●" Science Direct O Solid state Materials science ELSEVIER Current Opinion in Solid State and Materials Science 9(2005)219-229 Interfaces in oxide-fibre-based composites S.T. Mileik Solid State Physics Institute of the Russian Academy of Sciences, Chernogolovka, Moscow 142432, Russia Received 8 May 2006: accepted 20 May 2006 Abstract The behaviour of the fibre/matrix interface makes a composite being a material different from just a mixture of two constituents. Some important features of the interface have been known from the very beginning of research in the composite field whereas some new fea- tures are becoming known when deeper insight into the behaviour of composites occurred to be vital. Oxide-fibre-based composites pro- vide clear illustrations of the primary importance of the interface in composite microstructures. Dealing with single crystalline oxide fibres in various matrices opens a number of new possibilities as well as yields new findings about a role of the interfaces in composites A review of both the possibilities and findings is a subject of the present paper o 2006 Elsevier Ltd. All rights reserved Keywords: Composites; Oxide fibres; Interface: Creep; Fracture; Crystallisation 1. Introduction ites) being fully or partly ceramic should be considered Most efforts in using advance materials in machines Single crystal oxides have always attracted attention as a for transportation and power generation have been potential base for heat-resistant materials because of a always directed towards enhancing energy efficiency of number of the reasons the machines, which means decreasing fuel consumption per unit of useful work. Enhancing the use temperature 1. They are inherently resistant to oxidation of structural materials for hot parts of machines has 2. Single crystals of oxides are inherently strong. been and will be always a most efficient way to reach 3. They have high melting points, high elastic moduli, and the goal low density Most heat-resistant materials for heavily loaded struc tural elements such as Ni-base superalloys are approaching ertainly, single crystals of oxides are most promising their physical limit set by their microstructural stability and substances since polycrystalline oxides, first, are not to be melting points. Perhaps, this limit will be around 1100 suficiently stable at high temperatures(say, at tempera and a dependence of the maximum use temperature on year tures higher than 1200 C for alumina) and, secondly, poly of the alloy development is asymptotically going to this crystals are known to reveal low creep resistance at high limit. Since no other metal can be a real base for heat-resis- temperatures because of the same reasons as those for tant alloys due to well-known reasons, the only alternative non-stability. Hence, single crystalline oxide fibres are is to exploit ceramics. Homogeneous ceramics cannot be wanted as a base for heat-resistant composites used in heavily loaded structures because of their inherent However, to realise the potential of single crystalline brittleness. Hence, non-homogeneous materials (compos- oxide fibres in either a metal-based or oxide matrix, it necessary to develop an appropriate design of a composite, and the main component of the composite to be properly E-mail address: mileiko @issp ac constructed is the interface or interphase, which has a 1359-0286S. see front matter 2006 Elsevier Ltd. All rights reserved doi:l0.1016 j. cossms.2006.05.004
Interfaces in oxide–fibre-based composites S.T. Mileiko Solid State Physics Institute of the Russian Academy of Sciences, Chernogolovka, Moscow 142432, Russia Received 8 May 2006; accepted 20 May 2006 Abstract The behaviour of the fibre/matrix interface makes a composite being a material different from just a mixture of two constituents. Some important features of the interface have been known from the very beginning of research in the composite field whereas some new features are becoming known when deeper insight into the behaviour of composites occurred to be vital. Oxide–fibre-based composites provide clear illustrations of the primary importance of the interface in composite microstructures. Dealing with single crystalline oxide fibres in various matrices opens a number of new possibilities as well as yields new findings about a role of the interfaces in composites. A review of both the possibilities and findings is a subject of the present paper. 2006 Elsevier Ltd. All rights reserved. Keywords: Composites; Oxide fibres; Interface; Creep; Fracture; Crystallisation 1. Introduction Most efforts in using advance materials in machines for transportation and power generation have been always directed towards enhancing energy efficiency of the machines, which means decreasing fuel consumption per unit of useful work. Enhancing the use temperature of structural materials for hot parts of machines has been and will be always a most efficient way to reach the goal. Most heat-resistant materials for heavily loaded structural elements such as Ni-base superalloys are approaching their physical limit set by their microstructural stability and melting points. Perhaps, this limit will be around 1100 C and a dependence of the maximum use temperature on year of the alloy development is asymptotically going to this limit. Since no other metal can be a real base for heat-resistant alloys due to well-known reasons, the only alternative is to exploit ceramics. Homogeneous ceramics cannot be used in heavily loaded structures because of their inherent brittleness. Hence, non-homogeneous materials (composites) being fully or partly ceramic should be considered as future heat-resistant materials. Single crystal oxides have always attracted attention as a potential base for heat-resistant materials because of a number of the reasons: 1. They are inherently resistant to oxidation. 2. Single crystals of oxides are inherently strong. 3. They have high melting points, high elastic moduli, and low density. Certainly, single crystals of oxides are most promising substances since polycrystalline oxides, first, are not to be sufficiently stable at high temperatures (say, at temperatures higher than 1200 C for alumina) and, secondly, polycrystals are known to reveal low creep resistance at high temperatures because of the same reasons as those for non-stability. Hence, single crystalline oxide fibres are wanted as a base for heat-resistant composites. However, to realise the potential of single crystalline oxide fibres in either a metal-based or oxide matrix, it is necessary to develop an appropriate design of a composite, and the main component of the composite to be properly constructed is the interface or interphase, which has a 1359-0286/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.cossms.2006.05.004 E-mail address: mileiko@issp.ac.ru Current Opinion in Solid State and Materials Science 9 (2005) 219–229
