© 2012 ISIJ 510

ISIJ International, Vol. 52 (2012), No. 3, pp. 510–515

Variant Selection of Low Carbon High Alloy Steel in an Austenite Grain during Martensite Transformation

Shuoyuan ZHANG,1) Shigekazu MORITO2) and Yu-ichi KOMIZO3)

1) Japan Atomic Energy Agency, 1-1-1 Kohto, Sayo-cho, Sayo-gun, Hyogo, 679-5148 Japan. E-mail: zhang.shuoyuan@jaea.go.jp2) Department of Materials Science, Shimane University, 1060 Nishikawazu-cho, Matsue, Shimane, 690-8504 Japan. E-mail:mosh@riko.shimane-u.ac.jp 3) Joining and Welding Research Institute, Osaka University, 11-1 Mihogaoka, Ibaraki,Osaka, 567-0047 Japan. E-mail: komizo@jwri.osaka-u.ac.jp

(Received on September 6, 2011; accepted on October 21, 2011; originally published in Tetsu-to-Hagané,Vol. 97, 2011, No. 7, pp. 399–405)

In this study, the development of a lath martensite structure in low carbon high alloy steel wasobserved in situ using high-temperature laser scanning confocal microscopy. The crystallography of themartensite structure was analyzed using electron backscatter diffraction patterns. It was observed thatmartensite transformation starts from the prior austenite grain boundary. Then, the variant (Σ1) of anotherpacket belonging to the same bain correspondence was observed in the early stage of martensite trans-formation. Another block in the same packet was observed in the next stage of martensite transformation.Finally, transformation occurred among the neighbors of the transformed martensite block.

KEY WORDS: variant selection; martensite transformation; in situ observation; EBSD.

1. Introduction

In recent years, there have been demands for a decreasein the use of alloying elements and an improvement in themechanical properties of steels. Martensite is a microstruc-ture that is indispensable for the strengthening of steels.Thus, it is necessary to utilize a martensite microstructure tomeet these demands and to clarify the transformationbehavior of martensite in detail.

Martensite transformation is a diffusionless process and iscaused by an atomic correspondence between parent andproduct lattices. The strain due to martensite transformationis relaxed when an austenite is deformed plastically (Plasticaccommodation, PA) or elastically (elastic accommodation,EA), or when different variants are formed (self accommo-dation, SA). It has been reported that the microstructures ofshape memory alloys,1) non-ferrous alloys2) and thin-platemartensite3) are formed during the combination of SA vari-ants. These combinations appear well in a thermoelasticnon-ferrous martensite. It is believed that they can relax thetransformation stress predicted by the phenomenologicaltheory of martensite crystallography (PTMC). However, fer-rous lath martensite is not formed by such combinations. Itis observed that lath martensite contains some groups withalmost the same habit plane (packet). The packet consists ofsix types of variants. PTMC analysis reveals that the com-bination of variants in a sub-block in the block is not advan-tageous for SA.4) Although various techniques have beenused to analyze this structure, the reason as to why a lathmartensite microstructure is different from other martensitestructures is still unclear. The martensite transformationstrain should be relaxed by plastic deformation of the matrix

and/or martensite resulting plastic accommodation. More-over, the formation of variants plays an important role in theformation of a martensite structure.

In recent years, the electron backscatter diffraction(EBSD) technique has been adopted for analyzing the phasetransformation of microstructures.5) Subsequently, the crys-tallography of lath martensite structures was investigatedusing this technique.4,6–9) The crystal orientation relationshipof lath martensite,4,10,11) plastic accommodation in the parentaustenite9) and the variant selection in a grain boundary12)

have been discussed. The martensite structure after heattreatment has been observed in previous studies.13,14) How-ever, it is necessary to examine microstructure change insitu to understand microstructure evolution at elevated tem-peratures. The in situ SEM/EBSD technique has recentlybecome popular and has been used for in-situ observationsof the solid-phase transformation process of a metal materi-al.13,14) However, it is difficult to observe the high-speedmartensite transformation phenomenon during a continuousthermal cycle because of the present low time resolution ofthis technique. To overcome this problem, we developed ahigh-speed photography technique using high-temperaturelaser scanning confocal microscopy (LSCM) and observedthe solid-phase transformation in situ during a continuousthermal cycle.15–18)

In this study, the martensite transformation of low carbonhigh alloy steel was observed in situ by the LSCM techniqueduring a continuous cooling cycle. Then, the crystal orien-tation in the same areas was measured by EBSD. The trans-formation model of lath martensite is discussed using theobtained experimental results.

