You are on page 1of 12

Available online at www.sciencedirect.

com

Acta Materialia 58 (2010) 64526463 www.elsevier.com/locate/actamat

Texture development in near-a Ti friction stir welds


R.W. Fonda , K.E. Knipling
Naval Research Laboratory, Code 6356, 4555 Overlook Avenue, SW, Washington, DC 20375, USA Received 8 June 2010; received in revised form 6 August 2010; accepted 7 August 2010

Abstract The microstructures and crystallographic textures produced during friction stir welding of the near-a Ti-5111 titanium alloy were characterized as a function of welding speed. The textures produced were compared with ideal hexagonal close packed (hcp) shear textures and with predicted textures of hcp Burgers variants of ideal body-centered cubic (bcc) shear textures, showing that the deposited welds are dominated by the hcp P1 and bcc D1 textures. The hcp P1 shear texture was dominant at slow weld speeds, while the bcc D1 shear texture was dominant at the fast weld speed. This variation appears to result from a poor transmission of the shear deformation from the rotating tool to the deposited weld that develops at faster welding speeds. These observations are compared to other studies of friction stir welds in hcp and bcc materials reported in the literature. Published by Elsevier Ltd. on behalf of Acta Materialia Inc.
Keywords: Friction stir welding; Shear texture; Titanium alloys; Electron backscattering diraction (EBSD)

1. Introduction Friction stir welding (FSW) is a solid-state joining technique that was developed by the Welding Institute (TWI) in 1991 [1] and has rapidly developed into a commercially important joining process for aluminum alloys [2,3]. In FSW, a rotating tool is translated along a joint line to stir together the two sides of the joint. The rotating tool heats the surrounding material through frictional and adiabatic heating, softening it and enabling its transfer around the tool and into the tools wake. Using an appropriate tool design and welding parameters, the deposited material consolidates into a defect-free weld behind the tool (see Fig. 1). The rotating FSW tool introduces shear deformation into the surrounding material [4]. While there has been considerable work on the crystallographic textures that develop during shear deformation of face-centered cubic (fcc) materials [511], much less attention has been devoted to shear textures developed in body-centered cubic (bcc) materials. The four primary bcc shear textures, later desig-

Corresponding author. Tel./fax: +1 202 767 2622.

E-mail address: richard.fonda@nrl.navy.mil (R.W. Fonda).

nated D, E, F, and J (see Table 1), were initially reported by Backofen and Hundy [12] in torsion studies of Armco iron. Subsequent studies of this alloy only reported the presence of two [6] or three [8] of these textures, but a detailed analysis using crystal plasticity theory and experimental torsion tests of three iron alloys by Baczynski and Jonas [13] conrmed the four ideal orientations and described their locations along the torsion bers. That study also determined that the D2( 1 1 2)[1 1 1] shear texture component (in Baczynski and Jonas notation) dominates at elevated temperatures. More recently, Li et al. [14] observed these shear textures after simulated equal channel angular extrusion (ECAE) of a bcc material. They, however, revised the notation of the bcc ideal shear orientations in order to reinforce the similarities with the more established fcc shear textures. Their designations switched the D1 and D2 orientations and used E, E, J, and J in place of Baczynski and Jonas E2, E1, J2, and J1, respectively, emphasizing that the bcc shear planes and directions are related to the fcc shear planes and directions simply through an exchange of the hkl and uvw indices. Thus, the new bcc D1( 1 1 2)[1 1 1] shear texture bears an obvious resemblance to the fcc A1*(1 1 1)[ 1 1 2] orienta tion, the bcc E(1 1 0)[1 1 1] corresponds to the fcc

1359-6454/$36.00 Published by Elsevier Ltd. on behalf of Acta Materialia Inc. doi:10.1016/j.actamat.2010.08.007

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

6453

Fig. 1. Schematic of the friction stir welding process.

A(1  1 1)[1 1 0], and the bcc J(1 1 0)[1  1 2] corresponds to the fcc B(1  1 2)[1 1 0] orientation. Pole gures of these ideal bcc shear orientations are shown in Fig. 2. Li et al.s [14] revised texture designations are used throughout this paper unless otherwise specied.

