《复合材料 Composites》课程教学资源（学习资料）第二章 增强体_nicalon-1 A review of the development of three generations of small diameter silicon carbide fibres

40TH ANNIVERSARY J MATER SCI41(006)823-839 A review of the development of three generations of small diameter silicon carbide fibres A.R. BUNSELL A. IANT Ecole des Mines de Paris, Centre des Materiaux, BP87, 91003 Evry Cedex, France Three generations of small diameter ceramic fibres based on polycrystalline silicon carbide have been developed over a period of thirty years. This has been possible due to studies into the relationships between the microstructures and properties of the fibres a variety of techniques have been employed by research teams on three continents the fibres are made by he conversion of polymer precursors to ceramic fibres and all three generations are presently produced commercially. The nature of the precursor and the techniques used for cross-linking have been varied in order to optimise both properties and cost of manufacture. It has been possible to improve the characteristics of the fibres as the processes involved in the cross-linking of the precursor fibres have been better understood and the mechanisms governing both room temperature and high temperature behaviour determined. The result is that, although first generation fibres were limited by a low Young s modulus at room limiting the behaviour of bulk silicon carbide, the third generation fibres shows many ot tho oo temperature and by creep and instability of the structure at temperatures far lower than thos characteristics of stoichiometric silicon carbide. This remarkable improvement in characteristics has been due to a thorough understanding of the materials science governing the behaviour of these fibres which are reinforcements for ceramic matrix composite materials 2006 Springer Science Business Media, Inc. 1 Introduction The PCs obtained in this way could be melt spun to give Silicon carbide fibres, with diameters of around 15 um, weak fibres. Stabilisation was initially by cross-linking were first produced commercially in 1982 by Nippon Car- of the polymer by heating in air, just as in the carbon bon. This industrial production was the direct result of fibre production route using PAN precursors. This was research started in the 1970s and carried out by Professor followed by heating in vacuum at temperatures, generally Yajima and his team at the Tohoku University in Japan. around 1200oC and allowed the first generation of small The approach adopted to produce SiC fibres owed much diameter Sic fibres to be produced. The availability of to the experience gained from the development of carbon these Sic fibres brought rapid interest from the aerospace fibres which involves the spinning of polyacrylonitrile and aero-engine industries as they offered the possibil PAN) precursor fibres, which are stabilised by cro ity of producing ceramic fibre reinforced carbon and ce linking and then pyrolysed under controlled conditions ramic matrix composites materials, capable of being used to give carbon fibres. The starting polymer for produc- as structural materials to higher temperatures than those ing SiC fibres, by necessity, needed to contain silicon and attainable with the best nickel based super-alloys. The at- carbon atoms and the polycarbosilane(PCS), which was traction of silicon carbide is that it is a ceramic, which in chosen as the starting material, comprised these elements bulk form, has a Youngs modulus twice that of steel for arranged in a cyclic form consisting of six atoms, which less than half the density and can be used up to 1600c suggested their arrangement in B-SiC. The synthesis of Although oxidised at high temperature, bulk Sic under the PCS used dimethyldichlorosilane( CH3)2 SiCI which goes surface passive oxidation which protects the bulk of was converted into polydimethylsilane [(CH3) Si]n, by the specimen. However it was found that the characteris dechlorination with metal sodium and which in turn was tics of these first generation fibres were not those of bulk converted into a polycarbosilane polymer by heating in SiC. The fibres possessed a Youngs modulus less than an inert atmosphere at 4000C [1]. The chemical compo- half that expected. The fibres crept at 1000C and above sition of PCS can be simplified as -[SiCH3H-CH2In and degraded above 1250%C. The understanding of the 0022-2461 2006 Springer Science+Business Media, Inc DOI:10.1007/10853-0066566-z 823

40TH ANNIVERSARY J MATER SCI 4 1 (2 0 0 6 ) 8 2 3 –8 3 9 A review of the development of three generations of small diameter silicon carbide fibres A. R. BUNSELL, A. PIANT Ecole des Mines de Paris, Centre des Materiaux, BP87, 91003 Evry Cedex, France ´ Three generations of small diameter ceramic fibres based on polycrystalline silicon carbide have been developed over a period of thirty years. This has been possible due to studies into the relationships between the microstructures and properties of the fibres. A variety of techniques have been employed by research teams on three continents. The fibres are made by the conversion of polymer precursors to ceramic fibres and all three generations are presently produced commercially. The nature of the precursor and the techniques used for cross-linking have been varied in order to optimise both properties and cost of manufacture. It has been possible to improve the characteristics of the fibres as the processes involved in the cross-linking of the precursor fibres have been better understood and the mechanisms governing both room temperature and high temperature behaviour determined. The result is that, although first generation fibres were limited by a low Young’s modulus at room temperature and by creep and instability of the structure at temperatures far lower than those limiting the behaviour of bulk silicon carbide, the third generation fibres shows many of the characteristics of stoichiometric silicon carbide. This remarkable improvement in characteristics has been due to a thorough understanding of the materials science governing the behaviour of these fibres which are reinforcements for ceramic matrix composite materials.
