《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.1744-7402.2009.02469.x[1]

at.y. Appl Ceram. Technol, 7/ 3)304-315(2010) DOk:10.II1117447402200902469x International Journal o pplied Ceramic TECHNOLOGY ceramic Product D Cracking Resistance of Silicon Carbide Composites by Single- and Double-Notched Specimen Techniques Takashi Nozawaand Hiroyasu Tanigawa Japan Atomic Energy Agency, Tokai, Ibaraki 319-1195, Japan 66. This paper aims to evaluate fracture resistance of nanoinfiltration transient eutectic phase-sintered(NIT -reinforced silicon carbide matrix(SiC/SiC)composites, using the single-edge notched bend(SENB)test notch tensile(DNT)test. Both SENB and DNT test results identified notch insensitivity of NITE-SiC/SiC es. With the fact of notch insensitivity, unique proportional limit stress, and fracture strength were identified regardless of the presence of artificial notches. By applying the nonlinear fracture mechanics, micro- and macrocracking energies were separately stimated. Specifically, lower microcrack formation energy was identified compared with conventional low-stiffness composite system. Finally, superior crack resistance of NITE-SiC/SiC composites was clearly demonstrated Introduction line and near-stoichiometric SiC itself has inherently good chemical stability at high temperatures, strength Highly crystalline and near-stoichiometric silicon retention, specific strength, and low activation/low af- carbide fiber-reinforced silicon carbide matrix(SiC/SiC) ter-heat properties Of many composite types, new class composites are attractive materials for nuclear fusion "nuclear-grade"SiC/SiC composites namely (1)robust and advanced fission energy systems, as well as nonnu- and dense SiC/SiC composites produced by the nano- clear high-temperature industries, because high cryst infiltration transient eutectic(NITE) phase-sintered process and (2) high-purity SiC/SiC composites by This research was supported by a collaboration between the Japan Atomic Energy Ag the chemical vapor infiltration(CVD) and Institute of Energy Science and Technology Co. Ltd. This work was partly sponsor ising with perceived merits like excellent baseline me- the Japan Society for the Promotion of Science (SPS)under the aor 9860089, Grant-in-Aid for Young Scientists (Start-up). of KAKENHI chanical properties and proven radiation stability of microstructure and strength under certain irradiation 2009 The American Ceramic Society nditions. Specifically, the good gas tightness of the

Cracking Resistance of Silicon Carbide Composites by Single- and Double-Notched Specimen Techniques Takashi Nozawa* and Hiroyasu Tanigawa Japan Atomic Energy Agency, Tokai, Ibaraki 319-1195, Japan This paper aims to evaluate fracture resistance of nanoinfiltration transient eutectic phase-sintered (NITE)–silicon carbide fiber-reinforced silicon carbide matrix (SiC/SiC) composites, using the single-edge notched bend (SENB) test and the double￾notch tensile (DNT) test. Both SENB and DNT test results identified notch insensitivity of NITE–SiC/SiC composites. With the fact of notch insensitivity, unique proportional limit stress, and fracture strength were identified regardless of the presence of artificial notches. By applying the nonlinear fracture mechanics, micro- and macrocracking energies were separately estimated. Specifically, lower microcrack formation energy was identified compared with conventional low-stiffness composite system. Finally, superior crack resistance of NITE–SiC/SiC composites was clearly demonstrated. Introduction Highly crystalline and near-stoichiometric silicon carbide fiber-reinforced silicon carbide matrix (SiC/SiC) composites are attractive materials for nuclear fusion and advanced fission energy systems, as well as nonnu￾clear high-temperature industries, because high crystal￾line and near-stoichiometric SiC itself has inherently good chemical stability at high temperatures, strength retention, specific strength, and low activation/low af￾ter-heat properties. Of many composite types, new class ‘‘nuclear-grade’’ SiC/SiC composites namely (1) robust and dense SiC/SiC composites produced by the nano￾infiltration transient eutectic (NITE) phase-sintered process and (2) high-purity SiC/SiC composites by the chemical vapor infiltration (CVI) process are prom￾ising with perceived merits like excellent baseline me￾chanical properties1–3 and proven radiation stability of microstructure and strength under certain irradiation conditions.4 Specifically, the good gas tightness of the Int. J. Appl. Ceram. Technol., 7 [3] 304–315 (2010) DOI:10.1111/j.1744-7402.2009.02469.x Ceramic Product Development and Commercialization This research was supported by a collaboration between the Japan Atomic Energy Agency and Institute of Energy Science and Technology Co. Ltd. This work was partly sponsored by the Japan Society for the Promotion of Science (JSPS) under the contract of KAKENHI 19860089, Grant-in-Aid for Young Scientists (Start-up). *nozawa.takashi67@jaea.go.jp r 2009 The American Ceramic Society

