CATERTALIA Pergamon Acta mater.48(2000)46194628 www.elsevier.com/locate/actamat INTERFACIAL CHARACTERIZATION OF A SLURRY-CAST MELT-INFILTRATED SiC/SIC CERAMIC-MATRIX COMPOSITE .. J. breNnaN United Technologies Research Center, East Hartford, CT 06108, USA being developed for combustor applications under the High Speed Civil Transport(HSCT) Enabling Propul on Material(EPM) Program. A major part of this effort has dealt with the characterization and optimizatio cussed in this paper include an overview of the differences in composite microstructure between the EPM SiC/SiC material and a more conventional CVI SiC/SiC composite material, the microstructure/property relationships for the EPM SiC/SiC composite with two different types of Sic fiber(High- Nicalon and Sylramic ), and the effect of moist, high-temperature environments on the stability of the BN interface. o 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved. Keywords: Ceramic composites; Fibers; Interface; Microstructure; Mechanical properties 1 INTRODUCTION SiC-fiber-reinforced SiC matrix offers the best comb The Enabling Propulsion Material(EPM) Program is nation of high-temperature capability and high ther aimed at developing a gas turbine engine combustor mal conductivity. Chemical vapor deposited(CVD) liner so that an environmentally acceptable and econ- SiC has been reported to possess a thermal conduc- omically viable High Speed Civil Transport(HSCT) tivity of up to 325 Wim K [I], which is higher than for some metal superalloys. The most common aircraft can be achieved. One of the prime objectives SiC/SiC CMC system consists of a chemically vapor of this program has been to develop and demonstrat infiltrated(CVI) SiC matrix around a woven SiC-fiber a material system, design concept and manufacturing preform. The resultant microstructure of this CVI process that can meet the HSCT combustor s environ- Sic/Sic composite system, however, is not dense mental, thermal, structural, economic and durability requirements. Among these requirements are an No, instead it contains rather large regions of matrix emissions index <5 g/kg, a material that can with- porosity. This porosity lowers the thermal conduc- stand temperatures to 1200 C under a tensile stress tivity of the composite to an unacceptable level fo of up to 100 MPa, and be able to meet an 18,000 h the HSCT combustor liner. Therefore, it was decided life requirement. The combustor concepts under con- to concentrate SiC/SiC composite developmental sideration do not permit the use of film cooling. efforts on the system that consists of a woven SiC- which traditionally has been used to reduce combus- fiber preform( five-harness satin) that has a BN fiber tor liner temperatures to manageable levels for met- interface coating applied to it by CVI, followed by a allic combustor liners. Therefore. high-thermal-con- relatively thin(1-4 um)CVI coating of SiC. This ductivity materials with a significantly higher relatively porous rigidized preform is then infiltrated temperature capability than those of current metallic with a slurry of SiC particles, dried, and then infil- combustor liners are required for this application. trated with molten silicon metal. The result is a full Ceramic-matrix composites(CMCs)have been ident- dense matrix(except within the fiber tows)of Si/Sic. ified as having the highest potential to satisfy the as shown for the melt infiltration(MI) SiC/SiC com- design requirements and service conditions posite in Fig. 1, which is compared with a typical Of the CMC systems available, the system of an CVI SiC/SiC composite with its inherent matrix porosity. Both composites have a fiber volume frac tion of-0.35. Besides the higher thermal conductivity nnanjj@aol.com(J.J.Brennan)oftheMISiC/SiCcomposite,thelackofmatrix 1359-6454100/520.00@ 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved PI:S1359-6454(00)00248-2
Acta mater. 48 (2000) 4619–4628 www.elsevier.com/locate/actamat INTERFACIAL CHARACTERIZATION OF A SLURRY-CAST MELT-INFILTRATED SiC/SiC CERAMIC-MATRIX COMPOSITE J. J. BRENNAN* United Technologies Research Center, East Hartford, CT 06108, USA Abstract—An SiC-particulate, silicon-metal melt-infiltration-matrix composite reinforced with SiC fibers is being developed for combustor applications under the High Speed Civil Transport (HSCT) Enabling Propulsion Material (EPM) Program. A major part of this effort has dealt with the characterization and optimization of the boron nitride (BN) based fiber/matrix interface. BN was chosen as the primary interfacial material due to its inherently weak structure and thus good crack-deflecting ability, ease of deposition by chemical vapor infiltration (CVI) into woven fiber preforms, and relatively good environmental stability. Topics discussed in this paper include an overview of the differences in composite microstructure between the EPM SiC/SiC material and a more conventional CVI SiC/SiC composite material, the microstructure/property relationships for the EPM SiC/SiC composite with two different types of SiC fiber (High-Nicalon and Sylramic), and the effect of moist, high-temperature environments on the stability of the BN interface. 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved. Keywords: Ceramic composites; Fibers; Interface; Microstructure; Mechanical properties 1. INTRODUCTION The Enabling Propulsion Material (EPM) Program is aimed at developing a gas turbine engine combustor liner so that an environmentally acceptable and economically viable High Speed Civil Transport (HSCT) aircraft can be achieved. One of the prime objectives of this program has been to develop and demonstrate a material system, design concept and manufacturing process that can meet the HSCT combustor’s environmental, thermal, structural, economic and durability requirements. Among these requirements are an NOx emissions index ,5 g/kg, a material that can withstand temperatures to 1200°C under a tensile stress of up to 100 MPa, and be able to meet an 18,000 h life requirement. The combustor concepts under consideration do not permit the use of film cooling, which traditionally has been used to reduce combustor liner temperatures to manageable levels for metallic combustor liners. Therefore, high-thermal-conductivity materials with a significantly higher temperature capability than those of current metallic combustor liners are required for this application. Ceramic-matrix composites (CMCs) have been identified as having the highest potential to satisfy