Effect of Focal Depth of HAADF-STEM Imagingon the Solute Enriched Layers in Mg Alloys

Takanori Kiguchi1,+, Yohei Yamaguchi2, Shunya Tashiro2, Kazuhisa Sato1 and Toyohiko J. Konno1

1Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan2Department of Materials Science, Graduate School of Engineering, Tohoku University, Sendai 980-8579, Japan

This study has found that the Z-contrast of aberration-corrected high-angle annular dark field-scanning transmission electron microscopy(HAADF-STEM) images of solute enriched layers in Mg-TM-RE (TM:Zn, RE:Y, Gd) alloys is sensitive to the imaging conditions: the defocusand the focal depth. Consequently, the depth position of solute enriched layers in the projected direction shows strong effect on the Z-contrastaround each layer. The blurring of the Z-contrast is preferentially observed in both sides region along (0001)Mg basal planes. The boundaryposition between HCP- and FCC-type stacking sequence at edge regions does not change by defocus. The permissible defocus amount iscomparable to the focal depth. The aberration-corrected HAADF-STEM has short focal depth less than 10 nm, so that it is critical for interpretingimage contrast that the thin foil thickness and the depth position of the nano-size structure such as the solute enriched layers. In turn, it can bepresumed, by the blurring of the Z-contrast, that the depth position of the solute enriched layers in the projected direction. Thus, the focal depthof aberration-corrected HAADF-STEM is quite sensitive to the depth position of solute enriched layers in the thin foil of specimens and thedefocus of the electron probe, so that the imaging condition of aberration-corrected HAADF-STEM is critical in order to interpret correctly theZ-contrast of the images. [doi:10.2320/matertrans.MAW201505]

(Received May 1, 2015; Accepted July 9, 2015; Published August 28, 2015)

Keywords: magnesium alloy, long-period stacking ordered structure (LPSO), solute enriched layers, high-angle annular dark field-scanningtransmission electron microscopy (HAADF-STEM), focal depth, defocus, channeling

1. Introduction

Magnesium alloys show attractive properties as next-generation lightweight structural materials, such as chassisand parts of mobile devices, vehicles and airplanes. In thecurrent century, it has found that a novel structure, so-called“synchronized long period stacking ordered structure(LPSO)” in Mg-TM-RE (TM: transition metals, RE: rareearth metals) alloy systems that show excellent mechanicalproperties in comparison with the conventional Mg alloysand Al alloys such as duralmin.13)

To date, the structure and the morphology of LPSO havebeen clarified in atomic scale in the early studies mainly byZ-contrast imaging using high-angle annular dark field-scanning transmission electron microscopy (HAADF-STEM).410) Zhu et al. propose a model structure of 18R-type and 14H-type LPSOs in Mg96.7Zn0.8Y2.4Zr0.2 alloys.They inferred a well-ordered arrangement of Zn and Yatoms.6,7) Yokobayashi et al. found that the completelyordered L12-type structure in the orderdisorder (OD)structure in Mg-Al-Gd alloys.8) Egusa et al. showed a revisedcrystalline structure of 18R-type and 14H-type LPSO inMg97Zn1RE2 (RE = Y, Er) and Mg85Zn6Y9 alloys.9,10) Basedon the diffuse scattering pattern in the electron diffractionpatterns, they concluded that L12-type clusters align in in-plane short range order. We have already elucidated that thetransformation and the growth mechanism of LPSO poly-types in atomic scale, and that local strain field accompaniedwith LPSO and solute enriched layers.1113)

Recently, an aberration-corrected STEM is attractedtechnique for improving the quality of HAADF-STEMimages and the related microanalyses, such as an energydispersive X-ray spectroscopy (EDS) and electron energyloss spectroscopy (EELS), owing to much sharp and brilliant

electron probe. However, an aberration-corrected STEM usesa highly convergent electron probe, which strongly affects thelength of the electron channeling along an atomic column.Then, it should be critical that the interpretation of the imagecontrast around a nano-scale structure like solute enrichedlayers in LPSO alloys.