ISIJ International, Vol. 52 (2012), No. 3

511 © 2012 ISIJ

2. Experimental Procedure

In this study, martensite transformation was observed insitu during a continuous cooling cycle by high-temperatureLSCM. Figure 1 shows a schematic diagram of the setup ofa high-temperature LSCM system. The system consists ofan infrared image furnace and a laser scanning confocalmicroscope. Using a confocal system, only light incidentfrom the focal plane is permitted to reach the photon detec-tor. The image was recorded at a rate of 30 frames/s. Theinfrared light focus in the furnace covered a volume of diam-eter 10 mm and height 10 mm. The specimen in the furnacewas positioned in this volume. A high-purity alumina cruciblewas used, and the sample folder was placed on platinum standwith a B-type thermocouple installed at the back. The sampleswere machined to 5 mm in diameter and 1 mm in height, andthe observed plane was mirror polished. The specimens werecooled to room temperature after heating to 1 485°C at a rateof 33°C/s as shown in Fig. 2. The martensite transformationduring cooling was observed in situ observed. The rate ofcooling from 100°C to room temperature was –0.45°C/s.

The chemical compositions of the samples used in thisstudy are shown in Table 1. Low-carbon-13Cr%-9Ni% steelcontains a large amount of the retained austenite phase evenat room temperature because the start temperature for mar-tensite transformation is low.

After the LSCM observation, the microstructure at room

temperature was characterized by SEM/EBSD. The acceler-ating voltage of SEM (JEOL JSM-6400) was 20 kV and theelectron beam diameter was about 0.2 μ m. The step size was0.5 μm in these observations.

After the formation of a lath martensite block wasobserved in situ by high-temperature LSCM, the crystal ori-entation data corresponding to the microstructure of lathmartensite was analyzed. It is well known that there are 24variants in the case of a K-S orientation relationship(OR).4,7) The crystal orientation relationship based on vari-ant 1 (V1) is shown in Table 2. Figure 3 shows the coinci-dence grain boundary [Σ1, Σ3 coincidence site lattice(CSL)] for the 24 variants. Laths in a given packet have thesame plane parallel relationship between close-packed (CP)planes. Thus, there are four crystallographically differentpackets in a given austenite grain. In one packet, there aresix variants with different direction parallel relationships onthe same conjugate parallel CP plane. When V1 is used asa reference, it is considered that the misorientation betweenV1 and V4 (V8, V11, V13) is low angle (Σ1) and ORbetween V1 and V2 is a twin relationship (Σ3). Furthermore,OR between V7 and V8 is a twin relationship. On the otherhand, V13 and V16 become Σ1 CSL. Moreover, it wasshown that the variant connected by Σ1 CSL belonged to thesame Bain correspondence area (for example, V1-V4-V13-V16-V24-V21-V11-V8).

Given the three <001> directions, there are three crystal-lographic variants of the Bain strain. This means that a pack-

Fig. 1. Schematic diagram of the setup for in situ observations.

Table 1. Chemical compositions of the materials used (mass%).

C Si Mn P S Ni Cr Mo

0.018 0.29 0.8 0.003 0.002 9.05 12.95 0.49

Fig. 2. Thermal cycles applied to the specimens.

Table 2. 24 variants in the K-S relationship.

