Mechanical Properties of Ceramics and Composites

toughness are indicated by their ratio passing through a maximum at intermediate φ The benefits of finer SiC particles was also shown by Nakahira and colleagues [54] via a strength maximum at φ 0.1 of 500 MPa (D 2 m) and 400 MPa (D 8 m), and a decrease in strength with coarsening of the microstructure increasing both D and G.

Higher strengths with finer SiC particle size were also shown by Kim and Kim [55] for Al2O3-SiC composites made by the directed oxidation of Al metal (and leaving some residual metal, e.g. Fig. 8.20). Their starting strengths clearly decreased as the size of the added SiC particles increased (Fig. 15; Chap. 11, Sec. III.C).

Considerable research has been conducted on Al2O3 composites made with nanoscale SiC particles (despite this presenting some processing challenges), prompted by Niihara’s [56] work reporting strength increases from 320 MPa to > 1000 MPa with more limited toughness increases (from 3.2 to 4.7 MPa·m1/2) with addition of 5 v/o SiC. Thus Zhang et al. [57] showed strengths reaching maxima of 1100, 950, and 900 MPa respectively at SiC particle sizes of 0.05, 0.17, and 2.5 m, all at 3 v/o SiC starting from 550 MPa without SiC (all with highly polished surfaces). Strengths dropped rapidly and then more slowly as φ increased, reaching 800 MPa at 24 v/o, with much less differentiation between the different particle sizes. In contrast, (I) toughnesses for the finer two particle sizes varied only a limited amount from that of the pure Al2O3 ( 3.3 MPa·m1/2) while decreasing to a minimum of 3 MPa·m1/2 at 3 v/o, then increasing steadily to 4.7 MPa·m1/2 at 24 v/o, i.e. clearly different from and in the latter case opposite to the trends for strength. While many obtain less spectacular but still substantial increases in strength, some have also shown considerable improvement after annealing, e.g. Zhao et al. [58] showed increases from 760 to 1000 MPa with D = 0.15 m. It is clear that toughness changes have at best a limited effect on strength increases and that reductions of the Al2O3 G, e.g. to 200 nm, is an important factor in these increases (e.g. see Figs. 3.13–3.15). Reduction of flaw sizes [59] or their severity due to surface compressive stresses [60] and less relaxation of these on annealing or crack healing [61] have been indicated as additional factors.

Consider next Al2O3-TiC composites, for which there is much less data on this system in the literature, especially on its microstructural dependence, reflecting its earlier, mainly empirical, development a number of (e.g. 40) years ago for use in various wear and cutting tool applications. Rice [32] reviewed the literature and made measurements on some commercial bodies (commonly with 30 v/o TiC and D < 1 to 2 + m). While overall modest increases in toughness often correlated with modest increases in strength as φ increases, there was substantial variation, with approximate flaw sizes calculated for strength-to-toughness ratio (Fig. 9.10), ranging from reasonable values of several tens of microns to very questionable values of hundreds of microns. Though finer particle sizes are used,

points as transitions between the high and normal strengths. This data appears consistant with plastic deformation control of strength of BaTiO3, i.e. suppressed at very fine G resulting from SiC additions without higher temperature processing coarsening the structure, but approaching normal strength behavior with significant grain coarsening, with the larger grains generally playing a dominant role in strength. This intrepretation is supported by the close parallel behavior of hardness. However, some impact of increased E values (up to 50%) on strength as the level of SiC addition increases may also be operative, as may both conversion from a tetragonal to a pseudocubic structure as G is reduced below 1/2 to 2/3m. Modest ( 30 %) increases in toughness also probably contribute to strength increases but clearly cannot explain strength increases of up to > threefold.

Studies of concrete corroborate both the tradeoffs often seen between toughness and strength and the mechanism causing this tradeoff. Thus Strange and Bryant [64] reported that NB toughness tests of concretes gave increasing toughness as the aggregate (i.e. dispersed stones or stone fragments) sizes increased, but strengths decreased. This difference was attributed to crack nucleation from the aggregate particles thus limiting strengths, since such cracks, whose size is related to that of the aggregate particles, could initiate failure.

