Mechanical Properties of Bulk Ultrafine Grained Aluminum Fabricatedby Torsion Deformation at Various Temperatures and Strain Rates

Sunisa Khamsuk1, Nokeun Park1,2, Si Gao1, Daisuke Terada1,2, Hiroki Adachi3 and Nobuhiro Tsuji1,2,+

1Department of Materials Science and Engineering, Kyoto University, Kyoto 606-8501, Japan2Elements Strategy Initiative for Structural Materials (ESISM), Kyoto University, Kyoto 606-8501, Japan3Department of Materials Science and Chemistry, Graduate School of Engineering, University of Hyogo, Himeji 671-2280, Japan

A commercial purity aluminum was heavily deformed up to an equivalent strain of 4 at various temperatures and strain rates by torsiondeformation to produce specimens with various ultrafine grained (UFG) microstructures. The microstructures were characterized by electronbackscatter diffraction (EBSD) and transmission electron microscopy (TEM). The microstructural observation revealed that the torsion deformedspecimens had various mean grain sizes ranging from 0.38 to 8.6 µm. The grain size and dislocation density in the microstructures depended onthe deformation conditions organized by Zener­Hollomon parameter. The mechanical properties of the torsion deformed specimens wereinvestigated by tensile test at room temperature. It was found that the ultrafine grained specimens showed high strength which reached a valuealmost three times higher than that of the starting material. The strength of the UFG aluminum was higher than the level expected from the Hall­Petch relationship for conventionally coarse grained aluminum. The strengthening mechanisms in the UFG aluminum were discussed in terms ofsubstructures introduced during torsion deformation. [doi:10.2320/matertrans.MA201321]

(Received September 5, 2013; Accepted October 29, 2013; Published December 13, 2013)

Keywords: Hall­Petch relationship, ultrafine grains, commercial purity aluminum, torsion deformation

1. Introduction

Ultrafine grained (UFG) materials have become aninteresting subject in the research of materials sciencebecause of their unique microstructures and excellentmechanical properties.1) It was demonstrated in our previousstudy that torsion deformation can be one of the severeplastic deformation (SPD) processes to fabricate bulk UFGaluminum alloys.2) The torsion deformation has an advantagein accurately controlling deformation conditions (i.e., temper-atures and strain rates) and also allows a continuous imposeof large strain without interruption, compared with other SPDprocesses.3) However, the microstructure and mechanicalproperties of the UFG materials fabricated by torsiondeformation have not been clarified yet. One reason is thatthe torsion deformation has been rarely used for producingbulk UFG materials, and another reason is the difficulty tofabricate mechanical testing samples from torsion deformedspecimens. Although the mechanical properties of the UFGmaterials produced by torsion deformation have not beenwidely investigated, the ultrahigh strength could be expectedfor the UFG aluminum according to well-known Hall­Petchrelationship,4,5)

·y ¼ ·0 þ kyd�1

2 ð1Þwhere ·y is the yield stress, ·0 is the friction stress required tomove dislocations in very coarse grains, ky is the Hall­Petchcoefficient, and d is the average grain size. The Hall­Petchrelationship suggests that the yield stress should increaseproportionally with decreasing the minus square root of themean grain size, in other words, the yield stress of the UFGmaterial can be estimated from the microstructural parameter(grain size) by extrapolating the Hall­Petch relation forcoarse grained materials. Recently, it has been also reportedthat the Hall­Petch relationship is no longer valid for UFG

and nanostructured materials.6­11) For example, severalresearchers have reported that the UFG materials exhibitultrahigh strength, which is higher than that predicted fromthe Hall­Petch relationship for coarse grained materials.6­11)

On this background, it can be said that the strengtheningmechanism of the UFG materials is quite complex and stillunclear. Therefore, in the present study the authors try toclarify the following two issues. One objective of this work isto characterize the mechanical properties and microstructuresof the UFG materials fabricated by torsion deformation atvarious temperatures and strain rates. The other aim of thepresent study is to clarify the strengthening mechanism of theUFG materials on the basis of the microstructural parameters.

