J.Am. Cera.Soc,84141787-94(2001) journal Influence of Strong Fiber/Coating Interfaces on the mechanical Behavior and Lifetime of Hi-Nicalon/(PyC/SiC)/SiC Minicomposites Sebastien Bertrand, Rene Pailler, * and Jacques Lamon poratoire des Composites Thermostructuraux, UMR 5801(CNRS, SEP/SNECMA, UBL, CEA), 33600 Pessac. France Hi-Nicalon fiber-reinforced silicon carbide matrix minicom- PyC-based fiber coatings are the most common. Pyc displays a posites (Hi-Nicalon/SiC) with nanoscale multilayered (PyC layered microstructure, and it leads to high-strength-high- SiC)n fiber coatings(also referred to as interphases) have been toughness SiC/SiC composites. PyC, however, is not stable in an manufactured via pressure pulse chemical vapor infiltration oxidizing environment. Consequently, with a view to protect (P-CVI). Fiber/coating interfaces were strengthened by using treated fibers. The microstructures of the interphases as well ered Py C/SiC interphases has emerged as the propagation and deflection of cracks in the interfacial Multilayered PyC/SiC interphases in SiC/SiC composites have region were investigated by SEM and TEM. Interfacial shear been investigated by several authors. ,6., 7, 10-12 Composites with stress was estimated using various methods based on either the multilayered interphases display mechanical properties and life- width of hysteresis loops on unloading-reloading, crack spac- times at high temperatures that compare favorably with those of ing, or fitting of the force-deformation curve using their counterpart with a single-layer fiber coating. micromechanics-based model. Tensile behavior at room tem- The present paper inve tes the tensile mechanical behavior perature and lifetime in static fatigue in air at 700C were and lifetime of SiC/SiC minicomposites containing Hi-Nicalon related to the interphase/interface characteristics. fibers coated with nanoscale(PyC/SiC), multilayered interphases Strong fiber/coating interphases were obtained using treated fi- bers. Hi-Nicalon/SiC minicomposites reinforced with as-received L. Introduction fibers have been examined in a previous paper IL. Experimental ProceduR omposites with strong interfaces have been developed. The strong (1) Processing Conditions and Materials oating/fiber bond was obtained when the fibers had been prev Minicomposites are unidirectional composites reinforced with a ously treated. Features of the mechanical behavior of sic/sic single tow of fibers. The Hi-Nicalon tows consist of 500 filaments opposites with strong fiber/coating interfaces at room and el each having a diameter -13.5 um(+1.5 um). The minicompos vated temperatures have been examined in several papers ites were manufactured using either as-received-or as-treated Experiments as well as models have demonstrated that a strong Hi-Nicalon tows interface is beneficial to strength, toughness, lifetime, and creep he minicomposites have been produced via pressure pulse resistance. -By contrast, weak interfaces are shown to be chemical vapor infiltration(P-CVI). The P-CVI apparatus and the processing conditions were detailed. -The tows were mounted on SiC/SiC composites with strong fiber/coating bonds, SiC frames for deposition of the interphase and SiC matrix.The cracks are deflected within the coating cohesive failure tows were slightly twisted with a constant angle(I turn/5 cm)to short and branched multiple cracks. The associated ecrease the porosity and to increase the fiber volume fraction in short debonds and load transfers allow further cracking of the the minicomposites matrix via a scale effect related to the stressed volume of Two different interphases were deposited on the fibers(table D) uncracked matrix, leading to a higher density of matrix cracks (1)A(PyC/SiC)lo nanoscale multilayered coating consisting Sliding friction within the coating and multiple cracking of the of 10 Py C/SiC sequences. The thickness of each sublayer was matrix increase energy absorption, leading to toughening. Short e( PyC)=20 nm and e(SiC)=50 nm. PyC was deposited first on debonds and improved load transfers limit fiber overloading the fibers during matrix cracking, leading to strengthening of the composi (2)For comparison purposes, a single Pyc layer 100 nn The associated tensile stress-strain curve exhibits a wide curved domain, and the stress at matrix-cracking saturation is close The fiber volume fraction in the minicomposites was 40% ultimate failure. Values of the interfacial shear stress(T (+5%). The main properties of the minicomposite constituents are using various methods on SiC/SiC composites with sted in Table il. Additional data can be found elsewhere 2, based(PyC-based) fiber coating, range between 10 and weak interfaces, whereas they are 100-300 MPa interfaces. 3,4, 6 (2) Microstructural Characterization Surface analysis of the fibers was performed using electron spectroscopy(AES). The fiber/matrix interfacial F. Zok--contributing editor and longitudinal sections using SEM and HRTEM. Preparation of the thin foils was detailed elsewhere Manuscript No. 188914 Received November 29, 1999; approved November 2, Supported by SEP and CNRS through a grant given to S. Bertrand. Proprietary treatment(LCTS-SNECMA)
