《复合材料 Composites》课程教学资源（学习资料）第五章 陶瓷基复合材料_方石英作界面相

J.Am. Ceram.So,88间61521-1528(2005 DO:10.l111551-2916.2005.00303.x urna Toughening of Mullite/Cordierite Laminated Composites by Transformation Weakening of B-Cristobalite Interphases Waltraud M. Kriven*,**, f and Sang-Jin Lee* f Department of Materials Science and Engineering. University of Illinois at Urbana-Champaign, Urbana, Illinois 61801 An interesting concept for achieving graceful failure in oxide panied by a volume increase of (+)3.0% at 950C or(+)4.9% composites is discussed. It is based on crack deflection in a weak temperature. It is postulated that"transformation interphase between a matrix and reinforcement (e.g. fiber), weakening "of ceramic interphases can lead to overall toughen- around a fibrous core in a fibrous monolith, or in an interphase ing of a ceramic matrix composite. The interphases can separate in a laminated composite. The interphase can be phase trans- laminates, fibrous monolithic cores, or be placed at fiber/matrix formation weakened by a crystallographic unit cell, volume con- interfaces In thermally induced transformations, all interphases traction, and or shape change. Mullite/cordierite laminates with re pre-transformed before the approach of a crack, with some B-a-cristobalite, transformation-weakened interphase were consequent loss of overall strength of the material. In the ideal, investigated for interphase debonding behavior. The laminates shear stress-induced case, an oncoming crack induces a trans- were fabricated by stacking alternate tape-cast, green sheets of formation in its immediate environment, with strength only chemically doped p-cristobalite, which was synthesized by an minimally reduced throughout the bulk. Maximum toughening organic steric entrapment method, and a mullite/cordierite ma- is achieved, since the propagating crack needs to do work in trix mixture. The laminate showed fracture behavior dependir order to overcome the nucleation barrier and cause transforma na critical particle size effect. The grain size of polycrystalline tion, and onset of the other synergistic toughening mechanisms B-cristobalite was controlled by annealing time at 1300C.a(e.g. crack formation)occurs. t-pressed laminated composite, annealed for 10 h at 1300oC. The terminology of"transformation weakening "was first had an average grain size of -4 um and a 3-point flexure troduced by Kriven to describe the deleterious effect of strength of 131 MPa. Its work of fracture was 2.4 k 5.5% volume contraction in enstatite(Mgo- SiO] or MgSio non-catastrophic fracture behavior was demonstrated as it transforms from orthorhombic protoenstatite to monocline dentation response indicated crack deflection along the cristobal- ic clinoenstatite at 865C on quenching. Elastic tensor calcula lite debonding inte With nnealing time, the tions of the enstatite transformation strain 19,20 indicated an trength decreased due to the formation of internal macrocrack anisotro lume contraction, with a maximum 16.5% in the cristobalite layer, which occurred spontaneously during in the [c] axis of the product phase with respect to the parent thermally induced transformation crystal lattice. This gives rise to intragranular microcracks ori- ented perpendicular to the [c] monoclinic axis." Intragranular due to thermally induced transformation have been observed-in clinoenstatite grains which were grown HE brittleness and unreliability of ceramics remain difficult beyond a critical particle size of 7 un and unsolved problems. Attempts to impart"graceful fail a distinction can be drawn between thermally versus stress- ure"analogous to ductility in metals have been partially suc- induced transformations. Displacive transformations can be as- cessful with the use of composites. These are ceramic matrices sociated with a critical particle size in dense ceramics reinforced with fibers, particulates, platelets, or whisker-shaped tudies in zirconia transformation toughening, it is known that second phases. It is now well established that toughening results exceeding their critical particle size transform from debonding at the interface between matrix and reinforce usly on cooling through their transformation temperature. Op- ment phase. 2-7This is a crucial step leading to crack energy timally aged grains can be metastably retained down to room issipation, such as frictional sliding at the interface, while temperature, but can be induced to transform through the ac- transferring load-bearing forces on to fibers.-fo-i7using crack an interphase transformation-weakened composite, the differ- deflection mechanisms to operate in laminates The aim of this work was to introduce a new mechanism of ence between a shear- versus thermally-induced mechanism is nterfacial debonding in oxide ceramics, which has a small mis that in the latter, almost all of the interphase coating grains have atch of thermal expansion coefficient. It is postulated that an Iready transformed at room temperature, while in the former overall increase in toughness can be achieved by inducing a most of the grains are ideally at their optimum critical particle by a large, but negative volume change on cooling, or significant the critical resolved shear stress needed for transformation, 8- unit cell shape change. The proposed mechanism is schemat- The shear stress-induced transformation may be an effective ically illustrated in Fig. 1, and is based on the analogy with toughening mechanism in fully dense bodies. In the as-fabricat transformation toughening in zirconia(Zro,) which is accom- ed state, the composite has maximum bulk strength, and spe- cifically, transverse strength in directions perpendicular to fiber D. Marshalk-contributing editor lengths. Should a matrix crack approach the fiber, however, it induces transformation weakening in the interphase, but only in the immediate area of the crack. rather than in all of the inter uscript No 11138 Received June 25, 2004; approved December 14, 2004 mally induced transformation orted by the U.s. Air Force Office of Scientific Research, under Grant number In this paper, we demonstrate the feasibility of transforma on weakening as a viable debonding mechanism in ceramic matrix composites. A model system was chosen based on the res: Depa kent of Materials Science and Engincering Mokpo National cubic(B)to tetragonal(a) transformation in cristobalite(SiO2) The lattice correspondence operating in the cubic-tetragonal

Toughening of Mullite/Cordierite Laminated Composites by Transformation Weakening of b-Cristobalite Interphases Waltraud M. Kriven* , **,w and Sang-Jin Lee* ,z Department of Materials Science and Engineering, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801 An interesting concept for achieving graceful failure in oxide composites is discussed. It is based on crack deflection in a weak interphase between a matrix and reinforcement (e.g. fiber), around a fibrous core in a fibrous monolith, or in an interphase in a laminated composite. The interphase can be phase trans￾formation weakened by a crystallographic unit cell, volume con￾traction, and/or shape change. Mullite/cordierite laminates with a b-a-cristobalite, transformation-weakened interphase were investigated for interphase debonding behavior. The laminates were fabricated by stacking alternate, tape-cast, green sheets of chemically doped b-cristobalite, which was synthesized by an organic steric entrapment method, and a mullite/cordierite ma￾trix mixture. The laminate showed fracture behavior depending on a critical particle size effect. The grain size of polycrystalline b-cristobalite was controlled by annealing time at 13001C. A hot-pressed laminated composite, annealed for 10 h at 13001C, had an average grain size of B4 lm and a 3-point flexure strength of 131 MPa. Its work of fracture was 2.4 kJ/m2 but non-catastrophic fracture behavior was demonstrated. The in￾dentation response indicated crack deflection along the cristoba￾lite debonding interphase. With increasing annealing time, the strength decreased due to the formation of internal macrocracks in the cristobalite layer, which occurred spontaneously during thermally induced transformation. I. Introduction THE brittleness and unreliability of ceramics remain difficult and unsolved problems. Attempts to impart ‘‘graceful fail￾ure’’ analogous to ductility in metals have been partially suc￾cessful with the use of composites. These are ceramic matrices reinforced with fibers, particulates, platelets, or whisker-shaped second phases.1 It is now well established that toughening results from debonding at the interface between matrix and reinforce￾ment phase.2–7 This is a crucial step leading to crack energy dissipation, such as frictional sliding at the interface, while transferring load-bearing forces on to fibers,7–13 or causing crack deflection mechanisms to operate in laminates.14–17 The aim of this work was to introduce a new mechanism of interfacial debonding in oxide ceramics, which has a small mis￾match of thermal expansion coefficient. It is postulated that an overall increase in toughness can be achieved by inducing a phase transformation in an oxide coating which is accompanied by a large, but negative volume change on cooling, or significant unit cell shape change.18 The proposed mechanism is schemat￾ically illustrated in Fig. 1, and is based on the analogy with transformation toughening in zirconia (ZrO2) which is accom￾panied by a volume increase of (1) 3.0% at 9501C or (1) 4.9% at room temperature. It is postulated that ‘‘transformation weakening’’ of ceramic interphases can lead to overall toughen￾ing of a ceramic matrix composite. The interphases can separate laminates, fibrous monolithic cores, or be placed at fiber/matrix interfaces. In thermally induced transformations, all interphases are pre-transformed before the approach of a crack, with some consequent loss of overall strength of the material. In the ideal, shear stress-induced case, an oncoming crack induces a trans￾formation in its immediate environment, with strength only minimally reduced throughout the bulk. Maximum toughening is achieved, since the propagating crack needs to do work in order to overcome the nucleation barrier and cause transforma￾tion, and onset of the other synergistic toughening mechanisms (e.g. crack formation) occurs. The terminology of ‘‘transformation weakening’’ was first in￾troduced by Kriven19 to describe the deleterious effect of the
5.5% volume contraction in enstatite (MgO SiO2 or MgSiO3) as it transforms from orthorhombic protoenstatite to monoclin￾ic clinoenstatite at 8651C on quenching. Elastic tensor calcula￾tions of the enstatite transformation strain19,20 indicated an anisotropic volume contraction, with a maximum of
16.5% in the [c] axis of the product phase with respect to the parent crystal lattice. This gives rise to intragranular microcracks ori￾ented perpendicular to the [c] monoclinic axis.20 Intragranular microcracks due to thermally induced transformation have been previously observed21 in clinoenstatite grains which were grown beyond a critical particle size of 7 mm.22 A distinction can be drawn between thermally versus stress￾induced transformations. Displacive transformations can be as￾sociated with a critical particle size in dense ceramics.22–25 From studies in zirconia transformation toughening, it is known that grains exceeding their critical particle size transform spontane￾ously on cooling through their transformation temperature. Op￾timally aged grains can be metastably retained down to room temperature, but can be induced to transform through the ac￾tion of applied shear or tensile stresses.23,26–28 In the context of an interphase transformation-weakened composite, the differ￾ence between a shear- versus thermally-induced mechanism is that in the latter, almost all of the interphase coating grains have already transformed at room temperature, while in the former, most of the grains are ideally at their optimum critical particle size, ready to be stress-induced by a propagating crack or by the critical resolved shear stress needed for transformation.28–33 The shear stress-induced transformation may be an effective toughening mechanism in fully dense bodies. In the as-fabricat￾ed state, the composite has maximum bulk strength, and spe￾cifically, transverse strength in directions perpendicular to fiber lengths. Should a matrix crack approach the fiber, however, it induces transformation weakening in the interphase, but only in the immediate area of the crack, rather than in all of the inter￾phases throughout the bulk of the material, as occurs in ther￾mally induced transformation. In this paper, we demonstrate the feasibility of transforma￾tion weakening as a viable debonding mechanism in ceramic matrix composites. A model system was chosen based on the cubic (b) to tetragonal (a) transformation in cristobalite (SiO2). The lattice correspondence operating in the cubic-tetragonal 1521 Journal J. Am. Ceram. Soc., 88 [6] 1521–1528 (2005) DOI: 10.1111/j.1551-2916.2005.00303.x D. Marshall—contributing editor Supported by the U.S. Air Force Office of Scientific Research, under Grant number AFOSR-F49620-93-1-0227. *Member, American Ceramic Society. **Fellow, The American Ceramic Society. w Author to whom correspondence should be addressed. e-mail: kriven@uiuc.edu z Present address: Department of Materials Science and Engineering, Mokpo National University, Muan 534-729, Korea. Manuscript No. 11138. Received June 25, 2004; approved December 14, 2004.

