E驅≈3S Journal of the European Ceramic Society 20(2000)246.3-2468 Mullite/zirconia laminate composites for high temperature application Sang-Yeup Park a. * Biege Saruhan Hartmut Schneider b Department of Materials Engineering, Kangnung National University, Kangnung, Kangwondo 210-702, South Korea Institute for Materials Research, German Aerospace Center, 51147 Cologne, Germany Received 26 February 2000: received in revised form 3 May 2000; accepted 11 May 2000 Abstract ullite/zirconia laminate composites were studied for a better understanding of crack formation and crack propagation behavior at the interface of laminates. The laminate composites were fabricated by tape casting method and hot-pressing. It was demon- strated that the thermal mismatch stress was closely related to the crack formation and crack deflection behavior in the mullite/ zirconia laminate composites. Two kinds of cracks were observed in zirconia layers: one is channel cracks in layers subjected to tensile stress, the other is the edge cracks in layers subjected to compressive stress. Among three forms of zirconia(monoclinic, tetragonal, cubic) tetragonal zirconia was effective to deflect the crack at the interface due to the stress induced tetra- mono phase transformation. C 2000 Published by elsevier Science Ltd. All rights reserved. Keywords: Composites: Crack growth; Laminates; Mullite; Mullite-ZrO2: ZrO2 1. Introduction thermal expansion coefficients during cooling. Typical cracks observed in laminated composites are related to Oxide/oxide composites have been receiving growing one or a combination of the origins of mismatch stress attention, mainly due to their oxidation stability at high Recently, many researches have been made on oxide/ temperature applications. The key factor in providing oxide laminate composites, mainly on alumina/ zirconia damage tolerant behavior at high temperatures as well and several other systems. However, there are only as at room temperature is the control of interface prop- limited reports on mullite/zirconia laminate compo- erties in order to achieve weak bonding between lami- sites, O even though a large number of studies have been nate layers. 3 This depends upon avoiding chemical made on mullite/zirconia composites. Due to the high reactions and upon providing a crack deflection diffusion coefficient of atoms in oxides and the reactivity mechanism between individual layers of mullite with other oxides at high temperatures, there Oxide laminate composites have been fabricated by exists a limited number of candidate materials compa- various routes including tape casting, 4 slip casting, di tible with mullite. Zirconia is one of the most promising pressing, electrophoretic deposition. 7 However, various candidates due to its chemical stability vs. mullite at kinds of processing defects are commonly observed in high temperature purpose of this study is to the laminate composites during processing. Typical describe the crac mations during the processing of edge crack(or transverse crack) parallel to the interface, behaviors at the interface of the laminate. and debonding. These cracks can be generated during the processing due to the mismatch stresses betw layers. Origins of mismatch stresses are related to 2. Experimental procedure ferential drying, differential densification, and different Starting powders were mullite precursors composed of y-Al2O3 and Sio2(Siral, Condea, FRG; and mono- kr(S.Y. Park). clinic zirconia (TZ-0, Tosho, Japan), tetragonal zirconia 0955-2219/00/S.see front matter C 2000 Published by Elsevier Science Ltd. All rights reserved PII:S0955-2219(00)00157-6
Mullite/zirconia laminate composites for high temperature application Sang-Yeup Park a,*, Biege Saruhan b, Hartmut Schneider b a Department of Materials Engineering, Kangnung National University, Kangnung, Kangwondo 210-702, South Korea bInstitute for Materials Research, German Aerospace Center, 51147 Cologne, Germany Received 26 February 2000; received in revised form 3 May 2000; accepted 11 May 2000 Abstract Mullite/zirconia laminate composites were studied for a better understanding of crack formation and crack propagation behavior at the interface of laminates. The laminate composites were fabricated by tape casting method and hot-pressing. It was demonstrated that the thermal mismatch stress was closely related to the crack formation and crack de¯ection behavior in the mullite/ zirconia laminate composites. Two kinds of cracks were observed