Scripta Materialia 54 (2006) 11751180 www.actamat-journals.

com

Eect of strain rate on the ductility of a nanostructured aluminum alloy

B.Q. Han
a c

a,*

, J.Y. Huang b, Y.T. Zhu c, E.J. Lavernia

Department of Chemical Engineering and Materials Science, University of California, Davis, CA 95616, United States b Department of Physics, Boston College, Chestnut Hill, MA 02467, United States Materials Science and Technology Division, Los Alamos National Laboratory, Los Alamos, NM 87545, United States Received 13 September 2005; received in revised form 24 October 2005; accepted 17 November 2005 Available online 15 December 2005

Abstract Nanostructured materials normally exhibit enhanced ductility at higher strain rate. In this paper we report the observation that a cryomilled nanostructured 5083 Al alloy exhibits the opposite behavior, i.e., higher ductility at lower strain rates. This phenomenon was rationalized by a diusion-mediated stress relaxation mechanism that eectively delayed crack initiation events. 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Nanostructured; Aluminum alloy; Cryomilling; Ductility; Strain rate

1. Introduction The mechanical properties of nanostructured or ultrane-grained (UFG) materials are of interest within the context of nanotechnology as applied to structural materials [1]. There are a number of reported trends in behavior that deserve detailed study since they may provide insight into the underlying mechanisms. For example, a strong inuence of strain rate on mechanical properties has been observed in many nanostructured/UFG materials [24]. Most studies have reported an increase in tensile strength and ductility of nanostructured materials with increasing strain rates [3,57]. Deformation at higher strain rates has been proposed as a strategy for improving the ductility of nanostructured materials [3]. Indeed, face-centered cubic (fcc), body-centered cubic (bcc) and hexagonal close-packed (hcp) metals have been found to exhibit higher ductility at higher strain rates [6]. The accepted reason for improved ductility at higher strain rates is higher work hardening rates [8]. Work hardening delays necking, the onset of failure under tension for most metals and alloys. It is caused by the accumu*

Corresponding author. Tel.: +1 530 752 9568; fax: +1 530 752 9554. E-mail address: bqhan@ucdavis.edu (B.Q. Han).

lation of crystalline defects, such as dislocations, which make further deformation more dicult [9]. At the same time, dynamic recovery lowers the defect density, and consequently lowers the work hardening rate. Higher strain rates generate crystalline defects at a higher rate to compete with dynamic recovery and therefore increase the work hardening rate, which leads to higher ductility. Another strategy for improving the ductility of nanostructured fcc metals is to increase the strain rate sensitivity [3]. Despite the above ndings, published results sometimes reveal contradictory behavior. For example, a nanostructured Cu processed via ball milling was found to have a higher ductility at lower strain rates, while its tensile strength follows the same trend as other nanostructured materials, i.e., lower strength at lower strain rates [4]. The higher ductility at lower strain rates in the nanostructured Cu [4] cannot be explained by the above two strategies and therefore it likely involves a new mechanism. It is the objective of this study to investigate the mechanism of higher ductility at lower strain rate in nanostructured metals. We selected nanostructured 5083 Al alloy produced by cryomilling and subsequent consolidation via hot extrusion for this study as a result of recent interest in this synthesis route. For the past several years, cryomilling has evolved into a viable technology for processing UFG or

1359-6462/$ - see front matter 2005 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.scriptamat.2005.11.035

