Availableonlineatwww.sciencedirect.com SCIENCE E噩≈S Journal of the European Ceramic Society 24(2004)2339-234 www.elsevier.com/locate/jeurceramsoc Thermal shock resistance of fibrous monolithic Si3 N4/bn ceramics Young-Hag Koha, b,s*, Hae-Won Kima, Hyoun-Ee Kima, John W.Halloran a School of Materials Science and Engineering, Seoul National University, Seoul, 151-742, South Korea b Materials Science and Engineering Department, University of Michigan, Ann Arbor, MI48109-2136, USA Received 15 January 2003: received in revised form 25 June 2003: accepted 6 July 2003 Abstract Thermal shock resistance of fibrous monolithic Si3 N4/BN ceramic was investigated by measuring the strength retention after varying the temperature difference(AT)up to 1400C and was compared with that of monolithic Si3N4 Monolithic Si3 N4 showed catastrophic drop in flexural strength above AT of 1000C, while FM showed negligible reduction in flexural strength without critical temperature difference(AT). Two parameters, such as the resistance to crack initiation(R) and crack propagation(r). were used in order to explain the thermal shock behaviors of fibrous monolith and monolithic Si3N4. Furthermore crack interac- tions during flexural testing, such as delamination cracks and crack deflection, were characterized and were related to the work-of- racture (WoF) C) 2003 Elsevier Ltd. All rights reserved. Keywords: BN; Composites; Fibrous monoliths: Si3N4; Thermal shock 1. Introduction there are some possible methods to increase the thermal shock resistance of materials. For example, the addition Fibrous monoliths have been regarded as promising of ductile secondary phase into Al2O3 matrix increases materials for structural applications because of the the thermal shock resistance due to both reduced elastic noncatastrophic failure due to its unique modulus and increased fracture toughness. Also, flaw- architecture. 1-7 Fibrous monoliths are sintered or hot- tolerant material, such as fiber(or whisker)-reinforced pressed monolithic ceramics with a distinct fibrous tex- ceramics and laminated ceramics shows excellent ther ture consisting of strong cell and weak cell boundary mal shock resistance due to the increased resistance to that act as a easy crack path. One of the most promis- crack propagation through crack interactions with ing fibrous monoliths for high temperature applications toughening agents (fiber, whisker and weak inter is Si3N4/BN system because of its high strength and face) 6- However, so far, in spite of its importance for oxidation resistance at elevated temperature. -7 high temperature applications, no research has beer Since these composite materials are candidates as the done on thermal shock resistance of fibrous monolith high-temperature applications (e.g. in gas turbine In this paper, we have investigated thermal shock engines), it is inevitable to involve some kind of thermal resistance of fibrous monolithic Si3 N,/BN ceramics with shock loading. Most ceramics showed catastrophe temperature difference ranging from 800 to 1400C, by drops in mechanical properties, such as flexural measuring the retention of mechanical properties, such strength, elastic modulus, after thermal shock above the as flexural strength and work-of-fracture (WOF). For critical temperature(AT). -l5 This catastrophic drop the purpose of comparison, monolithic Si3 N4 was also in mechanical properties after thermal shock have lim- tested under the same conditions. ited the wide applications at high-temperatures Thermal shock resistance is dependent on several pri mary mechanical properties, such as fracture toughness, 2. Experimental fracture behavior, fracture strength, elastic modulus and coefficient of thermal expansion of material. 11, 2 Hence, 2. 1. Billet fabrication Fibrous monolithic Si3N4/BN ceramic was fabricated E-mail address: younghag(@engin. umich.edu (Y -H. Koh) using coextrusion process to produce a structure with 0955-2219S. see front matter C 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0955-2219(03)00644-7
Thermal shock resistance of fibrous monolithic Si3N4/BN ceramics Young-Hag Koha,b,*, Hae-Won Kima , Hyoun-Ee Kima , John W. Halloranb a School of Materials Science and Engineering, Seoul National University, Seoul, 151-742, South Korea bMaterials Science and Engineering Department, University of Michigan, Ann Arbor, MI 48109-2136, USA Received 15 January 2003; received in revised form 25 June 2003; accepted 6 July 2003 Abstract Thermal shock resistance of fibrous monolithic Si3N4/BN ceramic was investigated by measuring the strength retention after varying the temperature difference (T) up to 1400 C and was compared with that of monolithic Si3N4. Monolithic Si3N4 showed catastrophic drop in flexural strength above T of 1000 C, while FM showed negligible reduction in flexural strength without critical temperature difference (Tc). Two parameters, such as the resistance to crack initiation (R0 ) and crack propagation (R0000), were used in order to explain the thermal shock behaviors of fibrous monolith and monolithic Si3N4. Furthermore, crack interactions during flexural testing, such as delamination cracks and crack deflection, were characterized and were related to the work-offracture (WOF). # 2003 Elsevier Ltd. All rights reserved. Keywords: BN; Composites; Fibrous monoliths; Si3N4; Thermal shock 1. Introduction Fibrous monoliths have been regarded as promising materials for structural applications because of the noncatastrophic failure due to its unique architecture.17 Fibrous monoliths are sintered or hotpressed monolithic ceramics with a distinct fibrous texture consisting of strong cell and weak cell boundary that act as a easy crack path.1 One of the most promising fibrous monoliths for high temperature applications is Si3N4/BN system because of its high strength and oxidation resistance at elevated temperature.47 Since these composite materials are candidates as the high-temperature applications (e.g. in gas turbine engines), it is inevitable to involve some kind of thermal shock loading. Most ceramics showed catastrophic drops in mechanical properties, such as flexural strength, elastic modulus, after thermal shock above the critical temperature (Tc).1115 This catastrophic drop in mechanical properties after thermal shock have limited the wide applications at high-temperatures. Thermal shock resistance is dependent on several primary mechanical properties, such as fracture toughness, fracture behavior, fracture strength, elastic modulus and coefficient of thermal expansion of material.11,12 Hence, there are some possible methods to increase the thermal shock resistance of materials. For example, the addition of ductile secondary phase into Al2O3 matrix increases the thermal shock resistance due to both reduced elastic modulus and increased fracture toughness.14 Also, flawtolerant material, such as fiber (or whisker)-reinforced ceramics and laminated ceramics shows excellent thermal shock resistance due to the increased resistance to crack propagation through crack interactions with toughening agents (fiber, whisker and weak interface).1618 However, so far, in spite of its importance for high temperature applications, no research has been done on thermal shock resistance of fibrous monolith. In this paper, we have investigated thermal shock resistance of fibrous monolithic Si3N4/BN ceramics with temperature difference ranging from 800 to 1400 C, by measuring the retention of mechanical properties, such as flexural strength and work-of-fracture (WOF). For the purpose of comparison, monolithic Si3N4 was also tested under the same conditions. 2. Experimental 2.1. Billet fabrication Fibrous monolithic Si3N4/BN ceramic was fabricated using coextrusion process to produce a structure with 0955-2219/$ - see front matter # 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0955-2219(03)00644-7 Journal of the European Ceramic Society 24 (2004) 2339–2347 www.elsevier.com/locate/jeurceramsoc * Corresponding author. E-mail address: younghag@engin.umich.edu (Y.-H. Koh).
