Improvement of Uniform Elongation by Low Temperature Annealingin Al-2.5%Mg Alloy Processed by Accumulative Roll Bonding

Keizo Kashihara1,+1, Yoshikazu Komi1, Daisuke Terada2,+2 and Nobuhiro Tsuji2

1National Institute of Technology, Wakayama College, Gobo 644-0023, Japan2Kyoto University, Kyoto 606-8501, Japan

Al-2.5%Mg alloy (A5052) sheets were processed by accumulative roll bonding (ARB) for 1 to 7 cycles (equivalent strains from 0.8 to 5.6)at room temperature. The sheets processed by ARB for 7 cycles were then annealed isochronally for 30min at temperatures in the range from100°C to 400°C. Interestingly, it was found that the specimen annealed at 200°C followed by 7 cycles of ARB had the same level of yield stress(about 320MPa) but a larger uniform elongation than the specimen processed by 3 cycles of ARB. The improvement in uniform elongation bylow-temperature annealing is discussed in terms of the mechanism that the evolution of dislocation substructures inside ultra-fine grains causesplastic instability at a very early stage of the tensile test. Hardening by annealing was also observed in the specimen annealed at 100°C followedby 7 cycles of ARB. [doi:10.2320/matertrans.L-M2015806]

(Received December 26, 2014; Accepted March 21, 2015; Published May 15, 2015)

Keywords: aluminum magnesium alloy, accumulative roll bonding, severe plastic deformation, mechanical property, low temperatureannealing, electron backscatter diffraction, ultrafine grains

1. Introduction

Accumulative roll bonding (ARB) is a severe plasticdeformation method, and is capable of producing plates oflarge dimensions.1) ARB-processed materials with a grainsize below 1µm have extremely high strength in tensiletests but relatively low ductility, in particular, low uniformelongation. Therefore, as-ARB-processed materials show afracture at a very early stage of tensile tests, following tomacroscopic necking.2,3)

In ARB-processed materials annealed at various temper-atures, the uniform elongation recovers as deformedstructures are removed and the grain size increases duringannealing. Total elongation can become more than 10%, ifthe recrystallized grains larger than 1 µm having equiaxedmorphologies and free from dislocations are formed.4,5)

Since Al-Mg alloys, which are non-heat-treatable alumi-num alloys, are generally strengthened by solid solutionhardening and work hardening (dislocation strengthening),there is a great interest in hardening by ARB and softeningby annealing. There are several reports on the mechanicalproperties of ARB-processed Al-Mg alloys. For example,Tsuji et al. investigated the occurrence of super-plasticity inA5083 Al-Mg alloy, processed by ARB to an equivalentstrain of 4.0 at 200°C, then deformed under tensionat temperatures in the range from 200°C to 400°C.6)

Toroghinejad et al. examined the structure, strength andductility of A5083 Al-Mg alloy processed by ARB to anequivalent strain of 4.8 at room temperature.7) However, thereare few reports on the tensile properties of annealed Al-Mgalloys that were ARB-processed at room temperature.

The aim of the present study is to clarify the relationshipbetween tensile properties and microstructure during anneal-ing in an Al-2.5%Mg alloy (A5052) deformed by ARB atroom temperature.

2. Experimental Procedures

The starting material used in this study is a commercialA5052 alloy. The chemical composition is shown in Table 1.A5052 plates with dimensions of 0.5mm in thickness,120mm in length and 30mm in width were fully annealed at345°C for 30min. The average grain size after annealing was12.8 µm.

ARB was carried out at room temperature using a two-highrolling mill; the roll diameter was 70mm and the peripheralspeed was 100mm·s¹1. After degreasing and wire-brushingthe contact surfaces, two sheets were stacked (1mm overall),and then roll-bonded to 50% reduction in thickness(equivalent strain of 0.8) without use of lubricant. Thebonded sheet was cooled in water immediately after rolling.This process is denoted as 1 cycle of ARB. In this study, theARB process was repeated up to 7 cycles. The number ofARB cycles, total reduction in thickness, and equivalentstrain are shown in Table 2.

Specimens for tensile tests were cut from the ARB-processed sheets after 1, 3, 5, and 7 cycles using an electricdischarge apparatus. These specimens were denoted asspecimens 1c, 3c, 5c, and 7c, respectively. The specimencut from a fully annealed starting sheet was calledspecimen 0c. The tensile specimens had a gage length of10mm and a width of 5mm. The tensile axis of the specimen

Table 1 Chemical compositions of A5052 (mass%).

