Mechanical Properties of Ceramics and Composites

toughness from 800 to 1200°C (the latter being the limit of their toughness testing). This data thus also shows some strength variations at more modest temperatures before greater decreases at higher temperatures, though on a much more modest scale.

Baldoni et al. [19] showed that flexure strengths of their Al2O3-30 v/o TiC composite decreased gradually from 500 MPa at 22°C to 400 MPa at 1200°C. Their Si3N4-30 v/o TiC composite starting from 650 MPa at 22°C decreased modestly to a minimum at 800°C and then went to a sharp but modest maximum of 700 MPa at 1000°C, followed by a sharp drop to <400 MPa at 1200°C. These changes are in contrast to their toughness both showing minima at 800°C and then increasing substantially to the limit of their testing of 1200°C (Sec. B).

Turning to other all nonoxide composites, Endo et al. [52] showed that strengths of SiC-TiC hot pressed from 0.3–0.4 m particles with B4C+ C sintering aids had lower strengths for SiC rich bodies but higher strengths of TiC rich bodies with similar but more pronounced complexity at 1500 versus 22°C. Thus strengths with 0 TiC at 1500°C were 600 MPa (versus 1000 MPa at 22°C), increased to a maximum of 750 MPa at 20 v/o TiC, and then decreased (crossing the 22°C strengths at 50 v/o TiC at 600 MPa) to a minimum of 550 MPa at 60 v/o TiC, through a modest maximum at 80 v/o TiC and then to 400 MPa (versus 350 MPa at 22°C) at 100% TiC. Lin and Iseki [53] found that strengths of SiC (+ Al-B-C sintering aids) were constant at 510 MPa and then decreased by 30% at 1400°C, while similarly processed SiC with 10 and 30 v/o TiC ( 1.5) started at 430 MPa at 22°C and decreased fairly steadily to 300 MPa at 1400°C, with only limited acceleration of decrease from 1200 to 1400°C. (Such SiC+ TiC bodies made without Al-B-C additions had very low strengths of 100 MPa.) Subsequently they [54] showed similar trends for SiC with 2% AlN additions, but with 10+% lower strengths and similar results with 33 v/o TiC and 2% AlN, but higher strengths for 33 v/o TiC with 2% Ti versus AlN additions, i.e. the latter strengths were nearly identical in value and trend as for their SiC. Thus again temperature dependence can be complex and variable with effects of additives often being an important variable. McMurtry et al. [55] showed strengths of commercial αSiC+ 19 v/o TiB2 (1–5 m) being constant at 490 MPa to 1200°C, i.e. parallel to strengths of αSiC (both sintered with B-C additions) of 350°C. Note that the higher SiC+ TiB2 strengths probably reflect, at least in part, finer, more uniform SiC grain size due to grain growth inhibiting effects of the TiB2 and that there appears to be a limited SiC strength minimum at ≤ 600°C, and then a small increase, while SiC+ TiB2 indicates a possible limited maximum at ≤ 600°C, and then a small decrease in strength.

Turning to platelet composites, Mazdiyasni et al. [56] reported strengths for AlN with 0 to 10–30 w/o fine BN platelet particles generally decreasing with increasing BN content with such decreases being greater in magnitude at lower

versus higher levels of additions. Bodies fabricated without sintering aids (and having 1–6% residual porosity, generally increasing with BN content) all showed lower strengths at 1000 versus 22°C typically by 10–20%, i.e. generally consistent with expectations for decreases in E. Between 1000 and 1500°C strength decreases with no BN were pronounced, e.g. 45%, but they progressively decreased as the BN content increased, with the body with the highest (20 w/o) BN content actually showing a strength increase of 19–20% and somewhat higher overall strength than the 15 w/o BN body, despite greater residual porosity with 20 w/o BN. Bodies made with Y2O3 or CaH2 additives and extended to higher BN additions, with similar residual porosity levels, showed more complex trends with strength maxima at 1000–1250°C (often with a minimum at 1000°C with a 1250°C maximum), but generally with lower strengths, more so at the extremes of 22 and 1500°C, especially the former.