Only recently have there been any examinations of shear textures in hexagonal close packed (hcp) materials. The initial analysis of these textures was reported in 2007 by Beausir et al. [15], who used crystal plasticity theory to determine the ideal shear orientations and their persistence characteristics in hcp materials in simple shear, and compared those results to the experimental textures produced during torsion testing of a magnesium alloy. The possible hcp simple shear textures consist of ve bers, named B, P, Y, C1, and C2, and the end orientation of the P ber, named P1 (see Table 1). Beausir et al., however, demonstrated that the Y ber is only convergent from 30 6 u2 < 60 and that only one of the C bers is convergent. Furthermore, they indicated that the large rotation vectors around the Y and C bers would prevent any signicant intensities from developing around those bers. A similar analysis performed by Li [16] additionally predicted two possible twin-induced shear bers, which were labeled h5 and h6, and determined the most stable orientations along

Table 1 Ideal crystallographic orientations of bcc [14] and hcp [15,16] simple shear deformation textures with Euler angles (Bunges notation) from the u2 = 0 and u2 = 45 sections when applicable. Texture Description f 1 1 2g h1 1 1i f1 1  2g h1 1 1i {1 1 0} h1  1 1i {1 1 0} h 11 1i {1 1 0} h0 0 1i {1 1 0} h1  1 2i {1 1 0} h 11 2i {0 0 0 1} huvjwi {hkil} h1 1  2 0i f 1 1 0 0g h1 1  2 0i c-axis tilted 30 towards SP c-axis tilted 30 towards SP from SD c-axis tilted 30 towards SP from SD {hkil} h1 1  2 0i { 1 1 0 0} h1 1  2 0i Euler angles () u1 bcc bcc bcc bcc bcc bcc bcc hcp hcp hcp hcp hcp hcp hcp hcp D1 D2 E E F J J B-ber P-ber P1 Y-ber C1-ber C2-ber h5-ber {Li-2008-1031} H6-ber {Li-2008-1031} 54.7/234.7 144.7/324.7 125.3/305.3 35.3/215.3 90 270 0/180 90/270 90/210/330 30/150/270 0 0 0 0 60 120 85 140 U 45 90 45 90 35.3 35.3 45 90 54.7 54.7 90 090 0 30 90 90 90 90 u2 0 45 0 45 45 45 0 45 45 45 060 0 0 3060 060 060 060 060

Fig. 2. Pole gures showing the important ideal orientations associated with simple shear deformation of bcc materials (after Baczynski and Jonas [13] and Li et al. [14]).

6454

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

each of these ideal shear orientation bers. Pole gures of these ideal hcp shear orientations (except h5 and h6) are shown in Fig. 3. Investigations into the crystallographic textures that develop in friction stir welds have focused primarily on aluminum alloys [4,1727]. There has been comparatively little work on the development of weld microstructures during the FSW of steels [2830] and titanium alloys [3135]. Analysis of the shear textures produced during friction stir welding (or processing) of bcc and hcp materials has been similarly limited. The only reports known to the current authors are on the metastable bcc b-21S titanium alloy [31], the ab Ti6Al4V titanium alloy [33], pure iron [30], the near-a Ti-5111 titanium alloy [34], and the similar commercial purity a titanium alloy [35]. Reynolds et al. [31] studied the crystallographic texture in friction stir welds of a b titanium alloy, b-21S, which retains the metastable b phase after cooling. They observed that a rotation of $30 produced an excellent agreement with the shear textures reported by Rollett and Wright [36] for bcc tantalum. However, the unrotated FSW textures were later shown [34] to match the bcc D2( 1 1 2)[1 1 1] shear texture component in Baczynski and Jonas [13] notation, corresponding to a D1( 1 1 2)[1 1 1] shear texture in Li et al.s [14] notation. In friction stir welds of Ti6Al4V, Mironov et al. [33] measured the texture of the retained b phase in the deposited weld using electron backscatter diraction (EBSD). The ow patterns on the two sides of the weld were used to approximate the shape of the eective shear plane created by the tool shoulder and the tapered surface of the submerged pin. From this, they were able to rotate the pole gure acquired from the center of the transverse cross-section into an approximate shear frame of reference, with the shear direction (SD, aligned with the transverse direction at this location) aligned horizontally and the shear plane normal (SPN, perpendicular to the eective inclined surface of the tool) aligned vertically. Although the b volume fraction was very small, Mironov et al. reported that the resultant texture could be roughly described as the bcc J(1 1 0)[1  1 2] ideal simple shear texture.

