 Ceramic grade 15/10 0.5 ††† 90 6 37 2 MC-36 Ceramic grade 15/15 0.2 0.05 0.008 0.03 85 6 41 6 MC-32 Ceramic grade 15/20 1.3 0.3 0.05 0.2 83 6 15 6 MC-80 Hi-Nicalon 30/15 0.8 0.3 0.05 0.1 129 6 9 10 MC-85 Hi-Nicalon 15/30 1.4 0.4 0.067 0.2 122 6 15 10 MC-97 Hi-Nicalon 60/2 1.0 0.5 0.083 0.1 150 6 14 10 MC-114 Hi-Nicalon 120 C only 0.25 0.25 0.25 187 6 22 15 † Individual layers were not discernible. Fig. 3. SEM images of polished and etched cross sections of specimens of the Hi-Nicalon multilayer interface minicomposites with total carbon thicknesses of (a) 0.3 mm (MC-80) and (b) 0.5 mm (MC-97). Fig. 4. Representative tensile load–displacement curves for the HiNicalon minicomposites. The curves have been shifted for clarity. December 2000 Ceramic Composites with Multilayer Interface Coatings 3017
f the American Ceramic Sociery-l VoL. 83. No. 12 Table Ill. Specifications and Properties for the Fabricated Composite Tensile Samples Tensile specimen average Tensi Sample no Cv-1172 1225 2.41±0.036 0.833±0.015 69±27 4±1 CvI-1174 2.49±0.040 0.835±0.016 254±19 66+9 CVI-1175 Hi-Nicalon 28.5 2.61±0.018 0.872±0.007 260±24 300 200 50 ramic Grade Hi-Nicalon/Carbon Hi-Nicalon/Multilayer on/ Multilayer Interface Interface Carbon Nicalon Fibe 7. Comparison of average ultimate strength measurements of bricated and oxidized tensile bars obtained from the composite pla Fig. 5. TEM image of a specimen of the composite ceramicgrade Nicalon with a multilayer interface(CVI ealing for 100 h and then were fast-fractured at room temperature. The the alternating carbon and SiC layers. specimens had ground, exposed surfaces with no seal coat. The epresentative tensile curves for the oxidized specimens(Fig. 6) In monotonic tensile testing. the cessed specimens eveal little strain tolerance. As can be seen from the ultimate trength values(Table Ill and Fig. 7), less than one-third of the surfaces with uncorrelated behavior (i.e, fabric layers failed at as-fabricated strength was retained after exposure different locations ). Representative tensile curves are seen in Fig Specimens fabricated with Hi-Nicalon fibers were stronger than IV. Discussion those fabricated with ceramic-grade Nicalon, a condition that is haracteristic of the better strength retention of Hi-Nicalon, as was In the specimens of Naslain and co-workers, ",and in those also seen in the minicomposites(Table Ill and Fig. 7). In addition, examined here, cracks are ultimately directed between the fiber the higher overall density of the Hi-Nicalon samples likely and the first carbon layer when the fibers are not treated to increas contributes to the reduced standard deviation in the data(Table adhesion between the carbon laver and the fibers. Even in these materials. however. some crack deflection can be observed within In a test to determine resistance to oxidation, approximately half the multilayers, as seen in Fig 8, which shows a fiber and interface of the specimens of each of the samples were held at 950C in air 300 NIcalon/ Crack deflection within ade nicalon/ 100 Oxid bed Oxidized 0.0020.40.6081.0121.4 Fig. 8. SEM image of a polished cross section of a ceramic-grade Fig. 6. Representative stress-strain curves for as-fabricated and oxidized Nicalon multilayer interface minicomposite(MC-32)revealing some crack specimens obtained from composite plates. The curves have been shifted deflection within the interface and extension of cracks along the interface for clarity between the first carbon layer and the fiber