Huang and Nicholson [57] introduced up to 40 v/o Al2O3 platelets with aspect ratios up to 12 into Y-PSZ composites by diefollowed by isopressing or tape casting and lamination, then sinter-HIPing (see also Chap. 8, Sec. V.D). Tape casting gave substantial platelet orientation, which aided densification some and induced some elastic anisotropy and some improvements in strength and toughness, especially at elevated temperatures. Young’s and shear moduli at 22°C increased linearly with the v/o platelets, with little effect of aspect ratio; Poisson’s ratio also decreased linearly, while hardness HV (30 kg) increased somewhat more at the 40 v/o than linearly. IF fracture toughness at 22°C increased from 5 to a broad maximum of 7.9 and then decreased to 6 MPa·m1/2 at respectively 0, 15–30, and 40 v/o platelets, with similar trends with CNB tests, but with values respectively 20, 60, and 50% those from IF testing. Overall strengths at 22°C showed an opposite trend, i.e. overall decreasing by 30% at 5 and 40 v/o platelets. Opposite toughness and strength trends were also shown for effects of platelet aspect ratio, i.e. toughness increased 10%, while strength decreased 20% (and hardness decreased 5%) as the platelet aspect ratio increased from 1 to 12. Better correlation of CNB and strength results were found in elevated tests at 800 and 1300°C, showing higher values for the composites than the Y-PSZ matrix, especially at 800°C, with somewhat greater effects of platelet alignment at these temperatures than at 22°C.

Becher et al. [58] showed that strengths of Al2O3-SiC whisker composites at any given temperature generally increased in proportion to their dependence on whisker loading at 22°C to the limit of their testing (1200°C). All strengths decreased with increasing temperature, e.g. by 15% at 1000°C, consistent with E decreases, then more rapidly, e.g. by another 10–15% at 1200°C. Govila [59] obtained flexure strengths for Al2O3-15 w/o SiC whisker composites consistent with those of Becher et al. and showed very limited decreases to 600–800°C nominally parallel with those for Al2O3 alone and then progressively accelerating in decrease to 1/3 their value at 22°C at 1400°C, i.e greater decrease than for

the Al2O3 alone. The strength results of Yang and Stevens [60] for Al2O3-20 v/o SiC whisker composites were very similar in values and rates of change.

Turning to other oxide matrices, composites of 25 v/o SiC whiskers in a cordierite or an anorthite matrix were shown by Gadkaree [61] to decrease flexure strengths by 10% at 1000°C, i.e. as expected from changes in E, and then to accelerate in their decreases, e.g to 54% and 32% respectively of their E values at 22°C of 400 and 380 MPa respectively. Kumazawa et al. [62] showed that while strength of their mullite matrix was greater at 1300°C than at 22°C ( 560 versus 450 MPa), strengths of the composites were 480 MPa with 10–30 v/o SiC whiskers (i.e. 15% lower) and then decreased to 420 MPa at 40 v/o SiC at 1300°C. (Strengths at 22°C were higher at 480 MPa from 10 to 30 v/o SiC and then decreased to 410 MPa at 40 v/o SiC.) NB toughnesses were higher than those at 22°C, increasing from 2.7 to 3.6 MPa·m1/2 from 0 to 40 v/o SiC whiskers. Resulting strength-to-toughness ratios at 1300°C decreased from 207 to 114 m-1/2 over this range of whisker additions, similar to the trend for such values at 22°C (which were 10–20% higher overall, indicating reasonable flaw sizes of 20–80 microns).