2. Experimental Procedures

2.1 Torsion deformationThe material used in the present study was a commercial

purity aluminum (JIS 1100Al). The chemical composition ofthe starting material is summarized in Table 1. The torsionspecimens having a cylindrical gage part 4mm in length and8mm in diameter were machined from the as-receivedaluminum bars. The details about the dimension and shapeof the torsion specimens have been reported in the previousreport.2) The shear strain (£) and equivalent strain (¾eq) can becalculated by following equations:12)

£ ¼ 2³rN

Lð2Þ

¾eq ¼£ffiffiffi3

p ð3Þ

Table 1 Chemical composition of the 1100Al studied (mass%).

Si Fe Cu Mn Zn Ti Al

0.09 0.61 0.11 0.01 0.02 0.02 Bal.+Corresponding author, E-mail: nobuhiro-tsuji@mtl.kyoto-u.ac.jp

Materials Transactions, Vol. 55, No. 1 (2014) pp. 106 to 113Special Issue on Strength of Fine Grained Materials ® 60 Years of Hall­Petch®©2013 The Japan Institute of Metals and Materials

where r is the radial position, N is the number of rotation, L isthe gage length.

The specimens were heavily deformed by torsion defor-mation up to 1.1 rotations (corresponding to the maximumequivalent strain of 4 at the surface of the gage) at differenttemperatures ranging from room temperature (RT) to 400°Cand different strain rates from 10¹2 to 102 s¹1 to producespecimens with various ultrafine microstructures. Thedeformation was carried out at constant rotation speedcorresponding to the equivalent strain rate. The heatingrate used for all torsion deformation was 0.5°C·s¹1. Thedeformation temperature and strain rate are combined into asingle parameter, Zener­Hollomon (Z) parameter, determinedas,13)

Z ¼ _¾ expQ

RT

� �ð4Þ

where _¾ is the strain rate, Q is the apparent activation energyfor deformation, R is the gas constant and T is the absolutetemperature. The Q value of 156 kJ·mol¹1 was taken fromthe results of deformation in commercial purity aluminumreported by Jonas et al.13) and used for calculating the Zparameter in the present study. The deformation conditionsused in the present study are summarized in Table 2.

2.2 Microstructural analysisMicrostructural analysis was carried out using electron

back-scattering diffraction (EBSD) measurement, transmis-sion electron microscopy (TEM) and X-ray diffraction(XRD). Specimens for EBSD analysis were prepared bymechanical polishing followed by electro-polishing in asolution of 30 vol% nitric acid (HNO3) and 70 vol% methanol(CH3OH) at approximately ¹30°C with a voltage of 12V.EBSD observation was performed in a field-emissionscanning electron microscope (FE-SEM, Phillips FEIXL30S FEG) at an accelerating voltage of 15 kV. For TEManalysis, thin-foil specimens were prepared by twin-jetelectopolishing under the same condition and solution asthose for the EBSD sample preparation. TEM observationswere performed in a JEOL-2000EX TEM microscope with adouble-tilt stage holder at an operating voltage of 200 kV.The specimens for XRD investigation were prepared usingthe same procedures as those for the EBSD specimens. TheX-ray measurement was performed using Rigaku Smart-LabX-ray diffractometer with Cu K¡ radiation. All the observa-tions and analysis were carried out at the near surfaceposition of the gage part in the torsion specimens at which thecorresponding equivalent strain was 3.6.

2.3 Mechanical testMechanical properties of torsion deformed specimens

were characterized by tensile test. The tensile test specimenswere cut from near surface regions in the gage part oftorsion deformed specimens, as is illustrated in Fig. 1(a).In the near surface region of the torsion specimens, a highshear strain could be obtained, and that was also the positionof microstructural observations. The sheet-type tensile testspecimens had dimensions 2mm in gage length and 1mm ingage width. The cut specimens were mechanically polisheddown to a thickness of 0.4mm. The tensile test specimen isillustrated in Fig. 1(b).

