Calhoun: The NPS Institutional Archive DSpace Repository Faculty and Researchers Faculty and Researchers' Publications 2011 Development of nanocrystalline structure in Cu during friction stir processing (FSP) Su, Jian-Qing; Nelson, T.W.; McNelley, T.R.; Mishra, R.S. Elsevier J.-Q. Su, T.W. Nelson, T.R. McNelley, R.S. Mishra, "Development of nanocrystalline structure in Cu during friction stir processing (FSP)," Materials Science and Engineering, v. A528, (2011), pp. 5458-5464. http://hdl.handle.net/10945/55369 Downloaded from NPS Archive: Calhoun Author's personal copy Materials Science and Engineering A 528 (2011) 5458–5464 Contents lists available at ScienceDirect Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea Development of nanocrystalline structure in Cu during friction stir processing (FSP) Jian-Qing Su a,∗ , T.W. Nelson b , T.R. McNelley a , R.S. Mishra c a Department of Mechanical and Astronautical Engineering, Naval Postgraduate School, Monterey, CA 93943-5146, USA Department of Mechanical Engineering, Brigham Young University, Provo, UT 84604, USA c Department of Materials Science and Engineering, Missouri University of Science and Technology, Rolla, MO 65409, USA b a r t i c l e i n f o Article history: Received 9 December 2010 Received in revised form 7 March 2011 Accepted 7 March 2011 Available online 12 March 2011 Keywords: Friction stir processing (FSP) Orientation imaging microscopy (OIM) Grain refinement Nanocrystalline structure Microband a b s t r a c t The characteristics of microstructures at various locations behind the pin tool extraction site were studied in copper after FSP that had been conducted with continuous quenching to enhance cooling rates. The substructures initially formed around the pin tool consist of very small crystallites having sizes of a few tens of nanometers. It is proposed that the processing conditions result in formation of microband structures around pin tool in the presence of severe strain heterogeneity. The microbands appear as nanoscale elongated crystallites surrounded by high-angle boundaries. The elongated crystallites transform to nearly random oriented and equiaxed grain structures by shape adjustment during the initial stages of cooling from the peak temperature. Nanocrystalline structures ∼174 nm in size were produced in OFHC copper by FSP. © 2011 Elsevier B.V. All rights reserved. 1. Introduction Many methods for grain refinement and property enhancement in metals and alloys include plastic deformation of the material [1]. Two conditions should be met in order to produce nanocrystalline structures by plastic deformation: (a) a large concentration of nuclei or sub-structures of nano-scale size are formed in the deformed material, and (b) the nuclei or sub-structures can transform to new grains surrounded by high-angle boundaries without significant grain growth. It is usually thought unlikely that nano-scale grain sizes can be attained in metals and alloys by high-temperature plastic deformation because of the difficulty in attaining a sufficient concentration of refined sub-structure units as well as a strong growth tendency of such units as they transform from sub-structures to new grains at elevated temperature. Friction stir processing (FSP) is a recently developed thermomechanical processing technique based on the principles of friction stir welding (FSW) [2,3]. The material undergoes intense plastic deformation at temperatures near the peak temperature of the FSP/FSW thermo-mechanical cycle. The local peak temperatures attained in various aluminum and copper alloys have been ∗ Corresponding author. Present address: Department of Materials Science and Engineering, Missouri University of Science and Technology, Rolla, MO 65409, USA. Tel.: +1 573 341 4547; fax: +1 573 341 6934. E-mail address: jsu@mst.edu (J.-Q. Su). 