40TH ANNIVERSARY lower than 10-10 s-, which reduced as the temperature the methyl groups present in the PCS and PTC poly increased mers. The presence of oxygen in the precursor fibres Similar mechanical data have been reported for the lab- induced the out-gassing of carbon oxides between 400 oratory produced fibres produced from high-molecular- and 600 C at the beginning of the pyrolysis so reduc- weight PCS, both in tension and in creep[18] ing the final carbon content of the ceramic fibre most The creep data given in this paper, which come from important was the understanding that the presence of the authors'laboratory, have been obtained using ma- the oxygen produced an amorphous phase in which the hines dedicated to the high temperature testing of small small Sic grains and the free carbon aggregates were diameter filaments These machines have been described embedded elsewhere [19] and are capable of conducting tensile, re- The first fibre of this generation to be studied by XRD laxation, creep and, if necessary, fatigue tests on very and TEM was the Nicalon NLM-102 fibre with a 16 wt% small diameter filaments over a wide range of temper- content [17]. This study was followed by others on the atures. Other research teams have conducted creep tests same type of fibre and gave greater information on the using dead weights supported by the fibres. An alternative bonding of the oxygen atoms with neighbouring silicon and rapid way of ranking the creep or relaxation behaviour and carbon atoms [21, 22]. These studies revealed that of ceramic fibres is to put them into the form of a loop the fibre consisted of very small B-SiC grains with an around a former and to heat treat them. If the fibre re- average size, as calculated from the XRD patterns using mains perfectly elastic during the test its form returns to the Sherrer method, of 1.7 nm in size and these were its original, straight form on cooling and removal from embedded in an oxygen rich amorphous silicate phase the former. However if any anelastic behaviour occurs the The B-Sic grains were the only crystalline phase de- fibre retains a curvature. The degree of the induced cur- tected by XRD and electron diffraction. The early grade vature is taken as a mean of ranking the probable creep was later replaced by what was called the ceramic grade behaviour of the fibre compared to that of other fibres Nicalon NLM-202 fibre, the composition of which dif- [20] fered slightly from the earlier grade as it had a lower oxygen content of 12%. Studies on this fibre, which be- came the standard and most widely studied first gener- tion fibre, by TEM, allowed the structure of the free 3.1. Compositions and microstructures carbon to be determined and the mean composition of of first generation fibres the amorphous phase to be proposed [23]. Originally the Fig. 5 shows a typical failure surface of a first genera- oxygen was presumed to be included in an amorphous tion Sic fibre. There is no sign of any granular texture silica phase. Further studies showed that only a fraction so that the fibre appears to be glassy and possibly amor- of the oxygen was in the form of SiO2 and the rest was phous, as was originally concluded. However, wide angle present in the fibre in the form of a ternary phase SiO, Cy X-ray diffraction(XRD) studies and later examination by [21]. The fibre therefore seen to be composed of a transmission electron microscopy (TEM) showed that the continuum of SiO, C, tetrahedra with x +y=4.The first generation fibres were nano-crystalline with sizes of Nicalon 200 grade fibre with a 1l wt. oxygen content B-Sic grains generally being in the range 1.7 to 2 nm was determined as being composed of SiC4 tetrahedra although some variation occurred in early fibres due, pre- formed into small crystallites of 1.4 nm with a diamond sumably, to optimisation by the manufacturers in the du- like structure separated by two Sio, Cy tetrahedra (x+ ration of pyrolysis [17]. O), whereas the 100 grade fibres with 16 wt. of oxygen The compositions by weight percentages and densi- had a lower crystallinity. The Nicalon 200 grade fibre was s of the fibres are given in Table Il, together with the composed of 55 wt% B-SiC grains, 40% of an intergran carbon-to-silicon atomic weight ratios. It is clear that the ular phase with a mean composition of SiO,, 15 Co85 and first generation fibres were far from being stoichiomet- 5% of randomly oriented free carbon aggregates, I nm ric silicon carbide and that they contained an excess of in size. Coo2 lattice fringe images showed small stacks oxygen and carbon. This observation was the beginning of two fringes around 0.7 nm in size suggesting that the of an understanding of the easons why the fibres had basic structural unit(BSU) was a face-to-face associa- such different characteristics from those of bulk Sic. Al- tion of aromatic rings, called dicoronenes, in which the though the manufacture of these fibres from PCS and hydrogen-to-carbon atomic ratio is 0.5. With such a model PTC was analogous to that of carbon fibres made from for the microstructure of the fibre, a porosity level of PAN, an important difference quickly became evident, more than 2% was calculated [23]. Other authors pro which was that the oxygen, used to crosslink and re posed that the intergranular phase should be written as der infusible the precursors before pyrolysis, introduced Sio, Cl-xp which suggests that the composition could into the PAN fibres, is entirely removed at high temper- vary continuously from SiC to SiOz as the oxygen concen- atures but that used to stabilise the pcs and ptc fibres ation varied [24]. This gave a composition by weight of remains. The excess carbon in the SiC fibres came from 56% SiC, 10%C and 34% SiO.1 Co44