Tensile test was carried out using an universal test machineShimadzu model AG-I 100 kN at RT and an initial strain rateof 8.3 © 10¹4 s¹1. The tensile direction was perpendicular tothe shear direction in torsion. A CCD camera was used tomeasure the precise elongation during the tensile test. Tenimages per second were recorded by the CCD camera formeasuring the displacement of the gage part of the tensilespecimen.

3. Results and Discussion

3.1 Microstructures of torsion deformed specimensFigures 2(a)­2(f ) are EBSD grain boundary maps showing

microstructures of the starting material and the specimensdeformed by torsion to an equivalent strain of 3.6 under awide range of Z conditions from 1.28 © 1010 s¹1 to 2.21 ©1026 s¹1. In the grain boundary maps, the black and gray linesindicate high angle boundaries having misorientation (ª)larger than 15° ð15� 5 ªÞ and low angle boundaries withmisorientation of 2� 5 ª < 15�, respectively. The startingmaterial has a fully recrystallized structure having an averagegrain size of 28 µm (Fig. 2(a)). After deformation at Z =2.21 © 1026 s¹1, the microstructure consists of a lamellarboundary structure nearly parallel to the shear direction(Fig. 2(b)). The mean grain size (thickness) determined fromhigh angle boundaries (DHAGB) is 0.63 µm. In the specimendeformed at Z = 1.69 © 1015 s¹1, the microstructure mainlyconsists of elongated grains, most of the grains are surroundedby high angle grain boundaries, and the mean grain sizedetermined from high-angle boundaries increases to 1.16 µm(Fig. 2(c)). With decreasing Z value furthermore, the micro-structure changes to more equiaxed grains surrounded by highangle boundaries (Figs. 2(d)­2(f )). The mean grain sizeincreases with decreasing the Z value. The mean grain sizes ofthe specimens deformed at Z values from 1.28 © 1014 s¹1 to1.28 © 1010 s¹1 are in a range from 2.9 to 12.4 µm. When the

Table 2 Summary of the deformation conditions (strain rate and deforma-tion temperature) and Z parameter used in the present torsion deformation.

Temperature,T/°C

Strain rate,_¾/s¹1

Zener­Hollomon parameter,Z/s¹1

RT 10¹1 2.21 © 1026

200 10¹2 1.69 © 1015

300 10¹2 1.67 © 1012

400 102 1.28 © 1014

400 10¹2 1.28 © 1010Fig. 1 (a) Illustration of the side and top view of a specimen deformed by

torsion, showing the position where tensile specimens were taken. (b)Dimension and shape of the tensile test specimen.

Mechanical Properties of Bulk Ultrafine Grained Aluminum Fabricated by Torsion Deformation at Various Temperatures and Strain Rates 107

grain size was determined using all boundaries in the EBSDboundary maps (low angle boundaries + high angle ones:),the mean grain size (Dall) was naturally smaller than DHAGB,which is summarized in Table 3.

In order to examine more details in the microstructures,TEM observations were also performed. TEM micrographs ofthe specimens deformed at two different Z values are shownin Fig. 3. It is found that the microstructure of the specimendeformed at high Z value of Z = 2.21 © 1026 s¹1 consistsof elongated grains having large amount of dislocation asshown in Fig. 3(a). When the Z value decreases down to1.69 © 1015 s¹1, the microstructure consists of equiaxedgrains having low dislocation densities (Fig. 3(b)). Theresults coincide well with the EBSD microstructures shownin Fig. 2. Figure 4 shows XRD results of the specimens aftertorsion deformation by 1.1 rotations at various Z values. Itwas found that all the diffraction peaks were well matchedwith the diffraction pattern of aluminum with face-centered

cubic (Fcc) structure and no other phases were observed inthe XRD pattern as shown in Fig. 4(a). For a careful exami-nation of the changes in diffraction peaks with increasing the

Fig. 2 EBSD boundary maps of the starting material and the specimens torsion deformed to an equivalent strain of 3.6 under variousZ conditions: (a) nondeformed (starting) material, (b) Z = 2.21 © 1026 s¹1, (c) Z = 1.69 © 1015 s¹1, (d) Z = 1.28 © 1014 s¹1,(e) Z = 1.67 © 1012 s¹1 and (f ) Z = 1.28 © 1010 s¹1.