the design requirements and service conditions. Of the CMC systems available, the system of an * E-mail address: Brennanjj@aol.com (J.J. Brennan) 1359-6454/00/$20.00 2000 Acta Metallurgica Inc. Published by Elsevier Science Ltd. All rights reserved. PII: S13 59-6454(00)00248-2 SiC-fiber-reinforced SiC matrix offers the best combination of high-temperature capability and high thermal conductivity. Chemical vapor deposited (CVD) SiC has been reported to possess a thermal conductivity of up to 325 W/m K [1], which is higher than for some metal superalloys. The most common SiC/SiC CMC system consists of a chemically vapor infiltrated (CVI) SiC matrix around a woven SiC-fiber preform. The resultant microstructure of this CVI SiC/SiC composite system, however, is not dense; instead it contains rather large regions of matrix porosity. This porosity lowers the thermal conductivity of the composite to an unacceptable level for the HSCT combustor liner. Therefore, it was decided to concentrate SiC/SiC composite developmental efforts on the system that consists of a woven SiC- fiber preform (five-harness satin) that has a BN fiberinterface coating applied to it by CVI, followed by a relatively thin (1–4 µm) CVI coating of SiC. This relatively porous rigidized preform is then infiltrated with a slurry of SiC particles, dried, and then infiltrated with molten silicon metal. The result is a fully dense matrix (except within the fiber tows) of Si/SiC, as shown for the melt infiltration (MI) SiC/SiC composite in Fig. 1, which is compared with a typical CVI SiC/SiC composite with its inherent matrix porosity. Both composites have a fiber volume fraction of |0.35. Besides the higher thermal conductivity of the MI SiC/SiC composite, the lack of matrix
BRENNAN: INTERFACIAL CHARACTERIZATION Environmeatal degradation dimcult 2 Fig. I Microstructural differences between CVI and MI SiC/SiC composites porosity also leads to better environmental stability SiC fiber from Dow Corning Corp, Midland, MI and a higher proportional limit (matrix Both of these fibers have very low oxygen content, microcracking)stress. Figure 2 shows a thin-foil with the Hi-Nicalon fiber being carbon-rich while the transmission electron micrograph of an MI composite Sylramic fiber is very close to stoichiometic SiC ith Sylramic SiC fibers. The crystalline nature of Since the Sylramic fiber consists of 100-500 nm crys the fibers and the feathery CvI SiC layer can be seen tals of B-SiC, while the Hi-Nicalon fiber is a mixture clearly, in contrast to the more amorphous to turbos- of microcrystalline(2-20 nm) B-Sic plus turbostratic tratic nature of the bn interface coating domains of carbon, the Sylramic fiber has properties Once the fabrication method of the composite was much like those of sintered SiC; i.e., an elastic modu hosen, developmental efforts concentrated on the lus of -385 GPa and a thermal expansion coefficient type of SiC fiber to be utilized, the thickness and of -54x10-o/C [2]. The elastic modulus, therma chemistry of the BN fiber/matrix interface, and the expansion coefficient and thermal conductivity of the thermo-mechanical properties of the composite. Two Hi-Nicalon fiber are much lower than those of the different SiC fibers that were commercially available Sylramic fiber [2]. The following discussion covers were selected as the fiber candidates: Hi-Nicalon Sic the experimental results obtained for composites with fiber from Nippon Carbon Co., Japan, and Sylramic these two fibers, the rationale for the selection of th Sylramic fiber as the primary candidate, and the results of environmentally sensitive tests relating to the bn fiber/matrix interf 2. EXPERIMENTAl 2. 1. MI SiC/SiC composite fabrication The MI SiC/SiC composites utilized in the EPM program were initially fabricated by Carborundum nc, Niagara Falls, NY, and later in the program by REaver Allied Signal Composites, Inc, now called Ho Iss:ic Matris i. well Advanced Composites, Inc (HACD), Newark DE. The fabrication of these composites consisted of oating a two-dimensional(2D)woven SiC-fiber pre- form, usually five- or eight-harness satin, with an nterface coating of CVI BN, followed by an over- coating of CVI SiC. The rigidized preform was then subjected to infiltration by an SiC particulate slurry in order to fill the large residual porosity with SiC, followed by a melt infiltration of silicon metal which Fig. 2. TEM thin-foil analysis of the microstructure of a Syl. would fill the fine porosity left between the SiC grain particles. The fabrication and properties of similar MI
4620 BRENNAN: INTERFACIAL CHARACTERIZATION Fig. 1. Microstructural differences between CVI and MI SiC/SiC composites. porosity also leads to better environmental stability and a higher proportional limit (matrix microcracking) stress. Figure 2 shows a thin-foil transmission electron micrograph of an MI composite with Sylramic SiC fibers. The crystalline nature of the fibers and the feathery CVI SiC layer can be seen clearly, in contrast to the more amorphous to turbostratic nature of the BN interface coating. Once the fabrication method of the composite was chosen, developmental efforts concentrated on the type of SiC fiber to be utilized, the thickness and chemistry of the BN fiber/matrix interface, and the thermo-mechanical properties of the composite. Two different SiC fibers that were commercially available were selected as the fiber candidates: Hi-Nicalon SiC fiber from Nippon Carbon Co., Japan, and Sylramic Fig. 2. TEM thin-foil analysis of the microstructure of a Sylramic SiC fiber MI SiC/SiC composite. SiC fiber from Dow Corning Corp., Midland, MI. Both of these fibers have very low oxygen content, with the Hi-Nicalon fiber being carbon-rich while the Sylramic fiber is very close to stoichiometic SiC. Since the Sylramic fiber consists of 100–500 nm crystals of β-SiC, while the Hi-Nicalon fiber is a mixture of microcrystalline (2-20 nm) β-SiC plus turbostratic domains of carbon, the Sylramic fiber has properties much like those of sintered SiC; i.e., an elastic modulus of |385 GPa and a thermal expansion coefficient of |5.4×1026 /°C [2]. The elastic modulus, thermal expansion coefficient and thermal conductivity of the Hi-Nicalon fiber are much lower than those of the Sylramic fiber [2]. The following discussion covers the experimental results obtained for composites with these two fibers, the rationale for the selection of the Sylramic fiber as the primary candidate, and the results of environmentally sensitive tests relating to the BN fiber/matrix interface. 2. EXPERIMENTAL 2.1. MI SiC/SiC composite fabrication The MI SiC/SiC composites utilized in the EPM program were initially fabricated by Carborundum, Inc., Niagara Falls, NY, and later in the program by Allied Signal Composites, Inc., now called Honeywell Advanced Composites, Inc. (HACI), Newark, DE. The fabrication of these composites consisted of coating a two-dimensional (2D) woven SiC-fiber preform, usually five- or eight-harness satin, with an interface coating of CVI BN, followed by an overcoating of CVI SiC. The rigidized preform was then subjected to infiltration by an SiC particulate slurry in order to fill the large residual porosity with SiC, followed by a melt infiltration of silicon metal which would fill the fine porosity left between the SiC grain particles. The fabrication and properties of similar MI
