Materials Transactions, Vol. 45, No. 5 (24) pp. 1714 to 1721 #24 The Japan Institute of Light Metals Tear Toughness Evaluation of a Permanent Mold Cast A356 Aluminum Alloy Using a Small-size Specimen Hong Zhu* 1, Shinji Kumai, Toshikazu Tanaka* 2 and Akikazu Sato Department of Materials Science and Engineering, Tokyo Institute of Technology, Yokohama 226-852, Japan Tear toughness evaluation of a permanent mold cast A356 aluminum alloy was carried out by using two kinds of specimen with different size. One was equivalent to the specimen size designated in ASTM B871. The other one was about 3% as large as that of standard one in volume. Unit energies for tear fracture were obtained from load-displacement curves, and their specimen size, thickness and microstructure dependency were examined. Unit crack initiation energy (UEi) increased with increase in specimen thickness. Meanwhile, unit crack propagation energy (UEp) monotonically decreased in accordance with increase in specimen thickness. In order to make sure if the UEp values reflected the characteristics of local microstructure difference, the small-size tear specimens were collected from various parts in a single cast product. Larger UEp was obtained in the specimen with finer dendrite arm spacing (DAS). These findings suggest that the tear test using a smallsize specimen is useful for toughness evaluation of cast aluminum alloys. (Received September 3, 23; Accepted March 4, 24) Keywords: tear toughness, small-size specimen, unit energy, A356, solidification structure 1. Introduction Aluminum alloy castings have been used as structural materials for automotive engines. With improvement in casting quality their usage spread into nontraditional applications as a viable alternative to wrought or forged aluminum alloy products. In such applications, fracture toughness is of primary concern. Therefore, it is getting very important to evaluate fracture properties of the cast alloys. In general, in metallic materials with high strength and low ductility, toughness is evaluated by plane-strain fracture toughness, K IC. However, many aluminum alloys in practical use have moderate strength and high ductility. In this case, specimen size required for obtaining valid K IC values may reach the order of magnitude of a meter, and so the K IC test on such materials is not practical. From this point of view, J IC test is more suitable for aluminum alloys. However, its experimental procedure is still complicated. Besides such fracture toughness tests based on linearelastic or elastic-plastic fracture mechanics, there is a special demand for another type of toughness test from engineers and researchers who are working to develop aluminum castings with improved toughness. The toughness test that meets their demand should be not only reliable but also simple. It has been known that toughness of aluminum alloys can be estimated from tear resistance. Tear test is standardized in ASTM B871. 1) In the tear test, a sharp-notched plate specimen is subjected to static tensile loading until a crack develops at the root of the notch and travels across the width of the specimen. A load-displacement curve is recorded during the test. Several numerical results can be obtained from the load-displacement curve. Unit crack propagation energy (UEp) is the representative criterion of tear toughness. UEp is computed by dividing the measured energy for crack propagation by the net area of the specimen. * 1 Graduate Student, Tokyo Institute of Technology * 2 Graduate Student, Tokyo Institute of Technology. Present address: Sumitomo Light Metal Industries, Ltd., Nagoya 455-867, Japan Rating of fracture toughness has been made for various kinds of aluminum alloy products so far. 2 6) However, researches on cast aluminum alloy products have been very limited. Han et al. examined tear toughness of permanentmold cast and semi-liquid die-cast A356 aluminum alloys. 7) Recently, Kumai et al. 8) performed tear tests for permanentmold cast and direct-chill cast A356 aluminum alloys, and investigated the effect of solidification structure on both unit crack initiation energy (UEi) and UEp. Fine dendrite arm spacing was revealed to be effective for increase in UEp. They suggested that tear tests provide useful information concerning the effect of solidification structure on toughness, which is available to foundry engineers as a guide for further toughening of aluminum alloy castings. In these previous studies, a standard-size specimen designated in ASTM has been used. However, the cast product of interest in investigation is not always large enough, from which one can collect the standard-size specimen. In particular, it is difficult to obtain the standardsize specimen from a specific part of a commercial cast product. In such a case, reduction of the specimen size will be beneficial. It has been known that, however, numerical results of the tear test depend on not only specimen size but also specimen thickness. 1) Therefore, when we can elucidate the contribution of such specimen geometry on tear toughness and prove that the small-size specimen works properly for the test, the tear test will be available to foundry engineers and researchers as a tool for the development of aluminum alloy castings with improved toughness. In the present study, the effect of specimen thickness and specimen size on tear toughness was investigated using a permanent mold cast A356 aluminum alloy. In addition to that, small-size specimens sampled from various parts of the single cast product were tear-tested and correlation between UEp and solidification structure was discussed.