In the present study, it has been considered that the effectof the defocus and the focal length of the aberration-correctedHAADF-STEM on the Z-contrast of solute enriched layersthat is an elementary structure of LPSO predominantly inMgZn1Gd2 alloys.

2. Experimental Procedures

The Mg97Zn1Gd2 and Mg97Zn1Y2 (at%) alloy ingots wereprepared using a high-frequency induction casting method inan argon atmosphere. These compositions were just nominalones, and the average composition measured by theinductively coupled plasma-optical emission spectroscopy(ICP-OES) (ICPS-8100, Shimadzu) was much the samevalue. As-prepared ingots of Mg97Zn1Gd2 alloys were aged at553K during the times within 3.6 ksec in argon atmospherewithout any solution treatment at higher temperature and thenquenched in ice water in order to precipitate the soluteenriched layers. As-prepared ingots of Mg97Zn1Y2 alloyswere also aged at 933K for 3.6 ksec in argon atmospherefollowed by quenched in ice water in order to prepare thesmaller solute enriched layers rather than that of Mg97Zn1Gd2alloys.

The structural analysis was conducted using an aberration-corrected scanning transmission electron microscope (STEM)(JEM-ARM200F Cold FEG, 200 kV, JEOL). The estimatedconvergent semi-angle ¡ used for high-angle annular darkfield-scanning transmission electron microscopy (HAADF-STEM) was 13, 20, 27mrad in order to vary the focal depth.The estimated collection semi-angle ¢ range of HAADF-+Corresponding author, E-mail: tkiguchi@imr.tohoku.ac.jp

Materials Transactions, Vol. 56, No. 10 (2015) pp. 1633 to 1638©2015 The Japan Institute of Metals and Materials

http://dx.doi.org/10.2320/matertrans.MAW201505

STEM was set to 90175mrad. Selected area electrondiffraction (SAED) patterns were obtained using the samemicroscope. The thin foils of the alloys for the structuralanalysis were prepared using mechanical polishing followedby low-energy ion milling from 5 kV to 200V (PIPSmodel1691, Gatan). The HAADF-STEM images werefiltered using the local 2D-Wiener filtering to improve theimage quality without periodic artifacts (Filter Pro, HREMResearch).14,15) The thickness of the thin foils were estimatedusing the low-loss spectra of the electron energy lossspectroscopy (EELS) equipped with above the STEM(Enfinium ER, model 977, Gatan). The estimated thicknesst of the thin foil was about 100 nm using the low-loss spectraof EELS by the log-ratio method using the next eq. (1):

t ¼ � ln ItI0

� �; ð1Þ

where is the mean free path of the inelastic scattering, It theintegrated intensity of the low-loss region, and I0 that of zero-loss peak.16) The can be estimated approximately using nexteqs. (2):

¼ 106FðE0=EmÞLnð2¢E0=EmÞ

; F ¼ 1þ E0=1022ð1þ E0=511Þ2;

Em � 7:6Z0:36eff ; Zeff �

Xi

fiZ1:3iX

i

fiZ0:3i

; ð2Þ

where E0 is the kinetic energy of an electron 200 keV, ¢ thecollection semi-angle 12.5mrad, F the relativistic factor, fi theatomic fraction of each element of atomic number Zi.16)

3. Results and Discussion

3.1 Z-contrast of the solute enriched layersFirst, the observed shape of the solute enriched layers is

investigated under the HAADF-STEM conditions of ¡ =27mrad. Figure 1 shows a HAADF-STEM image ofMg97Zn1Y2 alloy with a SAED pattern including the areaobserving from ½2�1�10�Mg direction. This alloy was used inorder to form the relatively small solute enriched layers. Theline shape Z-contrast appears in the image. These contrastcorresponds to solute enriched layers with FCC-type stackingsequence from the early studies.6,17) The SAED patternindicates that the clearly appeared streaks around the basicreflections correspond to high density of planar defects, i.e.solute enriched layers. The length is distributing between aseveral tens to 100 nm. The lines No. 1, 2, and 3 show blurredZ-contrast in full width, while No. 4 and 5 show blurred Z-contrast partly. The difference of the Z-contrast can beascribed to the difference of the defocus of the convergentelectron beam, as we will describe later.