BRENNAN: INTERFACIAL CHARACTERIZATION SiC/SiC composites have been presented previously frictional sliding and pull-out of the fibers-which 3-9 contribute to the composite toughness [ 121being more extensive for the hi-Nicalon fibers than for the 2.2. Composite testing Sylramic fibers. The reason for this may be related to Tensile testing was performed in accordance with the higher elastic modulus of the Sylramic fiber, or, High-Speed Research/Enabling Propulsion Materials more likely, the much higher surface roughness of the (HSR/EPM) consensus standard. The testing method Sylramic fiber compared with the Hi-Nicalon fiber,as was similar to ASTM standard C-1275-95 [10]. Ten- shown in Fig. 4. The surface roughness of the two sile specimen design was a contoured, face-loaded fibers is a reflection of the relative Sic grain size of specimen geometry with an overall length of 152 mm the fibers Fibers with higher surface roughness have and a gage-section width of 10.16 mm. The tensile been found to have onounced influence on the specimen design used a gradual radius from the tab fiber sliding behavior in ceramic-matrix composites to the gage section to reduce stress concentrations to [131 avoid grip and tab failures[11]. All of the mechanical While the Hi-Nicalon MI SiC/SiC composite sys property definitions can be found in the ASTM stan- tem appears to have better toughness than the Syl- dard. One of the most crucial and subjective measure ments is the proportional limit. The proportional limit well in high-temperature fatigue testing. As shown in was determined using the offset method. A line, run- Fig. 5, a Hi-Nicalon fiber composite failed after 38h ning parallel to the elastic modulus slope, was gener- and 19 cycles in a 2 h hold-time tensile fatigue test ated at a strain axis offset of 0.00005 mm/mm. The at 1200oC and a stress of 160 MPa, whereas a Syl proportional limit was defined as the stress level at ramic fiber composite ran out to 1000 h, 500 cycles, which point the offset line intersected the load versus under the same test conditions. The stress level during elongation curve. A servo-hydraulic machine was this test was chosen to be somewhat higher than the used for the vast majority of tensile testing. Testing proportional limit, or matrix microcrack stress of the from room temperature to elevated temperature was composites, which at this temperature is-140 MPa done in the same rig and differed only by the presence for the Sylramic fiber composite and -125 MPa for of a small furnace that heated the sample gage plus the Hi-Nicalon fiber composite. The residual 1200C radius region to within 1% of the desired temperature. tensile properties of the fatigued Sylramic fiber com Fatigue testing was conducted at elevated tempera- posite are not significantly different from those meas- ture in load control. Testing was done in a programm as-fabricated Sylramic fiber composite able servo-hydraulic machine. The testing consisted From the fracture surfaces shown in Fig. >1.fatigued an of a 2 h dwell fatigue test with the load being relieved seen that the amount of fiber pull-out for to 5% of the hold load at the end of each cycle. This Sylramic fiber composite is similar to that shown in simulated service conditions for the proposed appli- Fig 3 for the 1200C tensile sample, but the fatigued cation. Most tests were run to failure or 500 h of Hi-Nicalon composite shows a region around the fatigue exposure. All samples that made 500 h of edges of the sample that appears very brittle exposure were then tested for residual tensile proper- This embrittlement of the Hi-Nicalon fiber com- es at room temperature to note if degradation had posite is even more apparent after tensile fatigue test occurred. A selected number of fatigue tests was run ing at a lower temperature of 650%C, as shown in Fig to many thousands of hours before failure occurred, 6. While both composites ran out to the test limit of as will be described later 546h, 273 cycles, the residual 650C strength and strain-to-failure of the Hi-Nicalon fiber composite were significantly degraded, while those for the Syl- 3. RESULTS AND DISCUSSION ramic fiber composite were not 3. 1. Fracture characteristics of Hi-Nicalon versus 3. 2. Fiber/matrix interfacial characteristics of Hi- Sylramic SiC fiber MI composites Nicalon versus Sylramic SiC fiber MI composites Figure 3 shows the fracture surfaces of 1200C ten- From transmission electron microscopy (TEM) sile samples of MI SiC/SiC composites with either thin-foil analysis of the interfacial region in an as- Sylramic or Hi-Nicalon fibers. From this figure, one processed Hi-Nicalon fiber MI composite, as show can see that the fracture surface of the hi-nicalon Fig. 7, it was found that the probable reason fo fiber composite is much more fibrous in nature than the easy debonding and long fiber pull-out in these that of the Sylramic fiber composite. The measured composites, as was shown in Fig. 3, is that a thin tensile strengths of the two composites are similar, at (40 nm) carbon-rich layer has formed between th 286 MPa; however, the strain-to-failure of the Hi- BN and the Hi-Nicalon fiber during the MI composite Nicalon fiber composite(0.48%) is over twice that processing. This carbon-rich layer is probably a result of the Sylramic fiber composite(0.21%). Thus, the of interaction of the oxygen in the bn layer "toughness", or inherent matrix crack-stopping (-6 at%)with the excess carbon in the Hi-Nicalon ability, of the Hi-Nicalon fiber composite is greater fiber at the MI composite processing temperature of an that of the Sylramic fiber composite, due to the >1400C. Thicker, but similar, carbon layers have
