The Kinetics of Isothermal Martensitic Transformation of Zirconia

Materials Transactions, Vol. 44, No. 9 (2003) pp. 1783 to 1789 #2003 The Japan Institute of Metals

The Kinetics of Isothermal Martensitic Transformation of Zirconia Containing a Small Amount of Yttria Jae-Hwan Pee1; * , Takahiro Akao1 , Shigeru Ohtsuka2 and Motozo Hayakawa1 1 2

Department of Mechanical Engineering, Tottori University, Tottori 680-8552, Japan Department of Mechanical Engineering, Yonago National College of Technology, Yonago 683-8502, Japan

The high temperature tetragonal phase of zirconia containing 1.40–1.60 mol% Y2 O3 were successfully quenched in at room temperature by rapid cooling of small spherical specimens. On these metastable specimens, the isothermal transformation behavior into the monoclinic phase was studied by dilatometric method. The transformation start time exhibited a C-curve in a TTT diagram. The C-curve shifted toward short time side as well as toward high temperature side with decreasing yttria content or increasing grain sizes. A study of the pre-aging effect during the incubation period on the subsequent isothermal transformation suggested consecutive growth of embryos, but alternative interpretation was also possible. The activation energy was evaluated to be
50 kJ/mol. (Received May 22, 2003; Accepted July 22, 2003) Keywords: zirconia-yttria alloy, martensitic transformation, isothermal transformation, time-temperature-transformation diagram, composition dependence, grain-size dependence, activation energy, dilatometric measurement

1.

Introduction

The phase transformation from the high temperature tetragonal to the monoclinic phase of zirconia has long been considered to be a typical athermal type martensitic transformation. Historically, Wolten1) was the first to suggest the athermal nature of the transformation. He conducted a high temperature X-ray experiment and observed a large hysteresis between the forward and backward transformation as well as stagnation of the transformation when the temperature was halted during the transformation. These observations led him to conclude that the transformation was an athermal type martensitic transformation. Since zirconia doped with a small amount of yttria behaves similarly except for the decreasing transformation temperature with increasing yttria content, the transformation mode has also been considered to be the same. As illustrated in Fig. 1 after Ref. 2), the Ms (martensite start temperature)

1500 Ms and A s , T/K

1300 As

1100 900

Ms

700 Isothermal Transformation

500 300 0.0

1.0

2.0

3.0

Y 2O3 content, x/mol% Fig. 1 The variation of Ms and As (reverse transformation start temperature) with Y2 O3 composition for bulk specimens measured with a continuous cooling and heating of 200 K/h2) along with the region of the isothermal t  m transformation4) after Ref. 2). *Graduate

Student, Tottori University.

decreases almost linearly with increasing yttria content up to 1.75 mol% Y2 O3 , but beyond which no transformation takes place not only during cooling to room temperature but also by a subsequent cooling to the liquid nitrogen temperature.2) This disappearance of the transformation is so sudden that it is difficult to be accounted for by a reduction of Ms with increasing amount of yttria. On the other hand, as also shown in Fig. 1, Y-TZP (Yttria doped Tetragonal Zirconia Polycrystals), which contain 2.6– 3.0 mol%Y2 O3 , are well known to exhibit isothermal transformation when annealed at around 550 K causing a sever degradation in the mechanical properties.3–5) The isothermal transformation takes place faster with decreasing yttria content. Thus, it was suggested that if the athermal nature of the transformation observed in the low yttria composition range is interpreted as a result of increased transformation rate due to the reduced yttria content so that the C-curve becomes intersected by the cooling curve in the TTT (Time Temperature Transformation) diagram, the above mentioned sudden disappearance of the transformation beyond 1.75 mol% Y2 O3 can be accounted for. Evidence for the validity of such interpretation, namely the isothermal transformation behavior in the lower yttria region, can be found in the literature.2,6,7) Among all, Hayakawa et al.2) found not only that the transformation temperature was dependent on the cooling rate but also the parent phase could be fully retained at room temperature for a certain composition range below 1.75 mol% Y2 O3 by a rapid cooling. The metastable tetragonal phase obtained by such a fast cooling readily transformed into the stable monoclinic phase by heating at around 550 K as expected from the isothermal nature of the transformation. More interestingly, the quenched metastable tetragonal phase underwent a burst type martensitic transformation into the monoclinic phase during a subzero cooling to the liquid nitrogen temperature. The occurrence of both isothermal and athermal modes of transformation in an alloy of a composition is very rare and no other system is known to transform at so different

1784

J.-H. Pee, T. Akao, S. Ohtsuka and M. Hayakawa

temperatures as in the present case. The isothermal nature of zirconia doped with a small amount of yttria has been recognized only recently.2,6,7) Even though it may be regarded as similar to the isothermal transformation of TZP, the essential nature of the transformation is not fully understood. In fact some researchers consider the isothermal transformation of TZP is bainitelike.8,9) In order to identify the rate controlling thermally activated process and clarify the transformation mode, quantitative measurements of the kinetics are necessary. In the present study, the kinetics of the isothermal transformation is studied with special interests in its dependence on the composition and grain size. 2.

