《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_Ceramic fibrous monolithic structures

Availableonlineatwww.sciencedirect.com DIRECT② COMPOSITE STRUCTURES ELSEVIER Composite Structures 66(2004)547-553 sevier. com/locate/compstruct Ceramic fibrous monolithic structures K C. Goretta T.A. Cruse, D. Singh, J.L. Routbort a,, A.R. de arellano-Lopez B.L. SI loffe Physico-Technical Institute, Politekhnicheskaya ul. 26, St. Petersburg 194021, Russia Available online 9 June 2004 Abstract High-strength ceramic fibers and composite structures that contain them are generally expensive. In a lower-cost approach for lbricating fibrous composites, reinforcing fiber-like cells that are distinct from a continuous matrix phase called the cell boundary can be formed in situ from powders Structures can be constructed by assembly and consolidation of filaments that consist of the cell phase and its surrounding cell boundary Fabrication of ceramic fibrous monoliths(FMs)is reviewed and mechanical properties of the most widely studied FMs ar discussed. Those based on Si, N4 cells within a bn cell boundary have achieved the best overall properties and uniformity of manufacture, but degrade severely at high temperatures in oxidizing environments. Those based on oxides are more stable, but are substantially weaker. Assessment of the future of FMs is offered, including cost reduction, fabrication practice, property improvement, and formation of complex structures Published by elsevier Ltd Keywords: Ceramics: Composites: Fibrous monoliths 1. Introduction mercially [9). Cermet FMs are now also produced commercially and have exhibited exceptional perfor- Powder-derived fibrous monoliths(FMs) generally mance as, for example, inserts for mining drill bits. In consist of strong cells that are surrounded by a weaker one study of rock drilling, all bits based on conventional cell boundary. The cells are typically a200-500 um wide Co-bonded WC fractured, whereas none of the bit [1-8]. FMs are produced most often by extrusion, fol- consisting of fibrous monolithic diamond/ Co-bonded lowed by lay up of filaments into laminates. The ex- WC failed. The superior performance of the FM bits truded filaments consist of a cell phase surrounded by a was attributed to the fact that the Fm composite sheath that forms a continuous cell boundary [2-8]. exhibited better wear resistance and had nearly twice the FMs exhibit graceful failure in flexure, with energy fracture toughness of the Co/wC cermet [101 ssipation arising from sliding of the cells, and Many ceramic FMs are still in the developmental branching and deflection of cracks [8]. Ceramic FMs stage. The most successful ceramic FMs are based on constitute lower-cost alternatives to conventional con- Si3 N4 cells and a Bn boundary [2-8, 11-21]. Substantial inuous-fiber ceramic composites in some applications, work on FMs is in progress to improve processing and a wide variety of ceramic FMs are available com- methods, lower fabrication costs, incorporate new ceramics, and produce new forms. Although the prop- erties of existing ceramic FMs have been studied in de- Corresponding author. TeL:+1-630-252-5065: fax: +1-630-252. tail for nearly a decade [3-8, 11-21], much work is needed to unravel the complex relationships among E-mail addresses: goretta@anl. gov (K.C. Goretta) anl.gov (T.A. Cruse), dsingh( @anl. gov (D. Singh),rou composition, processing, structure, and properties ov (J.L. Routbort), aralaus es (A.R. de Arellano-Lo This paper will summarize design considerations for ova.t(amail ioffe. ru (T.S. Orlova), smir bi@amail ioffe. ru (B. 1. Smir- FMs. how ceramic FMs are fabricated the structures n produced, and the basic me 0263-8223S front matter Published by Elsevier Ltd do: 10.1016/ struct200405002

Ceramic fibrous monolithic structures K.C. Goretta a , T.A. Cruse a , D. Singh a , J.L. Routbort a,*, A.R. de Arellano-Lopez b , T.S. Orlova c , B.I. Smirnov c a Energy Technology Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439-4838, USA b Departamento de Fisica de la Materia Condensada, Universidad de Sevilla, P.O. Box 1065, 41080 Sevilla, Spain c Ioffe Physico-Technical Institute, Politekhnicheskaya ul. 26, St. Petersburg 194021, Russia Available online 9 June 2004 Abstract High-strength ceramic fibers and composite structures that contain them are generally expensive. In a lower-cost approach for fabricating fibrous composites, reinforcing fiber-like cells that are distinct from a continuous matrix phase called the cell boundary can be formed in situ from powders. Structures can be constructed by assembly and consolidation of filaments that consist of the cell phase and its surrounding cell boundary. Fabrication of ceramic fibrous monoliths (FMs) is reviewed and mechanical properties of the most widely studied FMs are discussed. Those based on Si3N4 cells within a BN cell boundary have achieved the best overall properties and uniformity of manufacture, but degrade severely at high temperatures in oxidizing environments. Those based on oxides are more stable, but are substantially weaker. Assessment of the future of FMs is offered, including cost reduction, fabrication practice, property improvement, and formation of complex structures. Published by Elsevier Ltd. Keywords: Ceramics; Composites; Fibrous monoliths 1. Introduction Powder-derived fibrous monoliths (FMs) generally consist of strong cells that are surrounded by a weaker cell boundary. The cells are typically
