Mechanical Properties of Ceramics and Composites

FIGURE 6.2 Fracture toughness versus grain size (G) for ≥ 95% alumina bodies at 22, 1000, and 1300°C, where G values were available from Xu et al. [36], Grimes et al. [37], de With [31], and Dalgleish et al. [32]. (Note that the latter data was corrected for P = 0.01, 0.06, and 0.09 respectively for G = 2, 4, and 24 m using e–4P, but trends would still be similar without such corrections.) Despite differences in measurement techniques (indicated for toughness, G determinations not always specified) and some differences in fabrication, purity, and residual porosity, the data consistently indicates a decrease in toughness as G increases and an overall decrease in both toughness and its G dependence as T increases.

3 MPa·m1/2 at 22°C to a minimum of 2.3 MPa·m1/2 at 800°C and then increased to a maximum of 2.8 MPa·m1/2 at 1200°C. Strength behavior was opposite; starting at 175 MPa at 22°C, it increased to a maximum of 200 MPa at 800°C and then decreased to 160 MPa at the limit of testing of 1300°C. They also showed that toughness at 1200°C decreased 20% (along with increasing transgranular fracture) with increasing strain rate, similar to the 30% decrease for single crystals [38], but the former may reflect more reduction in bridging than in slip effects, which are the expected source of single crystal decreases. Baudin and Pena [44] subsequently showed that the 1200°C toughness of a stoichiometric spinel and two alumina rich ones, all having similar microstruc-

FIGURE 6.3 Comparison of the temperature dependences of fracture toughnesses of polycrystalline MgAl2O4 hot pressed with LiF additions, G 100 m, P0 [40], MgAl2O4 sintered without additives, G = 1.5 ± 0.8 m, P 0.02 [39] and stoichiometric crystals of three orientations (dashed lines), along with Young’s modulus (E) data for a dense polycrystalline body. Note the (1) limited toughness differences for the two polycrystalline bodies and the highest crystal values measured in different tests to several hundred degrees, (2) more rapid decrease of the body made with LiF as the LiF melting point is approached, then little or no decrease at higher temperatures in contrast to the lack of such a drop for the body made without LiF, but then a greater decrease for the latter at higher temperatures, and (3) similar initial rates of decrease for the three crystal orientations, reaching a minimum and then increasing substantially (but presumably decreasing again at higher temperatures).

tures, exhibited respectively intergranular fracture with bridging and mostly transgranular fracture.

Inoue and Matzke [45] measured toughness of sintered ThO2 (G 20 m with much of the 8% porosity being intragranular and a micron or more in dia.) using Hertzian cracks from spherical indenters of Al2O3 or steel from 22 to 388°C. These indenters gave toughness values of respectively 1.22 and 1.07 at 22°C decreasing to 0.77 MPa·m1/2 for both at 388°C, i.e. severalfold times the

planes. While such room temperature values are high in comparison to values from fractography by 60% and 100% respectively [51] (Chap. 2, Sec. III.D), the temperature dependences are of prime interest here. Guillou et al. [52] also

measured (Berkovich) indentation toughness of 6H SiC crystals for indents on

< - >

(0001) planes with cracks in 1010 directions showing 40% decrease by 600°C and then an bottoming out at 1.6 MPa·m1/2, as with Naylor and Page (Fig. 6.5). Thus there is agreement and there are similarities and differences in the SiC data.

Evans and Lange [53] showed that the DT toughness of commercial SiC hot pressed with Al2O3 addition was constant at 4 MPa·m1/2 to 1100°C and then decreased substantially, i.e. to 2.8 MPa·m1/2 at 1400°C. Henshall et al. [54] subsequently evaluated the NB toughness (and delayed failure) of the same SiC (G 1.5 m), giving toughness of 6.1 MPa·m1/2 to 1000°C and then decreasing to 4 MPa·m1/2 by 1773 K. While their absolute values are about 50% higher, the relative change in toughness with temperature were very similar. These changes correspond to a change from mixed interand transgranular fracture mode at 300 K to all intergranular fracture at > 1373 K, and the onset of de-