Figure 2 is an atomic-resolution HAADF-STEM imageobserved from ½2�1�10�Mg direction with a line profile ofMg97Zn1Gd2 alloys under the condition of ¡ = 27mrad.The image shows that all of these line shape Z-contrastcorresponds to solute enriched layers with FCC-type ABCAstacking sequence. However, the Z-contrast around eachsolute enriched layer seems different. The intensity ofbackground is high and the signal/background (S/B) ratio

of solute enriched layers are low in the region 1; Theintensity of background is high and the S/B ratio of soluteenriched layers are high in the region 2; The intensity ofbackground is low and the S/B ratio of solute enriched layersare also low in the region 3. The region 1 corresponds to thebroad Z-contrast in Fig. 1. These results infer that there aretwo factors that affect the Z-contrast of solute enriched layersand of background around them.

3.2 The effect of the defocus on Z-contrast of the soluteenriched layers

In this section, it is investigated that the effect of thedefocus on Z-contrast of the solute enriched layers in order toconsider the S/B ratio of solute enriched layers against thesurrounding background in Figs. 1, 2.

Figure 3 shows a through focus series of HAADF-STEMimages of solute enriched layers in Mg97Zn1Gd2 alloys

Fig. 1 HAADF-STEM image of the Mg97Zn1Y2 aged at 933K for 1 h withselected area electron diffraction pattern projected from ½2�1�10�Mgdirection. The arrows in the image indicates typical solute enrichedlayers with blurred Z-contrast.

Fig. 2 (a) HAADF-STEM image of the Mg97Zn1Gd2 aged at 553K for 1 h,projected from ½2�1�10�Mg direction, and (b) the line profile along the dottedline in (a).

T. Kiguchi, Y. Yamaguchi, S. Tashiro, K. Sato and T. J. Konno1634

projected from ½2�1�10�Mg direction under the condition that¡ = 27mrad. The defocus range is from 0 nm (just focus) to¹20 nm by ¹5 nm step. The just focus cannnot be decidedtechnically, so that it was decided in this study as the focuswhere atomic resolution images can be resolved most clearlyfor descriptive purposes. The estimated thickness of the thinfoil was about 100 nm using the low-loss spectra of EELSby the log-ratio method by the eq. (1) in the section 2.Therefore, the defocus range and the focal depth in this studyare shorter than the thin foil thickness. Z-contrast of the 2layers clearly reflects the atomic-scale structure until thedefocus of ¹10 nm. It is not recognized that the blurring ofthe background contrast around the solute enriched layers.These results infer that the defocus simply affects thesharpness of the solute enriched layers.

Here, focused on the edge structure of the left layer in eachimages. The dotted lines in images show the boundary ofthe stacking sequence change from FCC to HCP. Figure 3reveals that the edge structure and the boundary position donot change by the defocus. Figure 4 shows a HAADF-STEMimage obtained from the direction largely tilted from½2�1�10�Mg. This image clearly indicates the edge shape ofsolute enriched layers with an acute angle such as a utilityknife. This edge structure was also observed using 3D-electorn beam tomography.18)

The focal depth of the convergent electron beam inHAADF-STEM is described by the next equation:

Focal depth ¼ 1:77 � ¡2

; ð3Þ

where is the wave length of the incident electron beam, 2.51pm, and ¡ the convergent semi-angle of the incident electronbeam, 27mrad.1921) This equation shows that the focal depthis inversely proposed to the square of the numerical number ofoptics. Using the eq. (3), the focal depth estimates at 6 nmunder the given condition. This value well consists with thefact that the solute enriched layers in Fig. 3 are clearly imagedin an atomic resolution until the defocus ¹5 nm, while theybecome blurred under the defocus less than ¹10 nm. Inparticular, the atomic resolution image cannot be obtainedunder the defocus less than ¹15 nm any longer. This resultclaims that the correct imaging of Z-contrast is restricted to thedefocus value comparable with the focal depth of the incidentelectron beam, and infers that the channeling length of theelectron beam is shorter than thin foil thickness.

Then, the relationship between the incident electron beamand the edge structure in Fig. 3 should be in the schematicmodel in Fig. 4(b); the edge region of the solute soluteenriched layer locates at the incident surface side. Theelectron beam principally channels in the solute enrichedlayer when it is focused on the surface region. It channels inthe ¡-Mg matrix region and dechannels in the solute enrichedlayer when it is focused on the region away from the surface,which leads to the blurred tail of contrast around the soluteenriched layer.

3.3 The effect of the focal depth on Z-contrast of thesolute enriched layers

In this section, the effect of the focal depth on the Z-contrastof the solute enriched layers and the surrounding backgroundin the matrix under the fixed focus near just focus, which

Fig. 3 Through focus series of HAADF-STEM images of solute enrichedlayers in Mg97Zn1Gd2 alloys projected from ½2�1�10�Mg direction under thecondition that ¡ = 27mrad. The defocus ¦f of each image was (a) 0 nm,(b) ¹5 nm, (c) ¹10 nm, (d) ¹15 nm, and (e) ¹20 nm.

Fig. 4 (a) HAADF-STEM image obtained from the direction largely tiltedfrom ½2�1�10�Mg, and (b) schematic image of the effect of a defocus on thedepth position in focus within a thin foil.

Effect of Focal Depth of HAADF-STEM Imaging on the Solute Enriched Layers in Mg Alloys 1635

means that an atomic resolution image seems most clearly in atrough focus series under the given convergent semi-angle ¡.

First, it is considered that the effect of the focal depthacross the solute enriched layers, i.e. [0001]Mg direction.Figure 5 depicts HAADF-STEM images obtained at the fixedfocus near just focus under the condition of (a) ¡ = 13mrad,(b) ¡ = 20mrad, (c) ¡ = 27mrad, projected from ½2�1�10�Mgdirection. Figures 5(d) and (e) are the line profiles of theimage intensity distribution along the dotted lines AB and CDfor three convergent conditions, respectively. Figure 5(d)shows the definite intensity profiles of the solute enrichedlayers compared with the surrounding matrix for allconditions. This sharp profile corresponds to the FCC-typestacking sequence of the solute enriched layers.410) On theother hand, Fig. 5(e) shows the blurred intensity profiles ofthe solute enriched layers depending on the conditions. Thelarger ¡ results in much blurred intensity profile. The peakintensity of the center two layers of the solute enriched layersare buried in the background intensity under the conditionthat ¡ is larger than 20mrad, while the intensity profiles of thesolute enriched layers are clearly revealed under the condition¡ = 13mrad. This result shows that the shorter focal depthgives defocused and broad contrast of solute enriched layers,while the longer focal depth gives relatively sharp contrast ofsolute enriched layers. This behavior infers that the soluteenriched layer buried under the ¡-Mg matrix. The result infersthat the structure near the thin foil surface region is clearlyimaged, however, the structure below the depth position ofthe focal depth is defocused and forms image background.