0921-5093/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2011.03.043 found to be between 0.6Tm and 0.95Tm , values which correspond to the hot working temperature range for these metals and alloys [4]. Conventional thermo-mechanical processing methods, e.g., hot rolling, involve approximately uniform and isothermal straining of work piece materials. In contrast, the FSP/FSW thermo-mechanical cycle involves rapid transients and steep gradients in strain, strain rate and temperature. During FSP the material in contact with the pin tool experiences the largest strain, the highest strain rate and attains the highest peak temperature. The resulting microstructures then evolve after passage of the pin tool in a region of decreasing strain, strain rate and temperature. By combining enhanced cooling methods with FSP, nanocrystalline structures have been created in Al alloys [5–7]. However, the mechanisms governing the formation of such small grains were not established. Restoration mechanisms such as dynamic recovery or recrystallization that are understood to act during conventional thermo-mechanical processing are inadequate to explain refinement to nanoscale level during FSP of these alloys. To establish the mechanisms of formation of nanocrystalline structures during FSP it is necessary to examine the sub-structures formed in earliest stages of deformation nearby the pin tool. Orientation imaging microscopy (OIM) can provide quantitative measurement of grain structures at such locations and reveal microstructure evolution by evaluating grain characteristics in locations behind pin tool. Such information is difficult to obtain in highly refined structures in Al alloys but good quality elec- Author's personal copy J.-Q. Su et al. / Materials Science and Engineering A 528 (2011) 5458–5464 5459 two weeks of the FSP, grain structures of the FS processed sample were examined using an EDAX-TSL OIM system installed on a Philips XL-30S FEG scanning electron microscope. The scanning step size corresponds to a pixel size of 10 nm × 10 nm and was used for all regions behind pin tool. The error in the determination of the crystallographic orientation of each grain was less than 2◦ . A larger area OIM scan was performed to assess the final microstructure for statistical purposes. In this scan, the scanning step was 20 nm and the resulting data were subjected to clean-up procedures as follows: (1) grain dilatation with a grain tolerance angle of 5◦ , and (2) each grain contains at least two scan points. 3. Results and discussion Fig. 1. Schematic the FSP and the positions scanned by OIM in the processed material. tron backscattered diffraction (EBSD) patterns allow OIM for grains smaller than 100 nm in copper. The production of nanocrystalline structures during conventional high temperature deformation is generally not feasible even in alloys with multiple solute impurities and various particles to retard grain growth. It would likely be impossible in pure metals. In the present work, FSP was performed on commercial OFHC (C10100) copper with a purity of 99.99% and practical measures were employed to enhance cooling rates to preserve the deformation-induced features of microstructure. Then, microstructures at the pin tool surface as well as in various locations behind tool were examined to assess the formation and evolution mechanisms of nanocrystalline structures during FSP. 2. Experimental Polycrystalline copper sheet 2 mm in thickness was prepared by cold rolling and annealing of a commercial OFHC (C10100) copper plate having a purity of 99.99%. The initial Cu plate thickness of 6 mm was reduced to 2 mm in three rolling passes with 30 min anneals at 500 ◦ C after each rolling pass, providing a final annealed microstructure with a grain size of ∼70 !m. A smooth FSP tool with shoulder diameter, pin diameter and pin length of 7.5 mm, 2.5 mm and 2 mm, respectively, was fabricated in H13 steel and hardened to HRC 52. The tool was tilted ∼2.5◦ opposite the traveling direction during a single processing pass on the Cu sheet at 800 rpm and a travel speed of 120 mm min−1 . Dry ice was continuously placed on the surface of the processed material behind the tool to enhance cooling rates during the processing run. The pass was terminated prior to reaching the edge of the sheet by quickly lifting the rotating tool and the material surrounding the tool extraction site was then immediately quenched with additional dry ice in order to preserve the microstructures at locations around the tool. It should be mentioned that because