wwceramics. org/ACT Cracking Resistance of Silicon Carbide Composites dense NITE-SiC/SiC composites would provide an ad- would impact the failure behavior. Of particular impor ditional advantage to apply this to gas-cooled systems. tance is then to separately discuss micro-and macrocrack- Accordingly, several fusion blanket designs propose to ing behaviors. A simple guess indicates that, as the failure utilize this class of SiC/SiC composites. With the behavior generally depends on the type and distribution of completion of the "proof-of-principle"phase, the R&d the internal Aaws as a crack origin, there would be an ap- on SiC/SiC composites is now shifting to the more parent difference between dense NITE-SiC/SiC compos pragmatic phase like material data generation and en- ites and conventional porous composites such as CVI- gineering design. SiC/SiC composites. The failure initiation behavior would One of key features of nuclear-grade SiC/SiC com- also vary from the choice of fabric architecture and applied osites is high stiffness due to the utilization of high-crys- loading directions due to the inherent anisotropy of the talline SiC for both reinforcing fibers and matrix. When composites. Therefore, careful discussion is necessary applying this class of composite materials, crack propaga This study aims to identify crack resistance and tion behavior, that is, matrix crack pop-in and crack ex- damage tolerance of a new class of SiC/SiC composites, tension, therefore, needs to be clarified first and the failure that is, NITE-SiC/SiC composites. For this purpose, the criterion needs to be properly determined to deliver in- existing testing methodology for fracture toughness eval formation for the engineering design activities, as it has uation was first applied. Second, the developmental frac- been addressed that the inherent "brittle-like "fracture ture resistance method based on nonlinear fracture (so-called"quasi-ductile"fracture) gives significant diffi- mechanics was applied and the damage accumulation be lty in engineering component design. Note that this havior of composites was evaluated using various notched metals as this quasi-ductility occurs as a result of cumu- lative accumulation of irreversible permanent damages Historically, many efforts have been devoted for the crack propagation analysis by fracture toughness testing Experimental Procedure methodology using notched specimens. -14 Rice devel- Materials oped a path-independent integral method, so-called J integral, and demonstrated the validity of this method Two types of pilot-grade NITE-SiC/SiC for approximate estimates of strain concentrations at ites were produced by the Institute of Energy Science smooth-ended notch tips in elastic and elastic-plastic and Technology(Ube, Japan)(Table I). For the first Hashida composite, a -250-nm-thick pyrolytic carbon(PyC et al. further developed the fracture testing technic interphase as a form of fiber/matrix(F/M) interface was for ceramic matrix composites, which have matrix chemically vapor deposited(CVD)on the fiber surface toughness comparable with bridging toughness induced prior matrix densification (hereafter"NITE-Thick by fibers in the fracture process zone. One drawback of Coat"). In contrast, for the other type, a thin PyC in uate the postpeak part of the stress-displacement relation- case, the F/M interfacial coating was not successful.r the Hashida's approach is the limited applicability to eval- terphase(<50 nm)was designed. However, in the latter ship. Another attempt for the cracking energy evaluation formed, as clearly shown in micrographs in Table I. The has been carried out using unloading/reloading hysteresis major parts of fibers were uncoated, although thin PyC curves of the load-displacement relationship. 12-14 This was formed for limited number of fibers. In this study, technique separately distinguishes irreversible energies we designate this composite"NITE-Thin-Coat " For from elastic energy and crack formation energy. Similarl both types, highly crystalline and near-stoichiometric for fast fracture property evaluation, such hysteresis ap- Tyranno-SA third-grade SiC fibers were uni-direc- nportant composite tionally reinforced ith a fiber volume fraction of parameters can be obtained successfully. However, due 0.4-0.45. Typical micrographs show well-densified ma- to the complex failure behaviors of ites, it is some- trix indicating that the porosity of this material was very how questionable whether these techniques are fully ap- low(<5%). A secondary phase(white contrast in Table plicable to composites. Indeed, there is a general D), which was reportedly an oxide phase composed of understanding that damage accumulation by microcrack sintering additives such as Al2O3, SiO2, and Y20 nitiates prior the complete fracture of composites and this was localized in the matrix, specifically within intrabun-

dense NITE–SiC/SiC composites would provide an ad￾ditional advantage to apply this to gas-cooled systems.5 Accordingly, several fusion blanket designs propose to utilize this class of SiC/SiC composites.6–8 With the completion of the ‘‘proof-of-principle’’ phase, the R&D on SiC/SiC composites is now shifting to the more pragmatic phase like material data generation and en￾gineering design. One of key features of nuclear-grade SiC/SiC com￾posites is high stiffness due to the utilization of high-crys￾talline SiC for both reinforcing fibers and matrix. When applying this class of composite materials, crack propaga￾tion behavior, that is, matrix crack pop-in and crack ex￾tension, therefore, needs to be clarified first and the failure criterion needs to be properly determined to deliver in￾formation for the engineering design activities, as it has been addressed that the inherent ‘‘brittle-like’’ fracture (so-called ‘‘quasi-ductile’’ fracture) gives significant diffi- culty in engineering component design. Note that this quasi-ductility is totally different from the ductility of metals as this quasi-ductility occurs as a result of cumu￾lative accumulation of irreversible permanent damages. Historically, many efforts have been devoted for the crack propagation analysis by fracture toughness testing methodology using notched specimens.9–14 Rice devel￾oped a path-independent integral method, so-called J integral, and demonstrated the validity of this method for approximate estimates of strain concentrations at smooth-ended notch tips in elastic and elastic–plastic materials.9 Based on this J integral approach, Hashida et al. 11 further developed the fracture testing technique for ceramic matrix composites, which have matrix toughness comparable with bridging toughness induced by fibers in the fracture process zone. One drawback of the Hashida’s approach is the limited applicability to eval￾uate the postpeak part of the stress–displacement relation￾ship. Another attempt for the cracking energy evaluation has been carried out using unloading/reloading hysteresis curves of the load–displacement relationship.12–14 This technique separately distinguishes irreversible energies from elastic energy and crack formation energy. Similarly, for fast fracture property evaluation, such hysteresis ap￾proach has been widely adopted and important composite parameters can be obtained successfully.15 However, due to the complex failure behaviors of composites, it is some￾how questionable whether these techniques are fully ap￾plicable to composites. Indeed, there is a general understanding that damage accumulation by microcracks initiates prior the complete fracture of composites and this would impact the failure behavior. Of particular impor￾tance is then to separately discuss micro- and macrocrack￾ing behaviors. A simple guess indicates that, as the failure behavior generally depends on the type and distribution of the internal flaws as a crack origin, there would be an ap￾parent difference between dense NITE–SiC/SiC compos￾ites and conventional porous composites such as CVI– SiC/SiC composites. The failure initiation behavior would also vary from the choice of fabric architecture and applied loading directions due to the inherent anisotropy of the composites. Therefore, careful discussion is necessary. This study aims to identify crack resistance and damage tolerance of a new class of SiC/SiC composites, that is, NITE–SiC/SiC composites. For this purpose, the existing testing methodology for fracture toughness eval￾uation was first applied. Second, the developmental frac￾ture resistance method based on nonlinear fracture mechanics was applied and the damage accumulation be￾havior of composites was evaluated using various notched specimens. Experimental Procedure Materials Two types of pilot-grade NITE–SiC/SiC compos￾ites were produced by the Institute of Energy Science and Technology (Ube, Japan) (Table I). For the first composite, a B250-nm-thick pyrolytic carbon (PyC) interphase as a form of fiber/matrix (F/M) interface was chemically vapor deposited (CVD) on the fiber surface prior matrix densification (hereafter ‘‘NITE-Thick￾Coat’’). In contrast, for the other type, a thin PyC in￾terphase (o50 nm) was designed. However, in the latter case, the F/M interfacial coating was not successfully formed, as clearly shown in micrographs in Table I. The major parts of fibers were uncoated, although thin PyC was formed for limited number of fibers. In this study, we designate this composite ‘‘NITE-Thin-Coat.’’ For both types, highly crystalline and near-stoichiometric Tyrannot-SA third-grade SiC fibers were uni-direc￾tionally reinforced with a fiber volume fraction of 0.4–0.45. Typical micrographs show well-densified ma￾trix indicating that the porosity of this material was very low (o5%). A secondary phase (white contrast in Table I), which was reportedly an oxide phase composed of sintering additives such as Al2O3, SiO2, and Y2O3, 16 was localized in the matrix, specifically within intrabun￾www.ceramics.org/ACT Cracking Resistance of Silicon Carbide Composites 305

