《复合材料 Composites》课程教学资源（学习资料）第五章陶瓷基复合材料_j.1744-7402.2010.02485.x[1].pdf_大学文库

International Journal of Applied Ceramic Technolog-Sauder, Brusson and Lamon Vol. 7, No. 3, 2010 Pic The previous generations of SiC/SiC composites inal chemical bonding may be enhanced by mechanical ared via CVi and reinforced with nonstoichiomet- bonding as a result of thermally induced residual ic fibers(such as Nicalon or Hi-Nicalon fibers)possess stresses, which leads to gripping of the fiber by the ma- interesting mechanical properties like high trix. Radial gripping is enhanced when the interface is hness, high strength, and damage tolerance. How- rough. A thick fber coating(porous or pyrocarbon- ever,these fibers undergo shrinkage during irradiation based interphase) is used to reduce the clamping stresses by fast neutrons, which causes fiber/ matrix debonding induced by thermal expansion mismatch or surface and degradation of mechanical properties. By contrast, roughness. It has been reported that the SA3-1 the Hi-NicalonS and SA3 fibers exhibit microstructural forced SiC/SiC composites exhibit a brittle behavior stability because of a well-crystallized microstructure when the PyC interphase is very thin. .Then, it has and a lower impurity content, like CVD/CVI SiC. been shown experimentally that the failure strain in- Thus, it is necessary to investigate the mechanical be- creases with terphase thickness, whereas interfa- havior of CVI SiC/SiC composites reinforced by Hi- cial shear stress decreases. Various ranges of thickness NicalonS and SA3 fibers to check whether they can have been examined: 600 nm. hike mechanical properties and thermal conductivity sic gnis Morscher has ins estigfard fiber effets high operating (up to 1100C)and accidental polycrystalline fber. The Sylramic-reinforced minicom- (>1600C)temperatures, and posites had lower strain to failure and higher interfacial fast neutron and high fuence conditions. shear stress, compared with the Hi-Nicalon-reinforced Although both fibers exhibit comparable mechan- ones. This trend was attributed to the surface roughness ical properties, available data indicate that the tensile of the Sylramic fiber, which was reportedly quite resistance of composites is somewhat different: the significant: mean amplitude 69 nm against 4.3 nm for strain-to-failure of SA3 fiber-reinforced composites the Hi-Nicalon fiber. It is worth pointing out that (see for instance nanoinfiltration transient eutectic the mechanical behavior was not brittle, so that inter phase-sintered materials) is generally much smaller facial shear stress t in the range 60-160 MPa could be (<0. 3%)than that of the Hi-NicalonS-reinforced determined. The interfacial shear stress t increased ones(x 0.6-1%). This effect has not been satisfacto- with the applied stress, which was attributed to in- ly elucidated yet, although it was attributed to SA creased resistance to sliding with fber surface roughness in. Not much attention ha been paid to this surface roughness issue, and the cor- In order to analyze properly the role of surface relation between mechanical behavior and microstruc Important a ural features has not been well investigated. There are the reader of the elementary interface phenomena that some noticeable differences in the microstructure of are associated with matrix cracking under a uniaxial SA3 and Hi-NicalonS fibers. The average grain size of tensile load. When a crack initiates in the matrix SA3 fibers is reportedly 200 nm; the grain size increases pendicular to the fiber axis, its deflection results from from the core to the surface: grains as large as 400 nm coalescence with the interface crack that forms ahead of have been detected. By contrast, Hi-NicalonS contain the matrix crack tip, as a result of a polyaxial stress much smaller grains(R 20 nm)that are quite uni- state, 4. 5 and more particularly the stress component formly distributed. The presence of large grains results Orr perpendicular to the interface. Debonding is thus in a rough surface of SA3 fibers governed by the interface resistance to orr(opening The fiber/matrix interfacial domain is one of the strength). After deviation of the matrix crack, the load is decisive constituents of fiber-reinforced ceramic matrix transferred through the interface crack(composite be composites. Depending on the characteristics of this havior). The interface shear stress t reflects this load domain, the material either shows a brittle ceramic or a transfer. The higher the t, the shorter the interface crack damage-tolerant composite. Fiber failures can be pre- in the absence of roughness effects from the fiber sur- vented by crack deflection. For this purpose, the fber/ face. The sliding resistance may increase with the matrix interface must not be too strong. It can be tai applied stress because of roughness effects, leading lored via a pyrocarbon interphase. At the interface, ori to an increase in t