Choi and Salem [63] showed that NB and IF toughnesses of Si3N4+30 v/o SiC whiskers at 22°C and CNB toughness from 22 to 1200°C for both the composite and the matrix alone were all essentially constant at 5.7 MPa·m1/2 with no evidence of R-curve effects. This is consistent with both matrix and composite having identical strengths over the range 22–1400°C, decreasing only 5% at 800°C and then progressively more rapidly to 50% at 1400°C, which was attributed to SCG. Fractography showed fracture initiation commonly to occur from processing defects. Zhu et al. [64] reported that stress rupture of Si3N4+20 v/o SiC whisker composites resulted in mixed interand transgranular fracture and fracture mirrors and mist and hackle boundaries that increased in size as stress decreased, i.e. the same as or similar to normal brittle fracture at lower temperatures. However, fracture at 1200°C was all intergranular with cavity nucleation and growth resulting in a rough fracture region that increased in size as stress decreased, but crack propagation following this creep crack growth was smooth and mirrorlike, indicating catastrophic crack propagation. They reported that SiC whiskers were effective in increasing fracture resistance at 1200°C by crack arrest and crack bridging. Olagnon et al. [65] showed that strengths of their Si3N4-SiC whisker composites were lower at 20 v/o than 10 v/o whiskers ( 700 and 800 MPa respectively at 22 C) but were essentially the same ( 650 MPa) at 800–1300°C. Thus the 10 v/o composite lost strength faster at lower temperatures, e.g. by 25% to 1000°C, but both decreased faster at higher temperatures, e.g. to 200 MPa at 1300°C. They showed typical brittle fracture features of mirror, mist, and hackle surrounding fracture origins of specimens that failed at 1000°C and increasing areas of rough SCG (followed by smoother fracture) as test temperatures increased to 1200 and 1300°C.

Dusza and Sajgalik˘ [66] also measured both toughness and strength of Si3N4 plus 10 or 20 w/o βSi3N4 whisker composites from 22 to 1200°C. While the 20% whisker composite showed IF toughness decreasing only a few percent from the room temperature value of 6.3 MPa·m1/2 at 800, 1000, and 1200°C, they drastically decreased in strength from 510 MPa at 22°C to 300 MPa at 800°C, followed with much less decrease at higher temperatures. This was in contrast to the higher room temperature strength of 690 MPa (but similar toughness) for the 10% whisker composite decreasing to only 640 MPa at 800°C and to 450 MPa at 1200°C, where fracture at 800°C was brittle but occurred from high-tempera- ture SCG at higher temperatures. The lower properties of the 20% whisker composite were attributed to greater inhomogeneity of the whisker distribution.

The issue of possible effects of high-temperature crack healing was addressed by Moffatt et al. [67] by comparing effects of high temperature annealing and testing on a commercial Al2O3-17 v/o SiC whiskers (matrix G 5 m) and an alumina body similar to the matrix (G 8 m). They showed that IF toughness of the Al2O3 was 3.5 MPa·m1/2 at 22 and 800°C and then decreased linearly to 0.4 MPa·m1/2at 1400°C, while the whisker composite had only 1.8 MPa·m1/2 at 22°C, was modestly higher at 800 and 1000°C, and then accelerated, reaching 5.7 MPa·m1/2 at 1400°C. Both the Al2O3 and the whisker composite showed large increases in the apparent toughness of cracked specimens after annealing (with no recracking) and testing at room temperature, e.g. maximum values of 28 MPa·m1/2 for the Al2O3 after annealing at 1400°C and 120 MPa·m1/2 for the composite after annealing at 1200°C. However, recracking of specimens before testing returned them to their normal levels. On the other hand, annealing of cracked Al2O3 specimens and then testing them at elevated temperatures showed no changes. It was concluded that crack healing occurred in both materials but was more pronounced in the whisker composite, whereas annealing gave maximum results at 1200°C, but the effects of annealing were only significant at lower temperature property measurements.

Carroll and Dharani [68] reviewed literature on the strength dependence of ceramic whisker composites, noting that in general there was less improvement in ultimate strengths at higher temperatures versus those at 22°C. They presented a model for high-temperature failure focusing on thermal expansion coefficients, Poisson’s ratios of the constituents along with thermal history, and matrix–whisker friction.

Turning to directionally solidified ceramic eutectics, earlier work to fracture (i.e. area under the stress–strain curves) data of Hulse and Batt [69] showed substantial values for some systems for stressing parallel with the solidification direction (Fig. 11.11). They reported values of 100 J/m2 at 22 and 1500–1575°C for Al2O3-ZrO2(+Y2O3) and values for the CaO-MgO eutectic nearly this high at 22°C and rising over an order of magnitude by 1200°C. Kennard et al. [70,71] showed WOF decreasing some from 20 J/m2 at 22°C at

FIGURE 11.11 Plot of work-to-fracture for Al2O3-ZrO2(+Y2O3) and CaO-MgO eutectic composites of Hulse and Batt [67] and work of fracture (WOF) for MgOMgAl2O4 composite of Kennard et al. [70,71] versus test temperature. Note that use of these values to calculate toughnesses (an uncertain procedure, especially for the work to fracture values) yields 7, 7, and 4 MPa·m1/2 at room temperature. Increases in high temperature values, especially large ones, typically reflect effects of plastic deformation.

