J O U R N A L OF M A T E R I A L S S C I E N C E L E T T E R S 11
(1992) 1711-1714
Effect of grain size on the cyclic fatigue under rotary bending
behaviour
of sintered
AI203
H. N. KO
Nakanihon Automotive College, Sakahogi-cho, Kamo-gun, Gifu-ken, 505 Japan
It is important to know the cyclic fatigue behaviour of ceramics when using them as structural components. Although their behaviour has already been examined under various types of loading, elementary data are not sufficient. More information on factors affecting the fatigue strength seems to be necessary to clarify the fracture mechanism of ceramics and their reliability as structural materials. The effect of grain size on the fatigue strength of sintered A1203 has not been studied much in comparison with the effect on the static strength, and basic data are of limited availability [1]. Morebasic data are necessary to clarify the effect of the grain size on the fatigue behaviour of sintered A1203. Therefore, specimens with three grain sizes were prepared and rotary bending tests were performed at room temperature. The test results were compared with the static bending strength measured by a non-rotating fatigue machine. The material used was sintered A1203 obtained from Kyocera, Japan, with an alumina content of 99%. The material fabricated with SiO2 and MgO, etc., as additives (Kyocera, private communication). Three kinds of specimens with different grain sizes were prepared, changing the sintering temperature slightly by 25 °C. Each material microstructure is shown in Fig. 1. The average grain sizes, determined by the linear intercept method, were about 5, 8 and 19/zm. The material characteristics such as the fracture toughness and density were almost the same. The specimen had a cylindrical shape and its dimensions were the same as those adopted in the previous fatigue test on sintered A1203 [1-3]; the
diameter of the middle part was 8 mm and each end of the specimen had a larger diameter of 12 mm for chucking. The specimen was ground perpendicularly to its axis to make the finished surface smooth. The machine used was an Ono rotary bending fatigue testing machine operating at 3420 cycles min -1 . The loading type of the machine was four-point bending, and the stress state of the specimen was reversed bending. The fractured surfaces were examined by scanning electron microscopy (SEM). Fatigue testing for three kinds of specimens with different grain sizes was carried out generally for two specimens at each stress level and within the range 104-108 stress-cycles. Some specimens were tested at cyclic numbers of > 108 to examine the existence of a knee; 108 stress-cycles is equivalent to about 20 days under the present test conditions. Test points for samples with different grain sizes are plotted together on a semilogarithmic graph in Fig. 2; the
22 I IILIItll IIIIII I Fllrlrllr Ill! 2O I I ......
oI I,H,, blO
~ . ~
I
_
8
6 IIIIIIII l lllNI I ,, i i'r:ii::i [111111I I ll!lllll I rll 103
104
105
106
107
108
109
N
Figure2 S-N curvesfor materialsof grain size: (O) 19/zm, (A) 8/xm and (D) 5/xm.
FigureI Microstructuresof materialsof grain size(a) 19/xm,(b) 8/xm and (c) 5/xm.