The previous generations of SiC/SiC composites prepared via CVI and reinforced with nonstoichiometric fibers (such as Nicalon or Hi-Nicalon fibers) possess several interesting mechanical properties like high toughness, high strength, and damage tolerance.4 However, these fibers undergo shrinkage during irradiation by fast neutrons, which causes fiber/matrix debonding and degradation of mechanical properties. By contrast, the Hi-NicalonS and SA3 fibers exhibit microstructural stability because of a well-crystallized microstructure and a lower impurity content, like CVD/CVI SiC.1 Thus, it is necessary to investigate the mechanical behavior of CVI SiC/SiC composites reinforced by HiNicalonS and SA3 fibers to check whether they can meet stringent material requirements for fuel cladding: high mechanical properties and thermal conductivity under: high operating (up to 11001C) and accidental (416001C) temperatures, and fast neutron and high fluence conditions. Although both fibers exhibit comparable mechanical properties, available data indicate that the tensile resistance of composites is somewhat different: the strain-to-failure of SA3 fiber-reinforced composites (see for instance nanoinfiltration transient eutectic phase-sintered materials) is generally much smaller (o0.3%) than that of the Hi-NicalonS-reinforced ones ( 0.6–1%).5 This effect has not been satisfactorily elucidated yet, although it was attributed to SA3 fiber surface roughness in.5 Not much attention has been paid to this surface roughness issue, and the correlation between mechanical behavior and microstructural features has not been well investigated. There are some noticeable differences in the microstructure of SA3 and Hi-NicalonS fibers. The average grain size of SA3 fibers is reportedly 200 nm; the grain size increases from the core to the surface6 : grains as large as 400 nm have been detected. By contrast, Hi-NicalonS contain much smaller grains ( 20 nm) that are quite uniformly distributed. The presence of large grains results in a rough surface of SA3 fibers. The fiber/matrix interfacial domain is one of the decisive constituents of fiber-reinforced ceramic matrix composites. Depending on the characteristics of this domain, the material either shows a brittle ceramic or a damage-tolerant composite. Fiber failures can be prevented by crack deflection. For this purpose, the fiber/ matrix interface must not be too strong. It can be tailored via a pyrocarbon interphase. At the interface, original chemical bonding may be enhanced by mechanical bonding as a result of thermally induced residual stresses, which leads to gripping of the fiber by the matrix. Radial gripping is enhanced when the interface is rough.7–9 A thick fiber coating (porous or pyrocarbonbased interphase) is used to reduce the clamping stresses induced by thermal expansion mismatch or surface roughness.10 It has been reported that the SA3-reinforced SiC/SiC composites exhibit a brittle behavior when the PyC interphase is very thin.11,12 Then, it has been shown experimentally that the failure strain increases with PyC interphase thickness, whereas interfacial shear stress decreases.13 Various ranges of thickness have been examined: o60 nm, 60–300 nm, and 4600 nm. Greg Morscher has investigated fiber effects on SiC/BN/SiC minicomposites reinforced with a Sylramic polycrystalline fiber.14 The Sylramic-reinforced minicomposites had lower strain to failure and higher interfacial shear stress, compared with the Hi-Nicalon-reinforced ones. This trend was attributed to the surface roughness of the Sylramic fiber, which was reportedly quite significant: mean amplitude 69 nm against 4.3 nm for the Hi-Nicalon fiber. It is worth pointing out that the mechanical