Experimental Procedures

Series of specimens of different compositions, ZrO2 –1.30– 1.55 mol%Y2 O3 (1.30–1.55Y) were prepared from pure zirconia and ZrO2 –3.0 mol%Y2 O3 powders (TZ-0 and TZ3Y, TOSOH, Japan), both having 99.9% purity. Weighed powders for a given composition were mixed in a zirconia ball mill with ethanol for 50 h. After being dried, the mixed powder was lightly ground with a mortar and pestle and passed through a 100 mesh screen. Small spherical specimens were prepared from the mixed powders so that a fast cooling was possible to retain the high temperature tetragonal phase after sintering. A mixed powder was uniaxially pressed into a 1 mm-thick disk of 20 mm diameter in a steel die under 200 MPa. The sheet was diced with a knife-edge into cubes of 1 mm size. The cubes were granulated in a rotating cylinder whose inner walls were lined with emery paper. In about 24 h, beads of uniform diameter were obtained. They were placed in a platinum crucible and sintered at 1723–1873 K for 1 h in the ambient atmosphere. To avoid isothermal transformation of the specimens during cooling, they were taken out of the furnace when the furnace temperature reached 1273 K and rapidly cooled by immersing the outer surface of the crucible in water. Figure 2 shows SEM micrographs of a sintered specimen. The retention of the tetragonal phase after quenching was examined by X-ray diffraction method using Cu K
radiation. For the specimens sintered at 1723 K, a full retention of the tetragonal phase was possible with the specimens containing 1.40 mol% yttria and more. Specimens sintered at higher temperatures were less stable, but only specimens of fully tetragonal phase were used for the subsequent kinetic measurement. The densities of successfully quenched specimens were higher than 98% as measured by the Archimedes method. An average grain size of a sample was measured using the linear intercept method; the conversion factor 1.54 was multiplied to the average intercept length measured over 500 grains on the SEM micrographs taken from the crosssection of a sample. As seen in Fig. 2, equiaxed grains were uniformly distributed. The average grain sizes increased with increasing sintering temperature but nearly independent on the composition. For 1.55Y specimens, the grain sizes of 0.51, 0.63, 0.80, and 1.00 mm were obtained by sintering at 1723, 1773, 1823, and 1873 K, respectively. A handmade micro-dilatometer2) was used for monitoring the transformation behavior. It was a push rod type made of

Fig. 2 SEM micrographs of a dense spherical specimen of ZrO2 –1.6 mol% Y2 O3 prepared for rapid quenching experiments: (a) low magnification and (b) high magnification.

thin quarts rods; the specimen cell welded on one end of the supporting rod was heated by focused infrared gold image furnace (Shinku-Riko, Model MR-55W, Yokohama, Japan) and the displacement of the push rod was measured by a capacitance type transducer (MTI Accumeasure System, Model AS-2021SAI, USA). The maximum heating rate up to several hundred degree K was approximately 1000 K/min, but cooling rate was somewhat lower. Only specimens of good spherical shape and diameter between 600 and 700 mm were used for measurements. 3.

Results

Dilatometric data obtained from the isothermal holding of 1.50Y specimens sintered at 1723 K are shown in Figs. 3(a)– (f). To avoid overlapping of the dilatation curves, the data are plotted in three separate temperature ranges ((a)–(c)), and for each temperature range the early stage up to 300 s is replotted with an expanded time scale ((d)–(f)). Specimen’s temperature reached a preset temperature within a time between 30 and 50 s depending on the preset temperature. A small dilatation during this period was primarily due to the thermal expansion of the specimen, which increased with the holding temperature. Then dilatation due to the transformation followed a sigmoidal curve saturating at about 10 mm up

The Kinetics of Isothermal Martensitic Transformation of Zirconia Containing a Small Amount of Yttria

12

(a)

10

Dilatation, ∆ L/µm

Dilatation, ∆ L/µm

12

8 6

573 533

4

653

2

613

493

0

(d)