1176

B.Q. Han et al. / Scripta Materialia 54 (2006) 11751180

nanostructured materials for engineering applications [10 12]. However, in spite of numerous published studies on cryomilled materials, the inuence of strain rate on their ductility has not been investigated. As shown later, this material exhibits higher ductility at lower strain rate, similar to the nanostructured Cu produced by ball milling. 2. Materials and experimental procedures The cryomilled 5083 Al alloy (Al4.2 wt.%Mg0.67 wt.%Mn) was produced following the processing sequence of cryomilling, cold isostatic pressing and hot extrusion. Spray-atomized powder was mechanically milled for 8 h in a liquid nitrogen slurry. The cryomilled powder was canned and compacted by cold isostatic pressing at a pressure of $400 MPa. The compacts were extruded into round bars with a diameter of 19 mm at 823 K. High-resolution electron microscopy (HREM) of a JEOL 3000F microscope was operated at 300 kV. HREM samples were prepared via mechanical grinding using sandpapers with dierent grades to a thickness of approximately 50 lm. Final thinning and perforation were conducted with ion milling until a small perforation was made in the center. The as-extruded materials were machined along the extrusion direction into at dog-bone specimens with a gauge length of 12 mm, a width of 4 mm and a thickness of 2 mm. Tensile tests at various crosshead velocities from 0.012 mm/s to 0.0001 mm/s until failure were performed using a universal testing machine, loaded with a load cell with an accuracy within 0.5% of the indicated load and a dual-camera video extensometer to directly measure the displacement of the tensile gauge section with an accuracy of 5 lm. The fracture surface of the tensile specimens was studied using a scanning electron microscope (SEM) operated at 10 kV. A compression specimen of 5 mm cubes was loaded along the extrusion direction at a crosshead velocity of 0.0005 mm/s. 3. Experimental results The microstructure of the as-extruded cryomilled 5083 Al alloy is shown in Fig. 1(a), a section taken normal to the extrusion direction. The histogram of the grain size distribution of the as-extruded cryomilled 5083 Al alloy was reported in a previous study [13]. The mean grain size is about 207 nm, measured from %200 g. It is noteworthy that while most of grains are in the size range of 100 nm and 200 nm, some of the grains are smaller than 100 nm. Although micrometer-sized Al6Mn is usually observed in coarse-grained conventional 5083 Al alloy [14], this phase is absent in the cryomilled 5083 Al alloy. As shown in Fig. 1(a), a few particles (marked by arrows) with sizes from 2 nm to 10 nm are also observed in the microstructure. In a recent microstructure investigation of a cryomilled 5083 Al alloy, four types of second phases are identied, i.e., grain boundary MgO oxides, precipitates of Al12Mg2(CrMnFe) and Al12(FeMn)3Si, particles of Al6(CrMnFe), and AlSiO dispersoids [15]. We expect these second phases to be present in our current sample as well, given that the material and processing history are identical to the ones used in Ref. [15]. Although the volume fraction of these dispersoids in the microstructure is not high, they play a signicant role in stabilizing of microstructure and strengthening of cryomilled Al alloys [10,16]. Fig. 1(b) shows an atomic resolution HREM image of an area near a grain boundary. Marked by a white arrow is a stacking fault, which is frequently observed at grain interiors. Both screw and 60 dislocations are found, typical of fcc metals subjected to heavy deformation [17]. The dislocation density measured from inverse Fourier transformation images is about 8 1015 m2. Typical tensile stressstrain curves at strain rates of 103 s1 to 8.3 106 s1 of the extruded cryomilled 5083

Fig. 1. Microstructure of the as-extruded cryomilled 5083 Al alloy in a section normal to the extrusion direction.

B.Q. Han et al. / Scripta Materialia 54 (2006) 11751180

1177

Al alloys are shown in Fig. 2(a). For easy comparison of the stressstrain curves at dierent strain rates, they are plotted with a strain interval of 0.5%. The Youngs modulus calculated from the elastic portion of true stressstrain curves is $69.0 GPa, which is almost identical to the modulus value reported for pure Al [18]. The video extensome-