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 number of the purposes. In ductile-matrix composites the ever, the method does not allow an essential decrease in the interface is expected to be as strong as possible to allow fibre cost as compared to EFO transferring the load to the strong fibre; on the other hand Laser heated pedestal growth(LHPG)was applied for in brittle-matrix composites the interface(or interphase) is the first time to produce single-crystal Cr-doped Al2O expected to be sufficiently weak to trigger a mechanism of fibres. This was actually a floating-zone technique making the crack arrest and so providing sufficient fracture tough- use of small source rods locally melted at the end by a co ness to a composite structure. Hence, the microstructure laser [6]. There are some obvious advantages of the and properties of the oxide-fibre composites become pri- method: the absence of a crucible allows growing suffi- mary problems in the whole technology of future heat- ciently pure crystals, a small volume of the melts increa resistant compos thermal efficiency despite a low efficiency of the laser heat In the present paper, we start with a brief review of the ing and reduces mass exchange around the process zone fabrication technology schemes to produce oxide fibres that would be useful in growing such materials as mullite, focusing on the internal crystallisation method(ICM) which is characterised by a complicated phase diagram invented by the present author and his colleague, V I Kaz- But also the productivity rate of processes based on this min; then proceed with special features of some oxide fibres method can hardly be suitable to produce fibres to be used obtained by ICM, and continue with a brief outline of in structural material mechanical properties of ICM-fibres. a discussion of the properties of metal-matrix and oxide-matrix composites 2. 1. Internal crystallisation method (ICM) ith ICM-fibres focusing on the effect of the interface on mechanical behaviour of the composites is the main focus Internal crystallisation method invented in the authors of the paper laboratory was described in open literature for the first time nearly 15 years ago [7, 8]. Since then a main scheme of the 2. Single crystalline oxide fibres method as well as some variations of it have been published in a number of papers(see for instance Refs. [9-1l. Never- Certainly, the only way to produce either single crystal- theless, to make reading the present paper more convenient line oxides or those with typical eutectic microstructure is we are to describe here briefly the essence of the method to crystallise oxide melts. The following methods of crys- A schematic of the method is shown in Fig. 1. A molyb tallising oxide fibres are well known denum carcass with continuous channels in it. which is eas lly prepared by diffusion bonding or an assem blage of the 1. Edge Feeding Growth (EFG) wire and foil(step I in Fig. 1), is infiltrated with an oxide 2. Micro-pulling down(u-PD) melt(steps 2 and 3 )by the capillary force. The melt is then 3. Laser heated pedestal growth (LHPG) crystallised in the channels to form fibres in an oxide/ molybdenum block(step 4). This is a main scheme of the Strictly speaking all these methods are within a concept ICM, which can be varied to attain a particular goal. For of crystallising a melt by using a shaper, which was formu- example to ensure a homogeneous crystallographic orien- lated by Stepanov before the WWll [1]. Stepanov intro- tation of the fibres in a block, a seed is oriented in an duces a shaper to pre-determine a shape and size of the appropriate manner. Finally, the fibres are freed from the capillary column at the top of which the liquid/ solid inter- molybdenum carcass by dissolution of molybdenum in a face arises, although the authors were hardly aware about mixture of acids. It can be seen that the process of fibre Stepanov's ideas published in Russian. growth based on ICM is actually similar to growth of bulk EFG method was used for the first time to produce sap- single crystals; therefore, the fibre cost should be of the phire fibres by La Belle and Mlavsky [2), who lifted the same order of magnitudes as that of bulk crystals and they growth zone above the melt surface with a capillary tube can be used as reinforcements for structural composites in the crucible. the lower end of the tube is located near In the present context it is important to emphasize that the bottom of the crucible, and the growth zone is now the oxide/ molybdenum interface occurs to be ideal;an fixed relative to the heater independent of the level of the illustration is given in Fig. 2. It is also important to note melt surface which goes down with time. Both, a review that molybdenum foil is recrystallising in the process of of the corresponding techniques and discussion of the fibre melt infiltration and fibres are crystallizing in the channels growth parameters, structure and mechanical properties of Newly formed grains form steps on the surface, which are sapphire fibres are presented in Ref [3]. It appears that a approximately I um in height. The steps are replicated on stable growth takes place at rates no more than 0.5- the fibre surface as can be seen in Fig. 3. They can act as 1.0mm/ s, which makes productivity rate of the process stress concentrators, but perhaps, are not most dangerous. low, so that the cost of the fibres is too high to use them in structural applications 2.2. Fibres obtained by ICA Micro-pulling down method (u-PD)developed by Japanese researchers [4, 5] did actually turn up the eFg A family of single crystalline and eutectic oxide fibres scheme, which simplified slightly growth procedures. How- have been obtained by using ICM tion to sapphire
number of the purposes. In ductile–matrix composites the interface is expected to be as strong as possible to allow transferring the load to the strong fibre; on the other hand, in brittle–matrix composites the interface (or interphase) is expected to be sufficiently weak to trigger a mechanism of the crack arrest and so providing sufficient fracture toughness to a composite structure. Hence, the microstructure and properties of the oxide–fibre composites become primary problems in the whole technology of future heatresistant composites. In the present paper, we start with a brief review of the fabrication technology schemes to produce oxide fibres focusing on the internal crystallisation method (ICM) invented by the present author and his colleague, V.I. Kazmin; then proceed with special features of some oxide fibres obtained by ICM, and continue with a brief outline of mechanical properties of ICM-fibres. A discussion of the properties of metal–matrix and oxide–matrix composites with ICM-fibres focusing on the effect of the interface on mechanical behaviour of the composites is the main focus of the paper. 2. Single crystalline oxide fibres Certainly, the only way to produce either single crystalline oxides or those with typical eutectic microstructure is to crystallise oxide melts. The following methods of crystallising oxide fibres are well known: 1. Edge Feeding Growth (EFG) 2. Micro-pulling down (l-PD) 3. Laser heated pedestal growth (LHPG) Strictly speaking all these methods are within a concept of crystallising a melt by using a shaper, which was formulated by Stepanov before the WWII [1]. Stepanov introduces a shaper to pre-determine a shape and size of the capillary column at the top of which the liquid/solid interface arises, although the authors were hardly aware about Stepanov’s ideas published in Russian. EFG method was used for the first time to produce sapphire fibres by LaBelle and Mlavsky [2], who lifted the growth zone above the melt surface with a capillary tube in the crucible. The lower end of the tube is located near the bottom of the crucible, and the growth zone is now fixed relative to the heater independent of the level of the melt surface which goes down with time. Both, a review of the corresponding techniques and discussion of the fibre growth parameters, structure and mechanical properties of sapphire fibres are presented in Ref. [3]. It appears that a stable growth takes place at rates no more than 0.5– 1.0 mm/s, which makes productivity rate of the process low, so that the cost of the fibres is too high to use them in structural applications. Micro-pulling down method (l-PD) developed by Japanese researchers [4,5], did actually turn up the EFGscheme, which simplified slightly growth procedures. However, the method does not allow an essential decrease in the fibre cost as compared to EFG. Laser heated pedestal growth (LHPG) was applied for the first time to produce single-crystal Cr-doped Al2O3 fibres. This was actually a floating-zone technique making use of small source rods locally melted at the end by a CO2 laser [6]. There are some obvious advantages of the method: the absence of a crucible allows growing suffi- ciently pure crystals, a small volume of the melts increases thermal efficiency despite a low efficiency of the laser heating and reduces mass exchange around the process zone that would be useful in growing such materials as mullite, which is characterised by a complicated phase diagram. But also the productivity rate of processes based on this method can hardly be suitable to produce fibres to be used in structural materials. 