Now, what is the reason for the difference of the focaldepth dependence on the intensity profiles? We consider theeffect of the convergent semi-angle ¡ on the focal depthaccording to the eq. (3). Figure 6 depicts (a) the relationshipbetween the focal depth and the convergent semi-angle and(b) the schematic image of the focal depth difference in the

thin foils for STEM imaging. The convergent semi-angle ¡used in this study corresponds to the focal depth of 6 nm(27mrad), 10 nm (20mrad), and 25 nm (13mrad). The focalspread of the used STEM by chromatic aberration isestimated about 2 nm, which can be ignored for considering

Fig. 5 HAADF-STEM images obtained at the just focus under the condition of (a) ¡ = 13mrad, (b) ¡ = 20mrad, (c) ¡ = 27mrad,projected from ½2�1�10�Mg direction. Line profiles of the image intensity along the dotted lines (d) AB and (e) CD for three convergentconditions, respectively.

Fig. 6 (a) Relationship between the focal depth and the convergent semi-angle, and (b) schematic image of the focal depth difference in the thinfoils for STEM imaging.

T. Kiguchi, Y. Yamaguchi, S. Tashiro, K. Sato and T. J. Konno1636

the effect of the focal depth on our HAADF-STEMexperiments. Usually, the ¡ = 10mrad is adopted in conven-tional STEM without the aberration correction, which resultsin the focal depth of 45 nm. Figure 6(a) clearly shows that thefocal length steeply varies around ¡ = 10mrad. Thus, thefocal depth increases rapidly under the small ¡ less than20mrad. These difference affects on the effective channelinglength in the thin foil as shown in the schematic imageFig. 6(b), because it become easy that the incident electronsescape from the initial atomic column to the neighbor atomiccolumns under the larger ¡. Figures 5(a)(c) were takenunder the just focus, which means that the electron beamfocused on the incident surface of the thin foils. Then, thestructure near the surface is preferentially reflected in the Z-contrast, while the bottom region in the thin foils is defocusedowing to the dechanneling of the electron beam.

Based on the consideration, the depth information of thesolute enriched layers qualitatively. The solute enriched layerwith sharp intensity profile on AB locates on the surfaceregion of the thin foil less than 6 nm and continues down to themiddle or bottom regions, while that with blurred intensityprofile on CD locates on the middle or bottom regions.

Finally, it is considered that the effect of the focal depth onthe intensity profile at the edge of a solute enriched layer.Figure 7 shows that HAADF-STEM images obtained at thejust focus under the condition of (a) ¡ = 13mrad, (b) ¡ =20mrad, (c) ¡ = 27mrad, projected from ½2�1�10�Mg direction,and (d) the line profiles of the intensity of the edge region ofthe right layer for (a), (b) and (c) along ½01�10�Mg direction.Figure 7(d) indicates that the width of the edge regionremains unchanged, which is 24 nm. Comparing with Fig. 6,the focal depth preferentially affects on the image blurringinto [0001]Mg direction since the relationship between thedirection of the solute enriched layer and the contrastblurring; the contrast blurring in the edge direction overlapsthe solute enriched layer itself, while that in the out-of planedirection overlaps only with lower contrast ¡-Mg matrix.This result shows that the larger ¡ leads to the blurredcontrast just around the edge region, which means that theedge region elongates not along the thin foil surface but intothe internal of the foil.

Concluding these results, the effect of the focal depth andthe depth position in the thin foil on Z-contrast around soluteenriched layers in Fig. 8. This figure shows schematic imagesof the blurring of the Z-contrast around solute enriched layers:(a) the layers elongated along the foil surface, and (b) thelayers elongated into the foil internal region. The blurring ofthe Z-contrast is ignorable in the layers elongated along thefoil surface shown in Fig. 8(a), while it is predominant inthe layers elongated into the foil internal region shown inFig. 8(b). Thus, Z-contrast of the edge structure of soluteenriched layers located along the foil surface can beinterpreted intuitively. However, in general, the imagingcondition, the nanostructure in the specimen foils, and thedefocus of the electron beam should be well considered inorder to interpret correctly the Z-contrast of the images.

