Cyclic Response of SUS316L Stainless Steel Processed by ECAP

Yoshihisa Kaneko+1, Shingo Hayashi and Alexei Vinogradov+2

Department of Mechanical Engineering, Osaka City University, Osaka 558-8585, Japan

The cyclic response of the SUS316L stainless steel with the nanostructure induced by profuse twinning during equal channel angularpressing (ECAP) at 423K was investigated with the focus on the shape of the hysteresis loops and their evolution with cycling. The SUS316Lsteel with the nanostructured twins showed a very high stress amplitude if compared to their conventional counterparts. However, unlike theas-quenched specimens, the ECAPed SUS316L steel revealed considerable cyclic softening at plastic strain amplitudes above 2 © 10¹4. Besides,the ECAPed specimens showed characteristic stress asymmetry behavior during fatigue. At the initial stages of the fatigue tests, a tensile peakstress of the hysteresis loop was higher than a compressive one. Then, the tensile peak stresses gradually decreased with increasing cycles andfinally became lower than the compressive peak stress under plastic strain amplitudes higher than 2 © 10¹4. Since no stress asymmetry wasdetected for the as-quenched sample, the observed stress asymmetry behavior was attributed to the specific asymmetric response of thedirectionally sheared microstructure created during ECAP. [doi:10.2320/matertrans.MH201308]

(Received March 7, 2013; Accepted April 30, 2013; Published June 7, 2013)

Keywords: low-cycle fatigue, stainless steel, equal channel angular pressing, ultrafine grained materials, bulk nanostructured materials

1. Introduction

Ultrafine grain (UFG) materials manufactured by severeplastic deformation (SPD) exhibit a very promising combi-nation of mechanical properties. Indeed, under monotonicloads these materials have extraordinary strength, which isoften paired with affordable or even enhanced ductility whileunder cyclic loading they often show substantially improvedhigh-cyclic fatigue performance. Particularly, the mechanicaltwinning resulting in a nano-layered microstructure invarious metals and alloys has been recognized a key factorwhich enables a favorable combination of strength andductility. The nano-twinned microstructures and relatedsuperior mechanical properties were found in copperprocessed by electrodeposition, dynamic plastic deforma-tion1) and low-temperature ECAP followed by cryodrawingand cryorolling,2) and in the SUS316L austenitic stainlesssteel with low stacking fault energy.3,4) Another example ofthe twinning-based deformation strategy was given by Zhaoet al.5) for the low SFE steels with the TWIP (TwinningInduced Plasticity) effect.

The success of this strategy for improving the high cyclicfatigue properties was demonstrated in Ref. 4) for a stableSUS316L austenitic stainless steel: the fatigue limit of the 3-pass samples in the symmetric push-pull testing under con-stant stress amplitude was found to be of 550MPa which is byfar higher than that in its conventionally produced counterpart(of 200MPa). Due to its low SFE, this steel has propensity fortwinning, and, indeed, profuse deformation twinning wasactivated during ECAP processing at 423K so that after justthree ECAP passes by route Bc, a uniform nano-structure withgrain dimensions far below 100 nm on average was formeddue to finely spaced intersecting twin lamellar. While theimprovement of the monotonic strength with grain refinementis quite common, it is evident that the uniform elongationdoes not commonly improve as a result of SPD processing;