Tear Toughness Evaluation of a Permanent Mold Cast A356 Aluminum Alloy Using a Small-size Specimen 1715 2. Experimental Procedure 2.1 Materials The material used in the present study is A356 aluminum alloy, which belongs to the most widely used aluminumsilicon alloy system. Two types of permanent mold cast products (cast plates and cast bars) were provided for the test. For their production, special attention was paid to minimize both incorporation of casting defects and spread of microstructure between batches. 2.1.1 Cast plates Cast plates were fabricated with help of the Advanced Materials Research Laboratory, Hitachi Metals Ltd.. A356 alloy ingot was melted at 993 K. After degassing, Al- 1 mass%sr alloy was added to the melt for eutectic Si particle modification. The melt was cast into the permanent mold which was located in a pressure vessel, and cooled under 1 MPa pressure. The size of the plate was 2 2 (mm), as shown in Fig. 1(b). Chemical compositions were 7.1%Si,.34%Mg,.9%Fe,.16%Ti and.7%sr and bal. Al (mass%). The resultant cast plates were HIP (Hot Isostatic Pressing) treated with the applied pressure of MPa over a period of 3.6 ks at 773 K in argon gas atmosphere to reduce casting defects such as shrinkage and porosities. 2.1.2 Cast bars Cast bars were supplied by the research committee of light metallic alloys at the Japan Foundry Engineering Society. The fabrication procedure employed for the casting was as follows. A356 alloy ingots were melted at 973 K and degassed by injecting bubbles of argon gas through a rotating nozzle in order to reduce hydrogen content. After the first degassing, an Al-1 mass%sr alloy was added to the melt for the purpose of eutectic Si modification. The second degassing was performed, and then the melt was poured into a permanent mold which was maintained at 423 K. The present mold is standardized in JIS H522, 9) and has been widely R.25 R.25 5 8 35 (a) 2 5.7 (c) 8 8 8 8 32 6 1 14 14 14 14 6 t=2, 3.5, 5, 7, 9 t=2, 3.5, 5, 7 56 2 (b) 33 19 23 Fig. 1 Morphology of cast products and tear test specimens. (a) standardsize tear test specimen, (b) cast plate product, (c) small-size tear test specimen and (d) cast bar product. (d) (mm) 2 39 used for producing a sound aluminum cast product. The size of the cast bar is shown in Fig. 1(d). Chemical compositions were 7.1%Si,.39%Mg,.12%Fe,.11%Ti,.3%Cu,.1%Mn,.8%Sr, bal. Al (mass%). 2.2 Heat treatment The cast products were homogenized at 88 K for 14.4 ks (4 h) and water-quenched. After being maintained at room temperature for 43.2 ks (12 h), they were artificially aged at 433 K for 21.6 ks (6 h). 2.3 Tear test specimens Two kinds of specimens with different size were machined from the cast product. The larger one as shown in Fig. 1(a) is equivalent to the specimen size designated in ASTM B871 except for its thickness. In the present study, five variations in thickness were prepared; 2, 3.5, 5, 7 and 9 mm. The standardsize specimens were machined from the mid-central part of the cast plate, as shown in Fig. 1(b). The small-size specimen as shown in Fig. 1(c) is about 3% in volume as large as that of standard one with same thickness. Four thickness variations were prepared in this case; 2, 3.5, 5 and 7 mm. Small-size specimens were taken from the mid-central part of the cast bar, as shown in Fig. 1(d). In addition, small-size specimens were also collected from several specific parts of the single casting so that each specimen has different microstructure. The microstructural difference here means a relatively small difference in grain size, dendrite arm spacing and eutectic Si particle size, which are altered by local difference in cooling rates. The detailed explanation about small-size specimens is made in the following section (3.3). 2.4 Tear tests Tear tests were performed using an Instron-type testing machine. The specimen was subjected to tensile loading at a constant crosshead speed of 8:3 1 6 ms 1 at room temperature in air. Changes in load and crosshead displacement was recorded in order to obtain the load-displacement curve. We should mention that use of crosshead displacement is not recommended in general, because the displacement measurement includes all deformation in the test fixture and specimen clevis. The standard recommends use of displacement gages which are mounted on the specimen or the clevis. 1) In the present study, however, we dared to adopt crosshead displacement. The motivation for this decision was that foundry engineers and researchers require a simple and convenient toughness testing method. Load-displacement curves obtained in the present study are considered to be sufficiently useful for direct comparison and relative rating of tear toughness of the A356 cast products, so long as all tests employ exactly the same testing machine and loading system. Before starting the series of present tear toughness test, the authors carried out a number of tests using a displacement gage attached to the clevis. The obtained results will be shown in other reports. Throughout these comparative examinations between clip gage and crosshead displacement, we concluded that the present test is still useful to examine the difference among the specimens, i.e., the differences in
