5050 J Mater Sci(2007)42:5046-5056 150.0 23 produced a laboratory fiber (UF fiber)with the ◆·HNL similar composition as HNL fiber. There was no loss strength with heat treatments up to 1, 700C and then the strength decreased rapidly with further heat treat 0 ments up to 1,900C. He believed that strength is controlled by the residual tensile stresses, which expansion coefficient between SiC and C. This situa tion should be true. The coefficient of thermal expan sion of carbon/graphite(2.0-3.0 x 10/K)is less than SiC(3.9-4.0 x 10-/K). When fibers were cooled from high annealing temperature to room temperature, the nitial12001400160018002000 Sic grains want to contract, while carbon grain w Annealing temperature, C resist their contraction. This action-reaction will put Fig. 3 Tensile strength and its relation to the crystallite size for SiC in tension and carbon in compression. This residual SiC fibers annealed at elevated temperatures in Ar for 1 h tension stresses could have a contribution to the total stress loss. This case can be applied to each fiber type those temperature conditions above which perfor- which contains excess carbon, but here it should be mance degradation of the fibers occurred. The depen- more significant in HNL fiber because of high excess dence of strength on temperature in present study is in carbon(C/Si= 1.38). In both HNL fiber and TySA agreement with those of previous studies [5, 17, 18, 22]. fiber, the size of carbon grain increased with increasing Ichikawa et al. [18] reported that HNLS was quite annealing temperature [9, 10, 17]. The growth of the stable chemically after 1 h exposure in an argon gas at carbon corresponds to a decrease of localized spin 1, 800C, since no structural decomposition occurred centres [24. The growth of the carbon grain might and it exhibited a good strength of 1.9 GPa. The result in an increase of residual stress however, this crystallite size is about 35 nm. TEM observation shows evidence is insufficient because the magnitude of that this annealed HNLS fiber has a Sic grain size of residual stresses is strongly dependent on the volume approximately 200 nm, which is about 10 times larger fraction of carbon phase in a bulk material than that of the as-received fiber As for TysA fiber, this is a sintering fiber, which is Surface morphologies prepared by the reaction of a polyc h aluminiumacetylacetonate, and subsequently con- Figure 4 showed SEM morphologies of the fibers after verted into the Tyranno SA fiber, by decomposition with annealing at elevated temperatures in Ar for 1 h an evolution of CO and Sio (1,500C<T< 1, 700C) The HNL fibers annealed at temperatures below and sintering(about 1, 800C). TySA fiber retained 1, 400C had a smooth surface which is almost no most of its initial strength, because no significant grain difference from that of as received fibers( Fig. 4a) coarsening was observed even annealed at 1,900C. Annealing at 1,400 C caused slight coarsening of fiber Excellent strength retention has been observed in a surface( Fig 4b). Obvious changes in appearance were former work 5 observed for the fibers annealed at 1.780oC. These On the other hand, the HNL fiber has smaller crystal fibers showed a porous microstructure and large grain size comparing to that of HNLS fiber, but it showed deposition on the fiber surface(Fig. 4c). Such huge more rapid strength degradation than HNLS fiber crystals are not observed within bulk of the fiber, due above 1,400C annealing as shown in Fig 3; both to the presence of free carbon, which inhibit the grain HNLS and TysA fiber have near-stoichiometric com- boundary or/and gaseous diffusion position and high-crystallite structure, but they showed For the hnls fibers annealed below 1. 600 oC. their different strength retention. This observed phenome- microstructure did not vary compared to the as- non implied that other mechanisms must be responsi- received fibers(Fig. 4d). After annealing at 1, 600C, ble for strength degradation of Sic fibers besides the although the individual SiC grain grown on the fiber coarsening of crystallite size urface. but fiber surface still remained smooth and it One source for strength degradation is residual appeared no structure degradation(Fig. 4e). The fiber stresses, which were generated from phase transfor- annealed at 1, 780 C exhibited a rough surface with mation and the mismatch in the coefficient of thermal deposition of bulk SiC grains, but it still remained expansion between excess car bon and SiC grain Sacks relatively dense structure(Fig. 4f) 2 Springerthose temperature conditions above which perfor￾mance degradation of the fibers occurred. The