FIGURE 6.5 Fracture toughness of SiC as a function of test temperature. (A) For crystals: NB data of Henshall et al. [49] and indent data of Guillou et al. [52] for different (shown) crystal orientations. (B) Indent data of Naylor and Page [50] for hot pressed SiC. Note the (1) broad range of values at 22°C, which are somewhat to substantially higher in comparison to single crystal values from fractography and from comparison to polycrystalline values [51], (2) differing temperature trends in (A), limited in crystal values at 22°C, but they are nearly as high as most polycrystalline values, but with opposite temperature trends, and (3) the unusually high polycrystalline values in (B).

layed failure at ≥ 1273 K. Naylor and Page’s [50] indent tests of hot pressed SiC from the same source (and Al2O3 additions), while much higher in values, show a similar temperature trend.

In contrast to the above SiC hot pressed with Al additions, SiC densified with B, B+C, or B4C additions (i.e. commonly commercially sintered α=SiC) shows much less or no change of toughness (and often strength) to temperatures of 1500°C. Thus, Evans and Lange [53] showed that DT toughness of commercial sintered α-SiC increased slightly [ 10% to the limit of testing (1500°C) in contrast to SiC hot pressed with Al2O3 additions, as was noted above]. Similarly, Ghosh et al. [55] showed toughness measured by three techniques on the same commercial sintered α=SiC all being constant between 3 and 4 MPa·m1/2 over the 20–1400°C range. Srinivasan and Seshadri [56] also showed (NB) toughness constant at 4.8 MPa·m1/2 over this range when tested unoxidized in an inert atmosphere, but with varying increases in toughness above 600 –1000°C depending on testing in air or with prior oxidation of the samples (the latter giving greater increases). Popp and Pabst [57] corroborated that the toughness of commercial sintered α–SiC was nearly constant to the limits of their testing of 1200°C (actually showing a few percent decrease from the value of 3.5 MPa·m1/2 at 22°C) with negligible difference as a function of strain rates in an air atmosphere. This limited change was noted as correlating with the predominant transgranular fracture mode reported to at least 1400°C. However, they showed that while a reaction-processed SiC with 13% residual Si had a similarly constant toughness at 4 MPa·m1/2 over the same temperature range in air tested at high strain rates (1260 mm/min), it markedly increased by 200% from

900–1200°C at low strain rates (0.024 mm/min). These and the above results are generally consistent with those for high-temperature slow crack growth, which is discussed in the next section.

Kriegesmann et al. [58] corroborated that sintered and hot pressed SiC having mainly transgranular fracture at elevated temperatures retains toughness (and strengths) there, i.e. showed no decrease from values at 22°C to the limit of testing of 1400°C. On the other hand, such bodies having mainly intergranular fracture at elevated temperatures, while showing toughnesses (and strengths) essentially the same as at 22°C at 800°C, showed decreases of 20–40% at 1400°C. The bodies showing decreases included hot pressed material having the highest strengths at 22 °C (by 12–16% versus the other two hot pressed materials and by

70% versus the sintered SiC) and included bodies made from either α= or β= SiC powder. Bodies derived from either α or β powder showed that the SiC crystal phase was not the determinant of the retention or loss of toughness or strength. While the specific densification aids used were not disclosed, they did note that the use of B or B-based aids corresponded with transgranular fracture and toughness (and strength) retention at elevated temperatures versus alu- minum-based aids correlated with increased intergranular fracture and toughness

(and strength) decreases at elevated temperatures. These authors briefly discussed and showed micrographs of the microstructures of the bodies studied, showing that all four bodies had G values averaging 4 m, but with different distributions of sizes, shapes, or both. Thus of the two hot pressed bodies, both from α powder and with similar nominally equiaxed toughness (and strengths), the one with larger isolated, equiaxed grains (e.g. to 5+ m) had somewhat lower toughness ( 13%) but higher strength ( 4%) at 22°C and no loss of toughness (or strength) at elevated temperature. However, the body hot pressed from β powder that had tabular grains ( 2 m dia. and 5–10 m long) had the greatest strength at 22°C but 20% loss at 1400°C (toughness not measured).