behavior was not brittle, so that interfacial shear stress t in the range 60–160 MPa could be determined. The interfacial shear stress t increased with the applied stress, which was attributed to increased resistance to sliding with increasing sliding length. In order to analyze properly the role of surface roughness, it is important at this stage to remind the reader of the elementary interface phenomena that are associated with matrix cracking under a uniaxial tensile load. When a crack initiates in the matrix perpendicular to the fiber axis, its deflection results from coalescence with the interface crack that forms ahead of the matrix crack tip, as a result of a polyaxial stress state,14,15 and more particularly the stress component srr perpendicular to the interface. Debonding is thus governed by the interface resistance to srr (opening strength). After deviation of the matrix crack, the load is transferred through the interface crack (composite behavior). The interface shear stress t reflects this load transfer. The higher the t, the shorter the interface crack in the absence of roughness effects from the fiber surface. The sliding resistance may increase with the applied stress because of roughness effects,16 leading to an increase in t. 292 International Journal of Applied Ceramic Technology—Sauder, Brusson and Lamon Vol. 7, No. 3, 2010

wwceramics. org/ACT Infuence of Interface Characteristics on the Mechanical Properties The present paper compares SiC/SiC minicompos- Table il. Main Fibers and Tow Characteristics ites reinforced with Hi-NicalonS and SA3 fibers, with the primary emphasis on the correlation between me- chanical behavior and fiber/matrix interface characteris Fibers NicalonS SA3 tics. Minicomposites are unidirectional test specimens Density (g/cm) 2.98(2.94)3.1(2.95) reinforced by a single tow. The minicomposite approach Number of fibers ivestigation ha Average diameter(ur 13 7(699) dressed in several papers.7-9 Interface characteristics Specific mass(g/1000 m) 193(191)190(192) can be extracted from features of tensile stress-strain Young's modulus(GPa) 372(375)387(385) behavior Thermal conductivity Tows ntal Procedure Strain to failure(%) 0.73(0.02)0.68(0.04) Failure stress(MPa) 2477(75)2412(122) Material Manufacture standard dev pions fo are fiber properties measured in-house ow characteristics measured in-house. SiC/SiC minicomposites were manufactured using the CVI method. Hi-NicalonS(Nippon Carbon Co. Tensile Tests Ltd, Takauchi, Japan) or Tyranno SA3(UBE Indus- tries, Tokyo, Japan) tows were used as reinforcement Uniaxial tensile tests were performed at room tem- (Table D). The tows were coated by: perature at a constant strain rate(50 um/min). The load either a single layer of pyrocarbon(PyC): thick- was measured using a 500N load cell.Minicomposite ness=150 nm deformations were measured using two-parallel linear- or a multilayer containing five layers of Pyc variable differential transformer(LVDT)extensometers alternating with SiC: each layer was 30 nm thick. that were attached to the grips. Extensometers were lo- The thickness of the PyC layer on the fiber was cated on opposite sides of specimens, in order to ensure ignment of grips were a A 40-50% fiber volume fraction was designed. tubes using glue. The tubes were gripped into the testing Fractions of fibers and matrix within minicomposites machine Gauge length(distance between the inner ends were determined using image analysis of micrographs of of the tubes)was 25 mm. The gripping system compli- polished cross-sections(Table I1). The main fiber and ance is needed to take account of load train deforma- tow characteristics are summarized in Table Il. Note tion. For compliance calibration purpose, deformations that Hi-NicalonS and SA3 exhibit comparable mechan- of a few specimens were also measured using a digital image correlation technique. This technique is based on Table I. Characteristics of SiC/SiC Minicomposites(e= thickness of the PyC layers Name Fiber Interphase(s) thickness Section (mm) V(%) Single layer ex 150 nm e layer e 150 nm 43 Multilayer(x5)el=22=23=e4~ nm 0.140 e5~150nm M4 SA3 Multilayer(x5)el=22=63=e4- M5 HiS Multilayer(×5)el=c2=e3=姓 0.145 e5-150nm M6 His Single layer e 30 nm