C 2006 Springer Science + Business Media, Inc. 1. Introduction Silicon carbide fibres, with diameters of around 15 µm, were first produced commercially in 1982 by Nippon Car￾bon. This industrial production was the direct result of research started in the 1970s and carried out by Professor Yajima and his team at the Tohoku University in Japan. The approach adopted to produce SiC fibres owed much to the experience gained from the development of carbon fibres which involves the spinning of polyacrylonitrile (PAN) precursor fibres, which are stabilised by cross￾linking and then pyrolysed under controlled conditions to give carbon fibres. The starting polymer for produc￾ing SiC fibres, by necessity, needed to contain silicon and carbon atoms and the polycarbosilane (PCS), which was chosen as the starting material, comprised these elements arranged in a cyclic form consisting of six atoms, which suggested their arrangement in β-SiC. The synthesis of the PCS used dimethyldichlorosilane (CH3)2SiCl2 which was converted into polydimethylsilane [(CH3)2Si]n, by dechlorination with metal sodium and which in turn was converted into a polycarbosilane polymer by heating in an inert atmosphere at 400◦C [1]. The chemical compo￾sition of PCS can be simplified as –[SiCH3H-CH2]n –. The PCS obtained in this way could be melt spun to give weak fibres. Stabilisation was initially by cross-linking of the polymer by heating in air, just as in the carbon fibre production route using PAN precursors. This was followed by heating in vacuum at temperatures, generally around 1200◦C and allowed the first generation of small￾diameter SiC fibres to be produced. The availability of these SiC fibres brought rapid interest from the aerospace and aero-engine industries as they offered the possibil￾ity of producing ceramic fibre reinforced carbon and ce￾ramic matrix composites materials, capable of being used as structural materials to higher temperatures than those attainable with the best nickel based super-alloys. The at￾traction of silicon carbide is that it is a ceramic, which in bulk form, has a Young’s modulus twice that of steel for less than half the density and can be used up to 1600◦C. Although oxidised at high temperature, bulk SiC under￾goes surface passive oxidation which protects the bulk of the specimen. However it was found that the characteris￾tics of these first generation fibres were not those of bulk SiC. The fibres possessed a Young’s modulus less than half that expected. The fibres crept at 1000◦C and above and degraded above 1250◦C. The understanding of the 0022-2461
C 2006 Springer Science + Business Media, Inc. DOI: 10.1007/s10853-006-6566-z 823

40TH ANNIVERSARY material science involved in the processes governing this behaviour and the use of this knowledge to produce fibres CH with greatly enhanced properties have been the preoccu pation of a number of laboratories across the world for the last quarter of a century. The result has been the develop ment of three generations of fibres, the latest of which has produced fibres with properties approaching the limits of what is physically possible with silicon carbide C CH3 Figure I Repeat unit of polycarbosilane(PCS). The numbers allow easy 2. First generation fine Sic fibres The interest in silicon carbide as a reinforcement has prompted the development of several types of fibres in luding fibres with diameters usually greater than 100 um made by CVD onto a core filament and also monocrys- talline short filaments known as whiskers with diameters of the order of l um [2, 3]. The Cvd fibre is finding inter est as a reinforcement for titanium but its large diameter handling processes, which are commonly used with finer exploited in the production of SiC fibres. The numbers refer to the positions fibres used in the majority of composite materials. The Sic whiskers also present serious handling difficulties to- gether with worries about health related problems The announcement of the production of Sic based fi- by the cong diameters in the range of 10 to 20 um,made of polycarbosilane precursors, excited considerable interest amongst those people looking for reinforcements capable of operating in an oxidising atmo- sphere at over 1000oC, above the limits of nickel based alloys[4, 5]. Yajima and his colleagues explored a number of routes to produce polycarbosilane which could be used as a precursor for a ceramic fibre [5, 6]. The difficulty lay in the production of a form of polycarbosilane which Figure 3 The first generation of fine SiC fibres were made by cross-linking could be spun and converted into ceramic filaments. The the precursor PCS with oxygen. decomposition of polydimethylsilane(PDs) which was heated in an autoclave at 470oC for 14 h, was eventu- cross-linked. The PCs precursor was made infusible, in ally chosen as the route for the production of PCS as it the first generation of fibres, by crosslinking in air, in the gave a precursor which, although difficult to spin, could temperature range from 145 to 200oC, which introduced be spun from the melt and converted into a ceramic fibre. oxygen into the polymer, as shown in Fig 3. The con- The repeat