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 uniaxially aligned 250 micron cells of composition commercially available tester Grindo-sonic model Si3 N4(E-10, Ube Industries, Tokyo, Japan) with 6 MK4x, J W. Lemmon, St, Louis, MO, USA). 20 wt%Y2O3(99.9%, Johnson Matthey Electronics, MA USA)and 2 wt. Al,O3(HP-DBM, Reynolds, Bauxite, AK, USA), separated by 15-25 micron boron nitride 3. Results (HCP, Advanced Ceramics Corp Cleveland, OH, USA cell boundaries. Further details on the fabrication of 3. 1. Microstructure and mechanical properties before fibrous monoliths are described elsewhere. 4 For com thermal shock parison, monolithic Si3 N4 with 6 wt %Y2O3 and 2 wt% Al2O3 as sintering aids was also fabricated. The The typical microstructure of fibrous monolithic green billets were hot-pressed at 1740C under an Si3 N4/BN ceramic(FM)is shown in Fig. 1. Low mag applied pressure of 25 MPa for 2 h in a flowing N2 nification SEM micrographs of polished sections, shows atmosphere. The density of the specimens was measured three-dimensional representations of the sub-millimeter using the Archimedes method and the theoretical density structure of fibrous monoliths. The polycrystalline sili- of the specimens was estimated by the rule of mixture con nitride cells appear in dark contrast, while the con- tinuous boron nitride cell boundaries appear in bright 2.2. Specimen preparation contrast. The Si3 N4 cells are surrounded by the cell boundaries consisting of bn particles bonded with The thermal shock resistance was determined by yttriumaluminosilicate. measuring the retention of the flexural strength of The mechanical properties of monolithic Si3 N4 and water-quenched specimen. Specimens were machined FM samples are summarized in Table l. For FM, the nto a bar shape with dimensions of 3x4x45 mm and measured density (p) was slightly higher than theoretical ground with a 600-grit diamond wheel. The tensile side value(based on 82.5 vol. Si3 N4 cells and 17.5 vol% of the specimens was polished using diamond paste BN cell boundaries for fibrous monoliths), implying full down to 3 um, and subsequently chamfered to minimize densification of both Si3 N4 cell and BN cell boundary machining flaws. Also, the side surfaces of each speci- materials occurred. Elastic modulus (E)and flexural men were polished down to 30 um strength(MOR) of FM were slightly lower than those of monolithic Si3 N4, while apparent WoF increased 23. Thermal shock test remarkably due to the noncatastrophic failure through extensive crack interactions along the weak Bn cell Thermal shock test was carried out in a vertical tube boundaries furnace at temperatures between 800C and 1400C in The typical flexural responses of monolithic Si3 N4 and laboratory air. The furnace was heated at a heating rate FM are shown in Fig. 2. As expected, monolithic Sign of 10 C/min and maintained at exposure temperatures. showed higher strength but negligible apparent wo Polished specimens, suspended at the end of a platinum wire, were inserted into the hot-zone from the top and were soaked for 30 min to induce the homogeneous temperature distribution. After exposure, the specimens were quickly dropped into the water bath with a capa city of 5000 cc. The temperature of water bath did not increase notably after dropping the specimen 2.4. Mechanical test and characterization The flexural strength after thermal shock test was measured at room temperature by a four-point flexural configuration at a cross-head speed of 0.5 mm/min, and inner- and outer-spans of 20 and 40 mm, respectively The load versus crosshead deflection response and the work of fracture, calculated by determining the area under the load-crosshead deflection curve and dividing 250μm it by twice the cross-sectional area of the sample, are reported. Also, crack propagation during flexural Fig. I. Low magnification SEM micrographs of polished strength test after thermal shock was observed by an shows three-dimensional representations of the submillimeter of fibrous monoliths. The polycrystalline silicon nitride cells optical microscope and an SEM microscope. Elastic dark contrast and the continuous boron nitride cell boundaries are in moduli were measured by the impulse technique using a bright contrast. Courtesy of Bruce King)
uniaxially aligned 250 micron cells of composition Si3N4 (E-10, Ube Industries, Tokyo, Japan) with 6 wt.% Y2O3 (99.9%, Johnson Matthey Electronics, MA, USA) and 2 wt.% Al2O3 (HP-DBM, Reynolds, Bauxite, AK, USA), separated by 1525 micron boron nitride (HCP, Advanced Ceramics Corp., Cleveland, OH, USA) cell boundaries. Further details on the fabrication of fibrous monoliths are described elsewhere.1,4 For comparison, monolithic Si3N4 with 6 wt.% Y2O3 and 2 wt.% Al2O3 as sintering aids was also fabricated. The green billets were hot-pressed at 1740 C under an applied pressure of 25 MPa for 2 h in a flowing N2 atmosphere. The density of the specimens was measured using the Archimedes method and the theoretical density of the specimens was estimated by the rule of mixture. 2.2. Specimen preparation The thermal shock resistance was determined by measuring the retention of the flexural strength of water-quenched specimen. Specimens were machined into a bar shape with dimensions of 3445 mm and ground with a 600-grit diamond wheel. The tensile side of the specimens was polished using diamond paste down to 3 mm, and subsequently chamfered to minimize machining flaws. Also, the side surfaces of each specimen were polished down to 30 mm. 2.3. Thermal shock test Thermal shock test was carried out in a vertical tube furnace at temperatures between 800 C and 1400 C in laboratory air. The furnace was heated at a heating rate of 10 C/min and maintained at exposure temperatures. Polished specimens, suspended at the end of a platinum wire, were inserted into the hot-zone from the top and were soaked for 30 min to induce the homogeneous temperature distribution. After exposure, the specimens were quickly dropped into the water bath with a capacity of 5000 cc. The temperature of water bath did not increase notably after dropping the specimen. 2.4. Mechanical test and characterization The flexural strength after thermal shock test was measured at room temperature by a four-point flexural configuration at a cross-head speed of 0.5 mm/min, and inner- and outer-spans of 20 and 40 mm, respectively. The load versus crosshead deflection response and the work of fracture, calculated by determining the area under the load–crosshead deflection curve and dividing it by twice the cross-sectional area of the sample, are reported. Also, crack propagation during flexural strength test after thermal shock was observed by an optical microscope and an SEM microscope. Elastic moduli were measured by the impulse technique using a commercially available tester (Grindo-sonic model MK4x, J. W. Lemmon, St, Louis, MO, USA).20 3. Results 3.1. Microstructure and mechanical properties before thermal shock The typical microstructure of fibrous monolithic Si3N4/BN ceramic (FM) is shown in Fig. 1. Low magnification SEM micrographs of polished sections, shows three-dimensional representations of the sub-millimeter structure of fibrous monoliths. The polycrystalline silicon nitride cells appear in dark contrast, while the continuous boron nitride cell boundaries appear in bright contrast. The Si3N4 cells are surrounded by the cell boundaries consisting of BN particles bonded with yttriumaluminosilicate. The mechanical properties of monolithic Si3N4 and FM samples are summarized in Table 1. For FM, the measured density () was slightly higher than theoretical value (based on 82.5 vol.% Si3N4 cells and 17.5 vol.% BN cell boundaries for fibrous monoliths), implying full densification of both Si3N4 cell and BN cell boundary materials occurred. Elastic modulus (E) and flexural strength (MOR) of FM were slightly lower than those of monolithic Si3N4, while apparent WOF increased remarkably due to the noncatastrophic failure through extensive crack interactions along the weak BN cell boundaries. The typical flexural responses of monolithic Si3N4 and FM are shown in Fig. 2. As expected, monolithic Si3N4 showed higher strength but negligible apparent WOF Fig. 1. Low magnification SEM micrographs of polished sections, shows three-dimensional representations of the submillimeter structure of fibrous monoliths. The polycrystalline silicon nitride cells appear in dark contrast and the continuous boron nitride cell boundaries are in bright contrast. (Courtesy of Bruce King). 2340 Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 2341 because of catastrophic failure [Fig. 2(A)]. On the other 3. 