In monotonic tensile testing, the as-processed specimens showed nonlinear stress–strain behavior and very fibrous fracture surfaces with uncorrelated behavior (i.e., fabric layers failed at different locations). Representative tensile curves are seen in Fig. 6. Specimens fabricated with Hi-Nicalon fibers were stronger than those fabricated with ceramic-grade Nicalon, a condition that is characteristic of the better strength retention of Hi-Nicalon, as was also seen in the minicomposites (Table III and Fig. 7). In addition, the higher overall density of the Hi-Nicalon samples likely contributes to the reduced standard deviation in the data (Table III). In a test to determine resistance to oxidation, approximately half of the specimens of each of the samples were held at 950°C in air for 100 h and then were fast-fractured at room temperature. The specimens had ground, exposed surfaces with no seal coat. The representative tensile curves for the oxidized specimens (Fig. 6) reveal little strain tolerance. As can be seen from the ultimate strength values (Table III and Fig. 7), less than one-third of the as-fabricated strength was retained after exposure. IV. Discussion In the specimens of Naslain and co-workers,4,7 and in those examined here, cracks are ultimately directed between the fiber and the first carbon layer when the fibers are not treated to increase adhesion between the carbon layer and the fibers. Even in these materials, however, some crack deflection can be observed within the multilayers, as seen in Fig. 8, which shows a fiber and interface Table III. Specifications and Properties for the Fabricated Composite Tensile Samples Sample no. Interface Nicalon grade Infiltration time (h) Tensile specimen density g/cm3 Tensile specimen fractional density Unoxidized average strength (MPa) Oxidized average strength (MPa) CVI-1172 C/SiC Ceramic 12.25 2.41 6 0.036 0.833 6 0.015 169 6 27 54 6 18 CVI-1174 Carbon Hi-Nicalon 17 2.49 6 0.040 0.835 6 0.016 254 6 19 66 6 9 CVI-1175 C/SiC Hi-Nicalon 28.5 2.61 6 0.018 0.872 6 0.007 260 6 24 80 6 5 Fig. 5. TEM image of a specimen of the composite plate prepared from ceramic-grade Nicalon with a multilayer interface (CVI-1172) revealing the alternating carbon and SiC layers. Fig. 6. Representative stress–strain curves for as-fabricated and oxidized specimens obtained from composite plates. The curves have been shifted for clarity. Fig. 7. Comparison of average ultimate strength measurements of the as-fabricated and oxidized tensile bars obtained from the composite plates. Fig. 8. SEM image of a polished cross section of a ceramic-grade Nicalon multilayer interface minicomposite (MC-32) revealing some crack deflection within the interface and extension of cracks along the interface between the first carbon layer and the fiber. 3018 Journal of the American Ceramic Society—Besmann et al. Vol. 83, No. 12
December 2000 eramic Composites with Multilayer Interfa layer axes aligned parallel and the crack normal to the stress those of composites reinforced with ceramic-grade Nicalon Mac- direction. Naslain and co-workers have been able to synthesize roscopic tensile testing also indicated that multilayer interface materials in which cracking occurs within the interface coating by material has properties equivalent to composites having a single- use of a fiber treatment to increase bonding between the carbon carbon- layer interface. Oxidation protection based on the multi- layer and the fibers. , -The stronger bonding shifts the weakest layer C/Sic concept was not apparent from the materials and zone to either within a carbon layer or to the carbon-SiC layer testing reported here. However, fairly thick carbon- interface layers interfaces, resulting in greater energy absorption and higher were utilized (0. 1-0.2 um), and it has been demonstrated that macroscopic tensile strength. Naslain et al have also improved the thinner layers(<0. 