1716 H. Zhu, S. Kumai, T. Tanaka and A. Sato a AC khz,1a specimen at dendrite cell boundaries. Quantitative data of microstructural factors is as follows; grain size: 42 mm, dendrite arm spacing (DAS): 2 mm, eutectic Si particle size: 1.1 mm and aspect ratio of Si particle: 2.3. Tensile properties obtained using rectangular tensile specimens with shoulders (gage section: 16 5 4 (mm)) were as follows; :2 : 21 MPa, UTS: 268 MPa, elongation: 12.9%. Tear test specimens with different thickness were subdigital lock-in amplifier P V load,p potential drop, time,t time,t displacement, D displacement, D memory recorder P, V b m V time,t displacement, D Fig. 2 Schematic illustrations of tear test method and AC potential drop system for crack initiation detection. specimen size, thickness and microstructure. 5 m 2.5 Crack detection using AC potential drop method The crack initiation at the notch root was detected using an AC potential drop method. The measurement system is shown in Fig. 2. The constant alternating current (1 A, khz) was introduced to the specimen. The potential drop change due to the reduction of net cross section, which was caused by crack initiation and propagation, was detected using a digital lock-in amplifier. The crack initiation load was obtained from a pair of potential drop-time and load-time curves. c 2.6 Microstructure observation Optical observation of the microstructure was carried out for the polished cross-section of the heat-treated castings. An image analyzer was employed for quantitative evaluation of dendrite arm spacing (DAS) and size of eutectic Si particles. A polished cross-section was also anodized in a 2% HF solution at a voltage of 25 V and a current density of.2 mamm 2. The anodized sample exhibited a clear grain structure image under polarized light in an optical microscope. 3. Results and Discussion 3.1 The effect of specimen thickness on tear toughness for the standard-size specimen Figures 3(a) (c) show optical micrographs of the standardsize specimen machined from the cast plate. The solidification structure of the polycrystalline A356 alloy consists of primary dendrite branches and eutectic Si particles located Fig. 3 Microstructural features of the mid-central section of the cast plate product. (a) grain structure, (b) dendrite structure and (c) eutectic Si particles. 1 m
Tear Toughness Evaluation of a Permanent Mold Cast A356 Aluminum Alloy Using a Small-size Specimen 1717 Load, P/kN 25 2 15 1 5 2mm 3.5mm 5mm.5 1 1.5 2 2.5 3 Crosshead displacement, D/1 3 m Fig. 4 Load-displacement curves in tear tests for standard-size specimens with different thickness. Each arrow on the curve indicates the crack initiation point detected by the AC potential drop method. jected to tensile loading until a crack initiated at the notch root and then traveled across the width of the specimen. Figure 4 shows load-displacement curves of the standard-size specimens. Both maximum load and the area under the loaddisplacement curve increased with increasing specimen thickness. Instantaneous load drop, such as pop-in, was observed for the specimens with 5 mm thick and more. The pop-in load shifted toward the maximum load (P max ) with increasing thickness. The load drop at pop-in is considered to result from a sudden local crack generation and decrease in cross section area at a mid-thickness near the notch root. This relates to the degree of stress tri-axiality and this is larger in thicker specimens. The AC potential drop can detect the crack initiation much earlier than the occurrence of change in load-displacement curve. Small arrows on the load-displacement curves indicate 7mm 9mm the load where the potential drop change took place. Such a crack initiation load is called Pi hereafter. The Pi was also shifted to P max with increasing specimen thickness. The popin stress was located between Pi and P max in thick specimens. After pop-in, stable crack growth progressed toward both free surfaces while load increased up to P max. Then an unstable-crack started growing until it traveled across the width of the specimen. Figures 5(a) and (b) show schematic load-displacement curves in tear tests. The vertical line through the maximum load divided the load-displacement curve into two segments; crack initiation and