2. Strength retention hand, FM exhibited noncatastrophic failure due to its The thermal shock resistance was observed by mea- unique architecture, comprised of strong Si3 N4 cell and suring the retention of the flexural strength after ther weak BN cell boundary, resulting in high apparent mal shock test, as shown in Fig. 4. For monolithic WOF [Fig. 2(B). Moreover, the apparent strength Si3 N4, the traditional thermal shock behavior of brittle retention after the first failure was above 50% of origi material was observed, that is, the flexural strength nal strength, showing the noncatastrophic. This non- decreased rapidly after thermal shock with temperature catastrophic nature was attributed to the extensive difference of 1000C [Fig. 4(A)]. However, the flexural crack interactions, such as crack delaminations and strength of fM after thermal shock test was not chan crack deflections, as shown in Fig. 3. For fibrous ged much [Fig. 4(B)l, showing the excellent thermal monoliths, the crack propagates through the weak cell shock resistance. Moreover, there was no critical tem- boundaries to reduce the applied stress. Similar crack perature (AT), at which the strength decreases propagations have been observed in many different catastrophically, up to 1400C. kinds of fibrous monoliths 1-5 3. 22. fracture behavior 3.2. Mechanical properties after thermal shock lithic Si N4 and FM were not basically changed. After thermal shock. the fracture behaviors of mor When a material(monolithic Si3 N4 or FM)is sub jected to a rapid decrease in temperature(AT), the sur- face of the component is placed under tension and the interior under compression. If the tensile stress devel oped on the surface exceeds the strength of the material the cracks are generated, leading to a rapid drop in flexural strength 11-15 Fig. 3. Optical photograph of crack propagation of the fibrous monolithic Si]N4/ BN ceramic after flexural testing. Extensive crack (B)Fibrous Monolith interactions. such as crack delamination and crack deflection. were Crosshead Displacement I mm I ≈600. Fig. 2. Flexural response of (A) monolithic Si3 N4 and (B)fibrous (B)Fibrous Monolith monolithic Si3N4/BN ceramic before thermal shock test. Monolithic Si3N4 showed brittle fracture, while fibrous monolith showed graceful racture due to unique architecture. Note, retained apparent stress after first drop is above 50%(B) Table 0600800100012001400 Summarized mechanical properties of monolithic Si3N4 and fibrous monolithic Si3N4/BN ceramic Temperature Difference[C I p(g/cc) E(GPa) MOR (MPa) WoF (kJ/m) Fig. 4. Flexural strength of (A)monolithic Si3 N4 and (B)fibrous 3N4/bn ceramic after thermal shock with Monolithic si3N43.27±0.1318±4832±4 differene Flexural strength of monolithic Si3N Fibrous monolith3.09±0.1276±3416±34 5.94±1.34 strophically after thermal shock with AT=1000C; howe monolith showed negligible decrease in flexural strength
because of catastrophic failure [Fig. 2(A)]. On the other hand, FM exhibited noncatastrophic failure due to its unique architecture, comprised of strong Si3N4 cell and weak BN cell boundary, resulting in high apparent WOF [Fig. 2(B)]. Moreover, the apparent strength retention after the first failure was above 50% of original strength, showing the noncatastrophic. This noncatastrophic nature was attributed to the extensive crack interactions, such as crack delaminations and crack deflections, as shown in Fig. 3. For fibrous monoliths, the crack propagates through the weak cell boundaries to reduce the applied stress. Similar crack propagations have been observed in many different kinds of fibrous monoliths.15 3.2. Mechanical properties after thermal shock When a material (monolithic Si3N4 or FM) is subjected to a rapid decrease in temperature (T), the surface of the component is placed under tension and the interior under compression. If the tensile stress developed on the surface exceeds the strength of the material, the cracks are generated, leading to a rapid drop in flexural strength.1115 3.2.1. Strength retention The thermal shock resistance was observed by measuring the retention of the flexural strength after thermal shock test, as shown in Fig. 4. For monolithic Si3N4, the traditional thermal shock behavior of brittle material was observed, that is, the flexural strength decreased rapidly after thermal shock with temperature difference of 1000 C [Fig. 4(A)]. However, the flexural strength of FM after thermal shock test was not changed much [Fig. 4(B)], showing the excellent thermal shock resistance. Moreover, there was no critical temperature (Tc), at which the strength decreases catastrophically, up to 1400 C. 3.2.2. Fracture behavior After thermal shock, the fracture behaviors of monolithic Si3N4 and FM were not basically changed, as Table 1 Summarized mechanical properties of monolithic Si3N4 and fibrous monolithic Si3N4/BN ceramic Samples (g/cc) E (GPa) MOR (MPa) WOF (kJ/m2 ) Monolithic Si3N4 3.270.1 3184 83246 Negligible Fibrous monolith 3.090.1 2763 41634 5.941.34 Fig. 2. Flexural response of (A) monolithic Si3N4 and (B) fibrous monolithic Si3N4/BN ceramic before thermal shock test. Monolithic Si3N4 showed brittle fracture, while fibrous monolith showed graceful fracture due to unique architecture. Note, retained apparent stress after first drop is above 50% (B). Fig. 3. Optical photograph of crack propagation of the fibrous monolithic Si3N4/BN ceramic after flexural testing. Extensive crack interactions, such as crack delamination and crack deflection, were observed. Fig. 4. Flexural strength of (A) monolithic Si3N4 and (B) fibrous monolithic Si3N4/BN ceramic after thermal shock with temperature difference (T). Flexural strength of monolithic Si3N4 reduced catastrophically after thermal shock with T=1000 C; however, fibrous monolith showed negligible decrease in flexural strength. Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347 2341
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 shown in Fig. 5, that is, monolithic Si3N4 showed cata- ticles in the as-hot pressed material are already micro- strophic failure (not shown), while FM showed cracked. Hence, thermal shock damage seems to be noncatastrophic failure regardless of temperature d absorbed within the bn cell boundaries which would ference. Furthermore, with the increase in temperature decrease the cell boundary fracture resistance, enabling difference, more extensive crack interactions were easier crack deflection and higher WOF. The specimens observed. The increase in apparent stress after fist drop shocked with the highest temperature difference implies the midplane shear stress after thermal shock (AT=1400C)had the most extensive crack delamina- The apparent WOF of the FM specimens increased tions, as shown in Fig. 7(D). This remarkable increase remarkably after thermal shock test, as shown in Fig. 6. in crack delamination is attributed to not only pre- The WOF is higher if there is a large retained load once ferential crack propagation caused by thermal stress but fracture begins, and strongly depends upon the extent of also oxidized damage layer during exposure to air crack interactions and delamination. The thermally The change of surface morphology after thermal shocked specimens exhibited higher retained strength shock test is shown in Fig. 8. Up to the temperature and extensive crack delamination. Thermal shock difference of 1200oC, the surface was not damaged damage seems to be absorbed within the bn cell (not shown). However, with temperature difference of boundaries, which would decrease the cell boundary 1400C, the surface(both BN cell boundary and fracture resistance, enabling easier crack deflection and Si3N4 cell) was damaged to some extent due to the higher WOF oxidation 3.2.3. Crack propagation 3.2.4. Load-bearing capacit The increased crack interactions in the thermally The thermal stress developed on surface and interface shocked sample, manifested by crack path, are clearly of Si3 N4 and bn after the thermal shock affected the shown in Fig. 7A-D. After thermal shock, crack inter- flexural response of FM upon subsequent room tem- actions (crack delamination and crack deflection) perature testing, as shown in Fig. 9. The retained occurred more extensively compared to the specimen strength after the fist drop(Ist drop/lst peak) was not before the thermal shock(Fig. 3). Pronounced crack basically changed within the range between 40% and delamination occurred by the thermal shock of 800C 55%, meaning the excellent load-bearing capacity for [Fig. 7(A)], and further long crack delamination was actual applications. However, the normalized maximum observed after the thermal shock of 1200C [Fig. 7(C). strength (2nd peak/lst peak) increased after thermal The tendency for crack delamination in FM ceramics is shock test. This result means that the first peak was influenced by the interfacial crack resistance of the BN- caused by the crack initiation on the surface; thus, the ontaining cell boundary. 7. The increase in WOF after surface was slightly weakened due to the thermal stress thermal shock suggests that thermal shock reduces the Furthermore, the thermal stress developed in interface interfacial crack resistance of the cell boundary, which of Si, N4 and bn promoted extensive crack interactions is a composite of boron nitride and glass. The Bn par resulting in increased woF. A△T=800c (B)△T=1000c 0号 (c)△T=1200c D)△T=1400c 200400600800100012001400 Crosshead Displacement I mm] Temperature Difference[Cl lexural responses of fibrous monolithic Si3N4/BN Fig. 6. Work-of-fracture (woF) of fibrous monolithic Si3 N4/BN ermal shock with temperature difference(AT) of(A)80 800oC ceramic thermal shock with temperature difference (AT) oC,(C)1200C, and(D)1400C. All samples exhibited Fibrous monolith exhibited significant woF due to extensive crack