4. Conclusions

This study has investigated that the effect of the defocus

and the focal depth of the aberration-corrected HAADF-STEM on the Z-contrast of the solute enriched layers inMgZn1Y2, MgZn1Gd2 alloys, in order to consider the contrastaround the layers.(1) Z-contrast of HAADF-STEM images of the solute

enriched layers and the matrix around them is stronglyaffected on the depth position of the solute enrichedlayers, the focal depth, and the defocus of the incidentelectron beam.

(2) The permissible defocus amount for the atomic-resolution HAADF-STEM imaging is comparable tothe effect of the focal depth.

(3) If solute enriched layers locate along the foil surface,the blurring of the Z-contrast is preferentially observedin both sides region along the basal planes, while it ishardly observed out of the front edge. The boundaryposition between HCP- and FCC-type stacking se-quence does not change with defocus.

(4) The aberration-corrected HAADF-STEM has shortfocal depth less than 10 nm under the usually adoptedimaging condition, so that it is critical that the thin foil

Fig. 7 HAADF-STEM images obtained at the just focus under thecondition of (a) ¡ = 13mrad, (b) ¡ = 20mrad, (c) ¡ = 27mrad, projectedfrom ½2�1�10�Mg direction. (d) Line profiles of the intensity of the edgeregion of the right layer for (a), (b) and (c) along ½0�1�10�Mg direction.

Effect of Focal Depth of HAADF-STEM Imaging on the Solute Enriched Layers in Mg Alloys 1637

thickness and the depth position of the solute enrichedlayers. In turn, it can be presumed, by the blurring of theZ-contrast, that the depth position of the solute enrichedlayers.

(5) Thus, Z-contrast of the edge structure of solute enrichedlayers located along the foil surface can be interpretedintuitively, however, the imaging condition, soluteenriched layers in the thin foils, and the defocus ofthe electron beam should be well considered in order tointerpret correctly the Z-contrast around the soluteenriched layers.

Acknowledgements

This work was supported by a Grant-in-Aid for scientificResearch on Innovative Areas “Synchronized Long-PeriodStacking Ordered Structure ³ The Evolution of the MaterialScience for Innovative Development of the Next-generationLightweight Structure Materials ³” (23109006) from theMinistry of Education, Culture, Sports, Science and Tech-nology (MEXT), Japan. The ingots were supported byKawamura laboratory, Kumamoto University. The electro-discharge machining of the ingots was supported by Mr.Manabu Itoh, Institute for Materials Research, TohokuUniversity. This study was partly supported by HatakeyamaBunka Zaidan, and TonenGeneral Sekiyu Research &Development Encouragement & Assistance. The usage ofelectron microscopy was supported by Dr. Makoto Nagasako,Institute for Materials Research, Tohoku University.

REFERENCES

1) Y. Kawamura, K. Hayashi, A. Inoue and T. Masumoto: Mater. Trans.42 (2001) 11721176.

2) A. Inoue, Y. Kawamura, M. Matsushita, K. Hayashi and J. Koike:J. Mater. Res. 16 (2001) 18941900.

3) E. Abe, Y. Kawamura, K. Hayashi and A. Inoue: Acta Mater. 50 (2002)38453857.

4) T. Itoi, T. Seimiya, Y. Kawamura and M. Hirohashi: Scr. Mater. 51(2004) 107111.

5) M. Matsuda, S. Ii, Y. Kawamura, Y. Ikuhara and M. Nishida: Mater.Sci. Eng. A 393 (2005) 269274.

6) Y. M. Zhu, M. Weyland, A. J. Morton, K. Oh-ishi, K. Hono and J. F.Nie: Scr. Mater. 60 (2009) 980983.

7) Y. M. Zhu, A. J. Morton and J. F. Nie: ActaMater. 58 (2010) 29362947.8) H. Yokobayashi, K. Kishida, H. Inui, M. Yamasaki and Y. Kawamura:

Acta Mater. 59 (2011) 72877299.9) D. Egusa and E. Abe: Acta Mater. 60 (2012) 166178.10) E. Abe, A. Ono, T. Itoi, M. Yamasaki and Y. Kawamura: Philos. Mag.