2.1. Internal crystallisation method (ICM) Internal crystallisation method invented in the author’s laboratory was described in open literature for the first time nearly 15 years ago [7,8]. Since then a main scheme of the method as well as some variations of it have been published in a number of papers (see for instance Refs. [9–11]). Nevertheless, to make reading the present paper more convenient we are to describe here briefly the essence of the method. A schematic of the method is shown in Fig. 1. A molybdenum carcass with continuous channels in it, which is easily prepared by diffusion bonding of an assemblage of the wire and foil (step 1 in Fig. 1), is infiltrated with an oxide melt (steps 2 and 3) by the capillary force. The melt is then crystallised in the channels to form fibres in an oxide/ molybdenum block (step 4). This is a main scheme of the ICM, which can be varied to attain a particular goal. For example, to ensure a homogeneous crystallographic orientation of the fibres in a block, a seed is oriented in an appropriate manner. Finally, the fibres are freed from the molybdenum carcass by dissolution of molybdenum in a mixture of acids. It can be seen that the process of fibre growth based on ICM is actually similar to growth of bulk single crystals; therefore, the fibre cost should be of the same order of magnitudes as that of bulk crystals and they can be used as reinforcements for structural composites. In the present context it is important to emphasize that the oxide/molybdenum interface occurs to be ideal; an illustration is given in Fig. 2. It is also important to note that molybdenum foil is recrystallising in the process of melt infiltration and fibres are crystallizing in the channels. Newly formed grains form steps on the surface, which are approximately 1 lm in height. The steps are replicated on the fibre surface as can be seen in Fig. 3. They can act as stress concentrators, but perhaps, are not most dangerous. 2.2. Fibres obtained by ICM A family of single crystalline and eutectic oxide fibres have been obtained by using ICM. In addition to sapphire 220 S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 [7-9] and alumina-YAG-eutectics [7, 8, 10]. YAG with the 2. Coating fibres with a thin layer of various materials (100) orientation of the fibre axis [12] and single crystalline yields healing of surface defects and an essential mullite [13, 14]fibres have been produced. Crystallization of nhancement of the fibre strength. A CVd process was sapphire fibres [9] was conducted by using a seed to get used to coat the fibres to ensure an intimate contact fibres with the c-axis coinciding with the fibre axis: YAG on the interrace and mullite fibres have been crystallised without using seed, so their axis coincides with the (00 1) and c-directions, The second feature is of an obvious importance in the respectively(see Fig. 4) present context. There is a clear dependence of the fibre strength on the coating thickness presented in Fig. 6. How- 2.2.1. Strength of ICM-fibres ever, at the present time it is not clear as to the extent of the Room temperature strength of the fibres was measured maximum fibre strength that can be obtained. A maximum by bending a fibre over rigid cylinders of decreasing diam- can be expected as the coating layer also contains defects eters and counting an average distance between fibre gh temperature strength of the fibres has been mea breaks on each step of the experiment [15]. This yields a sured by testing oxide/molybdenum composites obtained dependence of ultimate strain on the fibre length provided by ICM in either tension [7, 8] or bending. Some results the Youngs modulus of the fibre material is also known. are shown in Fig. 7 The room temperature strength of ICM-fibres(see an example in Fig. 5)are characterised by two features [9, 10]: 2. 2. High temperature creep characteristics of ICM-fibres Creep tests of ICM-fibres are normally performed by 1. The strength/scale dependence is very strong, so the bending oxide-fibre/Mo-matrix specimens. The analysis Weibull exponent varies usually between 3 and 5. A rea- of the experimental data allows obtaining tensile creep son for it is certainly an existence of sufficiently rough characteristics of the fibres. The analysis [16] is based on defects, which are located mainly on the fibre surface. the following assumptions: Molybdenum carcass Oxide/molybdenum block oxide melt the carcass Fig. 1. Schematic of the internal crystallization method (ICM)
[7–9] and alumina–YAG-eutectics [7,8,10], YAG with the h100i orientation of the fibre axis [12] and single crystalline mullite [13,14] fibres have been produced. Crystallization of sapphire fibres [9] was conducted by using a seed to get fibres with the c-axis coinciding with the fibre axis; YAG and mullite fibres have been crystallised without using seed, so their axis coincides with the h001i and c-directions, respectively (see Fig. 4). 2.2.1. Strength of ICM-fibres Room temperature strength of the fibres was measured by bending a fibre over rigid cylinders of decreasing diameters and counting an average distance between fibre breaks on each step of the experiment [15]. This yields a dependence of ultimate strain on the fibre length provided the Young’s modulus of the fibre material is also known. The room temperature strength of ICM-fibres (see an example in Fig. 5) are characterised by two features [9,10]: 1. The strength/scale dependence is very strong, so the Weibull exponent varies usually between 3 and 5. A reason for it is certainly an existence of sufficiently rough defects, which are located mainly on the fibre surface. 2. Coating fibres with a thin layer of various materials yields healing of surface defects and an essential enhancement of the fibre strength. A CVD process was used to coat the fibres to ensure an intimate contact on the interface. The second feature is of an obvious importance in the present context. There is a clear dependence of the fibre strength on the coating thickness presented in Fig. 6. However, at the present time it is not clear as to the extent of the maximum fibre strength that can be obtained. A maximum can be expected as the coating layer also contains defects. High temperature strength of the fibres has been measured by testing oxide/molybdenum composites obtained by ICM in either tension [7,8] or bending. Some results are shown in Fig. 7. 2.2.2. High temperature creep characteristics of ICM-fibres Creep tests of ICM-fibres are normally performed by bending oxide–fibre/Mo–matrix specimens. The analysis of the experimental data allows obtaining tensile creep characteristics of the fibres. The analysis [16] is based on the following assumptions: Fig. 1. Schematic of the internal crystallization method (ICM). S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229 221