in zirconia layers; one is channel cracks in layers subjected to tensile stress, the other is the edge cracks in layers subjected to compressive stress. Among three forms of zirconia (monoclinic, tetragonal, cubic) tetragonal zirconia was eective to de¯ect the crack at the interface due to the stress induced tetra-mono phase transformation. # 2000 Published by Elsevier Science Ltd. All rights reserved. Keywords: Composites; Crack growth; Laminates; Mullite; Mullite-ZrO2; ZrO2 1. Introduction Oxide/oxide composites have been receiving growing attention, mainly due to their oxidation stability at high temperature applications.1 The key factor in providing damage tolerant behavior at high temperatures as well as at room temperature is the control of interface properties in order to achieve weak bonding between laminate layers.2,3 This depends upon avoiding chemical reactions and upon providing a crack de¯ection mechanism between individual layers. Oxide laminate composites have been fabricated by various routes including tape casting,4 slip casting,5 die pressing,6 electrophoretic deposition.7 However, various kinds of processing defects are commonly observed in the laminate composites during processing. Typical cracks observed in the laminate composites are channel crack (or tunnel crack) perpendicular to the interface, edge crack (or transverse crack) parallel to the interface, and debonding.8 These cracks can be generated during the processing due to the mismatch stresses between layers. Origins of mismatch stresses are related to differential drying, dierential densi®cation, and dierent thermal expansion coecients during cooling. Typical cracks observed in laminated composites are related to one or a combination of the origins of mismatch stress. Recently, many researches have been made on oxide/ oxide laminate composites, mainly on alumina/zirconia and several other systems.9 However, there are only limited reports on mullite/zirconia laminate composites,10 even though a large number of studies have been made on mullite/zirconia composites.11 Due to the high diusion coecient of atoms in oxides and the reactivity of mullite with other oxides at high temperatures, there exists a limited number of candidate materials compatible with mullite. Zirconia is one of the most promising candidates due to its chemical stability vs. mullite at high temperature. The purpose of this study is to describe the crack formations during the processing of mullite/zirconia laminate composites with dierent types of zirconia, and to demonstrate crack de¯ection behaviors at the interface of the laminate. 2. Experimental procedure Starting powders were mullite precursors composed of g-Al2O3 and SiO2 (Siral, Condea, FRG); and monoclinic zirconia (TZ-0, Tosho, Japan), tetragonal zirconia 0955-2219/00/$ - see front matter # 2000 Published by Elsevier Science Ltd. All rights reserved. PII: S0955-2219(00)00157-6 Journal of the European Ceramic Society 20 (2000) 2463±2468 * Corresponding author. E-mail address: sypark@knusun.kangnung.ac.kr (S.-Y. Park)
S.Y. Park et al. Journal of the European Ceramic Society 20(2000)2463-2468 with 3 mol%Y2O3(TZ-3Y, Tosho, Japan), cubic zir- to 1300C and 15 min holding time with controlled conia with 8 mol%Y2O3(TZ-8Y, Tosho, Japan). The heating and cooling rate. Thermal expansion coefficients of mullite precursor was calcined at 1 100C for 5 h in air. each layer were measured on a dilatometer( Bahr-thermo- Each slip was prepared by planetary ball milling with 70 analyse, Germany) for temperatures up to 1300C wt% solid loading. For the preparation of stable aqu eous slip, total amount of 2.5 wt. polyglycerin, cellu- lose. and silicon emulsion were added in order to 3. Results and discussion enhance the powder dispers reen dimensions of 60x60 mm were obtained by tape casting Phase composition and microstructure of as-pro- After tape casting, tapes were dried for 24 h at ambient cessed laminate reveal that the tapes produced from the condition. For the binder burn-out, dried tapes were calcined mullite precursor are fully crystallized into placed in air at 700C for 2 h with a slow heating rate of mullite after hot-pressing at 1300 C for 15 min under 15 MPa in air, as shown in Fig. 1. The ZrO2 tapes pro- Three combinations of laminate composites (mM: duced from unstabilized ZrO2 powder yield purely monoclinic ZrO2/mullite, tM: tetragonal ZrO2/mullite, monoclinic phase, whereas partially stabilized ZrO CM: cubic ZrO2/mullite)were