1522 Journal of the American Ceramic Society-Kriven and Lee Vol. 88. No. 6 the mixture was heated. The Pva solution wa MATRIX m 5 wt% PVa (degree of polymerization-1700, Air Products and CORE, Chemicals Inc, Allentown, PA)dissolved in water. The prop LAMINATE tions of the Pva to cation sources in the solution were adjusted TRANSFORMABLE n such a way that there were four times more positively charged OXIDE INTERPHASE valences from the cations than the negatively charged functiona ends of the organics. As the viscosity increased by evapo- ation of water, the mixture was vigorously stirred. The remain CRACK ing water was then dried, converting the gel into a solid. Finall the precursor was finely ground and calcined at 750C for I h The calcined powder was ball milled with zirconia media for 2 h. Iso-propyl alcohol was used as a solvent for milling. To Transformed grains observe the critical grain size, the ball-milled powder was uni- axially pressed at 20 MPa followed by iso-static pressing at 1 MPa for 10 min. The pellet-shaped green compacts were hot ressed by loading them in a graphite die and surrounding them with compatible oxides. Hot pressing was carried out at 28 MPa under an argon atmosphere, at a temperature of 1200C for I h After hot pressing, the samples were annealed in air at 1300C fiber reinforced. f tic diagram illustrating the mechanism of "transforma- for various times. Some stress-induced transformation was of ceramic interphases leading to overall toughening of fibrous monolithic or laminated ceramic matrix com- achieved by hand grinding the hot-pressed and annealed spec- nens on a # In this study, two powders with different particle sizes were prepared in order to observe the effect of initial powder particle size on phase stability. The first, transformation has been measured from single crystal studies. 4 designated as powder A, was calcined at 1100 C for 3 h and The fully expanded, high temperature B structure undergoes a furnace cooled. The second, designated as powder B, was as- reversible, displacive transformation to a collapsed a structure calcined powder, followed by I h of attrition milling Powders a n cooling at 265.C. This is accompanied by a volume decrease and B were annealed at 1300%C for 10 h of approximately 3. 2%. The temperature of the ae p in- Amorphous-type cordierite powder was also prepared by the version in cristobalite is variable and depends on the chemically ame method. The cordierite powder was synthesized from doped crystal structure of the starting material. In order Mg(NO3)2 6H2o(reagent grade, Aldrich Chemical Co. ) Al(- tabilize the B-cristobalite down to room temperature, it can be NO3) 9H,o(reagent grade, Aldrich Chemical Co. ) and Lu- chemically doped with"stuffing"cations.. In particular, in dox As-40 colloidal silica(40 wt% suspension in water, Du Pont the Cao-AlOxSio, system, the molar ratio of calcium oxide Chemicals). Commercial mullite powder(KM Mullite-101 Kyo. to alumina is one in which aluminum occupies a silicon tetra titsu, Nagoya, Japan), which had an average particle size of hedral site, while the calcium ion occupies all the interstitial non- 0.3 um and a specific surface area of 26 m/g, was used in the framework sites. The presence of foreign ions in the inter- mullite/cordierite mixture tices presumably inhibits the contraction of the structure during The mullite/cordierite mixtures having different cordierite thea→β cristobalite transformation content were characterized for relevant properties such as ther In this study, mixed mullite/cordierite laminates with B+a- mal expansion coefficient and flexural strength of hot-pressed cristobalite transformation weakened interphases were used for investigation of the phase transformation and fracture behavior. Laminates were fabricated by the tape casting process, ac- In the absence of commercially available, pure mullite fibers: a consisted of 25 vol% oxide powders, -63 vol% solvent, and the match of thermal expansion coefficients of mullite and 12 vol% organics. The 0.5 wt%(dry weight basis of oxide cristobalite, as well as to improve the sinterability of mullite, der) polyvinyl butyral(PVB, Monsanto, St. Louis, MO)was a mullite/cordierite mixture was chosen as the matrix phase. a added to the slurries as a dispersant. The solvent was composed tape casting technique was used to engineer a series of laminated of a mixture of trichloroethylene(CICH= CCl,, Aldrich Chem- composites in various sequences of stacked and hot pressed ical Co. )and ethanol (CH CH,OH, Aldrich Chemical Co tapes. The grain size of the polycrystalline B-cristobalite was while the organics included a binder(Pvb, monsanto)and plas- ontrolled by varying the annealing time at 1300C, and exam- ticizers(polyethylene glycol(PEG)2000 and dioctyl phthalate ination by scanning electron microscopy (SEM). In addition, the (DP), Aldrich Chemical Co. ) After pulverization, dispersion effects of the laminate design and heating conditions on the and mixing by ball-milling two times, the slurries were stirred in trength and toughness of the laminated composites were stud- vacuum. This helped in removing bubbles and adjusting the ied, both qualitatively by optical microscopy and indentation working viscosity. After aging for 2 days, the slurries were tape techniques, as well as quantitatively by measuring flexure cast using a doctor blade opening of 150-300 um, to obtain strengths and the works of fracture tape cast green sheets of 60-150 um thickness. Drying of the cast tapes was carried out under a saturated solvent atmosphere for 1 II. Experimental Procedure The green laminated composites had area dimensions of 25 mm x 51 mm after stacking of green sheets. Thermocompression (1) Preparation of Powder and laminate was performed at 10 MPa load for 30 min at 80.C, which was Chemically stabilized, amorphous-type silica powder was pre- the softening point of the organics. The organic additives were pared by the organic steric entrapment technique empl polyvinyl alcohol(PVA)solution as a polymeric carrier i-g removed by heating to 550C in an air atmosphere, using a two- step heating process. After these additives were burned out, the a clear sol was prepared from Ludox As-40 colloidal silica green laminates were iso-statically cold pressed at 170 MPa for 40 wt% suspension in water, Du Pont Chemicals, Wilmington, 10 min The laminated green bodies were hot pressed in the same DE): Al(NO3)3.9H20(reagent grade, Aldrich Chemical Co way as mentioned earlier. Milwaukee, Wn) and Ca(NO3)2. 4H2o(reagent grade, Aldrich All laminated composites after densification had a 30-layer Chemical Co. )in proportions to form a final composition of repetitive sequence of matrix and interphase. These were made Cao. 2Al2O3.80Sio2. After dissolving these reagents inin a separately optimized 5: 1 thickness ratio of matrix to inter DI water, the organic carrier, PVA solution, was added and phase, by stacking each green sheet. After hot pressing, the lam-