40TH ANNIVERSARY lower than 10−10 s−1, which reduced as the temperature increased. Similar mechanical data have been reported for the laboratory produced fibres produced from high-molecularweight PCS, both in tension and in creep [18]. The creep data given in this paper, which come from the authors’ laboratory, have been obtained using machines dedicated to the high temperature testing of small diameter filaments. These machines have been described elsewhere [19] and are capable of conducting tensile, relaxation, creep and, if necessary, fatigue tests on very small diameter filaments over a wide range of temperatures. Other research teams have conducted creep tests using dead weights supported by the fibres. An alternative and rapid way of ranking the creep or relaxation behaviour of ceramic fibres is to put them into the form of a loop around a former and to heat treat them. If the fibre remains perfectly elastic during the test its form returns to its original, straight form on cooling and removal from the former. However if any anelastic behaviour occurs the fibre retains a curvature. The degree of the induced curvature is taken as a mean of ranking the probable creep behaviour of the fibre compared to that of other fibres [20]. 3.1. Compositions and microstructures of first generation fibres Fig. 5 shows a typical failure surface of a first generation SiC fibre. There is no sign of any granular texture so that the fibre appears to be glassy and possibly amorphous, as was originally concluded. However, wide angle X-ray diffraction (XRD) studies and later examination by transmission electron microscopy (TEM) showed that the first generation fibres were nano-crystalline with sizes of β-SiC grains generally being in the range 1.7 to 2 nm although some variation occurred in early fibres due, presumably, to optimisation by the manufacturers in the duration of pyrolysis [17]. The compositions by weight percentages and densities of the fibres are given in Table II, together with the carbon-to-silicon atomic weight ratios. It is clear that the first generation fibres were far from being stoichiometric silicon carbide and that they contained an excess of oxygen and carbon. This observation was the beginning of an understanding of the reasons why the fibres had such different characteristics from those of bulk SiC. Although the manufacture of these fibres from PCS and PTC was analogous to that of carbon fibres made from PAN, an important difference quickly became evident, which was that the oxygen, used to crosslink and render infusible the precursors before pyrolysis, introduced into the PAN fibres, is entirely removed at high temperatures but that used to stabilise the PCS and PTC fibres remains. The excess carbon in the SiC fibres came from the methyl groups present in the PCS and PTC polymers. The presence of oxygen in the precursor fibres induced the out-gassing of carbon oxides between 400 and 600◦C at the beginning of the pyrolysis so reducing the final carbon content of the ceramic fibre. Most important was the understanding that the presence of the oxygen produced an amorphous phase in which the small SiC grains and the free carbon aggregates were embedded. The first fibre of this generation to be studied by XRD and TEM was the Nicalon NLM-102 fibre with a 16 wt.% content [17]. This study was followed by others on the same type of fibre and gave greater information on the bonding of the oxygen atoms with neighbouring silicon and carbon atoms [21, 22]. These studies revealed that the fibre consisted of very small β-SiC grains with an average size, as calculated from the XRD patterns using the Sherrer method, of 1.7 nm in size and these were embedded in an oxygen rich amorphous silicate phase. The β-SiC grains were the only crystalline phase detected by XRD and electron diffraction. The early grade was later replaced by what was called the ceramic grade Nicalon NLM-202 fibre, the composition of which differed slightly from the earlier grade as it had a lower oxygen content of 12%. Studies on this fibre, which became the standard and most widely studied first generation fibre, by TEM, allowed the structure of the free carbon to be determined and the mean composition of the amorphous phase to be proposed [23]. Originally the oxygen was presumed to be included in an amorphous silica phase. Further studies showed that only a fraction of the oxygen was in the form of SiO2 and the rest was present in the fibre in the form of a ternary phase SiOxCy [21]. The fibre was therefore seen to be composed of a continuum of SiOxCy tetrahedra with x + y = 4. The Nicalon 200 grade fibre with a 11 wt.% oxygen content was determined as being composed of SiC4 tetrahedra formed into small crystallites of 1.4 nm with a diamondlike structure separated by two SiOxCy tetrahedra (x = 0), whereas the 100 grade fibres with 16 wt.% of oxygen had a lower crystallinity. The Nicalon 200 grade fibre was composed of 55 wt.% β-SiC grains, 40% of an intergranular phase with a mean composition of SiO1.15C0.85 and 5% of randomly oriented free carbon aggregates, 1 nm in size. C002 lattice fringe images showed small stacks of two fringes around 0.7 nm in size suggesting that the basic structural unit (BSU) was a face-to-face association of aromatic rings, called dicoronenes, in which the hydrogen-to-carbon atomic ratio is 0.5. With such a model for the microstructure of the fibre, a porosity level of more than 2% was calculated [23]. Other authors proposed that the intergranular phase should be written as SiOxC1−x/2 which suggests that the composition could vary continuously from SiC to SiO2 as the oxygen concentration varied [24]. This gave a composition by weight of 56% SiC, 10% C and 34% SiO1.12C0.44. 828