of tilting of the tool, the material behind the tool quickly detached itself from the tool pin and remains intact during the rapid tool extraction, whereas the material ahead the tool might be pulled out with the lifted tool. The as-processed sheet was sectioned through the pin extraction site along the axis of the tool traverse in order to reveal the plane defined by the traversing direction and the sheet normal, as indicated in Fig. 1. Samples for OIM analysis were ground on progressively finer SiC papers to a 4000 grit finish and then polished with 0.05 !m colloidal silica in a vibratory polisher for 6 h. After Microstructures at the mid-depth of the stir zone and various locations behind the tool pin were examined by OIM. As illustrated in Fig. 1, these regions are denoted R1–R8, and are at progressively greater distances from the tool pin–stir zone interface. Locations R1-R7 were under the tool shoulder, and are ∼0, 0.01, 0.02, 0.06, 0.2, 1 and 2 mm away from the tool pin, respectively. Location R8 is located beyond the tool shoulder and exhibits essentially the final state of the microstructure. 3.1. Microstructure formed around pin tool The OIM inverse pole figure (IPF) map in Fig. 2a is from location R1 (Fig. 1) and shows the highly refined crystallites present nearby the tool pin–stir zone interface at the pin extraction site. These substructures were formed in the condition of peak temperature, largest strain and highest strain rate during FSP and comprise very small elongated crystallites with widths of 20–70 nm and lengths of 50–200 nm. The boundary statistics of these crystallites are presented in the misorientation angle distribution in Fig. 2b. The crystallite boundaries are mainly of high-angle in nature. Thus, the substructures formed around pin tool are nano-scale crystallites surrounded mostly by high-angle boundaries. Metals and alloys of high stacking fault energy (SFE) undergo restoration by dynamic recovery (DRV) when subjected to deformation at elevated temperature while those having low or medium SFE, such as copper [8], exhibit restoration by DRV and/or dynamic recrystallization (DRX). Derby [9] has further classified DRX mechanisms into rotational and migrational types. In migration recrystallization, new strain-free grains nucleate and grow within the deforming material when the strain reaches a critical level. In rotation recrystallization, a dynamically recovered substructure undergoes a gradual increase in misorientation, leading to the transformation of cell boundaries into sub-boundaries, and then into high-angle boundaries. Although it is unlikely that nanocrystalline structures will form in metals and alloys through DRX in conventional thermomechanical processing, very fine grained structures 0.1–0.3 !m in size have been reported in Cu subjected to high-strain and highstrain-rate deformation, e.g., after shock loading [10,11]. The peak temperature attained during such loading was estimated to be 0.4Tm –0.5Tm , at which the recrystallization (static or dynamic) occurred, when the material was initially at room temperature. Dynamic recrystallization was proposed to be responsible for the refined grain structures. The evolution of the deformation-induced microstructure during such high-strain, high-strain-rate deformation was described as: (1) elongated dislocation cell formation by dynamic recovery; (2) formation of elongated subgrains; and (3) break-up of elongated subgrains to form a microcrystalline structure [10,11]. The present results, as shown in Fig. 2, indicate that very refined substructures form around pin tool during the FSP. The possibility that these substructures are dislocation cells formed by Author's personal copy 5460 J.