Table 3 The structural parameters determined by EBSD measurement andXRD. In this table, Dall is the average (sub)grain size taking account of allboundaries having misorientation above 2°, DHABG is the mean grain sizemeasured from high angle grain boundaries (ª ² 15°), f HAGB is thefraction of high angle grain boundaries, and μ is the dislocation density.

Zener­Hollomonparameter, Z/s¹1 Dall/µm DHAGB/µm f HAGB μ/m¹2

Starting material 23.40 28.00 0.76 Not measured

2.21 © 1026 0.38 0.63 0.60 9.56 © 1013

1.69 © 1015 0.80 1.16 0.69 5.68 © 1013

1.28 © 1014 2.00 2.92 0.69 4.88 © 1013

1.67 © 1012 2.83 4.04 0.70 3.12 © 1013

1.28 © 1010 8.56 12.40 0.69 7.81 © 1012

Fig. 3 TEM micrographs of the specimens torsion deformed to an equivalent strain of 3.6 at (a) Z = 2.21 © 1026 s¹1 and(b) Z = 1.28 © 1014 s¹1.

S. Khamsuk et al.108

Z value in torsion deformation, the X-ray diffraction peak of(111) plane was selected and shown in Fig. 4(b). It can beclearly seen that the intensity of the X-ray diffraction peakdecreased and the peak was broadened with increasing Zvalue. Though the peak position seems to shift to lower angleside with decreasing the Z value, the accuracy and reason forthat are unclear at this moment. Generally, broadening of thepeak is caused by refinement of crystalline size and/orincreasing dislocation density. The dislocation density (μ)can be quantitatively estimated from the XRD pattern usingWilliamson­Hall method through following equations:14,15)

μ ¼ k¾2

Fb2ð5Þ

where k and F are constant values (16.1 for FCC materialand 1, respectively), b is the magnitude of Burgers vector(0.286 nm for aluminum), ¾ is the lattice strain described by{(¢cos ª/­)/(2sin ª/­)}, ¢ is the full-width at half maximumheight (FWHM) of the peaks, ­ is the wave length of X-rayand ª is the diffraction angle.

Figure 5 shows ¾ values obtained from plotting ¢cos ª/­of all the eight XRD peaks versus 2sin ª/­, where the latticestrain, ¾, is also indicated. The lattice strain values evaluatedfrom the XRD results were used for determining thedislocation density of all the specimens using eq. (5). Thedislocation density evaluated from the XRD results aresummarized in Table 3. It is found that the dislocationdensity increases with increasing Z value, which qualitativelycoincides with the TEM observations (Fig. 3). The disloca-tion densities evaluated from the XRD results are in the rangefrom 7.81 © 1012m¹2 to 9.56 © 1013m¹2.

(a)

(b)

Fig. 4 (a) X-ray diffraction patterns of the specimens after torsiondeformation by 1.1 rotations at various Z values. (b) X-ray diffractionpeaks of (111) plane of the specimens deformed at various Z values:(I) 2.21 © 1026 s¹1, (II) 1.69 © 1015 s¹1, and (III) 1.28 © 1014 s¹1.

Fig. 5 Williamson­Hall analysis of the specimens torsion deformed by 1.1 rotations at various Z values: (a) 2.21 © 1026 s¹1, (b) 1.69 ©1015 s¹1, (c) 1.28 © 1014 s¹1, and (d) 1.67 © 1012 s¹1. ¾ is the lattice strain used for calculating dislocation density.