0261-8028 © 1992 Chapman & Hall
1711
arrowed points indicate the specimens for which testing was stopped. It is known that the fatigue strength of a finer-grained material is higher than that of a larger-grained material. It can be considered that the grain size is the primary factor for increased fatigue strength, although some characteristics of materials, such as the grain boundaries, may be different. The life of each material increases remarkably as the stress amplitude decreases, and each material does not have a distinct knee at cyclic numbers of < 107; some specimens were fractured at near 2 x 107 stress-cycles. However, from the figure, the knee seems to exist at cyclic numbers above 108; the assumed fatigue limits seem to be about 7, 10 and 12 kg mm -2, respectively. A similar S - N curve, indicating the existence of a knee, was obtained in previous studies [1-3] for sintered A1203 with grain sizes of 10 and 28/xm. These rotary bending results suggest that fatigue testing should be carried out for at least more than several tens of days to confirm the basic fatigue strength of sintered A1203, as well as sintered Si3N4 [4-6]. A static bending test was also performed using an Ono rotary bending testing machine. After the specimen was attached to the machine, the applied stress was increased gradually without rotating the machine until the specimen was fractured. The mean static bending strength obtained from two specimens are shown in Table I; the results obtained in the present study are summarized in the table. The static strength of finer-grained material is higher than that of larger-grained material. The strength increase with grain size was ordinary in comparison with reported data [7, 8]. Comparing the assumed fatigue limit with static strength, the ratio of the fatigue limit to the static bending strength decreased as the grain size increased. A similar tendency was obtained on the previously sintered A1203; the ratios were 0.32 for material of 10/xm grain size and 0.15 for 28/xm, grain size. These results show that the ratio is high as the material is strong. A similar tendency is seen in the results for sintered Si3N4 [4-6]. The ratio is important from the practical viewpoint, so it seems necessary to obtain more information on ceramics. As it appeared possible to correlate the static bending strength with the fatigue strength, the test results were plotted together on a logarithmic graph; the S - N curves, including the static strength, are shown in Fig. 3. Each straight line in the figure was obtained by the least-squares method from the data excluding the arrowed points. As seen in the figure, each S - N curve can be represented by a straight line up to about 108 stress-cycles and can be expressed by the formula a n n = constant, where a is the stress amplitude and N is number of cycles to failure. The
I/
100
101
IJ
102
i!
103
J l,r
104
105
106
107
II
108
tl
109
N
Figure 3 S-N curves includingstatic bendingstrength for materials of grainsize: (©) 19/xm,(A) 8/xmand (E])5/xm.
straight line for the fatigue data was also similar to that shown in the figure. It seemed that the datum point of the straight line could be taken as the mean static strength. It is considered [5] that the S - N curve, including the static strength, is useful for estimating the fatigue strength of ceramics under rotary bending. The exponent n indicates the fatigue resistance of materials and seems to indicate the exponent in the subcritical crack velocity V against stress intensity factor KI relationship, V = A K ~ , where A is a material constant. The values for the present and previous materials are listed in Table II. As already pointed out in a previous letter [2], the values obtained under rotary bending whose stress state is reversed bending are small in comparison with the literature value; the value of n for sintered A1203 is generally greater than or near to 30 at room temperature. The value of n for sintered A1203 seems to be different because of the applied stress state; it is confirmed that the value of n for sintered Si3N4 is smaller under rotary bending than under static fatigue [9]. From Table II it should be emphasized that the value of n for finer-grained material is higher than that for larger-grained material. This tendency is seen in the results obtained under dynamic fatigue [10]. It is considered that in the same applied stress state the value of n for sintered A1203 is different due to the grain size. It was pointed out by Freiman et al. [11] that the grain size appeared to be the primary factor in determining the crack growth rates in alumina; materials with a large grain size had a greater resistance to crack growth. Similar results are seen in the crack growth data of T A B L E II n for sintered A1203 under rotary bending n
d (/zm)
Reference
23.4 19.7 15.5 19.3 12.7
5 8 19 10 28