BRENNAN: INTERFACIAL CHARACTERIZATION 4621 SiC/SiC composites have been presented previously [3–9]. 2.2. Composite testing Tensile testing was performed in accordance with High-Speed Research/Enabling Propulsion Materials (HSR/EPM) consensus standard. The testing method was similar to ASTM standard C-1275-95 [10]. Tensile specimen design was a contoured, face-loaded specimen geometry with an overall length of 152 mm and a gage-section width of 10.16 mm. The tensile specimen design used a gradual radius from the tab to the gage section to reduce stress concentrations to avoid grip and tab failures [11]. All of the mechanical property definitions can be found in the ASTM standard. One of the most crucial and subjective measurements is the proportional limit. The proportional limit was determined using the offset method. A line, running parallel to the elastic modulus slope, was generated at a strain axis offset of 0.00005 mm/mm. The proportional limit was defined as the stress level at which point the offset line intersected the load versus elongation curve. A servo-hydraulic machine was used for the vast majority of tensile testing. Testing from room temperature to elevated temperature was done in the same rig and differed only by the presence of a small furnace that heated the sample gage plus radius region to within 1% of the desired temperature. Fatigue testing was conducted at elevated temperature in load control. Testing was done in a programmable servo-hydraulic machine. The testing consisted of a 2 h dwell fatigue test with the load being relieved to 5% of the hold load at the end of each cycle. This simulated service conditions for the proposed application. Most tests were run to failure or 500 h of fatigue exposure. All samples that made 500 h of exposure were then tested for residual tensile properties at room temperature to note if degradation had occurred. A selected number of fatigue tests was run to many thousands of hours before failure occurred, as will be described later. 3. RESULTS AND DISCUSSION 3.1. Fracture characteristics of Hi-Nicalon versus Sylramic SiC fiber MI composites Figure 3 shows the fracture surfaces of 1200°C tensile samples of MI SiC/SiC composites with either Sylramic or Hi-Nicalon fibers. From this figure, one can see that the fracture surface of the Hi-Nicalon fiber composite is much more fibrous in nature than that of the Sylramic fiber composite. The measured tensile strengths of the two composites are similar, at |286 MPa; however, the strain-to-failure of the HiNicalon fiber composite (0.48%) is over twice that of the Sylramic fiber composite (0.21%). Thus, the “toughness”, or inherent matrix crack-stopping ability, of the Hi-Nicalon fiber composite is greater than that of the Sylramic fiber composite, due to the frictional sliding and pull-out of the fibers—which contribute to the composite toughness [12]—being more extensive for the Hi-Nicalon fibers than for the Sylramic fibers. The reason for this may be related to the higher elastic modulus of the Sylramic fiber, or, more likely, the much higher surface roughness of the Sylramic fiber compared with the Hi-Nicalon fiber, as shown in Fig. 4. The surface roughness of the two fibers is a reflection of the relative SiC grain size of the fibers. Fibers with higher surface roughness have been found to have a pronounced influence on the fiber sliding behavior in ceramic-matrix composites [13]. While the Hi-Nicalon MI SiC/SiC composite system appears to have better toughness than the Sylramic fiber composite system, it does not perform well in high-temperature fatigue testing. As shown in Fig. 5, a Hi-Nicalon fiber composite failed after 38 h and 19 cycles in a 2 h hold-time tensile fatigue test at 1200°C and a stress of 160 MPa, whereas a Sylramic fiber composite ran out to 1000 h, 500 cycles, under the same test conditions. The stress level during this test was chosen to be somewhat higher than the proportional limit, or matrix microcrack stress of the composites, which at this temperature is |140 MPa for the Sylramic fiber composite and |125 MPa for the Hi-Nicalon fiber composite. The residual 1200°C tensile properties of the fatigued Sylramic fiber composite are not significantly different from those measured for an as-fabricated Sylramic fiber composite. From the fracture surfaces shown in Fig. 5, it can be seen that the amount of fiber pull-out for the fatigued Sylramic fiber composite is similar to that shown in Fig. 3 for the 1200°C tensile sample, but the fatigued Hi-Nicalon composite shows a region around the edges of the sample that appears very brittle. This embrittlement of the Hi-Nicalon fiber composite is even more apparent after tensile fatigue testing at a lower temperature of 650°C, as shown in Fig. 6. While both composites ran out to the test limit of 546 h, 273 cycles, the residual 650°C strength and strain-to-failure of the Hi-Nicalon fiber composite were significantly degraded, while those for the Sylramic fiber composite were not. 3.2. Fiber/matrix interfacial characteristics of HiNicalon versus Sylramic SiC fiber MI composites From transmission electron microscopy (TEM) thin-foil analysis of the interfacial region in an asprocessed Hi-Nicalon fiber MI composite, as shown in Fig. 7, it was found that the probable reason for the easy debonding and long fiber pull-out in these composites, as was shown in Fig. 3, is that a thin (|40 nm) carbon-rich layer has formed between the BN and the Hi-Nicalon fiber during the MI composite processing. This carbon-rich layer is probably a result of interaction of the oxygen in the BN layer (|6 at%) with the excess carbon in the Hi-Nicalon fiber at the MI composite processing temperature of >1400°C. Thicker, but similar, carbon layers have
BRENNAN: INTERFACIAL CHARACTERIZATION SylramicTM 5HS 1200°CUTs=288MPa,Ef=0.48% 1200CUTS=286MPa,f=0.21 UTRC Fig 3. Fracture surface comparison between MI SiC/SiC composites with Hi-Nicalon and Sylramic SiC fibers Hi-Nicalon Fiber Sylramie M Fiber 1331510,8010u05 18.Bk I RMS 2.137 nM I 17,561 Fig 4. Fiber surface roughness differences between Sylramic and Hi-Nicalon SiC fibers
4622 BRENNAN: INTERFACIAL CHARACTERIZATION Fig. 3. Fracture surface comparison between MI SiC/SiC composites with Hi-Nicalon and Sylramic SiC fibers. Fig. 4. Fiber surface roughness differences between Sylramic and Hi-Nicalon SiC fibers