crack propagation. The area under the first segment of the curve is a measure of the energy necessary for the crack to initiate. The area under the second segment represents the energy necessary for the crack to propagate across the specimen. ASTM B871 1) designated three energetic parameters for evaluating tear toughness; Unit crack initiation energy (UEi), Unit crack propagation energy (UEp), and Unit total energy (UEt). Unit energy is computed by dividing the measured energy by the net area of the specimen according to the regulation as shown in Fig. 5(a). In the present study, the load corresponding to the crack initiation (Pi) was detected by the potential drop change. Unit crack initiation and propagation energies were also defined on the basis of Pi, as shown in Fig. 5(b). In order to distinguish these unit energies from those in Fig. 5(a), they are called True unit crack initiation energy, UEit and True unit crack propagation energy, UEpt, respectively. It should be noted that the crack size detected at Pi was very small. Unit energies are plotted as a function of specimen b t b t Load, P Maximum load, Pmax Load, P Crack initiation load, Pi Ep Ept Ei Eit Crosshead displacement, D Crosshead displacement, D Unit initiation energy: UEi=Ei/A Unit propagation energy: UEp=Ep/A Unit total energy: UEt=UEi+UEp True Unit initiation energy: UEit=Eit/A True Unit propagation energy: UEpt=Ept/A Ture unit total energy: UEtt=UEit+UEpt A: net area,a=bt b: width at root notch t: thickness Fig. 5 Schematic load-displacement curves in tear tests and definition of unit energies. (a) unit energies. (b) true unit energies.
1718 H. Zhu, S. Kumai, T. Tanaka and A. Sato 12 UEt UEp UEi Unit energy, UE/kNm 1 8 6 4 2 2 4 6 8 1 Thickness, t/mm (a) 12 UEtt UEpt UEit Ture unit energy, UEt/kNm 1 8 6 4 2 2 4 6 8 1 Thickness, t/mm (b) Fig. 6 Effects of specimen thickness on unit energies for standard-size specimens. (a) Unit energies; UEi, UEp and UEt. (b) True unit energies; UEit, UEpt and UEtt. thickness in Figs. 6(a) and (b). There is no significant change in UEt at any specimen thickness. In contrast, UEi increases as increasing specimen thickness, while UEp decreases as increasing specimen thickness. Thus, it was found that specimen thickness affects quantitative balance between UEi and UEp significantly. Similar specimen-thickness dependency was observed when we compared true unit energies, UEit, UEpt and UEtt, defined by Fig. 5(b). 3.2 The effect of specimen size and specimen thickness on tear toughness Figures 7(a) (c) show optical micrographs of the smallsize specimen, which was taken from the mid-central part of the cast bar product. Very few casting defect was detected. Average grain size, DAS, eutectic Si particle size and its aspect ratio were 476 mm, 26.5 mm, 1.8 mm, 1.5, respectively. Tensile properties obtained using round bar tensile specimens (gage section: 6.25 mm 32 mm) were as follows: :2 : 23 MPa, UTS: 31 MPa, elongation: 13%. Figure 8 shows load-displacement curves of small-size specimens. Thickness dependency was equivalent to that of standard-size specimens. Unit energies are plotted as a function of specimen thickness in Figs. 9(a) and (b). UEi increases as increasing specimen thickness. Meanwhile, UEp Fig. 7 Microstructural features of the mid-central section of the cast bar product. (a) grain structure, (b) dendrite structure and (c) eutectic Si particles. decreases as increasing specimen thickness. Equivalent specimen-thickness dependency was also observed in the small-size specimen when we compared true unit energies, UEit, UEpt and UEtt, defined by Fig. 5(b). Some previous studies treated the effect of specimen thickness on tear toughness. Komura and Taki investigated tear toughness of several medium strength aluminum wrought alloys. 4,5) They pointed out that the toughness value
Tear Toughness Evaluation of a Permanent Mold Cast A356 Aluminum Alloy Using a Small-size Specimen 1719 Load, P/kN 12 1 8 6 4 2 2mm 3.5mm 5mm.5 1 1.5 Crosshead displacement, D/1 3 m 7mm Fig. 8 Load-displacement curves in tear tests for small-size specimens with different thickness. Each arrow on the curve indicates the crack initiation point detected by the AC potential drop method. Unit energy, UE/kNm 1 Ture unit energy, UEt/kNm 1 12 8 6 4 2 12 8 6 4 2 UEt UEp UEi 2 4 6 8 Thickness, t/mm UEtt UEpt UEit 2 4 6 8 Thickness, t/mm Fig. 9 Effects of specimen thickness on unit energies for small-size specimens. (a) Unit energies; UEi, UEp and UEt. (b) True unit energies; UEit, UEpt and UEtt. was hardly influenced by the specimen thickness if it was less than 1 mm. Kobayashi et al. performed tear tests on a 217- T4 alloy using a standard-size specimen with different thickness (2.5, 5 and 12.5 mm). 