1 um) are required to obtain protective sealing composition and structure of the alternating layers through the use by the growth of silica. In addition, because untreated fibers were of pulsed chemical vapor deposition. o,2 used, debonding occurred between the first carbon layer and the The shapes of the tensile curves for the ceramic-grade Nicalon fiber, and thus the material behaved not very differently from nicomposite specimens with single-layer carbon interface(MC omposites having single-carbon-layer interfaces 22)are reproduced in the multilayer interface specimens, although with a possible reduction in ultimate load of 20%-30%. Either the relative thinness of the individual carbon layers or their irregular nature may be the cause of the lower ultimate load because they Acknowledgments rovide a less compliant layer to accommodate interface rough- ness.3,29 This effect is seen in the flexural-strength measurements We would like to acknowledge the useful of P F. Tortorelli, H. T, Lin, of Lowden and Stinton2and in the tensile test results of Naslain, 4 and R.A. Lowden. J C. McLaughlin performed much of the experimental infiltration in which there is limited measured displacement in the mechanical-property testing of materials with a relatively thin carbon interface layer. This finding may be supported by the References results from the minicomposite samples, in which the multilayer interface sample having the largest total thickness of carbon had R. A. Lowden and K. I Effect of Fiber Coatings on Interfa the maximum tensile strength (Table m) The results of the tensile strength measurements of the plate posites. Edited by C. G. Pantano and E. J. H. Chen. Materials Researc samples are consistent with those of Naslain and co-workers The plateau exhibited in their tensile curves after matrix cracking, which is absent in the curves generated in this work. may be th 5 essed by Cvl,C两hmB28(190 carbon thickness of 0.5 um)as compared with 0.25 um for the R. Naslain,"The Concep.? result of their substantially thicker interface layer(having a total Ceramic Composites, Scr. M ater,3l{81079-84(1994) Layered Interphases in SiC/SiC Pp. 23-39 in naterial prepared in this work. The thicker layer yields a lower High-Temperature Ceramic-Matrix Composites ll. Edited by A G. Evans and R. interfacial shear stress, allowing for greater slippage of the fibers American Ceramic Society. Westerville. OH. 1995 SE. Gourb aces in SiC/C CVD the matrix. The strengths of the composites prepared by Naslain ers, "Mater. Sci. Forum, 207(209J237-40(1996)- C) Mu. laye as Fibre C and co-workers from untreated ceramic-grade Nicalon fibers are higher than those reported here, although the present composites Oxidation Protection and Improved Mechanical Behavior; pp. 530-31 in prepared from Hi-Nicalon with single and multilayer interfaces iennial Conference on Carbon. American Carbon Society, University Park, had similar tensile behavior and ultimate strength PA,1997 7F, Rebillat, J. Lamon, R. Naslain, E. Lara-Curzio, M.K. Ferber, and T.M Because of their relatively large thickness, the multilayer Besmann, "Properties of Multilayered Interphases in SiC/SiC Chemical-Vapor- interfaces prepared in this study appeared to afford no benefit in Infiltrated Composites with"Weak and"Strong'Interfaces, ".Am. Ceram. Soc.,81 oxidation; all specimens experienced severe degradation of me- 192315-26(1998 chanical properties (i. e, low strength and strain tolerance). It is nd J. w. Bohlen, "Fiber Coatings for Ceramic Matrix Composites, Ceram. Eng. Sci. Proc., 13 [7-8]23-56(1992). likely that this condition resulted from the consumption of the and F. Heurtevent, "Method of manufacturing a carbon coating adjacent to the fibers via oxidation, followed by Composite Material with Lamellar Interphase Between Reinforced Fibers and Matrix silica formation that filled the resulting void space. 3,29The and Material Obtained, International Pat, No. wO 95/09136, Societe European relatively thick gap in the interlayers left by the oxidized carbon Propulsion, 19 may not have allowed sufficient rapid sealing of the exposed w.S. Steffier, "Multilayer Fiber Coating Comprising Altermate Fugitive Carbon surface because the layers were of the order of 0.1 to 0. 2 um thick and Ceramic Coating Material for Toughened Ceramic Composite Materials, U.S Pat.No.5455106,1995 (Fig. 6). The modeling and experimental efforts of Filipuzzi and G. Camus, R. Naslain, and J. Thebault, "Oxidation Mechanisms and co-workers, 2 and of Tortorelli and co-workers 4, 30 indicated Kinetics of 1-D-SiC/C/SiC Composite Materials: I, An Experimental Approach, that thicknesses of less than 0 I um are required. 