formed by alternating layers with 3 mol% Y2O3 tapes retained mainly tetragonal (7 layers of ZrO, and 8 layers of mullite). Alternating phase and partial content of monoclinic phase. Fully layers of mullite precursor/zirconia were hot-pressed at stabilized ZrO, with 8 mol%Y2O3 tapes retained only 1300 C, 15 MPa, for 15 min in air. After hot-pressing, cubic phase under the given condition. pecimens were cooled down with & C/min to room Fig. 2 shows SEM micrographs of the laminate com temperature without pressure. Hot-pressed specimens posites, fabricated with an alternating layer of mullite were polished with a diamond paste of 3 um, and then and monoclinic ZrO2(a), tetragonal ZrO2(b), cubic Vickers indentation was carried out. Indention load ZrO2(c). After hot-pressing, all samples display channel with 10 kg was enough to produce an observable crack. cracks parallel to the hot pressing direction. In the X-ray diffractometer (Siemens D-5000, Germany) was samples tM and cM, the channel cracks are in the Zro used for the phase identification of specimen. Micro- layer, while in the sample mM the channel cracks are in structure characterization of layered composites has the mullite layer. Additionally, in the sample mM, edge performed by SEM(Philips, 525M, Netherlands) on the cracks perpendicular to hot-pressing direction are polished and the fracture surfaces of the specimens. observed in the ro2 layers Dilatometer measurements have been carried out to Typical cracks originated from mismatch stress are study the densification behavior of each layer including dependent on the several factors, such as differential mullite precursor, mono-zirconia, tetra-zirconia, and drying, differential densification, and different thermal cubic zirconia. These measurements were carried out up expansion coefficients during cooling. During the drying 6000 2000 50.0060.0070.0080.00 Fig. I. X-ray diffraction patterns for(a) the sample hot-pressed at 1300C, 15 MPa for 15 min in air and (b) mullite precursor powder
with 3 mol% Y2O3 (TZ-3Y, Tosho, Japan), cubic zirconia with 8 mol% Y2O3 (TZ-8Y, Tosho, Japan). The mullite precursor was calcined at 1100C for 5 h in air. Each slip was prepared by planetary ball milling with 70 wt.% solid loading. For the preparation of stable aqueous slip, total amount of 2.5 wt.% polyglycerin, cellulose, and silicon emulsion were added in order to enhance the powder dispersion. Green tapes with dimensions of 6060 mm were obtained by tape casting. After tape casting, tapes were dried for 24 h at ambient condition. For the binder burn-out, dried tapes were placed in air at 700C for 2 h with a slow heating rate of 2C/min. Three combinations of laminate composites (mM: monoclinic ZrO2/mullite, tM: tetragonal ZrO2/mullite, cM: cubic ZrO2/mullite) were formed by alternating layers (7 layers of ZrO2 and 8 layers of mullite). Alternating layers of mullite precursor/zirconia were hot-pressed at 1300C, 15 MPa, for 15 min in air. After hot-pressing, specimens were cooled down with 8C/min to room temperature without pressure. Hot-pressed specimens were polished with a diamond paste of 3 mm, and then Vickers indentation was carried out. Indention load with 10 kg was enough to produce an observable crack. X-ray diractormeter (Siemens D-5000, Germany) was used for the phase identi®cation of specimen. Microstructure characterization of layered composites has performed by SEM (Philips, 525M, Netherlands) on the polished and the fracture surfaces of the specimens. Dilatometer measurements have been carried out to study the densi®cation behavior of each layer including mullite precursor, mono-zirconia, tetra-zirconia, and cubic zirconia. These measurements were carried out up to 1300C and 15 min holding time with controlled heating and cooling rate. Thermal expansion coecients of each layer were measured on a dilatometer (Bahr-thermoanalyse, Germany) for temperatures up to 1300C. 3. Results and discussion Phase composition and microstructure of as-processed laminate reveal that the tapes produced from the calcined mullite precursor are fully crystallized into mullite after hot-pressing at 1300C for 15 min under 15 MPa in air, as shown in Fig. 1. The ZrO2 tapes produced from unstabilized ZrO2 powder yield purely monoclinic phase, whereas partially stabilized ZrO2 with 3 mol% Y2O3 tapes retained mainly tetragonal phase and partial content of monoclinic phase. Fully stabilized ZrO2 with 8 mol% Y2O3 tapes retained only cubic phase under the