The present paper compares SiC/SiC minicomposites reinforced with Hi-NicalonS and SA3 fibers, with the primary emphasis on the correlation between mechanical behavior and fiber/matrix interface characteristics. Minicomposites are unidirectional test specimens reinforced by a single tow. The minicomposite approach to composite design and investigation has been addressed in several papers.17–19 Interface characteristics can be extracted from features of tensile stress–strain behavior.17–19 Experimental Procedure Material Manufacture SiC/SiC minicomposites were manufactured using the CVI method. Hi-NicalonS (Nippon Carbon Co. Ltd., Takauchi, Japan) or Tyranno SA3 (UBE Industries, Tokyo, Japan) tows were used as reinforcement (Table I). The tows were coated by: either a single layer of pyrocarbon (PyC): thickness 5 150 nm, or a multilayer containing five layers of PyC alternating with SiC: each layer was 30 nm thick. The thickness of the PyC layer on the fiber was 30 nm. A 40–50% fiber volume fraction was designed. Fractions of fibers and matrix within minicomposites were determined using image analysis of micrographs of polished cross-sections (Table II). The main fiber and tow characteristics are summarized in Table II. Note that Hi-NicalonS and SA3 exhibit comparable mechanical properties. Tensile Tests Uniaxial tensile tests were performed at room temperature at a constant strain rate (50 mm/min). The load was measured using a 500 N load cell. Minicomposite deformations were measured using two-parallel linearvariable differential transformer (LVDT) extensometers that were attached to the grips. Extensometers were located on opposite sides of specimens, in order to ensure alignment of grips. Minicomposite ends were affixed within metallic tubes using glue. The tubes were gripped into the testing machine. Gauge length (distance between the inner ends of the tubes) was 25 mm. The gripping system compliance is needed to take account of load train deformation. For compliance calibration purpose, deformations of a few specimens were also measured using a digital image correlation technique. This technique is based on Table I. Characteristics of SiC/SiC Minicomposites (e 5 thickness of the PyC layers) Name Fiber Interphase(s) thickness Section (mm2 ) Vf (%) M1 HiS Single layer eB150 nm 0.140 46 M2 SA3 Single layer eB150 nm 0.150 43 M3 HiS Multilayer ( 5) e1 5 e2 5 e3 5 e4B30 nm 0.140 46 e5B150 nm M4 SA3 Multilayer ( 5) e1 5 e2 5 e3 5 e4B30 nm 0.150 43 e5B150 nm M5 HiS Multilayer ( 5) e1 5 e2 5 e3 5 e4B30 nm 0.145 44 e5–150 nm M6 HiS Single layer eB30 nm 0.160 40 Table II. Main Fibers and Tow Characteristics Fibers HiNicalonS SA3 Density (g/cm3 ) 2.98 (2.94) 3.1 (2.95) Number of fibers 500 1600 Average diameter (mm) 13 7 (6.99) Specific mass (g/1000 m) 193 (191) 190 (192) Young’s modulus (GPa) 372 (375) 387 (385) Thermal conductivity (W m1 K1 ) 18 65 Tows Strain to failure (%) 0.73 (0.02) 0.68 (0.04) Failure stress (MPa) 2477 (75) 2412 (122) Data within parentheses are fiber properties measured in-house or standard deviations for tow characteristics measured in-house. www.ceramics.org/ACT Influence of Interface Characteristics on the Mechanical Properties 293

wwceramics. org/ACT Infuence of Interface Characteristics on the Mechanical Properties where SA is the width of the hysteresis loop, o is the applied stress in the unloading sequen sponds to 8A,Op the initial stress level at unloading, Ec the initial Youngs modulus of the minicomposite, Rf the fiber radius, v the Poissons ratio(v=Vm=ve, E the Youngs modulus of the matrix, Ef that of the fiber, and Vr the volume fraction of the fiber. Nis the number 100 of matrix cracks in the gauge length n was determined by ectron microsco (SEM)inspection of specimens after failure. In order to detect all the cracks, the specimens were etched using Murakami,s reagent, to achieve a selective attack of SiC t the surface of the cracks Acoustic emission was recorded during the 0.203040 ignals were analyzed by the number of counts for a 50 dB preset amplitude threshold. The fracture surface Fig. 2. Tensile stress-strain curves obtained for 2D wowen Hi- of the specimens was examined after the tests using the NicalonS/SiC composites. SEM. Fiber surface roughness was examined using atomic force microscopy (AFM)and the SEM stresses were negligibly small in the matrix. This will Results be discussed in a subsequent section. The first following conclusions can be drawn from Tensile Stress-Strain Bebavior of Hi-Nicalon S/SiC these features. on the basis of microstructure/behavion Minicomposite relations that have been established in previous pa pers. 17. They are completed by the data extracted he typical stress-strain curves shown in Fig. I re- from tensile stress-strain curves and summarized in fect the conventional composite behavior that has been Table ill: previously observed on SiC/SiC CVI minicomposite Saturation of matrix cracking was generally ob- einforced by Nicalon or Hi-NicalonS fibers 4. 