《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_Al2O3-SiC-高温

CERAMICS INTERNATIONAL ELSEVIER Ceramics International 25(1999)395-408 Review: High temperature deformation of Al2O3-based ceramic particle or whisker composites Q. Tai*, A. Mocellin LSG2M, UMR 7584, Ecole des Mines, Parc de saurupt, F-54042 Nancy Cedex, france Received 6 September 1997; accepted 17 November 1997 Abstract The major theoretical models for creep and the creep rate equations of ceramic materials and their dispersed phase composites e briefly reviewed. Then the literature on high temperature deformation behaviours of Al2O3-based oxide ceramic particle com- osites(Al2O3-ZrO2, Al2O3-Y3Al5O12, Al2O3-Tio2) and Al2O3-based non-oxide ceramic particle or whisker composites(Al2O3- SiC(w), Al2O3-SiC(p). Al2O3-TiCx Ni-x)since the mid 1980s is reviewed. Most studies have been concerned with the Al2O3-ZrO2 and Al2O3-SiC systems. The influences of various factors on the creep behaviours, the changes of the microstructure in the deformed specimens and the creep mechanisms of these composites are summarised and analysed. c 1999 Elsevier Science Limited and Techna S.r. l. all rights reserved 1. Introduction and reliabilities. Great emphasis is placed on their high temperature creep behaviours. In recent years, structural ceramic materials have The purpose of this paper is to recall briefly the major attracted much attention, because of their excellent theoretical models for creep and then to review the mechanical properties such as high strength, hardness, available information on the plastic deformation beha- anti-abrasion, chemical stability and heat resistance. viours of Al2O3-based ceramic composites. Composites There has been a recognition of the potential of struc- reinforced by long fibres are not discussed here since tural ceramics for use both in high temperature appli- their fabrication procedures markedly differ from those cations in advanced heat engine and heat exchangers based on powder processing which yield materials with and in ambient temperature applications in cutting dispersed phases tools, and wear parts. The disadvantage of ceramics is their low fracture toughness and poor mechanical relia- bility which so far have limited their practical applica 2. Deformation mechanisms s. me o Improve toughness and retain their high-temperature creep 2. 1. The rate equations for plastic deformation properties as well as enhance their mechanical reliability are a major challenge. One way to achieve these goals is The high temperature creep of single phase crystalline through the development of composite structures. That materials may be expressed by a relationship of the fol is to the ceramic matrices are added dispersed ceramic lowing form: particles, whiskers or fibres which reinforce the matrices and improve their mechanical properties Alumina-based ceramic composites such as Al_O3- E=A Zr02, Al O3-Y3AlsO12, Al2O3-SiC, Al,O3-TiC, Al,O TiCNI-x composites are widely studied, as to their where E is the steady state creep rate, A is a dimension ambient and high temperature mechanical properties less constant, D is the appropriate diffusion coefficient, G is the shear modulus, b is the magnitude of the Bur esponding author at Nanjing University of Chemical Tech- gers vector, k is Boltzmanns constant, Tis the absolute nology, 210009, Nanjing, People's Republic of China. temperature, d is the grain size, o is the applied stress, 0272-8842/99/$20.00@ 1999 Elsevier Science Limited and Techna S.r. L. All rights reserved PII:S0272-8842(98)00017-0

Review: High temperature deformation of Al2O3-based ceramic particle or whisker composites Q. Tai *, A. Mocellin LSG2M, UMR 7584, Ecole des Mines, Parc de Saurupt, F-54042 Nancy Cedex, France Received 6 September 1997; accepted 17 November 1997 Abstract The major theoretical models for creep and the creep rate equations of ceramic materials and their dispersed phase composites are brie¯y reviewed. Then the literature on high temperature deformation behaviours of Al2O3-based oxide ceramic particle com￾posites (Al2O3-ZrO2, Al2O3-Y3Al5O12, Al2O3-Tio2) and Al2O3-based non-oxide ceramic particle or whisker composites (Al2O3- SiC(w), Al2O3-SiC(p), Al2O3-TiCxN1-x) since the mid 1980s is reviewed. Most studies have been concerned with the Al2O3-ZrO2 and Al2O3-SiC systems. The in¯uences of various factors on the creep behaviours, the changes of the microstructure in the deformed specimens and the creep mechanisms of these composites are summarised and analysed. # 1999 Elsevier Science Limited and Techna S.r.l. All rights reserved. 1. Introduction In recent years, structural ceramic materials have attracted much attention, because of their excellent mechanical properties such as high strength, hardness, anti-abrasion, chemical stability and heat resistance. There has been a recognition of the potential of struc￾tural ceramics for use both in high temperature appli￾cations in advanced heat engine and heat exchangers and in ambient temperature applications in cutting tools, and wear parts. The disadvantage of ceramics is their low fracture toughness and poor mechanical relia￾bility which so far have limited their practical applica￾tions. Thus, methods to improve their fracture toughness and retain their high-temperature creep properties as well as enhance their mechanical reliability are a major challenge. One way to achieve these goals is through the development of composite structures. That is to the ceramic matrices are added dispersed ceramic particles, whiskers or ®bres which reinforce the matrices and improve their mechanical properties. Alumina-based ceramic composites such as Al2O3- Zr02, Al2O3-Y3Al5O12, Al2O3-SiC, Al2O3-TiC, Al2O3- TiCxN1-x composites are widely studied, as to their ambient and high temperature mechanical properties and reliabilities. Great emphasis is placed on their high temperature creep behaviours. The purpose of this paper is to recall brie¯y the major theoretical models for creep and then to review the available information on the plastic deformation beha￾viours of Al2O3-based ceramic composites. Composites reinforced by long ®bres are not discussed here since their fabrication procedures markedly di€er from those based on powder processing which yield materials with dispersed phases. 2. Deformation mechanisms 2.1. The rate equations for plastic deformation The high temperature creep of single phase crystalline materials may be expressed by a relationship of the fol￾lowing form: "_ ˆ A DGb kT b d  p  G  n …1† where "_ is the steady state creep rate, A is a dimension￾less constant, D is the appropriate di€usion coecient, G is the shear modulus, b is the magnitude of the Bur￾gers vector, k is Boltzmann's constant, T is the absolute temperature, d is the grain size,  is the applied stress, Ceramics International 25 (1999) 395±408 0272-8842/99/$20.00 #1999 Elsevier Science Limited and Techna S.r.l. All rights reserved PII: S0272-8842(98)00017-0 * Corresponding author at Nanjing University of Chemical Tech￾nology, 210009, Nanjing, People's Republic of China

Q. Tai. A. Mocellin/Ceramics International 25(1999)393-408 and p and n are constants termed the inverse grain size and exponent and the stress exponent, respectively. The dif- fusion coefficient D may be expressed as Do exp(-Q/ RD, where Do is a frequency factor, Q is the apparent PI=P2 PI activation energy, and R is the gas constant Vi is the volume fraction of phase i; ni is the viscosity For two phase composites, there are several equations undergoing Newtonian viscous flow which may express or predict their high temperature n= Vinl +v2772: qi is the phase'stress-concentration reep behaviours factor, Vig1+ V292=1; pi is internal stress caused by the In composites, where the second phase can be mismatch in creep strains between the phases sidered rigid, Raj and Ashby model [1] assumes that the VIPI+V2P2=0. hard second phase particles in the grain boundary of The Eqn. (5)is also valid for the case wherein one of matrix limit the grain boundary sliding and gives he phases is nondeformable by creep if diffusional mass transport around the purely elastic phase is taken into account R 2. 2. Theoretical models for plastic deformation where V is the second phase volume content, r is second phase grain radius, q and n are phenomenological There are several theoretical models for creep defor exponents and C is a constant mation. In general, they can be divided into two broad Chen model [2] considers the composites as a model categories: boundary mechanisms [5-ll and lattice system of a soft matrix containing equiaxed and rigid mechanisms [5, 12]. Boundary mechanisms rely on the inclusions. Based on a phenomenological constitutive presence of grain boundaries and occur only in poly- equation and a second phase continuum mechanics crystalline materials. They are associated with some model, his model gives: dependence on grain size so that p> l. Lattice mechan- isms are independent of the presence of grain bound E=(1-V)2+n2 () aries and occur both in single crystal and polycrystalline materials. They occur within the grain interiors and are where V is the second phase volume content, Eo is the independent of grain size, so p=0 strain rate of the reference matrix, n is the stress expo The boundary mechanisms can be subdivided into nent of the matrix four categories: diffusion creep, [5-7] interface reaction Ravichandran and Seetharaman model [3] considers controlled diffusion creep [8], grain boundary sliding that a rigid and noncreeping second phase distributes and grain rearrangement [5,6,9, 101, and cavitation creep uniformly in a continuous creeping matrix, and they and microcracking [11]. In diffusion creep where vacan- develop a simple continuum mechanics model to predict cies may flow from the zones experiencing tension to the steady state creep rates of composites those in compression either through the crystalline lat tice(Nabarro-Herring creep) or along the grain (1+C2 boundaries( Coble creep), the individual grains become (1+C) (4) elongated along the tensile axis Fig. 1). When grain +(1+C boundaries do not act as perfect sources or sinks for vacancies, the process of creating or annihilating point where C=l-I, v is the second phase volume content, n is the stress exponent of the matrix, A is the constant of the matrix For a two phase composite in which each phase undergoes diffusional creep, Wakashima and Liu give a viscoelastic constitutive equation corresponding to SUND/AR pring- dashpot model [4: A1-p(-分) where △EV1 Fig. 1. Diffusion flow by lattice(Nabarro-Herring creep)or by grain boundaries( Coble creep)