-Q. Su et al. / Materials Science and Engineering A 528 (2011) 5458–5464 Fig. 2. Characteristics of substructures formed around pin tool. (a) OIM image of the crystal structures, (b) statistic misorientation angle distribution, (c) misorientation along A–B (point-to-point), (d) misorientation along C–D (point-to-point) and (e) misorientation accumulation in an elongated crystallite along E–F (point-to-origin). dynamic recovery or subgrains developed from the cell structure must be considered. It has been shown that the dislocation cell (DC) size decreases with increasing plastic strain and a concurrent increase in dislocation density, and ultimately tends to a saturation value at very large plastic strains. For copper, this value is about 100 nm during room temperature deformation [12,13]. The deformation temperature is also a factor that influences the cell size and it has been established that the cell size increases with increasing deformation temperature in copper [14]. Thus, the minimum cell size that may be formed during hot deformation is expected to be larger than 100 nm in copper. However, the present results indicate that the substructures formed around pin initially are very small and on the order of 20–50 nm in size. This is apparent in upper right-hand corner of the image in Fig. 2a, the location closed to the pin–stir zone interface. Based on the Holt’s prediction [15], the relationship between the DC size (dDC ) and the applied shear stress (!) for OFHC copper is [16]: ! = 10.5 Gb dDC (1) where G is the shear modulus and b is the magnitude of the Burgers vector. If the finest substructures are assumed to be such dislocation cells, a shear stress of G/7.5 to G/18.6 is required for dDC sizes of 20–50 nm. Such a high shear stress at the pin–stir zone interface is implausible, especially considering that the local temperature is ≥0.5Tm . In addition, the crystallites are surrounded by high-angle boundaries and it is unlikely that these highly refined substructures are dislocation cells or subgrains that have evolved from even smaller cell structures. Recrystallization nuclei (static or dynamic) do not always evolve from a subgrain structure [17]. In conventional thermomechanical processing, the deformation-induced microstructure generally comprises of a mixture of the prior high-angle grain boundaries, shear bands, deformation bands, and low-angle subboundaries formed from accumulated dislocations. Such a mixed substructure provides numerous opportunities for nucleation. Thus, nucleation may occur at prior high angle grain boundaries, between two deformation bands (i.e., in transition bands) or within shear bands [18–21], resulting in a heterogeneous distribution of nuclei in the deformed matrix. In the FS processed material in this study (Fig. 2), the microstructure closest to the tool pin consists of a relatively homogeneous distribution of elongated crystallites. No features suggesting nuclei at prior high angle grain boundary or between or inside of band structures are apparent. Geometric DRX (GDRX) has been invoked as a mechanism for development of fine grained structures during thermo-mechanical processing [8,22]. The prior grains change shape congruently with the imposed strain while equiaxed subgrains form and remain equiaxed during deformation to large strains. The subgrain boundaries interact with the prior high-angle grain boundaries, leading to a serrated character of these boundaries. At very high strains the original grains will have been distorted to the point that their minimum, or thickness, dimension is on the order of 2–3 times the size of the equiaxed subgrains present in them. Then, GDRX occurs as opposite serrated boundaries begin to meet, leading to a pinching off of the original grains into short grain segments. Through this mechanism, up to one-third of the prior subgrain facets become high angle boundaries. However, the grain structures formed by GDRX still contain a large fraction of low-angle boundaries. In addition, neighboring grain segments from the same elongated, original grain will have a small lattice misorientation, especially in the immediate aftermath of GDRX. In contrast, the substructures formed around the tool pin during the FSP are surrounded almost entirely by high angle boundaries. As shown in Fig. 2c and d, neighboring, elongated crystallites exhibit high misorientations in both longitudinal and transverse directions. The forgoing considerations suggest that the microstructures formed around the tool pin during the FSP cannot be interpreted in Author's personal copy J.