Influence of Strong Fiber/Coating Interfaces on the Mechanical Behavior and Lifetime of Hi-Nicalon/(PyC/SiC)n/SiC Minicomposites Sebastien Bertrand, Rene Pailler,* and Jacques Lamon* Laboratoire des Composites Thermostructuraux, UMR 5801 (CNRS, SEP/SNECMA, UB1, CEA), 33600 Pessac, France Hi-Nicalon fiber-reinforced silicon carbide matrix minicomposites (Hi-Nicalon/SiC) with nanoscale multilayered (PyC/ SiC)n fiber coatings (also referred to as interphases) have been manufactured via pressure pulse chemical vapor infiltration (P-CVI). Fiber/coating interfaces were strengthened by using treated fibers. The microstructures of the interphases as well as the propagation and deflection of cracks in the interfacial region were investigated by SEM and TEM. Interfacial shear stress was estimated using various methods based on either the width of hysteresis loops on unloading–reloading, crack spacing, or fitting of the force–deformation curve using a micromechanics-based model. Tensile behavior at room temperature and lifetime in static fatigue in air at 700°C were related to the interphase/interface characteristics. I. Introduction FIBER/MATRIX interfaces in the most advanced ceramic matrix composites consist of a thin coating layer (micrometer-scale) of one or several materials deposited on the fiber. Recently, SiC/SiC composites with strong interfaces have been developed. The strong coating/fiber bond was obtained when the fibers had been previously treated.1–3 Features of the mechanical behavior of SiC/SiC composites with strong fiber/coating interfaces at room and elevated temperatures have been examined in several papers.3–8 Experiments as well as models have demonstrated that a strong interface is beneficial to strength, toughness, lifetime, and creep resistance.3–9 By contrast, weak interfaces are shown to be detrimental. In those SiC/SiC composites with strong fiber/coating bonds, the matrix cracks are deflected within the coating (cohesive failure mode) into short and branched multiple cracks.3,10 The associated short debonds and load transfers allow further cracking of the matrix via a scale effect related to the stressed volume of uncracked matrix,4,5 leading to a higher density of matrix cracks. Sliding friction within the coating and multiple cracking of the matrix increase energy absorption, leading to toughening. Short debonds and improved load transfers limit fiber overloading during matrix cracking, leading to strengthening of the composite. The associated tensile stress–strain curve exhibits a wide curved domain, and the stress at matrix-cracking saturation is close to ultimate failure. Values of the interfacial shear stress (t), measured using various methods on SiC/SiC composites with pyrocarbonbased (PyC-based) fiber coating, range between 10 and 20 MPa for weak interfaces, whereas they are 100–300 MPa for strong interfaces.3,4,6 PyC-based fiber coatings are the most common. PyC displays a layered microstructure, and it leads to high-strength–hightoughness SiC/SiC composites.3 PyC, however, is not stable in an oxidizing environment. Consequently, with a view to protect PyC-based interphases against oxidation, the concept of multilayered PyC/SiC interphases has emerged. Multilayered PyC/SiC interphases in SiC/SiC composites have been investigated by several authors.2,6,7,10–12 Composites with multilayered interphases display mechanical properties and lifetimes at high temperatures that compare favorably with those of their counterpart with a single-layer fiber coating. The present paper investigates the tensile mechanical behavior and lifetime of SiC/SiC minicomposites containing Hi-Nicalon fibers coated with nanoscale (PyC/SiC)n multilayered interphases. Strong fiber/coating interphases were obtained using treated fibers.† Hi-Nicalon/SiC minicomposites reinforced with as-received fibers have been examined in a previous paper.12 II. Experimental Procedures (1) Processing Conditions and Materials Minicomposites are unidirectional composites reinforced with a single tow of fibers. The Hi-Nicalon tows consist of 500 filaments, each having a diameter ;13.5 mm (61.5 mm). The minicomposites were manufactured using either as-received12 or as-treated Hi-Nicalon tows.† The minicomposites have been produced via pressure pulsed chemical vapor infiltration (P-CVI). The P-CVI apparatus and the processing conditions were detailed.12 The tows were mounted on SiC frames for deposition of the interphase and SiC matrix. The tows were slightly twisted with a constant angle (1 turn/5 cm) to decrease the porosity and to increase the fiber volume fraction in the minicomposites. Two different interphases were deposited on the fibers (Table I): (1) A (PyC/SiC)10 nanoscale multilayered coating consisting of 10 PyC/SiC sequences. The thickness of each sublayer was e(PyC) 5 20 nm and e(SiC) 5 50 nm. PyC was deposited first on the fibers. (2) For comparison purposes, a single PyC layer 100 nm thick. The fiber volume fraction in the minicomposites was ;40% (65%). The main properties of the minicomposite constituents are listed in Table II. Additional data can be found elsewhere.12,13 (2) Microstructural Characterization Surface analysis of the fibers was performed using Auger electron spectroscopy (AES). The fiber/matrix interfacial region was examined on failed minicomposites using SEM and on cross and longitudinal sections using SEM and HRTEM. Preparation of the thin foils was detailed elsewhere.14 F. Zok—contributing editor Manuscript No. 188914. Received November 29, 1999; approved November 2, 2000. Supported by SEP and CNRS through a grant given to S. Bertrand. *Member, American Ceramic Society. † Proprietary treatment (LCTS-SNECMA). J. Am. Ceram. Soc., 84 [4] 787–94 (2001) 787 journal
