mObERLYChAN et al: ROLES OF AMORPHOUS GRAIN BOUNDARIES 1633 interfaces promoted intergranular fracture; however, With the large amount of YAG(up to 20% by the more equiaxed shape of the submicron beta volume), these YAG-Sic ceramics could be con- grains did not allow for enhanced toughening due sidered as composites, with the YaG being a to grain bridging. Thus, high toughness in this cer-*weak" phase and an easy path for crack propa- amic required an amorphous grain boundary layer gation. In many SEM images the YAG formed a with a chemistry to weaken the grain boundary and nearly continuous network, and therefore would a grain shape with a high aspect ratio. Lack of any have provided a simple means for all crack propa- f these three ingredients produced an inherently gation to have occurred within the YAG. Thus brittle material when a continuous and relatively thick network of Analyses of the two reported Al2O3-SiC [1-3 YAG existed, the propagation of the crack through have indicated both similarities and disparities as the YAG suggested much of the strength of the ompared to the present ABC-SiC. Since the Al in composite was dependent on this weaker phase his ABC-sic partially reacted with oxides on the Although the fracture toughness was improved, the surface of Sic powders, the alumina-containing room temperature strength of YAG-Sic was amorphous phase at the grain boundaries of ABC- reduced as compared to Hexoloy-SA [7. In con- Sic could have acted similarly to the Al-containing trast, the lower concentration of sintering additives purities on the intergranular fracture surfaces of in the ABC-Sic enabled a much higher fracture Al2O3-SiC [1]. Although Suzuki [1] detected Al on strength to be realized in conjunction with higl tergranular fracture surfaces with Auger, he did toughness. not observe an amorphous phase by HR-TEM in Since the primary utility of these toughened sic did not include Auger analysis of fracture surfaces cations s is for high-temperature structural appli- nor HR-TEM of grain boundaries. The disparity in additives)on high temperature strength must be Suzukis aEs and HR-teM data could be attribu- considered. No high temperature mechanical testing ted to the fact that the Suzukis hr-tem images of these Yag-sic ceramics nor of ABC-Sic have did not depict one of the well-faceted 10001) grain yet been published. The decrease in strength of boundaries. Other research of grain boundaries in YAG at high temperatures [41], should correspond Sic typically did not resolve cross fringes to make to a decrease in strength of Sic containing 10-20% analysis of the boundary unambiguous [1, 23, 31- YAG, especially for an amorphous secondary 33]. ABC-SiC sintered at lowered temperatures phase. YAG-SiC has been proposed to offer good resolved no amorphous phase using HR- ductility at high temperature [7, but with a degra TEM [17, 21]. Yet, the analysis of amorphous layers dation in strength. The thin amorphous phase <I nm thick along these nonplanar boundaries in the ABC-Sic also has been expected to enhance fine-grained polycrystalline materials was recognized creep, as has been reported for Si3N4 [42] as subject to numerous difficulties [21]. Conflicting However, the residual triple junctions were typically lblished reports exhibiting both the presence and crystalline ternary phases [36]. Since Sic has a lack of amorphous grain boundaries could represent higher melting temperature than does Si3 N4,an respective differences between random and special improved creep resistance could be expected as orientation relationships across grain boundaries, compared to Si3 N4 at similar temperatures. Further d or differences in processing conditions. work has been initiated to determine which micro- Whenever AES of intergranular fracture surfaces structures would provide the best compromise of has detected substantial chemical impurities along high temperature strength and room temperature grain boundaries (i.e. much more than can be sol- toughness [43] ible in the crystal grain), the corresponding HR- The improved fracture toughness of YAG-SiC TEM images were believed inconclusive if they did was reported to be due to a corresponding increase not depict an amorphous nor a crystalline integra- in "short crack" formation [7]. This was ular phase. speculated [7] to occur (in part) as a result of stress The SEM analyses reported for the two YAG- arising from a mismatch in coefficient of thermal Sic ceramics(sintered with > 10% YAG) exhibited expansion(CTE) between the Sic and YAG. Such a similar rough crack path around the elongated a CTE mismatches would be a further reason for lim- Sic grains [4-7. The crack paths suggested more iting the volume fraction of a second phase. a tor- the presence of deflection rather than grain pullout. tuous crack path which occurred because it was However, a TEM analysis of the grain boundaries connecting-up numerous, pre-existing short cracks has not been reported for either YAG-Sic would provide a rough fracture surface, but would ceramic 14-7. Nor has it been determined whether also correlate to a lowering of the overall fractur the YAG secondary phase is fully crystalline or strength. amorphous. Since the volume fraction of YAG was In the ABC-Sic no pre-existing cracks were substantial, most Sic grains were coated with a observed after processing. In addition, the tortuous relatively thick layer of YAG, and SEM images crack paths, which were propagated during mechan- showed that the crack path remained in the YAG. ical testing, did not connect weaker secondaryinterfaces promoted intergranular fracture; however, the