Variant Parallel plane Parallel direction Rotation from variant 1 CSL

V1

(111)γ //(011)α

[-1,0,1]γ //[-1,-1,1]α None None

V2 [-1,0,1]γ //[-1,1,-1]α [0.58,–0.58,0.58]/60.0° Σ3

V3 [0,1,-1]γ //[-1,-1,1]α [0.00,–0.71,–0.71]/60.0° –

V4 [0,1,-1]γ //[-1,1,-1]α [0.00, 0.71,0.71]/10.5° Σ1

V5 [1,-1,0]γ //[-1,-1,1]α [0.00,0.71,0.71]/60° –

V6 [1,-1,0]γ //[-1,1,-1]α [0.00,–0.71,–0.71]/49.5° Σ11

V7

(1,-1,1)γ //(011)α

[1,0,-1]γ //[-1,-1,1]α [–0.58,–0.58,0.58]/49.5° Σ19b

V8 [1,0,-1]γ //[-1,1,-1]α [0.58,–0.58,0.58]/10.5° Σ1

V9 [-1,-1,0]γ //[-1,-1,1]α [–0.19,0.77,0.61]/50.5° –

V10 [-1,-1,0]γ //[-1,1,-1]α [–0.49,–0.46,0.74]/50.5° –

V11 [011]γ //[-1,-1,1]α [0.35,–0.93,–0.07]/14.9° Σ1

V12 [011]γ //[-1,1,-1]α [0.36,–0.71,0.60]/57.2° –

V13

(-1,1,1)γ //(011)α

[0,-1,-1]γ //[-1,-1,1]α [0.93,0.35,0.07]/14.9° Σ1

V14 [0,-1,-1]γ //[-1,1,-1]α [0.74,0.46,–0.49]/50.5° –

V15 [-1,0,-1]γ //[-1,-1,1]α [–0.25,–0.63,–0.74]/57.2° –

V16 [-1,0,-1]γ //[-1,1,-1]α [0.66,0.66,0.36]/20.6° –

V17 [110]γ //[-1,-1,1]α [–0.66,0.36,–0.66]/51.7° –

V18 [110]γ //[-1,1,-1]α [–0.30,–0.63,–0.72]/47.1° –

V19

(1,1,-1)γ //(011)α

[-1,1,0]γ //[-1,-1,1]α [–0.61,0.19,–0.77]/50.5° –

V20 [-1,1,0]γ //[-1,1,-1]α [–0.36,–0.60,–0.71]/57.2° –

V21 [0,-1,-1]γ //[-1,-1,1]α [0.96,0.00,–0.30]/20.6° –

V22 [0,-1,-1]γ //[-1,1,-1]α [–0.72,0.30,–0.63]/47.1° –

V23 [101]γ //[-1,-1,1]α [–0.74,–0.25,0.63]/57.2° –

V24 [101]γ //[-1,1,-1]α [0.91,–0.41,0.00]/21.1° –

© 2012 ISIJ 512

ISIJ International, Vol. 52 (2012), No. 3

et consists of three types of parallel blocks with differentorientations. In this study, a Bain map classified by areashaving the same orientation was used as the block map.20)

As a packet is a group of parallel laths with the same CPplane, the packet in the martensite structure was identifiedusing the CP map. Moreover, the variant in the martensitestructure was analyzed by the pole figure method.

3. Results and Discussion

3.1. In Situ Observation of Martensite TransformationThe As, Af, and Ms temperatures were measured during

the continuous thermal cycles by high-temperature LSCMand were found to be 719°C, 801°C and 131°C, respective-ly. However, Mf was not determined because it was lowerthan the room temperature in this experiment.

Figure 4 shows the in situ observations of the martensitetransformation behavior of an austenite grain during the

cooling cycle by high-temperature LSCM. As shown in Fig.4(a), Grain 1 and Grain 2 were observed as austenite grainsat 135°C. The surface relief of the martensite structure isshown in Fig. 4(b). At the lower part of Grain 1 in Fig. 4(b),the remained austenite can be clearly observed among theblocks that had transformed earlier (shown as white dottedlines). With decreasing temperature, the formation of theblocks and packets was clearly observed and groups of lathswere formed in packets, as shown by the arrows in Fig. 4(c).Finally, the remained austenite in each packet was graduallytransformed into martensite laths, as shown in Fig. 4(d).

3.2. Crystal Orientation Relationship between Marten-site and Retained Austenite

After cooling to room temperature, the crystal orientationof the martensite in the area in Fig. 4 was measured byEBSD. Figure 5 shows the crystal orientation in Grain 1.Figures 5(a) and 5(b) are orientation maps constructed fromthe inverse pole figure of the martensite and austenite,respectively. The orientation relationship between theretained austenite and martensite was obtained by compar-ing both phases at room temperature because Grain 1 con-

Fig. 3. Illustration of the 24 crystallographic variants (V1-V24) evolving in the {111} austenite plane in the K-S orienta-tion relationship. The triangles and rectangles indicate the {111} austenite (FCC) plane and the (011) martensite(BCC) plane, respectively.

Fig. 4. Overview of results of in situ observation martensite trans-formation by high-temperature laser scanning confocalmicroscopy.

Fig. 5. Corresponding crystal-orientation map measured by anEBSD analysis. (a) and (b) Inverse pole figure color map ofmartensite and austenite, respectively. Black lines showboundaries having a misorientation angle greater than 15°.

