Achieving High Strength and High Ductility in Friction Stir-Processed Cast Magnesium Alloy WEI YUAN, SUSHANTA K. PANIGRAHI, and RAJIV S. MISHRA Friction stir processing (FSP) is emerging as an effective tool for microstructural modification and property enhancement. As-cast AZ91 magnesium alloy was friction stir processed with one-pass and two-pass to examine the influence of processing conditions on microstructural evolution and corresponding mechanical properties. Grain refinement accompanied with development of strong basal texture was observed for both processing conditions. Ultrafinegrained (UFG) AZ91 was achieved under two-pass FSP with fine precipitates distributed on the grain boundary. The processed UFG AZ91 exhibited a high tensile strength of ~435 MPa (117 pct improvement) and tensile fracture elongation of ~23 pct. The promising combination of strength and ductility is attributed to the elimination of casting porosity, and high density of fine precipitates in an UFG structure with quite low dislocation density. The effects of grain size, precipitate, and texture on deformation behavior have been discussed. DOI: 10.1007/s11661-013-1744-5  The Minerals, Metals & Materials Society and ASM International 2013

I.

INTRODUCTION

MAGNESIUM alloys are attracting great interest from transportation industries for structural applications because of their high strength-to-weight ratio.[1,2] However, low ductility and room temperature formability limit the processing and application.[3] As a result, the main fabrication route of magnesium alloys remains casting.[1,2] Casting defects such as microscopic shrinkage porosity and inclusions adversely affect the strength and ductility of magnesium alloy. It is expected that the application will be expanded if the strength and ductility of magnesium alloys can be enhanced. Grain refinement has been a good option for enhancing the strength of magnesium alloys. Several severe plastic deformation (SPD) tools, such as equal channel angular extrusion/pressing,[4,5] high-pressure torsion,[6,7] accumulated roll bonding,[8] alternate biaxial reverse corrugated pressing,[9] differential speed rolling,[10,11] and friction stir processing (FSP)[12–14] have demonstrated capability in refining grain structure of magnesium alloys. However, grain refinement-induced strength enhancement is generally accompanied by the

WEI YUAN, Researcher, is with the Department of Materials Science & Engineering, Missouri University of Science and Technology, Rolla, MO 65409, and also with the Research & Development Division, Hitachi America Ltd., 34500 Grand River Avenue, Farmington Hills, MI 48335. SUSHANTA K. PANIGRAHI, Assistant Professor, is with the Department of Materials Science & Engineering, Missouri University of Science and Technology, and also with the Indian Institute of Technology Madras, Chennai 600036, India. RAJIV S. MISHRA, Professor, is with the Department of Materials Science & Engineering, Missouri University of Science and Technology, and also with the Center for Friction Stir Processing, Department of Materials Science and Engineering, University of North Texas, Denton, TX 76203. Contact e-mail: [email protected] Manuscript submitted September 9, 2012. METALLURGICAL AND MATERIALS TRANSACTIONS A

reduction in ductility, especially when the grain size is refined to an ultrafine regime, mainly because of the reduced work hardening.[15–17] Texture, which is always observed in magnesium alloys during intense shear deformation, plays an important role on room-temperature deformation behavior in addition to grain size as a result of limited independent slip systems available. Texture modification of wrought AZ magnesium alloys by SPD has shown promising room-temperature ductility because of the formation of a strong basal texture, which orientated the easy basal slip plane to the preferred orientation; however, the tensile yield strength was very low.[18–25] The special basal texture generation during SPD also led to the anisotropy in mechanical properties, especially in yield strength, which is directly related to the activation of deformation slip or twinning systems.[14,20,26] It is desirable to enhance strength as well as ductility in magnesium alloys. Recently, the FSP of cast alloys has shown effective elimination of cast porosity, extensive grain refinement, dissolution, and breakup of coarse second phase.[12,27–32] Precipitation has been shown to be an effective additive strengthening mechanism for magnesium alloys in addition to the grain refinementrelated strengthening. Feng and Ma[28] reported a high tensile strength of 337 MPa in AZ91 magnesium alloy via FSP with average grain size refined to ~15 lm and fine precipitates decorating the grain boundaries. By introducing nanosized precipitates in a fine-grained rare earth magnesium alloy EV31A through FSP, Freeney and Mishra[30] obtained significant enhancement in tensile strength and ductility. Xiao et al.[32] presented similar findings on a FSP fine-grained Mg-Gd-Y-Zr alloy with even higher tensile strength of 439 MPa, but a noticeable loss of elongation to 3.4 pct. Our previous study has achieved highly textured ultrafine-grained (UFG) AZ31 using FSP, which exhibited high tensile strength and ductility because of restricted basal slip.[26]

The current study aims to (a) achieve high strength and high ductility in commercial grade cast magnesium alloy through microstructural modification, (b) verify whether strong texture and strength anisotropy exist in FSP of high aluminum content precipitation strengthened magnesium alloy, and (c) explore the influence of grain size, precipitates, and texture on deformation behavior of processed magnesium alloy.