ter allows the measurement of displacement of the tensile gage section without mechanical contact with the tensile specimen. This method is much more accurate for measuring the strain as compared to the cross-head displacement method employed by most researchers. The high accuracy of the video extensometer is also demonstrated by the accurate determination of the Youngs modulus. Although only one sample was measured at each rate, the systematic variation of elongation with strain rate provides support to the premise that the measured results cannot be attributed to experimental scatter. The stressstrain curves reveal that the uniform deformation increases with decreasing strain rate. However, even at the lowest strain rate, the sample failed immediately after the peak stress. Inspection of the strain rate dependence of strength and elongation, as shown in Fig. 2(b), reveals that while the yield strength increases slightly, the ultimate tensile strength and elongation decreases with increasing strain rate. Fig. 2(c) shows the compression behavior and its comparison with the tensile behavior at the strain rate of 104 s1, which is plotted with a true strain interval of 0.01. The compression stressstrain curve showed a plastic strain of >0.2 with an approximately zero work hardening, typical of a nanostructured metal [19]. Fig. 3(a) shows a side view of the fracture surface of a tensile specimen at a strain rate of 103 s1. It reveals columnar features shearing out of the side surfaces. The top view of the fracture surface of a tensile specimen tested at a strain rate of 104 s1 (Fig. 3(b)) reveals several large cracks that are several hundred micrometers long. Closer inspection of the fracture surface of the specimen tested at a strain rate of 104 s1 (Fig. 3(c) and (d)) reveals many small cracks that are a few tens of micrometers long. Crack-free clusters surrounding the minor cracks are $20 lm in size, close to the size of cryomilled 5083 Al powders. The fracture surface of a tensile specimen tested at a strain rate of 8.3 106 s1 is shown in Fig. 4. The side view of the fracture surface shows similar columnar morphology (Fig. 4(a)). However, the top view of fracture surface (Fig. 4(b)) does not reveal the presence of any large cracks. As shown in Fig. 4(c) and (d), small cracks of a few tens micrometers were found on the fracture surfaces. Also, the density of the small cracks is much lower than that observed in Fig. 3. Although traditional dimple morphology was not observed, the fracture surface is quite similar to that of ductile fracture in mechanically alloyed Al alloys [12]. 4. Discussion Inspection of earlier studies on the mechanical behavior of nanocrystalline materials reveals that the residual porosity and impurity in nanocrystalline materials have a significant eect on their properties [20,21]. For instance, a nanocrystalline AlZr alloy processed by inert gas condensation and warm compaction has low tensile ductility

Fig. 2. (a) The tensile stressstrain curves, (b) the strain rate dependence of strength and elongation and (c) comparison of tension and compression stressstrain curves of the cryomilled 5083 Al alloy.

1178

B.Q. Han et al. / Scripta Materialia 54 (2006) 11751180

Fig. 3. Fractography of the cryomilled 5083 Al alloy at strain rates (a) 103 s1 and (b)(d) 104 s1.

Fig. 4. Fractography of the cryomilled 5083 Al alloy at a strain rate of 8.3 106 s1.

because of residual porosity [20]. The nanostructured 5083 Al alloy processed by cryomilling and hot extrusion has an improved but similar tensile behavior to these nanocrystalline alloys because of a higher consolidation density.

Inspection of the stressstrain curves of the 5083 Al alloy reveals that the ow stress increases during plastic deformation, but the samples are fractured in the work hardening region with very small elongation to failure. Although