International ournal of Applied Ceramic Technolog-Nozawa and Tanigawa Vol.7,No.3,2010 Table I. List of SiC/SiC Composites Tested NITE-Thick-Coat NITE-Thin-Coat PIP-Coat Typic microstructure Fiber Tyranno-SA3 Tyranno-SA3 Tyranno-SA3 Architecture UD Fiber volume fraction ~045 ~0.4 0.3 F/M interface 250nm Pyc None 150nm Pyc NITE-SIC NITE-SiC PIP-SiC Density 296g/cm3 319gm3212gcm3 Porosity 0.05 0.02 ~0.20 Proportional Limit 136 MPa(Tensile)204 MPa(Tensile)62 MPa(Tensile) Stress(PLS) Fracture Strength 146 MPa(Tensile) 674 MPa(Bend) 204 MPa(Tensile) 249 MPa(Tensile Elastic Modulus 336 GPa(Tensile) 27 GPa(Bend) 410 GPa(Tensile)45 GPa(Tensile) dles. This oxide phase, however, would not significantly Single-Edge Notched Bend Test(SENB) mpact test results at room temperature. For comparison, a two-dimensional (eight-harness) accumulation ehavior was evaluated by satin-woven SiC/SiC composite was fabricated by the the SENB technique. Various size st polymer impregnation and pyrolysis(PIP)method (Ube different initial notch depth(ao) were applied ("SENB- Industries, Ube, Japan)(hereafter"PIP-Coat"). Tyr- 1-4"in Fig. 1). Note that the width(W) to length(Lo) annoSA third-grade SiC fibers were applied. A ratio was fixed for all specimen types. For comparison, 100-nm-thick CVD-SiC and a subsequent x 150- unnotched bend specimens with a size of nm-thick PyC coating were formed on the fiber surface. 20 mm x 4 mm x 1.5 mm were tested. The number of This PIP process produces the SiC matrix of stoichio- test specimens is also listed in the table in Fig. 1. The metric composition(C/Si a 1)using a blend polymer fiber longitudinal direction for both specimen types of polycarbosilane and polymethylsilane. However, set perpendicular to the loading direction. Test coupons mall pores were distributed in the PIP-SiC matrix as including an artificial straight notch were machined shown in Table I, resulting in a comparably lower den- from the composite plate by a diamond saw with blade sity of x2. 1 g/cm. Details of the piP process are de thickness of <0.5 mm. The specimen surfaces were scribed in nozawa et al. I7 then polished using a standard metallographic tech

dles. This oxide phase, however, would not significantly impact test results at room temperature. For comparison, a two-dimensional (eight-harness) satin-woven SiC/SiC composite was fabricated by the polymer impregnation and pyrolysis (PIP) method (Ube Industries, Ube, Japan) (hereafter ‘‘PIP-Coat’’). Tyr￾annot-SA third-grade SiC fibers were applied. A B100-nm-thick CVD-SiC and a subsequent B150- nm-thick PyC coating were formed on the fiber surface. This PIP process produces the SiC matrix of stoichio￾metric composition (C/Si 1) using a blend polymer of polycarbosilane and polymethylsilane. However, small pores were distributed in the PIP–SiC matrix as shown in Table I, resulting in a comparably lower den￾sity of B2.1 g/cm3 . Details of the PIP process are de￾scribed in Nozawa et al.17 Single-Edge Notched Bend Test (SENB) Damage accumulation behavior was evaluated by the SENB technique. Various size specimens with a different initial notch depth (a0) were applied (‘‘SENB- 1–4’’ in Fig. 1). Note that the width (W) to length (L0) ratio was fixed for all specimen types. For comparison, unnotched bend specimens with a size of 20 mm 4 mm 1.5 mm were tested. The number of test specimens is also listed in the table in Fig. 1. The fiber longitudinal direction for both specimen types was set perpendicular to the loading direction. Test coupons including an artificial straight notch were machined from the composite plate by a diamond saw with blade thickness of o0.5 mm. The specimen surfaces were then polished using a standard metallographic technique Table I. List of SiC/SiC Composites Tested 306 International Journal of Applied Ceramic Technology—Nozawa and Tanigawa Vol. 7, No. 3, 2010