1200°C and then increasing to 40 J/m2 at 1600°C for the MgO-MgAl2O4 (Fig. 11.11). Mah et al. [72] reported the toughness of alumina-YAG eutectics of 4.4 MPa·m1/2 at 22°C, decreasing 10% by 1200°C, and then increasing 10% by 1500°C.

Hulse and Batt [69] reported strengths of their Al2O3-ZrO2(+Y2O3) decreasing from room temperature values of 700 MPa only 20% at 1600°C (and lower strengths for slower solidification, hence coarser eutectic structure), a similar trend at lower strength for their CaO·ZrO2-ZrO2 eutectic, and a possible increase in strength for their CaO-MgO eutectic (Fig. 11.12). They also showed that considerable ductility occurred in testing some of these composites at elevated temperature. Kennard et al. [70,71] showed strengths of their MgO-MgAl2O4 eutectic, though modest, also had limited decrease at higher temperature, with somewhat lower, but similar, strength trends for stressing normal to the solidification direction. Mah et al. [72] showed that strengths of their alumina-YAG eutectics decreased from room temperature values of 380 MPa by 25% to 280 MPa at 1500°C. Increases, especially sub-

11.12 Flexure strength for directionally solidified eutectics of Al2O3- ZrO2(+Y2O3), CaO·ZrO2-ZrO2 eutectic, and CaO-MgO of Hulse and Batt [69] versus test temperature. Note (1) tests were with the stress parallel with the solidification direction, except for the case where the stress was normal to this direction (marked perpendicular), and (2) there were lower strengths for slower solidification, hence coarser eutectic structure in the Al2O3-ZrO2(+Y2O3), system.

stantial ones, in higher temperature toughnesses are not necessarily reflected in the above strengths, since such toughness increases are typically due to plastic deformation, which does not necessarily increase strength. More recently, Waku et al. [73] reported that directionally solidified specimens of the alu- mina-YAG eutectic had strengths of 350–400 MPa at both 22 and 1800°C (only 25°C below melting!), with yielding observed at ≥1700°C.

Though the extensive and complex subjects of creep and stress rupture are beyond the scope of this book, a few brief comments are in order, since as strain rates decrease and especially as test temperatures increase, these processes start becoming factors in the temperature dependence of strength. There is often more creep data on some composite ceramics than on strength, e.g. on crystallized glasses, where earlier work by Barry et al. [74] and more extensively by James and Ashbee [75] are of value, but the recent broad survey by Wilkinson [76] is particularly valuable. This shows that rheological models give a reasonable perspective on much creep behavior, especially at lower and higher dispersed phase contents, i.e. below and above the percolation limits for the dispersed phase, with less certainty for the intermediate region transitioning from no to substantial percolation of the dispersed phase. Thus whiskers or needle shaped grains have much lower percolation limits (e.g. 12 v/o) so they can give greater increases in creep resistance, e.g. by of the order of 10 and 60 over respectively platelets and

equiaxed particles. Whisker composites also tend to have less dependence on volume fraction addition, processing, and matrix grain size above percolation limits. Thus the work of Lin et al. [77] is cited in noting that creep in Al2O3-SiC whisker composites at 1200°C varied as G-1 for the matrix versus G-2 to G-3 for Al2O3 and that the G dependence is not present at 1300°C. Desmarres et al. [78] observed that 30 v/o SiC whiskers in a SiAlYON matrix increased strengths by 10% and the temperature for more rapid decrease of strengths from 1000°C for the matrix alone to 1300°C with whiskers. The study of Jou et al. [79] on 50 v/o solid solutions of AlN and SiC also gave microstructural results of interest, namely highest creep rates in inhomogeneous samples (probably due to AlN rich areas), intermediate rates for finer G ( 3.1 m), and lower rates for coarser G ( 5.4 m) from 1400 to 1525°C and 50 to 120 MPa in bending.