200–500 lm wide [1–8]. FMs are produced most often by extrusion, fol￾lowed by lay upof filaments into laminates. The ex￾truded filaments consist of a cell phase surrounded by a sheath that forms a continuous cell boundary [2–8]. FMs exhibit graceful failure in flexure, with energy dissipation arising from sliding of the cells, and branching and deflection of cracks [8]. Ceramic FMs constitute lower-cost alternatives to conventional con￾tinuous-fiber ceramic composites in some applications, and a wide variety of ceramic FMs are available com￾mercially [9]. Cermet FMs are now also produced commercially and have exhibited exceptional perfor￾mance as, for example, inserts for mining drill bits. In one study of rock drilling, all bits based on conventional Co-bonded WC fractured, whereas none of the bits consisting of fibrous monolithic diamond/Co-bonded WC failed. The superior performance of the FM bits was attributed to the fact that the FM composite exhibited better wear resistance and had nearly twice the fracture toughness of the Co/WC cermet [10]. Many ceramic FMs are still in the developmental stage. The most successful ceramic FMs are based on Si3N4 cells and a BN boundary [2–8,11–21]. Substantial work on FMs is in progress to improve processing methods, lower fabrication costs, incorporate new ceramics, and produce new forms. Although the prop￾erties of existing ceramic FMs have been studied in de￾tail for nearly a decade [3–8,11–21], much work is needed to unravel the complex relationships among composition, processing, structure, and properties. This paper will summarize design considerations for FMs, how ceramic FMs are fabricated, the structures that have been produced, and the basic mechanical * Corresponding author. Tel.: +1-630-252-5065; fax: +1-630-252- 4798. E-mail addresses: goretta@anl.gov (K.C. Goretta), cruse@cmt. anl.gov (T.A. Cruse), dsingh@anl.gov (D. Singh), routbort@anl. gov (J.L. Routbort), aral@us.es (A.R. de Arellano-Lopez), orl￾ova.t@mail.ioffe.ru (T.S. Orlova), smir.bi@mail.ioffe.ru (B.I. Smir￾nov). 0263-8223/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.compstruct.2004.05.002 Composite Structures 66 (2004) 547–553 www.elsevier.com/locate/compstruct

K.C. Goretta et al. Composite Structures 66(2004)547-553 properties of FMs. It will also address considerations for have required hot pressing to achieve the desired extent production of useful and complex structures, including of consolidation [7, 24]. Hot isostatic pressing would be possibilities for production of low-cost, high-perfor- much less expensive than conventional hot pressing mance laminates and other structures but to date hot pressing has yielded superior products 2. Manufacture of fibrous monoliths 2.2. Materials considerations 2.1. Processing details Current ceramic FMs are based on a duplex micro- structure that consists of dense cells separated by a Following the pioneering work by Halloran and his ontinuous cell boundary. The cells, which may be oxide colleagues [2-7. FMs have been formed by a single or non-oxide, provide most of the strength of the FM The cell boundary provides the toughness by isolating processing sequence: blending of plastic masses of the the cells from each other and promoting dissipation of cell and cell-boundary phases; coextruding a duplex fil- ament consisting of a core and sheath, which form fracture energy by mechanisms such as pullout of the eventually the cell and cell boundary, optional bundling boundary/p deflection of a crack through the cell of duplex filaments and re-extruding to form a filament Structural ceramics that have become fm cells in- with many smaller-diameter cells surrounded by a con- clude oxides such as mullite(Al2O3SiO2), AlO3[22- tinuous cell boundary, laying up of the filaments into a 24]. ZrO2[25], ZrSiO4 [26], and the non-oxides SiC (41 green body; removing the plastic constituents by heat treatment; and consolidating to a fibrous monolithic Si3N4 [2-21 and various borides [7]. Oxides have the part by sintering or some type of hot pressing advantage of stability in oxidizing environments; non- Details on formulations of various plastic mixtures oxides have the advantage of substantially higher an be found in [2-7, 22-26). Successful coextrusion re- strength and superior creep resistance quires matching of the rheological properties of the core o dissipate fracture energy, the cell boundaries must and sheath mixtures. Failure to adequately match them either be weak themselves or poorly bonded to the cells d to pinching For long-term use at elevated temperatures, the cell of cells or other irregularities. boundary must exhibit minimal or no reaction with the Depending on how the filaments are laid up, an FMcan be a unidirectional(Fig. 1)or cross-ply laminate. Fila cells. There are a variety of approaches to achieving the ment laying up can be by hand or, for example, by direct cells required ability to impart toughness and thermal means such as solid freeform fabrication[27-29). Hand and chemical stability lay up is the more-expensive approach, but to date has The cell boundary is roughly equivalent to the matrix also imparted the highest uniformity to the resulting FM. of conventional continuous-fiber ceramic compos Consolidation by simple heat treatment is the least- ites. All-oxide