-Q. Su et al. / Materials Science and Engineering A 528 (2011) 5458–5464 terms of conventional DRV or DRX models alone. Instead, the highly refined crystallites are deformation-induced structures formed under unconventional processing conditions, i.e., there are steep gradients and as well as transients in strain, strain rate and temperature in the material in contact with and nearby the tool pin. Deformation-induced microstructures have been extensively investigated in deformed metals and alloys, and many microstructural features have been reported to be common to both cold and hot deformation, e.g., cell blocks, cells, dense dislocation walls, and subgrains [23]. It is also found that, during cold deformation, a microband structure starts to appear when the strain reaches a certain value [24–27]. With increasing strain, the propensity for microbands increases and the width of microbands decreases. It is recognized that a microband structure is often observed under conditions of highly localized microplastic deformation and the development of nonequilibrium structures [27,28]. The elongated substructures observed around the tool pin in this study are similar to the microband structures formed in metals during cold deformation [26,27]. In the current study, the use of a small tool and impressed cooling resulted in a rapid thermo-mechanical cycle and material flow within a very thin layer around the tool pin and, especially, abnormally steep gradients in strain and strain rate within the deformed volume. This likely results in severe strain heterogeneity in the deforming material and results in generation of microbands around the pin. In cold deformation, the initiation of microbands has been inferred to result from extensive dislocation activity [29,30]. It is proposed that the microbands are derived from walls of dislocations in the deformed matrix rather than by rearrangement of dislocations in the existing cell structure [26,29,30]. The microband walls are characterized by a denser dislocation arrangement than the lower-angle dislocation cell structures [31]. In a study of deformed bcc metal, a local lattice rotation model was also presented to explain the rotation of the microband groups [32]. The formation and evolution mechanisms of the microband structures observed here remain to be clarified but a steep gradient in strain in the deforming material is believed to be a factor contributing to substructure formation. Dislocations generated during plastic deformation can be divided into statistically stored dislocations (SSD) and geometrically necessary dislocations (GND) [33,34], and the total dislocation density ("T ) is a sum of the densities of SSD ("SSD ) and GND ("GND ): "T = "SSD + "GND (2) The SSDs reflect homogenous deformation and the GNDs are additional dislocations needed to accommodate the strain gradients [35], i.e., the presence of a steep strain gradient leads to an increased total dislocation density. The GNDs are also found to have an inhomogeneous distribution with very high local density values [36]. In this investigation, quantitative data on the strain gradient around the tool pin are not available. However, it is proposed that this gradient accounts for enhanced dislocation accumulation in the stir zone and the appearance of walls of high dislocation density in the development a refined microband structure. The microband structures observed in this study are surrounded mainly by high-angle boundaries. It is likely that the evolution of the initial microband dislocation walls may also include local lattice rotations. More work is needed in this area. The microbands formed around the tool pin appear to be distributed uniformly in the thin layer of deformed matrix surrounding the pin extraction site and are very refined even in their longitudinal direction. Under these FSP conditions the strain gradient and severe strain heterogeneity in local micro-regions resulted in generation of a high density sites for microband formation. The 5461 growth of the microbands was suppressed by each other leading to retention of highly refined and uniformly distributed structures. These microband features are small enough that there is no further substructure development in most of them. As shown in the example in Fig. 2e, the accumulated misorientation within an elongated microband is less than 1◦ which is within measurement error. Thus, these microbands appear to be individual, elongated crystallites surrounded by high-angle boundaries. 