F.Hardness, Compressive Strength, and Other Related Behavior

There is even less data for properties covered in this section than in previous sections, but this limited data is useful. Niihara et al. [18] showed that HV (4.9 N) of their composite of Al2O3 with 5 v/o of 2 m SiC particles started at the same value for Al2O3 alone of 20 GPa at 22°C and at their maximum temperature of 1400°C ( 1.5 GPa). However, while HV for the Al2O3 alone decreased very nearly linearly on a semilog plot, their composite was clearly bilinear, decreasing more slowly to 1100°C and then more rapidly to 1400°C, with the separation of the two thus being greatest at 1100°C, i.e. 7 and 3 GPa respectively for the composite and the Al2O3 alone. This trend for less hardness decrease in the composite at lower temperatures and more at higher temperatures versus their Al2O3 is very similar to their flexural strength data, except the strength actually slightly increased to a maximum at 1000°C before beginning its rapid decrease. In contrast to this, Tai and Watanabe [20] reported no clear difference in HV (196 N) between their Al2O3 alone and their composite of (WC+ 10 w/o Co) + 20 w/o Al2O3 ( 0.2 m) over the range tested (22–600°C).

One of the few studies of compressive strengths of ceramic composites versus temperature is one of May and Obi [80] on various crystallized bodies in the SiO2 + 25–35 m/o Li2O system with various heat treatments and amounts (0–3%) of P2O5 nucleating agent. As shown in Fig. 11.13, there is a broad diversity of strength levels and trends with temperature, including both some limited maxima at 100–200°C and minima at 500–600°C.

There have also been some compressive creep studies of ceramic composites, e.g. of crystallized glasses, which are also reviewed by Wilkinson [76], often showing substantial differences from tensile creep, e.g. in stress levels. Also, one study of directionally solidified Al2O3-YAG composites [81] in comparison with tests of similarly oriented Al2O3 and YAG single crystals at 1530°C and lower strain rates (10-4/ sec) showed that the composite creep behavior was close

11.13 Compressive fracture stress for various crystallized SiO2 + 25–35 m/o Li2O glasses as a function of temperature. (From Ref. 80. Published with permission of the British Ceramic Society.)

to or slightly less than that of sapphire, while at higher strain rates (10-5/ sec) the YAG reinforced the sapphire phase.

Verma et al. [82] showed that compressive strengths of their composites of SiC platlets (25–50 m dia., 1–2 m thick) in a borosilicate glass of matching expansion showed continuous, linear increases in strength at 625 and 700°C to 40 v/o SiC (the limits of compositions that could be fully densified). Substantial plastic flow of the composites occurred, since the glass had very low strength at the test temperatures, with both glass and composites thus exhibiting substantial strain rate sensitivity of strength. Thus while the strengths at elevated temperature were much less than at room temperature, increased viscosity due to the SiC platelets had a significant impact.

IV. DISCUSSION

The first of three major trends of this chapter noted for discussion is the variations that can occur in properties, especially strength, at modest temperatures

where testing is often neglected on the frequently incorrect assumption that no significant changes occur at these lower temperatures. Such lower temperature tensile strength behavior, though again given only limited consideration for monolithic ceramics, also revealed important, and sometimes dramatic changes in strength, indicative of underlying mechanisms (Chap. 6). For composites, such lower temperature strength changes are particularly pronounced for crystallized glasses, where substantial variations are seen in toughness and flexure strength (Figs. 11.1, 11.9), as well as compressive strength (Fig. 11.13), the latter being similar to the variety of flexural strength behavior at 22°C for various crystallization schedules in the same compositional system (Chap. 9, Sec. III. A).

Intermediate deviations were observed for toughness and strength of the TZP-Al2O3 composite of Tskuma et al. [6], mainly a minimum in both at 400°C (Fig. 11.10). Limited data for other composites often indicates normal strength decreases consistent with decreases in E as test temperature increases, but this is typically based on the lowest test temperature > 22°C being > 600–1000°C, i.e. near or at the upper limits where such variations primarily manifest themselves. While some observed variations are substantial, as noted above, some are often more limited, e.g. Al2O3-SiC(18), Si3N4-TiC(19), and SiC-TiB2 [55], but these are still potentially significant. For example, all these, especially the latter, must be considered against the typical 10–20% decreases expected from decreases in E, which often increases the net deviation.

