Available online at www.sciencedirect.com

ScienceDirect

Scripta Materialia 89 (2014) 25–28

www.elsevier.com/locate/scriptamat

Deformation twinning in a Mg–Al–Gd ternary alloycontaining precipitates with a long-period stacking-ordered

(LPSO) structure

Kyosuke Kishida,a,b,⇑ Atsushi Inoue,a Hideyuki Yokobayashia and Haruyuki Inuia,b

aDepartment of Materials Science and Engineering, Kyoto University, Sakyo-ku, Kyoto 606-8501, JapanbCenter for Elements Strategy Initiative for Structural Materials (ESISM), Kyoto University, Sakyo-ku, Kyoto 606-8501, Japan

Received 20 April 2014; revised 24 May 2014; accepted 13 June 2014Available online 2 July 2014

A magnesium alloy containing the Mg–Al–Gd long-period stacking-ordered (LPSO) platelet precipitates parallel to the basalplane was found to deform by c-axis tension twinning on {11�21}, whose passage causes the bending of the LPSO platelets, when Mggrains are oriented favorably for extension along the c-axis during plane-strain compression at room temperature. The bending ofthe LPSO platelets was found to be caused by the deformation twinning equivalent to {11�21} twinning in the Mg matrix.� 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Magnesium alloys; Deformation structure; Twinning; Transmission electron microscopy (TEM)

Recently, magnesium alloys containing Mg–TM(transition-metal)–RE (rare-earth) ternary precipitateswith long-period stacking-ordered (LPSO) structureshave attracted considerable attention as promising light-weight structural materials because of the simultaneousachievement of high strength (�600 MPa) and goodductility (�5%) [1]. It is usually reported that theseexcellent mechanical properties are achieved only afterextrusion at high temperatures above 350 �C [2].Although grain refinement of the Mg matrix in the vicin-ity of bent LPSO platelet precipitates as a result ofrecrystallization has been considered as a possible rea-son for the observed high strength and good ductility[3,4], the detailed mechanisms behind this have largelyremained unsolved. The lack of knowledge on funda-mental properties of the LPSO phase, such as crystalstructure, thermal stability and deformation mecha-nisms, is largely responsible for this. Hagihara et al.[5,6] were the first to make a systematic study on thedeformation mechanisms of the LPSO phase(Mg–5 at.% Zn–7 at.% Y) with the use of directionally

http://dx.doi.org/10.1016/j.scriptamat.2014.06.0191359-6462/� 2014 Acta Materialia Inc. Published by Elsevier Ltd. All rights

⇑Corresponding author at: Department of Materials Science andEngineering, Kyoto University, Sakyo-ku, Kyoto 606-8501,Japan.; e-mail: kishida.kyosuke.6w@kyoto-u.ac.jp

solidified (DS) ingots consisting of LPSO platelets withtheir platelet faces (parallel to the basal plane) nearlyparallel to the growth direction. They concluded thatbasal slip is by far the easiest slip system operative inmost orientations while deformation bands are formedwhen the compression axis is almost parallel to the basalplane so that the operation of basal slip is suppressed.They further concluded from transmission electronmicroscopy (TEM) observations that the deformationbands they observed are actually kinks formed bynumerous basal dislocations accumulated perpendicu-larly to the basal planes in a wall, as in the classicalexplanation for the kink formation in hexagonal close-packed (hcp) metals by Hess and Barrett [7]. Hagiharaet al. inferred that the kink formation is responsiblefor the bending of LPSO platelet precipitates observedin hot-extruded Mg alloys [5]. However, the rotationangles between two crystalline regions separated by akink boundary in Figs. 14 and 15 of Ref. [5] are as largeas 30–60�, which requires a basal dislocation every fewbasal planes in the kink wall. On top of that, it is verydifficult to see each individual dislocation in the kinkwall in Figs. 14 and 15 of Ref. [5]. These findings raisea serious question as to whether or not all these kinkbands are indeed formed by the accumulation of basaldislocations, which moved on closely spaced atomic

reserved.