given condition. Fig. 2 shows SEM micrographs of the laminate composites, fabricated with an alternating layer of mullite and monoclinic ZrO2 (a), tetragonal ZrO2 (b), cubic ZrO2 (c). After hot-pressing, all samples display channel cracks parallel to the hot pressing direction. In the samples tM and cM, the channel cracks are in the ZrO2 layer, while in the sample mM the channel cracks are in the mullite layer. Additionally, in the sample mM, edge cracks perpendicular to hot-pressing direction are observed in the ZrO2 layers. Typical cracks originated from mismatch stress are dependent on the several factors, such as dierential drying, dierential densi®cation, and dierent thermal expansion coecients during cooling. During the drying Fig. 1. X-ray diraction patterns for (a) the sample hot-pressed at 1300oC, 15 MPa for 15 min in air and (b) mullite precursor powder. 2464 S.-Y. Park et al. / Journal of the European Ceramic Society 20 (2000) 2463±2468
S.Y. Park et al. Journal of the European Ceramic Society 20(2000)2463-2468 period, rapid drying rate or high content of binder can In order to investigate crack formations in the lami be contributed to residual stress in the laminate. In this nates, dilatometer measurements of each layer were study, however, such a differential stress due to drying carried out up to 1300C for 15 min with a constant can be eliminated by the careful control of the drying heating rate (10 C/min). Dilatometer measurements ite. When the layers have different densification rates showed that mullite exhibited a high volume shrinkage d thermal expansion coefficients, thermal mismatch (66%)compared to monoclinic and cubic ZrO2(50 and stresses are generated during the densification process or 56%, respectively). Thus, the crack formation in the cooling process. Hillman et al. reported that two classes mullite layer is expected rather than in the Zro2 layer of cracks were observed in Al2O3/ZrO2 laminate com- during the densification of laminates. However, the posites. Cracks with a large opening displacement are crack formation in the mullite layer was observed only originated from drying and subsequent densification in the mM specimen, as shown in Fig. 2. Therefore, it is period, whereas cracks with a small opening displacement believed that the different densification rate is not the originated from thermal expansion mismatch during the main reason for the crack formations in the laminates cooling period. However, it is somewhat difficult to con- during densification clude which factor is more attributable to the crack forma- Therefore, we explored other possibilities resulting in tion either in the densification period or the cooling period. crack formations, such as thermal expansion mismatch during cooling. Thermal expansion coefficients(a)were measured individually on the samples that were pre- sintered at 1300C. These experiments demonstrate that mullite shows the lowest a(5x10-6/C) at 1300.C, m-zro2 while the a values of the tetragonal and cubic ZrO2 are 9×10-6C,10.5×10-6/°C, respectively, as shown in Fig. 3. However, the thermal expansion behavior of monoclinic ZrO2 showing a rapid shrinkage at 1150C was quite different compared to other zirconia. This different thermal behavior in monoclinic ZrO2 is closely related to the phase transformation of ZrO2, i.e. mono tetra transformation (volume contraction) at 1150.C during heating, and tetra-mono transformation(volume expansion) at 930C during cooling, as shown in Fig 4 Mullite Because the thermal expansion coefficient of ZrO2 is higher than that of mullite, it is believed that Zro2 lay ers contain tensile stress and form channel cracks during t-Zro cooling from the hot pressing temperature, as schema tically shown in Fig. 5. The higher density of channel 12.0 t-Zro2 3Y7 0.2mm c-7rO Mullite Mullite c-7rO 2 020040060080010001200 Fig. 3. Thermal expansion coefficient of specimens: mullite, mono- b)tetra-ZrO, and (c)cubiczrO2 ZrO,. tetra-ZrO, and cubic-ZrO