composites have achieved remarkable costly alternative, but it can be difficult to produce strengths and fracture toughnesses through engineering high-quality FMs by sintering Shrinkages during firing of porous matrices that minimize transmission of frac- must be matched to avoid cracking or formation of gy to the ceramic fibers [30,31]. Similarly unwanted voids and slumping must be minimized porous cell boundaries are effective in oxide FMs. They Oxide and some non-oxide FMs have been produced have been formed from, for example, large -grained by sintering, but all high-performance non-oxide FMs ceramic powders [22, 24, 26] or a mixture of ceramic particles and platelets [23, 32, 33]. To be fully effective in elevated-temperature application, the cell-boundary phase must be resistant to microstructural alteration with prolonged heating. Cell boundaries that incorpo rate a large fraction of large grains meet this require- ment. The cell boundary can be of the same composition as the cell [26] or unreactive [22-25 Oxide composites can also incorporate dense as matrices or cell boundaries, if the phases exhibi ibit little tendency to bond to the cells. Either a weak layer or very thin gap [32] can exists between the two constituents or the phases should inherently resist bonding strongly Monazite (LaPO4)and other similar compounds resist Cell boundary bonding to many structural ceramics [33-35] and have therefore been incorporated into conventional compos Fig. 1 Schematic diagram of unidirectional FM ites [34,35 or FMs [36, 37

properties of FMs. It will also address considerations for production of useful and complex structures, including possibilities for production of low-cost, high-perfor￾mance laminates and other structures. 2. Manufacture of fibrous monoliths 2.1. Processing details Following the pioneering work by Halloran and his colleagues [2–7], FMs have been formed by a single processing sequence: blending of plastic masses of the cell and cell-boundary phases; coextruding a duplex fil￾ament consisting of a core and sheath, which form eventually the cell and cell boundary; optional bundling of duplex filaments and re-extruding to form a filament with many smaller-diameter cells surrounded by a con￾tinuous cell boundary; laying upof the filaments into a green body; removing the plastic constituents by heat treatment; and consolidating to a fibrous monolithic part by sintering or some type of hot pressing. Details on formulations of various plastic mixtures can be found in [2–7,22–26]. Successful coextrusion re￾quires matching of the rheological properties of the core and sheath mixtures. Failure to adequately match them can lead to pinching of cells or other irregularities. Depending on how the filaments are laid up, an FM can be a unidirectional (Fig. 1) or cross-ply laminate. Fila￾ment laying upcan be by hand or, for example, by direct means such as solid freeform fabrication [27–29]. Hand lay up is the more-expensive approach, but to date has also imparted the highest uniformity to the resulting FM. Consolidation by simple heat treatment is the least￾costly alternative, but it can be difficult to produce high-quality FMs by sintering. Shrinkages during firing must be matched to avoid cracking or formation of unwanted voids and slumping must be minimized. Oxide and some non-oxide FMs have been produced by sintering, but all high-performance non-oxide FMs have required hot pressing to achieve the desired extent of consolidation [7,24]. Hot isostatic pressing would be much less expensive than conventional hot pressing, but to date hot pressing has yielded superior products. 2.2. Materials considerations Current ceramic FMs are based on a duplex micro￾structure that consists of dense cells separated by a continuous cell boundary. The cells, which may be oxide or non-oxide, provide most of the strength of the FM. The cell boundary provides the toughness by isolating the cells from each other and promoting dissipation of fracture energy by mechanisms such as pullout of the cells [18] or deflection of a crack through the cell boundary [15]. Structural ceramics that have become FM cells in￾clude oxides such as mullite (Al2O3 Æ SiO2), Al2O3 [22– 24], ZrO2 [25], ZrSiO4 [26], and the non-oxides SiC [4], Si3N4 [2–21], and various borides [7]. Oxides have the advantage of stability in oxidizing environments; non￾oxides have the advantage of substantially higher strength and superior creep resistance. To dissipate fracture energy, the cell boundaries must either be weak themselves or poorly bonded to the cells. For long-term use at elevated temperatures, the cell boundary must exhibit minimal or no reaction with the cells. There are a variety of approaches to achieving the cell’s required ability to impart toughness and thermal and chemical stability. The cell boundary is roughly equivalent to the matrix of conventional continuous-fiber ceramic compos￾ites. All-oxide composites have achieved remarkable strengths and fracture toughnesses through engineering of porous matrices that minimize transmission of frac￾ture energy to the ceramic fibers [30,31]. Similarly, porous cell boundaries are effective in oxide FMs. They have been formed from, for example, large-grained ceramic powders [22,24,26] or a mixture of ceramic particles and platelets [23,32,33]. To be fully effective in elevated-temperature application, the cell-boundary phase must be resistant to microstructural alteration with prolonged heating. Cell boundaries that incorpo￾rate a large fraction of large grains meet this require￾ment. The cell boundary can be of the same composition as the cell [26] or unreactive [22–25]. Oxide composites can also incorporate dense phases as matrices or cell boundaries, if the phases exhibit little tendency to bond to the cells. Either a weak layer or very thin gap[32] can exists between the two constituents or the phases should inherently resist bonding strongly. Monazite (LaPO4) and other similar compounds resist bonding to many structural ceramics [33–35] and have therefore been incorporated into conventional compos￾Fig. 1. Schematic diagram of unidirectional FM. ites [34,35] or FMs [36,37]. 