and p and n are constants termed the inverse grain size exponent and the stress exponent, respectively. The dif￾fusion coecient D may be expressed as Do exp (ÿQ/ RT), where Do is a frequency factor, Q is the apparent activation energy, and R is the gas constant. For two phase composites, there are several equations which may express or predict their high temperature creep behaviours. In composites, where the second phase can be con￾sidered rigid, Raj and Ashby model [1] assumes that the hard second phase particles in the grain boundary of matrix limit the grain boundary sliding and gives: "_ ˆ C n dprqV exp ÿ Q RT   …2† where V is the second phase volume content, r is second phase grain radius, q and n are phenomenological exponents and C is a constant. Chen model [2] considers the composites as a model system of a soft matrix containing equiaxed and rigid inclusions. Based on a phenomenological constitutive equation and a second phase continuum mechanics model, his model gives: "_ ˆ "_o…1 ÿ V† 2‡n=2 …3† where V is the second phase volume content, "_o is the strain rate of the reference matrix, n is the stress expo￾nent of the matrix. Ravichandran and Seetharaman model [3] considers that a rigid and noncreeping second phase distributes uniformly in a continuous creeping matrix, and they develop a simple continuum mechanics model to predict the steady state creep rates of composites: "_ ˆ A …1 ‡ C† 2 …1‡C† 1=n C   ‡ …1 ‡ C† 2 ÿ 1 2 4 3 5 n …4† where C ˆ 1 V 1=3 ÿ1, V is the second phase volume content, n is the stress exponent of the matrix, A is the constant of the matrix. For a two phase composite in which each phase undergoes di€usional creep, Wakashima and Liu give a viscoelastic constitutive equation corresponding to a spring-dashpot model [4]: " ˆ
E Eu 1 ÿ exp ÿ t  n o   ‡ 1    …5† where
E Eu  V1 p1 q1 ÿ 1   2 ˆ V2 p2 q2 ÿ 2 u  2 and   1 p1 ÿ p2 ˆ 1 p1 1 ÿ V1 1    ˆ ÿ 2 p2 1 ÿ V2 2    Vi is the volume fraction of phase i; i is the viscosity of phase i undergoing Newtonian viscous ¯ow,   V11 ‡ V22; qi is the phase `stress-concentration' factor, V1q1+V2q2=1; pi is internal stress caused by the mismatch in creep strains between the phases, V1p1 ‡ V2p2 ˆ 0. The Eqn. (5) is also valid for the case wherein one of the phases is nondeformable by creep if di€usional mass transport around the purely elastic phase is taken into account. 2.2. Theoretical models for plastic deformation There are several theoretical models for creep defor￾mation. In general, they can be divided into two broad categories: boundary mechanisms [5±11] and lattice mechanisms [5,12]. Boundary mechanisms rely on the presence of grain boundaries and occur only in poly￾crystalline materials. They are associated with some dependence on grain size so that p51. Lattice mechan￾isms are independent of the presence of grain bound￾aries and occur both in single crystal and polycrystalline materials. They occur within the grain interiors and are independent of grain size, so p=0. The boundary mechanisms can be subdivided into four categories: di€usion creep, [5±7] interface reaction controlled di€usion creep [8], grain boundary sliding and grain rearrangement [5,6,9,10], and cavitation creep and microcracking [11]. In di€usion creep where vacan￾cies may ¯ow from the zones experiencing tension to those in compression either through the crystalline lat￾tice (Nabarro±Herring creep) or along the grain boundaries (Coble creep), the individual grains become elongated along the tensile axis Fig. 1). When grain boundaries do not act as perfect sources or sinks for vacancies, the process of creating or annihilating point Fig. 1. Di€usion ¯ow by lattice (Nabarro±Herring creep) or by grain boundaries (Coble creep). 396 Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408

Q. Tai. A. Mocellin/Ceramics International 25(1999)395-408 defects may control the creep deformation. This is also been proposed by a number of authors but are not termed interface reaction-controlled diffusion creep. to be reviewed here Because this process involves diffusion of vacancies, the The lattice mechanisms can be briefly subdivided into grains are also elongated along the tensile axis In grain two categories [12]: dislocation climb and glide con- boundary sliding and grain rearrangement, there are trolled by climb on the one hand and dislocation climb several different models(Lifshitz sliding, Rachinger on the other(Fig 4). For the dislocation climb and glide sliding, Ashby and Verrall model, Gifkins model, etc. ) controlled by climb mechanism, in ceramics, the anion Lifshitz sliding occurs naturally as part of diffusion cation ratio ra/rc is 2, and the ceramics are either lacking some slip sys grains lying along the tensile axis. In contrast, during tems, or if five independent systems are available, not all Rachinger sliding, the grains slide, rearrange and retain may be active simultaneously [12]. The dislocation creep there is an increase in number of grains along the tensile tensile axis as weave the elongation of grains along the their original shapes, but exchange their neighbours, so mechanisms involve axis. In Ashby and Verrall model, which is two-dimen sional, during the deformation process, grains suffer a 2.3. Methods to identify deformation mechanism transient but complex shape change by diffusional transport Fig. 2). While in the Gifkins model, which In general, several mechanisms may contribute to the may be viewed as three-dimensional, during the defor- creep deformation at elevated temperature, but creep is mation process, grains move apart by grain boundary usually controlled by only one of these mechanisms. To sliding caused by the motion of grain boundary dis- identify which mechanism is dominant, there are several locations, resulting in a gap between the grains. When methods the gap is large enough, it is filled by an emerging grain Ashby [13]constructed deformation-mechanism maps from one layer to the next(Fig. 3)[10]. In these two by using rate-equations and sufficient data on the models, the grains almost retain their original shapes, materials. These maps for example show the fields of and there is an increase in number of grains along the stress and temperature in which each independent tensile axis. In the cavitation creep and microcracking, mechanism for plastic deformation is dominant. Knowl- extensive cavities form. They grow and link up forming edge of any two of the three variables(stress, temperature microcrack by grain boundary sliding. The principal and strain-rate) locates a point on the map, identifies mode of deformation is a damage mechanism. Varia- the dominant mechanism or mechanisms and gives the tions or refinements of the previous basic models have value of the third variable. Mohamed and Langdon [14] constructed deformation-mechanism maps with grain size as a variable. based on the deformation -mechanism maps constructed by Ashby and Mohamed et al., Heuer et al. [9], for example, have devised a stress-grain size deformation mechanism map for MgO-doped Al2O3 at 1500C(Fig. 5). They suggested that diffusional defor mation dominated for most grain sizes of interest, but Fig. 2. Grain boundary sliding and grain rearrangement by diffusion shby and verrall model)[10] 双 Fig 3. Grain boundary sliding and grain rearrangement. A gap forms between four grains and is filled by an emerging grain(Gifkins model) Fig 4. Dislocations move by (a)climb, (b)climb and glide

defects may control the creep deformation. This is termed interface reaction-controlled di€usion creep. Because this process involves di€usion of vacancies, the grains are also elongated along the tensile axis. In grain boundary sliding and grain rearrangement, there are several di€erent models (Lifshitz sliding, Rachinger sliding, Ashby and Verrall model, Gifkins model, etc.). Lifshitz sliding occurs naturally as part of di€usion creep, to maintain grain contiguity, the grains elongate along the tensile axis and maintain their adjacent neighbours, so there is no increase in the number of grains lying along the tensile axis. In contrast, during Rachinger sliding, the grains slide, rearrange and retain their original shapes, but exchange their neighbours, so there is an increase in number of grains along the tensile axis. In Ashby and Verrall model, which is two-dimen￾sional, during the deformation process, grains su€er a transient but complex shape change by di€usional transport Fig. 2). While in the Gifkins model, which may be viewed as three-dimensional, during the defor￾mation process, grains move apart by grain boundary sliding caused by the motion of grain boundary dis￾locations, resulting in a gap between the grains. When the gap is large enough, it is ®lled by an emerging grain from one layer to the next (Fig. 3) [10]. In these two models, the grains almost retain their original shapes, and there is an increase in number of grains along the tensile axis. In the cavitation creep and microcracking, extensive cavities form. They grow and link up forming microcrack by grain boundary sliding. The principal mode of deformation is a damage mechanism. Varia￾tions or re®nements of the previous basic models have also been proposed by a number of authors but are not to be reviewed here. The lattice mechanisms can be brie¯y subdivided into two categories [12]: dislocation climb and glide con￾trolled by climb on the one hand and dislocation climb on the other (Fig. 4). For the dislocation climb and glide controlled by climb mechanism, in ceramics, the anion/ cation ratio ra/rc is2, and the ceramics are either lacking some slip sys￾tems, or if ®ve independent systems are available, not all may be active simultaneously [12]. The dislocation creep mechanisms involve the elongation of grains along the tensile axis as well. 2.3. Methods to identify deformation mechanism In general, several mechanisms may contribute to the creep deformation at elevated temperature, but creep is usually controlled by only one of these mechanisms. To identify which mechanism is dominant, there are several methods. Ashby [13] constructed deformation-mechanism maps by using rate-equations and sucient data on the materials. These maps for example show the ®elds of stress and temperature in which each independent mechanism for plastic deformation is dominant. Knowl￾edge of any two of the three variables (stress, temperature and strain-rate) locates a point on the map, identi®es the dominant mechanism or mechanisms and gives the value of the third variable. Mohamed and Langdon [14] constructed deformation-mechanism maps with grain size as a variable. Based on the deformation-mechanism maps constructed by Ashby and Mohamed et al., Heuer et al. [9], for example, have devised a stress-grain size deformation mechanism map for MgO-doped Al2O3 at 1500C (Fig. 5). They suggested that di€usional defor￾mation dominated for most grain sizes of interest, but Fig. 2. Grain boundary sliding and grain rearrangement by di€usion (Ashby and Verrall model) [10]. Fig. 3. Grain boundary sliding and grain rearrangement. A gap forms between four grains and is ®lled by an emerging grain (Gifkins model) [10]. Fig. 4. Dislocations move by (a) climb, (b) climb and glide. Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408 397