period, rapid drying rate or high content of binder can be contributed to residual stress in the laminate. In this study, however, such a dierential stress due to drying can be eliminated by the careful control of the drying rate. When the layers have dierent densi®cation rates and thermal expansion coecients, thermal mismatch stresses are generated during the densi®cation process or cooling process. Hillman et al.12 reported that two classes of cracks were observed in Al2O3/ZrO2 laminate composites. Cracks with a large opening displacement are originated from drying and subsequent densi®cation period, whereas cracks with a small opening displacement originated from thermal expansion mismatch during the cooling period. However, it is somewhat dicult to conclude which factor is more attributable to the crack formation either in the densi®cation period or the cooling period. In order to investigate crack formations in the laminates, dilatometer measurements of each layer were carried out up to 1300C for 15 min with a constant heating rate (10C/min). Dilatometer measurements showed that mullite exhibited a high volume shrinkage (66%) compared to monoclinic and cubic ZrO2 (50 and 56%, respectively). Thus, the crack formation in the mullite layer is expected rather than in the ZrO2 layer during the densi®cation of laminates. However, the crack formation in the mullite layer was observed only in the mM specimen, as shown in Fig. 2. Therefore, it is believed that the dierent densi®cation rate is not the main reason for the crack formations in the laminates during densi®cation. Therefore, we explored other possibilities resulting in crack formations, such as thermal expansion mismatch during cooling. Thermal expansion coecients ( ) were measured individually on the samples that were presintered at 1300C. These experiments demonstrate that mullite shows the lowest (510ÿ6 / C) at 1300C, while the values of the tetragonal and cubic ZrO2 are 910ÿ6 / C, 10.510ÿ6 / C, respectively, as shown in Fig. 3. However, the thermal expansion behavior of monoclinic ZrO2 showing a rapid shrinkage at 1150C was quite dierent compared to other zirconia. This dierent thermal behavior in monoclinic ZrO2 is closely related to the phase transformation of ZrO2, i.e. monotetra transformation (volume contraction) at 1150C during heating, and tetra-mono transformation (volume expansion) at 930C during cooling, as shown in Fig. 4. Because the thermal expansion coecient of ZrO2 is higher than that of mullite, it is believed that ZrO2 layers contain tensile stress and form channel cracks during cooling from the hot pressing temperature, as schematically shown in Fig. 5. The higher density of channel Fig. 3. Thermal expansion coecient of specimens: mullite, monoZrO2, tetra-ZrO2, and cubic-ZrO2. Fig. 2. SEM micrographs of laminates composites: (a) mono-ZrO2, (b) tetra-ZrO2, and (c) cubic-ZrO2. S.-Y. Park et al. / Journal of the European Ceramic Society 20 (2000) 2463±2468 2465
S.Y. Park et al. Journal of the European Ceramic Society 20(2000)2463-2468 crac In specimen, In companise to those in he mM specimen are constrained from the tM specimen, is due to the larger thermal mismatch expansion at 933C, producing residual compressive stress. Obviously, residual mismatch stresses in the mM stress in Zro2 layers and tensile stress in mullite layers specimen are closely related to mono-tetra phase trans- The polished and the fractured surfaces of the laminate formation of ZrO?. Due to the volume increase during composites were investigated by SEM and as expected tetragonal to monoclinic phase transformation of no reaction zone was observed at the interface between unstabilized ZrO2 during cooling, the monoclinic ZrO2 ZrO2 and mullite. Only a sharp interface between ZrO and mullite layers was observed in the whole samples Indentation was performed into the mullite layer only. This appears straightforward since in our model ZrO, (Mono) system, mullite layers to the matrix of ceramic matrix Mullite (b) ZrO,(Tetra) lOoum taro T IF Mullite Fig. 4. Schematic demonstration of cracks in mullite/ZrO2 laminat mullite/mono-ZrO2,(b) mullite/tetra-ZrO2, and bic-ZrO,(or, Tensile stress: oc, compressive stress; IF. Zro 02004006008001000120014 Fig. 6. SEM micrographs of crack propagation at the mullite/Zro2 laminate composites:(a)mullite/mono-ZrO2, (b)mullite/ Fig. 5. Dimension change in mono-ZrO2 during thermal cycle tetra-ZrO2, and(c)mullite/cubic-ZrO2