10 8 653

6

613

533

2

1000

2000

3000

0

100

Time, t/s

(b)

10

12 673

693 703

6

300

(e)

10

683

8

200

Time, t/s

Dilatation, ∆ L/µm

12 Dilatation, ∆ L/µm

573

4

0

0

4 2 0

673

8

683

6

693

4 703

2 0

0

1000

2000

3000

0

100

Time, t/s

Dilatation, ∆ L/µm

10 8 713 733

4

300

200

300

(f)

(c)

6

200

Time, t/s

12

12 Dilatation, ∆ L/µm

1785

753

2 0

10 8 6

733

753

4

713

2 0

0

1000

2000

3000

Time, t/s

0

100 Time, t/s

Fig. 3 Dilatation curves of the isothermal transformation at various temperatures for the rapidly cooled specimens of 1.50Y sintered at 1723 K. (a) data for 493–653 K, (b) data for 673–703 K, (c) data for 713–753 K; (d), (e) and (f) show the expanded early stage of (a), (b) and (c), respectively.

to 653 K. Note that a complete transformation of a 700 mm diameter specimen would yield an 11.7 mm linear expansion assuming the lattice volume expansion was 5%. A well defined plateau was observed at low temperatures. But this became shorter and eventually disappeared when the dilatations due to the transformation and thermal expansion merge at higher temperatures, say 653 and 673 K. With further increase of the temperature, the plateau tended to reappear indicating a delay of the beginning of the transformation (at temperatures 683 and 703 K in Fig. 3(b)). At these temperatures, the transformation behavior was no longer sigmoidal shape and was quite different from those observed at lower temperatures (see 693 and 703 K in Fig. 3(b)). With further

increase of the holding temperature, the transformation diminished. As seen in Fig. 3(c), transformation seemed to be still in progress at 713 and 733 K. On the contrary, the curve at 753 K appeared to show a gradual shrinkage of the specimen. At first sight, this phenomenon might appear suggesting a partial transformation by intersecting the nose of C-curve during heating and a subsequent reverse transformation at the holding temperature. However, this was verified not the case by the fact that a heating of the same specimen up to 1173 K with the same heating rate showed no sign of reverse transformation around the expected As point (
1073 K) in case a partial transformation had occurred during the heating.

1786

J.-H. Pee, T. Akao, S. Ohtsuka and M. Hayakawa

800

800

6

400

2 493

200

0

Temperature, T/K

600

4

Temperature, T/K

Dilatation, ∆ L/µm

1.40Y 1.45Y

700

1.50Y 1.55Y

600 500

733 400

-2 0

500

1000

1500

10

100

1000

10000

Time, t/s

Time, t/s Fig. 4 Blank tests of the dilatometer at two different temperatures: 493 and 733 K.

Fig. 6 A TTT diagram showing C-curves for the transformation start time for specimens with different yttria contents and sintered at 1723 K.

To clarify the origin of this negative slope, a blank test was conducted with the dilatometer at some temperatures. The results at 493 and 733 K are shown in Fig. 4. At 493 K the initial expansion due to the heating of the holder was maintained at constant. But at 733 K, after the initial expansion, gradual shrinkage was observed for quite an extended time. This seems to be caused by the gradual temperature redistribution through the measuring system, where only specimen cell was heated by a focused beam. Since the apparent shrinking behavior of 753 K curve in Fig. 3(c) is the same as 733 K curve in Fig. 4, the former was interpreted as showing neither forward nor reverse transformation. The transformation-start times as well as the times for 10% and 50% transformation obtained from the above dilatometric curves were plotted in a TTT diagram in Fig. 5. The transformation start time was read from the point of a definite increment (roughly 2% of transformation) from the plateau. But this criterion became more or less uncertain when a flat plateau was absent. These times read from Fig. 3 were corrected for the time expended for heating to the preset temperature for isothermal holding. The temperature dependence of the transformation kinetics was represented by typical C-curves: flatter top and significant temperature dependence below the nose. The horizontal