transformation has been measured from single crystal studies.34 The fully expanded, high temperature b structure undergoes a reversible, displacive transformation to a collapsed a structure on cooling at 2651C. This is accompanied by a volume decrease of approximately 3.2%.35,36 The temperature of the a3b in￾version in cristobalite is variable and depends on the chemically doped crystal structure of the starting material.36,37 In order to stabilize the b-cristobalite down to room temperature, it can be chemically doped with ‘‘stuffing’’ cations.38,39 In particular, in the CaO–Al2O3–SiO2 system, the molar ratio of calcium oxide to alumina is one in which aluminum occupies a silicon tetra￾hedral site, while the calcium ion occupies all the interstitial non￾framework sites.40–42 The presence of foreign ions in the inter￾stices presumably inhibits the contraction of the structure during the a3b cristobalite transformation. In this study, mixed mullite/cordierite laminates with b-a￾cristobalite transformation weakened interphases were used for investigation of the phase transformation and fracture behavior. In the absence of commercially available, pure mullite fibers, a model laminate configuration was chosen. In order to optimize the match of thermal expansion coefficients of mullite and cristobalite, as well as to improve the sinterability of mullite, a mullite/cordierite mixture was chosen as the matrix phase. A tape casting technique was used to engineer a series of laminated composites in various sequences of stacked and hot pressed tapes. The grain size of the polycrystalline b-cristobalite was controlled by varying the annealing time at 13001C, and exam￾ination by scanning electron microscopy (SEM). In addition, the effects of the laminate design and heating conditions on the strength and toughness of the laminated composites were stud￾ied, both qualitatively by optical microscopy and indentation techniques, as well as quantitatively by measuring flexure strengths and the works of fracture. II. Experimental Procedure (1) Preparation of Powder and Laminate Chemically stabilized, amorphous-type silica powder was pre￾pared by the organic steric entrapment technique employing polyvinyl alcohol (PVA) solution as a polymeric carrier.43–45 A clear sol was prepared from Ludox AS-40 colloidal silica (40 wt% suspension in water, Du Pont Chemicals, Wilmington, DE); Al(NO3)3 9H2O (reagent grade, Aldrich Chemical Co., Milwaukee, WI) and Ca(NO3)2.4H2O (reagent grade, Aldrich Chemical Co.) in proportions to form a final composition of CaO. 2Al2O3 . 80SiO2. 42,46 After dissolving these reagents in DI water, the organic carrier, PVA solution, was added and the mixture was heated. The PVA solution was prepared from 5 wt% PVA (degree of polymerization-1700, Air Products and Chemicals Inc., Allentown, PA) dissolved in water. The propor￾tions of the PVA to cation sources in the solution were adjusted in such a way that there were four times more positively charged valences from the cations than the negatively charged functional ends of the organics.43–45 As the viscosity increased by evapo￾ration of water, the mixture was vigorously stirred. The remain￾ing water was then dried, converting the gel into a solid. Finally, the precursor was finely ground and calcined at 7501C for 1 h. The calcined powder was ball milled with zirconia media for 12 h. Iso-propyl alcohol was used as a solvent for milling. To observe the critical grain size, the ball-milled powder was uni￾axially pressed at 20 MPa followed by iso-static pressing at 170 MPa for 10 min. The pellet-shaped green compacts were hot pressed by loading them in a graphite die and surrounding them with compatible oxides. Hot pressing was carried out at 28 MPa under an argon atmosphere, at a temperature of 12001C for 1 h. After hot pressing, the samples were annealed in air at 13001C for various times. Some stress-induced transformation was achieved by hand grinding the hot-pressed and annealed spec￾imens on a #800 mesh SiC paper.27,47 In this study, two powders with different particle sizes were prepared in order to observe the effect of initial powder particle size on phase stability. The first, designated as powder A, was calcined at 11001C for 3 h and furnace cooled. The second, designated as powder B, was as￾calcined powder, followed by 1 h of attrition milling. Powders A and B were annealed at 13001C for 10 h. Amorphous-type cordierite powder was also prepared by the same method.48 The cordierite powder was synthesized from Mg(NO3)2 6H2O (reagent grade, Aldrich Chemical Co.); Al(- NO3)3 9H2O (reagent grade, Aldrich Chemical Co.), and Lu￾dox AS-40 colloidal silica (40 wt% suspension in water, Du Pont Chemicals). Commercial mullite powder (KM Mullite-101 Kyo￾titsu, Nagoya, Japan), which had an average particle size of 0.3 mm and a specific surface area of 26 m2 /g, was used in the mullite/cordierite mixture. The mullite/cordierite mixtures having different cordierite content were characterized for relevant properties such as ther￾mal expansion coefficient and flexural strength of hot-pressed samples. Laminates were fabricated by the tape casting process, ac￾cording to the procedure summarized in Fig. 2. The slurries consisted of B25 vol% oxide powders, B63 vol% solvent, and 12 vol% organics. The 0.5 wt% (dry weight basis of oxide pow￾der) polyvinyl butyral (PVB, Monsanto, St. Louis, MO) was added to the slurries as a dispersant. The solvent was composed of a mixture of trichloroethylene (ClCH 5 CCl2, Aldrich Chem￾ical Co.) and ethanol (CH3CH2OH, Aldrich Chemical Co.) while the organics included a binder (PVB, Monsanto) and plas￾ticizers (polyethylene glycol (PEG) 2000 and dioctyl phthalate (DP), Aldrich Chemical Co.). After pulverization, dispersion and mixing by ball-milling two times, the slurries were stirred in vacuum. This helped in removing bubbles and adjusting the working viscosity. After aging for 2 days, the slurries were tape cast using a doctor blade opening of 150–300 mm, to obtain tape cast green sheets of 60–150 mm thickness. Drying of the cast tapes was carried out under a saturated solvent atmosphere for 1 day. The green laminated composites had area dimensions of 25 mm 51 mm after stacking of green sheets. Thermocompression was performed at 10 MPa load for 30 min at 801C, which was the softening point of the organics. The organic additives were removed by heating to 5501C in an air atmosphere, using a two￾step heating process. After these additives were burned out, the green laminates were iso-statically cold pressed at 170 MPa for 10 min. The laminated green bodies were hot pressed in the same way as mentioned earlier. All laminated composites after densification had a 30-layer repetitive sequence of matrix and interphase. These were made in a separately optimized 5:1 thickness ratio of matrix to inter￾phase, by stacking each green sheet. After hot pressing, the lam￾Fig. 1. Schematic diagram illustrating the mechanism of ‘‘transforma￾tion weakening’’ of ceramic interphases leading to overall toughening of fiber reinforced, fibrous monolithic or laminated ceramic matrix com￾posites. 1522 Journal of the American Ceramic Society—Kriven and Lee Vol. 88, No. 6

June 2005 Transformation Weakening of Interphas Dispers CuKo radiation(40 kv, 40 mA). All XRD data was obtained solvent Solvent: ethanol and at room temperature after furnace cooling. The relative volume ratios of a- and B-cristobalite phases were determined by inte- ball milling for 24h grating the X-ray peak areas of (102)of a-cristobalite and (222) of p-cristobalite by the equation Add plasticizers+ binder Plasticizer: PEG/DP =[(102)2/(22l+1(102)×100 ballmilling for 15h which Va was volume fraction of a-cristobalite, 1(102) and /(222)B were peak intensities of (102) and(222)B respectively about 10000 to 15000 cP (F Flexural Testing: Four-point flexural testing was ageing for 48h erformed using a 10 mm inner span and a 20 mm outer pan, at a crosshead speed of 0.01 mm/min on a universal test- Tape casting and drying Casting rate: 1 cm/s ing machine(model 4502, Instron Corp, Canton, MA). A min- Drying conditio imum number of five bars were tested for each composition. The saturated atmosphere apparent work of fracture was obtained by dividing the area at room temp. under the load versus displacement curve by the cross-sectional of the sample atting, stacking and lamination Lamination condition (G) Indentation Test. A Vickers hardness test was ca 80°Cfr30 min under ar ried out with a micro-hardness tester 32 12. Mark V lab uniaxial compression Inc, East Granby, CT) under a 6 kg tation load in order to study crack propagation profiles and interaction with the Burnet RT400°c(cmin)2 a hold (H) Microstructure Characterization: The microstruc- 400-550°C(0.5°C/min)3 h hold ure of sintered cristobalite and the crack propagation behavior in the laminates after bend testing were observed by optical Cold isostatic pressing CIP condition roscopy(Nikon SMZ-2T, Tokyo, Japan)and sEM(Model 170 MPa for 10 min DS-130. International Scientific Instruments, Santa Clara, CA) To observe grain size and cracking, polished and then annealed ples were chemically etched in boiling phosphoric acid 1200° Cfor lh at28MPa for 30s Fig. 2. Flow chart of tape casting procedure for fabrication of lami- nated composites. IlL. Results and discussion (1) Powder Analysis and Development of Crystalline Phases the longitudinal axis of the specimens in the plane of the an agglomerated particle size range of 10-60 um. After ball- lamination. The bend-bars with dimensions of 30 mm milling, the powder had a narrower particle size distribution, x 3.0 mm x 2.5-3.0 mm were polished to a 15 um finish with forming small particles of approximately 0. 1-0. 4 um in size. The diamond paste. The specimens were then annealed at 1300"C for as-calcined powder had a BET specific surface area of 25 m/g The high specific surface area meant that the calcined powder was very porous. The ball-milled powder had a specific surface area of 75 m/g. The average particle size of the amorphous-type (2) characterization cordierite powder, after calcination at 750C for I h and ball- (A) Specific Surface Area Measurement: The specific milling for 12 h, was about 0.3 um, with a specific surface area of surface area of the calcined powders and ball-milled powders 80 m /g were compared by nitrogen gas absorption(Model ASAP 2400, Room temperature XRD spectra following the developmen Micromeritics, Norcross. GA) of crystalline phases of the calcined, amorphous-type silica pow (B) Thermal Expansion Coefficient Measurement: The der after various heating temperatures are shown in Fig 3.An variation of thermal expansion coefficient for polycrystalline amorphous phase was observed at 1000C. Above 1100.C, the cristobalite and mullite/cordierite mixtures were examined using B-cristobalite crystalline phase was detected and B-cristobalite a dilatometer(Netzsch Dilatometer, 402E, Selb, Germany) peaks were developed almost completely at 1200C. With heating up to 1100C at heating rate of 5C/min In the mul increasing temperature the amount of a-cristobalite phase lite/cordierite mixture, to obtain a relative density of above 95% increased gradually, while the amount of B-cristobalite de- in each specimen, the specimens had to be sintered at differer creased. At 1450.C, B-cristobalite remained as a minor phase temperatures. in the a-cristobalite matrix. The a form at high temperature was transformation resulting from the large using distilled water as a displacement liquid. The relative den- lite indicated that the aep transformation occurred at 180C y of each specimen was calculated from the theo density on heating and at 170C on cooling. The transformation tem- of mullite(3. 18 g/cm), cordierite(2.52 g/cm),an nd cristo perature was lower than that of pure cristobalite because of the (a phase: 2.33 g/cmand p phase: 2.26 g/cm) dopant effects. As the graphs show, the thermal expansion D) Average Grain Size Measurement: The average grain Defficient of B-cristobalite was approximately 1.5 x 10/C ize of sintered and annealed cristobalite was analyzed according and tended to decrease on heating. a change in thermal expan- to the Jeffries-Saltykov method sion coefficient was observed at the ae B transformation tem- (E) X-Ray Diffraction(XRD) Analysis: The develop- perature. It is known that the transformation of cristobalite nt of crystallinity in calcined, amorphous-type silica powder results in a large increase in the thermal expansion coefficient. 2 nd the phase change between a-and B-cristobalite hot-pressed This change was much less for the polycrystalline cristobalite es were studied as a function of he sintered at 1350 C than for the one at 1450C. The difference in annealing time using a Rigaku spectrometer(DMax automated the change of thermal expansion coefficient was attributed to the powder diffractor u/U.SA, Danvers, MA) with a-cristobalite content. In the case of the polycrystalline cristobal