BRENNAN: INTERFACIAL CHARACTERIZATION Hi-Nicalon 5HS SyIramicTM 5HS 1200"C, 160 MPa, 19 cycles, 38 hrs to failure 1200.C, 160 MPa, 501 cycles, 1002 hrs to runout (Residual 1200C UTS=248 MPa, Ef=0.21%) Fig. 5. Fracture surface comparison between Hi-Nicalon and Sylramic SiC fiber MI SiC/SiC composites after 200° C tensile fatigue been found in many ceramic-matrix composites con- much less fibrous and with much shorter fiber pull taining the oxygen-rich Nicalon SiC fibers [14]. This out. However, the mode of fracture does not change carbon-rich layer is not seen in CVI SiC-matrix com- after elevated-temperature fatigue testing, nor does posites with Hi-Nicalon fibers and BN interfaces, the residual tensile strength change significantly which are processed at much lower temperatures TEM thin-foil analysis was performed on both as-fab- (1000oC) than the MI composites ricated and high-temperature-fatigued Sylramic fiber TEM thin-foil analysis of the Hi-Nicalon fiber MI composites, with the results indicating that no carbon composite that was subjected to tensile fatigue at rich layer forms during composite processing, nor 650C, as was shown in Fig. 6, and resulted in a quite does a glassy silica layer form as a result of high weak and brittle composite, found that a glassy silica temperature fatigue testing. This is illustrated in Fig layer had replaced the carbon-rich layer between the 9 for the Sylramic fiber composite sample that was BN and Hi-Nicalon fiber surface. This glassy silica subjected to 650C tensile fatigue testineoon-nich layer, as shown in Fig. 8, appears to be a result of shown previously in Fig. 6. The lack of a car oxidation of the carbon-rich layer and then the Hi- layer next to fiber surface is undoubtedly due to the Nicalon fiber surface due to matrix cracks forming higher temperature stability of the stoichiometric, during the fatigue testing. This glassy layer appar- crystalline SiC Sylramic fiber, when compared with ently bonds the Bn strongly to the fiber and may the carbon-rich Hi-Nicalon SiC fiber, During fatigue weaken the Hi-Nicalon fiber itself. At higher tempera- at elevated temperatures, any matrix cracks that may tures, such as that experienced during the 1200oC form do not cause pipeline oxidation down the fatigue test, this glassy layer can actually start to con- fiber/BN interface without the presence of the carbon sume the BN layer, totally bonding up the fiber/matrix rich layer interface, as was seen around the periphery of the At high stresses during high-temperature tensile 1200 C fatigue fracture surface shown in Fig. 5 fatigue, cracks do form in the Sylramic fiber MI com- As shown previously, in contrast to the fibrous fast posites, as shown for a sample in Fig. 10 that was fracture surface of Hi-Nicalon fiber MI composites, tensile fatigued at 815C at a stress of 186 MPa, the fracture surface of Sylramic fiber composites is which is well above the matrix microcracking stress
BRENNAN: INTERFACIAL CHARACTERIZATION 4623 Fig. 5. Fracture surface comparison between Hi-Nicalon and Sylramic SiC fiber MI SiC/SiC composites after 1200°C tensile fatigue. been found in many ceramic-matrix composites containing the oxygen-rich Nicalon SiC fibers [14]. This carbon-rich layer is not seen in CVI SiC-matrix composites with Hi-Nicalon fibers and BN interfaces, which are processed at much lower temperatures (|1000°C) than the MI composites. TEM thin-foil analysis of the Hi-Nicalon fiber MI composite that was subjected to tensile fatigue at 650°C, as was shown in Fig. 6, and resulted in a quite weak and brittle composite, found that a glassy silica layer had replaced the carbon-rich layer between the BN and Hi-Nicalon fiber surface. This glassy silica layer, as shown in Fig. 8, appears to be a result of oxidation of the carbon-rich layer and then the HiNicalon fiber surface due to matrix cracks forming during the fatigue testing. This glassy layer apparently bonds the BN strongly to the fiber and may weaken the Hi-Nicalon fiber itself. At higher temperatures, such as that experienced during the 1200°C fatigue test, this glassy layer can actually start to consume the BN layer, totally bonding up the fiber/matrix interface, as was seen around the periphery of the 1200°C fatigue fracture surface shown in Fig. 5. As shown previously, in contrast to the fibrous fast fracture surface of Hi-Nicalon fiber MI composites, the fracture surface of Sylramic fiber composites is much less fibrous and with much shorter fiber pullout. However, the mode of fracture does not change after elevated-temperature fatigue testing, nor does the residual tensile strength change significantly. TEM thin-foil analysis was performed on both as-fabricated and high-temperature-fatigued Sylramic fiber composites, with the results indicating that no carbonrich layer forms during composite processing, nor does a glassy silica layer form as a result of hightemperature fatigue testing. This is illustrated in Fig. 9 for the Sylramic fiber composite sample that was subjected to 650°C tensile fatigue testing, as was shown previously in Fig. 6. The lack of a carbon-rich layer next to fiber surface is undoubtedly due to the higher temperature stability of the stoichiometric, crystalline SiC Sylramic fiber, when compared with the carbon-rich Hi-Nicalon SiC fiber. During fatigue at elevated temperatures, any matrix cracks that may form do not cause pipeline oxidation down the fiber/BN interface without the presence of the carbonrich layer. At high stresses during high-temperature tensile fatigue, cracks do form in the Sylramic fiber MI composites, as shown for a sample in Fig. 10 that was tensile fatigued at 815°C at a stress of 186 MPa, which is well above the matrix microcracking stress
BRENNAN: INTERFACIAL CHARACTERIZATION Hi-Nicalon 5HS SyIramieTM 5HS 650C, 160 MPa, 272 cycles, 545 hrs to runout 650C, 160 MPa, 273 cycles, 46 hrs to runout Residual 650C UTS=172 MPa, Ef=0.08%) (Residual 650C UTS=318 MPa, ef=0.23%) UTRC Fig. 6. Fracture surface comparison between Hi-Nicalon and Sylramic SiC fiber MI SiC/SiC composites after 650° tensile fatigue Carbon-Rich (near fiber Si Ire of Hi-Nicalon fiber/BN/MI SiC/SIC composite interfacial region(as-processed) of 140 MPa. This sample failed after 154 h(76 path, and not down the fiber/BN/matrix interface. At cycles), with the primary fracture surface s below the matrix microcracking stress, no some effects of oxidation, but the secondary are formed in the composite during testing, and such as that shown in Fig. 10 showing limit nly oxidation that occurs is confined to the sur- dation of the BN fiber coating only along the crack face of the composite. A Sylramic fiber MI composite