line at 743 K represents an approximate upper limit of the isothermal transformation of the present sample. Although the low temperature data were limited by the time of measurement, a partial transformation to the monoclinic phase was observed even at room temperature during an extended period of time. The isothermal transformation of this sample proceeds quite fast and the usual cooling rate employed for a bulk ceramic sample should have intersected the C-curve resulting in transformation, which seems to have been mistaken as athermal type transformation. Similar dilatometric measurements were conducted for specimens of different compositions and grain sizes to see the effect of the composition and grain size on the transformation kinetics. The results are summarized as the transformation start time in TTT diagrams. Figure 6 shows the effect of yttria content on the C-curve for the specimens sintered at 1723 K. Because the average grain sizes were essentially the same, the effect of grain size could be ignored. The C curves shifted toward the long time side as well as toward low temperature with increasing yttria content. Although this trend is qualitatively well expected from the stabilizing effect of yttria, it is worth noting that only 0.05 mol% addition of Y2 O3 content resulted in a significant delay in the transformation start time. It is also noted that the reduction in the nose temperature between 1.45Y and 1.55Y in Fig. 6 is comparable with the reduction in the Ms (transformation temperature during a continuous cooling) between the same compositions plotted in Fig. 1. Figure 7 shows the effect of grain size on the C-curve for specimens of 1.55Y sintered at different temperatures. The Ccurves shifted toward the short time side as well as toward higher temperature with increasing grain size. The behavior is consistent with often reported un-stabilizing effect of grain coarsening.10–16) As an Ms point is influenced by both T0 (equilibrium temperature between the chemical free energies of the parent and martensite phases) and a critical driving force (difference in the chemical free energies to trigger the nucleation of martensite), the position of a C-curve in a TTT diagram can be considered to be influenced by the same parameters. Although the stabilizing effect due to yttria content is expected through the reduction of T0 and that due to grain refinement must be through the increased driving force required for triggering the transformation, these effects

Temperature, T/K

800 2% 10% 50%

700 600 500 400 10

100

1000

10000

Time, t/s Fig. 5 A TTT-diagram showing C-curves for 2% transformation, 10% transformation and 50% transformation for 1.50Y specimen sintered at 1723 K.

The Kinetics of Isothermal Martensitic Transformation of Zirconia Containing a Small Amount of Yttria

Temperature, T/K

800

0.51µm 0.63µm 0.80µm 1.00µm

700 600 500 400 10

100

1000

10000

Time, t/s Fig. 7 A TTT diagram showing C-curves for the transformation start time for 1.55Y specimens with different grain size.

Temperature, T/K

800

(a)

1000

10000

Time, t/s

Temperature, T/K

(b)

0s 300s

700

600s 600 500 400 10

100

1000

or 600 T 523

ð1Þ

tTtotal ¼ tT þ tT0

400

800

300

600s

500

100

tT0 ¼

300s

600

10

tion start times for the subsequent aging as seen from the shifted C-curves in Fig. 8(a). The result seems to imply that some sort of preparation toward transformation or even subtle transformation below detection limit was in progress during the incubation period; certainly not a stage of simple waiting for a nucleation event governed by a statistical probability. It was then examined if such a change during the incubation period proceeded linearly with time until it reached the critical state at the transformation start time. If this hypothesis is correct, the effect of pre-aging for a certain period (300 or 600 s) at 523 K can be regarded as equivalent to the aging for tT0 at temperature T. That is:

where 523 and T are the incubation times (in unit s) at 523 K and TK, respectively. Thus, the effective transformation start time at T, tTtotal , for the specimens pre-aged at 523 K can be evaluated by adding the effective contribution by the aging, namely:

0s

700

1787

10000

Time, t/s Fig. 8 (a) C-curves of the transformation start time for 1.55Y specimens pre-aged at 523 K for 300 and 600 s in the incubation period as compared with that of no pre-aging. (b) Same as (a) but the transformation start times were corrected for the pre-aging effect (see text).

on the C-curves were indistinguishable. As seen in Fig. 3(a), the isothermal transformation behavior at lower temperatures was characterized by a well defined incubation period before a detectable amount of transformation was observed. To see if any significant change was occurring during this incubation period, the effect of preaging on the subsequent isothermal transformation was studied. Figure 8(a) shows the effect of the aging treatments at 523 K for 300 and 600 s on the subsequent transformation start times. Although these aging treatments were well within the incubation stage, they definitely shorten the transforma-

ð2Þ

where tT is the observed transformation start time at T for the specimen pre-aged at 523 K for 300 or 600 s. The corrected transformation start times are plotted in Fig. 8(b), where we see that the effective time expended before the transformation started after the pre-aging agreed well with those for the as sintered specimens at all temperatures. The result supports the initial hypothesis, namely a change toward transformation proceeded linearly with time during the incubation period. Finally, activation energies were evaluated from the presently obtained C-curves. Figures 9(a) and (b) show Arrhenius plots of the reaction rates, which are taken as the inverse of the transformation start time in Figs. 6 and 7. An activation energy was evaluated from the slope of the linear portion of the Arrhenius plot which corresponds to the data well below the nose temperature of C-curve.17,18) The data presented by open symbols which denote the data near and above nose of the C-curve were not used for the linear fit. Activation energies were about 50 kJ/mol, irrespective of yttria content and grain sizes. 4.

