lable at ScienceDirect

Acta Materialia 135 (2017) 112e123

Contents lists avai

Acta Materialia

journal homepage: www.elsevier .com/locate/actamat

Full length article

Micro-plasticity of medium Mn austenitic steel: Perfect dislocationplasticity and deformation twinning

Eun Jung Seo a, Jin Kyung Kim a, Lawrence Cho a, *, Javad Mola b, Chang Yeol Oh c,Bruno C. De Cooman a

a Materials Design Laboratory, Graduate Institute of Ferrous Technology, Pohang University of Science and Technology, 37673, Pohang, South Koreab Institute of Iron and Steel Technology, Technische Universit€at Bergakademie Freiberg, 09599, Freiberg, Germanyc National Institute for Nanomaterials Technology, Pohang University of Science and Technology, 37673, Pohang, South Korea

a r t i c l e i n f o

Article history:Received 23 March 2017Received in revised form24 May 2017Accepted 7 June 2017Available online 9 June 2017

Keywords:Medium Mn steelNano-indentationMicro-pillarTWIP

* Corresponding author.E-mail address: cyc616@postech.ac.kr (L. Cho).

http://dx.doi.org/10.1016/j.actamat.2017.06.0141359-6454/© 2017 Acta Materialia Inc. Published by E

a b s t r a c t

The micro-scale plastic deformation behavior of an austenitic Fe-1.2%C-7.0%Mn (in wt%) steel was studiedby means of nano-indentation and in situ compression of micro-pillars with selected crystallographicorientations. Transmission electron microscopy analysis reveals that the partial dislocation mediatedtwinning is preferred in a [001]-oriented single grain under compression. Twinning leads to large strainbursts during the nano-indentation and the micro-pillar compression. Perfect dislocation slip is thedominant deformation mechanism in a [111]-oriented single grain under compression. The observationsoffer strong support for the hypothesis that deformation twinning is a plasticity enhancing mechanismactivated during the deformation of medium Mn steel.

© 2017 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction

The strong global interest in medium Mn (4e7%) steel is due totheir excellent strength-ductility balance resulting from their spe-cific strain hardening behavior. Their properties are achieveddespite much lower alloying additions as compared to high Mntwinning induced plasticity (TWIP) steel. A medium Mn steeltypically achieves an ultimate tensile strength (UTS) more than1 GPa and a total elongation (TE) in the range of 30e40% [1e3]. Theexcellent mechanical properties of mediumMn steel are associatedwith a two-phasemicrostructure consisting of austenite and ferrite.

The deformation mechanism of the austenite phase in high MnTWIP steel has been the subject of numerous studies [4e6]. Thebeneficial influence of twins on the plasticity of FCC metals is welldocumented [7,8]. The stacking fault energy (SFE) of the austeniteaffects the deformation mechanism. Martensitic transformation,leading to transformation-induced plasticity (TRIP) effect, is thepredominant deformation mode when the SFE is less than15 mJ m�2. The TWIP effect becomes dominant when the SFE iswithin the 15e45 mJ m�2 range [9e11].

lsevier Ltd. All rights reserved.

It has been suggested that both TRIP and TWIP effects are acti-vated in the retained austenite of medium Mn steel [12e14]. Themechanical properties of the retained austenite in medium Mnsteels have however not yet been documented. One reason for thisis that the grain size of the retained austenite in intercriticallyannealed medium Mn steels is often very small, typically less than1 mm. The ultra-fine grain size of the retained austenite makes itchallenging to evaluate their intrinsic properties. Furthermore, thecarbon and Mn contents of the retained austenite in theintercritically-annealed medium Mn steel are very different fromthe nominal steel composition, as carbon and Mn partition fromferrite to austenite during intercritical annealing.

The present work aims at revealing the orientation-dependentdeformation mechanisms of a medium Mn austenitic steel. In thepresent study, a coarse-grained bulk sample of fully austeniticmediumMn steel was produced to study the micro-plasticity of thehigh carbon, medium Mn austenitic phase by means of both nano-indentation and in situ compression of micro-pillars with selectedcrystallographic orientations. Transmission electron microscopy(TEM) was used to investigate dislocation interactions and theevolution of the deformation microstructure after nano-indentation and micro-pillar compression tests.

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123 113

2. Experimental procedures

The composition of the hot rolled steel used in the present studywas Fe-1.2%C-7.0%Mn (in wt%). This corresponds to the typicalcomposition of the austenite phase in intercritically annealed me-dium Mn steels, when the annealing is carried out to obtain themaximum volume fraction of austenite at room temperature. Themicrostructure of the hot-rolled sheet steel was made fullyaustenitic by water quenching from 1150 �C. The selection of thegrains with specific orientations was determined by electronbackscatter diffraction (EBSD) in an FEI Quanta 3D FEG scanningelectron microscopy (SEM). The specimens used for the EBSDanalysis were prepared by electro-chemical polishing in a solutionof 5% HClO4þ 95% CH3COOH. Coarse grains with the surface normaloriented close to the [001] and [111] directions were chosen fornano-indentation and micro-pillar compression tests.

Nano-indentation tests were conducted using a Hysitron PI 85SEM Pico-Indenter equipped with a diamond Berkovich tip. Theload was applied to a maximum depth of 40 nm. The nano-indentation tests were performed at a displacement rate of 8 nm/s at the initial loading stage. TEM samples were taken from theindentation-tested area by the FIB lift-out technique in an FEIHelious Nanolab 650 dual beam FIB. The Oliver and Pharr analysismethod [15] was used to analyze the nano-indentation force-displacement data. The hardness (H) and the reduced modulus (Er)were derived from the following equations:

H ¼ Pmax

A(1)

and

Er ¼ 1b

ffiffiffip

p2

SffiffiffiA

p (2)

where Pmax is the maximum load applied during the indentation, Ais the projected area of contact between the indenter and thespecimen, b is a correction factor which depends on the geometryof the indenter (b¼ 1.034, for the Berkovich-type indenter), and S isthe unloading stiffness at the maximum depth of penetration. Sinceelastic deformation occurs in both the specimen and the indenter,Er is described by the following relationship

Er ¼ ð1� n2s ÞEs

þ ð1� n2i ÞEi

(3)

Es and ns are Young's modulus and Poisson's ratio for the spec-imen. Ei and ni are Young's modulus and Poisson's ratio for theindenter. For the diamond Berkovich indentor, Ei ¼ 1141 GPa andni ¼ 0.07.

In order to make accurate measurements by indentation ex-periments, the projected contact area, A, must be precisely known.The most commonly used method to calibrate the shape of theindenter tip was proposed by Oliver and Pharr [15,16]. A is empir-ically derived as a function of contact depth (hc), i.e A ¼ f (hc). Thevalue of hc is estimated from

hc ¼ hmax � 0:75Pmax

S(4)

where hmax is the maximum depth of penetration. The area func-tion of f (hc) was established by means of a tip calibration using afused silica standard. The area function relating the projectedcontact area (A) to the contact depth (hc) was determined as

A ¼ 24:5h2c � 5:2935� 104hc þ 4:7361� 106h1=2c

� 5:1554� 107h1=4c þ 1:3349� 108h1=8c

� 8:7237� 107h1=16c (5)

where the first term, 24:5h2c , describes an ideal Berkovich tip andthe other terms describe the deviation from the ideal tip geometry.

The micro-pillars of approximately 600 nm diameter with aheight-to-diameter aspect ratio of 3 were fabricated in an FEIHelious Nanolab 650 dual beam FIB operated at 30 kV. The taperangle of the micro-pillars was approximately 3�. In-situ compres-sion tests on the micro-pillars were performed in a SEM using aHysitron PI 85 SEM Pico-Indenter. A flat diamond punch tip with adiameter of 1 mmwas used. The load was applied up to a maximumdepth of 150 nm. The compression tests were performed at a strainrate of 0.02 s�1. The morphology of the deformed micro-pillars wasanalyzed in a ZEISS field emission Ultra-55 SEM before and after thecompression tests. TEM samples of the deformed pillars wereprepared by the FIB lift-out technique. TEM observations werecarried out in a JEOL JEM-2100F FE-TEM operating at 200 keV.