shown in Fig. 5, that is, monolithic Si3N4 showed catastrophic failure (not shown), while FM showed noncatastrophic failure regardless of temperature difference. Furthermore, with the increase in temperature difference, more extensive crack interactions were observed. The increase in apparent stress after fist drop implies the midplane shear stress after thermal shock. The apparent WOF of the FM specimens increased remarkably after thermal shock test, as shown in Fig. 6. The WOF is higher if there is a large retained load once fracture begins, and strongly depends upon the extent of crack interactions and delamination. The thermally shocked specimens exhibited higher retained strength and extensive crack delamination. Thermal shock damage seems to be absorbed within the BN cell boundaries, which would decrease the cell boundary fracture resistance, enabling easier crack deflection and higher WOF. 3.2.3. Crack propagation The increased crack interactions in the thermally shocked sample, manifested by crack path, are clearly shown in Fig. 7A–D. After thermal shock, crack interactions (crack delamination and crack deflection) occurred more extensively compared to the specimen before the thermal shock (Fig. 3). Pronounced crack delamination occurred by the thermal shock of 800 C [Fig. 7(A)], and further long crack delamination was observed after the thermal shock of 1200 C [Fig. 7 (C)]. The tendency for crack delamination in FM ceramics is influenced by the interfacial crack resistance of the BNcontaining cell boundary.7,8 The increase in WOF after thermal shock suggests that thermal shock reduces the interfacial crack resistance of the cell boundary, which is a composite of boron nitride and glass. The BN particles in the as-hot pressed material are already microcracked.1 Hence, thermal shock damage seems to be absorbed within the BN cell boundaries, which would decrease the cell boundary fracture resistance, enabling easier crack deflection and higher WOF. The specimens shocked with the highest temperature difference (T=1400 C) had the most extensive crack delaminations, as shown in Fig. 7(D). This remarkable increase in crack delamination is attributed to not only preferential crack propagation caused by thermal stress but also oxidized damage layer during exposure to air. The change of surface morphology after thermal shock test is shown in Fig. 8. Up to the temperature difference of 1200 C, the surface was not damaged (not shown). However, with temperature difference of 1400 C, the surface (both BN cell boundary and Si3N4 cell) was damaged to some extent due to the oxidation. 3.2.4. Load-bearing capacity The thermal stress developed on surface and interface of Si3N4 and BN after the thermal shock affected the flexural response of FM upon subsequent room temperature testing, as shown in Fig. 9. The retained strength after the fist drop (1st drop/1st peak) was not basically changed within the range between 40% and 55%, meaning the excellent load-bearing capacity for actual applications. However, the normalized maximum strength (2nd peak/1st peak) increased after thermal shock test. This result means that the first peak was caused by the crack initiation on the surface; thus, the surface was slightly weakened due to the thermal stress. Furthermore, the thermal stress developed in interface of Si3N4 and BN promoted extensive crack interactions, resulting in increased WOF. Fig. 5. Flexural responses of fibrous monolithic Si3N4/BN ceramic after thermal shock with temperature difference (T) of (A) 800 C, (B) 1000 C, (C) 1200 C, and (D) 1400 C. All samples exhibited graceful fractures. Fig. 6. Work-of-fracture (WOF) of fibrous monolithic Si3N4/BN ceramic after thermal shock with temperature difference (T). Fibrous monolith exhibited significant WOF due to extensive crack interactions. 2342 Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 2343 3 mm 3 mm (c) (D) mm 3 mm Fig. 7. Optical photographs of crack propagation of fibrous monolithic Si3 N4/BN ceramic after thermal shock with temperature difference(An)of (A)800C,(B)1000C,(C)1200C, and (D)1400C during flexural testing. All samples showed extensive crack interactions, such delaminations and crack deflections 4. Discussion aligned in the transverse direction. Therefore mechan- ical properties, such as elastic modulus, coefficient of The fracture strength of thermally shocked monolithic thermal expansion(CTE) and Poissons ratio, of FM Si3N4 is strongly dependent on the magnitude of tensile sample show anisotropy, as described in Table 2 stress developed on the surface, that is, if the tensile The elastic modulus of monolithic silicon nitride was stress exceeds its strength, the cracks are generated on sig d the transverse modulus of the the surface, resulting in catastrophic drop in flexural FM was less than half the longitudinal modulus, due to trength. However, for FM sample, the fracture the Bn, which has a small c-axis modulus. The long- strength of FM sample is less sensitive to surface flaws: itudinal thermal expansion of the Fm was slightly less therefore, the resistance to crack propagation is a more than monolithic silicon nitride, decreased by the a-axis critical factor than the resistance to crack initiation, BN, while the transverse thermal expansion of the FM which is critical for brittle monolithic Si3 N4. However, was much larger, increased by the c-axis BN. Poisson's pre-existing cracks on BN cell boundaries after hot- ratios were estimated from rule-of-mixture by taking pressing(T=1740oC)also affects the flexural response, 0.27 and 0.2 for Si3N4 and BN, respectively resulting in crack interactions. Therefore, some factors, The magnitude of thermal stress induced by the same such as the magnitude of thermal stress on surface, exposure will be different, depending on the cell align thermal shock resistance parameter and pre-existing ment (longitudinal and transverse direction). The tradi cracks tional approach to evaluate the thermal shock resistance is based on quenching the specimen from an elevated 4.1. Magnitude of thermal stress on the surface(ors) temperature into a quenching media and measuring the fracture strength of the material. Neglecting the heat Considering the structure of this uniaxial FM (Fig. 1), transfer and size effects, the maximum tensile stress it is noted that the elastic modulus and thermal expan- (ors) generated on the surface of the specimen can be sion coeficient is different in the transverse and long according itudinal directions. In addition, the hexagonal BN is ors =(Ea/(1-v).AT strongly textured, with the high stiffness/low expan sion a-axis aligned preferentially in the longitudinal where E, a and v represent the elastic modulus, the direction and the lower stiffness/higher expansion c-axis coefficient of thermal expansion(CTE)and Poissons