depen￾dence of strength on temperature in present study is in agreement with those of previous studies [5, 17, 18, 22]. Ichikawa et al. [18] reported that HNLS was quite stable chemically after 1 h exposure in an argon gas at 1,800 C, since no structural decomposition occurred and it exhibited a good strength of 1.9 GPa. The crystallite size is about 35 nm. TEM observation shows that this annealed HNLS fiber has a SiC grain size of approximately 200 nm, which is about 10 times larger than that of the as-received fiber. As for TySA fiber, this is a sintering fiber, which is prepared by the reaction of a polycarbosilane (PCS) with aluminiumacetylacetonate, and subsequently con￾verted into the Tyranno SA fiber, by decomposition with an evolution of CO and SiO (1,500 C <T < 1,700 C) and sintering (about 1,800 C). TySA fiber retained most of its initial strength, because no significant grain coarsening was observed even annealed at 1,900 C. Excellent strength retention has been observed in a former work [5]. On the other hand, the HNL fiber has smaller crystal size comparing to that of HNLS fiber, but it showed more rapid strength degradation than HNLS fiber above 1,400 C annealing as shown in Fig. 3; both HNLS and TySA fiber have near-stoichiometric com￾position and high-crystallite structure, but they showed different strength retention. This observed phenome￾non implied that other mechanisms must be responsi￾ble for strength degradation of SiC fibers besides the coarsening of crystallite size. One source for strength degradation is residual stresses, which were generated from phase transfor￾mation and the mismatch in the coefficient of thermal expansion between excess carbon and SiC grain. Sacks [23] produced a laboratory fiber (UF fiber) with the similar composition as HNL fiber. There was no loss in strength with heat treatments up to 1,700 C and then the strength decreased rapidly with further heat treat￾ments up to 1,900 C. He believed that strength is controlled by the residual tensile stresses, which developed as a result of the mismatch in thermal expansion coefficient between SiC and C. This situa￾tion should be true. The coefficient of thermal expan￾sion of carbon/graphite (2.0–3.0 · 10–6/K) is less than SiC (3.9–4.0 · 10–6/K). When fibers were cooled from high annealing temperature to room temperature, the SiC grains want to contract, while carbon grain will resist their contraction. This action-reaction will put SiC in tension and carbon in compression. This residual tension stresses could have a contribution to the total stress loss. This case can be applied to each fiber type, which contains excess carbon, but here it should be more significant in HNL fiber because of high excess carbon (C/Si = 1.38). In both HNL fiber and TySA fiber, the size of carbon grain increased with increasing annealing temperature [9, 10, 17]. The growth of the carbon corresponds to a decrease of localized spin centres [24]. The growth of the carbon grain might result in an increase of residual stress however, this evidence is insufficient because the magnitude of residual stresses is strongly dependent on the volume fraction of carbon phase in a bulk material. Surface morphologies Figure 4 showed SEM morphologies of the fibers after annealing at elevated temperatures in Ar for 1 h. The HNL fibers annealed at temperatures below 1,400 C had a smooth surface, which is almost no difference from that of as received fibers (Fig. 4a). Annealing at 1,400 C caused slight coarsening of fiber surface (Fig. 4b). Obvious changes in appearance were observed for the fibers annealed at 1,780 C. These fibers showed a porous microstructure and large grains deposition on the fiber surface (Fig. 4c). Such huge crystals are not observed within bulk of the fiber, due to the presence of free carbon, which inhibit the grain boundary or/and gaseous diffusion. For the HNLS fibers annealed below 1,600 C, their microstructure did not vary compared to the as￾received fibers (Fig. 4d). After annealing at 1,600 C, although the individual SiC grain grown on the fiber surface, but fiber surface still remained smooth and it appeared no structure degradation (Fig. 4e). The fiber annealed at 1,780 C exhibited a rough surface with deposition of bulk SiC grains, but it still remained a relatively dense structure (Fig. 4f). 0.0 1.0 2.0 3.0 4.0 1200 1400 1600 1800 2000 0.0 10.0 20.0 30.0 40.0 50.0 Crystallite size,nm Initial HNL HNLS TySA Annealing temperature, οC Tensile strength, /GPa Fig. 3 Tensile strength and its relation to the crystallite size for SiC fibers annealed at elevated temperatures in Ar for 1 h 5050 J Mater Sci (2007) 42:5046–5056 123