548 K.C. Goretta et al. / Composite Structures 66 (2004) 547–553

K.C. Goretta et al. Composite Structures 66(2004)547-553 549 The criteria for selecting a cell boundary for non- oxide FMs are largely the same as for oxides. By far the most widely used cell boundary to date has been hex agonal BN [2-8, 11-21], although graphite has also beer cell [13, 20). In most cases, Si,N, has been the cell Oxide s960 used [7]. In hot-pressed FMs, BN forms a dense, highly textured cell boundary that bonds only weakly to the sintering aids in the Si, N4(generally Al,O3 and Y,O3) that leach from the cell into the bn cell boundary 220 [20]. BN is al- most a perfect cell-boundary material, except for the fact that it can oxidize severely at elevated temperatures 020.250.3035 [38]. For long-term use at elevated temperatures, in most environments an environmental barrier will have to coat Displacement (mm) FMs that contain a bn-based cell bounda Fig. 2. Representative stress-displacement curve for ZrsioZrsioa Oxide and non-oxide FMs experience significant FM tested in flexure [24,391 fluctuations in temperature during processing and will likely do so during service. Most ceramics in FMs ex hibit anisotropic thermal-expansion coefficients [24 and Failure in flexure has been graceful,with significant the structures of FMs are strongly anisotropic [14 Thermal expansion will lead to residual stresses [20, 21 reference to possible applications in aerospace and therefore not all pairs of compatible ceramics can be structures, resistance to solid-particle erosion of ZrSio4/ formed into effective FMs. Excessive cracking caused ZrSiO4 FMs has been measured [40]. It was found that by thermal stresses can arise when thermal-expansion the FMs eroded much more rapidly than monolithic coefficients differ too greatly ZrSio4 because of rapid removal of the cell boundary and fracture and removal of large sections of cells. The differences in erosion rates were sufficiently large that in 3. Performance of fibrous monoliths erosive environments it was concluded that fms with porous cell boundaries would require a protective FMs generally exhibit adequate strength, good coating toughness, and graceful failure in flexure [7, 24]. Si3N4/ BN FMS have achieved the best overall properties and 3.2. Non-oxides consistency of structure, and are reliably manufactured on a commercial scale [9]. The properties of all FMs FMs consisting of Si3 N4 cells and a continuous BN have to date been limited to a rule of mixtures. The cell boundary have achieved excellent mechanical ceramic that constitutes the strong, creep-resistant cells properties [2-8, 11-21]. These FMs have consisted of 80- is itself stronger and more creep resistant than the FMs 85 vol % cells and 15-20 vol. cell boundary. Their into which it is incorporated because the cell boundary flexural strengths have exceeded 700 MPa and work- can support minimal load. Loss of strength and creep of-fracture values, although typically 3-7 kJ/m, have resistance in an FM VS. the properties of the cell is offset exceeded 10 kJ/m [3-7]. These values are more than by marked increases in an FM in toughness and flaw adequate for many structural applications ce. Mechanical proper rties that have been at- Out-of-plane [3-8, 12-17] and in-plane [18] fracture tained in FMs will be summarized briefly and elastic modulus [19] have been studied in detail, including effects of temperature [13, 15, 38]. Both unid 3. Oxides rectional and cross-ply laminates have been studied Softening of the glass in the bn phase or, in the presence Oxides have generally been produced in small bat- of oxygen, formation of a borate glass, leads to reduced es; we do not know of an oxide FM that has been properties at temperatures >800C. manufactured consistently on a commercial scale. AlO Thermal-expansion coefficients have been measured or ZrSio4 have been most widely used for the cell [22- and modeled 19, 21, 41]. Analytical solutions agreed well 24, 26, 36, 37, 39]. Typical structures have been 60-75 with those from finite-element analysis, although some vol% cell and 25-40 vol. cell boundary uncertainty was introduced because of the texture of the Basic mechanical properties, such as flexural strength BN cell boundary [21]. The thermal-expansion coeffi and fracture energy, have been measured. Strengths cients generally increased with temperature(measure- have been approximately 100-500 MPa, and work- ments were made to 1200C)and were larger for the of-fracture values have generally been al kJ/m- or less. through-thickness direction than in-plane [21]