788 Joumal of the American Ceramic Sociery--Bertrand et al. Vol 84. No 4 Table L. Description of Investigated Hi-Nicalon/SiC of 80N was applied using a dead weight that was progressively hung on the bottom grip via the displacement of the support at a constant speed. This force was -10% above the proportional limit (Table Ill). Acoustic emission showed that matrix cracking initi- atches Fibers, Interphase (nm) (nm) n ated below the proportional limit under a force around 50N. The first matrix cracks did not influence specimen compliance. At HN/(C/SIC)1o AR failure of the minicomposites, the chronometer was stopped by the HNT/(C/SIC)Io 50 10 falling dead weight, giving the lifetime HNT/C Il. Results and Discussion Pyc/siC b (1 Material Characterization Details on the microstructure of the nanoscale multilayered Table Il. Properties of Minicomposite Constituents (PyC/SiC)n fiber coatings have been provided. 2 AES depth profiles revealed the presence of an oxygen-enriched layer(15-50 Statistical nm)at the surface of the as-received Hi-Nicalon fibers (Table In This layer consisted of SiO, and free carbon. The first interfacial Constituents Composition (nm) (GPa) m (MPa) PyC sublayer deposited was bonded to this silicon/carbon/oxyge layer. Such sublayers have also been identified in composites Hi-Nicalon fibers Sio 15-502804.26 sponding fiber/matrix interactions were weak, and deflection of the Hi-Nicalon fibers(T) Free carbon 50-100 305 5.8 matrix cracks occurred at the fiber/interphase interfaces. 3, 10 PyC interphase Sic interphase 280 AEs depth profiles performed on treated Hi-Nicalon fibers SiC P-CVI matrix 4005.55.7 show that the surface of the fibers consists of a 50-100 nm thick TA R.= as-received, T. treated. Reference volume, I.=Im layer of free carbon(Fig. 1). The presence of such a superficial layer of free carbon increases the strength of the Py C coating/fiber interface in Nicalon/PyC/SiC composites ,,o The TEM micro- graph of Fig. I shows that the deposited Pyc is perfectly bonded ( Tensile Tests at Room Temperature to the fiber. By contrast, preexisting fiber/coating debonds are Uniaxial tension tests were performed at room temperature at a observed in the minicomposites reinforced with as-received Hi- Nicalon fibers I deformation rate of 50 um/mn using a machine and procedure metallic tubes that were then gripped into the testing machine. (2) Tensile Behavior at Room Temperature Gauge length was 20 mm. Load train compliance, Cs was Figure 2 shows that the force-deformation curves obtained for determined from tensile tests on fiber tows having various gauge HNT/(C/SiC)1o minicomposites reinforced with treated fibers engths(Cs= 0.3 um/N) exhibit typical features associated with strong fiber/matrix Unloading-reloading cycles were conducted on a few speci- bonds: ,3 mens of each batch to estimate the elastic modulus of the cracked (1) A wide curved domain up to ultimate failure attributed to minicomposites, residual strains at zero load, and T. After ultimate a high density of matrix cracks and short debonds, and failure, the test specimens were examined using SEM, and the (2) Rather narrow hysteresis loops, indicative of strong fiber/ crack spacings were measured. matrix interactions on unloading-reloading HN/(C/SiC)o minicomposites reinforced with as-received fi- (4 Static Fatigue Tests at 700C in Air bers display typical features associated with weak fiber/matrix The lifetime of the minicomposites under constant load was bonds, including measured in static fatigue at 700C in air. These temperature (1) A narrow curved domain reflecting longer debonds and a onditions were the worst, because the oxidation rate was high fo lower density of matrix cracks, and PyC and low for SiC. The minicomposite ends were glued within lumina tubes using an alumina-based ceramic adhesive. The gauge length(10 mm) was determined by the distance between the tubes. The tubes were gripped into the testing machine. The ses are used to describe the mechanical behavior of the minicomposites were positioned within the furnace hot zone where the temperature was uniform at 700C. The minicomposites were sites, as a result of the transverse cracks that locally reduce the stressed section to heated to the test temperature before loading. Then a constant force that of fibers Table Ill. Average Mechanical Properties of Minicomposites Tested Saturations FailureR Batches HN/(C/SIC)1o 351 71 0.12 135 0.30 19 0.83 HNT/(C/SiC)Io 0.79 34 750. 0.30 HNT/C 350 0.17 0.7 156 0.81 microcrack spacing in the internal matrix of the minicomposite. 'Microcrack spacing in the surface of the minicomposite. F
(3) Tensile Tests at Room Temperature Uniaxial tension tests were performed at room temperature at a deformation rate of 50 mm/mn using a machine and procedure detailed elsewhere.12 The minicomposite ends were glued within metallic tubes that were then gripped into the testing machine. Gauge length was 20 mm. Load train compliance, Cs, was determined from tensile tests on fiber tows having various gauge lengths (Cs 5 0.3 mm/N). Unloading–reloading cycles were conducted on a few specimens of each batch to estimate the elastic modulus of the cracked minicomposites, residual strains at zero load, and t. After ultimate failure, the test specimens were examined using SEM, and the crack spacings were measured. (4) Static Fatigue Tests at 700°C in Air The lifetime of the minicomposites under constant load was measured in static fatigue at 700°C in air. These temperature conditions were the worst, because the oxidation rate was high for PyC and low for SiC. The minicomposite ends were glued within alumina tubes using an alumina-based ceramic adhesive. The gauge length (10 mm) was determined by the distance between the tubes. The tubes were gripped into the testing machine. The minicomposites were positioned within the furnace hot zone where the temperature was uniform at 700°C. The minicomposites were heated to the test temperature before loading. Then a constant force of 80 N was applied using a dead weight that was progressively hung on the bottom grip via the displacement of the support at a constant speed. This force was ;10% above the proportional limit (Table III). Acoustic emission showed that matrix cracking initiated below the proportional limit under a force around 50 N.5 The first matrix cracks did not influence specimen compliance. At failure of the minicomposites, the chronometer was stopped by the falling dead weight, giving the lifetime. III. Results and Discussion (1) Material Characterization Details on the microstructure of the nanoscale multilayered (PyC/SiC)n fiber coatings have been provided.12 AES depth profiles revealed the presence of an oxygen-enriched layer (15–50 nm) at the surface of