8. I served for strains that were quite small when They compare fairly well with the behavior of 2D Hi- failure strain this refects the NicalonS/SiC composites(Fig. 2). They are markedly presence of long debond cracks and rather weak nonlinear, coupled with the following typical features Strain to failure compared fairly well with that of (a)an initial linear domain of elasticity, dry tows, which indicates that fiber contribution (b)a curved ulting from matrix cracking to mechanical behavior was altered neither by and associated fiber debonding. The boundary of this processing(which was expected) nor by local region is indicated by curve inflection at deformations stress concentrations c0.3% indicating saturation of matrix cracking, (c) a linear domain after saturation, attributed to The tensile characteristics reported in Table Ill as lastic deformation of fibers well as Fig. 1 clearly show that the behavior of mini- maximum composites was well reproducible whatever the interface force, indicating fber breaks, and There is no clear influence of Py C fiber coating thick (e)ultimate failure at deformations close to 0.7% ness or the number of layers in the interphase. which Interface shear stresses that have been extracted compare fairly well with the strains to failure from the hysteresis loops as well as matrix crack spac- measured on tows(Table II The onset of matrix cracking was identified by the ing distances at saturation(Table IID), indicate the same first acoustic emission signals, at deformations a.1%. trend and they support the above statements. The in- This strain corresponds to brittle failure of monolithic terface shear stresses are quite lowTable IID). Weak SiC, which suggests that thermally induced residual interfaces could be logically expected for this fber/in-

where dD is the width of the hysteresis loop, s is the applied stress in the unloading sequence that corresponds to dD, sp the initial stress level at unloading, Ec the initial Young’s modulus of the minicomposite, Rf the fiber radius, n the Poisson’s ratio (n 5 nm 5 nf), Em the Young’s modulus of the matrix, Ef that of the fiber, and Vf the volume fraction of the fiber. N is the number of matrix cracks in the gauge length. N was determined by scanning electron microscopy (SEM) inspection of specimens after failure. In order to detect all the cracks, the specimens were etched using Murakami’s reagent, to achieve a selective attack of SiC at the surface of the cracks. Acoustic emission was recorded during the tests. Signals were analyzed by the number of counts for a 50 dB preset amplitude threshold. The fracture surface of the specimens was examined after the tests using the SEM. Fiber surface roughness was examined using atomic force microscopy (AFM) and the SEM. Results Tensile Stress–Strain Behavior of Hi-NicalonS/SiC Minicomposites The typical stress–strain curves shown in Fig. 1 re- flect the conventional composite behavior that has been previously observed on SiC/SiC CVI minicomposites reinforced by Nicalon or Hi-NicalonS fibers.4,18,19 They compare fairly well with the behavior of 2D HiNicalonS/SiC composites (Fig. 2). They are markedly nonlinear, coupled with the following typical features (Fig. 1): (a) an initial linear domain of elasticity, (b) a curved region resulting from matrix cracking and associated fiber debonding. The boundary of this region is indicated by curve inflection at deformations 0.3% indicating saturation of matrix cracking, (c) a linear domain after saturation, attributed to elastic deformation of fibers, (d) a slight curvature preceding the maximum force, indicating fiber breaks, and (e) ultimate failure at deformations close to 0.7%, which compare fairly well with the strains to failure measured on tows (Table II). The