3.2. Microstructure evolution Fig. 3 presents OIM analysis of crystallite structures that evolve from the microbands formed around pin tool. These results are from the regions identified in Fig. 1. The elongated crystallites appear quickly to transform to, by coarsening mainly in the transverse direction, equiaxed grain structures behind the tool. Equiaxed crystallites have formed at location R3 at a distance of 0.02 mm from the pin tool–stir zone interface. With increasing distance behind the tool, the equiaxed grain structures appear to grow slightly in a region of decreasing temperature (R4–R8 in Fig. 3). In the present investigation, the final grain structures remained within the submicron regime. There is little difference in the misorientation angle distributions obtained in these different regions and the crystallites remain surrounded by mainly high-angle boundaries. The microtextures, however, vary during microstructure evolution. In this paper, all the pole figures are presented with the reference direction (RD) being aligned with the normal to the plate surface and the transverse direction (TD) aligned transverse to the process. As shown in Fig. 3 (location R1), the elongated crystal structures close pin tool exhibit a distinct fiber texture formed by rotation of a shear component about an axis approximately aligned with the tool axis. With transformation to equiaxed grain structures the texture becomes more random. This may reflect grain boundary migration or even rotation during the evolution. The mechanism of the transformation, however, is still not clear, and more work is needed in this area. A larger area OIM scan was performed for the final microstructure in order to obtain better statistical data. The OIM image in Fig. 4a reflects a predominantly high-angle boundary structure and equiaxed grains, with a few low angle boundaries. Inspection of the pole figure in Fig. 4b shows nearly a random texture of the FS processed microstructure although a weak shear component may still be discerned. The misorientation angle distribution (Fig. 4c) also suggests a randomly oriented assembly of grains. The corresponding grain size distribution is shown in Fig. 4d and e. Grain sizes achieved in the specimen ranged from 30 nm to 390 nm, and there are relatively few large grains in the distribution. Thus, the average grain sizes measured from about 990 grains are 109 nm by number-weighting and 174 nm by area-weighting, respectively. Fig. 4f and g summarizes the boundary character analysis based on the misorientation data. Three grain boundary categories are defined: low-angle boundaries (LAB), coincidence P P site lattice (CSL) boundaries ( 3– 31), and random high-angle grain boundaries (HAB). The grain boundary character data (Fig. 4f) reveal a high fraction (0.93) of HABs and approximately P 16% of the boundaries were classified as coincident site, or , boundaries while the remainder were LABs (∼0.07 fraction). Among the CSL P boundaries, the 3 CSL boundaries were present in the highest fraction (Fig. 4g). It is known that processing and annealing may introduce a large amount of twin boundaries in Cu [37–40]. The P results from this study show that the 3 boundaries are less than 8% of the total boundaries in the FS processed Cu. Although the formation and coarsening mechanism is not clear during P development of the nanocrystalline structures, the lack of 3 boundary suggests that twinning does not play an essential role in the process. Author's personal copy 5462 J.-Q. Su et al. / Materials Science and Engineering A 528 (2011) 5458–5464 Fig. 3. OIM analysis including crystal structures, pole figures and misorientation angle distributions for different regions behind pin tool from R1 to R8 as indicated in Fig. 1. Author's personal copy J.