Mechanical Properties of Bulk Ultrafine Grained Aluminum Fabricated by Torsion Deformation at Various Temperatures and Strain Rates 109

3.2 Mechanical properties of torsion deformedspecimens

The RT stress­strain curves of the specimens torsiondeformed by 1.1 rotations at various Z values are shown inFig. 6. The stress­strain curve of the starting (undeformed)specimen is also included in this figure for comparison andthe Z value and grain size are also indicated in Fig. 6.Here, Dall is indicated as the grain size. It was seen fromFig. 6(a) that the ultrafine grained materials showed veryhigh strength. For example, the yield stress (151MPa) andultimate tensile strength (195MPa) of the specimen withthe mean grain size of 0.38 µm were almost three timeshigher than those of the starting material (27 and 65MPa,respectively). However, it was also found that the flow curvesof the specimens with the mean grain sizes smaller than 1 µmreached the maximum stress at early stage of tensile test,resulting in a very limited uniform elongation. Similar resultshave been found in the UFG 1100Al produced by ARBprocess and subsequent annealing.16) For comparison, thestress­strain curves of the 1100Al ARB processed by 6cycles (equivalent strain of 4.8) at RT are shown in Fig. 6(b).The mean grain size (Dall)16) is also indicated in Fig. 6(b).It should be noted here that the UFG 1100Al fabricatedby torsion deformation exhibits different shapes of stress­strain curves from those of the UFG 1100Al produced byARB process and annealing. The ARB processed andannealed specimens having the mean grain sizes smaller

than 3.7 µm show discontinuous yielding characterized byyield-drop, while the UFG 1100Al produced by torsiondeformation does not show yield point phenomenon.

The 0.2% proof stress (·0.2), ultimate tensile strength(·UTS), uniform tensile elongation (eu) and total tensileelongation (et) were measured from all the stress­straincurves of the torsion deformed specimens and plotted as afunction of Z value or average grain size (Dall) in Fig. 7. It isfound that the strength increases with increasing Z value(Fig. 7(a)), in other words, with decreasing the grain size(Fig. 7(b)). The strength gradually increases with increasingZ, but it greatly increases at around Z = 1.69 © 1015 s¹1

(corresponding to the grain size of about 0.80 µm). Thiscorresponds to the change in microstructures from moreequiaxed morphologies to elongated lamellar morphologies,as was shown in Fig. 2. The tensile elongation significantlydecreases when the grain size is smaller than 1 µm. Similarresults have been observed in various kinds of UFG metallicmaterials fabricated by SPD and subsequent annealing, suchas pure aluminum,11) aluminum alloys7,17) and IF steel.18)

Actually, sudden drop of tensile elongation below grain sizeof 1 µm is also found in the ARB processed and annealed1100Al shown in Fig. 6(b). The low tensile ductility,especially the limited uniform elongation, of ultrafine grainedmetallic materials is attributed to high yield strength and lackof work hardening, which causes plastic instability (neckingin tensile deformation) occurring at very early stage of thetensile test. This can be described by the Considère’s criterionfor plastic instability, as

(a)

(b)

Fig. 7 The 0.2% proof stress (·0.2), tensile strength (·UST), uniformelongation (eu), and total elongation (et) of the torsion deformed 1100Al,plotted as a function of (a) Z parameter and (b) grain size (Dall).

(a)

(b)

Fig. 6 Room temperature stress­strain curves of the (a) 1100Al torsiondeformed by 1.1 rotations under various Z conditions, and (b) 1100AlARB processed by 6 cycles (equivalent strain of 4.8) at RT and annealedat various temperatures for 1.8 ks.17)

S. Khamsuk et al.110

d·

d¾� · ð6Þ

where · is the true flow stress, ¾ is the true strain and d·/d¾ isthe work hardening rate. UFG materials exhibit high flowstress (especially high yield strength) compared with thoseof the coarse grained materials. On the other hand, the workhardening is not enhanced by grain refinement. Therefore, thenecking starts at very low strain, which results in limitedtensile ductility.