3. Results

3.1. Nano-indentation tests

Fig. 1 shows a SEM micrograph and EBSD normal direction (ND)inverse pole figure (IPF) maps of the fully austenitic Fe-1.2%C-7.0%Mn steel. The average grain size was approximately 80 mm.Recrystallization twins were commonly observed. The twinboundaries are indicated by white lines in the IPF maps. Both nano-indentation and micro-pillar compression tests were performed incoarse austenite grains for which the surface normal was orientedclose to the [001] and [111] directions (Fig. 1). The influence of grainorientation on the twinning behavior of FCC metals and alloys iswell documented [17e20]. The stress on the leading and trailingpartials of dissociated dislocations depends on the orientation ofthe compression axis and the difference in the stress affects thestacking fault width [17]. Deformation twinning is favored in [001]-oriented crystals when subjected to compression deformation,while perfect dislocation glide is preferred in [111]-oriented crys-tals in compression [17,20]. The calculated maximum Schmid fac-tors for the perfect and partial dislocations with respect to the [001]and [111] compression axes are listed in Table 1.

Fig. 2 shows the crystallographic orientation dependence of thenano-indentation test results for the [001] and [111] directions. Thenano-indentation test results of the present study and the literaturedata [20e24] are summarized in Table 2. A hardness of 7.6 GPa wasobtained when the indentation load was applied along the [001]direction (Fig. 2(a)). Young's modulus, E[001], was estimated to be234.5 GPa. When the indentation was applied along the [111] di-rection (Fig. 2(b)), the hardness and Young's modulus, E[111], were8.5 GPa and 271.5 GPa, respectively. This difference, i.e. Young'smodulus in the <111> direction being higher than that in the<001> direction, is well-documented for austenitic alloys [20,25].

Due to the limitations of the microfabrication process, theindenter tip is not ideally sharp, but rounded. The geometry of theindenter tip can be approximated as spherical in the earliest stagesof contact where plastic yielding occurs [22e24,26]. With thisapproximation, it is possible to predict the expected elasticresponse using the Hertzian law for mechanical contacts, as re-ported in previous studies [22e24,26]. The elastic portion of theload-displacement curve (P-h curve) can be described by theHertzian elastic contact solution [27]:

Fig. 1. (a) SEM backscattered image of the austenitic Fe-1.2%C-7.0%Mn steel. (b) EBSD IPF map of the area labelled “b” in (a). (c) EBSD IPF map of the area labelled “c” in (a). Thecircles in (b) and (c) indicate the positions of [001]//ND-oriented and [111]//ND-oriented austenitic grains, respectively, where nano-indentation and micro-pillar compression testswere performed. Colors in the IPF maps indicates the crystal directions parallel to the viewing plane. The white lines in the IPF maps indicate the recrystallization twin boundaries.

Table 1Maximum Schmid factors for perfect dislocation glide, dissociated dislocation glide,and mechanical twinning.

Compression axis Schmid factors

Perfect dislocation Partialdislocations

Twinning partials

[001] 0.41 Trailing Leading 0.478 systems 0.23 0.47 4 systems

[111] 0.27 Trailing Leading 0.166 systems 0.31 0.16 6 systems

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123114

P ¼ 43Er

ffiffiffiffiffiffiffirh3

p(6)

P is the applied load, h is the corresponding indentation depth,and r is the radius of the indenter tip. The radius of the indenter tip

Fig. 2. Nano-indentation force-displacement curve for (a) the [001]//ND-oriented austenitic

Table 2Comparison of the results of the present work and literature data for the reduced modvarious steels, obtained by the nano-indentation test. Note that when no value for Es wa

Alloy Phase Orientation

High Mn TWIP Steel [20] Austenite [001][101][111]

316 Stainless Steel [21] Austenite e

Low C Medium Mn TRIP Steel [22] Austenite [001]Medium Mn TRIP Steel [23] Austenite e

Ferrite e

Interstitial Free Steel [24] Ferrite e

Medium Mn Steel [Present Study] Austenite [001][111]

used in the present study was determined to be 175 nm by the tipcalibration method. The elastic portion of the load-displacementpredicted by the Hertzian solution is indicated in Fig. 2(a) and (b).

In both nano-indentation force-displacement curves shown inFig. 2, the first pop-in loads, Pc, i.e. the load at which the firstdisplacement burst occurs, were noticeably different. The Pc in the[001] direction was 49.0 mN at a displacement of 5.9 nm. The Pc inthe [111] direction was much higher, 97.9 mN at a displacement of8.5 nm. The maximum elastic shear stress underneath a sphericalindenter, tmax can be calculated using the following relation [27]:

tmax ¼ 0:30

6PcE2rp3r2

!1=3

(7)

The calculated tmax for the [001] direction (8.0 GPa) was slightlylower than that for the [111] direction (10.1 GPa). It is interesting to

grain and (b) the [111]//ND-oriented austenitic grain in the mediumMn austenitic steel.

ulus (Er), Young's modulus (Es), hardness (H), and maximum shear stress (tmax) fors provided, Es was derived using Eq. (3), assuming a Poisson ratio of 0.3.

Er(GPa) Es(GPa) H(GPa) tmax(GPa)

156.2 164.6 e 2.38 ± 0.16162.6 172.4 e 2.59 ± 0.31170.8 182.6 e 2.79 ± 0.30220.0 247.7 2.7 e

e e e 9.4e 187.0 7.7 7.6e 187.0 3.4 e

210.0 234.0 e 11.3210.4 234.5 7.6 8.0236.7 271.5 8.5 10.1

Fig. 3. Deformation-induced twinning in the medium Mn austenitic grain indentation-tested along the [001] direction. (a) Cross-sectional TEM micrograph of the deformedaustenite grain, viewed along the [110] zone axis (inset: corresponding diffraction pattern showing the presence of reflections due to the twinned austenite). (b) DF image of themechanical twins in the area indicated by the rectangle in (a). (c) HR lattice image of the mechanical twins (inset: corresponding Fast Fourier transform).

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123 115

note that when the indentationwas tested along the [001] directioninwhich the twinning is expected to be favored, pop-ins weremorefrequently observed, as indicated by the arrows in Fig. 2(a). An

Fig. 4. Perfect dislocation slip in the medium Mn austenitic grain indentation-tested along tviewed along the ½110� axis (inset: selected area diffraction pattern of the deformed austeniteaustenite. TEM BF micrographs of the area of the deformed austenite, indicated by the rectang ¼ ð111Þ. Note that the straight features in (b) and (c) are not associated with stacking faulThe schematics in (c) to (d) illustrate the tetrahedron showing the orientation of {111} slip

indentation behavior with multiple pop-ins is clearly associatedwith deformation twinning.

In order to investigate the deformation mechanisms, TEM

he [111] direction. (a) Cross-sectional TEM micrograph of the deformed austenite grain,). (b) HR lattice image of dislocations on the ð111Þ and (111) slip planes in the deformedgle in (a) for three different two-beam conditions: (c) g ¼ (002), (d) g ¼ ð111Þ, and (e)

ts. The contrast is due to the perfect dislocations gliding on ð111Þ and (111) slip planes.planes, and the perfect dislocations on ð111Þ and (111) slip planes.

Fig. 5. (a) Compression stress-strain curve for a [001]//ND-oriented micro-pillar (see also Supplementary Movie 1). The large strain bursts in (a) are due to the rapid twin growth.(b) Compression stress-strain curve for a [111]//ND-oriented micro-pillar (see also Supplementary Movie 2). The small strain bursts in (b) are associated with perfect dislocation slip.

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123116

samples were prepared from the deformed region of the nano-indentation test by means of the FIB lift-out technique. Fig. 3shows TEM micrographs of deformation-induced nano-sizedtwins in the [001]//ND-oriented austenite grain. The viewing di-rection was along the [110] zone axis. The TEM sample consisted ofa single austenite grain of which the surface normal was orientedclose to the [001] direction. The FIB milling during the TEM samplepreparation did not induce phase transformation. The twin spots inthe diffraction pattern [110] zone axis clearly indicate the presenceof deformation twins. Streak-like patterns due to the nano-sizedtwins and the stacking faults were also visible in the diffractionpattern. Both the dark-field (DF) image and the high resolution (HR)lattice image show twins, with a thickness in the range of 4e10 nm.Although four equivalent twinning systems were in principlepossible according to the Schmid factor analysis of Table 1, only onetwin system, ð111Þ½112�, was activated.