element in the chemical structure of polycar- version of the cross-linked precursor fibre into a ceramic bosilane is given in Fig. 1. A steric view of this molecule fibre was explored again by Yajima et al. [7-9] and would show that the cycle of carbon and silicon atoms, others [10]. The cross-linked PCS fibres were insoluble with some bonds removed, is arranged in the form of a in all solvents. Heating the cross-linked precursor up to chair configuration, as illustrated in Fig. 2. This reflects 550C induced the evaporation of low molecular weight the arrangement seen in B-SiC. The numbering of the ar- components in the carbosilane which led to a consider- rangement of the atoms is so that a comparison can be able weight loss but resulted in an increase in molecular made to those in Fig. 1. The groups, shown in Fig. l, weight. Above this temperature and up to around 800oC, numbered 2, 3, 5 and 6 are in the same plane whilst 1 hydrogen and methane were lost from the side groups and 4 are out of this plane "Me"represents the methyl in the PCS leaving behind free carbon and cross-linking group(CH3). Unlike most polymers which are spun into was enhanced. Further heating to 1200oC showed that fibres, it was found that very high molecular weight was gas evolution was almost complete at 1000oC At 1050C not necessarily best and polymers with molecular weights hydrogen was again given off and the XRD patterns be- of around 1500 were eventually used for commercial pro- came sharper indicating greater regularity in the struc- duction. The precursor filaments were then spun from the ture. At and above 1300C the free carbon created by the melt in a nitrogen atmosphere at around 300C and then destruction of the methyl groups reacted with the Si-o 824

40TH ANNIVERSARY material science involved in the processes governing this behaviour and the use of this knowledge to produce fibres with greatly enhanced properties have been the preoccu￾pation of a number of laboratories across the world for the last quarter of a century. The result has been the develop￾ment of three generations of fibres, the latest of which has produced fibres with properties approaching the limits of what is physically possible with silicon carbide. 2. First generation fine SiC fibres The interest in silicon carbide as a reinforcement has prompted the development of several types of fibres in￾cluding fibres with diameters usually greater than 100 µm made by CVD onto a core filament and also monocrys￾talline short filaments, known as whiskers, with diameters of the order of 1 µm [2, 3]. The CVD fibre is finding inter￾est as a reinforcement for titanium but its large diameter makes it unsuitable for weaving and other types of fibre handling processes, which are commonly used with finer fibres used in the majority of composite materials. The SiC whiskers also present serious handling difficulties to￾gether with worries about health related problems. The announcement of the production of SiC based fi- bres having diameters in the range of 10 to 20 µm, made by the conversion of polycarbosilane precursors, excited considerable interest amongst those people looking for reinforcements capable of operating in an oxidising atmo￾sphere at over 1000◦C, above the limits of nickel based alloys [4, 5]. Yajima and his colleagues explored a number of routes to produce polycarbosilane which could be used as a precursor for a ceramic fibre [5, 6]. The difficulty lay in the production of a form of polycarbosilane which could be spun and converted into ceramic filaments. The decomposition of polydimethylsilane (PDS) which was heated in an autoclave at 470◦C for 14 h, was eventu￾ally chosen as the route for the production of PCS as it gave a precursor which, although difficult to spin, could be spun from the melt and converted into a ceramic fibre. The repeat element in the chemical structure of polycar￾bosilane is given in Fig. 1. A steric view of this molecule would show that the cycle of carbon and silicon atoms, with some bonds removed, is arranged in the form of a chair configuration, as illustrated in Fig. 2. This reflects the arrangement seen in β-SiC. The numbering of the ar￾rangement of the atoms is so that a comparison can be made to those in Fig. 1. The groups, shown in Fig. 1, numbered 2, 3, 5 and 6 are in the same plane whilst 1 and 4 are out of this plane. “Me” represents the methyl group (CH3). Unlike most polymers which are spun into fibres, it was found that very high molecular weight was not necessarily best and polymers with molecular weights of around 1500 were eventually used for commercial pro￾duction. The precursor filaments were then spun from the melt in a nitrogen atmosphere at around 300◦C and then Figure 1 Repeat unit of polycarbosilane (PCS). The numbers allow easy reference to Fig. 2. CH Si CH2 SiMe 1 2 3 6 H 4 5 CH2 SiMe 1 Figure 2 The steric conformation of the repeat unit of polycarbosilane exploited in the production of SiC fibres. The numbers refer to the positions shown in Fig. 1. Figure 3 The first generation of fine SiC fibres were made by cross-linking the precursor PCS with oxygen. cross-linked. The PCS precursor