The criteria for selecting a cell boundary for non￾oxide FMs are largely the same as for oxides. By far the most widely used cell boundary to date has been hex￾agonal BN [2–8,11–21], although graphite has also been used [7]. In hot-pressed FMs, BN forms a dense, highly textured cell boundary that bonds only weakly to the cell [13,20]. In most cases, Si3N4 has been the cell. Oxide sintering aids in the Si3N4 (generally Al2O3 and Y2O3) that leach from the cell into the BN cell boundary during hot pressing impart the bonding [20]. BN is al￾most a perfect cell-boundary material, except for the fact that it can oxidize severely at elevated temperatures [38]. For long-term use at elevated temperatures, in most environments an environmental barrier will have to coat FMs that contain a BN-based cell boundary. Oxide and non-oxide FMs experience significant fluctuations in temperature during processing and will likely do so during service. Most ceramics in FMs ex￾hibit anisotropic thermal-expansion coefficients [24] and the structures of FMs are strongly anisotropic [14]. Thermal expansion will lead to residual stresses [20,21] and therefore not all pairs of compatible ceramics can be formed into effective FMs. Excessive cracking caused by thermal stresses can arise when thermal-expansion coefficients differ too greatly. 3. Performance of fibrous monoliths FMs generally exhibit adequate strength, good toughness, and graceful failure in flexure [7,24]. Si3N4/ BN FMs have achieved the best overall properties and consistency of structure, and are reliably manufactured on a commercial scale [9]. The properties of all FMs have to date been limited to a rule of mixtures. The ceramic that constitutes the strong, creep-resistant cells is itself stronger and more creepresistant than the FMs into which it is incorporated because the cell boundary can support minimal load. Loss of strength and creep resistance in an FM vs. the properties of the cell is offset by marked increases in an FM in toughness and flaw tolerance. Mechanical properties that have been at￾tained in FMs will be summarized briefly. 3.1. Oxides Oxides have generally been produced in small bat￾ches; we do not know of an oxide FM that has been manufactured consistently on a commercial scale. Al2O3 or ZrSiO4 have been most widely used for the cell [22– 24,26,36,37,39]. Typical structures have been 60–75 vol.% cell and 25–40 vol.% cell boundary. Basic mechanical properties, such as flexural strength and fracture energy, have been measured. Strengths have been approximately 100–500 MPa, and work￾of-fracture values have generally been
1 kJ/m2 or less. Failure in flexure has been graceful, with significant strength retained after primary failure (Fig. 2). In reference to possible applications in aerospace structures, resistance to solid-particle erosion of ZrSiO4/ ZrSiO4 FMs has been measured [40]. It was found that the FMs eroded much more rapidly than monolithic ZrSiO4 because of rapid removal of the cell boundary and fracture and removal of large sections of cells. The differences in erosion rates were sufficiently large that in erosive environments it was concluded that FMs with porous cell boundaries would require a protective coating. 3.2. Non-oxides FMs consisting of Si3N4 cells and a continuous BN cell boundary have achieved excellent mechanical properties [2–8,11–21]. These FMs have consisted of 80– 85 vol.% cells and 15–20 vol.% cell boundary. Their flexural strengths have exceeded 700 MPa and work￾of-fracture values, although typically 3–7 kJ/m2, have exceeded 10 kJ/m2 [3–7]. These values are more than adequate for many structural applications. Out-of-plane [3–8,12–17] and in-plane [18] fracture and elastic modulus [19] have been studied in detail, including effects of temperature [13,15,38]. Both unidi￾rectional and cross-ply laminates have been studied. Softening of the glass in the BN phase or, in the presence of oxygen, formation of a borate glass, leads to reduced properties at temperatures >800
C. Thermal-expansion coefficients have been measured and modeled [19,21,41]. Analytical solutions agreed well with those from finite-element analysis, although some uncertainty was introduced because of the texture of the BN cell boundary [21]. The thermal-expansion coeffi- cients generally increased with temperature (measure￾ments were made to 1200
C) and were larger for the through-thickness direction than in-plane [21]. Fig. 2. Representative stress–displacement curve for ZrSiO4/ZrSiO4 FM tested in flexure [24,39]. K.C. Goretta et al. / Composite Structures 66 (2004) 547–553 549