II.

EXPERIMENTAL PROCEDURE

A high-pressure die-casting AZ91 magnesium alloy having a thickness of 4.2 mm was used. The nominal composition of this alloy was 9 wt pct aluminum and 1 wt pct zinc. To make friction stir passes, a tool with 12-mm diameter concave shoulder and 2.1-mm-stepped spiral pin was employed. Two types of FSP runs were made on the as-cast AZ91 with one-pass and two-pass runs. A tool rotation rate of 600 revolutions per minute (rpm) and a tool traverse speed of 1.7 mm/s were applied to the one-pass run. Two-pass FSP was carried out with 300 rpm and 1.7 mm/s overlapping the first pass. A tool tilt of 2.5 deg was adopted for all the FSP passes. Microstructures of the friction stir-processed samples, with cross section being perpendicular to the processing direction (PD), were examined using optical microscopy (OM), scanning electron microscope (SEM), and electron back-scattered diffraction (EBSD). For OM and SEM observations, samples were mechanically polished to 1 lm with alcohol-based diamond polishing compounds and etched using an acetic picral solution (4.5 g picric acid, 10 ml acetic acid, 10 ml water, and 70 ml ethanol). The specimens for EBSD characterization were further fine polished with 0.02 lm colloidal silica solution. The EBSD characterization was performed at the center of the processed AZ91 in a cross section perpendicular to the PD. The linear intercept method was used to obtain an average grain size. Transmission electron microscope (TEM) samples were prepared by focused ion milling in Helio Nano Lab 600 FIB/ FESEM. A TECNAI F20 TEM was used to examine

grain structure and morphology and distribution of precipitates under an operating voltage of 200 kV. Tensile property characterization in the current study was conducted using a computer-controlled, screw-driven tensile testing machine at a constant crosshead velocity. Tensile specimens have a gage length of 1.3 mm, a width of 1.0 mm (to eliminate the effect of texture variation) and a thickness of 0.6 mm. Specimens were machined along the PD and transverse direction (TD) from the center of the processed region and were polished with 1-lm finish and tested at room temperature at an initial strain rate of 1 9 103 s1. Aging treatment at 441 K (168 C) for 16 h was applied to two-pass AZ91 to characterize the effect of post aging on tensile property. At least three specimens for each condition were tested to evaluate the average property values.

III.

RESULTS

Figure 1 shows optical and SEM micrographs of AZ91 alloy in the as-cast condition. The low magnification of optical micrograph exhibits typical casting features with dendritic pattern of a-Mg matrix in white and second phase in gray.[29] The casting porosities in black were visible in certain locations with an area fraction of ~4.2 pct after averaging six randomly selected locations with same magnification using ImageJ software. The as-cast AZ91 varies in grain size with small grains of about 10 lm and large dendrites of about 120 lm. The secondary electron imaging at low magnification shows high density of coarse secondphase network distributed at grain boundaries. This second phase was identified as b-Mg17Al12 phase through energy dispersive X-ray (EDX) analysis. One such b-Mg17Al12 phase is shown in Figure 2 with a size of about 10 lm. The fraction of b-Mg17Al12 phase in ascast AZ91 was ~18.6 pct. The friction stir-processed AZ91 was sectioned in the TD for evaluation of the microstructure evolution. Figure 3 presents the typical cross sections of friction stir-processed AZ91 in one-pass and two-pass conditions.

Fig. 1—Optical micrograph and SEM image of as-cast AZ91. METALLURGICAL AND MATERIALS TRANSACTIONS A

In the current study, different material flow behavior can be observed for material processed with one-pass and two-pass. The material processed with two-pass shows a uniform macrostructure. However, a nonuniform structure can be observed for one-pass run, which separates the processed region into top and bottom parts. Figure 4 presents the microstructures of one-pass friction stir-processed AZ91 at top and bottom regions. Figure 4(a) is a low-magnification view of the top region and shows layer structure with alternate white and black contrast. A high-magnification examination indicates b-phase particle-rich layers, with some horizontally elongated, incompletely dissolved coarse b-phase being distributed alternately with a-Mg matrix. The cast dendritic structure was completely eliminated and substituted by much finer equiaxed grain structure. Furthermore, examination of the processed region showed complete elimination of cast porosity. A comparison between the top and bottom regions (Figures 4(b) and (c)) indicates similar grain size with an average grain size of about 2.8 ± 1.2 lm but distinct precipitate distribution. As shown in Figure 4(d), the initial coarse network type of Mg17Al12 phase were dissolved or broken into much smaller and almost equiaxed ones at the bottom region with majority of them distributed at the grain boundaries. Grain structure and precipitate morphology and distribution of two-pass friction stir-processed AZ91 were examined using TEM. Figures 5(a) and (b) show a bright-field image and scanning transmission electron