B4C hot pressed without additives (P 0.08, G 5 m) has been reported to have toughness decreases of 30% from 22 to 1200°C by Hollenberg and Walther [59]. While their toughness values appear low by ≥ 50%, the decrease with temperature is similar to, but possibly somewhat greater than, the decrease in E.

Next consider results for dense Si3N4. Naylor and Page [50], using an indentation technique, reported toughness of CVD material (i.e. made without additives, but quite possibly having some columnar or oriented grain structure or both) decreasing by nearly 50%, e.g. from 5.6 MPa·m1/2 at 22°C to 2.9 MPa·m1/2 at 800°C, i.e. greater than expected for E. Data for dense sintered or hot pressed Si3N4 and sialons also generally showed lower toughness at 1000–1100 than at 22°C, usually consistent with decreases in E, i.e. 10–20% [60,61]. However, tests continued to higher temperatures typically show a subsequent increase to toughness maxima at 1200–1400°C [63], with the temperature and extent of such maxima dependent on the type of toughness test, the strain rate of the test, and the nature of the material. Ohji et al. [63], using Si3N4 hot pressed with 3 and 5 wt% respectively of Al2O3 and Y2O3, showed the common tendency for toughness to decrease little or not at all to > 1000°C and then to show a modest maximum at 1200°C (while tensile strength had a small maximum at 1000°C); it had a substantial loading rate dependence, e.g. a maximum at a displacement rate of 10–3 mm/min at 1260°C.

Finally, consider the temperature dependence of toughness of other single crystals than Al2O3, MgAl2O4, and SiC presented earlier. NB data of Ingel et al. [64] for Y-PSZ and Y-CSZ showed the former decreasing to a minimum at 1000°C while the CSZ crystals probably reached a minimum at 600°C, both subsequently increasing substantially with further temperature increases beyond the minima (Fig. 6.6). The latter increase was attributed to plastic deformation, e.g. as shown by macroscopic deformation, especially in the CSZ, and higher toughness values at higher strain rates at high temperatures. This is in contrast to a similar but continuing decrease of the toughness of a commercial Mg-PSZ (and associated intergranular fracture). In all three cases the decreases are more than expected from E decreases, but possibly more for the two PSZ materials, which

FIGURE 6.6 Fracture toughness (NB) of ZrO2 single crystals partially and fully stabilized with Y2O3 (stress axis probably <110>), as well as for a commercial Mg-PSZ material, versus test temperature. (From Ingel et al. [64], published with permission of the Journal of the American Ceramic Society.) Also note indent data of Guillou et al. [52].

must reflect added decreases due to decreasing effects of transformation toughening. Note also (1) Ingel et al. reported DCB tests at 22°C giving toughnesses of 3–6 and 1.5 MPa·m1/2 for respectively PSZ and CSZ crystals, and (2) Guil-

lou et al. [52] showed indentation toughness of Ca-CSZ crystals (14 m/o CaO,

< - >

indented on {111} planes with 110 cracks decreasing from 1.32 to 0.48 MPa·m1/2 for T = 22 to 800°C, i.e. consistent in this trend with the Y-PSZ crystal data).

Ball and Payne’s [65], NB tests of quartz crystals of differing orientations showed toughness decreasing from 0.8 MPa·m1/2 to a minimum of 0.6 MPa·m1/2 at 200°C and then increasing with further temperature increase (indicating limited effects of differing orientations, Chap. 2, Sec. III.B). Guillou et al. [52] also measured Vickers indentation toughness of MgO crystals for {100}<100> fracture showing a decrease from 1.7 MPa·m1/2 at 22°C to a minimum at 250°C and then a maximum of 1.3 MPa·m1/2 at 250°C and then a maximum of 1.9 MPa·m1/2 at 350°C with an 20 N load, with less pro-

nounced changes, higher minimum, and lower (e.g. 10%) maximum values at an3 N load. Mah and Parthasarathy [66] showed that SENB toughness of YAG single crystals increased from 2.2 MPa·m1/2 at 22°C to 4.5 and 5.5 MPa·m1/2 at 1600°C respectively in air and vacuum, with most of the increases above 1200°C and no significant orientation dependence.