K.C. Goretta et al. Composite Structures 66(2004)547-553 Details of the various properties of ceramic FMs can be gleaned from the cited references. Compelling ques tions remain as to the extent to which these properties can be improved. Strength can perhaps be increased by well-controlled processing that reduces flaw severity, but so long as strengths can be predicted reliably, commer- cial applications for FMs are likely to be found. To improve mechanical properties substantially, we have examined two variations on conventional processing: (1) For all laminated structures, manipulation of residual stresses could perhaps lead to increased strength. (2) For tion of compression. creased pull-out of cells, and hence increased fracture toughness and, perhaps, strength [24]. Each of these High-temperature compressive creep of unidirec- improvements will be discussed briefly tional Si3,/BN has been studied in detail [42], however Most fractures in ceramics initiate at the surface be- only preliminary studies of cross-ply laminates have cause stresses are generally higher and flaws are most been performed [43]. The bN cell boundary can exhibit severe there. This is true of monolithic ceramics and microscopic plasticity, but does not truly creep. Its most likely, true of the cells in FMs. Compressive fracture strength is generally lower than the creep stress residual stresses at the surface of each cell might lead to hat the cells themselves can sustain. These FMs exhibit significantly higher strengths, but would have little effect good creep resistance in the longitudinal direction of the on fracture toughness. To fabricate such an FM,one cells, but creep rates are generally slightly faster than would need to produce a duplex cell surrounded by a hose of the host Si N, because the bn carries almost no continuous cell boundary. This structure can be formed load. Data from steady-state creep tests at 1400-1500C irectly by coextrusion of a three-layer filament. An indicated that stress was approximately proportional to easier alternative may be to coextrude a duplex filament strain rate; the activation energy was 600 kJ/mole [42] and coat it by a technology such as dip-coating or spray Some buckling of cells near the surface has been ob- coating[22] served(Fig 3)and, in compression, buckling would be a Residual stresses can be established on the surfaces of likely failure mode a duplex cell if the inner and outer materials have dif- Solid-particle erosion studies have indicated that ferent thermal-expansion coefficient. The main consid Si,N//BN FMs eroded much more rapidly than mono- erations for such a duplex cell are: compatibility lithic Si, N4, and, in fact, faster than would be predicted between species; ability to fabricate an FM from such a by any sort of rule of mixtures. Rapid loss of the Bn cell filament, in which the cell maintains its integrity; and the boundary and large-scale removal of the Si N4 cells was magnitude of the residual stresses. Preliminary work has responsible [44] which is similar to what occurred in been conducted on FMs based on the compatible oxides erosion of ZrSiO4-based FMs[40]. Erosive damage did mullite, AlO3, and Y2O3-stabilized ZrO2(YSZ) [22] not, on the basis of stress, lower the average strength of Al verage elastic properties for polycrystals of these three the FMs [44]. Independence of strength on erosive oxides are shown in Table I. Mullite has the lowest damage is attributable to the excellent tolerance to flaws thermal-expansion coefficient and YSZ the highest of ceramic Fms Values for the axial stress at the surface of the duple cell (P), the tangential stress at the surface of the sheath (o, 1), and the tangential stress at the outer sheath su 4. Structural application of FMs face(o,2) were estimated based on a flat-plate geometry [22], and can be been calculated for cylinders from an 4 laminates analysis by Hahn [451 Virtually all FMs studied to date have been either unidirectional or cross-ply laminates. Strength [2-7, 16- Table I 18, 23-26], stiffness [19, 21], and creep resistance [42, 43] Average coefficient of thermal expansion(x), and room-temperature have been bounded by a rule of mixtures. Thus, because Young's modulus (E)and Poisson's ratio (v) for FM constituents the cell boundaries are relatively weak, FMs are not in x(C-) E(GPa) general as strong, stiff, or creep resistant as is the cera- Mullite 59×10-6 0.25 mic of which the cells are composed. FMs are, however, AlO3 9.2×10-6 0.26 tougher, especially in flexural testing 11.6×10-6

High-temperature compressive creep of unidirec￾tional Si3N4/BN has been studied in detail [42], however, only preliminary studies of cross-ply laminates have been performed [43]. The BN cell boundary can exhibit microscopic plasticity, but does not truly creep. Its fracture strength is generally lower than the creepstress that the cells themselves can sustain. These FMs exhibit good creepresistance in the longitudinal direction of the cells, but creeprates are generally slightly faster than those of the host Si3N4 because the BN carries almost no load. Data from steady-state creeptests at 1400–1500
C indicated that stress was approximately proportional to strain rate; the activation energy was
600 kJ/mole [42]. Some buckling of cells near the surface has been ob￾served (Fig. 3) and, in compression, buckling would be a likely failure mode. Solid-particle erosion studies have indicated that Si3N4/BN FMs eroded much more rapidly than mono￾lithic Si3N4, and, in fact, faster than would be predicted by any sort of rule of mixtures. Rapid loss of the BN cell boundary and large-scale removal of the Si3N4 cells was responsible [44], which is similar to what occurred in erosion of ZrSiO4-based FMs [40]. Erosive damage did not, on the basis of stress, lower the average strength of the FMs [44]. Independence of strength on erosive damage is attributable to the excellent tolerance to flaws of ceramic FMs. 4. Structural application of FMs 4.1. Laminates Virtually all FMs studied to date have been either unidirectional or cross-ply laminates. Strength [2–7,16– 18,23–26], stiffness [19,21], and creepresistance [42,43] have been bounded by