The 0.2% proof stresses obtained from the tensile test ofthe torsion deformed specimens are also plotted as a functionof minus square root of the grain size (Dall) in Fig. 8 (Hall­Petch plot). The 0.2% proof stresses of the same materialreported by Kamikawa,16) i.e., the coarse grained specimensfabricated by conventional cold-rolling and annealingprocesses, and the UFG specimens fabricated by ARB andannealing processes (corresponding to Fig. 6(b)), are alsoplotted in Fig. 8 for comparison. One can clearly see fromthis figure that the UFG aluminum processed by torsiondeformation exhibits higher strength than that of theextrapolation of the Hall­Petch relationship for coarsegrained aluminum. In other words, the UFG aluminumexhibits extra-hardening. Extra-hardening has been reportedin the UFG materials including 1100Al fabricated by ARBand annealing processes,11,16) which is also plotted in Fig. 8.However, it is found in Fig. 8 that the extra-hardening ofthe UFG aluminum fabricated by torsion deformation issignificantly different from that of the ARB processed andannealed specimens, which seems to correspond with thedifference in the stress­strain behaviors between them showin Fig. 6. It has been clarified that the extra hardening ofthe UFG materials fabricated by SPD (ARB) and annealingprocesses is mainly attributed to the yield-point phenomenacharacteristically happening in UFG microstructures,19) andit is referred as dislocation source hardening.11) It is,however, not the case for the present UFG aluminumfabricated by torsion deformation, since the torsion speci-mens do not show obvious yield-point phenomena as shownin Fig. 6.

Probably the difference in the extra hardening behaviors iscaused by the difference in the microstructures of the UFGaluminum produced by torsion deformation and that by ARBprocess and annealing. It is known that the UFG aluminumfabricated by ARB process shows the lamellar morphologyhaving characteristics of deformation structures, i.e., elon-gated shapes, high density of dislocations and large numberof low angle boundaries.6) When the ARB processedaluminum is annealed at relatively low temperatures, theelongated morphology is maintained but the dislocationdensity decreases by recovery process with increasing theannealing temperature. Eventually, equiaxed grains free fromdislocations inside can be obtained.6,16) Such microstructuresare equivalent to fully recrystallized microstructures. On theother hand, the torsion deformed aluminum maintainsdeformation characteristics, although the morphology of theultrafine grains changes from elongated ones to moreequiaxed ones with decreasing Z value (Figs. 2 and 3). Infact, as shown in Table 3, the specimen torsion deformedunder the lowest Z condition (1.28 © 1010 s¹1) still maintainsthe dislocation density of 7.81 © 1012m¹2, which is signifi-cantly higher than that in fully recrystallized metallicmaterials (109­1010m¹2). Then, the extra hardening of thetorsion deformed aluminum is discussed quantitatively on thebasis of the substructures hereafter.

Now the strength of the UFG aluminum fabricated bytorsion deformation is estimated by adding various strength-ening mechanisms, in order to understand the extra hardeningin the torsion specimens. In metallic materials, several kindsof strengthening mechanisms can be considered, which aredislocation strengthening (strain hardening), grain boundarystrengthening, solution hardening and precipitation/disper-sion hardening.20) In the present commercial purity alumi-num, contributions of solution hardening and precipitation/dispersion strengthening can be ignored, as the amounts ofsolute atoms and precipitates are little. On the other hand,the EBSD, TEM and XRD results showed that the torsiondeformed aluminum contained fine-grained structure andhigh density of dislocations. Therefore, the yield strength(0.2% proof stress: ·0.2) of the torsion deformed specimens isestimated from the microstructural parameters, assuming theadditional law of the friction stress (·0), the contribution ofgrain boundary strengthening (·gb) and that of dislocationstrengthening (·d), i.e.,:

·0:2 ¼ ·0 þ ·gb þ ·d ð7ÞAccording to Hansen et al.,21) the contributions of grainboundary strengthening and dislocation strengthening in theUFG materials fabricated by SPD can be written as,

·0:2 ¼ ·0 þ kyd�1

2 þM¡Gbffiffiffiμ

p ð8Þwhere ky is the Hall­Petch slope, d is the mean grain size,M is Taylor factor (3.0),22) ¡ is a constant (0.24),23) G is theshear modulus (26GPa), b is the magnitude of Burgers vector(0.286 nm) and μ is the dislocation density. The value of22MPa is used as ·0.16) As ky, the value of 28MPaµm0.5

obtained by Kamikawa16) for coarse grained aluminum isused. DHAGB is used as the mean grain size (d) for thecalculation, which was summarized in Table 3 together withthe dislocation density (μ).