however, the material’s resistance to localised plastic flow inthe post-necking regime can increase remarkably giving riseto the overall considerable elongation to fracture,6­8) which isoften reported in the literature. Since the very early studies offatigue and cyclic behavior of SPD manufactured UFG metalsit was demonstrated that the compromised uniform elongationand the specific to these materials susceptibility to strainlocalization in deformation bands often significantly down-grade their low-cycle fatigue (LCF) properties. The phenom-enology of the cyclic response of UFG metals and alloys hasbeen well understood and discussed in several surveyspublished since 1997. In most cases the LCF properties werereported for sub-microcrystalline materials with equiaxedgrains of 200­300 nm diameters. The 1-pass ECAP of theSUS316L steel induces a specific bimodal microstructureconsisting of nano-twinned and strain hardened, but nottwinned regions. The LCF properties of materials having sucha particular microstructure have not been reported in theliterature as yet. In the present study, the cyclic softeningbehavior during testing under plastic strain control wasinvestigated for the SUS316L steel with the above mentionedmicrostructure produced during 1-pass ECAP. Specialattention was paid to the dependence of cyclic stress­strainresponse on plastic strain amplitude, which usually exerts astrong effect on dislocation motion and on resultant self-organized dislocation structure in coarse-grained materials.

2. Experimental Procedure

2.1 Process and materialsThe chemical composition of the SUS316L stainless steel

used in the present study is shown in Table 1. The SUS316Lsteel were annealed at 1353K for 5min and then quenchedin water. The material was shaped to 4 © 4 © 40mm3 billets.The ECAP processing was carried out with an ECAP die

Table 1 Chemical composition of the SUS316L stainless steel used(mass%).

C Si Mn P S Ni Cr Mo Fe

0.008 0.62 1.19 0.033 0.002 12.09 17.33 2.04 Bal.

+1Corresponding author, E-mail: kaneko@imat.eng.osaka-cu.ac.jp+2Present address: Laboratory for the Physics of Strength of Materialsand Intelligent Diagnostic Systems, Togliatti State University, Togliatti,445667, Russia

Materials Transactions, Vol. 54, No. 9 (2013) pp. 1612 to 1618Special Issue on Advanced Materials Science in Bulk Nanostructured Metals II©2013 The Japan Institute of Metals and Materials

having 4 © 4mm2 channel intersecting at right angle with asharp corner. During the ECAP process, the die temperaturewas kept at 423K by a heating device. Further details ofmaterial processing can be found in the reports by Uenoet al.3,4,9) In the present study, the billets were pressedthrough the die only 1 time.

The subsize specimens with gauge part of 1.5 © 1.5 ©2mm3 were shaped by spark erosion. The tensile axis andsurface normal of the specimens were parallel to extrusiondirection (ED) and transverse direction (TD), respectively.The specimen surfaces were mechanically and electrolyti-cally polished prior to the tests.

2.2 Fatigue testLow-cycle fatigue tests were carried out using a servo-

hydraulic machine (Shimadzu Servo Pulser EHF-LB10kN-10N). To ensure stress-free gripping, the flat I-shapedspecimens were fastened to the fatigue testing machine witha help of low melting-point alloy. Axial strain was measuredwith a strain gauge cemented at the gauge part. Cyclic tensileand compressive strains were applied to the specimens underconstant plastic strain amplitude. The plastic strain ampli-tudes ¾pl used in the study ranged from 1 © 10¹4 to 5 © 10¹3.The low-cycle fatigue tests were conducted in air at roomtemperature, and 0.2Hz frequency for all plastic strainamplitudes. The upper limit of fatigue cycles was set at30,000 cycles. The fatigue tests were terminated when a rapidreduction of the stress amplitude was detected in associationwith fatigue cracking.

3. Results

3.1 Cyclic softeningFigure 1 shows a typical microstructure of the processed

sample. Similarly to the results reported for the same typesteel,3,10) the very fine deformation twins were generated inapproximately a half volume of the billet.