the as-received Hi-Nicalon fibers (Table II). This layer consisted of SiO2 and free carbon. The first interfacial PyC sublayer deposited was bonded to this silicon/carbon/oxygen layer. Such sublayers have also been identified in composites reinforced with as-received Nicalon fibers (NL 202). The corresponding fiber/matrix interactions were weak, and deflection of the matrix cracks occurred at the fiber/interphase interfaces.1,3,10 AES depth profiles performed on treated Hi-Nicalon fibers show that the surface of the fibers consists of a 50–100 nm thick layer of free carbon (Fig. 1). The presence of such a superficial layer of free carbon increases the strength of the PyC coating/fiber interface in Nicalon/PyC/SiC composites.1,3,10 The TEM micrograph of Fig. 1 shows that the deposited PyC is perfectly bonded to the fiber. By contrast, preexisting fiber/coating debonds are observed in the minicomposites reinforced with as-received HiNicalon fibers.12 (2) Tensile Behavior at Room Temperature‡ Figure 2 shows that the force–deformation curves obtained for HNT/(C/SiC)10 minicomposites reinforced with treated fibers exhibit typical features associated with strong fiber/matrix bonds:1,3 (1) A wide curved domain up to ultimate failure attributed to a high density of matrix cracks and short debonds, and (2) Rather narrow hysteresis loops, indicative of strong fiber/ matrix interactions on unloading–reloading. HN/(C/SiC)10 minicomposites reinforced with as-received fibers display typical features associated with weak fiber/matrix bonds, including (1) A narrow curved domain reflecting longer debonds and a lower density of matrix cracks, and ‡ Forces instead of stresses are used to describe the mechanical behavior of the minicomposites, because derivation of a stress from the applied force is not straightforward or appropriate. The stress state is not uniform within the minicomposites, as a result of the transverse cracks that locally reduce the stressed section to that of fibers. Table I. Description of Investigated Hi-Nicalon/SiC Minicomposites Batches Fibers† Interphase Interphase characteristics e(PyC)‡ (nm) e(SiC)‡ (nm) n§ HN/(C/SiC)10 A.R. (20/50)10 20 50 10 HNT/(C/SiC)10 T. (20/50)10 20 50 10 HN/C A.R. (100/0)1 100 0 1 HNT/C T. (100/0)1 100 0 1 † A.R. 5 as-received, T. 5 treated. ‡ Thickness per sublayer. § Number of PyC/SiC bilayers. Table II. Properties of Minicomposite Constituents12,13 Constituents† Superficial layer Young’s Modulus (GPa) Statistical parameters Composition Thickness (nm) m so ‡ (MPa) Hi-Nicalon fibers (A.R.) SiO2 15–50 280 4.2 6 Hi-Nicalon fibers (T.) Free carbon 50–100 305 5.8 26 PyC interphase 12–80 SiC interphase 280 SiC P-CVI matrix 400 5.5 5.7 † A.R. 5 as-received, T. 5 treated. ‡ Reference volume, Vo 5 1 m3 . Table III. Average Mechanical Properties of Minicomposites Tested Batches Ec (GPa) Proportional limit§ Saturation§ Failure§ ls (mm) FE (N) εE (%) FS (N) εs (%) FR (N) εR (%) HN/(C/SiC)10 351 71 0.12 135 0.30 195 0.83 50† 175‡ HNT/(C/SiC)10 360 73 0.07 ;FR ;εR 183 0.79 35† 67‡ HN/C 342 75 0.09 100 0.30 171 0.69 40† 95‡ HNT/C 350 75 0.17 140 0.7 156 0.81 30† 100‡ † Microcrack spacing in the internal matrix of the minicomposite. ‡ Microcrack spacing in the surface of the minicomposite. § F 5 force; ε 5 deformation. 788 Journal of the American Ceramic Society—Bertrand et al. Vol. 84, No. 4
April 2001 Influence of Interfaces on Mechanical Behavior and Lifetime of Hi-Nicalon/(PyC/SiC) Sic 789 0.4 Ef Vf/Eo Fig 3. Relative elastic modulus versus applied E=elastic HN/C; D= HNT/C. eA= HN/(C/SiC)o; B uncracked minicom SiC)1o: C 010020030040050060070 indicates that the applied load is borne only by the fibers Therefore, fiber debonding is complete at this stage, and matrix cracking saturation has occurred. This minimum is not reached by 是100nm the modulus of HNT/(C/SiC)o minicomposites, indicating that saturation of matrix cracking has not occurred at ultimate failure. Figure 3 clearly shows that the behavior is significantly influ- Fig 1. TEM micrograph of fiber/coating interfacial region( batch C)and enced by the presence of treated fibers. Those minicomposites AES depth profile of treated Hi-Nicalon fibers. Pyc 1 shows deposited reinforced with as-received fibers display a significantly steep ublayer of PyC, "isotropic carbon"and "anisotropic carbon"indicate fiber modulus decrease. The minimum Err is observed at deformations superficial layer of free carbon. of -0 2%0.3%, which correspond to the strain at saturation indicated by the end of the curved domain of the force deforma tions curves(Fig. 2). For the HN/(C/SiC)o minicomposite, the minimum modulus becomes smaller than Ere This may be tainties in the data, including the modulus measurements By contrast, the minicomposites reinforced with treated fibers experience a gradual modulus decrease. The minimum is reached r larger deformations (20.7%), indicating a high strain at These trends suggest that debonding was more significant in those minicomposites reinforced with as-received fibers, which Strain(‰) implies the presence of weaker fiber/matrix bonds Fig. 2. Tensile force-deformation curves for Hi-Nicalon/SiC minicom- ( Matrix Cracking and Crack Deflection sites investigated in present paper. A= HN(C/SIC)o, B= HNT/(C/ SiC):C= HNC, D= HNT/C The crack spacing distance measured for the transverse crack (7)is always shorter in the internal matrix(Fig. 4). This effect seems to be related to tow g and may be attributed to the contribution of the radial compressive stress components that (2)Wide hysteresis loops, reflecting weaker fiber/matrix in teractions on unloading-reloading The above features can also be noticed on the force-deforma- tion curves obtained for those minicomposites with single-layer Pyc fiber coatings. However, a certain discrepancy is noticed in the force-deformation curves of some HNT/C minicomposites reinforced with treated fibers. The curved