Fig. 2—Presence of b-Mg17Al12 phase and a-magnesium matrix in the as-cast AZ91.

microscope (STEM) image at the same location. The two-pass FSP caused a significant grain refinement compared with the one-pass run with much finer and more equiaxed grain structure. As shown in Figure 6(a), the average grain size of two-pass friction stir-processed AZ91 was 0.54 ± 0.20 lm computed based on STEM images at six different locations. The grain boundaries under bright field imaging are not distinct, which might be related to the low misorienation angle nature of boundaries. The round-edged b-Mg17Al12 phase particles were mainly observed on the grain boundaries and grain triple junctions. Few round shaped finer precipitates were noticed in the grain interiors. The relatively large precipitates have ellipse shape; however, the fine precipitates retain a round shape. The size of precipitates was computed in terms of equivalent circular diameter (square root of the product of lengths in major and minor axes of the ellipse) assuming ellipse shape of precipitates. The size distribution of precipitates in twopass FSP is shown in Figure 6(b) with a phase fraction of 12.8 pct. The average size of precipitates was 113 ± 98 nm. Few dislocations can be detected in the processed region. Table I lists the EDX analysis of contents at various locations, the grain interior contained ~3.8 wt pct Al, and Zn was mainly concentrated on precipitates or adjacent locations. Figure 7 shows the {0002} and {10-10} pole figures of friction stir-processed AZ91 in as-cast and friction stirprocessed conditions. The location for friction stirprocessed specimens was at the centerline of the processed region at a depth of 1.0 mm. Pole figures of {0002} and {10-10} of as-cast AZ91 indicate a relatively random texture with maximum intensity of about 3. This random texture of cast AZ91 was modified to a strong basal fiber texture with alignment of the c-axes of the grains tilting about 25 to 30 deg away from the ND after FSP. The difference in degree of tilt between onepass and two-pass was negligible. The observation of texture evolution in AZ91 is similar to that has been reported in FSP of rolled AZ31 with a completely different starting textures.[26] Figure 8 presents the representative tensile engineering stress–strain curves of AZ91 in as-cast and friction stir-processed conditions tested at room temperature with an initial strain rate of 1 9 103 s1. At least three (six for as-cast AZ91) specimens for each condition were tested. The statistical value of mechanical properties, including tensile yield strength (YS), ultimate tensile strength (UTS), uniform elongation and total elongation to failure, are summarized in Table II. The asreceived AZ91 shows low tensile strength and poor

Fig. 3—Cross-sectional macrographs of friction stir-processed AZ91 with (a) single pass at 600 rpm/1.7 mm/s and (b) two-pass with 600 rpm/ 1.7 mm/s for the first pass and 300 rpm/1.7 mm/s for the second pass. METALLURGICAL AND MATERIALS TRANSACTIONS A

room-temperature ductility of less than 3 pct total elongation to failure for average. Majority of the as-cast AZ91 specimens failed abruptly. For friction stir-

processed AZ91 tested in TD, after one-pass, the tensile YS and total elongation increased from ~160 MPa to 302 MPa and ~3 pct to 19 pct, respectively. The tensile

Fig. 4—Micrographs show the grain structure and precipitate in one-pass friction stir-processed AZ91, (a, b) top region, and (c, d) bottom region.

Fig. 5—Grain structure and precipitates of two-pass friction stir-processed AZ91 in (a) TEM bright field image and (b) STEM image, location ‘A’ was marked in both images. METALLURGICAL AND MATERIALS TRANSACTIONS A

Fig. 6—Frequency distribution plots of (a) grain size and (b) precipitate size in two-pass friction stir-processed AZ91.

Table I. Location A B C D

Results of EDX Analysis of Fig. 5 (Weight Percent) Al

Zn

Mg

3.8 39.6 3.6 19.4

— 3.5 0.2 3.3

96.2 56.9 96.2 77.3

Fig. 8—Stress–strain curves of as-received and friction stir-processed AZ91.

Fig. 7—{0002} and {10-10} pole figures of AZ91 in (a) as-cast, (b) one-pass FSP and (c) two-pass FSP. METALLURGICAL AND MATERIALS TRANSACTIONS A

strength was further enhanced after two-pass FSP, with average tensile YS of ~390 MPa and UTS of ~435 MPa. The total elongation to failure for two-pass friction stirprocessed AZ91 was also improved to ~23 pct. Comparison of one-pass specimen with two-pass specimen shows a decrease in uniform elongation from 13.6 pct to 8.8 pct, but an increase in total elongation as grain size becomes finer. This observation is consistent with previous findings on friction stir-processed AZ31.[23] A significant anisotropy was also observed for friction stir-processed AZ91 in PD and TD. For processed AZ91 tested in PD, all stress–strain curves exhibit a common feature similar to that of the as-cast AZ91 specimens that fractured without developing a macroscopic necking, i.e., there is no pronounced post-uniform elongation. However, unlike as-cast AZ91, the tensile strength and total elongation were improved significantly after FSP with more than 10 pct total elongation. Contrary to the previous results on fine-grained and UFG AZ31,[23] the ductility in PD for AZ91 increased as the grain grew