Cyclic hardening curves are shown in Fig. 2 in termsof the mean stress amplitude ·a,m = (·max ¹ ·min)/2 as afunction of the cumulative plastic strain ¾pl,cum = 4N¾pl,where ·max and ·min are the tensile and compressive peakstresses of hysteresis loops, respectively, and N is the numberof cycles. For the sake of comparison, the cyclic responseof the as-quenched SUS316L specimen is plotted on thesame diagram. Interestingly, while most SPD-manufacturedmaterials exhibit pronounced cyclic softening even at fairlylow strain amplitudes, the stress amplitudes of the ECAPedspecimen fatigued at ¾pl = 1 © 10¹4 as well as those of theas-received specimen were almost constant. However, at theplastic strain amplitudes above 2 © 10¹4, the stress amplitudedecreased rapidly from the onset of cyclic deformation. Afterthis initial cyclic softening stage, the stress amplitudes werefluctuated at the plastic strain amplitudes of 2 © 10¹4 ¯¾pl ¯ 2 © 10¹3. For the test conducted at ¾pl = 1 © 10¹4, thefatigue test was stopped at 30,000 cycles. The other testswere terminated when a rapid decay in the stress amplitudewas noticed. This typically corresponds to the beginning offatigue crack initiation and after that the plastic strain controltest does not make sense since the plastic strain is actuallylocalized at the crack tip.

Figure 3 shows a typical behavior of the cyclic strainhardening rate derived from the corresponding cyclic hard-ening curve. Because of initial cyclic softening, the large partof the hardening rate was negative, particularly during theearly half of cycling. During the initial rapid softening stage,the negative hardening rate increased with increasing cycles.Then, a local maximum, which corresponds to an inflectionpoint of the hardening curve, was reached and the hardeningrate showed a trend to decrease again, i.e., the specimenexhibited secondary softening. The secondary softening wererecognized at the specimens fatigued at ¾pl ² 2 © 10¹4. In thelater stage of fatigue life, the hardening rate fluctuatedgreatly, reflecting the jerky character of the stress amplitude.

The cyclic stress­strain curve relating the stress amplitudeand the plastic strain amplitude is shown in Fig. 4 wherethe maximum stress amplitude and the stress amplitudeat the inflection point are plotted. As is expected fromthe monotonic stress­strain behavior, the stress amplitudeincreases with increasing plastic strain amplitude as apolynomial function. At ¾pl = 5 © 10¹3, the stress amplitudewas as high as 840MPa. The degree of the initial rapidsoftening can be described in terms of the difference betweenthe maximum stress amplitude and the stress amplitude at the

1μm

Fig. 1 A TEM photograph showing the microstructure of the SUS316Lstainless steel processed by 1-pass ECAP.

Cumulative Plastic Strain, εε pl,cum

0 10 20 30 40 50 60 70

Mea

n S

tres

s A

mp

litu

de,

σa,

m /M

Pa

0

200

400

600

800

1000

εpl=5x10-3

εpl=2x10-4

εpl=1x10-3 [Ref.4]

εpl=2x10-3

εpl=1x10-4

Not ECAPed (εpl=1x10-3) [Ref.4]

Fig. 2 Cyclic softening curves of the ECAPed SUS316L steel specimenscyclically deformed at plastic strain amplitudes from 1 © 10¹4 to5 © 10¹3. The stress amplitude plotted in the figure represents a meanvalue of the tensile and compressive peak stresses.

Cyclic Response of SUS316L Stainless Steel Processed by ECAP 1613

infection point. For the plastic strain amplitudes 2 © 10¹4 ¯¾pl ¯ 2 © 10¹3, the difference increased with the increasingstrain amplitude. However, at the higher plastic strainamplitude (¾pl = 5 © 10¹3), this difference is smaller.

3.2 Stress asymmetryThe prominent feature of the cyclic hysteresis loops

observed in the present ECAPed SUS316L steel is theirsignificant stress asymmetry in forward and backwarddirections. Figure 5 shows the behavior of the tensile andcompressive peak stresses with cycling under ¾pl = 2 © 10¹3

(c.f. Fig. 2 where the mean value between these two isplotted). At the very initial stage of fatigue, the tensile peakstress was higher than the compressive one. The highertensile stress can be considered as a manifestation of theBauschinger effect which is promoted by the directionality ofthe ECAP processing. This is because the shear direction ofthe ECAPed billet is equal to that of a strip specimenprestrained towards tensile direction. Then, the tensile peakstress decreased rapidly with increasing cycles, although thecompressive stress increased gradually.