domain seems to be narrower, and the hysteresis loops seem to be wider than expected The elastic modulus pertinent to the cracked minicomposites is derived from the slope of the linear portion of the reloading curve a minimum tangent modulus). The tangent to this linear portion intercepts the origin in most cases. For the HNT/C minicompos- 80 ites, it intercepts the abscissa on the negative side, suggesting that the fibers tend to contract as a result of the presence of tensile 120 thermally induced residual stresses in the fibers. The permanent train at zero load includes contributions from misfit relief an sliding. The larger permanent elongations at zero load exhibited by Radial position(arbitrary unit those minicomposites with as-received fibers(Fig. 2) suggest the presence of weaker fiber/matrix interactions when compared with minicomposites reinforced with treated fibers. Figure 3 shows the typical trends in the elastic modulus during Surface of the Center of the Surface of the the tensile tests. For most minicomposites, the modulus decreas minicomposite minicomposite minicomposite a minimum value that coincides with the quantity Ep r(Er is the Fig. 4. Example of distribution of Is spacing distances measured at fiber Youngs modulus, and Ve the fiber volume fraction), which various locations in matrix of
(2) Wide hysteresis loops, reflecting weaker fiber/matrix interactions on unloading–reloading. The above features can also be noticed on the force–deformation curves obtained for those minicomposites with single-layer PyC fiber coatings. However, a certain discrepancy is noticed in the force–deformation curves of some HNT/C minicomposites reinforced with treated fibers. The curved domain seems to be narrower, and the hysteresis loops seem to be wider than expected. The elastic modulus pertinent to the cracked minicomposites is derived from the slope of the linear portion of the reloading curve (minimum tangent modulus).12 The tangent to this linear portion intercepts the origin in most cases. For the HNT/C minicomposites, it intercepts the abscissa on the negative side, suggesting that the fibers tend to contract as a result of the presence of tensile thermally induced residual stresses in the fibers. The permanent strain at zero load includes contributions from misfit relief and sliding. The larger permanent elongations at zero load exhibited by those minicomposites with as-received fibers (Fig. 2) suggest the presence of weaker fiber/matrix interactions when compared with minicomposites reinforced with treated fibers. Figure 3 shows the typical trends in the elastic modulus during the tensile tests. For most minicomposites, the modulus decreases to a minimum value that coincides with the quantity Ef Vf (Ef is the fiber Young’s modulus, and Vf the fiber volume fraction), which indicates that the applied load is borne only by the fibers. Therefore, fiber debonding is complete at this stage, and matrixcracking saturation has occurred. This minimum is not reached by the modulus of HNT/(C/SiC)10 minicomposites, indicating that saturation of matrix cracking has not occurred at ultimate failure. Figure 3 clearly shows that the behavior is significantly influenced by the presence of treated fibers. Those minicomposites reinforced with as-received fibers display a significantly steep modulus decrease. The minimum Ef Vf is observed at deformations of ;0.2%–0.3%, which correspond to the strain at saturation indicated by the end of the curved domain of the force deformations curves (Fig. 2). For the HN/(C/SiC)10 minicomposite, the minimum modulus becomes smaller than Ef Vf . This may be attributed to the presence of broken or bent fibers, and/or uncertainties in the data, including the modulus measurements. By contrast, the minicomposites reinforced with treated fibers experience a gradual modulus decrease. The minimum is reached for larger deformations ($0.7%), indicating a high strain at saturation. These trends suggest that debonding was more significant in those minicomposites reinforced with as-received fibers, which implies the presence of weaker fiber/matrix bonds. (3) Matrix Cracking and Crack Deflection The crack spacing distance measured for the transverse crack (ls) is always shorter in the internal matrix (Fig. 4). This effect seems to be related to tow twisting and may be attributed to the contribution of the radial compressive stress components that Fig. 1. TEM micrograph of fiber/coating interfacial region (batch C) and AES depth profile of treated Hi-Nicalon fibers. “PyC 1” shows deposited sublayer of PyC; “isotropic carbon” and “anisotropic carbon” indicate fiber superficial layer of free carbon. Fig. 2. Tensile force–deformation curves for Hi-Nicalon/SiC minicomposites investigated in present paper. A 5 HN/(C/SiC)10; B 5 HNT/(C/ SiC)10; C 5 HN/C; D 5 HNT/C. Fig. 3. Relative elastic modulus versus applied deformation: E 5 elastic modulus given by minimum tangent modulus, E0 5 elastic modulus of uncracked minicomposite. A 5 HN/(C/SiC)10; B 5 HNT/(C/SiC)10; C 5 HN/C; D 5 HNT/C. Fig. 4. Example of distribution of ls spacing distances measured at various locations in matrix of minicomposite (minicomposite HN/(C/SiC)10). April 2001 Influence of Interfaces on Mechanical Behavior and Lifetime of Hi-Nicalon/(PyC/SiC)n/SiC 789