26 K. Kishida et al. / Scripta Materialia 89 (2014) 25–28

planes, in the wall perpendicular to the basal plane. Inthe present study, we investigate deformation micro-structures of a Mg–Al–Gd alloy containing a ternaryLPSO phase after plane-strain compression deformationat room temperature, in order to gain insight into themechanisms behind the bending of LPSO platelets dur-ing hot extrusion, excluding any high-temperatureeffects such as dynamic recrystallization. In our recentstudies [8,9], the crystal structure of the LPSO phasein the Mg–Al–Gd system has been found to be basedon the stacking of six-layer structural blocks (18R-type)as described by order–disorder (OD) theory [8–10], andgenerally possesses a one-dimensional disordered naturealong the stacking direction.

An ingot of a Mg–Al–Gd ternary alloy with a nomi-nal chemical composition of Mg–1.5 at.% Al–6.5 at.%Gd was produced by high-frequency induction-meltingmixtures of high-purity Mg, Al and Gd in a carbon cru-cible with a lid in vacuum. The ingot was heat-treated at525 �C for 64 h. Rectangular specimens for plane-straincompression tests measuring 2.2 mm � 2.3 mm �5.5 mm (in the x, y z directions, respectively, inFig. 1a) were cut from the heat-treated ingot by electricdischarge machining, and were polished mechanicallyand then electrolytically in a solution of perchloric acid,n-butyl alcohol and methanol (1:30:130 by volume)with 0.2 M of LiCl at �55 �C. Plane-strain compressiontests were performed at a nominal strain rate of5.5 � 10�4 s�1 at room temperature with a channel diesimilar to that designed by Chin et al. [12]. As seen inFigure 1a, the compression axis was set parallel to thez axis, while the channel die was placed to suppress spec-imen deformation in the y axis, allowing specimen exten-sion only in the x direction. Compression deformationwas made until the reduction measured in the z-axisreaches �15% with the specimen wrapped in polytetra-fluoroethylene film to reduce friction with the channeldie and compression jig. Deformation microstructures

Fig. 1. (a) Geometric configuration of plane-strain compression and(b–d) deformation microstructure of Mg–Al–Gd ternary ingotdeformed in plane-strain compression. (b) SEM backscattered electronimage, (c) EBSD orientation map and (d) TEM BF image. Arrows in(b) and (d) indicate boundaries between matrix and twinned regions.

were examined by scanning electron microscopy(SEM), TEM and scanning transmission electronmicroscopy (STEM). Specimens for TEM/STEM obser-vations were cut perpendicular to the y-axis of the spec-imen, polished mechanically and then ion-milled with 5(initial) or 2 (final) keV Ar ions.A SEM backscattered-electron image of Figure 1b depicts a deformation struc-ture of a Mg grain that is located on the cross-sectionperpendicular to the y direction in Figure 1a. For thisparticular Mg grain, the compression axis is approxi-mately parallel to [25 5 20 1], which is almost parallelto the basal plane. The x, y and z axis directions are indi-cated in Figure 1b. Many thin platelet precipitates of theLPSO phase, which appeared as bright bands in Fig-ure 1b with platelet faces parallel to the basal plane ofMg, are observed to be bent frequently, and appear asdeformation kinks after hot extrusion [4]. It is notewor-thy in Figure 1b that the bending angles observed forLPSO platelets are almost identical. Orientation map-ping by electron backscattered diffraction (EBSD)clearly indicates that the bending of LPSO platelets isformed as a result of the passage of deformation bands,as seen in Figure 1c. There are two sets of deformationbands: one with the boundary parallel to (11�21) and theother with the boundary parallel to (1121). Each of thesetwo sets of deformation bands has an identical orienta-tion, exhibiting a particular orientation relationshipwith the other areas (designated arbitrarily as thematrix). Orientation analysis by EBSD for the Mg grainrevealed that the matrix and the deformation bands withboundary planes parallel to (11�21) and (1121) are bothapproximately in a 34� rotation relationship around[�1100]. This is further confirmed by TEM observations,as shown in Figure 1d. A TEM bright-field image ofFigure 1d clearly indicates that the bending of LPSOplatelets is always accompanied by the passage of defor-mation bands with the boundary plane parallel either to(11�21) or to (1121). In addition, many dislocations areobserved to be generated in the Mg phase in both thematrix and deformation bands. Diffraction analysis by