Discussion

Isothermal nature of the tetragonal to monoclinic phase transformation was clearly shown for zirconia doped with 1.40–1.55 mol% Y2 O3 . The isothermal nature is the same as those observed in Y-TZP, which contains 2.5–3.0 mol% Y2 O3 . The isothermal nature seems to extend to pure zirconia as verified by the time dependence of the transformation observed through the transformation temperature.6) The reason why the t-m transformation of this composition range had long been considered to be typical athermal martensite seems to be due to its fast reaction. In fact, the time intervals of X-ray measurement used for monitoring the transformation by Wolten1) were longer than 30 min, by which time the amount of transformation must have well saturated to the value prescribed by the temperature. When the specimen’s temperature was reduced by a step, the amount of the m-

1788

J.-H. Pee, T. Akao, S. Ohtsuka and M. Hayakawa

2 (a)

Q 1.45Y = 50.1kJ/mol

0

Q 1.50Y = 51.1kJ/mol

ln(1/t)

-2

Q 1.55Y = 51.1kJ/mol

-4 -6 -8 -10 1.0

1.5

2.0 -1

-3

T /10 K

2.5

3.0

-1

2 (b)

Q1.00µm = 48.0kJ/mol

0

Q0.80µm = 51.3kJ/mol Q0.63µm = 51.0kJ/mol

ln(1/t)

-2

Q 0.51µm = 51.1kJ/mol

-4 -6 -8 -10 1.0

1.5

2.0 -1

-3

T /10 K

2.5

3.0

-1

Fig. 9 Arrhenius plots of the data (a) from Fig. 6 for different yttria contents and (b) from Fig. 7 for different grain sizes.

phase increased gradually but soon saturated to a level prescribed by the temperature. This behavior suggests that the transformation is more like martensite rather than bainite. The isothermal transformation behavior looked considerably different at temperatures below and above nose temperature. Below nose temperature it is characterized by a long incubation period followed by a well defined sigmoidal curve, while above nose temperature there was essentially no incubation period and the transformation proceeded more or less linearly with time. Although it was verified that no transformation took place during the heating of a specimen above the nose of C-curve, the continuous heating up to the holding temperature might have resulted in the apparent absence of the incubation period. The marked difference in the transformation rate, however, seems arise from the differences in the driving force and the degree of autocatalysis at temperatures below and above the nose temperatures. As seen from the position of the C-curves in Fig. 6, the transformation start time is sensitively dependent on the yttria content. With decreasing yttria content, not only the Ccurve shifted toward short time side but it shifted toward higher temperature side. Although the latter shift was small, the shift becomes evident if we compare with the nose temperature of 3Y-TZP (473K).9,19) The position of C-curve was also dependent on the average grain size of specimen; with increasing grain size the C-curve shifted toward short time side and toward higher temperature side. The unstabilization effect due to the variation of the grain size from 0.5 to 0.8 mm was nearly equivalent to that caused by the composition change from 1.55 to 1.45 mol% Y2 O3 .

At a temperature well below the nose, the transformation behavior is well characterized by a long incubation period where essentially no transformation was observed. In some theories of isothermal martensitic transformation,20,21) the incubation time is interpreted as the time before a first nucleation to occur which is governed certain statistical probability under a given activation energy and driving force. If this is the case, the probability of nucleation during a given period is constant and thus an interruption of aging before nucleation should have no effect on the following aging. But this is not the case observed in the present experiment, instead the effect of aging during incubation time was cumulative. This seems to suggest a continuous growth of embryos during incubation period. This may appear to support the embryo growth model proposed recently by Borgenstam and Hillert.22) However, the present method of monitoring the transformation was not very sensitive. Thus, a small amount of martensite may have already occurred even in the stage which we thought as an incubation period. Further study is necessary on this point. We have obtained
50 kJ/mol for the activation energy of the nucleation of the isothermal transformation of the present specimens; the value was the same within the accuracy of measurement irrespective of the composition and the grain sizes of the present specimens. Activation energy of 92 kJ/ mol was obtained for the isothermal transformation of 3YTZP under water vapor pressure of 3350 Pa;19) the value was interpreted as corresponding to the chemical reaction between water absorbed on the surface and Zr–O–Zr bonds. On the other hands the activation energy of O2 ions in YFSZ was reported to be
120 kJ/mol.23) The presently obtained activation energy 50 kJ/mol is considerably lower than the other two values. However, we cannot identify the corresponding rate controlling step for the isothermal transformation. 5.