4. Discussion The fracture strength of thermally shocked monolithic Si3N4 is strongly dependent on the magnitude of tensile stress developed on the surface, that is, if the tensile stress exceeds its strength, the cracks are generated on the surface, resulting in catastrophic drop in flexural strength. However, for FM sample, the fracture strength of FM sample is less sensitive to surface flaws; therefore, the resistance to crack propagation is a more critical factor than the resistance to crack initiation, which is critical for brittle monolithic Si3N4. However, pre-existing cracks on BN cell boundaries after hotpressing (T=1740 C) also affects the flexural response, resulting in crack interactions. Therefore, some factors, such as the magnitude of thermal stress on surface, thermal shock resistance parameter and pre-existing cracks, are discussed. 4.1. Magnitude of thermal stress on the surface (sTS) Considering the structure of this uniaxial FM (Fig. 1), it is noted that the elastic modulus and thermal expansion coefficient is different in the transverse and longitudinal directions. In addition, the hexagonal BN is strongly textured,19 with the high stiffness/low expansion a-axis aligned preferentially in the longitudinal direction and the lower stiffness/higher expansion c-axis aligned in the transverse direction. Therefore mechanical properties, such as elastic modulus, coefficient of thermal expansion (CTE) and Poisson’s ratio, of FM sample show anisotropy, as described in Table 2. The elastic modulus of monolithic silicon nitride was significantly higher, and the transverse modulus of the FM was less than half the longitudinal modulus, due to the BN, which has a small c-axis modulus. The longitudinal thermal expansion of the FM was slightly less than monolithic silicon nitride, decreased by the a-axis BN, while the transverse thermal expansion of the FM was much larger, increased by the c-axis BN. Poisson’s ratios were estimated from rule-of-mixture by taking 0.27 and 0.2 for Si3N4 21 and BN,22 respectively. The magnitude of thermal stress induced by the same exposure will be different, depending on the cell alignment (longitudinal and transverse direction). The traditional approach to evaluate the thermal shock resistance is based on quenching the specimen from an elevated temperature into a quenching media and measuring the fracture strength of the material. Neglecting the heat transfer and size effects, the maximum tensile stress (TS) generated on the surface of the specimen can be calculated according to:11 TS ¼ ð Þ E=ð Þ 1 DT ð1Þ where E, and represent the elastic modulus, the coefficient of thermal expansion (CTE) and Poisson’s Fig. 7. Optical photographs of crack propagation of fibrous monolithic Si3N4/BN ceramic after thermal shock with temperature difference (T) of (A) 800 C, (B) 1000 C, (C) 1200 C, and (D) 1400 C during flexural testing. All samples showed extensive crack interactions, such as crack delaminations and crack deflections. Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347 2343
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 1404)·2° peak/1"peak 0000 }- 100um 4 micrograph of fibrous monolithic Si3N4/BN ceramic Temperature Difference[CI al shock with temperature difference (An of 1400C. re at temperature up to 1200C, the surface was not Fig. 9. Retained apparent strength of fibrous monolithic Si3 N4/BN hile after exposure at 1400 oC, the surface layer was ceramics; (A)2nd peak/lst peak and(B)Ist drop/lst peak from load- damaged by the oxidation of both Bn cell boundaries and Si3 N4 cell deflection curve. After thermal shock, the retained strength of 2nd material peak/lst peak increased, implying that the fracture initiated from sur- face defects generated by thermal shock. Note, the retained strengths after first drop(b)of the all samples are higher than 40%, suggesting excellent load-bearing capacity ratio, respectively. AT is the temperature difference between exposure and water temperature 4.2. Thermal shock resistance parameter The normalized thermal stresses(oN Ts), that is, ther- mal stresses with longitudinal and transverse direction The conditions for crack and propaga were divided by that of monolithic Si3 N4 are estimated, have been extensively analysed by hasselman et al. as described in Table 2. Lower tensile stresses were Opposing property requirements prevail, depending on developed on the surface of FM samples with 80 and whether the material is required to be resistant to crack 55% for longitudinal and transverse direction, respec- initiation(for which high strength and low stifness are tively. Considering temperature difference of 1200oC, essential) or resistant to strength degradation following ne tensile stress of 2000 MPa was developed on the a severe thermal shock (in which case low strength and surface of monolithic Si3 N4; this value is high enough to high stifness are beneficial). We consider the crack develop the cracks on the surface, resulting in cata- initiation parameter R and crack propagation para strophic drop in flexural strength [Fig. 4(A)]. However, meter R estimated for monolithic Si3 N4 and FM actual thermal stress needs consideration of the heat sample consisting only longitudinal direction despite transfer depending on heat transfer coefficien there was anisotropy in thermal stress. These para- quenching medium, the thermal conductivity(k)and the meters can be expressed as characteristic dimension of the sample. Moreover, the r=kor(I-v)/(Ea) value from Eq. (1)only suggests the condition for crack nitiation which is critical for brittle material (i.e. monolithic Si3 N4) and not for crack propagation, which more important for tough material (i.e, FM sample) R"=KRc/(G2·(-) Therefore, a new parameter for describing thermal where, or is the fracture strength and Kic is the tough shock resistance should be considered ness and k is the thermal conductivity of the material The fracture strength of monolithic Si3N4 is almost twice that of FM sample(Table 1). The thermal con- ized thermal stress (ON. ductivities are 37 and 53 W/m K for monolithic Si3N4 eveloped on the surface of monolithic Si3 N4 and fibrous monolithic and FM sample with longitudinal direction. The calcu lated crack initiation parameter R(17.7 kw/m)of monolithic Si3 n4 is only slightly larger than that (15.8 (GPa) (10-C N,Is kW/) of FM sample, implying the condition for crack initiations are almost the same. Therefore, the large 318 difference in behavior can not be explained by resistance Fibrous monolith(longitudinal) 276 0.25 3.8 0.80 Fibrous monolith( transverse) 270.126.7 0.55 to crack initiation. The toughness of FM sample twice that of monolithic Si3N4.23 The calculated crack
ratio, respectively. T is the temperature difference between exposure and water temperature. The normalized thermal stresses (N,TS), that is, thermal stresses with longitudinal and transverse direction were divided by that of monolithic Si3N4, are estimated, as described in Table 2. Lower tensile stresses were developed on the surface of FM samples with 80 and 55% for longitudinal and transverse direction, respectively. Considering temperature difference of 1200 C, the tensile stress of 2000 MPa was developed on the surface of monolithic Si3N4; this value is high enough to develop the cracks on the surface, resulting in catastrophic drop in flexural strength [Fig. 4(A)]. However, actual thermal stress needs consideration of the heat transfer depending on heat transfer coefficient of the quenching medium, the thermal conductivity (k) and the characteristic dimension of the sample. Moreover, the value from Eq. (1) only suggests the condition for crack initiation which is critical for brittle material (i.e., monolithic Si3N4) and not for crack propagation, which is more important for tough material (i.e., FM sample). Therefore, a new parameter for describing thermal shock resistance should be considered. 4.2. Thermal shock resistance parameter The conditions for crack initiation and propagation have been extensively analysed by Hasselman et al.11,12 Opposing property requirements prevail, depending on whether the material is required to be resistant to crack initiation (for which high strength and low stiffness are essential) or resistant to strength degradation following a severe thermal shock (in which case low strength and high stiffness are beneficial). We consider the crack initiation parameter R0 and crack propagation parameter R0000 estimated for monolithic Si3N4 and FM sample consisting only longitudinal direction despite there was anisotropy in thermal stress. These parameters can be expressed as R0 ¼ k fð Þ 1 =ðÞ ð E 2Þ R0000 ¼ K2 IC= 2 f ð Þ 1- ð3Þ where, f is the fracture strength and KIC is the toughness and k is the thermal conductivity of the material. The fracture strength of monolithic Si3N4 is almost twice that of FM sample (Table 1). The thermal conductivities are 37 and 53 W/m K for monolithic Si3N4 and FM sample with longitudinal direction. The calculated crack initiation parameter R0 (17.7 kW/m) of monolithic Si3N4 is only slightly larger than that (15.8 kW/m) of FM sample, implying the condition for crack initiations are almost the same. Therefore, the large difference in behavior can not be explained by resistance to crack initiation. The toughness of FM sample was twice that of monolithic Si3N4. 