cracks in the cM specimen, in comparison to those in the tM specimen, is due to the larger thermal mismatch stress. Obviously, residual mismatch stresses in the mM specimen are closely related to mono-tetra phase transformation of ZrO2. Due to the volume increase during tetragonal to monoclinic phase transformation of unstabilized ZrO2 during cooling, the monoclinic ZrO2 layers in the mM specimen are constrained from expansion at 933C, producing residual compressive stress in ZrO2 layers and tensile stress in mullite layers. The polished and the fractured surfaces of the laminate composites were investigated by SEM and as expected no reaction zone was observed at the interface between ZrO2 and mullite. Only a sharp interface between ZrO2 and mullite layers was observed in the whole samples. Indentation was performed into the mullite layer only. This appears straightforward since in our model system, mullite layers to the matrix of ceramic matrix Fig. 4. Schematic demonstration of cracks in mullite/ZrO2 laminate composites: (a) mullite/mono-ZrO2, (b) mullite/tetra-ZrO2, and (c) mullite/cubic-ZrO2. (T, Tensile stress; C, compressive stress; IF, interface.) Fig. 5. Dimension change in mono-ZrO2 during thermal cycle. Fig. 6. SEM micrographs of crack propagation at the interface of mullite/ZrO2 laminate composites: (a) mullite/mono-ZrO2, (b) mullite/ tetra-ZrO2, and (c) mullite/cubic-ZrO2. 2466 S.-Y. Park et al. / Journal of the European Ceramic Society 20 (2000) 2463±2468
S.Y. Park et al. Journal of the European Ceramic Society 20(2000)2463-2468 2467 composites whereas ZrO, refers to the coatin layer in as shown in Fig. 6(b). XRD analysis of the Zro2 layer the mullite fiber/mullite matrix system. Cracks, coming in the tM laminate composite shows mainly tetragonal from the mullite layer, in the mM system are related to and a small content of monoclinic phase at the hot- crack bifurcation in ZrO, layers [Fig. 6(a). In this case, pressing condition. It might be expected that the tetra the cracks are deflected along the edge crack path in gonal Zro, located on the crack processing zone would ZrOz layer. Considering the stress distribution in the be transformed into monoclinic ZrO2 during the crack laminates, it can be expected that the indentation cracks propagation. In the case of the cM sample, however at the mullite layer will propagate straight through the cracks propagate through the mullite layer into the mullite layer, since the mullite layer locates in biaxial ZrO2 layer without any deflection behavior, as shown in tensile stress. However, the cracks which enter to the Fig. 6(c). It means that the interface of mullite/cubic ZrO, layer from the mullite layer propagate differently. ZrO, was not effective for crack deflection Recently Lange et al. 3 observed the crack bifurcation in Thermal shock resistance of mM, tM, and cM speci Al2O3: laminate composites when cracks entered mens were investigated by cyclic heating and cooling into thin Al2O3 layers sandwiched between Zr(12Ce)O2 between room temperature and 1300oC. During the layers. Because the Al2O3 layer is subjected to a com- cyclic test, 10 cycles were applied and specimens pressive stress due to thermal contraction upon cooling, annealed I h at 1300C. After cyclic test, tM, CM spec cracks entering the Al,O3 layers showed a deflection mens were remained without any changes, while mM behavior along the interface without passing through specimen was catastrophically degraded due to the the matrix. This kind of crack propagation observed in repeated tetra- mono phase transformation including a ZrO, layers of the mM laminate composites, due to the large volume expansion of mono-ZrO, layer, as shown formation during cooling. d compressive stress, developed from tetra- mono trans- in Fig. 7 Interestingly, crack propagation was different in the tM laminate composites. The cracks were deflected and 4. Conclusions arrested at the interface between mullite and zro, layers, This study has demonstrated that thermal mismatch stresses are closely related to the formation of cracks and the crack deflection behavior in the mullite/zirconia (a) Mullite/mono-zrO2 laminate composites. Two kinds of cracks were observed in zirconia layers; one is the channel cracks in layers subjected to tensile stress, another is the edge cracks in layers subjected to compressive stress. Channel cracks were formed during cooling due to high thermal expansion coefficients of zirconia, and edge crack was 1 cm formed due to the tetra- mono transformation of zirco nia during cooling. Among three forms of zirconia, partially stabilized type(tetragonal phase)was effective (b) Mullite/tetra-ZrO in deflecting the cracks at the interface due to the stress induced tetra-mono phase transformation. According to the Al2O3-SiOr-ZrOz equilibrium phase diagram mul- lite-ZrO2 are compatibles in solid state, therefore from a thermodynamic point of view no reaction is possible at the interface. The results obtained in this study sug- gest that partially stabilized zirconia coating on mullite fiber may provide damage tolerant for mullite/mullite tes (c) Mullite/cubic-Zro2 Acknowledgements This work was supported by the Korean Science Engineering Foundation(KOsEF) and the Alexander von humbolt foundation the authors would like to Fig. 7. Mullite/ZrO2 laminate composites after thermal cycles: (a) thank Professor W.A. Kaysser for useful discussion, B mullite/mono-ZrO2,,(b) mullite/ tetra-Zro2, and (c) mullite/cubic. Kanka for technical support, and J H. Song for the preparation of the manuscript