some basal slip could easily occur. Furthermore, when composite theory: isostrain and isostress model [15,16 grain size of Al_O3 was between 50 and 500 um and the (Fig. 6). Isostrain and isostress prediction diagrams can imposed stress was large enough, then dislocation climb be constructed by calculation. By comparing the and glide could occur. But in practice, the theory and experimental data with the isostrain and isostress pre- the experimental data used to construct the maps are diction, the model fitting the creep deformation can be poor or insufficient for many ceramic materials, thus determined. Since the isostress model is dominated by limiting their use in applications he least creep resistant phase and highest creep rate and At present, the main method to identify the possible the isostrain model is dominated by the most creep ate controlling mechanism is to compare the values of resistant phase and lowest creep rate, the phase con np and o obtained from experiments with theoretical trolling the creep behaviour can be determined predictions. Chokshi et al. [12 indicated that many For the two phase composites in which both phases ceramics exhibit stress exponents of N5, N3 or l deform inelastically, a self-consistent model was devel which appeared to be associated with dislocation glide oped [17, 18], which predicts the deformation behaviour and climb, climb from Bardeen-Herring sources, and of the composites when the viscoplastic laws of diffusion creep, respectively. The stress exponent of 2 phase are known. By self-consistent calculations, effec- might be due to the presence of a partially wetting grain tive strain rate sensitivity parameter and effective pre- boundary glassy phase or to control by an interface factor which are characteristic of the composite reaction[ 12] Besides, the stress exponent between I and behaviour can be obtained. Stress and strain rates in might be associated with grain boundary sliding and each phase are also attainable. From the comparison grain rearrangement, and a higher stress exponent between the model and the experiments, the possible (n>3)can also be associated with cavitation creep and deformation mechanisms of each phase can be deter microcracking. The inverse grain size exponent points to mined and the phase controlling the creep behaviour either a boundary mechanism (p>1)or a lattice can also be qualitatively determined mechanism (p=0). In general, a direct observation and analysis of the microstructure of the specimens after deformation is necessary to check preliminary conclu- 3. Plastic deformation behaviours of Al2O3-based sions drawn from the experimentally determined stress- ceramic composites train rate relationship For two phase composites, especially for those with In studies of the plastic deformation behaviours of duplex microstructures, authors are not only interested Al2O3-based ceramic composites, great attention is in the dominant mechanism of deformation, but also try concentrated on three aspects: strain rates, micro- to find out which phase controls the creep behaviour. structural changes and deformation mechanisms Creep of composites can be modelled by using standard As concerns the first of these items one of the main aims is to investigate the relationship between the creep rates and operating variables(imposed stress, grain size, temperature)and to evaluate the creep parameters(n, P, Q). The deformation behaviours are critically dependent Nabarro Climb, Dp DIffusional Crae :88: 3. MOCELILIOEmGEAY STRESS, MPa Fig. 6. Idealized composite microstructures:(a) isostress and(b)iso- Fig. 5. Deformation map for MgO-doped Al2O3 at 1500C [9]. strain orientations

some basal slip could easily occur. Furthermore, when grain size of Al2O3 was between 50 and 500 mm and the imposed stress was large enough, then dislocation climb and glide could occur. But in practice, the theory and the experimental data used to construct the maps are poor or insucient for many ceramic materials, thus limiting their use in applications. At present, the main method to identify the possible rate controlling mechanism is to compare the values of n,p and Q obtained from experiments with theoretical predictions. Chokshi et al. [12] indicated that many ceramics exhibit stress exponents of 5, 3 or 1, which appeared to be associated with dislocation glide and climb, climb from Bardeen±Herring sources, and di€usion creep, respectively. The stress exponent of 2 might be due to the presence of a partially wetting grain boundary glassy phase or to control by an interface reaction. [12] Besides, the stress exponent between 1 and 3 might be associated with grain boundary sliding and grain rearrangement, and a higher stress exponent (n53) can also be associated with cavitation creep and microcracking. The inverse grain size exponent points to either a boundary mechanism (p51) or a lattice mechanism (p=0). In general, a direct observation and analysis of the microstructure of the specimens after deformation is necessary to check preliminary conclu￾sions drawn from the experimentally determined stress± strain rate relationships. For two phase composites, especially for those with duplex microstructures, authors are not only interested in the dominant mechanism of deformation, but also try to ®nd out which phase controls the creep behaviour. Creep of composites can be modelled by using standard composite theory: isostrain and isostress model [15,16] (Fig. 6). Isostrain and isostress prediction diagrams can be constructed by calculation. By comparing the experimental data with the isostrain and isostress pre￾diction, the model ®tting the creep deformation can be determined. Since the isostress model is dominated by the least creep resistant phase and highest creep rate and the isostrain model is dominated by the most creep resistant phase and lowest creep rate, the phase con￾trolling the creep behaviour can be determined. For the two phase composites in which both phases deform inelastically, a self-consistent model was devel￾oped [17,18], which predicts the deformation behaviour of the composites when the viscoplastic laws of each phase are known. By self-consistent calculations, e€ec￾tive strain rate sensitivity parameter and e€ective pre￾factor which are characteristic of the composite behaviour can be obtained. Stress and strain rates in each phase are also attainable. From the comparison between the model and the experiments, the possible deformation mechanisms of each phase can be deter￾mined and the phase controlling the creep behaviour can also be qualitatively determined. 3. Plastic deformation behaviours of Al2O3-based ceramic composites In studies of the plastic deformation behaviours of Al2O3-based ceramic composites, great attention is concentrated on three aspects: strain rates, micro￾structural changes and deformation mechanisms. As concerns the ®rst of these items, one of the main aims is to investigate the relationship between the creep rates and operating variables (imposed stress, grain size, temperature) and to evaluate the creep parameters (n, p, Q). The deformation behaviours are critically dependent Fig. 5. Deformation map for MgO-doped Al2O3 at 1500C [9]. Fig. 6. Idealized composite microstructures: (a) isostress and (b) iso￾strain orientations. 398 Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408

on the physical and chemical properties of reinforcing grain growth or cavitation. Fridez [21] has reported particles or whiskers, their content, morphologies and that creep deformation accelerated grain growth. The distributions, and microstructures of composites change in the crystallographic texture of alumina was including grain sizes and shapes, pores, grain bound observed at higher stress range. [21, 23] The main defor aries, interfaces, as well as stress and temperature. Dur- mation mechanisms were diffusional creep, grain ing deformation, there often occur microstructural boundary sliding and sometimes basal slip. The diffu changes such as grain growth, changes of grain shape sional creep can become inter rface-controlled at low (grain elongation), texture development, formation of stresses, causing non-Newtonian creep behaviour. The ntermediate or intergranular phase, dislocation activity, cavitation was often caused by unaccommodated GBS vacancy nucleation and evolution, cavitation and evo- or basal slip and the basal slip can give rise to a defor ution, formation and development of microcracks, etc. mation texture creep deformation is an essential aspect which li The study of such microstructural changes accompar In Al2O3-based ceramic particle or whisker compo- sites, the creep behaviours of Al2O3-ZrO2 and Al_O3 ortant foundation for analysing creep deformation SiC composites have been most extensively investigated behaviours and creep mechanisms of composites In the following the creep behaviours of Al2O3-based we will first recall briefly the high oxide ceramic composites and Al2O3-based nonoxide temperature deformation behaviours of fine-grained ceramic composites will be discussed. The correspond- alumina and then review the high temperature defor- ing experimental data produced during the past 10 years mation behaviours of Al2O3-based ceramic composites. or so in both families of materials are summarised in High temperature deformation behaviours of fine- Tables I and 2, respectively grained alumina were widely investigated. [5-9][18-23] Most authors have shown that n=1-2, P=2-3, and 3. 1. Al2O3-based oxide ceramic particle composites 0=430-500 KJ mol- can represent deformation data at lower stress range. The stress exponent generally 3. 1.1. A1203-ZrOz composites decreased with increasing grain size. In some studies. Since Wakai et al. reported that a 3 mol% yttria stabi- non-steady state deformation has been reported due to lized zirconia exhibited superplasticity [24, Al2O3-ZrO Table l Deformation behaviours of Al2O3-based oxide ceramic particle composites Reference Additive Content Test Atm. a ranges e ranges T(C) Grain size (um) n p Q (vol%) type (MPa) (S-) AlO3 additiv KJ mol-) Wakai et al Zro 72.7C 10=4-10-31400-1500 Kellett et a ZrO,20Cair30-10010-4-10-21500 0.72.1 Wakai et al.(1988)[27 14.3Tair10-10010-8-10-51250-1450100.51.7-2.1 Wakai et al. (1988)[28 ZrO,727Tair10-14010-7-10-51250-14500.50.52.1 590-600 Nieh et al.(1989)[29] ZrO,72.7 r vacuun54010-5-10-21450-1650 ~0.5b Wakai et al.(1989)[301 ZrO, Tair2-10010-7-10-31250-14500.60.521-24-720-780 ZrO230.8Tair2-10010-7-10-31250-14501.00.619-24-680-740 ZrO314.3Tair2-10010-7-10-31250-14501.0 0.5172.1 540-760 Wang et al. (1991)B31 ZrO 5 C argon5-20010-5-10-41400-150028 3630 HfO25 C argon5-20010-510-41400-150027 5 C argon5-6010--10-41400-15003.8 Owen et al. (1994)[32] 7 T 4-10010-8-10-31327-14770.4 0.42.82.1 French et al. (1994)[15] TZZY 50Tair35-7510-810-61200-1350 air35-7510-810-71210-1350 Calderon-Mereno et al. (1995)[33] ZrO2 5.5C 0-15010-8-10-41300-145047 14 Calderon-Mereno et al. (1995)[33 ZrO2 5.5Cair10-15010-810-41300145026<1 Chevalier et al. (1997)[34 ZrO, 10B Bair50-20010-810-6120014002 0.5 760 Calderon-Mereno(1997)[35] 40Cair45-8510-510-41400-15002.31.61.7 Calderon-Mereno(1997)[35] Zro 649510-5-10 Calderon-Mereno(1997)[35] Zro 10Cair70-12310-5-10-415002 Flacher et al.(1997)[36 ZrO,5-20Cair20-13010-5-10-41300-14000.2 Clarisse et al. (1997)[3 ZrO, 0Cair4-20010-6-10-31275-14000.640.551-2 640-705 Clarisse et al. (1997)[37 ZrO2380Cair420010-7-10-31275-14001.120.801-2 Clarisse et al. (1997)[37] ZrO2380Cair420010-7-10-41275-14001 0.711-2 663-715 Duong et al. ( 1993)[16 YAG50Cair3-2010-8-10-51400-150010 612 YAG 3-2010-8-10-51400-15008 592 compression, T=tension, B=bendin Average grain