-Q. Su et al. / Materials Science and Engineering A 528 (2011) 5458–5464 5463 Fig. 4. Grain structures in the FS processed copper. (a) Grain map produced by OIM, (b) pole figure, (c) statistic grain boundary misorientation distribution, (d) grain size distribution in area fraction, (e) grain size distribution in number fraction, (f) grain boundary character distribution and (g) the occurrence frequency fractions of CSL boundaries. In this study, the formation of sub-micron and nano-scale crystalline structures reflects a direct transformation by coarsening of the highly refined microband structures formed around the tool pin. It should be noted that several recrystallization theories including continuous dynamic recrystallization, discontinuous dynamic recrystallization, geometric dynamic recrystallization and recrystallization via particle stimulated nucleation, which were developed to account for grain size control during conventional thermo-mechanical processing, have also been used to explain the formation of fine grain structures (usually 1–10 !m) during FSW/FSP [5,41–48]. Common FSW and FSP technologies inherently involve severe plastic deformation (SPD) as well as transients and gradients in strain, strain rate and temperature. Such gradients strongly depend on the processing conditions and governing Author's personal copy 5464 J.-Q. Su et al. / Materials Science and Engineering A 528 (2011) 5458–5464 factors include tool design, processing parameters and imposed cooling rates. Although it has not been quantitatively defined, shallower gradients and material flow in a thicker layer around the tool pin may result in a lesser degree of microband formation so that the deformation structures that are often observed in conventional thermo-mechanical processing, such as dislocation cells, deformation bands and elongated original grains, may also form in the deformed matrix around pin tool. These deformation features evolve to final structures through recognized mechanisms that have been reported in studies of FSW/FSP [5,37–44]. In the present study, under the exceptionally steep gradient and transient conditions resulting from use of a small tool and imposition of rapid cooling, the material flow in a very thin layer around pin tool resulted in severe strain heterogeneity giving rise to highly concentrated microplastic deformation at locations within this layer. Under such circumstances, a very high density of microband structures were formed having units as small as a few tens nanometer and containing no further substructures within them. These microbands appear as individual crystallites having high boundary misorientations and they rapidly transform to randomly oriented equiaxed grains by shape adjustment after passage of the tool. It is clear that, during FSP, the resulting microstructures are strongly depend on the processing conditions. 4. Conclusions FSP is an efficient technique for grain refinement in metals and alloys. Using small tools and imposing rapid cooling, nanocrystalline structures were successfully produced in OFHC copper in a single step. The resulting microstructures consist of equiaxed, nearly random oriented grains surrounded by high-angle boundaries. The grain size ranges mainly from 50 to 300 nm with an average size of about 109 nm (number-weighted) and 174 nm (area-weighed), respectively. The substructures initially formed around pin tool are very small, elongated crystallites. The elongated crystallites transform quickly to equiaxed grain structures by stable coarsening after passage of the tool. It is proposed that, for the processing conditions of the current research, that steep gradients in strain, strain rate and temperature cause severe strain heterogeneity giving rise to highly concentrated microplastic deformation in local micro-regions in the material around pin tool. Under such circumstances, a large amount of microbands are formed in the deformed matrix. The microband structures appear as nano-scale elongated crystallites surrounded by high angle boundaries. The small crystallite structures rapidly evolve to become equiaxed and randomly oriented grain structures behind the tool pin. Acknowledgements J.Q. Su would like to acknowledge the support of the US National Research Council (NRC) Research Associateship program. The authors acknowledge partial support by the Office of Naval Research