In order to quantify the extent of the stress asymmetry inthe hysteresis loop, let us consider a stress asymmetryparameter defined as the stress range "· = ·max ¹ ·min

related to the mean stress amplitude ·a,m. The behavior of

the stress asymmetry parameter "·/·a,m is shown in Fig. 6.A positive "·/·a,m value indicates that the tensile peak stressis higher than the compressive one. For all plastic strainamplitudes tested, the "·/·a,m values of ECAPed specimenswere positive at the onset of cyclic deformation. Moreover, at¾pl = 1 © 10¹4, the "·/·a,m value remained positive through-out the entire fatigue test, although some fluctuations wereobserved. For the fatigue tests at 2 © 10¹4 ¯ ¾pl ¯ 5 © 10¹4,the "·/·a,m value shifted from the positive (tensile) region tothe negative (compressive) one. It should be noted that thestress asymmetry is attributable to ECAP processing entirely.To the contrast, ideally symmetric hysteresis loops areobserved in the as-quenched SUS316L steel showing a closeto zero "·/·a,m value throughout the test.

3.3 Hysteresis loop shapeFigure 7 displays the evolution in the hysteresis loop shape

at ¾pl = 2 © 10¹4. In the initial deformation stage(¾pl,cum = 0.1), the hysteresis loop had a smooth lenticularshape. It can be noticed that concurrently with the evolutionof the stress asymmetry, the hysteresis loop becomesdistorted with increasing number of cycles.

To assess the change in the hysteresis loop quantitatively,we calculated the strain hardening rate in individual loopsselected arbitrarily at different stages of fatigue life. The

Plastic Strain Amplitude, εε pl

10-5 10-4 10-3 10-2

Str

ess

amp

litu

de,

σa

/MP

a

0

200

400

600

800

1000

Maximum stress amplitudeStress amplitude at inflection point

Fig. 4 Cyclic stress­strain curve of the ECAPed SUS316L stainless steel.Both the maximum stress amplitude and the stress amplitude at theinflection points of softening curves are plotted.

0 10 20 30 40 50 60 700

200

400

600

800

1000

Cumulative Plastic Strain, ε pl,cum

Str

ess

Am

plit

ud

e, σ

a /M

Pa compressive

tensile

Fig. 5 Tensile and compressive peak stresses under the plastic strainamplitude of 2 © 10¹3.

0 10 20 30 40 50 60 70

Str

ess

Asy

mm

etry

, Δ σ

/ σ

am

-1.5

-1.0

-0.5

0.0

0.5

1.0

1.5

εpl=5x10-3

εpl=2x10-4

εpl=1x10-3

εpl=2x10-3

εpl=1x10-4

Cumulative Plastic Strain, εpl,cum

As-quenched (εpl=1x10-3)

Fig. 6 Changes in the stress asymmetry parameter "·/·a,m during thefatigue tests.

Cumlative Plastic Strain, εεpl,cum

0 5 10 15 20Cyc

lic S

trai

n H

ard

enin

g R

ate,

dσ

/dε p

l,cu

m/ M

Pa

-40

-20

0

20

40

60

80

100

Mea

n S

tres

s A

mp

litu

de,

σa,

m /

MP

a

0

100

200

300

400

500

600

Hardening rate

Stress amplitude

(Inflection point)

(Local maximum)

Fig. 3 Cyclic strain softening rate in the ECAPed SUS316L steel fatiguedat the plastic strain amplitude of 2 © 10¹4.