The above deviations from the normal decrease of strength or toughness with temperature increase must reflect microstructural mismatch stress changes, commonly reductions; but more must be involved. Thus the initial decreases in strength and toughness in Li2O-SiO2 crystallized glasses with increasing temperature (Fig. 11.1) has been attributed to decreases in microstructural stress and resultant microcracking and the substantial increase in crack blunting by plastic flow. However, there are issues with both of these mechanisms and with alternative or additional mechanisms. Thus while decreasing microcracking is likely to reduce toughness, at least with some measures of toughness, decreases in strength with reduced microcracking may conceivably occur (mainly with very small microcracks, e.g. < a micron, which is possibly consistent with sizes in many crystallized glasses), though it is contrary to most data, raising the question of what the mechanism of the initial strength decrease is. Reductions of microstructural stresses as temperature increases can also be a factor for both polycrystalline composites and especially for single crystal, i.e. directionally solidified, composites, where plastic deformation at higher temperatures can be an important factor, as shown by observed ductility. However, microstructural stress reduction in these, and to a lesser extent in other bodies, can be complicated by differing changes in elastic anisotropy of crystalline phases as temperature increases, since such anisotropy may increase or decrease, or change from one direction to another as a function of temperature (Chap. 7, Fig. 11.14). Further,

while increasing plastic deformation at crack tips can clearly first increase and then significantly decrease strength, e.g. as in glass-based systems, it is more uncertain due to compositional, microstructural, and strain rate sensitivities. Further, the trends for strength and toughness are often opposite as plasticity and especially related SCG increase, with toughness often markedly increasing and strength significantly decreasing.

Three other mechanisms must also be considered. The first and broadest is the general decrease of E with increasing temperature, and resultant decreases in toughness and strength unless overcome or masked by other mechanisms. However, in such cases it still needs to be accounted for, i.e. adding to or subtracting from changes due to other mechanisms. The second mechanism is the often neglected possibility of SCG, e.g. increasing temperatures should progressively accelerate both crack tip reactions of the active species (commonly H2O) and desorption of the active species, which have opposite effects, resulting in possible toughness, and especially strength, minima. A key test for such effects would be inert atmospheres tests , and possibly tests as a function of increasing strain rate, which would increase strengths with SCG, and possibly decrease them with plasticity, effects. Many unknowns thus remain regarding such mechanisms, and more, and especially more comprehensive, data are needed to resolve these issues, e.g. comparison of toughness and strength behavior for other systems, e.g. for canasite systems (Fig. 11.1) and broader temperature and stressing conditions. The third mechanism, of at least partial crack healing, is often limited due to kinetics but may be a factor for less refractory materials, e.g. glasses, or via oxidation of nonoxide constituents.

Consider next the second trend, namely differing trends of different properties with temperature, especially for the same material, but also between materials. Comparison of Young’s modulus, strength, and fracture toughness have been noted above but deserve further attention along with comparison to other properties. Thus there is considerable similarity in the temperature dependences of strength and toughness of the TZP-Al2O3 composites of Tsukuma et al. [6], but also some differences. Similarly the deviations noted in E at 200°C may be related to the deviations to minima of toughness and strength at 400°C (and are clearly related to such effects seen in the E–T dependence of ZrO2 (Fig. 6.18)). Further, while there are similarities in the toughness trends for the two mullite-ZrO2 composites of Fig. 11.2, there are also significant differences, as is also the case where both strength and toughness were measured. The significant opposite trends of strength and toughness from 1000 to 1200°C most likely reflect opposite effects of deformation processes such as grain boundary sliding on toughness versus strength, i.e. enhancing crack propagation in the latter case and impeding it in the former. However, for this and other ZrO2 toughened systems, note that an added, simple monotonic decrease of both toughness and strength over and above that due to decreases in E until the ZrO2 transformation