Fig. 4 shows TEM micrographs of the [111]//ND-orientedaustenite grain after the nano-indentation test. The viewing di-rection was along the ½110� zone axis. There was no indication forthe formation of deformation twins in both the diffraction pattern(Fig. 4(a)) and the HR lattice image (Fig. 4(b)). Straight features,similar in appearance to stacking faults, were present along theð111Þ and (111) slip planes, as shown in the HR lattice image ofFig. 4(b) and the TEM two-beam bright field (BF) image shown inFig. 4(c). In order to characterize those defects, TEM observationswere carried out both near the ½211� zone axis and the ½211� zoneaxis. In these conditions, as shown in Fig. 4(d) and (e), only perfectdislocation contrast was visible, without any stacking fault contrast.This indicates that the defects formed on two different {111} slipplanes were perfect dislocations. The straight features observed inthe ½110� zone axis were therefore due to perfect dislocations on theð111Þ and (111) slip planes. The dislocation density was higher onthe ð111Þ plane as compared to the (111) plane (Fig. 4(b) and (c)).Slip on the (111) plane, which is perpendicular to the loading axis,should in principle not be activated. Some dislocations were

Table 3Comparison of the results of the present work and literature data obtained by the micromodulus, Eadj: adjusted Young's modulus obtained by Eq. (14)). Note that the micro-pilla

Alloy Compressionaxis

Deformationmechanism

YS(GPa)

t

(

High Mn TWIP Steel [37] [0 0 1] Twin 0.67 0High Mn TWIP Steel [29] [5 4 22] Twin 0.91 0

[16 6 23] Slip 1.01 0Medium Mn Steel [Present

Study][0 0 1] Twin 0.84 0[1 1 1] Slip 0.63 0

however also observed on the (111) planes. This may be due to thefact that the loading axis was not precisely parallel to the [111]direction.

3.2. Micro-pillar compression tests

Fig. 5 shows the crystallographic orientation dependence of themicro-pillar compression results for the [001] and [111] directions.The micro-pillar test results are summarized in Table 3. There weredistinct differences in the plastic deformation behaviors of the[001]- and [111]-oriented micro-pillars. Large-scale strain burstsfollowed by large stress drop were observed for the [001]-orientedmicro-pillar, in which the twinning is expected to be favored(Fig. 5(a) and SupplementaryMovie 1). The average strain burst sizewas 0.78 ± 0.56%. The accumulated burst strain was 4.7%. Incontrast, the [111]-oriented micro-pillar, in which the dislocationglide is favored, exhibited frequent, small-scale strain bursts(Fig. 5(b) and Supplementary Movie 2). The average strain burstsize and the accumulated burst strain were 0.29 ± 0.10% and 2.9%,respectively. The [111]-oriented micro-pillar had a smooth transi-tion from elastic to plastic flow and a steady rate of work hardening,as compared to the [001]-oriented micro-pillar. The stress for yieldonset in the [001]-oriented micro-pillar was slightly higher than inthe [111]-oriented single crystal. However, the stress levels at themaximum strain of approximately 9% (i.e. a displacement of150 nm) was similar for both cases, 2.2 GPa.

Supplementary video related to this article can be found athttp://dx.doi.org/10.1016/j.actamat.2017.06.014.

The yield stress was clearly larger for the [001]-oriented micro-pillar (0.84 GPa) as compared to the [111]-oriented micro-pillar(0.64 GPa). Here, the yield stress was considered to be the stress atwhich the elastic-plastic transition is observed, as indicated by thearrow in Fig. 5(a) and (b). Using the Schmid factor, m, for eachcompression direction (Table 1), the critical resolved shear stress,tcrss was calculated using the equation tcrss¼ m·sys where the sys is

-pillar compression (YS: yield stress, tcrss: critical resolved shear stress, Es: Young'sr diameter was 500e600 nm.

crss

GPa)Ave. strain burst size (%) Accumulated burst strain

(%)Es(GPa)

Eadj(GPa)

.32 e e e e

.44 e e e e

.47 e e e e

.40 0.78 ± 0.56 4.7 82 91.3

.17 0.29 ± 0.10 2.9 102 115.0

Fig. 6. SEM micrograph and cross-sectional TEM micrographs of medium Mn austenitic steel micro-pillars compressed along the [001] direction. (a) SEM image of the deformedmicro-pillar viewed at an angle of 52� to the normal of the pillar. The tetrahedron in (a) shows the orientation of the primary (111) slip plane and the leading twinning partialdislocation with Burgers vector a/6½112�. (b) TEM BF image of the deformed micro-pillar. (c) DF image of the ð111Þ½112�-type mechanical twins (inset: corresponding diffractionpattern showing the presence of reflections due to the twinned austenite). (d) TEM DF image of the ð111Þ½112�-type mechanical twin (inset: corresponding diffraction pattern). (e)TEM DF image showing the ferrite layer formed at the surface of the micro-pillar during the FIB micro-pillar preparation. The inset shows the corresponding diffraction pattern. Itindicates the K-S orientation relationship between the parent austenite (g) and the transformed ferrite (a).

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123 117

the yield stress. The tcrss for the [001] direction (0.40 GPa) wasmuch higher than tcrss for the [111] direction (0.17 GPa) (Table 3).Young's modulus of the specimen, Es, was estimated from thestiffness of the elastic portion of the compression stress-straincurves. The Es for the [001] direction (82 GPa) was lower thanthat for the [111] direction (102 GPa).

Fig. 6 shows a [001]-oriented austenite grain micro-pillardeformed in compression to a strain of 10%. Clear slip traces forthe (111) slip plane were visible in the SEM image of the deformedcrystal surface (Fig. 6(a)). These traces are due to the activation ofthe ð111Þ½112� twin system. The primary twin system activatedduring themicro-pillar compression, ð111Þ½112�, was different fromthe one activated during the nano-indentation, ð111Þ½112�, despitethe fact that both tests were carried out in the same austeniticgrain. In order to observe the deformation twins from the edge-ondirection, the TEM specimen of the deformed micro-pillar wastaken in a manner that the electron beam was aligned parallel tothe ½110� zone axis. Note that the TEM specimen taken after thenano-indentation was viewed along a different zone axis, the [110]zone axis (Fig. 3). Fig. 6(b) shows a TEM BF image of the deformedmicro-pillar. Fig. 6(c) and (d) show TEM DF images of the nano-sized twins in the deformed micro-pillar. Three relatively thicktwins of the ð111Þ½112� twin system were observed (Fig. 6(c)). Asingle ð111Þ½112� twin was also activated (Fig. 6(d)). The thicknessof the twins was in the range of 10e32 nm. A low density ofstacking faults and dislocations was also observed in the upper partof the deformed micro-pillar. It is important to note that the FIBmilling during the micro-pillar preparation induced the austenite-to-ferrite transformation at the surface of the micro-pillar. Fig. 6(e)shows that the deformed micro-pillar was covered by a layer offerrite. The average thicknesses of the surface transformed to ferritewas 34.6 ± 6.5 nm. It is clear that this ferrite layer was formedduring the FIB micro-pillar fabrication and not during the micro-pillar compression because no ferrite layer was observed in theTEM samples of the austenite grains after nano-indentation. Theinset in Fig. 6(e) shows a selected area diffraction pattern indicating

the Kurdjumov-Sachs (K-S) orientation relationship between theaustenitic matrix and the surface ferrite layer. The origin of thesurface transformation to ferrite will be discussed in section 4.4.

Fig. 7 shows a [111]-oriented austenite grain micro-pillardeformed to 10% engineering strain in compression. Slip tracesparallel to the ð111Þ slip plane are visible at the deformed crystalsurface (Fig. 6(a)). A very high density of dislocations with theprimary slip system of the ð111Þ planes was observed in thedeformed micro-pillar (Fig. 7(b)). Similarly to the observationsmade for the TEM sample of the indentation-tested, [111]-orientedgrain (Fig. 4), straight features corresponding to the perfect dislo-cations on the ð111Þ slip planes were frequently observed(Fig. 7(c)). Their Burgers vector was either a/2[101] or a/2[011].These perfect dislocations were out of contrast for the g ¼ ð111Þtwo-beam condition (Fig. 7(d)), indicating that the product of theirBurgers vector and the g vector was zero, i.e. [101]$ð111Þ ¼ 0 and[011]$ð111Þ ¼ 0. Together with the activation of the primary slip onthe (111) slip plane, a high density of tangled dislocations, which donot appear straight when viewed along the ½110� zone axis, was alsoobserved. Fig. 7(e) and (f) show that the FIB-induced austenite toferrite transformation was also observed in the [111]-oriented mi-cro-pillar. The deformed micro-pillar was surrounded by a layer ofthe transformed ferrite with a thickness of 22.3 ± 2.3 nm. The insetin Fig. 7(e) shows a selected area diffraction pattern indicating theNishiyama-Wasserman (N-W) orientation relationship betweenthe austenitic parent matrix and the transformed ferrite surface.