a rule of mixtures. Thus, because the cell boundaries are relatively weak, FMs are not in general as strong, stiff, or creepresistant as is the cera￾mic of which the cells are composed. FMs are, however, tougher, especially in flexural testing. Details of the various properties of ceramic FMs can be gleaned from the cited references. Compelling ques￾tions remain as to the extent to which these properties can be improved. Strength can perhaps be increased by well-controlled processing that reduces flaw severity, but so long as strengths can be predicted reliably, commer￾cial applications for FMs are likely to be found. To improve mechanical properties substantially, we have examined two variations on conventional processing: (1) For all laminated structures, manipulation of residual stresses could perhaps lead to increased strength. (2) For cross-ply laminates, changes in the cross section of a duplex filament should lead in the resulting FM to in￾creased pull-out of cells, and hence increased fracture toughness and, perhaps, strength [24]. Each of these improvements will be discussed briefly. Most fractures in ceramics initiate at the surface be￾cause stresses are generally higher and flaws are most severe there. This is true of monolithic ceramics and, most likely, true of the cells in FMs. Compressive residual stresses at the surface of each cell might lead to significantly higher strengths, but would have little effect on fracture toughness. To fabricate such an FM, one would need to produce a duplex cell surrounded by a continuous cell boundary. This structure can be formed directly by coextrusion of a three-layer filament. An easier alternative may be to coextrude a duplex filament and coat it by a technology such as dip-coating or spray￾coating [22]. Residual stresses can be established on the surfaces of a duplex cell if the inner and outer materials have dif￾ferent thermal-expansion coefficient. The main consid￾erations for such a duplex cell are: compatibility between species; ability to fabricate an FM from such a filament, in which the cell maintains its integrity; and the magnitude of the residual stresses. Preliminary work has been conducted on FMs based on the compatible oxides mullite, Al2O3, and Y2O3-stabilized ZrO2 (YSZ) [22]. Average elastic properties for polycrystals of these three oxides are shown in Table 1. Mullite has the lowest thermal-expansion coefficient and YSZ the highest. Values for the axial stress at the surface of the duplex cell (P), the tangential stress at the surface of the sheath ðrt;1Þ, and the tangential stress at the outer sheath sur￾face ðrt;2Þ were estimated based on a flat-plate geometry [22], and can be been calculated for cylinders from an analysis by Hahn [45]. Fig. 3. Scanning electron microscopy (SEM) photomicrograph of unidirectional Si3N4/BN FM crept at 1500
C; arrows indicate direc￾tion of compression. Table 1 Average coefficient of thermal expansion ðaÞ, and room-temperature Young’s modulus ðEÞ and Poisson’s ratio ðmÞ for FM constituents Oxide a (
C1) E (GPa) m Mullite 5.9 · 106 145 0.25 Al2O3 9.2 · 106 380 0.26 YSZ 11.6 · 106 205 0.23 550 K.C. Goretta et al. / Composite Structures 66 (2004) 547–553

K.C. Goretta et al. Composite Structures 66(2004)547-553 B2+B3 (a2-01)△T B1B3-2B2 1-n1v2.1+H E1+E2 B2=E E1 VEy /BB2+B31 Fig. 4. SEM phot graph of fractured section of 0/.. a2=(B-2人(2-a)△r Si3 N/BN FM in which kinking of cells is evident (arrow indicate sliding of cells, and A and B indicate pullout in each of two directions) where the subscript I refers to the core and the subscript 2 refers to the sheath. and v is the volume fraction Residual-stress values have been calculated for three volume fractions of core/sheath, with the core being 50 vol% mullite/50 vol. YSZ and the sheath being 50 vol% mullite/50 vol. AlO3 (Table 2). The residual stresses are a strong function of the sheath/core volume fraction. The stresses for a cell that is 90 vol. core are probably too large to be sustained without fracture Insufficient work has been done to determine the efficacy of this approach to fabricating FMs, and to our Fig. 5. SEM photomicrograph of fractured section of 0/90 cross-ply knowledge, this approach has been applied only to oxide Si,N/BN FM for which cell pullout lengths are short. FMs. Although residual stresses have been shown to be highly successful in preventing crack extension in brittle caused by cell kinking. Pullout lengths in cross-ply ramics(see, for example, [46), all of the handful of SiN /BN FMs have been measured to be on the same prototype mullite/Al_O3/YSZ FMs that we have pro. order as the cell width, 0.3 mm [18]. They could be duced to date have exhibited excessive cracking. Nev- increased substantially, and thus toughness increased ertheless, this approach probably merits additional markedly, if cell kinking could be minimized. Filaments attention with cross sections that fill space when stacked could do The second way in which mechanical properties of that. We have demonstrated this capability by simply existing FMs can be improved is specific to cross-ply fattening the filaments prior to stacking them into a laminates. Such laminates are probably the most com- green FM[47]. It is also possible to directly extrude cells mon form of FMs now produced. Because the filaments of various shapes that will minimize kinking. To date, now extruded are