Fig. 2. (a) BF-STEM image of a deformation twin and (b–d) SADpatterns taken from (b) Mg phase and (d) LPSO phase. (c) Schematicillustration of the SAD pattern of (b).

K. Kishida et al. / Scripta Materialia 89 (2014) 25–28 27

TEM/STEM has indicated that deformation bands areindeed deformation twins, as described below.

Figure 2a shows a magnified bright-field (BF) STEMimage of a region containing a deformation band withthe boundary planes macroscopically parallel to(11�21). A LPSO platelet precipitate is imaged as a darkband extending along the lateral direction in Figure 2a.The LPSO platelet precipitate is obviously bent by thepassage of the deformation band. The boundariesbetween the deformation band and the matrix are paral-lel to (11�21)1 and they are observed edge-on, since theincident beam direction is [�1100]. Selected-area electrondiffraction (SAED) patterns taken from regions contain-ing both the deformation band and matrix are shown inFigure 2(b) and (d) for the Mg and LPSO phases,respectively.

Analysis of the SAED pattern of Figure 2b indicatesthat the matrix and deformation band of the Mg phasehave a common (mutual) zone-axis orientation of[�1100]; these two regions are rotated with respect to eachother by �34� about the zone-axis orientation and theyare in a mirror relationship with respect to the boundaryplane of (11�21), as schematically shown in Figure 2c.These observations obviously indicate that deformationbands that make LPSO platelet precipitates bent aredeformation twins. Based on the above observations,the twinning elements for the deformation twins on{11�21} in the Mg phase are determined as follows:

K1 : f11�21g;K2 : f0001g;g1 : h�1�126i;g2

: h11�20i and s ¼ 0:616: ð1ÞThe same rotation relationship with a rotation angle

of �34� is confirmed to occur in the LPSO phase by ana-lyzing the SAED pattern of Figure 2d. The possibletwinning elements for the LPSO phase of 18R-type arethen deduced by assuming the simplest polytype of 1M(space group C2/m) among the possible OD structures[8,9,11], as follows:

K1 : f041g1M ;K2 : f001g1M ;g1 : h4 �3 12i1M ;g2 : h010i1M

ð2aÞ

K1 : f661g1M ;K2 : f001g1M ;g1 : h1 3 �2�4i1M ;g2 : h310i1M

ð2bÞ

K1 : f�6�65g1M ;K2 : f001g1M ;g1 : h17 3 24i1M ;g2 : h�3�10i1M

ð2cÞThree different sets of twinning elements are deduced

for three different orientation variants because of themonoclinic symmetry of the 1M polytype. These indicesare referred to the 1M polytype with the unit celldescribed as follows:

1 Since the unit cell of the LPSO phase in the Mg–Al–Gd system cannotunambiguously be determined because of the stacking disorder ofstructural blocks along the direction perpendicular to the close-packed planes, Miller indices to express directions and planes for theLPSO phase are referred to as those of the matrix phase of Mg withthe hcp structure unless otherwise stated.

a1M � 2ffiffiffi3p

aMg; b1M ¼ffiffiffi3p

a1M ; c1M

�ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiða1M=3Þ2 þ ð3cMgÞ2

q; b1M

� 180� � tan�1ð9cMg=a1MÞ;where aMg and cMg are the lattice parameters of Mg withthe hcp structure.