Si Fe Cu Mn Mg Cr Zn Ti Al

0.11 0.25 0.03 0.06 2.5 0.19 0.02 0.02 Bal.

Table 2 Number of ARB cycles, total reduction, and equivalent strain.

Number of cycles 1 3 5 7

Total reduction (%) 50 87.5 96.9 99.2

Equivalent strain 0.8 2.4 4 5.6

+1Corresponding author, E-mail: kashihara@wakayama-nct.ac.jp+2Present address: Chiba Institute of Technology, Tsudanuma 275-0016,Japan

Materials Transactions, Vol. 56, No. 6 (2015) pp. 803 to 807©2015 The Japan Institute of Light Metals

was parallel to the rolling direction (RD) of the sheet. Tensiletests were performed at an initial strain rate of 8.3 © 10¹4 s¹1

at room temperature, and tensile strain was measured by anextensometer. Two tensile tests were carried out for eachspecimen condition.

Specimen 7c was subjected to isochronal annealing treat-ments for 30min at temperatures in the range from 100°C to400°C (50°C steps). Vickers hardness tests were carried outon a rolling plane under a load of 0.98N. Hardness wasmeasured randomly at five points, and then the meanhardness value was calculated. The specimens annealed at100°C, 200°C, and 300°C were used for the tensile tests, andwere called specimens 7c-100, 7c-200, and 7c-300, respec-tively. The tensile tests were performed under the sameconditions as described above.

To examine the relationship between tensile properties andmicrostructures, specimens 3c and 7c-200 were examinedby electron backscatter diffraction (EBSD), using a field-emission scanning electron microscope (FE-SEM). Themicrostructure was observed in the mid-thickness area on alongitudinal section normal to the transverse direction of thesheet. The step size for EBSD was 40 nm. The present studyregards boundaries having misorientation larger than 15° ashigh angle boundaries.

3. Results

Figure 1 shows the stress-strain curves of the as-ARB-processed specimens 0c to 7c. The curves indicated that theflow stress increased and the total elongation decreased withincreasing strain. The stress-strain curves of the specimens7c-100, 7c-200, and 7c-300 are also shown. It is interestingthat yielding behavior changes from continuous one in thespecimen 0c to more discontinuous one having a slight yield-drop in the specimen 7c-300.

Figure 2 shows the ultimate tensile strength (·B), 0.2%proof stress (yield stress, ·0.2), total elongation (et), anduniform elongation (eu). With the first ARB cycle, theuniform and total elongations decrease abruptly, and thendecrease gradually with further ARB cycles. The yield stressis 107MPa and 405MPa in the specimens 0c and 7c,respectively. The ultimate tensile strength for the specimens0c and 7c are 210MPa and 422MPa, respectively. With 7cycles of ARB (an equivalent strain of 5.6), the as-ARB-processed A5052 alloy shows the yield stress 3.8 timeshigher than that of the fully annealed specimen 0c and theultimate tensile strength 2 times higher than the specimen 0c.The tendency shown in Fig. 2 agrees well with the ultimatetensile strength and total elongation in the ARB-processedA1100 aluminum previously reported.2)

Figure 3 shows the change in hardness for the 7cspecimens annealed for 30min at temperatures in the rangefrom 100°C to 400°C. Hardness increases at 100°C, thendecreases with increasing the annealing temperature. Thecurve indicates that recrystallization is almost completed at300°C.

Here, the strength and elongation of the as-ARB-processedspecimens are compared with those of the ARB-processedand annealed specimens. In Fig. 1, the specimen 7c-100 (as-annealed one) has the higher ultimate tensile strength than the

specimen 7c (as-ARB-processed). In addition, in Fig. 3, thehardness of the specimen 7c-100 is greater than that ofthe specimen 7c. These results suggest that hardening byannealing, which was reported by Huang et al.,8) occurred inthe specimen 7c-100.

Fig. 1 Stress-strain curves of the specimens processed by ARB to 0, 1, 3,5, and 7 cycles, and the specimens annealed at 100°C, 200°C, and 300°Cfollowing to 7 cycles of ARB.