4624 BRENNAN: INTERFACIAL CHARACTERIZATION Fig. 6. Fracture surface comparison between Hi-Nicalon and Sylramic SiC fiber MI SiC/SiC composites after 650°C tensile fatigue. Fig. 7. TEM microstructure of Hi-Nicalon fiber/BN/MI SiC/SiC composite interfacial region (as-processed). of |140 MPa. This sample failed after 154 h (76 cycles), with the primary fracture surface showing some effects of oxidation, but the secondary cracks such as that shown in Fig. 10 showing limited oxidation of the BN fiber coating only along the crack path, and not down the fiber/BN/matrix interface. At stresses below the matrix microcracking stress, no cracks are formed in the composite during testing, and the only oxidation that occurs is confined to the surface of the composite. A Sylramic fiber MI composite
BRENNAN: INTERFACIAL CHARACTERIZATION Hi-Nicalon Fiber BN Layer Fig. 8. TEM microstructure of Hi-Nicalon fiber/BN/MI SiC/SiC composite interfacial region(after 650C that deal with the Sylramic fiber/BN interfacial region. These two issues related to the roughness of the surface of the Sylramic fiber and its effect on composite properties, and the high-temperature moist ure susceptibility of the BN interface On occasion, composites tested under this prog exhibited rather weak and brittle tensile strengths, ompared with the usual composite properties. From BN Layer scanning electron microscopy( SEM)examination of the fracture surfaces of these composites it appeared that the Sylramic SiC fibers in the weak and brittle composites were not as dense as those in the strong and tough composites. On further examination of composite cross-sections, as shown in Fig. Il, the cvisc porosity in the fibers was found to actually consist of raphite particles that were concentrated in the center of the fibers This is an indication that the cessing of these particular fibers from their carbon Fig 9.TEM microstructure of Sylramic fiber/BN/MI SiC/SiC rich precursor did not proceed to completion. How ever. since tensile tests on these fibers indicated that they were almost as strong as fibers without the graphite inclusions, another reason for the poor com was fatigued at 1200C, 117 MPa stress, for a total posite properties must be present. From atomic force of 9870 h before failure occurred. No oxidation of microscopy(AFM) of the fiber surfaces, it was found the fibers or BN interface was found except in a that the graphite-containing Sylramic SiC fibers had region within -50 um of the tensile sample surface. much rougher surface topography [root-mean square 3.3. Issues of concern with Sylramic Sic fiber/BN roughness(RMS)-34 nm] than the dense fibers nterface MI SiC/SiC composites (RMS-10 nm), as shown in Fig. 12. It was thus con- luded that the rougher surface of these fibers made From the results discussed above. and others relat- it much more difficult for these fibers to debond and ing to thermal conductivity of the composites, the then pull out of the matrix compared with the system of Sylramic SiC fiber/BN interface/MI smoother fibers; i.e., the interfacial sliding stress was SiC/SiC composite was selected for further develop- too high [13]. The manufacturer of the Sylramic SiC ment under the HSCT/EPM program. During this fibers, Dow Corning Corp, took steps during sub- development phase, two issues of concern were raised sequent fiber processing to ensure that the sintering
BRENNAN: INTERFACIAL CHARACTERIZATION 4625 Fig. 8. TEM microstructure of Hi-Nicalon fiber/BN/MI SiC/SiC composite interfacial region (after 650°C tensile fatigue). Fig. 9. TEM microstructure of Sylramic fiber/BN/MI SiC/SiC composite interfacial region (after 650°C, 159 MPa tensile fatigue). was fatigued at 1200°C, 117 MPa stress, for a total of 9870 h before failure occurred. No oxidation of the fibers or BN interface was found except in a region within |50 µm of the tensile sample surface. 3.3. Issues of concern with Sylramic SiC fiber/BN interface MI SiC/SiC composites From the results discussed above, and others relating to thermal conductivity of the composites, the system of Sylramic SiC fiber/BN interface/MI SiC/SiC composite was selected for further development under the HSCT/EPM program. During this development phase, two issues of concern were raised that deal with the Sylramic fiber/BN interfacial region. These two issues related to the roughness of the surface of the Sylramic fiber and its effect on composite properties, and the high-temperature moisture susceptibility of the BN interface. On occasion, composites tested under this program exhibited rather weak and brittle tensile strengths, compared with the usual composite properties. From scanning electron microscopy (SEM) examination of the fracture surfaces of these composites it appeared that the Sylramic SiC fibers in the weak and brittle composites were not as dense as those in the strong and tough composites. On further examination of composite cross-sections, as shown in Fig. 11, the porosity in the fibers was found to actually consist of graphite particles that were concentrated in the center region of the fibers. This is an indication that the processing of these particular fibers from their carbonrich precursor did not proceed to completion. However, since tensile tests on these fibers indicated that they were almost as strong as fibers without the graphite inclusions, another reason for the poor composite properties must be present. From atomic force microscopy (AFM) of the fiber surfaces, it was found that the graphite-containing Sylramic SiC fibers had much rougher surface topography [root-mean square roughness (RMS)|34 nm] than the dense fibers (RMS|10 nm), as shown in Fig. 12. It was thus concluded that the rougher surface of these fibers made it much more difficult for these fibers to debond and then pull out of the matrix compared with the smoother fibers; i.e., the interfacial sliding stress was too high [13]. The manufacturer of the Sylramic SiC fibers, Dow Corning Corp., took steps during subsequent fiber processing to ensure that the sintering