23 The calculated crack Table 2 The values for calculating the normalized thermal stress (N,TS) developed on the surface of monolithic Si3N4 and fibrous monolithic Si3N4/BN ceramic Samples E (GPa) (106 / C) N,TS Monolithic Si3N4 318 0.27 4 1 Fibrous monolith (longitudinal) 276 0.25 3.8 0.80 Fibrous monolith (transverse) 127 0.12 6.7 0.55 Fig. 8. SEM micrograph of fibrous monolithic Si3N4/BN ceramic after thermal shock with temperature difference (T) of 1400 C. After exposure at temperature up to 1200 C, the surface was not damaged, while after exposure at 1400 C, the surface layer was damaged by the oxidation of both BN cell boundaries and Si3N4 cell material. Fig. 9. Retained apparent strength of fibrous monolithic Si3N4/BN ceramics; (A) 2nd peak/1st peak and (B) 1st drop /1st peak from load– deflection curve. After thermal shock, the retained strength of 2nd peak/1st peak increased, implying that the fracture initiated from surface defects generated by thermal shock. Note, the retained strengths after first drop (B) of the all samples are higher than 40%, suggesting excellent load-bearing capacity. 2344 Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 2345 propagation parameter R of FM sample is much lar- BN-rich cell boundaries. During thermal shock, the ger(>16 times) than that of monolithic Si3 N4, implying longitudinal thermal stress may fracture occasional that the crack propagation is more restricted for FM Si3 N4 cells, as shown in Fig. 10(B). The transverse ther sample, while the resistance to crack initiation is slightly mal stress most likely causes localized extension of the lower than that of monolithic Si3N4 pre-existing flaws in the BN-rich cell boundary and is Considering thermal shock parameters(R and R), unlikely to cause cracks within the Si3N4 cells. The FM sample is expected to show excellent thermal shock postulated BN-cell boundary cracks could not be resistance(see Fig. 4)because the resistance to crack observed because they would be observed by the pre- propagation(R")is much higher, while the resistance existing cracks in the BN and the rough topography of to crack initiation (R) is slightly lower compared to the cell boundary region. However, the extension of monolithic Si3 N4. Similar increased thermal shock BN-cell boundary cracks is believed to decreases the cell resistance have been observed for layered ceramic boundary fracture resistance(TBN), which is consisted structures where the cracks are deflected reliably at the with the observation of more extensive delamination interfaces. 16 after flexural testing of severely shocked sample The degree of delamination cracks is significantly 43. Pre-existing microcracks on BN cell boundaries the de of thermal stress due thermal shock that can extend the pre-existing cracks in We have previously observed the generation of cracks BN-rich cell boundary. The fraction of cell boundary within BN cell boundary layer after hot-pressing. The delamination was calculated by counting the ratio of CtE of BN in the basal plane is slightly negative from delaminated layers to the total number of cell boundary room temperature to 800 oC, about -2x10-6/oC, 24 layer from the SEM micrographs, as shown in Fig. 11 while, the Cte perpendicular to the basal plane is very Three cell boundaries(marked by arrows)were exten large and positive, about +40x10-/C. Therefore, sively delaminated and the rest one was remained ne bn contracts perpendicular to the basal plane (i.e, without delamination crack. The degree of delamination in the [0001] direction) during cooling. If the surround- cracks significantly increased after thermal shock, as ng Si3 n4 grains or glassy phase constrain the bn shown in Fig. 12. Before thermal shock, 40% cell platelets, large tensile stresses are developed perpend- boundaries were delaminated. However, after thermal cular to the basal plane, resulting in separating BN pla- shock with a temperature difference of 1400oC, almost telets into layers along the basal plane direction. Thus every BN cell boundary was delaminated. These results the as-fabricated specimens(before the thermal shock are attributed the decrease in cell boundary fracture treatment) had many pre-existing microcracks within resistance(TBN) through the extension of pre-existing he BN- rich cell boundary. The delaminated micro- cracks of the BN-rich cell boundaries, including the (a)Thermal Stress after Thermal Shock amount of the glass and the extent of pre-existing microcracks. determine the fracture resistance of the cell boundary (TBN) and the tendency for the crack deflec- tion and delamination Furthermore, shear stresses developed parallel to the basal plane made the surface of the platelets slide rela t ve to each other. Similarly, pre-existing mi were extended due to the anisotropy in CtE after ther- (b) Cracking due to Thermal Stress mal shock, dissipating the thermal stress; therefore, arrange 40.55 om crack propagations through bn cell boundaries became more favorable, resulting in high WOF (see Figs. 5 and Cell 7). Some researches have observed that pre-existing microcracks in BN platelets of Si3N-BN and Al_O3- t Cell Cracking Bn composites are beneficial to thermal shock cLonal & 0.80o SiN, Cell N-rich 4.4. Thermal shock induced cracks and crack Cell Boundary propagation during subsequent flexural testing After thermal shock, the Fm sample is placed under Fig. 10. (A)Therma after thermal shock in transverse and transverse and longitudinal stress depending on the fiber longitudinal direction Si3 Na cell between two BN-rich cell bo ting cracks in cell alignment, as sho wn In Fig. 10(A). The flaw tolerant boundaries(-.) and cell boundary cracks by thermal shock nature of the FM is related to crack deflection at the (),(ii) possible transverse cracks in Si, cell
propagation parameter R0000 of FM sample is much larger (>16 times) than that of monolithic Si3N4, implying that the crack propagation is more restricted for FM sample, while the resistance to crack initiation is slightly lower than that of monolithic Si3N4. Considering thermal shock parameters (R0 and R0000), FM sample is expected to show excellent thermal shock resistance (see Fig. 4) because the resistance to crack propagation (R0000) is much higher, while the resistance to crack initiation (R0 ) is slightly lower compared to monolithic Si3N4. Similar increased thermal shock resistance have been observed for layered ceramic structures where the cracks are deflected reliably at the interfaces.16 4.3. Pre-existing microcracks on BN cell boundaries We have previously observed the generation of cracks within BN cell boundary layer after hot-pressing.1 The CTE of BN in the basal plane is slightly negative from room temperature to 800 C, about 2106 / C,24 while, the CTE perpendicular to the basal plane is very large and positive, about +40106 / C.25 Therefore, the BN contracts perpendicular to the basal plane (i.e., in the [0001] direction) during cooling. If the surrounding Si3N4 grains or glassy phase constrain the BN platelets, large tensile stresses are developed perpendicular to the basal plane, resulting in separating BN platelets into layers along the basal plane direction. Thus the as-fabricated specimens (before the thermal shock treatment) had many pre-existing microcracks within the BN-rich cell boundary. The delaminated microcracks of the BN-rich cell boundaries, including the amount of the glass and the extent of pre-existing microcracks, determine the fracture resistance of the cell boundary (GBN) and the tendency for the crack deflection and delamination.810 Furthermore, shear stresses developed parallel to the basal plane made the surface of the platelets slide relative to each other. Similarly, pre-existing microcracks were extended due to the anisotropy in CTE after thermal shock, dissipating the thermal stress; therefore, crack propagations through BN cell boundaries became more favorable, resulting in high WOF (see Figs. 5 and 7). Some researches have observed that pre-existing microcracks in BN platelets of Si3N4–BN and Al2O3– BN composites are beneficial to thermal shock resistance.26,27 4.4. Thermal shock induced cracks and crack propagation during subsequent flexural testing After thermal shock, the FM sample is placed under transverse and longitudinal stress depending on the fiber alignment, as shown in Fig. 10(A). The flaw tolerant nature of the FM is related to crack deflection at the BN-rich cell boundaries. During thermal shock, the longitudinal thermal stress may fracture occasional Si3N4 cells, as shown in Fig. 10(B). The transverse thermal stress most likely causes localized extension of the pre-existing flaws in the BN-rich cell boundary and is unlikely to cause cracks within the Si3N4 