《复合材料 Composites》课程教学资源（学习资料）第二章 增强体_nicalon-2 Tensile Creep Behavior of SiC-Based Fibers With a Low Oxygen Content

J.Am. Ceran.Soe,9l46-115602007) Dol:10.111.1551-29162007.01535x C 2007 The American Ceramic Society urna Tensile Creep Behavior of SiC-Based Fibers With a Low Oxygen Content Cedric Sauder'and Jacques Lamon Laboratoire des Composites Thermostructuraux, UMR 5801: CNRS-Snecma-CEA-UBl, 33600 Pessac, France Commissariat a Energie atomique, DEC/SPUA/LMPC, 13108 Saint Paul les Durance, france The creep behavior of Hi-Nicalon, Hi-Nicalon S, and Tyranno During most of the tensile creep tests reported in the litera- SA3 fibers is investigated at temperatures up to 1700.C. Tensile ture, the fiber was not at a uniform temperature( cold-gripping tests were carried out on a high-capability fiber testing appara- method). Furthermore, the test duration rarely exceeds 48 h, for tus in which the fiber is heated uniformly under vacuum. Anal- practical reasons associated with the design of the experimental sis of initial microstructure and composition of fibers wa performed using various techniques. All the fibers experienced The cold grip-based technique presents a few important draw- a steady-state creep. Primary creep was found to be more or less backs. Fiber specimens are generally quite long (length averages ignificant depending on fiber microstructure Steady-state creep 100 mm) and there is a significant temperature gradient. Owing was shown to result from grain-boundary sliding. Activation en- to the temperature gradient, determination of creep strain from ergy and stress exponents were determined. Creep mechanisms fiber deformation is not straightforward. Heavy and tedious re discussed on the basis of activation energy and stress expo- calibration operations are required. Then, the use of long spec- nent data. Finally, tertiary creep was observed at very high imens is not recommended because of fiber diameter variation temperatures. Tertiary creep was related to volatilization of along the gauge. Specimen length must be selected with respect Sic. Results are discussed with respect to fiber microstructure. to characteristic diameter wave length 2.0 is about 160 mm for Tyranno SA and Hi-Nicalon fibers. Specimens with a uni- form diameter along the length can be obtained when the fiber gauge is significantly shorter than 2. L. Introduction In the hot grip-based technique, short specimens are used and the entire fiber can be at a uniform temperature. Some authors composites are designed to be used at high tempera- claim that the fiber may degrade due to the cement used for ures in various systems, including aerojet engines and gripping fiber ends. This difficulty can be overcome with recent stationary gas turbines for electrical power/steam cogeneration products. Furthermore, the results obtained g the hot grip- Furthermore, owing to the stability of SiC under neutron based technique appeared to be consistent with other available irradiation and to recent progress in the fabrication of stoichic data. 3 tes are candidates for nuclear Thus, in the present paper, some effort was directed toward applications, such as the structural component facing the the testing method, in order to overcome the difficulties associ- on fiber reinforcement, and more particularly on the ss depi plasma in fusion reactor blankets, or the control rod in tI ated with the cold grip-based technique, and to produce valu- Generation IV nuclear power plants able creep data on the last generation of Sic-based fibers. The structural performances of SiC/SiC composite of mechanical properties of fibers to temperature and environ IL. Fibers and Experimental Procedure ment. The present paper focuses on the creep behavior of Sic based fibers with a low oxygen content in an inert atmosphere. Description ofFibers These fibers are potential candidates for reinforcement of Sic Hi-Nicalon and Hi-Nicalon S(Nippon Carbon Co., Tokyo Sic composites for nuclear applications. Japan) and Tyranno SA3 (Ube Industry Ltd, Tokyo Several papers on the creep behavior of ceramic fibers apan) Sic-based fibers were investigated ( Table I). Two are available in the literature. Data have been produced different batches of Tyranno SA3 fibers were tested (they are on SiC-based fibers(see for instance Yu and colleagues). referred to as SA3(I)and SA3(2). Sylramic fibers were not The authors mainly used uniaxial tensile loading condition considered in this study because they contain boron, which or a qualitative technique such as bending stress relaxation. But only of these paper terested in the irradiation II mechanisms.4.6. itative of fibe d structure Testing tiny objects such as small-diameter ceramic fibers(the were performed using X-ray diffraction(XRD), Raman diameter may be as small as 10 um) at high temperatures for oscopy, transmission electron microscopy (TEM), electron long times is not straightforward. The results and analyses may for TEM were prepared following the method proposed by robe microanalysis(EPMA), and fractography. Specimens be biased as a result of the testing conditions. The testing meth- od thus warrants consideration in order to produce valuable Berger and Bunsel (2) Creep Tests The fiber samples were taken from tows(gauge length 25 mm) Graphite grips were affixed to sample ends using carbon-based cement C34 (from UCAR Co., Graftech International Ltd No. 22022. Received July 17, 2006: approved November 27, 2006. Parma, OH) rk was supported by CNRS and CEA and was accomplished as part of the CPR The creep tests were performed on a tensile device arch program. correspondence should be addressed. e-mail: lamon(@ Icts. designed for testing carbon fibers at temperature 3000C.Heating is generated by an 114

Tensile Creep Behavior of SiC-Based Fibers With a Low Oxygen Content Ce´dric Sauder* and Jacques Lamonw Laboratoire des Composites Thermostructuraux, UMR 5801: CNRS–Snecma–CEA–UB1, 33600 Pessac, France *Commissariat a` l’e´nergie atomique, DEC/SPUA/LMPC, 13108 Saint Paul les Durance, France The creep behavior of Hi-Nicalon, Hi-Nicalon S, and Tyranno SA3 fibers is investigated at temperatures up to 17001C. Tensile tests were carried out on a high-capability fiber testing appara￾tus in which the fiber is heated uniformly under vacuum. Anal￾ysis of initial microstructure and composition of fibers was performed using various techniques. All the fibers experienced a steady-state creep. Primary creep was found to be more or less significant depending on fiber microstructure. Steady-state creep was shown to result from grain-boundary sliding. Activation en￾ergy and stress exponents were determined. Creep mechanisms are discussed on the basis of activation energy and stress expo￾nent data. Finally, tertiary creep was observed at very high temperatures. Tertiary creep was related to volatilization of SiC. Results are discussed with respect to fiber microstructure. I. Introduction SIC/SIC composites are designed to be used at high tempera￾tures in various systems, including aerojet engines and stationary gas turbines for electrical power/steam cogeneration. Furthermore, owing to the stability of SiC under neutron irradiation and to recent progress in the fabrication of stoichio￾metric fibers, SiC/SiC composites are candidates for nuclear applications, such as the structural component facing the plasma in fusion reactor blankets,1,2 or the control rod in the Generation IV nuclear power plants. The structural performances of SiC/SiC composites depend on fiber reinforcement, and more particularly on the sensitivity of mechanical properties of fibers to temperature and environ￾ment. The present paper focuses on the creep behavior of SiC￾based fibers with a low oxygen content in an inert atmosphere. These fibers are potential candidates for reinforcement of SiC/ SiC composites for nuclear applications. Several papers on the creep behavior of ceramic fibers are available in the literature. Data have been produced on SiC-based fibers (see for instance Yu and colleagues3–8). The authors mainly used uniaxial tensile loading conditions, or a qualitative technique such as bending stress relaxation. But only some of these papers were interested in the creep mechanisms.4,6,8 Testing tiny objects such as small-diameter ceramic fibers (the diameter may be as small as 10 mm) at high temperatures for long times is not straightforward. The results and analyses may be biased as a result of the testing conditions. The testing meth￾od thus warrants consideration in order to produce valuable results. During most of the tensile creep tests reported in the litera￾ture, the fiber was not at a uniform temperature (cold-gripping method). Furthermore, the test duration rarely exceeds 48 h, for practical reasons associated with the design of the experimental setup. The cold grip-based technique presents a few important draw￾backs. Fiber specimens are generally quite long (length averages 100 mm) and there is a significant temperature gradient. Owing to the temperature gradient, determination of creep strain from fiber deformation is not straightforward.9 Heavy and tedious calibration operations are required. Then, the use of long spec￾imens is not recommended because of fiber diameter variation along the gauge. Specimen length must be selected with respect to characteristic diameter wave length l. 10 l is about 160 mm for Tyranno SA and Hi-Nicalon fibers.10 Specimens with a uni￾form diameter along the length can be obtained when the fiber gauge is significantly shorter than l. In the hot grip-based technique, short specimens are used and the entire fiber can be at a uniform temperature. Some authors claim that the fiber may degrade due to the cement used for gripping fiber ends. This difficulty can be overcome with recent products. Furthermore, the results obtained using the hot grip￾based technique appeared to be consistent with other available data.3 Thus, in the present paper, some effort was directed toward the testing method, in order to overcome the difficulties associ￾ated with the cold grip-based technique, and to produce valu￾able creep data on the last generation of SiC-based fibers. II. Fibers and Experimental Procedure (1) Description of Fibers Hi-Nicalon and Hi-Nicalon S (Nippon Carbon Co., Tokyo, Japan) and Tyranno SA3 (Ube Industry Ltd., Tokyo, Japan) SiC-based fibers were investigated (Table I). Two different batches of Tyranno SA3 fibers were tested (they are referred to as SA3 (1) and SA3 (2)). Sylramic fibers were not considered in this study because they contain boron, which makes them sensitive to significant degradation under neutron irradiation.11 Quantitative analyses of fibers composition and structure were performed using X-ray diffraction (XRD), Raman spec￾troscopy, transmission electron microscopy (TEM), electron probe microanalysis (EPMA), and fractography. Specimens for TEM were prepared following the method proposed by Berger and Bunsell.12 (2) Creep Tests The fiber samples were taken from tows (gauge length 25 mm). Graphite grips were affixed to sample ends using carbon-based cement C34 (from UCAR Co., Graftech International Ltd., Parma, OH). The creep tests were performed on a tensile device (Fig. 1) designed for testing carbon fibers at temperatures up to 30001C.13 Heating is generated by an electric current F. Wakai—contributing editor This work was supported by CNRS and CEA and was accomplished as part of the CPR ISMIR research program. w Author to whom correspondence should be addressed. e-mail: lamon@lcts. u-bordeaux1.fr Manuscript No. 22022. Received July 17, 2006; approved November 27, 2006. Journal J. Am. Ceram. Soc., 90 [4] 1146–1156 (2007) DOI: 10.1111/j.1551-2916.2007.01535.x r 2007 The American Ceramic Society 1146