12L. Filipuzzi and R. Naslain, "Oxidation Mechanisms and Kinetics of I-D-SiC/ C/SiC Composite Materials: Il, Modeling, J. Am. Ceran Soc., 77[21 467-80 V. Conclusions IR. D. James, R. A. Lowden, and K. L. More, "The Effects of Oxidation an In minicomposites, multilayer interfaces result in mechanical properties that may be equivalent to those of single-layer carbon interfaces. Properties are highly influenced by the total thickness Applications. Edited by M. D. Sacks. American Ceramic Society, Westerville, OH, of the carbon interface material, probably in accordance with the IP F. Tortorelli, S Nijhawan, L. Riester, and R. A Lowden, "Influence of Fiber need for a compliant interna surface roughness. Although there is some evidence for crack R. H. Jones, C. H. Henager Jr, and C. F, Windisch Jr,"High Temperature deflection within the multilayer interfacial coating, cracks are Corrosion and Crack Growth of SiC-SiC at Variable Oxygen Pressures, Mater. Sci edominantly directed along the interface between the fiber and rst carbon interface layer because of poor bonding between these Ibw. J. Lackey, S. Vaidyaraman, and K. L. More, "Laminated C-SiC Matrix materials. Thus, for the composites of this study, the additional Composites Produced by CVI, J. Am. Ceram Soc., 80[1]113-16( 1997) 17L. L. Snead, D. Steiner, and S J. Zinkle, "Measurement of the Effect ers largely provided no mechanical advantage, Gradual decay Ceramic Composite Interfacial Strength, J. Nucl. Mater, 191[1941 of the tensile load was observed after peak strength f for minicom- 556-70(1992) posites with presumably low interfacial shear stres K. L More E. Lara-Curzio H It has been demonstrated that the FCVi process can successfully deposit alternating layers of carbon and SiC Tensile strengths and Conference and Exposition on Co c=“ Materials and strain tolerances of composites with Hi-Nicalor Structures(Cocoa Beach, FL, January 1996)
layer axes aligned parallel and the crack normal to the stress direction. Naslain and co-workers have been able to synthesize materials in which cracking occurs within the interface coating by use of a fiber treatment to increase bonding between the carbon layer and the fibers.4,7,27 The stronger bonding shifts the weakest zone to either within a carbon layer or to the carbon–SiC layer interfaces, resulting in greater energy absorption and higher macroscopic tensile strength. Naslain et al. have also improved the composition and structure of the alternating layers through the use of pulsed chemical vapor deposition.6,28 The shapes of the tensile curves for the ceramic-grade Nicalon minicomposite specimens with single-layer carbon interface (MC- 22) are reproduced in the multilayer interface specimens, although with a possible reduction in ultimate load of 20%–30%. Either the relative thinness of the individual carbon layers or their irregular nature may be the cause of the lower ultimate load because they provide a less compliant layer to accommodate interface roughness.3,29 This effect is seen in the flexural-strength measurements of Lowden and Stinton29 and in the tensile test results of Naslain,4 in which there is limited measured displacement in the mechanical-property testing of materials with a relatively thin carbon interface layer. This finding may be supported by the results from the minicomposite samples, in which the multilayer interface sample having the largest total thickness of carbon had the maximum tensile strength (Table II). The results of the tensile strength measurements of the plate samples are consistent with those of Naslain and co-workers.4,27 The plateau exhibited in their tensile curves after matrix cracking, which is absent in the curves generated in this work, may be the result of their substantially thicker interface