was made infusible, in the first generation of fibres, by crosslinking in air, in the temperature range from 145 to 200◦C, which introduced oxygen into the polymer, as shown in Fig. 3. The con￾version of the cross-linked precursor fibre into a ceramic fibre was explored again by Yajima et al. [7–9] and others [10]. The cross-linked PCS fibres were insoluble in all solvents. Heating the cross-linked precursor up to 550◦C induced the evaporation of low molecular weight components in the carbosilanes which led to a consider￾able weight loss but resulted in an increase in molecular weight. Above this temperature and up to around 800◦C, hydrogen and methane were lost from the side groups in the PCS leaving behind free carbon and cross-linking was enhanced. Further heating to 1200◦C showed that gas evolution was almost complete at 1000◦C. At 1050◦C hydrogen was again given off and the XRD patterns be￾came sharper indicating greater regularity in the struc￾ture. At and above 1300◦C the free carbon created by the destruction of the methyl groups reacted with the Si–O 824

40TH ANNIVERSARY TABLE I Compositions of early varieties of first generation SiC fibres TABLE II Compositions, Youngs moduli and densities of first gener- produced by Nippon Carbon ation commercialised SiC fibres hemical composition Nippon Carbon Ube Industries Elemental composition %owt Fibre name Nicalon 200 Tyranno LOX-M C SiO, C Precursor PCS Cured by Oxidation xidation NLP10160 Si(wt%) NLM-1025 C(wt%) 31.6 NLP20254 O(wt%) 117 Ti(wt%) C/Si 1.31 1.36 group with the evolution of Co gas and forming a Si-c Young s modm ys (Pa bond. The nascent B-SiC grains, formed at slightly lower temperatures increased in size and the amorphous struc- ture evolved into a semi-crystalline structure consisting of nano sized B-SiC grains surrounded by a much less ordered phase made up of silicon, carbon and oxygen. [12]. The Tyranno fibres could be made with diameters Heating to 1500 C produced large grain growth, the evo- half that of the Nicalon fibres. The precursors of these first lution of carbon monoxide and the disintegration of the Tyranno fibres were also crosslinked in air. Ube Industries fibr used a code to indicate the oxygen content of the fibres The first fibres of this first generation, which were so that the fibre which was commercialised was known as made available by Nippon Carbon around 1982, were Tyranno LOX-M, with the letter M, which is the thirteenth the Nicalon 100 series but were replaced after about four letter in the alphabet, representing an oxygen content of years by the Nicalon 200 series which became the standard approximately 13%by weight grade for much of the ceramic matrix composite studies By the end of the 1980s the two Japanese companies subsequently undertaken. These fibres had diameters of were producing first generation fine diameter Sic around 15 um but showed variability in diameter along fibres and their compositions and densities are shown their length because of the difficulties of spinning the pre- in Table Il. The details of the composition and the cursor fibres, as can be seen from Fig 4. Table I shows nomenclature used to describe the fibres have changed the approximate chemical composition of these fibre slightly since their initial introduction so that the nicalon Yajima and his colleagues had considered several routes fibres mentioned in Table I are now grouped under the to making SiC fibres and one included the addition of simpler heading of Nicalon 100 or 200 series. Bulk Sic titanium to the PCS so as to give polytitanocarbosilane is however the second hardest material known and is (PTC)[11]. This precursor was obtained by the grafting crystalline, it possesses a Youngs modulus of around of titanium alkoxide, Ti(OR)4, in which=CnH2n+l, onto 400 GPa and a density of 3. 15 g/cm and can be used in the PCS chains. This linked the polymer chains together, air up to 1600oC. At this temperature passive oxidation of increasing molecular weight and its spinability. In 1987, the surface to Sio2 protects it from further degradation another Japanese company, Ube Industries announced the Table II reveals that the properties of the first generation production of Tyranno fibres made from PTC precursors of fine Sic fibres were not those of the bulk material and reported that they had better thermal and chemical This has been shown to be due to the non stoichiometric stability compared with the then-existent Nicalon fibres composition of the fibres, which are rich in carbon and contain oxygen, as shown in Tables I and Il. An alternative approach to producing polymer derived SiC fibres, similar to the first generation fibres which have been developed commercially, has been described by university researchers. In contrast to the manufactur ing technique described above, it has been shown that such fibres could be produced from precursor filaments made 864 from high-molecular-weight PCS [13]. The infusible PCS was prepared by pressure pyrolysis of polydimethylsilane Fine fibres were formed by the dry spinning of concen- trated PCs-based polymer solutions which were then py 78910111213141516118192021 rolised in an inert atmosphere in the temperature range of 1000 to 1200C. The fibres produced in this manner were Figure 4 Range of diameters observed with Nicalon NLM 202 fibre taken reported as possessing very similar properties to those of from the same tow [43 the commercialised first generation fibres [14] 825