D.Composites with Polycrystalline Nonoxide Matrices

Consider next data for composites of all nonoxide constituents, beginning with Si3N4-SiC bodies, which have had substantial investigation. Lange [65] showed strengths generally trending in the opposite direction from the SiC particle size dependence of toughness (Fig. 8.16), i.e. strength decreasing as D increased while toughness increased across the range of additions (Fig. 9.12). While data of Tanaka et al. [66] appears as a clear extension of the toughness–D trends of Lange, it is more uncertain in terms of extending the strength trends, but the differences in strength between the two studies probably reflects differences in finishing and testing and hence flaw populations. This possibility is heightened by considering the strength-to-toughness ratio ( the reciprocal of the flaw size), which very clearly indicates this data as an extension of Lange’s data and the inverse trend of strength and toughness as a function of D (Fig. 9.13). Strength decreasing with increasing SiC particle size is clearly shown in plotting strength versus SiC particle size for various volume fractions of SiC from Lange’s and other studies (Fig. 9.14). Data of Nakamura et al. [67] and others [68–72] supports this strength–particle size trend, showing strengths of 980 MPa for Si3N4 and progressively decreasing for 50 w/o SiC additions of increasing D from 0.3m (900 MPa) to 1 m (770 MPa). Also note that Pezzotti and Nishidas’ [69] showed strengths that, though decreasing with increasing SiC particle size, were higher than for their Si3N4 alone till SiC particle sizes of 35 m and were progressively lower for all larger SiC particle sizes. They also showed that

FIGURE 9.12 Strength versus volume fraction SiC particles of various sizes as shown from Lange [65] and Tanaka et al. [66]. Note the limited probable corrections for limited porosity (dashed lines and arrows). Contrast the dependence on D with that for toughness (Fig. 8.16). (From Ref. 50. Published with permission of the Journal of Materials Science.)

lower strengths with larger particles were associated with fracture initiation from larger SiC particles. Evaluation discussed below showed a D-1/2 dependence analogous to the G dependence of strengths of monolithic ceramics (Fig. 9.15).

Studies of composites of Si3N4 with nanosize SiC particles support the general trends for strengths to increase with finer SiC particle sizes and for there to be better correspondence of strength and toughness. Thus Sawaguchi et al. [70] reported strengths rising from 1 GPa with no SiC to 1.2 across the range of 10–50 v/o additions (and toughness rising from 5 to 7 MPa·m1/2 at φ= 0.1 and then decreasing to 5.2 MPa·m1/2 at φ= 0.5) for composites with SiC particles of 50 nm. Sasaki et al. [71] showed that addition of 300 nm SiC particles increased strengths from 780 MPa at φ= 0 to a maximum of 920 MPa at φ= 0.1, which then decreased substantially. This trend was similar to that of toughness, but the increase of the latter was 46% versus 18% for strength, and both were much greater than the 8% increase in E. They noted that the maxima in

FIGURE 9.13 The ratio of strength to toughness versus volume percent SiC particles for various particle sizes shown from data of Lange [65] and Tanaka et al. [66]. Note the implied flaw sizes (left scale which are the inverse of the ratio indicating reasonable sizes for the finest particle sizes, and clearly excessive for the larger particles) and Tanaka et al.’s data indicated as an extension of Lange’s. (From Ref. 50. Published with permission of the Journal of Materials Science.)

strength and toughness were associated with the occurrence of rod shaped Si3N4 grains in the range of 5–10 v/o SiC and the formation of larger fracture sources beyond such additions. Similarly Tian et al. [72] showed strengths increasing from 750 MPa at 0 v/o to a maximum of 950 MPa at 5 v/o ( 12% increase) and then decreasing to 650 MPa at 25 v/o, with a similar trend for toughness, but again with a greater ( 35%) increase to a maximum of 7.7 MPa·m1/2. Again, the use of very fine particles commonly resulted in substantial inhibition of matrix grain growth and incorporation of some SiC particles in the Si3N4 grains.