Fig. 8 Hall­Petch plot of the 0.2% proof stress of the 1100Al torsiondeformed under various Z conditions. The proof stress of coarse grained1100Al cold-rolled and annealed and that of 1100Al ARB processed andannealed (in Ref. 17)) are also plotted.

Mechanical Properties of Bulk Ultrafine Grained Aluminum Fabricated by Torsion Deformation at Various Temperatures and Strain Rates 111

In Fig. 9, the 0.2% proof stresses of the torsion deformedspecimens calculated according to eq. (8) using experimen-tally obtained DHAGB and μ are indicated as bars in theHall­Petch plot. The results are plotted for two kinds of themean grain size, (a) DHAGB and (b) Dall. The 0.2% proofstresses obtained from tensile testes are also plotted asclosed circles in Fig. 9 for comparison. It can be seen thatthe calculated 0.2% proof stress shows a relatively goodagreement with the experimental result, regardless of thekind of the mean grain size (DHAGB or Dall). The contributionof grain boundary strengthening (·gb) indicates the Hall­Petch relationship for coarse grained specimens. Therefore,it can be concluded that the extra hardening in the presentUFG aluminum fabricated by torsion deformation can beunderstood mainly by strengthening owing to the sub-structures observed within the microstructures (dislocationstrengthening: ·d).

Theoretically the similar calculation can be done for theARB processed and annealed specimens which showeddifferent tendency of extra-hardening in Fig. 8. However,the data of the ARB processed and annealed specimensshown in Fig. 8 are those previously studied in Ref. 17) andthe dislocation density data measured by the same method(XRD) are not available. However, ·0 + ·gb in Fig. 9coincides with the extrapolation of the Hall­Petch relation-ship for coarse grained materials. Furthermore, as was

shown in our previous studies (Refs. 6 and 16), themicrostructures of the ARB processed and annealedspecimens which have the mean grain sizes larger than1 µm are equivalent to fully recrystallized microstructures,so that the dislocation density within those grains must bevery low (probably below 1011m¹2). That is, at least aroundthe grain size of 1­2 µm, total lengths of the bars (·0 +·gb + ·d) for the ARB processed and annealed materialsmust be shorter than those for the torsion deformed materialsin Fig. 9, while the Hall­Petch line for the ARB materialslocates above that for the torsion materials in Fig. 8. Thisalso suggests that different strengthening mechanismsoperate in the torsion materials and the ARB materials,respectively.

4. Conclusions

In the present study, mechanical properties and micro-structures of the ultrafine grained aluminum fabricated bytorsion deformation were systematically investigated. Thestrengthening mechanisms of the ultrafine grained aluminumwere also discussed. The main results obtained are asfollows:(1) The various fine grain structures having mean grain

sizes ranging from 0.38 to 8.6 µm were obtained bytorsion deformation under various Z conditions. Theultrafine grained microstructures involved high densityof dislocations. The mean grain size decreased andthe dislocation density increased with increasing the Zvalue.

(2) The strength of the torsion deformed specimensincreased with increasing the Z value. The ultrafinegrained aluminum having submicrometer grain sizesexhibited very high strength. On the other hand, thetensile ductility of the submicrometer grained aluminumwas limited due to early plastic instability.

(3) The 0.2% proof stress of the ultrafine grained aluminumwas much higher than that predicted by conventionalHall­Petch relationship of aluminum extrapolated fromcoarse grained regions. The extra hardening behaviorof the torsion deformed aluminum was significantlydifferent from that of the ARB processed and annealedaluminum. The 0.2% proof stress of the torsionspecimens was evaluated from the microstructuralparameters experimentally obtained, and comparedwith the 0.2% proof stress experimentally obtained bytensile test. It was concluded that the extra hardeningin the torsion specimens was attributed mainly to thesubstructure strengthening.