6) They detected crack initiation load by using DC potential method. The crack initiation load was found to shift toward the maximum load as increasing specimen thickness. They also measured UEit and UEpt values on the basis of the crack initiation load and examined the effect of specimen thickness. They reported that both UEit and UEpt changed with specimen thickness. However, no explanation was made about it. In contrast to these studies, the experimental results in the (a) (b) present study exhibited a clear specimen thickness dependence of UEi and UEp. In particular, UEp decreases monotonically with increase in thickness. This result is consistent with Kaufman s report. 3) He described that there is a general trend for UEp to decrease with increasing specimen thickness. UEi is a measure of the energy necessary for the crack to initiation. It depends on morphology of the notch and the quality of the surface finish of the notch tip. In contrast to that, UEp is a measure of the energy necessary for the crack to propagate across the specimen. So, UEp is little affected by the geometrical and finishing condition of the notch tip. Therefore, UEt reflects the uncertainties of the UEi. Consequently, we give attention to UEp in the present study. 3.3 The effect of solidification structure on UEp Grain size, DAS, and size and distribution of eutectic Si particles are possible microstructural factors which affect the tear toughness for a sound A356 cast alloy including few casting defects. The effect of solidification structure on tear toughness has been examined by Han et al. using the standard-size tear specimen. 7) They prepared A356 cast materials by using different fabrication routes; sand mold cast, permanent mold cast, and semi-liquid die-cast. Microstructural difference among them was so large that they obtained clear difference in UEp values. Recently, Kumai et al. 8) examined the effect of solidification structure on tear toughness for several permanent mold cast (PM) and direct-chill cast (DC) products of the A356 aluminum alloy. Refinement of DAS and grain size increased both UEi and UEp. The significant increase of UEp was obtained in DC products with fine DAS. It was suggested that quantitative balance between UEi and UEp depend on the features of crack propagation, and the increased UEp in DC was due to the introduced slanted crack path. It is well known that the solidification structure is different in local even in the single cast product. The outer region of the cast generally exhibits finer DAS compared to that of the middle part. The small-size tear test specimen is so small that we can obtain the specimen including such a local microstructure selectively. Therefore, several kinds of specimens with different thickness were collected from the single cast bar product. Figure 1 shows schematically the location in the cast bar product, from which the tear specimen were collected. Microstructures of the cast bar product are shown in Figs. 11(a) (f). Alphabetic letters in the picture correspond to the local microstructure at which the specimen was collected. The specimen A(c) has a notch at the upper side of the cast. Crack propagation takes place downward through the ligament. Microstructural features through which the crack propagates correspond to (e) and (f) in Figs. 11. The specimen B(c) has a notch at the bottom side of the cast so that the crack propagates upward ((e) to (d) in Figs. 11). The specimens A(s) and B(s) are collected from the outer region of the cast with finer DAS. The crack in A(s) propagates through the ligament upward ((b) to (c) in Figs. 11). The crack in B(s) grows downward ((b) to (a) in Figs. 11). Tear tests were performed for these specimens. The obtained UEp values are shown in Fig. 12. Let us compare
172 H. Zhu, S. Kumai, T. Tanaka and A. Sato Fig.11 (b) (c) A(s) B(c) the UEp between A and B at first. As for A, the crack propagated from the notch toward the region where the DAS was getting finer. On the other hand, for B, the crack propagated mainly in the coarse microstructure region. Figure 12 shows that UEp values of A are larger than those Fig.11(b) (a) B(s) fine coarse coarse fine fine A(c) Fig.11 (e) Fig.11 (e) Fig. 1 Notation of the collected small-size specimens from various parts of the cast bar product. Microstructure of the shaded area in each specimen corresponds to the picture shown in Figs. 11. (f) (d) Unit crack propagation energy, UEp/kNm 1 35 3 25 2 15 1 5 2mm 3.5mm 5mm 7mm A(C) A(S) B(C) B(S) Specimen type Fig. 12 Effects of specimen thickness and local microstructure on UEp. a d m m b e m m c f m m Fig. 11 Local microstructure in the cast bar product. (a) Top-surface, (b) Middle-surface, (c) Bottom-surface, (d) Top-central, (e) Middlecentral, (f) Bottom-central.