Fig. 2 Ultimate tensile strength (·B), 0.2% proof stress (yield stress, ·0.2),total elongation (et), and uniform elongation (eu) of the ARB-processedspecimens.

Fig. 3 Change in hardness for the 7c specimens annealed for 30min attemperatures in the range from 100°C to 400°C.

K. Kashihara, Y. Komi, D. Terada and N. Tsuji804

To verify the effect of annealing on strength and ductility,the relationship between the yield stress and uniformelongation in the as-ARB-processed specimens and theannealed specimens is shown in Fig. 4. The yield stress ofthe specimens 3c and 7c-200 is 325MPa and 327MPa,respectively, while the uniform elongation is 1.4% in thespecimen 3c and 3.7% in the specimen 7c-200. Comparingthe uniform elongation of the specimen 7c-200 with that ofthe specimen 3c having the same level of the yield stress,it is obvious that the ductility of the specimen 7c-200 wasimproved by low-temperature annealing at 200°C. We willfocus on how the microstructures in the specimens 3c and7c-200 lead to the same level of yield stress, but produce adifference in uniform elongation.

Figure 5 shows high angle boundary maps obtained byEBSD analysis, depicting boundaries with misorientationlarger than 15° by black solid lines. The specimen 3cpossesses the microstructure in which original grains areelongated along RD. Finer grains are partly formed probablyby grain subdivision mechanism during the ARB.9) Themicrostructure of the specimen 7c-200 shows ultra-fine andpancake-shaped grains. Recrystallization has not yet oc-curred. The average boundary spacing (dND) along the normaldirection (ND) was measured by an intersection method. Theboundary spacing is 0.65 µm in the specimen 3c and 0.29 µmin the specimen 7c-200. Taylor factor M calculated from theEBSD data was 3.17 for the specimen 3c and 3.28 for thespecimen 7c-200.

4. Discussions

The mechanical properties of the specimen annealed at200°C following to 7 cycles of ARB (specimen 7c-200) werecompared with those of the specimen processed by 3 ARBcycles (specimen 3c, as-ARB-processed). The specimen 7c-200 had almost the same yield stress as the specimen 3c(about 320MPa). On the other hand, the uniform elongationof the specimen 7c-200 was larger than that of thespecimen 3c.

In general, ARB-processed materials have extremely highstrength. However, their uniform elongation is low because

macroscopic necking occurs at a very early stage of tensiletests.2,3) Ultra-fine grained structure developed during ARBprocessing contains dislocation substructures inside the ultra-fine grains, so that not only grain boundary strengtheningbut also dislocation strengthening should be considered inevaluating the strength of such materials. Tsuji et al. pointedout that the existence of dislocation substructures inside ultra-fine grains increases the flow stress of materials, but ratherdecreases the work hardening rate (d·/d¾).4) When workhardening is suppressed, plastic instability takes place at anearly stage of tensile tests and limits uniform elongation to alow value.

Based on the Hall-Petch equation, the effect of grainboundary strengthening for the specimens 3c and 7c-200is evaluated. The relationship between the yield stress ·0.2and the grain size d is given by the following Hall-Petchequation,10,11)

·0:2 ¼ ·0 þ kd�12 ð1Þ

where ·0 is the friction stress, and k is the Hall-Petch slope.Furukawa et al. reported that these coefficients are ·0 =62MPa and k = 0.149MPa·m1/2 for the ultra-fine grained Al-3%Mg alloy processed by equal-channel angular pressing.12)

The microstructures of the specimen 3c and specimen 7c-200 shown in Fig. 5 consisted of fine grains elongated alongRD. The aspect ratios of the grains (dRD/dND), where dRD anddND were the average high-angle boundary spacings alongRD and ND, respectively, were more than 10. It is difficult totranslate grains with very high aspect ratios into the sphericalgrain sizes as used in eq. (1). In addition, since the tensileaxis of the specimens is parallel to RD of the sheet, the effectof dND on the yield stress is stronger than that of dRD.Therefore, in this study, dND is conventionally used as thegrain size d in eq. (1).

Table 3 shows the yield stresses calculated from eq. (1)and those actually measured in the tensile tests for the

Fig. 4 Relationship between the 0.2% proof stress (yield stress) anduniform elongation of the specimens processed by ARB to 0, 1, 3, 5, and7 cycles, and of the specimens annealed at 100°C, 200°C, and 300°Cfollowing to 7 cycles of ARB.