Figure 3 shows an atomic-resolution high-angleangular dark-field (HAADF)-STEM image taken froma twin boundary region in the Mg–Al–Gd LPSO phaseprojected along the [1�100]. The ordered arrangementof brighter spots in the double-dagger pattern for thecentral four consecutive Gd-enriched layers is observedin both the matrix and twin regions. This confirms thatdeformation twinning occurs so that the ordered atomicarrangement in each of structural blocks in the ODstructure of the Mg–Al–Gd LPSO phase, which is char-acterized by the periodic arrangement of Al6Gd8 clusterswith the L12-type atomic arrangement on lattice pointsof a 2

ffiffiffi3p

aMg � 2ffiffiffi3p

aMg 2-D primitive hexagonal lattice[8,9], is preserved after deformation twinning. Of impor-tance to note is that the twin boundary is not perfectlyflat but deviated from the symmetric position on anatomic scale and that hardly any strain contrast arisingfrom dislocations is observed on the boundary betweenthe deformation band and matrix. These findings are incontrast to what was reported for kink bands by Hagi-hara et al. [5] for the compression behavior of direction-ally solidified (DS) LPSO phase, indicating thatdeformation bands observed in this study are not kinkbands that are formed by the mechanism proposed byHess and Barrett but are deformation twins [7].

In the OD structure of the Mg–Al–Gd LPSO phase,the stacking of structural blocks is virtually 1-D disor-dered, which can easily be recognized by investigatingthe relative shifts of Gd atom positions in the outer lay-ers of structural blocks, as described in our previouspapers [8,9]. Inspection of the matrix and twin regionsin the experimental HAADF-STEM image of the[1�100] incidence (Fig. 3) indicates that shifts of either 0or 1/3 occur preferentially for both the matrix and twin

Fig. 3. Atomic-resolution HAADF-STEM image of a boundaryregion of the deformation twin in the Mg–Al–Gd LPSO phase.

28 K. Kishida et al. / Scripta Materialia 89 (2014) 25–28

regions. This observation confirms that the OD struc-ture designated as the C1-type in our previous papers[8,9] is formed in the twin region as in the matrix, stillexhibiting 1-D disorder for the stacking of structuralblocks of the OD structure. The stacking sequence ofstructural blocks is not perfectly in mirror relation withrespect to the (11�21) twin boundary plane for the matrixand twin regions. Such a change in the stackingsequence of structural blocks during deformation twin-ning in the LPSO phase may be closely related to theformation (nucleation and growth) mechanisms ofdeformation twins of this type. The mechanism of defor-mation twinning involving atomic movement is cur-rently being investigated by the authors’ research group.

The present results clearly suggest that the deforma-tion twinning observed in this study may also occur dur-ing the hot extrusion process of Mg alloys containingLPSO platelet precipitates, if LPSO platelets are ori-ented so that the platelet face is parallel to the extrusiondirection. We believe that this is also the case for A-ori-ented specimens used in the study on the DS-processedMg–Zn–Y LPSO phase by Hagihara et al. [5,6], whenjudged from the relatively homogeneous shapes ofdeformation bands observed in compression approxi-mately along the basal plane. Most deformation bandswith relatively large rotation angles are believed to bedeformation twins, the same as those observed in thisstudy. Once deformation twins are formed, the basalplane inside the twins is rotated so as to be favorablyoriented for the activation of basal slip, which leads toan additional increase in the rotation angle of the twinregion with respect to the matrix, possibly by the kinkmechanism proposed by Hess and Barrett [7]. We thusbelieve that large rotation angles between the matrixand deformation bands and a rather wide distributionof the rotation angles reported by Hagihara et al. [5]should be interpreted as being due to the formation ofdeformation twins, and the subsequent lattice rotationsinside the twin region as caused by the motion of basaldislocations.