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 Interface where nm, On, and n are constants. The value of n can be chosen arbitrary. In what follows, mn=10h, which means that on is the stress to cause 1% creep strain for 100 h. We call this value as creep resistance of a material on 100-h time base Matrix A solution of a creep problem for a beam under bending yields a dependence of the deflection rate, f, of the beam at its centre on applied load 0. For a beam of rectangula cross-section of height 2h, width b we have Fibre Q (2) in case of 3. 0.5 um b7|1q in case of 4-point bending. Here L and Li are the distances Fibre in=a, Mn=nanbh- and M is the bending moment Matrix The solution was obtained by neglecting a contribution of shear deformations to the displacement that can be essential in case of 3-point bending Creep characteristics of ICM-fibres tested up to now Fig.2.TEM images of the sapphire-fibre/molybdenum-matrix interface fibres are presented in Fig. 8. Here alo stalline mullite in an oxide/ molybdenum block [32]. for Nextel 720 fibre (a-Al2O3 mullite) evaluated from experimental data presented in Ref [17] is shown. A num- ber of important conclusions can be now drawn; here we emphasize just three points 1. In temperature interval from 1100 to 1600C, values of the creep resistance of single crystalline YAG and mull- ite as well as that of alumina-YAG-eutectic fibres obtained by using ICM are nearly the YAG fibre looks slightly better than the others. Still, their creep resistance can be certainly enhanced by crystallising them in the(111) direction 2. Surprisingly enough, single crystalline mullite fibres pro- duced by ICM do not seem to be superior to, say, YAG fibres. Their creep resistance differs essentially from the experimental value(Dokko et al. [18] 3. Polycrystalline oxide fibres, available at th Fig. 3. A view of the flat surface of a sapphire fibre; a replica of the time, obviously lose their creep resistance below a tem- molybdenum foil can be seen with a grain size of 10 um. perature of 1200C certainly due to an intrinsic behav lour of 1. A contribution of the matrix, which is fully recrystallised molybdenum, to the creep resistance of oxide/molybde t hum composites at temperatures above 1000.C is negli-3.Oxide-fibre/metal-matrix composites A composite is characterised by identical creep behav- Metal-matrix composites reinforced with ICM-fibres our under tension and compression are obtained via liquid phase route [19, 20]. Hence, the 3. The creep law of the material is fibre/ matrix interface strength depends on wettability of an oxide with a metal melt; the issue is analysed in details (1) in a review paper[21].Ti- and Ni-based alloys as matrices are of an immediate practical interest. They represent also
1. A contribution of the matrix, which is fully recrystallised molybdenum, to the creep resistance of oxide/molybdenum composites at temperatures above 1000 C is negligible, less than 10 MPa [7]. 2. A composite is characterised by identical creep behaviour under tension and compression. 3. The creep law of the material is e_ ¼ gn r rn n ð1Þ where gn, rn, and n are constants. The value of gn can be chosen arbitrary. In what follows, gn = 104 h1 , which means that rn is the stress to cause 1% creep strain for 100 h. We call this value as creep resistance of a material on 100-h time base. A solution of a creep problem for a beam under bending yields a dependence of the deflection rate, _ f , of the beam at its centre on applied load Q. For a beam of rectangular cross-section of height 2h, width b we have: _ f ¼ gn 1 23nþ2 nnðn þ 2Þ Q rnh2 n L b n L h L ð2Þ in case of 3-point bending and _ f ¼ vn M Mn n L2 1 8 ð3Þ in case of 4-point bending. Here L and L1 are the distances between periphery and internal supports, respectively, vn ¼ gn h , Mn ¼ 2n 2nþ1 rnbh2 and M is the bending moment. The solution was obtained by neglecting a contribution of shear deformations to the displacement that can be essential in case of 3-point bending. Creep characteristics of ICM-fibres tested up to now including preliminary data for single crystalline mullite fibres are presented in Fig. 8. Here also creep resistance for Nextel 720 fibre (a-Al2O3 + mullite) evaluated from experimental data presented in Ref. [17] is shown. A number of important conclusions can be now drawn; here we emphasize just three points: 1. In temperature interval from 1100 to 1600 C, values of the creep resistance of single crystalline YAG and mullite as well as that of alumina–YAG-eutectic fibres obtained by using ICM are nearly the same. YAG fibre looks slightly better than the others. Still, their creep resistance can be certainly enhanced by crystallising them in the h111i direction. 2. Surprisingly enough, single crystalline mullite fibres produced by ICM do not seem to be superior to, say, YAG fibres. Their creep resistance differs essentially from the experimental value (Dokko et al. [18]). 3. Polycrystalline oxide fibres, available at the present time, obviously lose their creep resistance below a temperature of 1200 C certainly due to an intrinsic behaviour of grain boundaries. 3. Oxide–fibre/metal–matrix composites Metal–matrix composites reinforced with ICM-fibres are obtained via liquid phase route [19,20]. Hence, the fibre/matrix interface strength depends on wettability of an oxide with a metal melt; the issue is analysed in details in a review paper [21]. Ti- and Ni-based alloys as matrices are of an immediate practical interest. They represent also Fig. 2. TEM images of the sapphire–fibre/molybdenum–matrix interface in an oxide/molybdenum block [32]. Fig. 3. A view of the flat surface of a sapphire fibre; a replica of the molybdenum foil can be seen with a grain size of 10 lm. 222 S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 Fig 4. Single crystalline YAG(a)and mullite fibres(b)obtained by using ICM 8000 吵∴! Coated 2zum∽gz 8。8 FIBRE LENGTH/ mm Fig. 5. Room temperature bending strength of sapphire fibres as extracted Temperature/12001400-1600 and coated with silicon carbide Fig. 7. The temperature dependence of the bending and tensile strength of sapphire and some eutectic fibres 5000 two cases of the oxide/ metal composites from the point of view of the interface strength. Results of the study of these two cases will be presented below, but we start with a brief outline of micromechanical creep models, which are neces- sary to interpret test results in an appropriate way 3.1. Creep models We consider composites with creeping matrix and ini- tially continuous fibres. A continuous fibre means that its length is much larger than a critical fibre length. Obviously, there can be observed at least four creep regimes of such composite [22]: Coating thickness/micron 1. E: fibres are elastic and non-breaking 2. Br-NCr: fibres are elastic and brittle thickness. Solid points stand for fibres of batch vo453 with pyrocarbon 3. Cr: fibres are creeping and non-breaking bating: open points are for fibres of batch vo86 with SiC-coating 4. Br-Cr: fibres are creeping and brittle
two cases of the oxide/metal composites from the point of view of the interface strength. Results of the study of these two cases will be presented below, but we start with a brief outline of micromechanical creep models, which are necessary to interpret test results in an appropriate way. 3.1. Creep models We consider composites with creeping matrix and initially continuous fibres. A continuous fibre means that its length is much larger than a critical fibre length. Obviously, there can be observed at least four creep regimes of such composite [22]: 1. E: fibres are elastic and non-breaking. 2. Br–NCr: fibres are elastic and brittle. 3. Cr: fibres are creeping and non-breaking. 4. Br–Cr: fibres are creeping and brittle. Fig. 5. Room temperature bending strength of sapphire fibres as extracted and coated with silicon carbide. Fig. 6. Average sapphire fibre strength on a length of 1 mm versus coating thickness. Solid points stand for fibres of batch V0453 with pyrocarbon coating; open points are for fibres of batch V086 with SiC-coating. Fig. 7. The temperature dependence of the bending and tensile strength of sapphire and some eutectic fibres. Fig. 4. Single crystalline YAG (a) and mullite fibres (b) obtained by using ICM. S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229 223