onset of matrix cracking was identified by the first acoustic emission signals, at deformations 0.1%. This strain corresponds to brittle failure of monolithic SiC, which suggests that thermally induced residual stresses were negligibly small in the matrix. This will be discussed in a subsequent section. The first following conclusions can be drawn from these features, on the basis of microstructure/behavior relations that have been established in previous papers.4,17,18 They are completed by the data extracted from tensile stress–strain curves and summarized in Table III: Saturation of matrix cracking was generally observed for strains that were quite small when compared with failure strain. This reflects the presence of long debond cracks and rather weak interfaces. Strain to failure compared fairly well with that of dry tows, which indicates that fiber contribution to mechanical behavior was altered neither by processing (which was expected) nor by local stress concentrations. The tensile characteristics reported in Table III as well as Fig. 1 clearly show that the behavior of minicomposites was well reproducible whatever the interface. There is no clear influence of PyC fiber coating thickness or the number of layers in the interphase. Interface shear stresses that have been extracted from the hysteresis loops as well as matrix crack spacing distances at saturation (Table III), indicate the same trend and they support the above statements. The interface shear stresses are quite low (Table III). Weak interfaces could be logically expected for this fiber/inFig. 2. Tensile stress–strain curves obtained for 2D woven HiNicalonS/SiC composites. www.ceramics.org/ACT Influence of Interface Characteristics on the Mechanical Properties 295

terphase system because untreated fibers have been used.4,17,18 The current values of t fall within the range of data determined on SiC/SiC minicomposites reinforced with as-received Nicalon or Hi-Nicalon fi- bers.4,18,22 It is worth pointing out that t seems to depend on the thickness of the PyC fiber coating: comparable t was obtained for specimens M1, M3, and M5 for which the PyC fiber coating thickness was 150 nm. A larger t was obtained for specimen M6 with a thinner coating of 30 nm. Fracture surface images show a graceful composite failure with pullout fibers of comparable length (Fig. 3). Tensile Stress–Strain Behavior of SA3/SiC Minicomposites The tensile behavior of the SA3/SiC minicomposites was considerably erratic (Fig. 4): For some test specimens, it was comparable with that of Hi-NicalonS/SiC minicomposites (Fig. 3), exhibiting the conventional features of SiC/ SiC minicomposites. Note the curve inflection that was obtained at a larger strain (0.5%), which reflects a higher density of matrix cracks and stronger fiber/matrix interfaces. The larger Table III. Main Characteristics of the Tensile Behavior of Minicomposites. Hi-NicalonS/SiC E (GPa) rr (MPa) eR (%) rl(MPa) el (%) s (MPa) ds (lm) M1 400 (10) 870 (40) 0.61 (0.04) 250 (40) 0.07 (0.01) 9 (1) 350 (20) M3 390 (40) 980 (30) 0.69 (0.05) 350 (50) 0.10 (0.02) 10 (4) 340 (170) M5 420 (20) 880 (40) 0.63 (0.07) 410 (60) 0.10 (0.01) 16 (3) 220 (40) M6 410 (21) 830 (20) 0.60 (0.03) 320 (30) 0.07 (0.01) 33 (12) 173 (26) SA3/SiC E (GPa) rr(MPa) eR(%) rl (MPa) el (%) s (MPa) ds (lm) M2 420 (20) 855 (255) 0.38 (0.25) 290 (30) 0.08 (0.02) 280 () 40 () M4 415 (40) 600 (120) 0.15 (0.02) 430 (50) 0.11 (0.02) — — Standard deviations are given in parentheses. One test—not measured. E, Young’s modulus; sr, failure stress; eR, failure strain; s1, stress at the onset of matrix cracking; e1, strain at the onset of matrix cracking; t, interfacial shear stress; ds, crack spacing distance at saturation. Fig. 3. Tensile stress–strain curves obtained for SA3/SiC minicomposites: (a) M2 and (b) M4. 296 International Journal of Applied Ceramic Technology—Sauder, Brusson and Lamon Vol. 7, No. 3, 2010