790 Journal of the American Ceramic Society-Bertrand et al Vol 84. No 4 increase fiber/matrix interactions. These stresses are induced by the curved fibers that try to stretch under a tensile load The I data indicate that the fiber/matrix bond is strengthened in the minicomposites reinforced with treated fibers. Thus, the Is 1=E measured in the HNT/(C/SiC)Io minicomposites is significantl smaller than that in their counterpart with as-received fibers (HN/(C/SiC)o). Although the Is measured in the internal matrix b2 _(1+v)EE+(1-2)E falls within the same range for most minicomposites, that obtained for the minicomposites reinforced with treated fibers is the shortest where o is the applied stress in the unloading-reloading sequence (Table I) that corresponds to 84, o, the initial stress level at unloading, E Matrix cracking essentially involved transverse cracks the Youngs modulus of the minicomposite, R the fiber radius, v over, a few longitudinal cracks also were detected(Fig. 5) the Poisson's ratio (v=vm= v, Em the Youngs modulus of the cracks were not identified on the specimens inspected before matrix, Er that of fiber, and ve the fiber volume fraction. T was testing. They were created during the tensile tests derived from the SA-o data measured during the last unloading- Unlike the minicomposites with as-received fibers in which reloading sequence before ultimate failure of the minicomposites ignificant fiber/coating debonds were observed before testing, The number of matrix cracks in gauge length, N, was determined those minicomposites with treated fibers did not exhibit such from SEM inspection of the minicomposites after failure preexisting interface features. (2) The spacing distance of the matrix cracks at saturation. Figure 6 illustrates the significant differences observed in the Estimates of T are given by the following equations deflection of matrix cracks, depending on the fiber/coating bond In those minicomposites reinforced with as-received fibers, deflec- sRe tion of the matrix cracks occurs in the fiber/interphase interface and also in the matrix/interphase interface(Fig. 6). It is worth 24(1+ pointing out that the crack-opening displacement is rather large (0. 2 um). In those minicomposites reinforced with treated fibers, rrl matrix cracks are deflected within the coating in the PyC sublayers (Fig. 6). Crack branching can also be noticed on Fig. 6. Finally, the crack reaches the PyC sublayer bonded to the fiber surface. Unlike where o, is the applied stress at matrix-cracking saturation. in minicomposites with as-received fibers, the crack-openin 3) The force-deformation curves. A curve fitting procedur displacement is now very small (-10 nm been detailed and validated in previous works. s This model has based on a model of the tensile behavior was used e T adj (4 Interfacial Shear Stress (T) ment involved predictions of the force-deformation curves(Fig. results of tensile tests and crack examination, estimates of the T ters given in Table t properties and the flaw-strength parame- To assess the trends that have been identified based 7)from the cons were extracted from various data provided by the tensile tests The T estimates given in Table Iv evidence an unquestionable (1) The width of the hysteresis loops(8A)measured durin strengthening of the fiber/coating bond associated with the treated equation(established elsewhere for microcomposites 2. O/lowing fibers. The different methods agree satisfactorily in indicating this trend, despite certain discrepancies that can be noticed. The scatter in T measurements previously noticed does not affect this trend biN(I-a1Ve-Ro Therefore the results can be summarized as follows .t data are 2:E ( closer to 200 MPa for minicomposites reinforced with treated fibers, and they are close to 100 MPa for those reinforced with as-received fibers These trends compare satisfactorily with those established on Nicalon/SiC composites. This agreement can be attributed to the presence of a silicon/carbon/oxygen layer at the surface of as- received Nicalon and Hi-Nicalon fibers, and a free carbon layer at the surface of treated Nicalon and Hi-Nicalon fibers. which has been shown to dictate the fiber/coating bond (5) Lifetime in Static Fatigue at 700C n did not fail after 140 and 200 h. Further comparison of the data showed that the multilayered coating improved lifetime: from 2 to 20 h for those minicomposites reinforced with as-received fibers The influence might appear to be less clear for those minicompos ites reinforced with treated fibers because of scatter in the data 1.5pm for the HNT/(C/SiC)o minicomposites However, some HNT/ (C/SiC)o minicomposites were not broken after 200 and 140 h, was 114 h. This trend agreed with previous results reported elsewhere. 7, 8 The results suggest an effect of the fiber/coating bond. Strong nIcon matrix cracks. SEM images(Fig. 9)of the interfacial regions after the static fatigue tests show that PyC layers have disappeared Fig. 5. SEM images of longitudinal cracks detected in Hi-Nicalon/SiC However, some interesting differences can be noticed, depending minicomposite after tensile tests. on the batch, that can be related to the location of the debond crack
increase fiber/matrix interactions. These stresses are induced by the curved fibers that try to stretch under a tensile load. The ls data indicate that the fiber/matrix bond is strengthened in the minicomposites reinforced with treated fibers. Thus, the ls measured in the HNT/(C/SiC)10 minicomposites is significantly smaller than that in their counterpart with as-received fibers (HN/(C/SiC)10). Although the ls measured in the internal matrix falls within the same range for most minicomposites, that obtained for the minicomposites reinforced with treated fibers is the shortest (Table III). Matrix cracking essentially involved transverse cracks. Moreover, a few longitudinal cracks also were detected (Fig. 5). Such cracks were not identified on the specimens inspected before testing. They were created during the tensile tests. Unlike the minicomposites with as-received fibers in which significant fiber/coating debonds were observed before testing, those minicomposites with treated fibers did not exhibit such preexisting interface features. Figure 6 illustrates the significant differences observed in the deflection of matrix cracks, depending on the fiber/coating bond. In those minicomposites reinforced with as-received fibers, deflection of the matrix cracks occurs in the fiber/interphase