ISIJ International, Vol. 52 (2012), No. 3

513 © 2012 ISIJ

tained bulk retained austenite.From the crystal orientation of retained austenite, OR

between the parent (austenite) and transformed phase (mar-tensite) is summarized in Fig. 6. The results show that theCP plane [(111) γ and (011) α ’] and CP direction ([-101]γand [-1-11]α ’) were almost parallel with a decentralizationof 3° or less. It can be concluded that the martensiteobserved here had a K-S relationship21) with the austenite.

3.3. Microstructure of Martensite on the Sample Sur-face and in Bulk

Martensite transformation on the surface of specimenswas observed by high-temperature LSCM. The martensitestructure was influenced by the restraint exerted by the sam-ple surface. The effect of the surface should be examinedbecause the packet parallel to the CP plane can be easilyobserved.22) It is known that, the ideal six variants in a pack-et of lath martensite can relax the shear strain but cannotaffect the change in volume.

As there is no restraint in the normal direction of the sur-face, a packet that can accommodate cubical expansion inthis direction might be produced. Thus, the martensitemicrostructure at the surface was compared with that in thesample.

The Bain map and the CP map for the martensite structurewith prior austenite and Grain 1 and Grain 2 observed onboth unpolished and polished surfaces of the samples areshown in Fig. 7. The black and white lines indicate high-angle misorientation and the Σ3 twin-grain boundary,respectively. In Fig. 7(b), the red (CP1), yellow (CP2) andgreen areas(CP4) are adjacent packets, as indicated by thepink circles. It was assumed that the boundaries of thesepackets were connected by the same Bain group [Fig. 7(a)].Furthermore, numerous Σ3 twin-grain boundaries wereincluded in the same packet.

The results of the high-temperature LSCM revealed thatthe block groups [arrows in Fig. 4(c)] that considerablytransformed in Grain 1 belong to the same Bain correspon-dence area [Fig. 7(a)]. Same result was obtained in Grain 2as shown in Fig. 7(c).

To confirm the surface effect during martensite transfor-mation, the crystal orientation of the martensite structure inthe sample that was polished to half its original height wasexamined [Figs. 7(e) and 7(f)). The boundary between thetwo packets, CP3 (blue) and CP4 (green) was connected bythe same Bain group B2. The result was similar to that ofthe surface observation. Furthermore, Fig. 7(e) indicated

that adjacent packets were connected by variants belongingto the same Bain group.

The observation of the martensite microstructure in thesample revealed that adjacent packets were connected by thesame variants belonging to the same Bain group, which issimilar to the results obtained for the surface. In the nextsection, the variant selection observed in a parent austenitegrain (Grain 1) is discussed based on the results of variantanalysis of the martensite microstructure.

3.4. Crystal Orientation of the Martensite Microstruc-ture at Room Temperature

Variants of the martensite microstructure in Grain 1 wereanalyzed in detail by the pole figure method, as shown inFig. 8. The misorientations of the variants with a twin rela-tionship (Σ3), such as V1-V2, V7-V8, V15-V16, and V23-V24 were observed in Grain 1.

Next, the time series of the variant formation was ana-lyzed to discuss the formation of packets. Figure 9 showsthe in situ observation results of the martensite transforma-tion in Grain 1 [the same area in Fig. 4(c)]. The variant anal-ysis clearly indicated that the block marked with the whitedotted line in Fig. 9(g) was V16 [Fig. 9(b)]. After V16 wasproduced, V15-which has a twin relationship with V16-wasformed from the remained austenite [Figs. 9(c) and 9(d)].

V8 grew from the grain boundaries and V7 was trans-formed in an adjacent area. The orientation between V7 andV8 is also a twin relationship. Moreover, V1 was observedas shown in Fig. 9(b). Figure 10 shows the {111} pole fig-ure of the martensite plates. It was observed that the threevariants V1, V8 and V7 intersect at the (-1, -1, 1) pole point.In other words, V1 and V7 were rotated by 10.5° (V1-V8)and 60° (V8-V7) from V8, respectively, at the centre of theCP direction [-1, -1, 1]. It has been reported that V1-V8 hasa Σ1 relationship and small misorientations in the bainitetransformed from the prior austenite grain boundary.24,25)

Moreover, V1, V7 and V8 have the same relationship of CPplane as austenite. This means that laths with these threetypes of variants can grow in the same direction,24) whichresults in the accommodation of transformation stress andstrain.