more equiaxed shape of the submicron beta grains did not allow for enhanced toughening due to grain bridging. Thus, high toughness in this cer￾amic required an amorphous grain boundary layer with a chemistry to weaken the grain boundary and a grain shape with a high aspect ratio. Lack of any of these three ingredients produced an inherently brittle material. Analyses of the two reported Al2O3±SiC [1±3] have indicated both similarities and disparities as compared to the present ABC±SiC. Since the Al in this ABC±SiC partially reacted with oxides on the surface of SiC powders, the alumina-containing amorphous phase at the grain boundaries of ABC± SiC could have acted similarly to the Al-containing impurities on the intergranular fracture surfaces of Al2O3±SiC [1]. Although Suzuki [1] detected Al on intergranular fracture surfaces with Auger, he did not observe an amorphous phase by HR-TEM in the same material. Other reported Al2O3±SiC [2, 3] did not include Auger analysis of fracture surfaces nor HR-TEM of grain boundaries. The disparity in Suzuki's AES and HR-TEM data could be attribu￾ted to the fact that the Suzuki's HR±TEM images did not depict one of the well-faceted {0001} grain boundaries. Other research of grain boundaries in SiC typically did not resolve cross fringes to make analysis of the boundary unambiguous [1, 23, 31± 33]. ABC±SiC sintered at lowered temperatures resolved no amorphous phase using HR￾TEM [17, 21]. Yet, the analysis of amorphous layers <1 nm thick along these nonplanar boundaries in ®ne-grained polycrystalline materials was recognized as subject to numerous diculties [21]. Con¯icting published reports exhibiting both the presence and lack of amorphous grain boundaries could represent respective di€erences between random and special orientation relationships across grain boundaries, and/or di€erences in processing conditions. Whenever AES of intergranular fracture surfaces has detected substantial chemical impurities along grain boundaries (i.e. much more than can be sol￾uble in the crystal grain), the corresponding HR￾TEM images were believed inconclusive if they did not depict an amorphous nor a crystalline intergra￾nular phase. The SEM analyses reported for the two YAG± SiC ceramics (sintered with >10% YAG) exhibited a similar rough crack path around the elongated a SiC grains [4±7]. The crack paths suggested more the presence of de¯ection rather than grain pullout. However, a TEM analysis of the grain boundaries has not been reported for either YAG±SiC ceramic [4±7]. Nor has it been determined whether the YAG secondary phase is fully crystalline or amorphous. Since the volume fraction of YAG was substantial, most SiC grains were coated with a relatively thick layer of YAG, and SEM images showed that the crack path remained in the YAG. With the large amount of YAG (up to 20% by volume), these YAG±SiC ceramics could be con￾sidered as composites, with the YAG being a ``weak'' phase and an easy path for crack propa￾gation. In many SEM images the YAG formed a nearly continuous network, and therefore would have provided a simple means for all crack propa￾gation to have occurred within the YAG. Thus, when a continuous and relatively thick network of YAG existed, the propagation of the crack through the YAG suggested much of the strength of the composite was dependent on this weaker phase. Although the fracture toughness was improved, the room temperature strength of YAG±SiC was reduced as compared to Hexoloy±SA [7]. In con￾trast, the lower concentration of sintering additives in the ABC±SiC enabled a much higher fracture strength to be realized in conjunction with high toughness. Since the primary utility of these toughened SiC ceramics is for high-temperature structural appli￾cations, the e€ects of microstructure (and residual additives) on high temperature strength must be considered. No high temperature mechanical testing of these YAG±SiC ceramics nor of ABC±SiC have yet been published. The decrease in strength of YAG at high temperatures [41], should correspond to a decrease in strength of SiC containing 10±20% YAG, especially for an amorphous secondary phase. YAG±SiC has been proposed to o€er good ductility at high temperature [7], but with a degra￾dation in strength. The thin amorphous phase in the ABC±SiC also has been expected to enhance creep, as has been reported for Si3N4 [42]. However, the residual triple junctions were typically crystalline ternary phases [36]. Since SiC has a higher melting temperature than does Si3N4, an improved creep resistance could be expected as compared to Si3N4 at similar temperatures. Further work has been initiated to determine which micro￾structures would provide the best compromise of high temperature strength and room temperature toughness [43]. The improved fracture toughness of YAG±SiC was reported to be due to a corresponding increase in ``short crack'' formation [7]. This was speculated [7] to occur (in part) as a result of stress arising from a mismatch in coecient of thermal expansion (CTE) between the SiC and YAG. Such CTE mismatches would be a further reason for lim￾iting the volume fraction of a second phase. A tor￾tuous crack path which occurred because it was connecting-up numerous, pre-existing short cracks would provide a rough fracture surface, but would also correlate to a lowering of the overall fracture strength. In the ABC±SiC no pre-existing cracks were observed after processing. In addition, the tortuous crack paths, which were propagated during mechan￾ical testing, did not connect weaker secondary MOBERLYCHAN et al.: ROLES OF AMORPHOUS GRAIN BOUNDARIES 1633