wwceramics. org/ACT Cracking Resistance of Silicon Carbide Composites ○ Materials spec. o m, ITit me. L. Depth, a ofs Fig. I. Drawing of the single-edge notched bend (SENB)test NITE-Thick-Coat with a surface finish of l um for the crack extension observation. The radius of the notch root was approx- mately 150 um. The SENB tests were conducted at Fig. 2. Drawing of the double-notch tensile(DNT) test specimen room temperature in ambient air using an electrome- chanical testing machine ng a threepoint bend fixture with a support span(L) of 16 mm for SENB-1 3 and 32 mm for SENB-4. diamond saw. The radius of the notch root was The crack opening displacement(COD) was measured 150 um. The DNT tests were conducted at room by the clip-on-type gauge. The constant crosshead dis- temperature using an electromechanical testing machine placement rate was 0.1 mm/min. Unloading/reloading with a wedge-type gripping device. The COd was mea sequences were applied to evaluate the damage accumu- sured by a pair of Cod gauges. An average of two lation behavior during the test. In this paper, test results readings was applied in the calculation. A constant of NITE-Thick-Coat are reported. No data are pres- crosshead displacement rate was 0.5 mm/min. Unload ently available for NITE-Thin-Coat. Furthermore, en- ing/ reloading sequences were applied to evaluate the tire tests for PIP-Coat were invalid because of the damage accumulation behavior during the tests. The compressive failure at the loading point. Details are de- DNT tests were applied to all SiC/SiC composite types scribed in nozawa et al listed in table i Double-Notch Tensile(DNT) Test Figure 2 shows a photo image of the test setup and Microstructural Observation a drawing of the dNT test specimen. Notched tensile A crack extension length from the initial artificial specimens with a different initial notch depth(ao)were tch was measured using optical microscopy. For this used. For comparison, unnotched rectangular tensile purpose, replica films of the surface microstructure of the specimens (40 mm x 4mm x 1.5 mm) with a gauge specimen were taken at each unloading point for SeNB length of 15 mm were tested. The number of test spec- tests. For DNT tests, the crack extension length could not mens are also listed in Fig. 2. The fiber longitudinal be measured primarily due to unstable crack extension just direction was set parallel to the loading direction. Test prior fracture. Fracture surfaces were observed using field- hined from the composite plate by mission scanning electron microscopy

with a surface finish of B1 mm for the crack extension observation. The radius of the notch root was approx￾imately 150 mm. The SENB tests were conducted at room temperature in ambient air using an electrome￾chanical testing machine. Test specimens were loaded using a threepoint bend fixture with a support span (L) of 16 mm for SENB-1 B3 and 32 mm for SENB-4. The crack opening displacement (COD) was measured by the clip-on-type gauge. The constant crosshead dis￾placement rate was 0.1 mm/min. Unloading/reloading sequences were applied to evaluate the damage accumu￾lation behavior during the test. In this paper, test results of NITE-Thick-Coat are reported. No data are pres￾ently available for NITE-Thin-Coat. Furthermore, en￾tire tests for PIP-Coat were invalid because of the compressive failure at the loading point. Details are de￾scribed in Nozawa et al.18 Double-Notch Tensile (DNT) Test Figure 2 shows a photo image of the test setup and a drawing of the DNT test specimen. Notched tensile specimens with a different initial notch depth (a0) were used. For comparison, unnotched rectangular tensile specimens (40 mm 4 mm 1.5 mm) with a gauge length of 15 mm were tested. The number of test spec￾imens are also listed in Fig. 2. The fiber longitudinal direction was set parallel to the loading direction. Test coupons were machined from the composite plate by a diamond saw. The radius of the notch root was B150 mm. The DNT tests were conducted at room temperature using an electromechanical testing machine with a wedge-type gripping device. The COD was mea￾sured by a pair of COD gauges. An average of two readings was applied in the calculation. A constant crosshead displacement rate was 0.5 mm/min. Unload￾ing/reloading sequences were applied to evaluate the damage accumulation behavior during the tests. The DNT tests were applied to all SiC/SiC composite types listed in Table I. Microstructural Observation A crack extension length from the initial artificial notch was measured using optical microscopy. For this purpose, replica films of the surface microstructure of the specimen were taken at each unloading point for SENB tests. For DNT tests, the crack extension length could not be measured primarily due to unstable crack extension just prior fracture. Fracture surfaces were observed using field￾emission scanning electron microscopy. Fig. 2. Drawing of the double-notch tensile (DNT) test specimen. Fig. 1. Drawing of the single-edge notched bend (SENB) test specimen. www.ceramics.org/ACT Cracking Resistance of Silicon Carbide Composites 307