3.5. Variant Selection Mechanism of Martensite Trans-formation

The strain due to lath martensite transformation is relaxedby the PA or SA process. When the carbon content isincreased, the martensite transformation start temperature isdecreased,19) and PA does not occur readily in the remainedaustenite. Thus, SA with the formation of different variantsis expected in this case.23) On the other hand, the accommo-dation of transformation strain by combining variants withsmall misorientations that belong to different packets hasbeen reported in bainite laths with the same Bain correspon-dence.24,25) In this study, many combinations of variants withthe same Bain correspondence were observed. This suggest-ed that variants belonging to the same Bain group weretransformed by a similar accommodation mechanism.26,27)

Next, a martensite transformation model with respect tothe time series was proposed. It was observed that marten-site transformation starts from the prior austenite grainboundary and proceeds along the growth direction of the

Fig. 6. Distribution of close-packed directions and close-packedplanes of retained austenite and martensite, respectively: (a)and (b) near the K-S side. Corresponding to 3° betweenmajor grid lines.

© 2012 ISIJ 514

ISIJ International, Vol. 52 (2012), No. 3

block initially formed in the prior austenite grain. Then thevariant (Σ1) of another packet belonging to the same Baincorrespondence was observed in the early stage of marten-site transformation. Another block in the same packet wasobserved in the next stage of martensite transformation.Finally, transformation occurred among the neighbors of thetransformed martensite block.

In previous studies,28–30) we analyzed the crystallographyof martansite and austenite structures by X-ray diffractionusing Synchrotron radiation. The austenite phase accommo-

Fig. 7. Typical orientation imaging maps of Grain 1 in the unpol-ished sample [(a), (b)], Grain 2 in the unpolished sample[(c),(d)] and 500-μm polished sample [(e), (f)]. (a), (c) and (e) areBain maps and (b), (d), and (f) are close-packed maps. Theblack and white lines show boundaries having misorientationangles greater than 15° and Σ3 CSL boundaries, respectively.

Fig. 8. Typical orientation imaging maps of Grain 1. (a) Variantcolor map with a color key according to local orientation (onecolor for each of the three Euler angles) and (b) {001} polefigure showing orientation of the martensite corresponding tothe image quality (IQ) map. The black, white and green linesshow boundaries having misorientation angles greater than15° and CSL boundaries Σ3 and Σ1, respectively.

Fig. 9. (a)–(f) Overview of the results of in situ observations ofmartensite transformation in Grain 1 by high-temperaturelaser scanning confocal microscopy. (g), (h) Typical orien-tation imaging maps of Grain 1. The black lines show theboundaries having misorientation angles greater than 15°.(h) IQ map.

Fig. 10. {111} pole figure showing orientations of martensite plates.

Fig. 11. Schematic illustration describing martensite transforma-tion in prior austenite with a color key according to theBain zone.

ISIJ International, Vol. 52 (2012), No. 3

515 © 2012 ISIJ

dated by a rotation of less than 3° of the crystallite immedi-ately after martensite transformation. It was believed thatthe plastic accommodation mechanism of austenite wasdominant in the first stage of martensite transformationbecause a large amount of austenite remained.

A model of lath martensite transformation is proposedfrom the results as shown in Fig. 11. Initially, in one packet,a block that has a preferred variant from the energy view-point is transformed from the austenite grain boundary.Then, the block of another packet appears from another aus-tenite grain boundary. The two blocks collide as shown inFig. 11(b). It can be considered that an initial block appearsfrom a packet boundary in competition with a formationfrom a grain boundary. Next a different packet with thesame Bain correspondence appears as shown in Fig. 11(c).Another variant of another packet that belongs to the sameBain correspondence is formed from the deformed austeniteto immediately relax the transformation strain [Figs. 11(c)and 4(b), 4(c) and 9(b)]. Moreover, because this bock has apreferred variant from the energy viewpoint, it is suggestedthat the other block is formed apart from the first block torelax the transformation stress [Fig. 11(c)]. In the next stage,the self-accommodation variant (another block in the samepacket) is dominant because the amount of austenite isdecreased. Therefore, the martensite phase is formed aroundthe initial block as a self-accommodation variant [Figs.11(d), 11(e) and 9(c), 9(d)]. Finally, transformation occursamong the neighbors of the transformed martensite block.

4. Conclusions

Martensite transformation in low carbon high alloy steelwas observed in situ by high-temperature LSCM duringcontinuous cooling, and the crystallography of the marten-site structure was analyzed by EBSD. The main conclusionsare as follows:

(1) Initially, a block was transformed from the prioraustenite grain boundary. Then, a different packet with thesame Bain correspondence appeared in the early stage ofmartensite transformation. It can be assumed that anothervariant of another packet that belongs to the same Bain cor-respondence is formed from the deformed austenite toimmediately relax the transformation strain. Moreover,because this bock has a preferred variant from the energyviewpoint, it is suggested that the other block is formedapart from the first block to relax the transformation stress.