April 2007 Sic-Based Fibers and Low Oxygen Conten 1147 Table I. Properties of Sic Fibers Investigated in the Present Study ippon Carbon Co., Japan Ube Industries Ltd, Japan Hi-Nicalon Hi-Nicalon S 43(2) Batch no 225103 320203 M-010071 Diameter (um) Density(g/cm 3.0 3.l Tensile strength(GPa) 2.516 Tensile modulus(GPa) Grains size(nm) X-ray diffraction 5-10 60-70 60-70 Transmission electron microscopy 50400 Chemical composition(wt%/at. % 621/41.3 84/48.l 666/46.0(edge) 69.1/49(edge) 0.3/395(core) 66.1/45.6(core) 37.7/58.5 0.5/50.7(edge) 39.2/60.l(core) 33.5/54.l(core) 1. 16(edg 1.03(edge) .52(c 1.19(core) Fig 1. Schematic diagram of the high-temperature fiber-testing apparatus. circulating through the fiber, under secondary vacuum(residual The fiber was first kept stress-free at the test temperature for pressure <*Pa). In such an environment, active oxidation is 30 min. Then, the stress was applied. This loading step took less infinitively slow. The temperature of the fiber was measured than 10 s. The diameter of each fiber was measured in situ using using a bichromatic pyrometer (IrCON. Niles, IL). The tem- a laser mounted on the testing apparatus. It is given by the av- perature profiles showed that the temperature is uniform over erage of several measurements along the gauge length t To en- more than 95% of the gauge length. Furthermore, it appeared sure a good reproducibility of the results, only those specimens hat grips remained at a temperature close to room temperature with quite uniform diameters along the gauge were tested For during the tests(<50.C). Thus, the loading frame compliance these specimens, the diameters measured along the fiber differed was not affected during the tests fiber deformations can be from the average by <3% derived from grip displacement. Loading frame compliance was Fiber deformations were derived from grip displac taken to be equal to that estimated at room temperature. Com- Data were corrected to account for deformation of the putations of temperature distributions for various thermal con- frame. The loading frame compliance was estimated us ductivities showed that the temperature gradient from the core to the surface of the fiber is<2°at1000°C13 'The cross sections of fibers are circular. The diameter is variable along the gauge

circulating through the fiber, under secondary vacuum (residual pressure B10
4 Pa). In such an environment, active oxidation is infinitively slow.14 The temperature of the fiber was measured using a bichromatic pyrometer (IRCON, Niles, IL). The tem￾perature profiles showed that the temperature is uniform over more than 95% of the gauge length.13 Furthermore, it appeared that grips remained at a temperature close to room temperature during the tests (o501C). Thus, the loading frame compliance was not affected during the tests. Fiber deformations can be derived from grip displacement. Loading frame compliance was taken to be equal to that estimated at room temperature. Com￾putations of temperature distributions for various thermal con￾ductivities showed that the temperature gradient from the core to the surface of the fiber is o21C at 10001C.13 The fiber was first kept stress-free at the test temperature for 30 min. Then, the stress was applied. This loading step took less than 10 s. The diameter of each fiber was measured in situ using a laser mounted on the testing apparatus. It is given by the av￾erage of several measurements along the gauge length.z To en￾sure a good reproducibility of the results, only those specimens with quite uniform diameters along the gauge were tested. For these specimens, the diameters measured along the fiber differed from the average by o3%. Fiber deformations were derived from grip displacement. Data were corrected to account for deformation of the loading frame. The loading frame compliance was estimated using the Quartz chamber x y Camera z Mirror Light Laser y z Mirror Mirror Vacuum captor displacement table Pyrometer Fig. 1. Schematic diagram of the high-temperature fiber-testing apparatus. Table I. Properties of SiC Fibers Investigated in the Present Study Suppliers Nippon Carbon Co., Japan Ube Industries Ltd., Japan Type of fiber Hi-Nicalon Hi-Nicalon S Tyranno SA3 (1) Tyranno SA3 (2) Batch no. 225103 320203 M-0110071 M-0304041 Diameter (mm) 1416 1316 7.5 7.2 Density (g/cm3 ) 2.7416 3.016 3.0 3.1 Tensile strength (GPa) 2.816 2.516 2.816 2.8 Tensile modulus (GPa) 290 375 325 380 Grains size (nm) X-ray diffraction 5–10 20 60–70 60–70 Transmission electron microscopy 5–10 10–50 50–400 50–400 Chemical composition (wt%/at.%) Si 62.1/41.3 68.4/48.1 66.6/46.0 (edge) 69.1/49 (edge) 60.3/39.5 (core) 66.1/45.6 (core) C 37.7/58.5 31.3/51.5 33/53.6 (edge) 30.5/50.7 (edge) 39.2/60.1 (core) 33.5/54.1 (core) O 0.2/0.2 0.3/0.3 0.2/0.2 0.1/0.1 Al — — 0.3/0.2 0.3/0.2 C/Si (at.%) 1.41 1.07 1.16 (edge) 1.03 (edge) 1.52 (core) 1.19 (core) z The cross sections of fibers are circular. The diameter is variable along the gauge. April 2007 SiC-Based Fibers and Low Oxygen Content 1147