International ournal of Applied Ceramic Technolog-Nozawa and Tanigawa Vol.7,No.3,2010 NITE-Thin-Coat NITE-Thick-Coat 0001002003004005006 Strain [%] Strain [%J tensile fracture behaviors of a) nanoinfltration transient eutectic(NITEr and(b) polymer-impregnation Results ductility. The big difference was in the high fracture strain (0.9%)compared with those of NITE-SiC/SiC com- Tensile Properties posites(<0.1%). Additionally, the proportional limit Figure 3 shows typical tensile stress versus strain stress(PLS), that is, an equivalent stress when matrix cracking initiates, was low ( 30 MPa)due to the brittle curves of SiC/SiC composites by the standard tensile PIp-SiC matrix For both types of NITE-SiC/SiC com- rest and reduced data are listed in Table l. the nite- Thin-Coat composite exhibited brittle fracture. The scan- posites, the PLS was quite high(<150 MPa)due to im- brittle surface probably due to the strong bonding at the by the NITe prolif Ensity of the SiC matrix produced proved stiffness and der ning electron microscope image(Fig 4a)clearly shows the F/M interface. No fiber pullout was observed in the frac- SENB Test Results ture surface. In contrast, NITE-Thick-Coat showed quasi- ductility coupled with fiber pullouts due to the cumulative Figure 5 shows a typical load versus COD curve debonding until fracture(Fig. 4b). However, the total during SENB tests for NITE-Thick-Coat and Fig.6ex- elongation was not so significant even in this case. In contrast, the PIP-Coat composite showed better quasi- ing stage of the SENB-3 specimen. From Fig NITE-Thin ick-Coa 50 Fig 4. Typical fracture surface images of nanoinfilration transient-eutectic(NITESiC/SiC composites

Results Tensile Properties Figure 3 shows typical tensile stress versus strain curves of SiC/SiC composites by the standard tensile test and reduced data are listed in Table I. The NITE￾Thin-Coat composite exhibited brittle fracture. The scan￾ning electron microscope image (Fig. 4a) clearly shows the brittle surface probably due to the strong bonding at the F/M interface. No fiber pullout was observed in the frac￾ture surface. In contrast, NITE-Thick-Coat showed quasi￾ductility coupled with fiber pullouts due to the cumulative debonding until fracture (Fig. 4b). However, the total elongation was not so significant even in this case. In contrast, the PIP-Coat composite showed better quasi￾ductility. The big difference was in the high fracture strain (B0.9%) compared with those of NITE–SiC/SiC com￾posites (o0.1%). Additionally, the proportional limit stress (PLS), that is, an equivalent stress when matrix￾cracking initiates, was low (B30 MPa) due to the brittle PIP–SiC matrix. For both types of NITE–SiC/SiC com￾posites, the PLS was quite high (B150 MPa) due to im￾proved stiffness and density of the SiC matrix produced by the NITE process. SENB Test Results Figure 5 shows a typical load versus COD curve during SENB tests for NITE-Thick-Coat and Fig. 6 ex￾hibits typical specimen surface micrographs in each load￾ing stage of the SENB-3 specimen. From Fig. 5, Fig. 4. Typical fracture surface images of nanoinfiltration transient-eutectic (NITE)–SiC/SiC composites. Fig. 3. Typical tensile fracture behaviors of (a) nanoinfiltration transient-eutectic (NITE)– and (b) polymer-impregnation and pyrolysis (PIP)–SiC/SiC composites. 308 International Journal of Applied Ceramic Technology—Nozawa and Tanigawa Vol. 7, No. 3, 2010

wwceramics. org/ACT Cracking Resistance of Silicon Carbide Composites served up to the maximum applied load(ig. 6b). How- NITE- Thick-Coat apparent energy consumpton in 031wmmW12) displacement curve in Fig. 5 implies probable accumula- tion of microcracks in the matrix. At the maximum load SENB-1b(W=4mm, t=lmm, a/=0.5) rapid crack propagation in a direction perpendicular to the longitudinal fiber direction occurred coupled with load drop(Fig. 6c). By applying, further loading many 10.2030405060.70 macrocracks, which have branched parallel to the longi Crack Opening Displacement (mm tudinal fiber direction, were observed with fber sliding at the F/M interface and fiber breaks(Fig. 6d and e). Figure 6f shows the surface image of the specimen after the test, SENB-2a (w=4mm, t=2mm, a/w=0. 25) indicating no breakage into pieces due to the huge con- SENB-2b(=4mm, t=2mm, ao /w=0.5) tribution of the strong frictional shear at the F/m inter face. In summary, three apparent damage accumulation stages in fracture behavior in the SENB test were identi fied as follows(1)initial elastic segment followed by a nonlinear stage due to microcrack formation,(2)macro- 0.10.20.3040.50.60.708 crack extension with considerable microcrack formation Crack Opening Displacement (mm and (3)load transferring by friction at the F/M interface, progressive crack brand Figure 7 plots the crack extension (Aa) measured using the replica film method for NITE-Thick-Coat. Of SENB-3a(w84mm, tamm, a/w-D. 25) particular emphasis is that unique trend was obtained by sSENB-3b(W-4mm, tamm, a /w-0. 5) normalizing specimen width(W), clearly indicating no specimen geometry effect on the crack extension behavior From Fig. 7, no crack extension was obtained in the first stage. In the second stage, it is apparent that the macro- 00.10.2030.40.5060.708 Crack Opening Displacement (mm] crack length rapidly increased with increasing displace ment. Of particular emphasis is that the crack propagation rate was almost a constant(A(a/W)/Ax= 3. 13). Con SENB-4 (W=Bmm, t=4mm, a /w=0.5) pared with the rapid crack extension in the second stage,a mild increase of the crack length was obtained in the third stage,that is, in the mixed failure accumulation process. From Fig. 5, two characteristic parameters namely: PLS as an initiation load of microcracking and ultimate flexural strength(UFS)as an initiation load of macro- cra obtained. figure 8 shows the Crack Opening Displacement [mm] rameters normalized as a Aexural stress form as: Fig. 5. Load us crack opening displacement curves in single-edge 3PL notched bend (sENB) tests for NITE-Thick-Coat where the applied load was(P), support span(L), specimen artaine linearity was obtained prior the maximum load was width(W), and specimen thickness (o). Although there ined. From Fig. 6, it is clearly identified that there were limited data sets, an important finding is that both were no a cracks initiated at the initial linear normalized PLS and fexural strength were nearly propor- ment and microcracks were initiated in the hber bundles tional to(1-do/W), indicating probable notch insensitivity parallel to the longitudinal fiber direction at the turning Regarding the notch insensitivity of NITE-SiC/SiC point from linear to nonlinear fracture behavior(Fig 6a). posites, the slope of these trends can produce a unique Beyond this load level, no further visible cracks were ob- proportional limit of flexural stress of 244 MPa,and