(a)

(b)

Fig. 5 High angle boundary maps depicting boundaries having misor-ientation larger than 15° of (a) the specimen 3c and (b) the specimen 7c-200 obtained from EBSD data. Black solid lines indicate high angleboundaries. RD indicates rolling direction; ND, normal direction.

Improvement of Uniform Elongation by Low Temperature Annealing in Al-2.5%Mg Alloy Processed by Accumulative Roll Bonding 805

specimen 3c and the specimen 7c-200. The calculated yieldstress is 246.8MPa for the specimen 3c and 338.7MPa forthe specimen 7c-200. That is, the calculated value forspecimen 3c is lower by ¦·0.2 = 91.9MPa.

The tensile tests showed the almost identical yield stressesfor the specimens 3c and 7c-200, as also shown in Table 3.However, in the calculations using eq. (1), the yield stress ofthe as-ARB-processed specimen (3c) was lower than that ofthe as-annealed specimen (7c-200). This means that thepredicted difference of the yield stress (i.e.¦·0.2 = 91.9MPa)is attributable to the difference in dislocation strengtheningbetween the specimen 3c and the specimen 7c-200.

According to Kamikawa et al., the yield stress of ARB-processed materials is expressed by the following equation,13)

·0:2 ¼ ·0 þM¡Gbffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiµ0 þ µdis

pþ kd�

12 ð2Þ

where µ0 is the dislocation density within ultrafine grains andµdis is the dislocation density stored in low-angle dislocationboundaries. The sum of these two dislocation densitiescontributes to dislocation strengthening. M is the Taylorfactor (M = 3.06 in random texture, M = 3.17 for thespecimen 3c), ¡ is a constant (¡ = 0.24), G is the shearmodulus (G = 26GPa), and b is the magnitude of theBurgers vector (b = 0.286 nm).

The relative difference in dislocation strengthening,corresponding to ¦(µ0 + µdis), is calculated from eq. (2),using the difference of the yield stress (¦·0.2 = 91.9MPa)obtained from eq. (1). The value of ¦(µ0 + µdis) for thespecimen 3c is estimated at 2.64 © 1014m¹2.

In this study, no direct observation of dislocation sub-structure was carried out. However, based on the comparisonbetween the measured and calculated yield stresses, it isobvious that the effect of dislocation strengthening is greaterin the specimen 3c than in the specimen 7c-200.

The larger uniform elongation for the specimen 7c-200than for the specimen 3c can be explained using theoccurrence of plastic instability, caused by the evolution ofa dislocation substructure inside ultra-fine grains.4) If thedislocation substructures that limit further work-hardeningdisappear in low-temperature annealing, the ductility wouldbe improved, like that in the specimen 7c-200. The presentresult indicates that the improvement of ductility can beexpected not only in the materials consisting of fullyrecrystallized equiaxed grains larger than 1 µm, as has beenreported in previous studies,4,5) but also in the materialsexhibiting highly elongated grains having boundary spacingless than 1 µm, as shown in this study.

Finally, we estimate the value of µ0 + µdis in eq. (2) for thespecimen 7c-200 under the assumption that by the annealing

at 200°C, thermally unstable dislocations introduced bydeformation move and disappear or rearrange to form low-angle dislocation boundaries, which are incorporated into apart of the low dislocation density (µdis). Figure 6 showsa low and high angle boundary map for the specimen7c-200 (the step size is 10 nm). Low angle boundaries withmisorientation from 2° to 15° and high angle boundarieslarger than 15° are depicted by red and black lines,respectively. The length of low angle boundaries was165.7 µm in the observation area of 25 µm2. Since EBSDpatterns are taken within the depth of 40 nm from thespecimen surface,14) µdis is estimated at 1.66 © 1014m¹2. Thisvalue corresponds to 75.4MPa in stress, based on eq. (2),(¡ = 0.24, G = 26GPa, b = 0.286 nm, and M = 3.28).Consequently, the sum of stresses for the specimen 7c-200,including the friction stress and the Hall-Petch stress,becomes 414.1MPa; it was larger than the measured yieldstress (327MPa in Table 3). This difference may be ascribedto two factors. First, the present study used Al-2.5%Mgalloy, whereas Furukawa et al. used Al-3%Mg alloy;12) thisoverestimates the value of ·0 for the present calculation, dueto the effect of solute concentration. Second, the boundaryspacing along ND (dND) is used as the grain size (d) for thepresent calculation, and the grain size is underestimated thanthe actual grain size in the microstructure heavily elongatedalong RD (Fig. 5).