on the physical and chemical properties of reinforcing particles or whiskers, their content, morphologies and distributions, and microstructures of composites including grain sizes and shapes, pores, grain bound￾aries, interfaces, as well as stress and temperature. Dur￾ing deformation, there often occur microstructural changes such as grain growth, changes of grain shape (grain elongation), texture development, formation of intermediate or intergranular phase, dislocation activity, vacancy nucleation and evolution, cavitation and evo￾lution, formation and development of microcracks, etc. The study of such microstructural changes accompany￾ing creep deformation is an essential aspect which is an important foundation for analysing creep deformation behaviours and creep mechanisms of composites. In this chapter, we will ®rst recall brie¯y the high temperature deformation behaviours of ®ne-grained alumina and then review the high temperature defor￾mation behaviours of Al2O3-based ceramic composites. High temperature deformation behaviours of ®ne￾grained alumina were widely investigated. [5±9] [18±23] Most authors have shown that n=1±2, p=2±3, and Q=430±500 KJ molÿ1 can represent deformation data at lower stress range. The stress exponent generally decreased with increasing grain size. In some studies, non-steady state deformation has been reported due to grain growth or cavitation. Fridez [21] has reported that creep deformation accelerated grain growth. The change in the crystallographic texture of alumina was observed at higher stress range. [21,23] The main defor￾mation mechanisms were di€usional creep, grain boundary sliding and sometimes basal slip. The di€u￾sional creep can become interface-controlled at low stresses, causing non-Newtonian creep behaviour. The cavitation was often caused by unaccommodated GBS or basal slip and the basal slip can give rise to a defor￾mation texture. In Al2O3-based ceramic particle or whisker compo￾sites, the creep behaviours of Al2O3-ZrO2 and Al2O3- SiC composites have been most extensively investigated. In the following the creep behaviours of Al2O3-based oxide ceramic composites and Al2O3-based nonoxide ceramic composites will be discussed. The correspond￾ing experimental data produced during the past 10 years or so in both families of materials are summarised in Tables 1 and 2, respectively. 3.1. Al2O3-based oxide ceramic particle composites 3.1.1. Al2O3-ZrO2 composites Since Wakai et al. reported that a 3 mol% yttria stabi￾lized zirconia exhibited superplasticity [24], Al2O3-ZrO2 Table 1 Deformation behaviours of Al2O3-based oxide ceramic particle composites Reference Additive Content (vol%) Test type a Atm.  ranges (MPa)  ranges (Sÿ1 ) T ( C) Grain size (m) Al2O3 additive np Q (KJ molÿ1 ) Wakai et al. (1986) [25] ZrO2 72.7 C air ± 10ÿ4 ±10ÿ3 1400±1500 ± ± 1.2±2.0 ± 620 Kellett et al. (1986) [26] ZrO2 20 C air 30±100 10ÿ4 ±10ÿ2 1500 1.1 0.7 2.1 ± ± Wakai et al. (1988) [27] ZrO2 14.3 T air 10±100 10ÿ8 ±10ÿ5 1250±1450 1.0 0.5 1.7±2.1 ± 750 Wakai et al. (1988) [28] ZrO2 72.7 T air 10±140 10ÿ7 ±10ÿ5 1250±1450 0.5 0.5 2.1 ± 590±600 Nieh et al. (1989) [29] ZrO2 72.7 T vacuum 5±40 10ÿ5 ±10ÿ2 1450±1650 0.5b 2± ± Wakai et al. (1989) [30] ZrO2 50 T air 2±100 10ÿ7 ±10ÿ3 1250±1450 0.6 0.5 2.1±2.4 ± 720±780 ZrO2 30.8 T air 2±100 10ÿ7 ±10ÿ3 1250±1450 1.0 0.6 1.9±2.4 ± 680±740 ZrO2 14.3 T air 2±100 10ÿ7 ±10ÿ3 1250±1450 1.0 0.5 1.7±2.1 ± 640±760 Wang et al. (1991) [31] ZrO2 5 C argon 5±200 10ÿ5 ±10ÿ4 1400±1500 2.8 ± ± 3 630 HfO2 5 C argon 5±200 10ÿ5 ±10ÿ4 1400±1500 2.7 ± ± 3 685 TiO2 5 C argon 5±60 10ÿ5 ±10ÿ4 1400±1500 3.8 ± ± 2 570 Owen et al. (1994) [32] ZrO2 72.7 T air 4±100 10ÿ8 ±10ÿ3 1327±1477 0.4 0.4 2.8 2.1 585 French et al. (1994) [15] ZrO2 50 T air 35±75 10ÿ8 ±10ÿ6 1200±1350 2.3 1.8 ± 633 YAG 50 T air 35±75 10ÿ8 ±10ÿ7 1210±1350 2.0 2.6 ± 695 Calderon-Mereno et al. (1995) [33] ZrO2 5.5 C air 10±150 10ÿ8 ±10ÿ4 1300±1450 4.7 ± 1.4 ± 580 Calderon-Mereno et al. (1995) [33] ZrO2 5.5 C air 10±150 10ÿ8 ±10ÿ4 1300±1450 2.6 <1 1.8 ± 540 Chevalier et al. (1997) [34] ZrO2 10 B air 50±200 10ÿ8 ±10ÿ6 1200±1400 2 0.5 2.5 ± 760 Calderon-Mereno (1997) [35] ZrO2 40 C air 45±85 10ÿ5 ±10ÿ4 1400±1500 2.3 1.6 1.7 ± ± Calderon-Mereno (1997) [35] ZrO2 20 C air 64±95 10ÿ5 ±10ÿ4 1500 2.3 1.6 1.4 ± ± Calderon-Mereno (1997) [35] ZrO2 10 C air 70±123 10ÿ5 ±10ÿ4 1500 2.3 1.6 1.2 ± ± Flacher et al. (1997) [36] ZrO2 5±20 C air 20±130 10ÿ5 ±10ÿ4 1300±1400 0.2Ä 2 ± 650 Clarisse et al. (1997) [37] ZrO2 80 C air 4±200 10ÿ6 ±10ÿ3 1275±1400 0.64 0.55 1±2 ± 640±705 Clarisse et al. (1997) [37] ZrO2 80 C air 4±200 10ÿ7 ±10ÿ3 1275±1400 1.12 0.80 1±2 ± 642±723 Clarisse et al. (1997) [37] ZrO2 80 C air 4±200 10ÿ7 ±10ÿ4 1275±1400 1.40 0.71 1±2 ± 663±715 Duong et al. (1993) [16] YAG 50 C air 3±20 10ÿ8 ±10ÿ5 1400±1500 10 3 1.1 ± 612 YAG 75 C air 3±20 10ÿ8 ±10ÿ5 1400±1500 8 3 1.1 ± 592 a C=compression, T=tension, B=bending. b Average grain size. Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408 399

Q. Tai. A. Mocellin/Ceramics International 25(1999)395-408 omposites have been extensively studied [15, 25-40 Most studies on Al,Ox-ZrO, composites indicated Wakai and Kano [28]demonstrated that a 3 mol% yttria that the introduction of rO2 in Al2O3 made the creep stabilized zirconia with 27. 3 vol% alumina composite rate of the composites lower than that of either of their (3Y20A)was capable of large superplastic elongations single-phase constituents. [15, 27[30-39] Wakai et al of more than 200%, while Nieh et al. [29] reported a [27, 30] and Chen [39] attributed it to the suppression of tensile elongation of 500% at 1650C for 3Y20A. The interface reactions controlled diffusion creep of Al2O3 superplasticity of the Al2O3-ZrO2 composites has been or hindrance of grain-boundary movement by reported by many other ceramic workers granular ZrO2 particles. French et al. [15] indicated that Inspection of all the available data of the Al2O3-ZrO2 the decrease of the creep rate was caused by a strong composites shown in Table I indicates that the values of segregation of Yt(and possibly Zr*t) to the alumina n range from about 1. 5 to 2.5. The activation energies grain boundaries, which resulted in a decrease of AlO3 tend to lie in the range 600-700 KJ mol-I. The values of in grain boundary dislocation mobility and interface inverse grain size exponent were only reported by Wang reaction controlling diffusional creep rate. The effect of et al. [3l] and Owen et al. [23](p=3, for 5 vol% ZrO content on the creep rate of composites was first p=2. 1 for 3Y20A) The differences in the values of n, 2 studied by Wakai et al. [30, 38](Fig. 7). The results and p are probably caused by the conditions of test, the showed that when the ZrO2 content was more than different stress, strain rate and temperature ranges, the 14.3 vol% the creep rate decreased with decrease of the ZrO2 content, the levels of impurities, and the micro- ZrO2 content. The values of n remained almost structure of the composites unchanged. Recent investigations 35-37 Table 2 Deformation behaviours of Al2O3-based non-oxide ceramic particle or whisker dditive Content Test Atm. g ranges Grain size (vol %) type (MPa) (KJ mol-) Chokshi et al 40-10010-7-10 1500 Porter et al 40-20010-7-10-31450-1600 (1987)[42] Lipetzky et al. 30-25010=10-10-51200-1300 450-500 (1988)43 9.3,18B 40-20010-1-10-4/4 40-20010-710 (1988)[44 B 6.3 37-30010-910 1990)[45] De arellano.l 6-30 C argon,air100-30010-8-10-41300-1500 (1991)[47 BBBB 37-30010-910-6 Lipetzky et al. 25-25010-910 13040 269,655 (1991)[48 25-25010--10 210.966 14-1.7 (1992)[49] DeArellano-Lopez 91400 0.9-1.9 etal.(1993)[50 Sic wy 15-30C 8041010-6-10-40491400 2.4-5.9 10-10010-7-10-41400-1550 820-830 (1995)[51] 50-23010-810-7 12-8 1996)[52] Ohji et al 994)[53 Deng et al BBTBBB aaaa 50-23010-8-10-71300128 50-15010-10-1 1200-1300 40-12510-910-51160-1400<50.6 (1996)[54 40-12510-910-51160-1400 Katsumura et al CosNo.so)54 B argon50-30010-6-10-31300-1400081.42022-281.5 (1987)[55 TiCal T vacuum 7010-5-10-21300-1550 3.2-4.1 (1991)