1148 Journal of the American Ceramic Society--Sauder and Lamon Vol. 90. No. 4 G-Nicalon Hi-Nicalon S 1.4 1.4 。t。↑。:0.8 Cs-o 0.6 04 42 7=65-4-3=2-101234567 76=5-4-3=2-101 (c) 040+- 08茜 0.6a 0.6a Fig. 2. Si, C, O, and Al atomic concentrations along the diameter of(a) Hi-Nicalon(b), Hi-Nicalon S(c), Tyranno SA3(1)and (d), Tyranno SA3(2) fiber as measured by electron probe microanalyst conventional calibration technique based on tensile tests on fi- Hi-Nicalon S is made up of clusters of Sic grains(Fig bers having various gauge lengths. As indicated above, as the Grain size averages 20 nm (Table D). The largest grains were 50 grips remained at a temperature close to room temperature dur m in size. Carbon is located between the SiC grains(Fig 4) ing the tests, the loading frame compliance estimated at room Grain boundaries do not appear clearly(Fig 4) The concentration in C and Si was not found to be uniform in Most of the creep tests were range Is the sa3 fibers(Fig. 2). There is a larger amount of free carbon for analysis of crept fibers. Stresses in the 50-850MPa resent in the core. The sa3(1)fiber contained a larger amount were applied, whereas the temperatures hel150° of free carbon when compared with more recent SA3(2)(Table 1700C range. Tests were performed for as long as 350 h in D). Elemental composition in SA3 (2) is closer to stoichiomet order to identify the different creep stages suggesting that fibers of this second batch have been improved the sa3()fibers and 100 nm in the sA3(2)ones. Aluminum II. Results and discussion was identified (Table D). According to Ishikawa, Al aggregates (1) Structure and Composition of Fibers The grain size is much larger when compared with Hi-Nica Table i summarizes the results of microstructural lon and Hi-Nicalon S fibers. a difference in the grain size can be All the fibers contain a small amount of oxygen (0.2%) noted from the micrographs shown in Fig. 5. The size of B-sic Higher oxygen contents were reported for Hi-Nicalon fibers rains averaged 200 nm (table D). The largest grains were 400 (0.6-0.9w/o) nm in size. The grains displayed stack faults(Fig. 3). This ex- There is a larger amount of free carbon in the hi-nicalo lains the discrepancy in grain sizes estimated using XRD and fiber when compared with Hi-Nicalon S. Hi-Nicalon S is stoi TEM (Table I). Grain size was larger from the core to the sur- chiometric, but the results indicate an excess of carbon. Figures face of the fibers. As opposed to Hi-Nicalon S fibers, grain 2(a)and(b) show that the element concentration is uniform in boundaries are clearly marked(Fig 4). Carbon shows a turbos- both fibers. However, a carbon-rich phase was detected using tratic structure. It is located between B-Sic grains(Fig 4) EPMA. It was located in the surface. over a thickness 100 nm. ed,6., Hi-Nicalon fiber microstructure is well document Thus, data from the literature can be reported here. (2) Creep behavior faulted.Grain size averaged 5 nm(Table 1). The largest grains Steady-state creep was observed after a more or less long a 9 Hi-Nicalon fibers consist of fine p-SiC grains, which may be The typical creep curves that were obtained are shown in Fig. were 20 nm in size (Table D). The carbon phase(turbostratic mary creep stage, depending on the fiber: after about 140 h for carbon) consists of distorted stacks of 5-10 graphitic planes, 2-5 Hi-Nicalon fibers at 1200C, about 72 h for SA3 (1)fiber at 200C, about 8 h for SA3(2)fiber at 1250C, and about h for

conventional calibration technique based on tensile tests on fi- bers having various gauge lengths.15 As indicated above, as the grips remained at a temperature close to room temperature dur￾ing the tests, the loading frame compliance estimated at room temperature was pertinent. Most of the creep tests were interrupted before fiber failure, for analysis of crept fibers. Stresses in the range 150–850 MPa were applied, whereas the temperatures were in the 11501– 17001C range. Tests were performed for as long as 350 h in order to identify the different creep stages. III. Results and Discussion (1) Structure and Composition of Fibers Table I summarizes the results of microstructural analyses. All the fibers contain a small amount of oxygen (0.2%). Higher oxygen contents were reported for Hi-Nicalon fibers6,16 (0.6–0.9 w/o). There is a larger amount of free carbon in the Hi-Nicalon fiber when compared with Hi-Nicalon S. Hi-Nicalon S is stoi￾chiometric, but the results indicate an excess of carbon. Figures 2(a) and (b) show that the element concentration is uniform in both fibers. However, a carbon-rich phase was detected using EPMA. It was located in the surface, over a thickness o100 nm. The Hi-Nicalon fiber microstructure is well document￾ed.6,17,18 Thus, data from the literature can be reported here. Hi-Nicalon fibers consist of fine b-SiC grains, which may be faulted.18 Grain size averaged 5 nm (Table I). The largest grains were 20 nm in size (Table I). The carbon phase (turbostratic carbon) consists of distorted stacks of 5–10 graphitic planes, 2–5 nm long. Hi-Nicalon S is made up of clusters of SiC grains (Fig. 3). Grain size averages 20 nm (Table I). The largest grains were 50 nm in size. Carbon is located between the SiC grains (Fig. 4). Grain boundaries do not appear clearly (Fig. 4). The concentration in C and Si was not found to be uniform in the SA3 fibers (Fig. 2). There is a larger amount of free carbon present in the core. The SA3 (1) fiber contained a larger amount of free carbon when compared with more recent SA3 (2) (Table I). Elemental composition in SA3 (2) is closer to stoichiometry, suggesting that fibers of this second batch have been improved. A carbon-rich phase was detected on the surface, over 300 nm in the SA3 (1) fibers and 100 nm in the SA3 (2) ones. Aluminum was identified (Table I). According to Ishikawa,19 Al aggregates at grain boundaries. The grain size is much larger when compared with Hi-Nica￾lon and Hi-Nicalon S fibers. A difference in the grain size can be noted from the micrographs shown in Fig. 5. The size of b-SiC grains averaged 200 nm (Table I). The largest grains were 400 nm in size. The grains displayed stack faults (Fig. 3). This ex￾plains the discrepancy in grain sizes estimated using XRD and TEM (Table I). Grain size was larger from the core to the sur￾face of the fibers. As opposed to Hi-Nicalon S fibers, grain boundaries are clearly marked (Fig. 4). Carbon shows a turbos￾tratic structure. It is located between b-SiC grains (Fig. 4). (2) Creep Behavior The typical creep curves that were obtained are shown in Fig. 6. Steady-state creep was observed after a more or less long pri￾mary creep stage, depending on the fiber: after about 140 h for Hi-Nicalon fibers at 12001C, about 72 h for SA3 (1) fiber at 12001C, about 8 h for SA3(2) fiber at 12501C, and about 8 h for 0 10 20 30 40 50 60 70 –7 –6 –5 –4 –3 –2 –1 x (µm) –7 –6 –5 –4 –4 –3 –3 –2 –2 –1 –1 0 0 1 1 2 2 3 3 4 4 0 12 34 5 6 7 567 x (µm) x (µm) –4 –3 –2 –1 0 1 2 3 4 x (µm) at. % (C et Si) 0 10 20 30 40 50 60 70 at. % (C et Si) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 at. % (O) at. % (C, Si) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 at. % (O et AI) at. % (O et AI) C Si O 0 0.2 0.4 0.6 0.8 1 1.2 1.4 at. % (O) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 C Si O 0 10 20 30 40 50 60 70 at. % (C, Si) C Si O Al 0 10 20 30 40 50 60 70 C Si O Al Hi-Nicalon Hi-Nicalon S (a) (b) (c) (d) SA3(1) SA3(2) Fig. 2. Si, C, O, and Al atomic concentrations along the diameter of (a) Hi-Nicalon (b), Hi-Nicalon S (c), Tyranno SA3(1) and (d), Tyranno SA3 (2) fiber as measured by electron probe microanalysis. 1148 Journal of the American Ceramic Society—Sauder and Lamon Vol. 90, No. 4