Y. Kaneko, S. Hayashi and A. Vinogradov1614

strain hardening rates measured within the hysteresis loops at¾pl,cum = 0.1 and 9 are exemplified in Fig. 8 for the specimentested at ¾pl = 2 © 10¹4. On the initial stage, ¾pl,cum = 0.1, thestrain hardening rate decreased monotonically during eithertensile or compressive direction of loading. As a matter offact, the hardening rate curves at both tensile and compres-sive halves of the hysteresis were almost the same within theregular scatter. At the later deformation stage, ¾pl,cum = 9, it isclear that the further reduction in the hardening rate occurredwith inflection points on the hardening curves in both loadingdirections. This kind of the loop distortion is indicative of the

beginning of the secondary softening stage. It is important tonotice that the hardening rate curve during tensile straining atthis deformation stage was different from that in the oppositedirection, i.e., the asymmetric strain hardening behavior isevident at the latter stage of fatigue of the specimen tested at¾pl = 2 © 10¹4.

Figure 9 shows the change in the hysteresis loop shapeat the large plastic strain amplitude, ¾pl = 5 © 10¹3. Thehysteresis loops at ¾pl,cum = 3.6 and 13.6 have shape whichis typically observed in ductile metallic materials with aconventional grain size. At ¾pl,cum = 16, the loop distortionis observed only during tensile straining. The shape of thedistorted loops appears to be apparently different from theloop distortions observed at ¾pl = 2 © 10¹4. Because it isseen only during tension, this kind of distortion is most likelyappeared as a result of fatigue cracking. Hence, the secondarysoftening, which is observed at the relatively high plasticstrain amplitude ¾pl = 5 © 10¹3, Fig. 2, should attributed tothe fatigue cracking. However, it is unlikely that fatiguecracking could cause secondary cyclic softening at 2 ©10¹4 ¯ ¾pl ¯ 2 © 10¹3 since the stress amplitudes increasedagain after the secondary softening at these strain amplitudes.Furthermore, no characteristic changes peculiar to fatiguecracking was detected in hysteresis loops.

4. Discussion

4.1 Cyclic softeningCyclic softening has been systematically reported to occur

in many fcc UFG metals and alloys with relatively highSFE.11­14) On the other hand, hcp metals such as Ti appearedimmune from softening under cyclic deformation.7,14,15)

Str

ess,

σ /M

Pa

-1000

-500

0

500

1000

Strain, ε / x10-3-4 -2 0 2 4

-1000

-500

0

500

1000

-4 -2 0 2 4

Strain, ε / x10-3

Str

ess,

σ /M

Pa

(a) εpl,cum=0.1 (b) εpl,cum=7

(c) εpl,cum=9 (d) εpl,cum=20

Fig. 7 Typical shape of a hysteresis loop at ¾pl = 2 © 10¹4.

Cumulative Total Strain, εεcum

0.000 0.002 0.004 0.006 0.008 0.010 0.012

Str

ain

Har

den

ing

Rat

e, d

σ/d

ε /G

Pa

0

50

100

150

200

250

300

Tensile Straining Compressive Straining

εpl,cum=0.1εpl,cum=9

Fig. 8 Strain hardening rate in a randomly selected hysteresis loop at theplastic strain amplitude of 2 © 10¹4. The cumulative total strain iscalculated, assuming that the original point corresponds to the hysteresisloop edge on a compressive side.

Cyclic Response of SUS316L Stainless Steel Processed by ECAP 1615

Microstructural changes such as fatigue induced dynamicgrain coarsening11,12) and dislocation annihilation13) havelong been recognized as a root cause for cyclic strainsoftening of UFG materials manufactured by SPD. De-pendence of the cyclic softening on the plastic strainamplitude was reported in UFG copper produced by ECAP,and the cyclic softening ratio was associated with the extentof grain coarsening.12) However, careful microstructuralobservations of the present SUS316L by electron channelingcontrast (ECC) technique in a scanning electron microscopedid not reveal any signatures of grain coarsening after fatigue.In coarse-grained SUS316L steel fatigued, the TEMobservations have shown the occurrence of dislocation self-organization which should involve dislocation cross slipprocess.16) Similarly, for cyclic deformation of the ECAPedSUS316L steel, the cross slip which promotes dislocationannihilation is also expected. It is likely that the initial rapidsoftening observed at 2 © 10¹4 ¯ ¾pl ¯ 5 © 10¹3 was causedprimarily by the reduction of excess dislocation densitystored in the specimen during the ECAP. The dislocationstorage during processing and the subsequent dislocationannihilation during LCF are most reasonably to occur in theregions where the nanostructured twins were not favored.