composites have been extensively studied [15, 25±40]. Wakai and Kano [28] demonstrated that a 3 mol% yttria stabilized zirconia with 27.3 vol% alumina composite (3Y20A) was capable of large superplastic elongations of more than 200%, while Nieh et al. [29] reported a tensile elongation of 500% at 1650C for 3Y20A. The superplasticity of the Al2O3-ZrO2 composites has been reported by many other ceramic workers. Inspection of all the available data of the Al2O3-ZrO2 composites shown in Table 1 indicates that the values of n range from about 1.5 to 2.5. The activation energies tend to lie in the range 600±700 KJ molÿ1 . The values of inverse grain size exponent were only reported by Wang et al. [31] and Owen et al. [23] (p=3, for 5 vol% ZrO2; p=2.1 for 3Y20A). The di€erences in the values of n, Q and p are probably caused by the conditions of test, the di€erent stress, strain rate and temperature ranges, the ZrO2 content, the levels of impurities, and the micro￾structure of the composites. Most studies on Al2O3-ZrO2 composites indicated that the introduction of ZrO2 in Al2O3 made the creep rate of the composites lower than that of either of their single-phase constituents. [15,27] [30±39] Wakai et al. [27,30] and Chen [39] attributed it to the suppression of interface reactions controlled di€usion creep of Al2O3 or hindrance of grain-boundary movement by inter￾granular ZrO2 particles. French et al. [15] indicated that the decrease of the creep rate was caused by a strong segregation of Y3+ (and possibly Zr4+) to the alumina grain boundaries, which resulted in a decrease of Al2O3 in grain boundary dislocation mobility and interface reaction controlling di€usional creep rate. The e€ect of ZrO2 content on the creep rate of composites was ®rst studied by Wakai et al. [30,38] (Fig. 7). The results showed that when the ZrO2 content was more than 14.3 vol% the creep rate decreased with decrease of the ZrO2 content. The values of n remained almost unchanged. Recent investigations [35±37] showed Table 2 Deformation behaviours of Al2O3-based non-oxide ceramic particle or whisker composites Reference Additive Content (vol%) Test type Atm.  ranges (MPa)  ranges (Sÿ1 ) T ( C) Grain size (m) Al2O3additive np Q (KJ molÿ1 ) Chokshi et al. (1985) [41] SiC(w)a 18 B air 40±100 10ÿ7 ±10ÿ4 1500 45 ± 5.2 ± ± Porter et al. (1987) [42] SiC(w) 5±20 B air 40±200 10ÿ7 ±10ÿ3 1450±1600 ± ± 5 ± 450 Lipetzky et al. (1988) [43] SiC(w) 33 B air 30±250 10ÿ10±10ÿ5 1200±1300 1±2 ± 1,5 ± 450±500 Xia et al. SiC(w) 9.3,18 B air 40±200 10ÿ7 ±10ÿ4 1400±1550 1±2 ± 3.8 ± ± (1988) [44] SiC(w) 30 B air 40±200 10ÿ7 ±10ÿ4 1400±1550 1±2 ± 6.3 ± ± Lin et al. SiC(w) 20 B air 37±300 10ÿ9 ±10ÿ5 1200±1400 2 ± 2 ± ± (1990) [45] SiC(w) 20 B air >125 10ÿ9 ±10ÿ5 1400 2 ± 7±8 ± ± DeArellano-Lopez et al. (1990) [46] SiC(w) 6±30 C argon, air 100±300 10ÿ8 ±10ÿ4 1300±1500 ± ± 1.2±1.8 ± 620 Lin et al. SiC(w) 30±50 B air 37±300 10ÿ9 ±10ÿ5 1300 1-2 ± 6 ± ± (1991) [47] SiC(w) 30±50 B air 37±300 10ÿ9 ±10ÿ5 1200 1-2 ± 3 ± ± SiC(w) 10 B air 37±300 10ÿ9 ±10ÿ6 1300 8 ± 4 ± ± SiC(w) 10 B air 37±300 10ÿ9 ±10ÿ6 1200 8 ± 2 ± ± Lipetzky et al. SiC(w) 33 C air 25±250 10ÿ9 ±10ÿ6 1200±1400 1±2 ± 1,3 ± 269, 655 (1991) [48] SiC(w) 33 C nitrogen 25±250 10ÿ9 ±10ÿ6 1200±1400 1±2 ± 1,3 ± 210, 966 Swan et al. (1992) [49] SiC(w) 30 C air 25±200 10ÿ8 ±10ÿ5 1200±1350 1±2 ± 1.4±1.7 ± 370 DeArellano-Lopez SiC(w) 5±30 C argon 10±240 10ÿ7 ±10ÿ5 0491400 ± ± 0.9±1.9 ± ± et al. (1993) [50] SiC(w) 15±30 C argon 80±410 10ÿ6 ±10ÿ4 0491400 ± ± 2.4±5.9 ± ± Xia et al. (1995) [51] SiC(w) 9.3 B air 10±100 10ÿ7 ±10ÿ4 1400±1550 1±2 ± 3.8 ± 820±830 Lin et al. SiC(w) 10 B air 50±230 10ÿ8 ±10ÿ7 1200 1.2±8 ± 2 1 ± (1996) [52] SiC(w) 10 B air 50±230 10ÿ8 ±10ÿ7 1300 1.2±8 ± 3.5 ± ± Ohji et al. SiC(p) 17 T air 50±150 10ÿ10±10ÿ7 1200±1300 2 0.1 3.1 ± ± (1994) [53] SiC(p) 17 B air 100±200 10ÿ11±10ÿ9 1200 2 0.1 2.2 ± ± Deng et al. SiC(p) 10 B air 40±125 10ÿ9 ±10ÿ5 1160±1400 <5 0.6 4.27 ± 444 (1996) [54] SiC(p) 10 B air 40±125 10ÿ9 ±10ÿ5 1160±1400 55 2.7 4.75 ± 666 Katsumura et al. (1987) [55] TiC0.5N0.5(p) 54 B argon 50±300 10ÿ6 ±10ÿ3 1300±1400 0.8 1.4 2.0±2.2 ± 281.5 Nagano et al. (1991) [56] TiC(p) 26 T vacuum 8±70 10ÿ5 ±10ÿ2 1300±1550 1.2Ä 3.2±4.1 ± 853 a : (w)=whiskers, (p)=particles 400 Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408

Q. Tai. A. Mocellin/Ceramics International 25(1999)395-408 almost the same results as those obtained by Wakai et Microstructural characterization of the deformed al. But the results obtained by Clarisse et al. [37] showed specimens generally revealed remarkable structural sta that when Zro2 content exceeded that of Al2O3, the bility as long as the second phase was present at a small creep rate of the composites was somewhat higher than volume fraction, typically no more than 10%[2]. Chen that of single-phase AlO3 due to a more ductile Zro2 attributed this to the particle pinning effect [2]. The phase which played a role in the deformation of the com- measurements showed that there was no or very limited posites. The values of n obtained by Calderon-Moreno concurrent grain growth during deformation [35]showed a slight increase with the ZrOz content [15, 28, 29, 32, 33] and there was no significant change in Wang et al. [31] and Owen et al. [32] studied the the aspect ratio of the grains, which essentially retained influence of grain size on the creep behaviours of their equiaxed shapes [29, 33]. Wakai et al. [28] and Al2O, composites. The values of p obtained by Owen et al. [32] observed that in 3Y20A there was very Wang et al. was about 3, while that obtained by Owen little increase in the aspect ratio of the two phases fol- et al. was 2. 1. The difference of the values of p is mainly lowing deformation, the grains were elongated slightly in caused by the content of ZrOz. In the experiment of the tensile direction. The grain aspect ratio of ZrO2 Wang et al. the ZrO2 content was 5 vol%, but in the grains was 1. 15-1. 16, and that of Al2O3 grains was 1.3- experiment of Owen et al. the ZrO2 content was 1.5. This implies that the intragranular strain of ZrO2 72.7 vol%, and in this two phase composite the values grains is smaller than that of Al_O3 grains. In some of p of ZrO,, AlO3 phase and their volume average experiments, neither significant intragranular dislocation obtained by Owen et al. were 2.0, 2.6 and 2. 1 respec- activity nor significant cavitation was observed [15, 3 tively(Fig. 8). So, the addition of Zro, decreases the but in most experiments, cavitation was noted [28, 33, 34] values of p of Al_O - ZrO2 composites. The influence of Cavities tended to nucleate at triple point junctions grain size on the creep behaviours was also investigated associated with an alumina grain and then they grew by other authors [33, 34, 37]. The impurity content of quite quickly along those Al2O3-Al2O3, ZrO2-ZrOz and AlO3 and ZrO? and the level of segregation at grain Al2O3-ZrO2 interfaces normal to the tensile axis. the boundaries may affect strongly the creep behaviours of cavity volume fraction and size were dependent on the he composites. Generally, the higher the content of Si, strain rate, both decreasing with decreasing strain rate Fe, Na impurity, the higher the creep rate of the com- posite. Although French et al. [15 indicated that o Many authors ruled out the deformation mechanisms intragranular dislocation or significant contributions Al2O3/c-ZrO2(8 mol%Y2O3)composite the segregation of y+ to the Al2O3 grain boundaries slowed the creep te of this composite the results obtained by Chevalier et al. [34] showed that magnesia-stabilized ZrO2-Al2O3 composite decreased the creep rate while yttria-stabi lised zirconia was not favourable for creep resistance of composite. The authors sium-containing grain boundary glassy phase was favourable for creep resistance of the composite. The role of the y+ and Mg2+ at grain boundaries needs to be more systematically investigated rrw ▲3TG Llm) l Fig. 8. Variation in creep rate with grain size for Al2Or-727 vol% ZrO2 composite. The grain size may be defined in terms of the ZrO2 Fig. 7. Influence of ZrO2 content on the strain rate in Al2O3-zrO2 or Al2O3 phases or their volume average. p:(0)ZrO2, 2.0:(A)Al2O3, composites [38]. 6:(o)Av.2.1[2J