inates were cut into bend-bars. The cutting direction was along the longitudinal axis of the specimens in the plane of the lamination. The bend-bars with dimensions of 30 mm 3.0 mm 2.5–3.0 mm were polished to a 15 mm finish with diamond paste. The specimens were then annealed at 13001C for different times. (2) Characterization (A) Specific Surface Area Measurement: The specific surface area of the calcined powders and ball-milled powders were compared by nitrogen gas absorption (Model ASAP 2400, Micromeritics, Norcross, GA). (B) Thermal Expansion Coefficient Measurement: The variation of thermal expansion coefficient for polycrystalline cristobalite and mullite/cordierite mixtures were examined using a dilatometer (Netzsch Dilatometer, 402E, Selb, Germany), heating up to 11001C at heating rate of 51C/min. In the mul￾lite/cordierite mixture, to obtain a relative density of above 95% in each specimen, the specimens had to be sintered at different temperatures. (C) Relative Density Measurement: The density of the sintered specimens was estimated by the Archimedes method using distilled water as a displacement liquid. The relative den￾sity of each specimen was calculated from the theoretical density of mullite (3.18 g/cm3 ), cordierite (2.52 g/cm3 ), and cristobalite (a phase: 2.33 g/cm3 and b phase: 2.26 g/cm3 ). (D) Average Grain Size Measurement: The average grain size of sintered and annealed cristobalite was analyzed according to the Jeffries–Saltykov method.49 (E) X-Ray Diffraction (XRD) Analysis: The develop￾ment of crystallinity in calcined, amorphous-type silica powder and the phase change between a- and b-cristobalite hot-pressed samples were studied as a function of heating temperature and annealing time using a Rigaku spectrometer (DMax automated powder diffractometer, Rigaku/U.S.A., Danvers, MA) with CuKa radiation (40 kV, 40 mA). All XRD data was obtained at room temperature after furnace cooling. The relative volume ratios of a- and b-cristobalite phases were determined by inte￾grating the X-ray peak areas of (102) of a-cristobalite and (222) of b-cristobalite by the equation:24 Va ¼ ½Ið102Þa=Ið222Þb þ Ið102Þa 100 in which Va was volume fraction of a-cristobalite, I(102)a and I(222)b were peak intensities of (102)a and (222)b respectively. (F) Flexural Testing: Four-point flexural testing was performed using a 10 mm inner span and a 20 mm outer span, at a crosshead speed of 0.01 mm/min on a universal test￾ing machine (model 4502, Instron Corp., Canton, MA). A min￾imum number of five bars were tested for each composition. The apparent work of fracture was obtained by dividing the area under the load versus displacement curve by the cross-sectional area of the sample.50 (G) Indentation Test: A Vickers hardness test was car￾ried out with a micro-hardness tester (Zwick 3212, Mark V Lab. Inc., East Granby, CT) under a 6 kg indentation load, in order to study crack propagation profiles and interaction with the microstructure. (H) Microstructure Characterization: The microstruc￾ture of sintered cristobalite and the crack propagation behavior in the laminates after bend testing were observed by optical mi￾croscopy (Nikon SMZ-2T, Tokyo, Japan) and SEM (Model DS-130, International Scientific Instruments, Santa Clara, CA). To observe grain size and cracking, polished and then annealed samples were chemically etched in boiling phosphoric acid for 30 s. III. Results and Discussion (1) Powder Analysis and Development of Crystalline Phases The calcined, amorphous-type silica powder was very soft, with an agglomerated particle size range of 10–60 mm. After ball￾milling, the powder had a narrower particle size distribution, forming small particles of approximately 0.1–0.4 mm in size. The as-calcined powder had a BET specific surface area of 25 m2 /g. The high specific surface area meant that the calcined powder was very porous. The ball-milled powder had a specific surface area of 75 m2 /g. The average particle size of the amorphous-type cordierite powder, after calcination at 7501C for 1 h and ball￾milling for 12 h, was about 0.3 mm, with a specific surface area of 80 m2 /g. Room temperature XRD spectra following the development of crystalline phases of the calcined, amorphous-type silica pow￾der after various heating temperatures are shown in Fig. 3. An amorphous phase was observed at 10001C. Above 11001C, the b-cristobalite crystalline phase was detected and b-cristobalite peaks were developed almost completely at 12001C. With increasing temperature the amount of a-cristobalite phase increased gradually, while the amount of b-cristobalite de￾creased. At 14501C, b-cristobalite remained as a minor phase in the a-cristobalite matrix. The a form at high temperature was due to the spontaneous transformation resulting from the large particle size.22,51 Dilatometry curves for the stabilized cristoba￾lite indicated that the a3b transformation occurred at 1801C on heating and at 1701C on cooling. The transformation tem￾perature was lower than that of pure cristobalite because of the dopant effects.36 As the graphs show, the thermal expansion coefficient of b-cristobalite was approximately 1.5 10
6 /1C and tended to decrease on heating. A change in thermal expan￾sion coefficient was observed at the a3b transformation tem￾perature. It is known that the transformation of cristobalite results in a large increase in the thermal expansion coefficient.52 This change was much less for the polycrystalline cristobalite sintered at 13501C than for the one at 14501C. The difference in the change of thermal expansion coefficient was attributed to the a-cristobalite content. In the case of the polycrystalline cristoba￾Fig. 2. Flow chart of tape casting procedure for fabrication of lami￾nated composites. June 2005 Transformation Weakening of Interphases 1523

1524 Journal of the American Ceramic Society-Kriven and Lee Vol. 88. No 6 β: B-cristobalt a: a-cristobalite p(11)(101) P(220)a1(200) P(222) 1450°C 1350c 500m Fig. 5. Scanning electron photograph of fragile cristobalite sample hot- 200°C essed at 1200C and annealed at 1300C for 50 h The sample was not polished and etched. 1100°C lite with the(101)and (200) peaks from a-cristobalite, which are the high intensity peaks of the cristobalite system, because the 000°C peaks were located at almost the same 20 value(Fig. 3). There- fore, the relative amounts of a- and B-cristobalite phase was compared from the relative intensity of (102) and(222)B peaks. In general, the relative volume ratios of a- and B-cristobalit 45 50 phases and the average grain size increased with increasing an- 20(CuKa) lealing time. The increase continued up to an annealing time of 50 h. The rate of grain growth decreased gradually from 12 h ically stabilized, amorphous, silica powder after heating at various tem- annealing time of 50 h resulted in fragile samples with extensive peratures. The holding time at each temperature was I h. cracks. This was because the thermally-induced transformation occurred spontaneously, primarily due to the critical particle size effect. The sample annealed for 50 h showed almost 80 vol% lite sintered at 1350C, a-cristobalite was present in the B- a-cristobalite. Figure 5 is an SEM micrograph of the fragile cristobalite matrix, whereas, for the polycrystalline cristobalite cristobalite sample which was hot-pressed and annealed for sintered at 1450.C, the a-cristobalite was the matrix phase 50 h. Extensive "macrocracks were observed over the whole (Fig. 3). The amorphous-type, cordierite powder crystallized sample surface. to a-cordierite at 1250C The polished and etched surface micrographs of cristobalite annealed at 1300oC for various times are shown in Fig. 6. Some macrocracks were detected in the microstructure having an (2) Critical Grain Size average grain size of 5 um, at annealing times of 12 h. The hot- The formation of a-cristobalite on cooling was affected by va ressed cristobalite annealed for 30 h consisted of about 72%a- to a phase occurred Spontaneous transformation from B phase cristobalite as shown in Fig. 4. It also exhibited macrocracks in cristobalite grain size stabilized the p phase. The grain size of the hemically doped B-cristobalite was controlled by annealing the influence of stress. Shear stress-induced阝→∝- cristobalite Figure 4 shows the plot of volume fraction of a-cristobalite conversion for annealed and ground specimens at various an- erage grain sizes for hot-pressed cristobalite samples as a nealing times are compared with the unground specimens. It is tion of annealing time at 1300C. It was impossible to com recognized that grinding involves a complex stress state. Fur- pare the intensity of the(111)and (220) peaks from B-cristoba thermore, displacive transformations can be induced by shear deformation that can be induced by grinding. This has been widely demonstrated in ceramics such density of as hot-pressed cristobalite specim The mineral rhombohedral calcite(CaCO3), for ex- (98% relative density) ample, can be cyclically transformed to orthorhombic aragonite by repeated grinding in a planetary mill. Figure 7 displays the esult for hand-ground cristobalite. In view of the Gaussian dis- tribution of grain sizes around the critical particle size for shear- induced transformation the two effects of shear-induced and thermally-induced transformation are superimposed. However the tendency exists for an increase in stress-induced a-phase wit increasing annealing time up to maximum. Specifically, a max oRatio of a/B-cristobalite phase imum increase of about 17 vol% a-cristobalite was calculated in the polycrystalline sample annealed at 1300C for 10 h. This in- dicates that the optimum range of critical grain size for shear Annealing Time at 1300C(h) stress-induced transformation was approximately 4-5 um. The increase in the amount of a-cristobalite in the ground specimens Ime fraction of a-cristobalite and over that in the only annealed specimens, decreased with in pressed alite samples as a function g time at I300°C. creasing annealing time above 10 h. This may be attributed to