April 2007 Sic-Based Fibers and Low Oxygen Conten 1149 (b) Hi-Nicalon S Hi-Nicalon S at 1350C. The creep results reported by most au- thors were obtained during much shorter tests(1600.C(tertiary creep). Creep curves were fitted by the following well-accepted equations of deformations in the primary and in the secondary stages. Tertiary creep is examined in a subsequent section (1) Ep=oA[1-exp(pr) Es= Bot where subscripts e, p, and s refer, respectively, to elastic regime primary, and secondary creep. o is the stress on the fiber, Ec the initial fiber Youngs modulus, I is time, and A, B, n, and p re constants Figure 6 shows that an excellent agreement was obtained for all the fibers. Note that Hi-Nicalon s and sa3 fibers are less 500nm 50nm Based on microstructure analysis, the fibers can be considered SC(251A) to be a mixture of wrinkled carbon-layer packets and Sic grains a possible controlling creep mechanism may involve grain- boundary sliding, carbon diffusion, dewrinkling, deform and sliding of carbon crystallites.6 (3) Creep Mechanisms-Primary Creep Primary creep can be attributed to viscoelastic deformation of carbon at grain boundaries. The viscoelasticity of carbon has SiCu(131A) been discussed by Kelly20 and it has been observed by Sauder et al. on various carbon fibers at high temperatures. Because Fig 3. Microstructure and electron diffraction pattern of (a) Tyranno of the very weak interaction between layer planes, each basal SA3 (2)and(b) Hi-Nicalon S fiber(effective beam size =2. 15 um). plane can deform as a separate unit in two dimensions, which Hi-Nicalon S sA3(2) ig. 4. Lattice fringe images showing the presence of turbostratic carbon at the Sic grain boundary for(a) Hi-Nicalon S and(b) Tyranno SA3(2)fiber

Hi-Nicalon S at 13501C. The creep results reported by most au￾thors were obtained during much shorter tests (o48 h). Thus, it may be anticipated that their tests were not sufficiently long, so that true secondary creep stage was probably not reached. Figure 7 shows a typical creep curve obtained at incremental temperature steps. It can be noted that creep accelerated at tem￾peratures 416001C (tertiary creep). Creep curves were fitted by the following well-accepted equations of deformations in the primary and in the secondary stages. Tertiary creep is examined in a subsequent section: ee ¼ s Eo (1) ep ¼ sA½1
exp ð
ptÞ (2) es ¼ Bsn t (3) e ¼ ee þ ep þ es (4) where subscripts e, p, and s refer, respectively, to elastic regime, primary, and secondary creep. s is the stress on the fiber, Eo is the initial fiber Young’s modulus, t is time, and A, B, n, and p are constants. Figure 6 shows that an excellent agreement was obtained for all the fibers. Note that Hi-Nicalon S and SA3 fibers are less sensitive to creep than Hi-Nicalon. Based on microstructure analysis, the fibers can be considered to be a mixture of wrinkled carbon-layer packets and SiC grains. A possible controlling creep mechanism may involve grain￾boundary sliding, carbon diffusion, dewrinkling, deformation, and sliding of carbon crystallites.6 (3) Creep Mechanisms—Primary Creep Primary creep can be attributed to viscoelastic deformation of carbon at grain boundaries. The viscoelasticity of carbon has been discussed by Kelly20 and it has been observed by Sauder et al. 21 on various carbon fibers at high temperatures. Because of the very weak interaction between layer planes, each basal plane can deform as a separate unit in two dimensions, which carbon carbon SiC 10 nm 7 nm SiC SiC SiC SiC SA3(2) (a) (b) Hi-Nicalon S Fig. 4. Lattice fringe images showing the presence of turbostratic carbon at the SiC grain boundary for (a) Hi-Nicalon S and (b) Tyranno SA3 (2) fiber. Fig. 3. Microstructure and electron diffraction pattern of (a) Tyranno SA3 (2) and (b) Hi-Nicalon S fiber (effective beam size 5 2.15 mm). April 2007 SiC-Based Fibers and Low Oxygen Content 1149

l150 Journal of the American Ceramic Society--Sauder and Lamon Vol. 90. No. 4 sA3(1) (2) Fig. 5. Scanning electron microscope micrographs of the cross sections of (a) Hi-Nicalon. (b)Hi-Nicalon S(c) Tyranno SA3(1), and(d) Tyranno SA3 produces substantial basal plane shear. Furthermore, the mag- It is worth mentioning that primary creep was more signifi nitude of the viscoelastic response of carbon fibers under tension cant in those fibers that contained a large amount of carbon(Hi- depends on the orientation of the graphitic planes. It increases Nicalon). This supports the above carbon deformation-driven with the fraction of graphitic planes with a large angle to loading mechanism direction(isotropic carbon). By contrast, it is limited in the ani Although the Hi-Nicalon and SA3 (1) fibers possessed the sotropic fibers, in which most of the graphitic planes are orient same fraction of carbon, Hi-Nicalon fiber showed a larger sen- ed parallel to the loading direction. In SiC fibers, the orientation sitivity to creep. This discrepancy can be attributed to grain size of graphitic planes is influenced by grain-boundary distribution. which is much smaller in the Hi- Nicalon fiber. It could also Thus, graphitic planes can take on all orientations. Further- be related to the structure of carbon, which displayed a better more, it was indicated earlier that the carbon present in these rganization in the SA3(1) fiber (smaller distance between iC fibers consists of stacks of a few graphitic planes. As a con- two successive layers: doo), as a result of a higher-processing sequence, primary creep may involve deformation of carbon at temperature. A low doog implies a larger stiffness and smaller grain boundaries and grain sliding due to basal plane shear deformations

produces substantial basal plane shear.20 Furthermore, the mag￾nitude of the viscoelastic response of carbon fibers under tension depends on the orientation of the graphitic planes.21 It increases with the fraction of graphitic planes with a large angle to loading direction (isotropic carbon). By contrast, it is limited in the ani￾sotropic fibers, in which most of the graphitic planes are orient￾ed parallel to the loading direction. In SiC fibers, the orientation of graphitic planes is influenced by grain-boundary distribution. Thus, graphitic planes can take on all orientations. Further￾more, it was indicated earlier that the carbon present in these SiC fibers consists of stacks of a few graphitic planes. As a con￾sequence, primary creep may involve deformation of carbon at grain boundaries and grain sliding due to basal plane shear. It is worth mentioning that primary creep was more signifi- cant in those fibers that contained a large amount of carbon (Hi￾Nicalon). This supports the above carbon deformation-driven mechanism. Although the Hi-Nicalon and SA3 (1) fibers possessed the same fraction of carbon, Hi-Nicalon fiber showed a larger sen￾sitivity to creep. This discrepancy can be attributed to grain size, which is much smaller in the Hi-Nicalon fiber. It could also be related to the structure of carbon, which displayed a better organization in the SA3 (1) fiber (smaller distance between two successive layers: d002), as a result of a higher-processing temperature. A low d002 implies a larger stiffness and smaller deformations. (a) Hi-Nicalon (b) Hi-Nicalon S SA3 (1) SA3 (2) (c) (d) Zoom Fig. 5. Scanning electron microscope micrographs of the cross sections of (a) Hi-Nicalon, (b) Hi-Nicalon S, (c) Tyranno SA3 (1), and (d) Tyranno SA3 (2) fibers. 1150 Journal of the American Ceramic Society—Sauder and Lamon Vol. 90, No. 4