Table II. Condition

YS (MPa)

As-Cast One-pass_TD One-pass_PD Two-pass_TD Two-pass_PD Two-pass+ aging_TD Two-pass + aging_PD

162 302 203 390 258 381 234

± ± ± ± ± ± ±

A Summary of Tensile Properties in Different Conditions

12 3 10 5 6 4 8

UTS (MPa) 201 379 285 435 351 422 329

± ± ± ± ± ± ±

15 4 8 1 6 7 6

finer. The discrepancy is likely to be due to the better microstructural homogenization in two-pass AZ91 compared with the one-pass. The observation that TD has higher tensile strength as well as higher ductility than those of PD for the current fine-grained and UFG AZ91 is similar to the findings of friction stir-processed AZ31.[23] Aging of two-pass FSP did not enhance the strength in either PD or TD, on the contrary, reduced the ductility, especially in PD.

IV.

DISCUSSION

A. Microstructural and Texture Evolution The as-cast structure was modified by FSP with the complete elimination of coarse dendritic grains and casting porosity. Accompanying the grain refinement, the coarse network type of precipitates was dissolved and refined into much finer scale. However, distinct difference in processed region was observed between one-pass and two-pass FSP of as-cast AZ91, with heterogeneous layer structure for one-pass and more uniform structure for two-pass. Similar observation of b-phase particles rich bands has been reported during the FSPs of cast AZ91 and AZ80 magnesium alloys, which was attributed to the lower strain rate/strain and temperature conditions.[27,31] A high tool rotation rate was suggested to dissolve b-phase and in turn achieve uniform microstructure.[27] Processing parameters have shown direct influence on the thickness of heterogeneous layer during the FSP of cast AZ91. Previous observation indicated that the top layer was thinner as the tool rotation rate increased for a constant tool traverse speed (images not shown). Another factor causing the difference in the macrostructures between one-pass and two-pass FSPs was the characteristic of starting microstructure for processing. During the FSP, material in front of the tool is driven upward. The upward-moving material on the leading side then sweeps around the pin in the rotation direction;[33–35] however, on the trailing side, the layer rotating with the tool is decelerated and is deposited in the void in the wake of the pin with the assistance of tool shoulder.[34] The heterogeneous layer structure is likely related to the sweeping action of the shoulder as it comes into contact with the as-cast material first and transports it toward the trailing edge around the rotating pin. Because of the significant difference

Uniform Elongation (pct) 1.9 13.6 10.5 8.8 15.4 7.5 8.1

± ± ± ± ± ± ±

1.4 0.8 1.3 0.3 0.4 0.4 0.5

Total Elongation (pct) 2.7 19.1 10.5 23.4 15.6 21.2 8.1

± ± ± ± ± ± ±

1.1 1.4 1.3 0.2 0.5 0.2 0.5

between the as-cast and the recrystallized microstructures, the transported cast material from the front of the pin tool might not undergo full dissolution of precipitates and exhibited the difference between top and bottom regions. For the two-pass FSP, the starting microstructure for the second pass is recrystallized grain structure with remaining b-phase particles, which eliminated the variation in structure in the processed region of the two-pass run. Pre-solution treatment of cast magnesium alloy to homogenize the microstructure has been proven to result in uniform microstructure in single pass FSP.[29,31] As shown in Figures 4 and 5, FSP eliminated cast porosity, network type of coarse b-phase, and significantly refined the grain structure as a result of dynamic recrystallization.[12,27,36–39] Average grain sizes of ~2.8 and 0.5 lm were achieved in one-pass and twopass runs, respectively. UFG structure has been reported by FSP of various metals and alloys, which generally requires low frictional heat input, high strain (rate) and extra cooling to recrystallize the grain structure and suppress grain growth during postFSP.[13,40,41] The very low dislocation density in two-pass friction stir-processed AZ91 might indicate discontinuous dynamic recrystallization which sweeps off dislocation during grain boundary migration, resulting in very low dislocation density in the recrystallized grains. The current UFG structure via two-pass FSP was mainly attributed to low heat input and homogeneous distribution of fine b-phase precipitates, which pin grain boundary migration and inhibit grain growth. The as-cast AZ91 had a random texture as expected (Figure 7(a)). However, a strong basal texture developed during one-pass and two-pass FSPs. Similar to the previous results in FSP of AZ31, a basal fiber texture for both one-pass and two-pass runs shows a tilt of c-axis toward the PD by about 25 through 30 deg at the centerline of the processed region.[23,42] The formation of basal texture has been well documented as a result of shear deformation during FSP, which aligns the easy basal slip planes roughly parallel to pin column surface.[24–26,43] It has been suggested that rare earth elements can weaken the deformation texture[44,45] through particle-stimulated nucleation of recrystallization.[46] However, from the current results, it can be concluded that the b-phase particles have limited effect on texture weakening with the evidence of much similar strong basal texture between AZ91 and AZ31 both obtained with FSP. METALLURGICAL AND MATERIALS TRANSACTIONS A