Acknowledgements

This study was financially supported by the Grant-in-Aidfor Scientific Research on Innovative Area, “Bulk Nano-structured Metals” (area No. 2201), the Grant-in-Aid forScientific Research (A) (No. 24246114), and the ElementsStrategy Initiative for Structural Materials (ESISM), allthrough the Ministry of Education, Culture, Sports, Scienceand Technology (MEXT), Japan (contract No. 22102002),and the supports are gratefully appreciated.

(a)

(b)

Fig. 9 A comparison between the experimental and the calculated valuesof the 0.2% proof stress. Plotted for two kinds of mean grain sizes shownin Table 2: (a) DHAGB and (b) Dall.

S. Khamsuk et al.112

REFERENCES

1) M. A. Meyers, A. Mishra and D. J. Benson: Prog. Mater. Sci. 51 (2006)427­556.

2) S. Khamsuk, N. Park, H. Adachi, D. Terada and N. Tsuji: J. Mater. Sci.47 (2012) 7841­7847.

3) H. J. McQueen, S. Spigarelli, M. E. Kassner and E. Evangelista: HotDeformation and Processing of Aluminum Alloys, (Taylor & FrancisGroup, 2011) p. 59.

4) E. Q. Hall: Acta Mater. 64 (1951) 747­753.5) N. Petch: Prog. Metal Phys. 5 (1954) 1­52.6) N. Tsuji, Y. Ito and Y. Minamino: Scr. Mater. 47 (2002) 893­899.7) C. Y. Yu, C. K. Kao and C. P. Chang: Acta Mater. 53 (2005) 4019­

4028.8) R. Masumura, P. M. Hazzledine and C. S. Pande: Acta Mater. 46 (1998)

4527­4534.9) P. G. Sanders, J. Eastman and J. R. Weertman: Acta Mater. 45 (1997)

4019­4025.10) A. H. Chokshi, A. Rosen, J. Karch and H. Gleiter: Scr. Metall. 23

(1989) 1679­1683.11) N. Kamikawa, X. Huang, N. Tsuji and N. Hansen: Acta Mater. 57

(2009) 4198­4208.12) S. C. Shrivastava, J. J. Jonas and G. Canova: Mech. Phys. Solids 30

(1982) 75.13) J. J. Jonas, C. M. Sellars and W. J. Tegart: Int. Mater. Rev. 14 (1969)

1­24.14) G. K. Williamson and W. H. Hall: Acta Mater. 1 (1953) 22­31.15) G. K. Williamson and R. E. Smallman: Philos. Mag. 1 (1956) 34­46.16) N. Kamikawa: Ph.D. Thesis, Osaka University, (2005).17) X. Huang, N. Kamikawa and N. Hansen: Mater. Sci. Eng. A 483­484

(2008) 102­104.18) R. Ueji, N. Tsuji, Y. Minamino and Y. Koizumi: Sci. Tech. Adv. Mater.

5 (2004) 153­162.19) S. Gao, M. Chen, N. Kamikawa, A. Shibata and N. Tsuji: Mater. Trans.

55 (2014), No. 1, in press.20) W. D. Callister and D. G. Rethwisch: Materials Science and

Engineering, Chapter 7. Dislocations and Strengthening Mechanisms,7th Edition, (New York Wiley, 2007) pp. 197­233.

21) N. Hansen, X. Huang, R. Ueji and N. Tsuji: Mater. Sci. Eng. A 387­389 (2004) 191­194.

22) J. Gubicza, G. Dirras, P. Szommer and B. Bacroix: Mater. Sci. Eng. A458 (2007) 385­390.

23) N. Hansen and X. Huang: Acta Mater. 46 (1998) 1827­1836.

Mechanical Properties of Bulk Ultrafine Grained Aluminum Fabricated by Torsion Deformation at Various Temperatures and Strain Rates 113