lite sintered at 13501C, a-cristobalite was present in the b￾cristobalite matrix, whereas, for the polycrystalline cristobalite sintered at 14501C, the a-cristobalite was the matrix phase (Fig. 3). The amorphous-type, cordierite powder crystallized to a-cordierite at 12501C.48 (2) Critical Grain Size The formation of a-cristobalite on cooling was affected by var￾ying the grain size.24 Spontaneous transformation from b phase to a phase occurred at larger grain sizes. In contrast, a small b￾cristobalite grain size stabilized the b phase. The grain size of the chemically doped b-cristobalite was controlled by annealing time. Figure 4 shows the plot of volume fraction of a-cristobalite and average grain sizes for hot-pressed cristobalite samples as a function of annealing time at 13001C. It was impossible to com￾pare the intensity of the (111) and (220) peaks from b-cristoba￾lite with the (101) and (200) peaks from a-cristobalite, which are the high intensity peaks of the cristobalite system, because the peaks were located at almost the same 2y value (Fig. 3). There￾fore, the relative amounts of a- and b-cristobalite phase was compared from the relative intensity of (102)a and (222)b peaks. In general, the relative volume ratios of a- and b-cristobalite phases and the average grain size increased with increasing an￾nealing time. The increase continued up to an annealing time of 50 h. The rate of grain growth decreased gradually from 12 h annealing time onwards. Large grain sizes of about 8.5 mm at an annealing time of 50 h resulted in fragile samples with extensive cracks. This was because the thermally-induced transformation occurred spontaneously, primarily due to the critical particle size effect.24,51 The sample annealed for 50 h showed almost 80 vol% a-cristobalite. Figure 5 is an SEM micrograph of the fragile cristobalite sample which was hot-pressed and annealed for 50 h. Extensive ‘‘macrocracks’’ were observed over the whole sample surface. The polished and etched surface micrographs of cristobalite annealed at 13001C for various times are shown in Fig. 6. Some macrocracks were detected in the microstructure having an average grain size of 5 mm, at annealing times of 12 h. The hot￾pressed cristobalite annealed for 30 h consisted of about 72% a￾cristobalite as shown in Fig. 4. It also exhibited macrocracks in the microstructure having an average grain size of 7 mm. The transformation of b to a cristobalite was susceptible to the influence of stress. Shear stress-induced b-a-cristobalite conversion for annealed and ground specimens at various an￾nealing times are compared with the unground specimens. It is recognized that grinding involves a complex stress state. Fur￾thermore, displacive transformations can be induced by shear deformation that can be induced by grinding. This has been widely demonstrated in ceramics such as zirconia and in ensta￾tite.26,27,53 The mineral rhombohedral calcite (CaCO3), for ex￾ample, can be cyclically transformed to orthorhombic aragonite by repeated grinding in a planetary mill.54 Figure 7 displays the result for hand-ground cristobalite. In view of the Gaussian dis￾tribution of grain sizes around the critical particle size for shear￾induced transformation, the two effects of shear-induced and thermally-induced transformation are superimposed. However, the tendency exists for an increase in stress-induced a-phase with increasing annealing time up to maximum. Specifically, a max￾imum increase of about 17 vol% a-cristobalite was calculated in the polycrystalline sample annealed at 13001C for 10 h. This in￾dicates that the optimum range of critical grain size for shear stress-induced transformation was approximately 4–5 mm. The increase in the amount of a-cristobalite in the ground specimens over that in the only annealed specimens, decreased with in￾creasing annealing time above 10 h. This may be attributed to Fig. 4. Volume fraction of a-cristobalite and average grain sizes of hot￾pressed cristobalite samples as a function of annealing time at 13001C. Fig. 3. Room temperature X-ray diffraction spectra of calcined, chem￾ically stabilized, amorphous, silica powder after heating at various tem￾peratures. The holding time at each temperature was 1 h. Fig. 5. Scanning electron photograph of fragile cristobalite sample hot￾pressed at 12001C and annealed at 13001C for 50 h. The sample was not polished and etched. 1524 Journal of the American Ceramic Society—Kriven and Lee Vol. 88, No. 6

June 2005 Transformation Weakening of Interphas (a) c 5.0 (d) 5.0um Fig. 6. Scanning electron micrographs of the polished and etched surfaces for hot pressed cristobalite according to different annealing times of (a)Oh, b)10 h, (c)12 h, and(d)30 h(where arrows indicate"macrocracks") the fact that the amount of B-cristobalite, which can be trans- calcined powder (powder A)was attrition milled for I h. The formed by shear stress, was decreased by the spontaneous, ther- attrition-milled powder (powder B), in which the particle size mally-induced transformation which occurred in the oversized was reduced to less than I um, was also annealed at the sam rains temperature and same times as for the as-calcined powder A The critical grain size in the chemically doped B-cristobalite The as-calcined powder consisted of agglomerates composed of can be confirmed by the effect of initial powder size. The as- tiny crystallites in the B-f The volume fraction of d-cristoba- lite was measured for the two annealed powders. Powder A showed about 40 vol a-cristobalite. In contrast, powder B showed less than 10 vol a-cristobalite. The results of the effect of powder size, therefore, were consistent with the grain size trition milling the agglomerates were milled into tiny crystallites At higher annealing temperatures, sintering and grain growth ould be favored in agglomerates, but would be difficult in in- dividually separated crystallites. The volume fraction of a- 二 stobalite was measured for the two annealed powders. Pov der a showed about 40 vol% a-cristobalite. In contrast, powder B showed less than 10 vol% a-cristobalite. The results of the consistent with the grain ● Before grinding O After grinding (3) Composition of laminate Table I shows the variation of thermal expansion coefficient and flexural strength for the mullite/cordierite mixtures as a function Annealing Time at 1300C (h) ficient and flexural strength decreased as cordierite content was Fig. 7. Grinding effect on the relative volume ratio of l- and B- creased. To match the thermal expansion coefficient to the cristobalite phases for hot-pressed nemically doped B-cristobalite (1. 5 x 10/C), the mullite/cor Ing time at1300°C dierite layer should also have a low thermal expansion coeffi

the fact that the amount of b-cristobalite, which can be trans￾formed by shear stress, was decreased by the spontaneous, ther￾mally-induced transformation which occurred in the oversized grains. The critical grain size in the chemically doped b-cristobalite can be confirmed by the effect of initial powder size. The as￾calcined powder (powder A) was attrition milled for 1 h. The attrition-milled powder (powder B), in which the particle size was reduced to less than 1 mm, was also annealed at the same temperature and same times as for the as-calcined powder A. The as-calcined powder consisted of agglomerates composed of tiny crystallites in the b-form. The volume fraction of a-cristoba￾lite was measured for the two annealed powders. Powder A showed about 40 vol % a-cristobalite. In contrast, powder B showed less than 10 vol % a-cristobalite. The results of the effect of powder size, therefore, were consistent with the grain size effect controlling thermally-induced transformation. During at￾trition milling the agglomerates were milled into tiny crystallites. At higher annealing temperatures, sintering and grain growth would be favored in agglomerates, but would be difficult in in￾dividually separated crystallites. The volume fraction of a￾cristobalite was measured for the two annealed powders. Pow￾der A showed about 40 vol% a-cristobalite. In contrast, powder B showed less than 10 vol% a-cristobalite. The results of the effect of powder size therefore, were consistent with the grain size effect controlling thermally-induced transformation. (3) Composition of Laminates Table I shows the variation of thermal expansion coefficient and flexural strength for the mullite/cordierite mixtures as a function of cordierite content. As expected, the thermal expansion coef- ficient and flexural strength decreased as cordierite content was increased. To match the thermal expansion coefficient to the chemically doped b-cristobalite (1.5 10
6 /1C), the mullite/cor￾dierite layer should also have a low thermal expansion coeffi- Fig. 6. Scanning electron micrographs of the polished and etched surfaces for hot pressed cristobalite according to different annealing times of (a) 0 h, (b) 10 h, (c) 12 h, and (d) 30 h (where arrows indicate ‘‘macrocracks’’). Fig. 7. Grinding effect on the relative volume ratio of a- and b￾cristobalite phases for hot-pressed cristobalite as a function of anneal￾ing time at 13001C. June 2005 Transformation Weakening of Interphases 1525

1526 Journal of the American Ceramic Society-Kriven and Lee Vol. 88. No 6 Table I. Variation of Thermal Expansion Coefficient and Flexural Strength for Mullite/Cordierite Mixture as a Function of Cordierite content Cordierite content(wt%) 100 Thermal expansion coefficient(×10-°C Flexural strength(MPa) 229+12205+30 172+18 4l1+18 116+2 108+23 For each cordierite content, seven to eight samples were tested in flexure, and the results represent the mean and amount of largest scatte Table Il. Bulk Density, Volume Fraction of a-Cristobalite, Average Grain Size, Strength, and Work of Fracture for Hot-Pressed Laminates at Various Annealing Times Annealing time at 1300C 10 Bulk density (g/cm) 2.68 2.67 2.63 2.65 Volume fraction of a-cristobalite (% 22 47 ize(um) l.2 Strength, Omax(MPa) 171+9 124+17 109±4 82+17 Work of fracture(kJ/m) 120 rElative density of each component in the laminated composite after hot pressing(cristobalite: 98%, cordierite: 98%, 60 wt% mullite/40 wt% cordierite mixture: 96%) For each annealing time, four to six samples were tested in flexure, and the results represent the mean and amount of largest scatter. The work-of-fracture was calculated samples showing cient Thermal expansion compatibility is an important factor in is interesting to note that the 12 h annealed sample appears to fabricating a stable, laminated structure with minimal thermal have exceeded the critical particle size for stress-induced trans- stresses at the interface. To design a laminated composite with formation and contained visible, thermally-induced macro- minimum thermal expansion difference between laminates and cracks(Fig. 6(c) yet retain reasonable strength, the 40 wt% cordierite content Figure presents load-deflection curves under un-notched, 4- was therefore selected for the mullite/cordierite matrix layers. point flexural testing, as a function of annealing time at 1300C. For the un-annealed bend bar. the curve showed brittle fracture. For the 10 h annealing case, the curve showed non-catastrophic (4) Fracture Behavior fracture. The step-wise load drops were characteristic of graceful To study the fracture behavior, the hot-pressed laminated com- failure. This implied that the matrix crack was debonding the posites having the same matrix to interphase thickness ratio of interphase, giving a relatively significant increase in work of 5: I were 4-point, flexural tested after various annealing times rain size, strength, and work of fracture for the laminates ob- tained at different annealing times are listed in table il. the bulk density was not changed following annealing at 1300C because the laminates appeared to have achieved their maxi lum density after the hot-pressing process. The strength de- creased when the annealing time was increased. The highest cristobalite ork of fracture of 2.38 kJ/m was observed in the laminate having a strength of 131 MPa at annealing times of 10 h. The 10 h annealed sample appeared to be optimal to maximize the volume of B-phase susceptible to shear stress-induced phase transformation. The average grain sizes in both the 10 and I h annealed samples were similar as expected, being 4.2 and 5.0 um, respectively, as summarized in Table Il. However, compa ison of their corresponding microstructures in Figs. 6(b)and(c), respectively, shows that there was a wide distribution of grain sizes and so only a trend can be observed. In general however, it 0.5m 0.25 (b) 0.15 O: Strength, Omax(MP 0.05 0.0 0.10 15 0.20.25 0.35 Fig 8. Load-deflection curves for laminates of 15 mullite/ cord matrix layers separated by cristobalite interphases under 4-point ile Fig 9. Optical micrographs of crack propagation in laminated samples testing as a function of annealing time at 13000 annealed at 1300C for(a)O h and(b)10 h after 4-point flexural testing