5. Conclusions

(1) Highly elongated lamellar grain structures were ob-served in the Al-2.5%Mg specimens processed by 3cycles of ARB (specimen 3c), and the specimenannealed at 200°C following to 7 cycles of ARB(specimen 7c-200). The average boundary spacingalong the normal direction of the sheets was 0.65 µmin the specimen 3c and 0.29 µm in the specimen 7c-200.

(2) The specimen 7c-200 had the same level of yieldstress as the specimen 3c (about 320MPa), while thespecimen 7c-200 showed larger uniform elongation.Recrystallization did not occur in the specimen 7c-200.

Table 3 The 0.2% proof stress (·0.2 = yield stress), calculated from theHall-Petch equation (1), and that measured in tensile tests of thespecimen 3c and the specimen 7c-200. dND denotes the high-angleboundary spacing along the normal direction; an EBSD step size of 40 nmwas used in determining dND.

Calculated value Measured value

Specimen dND ·0.2 ·0.2

3c 0.65 µm 246.8MPa 325MPa

7c-200 0.29 µm 338.7MPa 327MPa

Fig. 6 A low and high angle boundary map for the specimen 7c-200. Redlines indicate low angle boundaries with misorientation from 2° to 15°,and black lines high angle boundaries larger than 15°.

K. Kashihara, Y. Komi, D. Terada and N. Tsuji806

That is, it was found that ductility of the ARB processedmaterial could be improved by low-temperature anneal-ing at 200°C, even though the elongated lamellarstructures having boundary spacing less than 1 µm weremaintained.

(3) The improvement in the uniform elongation by low-temperature annealing could be explained using themechanism that low-temperature annealing suppressedthe early plastic instability.

(4) The occurrence of hardening by annealing8) wasconfirmed in the specimen annealed at 100°C after 7cycles of ARB.

Acknowledgements

K.K. acknowledges financial support from the Light MetalEducational Foundation, Inc. and technical help by F.Murabayashi and T. Maeda. D.T. and N.T. thank the Grant-in-Aid for Scientific Research on Innovative Area, “BulkNanostructured Metals”, from the Ministry of Education,Culture, Sports, Science and Technology of Japan, for theirsupport.

REFERENCES

1) Y. Saito, H. Utsunomiya, N. Tsuji and T. Sakai: Acta Mater. 47 (1999)579­583.

2) Y. Saito, N. Tsuji, H. Utsunomiya, T. Sakai and R. G. Hong: Scr. Mater.39 (1998) 1221­1227.

3) N. Tsuji: Nanostructured Metals and Alloys, ed. by S. H. Wang,(Woodhead Publishing, 2011) p. 53.

4) N. Tsuji, Y. Ito, Y. Saito and Y. Minamino: Scr. Mater. 47 (2002) 893­899.

5) N. Kamikawa, N. Tsuji and Y. Saito: Tetsu-to-Hagane 89 (2003) 273­280.

6) N. Tsuji, K. Shiotsuki and Y. Saito: Mater. Trans. JIM 40 (1999) 765­771.

7) M. R. Toroghinejad, F. Ashrafizadeh and R. Jamaati: Mater. Sci. Eng. A561 (2013) 145­151.

8) X. Huang, N. Hansen and N. Tsuji: Science 312 (2006) 249­251.9) N. Hansen and D. J. Jensen: Philos. Trans. Roy. Soc. Lond. A 357

(1999) 1447­1469.10) E. O. Hall: Proc. Phys. Roy. Soc. B 64 (1951) 747­753.11) N. J. Petch: J. Iron Steel Inst. 174 (1953) 25­27.12) M. Furukawa, Z. Horita, M. Nemoto, R. Z. Valiev and T. G. Langdon:

Philos. Mag. A 78 (1998) 203­215.13) N. Kamikawa, X. Huang, N. Tsuji and N. Hansen: Acta Mater. 57

(2009) 4198­4208.14) S. Suzuki: Materia Japan 40 (2001) 612­616.

Improvement of Uniform Elongation by Low Temperature Annealing in Al-2.5%Mg Alloy Processed by Accumulative Roll Bonding 807