40TH ANNIVERSARY The existence of porosity in the first generation ceramic If improved fibres were to be made, which could go fibres was proposed so as to account for the outgassing some way to meeting the promise of a pure SiC, it was of the hundreds of volumes of gas per volume of fibre, clear that a more stoichiometric composition had to be which occurs during the transformation from an organic achieved. As mentioned above. it had been seen that the to a mineral structure, which occurs during the manu- presence of oxygen in the fibre resulted in a poorly or- facture of the fibre [25]. It was reasoned that nanometric ganised phase in which the Sic grains were embedded channels must exist in the fibres during the transformation It was clear that the oxygen content had to be consider stage as diffusion through a solid phase would be too slow. ably reduced if improvements were to be made. As the These nanochannels collat further heating above the primary reason that the oxygen was in the fibres was that temperature at which the gas is evolved. A porosity of it was introduced to render the precursor fibres infusible 6.2% was shown to exist experimentally by X-ray scatter- it was clear that other means of cross-linking the polymer ing measurements on Nicalon NLM-202 fibres pyrolized precursors had to be investigated at1400°C[26,27 The result of the studies on the first generation fibres as the conclusion that it was the non-stoichiometric com- 4. Second generation small diameter SiC fibres position of the fibres which were limiting their physical The route adopted to reduce the oxygen content in the characteristics. The presence of the amorphous intergran- second generation of small diameter SiC fibres was to in the fibres after pyrolysis making a Si-o-C phase. The render the PCS precursor fibre infusible. This was possible low fraction of a granular Sic phase accounted for the by using different types of irradiation which could interact Youngs modulus of the fibres being only half that of with the precursor polymers to produce free radicals and bulk SiC. The amorphous phase also explained why the gaseous products by the scission of the chemical bonds fibres began to lose strength and creep at temperatures of Si-CH3, Si-H and C-H. This allowed Si-Si and Si-C around 1000 or 1100oc whereas bulk sic would be bonds to be formed. a number of different types of radi pected to resist to higher temperatures [28]. Heating to ation were investigated by both fibre producers working above 1500.C induced rapid grain growth and outgassing in collaboration with the Japanese Atomic Energy Re- of Sio and Co which came from the oxygen in the amor- search Institute. Gamma irradiation was investigated but phous Si-o-C phase and also the free carbon in the fibre, finally electron radiation in a helium atmosphere was used as described by the following relationship to make the second generation fibres. The cross-linking step was followed by heat treatment at 327 C for a short time to eliminate the remaining free radicals which SiC,O,- SiC()+ Sio(g)+Co(g) apped in the irradiated precursor fibre [35]. An example of how the hydrogen atom, bonded to the silicon atom, as It had been noticed in early studies that the rate of shown in Fig. 2, is removed by electron bombardment, so strength loss of first generation fibres was lower in as to allow direct bonding between the two silicon atoms idizing atmospheres than in an inert argon atmosphere in neighbouring molecules, is shown in Fig. 6[36] [17, 29, 30]. It was concluded that heating the fibres in air Below 550 C cross-linking between main chains dom- produced a silica coating which hindered outgassing of inates and is induced by the dehydrogenation condensa the products of the decomposition of the Si-o-C phase; tion of the Si-H groups. From 550oC to 800C, the side this was further studied by heating fibres made from both chains on the cross-linked polymer begin to decompose the pCS and PtC precursors over a wide range of oxygen and CH4 and H2 are given off producing an inorganic fi partial pressures and in carbon dioxide [31, 32]. These bre. Above 800C and up to 1000C, hydrogen is given authors also showed that the formation of a barrier coat- off, most probably associated with the decomposition ing of Sic laid down by chemical vapour deposition on the fibre surface suppressed gas evolution and slowed the degradation process. They also showed that, what had been presumed in earlier studies was actually the case and that a solid layer of Sio2 formed on the surface of the Nicalon fibres and a Sioz layer with a small amount of TiO2 was formed on the surfaces of the Tyranno fibres It should however be noted that increasing the pressure an argon environment also hindered the onset of decomposition. When the first generation fibres wer heated in Argon at a pressure of 138 MPa the onset of Figure 6 Direct cross-linking of the PCS precursor polymer by irradiation fibre weight loss increased from 1200 to 1500C [34]