BRENNAN: INTERFACIAL CHARACTERIZATION BN=B Fig 10 Crack propagation and subsequent oxidation for 815C tensile fatigue of Sylramic fiber MI SiC/SiC composite above the matrix microcrack stress(186 MPa, 54 h to failure) tn entire to Porosity concentrated in center of fibers Porosity is actually free carbon Fig. 11. Microstructural analysis of MI SiC/SiC composite with porous Sylramic SiC fibers(porosity actually of the Sylramic fibers went to completion. This cern was the potential moisture susceptibility of the appeared to solve the problem of rough surface Syl- BN interface between the Sylramic SiC fibers and the ramic fibers leading to poor composite properties MI SiC matrix at elevated temperatures. High-tem Since the goal of the HSCT/EPM SiC/SiC com- perature gas turbine combustor environments can posite program was the development of a high-tem- contain very high moisture contents under very high rature gas turbine combustor liner result of the fuel/air combustio
4626 BRENNAN: INTERFACIAL CHARACTERIZATION Fig. 10. Crack propagation and subsequent oxidation for 815°C tensile fatigue of Sylramic fiber MI SiC/SiC composite above the matrix microcrack stress (186 MPa, 54 h to failure). Fig. 11. Microstructural analysis of MI SiC/SiC composite with porous Sylramic SiC fibers (porosity actually free carbon). of the Sylramic fibers went to completion. This appeared to solve the problem of rough surface Sylramic fibers leading to poor composite properties. Since the goal of the HSCT/EPM SiC/SiC composite program was the development of a high-temperature gas turbine combustor liner, an area of concern was the potential moisture susceptibility of the BN interface between the Sylramic SiC fibers and the MI SiC matrix at elevated temperatures. High-temperature gas turbine combustor environments can contain very high moisture contents under very high pressures as a result of the fuel/air combustion pro-
BRENNAN: INTERFACIAL CHARACTERIZATION 4627 Dense Fiber Porous Fiber rv Fig. 12. AFM surface roughness analysis of dense versus porous Sylramic SiC fibers(porous fibers exhibit much rougher surfaces) cess. From tests conducted on minicomposites with Hi-Nicalon sic fibers and cvi SiC matrices with a number of different bn interfaces at 700-800oC in furnaces with water contents up to 90%, it has been found that accelerated oxidation and recession of the BN can occur [15]. Therefore, a series of furnace tests and tensile fatigue tests was conducted with the Syl ramic SiC fiber/BN interface MI SiC/SiC composites nder moist, high-temperature environments Figure 13 shows the fracture surface of an MI SiC/SiC composite after tensile fatigue testing at 760 C under a maximum stress of 1 17 MPa in a 90% 10. 0 ky 10um steam atmosphere. The sample failed after 44 h(22 cycles), which is significantly less than the thousands of hours to failure in a normal air environment. as was discussed earlier. From Fig. 13. it can be seen that the Bn fiber/matrix interfacial coating is entirel oxidized away, due to the volatile HBO phases that can form in high-temperature moist environments [15]. In order to prevent this, experiments have begun to investigate the moisture stability of Si-doped BN since this approach was found to result in the forma- nantly silica glass instead of boria, which tended to seal the interphase and resulted in minimal BN interphase recession [ 15]. Early results obtained on the Si-doped BN interface approach have Fig. 13. Fracture surface of an MI SiC/SiC indicated that the high-temperature moist environ- tensile fatigue testing at 760"C in a 90% steam atmosphere mental stability of the MI SiC/SiC composites under (BN interfacial layer oxidized ). fatigue test conditions is improved for atmospheres with relatively low moisture content(20%H]O), but not significantly improved for atmospheres with high 4. CONCLUSIONS moisture content(90% H_). Additional work on Under the HSCT/EPM ceramic composite combus environmentally stable interfaces for MI SiC/SiC tor program, a slurry-cast, Si-melt-infiltrated SiC opposites is recommended. fiber-reinforced Si/SiC matrix composite with a BN
BRENNAN: INTERFACIAL CHARACTERIZATION 4627 Fig. 12. AFM surface roughness analysis of dense versus porous Sylramic SiC fibers (porous fibers exhibit much rougher surfaces). cess. From tests conducted on minicomposites with Hi-Nicalon SiC fibers and CVI SiC matrices with a number of different BN interfaces at 700–800°C in furnaces with water contents up to 90%, it has been found that accelerated oxidation and recession of the BN can occur [15]. Therefore, a series of furnace tests and tensile fatigue tests was conducted with the Sylramic SiC fiber/BN interface MI SiC/SiC composites under moist, high-temperature environments. Figure 13 shows the fracture surface of an MI SiC/SiC composite after tensile fatigue testing at 760°C under a maximum stress of 117 MPa in a 90% steam atmosphere. The sample failed after 44 h (22 cycles), which is significantly less than the thousands of hours to failure in a normal air environment, as was discussed earlier. From Fig. 13, it can be seen that the BN fiber/matrix interfacial coating is entirely oxidized away, due to the volatile HBO phases that can form in high-temperature moist environments [15]. In order to prevent this, experiments have begun to investigate the moisture stability of Si-doped BN, since this approach was found to result in the formation of predominantly silica glass instead of boria, which tended to seal the interphase and resulted in minimal BN interphase recession [15]. Early results obtained on the Si-doped BN interface approach have indicated that the high-temperature moist environmental stability of the MI SiC/SiC composites under fatigue test conditions is improved for atmospheres with relatively low moisture content (20% H2O), but not significantly improved for atmospheres with high moisture content (90% H2O). Additional work on environmentally stable interfaces for MI SiC/SiC composites is recommended. Fig. 13. Fracture surface of an MI SiC/SiC composite after tensile fatigue testing at 760°C in a 90% steam atmosphere (BN interfacial layer oxidized). 4. CONCLUSIONS Under the HSCT/EPM ceramic composite combustor program, a slurry-cast, Si-melt-infiltrated SiC- fiber-reinforced Si/SiC matrix composite with a BN