Lett. 91 (2011) 690696.11) T. Kiguchi, Y. Ninomiya, K. Shimmi, K. Sato and T. J. Konno: Mater.

Trans. 54 (2013) 668674.12) T. Kiguchi, S. Matsunaga, K. Sato and T. J. Konno: Mater. Trans. 55

(2014) 13771382.13) S. Matsunaga, T. Kiguchi, K. Sato and T. J. Konno: Mater. Trans. 56

(2015) 923927.14) T. Kogure, P. H. C. Eilers and K. Ishizuka: Microscopy and Analysis

(2008) S11S14.15) K. Ishizuka, P. H. C. Eilers and T. Kogure: Microsc. Today 15(5)

(2007) 1620.16) R. F. Egerton: Electron Energy Loss Spectroscopy in the Electron

microscopy, 3rd ed., (Plenum Press, New York, NY, 2011).17) M. Yamasaki, M. Sasaki, M. Nishijima, K. Hiraga and Y. Kawamura:

Acta Mater. 55 (2007) 67986805.18) K. Sato, S. Matsunaga, S. Tashiro, Y. Yamaguchi, T. Kiguchi and T. J.

Konno: Mater. Trans. 56 (2015) 928932.19) P. D. Nelist: Electron Energy Loss Spectroscopy in the Electron

microscopy, ed. by S. J. Pennycook and P. D. Nellist, (Springer, NewYork, NY, 2011) pp. 91115.

20) P. D. Nellist, E. C. Cosgriff, G. Behan and A. I. Kirkland: Microsc.Microanal. 14 (2008) 8288.

21) N. Tanaka and K. Saitoh: Scanning Transmission Electron Microscopyof Nanomaterials, ed. by N. Tanaka, (Imperial College Press, London,UK, 2015) pp. 7576.

Fig. 8 Schematic images of the blurring of the Z-contrast around solute enriched layers: (a) the layers elongated along the foil surface, and(b) the layers elongated into the foil internal region.

T. Kiguchi, Y. Yamaguchi, S. Tashiro, K. Sato and T. J. Konno1638

http://dx.doi.org/10.2320/matertrans.42.1172http://dx.doi.org/10.2320/matertrans.42.1172http://dx.doi.org/10.1557/JMR.2001.0260http://dx.doi.org/10.1016/S1359-6454(02)00191-Xhttp://dx.doi.org/10.1016/S1359-6454(02)00191-Xhttp://dx.doi.org/10.1016/j.scriptamat.2004.04.003http://dx.doi.org/10.1016/j.scriptamat.2004.04.003http://dx.doi.org/10.1016/j.msea.2004.10.040http://dx.doi.org/10.1016/j.msea.2004.10.040http://dx.doi.org/10.1016/j.scriptamat.2009.02.029http://dx.doi.org/10.1016/j.actamat.2010.01.022http://dx.doi.org/10.1016/j.actamat.2011.08.011http://dx.doi.org/10.1016/j.actamat.2011.09.030http://dx.doi.org/10.1080/09500839.2011.609149http://dx.doi.org/10.1080/09500839.2011.609149http://dx.doi.org/10.2320/matertrans.MI201221http://dx.doi.org/10.2320/matertrans.MI201221http://dx.doi.org/10.2320/matertrans.MAW201413http://dx.doi.org/10.2320/matertrans.MAW201413http://dx.doi.org/10.2320/matertrans.MH201404http://dx.doi.org/10.2320/matertrans.MH201404http://dx.doi.org/10.1016/j.actamat.2007.08.033http://dx.doi.org/10.2320/matertrans.MH201406http://dx.doi.org/10.1017/S1431927608080057http://dx.doi.org/10.1017/S1431927608080057