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 the interface strength and equal to that of the matrix. Note that Eq (6) was obtained assuming steady state creep fol- lowed the ending of the fibre breakage process on the tran- sitional stage of creep. It occurs that the larger the value of the shorter is the fibre length in the steady-state condition and, hence, the larger is the stress carried by the fibre For regime Cr, creep-rate/stress dependence is obvious: 0=a Ve+o · Alumina- YAG where nf, af, and n are constants in a power low for the fibre Mullite o- Nextel 720 1000110012001300140015001600 Regime br-Cr is a combination of the two regimes just Temperature/C described Note that while evaluating creep characteristics of fibres Fig. 8. Creep polycrystalline mt ICM-fibres in comparison with a Section 2.2.2, Eq. (7)was used. For metal-matrix compos- based commercial fibre presented in Ref [17] ites, either regime E or regime Br-NCr is characteristic, the latter one being of most interest. Hence, normally Eq (6) et the creep law of the matrix will be as given by Eq. used to analyse experimental data (1), just it is convenient to denote the Om, and In 3. 2. TiAl-based-inatrix: strong interface For regime E, the dependence of fibre stress on time at a Hrer=hn+HE∥Em, where Ef and Em are Youngs Hence. the value of parameter a in E4625!多2“分 onstant composite stress, o, will obviously be Titanium wets oxides perfectly; it can be seen on a cr section of the sapphire-fibre/Ti-48Al-matrix compe c()= presented in Fig. 9[20]. During the contact between the ym+m(m-1) "e a fibre and liquid matrix dissolution of the fibre has occurred. Initially sharp corners of the ICM-fibres becom (4) founded. An attempt to measure the interface strength push out was not successful: it occurred to be too st moduli of the fibre and matrix, respectively, Vr and F Creep experiments and their analysis have yielded values are their volume fractions of n and on. Note that at nln=10*- the value of on is the The creep strain of the composite is actually elastic stress to cause 1% of creep strain for 100 h. This character- deformation of the fibre changing with time istic stress is presented in Fig. 10 for temperatures up to 1100C. It can be seen that on a 120 MPa at a temperature Erve at t→∞ as high as 1050C. This means that the effective stress on the fibres is about 500 MPa. This nearly corresponds to For a composite with initially continuous brittle fibres the effective strength of sapphire fibres in a molybdenum creeping in regime Br-NCr, the dependence between creep matrix(Fig. 7) rate and stress is written [16]as It is important to point out two similarities between these two cases. First, in both cases the fibre/matrix inter- (2)+(.= fibre/matrix microstructure of high shear strength for the fibre strength. do is the average strength of a fibre of composites under consideration remain to be revealed. Sec- ond, the effective fibre strength is certainly higher than the length lo and characteristic cross-sectional size d, fibre strength determined by testing separate fibres. healing q=m+B+mB, and fibre surface defects that can be filled with the matrix due to fibre-matrix and matrix-fibre wetting in the alumina- molybdenum and TiAl-sapphire systems, respectively, can be one reason for it. Moreover, in the latter case sharp defects can be rounded as a result of the fibre material dis- An important parameter, a, is a continuity factor describing solution in the molten matrix. the fibre/matrix interface strength, a-0 if there is no Finally, it should be noted that to make a conclusion on bonding at the interface, a= l for an ideal bonding when a possibility to oxide-fibre/Ti-Al composites to become
Let the creep law of the matrix will be as given by Eq. (1), just it is convenient to denote the constants as gm, rm, and m. For regime E, the dependence of fibre stress on time at a constant composite stress, r, will obviously be rðfÞ ðtÞ ¼ r V f 1 V m V V m þ gmðm 1Þ V fEf rm r rm m1 t 2 6 4 3 7 5 1 m1 8 >>>: 9 >>= >>; ð4Þ Here V = Vm + VfEf/Em, where Ef and Em are Young’s moduli of the fibre and matrix, respectively, Vf and Vm are their volume fractions. The creep strain of the composite is actually elastic deformation of the fibre changing with time eðtÞ ¼ rðfÞ ðtÞ Ef ;e ! r EfV f at t ! 1: ð5Þ For a composite with initially continuous brittle fibres creeping in regime Br–NCr, the dependence between creep rate and stress is written [16] as r ¼ krm rðfÞ 0 krm !b l0 d 2 4 3 5 mþ1 q e_ gm 1 q V f þ rm e_ gm 1 m V m ð6Þ where b is the exponent in the Weibull distribution for the fibre strength, rðfÞ 0 is the average strength of a fibre of length l0 and characteristic cross-sectional size d, q = m + b + mb, and k ¼ a 2 3 1 m m 2m þ 1 2 ffiffiffi 3 p p !1 2 1 2 4 3 5 1 m : An important parameter, a, is a continuity factor describing the fibre/matrix interface strength, a ! 0 if there is no bonding at the interface, a = 1 for an ideal bonding when the interface strength and equal to that of the matrix. Note that Eq. (6) was obtained assuming steady state creep followed the ending of the fibre breakage process on the transitional stage of creep. It occurs that the larger the value of a the shorter is the fibre length in the steady-state condition and, hence, the larger is the stress carried by the fibre. For regime Cr, creep-rate/stress dependence is obvious: r ¼ rf e_ gf 1 n V f þ rm e_ gm 1 m V m ð7Þ where gf, rf, and n are constants in a power low for the fibre similar to Eq. (1). Regime Br–Cr is a combination of the two regimes just described. Note that while evaluating creep characteristics of fibres, Section 2.2.2, Eq. (7) was used. For metal–matrix composites, either regime E or regime Br–NCr is characteristic, the latter one being of most interest. Hence, normally Eq. (6) is used to analyse experimental data. 3.2. TiAl-based-matrix: strong interface Titanium wets oxides perfectly; it can be seen on a crosssection of the sapphire–fibre/Ti–48Al–matrix composite presented in Fig. 9 [20]. During the contact between the fibre and liquid matrix dissolution of the fibre has occurred. Initially sharp corners of the ICM-fibres become rounded. An attempt to measure the interface strength by push out was not successful: it occurred to be too strong. Hence, the value of parameter a in Eq. (6) is equal 1. Creep experiments and their analysis have yielded values of n and rn. Note that at gn = 104 h1 the value of rn is the stress to cause 1% of creep strain for 100 h. This characteristic stress is presented in Fig. 10 for temperatures up to 1100 C. It can be seen that rn 120 MPa at a temperature as high as 1050 C. This means that the effective stress on the fibres is about 500 MPa. This nearly corresponds to the effective strength of sapphire fibres in a molybdenum matrix (Fig. 7). It is important to point out two similarities between these two cases. First, in both cases the fibre/matrix interface strength is extremely high. For the molybdenum matrix, which is wetted with molten alumina well, the situation is clear. Particular mechanisms of formation of the fibre/matrix microstructure of high shear strength for the composites under consideration remain to be revealed. Second, the effective fibre strength is certainly higher than the fibre strength determined by testing separate fibres. Healing fibre surface defects that can be filled with the matrix due to fibre–matrix and matrix–fibre wetting in the alumina– molybdenum and TiAl–sapphire systems, respectively, can be one reason for it. Moreover, in the latter case sharp defects can be rounded as a result of the fibre material dissolution in the molten matrix. Finally, it should be noted that to make a conclusion on a possibility to oxide–fibre/Ti–Al composites to become Fig. 8. Creep resistance of some ICM-fibres in comparison with a polycrystalline mullite-based commercial fibre presented in Ref. [17]. 