40TH ANNIVERSARY T A B L E I Compositions of early varieties of first generation SiC fibres produced by Nippon Carbon Elemental composition %wt Chemical composition %wt Fibre type Si C O SiC SiO2 C NLP-101 60 27 13 69 24 7 NLM-102 54 34 12 63 21.5 15.5 NLP-202 54 37 9 66 17 17 group with the evolution of CO gas and forming a Si–C bond. The nascent β-SiC grains, formed at slightly lower temperatures increased in size and the amorphous struc￾ture evolved into a semi-crystalline structure consisting of nano sized β-SiC grains surrounded by a much less ordered phase made up of silicon, carbon and oxygen. Heating to 1500◦C produced large grain growth, the evo￾lution of carbon monoxide and the disintegration of the fibre. The first fibres of this first generation, which were made available by Nippon Carbon around 1982, were the Nicalon 100 series but were replaced after about four years by the Nicalon 200 series which became the standard grade for much of the ceramic matrix composite studies subsequently undertaken. These fibres had diameters of around 15 µm but showed variability in diameter along their length because of the difficulties of spinning the pre￾cursor fibres, as can be seen from Fig. 4. Table I shows the approximate chemical composition of these fibres. Yajima and his colleagues had considered several routes to making SiC fibres and one included the addition of titanium to the PCS so as to give polytitanocarbosilane (PTC) [11]. This precursor was obtained by the grafting of titanium alkoxide, Ti(OR)4, in which R=CnH2n+1, onto the PCS chains. This linked the polymer chains together, increasing molecular weight and its spinability. In 1987, another Japanese company, Ube Industries announced the production of Tyranno fibres made from PTC precursors and reported that they had better thermal and chemical stability compared with the then-existent Nicalon fibres 0 2 4 6 8 10 12 14 16 18 20 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 24 25 Diameter (µm) % Figure 4 Range of diameters observed with Nicalon NLM 202 fibre taken from the same tow [43]. T AB L E I I Compositions, Young’s moduli and densities of first gener￾ation commercialised SiC fibres Producer Nippon Carbon Ube Industries Fibre name Nicalon 200 Tyranno LOX-M Precursor PCS PTC Cured by Oxidation Oxidation Si (wt%) 56.6 54 C (wt%) 31.7 31.6 O (wt%) 11.7 12.4 Ti (wt%) 0 2.0 C/Si 1.31 1.36 Young’s modulus (GPa) 200 185 Density (g/cm3) 2.55 2.37 [12]. The Tyranno fibres could be made with diameters half that of the Nicalon fibres. The precursors of these first Tyranno fibres were also crosslinked in air. Ube Industries used a code to indicate the oxygen content of the fibres so that the fibre which was commercialised was known as Tyranno LOX-M, with the letter M, which is the thirteenth letter in the alphabet, representing an oxygen content of approximately 13% by weight. By the end of the 1980s the two Japanese companies were producing first generation fine diameter SiC fibres and their compositions and densities are shown in Table II. The details of the composition and the nomenclature used to describe the fibres have changed slightly since their initial introduction so that the Nicalon fibres mentioned in Table I are now grouped under the simpler heading of Nicalon 100 or 200 series. Bulk SiC is however the second hardest material known and is crystalline, it possesses a Young’s modulus of around 400 GPa and a density of 3.15 g/cm3 and can be used in air up to 1600◦C. At this temperature passive oxidation of the surface to SiO2 protects it from further degradation. Table II reveals that the properties of the first generation of fine SiC fibres were not those of the bulk material. This has been shown to be due to the non stoichiometric composition of the fibres, which are rich in carbon and contain oxygen, as shown in Tables I and II. An alternative approach to producing polymer derived SiC fibres, similar to the first generation fibres which have been developed commercially, has been described by university researchers. In contrast to the manufactur￾ing technique described above, it has been shown that such fibres could be produced from precursor filaments made from high-molecular-weight PCS [13]. The infusible PCS was prepared by pressure pyrolysis of polydimethylsilane. Fine fibres were formed by the dry spinning of concen￾trated PCS-based polymer solutions which were then py￾rolised in an inert atmosphere in the temperature range of 1000 to 1200◦C. The fibres produced in this manner were reported as possessing very similar properties to those of the commercialised first generation fibres [14]. 825