4. Discussion

4.1. Orientation dependent deformation behavior

The motion of isolated Shockley partial dislocations with thesame a/6<112>-type Burgers vector on successive {111}-typeplanes is the origin of overlapping stacking faults and deformation-induced twins. The SFE plays an important role in deformationtwinning. The room temperature SFE of the Fe-1.2%C-7.0%Mn used

Fig. 7. SEM micrograph and cross-sectional TEM micrographs of medium Mn austenitic steel micro-pillars compressed along the [111] direction. (a) SEM image of the deformedmicro-pillar viewed at an angle of 52� to the normal of the pillar. The tetrahedron in (a) indicates the orientation of the primary ð111Þ slip plane. The deformation is by the motion ofperfect dislocations with Burgers vector a/2[101] or a/2[011]. (b) TEM BF image of the deformed micro-pillar containing a high density of perfect dislocations. (c) Two-beam(g ¼ ð111Þ) TEM BF micrographs of the area indicated by the rectangle in (b), showing perfect dislocations on the ð111Þ slip plane. (d) Two-beam (g ¼ ð111Þ) TEM BF micro-graphs of the same region. The perfect dislocations on ð111Þ slip plane are out of contrast for g ¼ ð111Þ. (e) TEM DF image taken using the ð011Þa0 reflection, showing the ferrite layerformed at the surface of the micro-pillar during the FIB micro-pillar preparation (inset: corresponding diffraction pattern showing the N-W orientation relationship between theparent austenite (g) and the transformed ferrite (a)). (f) TEM DF image taken using the (002)g reflection, showing the austenite matrix in the deformed micro-pillar.

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123118

in the present study is estimated to be 34 mJ m�2, 28 mJ m�2 and28 mJ m�2 based on the thermodynamic models proposed byCurtze and Kuokkala [6], Saeed-Akbari et al. [10], and Dumay et al.[28], respectively. An g/ε interface energy of 10mJm�2 was used forthe calculation of the SFE. As reported in Refs. [9,11], the TWIP effectis the predominant plasticity-enhancing deformation mode whenthe SFE is in the range of 15e45 mJ m�2. The previous reports ondeformation behavior of austenitic steels [9,11] are consistent withthe observations of the present study that twinning or dislocationgliding is activated during the deformation of the medium Mnaustenitic steel and that no austenite to a'- or ε-martensite trans-formation is observed. It should be pointed out that the modelsdeveloped for the SFE calculation were derived for steels with arelatively high Mn content ranging from 10 to 35 wt% and thatprecise experimental values of the SFE for medium Mn austeniticsteels are unknown.

It was found that the deformation mode during the nano-indentation and the micro-pillar compression results wasstrongly influenced by the grain orientation. Deformation twinningwas the preferred deformation mode in the [001]//ND-orientedsingle grain under compression. Perfect dislocation slip was thedominant deformationmechanism in the [111]//ND-oriented singlegrain deformed in compression. These observations are consistentwith the results of previous studies [17,20]. In particular, Karamenet al. [17] studied the stress-strain behavior of the bulk sample ofhigh Mn Hadfield steel (Fe-12.34Mn-1.03C in wt%) single crystalsfor selected crystallographic orientations (½111�, [001] and ½123�)under tension and compression. They reported that twinning wasthe dominant deformation mechanism in [0 0 1] crystals subjectedto compression, while multiple slip was determined to be thedominant deformationmechanism in ½111� crystals in compression.

Similar observations have been reported for the small-scalecompression tests. Kang et al. [20] reported an evidence of defor-mation twinning in a high Mn TWIP steel (Fe-0.58C-17.5Mn-1.5Alin wt%) when the nano-indentation was applied along the [001]direction of the crystal. They reported that no mechanical twinwasobserved in the grain indention-tested along the [111] orientation.

Choi et al. [29] reported similar trends of the orientation-dependent deformation behavior in high Mn TWIP steel (Fe-0.6C-22Mn in wt%) micro-pillars.

The orientation dependent deformation behavior observed inthe present study can be explained by a Schmid factor analysis, asshown by Karaman et al. [17]. Table 1 lists the calculated Schmidfactors for slip and deformation twinning as a function of thecompressive loading directions. When the compressive loading isalong the [001] direction, the Schmid factor for the leading partialdislocation (mlp) is larger than that for the trailing partial disloca-tion (mtp). This effectively promotes the creation of wide stackingfaults and twin formation. The mlp is also larger than the Schmidfactor for the perfect dislocation slip (mslip). In contrast, when theloading axis was parallel to the [111] direction, the twin formationis suppressed since mlp < mtp. If it is assumed that the criticalresolved shear stresses, tcrss, for both twinning and slip areapproximately equal, further prediction of deformation behavior ispossible, as shown by Gutierrez-Urrutia et al. [18]. In compressionalong the [001] direction, deformation twinning dominatesbecause the Schmid factor for the twin system is higher than thatfor slip, i.e.mlp >mslip. In compression along the [111] direction, theperfect dislocation slip should be dominated as mlp < mslip. How-ever, in principle, tcrss for twinning and slip are not equal and thus acareful analysis is needed. tcrss can be derived by several ap-proaches, as will be discussed in section 4.2.

In general, the activated slip/twin systems are those predictedby Schmid's law. If the compressive axis is precisely along the [111]direction, there are six slip systems with an equivalent Schmidfactor (Table 1). However, a slight misorientation of the compres-sive axis from the exact [111] direction causes only one specific slipsystem to be activated. Table 4 lists the calculated maximumSchmid factor for slip/twin systems considering the actual crystalorientation determined by EBSD and the experimentally observedtwin/slip systems. When the compressive axis was along the [28 3227] direction, the highest Schmid factor was found for theð111Þ½011� slip system. This slip system corresponds to the exper-imentally observed slip system. When the compressive axis was

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123 119

along the [1 1 50] direction, the highest Schmid factor was for theð111Þ½112� twin system. In the case of the micro-pillar test results,the predicted twin system was consistent with the observed twinsystem. A different twin system, ð111Þ½112�, was however activatedin the case of the nano-indentation results. The deviation of Schmidfactor analysis for twinning in the case of the nano-indentationresults could be due to the following reasons: (1) the locationwhere the maximum resolved shear stress is reached can be awayfrom the loading axis in nano-indentation test, as it is governed byindenter shape, elastic anisotropy, and operative slip systems, asreported by Li et al. [30]. (2) It is also possible that the alignmentbetween the loading axis and the specimen was not perfect.

The multiple pop-ins recorded during the nano-indentation inthe [001] direction (Fig. 2(a)) were interpreted as being due todeformation twinning. In the micro-pillar compression tests, theorientation-dependence of mechanical behavior was more clearlyshown. The large strain bursts during the compression of the [001]-oriented micro-pillar were clearly due to the rapid twin growth.These strain bursts plasticity are different from those appearing inmicro-pillar associated with the continuous operation of a fewdislocation sources [31e33], or the destruction of jammed dislo-cation configurations [34,35]. Similar strain bursts due to the twingrowth in micro-pillars of a high Mn TWIP steel [36,37] and asingle-crystal Ti alloy (HCP) [38] have been reported. The per-centage of the plastic strain contributed by deformation twinning,εt, was estimated by means of the following equation [36]:

εt ¼ T$gt$cosaH

(8)

where T is the twin thickness, gt is the twinning shear strain equalto

ffiffiffi2

p=2, a is the angle between the pillar axis and the activated

twinning direction, and H is the height of the pillar before thecompression. The value of εt was 3.3%, which is smaller than theaccumulated burst strain of 4.7% (Table 3). This difference may bedue to the exclusion of the plasticity contributed by the dislocationglide and the formation of stacking faults. When the loading axiswas along the [111] direction, i.e. when a slip-dominant deforma-tion was favored, the accumulated burst strain was 2.9%. This wasmuch smaller as compared to [001] direction, despite the fact thatgreater number of strain bursts took place during the micro-pillarcompression. The observations offer a strong support for the hy-pothesis that deformation twinning is activated as a plasticityenhancing mechanism during the deformation of medium Mnsteels.