round, when stacked and pressed in however, insufficient work has been done to prove the the green state, two types of distortions occur. The cross efficacy of this approach. sections of the individual filaments (cells plus cell boundaries) flatten. In cross-ply laminates, the filaments 4.2 Three-dimensional structures Iso become kinked along their lengths as alternating layers fill the voids that are present because stacked Polymer- and metal-matrix composites can readily cylinders do not completely fill space. Resulting FMs be fabricated into complex shapes. For complex forms exhibit a unique pattern of fracture. As shown in Fig. 4, fracture of FMs involves sliding cells, one may apply two types of fabrication methods and pullout of individual cells. However, pullout lengths (1) For composites that contain either fibers or Fm cells are rather short(Fig. 5)because of stress concentrations one may weave the reinforcing phase [48; or, for FMs and other composites that do not incorporate pre- Table 2 existing continuous fibers, one can pattern the rein Calculated residual stresses in duplex FM cell forcing phase into a three-dimensional configuration g,,(GPa) a,(GPa) directly [49].(2)One can also attempt to join simpler forms in which the reinforcing phase is distributed in a 90/10 -6.7 two-dimensional pattern [50, 511 8020 For FMs, weaving encounters a series of problems 7030 trengths are governed by a rule of mixtures and any

P ¼ b2 þ b3 b1b3  2b2 2 ðr2  r1ÞDT rt;1 ¼  1 þ V1 V2 P; rt;2 ¼  2V1 V2 P b1 ¼ 1  m1 E1 þ m2 E2 þ 1 þ V2 V1E2 ; b2 ¼ m1 E1 þ V1m2 V2E2 ; b3 ¼ 1 E1 þ V1 V2E2 rz;2 ¼  V1 V2 b1 b2 b2 þ b3 b1b3  2b2 2  1 b2 ! ða2  a1ÞDT where the subscript 1 refers to the core and the subscript 2 refers to the sheath, and V is the volume fraction. Residual-stress values have been calculated for three volume fractions of core/sheath, with the core being 50 vol.% mullite/50 vol.% YSZ and the sheath being 50 vol.% mullite/50 vol.% Al2O3 (Table 2). The residual stresses are a strong function of the sheath/core volume fraction. The stresses for a cell that is 90 vol.% core are probably too large to be sustained without fracture. Insufficient work has been done to determine the efficacy of this approach to fabricating FMs, and to our knowledge, this approach has been applied only to oxide FMs. Although residual stresses have been shown to be highly successful in preventing crack extension in brittle ceramics (see, for example, [46]), all of the handful of prototype mullite/Al2O3/YSZ FMs that we have pro￾duced to date have exhibited excessive cracking. Nev￾ertheless, this approach probably merits additional attention. The second way in which mechanical properties of existing FMs can be improved is specific to cross-ply laminates. Such laminates are probably the most com￾mon form of FMs now produced. Because the filaments now extruded are round, when stacked and pressed in the green state, two types of distortions occur. The cross sections of the individual filaments (cells plus cell boundaries) flatten. In cross-ply laminates, the filaments also become kinked along their lengths as alternating layers fill the voids that are present because stacked cylinders do not completely fill space. Resulting FMs exhibit a unique pattern of fracture. As shown in Fig. 4, fracture of FMs involves sliding and pullout of individual cells. However, pullout lengths are rather short (Fig. 5) because of stress concentrations caused by cell kinking. Pullout lengths in cross-ply Si3N4/BN FMs have been measured to be on the same order as the cell width,
0.3 mm [18]. They could be increased substantially, and thus toughness increased markedly, if cell kinking could be minimized. Filaments with cross sections that fill space when stacked could do that. We have demonstrated this capability by simply flattening the filaments prior to stacking them into a green FM [47]. It is also possible to directly extrude cells of various shapes that will minimize kinking. To date, however, insufficient work has been done to prove the efficacy of this approach. 4.2. Three-dimensional structures Polymer– and metal–matrix composites can readily be fabricated into complex shapes. For complex forms of ceramic–matrix composites with continuous fibers or cells, one may apply two types of fabrication methods: (1) For composites that contain either fibers or FM cells, one may weave the reinforcing phase [48]; or, for FMs and other composites that do not incorporate pre￾existing continuous fibers, one can pattern the rein￾forcing phase into a three-dimensional configuration directly [49]. (2) One can also attempt to join simpler forms in which the reinforcing phase is distributed in a two-dimensional pattern [50,51]. For FMs, weaving encounters a series of problems. Strengths are governed by a rule of mixtures and any Table 2 Calculated residual stresses in duplex FM cell Core/sheath ratio rz;2 (GPa) rt;1 (GPa) rt;2 (GPa) 90/10 )6.7 )7.1 )2.0 80/20 )1.7 )1.9 0.5 70/30 )0.8 )1.0 1.8 Fig. 4. SEM photomicrograph of fractured section of 0
/90
cross-ply Si3N4/BN FM in which kinking of cells is evident (arrow indicate sliding of cells, and A and B indicate pullout in each of two directions). Fig. 5. SEM photomicrograph of fractured section of 0
/90
cross-ply Si3N4/BN FM for which cell pullout lengths are short. K.C. Goretta et al. / Composite Structures 66 (2004) 547–553 551