cells. The postulated BN-cell boundary cracks could not be observed because they would be observed by the preexisting cracks in the BN and the rough topography of the cell boundary region. However, the extension of BN-cell boundary cracks is believed to decreases the cell boundary fracture resistance (GBN), which is consisted with the observation of more extensive delamination after flexural testing of severely shocked sample. The degree of delamination cracks is significantly dependent on the magnitude of thermal stress due to thermal shock that can extend the pre-existing cracks in BN-rich cell boundary. The fraction of cell boundary delamination was calculated by counting the ratio of delaminated layers to the total number of cell boundary layer from the SEM micrographs, as shown in Fig. 11. Three cell boundaries (marked by arrows) were extensively delaminated and the rest one was remained without delamination crack. The degree of delamination cracks significantly increased after thermal shock, as shown in Fig. 12. Before thermal shock, 40% cell boundaries were delaminated. However, after thermal shock with a temperature difference of 1400 C, almost every BN cell boundary was delaminated. These results are attributed the decrease in cell boundary fracture resistance (GBN) through the extension of pre-existing Fig. 10. (A) Thermal stress after thermal shock in transverse and longitudinal direction and (B) A schematic of single Si3N4 cell between two BN-rich cell boundaries, illustrating (i) pre-existing cracks in cell boundaries (- - -) and cell boundary cracks extended by thermal shock (—), (ii) possible transverse cracks in Si3N4 cell. Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347 2345
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 0.8 06 Pull-out Length 204 200um v”△T=1400° Fig. Il. SEM micrograph of the thermally shocked sample after flex. 0 ural testing, illustrating the delamination cracks(marked by arrow Pull-out Length mm and pull-out length Fig 13. Cumulative distribution function versus pull-out length of the samples before and after thermal shock with a various temperature difference△T) fracture resistance(T BN). The FM samples after thermal shock with the temperature difference of 1200C, 70% of the delamination cracks showed the long pull-out length(>1000 um). However, after thermal shock with the temperature difference of 1400C, the pull-out length was decreased again even tough almost every BN- rich cell boundary was delaminated (see Fig. 12) This result is attributed to the damage of both Si3N4 40 cells and bn cell boundaries due to the oxidation in other words, when the Si3 N4 cell is strong (i.e, few flaws on the surface), the delamination cracks extends to long 200400600800100012001400 distance before kinking out of BN-rich cell boundary. Temperature Difference [Cl However, when the Si3 N4 cell is damaged due to the oxidation (i.e, many flaws on the surface, see Fig. 8) Fig.12.Fraction of crack delamination in the thermally shocked amounts of the delamination cracks kink out of the BN sample after flexural testing as a function of temperature difierence rich cell boundary after propagating only a short dis (An. Temperature difference(An) of 0 oC represents the sample before thermal shoc tance, in which the flaws are present on Si3N4 cell, while almost every cell boundary is delaminated due to the cracks on BN-rich cell boundaries, consequently pro- decrease in cell boundary fracture resistance moting the delamination cracks The magnitude of thermal stress is expected to change the length of the delamination cracks however it is 5. Conclusions difficult to quantify the length of delamination cracks, because it is not easy to discern the crack tip in the BN Excellent thermal shock resistance was observed for rich cell boundary. Therefore, the delamination dis- fibrous monolithic Si3 N4/BN ceramics. Monolithic tance, defined as pull-out length(see Fig. 11), can be Si3 n4 showed a catastrophic drop in flexural strength measured from the distance between through-thickness with temperature difference of 1000C, meaning that cracks in adjacent Si3N4 layers. A cumulative distribu- the tensile stress was developed on the surface exceeding tion plot of pull-out lengths is shown in Fig 13 for each its fracture strength, and thus cracking the surface. of the samples before and after thermal shock. Before However, fibrous monolithic showed negligible reduc- thermal shock, the FM sample showed the amounts of tion in flexural strength, and remarkable increase in short pull-out length(<100 um). Almost half of the work-of-fracture(WOF). Such excellent thermal shock delamination cracks kinked out of the BN cell boundary resistance was attributed to high resistance to crack after propagating only a short distance. However, after propagation(rm) through crack interactions with weak thermal shock, the pull-out length was significantly cell boundaries. The remarkable increase in WOF increased, implying that the decrease in cell boundary delamination cracks were attributed to the reduction
cracks on BN-rich cell boundaries, consequently promoting the delamination cracks. The magnitude of thermal stress is expected to change the length of the delamnination cracks; however, it is difficult to quantify the length of delamination cracks, because it is not easy to discern the crack tip in the BNrich cell boundary. Therefore, the delamination distance, defined as pull-out length (see Fig. 11), can be measured from the distance between through-thickness cracks in adjacent Si3N4 layers. A cumulative distribution plot of pull-out lengths is shown in Fig. 13 for each of the samples before and after thermal shock. Before thermal shock, the FM sample showed the amounts of short pull-out length (1000 mm). However, after thermal shock with the temperature difference of 1400 C, the pull-out length was decreased again even tough almost every BN-rich cell boundary was delaminated (see Fig. 12). This result is attributed to the damage of both Si3N4 cells and BN cell boundaries due to the oxidation. In other words, when the Si3N4 cell is strong (i.e., few flaws on the surface), the delamination cracks extends to long distance before kinking out of BN-rich cell boundary. However, when the Si3N4 cell is damaged due to the oxidation (i.e., many flaws on the surface, see Fig. 8), amounts of the delamination cracks kink out of the BNrich cell boundary after propagating only a short distance, in which the flaws are present on Si3N4 cell, while almost every cell boundary is delaminated due to the decrease in cell boundary fracture resistance. 5. Conclusions Excellent thermal shock resistance was observed for fibrous monolithic Si3N4/BN ceramics. Monolithic Si3N4 showed a catastrophic drop in flexural strength with temperature difference of 1000 C, meaning that the tensile stress was developed on the surface exceeding its fracture strength, and thus cracking the surface. However, fibrous monolithic showed negligible reduction in flexural strength, and remarkable increase in work-of-fracture (WOF). Such excellent thermal shock resistance was attributed to high resistance to crack propagation (R0000) through crack interactions with weak cell boundaries. The remarkable increase in WOF and delamination cracks were attributed to the reduction in Fig. 11. SEM micrograph of the thermally shocked sample after flexural testing, illustrating the delamination cracks (marked by arrow) and pull-out length. Fig. 12. Fraction of crack delamination in the thermally shocked sample after flexural testing as a function of temperature difference (T). Temperature difference (T) of 0 C represents the sample before thermal shock. Fig. 13. Cumulative distribution function versus pull-out length of the samples before and after thermal shock with a various temperature difference (T). 2346 Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347