BRENNAN: INTERFACIAL CHARACTERIZATION interface was developed that exhibits improved ther. saep tes anca s pnost anin g i he god se er this roeraney SiC/SiC composite for advanced gas turbine combus- tor applications. The fiber selected, Sylramic crys REFERENCES talline stoichiometric SiC. was found to exhibit improved thermal and environmental stability in the I. Wang, H, Singh, R and Goela,JJ.Am. CeramSoc. melt-infiltrated(MD)composites compared with the 2. Lipowitz, I. Rabe, I. Zangvil, A. and Xu, Y. Ceram. Eng carbon-rich Hi-Nicalon type SiC fiber. The instability Sci.Proc,l997,18(3),147 of the Hi-Nicalon fiber in the MI composite was 3. Srinivasan, G, Venkateswaran, V and Lau, S.Ceram of a thin, carbon-rich cial layer between the fiber and the BN fiber coating 4. Xu Y, Cheng, L. and Zhang, L. Carbon, 1999,37(8) during composite fabrication which, on subsequent 5.Luthra, K Singh, R and Brun, M. Am. Ceram Soc. Bull. exposure in oxidizing environment under stress, oxid- 993,72(7),79 ized away leading to silica formation and a strong 6. Luthra, K, Singh, R an M interfacial bonding state. The high proportional lim dings of the 6th Europe or matrix microcracking stress, of the Sylramic SiC 0-27 September 1993, ed. R. Naslain, J. Lamon and D fiber MI composites results in a high tensile fatigue Doumeing, Woodhead Publishing Ltd, Cambridge, UK, limit with very little or no degradation in composite 993,p.429 properties. 7. Corman, G, Brun, M, Meschter, P and Luthra, K. in Pr Issues still of concern for use of these composites ceedings of the 39th International SAMPE Symposium, Society for the Advancement of Materials and Process in advanced gas turbine combustor applications include the occasional weak and brittle composite due 8. Corman, G, Heinen, J and Goetz, R, ASME Paper 95. to incompletely sintered Sylramic SiC fibers witl G7-387. Presented at the 40th ASME International Gas rougher than normal surfaces, and the degradation of ne Congress and Exposition, Hous- composite properties in hot moist environments due 9. Corman, G, Brun, M. and Luthra, K ASME Paper 99. to accelerated oxidation of the bn fiber/matrix inter GT-234. Presented at the 44th ASME Gas Turbine and face. While the former concern has been reduced by Aeroengine Technical Congress, Exposition and Users changes in the processing of the Sylramic SiC fibers Symposium, Indianapolis, IN,7-10 June 1999 which ensure that sintering goes to completion, the 10. Anmual Book of ASTM Standards Section 15, vol. 15.01 atter concern has only been partially alleviated by adelphia, PA, 2000, p. 333 doping the Bn with Si. Additional work is necessary 11. Wortham, D. NASA CR-185261, National Aeronautics and to solve the high-temperature moisture degradation of Space Administration(NASA), Washington, DC, 1990 the Bn interface in these composites 2. Weihs, T, Sbaizero, O, Luh, E. and Nix, W. J.Am Ceram. Soc.. 1991. 74. 535 13. Liu, H, Zhou, L. and Mai, YJ.Am. oC,1995, acknowled pon work supported by Nas d like to 14. Cooper, R and Chyung, K.J. Mater. So 2.3148 wn and R. Wong of 15. Morscher, G, Bryant, D and Tressler, R ng. Sci. UTRC fo of the composite Proc., 1997, 18(3,525
4628 BRENNAN: INTERFACIAL CHARACTERIZATION interface was developed that exhibits improved thermal and mechanical properties over conventional CVI SiC/SiC composite for advanced gas turbine combustor applications. The fiber selected, Sylramic crystalline stoichiometric SiC, was found to exhibit improved thermal and environmental stability in the melt-infiltrated (MI) composites compared with the carbon-rich Hi-Nicalon type SiC fiber. The instability of the Hi-Nicalon fiber in the MI composite was related to the formation of a thin, carbon-rich interfacial layer between the fiber and the BN fiber coating during composite fabrication which, on subsequent exposure in oxidizing environment under stress, oxidized away leading to silica formation and a strong interfacial bonding state. The high proportional limit, or matrix microcracking stress, of the Sylramic SiC fiber MI composites results in a high tensile fatigue limit with very little or no degradation in composite properties. Issues still of concern for use of these composites in advanced gas turbine combustor applications include the occasional weak and brittle composite due to incompletely sintered Sylramic SiC fibers with rougher than normal surfaces, and the degradation of composite properties in hot moist environments due to accelerated oxidation of the BN fiber/matrix interface. While the former concern has been reduced by changes in the processing of the Sylramic SiC fibers which ensure that sintering goes to completion, the latter concern has only been partially alleviated by doping the BN with Si. Additional work is necessary to solve the high-temperature moisture degradation of the BN interface in these composites. Acknowledgements—This paper is based upon work supported by NASA contract NAS3-26385. The author would like to thank G. McCarthy, B. Laube, B. Brown and R. Wong of UTRC for the microstructural analyses of the composite samples, and G. Linsey and G. Ojard of Pratt & Whitney for their technical support during the course of this program. REFERENCES 1. Wang, H., Singh, R. and Goela, J. J. Am. Ceram. Soc., 1995, 78, 2437. 2. Lipowitz, J., Rabe, J., Zangvil, A. and Xu, Y. Ceram. Eng. Sci. Proc., 1997, 18(3), 147. 3. Srinivasan, G., Venkateswaran, V. and Lau, S. Ceram. Eng. Sci. Proc., 1995, 16(4), 63. 4. Xu, Y., Cheng, L. and Zhang, L. Carbon, 1999, 37(8), 1179. 5. Luthra, K., Singh, R. and Brun, M. Am. Ceram. Soc. Bull., 1993, 72(7), 79. 6. Luthra, K., Singh, R. and Brun, M. in High Temperature Ceramic Composites, Proceedings of the 6th European Conference on Composite Materials, Bordeaux, France, 20–27 September 1993, ed. R. Naslain, J. Lamon and D. Doumeing, Woodhead Publishing Ltd, Cambridge, UK, 1993, p. 429. 7. Corman, G., Brun, M., Meschter, P. and Luthra, K. in Proceedings of the 39th International SAMPE Symposium, Society for the Advancement of Materials and Process Engineering, Covina, CA, 1994, p. 2300. 8. Corman, G., Heinen, J. and Goetz, R., ASME Paper 95- GT-387. Presented at the 40th ASME International Gas Turbine and Aeroengine Congress and Exposition, Houston, TX, 5–8 June 1995. 9. Corman, G., Brun, M. and Luthra, K., ASME Paper 99- GT-234. Presented at the 44th ASME Gas Turbine and Aeroengine Technical Congress, Exposition and Users Symposium, Indianapolis, IN, 7–10 June 1999. 10. Annual Book of ASTM Standards Section 15, vol. 15.01. American Society for Testing and Matetrials (ASTM), Philadelphia, PA, 2000, p. 333. 11. Wortham, D. NASA CR-185261, National Aeronautics and Space Administration (NASA), Washington, DC, 1990. 12. Weihs, T., Sbaizero, O., Luh, E. and Nix, W. J. Am. Ceram. Soc., 1991, 74, 535. 13. Liu, H., Zhou, L. and Mai, Y. J. Am. Ceram. Soc., 1995, 78, 560. 14. Cooper, R. and Chyung, K. J. Mater. Sci., 1987, 22, 3148. 15. Morscher, G., Bryant, D. and Tressler, R. Ceram. Eng. Sci. Proc., 1997, 18(3), 525