cient. Thermal expansion compatibility is an important factor in fabricating a stable, laminated structure with minimal thermal stresses at the interface. To design a laminated composite with minimum thermal expansion difference between laminates and yet retain reasonable strength, the 40 wt% cordierite content was therefore selected for the mullite/cordierite matrix layers. (4) Fracture Behavior To study the fracture behavior, the hot-pressed laminated com￾posites having the same matrix to interphase thickness ratio of 5:1 were 4-point, flexural tested after various annealing times. The bulk density, volume fraction of a-cristobalite, average grain size, strength, and work of fracture for the laminates ob￾tained at different annealing times are listed in Table II. The bulk density was not changed following annealing at 13001C because the laminates appeared to have achieved their maxi￾mum density after the hot-pressing process. The strength de￾creased when the annealing time was increased. The highest work of fracture of 2.38 kJ/m2 was observed in the laminate having a strength of 131 MPa at annealing times of 10 h. The 10 h annealed sample appeared to be optimal to maximize the volume of b-phase susceptible to shear stress-induced phase transformation. The average grain sizes in both the 10 and 12 h annealed samples were similar as expected, being 4.2 and 5.0 mm, respectively, as summarized in Table II. However, compar￾ison of their corresponding microstructures in Figs. 6(b) and (c), respectively, shows that there was a wide distribution of grain sizes and so only a trend can be observed. In general however, it is interesting to note that the 12 h annealed sample appears to have exceeded the critical particle size for stress-induced trans￾formation and contained visible, thermally-induced macro￾cracks (Fig. 6(c)). Figure 8 presents load-deflection curves under un-notched, 4- point flexural testing, as a function of annealing time at 13001C. For the un-annealed bend bar, the curve showed brittle fracture. For the 10 h annealing case, the curve showed non-catastrophic fracture. The step-wise load drops were characteristic of graceful failure. This implied that the matrix crack was debonding the interphase, giving a relatively significant increase in work of Table I. Variation of Thermal Expansion Coefficient and Flexural Strength for Mullite/Cordierite Mixture as a Function of Cordierite Content Cordierite content (wt%) 0 20 40 60 80 100 Thermal expansion coefficient ( 10
6 /1C) 4.5 3.8 3.1 2.6 2.0 1.4 Flexural strengthw (MPa) 229712 205730 172718 141718 116726 108723 w For each cordierite content, seven to eight samples were tested in flexure, and the results represent the mean and amount of largest scatter. Table II. Bulk Density, Volume Fraction of a-Cristobalite, Average Grain Size, Strength, and Work of Fracture for Hot-Pressed Laminates at Various Annealing Times Annealing time at 13001C 0 10 12 36 Bulk density (g/cm3 ) w 2.68 2.67 2.63 2.65 Volume fraction of a-cristobalite (%) 22 47 52 77 Average grain size (mm) 1.2 4.2 5.0 7.3 Strength,z smax (MPa) 17179 124717 10974 82717 Work of fracture (kJ/m2 ) 1.20 2.38 2.00 1.82 w Relative density of each component in the laminated composite after hot pressing (cristobalite: 98%, cordierite: 98%, 60 wt% mullite/40 wt% cordierite mixture: 96%). z For each annealing time, four to six samples were tested in flexure, and the results represent the mean and amount of largest scatter. The work-of-fracture was calculated from samples showing the maximum values. Fig. 8. Load–deflection curves for laminates of 15 mullite/cordierite matrix layers separated by cristobalite interphases under 4-point flexural testing, as a function of annealing time at 13001C. Fig. 9. Optical micrographs of crack propagation in laminated samples annealed at 13001C for (a) 0 h and (b) 10 h after 4-point flexural testing. 1526 Journal of the American Ceramic Society—Kriven and Lee Vol. 88, No. 6

June 2005 Transformation Weakening of Interphas 1527 (a) Cristobalite 6难编 mcloone/cordierite mulmte/cordierite ∮)21 100ul 100u Fig. 11. Scanning electron micrograph of indentation crack pattern (b) hot-pressed laminate annealed for 10 h at 1300.C. Residual stresses at interfaces due to thermal expansion mis- u natch can have an effect on fracture behavior. In this laminated composite, however, the thermal expansion coefficient of the matrix, which had a 40 wt% cordierite content, was quite close to that of the B-cristobalite interphase. Therefore, the effect of residual stress on fracture behavior was considered to be To examine the interaction between crack propagation and the laminated microstructure. Vickers indentation cracks were introduced and the sEM micrograph is shown in Fig. 11. In- the lar 5护门axxH以比m layer into the adjacent matrix layer. 100um Fig 10. Scanning electron micrographs of hot-pressed laminates ealed at 1300.C for(a) 10 h and(b)36 h A pronounced, Vickers in- An oxide-laminated composite consisting of a mullite/cordierite duced macrocrack was observed to have propagated in the cristobalite rix separated by nterphase in the laminate annealed for 36 h terphase has successfully been engineered. Matrix-crack deflec- tion by shear stress-induced phase transformation was observed at a critical doped cristobalite grain size of 4-5 um, resulting in fracture. The optical micrographs confirming crack deflection a comparatively high work of fracture. The best mechanical ehavior was exhibited by an interphase annealed for 10 h at omparison with the un-annealed laminate, the crack was 1300.C. The macrocracks, which were caused by spontaneous visibly deflected along the interphase, especially in the central thermally-induced phase transformation during the cooling shear region of the cristobalite layer in the composite anneale process, provided an easier propagation path for crack deflec- for 10 h tion without requiring significant crack energy for propagation The thermally-induced cracks in the hot-pressed laminated cristobalite layer before conducting flexural strength tests are and hot pressing were used, but the concept of transformation shown in Fig. 10. In the bulk sample annealed for 10 h, ne wakening and debonding of interphases leading to overall macrocracks were observed as seen in Fig. 6(b). Similarly in the graceful failure has essentially been demonstrated. In continu- laminate annealed for 10 h. no macrocracks were observed ous fiber reinforced CMCs. fibrous monoliths. or laminated (Fig. 10(a)), although some transverse, intergranular micro- composites, significant toughening can be achieved by shear-in- cracking appears to be present in the cristobalite layers. These duced, transformation weakening, causing debonding of inter may be specimen preparation artifacts due to shear-induced ses. As such, this mechanism has the potential for developing ransformation resulting from grinding and polishing of samples fully dense composites, which initially have both high strength for SEM examination and hence release of matrix constraint. In s well as potentially high toughness comparison, however, the 36 h annealed sample, having an av- erage grain size of x7.3 um clearly contained macrocracks propagating within the cristobalite layer(Fig. 10(b)). The 36 h nnealed laminated sample had a lower strength and work of fracture than did the laminate annealed at 10 h as seen in fig. 8 hich is supported by the U.S. Department of Energy under Grant DEF This observation is consistent with the hypothesis that the 10 h ER45439, at the University of Illinois at Urbana-Champaign nnealed sample, being close to the critical particle size for stress-induced transformation, was able to absorb more fracture energy than could the 36 h annealed sample, which was over aged and hence required little or no fracture energy to nucleate the transformation ACEm图2B pment of High Toughness Ceramics

fracture. The optical micrographs confirming crack deflection in the laminated samples are seen in Figs. 9(a) and (b). In comparison with the un-annealed laminate, the crack was visibly deflected along the interphase, especially in the central shear region of the cristobalite layer in the composite annealed for 10 h. The thermally-induced cracks in the hot-pressed laminated cristobalite layer before conducting flexural strength tests are shown in Fig. 10. In the bulk sample annealed for 10 h, no macrocracks were observed as seen in Fig. 6(b). Similarly in the laminate annealed for 10 h, no macrocracks were observed (Fig. 10(a)), although some transverse, intergranular micro￾cracking appears to be present in the cristobalite layers. These may be specimen preparation artifacts due to shear-induced transformation resulting from grinding and polishing of samples for SEM examination, and hence release of matrix constraint. In comparison, however, the 36 h annealed sample, having an av￾erage grain size of B7.3 mm clearly contained macrocracks propagating within the cristobalite layer (Fig. 10(b)). The 36 h annealed laminated sample had a lower strength and work of fracture than did the laminate annealed at 10 h, as seen in Fig. 8. This observation is consistent with the hypothesis that the 10 h annealed sample, being close to the critical particle size for stress-induced transformation, was able to absorb more fracture energy than could the 36 h annealed sample, which was over￾aged and hence required little or no fracture energy to nucleate the transformation. Residual stresses at interfaces due to thermal expansion mis￾match can have an effect on fracture behavior. In this laminated composite, however, the thermal expansion coefficient of the matrix, which had a 40 wt% cordierite content, was quite close to that of the b-cristobalite interphase. Therefore, the effect of residual stress on fracture behavior was considered to be negligible. To examine the interaction between crack propagation and the laminated microstructure, Vickers indentation cracks were introduced and the SEM micrograph is shown in Fig. 11. In￾dent-induced cracks, in the laminated hot-pressed and annealed for 10 h, displayed a propagation path through the mullite/cor￾dierite layer. However, the crack did not cross the cristobalite layer into the adjacent matrix layer. IV. Conclusions An oxide-laminated composite consisting of a mullite/cordierite matrix separated by a transformation-weakened cristobalite in￾terphase has successfully been engineered. Matrix-crack deflec￾tion by shear stress-induced phase transformation was observed at a critical doped cristobalite grain size of 4–5 mm, resulting in a comparatively high work of fracture. The best mechanical behavior was exhibited by an interphase annealed for 10 h at 13001C. The macrocracks, which were caused by spontaneous thermally-induced phase transformation during the cooling process, provided an easier propagation path for crack deflec￾tion without requiring significant crack energy for propagation. Conventional ceramic processing techniques of tape casting and hot pressing were used, but the concept of transformation weakening and debonding of interphases leading to overall graceful failure has essentially been demonstrated. In continu￾ous fiber reinforced CMCs, fibrous monoliths, or laminated composites, significant toughening can be achieved by shear-in￾duced, transformation weakening, causing debonding of inter￾phases. As such, this mechanism has the potential for developing fully dense composites, which initially have both high strength as well as potentially high toughness. Acknowledgment Use is acknowledged of some of the facilities maintained in the Center for Microanalysis of Materials, of the Frederick Seitz Materials Research Laboratory, which is supported by the U.S. Department of Energy under Grant DEFG02-91- ER45439, at the University of Illinois at Urbana-Champaign. References 1 A. G. Evans, ‘‘Perspectives on the Development of High Toughness Ceramics,’’ J. Am. Ceram. Soc., 73 [2] 187–206 (1990). Fig. 10. Scanning electron micrographs of hot-pressed laminates an￾nealed at 13001C for (a) 10 h and (b) 36 h. A pronounced, Vickers in￾duced macrocrack was observed to have propagated in the cristobalite interphase in the laminate annealed for 36 h. Fig. 11. Scanning electron micrograph of indentation crack pattern in hot-pressed laminate annealed for 10 h at 13001C. June 2005 Transformation Weakening of Interphases 1527