In the prestrained course-grained copper specimen, cyclicsoftening occurred at the plastic strain amplitude above¾pl = 8 © 10¹5, which was referred to as a cyclic softeningthreshold.17) The cyclic hardening curves in Fig. 2 indicatethat the softening threshold amplitude in the ECAPedSUS316L steel should exist at ¾pl between 1 © 10¹4 and2 © 10¹4. Below the softening threshold amplitude, themicrostructure would be unaltered during fatigue cycling.The positive "·/·a,m value at ¾pl = 1 © 10¹4, Fig. 6, isconsistent with the absence of cyclic softening, given that theinitial microstructure produced by the ECAP is maintained.

For the plastic strain amplitudes of 1 © 10¹4 ¯ ¾pl ¯2 © 10¹3, the degree of initial cyclic softening increaseswith increasing plastic strain amplitude as shown in Fig. 4.Above the softening threshold, the dislocation annihilationoccurs.

The degree of cyclic softening at ¾pl = 5 © 10¹3 wasinsignificant in comparison with that during fatigue tests atthe smaller strain amplitude (¾pl = 1 © 10¹3 or 2 © 10¹3).Two different reasons can be given in account for the

observed small cyclic softening at ¾pl = 5 © 10¹3. The first isthe early fatigue cracking. It is feasible that in the materials,which have been significantly hardened during processingand which posses only limited ductility and limited resistanceto fatigue crack propagation, the cracks develop at higherstrain amplitudes before the specimen becomes sufficientlysoftened during the initial cyclic stage. The second possibilityis that the dislocation multiplication overrides the effectof dislocation recovery at such a large strain amplitude as5 © 10¹3 (cf. also the cyclic hardening curve correspondingto ¾pl = 5 © 10¹3, Fig. 2, where slight hardening is appreci-able after the initial rapid softening). It is therefore plausibleto suppose that the overall complexity of the cyclic softeningbehavior is caused by a synergistic interplay betweendislocation annihilation and multiplication which occurconcurrently during cyclic deformation.

4.2 Stress asymmetryThe specific asymmetry of the hysteresis loop for ECAP

materials, cf. Figs. 5 and 7, is worthy in deep discussion.In the range of plastic strain amplitudes of 2 © 10¹4 ¯¾pl ¯ 2 © 10¹3, the particularly significant reduction of thestress amplitude is observed on the tensile part of thehysteresis loop during cycling, Fig. 5, with an evidentcrossover around ¾pl,cum = 0.25 (i.e., at around N of 30 initialcycles). The specimens showing such specific stressasymmetry had two common characteristics: (i) the distortedhysteresis loops and (ii) the fluctuations of the stressamplitude at the later fatigue stage. Similar hysteresis loopshaving a typical sigmoidal shape have been reported forthe ZK60A wrought magnesium alloy which deforms withan aid of mechanical twinning during fatigue.18) Thus, it isplausible that the characteristic distortion of hysteresis loopsin the present SUS316L steel is originated from the effectof twinning during fatigue. In fact, profuse twinning iscommonly observed in this steel during deformation.3,4)

After rapid cyclic softening caused by dislocationrecovery, the twinning deformation can come into play toaccommodate the imposed plastic strain of later deformationstages. The observed significant fluctuations of the stressamplitude, Figs. 2 and 3, can be caused by discontinuousmicrostructural evolution involving coalescence of nano-twinbands introduced during ECAP.