The {11�21} deformation twin identified here in theMg matrix corresponds to the c-axis tension twin, whichhas hardly ever been reported in Mg and its alloys [13].In the polycrystalline specimen tested in the presentstudy under a plane-strain compression condition,deformation twins are confirmed to be activated in Mggrains with the loading axis orientation z and the exten-sion direction x being nearly parallel to the basal planeand the c-axis, respectively. Since basal slip is by far theeasiest slip system in Mg, it is easily imagined that thisgeometrical situation can be realized for many Mggrains during hot extrusion so that {11�21} deformationtwins are formed in suitable Mg grains. In view of thefact that {11�21} deformation twins have hardly ever

been observed in Mg and its alloys, the LPSO phasemay play a decisive role in the nucleation of {11�21}twins. In fact, our preliminary study on the compressiondeformation behavior of the Mg–Zn–Y LPSO phasewith the use of single-crystalline micropillars hasrevealed the activation of similar deformation twins withtheir habit planes close to {11�21} in the Mg–Zn–YLPSO phase [14].

In summary, the deformation microstructures of aMg–Al–Gd alloy with LPSO platelet precipitates afterplane-strain compression at room temperature wereinvestigated. The c-axis tension twins on {11�21}, whichhave hardly ever been reported in Mg and its alloys,were confirmed to be activated in both the Mg matrixand the Mg–Al–Gd LPSO platelet precipitates. Threedifferent sets of twinning elements for the deformationtwins in the Mg–Al–Gd LPSO phase are deduced forthree different orientation variants by assuming the sim-plest MDO polytypes with the monoclinic symmetry(1M polytype) as shown in Eqs. (2a)–(2c).

This work was supported by Grants-in-Aid forScientific Research from the Ministry of Education, Cul-ture, Sports, Science and Technology (MEXT), Japan(Nos. 23360306, 23109002, 26109712 and 26289258)and in part by the Elements Strategy Initiative for Struc-tural Materials (ESISM) from MEXT, Japan.

[1] Y. Kawamura, K. Hayashi, A. Inoue, T. Masumoto,Mater. Trans. 42 (2001) 1172.

[2] Y. Kawamura, M. Yamasaki, Mater. Trans. 48 (2007)2986.

[3] K. Hagihara, A. Kinoshita, Y. Sugino, M. Yamasaki, Y.Kawamura, H.Y. Yasuda, Y. Umakoshi, Acta Mater. 58(2010) 6282.

[4] M. Yamasaki, K. Hashimoto, K. Hagihara, Y. Kawam-ura, Acta Mater. 59 (2011) 3646.

[5] K. Hagihara, N. Yokotani, Y. Umakoshi, Intermetallics18 (2010) 267.

[6] K. Hagihara, Y. Sugino, Y. Fukusumi, Y. Umakoshi, T.Nakano, Mater. Trans. 52 (2011) 1096.

[7] J.B. Hess, C.S. Barrett, Trans. Am. Inst. Min. Metall.Eng. 185 (1949) 599.

[8] H. Yokobayashi, K. Kishida, H. Inui, M. Yamasaki, Y.Kawamura, Acta Mater. 59 (2011) 7287.

[9] K. Kishida, H. Yokobayashi, A. Inoue, H. Inui, in: MRSSymp. Proc., vol. 1516, 2013, p. 291.

[10] K. Dornberger-Schiff, Acta Cryst. 9 (1956) 593.[11] K. Kishida, H. Yokobayashi, H. Inui, Philos. Mag. 93

(2013) 2826.[12] G.Y. Chin, E.A. Nesbitt, A.J. Williams, Acta Metall. 14

(1966) 467.[13] N. Stanford, Philos. Mag. Lett. 88 (2008) 379.[14] A. Inoue, K. Kishida, H. Inui, K. Hagihara, in: MRS

Symp. Proc., vol. 1516, 2013, p. 151.