B.Q. Han et al. / Scripta Materialia 54 (2006) 11751180

1179

the tensile elongation is small, we believe that it is a result of plastic deformation mediated by dislocation activity. This conclusion is drawn from the similarity between the tensile curve and the compression curve at the same strains. Careful examination of Fig. 2(c) reveals that the early strain hardening behavior in the tensile stressstrain curve is similar to that in the compression stressstrain curve. This similarity provides support to the notion that both were inuenced by the same mechanism. In other words, strain hardening in both the tensile and compression stressstrain curves was caused by dislocation accumulation. The large plastic deformation that was observed during compression indicates that dislocation activity played a signicant role during compression deformation. Fig. 2 clearly shows that the nanostructured 5083 Al alloy has higher tensile ductility and lower yield strength when tested at lower strain rates. SEM images (Fig. 3) of the fracture surfaces at fast strain rates of 103 s1 and 104 s1 reveal extensive cracks. In contrast, the specimens tested at a low strain rate of 106 s1 have fewer and smaller cracks. These observations indicate that the sample fracture was controlled by crack initiation and growth, and that the primary reason for improved ductility at lower strain rates is the delayed crack initiation/lower growth rate. The nanostructured 5083 Al alloy also shows a higher ultimate tensile strength at lower strain rate (Fig. 2), although its yield strength decreased with decreasing strain rate. The higher ow stress at slower strain rates in a conventional 5083 Al alloy is attributed to the eect of dynamic strain aging (DSA), or the PortevinLe Chatelier eect, which is caused by the interactions of dislocations with solute atoms [2224]. During DSA, a lower strain rates allows fast diusing solute atoms to diuse to dislocations in motion, which exerts a dragging force on moving dislocations and raise the ow stress. The DSA should cause negative strain rate sensitivity [22], which is indeed observed in the nanostructured 5083 Al alloy studied here (data to be presented in a future publication). As discussed above, the nanostructured 5083 alloy exhibits both improved ductility and DSA at lower strain rates. A similar phenomenon has also bee reported in a 5086 Al alloy [23]. This raises one question: is DSA responsible for the improved ductility at lower strain rate? For DSA to occur, it is necessary to have fast diusion of the solute atoms. Cryomilling introduces small impurity atoms such as H, O, C and N [10,16,25], which can diuse readily because of their small size. These impurity atoms, as well as solute elements, should certainly promote the DSA behavior during mechanical testing. However, diusion of these atoms to moving dislocations makes it more dicult for the dislocation to move, and may not help prevent crack initiation and growth. In practice, the higher ow stress caused by DSA may actually increase the driving force for crack initiation. On the basis of the present experimental ndings, as well as on others reported in the literature [4], we propose a diffusion-mediated stress relaxation mechanism to explain the

observed higher ductility at lower strain rates. The rationale used to arrive at this proposed mechanism is discussed below. Cracks should initiate at local stress concentration sites such as nanoscale dispersoids, grain boundaries, interfaces, microcracks and other defects. Stress concentration often exists at these sites, making their local stresses much higher than the applied external stresses. Lower strain rates will allow solute atoms, especially faster diusing interstitial atoms, to diuse into these sites and relax the local stress so that it does not reach the critical stress level required for crack initiation. The high density of dislocations near the pre-crack sites can act as fast diusion channels for the solute atoms and therefore should help to prevent crack initiation. If the strain rate is suciently low and the initiation/growth of microcracks is suciently suppressed, the nearly-perfectly plastic behavior as reported in tension of nanostructured Cu [3,4,26] is expected to appear in tension for the present cryomilled 5083 Al alloy. As discussed above, the proposed diusion-mediated stress relaxation mechanism is dierent from the reported DSA mechanism. The diusion-mediated stress relaxation is caused by the interaction of fast diusing atoms with stationary stress concentration sites, while DSA is caused by the interaction of fast diusion atoms with moving dislocations. This dierence explains why the phenomena of DSA and improved ductility at lower strain rate are not always observed simultaneously. The diusion-mediated stress relaxation mechanism has a similarity with the DSA mechanism in that both are caused by the fast diusion of solute atoms, which explains why these two phenomena often occur together, as in the present study. The proposed diusion-mediated stress relaxation mechanism can also explain the higher ductility at lower strain rate reported in ball-milled Cu [4]. In ball-milled Cu [4], the same trend of higher ductility at lower strain rate is observed (elongation increases from 5.1% to 11.1% when strain rate decreases from 102 s1 to 104 s1). However, its ultimate tensile strength was lower when tested at lower strain rate, contradicting the reported trend for cryomilled Al. Ball milling usually introduces large amount of defects that can act as potential crack initiation sites. At the same time, it also introduces some fast diusing impurities such as H, N, O, and C, which, together with a high density of dislocations, can help relieve stress concentration and thereby prevent crack initiation via the proposed mechanism. It should be noted, however, that although the proposed mechanism can explain the improved ductility at lower strain rate observed in the current investigation and reported in the literature, more experimental studies are needed to further verify its validity. 5. Conclusions Nanostructured 5083 Al alloy produced by cryomilling and hot extrusion is found to have higher ductility at lower