-15 -10 -5 0 5 10 15

Str

ess,

σ /

MP

a

-1000

-500

0

500

1000

Strain, ε / x10-3-15 -10 -5 0 5 10 15

Strain, ε / x10-3

(a) εpl,cum=3.6 (b)εpl,cum=13.6

εpl,cum=16

Fig. 9 Hysteresis loops under the plastic strain amplitude of 5 © 10¹3.

Y. Kaneko, S. Hayashi and A. Vinogradov1616

Microstructural details of the microscopic mechanismgoverning the asymmetric stress behavior, which is alsoaccompanied by cyclic softening at the tensile part and cyclichardening at the compressive part, are still unclear. Asimplified picture can be drawn to understand the drift ofhysteresis loop towards the compressive side, Fig. 7, fromthe mechanistic viewpoint assuming that the irreversibleelongation occurs in tensile direction. The illustration for thispossible irreversible elongation is shown in Fig. 10. In thepresent fatigue testing system, which enables the plasticstrain control, both the tensile and compressive stressamplitudes are adjusted at every half cycle to keep bothedges of the hysteresis loop corresponding to the targetelastic lines within a preset tolerance. If a specimen isirreversibly elongated in tensile direction, Fig. 10(b), thetensile and compressive peak stresses are automaticallydrifted to the compressive side in the next cycle, Fig. 10(c).In this manner, the irreversible elongation can cause theobserved stress asymmetry.

The irreversible elongation should be accompanied by theunusual cyclic deformation whereby a plastic deformationmechanism controlling tensile straining is certainly differentfrom that in compressive straining. The hardening ratebehavior shown in Fig. 8 (¾pl,cum = 9), indicating that thestress­strain hardening curves were different in tensile andcompressive parts of the hysteresis loop, is consistent withthe presence of the asymmetric deformation mechanism.

A microstructurally-based modeling of the asymmetriccyclic response of UFG and bulk-nano materials manufac-tured by ECAP is still lucking and this should rely heavily onthe specifics of the simple shear texture, which is inevitablycreated during pressing through the ECAP die and whichinevitably determines the dislocation behavior when theexternal force is applied in a certain direction. Thesignificance of the crystallographic texture and dislocation-based modeling for understanding of the mechanical behaviorof SPD-produced bulk-nano metals has been highlighted onmany occasions (see, for example, a recent review15) andreferences therein). This challenging task will be in focus offurther investigations.

5. Summary and Conclusions

(1) The cyclic response of the SUS316L steel processedby one ECAP pass has been investigated with the focus on

the cyclic softening behavior in dependence on the appliedplastic strain amplitude ¾pl. Cyclic softening was almostabsent at the plastic strain amplitude as low as of 1 © 10¹4.At higher ¾pl ranging between 2 © 10¹4 and 2 © 10¹3

pronounced cyclic softening was observed. The extent ofinitial cyclic softening increased with increasing plastic strainamplitude below ¾pl = 2 © 10¹3. However, cyclic softeningwas lowered again at the highest plastic strain amplitude¾pl = 5 © 10¹3.

(2) The significantly strengthened by ECAP specimens ofthe SUS316L steel exhibited notable hardening in the cyclicstress­strain curve: the maximum stress amplitude increasedreasonably with increasing plastic strain amplitude from300MPa at ¾pl = 1 © 10¹4 to 840MPa at ¾pl = 5 © 10¹3.

(3) The most intriguing finding of the present work is thatthe ECAPed SUS316L steel exhibits significant asymmetryin the tensile and compressive parts of the hysteresis loops.Either tensile or compressive stress component can benotably higher than the other, depending on the imposedplastic stress amplitude and the number of cycles. Thecomplexity of stress asymmetry is further gained from theobserved cyclic softening. The observed stress asymmetry isfully attributed to the specific micro structure and texturecreated during ECAP.