1528 Journal of the American Ceramic Society--Kriven and Lee Vol. 88. No 6 -D. B. Marshall b.n. cox and a. Evans.“TheM Vol 12, Science and Technology of Zirconia ll, Edited by N. Claussen, M. Role. racking in Brittle Matrix Fiber Composites, Acta Met. Mater, 33(11]2013-2 and A H. Heuer. American Ceramic Society, Westerville, OH, 1984. Reinfored Ceramics. "/Mech. Phys. Solids, 34(2 167-89 [1g racture in Fiber and Other Non posite Ce- amics, Edited by R. T Tressler. G. L Messing. C. G. Pantano, and E Newnham Dissimilar Elastic Materials, "Int J Solids Struct, 25, 1053-67(1989). 3w. M. Kriven, "The Transformation Mechanism of Spherical Zirconia Par- SA. G. Evans and D. BM Matrix Composites, "Acta Met. Mater. 37[10]2567-831989 ology of Zircon ia ii. Edited by N. claussen, M. Rahle, and A. H. Heuer iber-Matrix Interface in Ceramic Composites, Am. Ceram. SA. H. Heuer and M. ihle."On the Nucleation of the Martensitic Transfor M. Prevo and J. Brennan"H ength Silicon Carbide Fiber-Rein- w.M. Krven, ""Martensitic Toughening of Ceramics, Mater. Sci. E J. Mater.Sci.,15[2]463-8(1980 A127,249-55(1990) erization of glass and Glass- R. L. Withers, J. G. Thompson, and T.R. Welberry. "The Structure and Ceramics, Edited by R. eD. A. peacor, "High-Temperature Single-Crystal Study of the Cristobalite "H. C. Cao E. Bischof o. Sbaizero. M. Ruhle.AG. Evans. D. B. Marshall 8.274-98(1973) and J. J. Brennan, ""Effect of Interfaces on the Properties of Fiber-Reinforced G. Hill and R. Roy, ca structure Studies: v. The Variable Inversion in Soc,4l2532-7(1958) uang.Y.Xu. D. Zhu and w.M. River N. Fenner, "Stability Relations of the Silica Minerals. Am. J. Sci./4th B-SiAION Composites Reinforced with SiC Monofilaments, "J Mater. Sci. Eng M. Huang, D. Zhu, Y. Xu, T Mackin, and w. M. Kriven, ""Interfacial Soc. Bull, 36[[4]142-8(1957) males in Tridymite and Cristobalite, "Am. Ceram. Properties of SiC Monofilament Reinforced B'-SiAION Composites, "J. Mater tives of the Silica Structures. Am. Mineral 39 Sci.Eng.A,201.15968(19 78600-14(1954 1. Huang. D. Zhu, Y. Xu, w.M.Riven, and C. Y. Yuh, "SiCtyo-Si- A. J. Perrotta. D. K. Grubbs. E S. Martin. and N.R. Dando,"Chemical roperties and Oxidation Retained Properties, J. Mater. Sci. Stabilization of B-Cristobalite, "J. Am. Ceram. Soc. 72[31441-7(1989 A,220[1-2]17484(1996 E. S. Thomas, J. G. Thompson, R. L Withers, M. Sterns, Y. Xiao, and D. H Kuo, w.M. Kriven, and T J Mackin, Control of Interfacial Prop- R J. Kirkpatrick, Further Investigation of the Stabilization of P-Cristobalite Am paris D. wei eidner, J. D. Jorgensen, and M. A Saltzberg, ""Pressur " W. J. Clegg. K. Kendall. N. M. Alford, D. Birchall and T. W.Button. Induced Phase Transition and Pressure Dependence of Crystal Structure in Low "A Simple Way to Make Tough Ceramics, "Nature, 347, 455-7(1990) Clegg. "The Fabrication and Failure of Laminar Ceramic Composites. M. A Gulgun and w. M. Kriven, " A Simple Solution-polyn D H. Kuo and w. M. Kriven, "A Strong and Damage Tolerant Oxide Lam- Technology, and Commercialization of Powder Synthesis and Shape Forming Proc- Interfacial Engineering for Optimized Prop- Society. Westerville, OH. 199 cMkn、:埋a1g9 2.“3mM0 ide spow, des a Ao gleans上m间rm ynthesis: Ceramics, Glasses, Composites Il 19w. M. Kriven, "Possible Alternative Transformation Tougheners to Zirconia: 4M. A. Galgan, W. M. Kriven, and M. H. Nguyen, "Processes for Preparing Crystallographic Aspects,J. An Ceram Soc. 71[12] 1021-30(1988). Noy 19th ns and Their Applications in E.S. Thomas, J.G. Thompson, R L Withers, M. Sterns, Y. Xiao, and RJ 0l-100199 on of the Stabilization of p-Cristobalite 2W. E. Lee and A. H. Heuer, "On the Polymorphism of Enstatite, "J. Am. J. Am. Ceram. Soc., 77[1]49-56(1994) -E C. Bloor. ""Conversion in Steatite Ceramics, ""J. Br. Ceram Soc. 2. 309-16 43. S. Reed and A. M. Lejus, "Effect of Grinding and Polishing on Near-Sur- face Phase Transforma (pAH.Heuer,NClaussen,WM.Kriven,and M.Rihle Riven.“ Cryst bulit Amorphous Cordierite Powder P by a PVA Solution-Polymerization Route. "J.Am. Ceram. Soc., 82[8]2049-55(1999) 4R. T. DeHoff and F N. Rhines, Quantitative Microscopy. McGraw-Hill,New M. Kriven, C.J. Chan, and E. A. Barinek "The Particle-Size Elect ol SoH. G. Tattersall and g.ta Ceramics, Vol. 24, Science and Technology of Zirconia /ll, Edited by S. Somiya, N in Metals, Ceramics and Other Materials, "J. Mater. Sci, 1, 296-301(1966). SC. M. Huang. D H. Kuo. Y J. Kim and W. M. Kriven, "Phase Stability of -C J. Chan. w. M. Kriven. andJ. F. Yo Chemically Derived Enstatite(MgSiO3) Powders, J. Am. Ceram. Soc.. 77[101 7 =D. Zhu and w. M. Krimen "hear nduced airman sformation in Enstatite. 32Y. Imanaka, S. Aoki, N. Kamehara, and K. Niwa, "Crystallization of c.Eng.Proc.,17A,383-90(1996) Low Temperature Fired Glass/ Ceramic Composite, Yogyo Kyokai Shi, 95[111 2D. Zhu and w. M. Kriven "Shear Induced Transformation and Plasticity in ental Study on the Polymorphism of Enstatite, "Am. 2M. Ruhle and W.M. Kriven, ""Stress-Induced Transformations in Composite Minera,59.,345-52(l sotH.Heaumrand me. unise Phase: Thns o3malions n zr containing ce. Mucture of cadium. arbon at." I. Cheme. soac. Faraday igns he. 61-6 ramics: Il. The Martensitic Reaction t-Zro2: pp. 14-32 in Advances in Ceramics, 975