April 2007 SiC-Based Fibers and Low Oxygen Conten 1151 Hi-Nicalot sA3(1) wii E=9.7×10-9s-1 80.5+- E=3.8×10s 15 4 T=1200°C T=1200°c g=850 MPa d= 500 MPa 04896144192240288336 04896144192240288336 Time(h) sA3(2) Hi-Nicalon S experiment 0.6 0.5 0.8 0.4 E=13×10-8s-1 0.6 --=325×10-8s-1 T=1350°C d=850 MP 0.1 d=850 MPa 0.0 Time(h) Time(h) Fig. 6. Creep behavior of (a)Hi-Nicalon. (b) Tyranno SA3(1),(c)SA3(2), and(d) Hi-Nicalon S fibers The Hi-Nicalon fiber contains a significant amount o (4) Secondary Creep: Mechanism nsity SiC. This SiC not stable. as a result of the duration of pyrolysi during fiber pi rowth thus starts at I in the Hi-Nicalon fiber. It causes primary creep stage. Extensive investigation of Hi-Nicalon a decrease in creep rate. This phenomenon may explain the im- would have required long experiments, as indicated above perfect fit of the experimental creep curve by Eqs.(1H4) Creep rate in the steady state is expressed by the following (Fig. 6(a)) general relationship: 8= Dde o"with D=D。exp (是) where is a material constant, D is the diffusivity, Q is the ap- parent activation energy, R is the universal gas constant, Tis temperature in Kelvin, de is the grain size o is the applied stress, and m and n are exponents. Different values of m and n corres 1650°c pond to different mechanisms. Values of n and o may be ob- ;1550°c;16 tained experimentally and used to infer a rate-controlling 145c1500c: The creep rate depends on temperature through an Arrheniu xponential term(Eq. (5). Figure 8 shows typical Arrhenius creep rate plots obtained for the sA3(2)fiber Activation energy 0 12 24 36 48 60 72 84 96 was determined by fitting Eq(5)to creep rates determined at vanous temperatures 7. Creep of SA3 (2) fiber under a stress of 150 MPa and in the 0 -1700C temperature range 6. 3 The largest apparent activation energy was determined for Nicalon S fibers (Table ID). Lower similar values were ob- ned for both SA3(1)and SA3(2)fibers. Knowing that the

The Hi-Nicalon fiber contains a significant amount of low￾density SiC. This SiC phase is not stable, as a result of the short duration of pyrolysis step during fiber processing.5 Grain growth thus starts at 12001C in the Hi-Nicalon fiber. It causes a decrease in creep rate. This phenomenon may explain the im￾perfect fit of the experimental creep curve by Eqs. (1)–(4) (Fig. 6(a)). (4) Secondary Creep: Mechanism Stationary creep was investigated essentially on Hi-Nicalon S and Tyranno SA3 fibers, which displayed a reasonably short primary creep stage. Extensive investigation of Hi-Nicalon would have required long experiments, as indicated above. Creep rate in the steady state is expressed by the following general relationship: e : ¼ fDdm g snwith D ¼ Do exp
Q RT
(5) where f is a material constant, D is the diffusivity, Q is the ap￾parent activation energy, R is the universal gas constant, T is temperature in Kelvin, dg is the grain size, s is the applied stress, and m and n are exponents. Different values of m and n corres￾pond to different mechanisms. Values of n and Q may be ob￾tained experimentally and used to infer a rate-controlling mechanism for creep. The creep rate depends on temperature through an Arrhenius exponential term (Eq. (5)). Figure 8 shows typical Arrhenius creep rate plots obtained for the SA3 (2) fiber. Activation energy was determined by fitting Eq. (5) to creep rates determined at various temperatures. The largest apparent activation energy was determined for Hi-Nicalon S fibers (Table II). Lower similar values were ob￾tained for both SA3 (1) and SA3 (2) fibers. Knowing that the Hi-Nicalon (a) (b) (c) (d) SA3(1) SA3 (2) Hi-Nicalon S 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 48 96 144 192 240 288 336 Time (h) Strain (%) experiment prediction · 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 12 24 36 48 60 Time (h) Strain (%) experiment prediction · 0 0.5 1 1.5 2 2.5 3 3.5 0 48 96 144 192 240 288 336 Time (h) Strain (%) experiment prediction · T = 1200°C 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 12 24 36 48 60 Time (h) Strain (%) experiment prediction ·
= 3.25 × 10–8 s–1
= 1.3 × 10–8 s–1
= 9.7 × 10–9 s–1
= 3.8 × 10–9 s–1 T = 1250°C σ = 850 MPa T = 1350°C σ = 850 MPa σ = 850 MPa T = 1200°C σ = 500 MPa Fig. 6. Creep behavior of (a) Hi-Nicalon, (b) Tyranno SA3 (1), (c) SA3 (2), and (d) Hi-Nicalon S fibers. 0 1 2 3 4 5 6 7 8 0 12 24 36 48 60 72 84 96 Time (h) Strain (%) 1350°C 1400°C 1450°C 1500°C 1550°C 1600°C 1650°C 1700°C Fig. 7. Creep of SA3 (2) fiber under a stress of 150 MPa and in the 13501–17001C temperature range. April 2007 SiC-Based Fibers and Low Oxygen Content 1151

l152 Journal of the American Ceramic Society--Sauder and Lamon Vol. 90. No. 4 1/T巛K-1 Stress(MPa) 52E045.6E0460E0464E0468E0472E04 1.00E-05 4 SA3(2)fiber(3DD Mpa) 10Eo6·s31eoco 00E-06 100E07 1.0E-08 D0E08 鲤 1364 kJ. mo 370 KJ. mor 1.0E09 16001500°1400°1300 200° Fig 9. Logarithmic plot of strain rate versus applied stress for SA3(2) Fig 8. Steady-state creep rate versus reciprocal temperature for SA3 The apparent activation energy determined on Hi-Nicalon S (770 kJ/mol, Table In is consistent with that corresponding to diffusion of carbon within grains(840 kJ/mor--) It seems too main difference between both SA3 fibers lies in the amount of low for silicon diffusion(910 kJ/mol) Diffusion within grains free carbon, this result indicates that secondary creep was not could be related to the feature of the microstructure. As men- influenced by the amount of carbon. tioned in section Ill (1), grain boundaries were not clearly de- Determination of stress exponent by fitting Eq(5)to creep fined. It may be thought that they cannot be preferred paths fo rates determined at different stresses(Fig 9) yields stress expo- diffusion. nents of 2.5 for all the fibers. It is worth mentioning that plots of The activation energy determined for SA3 fibers(370 kJ /mol train rates determined under various temperatures and applied Table ID)is close to that corresponding to diffusion of Al stresses(Figs.8-10), and creep parameters reported in Table ll grain boundaries in polycrystalline SiC (360-460 kJ/molss-y do not exhibit a noticeable discrepancy, which demonstrates the This suggests the contribution of diffusion of Al at grain bound pertinence of experimental work. ies. It is also close to half of that corresponding to diffusion of In polycrystalline SiC ceramics, n= I and creep results from or Si within grains for Hi-Nicalon S(Table ID). Hence, the diffusional processes either at grain boundaries(Coble creep. contribution of diffusion of C or Si at grain boundaries canno m=3)or within grains( Nabarro Herring creep m=2). A discarded activation energy corresponds to Nabarro Herring creep. The Figure 10 compares creep rates for Hi-Nicalon S and SA3(2) creep rate is very sensitive to grain size(Eq. 5): m=2 or 3. fibers under 850 MPa. It appears unambiguously that creep de Identical trends have been reported for both a- and B-sic formations are larger in SA3 fibers, despite the presence of larger Diffusional creep in polycrystalline SiC occurs by diffusion of rains. This trend can be attributed to the presence of Al. Alu minum is known to favor diffusion at a high temperature. This or by diffusion of impurities at grain boundaries. There is a causes an increase in the diffusion coefficient d(Eq. (5)). At very limited amount of data on diffusion of carbon and silicon temperatures above 1500oC, the SA3 fiber becomes more creep elements in Sic. There is no consensus about the diffusion rate. resistant than hi-Nicalon s Depending on the author, the diffusion rate of carbon could be 100 times faster than that of silicium, 26, 27 or it could be slower. 28 Activation energies for diffusion of carbon and silicium within (5) Tertiary Creep Stage igures 7 and ll clearly indicate that, at very high temperatures, Fi been estimated to be 84029 and 910 kJ/mol, 30 respectively. It the creep rate accelerates into a tertiary stage. In order to iden- is worth keeping in mind that creep of polycrystalline Sic tify the creep mechanism, tests were interrupted during the third cannot be controlled by dislocation moti <1700.C.31,32 For Nicalon SiC fibers tested in CO nel.the scanning electron microscopy. Figures 1l and 12 show the typ- activation energy was found to be consistent with the activation ical microstructure of fibers that was observed. It consists of two energies of thermally activated viscous flow of glasses at high distinct parts: the core and an annular region. The microstruc- ure of the core was unaffected, whereas Auger analysis showed between 2 and 3 that have been determined fo low oxygen content SiC fibers" correspond to grain-bound- 1/T(K-) ary sliding in the absence of a glassy phase. 52E-0456E046.0E-0464E-0 68E-0472E04 Creep of the fibers of this study (na 2.5)may be attributed 1.00E405 重理理理理 grain-boundary sliding, without grain elongation and glassy phase(Rachinger mechanism). In polycrystalline ceramics accommodation results from diffusion and fold formation at 1.00E-06 riple junctions. In SiC fibers, it probably involves carbon deformation 00E07 Table IL. Steady-State Creep Parameters 1.00E08 range(C) o(k/mor) 1.00E09 SA3(1 l150-1500 0-370)2.5(2.35-2.6) 1400°1300%1200° SA3(2) l150-1500 370)2.5(2.3-2.6 Hi-Nicalon s 1300-1500 770)2.6(2.42.9) state creep rate versus reciprocal temperature for SA3 and Hi- Nicalon S fibers under a stress of 850 MPa