composites whereas ZrO2 refers to the coatin layer in the mullite ®ber/mullite matrix system. Cracks, coming from the mullite layer, in the mM system are related to crack bifurcation in ZrO2 layers [Fig. 6(a)]. In this case, the cracks are de¯ected along the edge crack path in ZrO2 layer. Considering the stress distribution in the laminates, it can be expected that the indentation cracks at the mullite layer will propagate straight through the mullite layer, since the mullite layer locates in biaxial tensile stress. However, the cracks which enter to the ZrO2 layer from the mullite layer propagate dierently. Recently Lange et al.13 observed the crack bifurcation in Al2O3/ZrO2 laminate composites when cracks entered into thin Al2O3 layers sandwiched between Zr(12Ce)O2 layers. Because the Al2O3 layer is subjected to a compressive stress due to thermal contraction upon cooling, cracks entering the Al2O3 layers showed a de¯ection behavior along the interface without passing through the matrix. This kind of crack propagation observed in ZrO2 layers of the mM laminate composites, due to the compressive stress, developed from tetra-mono transformation during cooling. Interestingly, crack propagation was dierent in the tM laminate composites. The cracks were de¯ected and arrested at the interface between mullite and ZrO2 layers, as shown in Fig. 6(b). XRD analysis of the ZrO2 layer in the tM laminate composite shows mainly tetragonal and a small content of monoclinic phase at the hotpressing condition. It might be expected that the tetragonal ZrO2 located on the crack processing zone would be transformed into monoclinic ZrO2 during the crack propagation. In the case of the cM sample, however, cracks propagate through the mullite layer into the ZrO2 layer without any de¯ection behavior, as shown in Fig. 6 (c). It means that the interface of mullite/cubic ZrO2 was not eective for crack de¯ection. Thermal shock resistance of mM, tM, and cM specimens were investigated by cyclic heating and cooling between room temperature and 1300C. During the cyclic test, 10 cycles were applied and specimens annealed 1 h at 1300C. After cyclic test, tM, cM specimens were remained without any changes, while mM specimen was catastrophically degraded due to the repeated tetra-mono phase transformation including a large volume expansion of mono-ZrO2 layer, as shown in Fig. 7. 4. Conclusions This study has demonstrated that thermal mismatch stresses are closely related to the formation of cracks and the crack de¯ection behavior in the mullite/zirconia laminate composites. Two kinds of cracks were observed in zirconia layers; one is the channel cracks in layers subjected to tensile stress, another is the edge cracks in layers subjected to compressive stress. Channel cracks were formed during cooling due to high thermal expansion coecients of zirconia, and edge crack was formed due to the tetra-mono transformation of zirconia during cooling. Among three forms of zirconia, partially stabilized type (tetragonal phase) was eective in de¯ecting the cracks at the interface due to the stress induced tetra-mono phase transformation. According to the Al2O3±SiO2±ZrO2 equilibrium phase diagram mullite±ZrO2 are compatibles in solid state, therefore from a thermodynamic point of view no reaction is possible at the interface. The results obtained in this study suggest that partially stabilized zirconia coating on mullite ®ber may provide damage tolerant for mullite/mullite composites. Acknowledgements This work was supported by the Korean Science Engineering Foundation (KOSEF) and the Alexander von Humbolt Foundation. The authors would like to thank Professor W.A. Kaysser for useful discussion, B. Kanka for technical support, and J.H. Song for the preparation of the manuscript. Fig. 7. Mullite/ZrO2 laminate composites after thermal cycles: (a) mullite/mono-ZrO2, (b) mullite/tetra-ZrO2, and (c) mullite/cubicZrO2. S.-Y. Park et al. / Journal of the European Ceramic Society 20 (2000) 2463±2468 2467