interface and also in the matrix/interphase interface (Fig. 6). It is worth pointing out that the crack-opening displacement is rather large (0.2 mm). In those minicomposites reinforced with treated fibers, matrix cracks are deflected within the coating in the PyC sublayers (Fig. 6). Crack branching can also be noticed on Fig. 6. Finally, the crack reaches the PyC sublayer bonded to the fiber surface. Unlike in minicomposites with as-received fibers, the crack-opening displacement is now very small (;10 nm). (4) Interfacial Shear Stress (t) To assess the trends that have been identified based on the results of tensile tests and crack examination, estimates of the t were extracted from various data provided by the tensile tests: (1) The width of the hysteresis loops (dD) measured during unloading–reloading cycles, which is related to t by the following equation (established elsewhere for microcomposites15): t 5 b2N~1 2 a1Vf! 2 Rf 2V f 2 Em S sp 2 dDDS s sp DS1 2 s sp D (1) with a1 5 Ef Ec (2) b2 5 ~1 1 n! Em@Ef 1 ~1 2 2n!Ec# Ef@~1 1 n!Ef 1 ~1 2 n!Ec# where s is the applied stress in the unloading–reloading sequence that corresponds to dD, sp the initial stress level at unloading, Ec the Young’s modulus of the minicomposite, Rf the fiber radius, n the Poisson’s ratio (n5nm 5 nf ), Em the Young’s modulus of the matrix, Ef that of fiber, and Vf the fiber volume fraction. t was derived from the dD–s data measured during the last unloading– reloading sequence before ultimate failure of the minicomposites. The number of matrix cracks in gauge length, N, was determined from SEM inspection of the minicomposites after failure. (2) The spacing distance of the matrix cracks at saturation. Estimates of t are given by the following equations:16,17 t 5 ssRf 2VflsS1 1 EfVf EmVm D (3) t 5 ssRfVm 2Vfls (4) where ss is the applied stress at matrix-cracking saturation. (3) The force–deformation curves. A curve fitting procedure based on a model of the tensile behavior was used. This model has been detailed and validated in previous works.5,12 The t adjustment involved predictions of the force–deformation curves (Fig. 7) from the constituent properties and the flaw–strength parameters given in Table II. The t estimates given in Table IV evidence an unquestionable strengthening of the fiber/coating bond associated with the treated fibers. The different methods agree satisfactorily in indicating this trend, despite certain discrepancies that can be noticed. The scatter in t measurements previously noticed12 does not affect this trend. Therefore, the results can be summarized as follows: t data are closer to 200 MPa for minicomposites reinforced with treated fibers, and they are close to 100 MPa for those reinforced with as-received fibers. These trends compare satisfactorily with those established on Nicalon/SiC composites.3 This agreement can be attributed to the presence of a silicon/carbon/oxygen layer at the surface of asreceived Nicalon and Hi-Nicalon fibers, and a free carbon layer at the surface of treated Nicalon and Hi-Nicalon fibers, which has been shown to dictate the fiber/coating bond. (5) Lifetime in Static Fatigue at 700°C The lifetime data are shown in graphical form in Fig. 8. The lifetimes obtained for those minicomposites reinforced with treated fibers were unambiguously the longest. Some specimens did not fail after 140 and 200 h. Further comparison of the data showed that the multilayered coating improved lifetime: from 2 to ;20 h for those minicomposites reinforced with as-received fibers. The influence might appear to be less clear for those minicomposites reinforced with treated fibers because of scatter in the data for the HNT/(C/SiC)10 minicomposites. However, some HNT/ (C/SiC)10 minicomposites were not broken after 200 and 140 h, whereas the maximum lifetime of those HNT/C minicomposites was 114 h. This trend agreed with previous results reported elsewhere.7,8 The results suggest an effect of the fiber/coating bond. Strong bonds have been shown to reduce debonding and crack-opening displacement that limit the amount of oxygen migrating within the matrix cracks. SEM images (Fig. 9) of the interfacial regions after the static fatigue tests show that PyC layers have disappeared. However, some interesting differences can be noticed, depending on the batch, that can be related to the location of the debond crack Fig. 5. SEM images of longitudinal cracks detected in Hi-Nicalon/SiC minicomposite after tensile tests. 790 Journal of the American Ceramic Society—Bertrand et al. Vol. 84, No. 4
pril 2001 Influence of Interfaces on Mechanical Behavior and Lifetime of Hi-Nicalon/(PyC/SiC) Sic 791 M (Fiber surface Sic 1 um (Fiber surface) Fiber Fibe Fig icrographs showing deflection of matrix microcracks in interfacial region of minicomposites: (a) SEM micrograph of a HN/(C/SIC)o minicomposite,(b)TEM micrograph of a HNT/(C/SiC)o minicomposite in the interfacial region. In the HN/(C/SiC)o minicomposites, on (a)250pT「·T·冂 the Pyc sublayer that lay on the fibers has been eliminated whereas, in the HNT/(C/SiC)o minicomposites, all the PyC layers exp in which matrix cracks are found to be deflected are affected V. Discussion 2150 It has been confirmed again that multilayered interphases are not detrimental to the mechanical behavior of sic/sic minicom- 100 posites, although these contain stiff sublayers of SiC and very thin (nanometer-scale)PyC sublayers The presence of rather strong fiber/coating interfaces in those minicomposites reinforced with treated Hi-Nicalon fibers has been Batcha revealed by a set of data, including features of the force deformation curves. location of the debond crack in the interfacial 002040.60.8 region, composition of the fiber surface, and estimates of T. Deformation (%) The force-deformation curves pertinent to those minicompos- ites reinforced with treated fibers exhibited the features previousl observed on Nicalon/SiC minicomposites and two-dimensional woven composites reinforced with treated fibers(NLM 202) including(i) a wide curved domain, (ii)a large stress/strain at saturation of matrix cracking that coincides with ultimate failures, (b)250 (iii) small residual deformations at zero load, and (iv) a gentle modulus decrease during tensile tests. The matrix cracks were deflected within each Pyc sublayer of the interphase in those minicomposites reinforced with treated fibers and at the fiber surface in those minicomposites reinforced 150 experiment vith as-received fibers. This debond patten was similar to that observed on Nicalon/SiC composites. 