nonlinearity was obtained prior the maximum load was attained. From Fig. 6, it is clearly identified that there were no apparent cracks initiated at the initial linear seg￾ment and microcracks were initiated in the fiber bundles parallel to the longitudinal fiber direction at the turning point from linear to nonlinear fracture behavior (Fig. 6a). Beyond this load level, no further visible cracks were ob￾served up to the maximum applied load (Fig. 6b). How￾ever, apparent energy consumption in the load– displacement curve in Fig. 5 implies probable accumula￾tion of microcracks in the matrix. At the maximum load, rapid crack propagation in a direction perpendicular to the longitudinal fiber direction occurred coupled with load drop (Fig. 6c). By applying, further loading many macrocracks, which have branched parallel to the longi￾tudinal fiber direction, were observed with fiber sliding at the F/M interface and fiber breaks (Fig. 6d and e). Figure 6f shows the surface image of the specimen after the test, indicating no breakage into pieces due to the huge con￾tribution of the strong frictional shear at the F/M inter￾face. In summary, three apparent damage accumulation stages in fracture behavior in the SENB test were identi- fied as follows (1) initial elastic segment followed by a nonlinear stage due to microcrack formation, (2) macro￾crack extension with considerable microcrack formation, and (3) load transferring by friction at the F/M interface, coupled with progressive crack branching and fiber breaks. Figure 7 plots the crack extension (Da) measured using the replica film method for NITE-Thick-Coat. Of particular emphasis is that unique trend was obtained by normalizing specimen width (W), clearly indicating no specimen geometry effect on the crack extension behavior. From Fig. 7, no crack extension was obtained in the first stage. In the second stage, it is apparent that the macro￾crack length rapidly increased with increasing displace￾ment. Of particular emphasis is that the crack propagation rate was almost a constant (D(a/W)/Dx 5B3.13). Com￾pared with the rapid crack extension in the second stage, a mild increase of the crack length was obtained in the third stage, that is, in the mixed failure accumulation process. From Fig. 5, two characteristic parameters namely: PLS as an initiation load of microcracking and ultimate flexural strength (UFS) as an initiation load of macro￾cracking were obtained. Figure 8 shows these two pa￾rameters normalized as a flexural stress form as: s0 ¼ 3PL 2tW 2 ð1Þ where the applied load was (P), support span (L), specimen width (W), and specimen thickness (t). Although there were limited data sets, an important finding is that both normalized PLS and flexural strength were nearly propor￾tional to (1-a0/W) 2 , indicating probable notch insensitivity. Regarding the notch insensitivity of NITE–SiC/SiC com￾posites, the slope of these trends can produce a unique proportional limit of flexural stress of B244MPa, and an Fig. 5. Load vs. crack opening displacement curves in single-edge notched bend (SENB) tests for NITE-Thick-Coat. www.ceramics.org/ACT Cracking Resistance of Silicon Carbide Composites 309

310 International ournal of Applied Ceramic Technolog-Nozawa and Tanigawa Vol 7, No 3, 2010 1mm 1mm 1mm 1mm 1mm 500um Fig. 6. Specimen surface images of the single-edge notched bend (SENB) specimen during the tests. UFS of 615 MPa, respectively, regardless of the presence load drop occurred with complete composite breaks. It of initial notch machining. Specifically, it should be em- is apparent that the damage was induced before macro- asized that there seemed no significant size dependency. cracking for NITE-Thick-Coat and PIP-Coat. The NITE-Thin-Coat material contrarily failed in a brittle DNT Test Results manner. Unstable fracture of the DNT specimens was similar to that of standard tensile test results Similar to Figure 9 shows DNT test results for various the SEnB test results, the microcracking load(ppls) SiC composites. At the maximum load applied,a king load (Pmax) normalized by th

UFS of B615MPa, respectively, regardless of the presence of initial notch machining. Specifically, it should be em￾phasized that there seemed no significant size dependency. DNT Test Results Figure 9 shows DNT test results for various SiC/ SiC composites. At the maximum load applied, a rapid load drop occurred with complete composite breaks. It is apparent that the damage was induced before macro￾cracking for NITE-Thick-Coat and PIP-Coat. The NITE-Thin-Coat material contrarily failed in a brittle manner. Unstable fracture of the DNT specimens was similar to that of standard tensile test results. Similar to the SENB test results, the microcracking load (PPLS) and the macrocracking load (Pmax) normalized by the Fig. 6. Specimen surface images of the single-edge notched bend (SENB) specimen during the tests. 310 International Journal of Applied Ceramic Technology—Nozawa and Tanigawa Vol. 7, No. 3, 2010