under contract number N00014-09-WR20201, with Drs. J. Deloach and R. Fonda as Scientific Officers. References [1] C.Y. Barlow, N. Hansen, in: J.R. Groza, J.F. Schackelford, E.J. Lavernia, M.T. Powers (Eds.), Materials Processing Handbook, CRC Press, Boca Raton, FL, 2007 (Chapter 12-1). [2] W.M. Thomas, E.D. Nicholas, J.C. Needham, M.G. Murch, P. Templesmith, C.J. Dawes, Friction Stir Butt Welding, G.B. Patent Application No. 9125978.8, Dec. 1991; U.S. Patent No. 5460317, Oct. 1995. [3] R.S. Mishra, M.W. Mahoney, Mater. Sci. Forum 507 (2001) 357–359. [4] R.S. Mishra, Z.Y. Ma, Mater. Sci. Eng. R 50 (2005) 1–78. [5] C.G. Rhodes, M.W. Mahoney, W.H. Bingel, M. Calabrese, Scr. Mater. 48 (2003) 1451–1455. [6] J.Q. Su, T.W. Nelson, C.J. Sterling, J. Mater. Res. 18 (2003) 1757–1760. [7] J.Q. Su, T.W. Nelson, C.J. Sterling, Philos. Mag. 86 (2006) 1–24. [8] F.J. Humphreys, M. Hatherly, Recrystallization and Related Annealing Phenomena, Elsevier, Oxford, 2004. [9] B. Derby, Acta Metall. Mater. 39 (1991) 955–962. [10] U.R. Andrade, M.A. Meyers, K.S. Vecchio, A.H. Chokshi, Acta Metall. Mater. 42 (1994) 3183–3195. [11] M.A. Meyers, V.F. Nesterenko, J.C. LaSalvia, Y.B. Xu, Q. Xue, J. Phys. IV France 10 (2000) 51–56. [12] J.D. Embury, in: A. Kelly, R.B. Nicholson (Eds.), Strengthening Mechanisms in Crystals, Wiley, New York, NY, 1971, pp. 331–402. [13] K. Wang, N.R. Tao, G. Liu, J. Lu, K. Lu, Acta Mater. 54 (2006) 5281–5291. [14] T. Tabata, K. Takagi, H. Fujita, Trans. Jpn. Inst. Met. 16 (1975) 569–579. [15] D.L. Holt, J. Appl. Phys. 41 (1970) 3197–3201. [16] M.R. Staker, D.L. Holt, Acta Metall. 20 (1972) 569–579. [17] B. Hutchinson, Scr. Metall. Mater. 27 (1992) 1471–1475. [18] P.A. Beck, P.R. Sperry, J. Appl. Phys. 21 (1950) 150–152. [19] S.P. Bellier, R.D. Doherty, Acta Metall. 25 (1977) 521–538. [20] I.L. Dillamore, P.L. Morris, C.J.E. Smith, W.B. Hutchinson, Proc. R. Soc. A329 (1972) 405–420. [21] A.A. Ridha, W.B. Hutchinson, Acta Metall. 30 (1982) 1929–1939. [22] H.J. McQueen, O. Knustad, N. Ryum, J.K. Solberg, Scr. Metall. 19 (1985) 73–78. [23] N. Hansen, Metall. Mater. Trans. 32A (2001) 2917–2935. [24] A.S. Malin, M. Hatherly, Met. Sci. 13 (1979) 463–472. [25] M. Hatherly, A.S. Malin, Scr. Metall. 18 (1984) 449–454. [26] Q.Z. Chen, B.J. Duggan, Metall. Mater. Trans. 35A (2004) 3423–3430. [27] J.C. Sanchez, L.E. Murr, K.P. Staudhammer, Acta Mater. 45 (1997) 3223–3235. [28] A.E. Romanov, V.I. Vladimirov, in: F.R.N. Nabarro (Ed.), Dislocations in Solids 9, North-Holland, Amsterdam, 1992, pp. 191–302. [29] W.B. Hutchinson, Philos. Trans. R. Soc. Lond. A357 (1999) 1471–1485. [30] M. Hatherly, in: R.C. Gifkins (Ed.), ICSMA 6, Pergamon Press, Oxford, United Kingdom, 1983, p. p1181. [31] J.C. Huang, G.T.I.I.I. Gray, Acta Metall. Mater. 37 (1989) 3335–3347. [32] D. Dorner, Y. Adachi, K. Tsuzaki, Scr. Mater. 57 (2007) 775–778. [33] M.F. Ashby, Philos. Mag. 21 (1970) 399–424. [34] D.A. Hughes, N. Hansen, D.J. Bammann, Scr. Mater. 48 (2003) 147–153. [35] N.A. Fleck, G.M. Muller, M.F. Ashby, J.W. Hutchinson, Acta Metall. Mater. 42 (1994) 475–487. [36] E. Demir, D. Raabe, N. Zaafarani, S. Zaefferer, Acta Mater. 57 (2009) 559–569. [37] K. Han, R.P. Walsh, A. Ishmaku, V. Toplosky, L. Brandao, J.D. Embury, Philos. Mag. 84 (2004) 3705–3716. [38] C.X. Huang, K. Wang, S.D. Wu, Z.F. Zhang, G.Y. Li, S.X. Li, Acta Mater. 54 (2006) 655–665. [39] H. Miura, T. Sakai, S. Andiarwanto, J.J. Jonas, Philos. Mag. 85 (2005) 2653–2669. [40] Ph. Gerber, T. Baudin, R. Chiron, B. Bacroix, Mater. Sci. Forum 495–497 (2005) 1303–1308. [41] K.V. Jata, S.L. Semiatin, Scr. Mater. 43 (2000) 743–749. [42] J.Q. Su, T.W. Nelson, R. Mishra, M. Mahoney, Acta Mater. 51 (2003) 713–729. [43] J.Q. Su, T.W. Nelson, C.J. Sterling, Mater. Sci. Eng. A405 (2005) 277–286. [44] K.V. Jata, K.K. Sankaran, J.J. Ruschau, Metall. Mater. Trans. A31 (2000) 2181–2192. [45] T.R. McNelley, S. Swaminathan, J.Q. Su, Scr. Mater. 58 (2008) 349–354. [46] Y.S. Sato, M. Urata, H. Kokawa, Metall. Mater. Trans. A33 (2002) 625–635. [47] R.W. Fonda, J.F. Bingert, K.J. Colligan, Scr. Mater. 51 (2004) 243–248. [48] K.A.A. Hassan, B.P. Wynne, P.B. Prangnell, Fourth International Symposium on FSW, TWI, Park City, USA, 2003 (CD ROM).