almost the same results as those obtained by Wakai et al. But the results obtained by Clarisse et al. [37] showed that when ZrO2 content exceeded that of Al2O3, the creep rate of the composites was somewhat higher than that of single-phase Al2O3 due to a more ductile ZrO2 phase which played a role in the deformation of the com￾posites. The values of n obtained by Calderon±Moreno [35] showed a slight increase with the ZrO2 content. Wang et al. [31] and Owen et al. [32] studied the in¯uence of grain size on the creep behaviours of Al2O3±ZrO2 composites. The values of p obtained by Wang et al. was about 3, while that obtained by Owen et al. was 2.1. The di€erence of the values of p is mainly caused by the content of ZrO2. In the experiment of Wang et al. the ZrO2 content was 5 vol%, but in the experiment of Owen et al. the ZrO2 content was 72.7 vol%, and in this two phase composite the values of p of ZrO2, Al2O3 phase and their volume average obtained by Owen et al. were 2.0, 2.6 and 2.1 respec￾tively (Fig. 8). So, the addition of ZrO2 decreases the values of p of Al2O3±ZrO2 composites. The in¯uence of grain size on the creep behaviours was also investigated by other authors [33,34,37]. The impurity content of Al2O3 and ZrO2 and the level of segregation at grain boundaries may a€ect strongly the creep behaviours of the composites. Generally, the higher the content of Si, Fe, Na impurity, the higher the creep rate of the com￾posite. Although French et al. [15] indicated that in Al2O3/c-ZrO2 (8 mol% Y2O3) composite the segregation of Y3+ to the Al2O3 grain boundaries slowed the creep rate of this composite, the results obtained by Chevalier et al. [34] showed that magnesia-stabilized ZrO2-Al2O3 composite decreased the creep rate while yttria-stabi￾lised zirconia was not favourable for creep resistance of the composite. The authors suggested that a magne￾sium-containing grain boundary glassy phase was favourable for creep resistance of the composite. The role of the Y3+ and Mg2+ at grain boundaries needs to be more systematically investigated. Microstructural characterization of the deformed specimens generally revealed remarkable structural sta￾bility as long as the second phase was present at a small volume fraction, typically no more than 10% [2]. Chen attributed this to the particle pinning e€ect [2]. The measurements showed that there was no or very limited concurrent grain growth during deformation [15,28,29,32,33] and there was no signi®cant change in the aspect ratio of the grains, which essentially retained their equiaxed shapes [29,33]. Wakai et al. [28] and Owen et al. [32] observed that in 3Y20A there was very little increase in the aspect ratio of the two phases fol￾lowing deformation, the grains were elongated slightly in the tensile direction. The grain aspect ratio of ZrO2 grains was 1.15±1.16, and that of Al2O3 grains was 1.3± 1.5. This implies that the intragranular strain of ZrO2 grains is smaller than that of Al2O3 grains. In some experiments, neither signi®cant intragranular dislocation activity nor signi®cant cavitation was observed [15,32], but in most experiments, cavitation was noted [28,33,34]. Cavities tended to nucleate at triple point junctions associated with an alumina grain and then they grew quite quickly along those Al2O3±Al2O3, ZrO2±ZrO2 and Al2O3±ZrO2 interfaces normal to the tensile axis. The cavity volume fraction and size were dependent on the strain rate, both decreasing with decreasing strain rate. Many authors ruled out the deformation mechanisms of intragranular dislocation or signi®cant contributions Fig. 7. In¯uence of ZrO2 content on the strain rate in Al2O3-ZrO2 composites [38]. Fig. 8. Variation in creep rate with grain size for Al2O3±72.7 vol% ZrO2 composite. The grain size may be de®ned in terms of the ZrO2, or Al2O3 phases or their volume average. p: (&) ZrO2, 2.0; (
) Al2O3, 2.6; (o) Av. 2.1 [32]. Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408 401

Q. Tai. A. Mocellin/Ceramics International 25(1999)393-408 from diffusion creep by comparison of the stress expo-- nent with theoretical predictions and the observation of 1250°c microstructures of Al,OxZrO composites. They attributed the deformation to some form of grain 10 boundary sliding and grain rearrangement process [27 YAG 29, 32-37. Owen et al. indicated that in their study of 3Y20A the grain boundary sliding was mostly of the 10 AY50 Rachinger type [32]. Calderon-Moreno et al. pointed F out that grain boundary sliding with a mix of irc d (intertface-reaction controlled) and TMc (transport of 10 ism in their study [33]. Wakai et al. [30] indicated that 5070100 200 when the content of Zro, was very small and the size of Stress. MPa ZrO2 grains was much smaller than that of Al2O3 grains, the interface-reaction controlled diffusion creep Fig. 9. Variation in strain rate with stress for AlO3, Y3Al5O12 and could be the main creep mechanism. Clarisse et al. pro- Al203-50vol% Y3AlsOi2 composite [51 posed that the main mechanism was grain boundary sliding accommodated by either grain boundary difft sion at high stress or an interface reaction at low stress 695 KJ mol-, respectively. While those obtained by in their study [37]. While in the study of Chevalier et al. Duong et al. were 1. 1, 612 KJ mol-, respectively. The [34]. it is proposed that cavitation and microcracking by difference is possibly caused by different test types and grain boundary sliding was the main mechanism. This is range of applied stress. The formers performed their possibly related to their type of test (bending) and tests under tension at higher stress(25-75 MPa), while higher effective stresses he latters tests under uniaxial compression at lower From those investigations mentioned above, it can stress(3-20 MPa). The experiments of Duong et al be seen that the introduction of ZrO, in Al,O3 can showed that the grain size rather than Y3AlsO1? content improve the creep resistance of Al2O3 matrix either by played an important role in the creep behaviour, and the suppression of interface reactions controlled difft the value of the stress exponent was independent of sion creep of Al2O3 or by the hindrance of grain- tempera boundary movement. In some conditions, the creep rate The observation of microstructure of the AlO3- increases but the value of n remains almost unchanged Y3AlsO12 composites showed that no dynamic grain or increases slightly with increasing the ZrO2 content. growth occurred as a result of the deformation [15,16] The addition of ZrO2 can decrease the values of p of The grains remained fairly equiaxed after deformation Al2O, composites. The impurities of Si, Fe, Na at The evidence of cavitation was observed by Duong et grain boundaries favour grain boundary sliding, a al, the amount of cavitation was estimated to be higher level of such impurities results in a higher creep between 2 and 5%[16]. While French et al. did not rate of the composite. But the effect of Y3+ or Mg2+ on observe significant cavitation. The former authors per- the creep rate of the composites is ambiguous, further formed their tests at more elevated temperatures Investigation is needed. Microstructures of the temperature may favour the nucleation and sites show considerable stability due to the ZrO2 particle growth of cavities pinning effect and there is no or very limited concurrent As far as the deformation mechanism is concerned grain growth during deformation. Cavitation caused by Duong et al. suggested that the creep behaviour was unaccommodated grain boundary sliding often occurs controlled by a diffusional Nabarro-Herring mechan during deformation. The main creep mechanism is grain ism [16. French et al. proposed a diffusional creep con- boundary sliding and grain rearrangement trolled by an interface reaction (source-controlled) mechanism [15] The creep data correlated well with the 3.1.2. A72O3-Y3AlsO12 composites predicted behaviour based on the isostress model [16] The studies of the behaviour of the Al2O3- Thus, the authors suggested that the creep behaviour Y3AlsO12 system showed that the strain rate of the was most probably controlled by Y3Al5O12 in their composites was lower than for pure AlO3 or Y3Al5O12 conditions [15, 16(Fig. 9). French et al. attributed it to the strong segregation of Y3+ to interfaces hindering interface 3. 1.3. A120rTiO2 composite reaction controlling diffusional rate [15]. Duong et al A study of the creep behaviour of Al2O3-TiO2 com indicated that the creep rate was controlled by the posite was reported by Wang et al. [31] The addition of Y3AlsO12 phase which is more creep-resistant. [16] The 5 vol% TiO2 into Al2O3 matrix obviously enhanced values of n and obtained by French et al. were 2.6, its creep rate. This composite exhibited excellent