Wu et al. [39] studied the mechanism of deformation twinningin a high Mn TWIP steel (Fe-0.6C-22.0Mn inwt%) micro-pillar. Theyreported that twin nucleation and growth occur via the emission ofpartial dislocations from the pillar surface on consecutive {111}planes. Their result is consistent with the results of the presentstudy (Fig. 6) that the twin nucleation in micro-pillars invariablyoccurred near the pillar surfaces. The twin nucleation is possiblyaided by the defects created during the ion-milling step. On thisbasis, Liang et al. [37] have proposed a mechanism for the forma-tion of twins in a high Mn TWIP steel micro-pillar. The mechanismproposed by Liang et al. [37] involves the dissociation of radiation-

Table 4Orientation dependence of the Schmid factor and the primary slip/twin systems for the

Compression Axis Actual axis Maximum Schmid factors and the s

Perfect dislocation slip

[001] [1 1 50] 0.416

ð111Þ½101� andð111Þ½011�[111] [28 32 27] 0.313

ð111Þ½011�

induced a/3<111> vacancy Frank loops into sessile a/6<110> dis-locations and twin-emitting a/6<112> Shockley partials on a con-jugate {111} plane.

In the present study, the thickness of the deformation twins inthe deformed micro-pillar ranged from 10 to 32 nm (Fig. 6).Considering the planar spacing of {111} plane in austenite, the glideof a/6<112>-type twinning partial dislocations on the approxi-mately 50e150 successive {111} planes is required to form theobserved deformation twins. Twin growth can occur in anavalanche-like manner, e.g. via the rapid and coherent partialdislocation emission adjacent to twin boundary as reported byLiang and Huang [36]. Liang and Huang observed a large strainburst due to twin growth in a high Mn TWIP steel (Fe-0.6C-22.0Mnin wt%) micro-pillar, similar to the one observed in the presentstudy (Fig. 5(a)). They reported that the average rate of partialdislocation emission was as high as ~3300 s�1. Yu et al. [38] alsoobserved a similar burst-like twinning behavior in a Ti alloy (HCP)micro-pillar. They described deformation twinning as a perfectlycoherent “stimulated slip” phenomenon, which is in contrast to theless coherent “spontaneous slip” of ordinary dislocation plasticity.Their view was based on the fact that deformation twinning ischaracterized by perfectly correlated layer-by-layer shearingbecause all twinning partial dislocations on the adjacent parallelplanes must have the same Burgers vector. This may explain whythe deformation twinning causes large strain bursts. On the otherhand, inelastic shear activities are randomly dispersed among slipplanes in perfect dislocation slip process [38].

4.2. Critical resolved shear stress for twinning and dislocation slip

The preference for twinning or slip can be explained bycomparing the critical resolved shear stress for a perfect disloca-tion, tperfect, and the critical resolved shear stress to initiate theShockley partial twinning dislocation tpartial to create a deformationtwin. With an approximation of the source size equal to the grainsize, tperfect required to activate the dislocation source is given byclassical dislocation theory [40],

tperfect ¼2aGbperfect

D(9)

where G is shear modulus, bperfect is the magnitude of the Burgersvectors of the perfect dislocation, D is grain size, and a is a constantdescribing the character of the dislocation (a ¼ 0.5 and 1.5 for edgeand screw dislocations, respectively). tpartial can be obtained by anequation similar to Eq. (9), which takes into account the formationof the stacking fault. The stress for the twin growth from existingtwin nuclei is determined by a balance between the repulsive forcesof two Shockley partials and the attractive force of the SFE and it isgiven by g/bpartial. This leads to the following equation for tpartial[40]:

tpartial ¼2aGbpartial

Dþ g

bpartial(10)

medium Mn austenitic steel.

lip/twin system Observed primary slip/twin system

Twinning leading partials Nano-indentation Micro-pillar

0.480

ð111Þ½112�ð111Þ½112� twin ð111Þ½112� twin

0.193ð111Þ½121�

ð111Þ½101� or ð111Þ½011�perfect dislocation glide

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123120

where g is the SFE, and bpartial is the magnitude of the Burgersvector of the partial dislocation. Equations similar to Eq. (10) havebeen proposed [18,41,42]. Application of Eq. (9) to the Fe-1.2%C-7.0%Mn steel used in the present study yields tperfect ¼69 MPa (G ¼ 81 GPa, bperfect ¼ 0.256 nm, az 1, and D¼ 0.6 mm). Eq.(10) yields tpartial ¼ 229 MPa (g ¼ 28 mJ m�2, bpartial ¼ 0.148 nm). Inboth cases, the grain size was taken equal to the micro-pillardiameter. These estimates reveal that tpartial for deformationtwinning should be considerably larger than tperfect for the perfectdislocation glide. This trend is in an agreement with the results ofthe present study that the critical resolved shear stress, tcrss, for the[001]-orientedmicro-pillar, which is equivalent to tpartial, wasmorethan 2 times tperfect for the [111]-oriented micro-pillar (Table 3).

The homogeneous dislocation nucleation model was initiallyproposed by Hirth and Lothe [43], and extended by Michalske andHouston [44]. In the model, the energies for the activation of aperfect dislocation Frank-Read source (Eperfect) and a partial dislo-cation twin source (Epartial) in the presence of the applied shearstress, are expressed as [43,44]:

Eperfect ¼ �pR2tbperfect þ12Gb2perfectR

2� n

2ð1� nÞ�ln

4Rr0

� 2�

(11)

and

Epartial ¼ �pR2tbpartial þ12Gb2partialR

2� n

2ð1� nÞ�ln

4Rr0

� 2�þ pR2g

(12)

where, R is a dislocation loop radius, t is the applied shear stress, nis the Poisson's ratio, and r0 is the core radius of the dislocation,which is assumed to be equal to the Burgers vector. The maximumenergy barrier occurs at a critical dislocation loop size corre-sponding to the condition, vE=vRjR¼Rc

¼ 0. With this condition, thecritical dislocation loop size and the maximum energy barrier canbe calculated for a given applied shear stress. Fig. 8(a) and (b) showthe applied shear stress dependence of the critical dislocation loopsize and critical nucleation energy. The tcrss measured for themicro-pillars in this work is indicated in Fig. 8(a) and (b). It isshown that, as the applied shear stress increases, the critical loopsize and the nucleation energy of partial and perfect dislocationloops become comparable. Fig. 8(b) shows that the creation of thepartial dislocation loop is energetically favorable when the appliedshear stress > 270 MPa. When the applied shear stress is equal tothe tcrss for the [001]-oriented micro-pillar (400 MPa), Epartial islower than Eperfect, as indicated in Fig. 8(b). In this condition, thenucleation of partial dislocation loop is energetically favorable.When the applied shear stress equals the tcrss for the [111]-orientedmicro-pillar (170 MPa), Eperfect is significantly lower than Epartial(Fig. 8(b)), implying that the perfect dislocation glide is favorable. Inother words, the preference for twinning in a [001]-orientedmicro-pillar and the preference for perfect dislocation glide in a [111]-oriented micro-pillar are explained in terms of the energy of thedislocation nucleation.

In another approach to estimate the twinning stress, anempirical Hall-Petch-type relation has been used to describe thegrain-size dependence of the twinning stress [45]:

s ¼ s0 þKTffiffiffiffiD

p (13)

where s0 is the stress at which yielding occurs at very large grainsizes, and KT is the Hall-Petch slope for twinning. KT has been re-ported to be consistently greater than KS, the Hall-Petch slope fordeformation controlled by slip [45]. This implies that the grain size

impact on twinning is more than its impact on slip. In FCC Cu, KT istwo times KS [46]. Similar observations have been made for BCC[47e49] and HCP alloys [50]. A similar trend applies to micro-pillars. Yu et al. [38] studied the size-dependent deformationtwinning behavior in HCP Ti alloy micro-pillars. They reported thatthe deformation twinning was entirely replaced by dislocationplasticity when themicro-pillar diameter was less than onemicron.They explained that this is due to the fact that the stress requiredfor deformation twinning increases drastically with decreasingsample size. If this argument is applied to the results of the presentstudy, in the case of the micro-pillar compression test, tpartial islikely to be much larger than tperfect because the stress required fordeformation twinning is drastically increased in a small sample.