40TH ANNIVERSARY TABLE III Details of manufacture, elemental composition and approximate cost of all three generations of SiC based fibres maximum Cross linki Elemental composition Trade mark Manufacture temperature (wt%) (g/cm)(um)(US S/g First Gen. Nicalon 200 Nippon Carbon Oxygen 1200°C 56si+32C+120 2.5514 2000 Tyranno LOX-M Ube Ind. 1200°C 54Si+32C+120+2T24811 Second Hi-Nicalon Electron °C 1300° 625Si+37C+0502.7412 8000 Gen irradiation Tyranno LOX-E Ube Ind Electron 1300°C 55Si+37.5+5.50+2T2.3911 radiation Tyranno ZM Ube Ind Oxygen 1300°C 57Si+345C+7.50+248 11 1500 Tyranno ZE Ube Ind Electron 1300°C 585Si+38.5C+20+2.5511 Third Gen. Tyranno SA 68Si+32C+0.6Al Tyranno SA 3 Ind 68Si+32C+06Al Sylramic COI ceramics 67Si+29c+0.80+3.0 23B+04N+2.1T Sylramic iB COI Ceramics Oxygen >1700°CN/A 3.05 002 Hi-Nicalon Type-S Nippon Carbon Electron 1500°C 69Si+31C+0.20 3.05 3000 irradiation TABLE IV Details of mechanical and thermal properties of all three generations of SiC based fibres Thermal expansion Room om temperature coefficient, ppm/C(to emal conductivity W/m tempera gs modulus Trade mark Manufacturer 1000°C)[1 K[15 rength(GPa)(GPa) First gen Nicalon 200 Nippon Carbon 3.2 Second Gen. Hi-Nicalon Nippon Carbon 3.5 Tyranno LOX-E Ube Ind Tyranno ZM Third gen Tyranno SA3 Sylramic iBN Hi-Nicalon Type-s Nippon Carbon NA 3. Mechanical behaviour of first generation Sic The behaviour of the first generation fibres remains lin- fibres early elastic up to 1250C but short term strength begins to At room temperature the fibres show linearly elastic be- fall around 1000oC. There is a difference, particularly for haviour. The variation in fibre diameter along individual the Tyranno LOX-M fibre, when tested in air or in argon, fibres makes the measurement of stress and modulus in- with an earlier onset of strength reduction being observed herently difficult which explains some discrepancies in the when the fibres are tested in air indicating a higher sen- published data for these fibres which in any case have been sitivity to carbon oxidation of the surfaces. However the improved since their introduction. Some typical property oxidation of the Nicalon 100 series fibres could be bene- data for first generation SiC fibres can be found in Tables ficial particularly under long term loading conditions as it II and IV. It should be noted that the Tyranno LOX-M, al- slowed internal decomposition of the fibres [17]. Growth nerally available. Fig. 5 shows of silica is observed on the surfaces of both fibre the fracture morphology of a first generation Nicalon fi- they are heated in air at 1200C and above in air. This bre broken in tension [15]. It can be seen that the fracture layer can have an irregular thickness along the fibre and suggests a glassy structure of the fibre. A critical stress pores are formed at the silica/SiC fibre interface and pores intensity factor K lc of 2 MPa.m was determined for the can be formed at 1450C which induce local decohesion Nicalon NLM-202 fibre which is more characteristic of of the silica layer from the fibre. These pores are pro glass than bulk Sic [16] duced by the outgassing of carbon monoxide from the

40TH ANNIVERSARY T A B L E I I I Details of manufacture, elemental composition and approximate cost of all three generations of SiC based fibres Trade mark Manufacturer Cross linking method Approximate maximum production temperature Elemental composition (wt%) Density (g/cm3) Average diameter (µm) Cost (US $/kg) First Gen. Nicalon 200 Nippon Carbon Oxygen 1200◦C 56Si + 32C + 12O 2.55 14 2000 Tyranno LOX-M Ube Ind. Oxygen 1200◦C 54Si + 32C +12O + 2Ti 2.48 11 1250 Second Gen. Hi-Nicalon Nippon Carbon Electron irradiation 1300◦C 62.5Si + 37C +O.5O 2.74 12 8000 Tyranno LOX-E Ube Ind. Electron irradiation 1300◦C 55Si + 37.5 +5.5O + 2Ti 2.39 11 N/A Tyranno ZM Ube Ind. Oxygen 1300◦C 57Si + 34.5C + 7.5O + 1Zr 2.48 11 1500 Tyranno ZE Ube Ind. Electron irradiation 1300◦C 58.5Si + 38.5C + 2O + 1Zr 2.55 11 N/A Third Gen. Tyranno SA 1 Ube Ind. Oxygen >1700◦C 68Si + 32C +0.6Al 3.02 11 N/A Tyranno SA 3 Ube Ind. Oxygen >1700◦C 68Si + 32C +0.6Al 3.1 7.5 5000 Sylramic COI ceramics Oxygen >1700◦C 67Si + 29C +0.8O + 2.3B +0.4N +2.1Ti 3.05 10 10000 Sylramic iBN COI Ceramics Oxygen >1700◦C N/A 3.05 10 >10000 Hi-Nicalon Type-S Nippon Carbon Electron irradiation >1500◦C 69Si + 31C + 0.2O 3.05 12 13000 T A B L E I V Details of mechanical and thermal properties of all three generations of SiC based fibres Trade mark Manufacturer Thermal expansion coefficient, ppm/◦C (to 1000◦C) [15] Room temperature axial thermal conductivity W/m K [15] Room temperature strength (GPa) Room temperature Young’s modulus (GPa) First Gen. Nicalon 200 Nippon Carbon 3.2 3 3 200 Tyranno LOX-M Ube Ind. 3.1 1.5 3.3 185 Second Gen. Hi-Nicalon Nippon Carbon 3.5 8 2.8 270 Tyranno LOX-E Ube Ind. NA NA 2.9 200 Tyranno ZM Ube Ind. NA 2.5 3.4 200 Tyranno ZE Ube Ind. NA NA 3.5 233 Third Gen. Tyranno SA1 Ube Ind. NA 65 2.8 375 Tyranno SA3 Ube Ind. NA 65 2.9 375 Sylramic COI Ceramics 5.4 46 3.2 400 Sylramic iBN COI Ceramics 5.4 >46 3.5 400 Hi-Nicalon Type-S Nippon Carbon NA 18 2.5 400 3. Mechanical behaviour of first generation SiC fibres At room temperature the fibres show linearly elastic be￾haviour. The variation in fibre diameter along individual fibres makes the measurement of stress and modulus in￾herently difficult which explains some discrepancies