main difference between both SA3 fibers lies in the amount of free carbon, this result indicates that secondary creep was not influenced by the amount of carbon. Determination of stress exponent by fitting Eq. (5) to creep rates determined at different stresses (Fig. 9) yields stress expo￾nents of 2.5 for all the fibers. It is worth mentioning that plots of strain rates determined under various temperatures and applied stresses (Figs. 8–10), and creep parameters reported in Table II do not exhibit a noticeable discrepancy, which demonstrates the pertinence of experimental work. In polycrystalline SiC ceramics, n 5 1 and creep results from diffusional processes either at grain boundaries (Coble creep, m 5 3) or within grains (Nabarro Herring creep m 5 2). A high￾activation energy corresponds to Nabarro Herring creep. The creep rate is very sensitive to grain size (Eq. 5): m 5 2 or 3. Identical trends have been reported for both a- and b-SiC. Diffusional creep in polycrystalline SiC occurs by diffusion of carbon22 or silicon23 at grain boundaries22,23 or within grains,24 or by diffusion of impurities at grain boundaries.25 There is a very limited amount of data on diffusion of carbon and silicon elements in SiC. There is no consensus about the diffusion rate. Depending on the author, the diffusion rate of carbon could be 100 times faster than that of silicium,26,27 or it could be slower.28 Activation energies for diffusion of carbon and silicium within grains in b-SiC made via chemical vapor deposition have been estimated to be 84029 and 910 kJ/mol,30 respectively. It is worth keeping in mind that creep of polycrystalline SiC cannot be controlled by dislocation motions at temperatures o17001C.31,32 For Nicalon SiC fibers tested in CO, n1, the activation energy was found to be consistent with the activation energies of thermally activated viscous flow of glasses at high temperatures.33 n exponents between 2 and 3 that have been determined for low oxygen content SiC fibers34–36 correspond to grain-bound￾ary sliding in the absence of a glassy phase.37 Creep of the fibers of this study (n2.5) may be attributed to grain-boundary sliding, without grain elongation and glassy phase (Rachinger mechanism). In polycrystalline ceramics, accommodation results from diffusion and fold formation at triple junctions.37 In SiC fibers, it probably involves carbon deformation. The apparent activation energy determined on Hi-Nicalon S (770 kJ/mol, Table II) is consistent with that corresponding to diffusion of carbon within grains (840 kJ/mol28–29). It seems too low for silicon diffusion (910 kJ/mol30). Diffusion within grains could be related to the feature of the microstructure. As men￾tioned in section III (1), grain boundaries were not clearly de- fined. It may be thought that they cannot be preferred paths for diffusion. The activation energy determined for SA3 fibers (370 kJ/mol, Table II) is close to that corresponding to diffusion of Al at grain boundaries in polycrystalline SiC (360–460 kJ/mol38–40). This suggests the contribution of diffusion of Al at grain bound￾aries. It is also close to half of that corresponding to diffusion of C or Si within grains for Hi-Nicalon S (Table II). Hence, the contribution of diffusion of C or Si at grain boundaries cannot be discarded. Figure 10 compares creep rates for Hi-Nicalon S and SA3 (2) fibers under 850 MPa. It appears unambiguously that creep de￾formations are larger in SA3 fibers, despite the presence of larger grains. This trend can be attributed to the presence of Al. Alu￾minum is known to favor diffusion at a high temperature. This causes an increase in the diffusion coefficient D (Eq. (5)). At temperatures above 15001C, the SA3 fiber becomes more creep resistant than Hi-Nicalon S. (5) Tertiary Creep Stage Figures 7 and 11 clearly indicate that, at very high temperatures, the creep rate accelerates into a tertiary stage. In order to iden￾tify the creep mechanism, tests were interrupted during the third stage and the cross section of the fibers was examined using scanning electron microscopy. Figures 11 and 12 show the typ￾ical microstructure of fibers that was observed. It consists of two distinct parts: the core and an annular region. The microstruc￾ture of the core was unaffected, whereas Auger analysis showed 1600 °C 1500 °C 1400 °C 1300 °C 1200 °C 1.00E-09 1.00E-08 1.00E-07 1.00E-06 1.00E-05 5.2E-04 5.6E-04 6.0E-04 6.4E-04 6.8E-04 7.2E-04 1/T (K ) 370 kJ.mol 372 kJ.mol 364 kJ.mol ε (s ) . Fig. 8. Steady-state creep rate versus reciprocal temperature for SA3 (2) fiber. Table II. Steady-State Creep Parameters Fibers Temperature range (1C) Activation energy Q (kJ/mol
1 ) n SA3 (1) 1150–1500 370 (360–370) 2.5 (2.35–2.6) SA3 (2) 1150–1500 370 (360–370) 2.5 (2.3–2.6) Hi-Nicalon S 1300–1500 770 (750–770) 2.6 (2.4–2.9) 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 100 Stress (MPa) 1000 ε (s ) . Fig. 9. Logarithmic plot of strain rate versus applied stress for SA3 (2) fiber. 1400 °C 1300 °C 1200 °C 1.00E-09 1.00E-08 1.00E-07 1.00E-06 1.00E-05 5.2E-04 5.6E-04 6.0E-04 6.4E-04 6.8E-04 7.2E-04 1/T (K ) ε (s ) . Fig. 10. Steady-state creep rate versus reciprocal temperature for SA3 (2) and Hi-Nicalon S fibers under a stress of 850 MPa. 1152 Journal of the American Ceramic Society—Sauder and Lamon Vol. 90, No. 4