Tear Toughness Evaluation of a Permanent Mold Cast A356 Aluminum Alloy Using a Small-size Specimen 1721 Ultimate tensile strenth, UTS/MPa Proof stess,.2/mpa 35 3 25 2 15 5 Fig. 13 UEp (t=2mm) UEp (t=3.5mm) UEp (t=5mm) UEp (t=7mm) 15 2 25 3 35 4 45 5 Dendrite arm spacing, DAS/ m of B. Comparison between (c) and (s) was also made. A(s) and B(s) have larger UEp compared to A(c) and B(c). The grain morphology is different between the Fig. 11(a) and Fig. 11(c). The former consists of equal-axis grains and the latter includes columnar grains. Quantitative evaluation of the grain size is difficult because of such morphology difference in grain structure. Therefore, only the effect of DAS on UEp was treated in the present study. The obtained UEp was plotted against DAS in Fig. 13 with proof stress, UTS and elongation of the specimen. Tensile properties were obtained for the casting fabricated under the same casting condition. These results are seemed to be quite reasonable considering from the general relationship between material s toughness and fineness of the DAS in cast aluminum products. The tear specimens with finer DAS showed larger UEp. The experimental results shown in Fig. 12 suggest that the UEp is a useful measure for evaluating tear toughness of the cast material. Furthermore, it is also demonstrated that the UEp is sensitive to relatively small microstructural difference in the single cast. 4. Conclusions Tear toughness evaluation was performed on permanentmold cast A356 alloys. Unit energies for crack initiation and propagation were examined and their specimen size, specimen thickness, and microstructure dependency were discussed. UTS σ.2 elongation 35 3 25 2 15 1 5 Unit propagation energy, UEp/kNm 1 elongation, /% Effects of DAS on the tensile properties and tear toughness. Comparable specimen thickness dependency was obtained for unit energies in standard-size and small-size specimens. Unit crack initiation energy (UEi) increased with increase in specimen thickness. Meanwhile, unit crack propagation energy (UEp) monotonically decreased in accordance with increase in specimen thickness. Such a simple specimen thickness dependency is desirable when one use the tear test as a practical toughness evaluation method for cast aluminum alloys. Small-size tear test specimens were collected from the various parts in the single cast product. UEp values reflected the characteristics of local microstructure. The tear specimen with finer DAS showed larger UEp. These findings obtained in the present study suggest that the tear test using a smallsize specimen is useful for toughness evaluation of the cast aluminum alloys even though it is still one of the comparative tests. Acknowledgements The present authors wish to express their appreciation to the research committee of light metallic alloys at the Japan Foundry Engineering Society for supplying cast products and offering useful discussion. The authors also acknowledge Hitachi Metals Ltd. for producing cast products and The Light Metal Education Foundation, Inc. for providing partial financial support for this work. REFERENCES 1) ASTM standard, Designation: B871-96, Standard Test Method for Tear Testing of Aluminum alloy Products, 62 68. 2) J. G. Kaufman and A. H. Knoll: Materials Research & Standards, April (1964) 151 155. 3) J. G. Kaufman: Fracture Resistance of Aluminum Alloys, The Aluminum Association, ASM International, ISBN: -8717-732-2 (21) 38 74. 4) S. Komura and H. Taki: J. JILM 24, 9 (1974) 399 45. 5) S. Komura and H. Taki: J. JILM 24, 9 (1974) 46 41. 6) T. Kobayashi, M. Niinomi and K. Ikeda: J. JILM, 38, 1 (1988) 9 15. 7) S. W. Han, S. W. Kim and S. Kumai: Fatigue Fract. Engng Mater. Struct. 27, 1 (24) 9 17. 8) S. Kumai, T. Tanaka, H. Zhu and A. Sato: Mater. Trans. 45 (24) 176 1713. 9) JIS Handbook, Non-ferrous metals, (Japanese Standard Association, 21) 957 965.