wwceramics. org/ACT Infuence of Interface Characteristics on the Mechanical Properties and that it tends to be quite brittle when the fiber coat ing becomes thinner Table iv also reports the interface shear stress mea sured on the damage-tolerant specimens. It is much posites, with the same PyC fiber coating thickness. It could not be determined on the quite brittle test spec- imens, due to the lack of hysteresis loops. Figure 5 shows that the fracture surface of the damage-tolerant SA3/SiC minicomposites contained quite long pullout fibers. A few short pullout fibers can be observed in the fracture surface of the quite brittle minicomposites. This confirms that matrix cracks formed before ultimate fail- and that fiber/matrix debonding occurred. Th 0 200M pening width of hysteresis loops can also be noticed, as well as tow-controlled ultimate failure Fiber Surface Roughness at a deformation 0.6% For the other specimens from batch M2 and for Figure 6 shows images of the surface of pullout all the rom batch ma. the behar fibers during tensile tests. They demonstrate that devi was essentially linear and brittle like. But it is ation of the matrix crack occurred at the fiber surface in worth pointing out that the strain to failure was both Hi-NicalonS/SiC and A3/SiC. minicomposites slightly larger than 0.1% and that acoustic emis- Thus, it can be concluded that bonding of SA3 fibers signals initiated before failure, which indi- was not significantly stronger than that of Hi-NicalonS cates that the first matrix crack did not cause fibers, and it can be reasonably conjectured that the in- minIco terface opening strength was comparable in both sys tems. It was not sufficiently high to cause deviation Table IV shows that for those specimens from within the interphase, as it has been observed when th batch M2 that exhibited the conventional features of surface of SiC fibers has been treated to strengthen the omposite behavior, the tensile characteristics are similar Then, the surface of SA3 fibers appears to be rather to those of Hi-NicalonS/SiC minicomposites. This rough, with peaks and valleys confirms that SA3 fibers were not degraded during pro- cessing because SA3 fibers strength Micrographs of cross-sections as well as AFM im- 1400..2324 For those specimens from batch M4, the ages of fiber surface highlight this topography(Figs. 6 failure characteristics are very close to those at the onset and 7). Table IV summarizes the fiber surface roughness highly dependent on thickness of the Py C fiber coating, tude observed in the analyzed portion of the fiber sur- (r Table IV. Characteristics of Fiber Surface Rough- face)clearly indicate that the surface roughness ness Determined Using AFM and the Magnitude of amplitude of SA3 fibers is much more significant than Clamping Stresses oR (Eq(4)) that of Hi-NicalonS fibers. Note that Rmax may be as high as 60 nm for the SA3 fibers, which is much larger RRMs(nm) Rmax(nm) OR(MPa) than the thickness of the PyC coating(epyc) in the M4 It should also be noticed that the diam- 2.33 eter is three times as small as that of Hi-Nicalot Fiber SA3 558 930 whereas the roughness amplitude is about four times as

opening width of hysteresis loops can also be noticed, as well as tow-controlled ultimate failure at a deformation B0.6%. For the other specimens from batch M2 and for all the specimens from batch M4, the behavior was essentially linear and brittle like. But it is worth pointing out that the strain to failure was slightly larger than 0.1% and that acoustic emission signals initiated before failure, which indicates that the first matrix crack did not cause minicomposite failure. Table IV shows that for those specimens from batch M2 that exhibited the conventional features of composite behavior, the tensile characteristics are similar to those of Hi-NicalonS/SiC minicomposites. This confirms that SA3 fibers were not degraded during processing because SA3 fibers strength is stable up to 14001C.23,24 For those specimens from batch M4, the failure characteristics are very close to those at the onset of matrix cracking. It seems obvious that the behavior is highly dependent on thickness of the PyC fiber coating, and that it tends to be quite brittle when the fiber coating becomes thinner. Table IV also reports the interface shear stress measured on the damage-tolerant specimens. It is much larger than t obtained for Hi-NicalonS/SiC minicomposites, with the same PyC fiber coating thickness. It could not be determined on the quite brittle test specimens, due to the lack of hysteresis loops. Figure 5 shows that the fracture surface of the damage-tolerant SA3/SiC minicomposites contained quite long pullout fibers. A few short pullout fibers can be observed