1180

B.Q. Han et al. / Scripta Materialia 54 (2006) 11751180 [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] Lu L, Li SX, Lu K. Scripta Mater 2001;45:1163. Wang YM, Ma E, Valiev RZ, Zhu YT. Adv Mater 2004;16:328. Wang YM, Ma E. Appl Phys Lett 2003;83:3165. Zhu YT, Langdon TG. JOM 2004:58. Zhu YT, Liao X. Nature Mater 2004;31:351. Tellkamp VL, Melmed A, Lavernia EJ. Metall Mater Trans A 2001;32A:2335. Hayes RW, Rodriguez R, Lavernia J. Acta Mater 2001;49:4055. Han BQ, Lee Z, Nutt SR, Lavernia EJ, Mohamed FA. Metall Mater Trans A 2003;34A:603. Han BQ, Lee Z, Witkin D, Nutt SR, Lavernia EJ. Metall Mater Trans A 2005;36A:957. Blum W, Zhu Q, Merkel R, McQueen HJ. Mater Sci Eng 1996; A205:23. Lucadamo G, Yang NYC, SanMarchi C, Lavernia EJ. In: Proc of MRS Symp, vol. 882E, 2005: EE5.2.1. Zhou F, Lee J, Dallek S, Lavernia EJ. J Mater Res 2001;16:3451. Huang JY, Zhu YT, Jiang HG, Lowe TC. Acta Mater 2001;49:1497. Frost HJ, Ashby MF. Deformation-Mechanism Maps: The Plasticity and Creep of Metals and Ceramics. Oxford: Pergamon Press; 1982. Jia D, Wang YM, Ramesh KT, Ma E, Zhu YT, Valiev RZ. Appl Phys Lett 2001;79:611. Sanders PG, Eastman JA, Weertman JR. Acta Mater 1997;45:4019. Sanders PG, Fougere GE, Thompson LJ, Eastman JA, Weertman JR. Nanostruct Mater 1997;8:243. Balik J, Lukac P. Acta Metall Mater 1993;41:1447. Wagenhofer M, Erickson-Natishan M, Armstrong RW, Zerilli FJ. Scripta Mater 1999;41:1177. Clausen AH, Borvik T, Hopperstad OS, Benallal A. Mater Sci Eng A 2004;364:260. Han BQ, Lavernia EJ, Mohamed FA. Metall Mater Trans A 2005;36A:345. Champion Y, Langlois C, Guerin-Mailly S, Langlois P, Bonnentien JL, Hytch MJ. Science 2003;300:310.

strain rates. This phenomenon cannot be explained by either dynamic recovery or dynamic strain aging. The experimental observations made in the present study and reported in the literature suggest that a new mechanism, diusion-mediated stress relaxation, is responsible for the observed higher ductility at lower strain rates. In this new mechanism, fast diusing atoms relieve stressconcentration so that the crack initiations are suppressed. This mechanism predicts that material systems which contain fast diusing solutes and whose failure is controlled by defectinitiated cracks are likely to exhibit higher ductility at lower strain rates. The mechanism can explain properly currently available experimental observations and is to be further validated with additional investigations. Acknowledgements The support from the Oce of Naval Research under Grant No. N00014-03-1-0149 with Dr. Lawrence Kabaco as program ocer is gratefully acknowledged. References
[1] Milligan WW. In: Milne I, Ritchie RO, Karihaloo B, editors. Comprehensive structural integrity, vol. 8. Oxford: Elsevier; 2003. p. 529. [2] Torre FHD, Pereloma EV, Davies CHJ. Scripta Mater 2004;51:367. [3] Wang YM, Ma E. Acta Mater 2004;52:1699. [4] Cheng S, Ma E, Wang YM, Kecskes LJ, Youssef KM, Koch CC, et al. Acta Mater 2005;53:1521.