layer (having a total carbon thickness of 0.5 mm) as compared with 0.25 mm for the material prepared in this work. The thicker layer yields a lower interfacial shear stress, allowing for greater slippage of the fibers in the matrix. The strengths of the composites prepared by Naslain and co-workers from untreated ceramic-grade Nicalon fibers are higher than those reported here, although the present composites prepared from Hi-Nicalon with single and multilayer interfaces had similar tensile behavior and ultimate strength. Because of their relatively large thickness, the multilayer interfaces prepared in this study appeared to afford no benefit in oxidation; all specimens experienced severe degradation of mechanical properties (i.e., low strength and strain tolerance). It is likely that this condition resulted from the consumption of the carbon coating adjacent to the fibers via oxidation, followed by silica formation that filled the resulting void space.13,29 The relatively thick gap in the interlayers left by the oxidized carbon may not have allowed sufficient rapid sealing of the exposed surface because the layers were of the order of 0.1 to 0.2 mm thick (Fig. 6). The modeling and experimental efforts of Filipuzzi and co-workers11,12 and of Tortorelli and co-workers14,30 indicated that thicknesses of less than 0.1 mm are required. V. Conclusions In minicomposites, multilayer interfaces result in mechanical properties that may be equivalent to those of single-layer carbon interfaces. Properties are highly influenced by the total thickness of the carbon interface material, probably in accordance with the need for a compliant interlayer material to accommodate fibersurface roughness. Although there is some evidence for crack deflection within the multilayer interfacial coating, cracks are predominantly directed along the interface between the fiber and first carbon interface layer because of poor bonding between these materials. Thus, for the composites of this study, the additional layers largely provided no mechanical advantage. Gradual decay of the tensile load was observed after peak strength for minicomposites with presumably low interfacial shear stress. It has been demonstrated that the FCVI process can successfully deposit alternating layers of carbon and SiC. Tensile strengths and strain tolerances of composites with Hi-Nicalon were superior to those of composites reinforced with ceramic-grade Nicalon. Macroscopic tensile testing also indicated that multilayer interface material has properties equivalent to composites having a singlecarbon-layer interface. Oxidation protection based on the multilayer C/SiC concept was not apparent from the materials and testing reported here. However, fairly thick carbon-interface layers were utilized (0.1–0.2 mm), and it has been demonstrated that thinner layers (,0.1 mm) are required to obtain protective sealing by the growth of silica. In addition, because untreated fibers were used, debonding occurred between the first carbon layer and the fiber, and thus the material behaved not very differently from composites having single-carbon-layer interfaces. Acknowledgments We would like to acknowledge the useful comments of P. F. Tortorelli, H. T. Lin, and R. A. Lowden. J. C. McLaughlin performed much of the experimental infiltration work, and T. S. Geer prepared the metallographic samples. References 1 R. A. Lowden and K. L. More, “The Effect of Fiber Coatings on Interfacial Shear Strength and the Mechanical Behavior of Ceramic Composites”; pp. 205–14 in Materials Research Society Symposium Proceedings, Vol. 170, Interfaces in Composites. Edited by C. G. Pantano and E. J. H. Chen. Materials Research Society, Pittsburgh, PA, 1990. 2 R. Naslain, “Fiber-Matrix Interphases and Interfaces in Ceramic-Matrix Composites Processed by CVI,” Compos. Interfaces, 1 [3] 253–86 (1993). 3 R. J. Kerans, “Issues in the Control of Fiber-Matrix Interface Properties in Ceramic Composites,” Scr. Metall. Mater., 31 [8] 1079–84 (1994). 