It should be mentioned that Choi et al. [29] reported the resultsin contradiction with the present results. They reported that thevalues of the tcrss of micro-pillars oriented for deformation twin-ning and micro-pillars oriented for dislocation glide were similar.They suggested that the elasto-plastic transition observed for theTWIP steel micro-pillars was due to dislocation glide, independentof the fact that the orientation may favor twinning. The reason forthe apparent contradiction is not clear, but the discrepancy mayhave originated partly from difference in the SFE. The SFE of theTWIP steel used in the work of Choi et al. [29] (23 mJ m�2) wassmaller than that of the steel used in the present work(~28 mJ m�2). Choi et al. [29] observed ε-martensite in thedeformed TWIP steel micro-pillar, suggesting the SFE was clearlylower as compared to the steel used for the present work. If the SFEis smaller, the stress required for the twin growth decreases,leading to a decrease of tpartial, according to Eq. (10).

The value for the tcrss obtained by the micro-pillar compressiontests (�0.40 GPa) was considerably lower than the value for thetmax obtained by nano-indentation (~10 GPa z G/8). Note that tcrssand tmax obtained in the present study are not physically same. tmaxis themaximum shear stress, which lies along the loading direction,i.e. it is not that resolved in specific slip direction. Bei et al. [51] alsoreported that the tcrss of a Mo alloy single crystal obtained frommicro pillar tests (zG/26) was a factor of 3 lower than tmax ob-tained from the nano-indentation tests (zG/8). They argued thatthis difference was due to the different stresses required to ho-mogeneously nucleate a full dislocation loop within a bulk sampleduring nano-indentation and to heterogeneously nucleate half orquarter dislocation loops at the free surfaces and edges of micro-pillars. Liang et al. [37] suggested that the ion-radiation-induceddefects due to the FIB micro-pillar fabrication cause a local vol-ume change in the damaged subsurface layer and induces sub-stantial internal stress into this layer due to the constraint fromthose parts of the micro-pillar which are not affected by the ion-beam damage. They argue that the ion-radiation-induced internalstresses present in the damaged subsurface layer of micro-pillarsfacilitate the dissociation of vacancy Frank loops which act astwinning nuclei. In the present case, the phase transformation atthe surface of the pillar created both surface internal stresses, dueto the lattice expansion associated with the austenite-to-ferritetransformation, and a high density of the transformationdislocations.

4.3. Young's modulus obtained by nano-indentation and micro-pillar test

The measured values of Young's modulus were significantlydifferent for nano-indentation and micro-pillar compression. Forexample, Es for the [001] direction was measured to be 234.5 GPaand 82 GPa by nano-indentation and micro-pillar compression,respectively. Pharr [16] reported that when the pile-up whichforms around the indentation impression is prevalent in the

Fig. 8. Applied shear stress dependence of (a) the critical size with the effective source size, and (b) the critical energy of nucleation of a perfect dislocation half loop (Eperfect) and apartial dislocation half loop (Epartial). The partial dislocation loop is unstable at an applied shear stress lower than 270 MPa. The experimentally obtained critical resolved shearstresses, tcrss, for the [100]- and [111]-oriented micropillar are indicated in (b). For t½111�crss , Eperfect < Epartial. For t

½100�crss , Eperfect>Epartial.

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123 121

material, the Oliver and Pharr method for the nano-indentationdata analysis may lead to an overestimation of the modulus by asmuch as 50%. He reported that pile-ups in the material will make asignificant contribution when hf/hmax > 0.7. Here, hf is the finaldisplacement after complete unloading and hmax is the maximumdepth of penetration. In the present work, the value of Es obtainedby the nano-indentation may have been overestimated as hf/hmaxwas approximately 0.71.

On the other hand, the values of Young's modulus obtained bymicro-pillar compression were considerably lower than the valueof 159 GPa reported for an austenitic Fe-1.5 wt% C [52]. Young'smodulus is 174.7 GPa in the bulk sample of highMnTWIP steel [53].The underestimation of Young's modulus in micro-pillarcompression has been reported [54e56]. Greer et al. [55] re-ported that this anomalously soft initial response may partially bedue to the compliance of the substrate supporting the micro-pillar.Choi et al. [57] reported that the underestimation of Young'smodulus is due to the substrate deformation and that this effectdecreases with increasing values of aspect ratio. They provided thefollowing equation, derived based on the Sneddon correction [58]:

1Eadj

¼ 1E� 1E�

p

4dL0

(14)

where Eadj is the adjusted Young's modulus, E is the apparentmodulus measured by the micro-pillar test, E* is the reduced(indentation) modulus, d is micro-pillar diameter, and L0 is theinitial micro-pillar height. The values of the Eadj are listed in Table 3.These values are still too small. It is possible, as noted by Dimiduket al. [56], that the misalignment between the loading platen andthe specimen leads to the gradual elastic loading of the specimenand anomalous soft initial response of the specimen.

4.4. FIB-induced FCC-to-BCC transformation

Knipling et al. [59] reported that FIB milling resulted in theaustenite-to-ferrite transformation in commercial stainless steels.Basa et al. [60] reported a very similar FIB-induced phase trans-formation in the highly stable austenite phase of a super duplexstainless steel. They concluded that the product BCC phase was notformed by a martensitic transformation based on the observationthat the invariant plane strain and plastic deformation, which areexpected in the case of a martensitic transformation, were notobserved. They argued that the austenite-to-ferrite transformationwas chemically-induced, i.e. it was due to the local enrichment of

gallium, which is a strong ferrite stabilizer. Babu et al. [61] alsoobserved the FIB-induced austenite-to-ferrite transformation in316L austenitic stainless steel. They concluded that in addition tothe local gallium enrichment, the strain associated with the ionimplantation also promoted the phase transformation during theFIB irradiation.

The FIB-induced austenite-to-ferrite transformation was alsoobserved in the present work. A BCC ferrite layer with a thickness inthe range of 20 nme30 nmwas formed at the surface of the micro-pillars during the FIB pillar microfabrication (Figs. 6(e) and 7(e-f)).Selected area diffraction patterns indicated both the K-S and N-Worientation relationship between austenite and ferrite. The productBCC layer is very likely not due to a martensitic transformation asthe characteristic features of high carbon martensite were notobserved by TEM analysis. This indicates that this transformationwas mainly chemically-driven, as suggested by Basa et al. [60].

The mechanism of the phase transformation during the FIBmicrofabrication is different depending on the austenite stability. Inthe case of the highly stable austenite in the mediumMn steel usedin the present study, the transformation was mainly chemically-driven. That is, during the gallium ion implantation, the composi-tion of austenite was gradually changed and the austenite becamemetastable. Once a critical Ga content was reached, the austenitespontaneously transformed to ferrite. On the other hand, if theaustenite is metastable, the phase transformation can be triggeredmainly by the stress generated by the ion implantation. Forexample, the mechanically-induced martensitic transformationwas observed during the FIB milling of the metastable retainedaustenite phase in a quenching and partitioning (Q&P) processedsteel (Supplementary materials). Invariant plane-strain surfacereliefs associated with the martensitic transformation were clearlyobserved in the retained austenite phase immediately after a singlescan of FIB (Fig. S1). The structure of the BCC product phase wasclearly identified by the EBSD analysis. Since this transformationtook place after a single FIB scan with dwell time of microsecondtime scale, the FIB-induced heating was negligible, i.e. phasetransformation by diffusional mechanism should not be consid-ered. As the contents of carbon and Mn, austenite stabilizers, wererelatively low in the Q&P processed steel (Fe-0.4C-4.0Mn-1.6Si-1.0Cr inwt%), the stability of the retained austenite in this steel wasclearlymuch lower as compared to the austenite in themediumMnsteel (Fe-1.2C-7.0Mn in wt%) used in the present study. The presentresults therefore suggest that the FIB-induced phase trans-formation should be seen as chemically-induced austenite-to-ferrite phase transformation when the austenite is highly stable.

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123122

For an austenite phase with low stability, martensitic trans-formation due to the strain induced by the ion implantation willtake place.

A large density of dislocations is created during the austenite-to-martensite (g/a0) transformation. Three families of disloca-tions are involved: (a) a/2<110>g-type transformation dislocationsassociated with the structural ledges, (b) a/2 <111>a'-type screwmisfit dislocations at the austenite-martensite interphase bound-ary, and (c) a/2<110>g-type dislocations in austenite, resulting fromthe deformation of the austenite during the martensite trans-formation. The dislocations in the austenite are subsequentlyinherited by the martensite. The dislocations in lath martensite aremainly screw dislocations [62e65]. Shibata et al. also [66] observedtwo sets of {112}<111>a0-type glissile screw dislocations on theinterphase boundary and suggested that such interfacial disloca-tions exist every third layer of the parallel close packed planes(111)g//(011)a0. Moritani et al. [63] compared the interphaseboundary structures of the lath martensite and the bainitic ferritelath. They observed similar accommodation screw dislocations inboth cases. This implies that, regardless of whether the trans-formed layer observed at the surface of the micro-pillar is ferrite ormartensite, the transformation creates a large density of screw-type dislocation segments. The micro-pillar tests are thereforeinfluenced by the preexisting dislocations at the surface.