wwceramics. org/ACT Cracking Resistance of Silicon Carbide Composites 311 sensitivity. In short, the averaged stress applied on the [ No ligament area then determines the composite strength This result was quite consistent with that of SENB test results. In contrast, the limited number of dara indicate the apparent notch sensitivity of NITE-Thin-Coat, whereby the strength of NITE-Thin-Coat exhibits non linear parabolic relation to 1-do/W. This would be rea- interfile Biven that the strong bonding at the F/M eventually produces brittleness of a material like ceram- w/ crack branching ics. In general, brittle ceramics are subject to surface Haws. she 田SENB-2 sion,the unique PLS of 133 MPa and the ultimate tensile strength(UTS)of 143 MPa were determined for NITE-Thick-Coat, regardless of the presence of notches. The same is true for the PIP-Coat, yielding 0.4 0.5 0.6 the unique PLS of 59 MPa and UTS of 234 MPa. Fig. 7. Crack extension to width ratio versus crack opening Discussion displacement for NITE- Thick-Coat Fracture Toughness by Linear Fracture Mechanics initial ligament area as stress forms were plotted as a Approach function of the initial notch depth to specimen width One classical approach to discuss fracture energy is normalized microcracking load and macrocracking load the fracture toughness, defined by the linear elastic frac were proportional to 1-do/W for the cases of NITE- ture mechanics (LEFM) and integral that has been hick-Coat and PIP-Coat, clearly indicating notch in- widely applied. According to the literature, the integral can be defined as a simplified form 20 80 J 口Bend.PLs where u is total work obtained from the load versus SENB-1, PLS load line displacement curve. Generally, the use of SENB-2 PLS ou=615±85 LEFM in complex composite materials particularly in NB-3 UFS the in-plane direction is inappropriate. However, as- V SENB-4 suming that the present SiC/SiC composites follow this 乏Na V SENB-4, UFS equation consistently, integral for NITE-Thick-Coat was expressed as a function of Aa. Specifically, by plot ting as a function of Aa/W, a unique trend was obtained (Fig. 11). The value of/ integral for NITE-Thick-Coat gradually increased at pus=244±45MPa first with increasing Aa/W and the inclement of the/ in- tegral was accelerated when Aa/W>0. 2(in the third stag in Fig. 7). One possible explanation is that the integral considers entire energy consumption during the test in- 0.6 cluding microcrack formation, friction at the F/M inter- face thermal strain relief, fiber breaking, et As the late

initial ligament area as stress forms were plotted as a function of the initial notch depth to specimen width ratio, 1-a0/W (Fig. 10). It is worth noting that both normalized microcracking load and macrocracking load were proportional to 1a0/W for the cases of NITE￾Thick-Coat and PIP-Coat, clearly indicating notch in￾sensitivity. In short, the averaged stress applied on the ligament area then determines the composite strength. This result was quite consistent with that of SENB test results. In contrast, the limited number of data indicate the apparent notch sensitivity of NITE-Thin-Coat, whereby the strength of NITE-Thin-Coat exhibits non￾linear parabolic relation to 1a0/W. This would be rea￾sonable given that the strong bonding at the F/M interface prohibits crack branching at the interface, eventually produces brittleness of a material like ceram￾ics. In general, brittle ceramics are subject to surface flaws, showing significant notch sensitivity. In conclu￾sion, the unique PLS of B133 MPa and the ultimate tensile strength (UTS) of B143 MPa were determined for NITE-Thick-Coat, regardless of the presence of notches. The same is true for the PIP-Coat, yielding the unique PLS of B59 MPa and UTS of B234 MPa. Discussion Fracture Toughness by Linear Fracture Mechanics Approach One classical approach to discuss fracture energy is the fracture toughness, defined by the linear elastic frac￾ture mechanics (LEFM) and J integral that has been widely applied. According to the literature, the J integral can be defined as a simplified form10: J ¼ 2U t Wð Þ  a ð2Þ where U is total work obtained from the load versus load line displacement curve. Generally, the use of LEFM in complex composite materials particularly in the in-plane direction is inappropriate. However, as￾suming that the present SiC/SiC composites follow this equation consistently, J integral for NITE-Thick-Coat was expressed as a function of Da. Specifically, by plot￾ting as a function of Da/W, a unique trend was obtained regardless of the specimen size (Fig. 11). The value of J integral for NITE-Thick-Coat gradually increased at first with increasing Da/W and the inclement of the J in￾tegral was accelerated when Da/W40.2 (in the third stage in Fig. 7). One possible explanation is that the J integral considers entire energy consumption during the test in￾cluding microcrack formation, friction at the F/M inter￾face, thermal strain relief, fiber breaking, etc. As the latest SiC/SiC composites possesses strong bonding and im￾proved friction resistance due to the rough fiber surface, Fig. 7. Crack extension to width ratio versus crack opening displacement for NITE-Thick-Coat. Fig. 8. Strengths as a function of the notch depth to width ratio in single-edge notched bend (SENB) tests. www.ceramics.org/ACT Cracking Resistance of Silicon Carbide Composites 311