224 S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 structural materials, a number of questions must be 人143MPa answered. Stress rupture and oxidation resistance at high temperatures as well as crack resistance at room tempera ture are most important ones. At the present time, there can be seen some ways to make oxide-fibre/Ti-Al compos- ites that are not too brittle. An illustration of a possibility of one way to do this is presented in Fig. Il 3.3. Ni-based-matrix: variable interface An effectiveness of the usage of ICM-fibres in Ni-based- matrix was discussed in details in Refs. [16, 19, 32]. The main finding of that research is a synergetic influence of he fibre/matrix interface on the creep behaviour of the composites. This means that the interface determines the composite creep resistance in two ways and the output is Fig. Il. Load versus displacement for 3-point bending of two oxide-fibre/ TiAl-matrix composite specimens of a special macrostructure at room emperature larger than just a linear sum. The first way is sufficiently obvious: increasing the fibre/matrix strength yields increasing stresses in the fibre d way is not so obvious. It emerges from an in between the fibre surface and matrix material The matrix, especially in a liquid state, can cause dam age of the fibre surface and then the fibre being extracted from the matrix and tested as an isolated object reveals an essential decrease in the strength. In the case of sapphire fibre in nickel-based-matrix the fibre degradation of such 65m kind was observed in a number of works [23, 24]. Their authors used this observation to speak out a doubt in the Fig9.SEM micrograph of a cross-section of the sapphire-fibre/Ti-48AI- future of oxide/Ni systems as heat resistant composites However. the behaviour of a fibre in matrix differs from that of a separate fibre and the difference can be impressive Actually, an example of such difference was presented above, in Figs. 5 and 6, where the coating acted as a matrix. Hence, if the fibre/matrix interface is ideal, like the coating is, then strength characteristics of the fibre in matrix will be much higher than those of fibres tested separately. The idea can also be proved by an analysis of results of testing com- osites of a definite fibre/n matrix with interface strength. The analysis is supposed to be based on comparison of experimental results and those obtained by calculations made on a basis of the micro-mechanical model The main problem with this composite system is actu lly the interface bond. From the very beginning of the history of modern composites, poor wetting of oxide (in 40 particular, sapphire) with molten nickel has been known [25]. Also it is now clear that appropriate design of the interface is possible [24], which will yield a sufficiently strong bonding at the interface. The importance of the Fig. 10. Stress to cause I% creep stain for 100 h versus testing temperature interface bonding in designing heat-resistant composites for the sapphire-fibre/Ti-48AI-matrix composite with fibre volume of oxide/Ni-based-matrix is illustrated by experiments in fractions between 0.20 and 0. 25 [20]. hich creep tests [16] have been complemented with
structural materials, a number of questions must be answered. Stress rupture and oxidation resistance at high temperatures as well as crack resistance at room temperature are most important ones. At the present time, there can be seen some ways to make oxide–fibre/Ti–Al composites that are not too brittle. An illustration of a possibility of one way to do this is presented in Fig. 11. 3.3. Ni-based-matrix: variable interface An effectiveness of the usage of ICM-fibres in Ni-basedmatrix was discussed in details in Refs. [16,19,32]. The main finding of that research is a synergetic influence of the fibre/matrix interface on the creep behaviour of the composites. This means that the interface determines the composite creep resistance in two ways and the output is larger than just a linear sum. The first way is sufficiently obvious: increasing the fibre/matrix interface strength yields increasing stresses in the fibre. The second way is not so obvious. It emerges from an interaction between the fibre surface and matrix material. The matrix, especially in a liquid state, can cause damage of the fibre surface and then the fibre being extracted from the matrix and tested as an isolated object reveals an essential decrease in the strength. In the case of sapphire fibre in nickel-based-matrix the fibre degradation of such kind was observed in a number of works [23,24]. Their authors used this observation to speak out a doubt in the future of oxide/Ni systems as heat resistant composites. However, the behaviour of a fibre in matrix differs from that of a separate fibre and the difference can be impressive. Actually, an example of such difference was presented above, in Figs. 5 and 6, where the coating acted as a matrix. Hence, if the fibre/matrix interface is ideal, like the coating is, then strength characteristics of the fibre in matrix will be much higher than those of fibres tested separately. The idea can also be proved by an analysis of results of testing composites of a definite fibre/matrix system with changing interface strength. The analysis is supposed to be based on comparison of experimental results and those obtained by calculations made on a basis of the micro-mechanical model. The main problem with this composite system is actually the interface bond. From the very beginning of the history of modern composites, poor wetting of oxide (in particular, sapphire) with molten nickel has been known [25]. Also it is now clear that appropriate design of the interface is possible [24], which will yield a sufficiently strong bonding at the interface. The importance of the interface bonding in designing heat-resistant composites of oxide/Ni-based-matrix is illustrated by experiments in which creep tests [16] have been complemented with Fig. 9. SEM micrograph of a cross-section of the sapphire–fibre/Ti–48Al– matrix composite [20]. Fig. 10. Stress to cause 1% creep stain for 100 h versus testing temperature for the sapphire–fibre/Ti–48Al–matrix composite with fibre volume fractions between 0.20 and 0.25 [20]. Fig. 11. Load versus displacement for 3-point bending of two oxide–fibre/ TiAl–matrix composite specimens of a special macrostructure at room temperature. S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229 225
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 070 developed before the observation had been done. Reasons for the formation of such interface are not clear: however Matrix we can assume that the nucleation of bonding at some sites of the interface takes place followed by diffusion of necessary elements from the matrix and depletion of the matrix with the elements responsible for the bonding The interface strength measured by a modified push-out technique presented in Fig. 13a starts to decrease with the fibre volume fraction reaches 0. 