in the published data for these fibres which in any case have been improved since their introduction. Some typical property data for first generation SiC fibres can be found in Tables III and IV. It should be noted that the Tyranno LOX-M, al￾though still made, is not generally available. Fig. 5 shows the fracture morphology of a first generation Nicalon fi- bre broken in tension [15]. It can be seen that the fracture suggests a glassy structure of the fibre. A critical stress intensity factor K1c of 2 MPa.m1/2 was determined for the Nicalon NLM-202 fibre which is more characteristic of glass than bulk SiC [16]. The behaviour of the first generation fibres remains lin￾early elastic up to 1250◦C but short term strength begins to fall around 1000◦C. There is a difference, particularly for the Tyranno LOX-M fibre, when tested in air or in argon, with an earlier onset of strength reduction being observed when the fibres are tested in air indicating a higher sen￾sitivity to carbon oxidation of the surfaces. However the oxidation of the Nicalon 100 series fibres could be bene- ficial particularly under long term loading conditions as it slowed internal decomposition of the fibres [17]. Growth of silica is observed on the surfaces of both fibres when they are heated in air at 1200◦C and above in air. This layer can have an irregular thickness along the fibre and pores are formed at the silica/SiC fibre interface and pores can be formed at 1450◦C which induce local decohesion of the silica layer from the fibre. These pores are pro￾duced by the outgassing of carbon monoxide from the 826

40TH ANNIVERSARY grain growth depended on the temperature but the time to stabilisation corresponded to the period of primary creep observed at higher loads. At temperatures above 1150oC, for the Nicalon 200 fibres and above 1050. for the served that it was possible to measure an initial shrinkag followed by positive creep showing that two mechanisms were in competition. The variation of steady state creep rate with temperature, T, and applied stress, o, can often be modelled by: 5um where A is a constant depending on the material. The stress exponent, n, and the activation energy for creep, 2, can be deduced from creep experiments and their val- Figure 5 Fracture morphology of a first generation Nicalon fibre [15]- ues can suggest the processes involved. Below 1200 C, the stress component is near to unity for Nicalon 200 fi interior of the fibre. Failure surfaces remain brittle at high bres revealing Newtonian creep, most probably caused by temperature but new types of defects are seen compared grain boundary sliding and controlled by the oxygen rich to those found at room temperature. Local chemical in- homogeneities at the fibres surfaces such as carbon rich increases to 2 as the intergranular phase decomposes and grain sliding becomes more difficult. zones are preferentially decomposed or oxidized giving The two mechanisms which were in competition were rise to porous weak regions The fibres were seen to creep from around 1000C[17] therefore grain growth and perfection and grain bound although, when first observed, this was considered a con- ary sliding. The results are that the creep curves of first troversial observation as most studies indicated that the generation fibres show primary creep which lasts several fibres shrank on heating above this temperature. These hours followed by secondary creep with no tertiary creep stage up to 1250C Shrinkage could be reduced in these latter conclusions were based on fibres heat treated un- fibres by heat treatment at 1200 C in Argon for 5 h which der no imposed load however the creep observations were induces densification and an increase in Youngs modulus made on fibres subjected to loads at temperature. This of around 15 GPa. The creep activation energy was found still leads to some confusion in the literature with some researchers referring to fibre stability as reflecting the to be around 250kJ/mol characteristics of the fibres, tested at room temperature As the applied stress was increased, at a given tem- after heat treatment. The present authors prefer to consider perature, the period during which shrinkage was ob- the characteristics under load at high temperature, which served decreased until a stress was reached at which only positive creep was measured although, as inti better reflects possible end use conditions. The strength mated above, the processes inducing shrinkage still and elastic moduli characteristics of a material are or- occurred and controlled the period of primary creep ten altered at high temperature with respect to those at Table v shows the stresses as a function of temperature room temperature, because of reversible mechanisms at below which shrinkage has been observed for the two the level of the finest structure of the material and so do not become apparent when the temperature is lowered. fibres Exposure to high temperatures can also, of course, induce Creep curves of the first generation fibres revealed the irreversible changes in the fibre structure which also need existence of stress level thresholds, defined as a creep rate to be understood. The testing conditions for defining sta bility at temperature will be explained as necessary in this TABLE V Maximum stress at which shrinkage has been detected in first ge eneration SiC based fibres Further studies on the first generation fibres revealed that. under low loads. the fibres did shrink but, under Temperature(C) Nicalon 200 Tyranno LOX-M higher loads they crept. This behaviour was observed from around 1000oc. studies on the nicalon 100 series fibres which first revealed this behaviour showed that shrinkage 150 0.34Gl could be attributed to B-Sic grain growth in the fibres 1350 0.18 0. 08 GPa hich stabilised with grain sizes around 3 nm. The rate of 1450 006Gl 0.19 827