4 R. Naslain, “The Concept of Layered Interphases in SiC/SiC”; pp. 23–39 in High-Temperature Ceramic-Matrix Composites II. Edited by A. G. Evans and R. Naslain. American Ceramic Society, Westerville, OH, 1995. 5 F. Gourbilleau, G. Nouet, and M. Ducarroir, “Interfaces in SiC/C CVD Multilayers,” Mater. Sci. Forum, 207 [209] 237–40 (1996). 6 F. Heurtevent, R. Pailler, and X. Bourrat, “(PyC/SiC) Multilayer as Fibre Coating: Oxidation Protection and Improved Mechanical Behavior”; pp. 530–31 in Carbon ’97, 23rd Biennial Conference on Carbon. American Carbon Society, University Park, PA, 1997. 7 F. Rebillat, J. Lamon, R. Naslain, E. Lara-Curzio, M. K. Ferber, and T. M. Besmann, “Properties of Multilayered Interphases in SiC/SiC Chemical-VaporInfiltrated Composites with ‘Weak’ and ‘Strong’ Interfaces,” J. Am. Ceram. Soc., 81 [9] 2315–26 (1998). 8 H. W. Carpenter and J. W. Bohlen, “Fiber Coatings for Ceramic Matrix Composites,” Ceram. Eng. Sci. Proc., 13 [7–8] 23–56 (1992). 9 S. Goujard, P. Dupel, R. Pailler, and F. Heurtevent, “Method of Manufacturing a Composite Material with Lamellar Interphase Between Reinforced Fibers and Matrix, and Material Obtained,” International Pat. No. WO 95/09136, Societe European Propulsion, 1995. 10W. S. Steffier, “Multilayer Fiber Coating Comprising Alternate Fugitive Carbon and Ceramic Coating Material for Toughened Ceramic Composite Materials,” U.S. Pat. No. 5455106, 1995. 11L. Filipuzzi, G. Camus, R. Naslain, and J. Thebault, “Oxidation Mechanisms and Kinetics of 1-D-SiC/C/SiC Composite Materials: I, An Experimental Approach,” J. Am. Ceram. Soc., 77 [2] 459–66 (1994). 12L. Filipuzzi and R. Naslain, “Oxidation Mechanisms and Kinetics of 1-D-SiC/ C/SiC Composite Materials: II, Modeling,” J. Am. Ceram. Soc., 77 [2] 467–80 (1994). 13R. D. James, R. A. Lowden, and K. L. More, “The Effects of Oxidation and Combustion Environments on the Properties of Nicalon/SiC Composites”; pp. 925–35 in Ceramic Transactions, Vol. 19, Advanced Composite Materials: Processing, Microstructures, Bulk and Interfacial Properties, Characterization Methods, and Applications. Edited by M. D. Sacks. American Ceramic Society, Westerville, OH, 1991. 14P. F. Tortorelli, S. Nijhawan, L. Riester, and R. A. Lowden, “Influence of Fiber Coatings on the Oxidation of Fiber-Reinforced SiC Composites,” Ceram. Eng. Sci. Proc., 14 [7–8] 358–66 (1993). 15R. H. Jones, C. H. Henager Jr., and C. F. Windisch Jr., “High Temperature Corrosion and Crack Growth of SiC–SiC at Variable Oxygen Pressures,” Mater. Sci. Eng. A, 198, 103–12 (1995). 16W. J. Lackey, S. Vaidyaraman, and K. L. More, “Laminated C-SiC Matrix Composites Produced by CVI,” J. Am. Ceram. Soc., 80 [1] 113–16 (1997). 17L. L. Snead, D. Steiner, and S. J. Zinkle, “Measurement of the Effect of Radiation Damage to Ceramic Composite Interfacial Strength,” J. Nucl. Mater., 191 [194] 556–70 (1992). 18K. L. More, E. Lara-Curzio, H. T. Lin, P. J. Tortorelli, R. Shinavsky, and W. S. Steffier, “Multilayered SiC Fiber Coatings in NicalonTM Fiber-Reinforced Ceramic Matrix Composites”; unpublished poster presented at the 20th Annual Cocoa Beach Conference and Exposition on Composites, Advanced Ceramics, Materials and Structures (Cocoa Beach, FL, January 1996). December 2000 Ceramic Composites with Multilayer Interface Coatings 3019
an Ceramic societ VoL. 83. No. 12 (S. Bertrand, O. Boisron,R. Pailler, J. Lamon, and R. Naslain,"(PyC/SiC) and N. P. Bansal and Y. L. Chen, "Chemical, Mechanical, and Microstructural SiCh Nanoscale- Multilayer Interphases by Pressure Pulsed-CVI, Key Eng haracterization of Low-Oxygen Containing Silicon Carbide Fibers with Ceramic Mater.,16415]357-601999) Coatings, "J. Mater. Sci., 33, 5277-89(1998). 