312 International ournal of Applied Ceramic Technolog-Nozawa and Tanigawa Vol.7,No.3,2010 3.5 DNT-2a(W=10mm, t=1.7mm, ao/w=0. 20) DNT-2b(W=10mm, t=1.7mm, a0/w=0. 36) DNT-2c(W=10mm, t=1.7mm, a/=0.58) PIP- Coat 0.35 Crack opening displacement (mm r DNT-3a(W=4mm, t=1. 5mm, ao/w=0. 30) DNT-4(W=7.5mm, t=1.5mm, ao/W=0.51 DNT-3b(W=4mm, +1.5mm, a/=0.60)./ NITE-Thick-Coat NITE-Thin-Coat Fig 9. Load versus crack opening displacement curves in double-notch tensile(DNT) tests the contribution from the F/M interfacial interaction was huge enough to increase the total work. New surface cre- NTE- Thick-o on by crack branching might also influence the results PIP-Coat(W-1Omm) Failure Energy Analysis by Nonlinear fracture 个个 Mechanics Appr As aforementioned, test results apparently show quasi-ductility of composites, that is, energy consump tion during irreversible damage accumulation beyond matrix cracking. Therefore, considering the contribu- tions separately from irreversible energies such as inter- facial friction, thermal-residual strain energy, and fiber breaks become a key. Because of this quasi-ductility king an analytical model based on the nonlinear frac ture mechanics is reasonable and the analytical 1-a0/w method has been developed by the e authors. In this Fig. 10. Strengths as a function of the notch depth to width ratio model, the total work during tched specimen in double-notch tensile(DNT) tests

the contribution from the F/M interfacial interaction was huge enough to increase the total work. New surface cre￾ation by crack branching might also influence the results. Failure Energy Analysis by Nonlinear Fracture Mechanics Approach As aforementioned, test results apparently show quasi-ductility of composites, that is, energy consump￾tion during irreversible damage accumulation beyond matrix cracking. Therefore, considering the contribu￾tions separately from irreversible energies such as inter￾facial friction, thermal-residual strain energy, and fiber breaks become a key. Because of this quasi-ductility, taking an analytical model based on the nonlinear frac￾ture mechanics11,12 is reasonable and the analytical method has been developed by the authors.18 In this model, the total work during the notched specimen Fig. 9. Load versus crack opening displacement curves in double-notch tensile (DNT) tests. Fig. 10. Strengths as a function of the notch depth to width ratio in double-notch tensile (DNT) tests. 312 International Journal of Applied Ceramic Technology—Nozawa and Tanigawa Vol. 7, No. 3, 2010

wwceramics. org/ACT Cracking Resistance of Silicon Carbide Composites 313 where load-line displacement is x. Note that the crack surface formation energy includes micro-and macro- E: SENB-2 d SENB-3 Following the definition specified in Fig. 12, the ■SENB4 rack surface for rmation energy was plotted as a function of displacement(Fig. 13). From Fig. 13, it is apparent that the crack surface formation energy rapidly increased when damage accumulated. Specifically, the crack sur face formation energy seems proportional to the load- line displacement in the second stage. A constant crack surface formation energy rate is then obtained by linear SU fit. In contrast, the crack length change as a function of load-line displacement was obtained from Fig. 7. Sub- iting these data into the Eq. (4), a fracture Coat. Note that no significant size effect was obtained 0. 3 0.4 0.5 Of particular emphasis is that the result clearly indicates good coincidence with the critical value of the J integral Fig I1. )integral versus crack extension when Aa/W=0 in Fig. 11, which was determined by test(w) is expressed as the work until the peak load. For both cases, these en- ergy parameters assigned to the crack surface formation w=Ue+ Ur+Ur+r icrocrack formation but no contribution from interfacial friction was considered where elastic energy(Ue), friction energy at the interface One drawback of this analysis is that the fracture re- (Uf), residual strain energy(Ur), and crack surface for- sistance( G) defined in this study cannot perfectly distin mation energy(T)are defined in Fig. 12. Then, the ImIcro- and fracture resistance(G can be defined as G、or1orax 0.15 tda t ax da DOABC: Crack formation energy (r) AOCP: Residual strain energy (U) O SENB APCQ: Friction energy (U) E SENB-2 △QcR: Elastic 1 SENB-3 ■SENB4 B: initiation of macro-cracking Micro- crack formation energy 品005 P R S 0.2 0.3 040.5 0.6 Fig 12. Definition: (a)elastic energy, (b) permanent strain Fig 13. Crack surface formation energy consumed by single-edge energy,(c) friction energy, and (d) crack surface formation energ. notched bend (SENB)tests

test (w) is expressed as w ¼ Ue þ Ufr þ Ur þ G ð3Þ where elastic energy (Ue), friction energy at the interface (Ufr), residual strain energy (Ur), and crack surface for￾mation energy (G) are defined in Fig. 12. Then, the fracture resistance (G) can be defined as G ¼ ›G t›a ¼ 1 t ›G ›x ›x ›a ð4Þ where load-line displacement is x. Note that the crack surface formation energy includes micro- and macro￾crack-forming energies together. Following the definition specified in Fig. 12, the crack surface formation energy was plotted as a function of displacement (Fig. 13). From Fig. 13, it is apparent that the crack surface formation energy rapidly increased when damage accumulated. Specifically, the crack sur￾face formation energy seems proportional to the load￾line displacement in the second stage. A constant crack surface formation energy rate is then obtained by linear fit. In contrast, the crack length change as a function of load-line displacement was obtained from Fig. 7. Sub￾stituting these data into the Eq. (4), a fracture resistance of B3.8 kJ/m2 was finally obtained for NITE-Thick￾Coat. Note that no significant size effect was obtained. Of particular emphasis is that the result clearly indicates good coincidence with the critical value of the J integral when Da/W 5 0 in Fig. 11, which was determined by the work until the peak load. For both cases, these en￾ergy parameters assigned to the crack surface formation including microcrack formation but no contribution from interfacial friction was considered. One drawback of this analysis is that the fracture re￾sistance (G) defined in this study cannot perfectly distin￾guish contributions from micro- and macrocrack Fig. 11. J integral versus crack extension length to width ratio. Fig. 12. Definition: (a) elastic energy, (b) permanent strain energy, (c) friction energy, and (d) crack surface formation energy. Fig. 13. Crack surface formation energy consumed by single-edge notched bend (SENB) tests. www.ceramics.org/ACT Cracking Resistance of Silicon Carbide Composites 313