B. Strengthening in Friction Stir-Processed AZ91 Magnesium Alloy The strength of high aluminum content magnesium alloy is influenced by the effect of alloy content on solid solution strengthening, the size and volume fraction of b-phase particles, as well as the grain size and existing microstructural defects. The as-cast AZ91 exhibited quite low tensile strength, poor elongation and large variation in these properties. These are attributed to the presence of cast porosities and network type coarse Mg17Al12 phase at the grain boundaries that tend to nucleate cracks or undergo debonding from the matrix during deformation.[28] FSP resulted in a significant improvement in both tensile strength and ductility. This is mainly because of porosity elimination, grain refinement, and significant breakup and dissolution of the coarse Mg17Al12 phase.[28,30,31] Different from those reported by Feng et al.[31] on AZ80 that one-pass FSP led to even less ductility of ~3 pct (in PD) than that of as-cast, the current one-pass FSP enhanced not only the tensile strengths but also the ductility, although heterogeneous microstructure was observed in both cases. The discrepancy results from different thermal cycles which resulted in different grain sizes and fractions of remaining undissolved precipitates. The finer grain size and lower amount of remaining coarse precipitates observed in the one-pass specimens of the current study contributed to the better tensile properties. The two-pass FSP in PD further increased the tensile strength and ductility compared with the one-pass FSP (as shown in Figure 8 and Table II) because of the further refinement of grains, precipitates, and elimination of heterogeneous microstructure as evidenced in Figures 4 and 5. The TD testing exhibited much higher tensile strength than that in PD for both one-pass and two-pass FSPs, with extremely high tensile YS of ~390 MPa and UTS of ~435 MPa in two-pass FSP. The strength anisotropy was attributed to the strong basal texture generated during the FSP, which tilted the c-axis toward PD, and in turn promoted easy basal slip in PD and difficult nonbasal slip in TD.[26] High-strength AZ91 has also been reported during high ratio differential speed rolling, which appears difficult to interpret from strengthening mechanisms point of view because of the extremely fine grains of 280 nm and distribution of fine b-phase particles (88 nm) on the grain boundaries.[47] For the currently obtained high-strength AZ91, EDX analysis (Table I) of grain interior of two-pass AZ91 showed low aluminum supersaturating with a content of ~3.8 wt pct, which was slightly higher than the content in AZ31 magnesium alloy. Considering the much similar c-axis tilts in friction stir-processed AZ91 and AZ31, and the fact that the variation in basal c-axis tilt was about 5 to 10 deg, it might be rational to compute the tensile YS of the current friction stir-processed AZ91 by adopting the empirical grain boundary strengthening formula previously developed for friction stir-processed AZ31 with additional strengthening from b-phase particles,[42] since the dislocation density was low for both friction stir-processed AZ31 and AZ91 during dynamic METALLURGICAL AND MATERIALS TRANSACTIONS A

recrystallization. The Hall–Petch parameters (friction stress r0 and stress concentration factor ky) for grainboundary strengthening of friction stir-processed AZ31 with grain size between 0.7 and 10.2 lm is presented in Table III. The theoretically computed tensile YS values without considering particle strengthening for two-pass friction stir-processed UFG AZ91 were 163 MPa and 367 MPa in PD and TD, respectively. The strengthening contribution from interaction of b-phase particles with basal dislocation through Orowan looping was developed by Nie[48] and can be expressed as:   Gb d  ln p ; ½1 DsOrowan ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi b p ffi  1 dp 2p ð1  mÞ 0:953 f