April 2007 SiC-Based Fibers and Low Oxygen Conten 1153 16 14 12 E0.8 Test st SEM observation 04 02 0 0 12 Temps(h) x35.0 Fig. 11. Creep test at 1450C and under a stress of 500 MPa for Hi-Nicalon S fiber, and scanning electron microscope micrographs of cross section after test interruption(cross sections were obtained by cutting fibers using a blade) that the annular the fb was made of pure carbon. Furthermore, more serious effect. Under such vacuu oxidation the diameter of the fiber was unchanged. These results suggest products (Sio (g)and co (g)) would be hat silicon volatilized and that this phenomenon advanced from ed, so that the fiber would be completely le authors the surface toward the core. Data in the literature support this who observed degradation of SiC fibers aperature assumption. Thus, Fig. 13 shows that Sic decomposes at tem- under argon" attributed this phenomenon to active oxidation. peratures above 1400C when the pressure is identical to that in But they did not go into an in-depth investigation to ascertain the chamber(10 Pa). Furthermore, the authors have shown that gaseous Si (g) is produced preponderantly under these con- The creep rate acceleration would result from a change in the ditions.4- The experimental conditions were favorable to sil- stress state in the SiC fiber, caused by the annular degradation of icon volatilization. Thermodynamic equilibrium could not be fiber. During Si volatilization, stiff SiC is replaced by a porous reached and the pressure of gaseous si (g) remained above the carbon material that is much more compliant. As a conse- uilibrium value in the chamber. quence, the load is carried preponderantly by the Sic core. The pressure in the chamber (10 Pa) is smaller than As annular degradation proceeds, there is an increase in stress, that of si (g) at temperatures above 1400C, according to according to the following equation: Fig. 13. As a consequence, gaseous species can be eliminated; (i The chamber wall was covered with a deposit after the gaseous products condense The above phenomenon cannot be attributed to active oxi- a(1) FoRc dation from residual oxygen. Active oxidation would have a

that the annular region was made of pure carbon. Furthermore, the diameter of the fiber was unchanged. These results suggest that silicon volatilized and that this phenomenon advanced from the surface toward the core. Data in the literature support this assumption. Thus, Fig. 13 shows that SiC decomposes at tem￾peratures above 14001C when the pressure is identical to that in the chamber (10
4 Pa). Furthermore, the authors have shown that gaseous Si (g) is produced preponderantly under these con￾ditions.41–48 The experimental conditions were favorable to sil￾icon volatilization. Thermodynamic equilibrium could not be reached and the pressure of gaseous Si (g) remained above the equilibrium value in the chamber: (i) The pressure in the chamber (10
4 Pa) is smaller than that of Si (g) at temperatures above 14001C, according to Fig. 13. As a consequence, gaseous species can be eliminated; (ii) The chamber wall was covered with a deposit after the tests, indicating that gaseous products condensed. The above phenomenon cannot be attributed to active oxi￾dation from residual oxygen. Active oxidation would have a more serious effect. Under such vacuum conditions, oxidation products (SiO (g) and CO (g)) would be continuously eliminat￾ed, so that the fiber would be completely destroyed. The authors who observed degradation of SiC fibers at a high temperature under argon49 attributed this phenomenon to active oxidation. But they did not go into an in-depth investigation to ascertain their interpretation. The creep rate acceleration would result from a change in the stress state in the SiC fiber, caused by the annular degradation of fiber. During Si volatilization, stiff SiC is replaced by a porous carbon material that is much more compliant. As a conse￾quence, the load is carried preponderantly by the SiC core. As annular degradation proceeds, there is an increase in stress, according to the following equation: sðtÞ ¼ soR2 c ðRc
eðtÞÞ2 (6) 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 4 8 12 16 20 24 Temps (h) Strain (%) Test stop for SEM observation Fig. 11. Creep test at 14501C and under a stress of 500 MPa for Hi-Nicalon S fiber, and scanning electron microscope micrographs of cross section after test interruption (cross sections were obtained by cutting fibers using a blade). April 2007 SiC-Based Fibers and Low Oxygen Content 1153

l154 Journal of the American Ceramic Society--Sauder and Lamon Vol. 90. No. 4 Global cross section .kv×1a.k3白m Core observation Edge observation 75g edge magnification(cross sections were obtained by cutting fibers using a Daue presented in Fig. 7. (a)global cross section, (b)core magnification, (c) Fig 12. Scanning electron microscope micrographs of SA3(2)after the ere g is the tress,Re is the fiber diameter, and e(t) is (6) TEM Analysis of Crept Fibers the thickness of ular carbon layer TEM analysis was performed on SA3 (2)fibers tested at 1600oC In a first step a simple linear time dependence was selected and Hi-Nicalon S fibers tested at 1500.C. Creep tests were in- terrupted when the total deformation of the fiber reached 6% The following features were observed (7 o Carbon was highly preponderant in the superficial region (the thickness was close to 500 nm) The degradation rate was estimated from the thickness of the (i Cavities were not detected at triple junctions. As indi- annular layer determined from micrographs of fibers after in- cated above, it may be thought that accommodation is made ed creep tests. k=2.10-m/s was estimated for the Hi- possible because of carbon anisotropy Stress relation(6)was introduced into Eqs. (3)and (4). Figure 14 shows the creep curve that was predicted for Hi-Nicalon S a good agreement with the experimental results wa IV. Conclusion btained which supports the analysis. Nevertheless, a slight The tensile creep behavior of Sic fibers with a low oxygen con- discrepancy can be noticed, which may result from approxi- tent was investigated up to very high temperatures. Tests of long mations in the annular degradation law and in load sharing duration were carried out on a high-performance testing device. The contribution of the annular carbon layer was negle The three stages of creep were evidenced annular degradation law and contribution of the carbon layer / primary creep was particularly long in the Hi-Nicalon fibers Refinement, if necessary, would introduce a more complex in load sharing ca lasted more than 144 h. Primary creep was attributed to vis- elastic deformation of carbon at grain boundaries. Primary

where so is the applied stress, Rc is the fiber diameter, and e(t) is the thickness of the annular carbon layer. In a first step, a simple linear time dependence was selected for e(t): eðtÞ ¼ kt (7) The degradation rate was estimated from the thickness of the annular layer determined from micrographs of fibers after in￾terrupted creep tests. k 5 2.10
12 m/s was estimated for the Hi￾Nicalon S fiber. Stress relation (6) was introduced into Eqs. (3) and (4). Figure 14 shows the creep curve that was predicted for Hi-Nicalon S. A good agreement with the experimental results was obtained, which supports the analysis. Nevertheless, a slight discrepancy can be noticed, which may result from approxi￾mations in the annular degradation law and in load sharing. The contribution of the annular carbon layer was neglected. Refinement, if necessary, would introduce a more complex annular degradation law and contribution of the carbon layer in load sharing. (6) TEM Analysis of Crept Fibers TEM analysis was performed on SA3 (2) fibers tested at 16001C and Hi-Nicalon S fibers tested at 15001C. Creep tests were in￾terrupted when the total deformation of the fiber reached 6%. The following features were observed: (i) Carbon was highly preponderant in the superficial region (the thickness was close to 500 nm). (ii) Cavities were not detected at triple junctions. As indi￾cated above, it may be thought that accommodation is made possible because of carbon anisotropy. IV. Conclusion The tensile creep behavior of SiC fibers with a low oxygen con￾tent was investigated up to very high temperatures. Tests of long duration were carried out on a high-performance testing device. The three stages of creep were evidenced. Primary creep was particularly long in the Hi-Nicalon fibers. It lasted more than 144 h. Primary creep was attributed to vis￾coelastic deformation of carbon at grain boundaries. Primary Fig. 12. Scanning electron microscope micrographs of SA3 (2) after the creep test presented in Fig. 7, (a) global cross section, (b) core magnification, (c) edge magnification (cross sections were obtained by cutting fibers using a blade). 1154 Journal of the American Ceramic Society—Sauder and Lamon Vol. 90, No. 4