Acknowledgments

This study was financially supported by the Grant-in-Aidfor Scientific Research on Innovative Areas “Bulk nano-structured metals” through MEXT Japan under contractNo. 22102006, and by that on Challenging ExploratoryResearch through JSPS under contract No. 24656091.Special thanks to Mr. N. Tabata for his assistance with theTEM observation. One of the authors (A.V.) wishes tothank the Russian Ministry of Education and Science forpartial support of this research through the Grant-in-AidNo. 11.G34.31.0031.

REFERENCES

1) K. Lu, L. Lu and S. Suresh: Science 324 (2009) 349­352.2) Y. H. Zhao, J. E. Bingert, X. Z. Liao, B. Z. Cui, K. Han, A. V.

Sergueeva, A. K. Mukherjee, R. Z. Valiev, T. G. Langdon and Y. T. T.Zhu: Adv. Mater. 18 (2006) 2949­2953.

3) H. Ueno, K. Kakihata, Y. Kaneko, S. Hashimoto and A. Vinogradov:J. Mater. Sci. 46 (2011) 4276­4283.

irreversibleelongation

plastic straincontrol

target elastic lines(a) (b) (c)

Fig. 10 A schematic illustration of irreversible elongation model which can give rise to the tensile-compressive stress asymmetry ofhysteresis loops.

Cyclic Response of SUS316L Stainless Steel Processed by ECAP 1617

4) H. Ueno, K. Kakihata, Y. Kaneko, S. Hashimoto and A. Vinogradov:Acta Mater. 59 (2011) 7060­7069.

5) Y. H. Zhao, Y. T. Zhu, X. Z. Liao, Z. Horita and T. G. Langdon: Appl.Phys. Lett. 89 (2006) 121906.

6) L. W. Meyer, K. Sommer, T. Halle and M. Hockauf: Nanomaterials bySevere Plastic Deformation, ed. by H. J. Maier and Y. Estrin, (TransTech Publications Ltd., Goslar, Germany, 2008) pp. 815­820.

7) A. Y. Vinogradov, V. V. Stolyarov, S. Hashimoto and R. Z. Valiev:Mater. Sci. Eng. A 318 (2001) 163­173.

8) A. Vinogradov, S. Hashimoto and V. I. Kopylov: Mater. Sci. Eng. A355 (2003) 277­285.

9) H. Ueno, K. Kakihata, A. Vinogradov, Y. Kaneko and S. Hashimoto:Tribologist 56 (2011) 506­513.

10) Y. Kaneko, N. Tabata, H. Ueno and A. Vinogradov: Proc. Int.

Workshop on Bulk Nanostructured Metals, (2012) pp. 137­140.11) S. R. Agnew and J. R. Weertman: Mater. Sci. Eng. A 244 (1998) 145­

153.12) H. W. Höppel, Z. M. Zhou, H. Mughrabi and R. Z. Valiev: Philos. Mag.

A 82 (2002) 1781­1794.13) C. Xu, Q. Wang, M. Zheng, J. Li, M. Huang, Q. Jia, J. Zhu, L. Kunz

and M. Buksa: Mater. Sci. Eng. A 475 (2008) 249­256.14) Y. Estrin and A. Vinogradov: Int. J. Fatigue 32 (2010) 898­907.15) Y. Estrin and A. Vinogradov: Acta Mater. 61 (2013) 782­817.16) J. Polák: Compr. Struct. Integrity 4 (2007) 504­537.17) C. E. Feltner and C. Laird: Acta Metall. 15 (1967) 1633­1653.18) L. Wu, A. Jain, D. W. Brown, G. M. Stoica, S. R. Agnew, B. Clausen,

D. E. Fielden and P. K. Liaw: Acta Mater. 56 (2008) 688­695.

Y. Kaneko, S. Hayashi and A. Vinogradov1618