where G is the shear modulus (17 GPa for magnesium), b is the Burgers vector (3.2 9 1010 m), v is the Poisson’s ratio (0.35),[49] f is the phase fraction (0.128) of b-phase, dp is the mean planar diameter of the point obstacles (1.13 9 107 m). The computed contribution from particle strengthening to resolved shear stress is 34 MPa. A Taylor factor of 2.3 for basal slip dominated deformation in PD was achieved from EBSD data by assuming Sachs polycrystal deformation model (intergranular stress equilibrium). Therefore, the increase in tensile YS due to b-phase particles was ~78 MPa. The overall computed tensile YS in PD was 241 MPa, slightly less than the experimentally determined value of 258 MPa. Assuming similar particle–dislocation interaction applies to nonbasal dislocation dominated deformation in TD, by simply replacing the Taylor factor to 2.2 for nonbasal slip according to EBSD data analysis, an increase in tensile YS due to b-phase particles was found to be ~75 MPa. This overestimates the tensile YS to 442 MPa. These results indirectly indicate that b-phase particles in two-pass friction stirprocessed UFG AZ91 have more effect on inhibiting basal dislocation slip than nonbasal dislocation slip, at least in the current study with this distribution of precipitates. The slight variation in tensile YS and significant reduction in ductility in PD, but hardly any change in TD after aging might support this hypothesis. The negligible effect of aging on property in current UFG AZ91 compared with the significant strengthening reported in fine-grained AZ91 by Feng et al.[28] was due to the difference in Al contents in matrix, about 3.8 pct for current study and 7.5 pct in Reference 28. It may be speculated that grain boundary strengthening, solid solution strengthening (initially embedded in Hall–Petch formula for friction stir-processed AZ31) and particle strengthening contributed to the overall strength at Table III. Loading PD TD

Hall–Petch Parameters Developed for Friction Stir-Processed AZ31[42] r0 (MPa)

ky (MPa lm1/2)

39 145

88 157

yielding in PD, and mainly grain boundary strengthening and solid solution strengthening in TD. C. Effects of Grain Size and Precipitate on Work Hardening and Ductility The ductilities were significantly improved after FSPs in both PD and TD compared with the as-cast condition as shown in Figure 8 and Table II. The lower ductility in the as-cast AZ91 was due to the presence of cast porosity and network type of coarse b-phase, which could be the stress concentration sites during loading: as a consequence, crack nucleation, void coalescence, or debonding between a-Mg matrix and b-phase led to premature fracture. FSP eliminated casting defects and coarse b-phase, and thus enhanced the tensile ductility. The increase in ductility from one-pass to two-pass was related to the particle refinement and microstructure homogenization. Grain refinement to ultrafine scale generally leads to reduction in ductility as a result of reduced work hardening in finer grain size. It is also reported that tensile ductility could be enhanced in the UFG materials with lower dislocation density, which can provide more room for dislocation storage.[50] The current two-pass UFG AZ91 shows extremely high tensile strength as well as high ductility, with a relatively low uniform elongation but high post-uniform elongation in TD. The high post-uniform elongation was likely related to the enhanced strain rate sensitivity as grain size becomes finer.[51,52] Although the uniform elongation in two-pass friction stir-processed UFG AZ91 in TD was relatively low compared with the PD or one-pass FSP, it is much higher than that generally reported in UFG magnesium alloy,[15,23] especially with such a high tensile strength. This observation suggests that plastic instability was delayed in current UFG AZ91 through either enhanced work hardening ability or reduced rate of dislocation recovery. Figure 9 plots the work hardening rate h (=dr/de) derived from true stress verses true strain as a function of true stress/net flow stress for friction

stir-processed AZ91 in both PD and TD, as well as friction stir-processed UFG AZ31(having ~2 pct uniform elongation in TD and ~10 pct uniform elongation in PD).[23] After a short period of elastoplastic transition, the work hardening rate decreases linearly and steadily for all testing conditions because of the balance between dislocation storage and annihilation, which is the stage III hardening range.[53] Stage III can be expressed using Voce equation, which is given as dr @Gbk1 k2 r ¼  de 2 2

½2

where k1 is a constant and k2 is twice of the slope of the strain hardening rate versus true stress plot and is an indication of the rate of recovery. The first term on the right-hand side is associated with the dislocation storage discussed above and the second term is associated with annihilation of dislocations. The stage III is different for one-pass and two-pass FSP in PD and TD, evidenced from the various slopes of stage III. The two-pass FSP in TD shows the largest slope, followed by one-pass in PD and one-pass in TD. The two-pass FSP in PD exhibits lowest slope, i.e., the lowest rate of dislocation recovery. As shown in Figure 9(b), the degree of work hardening was smallest for two-pass in TD, and largest for one-pass in TD and two-pass in PD, with one-pass in PD in between. The largest degree of work hardening and the lowest rate of dislocation recovery in PD for two-pass AZ91 contribute to the highest uniform elongation. On the contrary, the lowest in TD for twopass AZ91. Comparing the UFG AZ91 with UFG AZ31, we find that both have submicron grain sizes (0.5 lm for AZ91 and 0.8 lm for AZ31), similar inclined c-axis tilts of about 25 through 30 deg, and low dislocation densities, but completely different work hardening behaviors. The initial work hardening rate was much higher for UFG AZ31, but it quickly dropped below that of UFG AZ91 in a short net flow stress region. The recovery rate determined by linear fit in stage III was much higher for

Fig. 9—Work hardening rate as a function of true stress and net flow stress for friction stir-processed AZ91 in PD and TD, work hardening behavior of UFG AZ31 was also included.[23] METALLURGICAL AND MATERIALS TRANSACTIONS A

UFG AZ31 than that for UFG AZ91 in both PD and TD. It is quite possible that the fine b-phase precipitates enhanced effective dislocation storage, which retarded the drop of work hardening rate and resulted in the improved uniform elongation in AZ91, although the grain size was smaller than that of AZ31.