10 Deviation within the interphase resulted from the presence of a stronger bond between 100 the interphase and the fib In those minicomposites reinforced with as-received fibers, weakening of the fiber/interphase region has been evidenced usin SEM. The preexisting debond cracks were attributed to the lateral contraction of as-received Hi-Nicalon fibers during minicomposite Batch C rocessing. Yun et al. have shown that the lateral contraction of as-received Hi-Nicalon fibers during processing of composites can 0 xceed 1% at 1200C. The contraction was associated with fiber Deformation(%) shrinkage and with the transformation of the Sic amorphous phase nto a B-sic crystallized phase. Fig. 7. Comparison of predicted and experimental force deformation curves for SiC/SiC minicomposites(a) HN/(C/SiC)o and(b)HNT/(C/SiC)o The interfacial bond was characterized by the interfacial shear tress (T). As previously mentioned, all the methods indicated tha the fiber/coating bond was stronger in those minicomposites fibers and T > 200 MPa for those reinforced with treated fibers reinforced with treated fibers. However, the range of t data was In study, T 100 MPa for those Hi-Nicalon/SiC arrower than that observed on Nicalon/Sic two-dimensional InIc tes reinforced with as-received fibers. and T s 200 woven composites. In those Nicalon/SiC composites, the respe MPa e reinforced with treated fibers. A contribution of tive t values were different by more than I order of magnitude: twisting could be expected. Twisting generated radial compressive T 10 MPa for those composites reinforced with as-received stresses that enhanced fiber/matrix interactions and can tend to
in the interfacial region. In the HN/(C/SiC)10 minicomposites, only the PyC sublayer that lay on the fibers has been eliminated; whereas, in the HNT/(C/SiC)10 minicomposites, all the PyC layers in which matrix cracks are found to be deflected are affected. IV. Discussion It has been confirmed again that multilayered interphases are not detrimental to the mechanical behavior of SiC/SiC minicomposites, although these contain stiff sublayers of SiC and very thin (nanometer-scale) PyC sublayers. The presence of rather strong fiber/coating interfaces in those minicomposites reinforced with treated Hi-Nicalon fibers has been revealed by a set of data, including features of the force– deformation curves, location of the debond crack in the interfacial region, composition of the fiber surface, and estimates of t. The force–deformation curves pertinent to those minicomposites reinforced with treated fibers exhibited the features previously observed on Nicalon/SiC minicomposites and two-dimensional woven composites reinforced with treated fibers (NLM 202) including (i) a wide curved domain, (ii) a large stress/strain at saturation of matrix cracking that coincides with ultimate failures, (iii) small residual deformations at zero load, and (iv) a gentle modulus decrease during tensile tests. The matrix cracks were deflected within each PyC sublayer of the interphase in those minicomposites reinforced with treated fibers and at the fiber surface in those minicomposites reinforced with as-received fibers. This debond pattern was similar to that observed on Nicalon/SiC composites.3,10 Deviation within the interphase resulted from the presence of a stronger bond between the interphase and the fiber.6 In those minicomposites reinforced with as-received fibers, weakening of the fiber/interphase region has been evidenced using SEM. The preexisting debond cracks were attributed to the lateral contraction of as-received Hi-Nicalon fibers during minicomposite processing. Yun et al.20 have shown that the lateral contraction of as-received Hi-Nicalon fibers during processing of composites can exceed 1% at 1200°C. The contraction was associated with fiber shrinkage and with the transformation of the SiC amorphous phase into a b-SiC crystallized phase.21 The interfacial bond was characterized by the interfacial shear stress (t). As previously mentioned, all the methods indicated that the fiber/coating bond was stronger in those minicomposites reinforced with treated fibers. However, the range of t data was narrower than that observed on Nicalon/SiC two-dimensional woven composites. In those Nicalon/SiC composites, the respective t values were different by more than 1 order of magnitude: t ' 10 MPa for those composites reinforced with as-received fibers and t . 200 MPa for those reinforced with treated fibers.3 In the present study, t ' 100 MPa for those Hi-Nicalon/SiC minicomposites reinforced with as-received fibers, and t ' 200 MPa for those reinforced with treated fibers. A contribution of twisting could be expected. Twisting generated radial compressive stresses that enhanced fiber/matrix interactions and can tend to Fig. 7. Comparison of predicted and experimental force deformation curves for SiC/SiC minicomposites (a) HN/(C/SiC)10 and (b) HNT/(C/SiC)10. Fig. 6. Micrographs showing deflection of matrix microcracks in interfacial region of minicomposites: (a) SEM micrograph of a HN/(C/SiC)10 minicomposite, (b) TEM micrograph of a HNT/(C/SiC)10 minicomposite. April 2001 Influence of Interfaces on Mechanical Behavior and Lifetime of Hi-Nicalon/(PyC/SiC)n/SiC 791