25, certainly due to Fibre shortage of elements that provide interface bonding. That corresponds to a decrease in the creep resistance, Fig. 13b. The creep-resistance/fibre-V deviates clearly fro lation of the depend 33701q results we need values of the parameters involved in Eq (6). A set of these parameters was chosen from the results of direct independent tests(see Table 1)except written in bold To describe the creep behaviour of composites with large fibre volume fractions and a weak matrix/ matrix interface t'a 20 MPa(a=0.01), a value fibre strength equal to 150 MPa, which is characteristic value for fibres tested separately, can be accepted. The calculation does correspond to the experiment To describe the behaviour of the fibre volume fractions(strong interface), we need, obvi- A-field ously, to adjust the value of a, which describes the inter face strength. However, varying only the value of a does Fig12.TEM images of the fibre/matrix interface in a alumina-YAG- not yield a satisfactory result; to make the dependences for various o to be consistent with the experimental points, it is necessary to assume variations of the fibre strength characteristics together with variations of the observation of the interface microstructure [19] and mea- interface strength. Such assumption is not an artifi surements of strength [2 procedure. The first experimental fact supporting an First of all, it should be emphasized that the fibre/ of the fibre strength characteristics being affected by the matrix interface in Ni-matrix composites(Fig. 12)does matrix via the interface was mentioned above. Secondly drastically differ from what is observed in Mo- or Ti- TEM-observations of the microstructure of the interface based composites(Figs. 2 and 9, respectively). It is inter-( Fig. 12)reveal just a partial contact on the matrix/matrix esting that the interface looks like an island-type model interface in these composites. It should be noted that the Calculation: Weak interface 75 Fibre volume fraction Fibre volume fraction Fig 13. Alumina-YAG-eutectic-matrix/Ni-based-matrix composites: (a) the interface strength versus fibre volume fraction and(b)stress to cause 1% creep strain for 100 h at 1150C versus fibre volume fraction
observation of the interface microstructure [19] and measurements of strength [26]. First of all, it should be emphasized that the fibre/ matrix interface in Ni–matrix composites (Fig. 12) does drastically differ from what is observed in Mo- or Tibased composites (Figs. 2 and 9, respectively). It is interesting that the interface looks like an island-type model developed before the observation had been done. Reasons for the formation of such interface are not clear; however, we can assume that the nucleation of bonding at some sites of the interface takes place followed by diffusion of necessary elements from the matrix and depletion of the matrix with the elements responsible for the bonding. The interface strength measured by a modified push-out technique presented in Fig. 13a starts to decrease with the fibre volume fraction reaches 0.25, certainly due to a shortage of elements that provide interface bonding. That corresponds to a decrease in the creep resistance, Fig. 13b. The creep-resistance/fibre-volume-fraction curve deviates clearly from an extrapolation of the dependence observed at low fibre volume fractions. To analyse these results we need values of the parameters involved in Eq. (6). A set of these parameters was chosen from the results of direct independent tests (see Table 1) except for that written in bold. To describe the creep behaviour of composites with large fibre volume fractions and a weak matrix/matrix interface, s* 20 MPa (a = 0.01), a value fibre strength equal to 150 MPa, which is characteristic value for fibres tested separately, can be accepted. The calculation does correspond to the experiment. To describe the behaviour of the composites at low fibre volume fractions (strong interface), we need, obviously, to adjust the value of a, which describes the interface strength. However, varying only the value of a does not yield a satisfactory result; to make the dependences for various a to be consistent with the experimental points, it is necessary to assume variations of the fibre strength characteristics together with variations of the interface strength. Such assumption is not an artificial procedure. The first experimental fact supporting an idea of the fibre strength characteristics being affected by the matrix via the interface was mentioned above. Secondly, TEM-observations of the microstructure of the interface (Fig. 12) reveal just a partial contact on the matrix/matrix interface in these composites. It should be noted that the Fig. 12. TEM images of the fibre/matrix interface in a alumina–YAGeutectic–fibre/Ni-based-matrix composite specimen. Fig. 13. Alumina–YAG-eutectic–matrix/Ni-based-matrix composites: (a) the interface strength versus fibre volume fraction and (b) stress to cause 1% creep strain for 100 h at 1150 C versus fibre volume fraction. 226 S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 Table I creep properties that have been really observed for a fam- Constituents prop used in model calculations ily of ICM-oxide-matrix/nickel-based-matrix composites Interface property Fibre properties at 1150C. The upper line corresponds to an upper limit (o= l mm. of the he creep resistance of ICM-oxide-m d=0.1mm) metal-matrix composites at 1150C. It should be empha o(D(MPa) sized that the experimental point obtained earlier for interface, Vr>0.25 0.01 interface, Vr<0.15 0.5 33 alumina-YAG-eutectic-matrix/molybdenum-matrix com- posite at the same temperature and sapphire-fibre/TiAl- matrix composite at 900-1000C lie nearly on this line Matrix properties Note that both composites are characterised by an ideal 41.5 MPa Inter Note: The fibre strength at a= l corresponds to data presented in Fig. 7. The experiments and results of calculations based on the micro-mechanical model allow expecting a maximum use stronger the interface, the larger portion of the fibre sur- temperature for the oxide-fibre/Ni-based-matrix system face is in contact with the matrix. A continuous interface to be at least 1150C [16]. The criterion for the maximum was observed in the Mo-based composite. Therefore, the temperature is that specific creep resistance of a composite assumption made in modelling creep behaviour, which is equal to that of a superalloy with density of 8 g/cm at yields Eq. (6), namely introducing continuity factor a, 150 MPa has a physical meaning rather than produces just a phe- A further increase in the use temperature of composites nomenological parameter in the mechanical model. With in the oxide/Ni-system depends on the ability to appropri- increasing the value of a a larger portion of the surface ately control the defects are healed that yields an increase in the fibre effec- tive strength. This effect is added to a normal strength/ 4. Oxide-fibre/oxide-matrix composites scale effect It is informative to replot the calculated dependencies The fibre/ matrix interface in such composites made of presented in Fig. 13b together with that for an ideal inter- brittle continuances entirely determines their applicability face (a= 1), which corresponds also to an increase in the as structural materials. The problem is of importance since characteristic fibre strength to a value determined by test- such composites being reinforced with single crystalline ing oxide/molybdenum composites(Fig. 7). This is done fibres promised the use temperatures up to.C(see weak and strong (real) interfaces does actually represent the macrostructures of brittle-fibre/brittle-matrix compos- tes with enhanced fracture toughness. The first one is based on introducing a"weak"interphase between the fibre and matrix, which deviates a macrocrack and pro- vides the energy dissipation due to pull-out(see, for exam- 240 Eutectic-fibre/Mo-matrix Sapphire-fibre/TiAl-matrix, 900-1000"C ple [27]). The second one used mostly for oxide/oxide composites is actually nearly the same, just the weak inter- 200 face occurs as a result of using a porous matrix [28, 29 The ICM-fibres can obviously be used in both two ways The measurements of values of the critical stress intensity factor K" for sapphire/carbon-interphase/alumina compos- ites with fibre volume fraction of 0. 12-0. 28 yield values of critical stress intensity factor equal to 11.4+0.9 MPa m"/2 as compared with corresponding values for un-reinforced natrix o obtained under the same conditions of 5.9+1.1 MPa m/2[30]. Results of the experimental study of creep behaviour of such model composites at 1200C [31] show that their creep behaviour could be described by Eq(6). The calculated dependence of the creep resis- tance(stress to cause 1% creep strain for 100 h)on fibre volume fraction is presented in Fig. 15. The experimental Fibre volume fraction data for four specimens tested are given in Table 2, and typical microstructures of the composites are presented in Fig. 14. Creep resistance versus fibre volume fraction of ICM-oxide- matrix/nickel Fig. 16. Specimen A2053 is characterised by extremely iven in Table le l. Also expe perimental points for the alumina-YAG-eutectic- non-homogeneous fibre packing: it causes a decrease fibre/TiAl-matrix(see trix composite taken from Ref [8]and sapphire- the creep resistance. It should be noted that the effective Fig. 10)composit creep properties of the matrix occur to be quite moderate