V.

CONCLUSIONS

The as-cast AZ91 magnesium alloy was friction stir processed with one-pass and two-pass runs. FSP eliminated porosity and significantly refined the average grain sizes to 2.8 and 0.5 lm for one-pass and two-pass runs, respectively, but also introduced strong basal fiber texture. The network type of coarse precipitates was dissolved and broken into fine particles mainly distributed at the grain boundaries. As the grain size reduced during second FSP pass, the precipitates also became finer. FSP remarkably enhanced the tensile property of the as-cast AZ91. Two-pass FSP improved the tensile YS to ~390 MPa, UTS to ~435 MPa (117 pct improvement) in TD; and YS to ~258 MPa, and UTS to ~340 MPa (62 pct improvement) in PD. The tensile failure elongations were also enhanced from 3.5 pct to 23 pct and to 16 pct in TD and PD, respectively. The promising combination of strength and ductility was attributed to the elimination of porosity in UFG structure with low dislocation density and fine precipitates. Precipitates appear to influence basal dislocations more than nonbasal dislocations in current UFG AZ91.

ACKNOWLEDGMENTS The authors gratefully acknowledge the support provided by the National Science Foundation through grants NSF-EEC-0531019 and NSF-IIP-1157754.

REFERENCES 1. B.L. Mordike and T. Ebert: Mater. Sci. Eng. A, 2001, vol. 302, pp. 37–45. 2. B.B. Clow: Adv. Mater. Process., 1996, vol. 150, p. 33. 3. S.R. Agnew, J.W. Senn, and J.A. Horton: JOM, 2006, vol. 58, pp. 62–69. 4. M. Mabuchi, H. Iwasaki, K. Yanase, and K. Higashi: Scripta Mater., 1997, vol. 36, pp. 681–86. 5. A. Yamashita, Z. Horita, and T.G. Langdon: Mater. Sci. Eng. A, 2001, vol. 300, pp. 142–47. 6. A. Galiyev and R. Kaibyshev: Mater. Trans., 2001, vol. 42, pp. 1190–99. 7. M. Kai, Z. Horita, and T.G. Langdon: Mater. Sci. Eng. A, 2008, vol. 488, pp. 117–24. 8. M.T. Pe´rez-Prado, J.A. Del Valle, and O.A. Ruano: Scripta Mater., 2004, vol. 51, pp. 1093–97. 9. Q. Yang and A.K. Ghosh: Acta Mater., 2006, vol. 54, pp. 5147–58. 10. H. Watanabe, T. Mukai, and K. Ishikawa: J. Mater. Sci., 2004, vol. 39, pp. 1477–80. 11. W.J. Kim, J.D. Park, J.Y. Wang, and W.S. Yoon: Scripta Mater., 2007, vol. 57, pp. 755–58.