from di€usion creep by comparison of the stress expo￾nent with theoretical predictions and the observation of microstructures of Al2O3±ZrO2 composites. They attributed the deformation to some form of grain boundary sliding and grain rearrangement process [27± 29,32±37]. Owen et al. indicated that in their study of 3Y20A the grain boundary sliding was mostly of the Rachinger type [32]. Calderon±Moreno et al. pointed out that grain boundary sliding with a mix of IRC (interface±reaction controlled) and TMC (transport of matter controlled) was the main deformation mechan￾ism in their study [33]. Wakai et al. [30] indicated that when the content of ZrO2 was very small and the size of ZrO2 grains was much smaller than that of Al2O3 grains, the interface±reaction controlled di€usion creep could be the main creep mechanism. Clarisse et al. pro￾posed that the main mechanism was grain boundary sliding accommodated by either grain boundary di€u￾sion at high stress or an interface reaction at low stress in their study [37]. While in the study of Chevalier et al. [34], it is proposed that cavitation and microcracking by grain boundary sliding was the main mechanism. This is possibly related to their type of test (bending) and higher e€ective stresses. From those investigations mentioned above, it can be seen that the introduction of ZrO2 in Al2O3 can improve the creep resistance of Al2O3 matrix either by the suppression of interface reactions controlled di€u￾sion creep of Al2O3 or by the hindrance of grain￾boundary movement. In some conditions, the creep rate increases but the value of n remains almost unchanged or increases slightly with increasing the ZrO2 content. The addition of ZrO2 can decrease the values of p of Al2O3±ZrO2 composites. The impurities of Si, Fe, Na at grain boundaries favour grain boundary sliding, a higher level of such impurities results in a higher creep rate of the composite. But the e€ect of Y3+ or Mg2+ on the creep rate of the composites is ambiguous, further investigation is needed. Microstructures of the compo￾sites show considerable stability due to the ZrO2 particle pinning e€ect and there is no or very limited concurrent grain growth during deformation. Cavitation caused by unaccommodated grain boundary sliding often occurs during deformation. The main creep mechanism is grain boundary sliding and grain rearrangement. 3.1.2. Al2O3-Y3Al5O12 composites The studies of the creep behaviour of the Al2O3± Y3Al5O12 system showed that the strain rate of the composites was lower than for pure Al2O3 or Y3Al5O12 [15,16] (Fig. 9). French et al. attributed it to the strong segregation of Y3+ to interfaces hindering interface reaction controlling di€usional rate [15]. Duong et al. indicated that the creep rate was controlled by the Y3Al5O12 phase which is more creep-resistant. [16] The values of n and Q obtained by French et al. were 2.6, 695 KJ molÿ1 , respectively. While those obtained by Duong et al. were 1.1, 612 KJ molÿ1 , respectively. The di€erence is possibly caused by di€erent test types and range of applied stress. The formers performed their tests under tension at higher stress (25±75 MPa), while the latter's tests under uniaxial compression at lower stress (3±20 MPa). The experiments of Duong et al. showed that the grain size rather than Y3Al5O12 content played an important role in the creep behaviour, and the value of the stress exponent was independent of temperature. The observation of microstructure of the Al2O3± Y3Al5O12 composites showed that no dynamic grain growth occurred as a result of the deformation. [15,16] The grains remained fairly equiaxed after deformation. The evidence of cavitation was observed by Duong et al., the amount of cavitation was estimated to be between 2 and 5% [16]. While French et al. did not observe signi®cant cavitation. The former authors per￾formed their tests at more elevated temperatures. Higher temperature may favour the nucleation and growth of cavities. As far as the deformation mechanism is concerned, Duong et al. suggested that the creep behaviour was controlled by a di€usional Nabarro±Herring mechan￾ism [16]. French et al. proposed a di€usional creep con￾trolled by an interface reaction (source±controlled) mechanism. [15] The creep data correlated well with the predicted behaviour based on the isostress model [16]. Thus, the authors suggested that the creep behaviour was most probably controlled by Y3Al5O12 in their conditions. 3.1.3. Al2O3±TiO2 composite A study of the creep behaviour of Al2O3±TiO2 com￾posite was reported by Wang et al. [31] The addition of 5 vol% TiO2 into Al2O3 matrix obviously enhanced its creep rate. This composite exhibited excellent Fig. 9. Variation in strain rate with stress for Al2O3, Y3Al5O12 and Al2O3-50 vol% Y3Al5O12 composite [15]. 402 Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408

Q. Tai. A. Mocellin/Ceramics International 25(1999)393-408 ties because of its lower flow stress and stronger interfacial cohesion. The apparent activa tion energy was 570 KJ mol-I, which is higher than that of Al,O3(420 KJ mol-)obtained by the authors. The authors suggested that an increase in the boundary dif- usion activation energy due to dopants was correlated to the corresponding decrease in the interfacial energy that is the cohesion at the interface increased with the activation energy. Observation of the influence of grain size on the creep behaviour showed that the addition of TiO, into Al,O3 made the values of p decrease from 3 to 2. This means that the addition of titanium increased the lattice cation vacancy concentration of Al2O3 matrix, so that the lat tice diffusion of aluminium overtook boundary diffu- 100 sion, thus resulting in the change of the values of p STRESS( MPa) It must be pointed out that the Al2O3-TiO2 compo- Fig. 10. Influence of stress on stress exponent in Al2 w, com- site shows a poor high temperature fluxural strength posites(+)SiC content: 0, 10-40 MPa, n=1.3;(A)SiC content because titania forms a nearly continuous network of 20-240MPa,n=18;(■) Sic content:15,30-80MPa,n=0.8,125- aluminium- titanate at the grain boundary in the as-sin 410MPa,n=3.4;(●) SiC content:30,30-165MPa,n=0.9,170-370 tered specimens and this compound deteriorates the MPa,n=5.9[50 fracture strength of the Al2O3-TiO2 composite 3. 2. A2O based non-oxide ceramic particle or whisker rate. De Arellano-Lopez et al. [50] introduced a critical composites stress level ae which depended on the content and impurities of whiskers, as well as the test conditions 3.2.1.A120r-SiC(w)composites When stresses were larger than ac, the stress exponent Since the initial work on the creep behaviour of changed from its lower value to higher value. Higher Al2O SiC(w) composites reported by Chokshi and Por- content, higher purity of whiskers, lower impurity of the ter [41], many studies on such materials have been specimens and an inert atmosphere in tests might lead to made. Generally the SiC whisker diameters were in the a higher value of de range 0 I to I um and their aspect ratio about 10-20. Experiments on the influence of temperature on the Almost all the studies on Al2O3-SiC(w) composites creep behaviours showed that a higher temperature ( Table 2) have shown that their creep resistance was often resulted in a higher value of n [41-48, 51]. The generally far superior to that of the unreinforced matrix stress exponents for Al2O3-33 vol% SiC(w) obtained by [41-52. The creep rate was found to be one or two Lipetzky et al changed from I at 1200C to 3 at 1300- orders of magnitude lower than that of Al2O3 matrix 1400 C[48](Fig. 11). The stress exponents for different [45, 49. Most authors attributed the improvement in content(10-50 vol%)SiC(w) composites obtained by Lin reep resistance to the whiskers which act as hard pin- et al. varied from 2-3 at 1200C to 3. 5-6 at 1300C ning particles on the grain boundary surfaces and as (45, 47, 52 Higher values of n(3.8-6.3) were also hard particles penetrating across the boundary planes, obtained by Chokshi et al., [41] Porter et al. [42]and Xia thus retarding creep deformation by grain boundary et al. [44, 51]. Their experiments were all performed at sliding [45-50, 52 higher temperatures (1400-1600C) using four-point Inspection of the values of the stress exponent of bending creep tests. Although the values of n obtained Al2O-SiCw) composites shown in Table 2 indicates by DeArellano-Lopez et al. [146] and Swan et al. [491 that the values of n vary from I to 7-8, depending upon using compression tests were lower(1-1. 8), their results the method of test, the stress, temperature ranges, the also showed that a higher temperature favoured a SiC(w) content, and so on. The variation of the stress higher value of n. The increase of n at higher tempera- range often resulted in the change of the values of n ture is attributed to two factors: more extensive cavita [43, 45, 50]. From levels of about 1-2 at lower stresses, tion and crack caused by the stress concentrations they increased to 5-7 at higher stresses(Fig. 10). This is resulted from thermal mismatch and more glassy phases attributed to a change in the creep mechanism. Higher at grain boundaries caused by the thermal oxidation of stresses often lead to extensive cavitation occurring SiC whiskers within glass pockets at interfaces and grain boundaries, The effect of the content of SiC whiskers on the creep and crack generation which causes matrix grains to behaviours was studied by some authors. When the separate from the whiskers, thus increasing the creep whisker content was <20 vol%, the creep resistance

superplastic properties because of its lower ¯ow stress and stronger interfacial cohesion. The apparent activa￾tion energy was 570 KJ molÿ1 , which is higher than that of Al2O3 (420 KJ molÿ1 ) obtained by the authors. The authors suggested that an increase in the boundary dif￾fusion activation energy due to dopants was correlated to the corresponding decrease in the interfacial energy, that is the cohesion at the interface increased with the activation energy. Observation of the in¯uence of grain size on the creep behaviour showed that the addition of TiO2 into Al2O3 made the values of p decrease from 3 to 2. This means that the addition of titanium increased the lattice cation vacancy concentration of Al2O3 matrix, so that the lat￾tice di€usion of aluminium overtook boundary di€u￾sion, thus resulting in the change of the values of p. It must be pointed out that the Al2O3±TiO2 compo￾site shows a poor high temperature ¯uxural strength because titania forms a nearly continuous network of aluminium-titanate at the grain boundary in the as-sin￾tered specimens and this compound deteriorates the fracture strength of the Al2O3±TiO2 composite. 3.2. Al2O3±based non-oxide ceramic particle or whisker composites 3.2.1. Al2O3±SiC(w) composites Since the initial work on the creep behaviour of Al2O3±SiC(w) composites reported by Chokshi and Por￾ter [41], many studies on such materials have been made. Generally the SiC whisker diameters were in the range 0.1 to 1m and their aspect ratio about 10±20. Almost all the studies on Al2O3±SiC(w) composites (Table 2) have shown that their creep resistance was generally far superior to that of the unreinforced matrix [41±52]. The creep rate was found to be one or two orders of magnitude lower than that of Al2O3 matrix [45,49]. Most authors attributed the improvement in creep resistance to the whiskers which act as hard pin￾ning particles on the grain boundary surfaces and as hard particles penetrating across the boundary planes, thus retarding creep deformation by grain boundary sliding [45±50,52]. Inspection of the values of the stress exponent of Al2O3±SiC(w) composites shown in Table 2 indicates that the values of n vary from 1 to 7±8, depending upon the method of test, the stress, temperature ranges, the SiC(w) content, and so on. The variation of the stress range often resulted in the change of the values of n [43,45,50]. From levels of about 1±2 at lower stresses, they increased to 5±7 at higher stresses (Fig. 10). This is attributed to a change in the creep mechanism. Higher stresses often lead to extensive cavitation occurring within glass pockets at interfaces and grain boundaries, and crack generation which causes matrix grains to separate from the whiskers, thus increasing the creep rate. DeArellano-Lopez et al. [50] introduced a critical stress level c which depended on the content and impurities of whiskers, as well as the test conditions. When stresses were larger than c, the stress exponent changed from its lower value to higher value. Higher content, higher purity of whiskers, lower impurity of the specimens and an inert atmosphere in tests might lead to a higher value of c. Experiments on the in¯uence of temperature on the creep behaviours showed that a higher temperature often resulted in a higher value of n [41±48,51]. The stress exponents for Al2O3-33 vol% SiC(w) obtained by Lipetzky et al. changed from 1 at 1200C to 3 at 1300± 1400C [48] (Fig. 11). The stress exponents for di€erent content (10±50 vol%) SiC(w) composites obtained by Lin et al. varied from 2±3 at 1200C to 3.5±6 at 1300C. [45,47,52] Higher values of n (3.8±6.3) were also obtained by Chokshi et al., [41] Porter et al. [42] and Xia et al. [44,51]. Their experiments were all performed at higher temperatures (1400±1600C) using four-point bending creep tests. Although the values of n obtained by DeArellano-Lopez et al. [46] and Swan et al. [49] using compression tests were lower (1±1.8), their results also showed that a higher temperature favoured a higher value of n. The increase of n at higher tempera￾ture is attributed to two factors: more extensive cavita￾tion and crack caused by the stress concentrations resulted from thermal mismatch and more glassy phases at grain boundaries caused by the thermal oxidation of SiC whiskers. The e€ect of the content of SiC whiskers on the creep behaviours was studied by some authors. When the whisker content was 420 vol%, the creep resistance Fig. 10. In¯uence of stress on stress exponent in Al2O3-SiC(w) com￾posites (+) SiC content: 0, 10±40 MPa, n=1.3; (~) SiC content: 5, 20±240 MPa, n=1.8; (&) SiC content: 15, 30±80 MPa, n=0.8, 125± 410 MPa, n=3.4; (*) SiC content: 30, 30±165 MPa, n=0.9, 170±370 MPa, n=5.9 [50]. Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408 403