40TH ANNIVERSARY The existence of porosity in the first generation ceramic fibres was proposed so as to account for the outgassing of the hundreds of volumes of gas per volume of fibre, which occurs during the transformation from an organic to a mineral structure, which occurs during the manufacture of the fibre [25]. It was reasoned that nanometric channels must exist in the fibres during the transformation stage as diffusion through a solid phase would be too slow. These nanochannels collapse on further heating above the temperature at which the gas is evolved. A porosity of 6.2% was shown to exist experimentally by X-ray scattering measurements on Nicalon NLM-202 fibres pyrolized at 1400◦C [26, 27]. The result of the studies on the first generation fibres was the conclusion that it was the non-stoichiometric composition of the fibres which were limiting their physical characteristics. The presence of the amorphous intergranular phase was clearly due to the oxygen which remained in the fibres after pyrolysis making a Si–O–C phase. The low fraction of a granular SiC phase accounted for the Young’s modulus of the fibres being only half that of bulk SiC. The amorphous phase also explained why the fibres began to lose strength and creep at temperatures around 1000 or 1100◦C whereas bulk SiC would be expected to resist to higher temperatures [28]. Heating to above 1500◦C induced rapid grain growth and outgassing of SiO and CO which came from the oxygen in the amorphous Si–O–C phase and also the free carbon in the fibre, as described by the following relationship. SiCxOy → SiC (s) + SiO(g) + CO(g) It had been noticed in early studies that the rate of strength loss of first generation fibres was lower in oxidizing atmospheres than in an inert argon atmosphere [17, 29, 30]. It was concluded that heating the fibres in air produced a silica coating which hindered outgassing of the products of the decomposition of the Si–O–C phase; this was further studied by heating fibres made from both the PCS and PTC precursors over a wide range of oxygen partial pressures and in carbon dioxide [31, 32]. These authors also showed that the formation of a barrier coating of SiC laid down by chemical vapour deposition on the fibre surface suppressed gas evolution and slowed the degradation process. They also showed that, what had been presumed in earlier studies was actually the case and that a solid layer of SiO2 formed on the surface of the Nicalon fibres and a SiO2 layer with a small amount of TiO2 was formed on the surfaces of the Tyranno fibres [33]. It should however be noted that increasing the pressure of an argon environment also hindered the onset of decomposition. When the first generation fibres were heated in Argon at a pressure of 138 MPa the onset of fibre weight loss increased from 1200 to 1500◦C [34]. If improved fibres were to be made, which could go some way to meeting the promise of a pure SiC, it was clear that a more stoichiometric composition had to be achieved. As mentioned above, it had been seen that the presence of oxygen in the fibre resulted in a poorly organised phase in which the SiC grains were embedded. It was clear that the oxygen content had to be considerably reduced if improvements were to be made. As the primary reason that the oxygen was in the fibres was that it was introduced to render the precursor fibres infusible, it was clear that other means of cross-linking the polymer precursors had to be investigated. 4. Second generation small diameter SiC fibres The route adopted to reduce the oxygen content in the second generation of small diameter SiC fibres was to eliminate the oxygen induced cross-linking step used to render the PCS precursor fibre infusible. This was possible by using different types of irradiation which could interact with the precursor polymers to produce free radicals and gaseous products by the scission of the chemical bonds of Si–CH3, Si–H and C–H. This allowed Si–Si and Si–C bonds to be formed. A number of different types of radiation were investigated by both fibre producers working in collaboration with the Japanese Atomic Energy Research Institute. Gamma irradiation was investigated but finally electron radiation in a helium atmosphere was used to make the second generation fibres. The cross-linking step was followed by heat treatment at 327◦C for a short time to eliminate the remaining free radicals which were trapped in the irradiated precursor fibre [35]. An example of how the hydrogen atom, bonded to the silicon atom, as shown in Fig. 2, is removed by electron bombardment, so as to allow direct bonding between the two silicon atoms in neighbouring molecules, is shown in Fig. 6 [36]. Below 550◦C cross-linking between main chains dominates and is induced by the dehydrogenation condensation of the Si–H groups. From 550◦C to 800◦C, the side chains on the cross-linked polymer begin to decompose and CH4 and H2 are given off producing an inorganic fi- bre. Above 800◦C and up to 1000◦C, hydrogen is given off, most probably associated with the decomposition of Figure 6 Direct cross-linking of the PCS precursor polymer by irradiation curing. 829