April 2007 SiC-Based Fibers and Low Oxygen Conten 1155 6.5 3.5 25 1.E+00 SiC +s 41 zrC45 zrC46 fC43 1.E-01 c 1E03 1E04 1E05 1E06 1E09 1E-10 1300c 1500°c1700c1900°c2200°c2500°c3000°c Fig 13. Vapor pressure evolution versus reciprocal temperature for various carbides ke to thank J M. Goyheneche x. Bourrat, and B mar (CEA) for valuable discussions. The authors are grateful to Dr. T. Ishikawa providing Tyranno SA3 samples, and to Nippon Carbon for the supply of Hi- 0.8 References Materials research Fenici. ""Critical Issues and Current Status of SiC/SiC Composites for Fusion. 283-287,128-3702000 PH M. Yun, J. C. Goldsby, and J.A. DiCarlo, "Tensile Creep and Stress-Rup- tre Behavior of Polymer Derived SiC Fibers, Ceramic TransactioNs, 46, 17-28 Temps(h) M. Yun and J. A DiCarlo, "Comparison of The Tensile, Creep, and Rup- re Strength Properties of Stoichiometne SiC Fibers; Report NASA/TM-1999- 14. Comparison of experiment and prediction of creep behavior at 1450C and under a stress of 500 MPa for the Hi-Nicalon S fiber R. Bunsell and A Piant. ""A Review of the Development of Three Gener- ions of Small Diameter Silicon Carbide Fibers. ". Mater. Sci. 41. 823-39 (2006 R. Bodet, X. Bourrat, J. Lamon, and R. Naslain, Tensile Creep Behavior of a creep is enhanced by the amount of free The Hi-Nicalc fiber experienced much larger deform han hi-Nicalon s on Between micro- ucture and Mechanical Behaviour at High Tempe a SiC Fibre with and sa3 fibers. Furthermore a shorter ry creep stage was Low Oxygen Content(Hi-Nicalon), "J Mater. Sci., 347(1997) observed on both latter fibers Rupture Failure of Ceramic Fibers and Composites, Ceramic Transactions, 99 The determination of creep constants, including the stress 119-134(19) and B. F. Dss in Mechanica s. Lohr, and R. Morell. Elsevier exponent(na2.5)and apparent activation energy, suggests the Tensile Creep of Ceramics: The Development ollowing mechanisms of secondary creep Grain-boundary sliding without grain elongation and glassy phases(Rachinger type). Accommodation was due to Aio( I Davies Fnrec o ompliance of carbon. ater.Sci,40[23l6187-93(20 (2) Diffusion of Al, C, or Si at grain boundaries in the SA3 rs, Ceram. Eng. Sci. Proc., 18[3]579-589(1 Diffusion of carbon or silicon within the grain in the Hi- 12M. H. Berger and A. R. Bunsell, ""Thin Foil Preparation of Small Diameter calon s fibe Ceramic or Glass Fibres for Observation by Transmission Electron Microscopy. J. Mater. Sci. Left, 12 825-8( 1993). But diffusion of impurities was not established, due to the Sauder, J. Lamon, and R. Pailler, ""Thermomechanical Properties of Car- ucity of data in the literature on diffusion of impurities with crystal ry creep was shown to be due to an increase in stress as silane-Derived Silicon Carbide Fibers Under Reduced Pressures. "J. Am. Ceram. he load-bearing fiber area is reduced by volatilization of si Soc,84]566-7002001)

creep is enhanced by the amount of free carbon. The Hi-Nicalon fiber experienced much larger deformations than Hi-Nicalon S and SA3 fibers. Furthermore, a shorter primary creep stage was observed on both latter fibers. The determination of creep constants, including the stress exponent (n2.5) and apparent activation energy, suggests the following mechanisms of secondary creep: (1) Grain-boundary sliding without grain elongation and glassy phases (Rachinger type). Accommodation was due to compliance of carbon. (2) Diffusion of Al, C, or Si at grain boundaries in the SA3 fiber. (3) Diffusion of carbon or silicon within the grain in the Hi￾Nicalon S fiber. But diffusion of impurities was not established, due to the paucity of data in the literature on diffusion of impurities within SiC polycrystals. Tertiary creep was shown to be due to an increase in stress as the load-bearing fiber area is reduced by volatilization of Si. Acknowledgments The authors would like to thank J. M. Goyheneche, X. Bourrat, and B. Marini (CEA) for valuable discussions. The authors are grateful to Dr. T. Ishikawa for providing Tyranno SA3 samples, and to Nippon Carbon for the supply of Hi￾Nicalon S fiber. References 1 T. Muroga, M. Gasparotto, and S. J. Zinkle, ‘‘Overview of Materials Research for Fusion Reactors,’’ Fusion Eng. Des., 61–62, 13–25 (2002). 2 A. Hasegawa, A. Kohyama, R. H. Jones, L. L. Snead, B. Riccardi, and P. Fenici, ‘‘Critical Issues and Current Status of SiC/SiC Composites for Fusion,’’ J. Nucl. Mater., 283–287, 128–37 (2000). 3 H. M. Yun, J. C. Goldsby, and J. A. DiCarlo, ‘‘Tensile Creep and Stress-Rup￾ture Behavior of Polymer Derived SiC Fibers,’’ Ceramic Transactions, 46, 17–28 (1994). 4 H. M. Yun and J. A. DiCarlo, ‘‘Comparison of The Tensile, Creep, and Rup￾ture Strength Properties of Stoichiometric SiC Fibers’’; Report NASA/TM-1999- 209284, 1999. 5 A. R. Bunsell and A. Piant, ‘‘A Review of the Development of Three Gener￾ations of Small Diameter Silicon Carbide Fibers,’’ J. Mater. Sci., 41, 823–39 (2006). 6 R. Bodet, X. Bourrat, J. Lamon, and R. Naslain, ‘‘Tensile Creep Behavior of a Silicon Carbide-Based Fiber With a Low Oxygen Content,’’ J. Mater. Sci., 30, 661–77 (1995). 7 G. Chollon, R. Pailler, R. Naslain, and P. Olry, ‘‘Correlation Between Micro￾structure and Mechanical Behaviour at High Temperatures of a SiC Fibre With a Low Oxygen Content (Hi-Nicalon),’’ J. Mater. Sci., 32, 1133–47 (1997). 8 J. A. DiCarlo and H. M. Yun, ‘‘Microstructural Factors Affecting Creep￾Rupture Failure of Ceramic Fibers and Composites,’’ Ceramic Transactions, 99, 119–134 (1998). 9 F. A. Kandil and B. F. Dyson, ‘‘Tensile Creep of Ceramics: The Development of a Testing Facility’’; pp. 151 in Mechanical Testing of Engineering Ceramics at High Temperatures, Edited by B. F. Dyson, R. D. Lohr, and R. Morell. Elsevier Applied Science, London, 1989. 10I. J. Davies, ‘‘Effect of Variable Radius on the Initial Creep Rate of Ceramic Fibres,’’ J. Mater. Sci., 40 [23] 6187–93 (2005). 11G. A. Newsome, ‘‘The Effect of Neutron Irradiation on Silicon Carbide Fi￾bers,’’ Ceram. Eng. Sci. Proc., 18 [3] 579–589 (1997). 12M. H. Berger and A. R. Bunsell, ‘‘Thin Foil Preparation of Small Diameter Ceramic or Glass Fibres for Observation by Transmission Electron Microscopy,’’ J. Mater. Sci. Lett., 12, 825–8 (1993). 13C. Sauder, J. Lamon, and R. Pailler, ‘‘Thermomechanical Properties of Car￾bon Fibers at High Temperatures (Up to 20001C),’’ Compos. Sci. Technol., 62 [4] 499–504 (2002). 14T. Shimoo, H. Takeuchi, and K. Okamura, ‘‘Thermal Stability of Polycarbo￾silane-Derived Silicon Carbide Fibers Under Reduced Pressures,’’ J. Am. Ceram. Soc., 84 [3] 566–70 (2001). 1.E-10 1.E-09 1.E-08 1.E-07 1.E-06 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 7 6.5 6 5.5 5 4.5 4 3.5 3 2,5 2 104/T(K) Vapor pressure (atm) SiC + Si41 SiC + C41 SiC42 TiC43 B4C44 WC43 ZrC45 ZrC46 HfC43 C47 1300°C 1500°C 1700°C 1900°C 2200°C 2500°C 3000°C Fig. 13. Vapor pressure evolution versus reciprocal temperature for various carbides. 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0 4 8 12 16 20 24 Temps (h) Strain (%) Experiment prediction Fig. 14. Comparison of experiment and prediction of creep behavior at 14501C and under a stress of 500 MPa for the Hi-Nicalon S fiber. April 2007 SiC-Based Fibers and Low Oxygen Content 1155