The above review shows that crystal toughnesses often increase substantially at high temperatures due to plastic flow and less disparity between different fracture toughness tests and some of these with strength results, especially in their G dependence, as is commonly found in testing at or near room temperature. However, as is discussed later, this often has little or no relevance to hightemperature toughness of polycrystalline bodies, where grain effects, especially grain boundary sliding and failure, dominate, e.g. as indicated by the continued decrease in polycrystalline PSZ toughness (Fig. 6.6). Toughness and strength testing of bodies, especially single crystals showing substantial crack tip and even bulk plasticity (e.g. Fig. 6.6), indicate more consistent agreement regarding the onset and effects of such plasticity. However, caution and the need for more comparative testing is indicated by results of Hirsch and Roberts [67], who used toughness tests to determine the ductile–brittle transition in single crystals of Si. They found that this transition varied by up to 250°C between DCB and indentation fracture tests. The presence of dislocations at or near the crack tip due to the indent was an important factor in lowering the ductile–brittle transition temperature, e.g. polishing off most of the indent raised the transition temperature 50°C and abrading the area lowered the temperature 40°C.

B.Crack Propagation as a Function of G and Temperature

Slow crack growth (SCG) measurements at elevated temperatures typically use the same types of tests as are used for SCG due to environmental effects at room temperature and the same equation, i.e. Eq. (2.3), with the exponent n thus again being used as a value to characterize the process, as was discussed by Evans [68]. However, the first and most general of two aspects of such data is serious limitations of specific effects as a function of G or other grain parameters, and quantitative effects of grain boundary phases. Second is the transition from one mechanism to another as T increases, which, while aided by the common use of the same equation for various mechanisms of slow crack growth, has been widely neglected. Thus data is almost exclusively for room temperatures or at high temperatures where SCG due to grain boundary sliding dominates. However, Evans and Lange [53] showed that DT toughness of commercial SiC hot pressed with Al2O3 addition (having room temperature SCG due to moisture giving n 80 with activation energies characteristic of stress corrosion processes) was disappearing, i.e. n > 200, at 600°C. On the other hand, SCG was again clearly evident by 1400°C (n 21), but of a dif-

FIGURE 6.8 Examples of high-temperature slow crack growth in true tensile tested hot pressed SiC. SEMs showing regions of high temperature SCG (rougher areas outlined by dashed white lines) exposed by subsequently completing fracture at room temperature, with indicated temperatures and stresses. (From Ref. 69, published with permission of Brook Hill Publishing Company.)

that examining specimen surfaces after high-temperature stressing can also reveal other sites and approximate extents of high-temperature slow crack growth.

Turning to specific cases, static and cyclic crack growth studies of a commercial (AD-998) alumina (G 5 m) at 1200°C by Lin et al. [70] again showed mainly intergranular fracture, usually along a single crack due to SCG, but under some conditions growth of more than one crack was indicated. They also indicated some contribution of glassy phase bridging despite the very low level of such material in this 99.8% pure body. Horibe and Sumita [71] studied high temperature static and dynamic crack propagation fracture stress of two SiC bodies densified with B + C additions for two powders differing mainly in their contents of SiO2 (0.34 vs. 0.53 w/o) and Al (0.03 vs. 0.39 w/o). While some results are complex, their overall conclusion was that the higher Al content correlated with greater creep and lower strength at 1500°C, i.e. consistent with effects of Al2O3 additions in the previous section.

Baumgartner reported SCG of dense sintered TiB2 in molten Al at 970°C (of interest for possible use of TiB2 as electrodes for Al refining) based on both DT [72] and dynamic fatigue [73] tests (see Fig. 20 for σ–G–1/2 data). DT tests showed no SCG in purer, finer G (< 10 m, P 0.01) but definite SCG in larger G ( 20 m, P 0.02) with more impurities (especially 0.3 w/o oxygen), where the intergranular penetration of Al was accompanied by cell impurities of