5. Conclusions

The micro-scale plastic deformation behavior of austenitic Fe-1.2%C-7.0%Mn steel was studied in the [001] and [111] crystallo-graphic orientations by means of nano-indentation and in situmicro-pillar compression tests.

1. The SFE of the steel was estimated to be in the range of28e34 mJ m�2. The calculated SFE range is consistent with theobservations that, depending on the orientation, only twinningor dislocation gliding was activated during the deformation ofthe medium Mn austenitic steel.

2. Deformation twinning was the preferred deformation mode inthe [001]//ND-oriented single grain under compression. Perfectdislocation slip was the dominant deformation mechanism inthe [111]//ND-oriented single grain deformed in compression.The activated slip/twin systemswere, in general, consistent withthose predicted by the Schmid's law. The rapid twin growth ledto large strain bursts during the compression of the [001]//ND-oriented micro-pillar. The observations offer strong support forthe hypothesis that deformation twinning is a plasticityenhancing mechanism activated during the deformation ofmedium Mn steel.

3. The critical resolved shear stress for [001]//ND-oriented micro-pillar, tpartial, was almost two times that for [111]//ND-orientedmicro-pillar, tperfect. These results are in agreement with thecalculation of the critical resolved shear stress based on theclassical dislocation theory.

4. The FIB milling during the micro-pillar preparation induced theaustenite-to-ferrite transformation at the surface of the micro-pillar. It is suggested that the mechanism of the FIB-inducedphase transformation is different depending on the austenitestability. When the austenite is highly stable as the case for highcarbon medium Mn steel (Fe-1.2%C-7.0%Mn), the FIB-inducedphase transformation was mainly chemically-induced due tothe local enrichment of gallium, which is a strong ferrite stabi-lizer. For an austenite phasewith low stability, as the case for theaustenite in Q&P processed steel, martensitic transformationdue to the strain induced by the ion implantation took place.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.actamat.2017.06.014.

References

[1] J. Shi, X. Sun, M. Wang, W. Hui, H. Dong, W. Cao, Enhanced work-hardeningbehavior and mechanical properties in ultrafine-grained steels with large-fractioned metastable austenite, Scr. Mater 63 (8) (2010) 815e818.

[2] S. Lee, B.C. De Cooman, On the selection of the optimal intercritical annealingtemperature for medium Mn TRIP steel, Metall. Mater. Trans. A 44 (11) (2013)5018e5024.

[3] R. Zhang, W.Q. Cao, Z.J. Peng, J. Shi, H. Dong, C.X. Huang, Intercritical rollinginduced ultrafine microstructure and excellent mechanical properties of themedium-Mn steel, Mater. Sci. Eng. A 583 (2013) 84e88.

[4] J.-K. Kim, B.C. De Cooman, Stacking fault energy and deformation mechanismsin Fe-xMn-0.6C-yAl TWIP steel, Mater. Sci. Eng. A 676 (2016) 216e231.

[5] O. Bouaziz, S. Allain, C.P. Scott, P. Cugy, D. Barbier, High manganese austenitictwinning induced plasticity steels: a review of the microstructure propertiesrelationships, Curr. Opin, Solid State Mater. Sci. 15 (4) (2011) 141e168.

[6] S. Curtze, V.T. Kuokkala, Dependence of tensile deformation behavior of TWIPsteels on stacking fault energy, temperature and strain rate, Acta Mater 58(15) (2010) 5129e5141.

[7] R. Liu, Z.J. Zhang, L.L. Li, X.H. An, Z.F. Zhang, Microscopic mechanismscontributing to the synchronous improvement of strength and plasticity (SISP)for TWIP copper alloys, Sci. Rep. 5 (2015).

[8] L. Lu, X. Chen, X. Huang, K. Lu, Revealing the maximum strength in nano-twinned copper, Science 323 (5914) (2009) 607e610.

[9] L. Remy, A. Pineau, Twinning and strain-induced FCC/ HCP transformation inthe Fe-Mn-Cr-C system, Mater. Sci. Eng. 28 (1) (1977) 99e107.

[10] A. Saeed-Akbari, J. Imlau, U. Prahl, W. Bleck, Derivation and variation incomposition-dependent stacking fault energy maps based on subregular so-lution model in high-manganese steels, Metall. Mater. Trans. A 40 (13) (2009)3076e3090.

[11] S. Allain, J.-P. Chateau, O. Bouaziz, S. Migot, N. Guelton, Correlations betweenthe calculated stacking fault energy and the plasticity mechanisms inFeeMneC alloys, Mater. Sci. Eng. A 387 (2004) 158e162.

[12] L. Cho, E.J. Seo, B.C. De Cooman, Near-Ac3 austenitized ultra-fine-grainedquenching and partitioning (Q&P) steel, Scr. Mater. 123 (2016) 69e72.

[13] M.I. Latypov, S. Shin, B.C. De Cooman, H.S. Kim, Micromechanical finiteelement analysis of strain partitioning in multiphase medium manganeseTWIPþ TRIP steel, Acta Mater 108 (2016) 219e228.

[14] E.J. Seo, L. Cho, Y. Estrin, B.C. De Cooman, Microstructure-mechanical prop-erties relationships for quenching and partitioning (Q&P) processed steel,Acta Mater 113 (2016) 124e139.

[15] W.C. Oliver, G.M. Pharr, An improved technique for determining hardness andelastic modulus using load and displacement sensing indentation experi-ments, J. Mater. Res. 7 (06) (1992) 1564e1583.

[16] G.M. Pharr, Measurement of mechanical properties by ultra-low load inden-tation, Mater. Sci. Eng. A 253 (1e2) (1998) 151e159.

[17] I. Karaman, H. Sehitoglu, K. Gall, Y.I. Chumlyakov, H.J. Maier, Deformation ofsingle crystal Hadfield steel by twinning and slip, Acta Mater 48 (6) (2000)1345e1359.

[18] I. Gutierrez-Urrutia, S. Zaefferer, D. Raabe, The effect of grain size and grainorientation on deformation twinning in a Fee22wt.% Mne0.6 wt.% C TWIPsteel, Mater. Sci. Eng. A 527 (15) (2010) 3552e3560.

[19] I.-C. Jung, L. Cho, B.C. De Cooman, In situ observation of the influence of Al ondeformation-induced twinning in TWIP steel, ISIJ Int. 55 (4) (2015) 870e876.

[20] S. Kang, Y.-S. Jung, B.-G. Yoo, J.-I. Jang, Y.-K. Lee, Orientation-dependentindentation modulus and yielding in a high Mn twinning-induced plasticitysteel, Mater. Sci. Eng. A 532 (2012) 500e504.

[21] J. Sort, A. Concustell, E. Men�endez, S. Suri~nach, M. Bar�o, J. Farran, J. Nogu�es,Selective generation of local ferromagnetism in austenitic stainless steel usingnanoindentation, Appl. Phys. Lett. 89 (3) (2006) 032509.

[22] T.-H. Ahn, C.-S. Oh, D. Kim, K. Oh, H. Bei, E.P. George, H. Han, Investigation ofstrain-induced martensitic transformation in metastable austenite usingnanoindentation, Scr. Mater 63 (5) (2010) 540e543.

[23] B. He, M. Huang, Z. Liang, A. Ngan, H. Luo, J. Shi, W. Cao, H. Dong, Nano-indentation investigation on the mechanical stability of individual austenitegrains in a medium-Mn transformation-induced plasticity steel, Scr. Mater 69(3) (2013) 215e218.

[24] T. Ohmura, K. Tsuzaki, F. Yin, Nanoindentation-induced deformation behaviorin the vicinity of single grain boundary of interstitial-free steel, Mater. Trans.46 (9) (2005) 2026e2029.

[25] J.C. Stinville, C. Tromas, P. Villechaise, C. Templier, Anisotropy changes inhardness and indentation modulus induced by plasma nitriding of 316Lpolycrystalline stainless steel, Scr. Mater 64 (1) (2011) 37e40.