Y.H. Koh et al. Journal of the European Ceramic Society 24(2004)2339-2347 2347 cell boundary fracture resistance by extension of pre- initiation and crack propagation in brittle ceramics. J. Am. ng microcracks on BN-rich cell boundaries. 12. Hasselman. D. P. H, Thermal stress resistance parameters for brittle refractory ceramics: a compendium. Am. Ceram. Soc Bl.1970.,49,1033-1037 Acknowledgements 13. Wang. H. and Singh, R. N. Thermal shock behavior of ceramics and ceramic composites. Int. Mater. Rev., 1994. 39. 228-244 This research was supported by a grant fro 14. Aldridge. M. and Yeomans. J. A, The thermal shock behavior of Center for Advanced Materials Processing(CAMP) of ductile particle toughened alumina composites. J. Eur. Ceram. e 21st Century Frontier R&D Program funded by the Soc.,1998,19,1769-1775 15. Hirano, T and Ihara, K, Thermal shock resistance of Si3N4/ Ministry of Science and Technology, Republic of SiC nanocomposites fabricated from amorphous Si-C-n pre- cursor powders. Mater. Lett., 1996, 26, 285-289 6. Vandeperre L.J., Kristofferson, A, Carlstrom, E. and Clegg, w.J., Thermal shock of layered ceramic structures with crack- References deflecting interfaces.J. Am. Ceram. Soc. 2001. 84. 104-110 17. Lee, w.J. and Case, E. D, Cyclic thermal shock in SiC-whisker 1. Kovar. D. King. B. H. Trice. R. W. and Halloran. J. w reinforced alumina composite. Mater. Sci Eng, 1989. A119, 113- Fibrous monolithic ceramics. J. Am. Ceram. Soc.. 1997 80 2471 18. Schneibel. H. Sabol, S. M. Morrison, J, Ludeman, E. and 2487 2. Baskaran, S, Nunn. S. D, Popovic, D. and Halloran. J. H Carmichael, C. A, Cyclic thermal shock resistance of several Fibrous monolithic ceramics L. fabrication. microstructure advanced ceramics and ceramic composites. J. Am. Ceram. Soc., and indentation behavior. Am. Ceram. Soc. 1993. 76. 2209- 1998,81,1888-1892 2216 19. Lienard. s. Y Kovar d. moon .r.. bowman k.. and 3. Baskaran S and Halloran. J. H. Fibrous monolithic ceramics: Halloran, J. H, Texture development in Si3N4/BN fibrous J. Mater.Sci,2000,35,3365-3371 Il. flexural strength and fracture behavior of the silicon carbide, graphite system. J. Am. Ceram. Soc., 1993. 76. 2217-2224 20. 494-92a Standard Practice for Measuring Ultrasonic Velocity in 4. King. Influence of Architecture on the Mechanical Properties laterals. In Annual Book of ASTM Standards, Section 3, Vol. of Fibrous Monolithic Ceramics. PhD thesis. University of Michi Nondestructive Testing Methods. ed. P. C. Fazio et al. amer gan, Ann Arbor, MI. 1997 ican Society for Testing and Materials, Philadelphia, PA, 1994, 5. Trice. R. W. and Halloran, J. H, Effect of sintering aid compo- pp.179-190 tion on the processing of Si,N4/BN fibrous monolithic 21. Hampshire, S, Nitride ceramics In Engineered Materials Hand- J.Am. Ceran.Soc,1999,82,2943-2947 8l4-820 6. Trice, R.W. and Halloran, J H Elevated-temperature mechan- 22. Killey, A. and MacMillan, N.H., Strong Solids, 3 ed. Oxford ical properties of silicon nitride/ boron nitride fibrous ceramics. J. Am. Ceram. Soc. 2000.83 311-316 23. Koh, Y.H., Kim, H. w. and Kim, H. E, Mechanical properties 7. Trice. R. w. and Halloran. J H. Influence of microstructure and Si3N4/B ceramics with different cell temperature on the interfacial fracture energy of silicon nitride/ boron nitride fibrous monolithic ceramics. 4m. Ceram. Soc 24. Kelly, B. T, The anisotropic thermal expansion of boron nitride 999.82.2502-2508. ll. Interpretation by the semi-continuum modeL. Phil. Mag, 8. Kovar. D. Thouless. M. D. and Halloran. J H. Crack deflection 1975,32,859-867. nd propagation in layered silicon nitride/boron nitride ceramics. 25. Yates, B, Ovary, M. J and Pirgon, O, The anisotropic thermal J.Am. Ceran.Soe.,1998,81,1004-1012 pansion of boron nitride I. Experimental results and their ana- 9.He,M.Yand Hutchinson, J H, crack deflection at an interface lysis.Phil.Mag-,1975,32,847-857 between dissimilar elastic materials. Int. J. Solids Structures, 26. Goeuriot-Launay, D, Brayet, G. and Thevenot, F, Boron 1989,25,1053-1067 nitride effect on the thermal shock resistance of an alumina based 10. Phillipps, A.J., Clegg, W.J. and Clyne, T.w. Fracture beha- ceramic composite. J.Mater. Sci. Lett., 1986, 5, 940-942. 27. Lutz. E. H. and Swain. M. V. Fracture toughness and thermal propagation. Acta Metall. Mater., 1993, 41, 805-817 shock behavior of silicon nitride-boron nitride ceramics. J. Am I1. Hasselman, D. P. H, Unified theory of thermal shock fracture Ceran.Soc.1992,75,67-70
cell boundary fracture resistance by extension of preexisting microcracks on BN-rich cell boundaries. Acknowledgements This research was supported by a grant from the Center for Advanced Materials Processing (CAMP) of the 21st Century Frontier R&D Program funded by the Ministry of Science and Technology, Republic of Korea. References 1. Kovar, D., King, B. H., Trice, R. W. and Halloran, J. W., Fibrous monolithic ceramics. J. Am. Ceram. Soc., 1997, 80, 2471– 2487. 2. Baskaran, S., Nunn, S. D., Popovic, D. and Halloran, J. H., Fibrous monolithic ceramics: I, fabrication, microstructure, and indentation behavior. J. Am. Ceram. Soc., 1993, 76, 2209– 2216. 3. Baskaran, S. and Halloran, J. H., Fibrous monolithic ceramics: II, flexural strength and fracture behavior of the silicon carbide/ graphite system. J. Am. Ceram. Soc., 1993, 76, 2217–2224. 4. King, B. H. Influence of Architecture on the Mechanical Properties of Fibrous Monolithic Ceramics. PhD thesis, University of Michigan, Ann Arbor, MI, 1997. 5. Trice, R. W. and Halloran, J. H., Effect of sintering aid composition on the processing of Si3N4/BN fibrous monolithic ceramics. J. Am .Ceram. Soc., 1999, 82, 2943–2947. 6. Trice, R. W. and Halloran, J. H., Elevated-temperature mechanical properties of silicon nitride/boron nitride fibrous monolithic ceramics. J. Am. Ceram. Soc., 2000, 83, 311–316. 7. Trice, R. W. and Halloran, J. H., Influence of microstructure and temperature on the interfacial fracture energy of silicon nitride/ boron nitride fibrous monolithic ceramics. J. Am. Ceram. Soc., 1999, 82, 2502–2508. 8. Kovar, D., Thouless, M. D. and Halloran, J. H., Crack deflection and propagation in layered silicon nitride/boron nitride ceramics. J. Am. Ceram. Soc., 1998, 81, 1004–1012. 9. He, M. Y. and Hutchinson, J. H., crack deflection at an interface between dissimilar elastic materials. Int. J. Solids Structures, 1989, 25, 1053–1067. 10. Phillipps, A. J., Clegg, W. J. and Clyne, T. W., Fracture behaviour of ceramic laminates in bending: I. Modelling of crack propagation. Acta Metall. Mater., 1993, 41, 805–817. 11. Hasselman, D. P. H., Unified theory of thermal shock fracture initiation and crack propagation in brittle ceramics. J. Am. Ceram. Soc., 1969, 52, 600–604. 12. Hasselman, D. P. H., Thermal stress resistance parameters for brittle refractory ceramics: a compendium. Am. Ceram. Soc. Bull., 1970, 49, 1033–1037. 13. Wang, H. and Singh, R. N., Thermal shock behavior of ceramics and ceramic composites. Int. Mater. Rev., 1994, 39, 228–244. 14. Aldridge, M. and Yeomans, J. A., The thermal shock behavior of ductile particle toughened alumina composites. J. Eur. Ceram. Soc., 1998, 19, 1769–1775. 15. Hirano, T. and Niihara, K., Thermal shock resistance of Si3N4/ SiC nanocomposites fabricated from amorphous Si–C–N precursor powders. Mater. Lett., 1996, 26, 285–289. 16. Vandeperre, L. J., Kristofferson, A., Carlstro¨m, E. and Clegg, W. J., Thermal shock of layered ceramic structures with crackdeflecting interfaces. J. Am. Ceram. Soc., 2001, 84, 104–110. 17. Lee, W. J. and Case, E. D., Cyclic thermal shock in SiC-whisker reinforced alumina composite. Mater. Sci. Eng., 1989, A119, 113– 126. 18. Schneibel, J. H., Sabol, S. M., Morrison, J., Ludeman, E. and Carmichael, C. A., Cyclic thermal shock resistance of several advanced ceramics and ceramic composites. J. Am. Ceram. Soc., 1998, 81, 1888–1892. 19. Lienard, S. Y., Kovar, D., Moon, R. J., Bowman, K. J. and Halloran, J. H., Texture development in Si3N4/BN fibrous monolithic ceramics. J. Mater. Sci., 2000, 35, 3365–3371. 20. 494-92a Standard Practice for Measuring Ultrasonic Velocity in Materials. In Annual Book of ASTM Standards, Section 3, Vol. 03, Nondestructive Testing Methods, ed. P. C. Fazio et al. American Society for Testing and Materials, Philadelphia, PA, 1994, pp. 179–190. 21. Hampshire, S., Nitride ceramics. In Engineered Materials Handbook, Vol. 4, Ceramics and Glasses. ASM International, 1991, pp. 814–820. 22. Killey, A. and MacMillan, N. H., Strong Solids, 3 ed. Oxford University Press, 1986. 23. Koh, Y. H., Kim, H. W. and Kim, H. E., Mechanical properties of fibrous monolithic Si3N4/BN ceramics with different cell boundary thicknesses. J. Ecers. Ceram. Soc. (in press). 24. Kelly, B. T., The anisotropic thermal expansion of boron nitride II. Interpretation by the semi-continuum model. Phil. Mag., 1975, 32, 859–867. 25. Yates, B., Ovary, M. J. and Pirgon, O., The anisotropic thermal expansion of boron nitride I. Experimental results and their analysis. Phil. Mag., 1975, 32, 847–857. 26. Goeuriot-Launay, D., Brayet, G. and Thevenot, F., Boron nitride effect on the thermal shock resistance of an alumina based ceramic composite. J. Mater. Sci. Lett., 1986, 5, 940–942. 27. Lutz, E. H. and Swain, M. V., Fracture toughness and thermal shock behavior of silicon nitride–boron nitride ceramics. J. Am. Ceram. Soc., 1992, 75, 67–70. Y.-H. Koh et al. / Journal of the European Ceramic Society 24 (2004) 2339–2347 2347