2T. M. Besmann, B. W. Sheldon,R. A Lowden, and D P.Stinton,"Vapor-Phase SM. Takeda, J. Sakamoto, A. Saeki, Y Imai, and H. Ichikawa, "High Performance erties of Continuous-Filament Silicon Carbide Fiber Hi-Nicalon for Ceramic Matrix Composites, Ceram. Eng. Sci. 253,1104-109(1991 Proc,l64]3744(1995) R. Naslain, C. Robin-Brosse, S. Gouard, and J. M. Jouin, "The Minicomposite N. Lissart and J. Lamon, "Damage and Failure in Ceramic Matrix Minicompos- Concept in the Field of Ceramic Matrix Composites: Advantages and Limits"; pp ites: Experimental Study and Model, Acta Mater., 45[3] 1025-44( 1997). 198-208 in Progress in Adranced Materials and Mechanics, Edited by w. Tzuchiang 27C. Droillard and J. Lamon, "Fracture Toughness of 2-D Woven SiC/SiC and T. w. Chou. Peking University Press, Peking, China, 1996. CVI-Composites with Multilayer Interphases, R. Naslain, J. Lamon, R. Pailler, X. Bourat, A. Guette, and F. Langlais, (1996 2P. Dupel, J. L. Bobet, Pailler, and J. Lamon, "Influence d' Interphases Pyrocar- Micro/Minicomposites: A Useful Approach to the Design and Development of Non-oxide CMCs, Composites, Part A, 30, 537-47(1999). 3uStandard Test Method for Monotonic Tensile Strength Testing of Continuous 2R. A. Lowden and D. P. Stinton, "Interface Modification in Nicalon/Sic Fiber-Reinforced Advanced Ceramic ross-Section Specimens at Composites, "Ceram. Eng. Sci. Proc., 9[7-8]705-21(1988) Ambient Temperatures, ASTM Standard C1275-95, Anmual Book of ASTM Stan- 3P. F. Tortorelli and K. L. More, "Time Dependence of Oxidation-Induced ds, Vol. 15.01, American Society for Testing and Materials, Conshohocken, PA, microstructural Changes in Nicalon- and Nextel-Reinforced SiC, Cera. Eng. Sci Proc,17366-74(1996)
19S. Bertrand, O. Boisron, R. Pailler, J. Lamon, and R. Naslain, “(PyC/SiC)n and (BN/SiC)n Nanoscale-Multilayer Interphases by Pressure Pulsed-CVI,” Key Eng. Mater., 164 [5] 357–60 (1999). 20T. M. Besmann, B. W. Sheldon, R. A. Lowden, and D. P. Stinton, “Vapor-Phase Fabrication and Properties of Continuous-Filament Ceramic Composites,” Science, 253, 1104–109 (1991). 21R. Naslain, C. Robin-Brosse, S. Goujard, and J. M. Jouin, “The Minicomposite Concept in the Field of Ceramic Matrix Composites: Advantages and Limits”; pp. 198–208 in Progress in Advanced Materials and Mechanics. Edited by W. Tzuchiang and T. W. Chou. Peking University Press, Peking, China, 1996. 22R. Naslain, J. Lamon, R. Pailler, X. Bourat, A. Guette, and F. Langlais, “Micro/Minicomposites: A Useful Approach to the Design and Development of Non-oxide CMCs,” Composites, Part A, 30, 537–47 (1999). 23“Standard Test Method for Monotonic Tensile Strength Testing of Continuous Fiber-Reinforced Advanced Ceramics with Rectangular Cross-Section Specimens at Ambient Temperatures,” ASTM Standard C1275-95, Annual Book of ASTM Standards, Vol. 15.01, American Society for Testing and Materials, Conshohocken, PA, 1995. 24N. P. Bansal and Y. L. Chen, “Chemical, Mechanical, and Microstructural Characterization of Low-Oxygen Containing Silicon Carbide Fibers with Ceramic Coatings,” J. Mater. Sci., 33, 5277–89 (1998). 25M. Takeda, J. Sakamoto, A. Saeki, Y. Imai, and H. Ichikawa, “High Performance Silicon Carbide Fiber Hi-Nicalon for Ceramic Matrix Composites,” Ceram. Eng. Sci. Proc., 16 [4] 37–44 (1995). 26N. Lissart and J. Lamon, “Damage and Failure in Ceramic Matrix Minicomposites: Experimental Study and Model,” Acta Mater., 45 [3] 1025–44 (1997). 27C. Droillard and J. Lamon, “Fracture Toughness of 2-D Woven SiC/SiC CVI-Composites with Multilayer Interphases,” J. Am. Ceram. Soc., 79 [4] 849–58 (1996). 28P. Dupel, J. L. Bobet, Pailler, and J. Lamon, “Influence d’Interphases Pyrocarbone Deposees sur les Caracteristiques Mecaniques de Materiaux Composites Unidirectionnels,” J. Phys. III France, 5, 937–51 (1995). 29R. A. Lowden and D. P. Stinton, “Interface Modification in Nicalon/SiC Composites,” Ceram. Eng. Sci. Proc., 9 [7–8] 705–21 (1988). 30P. F. Tortorelli and K. L. More, “Time Dependence of Oxidation-Induced Microstructural Changes in Nicalon- and Nextel-Reinforced SiC,” Ceram. Eng. Sci. Proc., 17 [4] 366–74 (1996). M 3020 Journal of the American Ceramic Society—Besmann et al. Vol. 83, No. 12