in the fracture surface of the quite brittle minicomposites. This confirms that matrix cracks formed before ultimate failure, and that fiber/matrix debonding occurred. The damage-tolerant composite behavior exists, but it is not reflected by the stress–strain curve because it is very limited. Fiber Surface Roughness Figure 6 shows images of the surface of pullout fibers during tensile tests. They demonstrate that deviation of the matrix crack occurred at the fiber surface in both Hi-NicalonS/SiC and SA3/SiC minicomposites. Thus, it can be concluded that bonding of SA3 fibers was not significantly stronger than that of Hi-NicalonS fibers, and it can be reasonably conjectured that the interface opening strength was comparable in both systems. It was not sufficiently high to cause deviation within the interphase, as it has been observed when the surface of SiC fibers has been treated to strengthen the fiber/coating bond.25 Then, the surface of SA3 fibers appears to be rather rough, with peaks and valleys. Micrographs of cross-sections as well as AFM images of fiber surface highlight this topography (Figs. 6 and 7). Table IV summarizes the fiber surface roughness amplitudes determined using the AFM. Both RRMS (root mean square) and Rmax (maximum peak amplitude observed in the analyzed portion of the fiber surface) clearly indicate that the surface roughness amplitude of SA3 fibers is much more significant than that of Hi-NicalonS fibers. Note that Rmax may be as high as 60 nm for the SA3 fibers, which is much larger than the thickness of the PyC coating (ePyC) in the M4 test specimens. It should also be noticed that the diameter is three times as small as that of Hi-NicalonS, whereas the roughness amplitude is about four times as Fig. 4. Scanning electron micrograph of the fracture surface of a Hi-NicalonS/SiC minicomposite test specimen. Table IV. Characteristics of Fiber Surface Roughness Determined Using AFM and the Magnitude of Clamping Stresses rR (Eq. (4)) RRMS (nm) Rmax (nm) rR (MPa) Fiber HiS 2.33 15.2 130 Fiber SA3 8.04 55.8 930 www.ceramics.org/ACT Influence of Interface Characteristics on the Mechanical Properties 297

large. In the Hi-NicalonS/SiC minicomposites, Rmax is much smaller than ePyC. The radial stress induced by the fiber surface roughness amplitude in the interface crack during tensile loading was estimated using the following equation10: sR ¼ EmEf Efð1 þ nmÞ þ Emð1 nfÞ A Rf ð4Þ where A is the amplitude of lateral displacements induced by surface roughness (Fig. 8) and Rf is the fiber radius. Setting A to 2RRMS yields sR 5 930 MPa for SA3/ SiC and sR 5 130 MPa for Hi-NicalonS. These are rough estimates of the average values of sR in the absence of a thin interphase. Then, setting A to hmax yields upper bounds: sR 5 3080 MPa for SA3/SiC and sR 5 850 MPa for Hi-NicalonS/SiC. These values are useful for comparison purposes. They clearly indicate that significant effects of fiber surface roughness are expected with the SA3 fibers, when compared with HiNicalonS. The influence of PyC fiber coating can be anticipated using Eq. (5): A h ePyCð1 ePyCÞ ð5Þ where h is the peak amplitude and ePyC is interphase deformation during sliding, ePyCo1%. However, it may depend on the PyC structure. But this should not affect the analysis drastically, provided ePyC does not exceed 1%. Equations (4) and (5) indicate that thicker coatings decrease A and sR. The clamping stress sR-0 when the coating thickness exceeds the peak amplitude. Thus, it appears that differences in minicomposite behaviors are related to contribution of fiber surface roughness in the interface crack rather than to fiber/ matrix interface opening strength. Residual Stresses Residual stresses build up during cooling down from the processing temperature ( 10001C) depending on the thermal expansion mismatch. Figure 9 shows that fibers and the CVI matrix have comparable coefficients of thermal expansion,26 suggesting Fig. 5. Micrographs of fracture surfaces of SA3/SiC minicomposites: (a) M2 with composite behavior, (b) M2 with brittle-like behavior, and (c) M4. 298 International Journal of Applied Ceramic Technology—Sauder, Brusson and Lamon Vol. 7, No. 3, 2010

《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_j.1744-7402.2010.02485.x[1]

《复合材料 Composites》课程教学资源（学习资料）第五章陶瓷基复合材料_j.1744-7402.2010.02485.x[1]