METALLURGICAL AND MATERIALS TRANSACTIONS A

12. R.S. Mishra and Z.Y. Ma: Mater. Sci. Eng. R, 2005, vol. 50, pp. 1–78. 13. C.I. Chang, X.H. Du, and J.C. Huang: Scripta Mater., 2007, vol. 57, pp. 209–212. 14. G. Bhargava, W. Yuan, S.S. Webb, and R.S. Mishra: Metall. Mater. Trans. A, 2010, vol. 41A, pp. 13–17. 15. S.X. Ding, W.T. Lee, C.P. Chang, L.W. Chang, and P.W. Kao: Scripta Mater., 2008, vol. 59, pp. 1006–09. 16. W.J. Kim and Y.G. Lee: Mater. Sci. Eng. A, 2011, vol. 528, pp. 2062–66. 17. W.J. Kim, Y.G. Lee, M.J. Lee, J.Y. Wang, and Y.B. Park: Scripta Mater., 2011, vol. 65, pp. 1105–08. 18. T. Mukai, M. Yamanoi, H. Watanabe, and K. Higashi: Scripta Mater., 2001, vol. 45, pp. 89–94. 19. J. Koike, T. Kobayashi, T. Mukai, H. Watanabe, M. Suzuki, K. Maruyama, and K. Higashi: Acta Mater., 2003, vol. 51, pp. 2055–65. 20. S.R. Agnew, J.A. Horton, T.M. Lillo, and D.W. Brown: Scripta Mater., 2004, vol. 50, pp. 377–81. 21. W.J. Kim, C.W. An, Y.S. Kim, and S.I. Hong: Scripta Mater., 2002, vol. 47, pp. 39–44. 22. H.W. Lee, T.S. Lui, and L.H. Chen: J. Alloys Compd., 2009, vol. 475, pp. 139–44. 23. W. Yuan and R.S. Mishra: Mater. Sci. Eng. A, 2012, vol. 558, pp. 716–24. 24. Y.N. Wang, C.I. Chang, C.J. Lee, H.K. Lin, and J.C. Huang: Scripta Mater., 2006, vol. 55, pp. 637–40. 25. W. Woo, H. Choo, D.W. Brown, P.K. Liaw, and Z. Feng: Scripta Mater., 2006, vol. 54, pp. 1859–64. 26. W. Yuan, R.S. Mishra, B. Carlson, R.K. Mishra, R. Verma, and R. Kubic: Scripta Mater., 2011, vol. 64, pp. 580–83. 27. Z.Y. Ma, A.L. Pilchak, M.C. Juhas, and J.C. Williams: Scripta Mater., 2008, vol. 58, pp. 361–66. 28. A.H. Feng and Z.Y. Ma: Scripta Mater., 2007, vol. 56, pp. 397– 400. 29. P. Cavaliere and P.P. De Marco: J. Mater. Process. Technol., 2007, vol. 184, pp. 77–83. 30. T.A. Freeney and R.S. Mishra: Metall. Mater. Trans. A, 2009, vol. 41A, pp. 73–84. 31. A.H. Feng, B.L. Xiao, Z.Y. Ma, and R.S. Chen: Metall. Mater. Trans. A, 2009, vol. 40A, pp. 2447–56. 32. B.L. Xiao, Q. Yang, J. Yang, W.G. Wang, G.M. Xie, and Z.Y. Ma: J. Alloys Compd., 2011, vol. 509, pp. 2879–84. 33. H.N.B. Schmidt, T.L. Dickerson, and J.H. Hattel: Acta Mater., 2006, vol. 54, pp. 1199–1209. 34. K. Colligan: Welding J., 1999, vol. 78, pp. 229–37s. 35. T.U. Seidel and A.P. Reynolds: Metall. Mater. Trans. A, 2001, vol. 32A, pp. 2879–84. 36. J.A. Esparza, W.C. Davis, E.A. Trillo, and L.E. Murr: J. Mater. Sci. Lett., 2002, vol. 21, pp. 917–20. 37. S.H.C. Park, Y.S. Sato, and H. Kokawa: J. Mater. Sci., 2003, vol. 38, pp. 4379–83. 38. W.B. Lee, Y.M. Yeon, and S.B. Jung: J. Mater. Sci. Technol., 2003, vol. 19, pp. 785–90. 39. U.F.H.R. Suhuddin, S. Mironov, Y.S. Sato, H. Kokawa, and C.W. Lee: Acta Mater., 2009, vol. 57, pp. 5406–18. 40. C.I. Chang, X.H. Du, and J.C. Huang: Scripta Mater., 2008, vol. 59, pp. 356–59. 41. J.Q. Su, T.W. Nelson, and C.J. Sterling: Scripta Mater., 2005, vol. 52, pp. 135–40. 42. W. Yuan, S.K. Panigrahi, J.Q. Su, and R.S. Mishra: Scripta Mater., 2011, vol. 65, pp. 994–97. 43. S.H.C. Park, Y.S. Sato and H. Kokawa: Metall. Mater. Trans. A, 2003, vol. 34A, pp. 987–94. 44. M.R. Barnett, M.D. Nave, and C.J. Bettles: Mater. Sci. Eng. A, 2004, vol. 386, pp. 205–11. 45. J. Bohlen, M.R. Nu¨rnberg, J.W. Senn, D. Letzig, and S.R. Agnew: Acta Mater., 2007, vol. 55, pp. 2101–12. 46. E.A. Ball and P.B. Prangnell: Scripta Metall., 1994, vol. 31, pp. 111–16. 47. W.J. Kim, H.G. Jeong, and H.T. Jeong: Scripta Mater., 2009, vol. 61, pp. 1040–43. 48. J.F. Nie: Scripta Mater., 2003, vol. 48, pp. 1009–15. 49. H.J. Frost and M.F. Ashby: Deformation-Mechanism Maps, Pergamon Press, Oxford, 1982, p. 44.

50. Y.H. Zhao, J.F. Bingert, Y.T. Zhu, X.Z. Liao, R.Z. Valiev, Z. Horita, T.G. Langdon, Y.Z. Zhou, and E.J. Lavernia: Appl. Phys. Lett., 2008, vol. 92, p. 081903. 51. Q. Yang and A.K. Ghosh: Acta Mater., 2006, vol. 54, pp. 5159–70.

52. W.J. Kim, H.W. Lee, S.J. Yoo, and Y.B. Park: Mater. Sci. Eng. A, 2011, vol. 528, pp. 874–79. 53. U.F. Kocks and H. Mecking: Prog. Mater Sci., 2003, vol. 48, pp. 171–273.

METALLURGICAL AND MATERIALS TRANSACTIONS A