Conclusions

The kinetics of isothermal martensitic transformation of zirconia doped with small amounts of yttria was studied quantitatively. The transformation behavior at temperatures below the nose of C-curve is characterized by a long incubation period followed by a sigmoidal curve. On the other hand, at higher temperatures there was little incubation time and the transformation proceeds more steadily with time. The amount of yttria and grain size gave significant influence on the location of the C-curve, namely the stability against the transformation. The effect of aging during the incubation period was found to be cumulative. However, a caution is necessary to properly interpret this because of the limited sensitivity of the measurement of the transformation. The activation energy of 50 kJ/mol was obtained for the nucleation of the isothermal transformation of the present specimens; they are considerably lower than those corresponding to the chemical reaction between H2 O and ZrO2 on the specimen’s surface and the diffusion of O2 ions in YFSZ.

The Kinetics of Isothermal Martensitic Transformation of Zirconia Containing a Small Amount of Yttria

REFERENCES 1) G. M. Wolten: J. Am. Ceram. Soc. 46 (1963) 418–422. 2) M. Hayakawa, K. Nishio, J. Hamakita and T. Onda: Mater. Sci. and Eng. A273–275 (1999) 213–217. 3) K. Kobayashi, H. Kuwajima and T. Masaki: Solid State Ionics 3/4 (1981) 489–493. 4) T. Sato, S. Ohtaki and M. Shimada: J. Mater. Sci. 20 (1985) 1466– 1470. 5) F. F. Lange, G. L. Dunlop and B. I. Davis: J. Am. Ceram. Soc. 69 (1986) 237–240. 6) M. Hayakawa, L. Wang and H. Tottori: 20th ASM Heat Treating Soc. Conf. Proc., St. Louis, Missouri, October 2000 (ASM International Materials Park, 2001) pp. 466–469. 7) H. Tsubakino and N. Matsuura: J. Am. Ceram. Soc. 85 (2002) 2102– 2106. 8) N. Nakanishi: J. Jpn. Soc. Powder and Powder Metallurgy 33 (1986) 356–360. 9) N. Nakanishi and T. Shigematsu: Mater. Trans., JIM 32 (1991) 778– 784. 10) R. C. Garvie and M. V. Swain: J. Mater. Sci. 20 (1985) 1193–1200. 11) F. F. Lange: J. Mater. Sci. 17 (1982) 225–234.

1789

12) C. A. Andersson and T. K. Gupta: Science and Technology of Zirconia, ed. by A. Heuer and L. W. Hobbs (Am. Ceram. Soc., Columbus, Ohio, 1981) pp. 184–201. 13) I.-W. Chen and Y.-H. Chiao: Acta Metall. 31 (1983) 1627–1638. 14) A. H. Heuer, N. Claussen, W. M. Kriven and M. Ru¨hle: J. Am. Ceram. Soc. 65 (1982) 642–650. 15) M. Hayakawa, H. Shinmen and M. Oka: Proc. ICOMAT-92, ed. by C. M. Wayman and J. Perkins, (Monterey Institute of Advanced Studies, Monterey, CA, 1993) pp. 677–682. 16) P. F. Becher and M. V. Swain: J. Am. Ceram. Soc. 75 (1992) 493–502. 17) G. V. Kurdjumov and O. P. Maksimova: Dokl Akad. Nauk SSSR 61 (1948) 83–86. 18) G. V. Kurdjumov and O. P. Maksimova: Dokl Akad. Nauk SSSR 73 (1950) 95–98. 19) T. Sato, S. Ohtaki, T. Endo and M. Shimada: J. Am. Ceram. Soc. 68 (1985) C-320-C-322. 20) S. R. Pati and M. Cohen: Acta Metall. 17 (1969) 189–199. 21) T. Kakeshita, K. Kuroiwa, K. Shimizu, T. Ikeda, A. Yamagishi and M. Date: Mater. Trans., JIM 34 (1993) 423–428. 22) A. Borgenstam and M. Hillert: Acta Mater. 48 (2000) 2777–2785. 23) Y. Ikuma, Y. Tsubaki and T. Masaki: J. Ceram. Soc. Japan (Seramikkusu Ronbunshi), 99 (1991) 101–103.