Present study Present study Present study Previous study, [1] Previous study, [1-3]
T A B L E I Assumed fatigue limit and static strength
1712
F tl
Grain size, d (~m)
Fatigue limit (kg mm -2)
Static strength (kg ram-Z)
Ratio fatigue limit/static strength
5 8 19
12 10 7
28.5 25.5 23
0.42 0.39 0.3
compact tension sintered A1203 specimens with an alumina content of 99.9% under repeated unidirectional stress [12]; the value of n for material of 19/xm grain size is lower than that for material of 1/xm grain size. It is considered that the difference in n for the present materials is due to the crack growth behaviour from the initial flaw; the mechanism is the subject for a future study. The typical appearance of the fractured specimen after fatigue testing is shown in Fig. 4a and b. The fractured surface is perpendicular to the specimen axis. In many cases the last failure parts of specimens were fractured with branching (Fig. 4b), whereas some specimens were fractured straight (Fig. 4a), i.e. without branching. Sintered Si3N4, the fatigue strength of which was much higher than that of the present material, had a fractured shape similar to an inverted T or Y [4, 6]. The fractured shape seems to be different because of the strength of the material or the applied stress; the present specimens were fractured with clearer branching, as the fracture stress was high. To examine the fatigue fracture features, fractured surfaces after fatigue and static tests were observed by SEM. It was known from macroscopic observation that each overall fractured surface after fatigue testing was very uniform and smooth in comparison with that after static testing. Mirrors were not revealed on the fractured surfaces after the two tests. To observe mirrors it may be necessary to prepare material of much finer grain size, viz. 1-2/xm, as it was pointed out by Kirchner and Gruver [13] that large-grained material does not have mirrors. The fracture seemed tO propagate
radially from a certain part near the specimen surface. Fig. 5a-c shows typical microphotographs near the fracture-initiation part of the fractured surface after fatigue testing. The fracture occurred mainly as a transgranular process and the fracture features were similar to those near the middle of the fractured whole surface. Moreover, microscopic observations for the fractured surface after fatigue test were similar to those after static testing and the fracture features after the two tests could not be distinguished; no characteristic sign of fatigue fracture could be found. Within the limits of the experiment, the observations on finer-grained material were similar to those on larger-grained material, and no remarkable difference could be found; many pores with various shapes and sizes were observed, and the pore size for finer-grained material seemed to be a little smaller than that for larger-grained material. It was not easy to detect a fracture origin clearly, even under microscopic observation. However, it seems to be generally true that the combination of larger or a greater number of pores combined with one of a few larger grains is typically the fracture origin; this is the view of Rice [14], based on his minute observation on A1203 samples [1-3] similar to the present materials.
Acknowledgements I thank Dr O. Kamigaito of Toyota Central Research and Development Laboratories for valuable advice and suggestions. I also thank Kyocera Kagoshima Factory for their support.
Figure 4 Fractured specimens after fatigue testing: (a) without and (b) with branching.
Figure5 Typical micrographs near the fracture initiation part of the fracture surface after fatigue testing: (a) d = 19 ~m, a = 8 k g m m -2 , N = 7.99 x 106; (b) d = 8/zm, a = 11 k g m m -2, N = 1.01 x 107; (c) d = 5/xm, a = 19 k g m m -2, N = 1.38 x 105.
1713
References 1. 2. 3. 4. 5.
6. 7. 8.
9. 10.
H.N. KO, J. Mater. Sci. Lett. 8 (1989) 1438. Idem, ibid. 5 (1986) 464. Idem, ibid. 6 (1987) 801. Idem, ibid. 6 (1987) 175. Idem, in Proceedings of the MRS International Meeting on Advanced Materials, 5, Tokyo, Japan, 1988 (Materials Research Society, Pittsburgh, Pennsylvania, 1989) p. 43. Idem, J. Mater. Sci. Lett. 10 (1991) 545. R.R. MATHESON, Ceram. Age 79 (1963) 54. O. JOHARI and N. N. PARIKH, in "Fracture mechanics of ceramics", Vol. 1 (Plenum Press, New York, 1974) p. 399. H.N. KO, J. Ceram. Soc. Jpn. 99 (1991) 533. A . J . GESING and R. C. BRADT, in "Fracture mechanics
1714
of ceramics", gol. 5 (Plenum Press, New York, 1983) p. 569. 11. S.W. FREIMAN, K. R. McKINNEY and H. L. SMITH, in "Fracture mechanics of ceramics", Vol. 2 (Plenum Press, New York, 1974) p. 659. 12. A. UENO, H. KISHIMOTO, H. KAWAMOTO and S. OHGAWARA, J. Soc. Mater. Sci. Jpn. (1990) 153. 13. H. P. KIRCHNER and R. M. GRUVER, in "Fracture mechanics of ceramics", Vol. 1 (Plenum Press, New York, 1974) p. 309. 14. R.W. RICE, Private communication (1990).
Received 2 January 1992 and accepted 6 July 1992