(2) Another block in the same packet was observed in

the next stage of martensite transformation. This suggeststhat the martensite phase was formed around the initialblock as a self-accommodation variant. Finally, transforma-tion occurred among the neighbors of the transformed mar-tensite block.

(3) A model for lath martensite transformation was pro-posed on the basis of the accommodation mechanism of thetransformation strain.

REFERENCES

1) T. A. Schroeder and C. M. Wayman: Acta Metall., 25 (1977).2) T. Saburi and C. M. Wayman: Acta Metall., 27 (1979), 979.3) H. Okamoto, M. Oka and I. Tamura: Trans. Jpn. Inst. Met., 19

(1978), 674.4) S. Morito, H. Tanaka, R. Konishi, T. Furuhara and T. Maki: Acta

Mater., 51 (2003), 1789.5) A. F. Gourgues-Lorenzon: Int. Mater. Rev., 52 (2007), 65.6) A. F. Gourgues, H. M. Flower and T. C. Lindley: Mater. Sci. Tech-

nol., 16 (2000), 26.7) H. Kitahara, R. Ueji, N. Tsuji and Y. Minamino: Acta Mater., 54

(2006), 1279.8) S. Morito, X. Huang, T. Furuhara, T. Maki and N. Hansen: Acta

Mater., 54 (2006), 5323.9) G. Miyamoto, A. Shibata, T. Maki and T. Furuhara: Acta Mater., 57

(2009), 1120.10) H. Kitahara, R. Ueji, M. Ueda, N. Tsuji and Y. Minamino: Mater.

Charact., 54 (2005), 378.11) S. Morito, S. Shibata and T. Maki: CAMP-ISIJ, 18 (2005), 1304.12) S. Morito, J. Nishikawa, T. Ohba, T. Furuhara and T. Maki: Proc. of

the Int. Conf. on Martensitic Transformations, ICOMAT-08, TMS,Warrendale, PA, (2008), 649.

13) G. G. E. Seward, S. Celotto, D. J. Prior, J. Wheeler and R. C. Pond:Acta Mater., 52 (2004), 821.

14) T. Fukino and S. Tsurekawa: Mater. Trans., 49 (2008), 2770.15) H. Terasaki and Y. Komizo: Sci. Technol. Weld. Join., 11 (2006),

561.16) D. Zhang, H. Terasaki and Y. Komizo: Acta Mater., 58 (2010), 1369.17) Y. Komizo and H. Terasaki: Sci. Technol. Weld. Join., 16 (2011), 56.18) Y. Komizo and H. Terasaki: Sci. Technol. Weld. Join., 16 (2011), 61.19) G. H. Eichelman and F. C. Hull: Trans. Am. Soc. Met., 45 (1953), 77.20) N. Takayama, G. Miyamoto and T. Furuhara: CAMP-ISIJ, 22 (2009),

534, CD-ROM.21) P. M. Kelly, A. Jostsons and R. G. Blake: Acta Metall. Mater., 38

(1990), 1075.22) N. Shibuta, S. Nambu, K. Asakura, J. Inoue and T. Koseki: CAMP-

ISIJ, 23 (2010), 1161, CD-ROM.23) S. Morito: Tetsu-to-Hagané, 95 (2009), 90.24) A. Lambert-Perlade, A. F. Gourgues and A. Pineau: Acta Mater., 52

(2004), 2337.25) T. Furuhara, H. Kawata, S. Morito and T. Maki: Mater. Sci. Eng. A-

Struct. Mater. Prop. Microstruct. Process., 431 (2006), 228.26) T. Furuhara: Tetsu-to-Hagané, 95 (2009), 97.27) G. Miyamoto: Mater. Jpn., 49 (2010), 332.28) S. Zhang, H. Terasaki and Y. I. Komizo: Proc. of 20th the Int.

Offshore and Polar Engineering Conf., ISOPE, Int. Society ofOffshore and Polar Engineers, Cupertino, CA, (2010), 4, 308.

29) S. Y. Zhang, H. Terasaki and Y. Komizo: Tetsu-to-Hagané, 96(2010), 64.

30) S. Y. Zhang, H. Terasaki and Y. Komizo: Tetsu-to-Hagané, 96(2010), 691.