2 D. B. Marshall, B. N. Cox, and A. G. Evans, ‘‘The Mechanics of Matrix Cracking in Brittle Matrix Fiber Composites,’’ Acta Met. Mater., 33 [11] 2013–21 (1985). 3 B. Budiansky, J. W. Hutchinson, and A. G. Evans, ‘‘Matrix Fracture in Fiber Reinforced Ceramics,’’ J. Mech. Phys. Solids, 34 [2] 167–89 (1986). 4 M. Y. He and J. W. Hutchinson, ‘‘Crack Deflection at an Interphase Between Dissimilar Elastic Materials,’’ Int. J. Solids Struct., 25, 1053–67 (1989). 5 A. G. Evans and D. B. Marshall, Overview No. 85, ‘‘The Mechanical Behavior of Ceramic Matrix Composites,’’ Acta Met. Mater., 37 [10] 2567–83 (1989). 6 R. J. Kerans, R. S. Hay, N. J. Pagano, and T. A. Parthasarathy, ‘‘The Role of Fiber–Matrix Interface in Ceramic Composites,’’ Am. Ceram. Soc. Bull., 68 [2] 429–42 (1993). 7 K. M. Prevo and J. J. Brennan, ‘‘High-Strength Silicon Carbide Fiber-Rein￾forced Glass-Matrix Composites,’’ J. Mater. Sci., 15 [2] 463–8 (1980). 8 J. J. Brennan, ‘‘Interfacial Characterization of Glass and Glass-Ceramic Matrix/Nicalon SiC Fiber Composites’’; pp. 549–60 in Tailing Multiphase and Composite Ceramics, Edited by R. T. Tressler, G. L. Messing, C. G. Pantano, and R. E. Newnham. Plenum Press, New York, 1986. 9 H. C. Cao, E. Bischoff, O. Sbaizero, M. Ru¨hle, A. G. Evans, D. B. Marshall, and J. J. Brennan, ‘‘Effect of Interfaces on the Properties of Fiber-Reinforced Ceramics,’’ J. Am. Ceram. Soc., 73 [6] 1691–9 (1990). 10C. M. Huang, Y. Xu, D. Zhu, and W. M. Kriven, ‘‘Combustion Synthesized b0 -SiAlON Composites Reinforced with SiC Monofilaments,’’ J. Mater. Sci. Eng. A, 188, 341–51 (1994). 11C. M. Huang, D. Zhu, Y. Xu, T. Mackin, and W. M. Kriven, ‘‘Interfacial Properties of SiC Monofilament Reinforced b0 -SiAlON Composites,’’ J. Mater. Sci. Eng. A, 201, 159–68 (1995). 12C. M. Huang, D. Zhu, Y. Xu, W. M. Kriven, and C. Y. Yuh, ‘‘SiCf/O’-Si￾AlON Composites: Properties and Oxidation Retained Properties,’’ J. Mater. Sci. A, 220 [1–2] 174–84 (1996). 13D. H. Kuo, W. M. Kriven, and T. J. Mackin, ‘‘Control of Interfacial Prop￾erties Through Fiber Coatings: Monazite Coatings in Oxide/Oxide Composites,’’ J. Am. Ceram. Soc., 80 [12] 2987–96 (1997). 14W. J. Clegg, K. Kendall, N. M. Alford, D. Birchall, and T. W. Button, ‘‘A Simple Way to Make Tough Ceramics,’’ Nature, 347, 455–7 (1990). 15W. J. Clegg, ‘‘The Fabrication and Failure of Laminar Ceramic Composites,’’ Acta Metall., 40 [11] 3085–93 (1992). 16D. H. Kuo and W. M. Kriven, ‘‘A Strong and Damage Tolerant Oxide Lam￾inate,’’ J. Am. Ceram. Soc., 80 [9] 2421–4 (1997). 17D. H. Kuo and W. M. Kriven, ‘‘Interfacial Engineering for Optimized Prop￾erties,’’ Mater. Res. Soc. Symp., 458, 477–88 (1997). 18W. M. Kriven and S. J. Lee, ‘‘Toughening of Ceramic Composites by Trans￾formation Weakening of Interphases’’; US Patent No. 6,361,888 B1, issued Mar 26th 2002. 19W. M. Kriven, ‘‘Possible Alternative Transformation Tougheners to Zirconia: Crystallographic Aspects,’’ J. Am. Ceram. Soc., 71 [12] 1021–30 (1988). 20W. M. Kriven, ‘‘Displacive Phase Transformations and Their Applications in Structural Ceramics,’’ J. Phys. IV, Colloque, C8, 101–10 (1995). 21W. E. Lee and A. H. Heuer, ‘‘On the Polymorphism of Enstatite,’’ J. Am. Ceram. Soc., 70 [5] 349–60 (1987). 22E. C. Bloor, ‘‘Conversion in Steatite Ceramics,’’ J. Br. Ceram. Soc., 2, 309–16 (1964). 23A. H. Heuer, N. Claussen, W. M. Kriven, and M. Ru¨hle, ‘‘Stability of Te￾tragonal ZrO2 Particles in Ceramic Matrices,’’ J. Am. Ceram. Soc., 65 [12] 642–50 (1982). 24W. M. Kriven, C. J. Chan, and E. A. Barinek, ‘‘The Particle-Size Effect of Dicalcium Silicate in a Calcium Zirconate Matrix’’; pp. 145–55 in Advances in Ceramics, Vol. 24, Science and Technology of Zirconia III, Edited by S. Somiya, N. Yamamoto, and H. Yanagida. American Ceramic Society, Westerville, OH, 1988. 25C. J. Chan, W. M. Kriven, and J. F. Young, ‘‘Physical Stabilization of the b￾g Transformation in Dicalcium Silicate,’’ J. Am. Ceram. Soc., 75 [6] 1621–7 (1992). 26D. Zhu and W. M. Kriven, ‘‘Shear Induced Transformation in Enstatite,’’ Ceram. Sci. Eng. Proc., 17 A, 383–90 (1996). 27D. Zhu and W. M. Kriven, ‘‘Shear Induced Transformation and Plasticity in Enstatite,’’ J. Am. Ceram. Soc., to be submitted. 28M. Ru¨hle and W. M. Kriven, ‘‘Stress-Induced Transformations in Composite Zirconia Ceramics,’’ Ber. Bunsen. Phys. Chem., 87, 222–8 (1983). 29A. H. Heuer and M. Ru¨hle, ‘‘Phase Transformations in ZrO2-Containing Ce￾ramics: II, The Martensitic Reaction t-ZrO2’’; pp. 14–32 in Advances in Ceramics, Vol. 12, Science and Technology of Zirconia II, Edited by N. Claussen, M. Ru¨hle, and A. H. Heuer. American Ceramic Society, Westerville, OH, 1984. 30W. M. Kriven, ‘‘Displacive Transformation Mechanism in Zirconia Ceramics and Other Non-Metals’’; pp. 223–37 in Tailoring Multiphase and Composite Ce￾ramics, Edited by R. T. Tressler, G. L. Messing, C. G. Pantano, and E. Newnham. Plenum, New York, 1986. 31W. M. Kriven, ‘‘The Transformation Mechanism of Spherical Zirconia Par￾ticles in Alumina’’; pp. 64–77 in Advances in Ceramics, Vol. 12, Science and Tech￾nology of Zirconia II, Edited by N. Claussen, M. Ru¨hle, and A. H. Heuer. American Ceramic Society, Westerville, OH, 1984. 32A. H. Heuer and M. Ru¨hle, ‘‘On the Nucleation of the Martensitic Transfor￾mation in Zirconia,’’ Acta Metall. Mater., 33, 2101–12 (1985). 33W. M. Kriven, ‘‘Martensitic Toughening of Ceramics,’’ Mater. Sci. Eng., A 127, 249–55 (1990). 34R. L. Withers, J. G. Thompson, and T. R. Welberry, ‘‘The Structure and Microstructure of a-Cristobalite and its Relationship to b-Cristobalite,’’ Phys. Chem. Miner., 16, 517–23 (1989). 35D. A. Peacor, ‘‘High-Temperature Single-Crystal Study of the Cristobalite Inversion,’’ Z. Kristallogr., 138, 274–98 (1973). 36V. G. Hill and R. Roy, ‘‘Silica Structure Studies: V, The Variable Inversion in Cristobalite,’’ J. Am. Ceram. Soc., 41 [12] 532–7 (1958). 37C. N. Fenner, ‘‘Stability Relations of the Silica Minerals,’’ Am. J. Sci. [4th Series], 36 [214] 331–84 (1913). 38W. Eitel, ‘‘Structural Anomalies in Tridymite and Cristobalite,’’ Am. Ceram. Soc. Bull., 36 [4] 142–8 (1957). 39M. J. Buerger, ‘‘Stuffed Derivatives of the Silica Structures,’’ Am. Mineral, 39 [7–8] 600–14 (1954). 40A. J. Perrotta, D. K. Grubbs, E. S. Martin, and N. R. Dando, ‘‘Chemical Stabilization of b-Cristobalite,’’ J. Am. Ceram. Soc., 72 [3] 441–7 (1989). 41E. S. Thomas, J. G. Thompson, R. L. Withers, M. Sterns, Y. Xiao, and R. J. Kirkpatrick, ‘‘Further Investigation of the Stabilization of b-Cristobalite,’’ J. Am. Ceram. Soc., 77 [1] 49–56 (1994). 42J. B. Parise, D. J. Weidner, J. D. Jorgensen, and M. A. Saltzberg, ‘‘Pressure￾Induced Phase Transition and Pressure Dependence of Crystal Structure in Low (a) and Ca/Al-Doped Cristobalite,’’ J. Appl. Phys., 75 [3] 1361–6 (1994). 43M. A. Gu¨lgu¨n and W. M. Kriven, ‘‘A Simple Solution-Polymerization Route for Oxide Powder Synthesis’’; pp. 57–66 in Ceramics Transactions, Vol. 62, Science, Technology, and Commercialization of Powder Synthesis and Shape Forming Proc￾esses, Edited by J. J. Kingsley, C. H. Schilling, and J. H. Adair. American Ceramic Society, Westerville, OH, 1996. 44W. M. Kriven, S. J. Lee, M. A. Gu¨lgu¨n, M. H. Nguyen, and D. K. Kim, ‘‘Synthesis of Oxide Powders Via Polymeric Steric Entrapment,’’ (invited review paper) in Innovative Processing/Synthesis: Ceramics, Glasses, Composites III,’’ Ceram. Trans., 108, 99–110 (2000). 45M. A. Gu¨lgu¨n, W. M. Kriven, and M. H. Nguyen, ‘‘Processes for Preparing Mixed Oxide Powders’’; US Patent No. 6,482,387, issued Nov 19th 2002. 46E. S. Thomas, J. G. Thompson, R. L. Withers, M. Sterns, Y. Xiao, and R. J. Kirkpatrick, ‘‘Further Investigation of the Stabilization of b-Cristobalite,’’ J. Am. Ceram. Soc., 77 [1] 49–56 (1994). 47J. S. Reed and A. M. Lejus, ‘‘Effect of Grinding and Polishing on Near-Sur￾face Phase Transformation in Zirconia,’’ J. Mater. Res. Bull., 12, 949–54 (1977). 48S. J. Lee and W. M. Kriven, ‘‘Crystallization and Densification of Nano-Size, Amorphous Cordierite Powder Prepared by a PVA Solution-Polymerization Route,’’ J. Am. Ceram. Soc., 82 [8] 2049–55 (1999). 49R. T. DeHoff and F. N. Rhines, Quantitative Microscopy. McGraw-Hill, New York, 1968. 50H. G. Tattersall and G. Tappin, ‘‘The Work of Fracture and its Measurement in Metals, Ceramics and Other Materials,’’ J. Mater. Sci., 1, 296–301 (1966). 51C. M. Huang, D. H. Kuo, Y. J. Kim, and W. M. Kriven, ‘‘Phase Stability of Chemically Derived Enstatite (MgSiO3) Powders,’’ J. Am. Ceram. Soc., 77 [10] 2625–31 (1994). 52Y. Imanaka, S. Aoki, N. Kamehara, and K. Niwa, ‘‘Crystallization of Low Temperature Fired Glass/Ceramic Composite,’’ Yogyo Kyokai Shi, 95 [11] 1119–21 (1987). 53J. R. Smyth, ‘‘Experimental Study on the Polymorphism of Enstatite,’’ Am. Mineral, 59, 345–52 (1974). 54J. M. Craido and J. M. Trillo, ‘‘Effects of Mechanical Grinding on the Texture and Structure of Calcium Carbonate,’’ J. Chem. Soc. Faraday Trans., 117, 961–6 (1975). & 1528 Journal of the American Ceramic Society—Kriven and Lee Vol. 88, No. 6