40TH ANNIVERSARY Figure 5 Fracture morphology of a first generation Nicalon fibre [15]. interior of the fibre. Failure surfaces remain brittle at high temperature but new types of defects are seen compared to those found at room temperature. Local chemical in￾homogeneities at the fibres’ surfaces such as carbon rich zones are preferentially decomposed or oxidized giving rise to porous weak regions. The fibres were seen to creep from around 1000◦C [17], although, when first observed, this was considered a con￾troversial observation as most studies indicated that the fibres shrank on heating above this temperature. These latter conclusions were based on fibres heat treated un￾der no imposed load however the creep observations were made on fibres subjected to loads at temperature. This still leads to some confusion in the literature with some researchers referring to fibre stability as reflecting the characteristics of the fibres, tested at room temperature, after heat treatment. The present authors prefer to consider the characteristics under load at high temperature, which better reflects possible end use conditions. The strength and elastic moduli characteristics of a material are of￾ten altered at high temperature with respect to those at room temperature, because of reversible mechanisms at the level of the finest structure of the material and so do not become apparent when the temperature is lowered. Exposure to high temperatures can also, of course, induce irreversible changes in the fibre structure which also need to be understood. The testing conditions for defining sta￾bility at temperature will be explained as necessary in this paper. Further studies on the first generation fibres revealed that, under low loads, the fibres did shrink but, under higher loads they crept. This behaviour was observed from around 1000◦C. Studies on the Nicalon 100 series fibres which first revealed this behaviour showed that shrinkage could be attributed to β-SiC grain growth in the fibres which stabilised with grain sizes around 3 nm. The rate of grain growth depended on the temperature but the time to stabilisation corresponded to the period of primary creep observed at higher loads. At temperatures above 1150◦C, for the Nicalon 200 fibres and above 1050◦C, for the Tyranno LOX-M fibres, and under low loads, it was ob￾served that it was possible to measure an initial shrinkage followed by positive creep showing that two mechanisms were in competition. The variation of steady state creep rate with temperature, T, and applied stress, σ, can often be modelled by: ε˙ = Aσn exp
−Q RT where A is a constant depending on the material. The stress exponent, n, and the activation energy for creep, Q, can be deduced from creep experiments and their val￾ues can suggest the processes involved. Below 1200◦C, the stress component is near to unity for Nicalon 200 fi- bres revealing Newtonian creep, most probably caused by grain boundary sliding and controlled by the oxygen rich intergranular phase. Above 1200◦C, the stress component increases to 2 as the intergranular phase decomposes and grain sliding becomes more difficult. The two mechanisms which were in competition were therefore grain growth and perfection and grain bound￾ary sliding. The results are that the creep curves of first generation fibres show primary creep which lasts several hours followed by secondary creep with no tertiary creep stage up to 1250◦C. Shrinkage could be reduced in these fibres by heat treatment at 1200◦C in Argon for 5 h which induces densification and an increase in Young’s modulus of around 15 GPa. The creep activation energy was found to be around 250 kJ/mol. As the applied stress was increased, at a given tem￾perature, the period during which shrinkage was ob￾served decreased until a stress was reached at which only positive creep was measured although, as inti￾mated above, the processes inducing shrinkage still occurred and controlled the period of primary creep. Table V shows the stresses as a function of temperature below which shrinkage has been observed for the two fibres. Creep curves of the first generation fibres revealed the existence of stress level thresholds, defined as a creep rate T A B L E V Maximum stress at which shrinkage has been detected in first generation SiC based fibres Temperature (◦C) Nicalon 200 Tyranno LOX-M 1050 - 1.00 GPa 1150 - 0.65 GPa 1250 0.34 GPa 0.40 Gpa 1350 0.18 GPa 0.08 GPa 1450 0.06 GPa 0.19 827