468 S.Y. Park et al. Journal of the European Ceramic Society 20(2000)2463-2468 References 8. Cai. P.Z. Green. D.J. and Messing. G. M.. Constrained densi- fication of alumina/zirconia hybrid laminates, I: experimental 1. Morgan, P. E D. and Marshall, D. B. Functional interfaces for observations of processing defects. J. Am. Ceram. Soc., 1997, Kide/oxide composites. Mater. Sci. Eng. 4, 1993, 162, 15-25 808),19291939 2. Kuo, D. H. and Kriven, W. M, Chemical stability, micro- 9. Russo, C. J. Harmer. M. P, Chan. H. M. and Miller. G.A. structure and mechanical behavior of LaPOa-containing cera Design of a laminated ceramic composite for improve strength mics. Mater. Sci. Eng, 1996. A120. 123-134 and toughness. J. Am. Ceram Soc., 1992. 75(12). 3396-340 3. Morgan. P. E. D, Marshall. D. B. and Hously, R. M, High 10. Melendo, M. J. Clauss. C, Rodriguea, A. D. Herencia, A temperature stability of monazite-alumina composites. Mater and Moya, J. S, Microstructure and high-temperature mechan- Sci.Eng.,1995,A195,215-222 ical behavior of alumina/alumina-ytria-stabilized tetragonal zir- 4. Plucknett. K.P. Caceres C.H. Huhges C. and wilkinson. D. s. conia multilayer composites. J. Am. Ceram. Soc., 1997, 80(8) Processing of tape-cast laminates prepared from fine alumina/zir- conia powder. J. Am. Ceram. Soc., 1993, 7(8), 2145-2153 I1. Moya, M.S. and Osendi, M. I, Microstructure and mechanical 5. Requena, J, Moreno, R and Moya, J.S., Alumina and alumina properties of mullitezirconia composites. J. Mater. Sci, 1984. zirconia multilayer composites obtained by slip casting. J. Am 19(8),2909-2914 Ceran.Soc.1997,78(7),15ll-1513 12. Hillman, C. Suo, Z. and Lange. F. F, Cracking of laminates 6. Wang, H and Hu, X, Surface properties of ceramic laminates fab- subjected to biaxial tensile stresses. J. Am. Ceram. Soc., 1996, ricated by die pressing. J. Am. Ceram. Soc., 1996, 79(2),553-556 79(8),2127-2133 7. Sarkar, P, Huang, X. and Nicholson, P.s., Structural ceramic 13. Oechsner. M.. Suo. Z. and Lange. F. F. Crack bifurcation in microlaminates by electrophoretic deposition. J. Am. Ceram. laminar ceramic composites. J. A. Ceram. Soc., 1834, 79(7) Soc.1992,75(10),2907-2909 1838
References 1. Morgan, P. E. D. and Marshall, D. B., Functional interfaces for oxide/oxide composites. Mater. Sci. Eng. A, 1993, 162, 15±25. 2. Kuo, D. H. and Kriven, W. M., Chemical stability, microstructure and mechanical behavior of LaPO4-containing ceramics. Mater. Sci. Eng., 1996, A120, 123±134. 3. Morgan, P. E. D., Marshall, D. B. and Hously, R. M., High temperature stability of monazite-alumina composites. Mater. Sci. Eng., 1995, A195, 215±222. 4. Plucknett, K. P., CaceÂres, C. H., Huhges, C. and Wilkinson, D. S., Processing of tape-cast laminates prepared from ®ne alumina/zirconia powder. J. Am. Ceram. Soc., 1993, 7(8), 2145±2153. 5. Requena, J., Moreno, R. and Moya, J. S., Alumina and alumina/ zirconia multilayer composites obtained by slip casting. J. Am. Ceram. Soc., 1997, 78(7), 1511±1513. 6. Wang, H. and Hu, X., Surface properties of ceramic laminates fabricated by die pressing. J. Am. Ceram. Soc., 1996, 79(2), 553±556. 7. Sarkar, P., Huang, X. and Nicholson, P. S., Structural ceramic microlaminates by electrophoretic deposition. J. Am. Ceram. Soc., 1992, 75(10), 2907±2909. 8. Cai, P. Z., Green, D. J. and Messing, G. M., Constrained densi- ®cation of alumina/zirconia hybrid laminates, I: experimental observations of processing defects. J. Am. Ceram. Soc., 1997, 80(8), 1929±1939. 9. Russo, C. J., Harmer, M. P., Chan, H. M. and Miller, G. A., Design of a laminated ceramic composite for improve strength and toughness. J. Am. Ceram. Soc., 1992, 75(12), 3396±3400. 10. Melendo, M. J., Clauss, C., Rodriguea, A. D., Herencia, A. J. and Moya, J. S., Microstructure and high-temperature mechanical behavior of alumina/alumina-ytria-stabilized tetragonal zirconia multilayer composites. J. Am. Ceram. Soc., 1997, 80(8), 2126±2130. 11. Moya, M. S. and Osendi, M. I., Microstructure and mechanical properties of mullite-zirconia composites. J. Mater. Sci., 1984, 19(8), 2909±2914. 12. Hillman, C., Suo, Z. and Lange, F. F., Cracking of laminates subjected to biaxial tensile stresses. J. Am. Ceram. Soc., 1996, 79(8), 2127±2133. 13. Oechsner, M., Suo, Z. and Lange, F. F., Crack bifurcation in laminar ceramic composites. J. Am. Ceram. Soc., 1834, 79(7), 1838. 2468 S.-Y. Park et al. / Journal of the European Ceramic Society 20 (2000) 2463±2468