stronger the interface, the larger portion of the fibre surface is in contact with the matrix. A continuous interface was observed in the Mo-based composite. Therefore, the assumption made in modelling creep behaviour, which yields Eq. (6), namely introducing continuity factor a, has a physical meaning rather than produces just a phenomenological parameter in the mechanical model. With increasing the value of a a larger portion of the surface defects are healed that yields an increase in the fibre effective strength. This effect is added to a normal strength/ scale effect. It is informative to replot the calculated dependencies presented in Fig. 13b together with that for an ideal interface (a = 1), which corresponds also to an increase in the characteristic fibre strength to a value determined by testing oxide/molybdenum composites (Fig. 7). This is done in Fig. 14. A field between the lines corresponding to weak and strong (real) interfaces does actually represent creep properties that have been really observed for a family of ICM-oxide–matrix/nickel-based-matrix composites at 1150 C. The upper line corresponds to an upper limit of the creep resistance of ICM-oxide–matrix/ metal–matrix composites at 1150 C. It should be emphasized that the experimental point obtained earlier for alumina–YAG-eutectic–matrix/molybdenum–matrix composite at the same temperature and sapphire–fibre/TiAl– matrix composite at 900–1000 C lie nearly on this line. Note that both composites are characterised by an ideal interface. The experiments and results of calculations based on the micro-mechanical model allow expecting a maximum use temperature for the oxide–fibre/Ni-based-matrix system to be at least 1150 C [16]. The criterion for the maximum temperature is that specific creep resistance of a composite is equal to that of a superalloy with density of 8 g/cm3 at 150 MPa. A further increase in the use temperature of composites in the oxide/Ni-system depends on the ability to appropriately control the fibre/matrix interface. 4. Oxide–fibre/oxide–matrix composites The fibre/matrix interface in such composites made of brittle continuances entirely determines their applicability as structural materials. The problem is of importance since such composites being reinforced with single crystalline fibres promised the use temperatures up to 1600 C (see Fig. 8). There have been known two ways of organizing the macrostructures of brittle–fibre/brittle–matrix composites with enhanced fracture toughness. The first one is based on introducing a ‘‘weak’’ interphase between the fibre and matrix, which deviates a macrocrack and provides the energy dissipation due to pull-out (see, for example [27]). The second one used mostly for oxide/oxide composites is actually nearly the same, just the weak interface occurs as a result of using a porous matrix [28,29]. The ICM-fibres can obviously be used in both two ways. The measurements of values of the critical stress intensity factor K* for sapphire/carbon-interphase/alumina composites with fibre volume fraction of 0.12–0.28 yield values of critical stress intensity factor equal to 11.4 ± 0.9 MPa m1/2 as compared with corresponding values for un-reinforced matrix obtained under the same conditions of 5.9 ± 1.1 MPa m1/2 [30]. Results of the experimental study of creep behaviour of such model composites at 1200 C [31] show that their creep behaviour could be described by Eq. (6). The calculated dependence of the creep resistance (stress to cause 1% creep strain for 100 h) on fibre volume fraction is presented in Fig. 15. The experimental data for four specimens tested are given in Table 2, and typical microstructures of the composites are presented in Fig. 16. Specimen A2053 is characterised by extremely non-homogeneous fibre packing; it causes a decrease in the creep resistance. It should be noted that the effective creep properties of the matrix occur to be quite moderate. Table 1 Constituents properties used in model calculations Interface property Fibre properties (lo = 1 mm, d = 0.1 mm) a brðfÞ 0 (MPa) Weak interface, Vf > 0.25 0.01 3 150 Strong interface, Vf < 0.15 0.5 3 450 Ideal interface 1 3 600 Matrix properties gm = 104 h1 , m = 2.8, rm = 41.5 MPa Note: The fibre strength at a = 1 corresponds to data presented in Fig. 7. Fig. 14. Creep resistance versus fibre volume fraction of ICM-oxide– matrix/nickel-based-matrix composites. Parameters in the creep model are given in Table 1. Also experimental points for the alumina–YAG-eutectic– matrix/molybdenum–matrix composite taken from Ref. [8] and sapphire– fibre/TiAl–matrix (see Fig. 10) composites are shown. S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229 227
S.T. Mileiko Current Opinion in Solid State and Materials Science 9(2005)219-229 120 n=7(calculated value) 100. 6 MPa The average fibre volume fraction is 0.26 Fig. 15. The calculated creep resistance of sapphire-fibre/alumina-matri composites at1200°C Fig 17. A load versus displacement curve obtained in testing YAG-fibre/ YAG-matrix specimen in 3-point bending. An increase in the slope of the curve is an effect caused by a testing machine and not by a special behaviour of a composite Hence, really high creep resistant composites of such types Experimental creep characteristics of sapphire-fibre/alumina-matrix com- are expected when YAG and mullite fibres are used in poly crystalline YAG and mullite matrices eep resistance(MPa) However, stability of macrostructures of both types A2005 very high temperatures for long time can hardly be obtained due to fibre/interphasematrix interaction and A2034 sintering the porous matrix. Hence, the authors' lab is A2053 now looking for new ways of organising non-brittle a2005 a2053 a2034 Fig. 16. Macrostructure of some sapphire/alumina specimens
Hence, really high creep resistant composites of such types are expected when YAG and mullite fibres are used in polycrystalline YAG and mullite matrices. However, stability of macrostructures of both types at very high temperatures for long time can hardly be obtained due to fibre/interphase/matrix interaction and sintering the porous matrix. Hence, the authors’ lab is now looking for new ways of organising non-brittle Table 2 Experimental creep characteristics of sapphire–fibre/alumina–matrix composite specimens Specimen number n Creep resistance (MPa) A2005 6.01 94.9 A2007 5.60 77.8 A2034 5 64.3 A2053 2.85 38.1 Fig. 16. Macrostructure of some sapphire/alumina specimens. Fig. 15. The calculated creep resistance of sapphire–fibre/alumina–matrix composites at 1200 C. Fig. 17. A load versus displacement curve obtained in testing YAG–fibre/ YAG–matrix specimen in 3-point bending. An increase in the slope of the curve is an effect caused by a testing machine and not by a special behaviour of a composite. 228 S.T. Mileiko / Current Opinion in Solid State and Materials Science 9 (2005) 219–229