[26] A. Gouldstone, H.-J. Koh, K.-Y. Zeng, A. Giannakopoulos, S. Suresh, Discrete andcontinuous deformation during nanoindentation of thin films, Acta Mater 48(9) (2000) 2277e2295.

[27] K.L. Johnson, Contact Mechanics, Cambridge university press, Cambridge, UK,1985, pp. 84e106.

E.J. Seo et al. / Acta Materialia 135 (2017) 112e123 123

[28] A. Dumay, J.-P. Chateau, S. Allain, S. Migot, O. Bouaziz, Influence of additionelements on the stacking-fault energy and mechanical properties of anaustenitic FeeMneC steel, Mater. Sci. Eng. A 483 (2008) 184e187.

[29] W.S. Choi, B.C. De Cooman, S. Sandl€obes, D. Raabe, Size and orientation effectsin partial dislocation-mediated deformation of twinning-induced plasticitysteel micro-pillars, Acta Mater 98 (2015) 391e404.

[30] T. Li, Y. Gao, H. Bei, E.P. George, Indentation Schmid factor and orientationdependence of nanoindentation pop-in behavior of NiAl single crystals,J. Mech. Phys. Solids 59 (6) (2011) 1147e1162.

[31] Y. Cui, P. Lin, Z. Liu, Z. Zhuang, Theoretical and numerical investigations ofsingle arm dislocation source controlled plastic flow in FCC micropillars, Int. J.Plast. 55 (2014) 279e292.

[32] J.R. Greer, W.D. Nix, Nanoscale gold pillars strengthened through dislocationstarvation, Phys. Rev. B 73 (24) (2006) 245410.

[33] Z. Shan, R.K. Mishra, S.S. Asif, O.L. Warren, A.M. Minor, Mechanical annealingand source-limited deformation in submicrometre-diameter Ni crystals, Nat.Mater 7 (2) (2008) 115e119.

[34] F.F. Csikor, C. Motz, D. Weygand, M. Zaiser, S. Zapperi, Dislocation avalanches,strain bursts, and the problem of plastic forming at the micrometer scale,Science 318 (5848) (2007) 251e254.

[35] Z.-J. Wang, Q.-J. Li, Z.-W. Shan, J. Li, J. Sun, E. Ma, Sample size effects on thelarge strain bursts in submicron aluminum pillars, Appl. Phys. Lett. 100 (7)(2012) 071906.

[36] Z.Y. Liang, M.X. Huang, Deformation twinning in small-sized face-centredcubic single crystals: experiments and modelling, J. Mech. Phys. Solids 85(2015) 128e142.

[37] Z.Y. Liang, J.T.M. De Hosson, M.X. Huang, Size effect on deformation twinningin face-centred cubic single crystals: experiments and modelling, Acta Mater129 (2017) 1e10.

[38] Q. Yu, Z.-W. Shan, J. Li, X. Huang, L. Xiao, J. Sun, E. Ma, Strong crystal size effecton deformation twinning, Nature 463 (7279) (2010) 335e338.

[39] S. Wu, H. Yen, M. Huang, A. Ngan, Deformation twinning in submicron andmicron pillars of twinning-induced plasticity steel, Scr. Mater. 67 (7) (2012)641e644.

[40] M. Chen, E. Ma, K.J. Hemker, H. Sheng, Y. Wang, X. Cheng, Deformationtwinning in nanocrystalline aluminum, Science 300 (5623) (2003)1275e1277.

[41] H. Suzuki, C. Barrett, Deformation twinning in silver-gold alloys, Acta Metall. 6(3) (1958) 156e165.

[42] D.R. Steinmetz, T. J€apel, B. Wietbrock, P. Eisenlohr, I. Gutierrez-Urrutia,A. SaeedeAkbari, T. Hickel, F. Roters, D. Raabe, Revealing the strain-hardeningbehavior of twinning-induced plasticity steels: theory, simulations, experi-ments, Acta Mater 61 (2) (2013) 494e510.

[43] J.P. Hirth, J. Lothe, Theory of Dislocations, Wiley, New York, 1982.[44] T. Michalske, J. Houston, Dislocation nucleation at nano-scale mechanical

contacts, Acta Mater 46 (2) (1998) 391e396.[45] M. Meyers, O. V€ohringer, V. Lubarda, The onset of twinning in metals: a

constitutive description, Acta Mater 49 (19) (2001) 4025e4039.[46] O. Vohringer, Flow stress for twinning of alpha-Cu-alloys, Z. fur Met. 67 (8)

(1976) 518e524.[47] D. Lohe, O. Vohringer, On the influence of mechanical twinning on defor-

mation behaviour of ferritic iron base alloys, Z. Met. 77 (9) (1986) 557e563.

[48] M. Marcinkowski, H. Lipsitt, The plastic deformation of chromium at lowtemperatures, Acta Metall. 10 (2) (1962) 95e111.

[49] T. Lindley, R. Smallman, The plastic deformation of polycrystalline vanadiumat low temperatures, Acta Metall. 11 (5) (1963) 361e371.

[50] S. Song, G. Gray, Influence of temperature and strain rate on slip and twinningbehavior of Zr, Metall. Mater. Trans. A 26 (10) (1995) 2665e2675.

[51] H. Bei, Y. Gao, S. Shim, E.P. George, G.M. Pharr, Strength differences arisingfrom homogeneous versus heterogeneous dislocation nucleation, Phys. Rev. B77 (6) (2008) 060103.

[52] A. Oila, S. Bull, Atomistic simulation of FeeC austenite, Comput. Mater. Sci. 45(2) (2009) 235e239.

[53] J. Kim, S.-J. Lee, B.C. De Cooman, Effect of Al on the stacking fault energy ofFee18Mne0.6C twinning-induced plasticity, Scr. Mater 65 (4) (2011)363e366.

[54] D.E. Hurtado, M. Ortiz, Surface effects and the size-dependent hardening andstrengthening of nickel micropillars, J. Mech. Phys. Solids 60 (8) (2012)1432e1446.

[55] J.R. Greer, W.C. Oliver, W.D. Nix, Size dependence of mechanical properties ofgold at the micron scale in the absence of strain gradients, Acta Mater 53 (6)(2005) 1821e1830.

[56] D. Dimiduk, M. Uchic, T. Parthasarathy, Size-affected single-slip behavior ofpure nickel microcrystals, Acta Mater 53 (15) (2005) 4065e4077.

[57] I.-S. Choi, Y. Gan, D. Kaufmann, O. Kraft, R. Schwaiger, Measurement ofYoung's modulus of anisotropic materials using microcompression testing,J. Mater. Res. 27 (21) (2012) 2752e2759.

[58] I.N. Sneddon, The relation between load and penetration in the axisymmetricBoussinesq problem for a punch of arbitrary profile, Int. J. Eng. Sci. 3 (1) (1965)47e57.

[59] K.E. Knipling, D.J. Rowenhorst, R.W. Fonda, G. Spanos, Effects of focused ionbeam milling on austenite stability in ferrous alloys, Mater. Charact. 61 (1)(2010) 1e6.

[60] A. Basa, C. Thaulow, A. Barnoush, Chemically induced phase transformation inaustenite by focused ion beam, Metall. Mater. Trans. A 45 (3) (2014)1189e1198.

[61] R.P. Babu, S. Irukuvarghula, A. Harte, M. Preuss, Nature of gallium focused ionbeam induced phase transformation in 316L austenitic stainless steel, ActaMater 120 (2016) 391e402.

[62] B.P.J. Sandvik, C.M. Wayman, Characteristics of lath martensite: Part I. crys-tallographic and substructural features, Metall. Trans. A 14 (4) (1983)809e822.

[63] T. Moritani, N. Miyajima, T. Furuhara, T. Maki, Comparison of interphaseboundary structure between bainite and martensite in steel, Scr. Mater 47 (3)(2002) 193e199.

[64] D. Akama, T. Tsuchiyama, S. Takaki, Change in dislocation characteristics withcold working in ultralow-carbon martensitic steel, ISIJ Int. 56 (9) (2016)1675e1680.

[65] D. Sulistiyo, L. Cho, E. Seo, B. De Cooman, Internal friction analysis of lathmartensite in press hardened steel, Mater. Sci. Technol. (2016) 1e14.

[66] A. Shibata, T. Furuhara, T. Maki, Interphase boundary structure and accom-modation mechanism of lenticular martensite in FeeNi alloys, Acta Mater 58(9) (2010) 3477e3492.