Petrovic et al. [73] showed that strengths of Si3N4 with additions of 3 m MoSi2 particles (+ 1 w/o MgO densification aid) were constant at 1 GPa but had a modest maximum of 5% at 30–40 v/o MoSi2 and then a definitive decrease to 900 MPa at 50 v/o, while the addition of 10 m MoSi2 particles resulted in a decrease in strength generally increasing as φ increased, e.g. to 700 MPa. Addition of 10 m MoSi2 particles (+ 5 w/o MgO densification aid), while

FIGURE 9.14 Strength versus SiC particle size for various indicated volume percent of SiC particles in Si3N4 matrix composites from various investigators [65–71]. This shows a clear trend for significant decreases with increasing particle sizes, opposite of toughness trends. (From Ref. 50. Published with permission of the Journal of Materials Science.)

giving a modest maximum of strength at 10 v/o of 700 MPa (up from 600 MPa at φ=0), then substantially decreased to 300 MPa at 50 v/o. This strength behavior is in marked contrast to that of (I) toughness, which showed all three compositions reaching maxima of > 8 and > 6 MPa·m1/2 for composites with the 10 m MoSi2 particles and respectively for 1 and 5 w/o MgO, and a maximum of5.5 MPa·m1/2 for the composite with 3 m MoSi2 particles, i.e strength and toughness had essentially opposite dependences on φ and D.

Mah et al. [74] reported that strengths of their composites of Si3N4 with TiC particles (average and maximum size 8 and 30 m) decreased from 700 MPa at 0 v/o to 450 MPa at 50 v/o, contrary to the sharp (IF) toughness maximum of 7 MPa·m1/2 at 20 v/o (from a baseline of 4–5 MP·am1/2), as well as a progressive increase in E with increasing TiC content. Reasonable flaw sizes were generally indicated by strengths and toughnesses, especially for other than the high toughness values, and were consistent with sizes of larger TiC particles or clusters of them found on fractures surfaces, probably as fracture origins. Given the particle size and range and the Si3N4-TiC expansion difference, microcracking is the likely cause of strength reduction and probably of the toughness

FIGURE 9.15 Strength versus the inverse square root of SiC particle size (D) at 22°C. Note the clear bilinear dependence on D-1/2 similar to that for glass matrix composites (Figs. 9.2, 9.17) and monolithic ceramics (Fig, 3.1, Chapters 3,11). (From Refs. 55,65,69.)

maxima, hence being another example of opposite trends of strength and toughness due to microcracking.

Turning to SiC-TiB2 composites, while Yoon and Kang [75] showed (IF) toughness was independent of TiB2 particle size (at 5.3 MPa·m1/2), strength decreased from 580 MPa at D 2 m to 350 MPa at D = 12 m. However, strengths of specimens of Cho et al. [76] made with TiB2 particles 1–15 m were nearly independent of TiB2 content to the limit of testing (70 w/o) but with a probable limited minimum of 500 MPa at 50 w/o (in contrast to a clear (I) toughness maximum of 4.2 MPa·m1/2 at 50 w/o). While McMurtry et al. [77] did not report the dependence of strength on TiB2 particle size, they showed that while toughness for φ= 0.16 and D 2 m increased from 6.7 to 9 MPa·m1/2 as uniformity of mixing increased, strengths were unchanged. Magley et al. [78] provided evidence of stress-induced microcracking in SiC–15 v/o TiB2 by measuring residual stress in bodies before and after strength testing by x-ray techniques. After failure, residual stresses were 60% lower, which was attributed to stress-induced microcracking.

Cho et al.’s [76] SiC-TiC composites, similar to their SiC-TiB2 composites

above, showed a more pronounced strength minimum of 400 (from 600 ) MPa·m1/2 versus a toughness maximum of 4.5 (from 2.7) MPa·m1/2 at 50 w/o. The strengths of Endo et al. [79] were also 600 MPa at φ= 0 but went through a modest maximum of 760 MPa at 30 w/o and then tended to decrease to 400 MPa at 100% TiC, in contrast to a pronounced maximum in (NB) toughness of 6 MPa·m1/2 (versus 2 and 3 MPa·m1/2 for pure SiC and TiC respectively). Greater differences were shown by Lin and Iseki’s [80] SiC-TiC toughnesses, which increased from 4.5 to 9 MPa·m1/2 as TiC content increased from 0 to 40 v/o; their strengths decreased from 530 MPa at 0 v/o to 430 MPa at 10 v/o and then remained constant.