Q. Tai. A. Mocellin/Ceramics International 25(1999)395-408 Microstructural observation of creep specimens gen 1400°c erally revealed more or less cavitation at interfaces grain boundaries and triple joint junctions. [ 43, 45- 苏 1200C 60, 52 Unaccommodated grain boundary sliding wa considered to be responsible for the formation of cav- ities. Higher stresses and higher temperatures often increased the amount of intergranular voids, cavities and cracks, and sometimes caused the separation of interfaces and grain boundaries [43, 48]. The Al2O3- SiC(w) composites exhibited higher number density and 言 smaller cavities than those of monolithic AlO3 [45] One study showed that 30 and 50 vol% SiC(w)comp sites exhibited cavity number density 2 orders of mag- nitude higher than that of Al2O3-20 vol%S 10 composite. The average cavity size of 30 and 50 vol% Appfied Stress(MPa) SiC(w) composites was 0.4 um, while that of 20 vol% composite was 0.05 um [47]. In the four-point bending Fig. 11. Influence of temperature on stress exponent in an AlOr tests, the cavity number density was approximately 1 to 2 orders of magnitude in the compressive surface region less than that in the tensile surface region [47], and at higher temperature, cracklike cavities tended to form increased with increasing whisker content [42, 46, 47] and extensive macroscopic tensile surface cracks, and the the values of n were lower [46, 47, but when the whisker number of surface cracks increased with the grain size of content exceeded 30 vol%, the creep rates and the values AlO3 [52] of n increased due to(1) the promotion of creep cavita Although a glassy phase was present in very limit tion and crack generation from the higher number den- amounts in some as-sintered composites, many authors sity of nucleation sites, and (2)more extensive [43, 46, 47, 49, 52] observed that after creep deformation formation of grain boundary glassy phase in air ambient a SiO2-rich glassy phase around the The effect of test ambient on creep behaviours is an whiskers had formed due to the oxidation of the latter important issue since the Sic whiskers are easily oxi- In most cases, the glassy phase penetrated along grain dized in air at elevated temperature Experiments [46, 48 boundaries and interfaces and accumulated at triple showed that although the stress exponents were almost grain junctions throughout the materials, which pre the same in different ambients, the creep rates were sumably facilitated grain boundary sliding and higher in air than in inert atmosphere. Lin et al. [52 increased the creep rate have investigated the effect of matrix grain size(varying Dislocation networks were observed by some authors from 1. 2 to 8.0 um)on the creep behaviours. At 1200C [42, 43, 46, 49, 51]. But in most cases, the dislocation den the creep rate exhibited an inverse grain size exponent of sity was very low [42, 43, 46, 49] and the dislocation net approximately l, but at 1300C the creep rate was not works were also observed in the as-sintered specimens sensitive to the grain size due to enhanced nucleation They probably formed as a result of the large residual and coalescence of creep cavities and the development stresses resulting from thermal expansion mismatch of macroscopic cracks as the grain size increased. In during the fabrication processing. Contrary to other addition, the creep rate of Al2O3-SiC(w) composites was authors, Xia et al. [51] observed extensive dislocation accelerated by the introduction of certain additives (i.e. networks in the crept specimens, and higher dislocation Y2O3)[47. The presence of an intergranular glassy densities were observed in specimens deformed to large phase introduced by the Y2O3 additive facilitated creep strains deformation resulting in an order of magnitude increase Some authors observed the grain offset and rotation in creep rate. Finally, it should be pointed out that of Al2O3 [45, 52. As far as the creep mechanism is con- the method of creep testing may also influence the cerned, in early studies, a dislocation-controlled creep experimental results. In four-point bending creep mechanism was proposed by some authors [(41, 42, 44 tests, the creep rate is strongly affected by the surface But no sufficient microstructural information was pre- tensile properties of the specimens. The growth of sented to support this assertion, and this mechanism for surface cracks can result in a creep exponent of 2 the deformed Al2O3-SiC(w) composites remains doubt or higher when the growth rate of microcrack obeys ful. A more recent study [51]revealed extensive disloca- a power-law dependence on the local normal stress. tion activity and very little evidence for the development This mode of testing can result in large errors in the of initial cavitation. The authors proposed an intra- stress exponent [49] granular dislocation mechanism controlled by lattice

increased with increasing whisker content [42,46,47] and the values of n were lower [46,47], but when the whisker content exceeded 30 vol%, the creep rates and the values of n increased due to (1) the promotion of creep cavita￾tion and crack generation from the higher number den￾sity of nucleation sites, and (2) more extensive formation of grain boundary glassy phase. The e€ect of test ambient on creep behaviours is an important issue since the SiC whiskers are easily oxi￾dized in air at elevated temperature. Experiments [46,48] showed that although the stress exponents were almost the same in di€erent ambients, the creep rates were higher in air than in inert atmosphere. Lin et al. [52] have investigated the e€ect of matrix grain size (varying from 1.2 to 8.0 m) on the creep behaviours. At 1200C the creep rate exhibited an inverse grain size exponent of approximately 1, but at 1300C the creep rate was not sensitive to the grain size due to enhanced nucleation and coalescence of creep cavities and the development of macroscopic cracks as the grain size increased. In addition, the creep rate of Al2O3±SiC(w) composites was accelerated by the introduction of certain additives (i.e. Y2O3) [47]. The presence of an intergranular glassy phase introduced by the Y2O3 additive facilitated creep deformation resulting in an order of magnitude increase in creep rate. Finally, it should be pointed out that the method of creep testing may also in¯uence the experimental results. In four-point bending creep tests, the creep rate is strongly a€ected by the surface tensile properties of the specimens. The growth of surface cracks can result in a creep exponent of 2 or higher when the growth rate of microcrack obeys a power-law dependence on the local normal stress. This mode of testing can result in large errors in the stress exponent [49]. Microstructural observation of creep specimens gen￾erally revealed more or less cavitation at interfaces, grain boundaries and triple joint junctions. [43,45± 50,52] Unaccommodated grain boundary sliding was considered to be responsible for the formation of cav￾ities. Higher stresses and higher temperatures often increased the amount of intergranular voids, cavities and cracks, and sometimes caused the separation of interfaces and grain boundaries [43,48]. The Al2O3± SiC(w) composites exhibited higher number density and smaller cavities than those of monolithic Al2O3 [45]. One study showed that 30 and 50 vol% SiC(w) compo￾sites exhibited cavity number density 2 orders of mag￾nitude higher than that of Al2O3-20 vol% SiC(w) composite. The average cavity size of 30 and 50 vol% SiC(w) composites was 0.4 m, while that of 20 vol% composite was 0.05 m [47]. In the four-point bending tests, the cavity number density was approximately 1 to 2 orders of magnitude in the compressive surface region less than that in the tensile surface region [47], and at higher temperature, cracklike cavities tended to form extensive macroscopic tensile surface cracks, and the number of surface cracks increased with the grain size of Al2O3 [52]. Although a glassy phase was present in very limited amounts in some as-sintered composites, many authors [43,46,47,49,52] observed that after creep deformation in air ambient a SiO2±rich glassy phase around the whiskers had formed due to the oxidation of the latter. In most cases, the glassy phase penetrated along grain boundaries and interfaces and accumulated at triple grain junctions throughout the materials, which pre￾sumably facilitated grain boundary sliding and increased the creep rate. Dislocation networks were observed by some authors [42,43,46,49,51]. But in most cases, the dislocation den￾sity was very low [42,43,46,49] and the dislocation net￾works were also observed in the as-sintered specimens. They probably formed as a result of the large residual stresses resulting from thermal expansion mismatch during the fabrication processing. Contrary to other authors, Xia et al. [51] observed extensive dislocation networks in the crept specimens, and higher dislocation densities were observed in specimens deformed to large strains. Some authors observed the grain o€set and rotation of Al2O3 [45,52]. As far as the creep mechanism is con￾cerned, in early studies, a dislocation-controlled creep mechanism was proposed by some authors [41,42,44]. But no sucient microstructural information was pre￾sented to support this assertion, and this mechanism for the deformed Al2O3±SiC(w) composites remains doubt￾ful. A more recent study [51] revealed extensive disloca￾tion activity and very little evidence for the development of initial cavitation. The authors proposed an intra￾granular dislocation mechanism controlled by lattice Fig. 11. In¯uence of temperature on stress exponent in an Al2O3± 33 vol% SiC(w) composite (at 1200C, n=1; at 1300, 1400C, n=3) [48]. 404 Q. Tai. A. Mocellin / Ceramics International 25 (1999) 395±408