In contrast to opposite strength and toughness trends with composite composition, data of Li and Watanabe [81] on SiC with AlN show a maximum in strength at 5 v/o AlN, similar to the maximum in toughness at 10 v/o AlN. Much and possibly all of the rise in strength appears to be due to substantial initial reduction in SiC grain size as AlN was added. The subsequent decrease in strength with larger AlN additions must reflect, at least in part, the decrease in E as AlN content increases. Such effects of composition on E and especially grain size could well differ between strength and toughness and thus be a possible explanation for some differences between them.

Finally, note two aspects of flaw and particle size dependence, starting with strength-to-toughness ratios for the compositions. The above SiC-AlN data shows no clear trend, the ratio averaging 136 m1/2, implying a reasonable flaw size of 50 m, while SiC-TiC data gives values ranging from too low to too high [32,81]. Other composite systems previously evaluated [32] gave either scattered values giving reasonable flaw sizes or (for Si3N4-TiC) values decreasing substantially as φ increased to either reasonable or larger indicated flaw sizes. Second, the limited data for particle sizes ranging from below to above expected flaw sizes clearly indicate a two-branch behavior (Fig. 9.15) as has been found for glass matrix composites (Figs. 9.2, 9.17) and monolithic ceramics (Fig. 3.1). Thus at finer particle sizes there is limited but generally some increase in strength as D decreases, consistent with substantial other data above showing such behavior in this D range of a few to several microns. However, once the particle size (or particle cluster size) reaches the flaw size, the particles become the flaws, giving greater particle size dependence to the composites. Note that matrix grain size still has an effect on strength, as is commonly observed above, since it impacts the size of machining flaws [e.g. via Eq. (3.2)], as well as effects on local fracture toughness.

E.Platelet and Whisker Composites

Consider first whisker composites, where again much of the emphasis has been on toughness, so there is less data on strength. However, since whisker compos-

ites have been extensively investigated, there is reasonable strength data available. Thus Table 8.2 presents a summary of data for whisker composites that shows strength increases commonly being similar to toughness increases, and in the case of composites with lower Young’s moduli, strength increases are often similar to increases in E. One exception to this is composites of SiC whiskers in Si3N4 matrices, which along with composites of SiC whiskers in Al2O3 are discussed further below.

The comparisons in Table 8.2 address primarily composition and not specifics such as the effects of whisker dimensions, some of this information is available for Al2O3- SiC whisker composites. As noted in Chap. 8, Sec. V.D, toughness commonly increased with diameter, sometimes with length, or both (i.e. with aspect ratio) of the SiC whiskers and was affected by whisker surface character. Baek and Kim [82] showed that doubling whisker lengths from 9 to 18 m increased strengths with 20 v/o whiskers 20 % (from 380 to 450 MPa), i.e. between the increases in toughness measured by NB and CNB techniques, but closer to the former. They also reported that such increases were consistent with a model for whisker pullout. However, Yasuda et al. [83] reported strengths of composites with the same whisker contents decreasing by 30% over the same range of whisker length increase, i.e. from 700 to 500 MPa (the latter also being the matrix strength without whiskers), with further decreases to 400 MPa at whisker lengths of 50 m. These observations are consistent with their observed negative effects of increased whisker length on toughness (while increased diameter was reported to improve toughness, its effect on strength were not reported). However, Krylov et al. [84] reported significant (e.g. 140%) increases in strength as whisker diameters increased from 0.1 to 2 m, with > 1/2 the increase occurring by 0.5 m, i.e. with the benefits saturating or reaching a maximum at 2 m (as was also the case for toughness increases, though they were only 1/4 those of strength). A maximum in benefits with increasing whisker dimensions would be expected due to the onset of microcracking, which was indicated by Tiegs and Bowman [51,85] as the source of low strength ( 180 MPa) in composites with 20 v/o SiC whiskers averaging 5 m in diameter. The character of whisker surfaces also appears to be a factor in strength, often following that of toughness as reported by Tiegs and Bowman [51] for various whisker treatments. However, Steyer and Faber [86] reported that increasing the thickness of the carbon coating on the whiskers progressively reduced strengths, e.g. from 600 to 400 MPa, as coating thickness went from 0 to 50 Å. On the other hand, a 20 Å carbon coating of the whiskers gave toughnesses that were independent of indent load (versus uncoated whiskers showing toughness starting from 20% lower, increasing to 20% higher as indent load increased).Yang and Stevens [87] showed that the presence of an interfacial silicate film was detrimental to whisker composite properties.