Mironov et al. [30] employed a similar approach in determining the shear textures produced during friction stir processing of pure iron. Rotating the textures observed in three dierent regions of the weld transverse cross-section into the shear reference frame, these researchers observed 1 1 2)[1 1 1] consistent textures that were identied as D2( (using Li et al.s notation). More recently, Knipling and Fonda [34] examined the texture produced around the friction stir welding tool and in the deposited weld of plan-view sections of friction stir welds in Ti-5111, a near-a alloy which retains nearly no bcc b at room temperature. They discovered that rotating the observed texture from any of these locations to align the tool tangent with SD causes the predominant h1 1  2 0ia direction to align with the horizontal axis of the pole gure, and that subsequent rotations about this SD axis to account for the taper angle on the truncated conical tool nearly align a h1 0  1 0ia direction with the presumed SPN direction (normal to the tool surface) and similarly align the predominant {0 0 0 1}a perpendicular to both SD and SPN (along the inclined surface of the tool). The observed texture was identied as originating from a bcc simple shear texture, after transformation from the high temperature bcc b phase to hcp a according to the Burgers orientation relationship (OR) [37], which aligns the close-packed planes and directions such that {1 1 1}b || {0 0 0 1}a and h1 1 0ib || h1 1  2 0ia. The observed texture corresponds to a bcc D2( 1 1 2)[1 1 1] shear texture using Baczynski and Jonas notation, or a D1( 1 1 2)[1 1 1] shear texture in Li et al.s notation. There was also evidence of an hcp P1 (1 0  1 0)[1 1  2 0] shear texture component. Finally, Mironov et al. [35] studied friction stir welds in a commercial purity a titanium alloy and demonstrated that the dominant texture in the deposited weld was the hcp P1 (1 0  1 0)[1 1  2 0] shear texture. While these ve studies involved friction stir welds in different alloys, each using a unique welding condition, it is remarkable that ve dierent textures were reported in the deposited welds. Thus, the current study was initiated to determine the eect of welding parameter variations on the crystallographic texture produced in the deposited weld of friction stir welds in the near-a Ti-5111 titanium alloy.

Fig. 3. Pole gures showing the important ideal orientations associated with simple shear deformation of hcp materials (after Beausir et al. [15,16]).

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

6455

2. Experimental procedure The three friction stir welds fabricated for this study were prepared by the Edison Welding Institute with 6 mm (0.25 in.) thick Ti-5111 plate. This Ti5Al1Sn 1Zr1V0.8Mo (wt.%) alloy is a near-a titanium alloy developed for marine applications that require superior toughness, good weldability, and good stress-corrosion cracking resistance [38]. The FSW tool was from EWIs DuraStir line, a tungsten-based alloy tool with a geometry consisting of a narrow shoulder and a truncated conical shape. The three welds were prepared at dierent weld travel speeds of 0.4, 1 and 1.7 mm s1 (1, 2.5 and 4 ipm) with a constant tool rotation rate of 234 rpm. The heat of welding generated by the rotating tool varies inversely with the welding speed. This is reected in the energy input per unit length (see Table 2) that was estimated from the machine torque and welding parameters. The welds prepared with slow, medium, and fast speeds thus correspond to hot, moderate, and cold welding conditions and are designated HP, MP, and CP, respectively, in Table 2. The end of the weld was preserved by extracting the tool immediately upon completion of the welding and quenching it with cold water. This stop-action weld [19,22,39] was intended to freeze-in the microstructures surrounding the tool and provide a static representation of the microstructures that were present during the welding process. The weld exit hole was sectioned in plan-view near the plate mid-thickness for further examination. The midthickness sectioning plane minimizes inuences from both the shoulder and the base of the tool, thereby revealing the microstructural evolution introduced by the conical tool. These plan-view sections of the exit holes were examined by backscattered electron (BSE) imaging and EBSD in a JEOL JSM-7001F SEM equipped with a Hikari EBSD detector (TSL/EDAX, Draper, UT) to reveal the grain structures and textures of the deposited welds as a function of distance from the FSW tool. Texture determinations of the deposited welds were obtained at 1 lm step size from large scans (850 850, 900 400, and 1325 636 lm2 for the CP, MP, and HP welds, respectively). It is critical to identify the crystal basis orientation, the orientation of the shear deformation, and the sense of shear produced by the rotating tool to ensure consistency among all data and the literature. This study, and previous crystallographic orientation studies in the literature [15,16,33,35],
Table 2 Friction stir welding parameters and weld designations. Welding condition Hot Moderate Cold Designation HP MP CP RPM 234 234 234 Weld speed 0.4 mm/s (1 ipm) 1 mm/s (2.5 ipm) 1.7 mm/s (4 ipm) Line energy (kJ/ mm) 11.9 4.9 3.1

use a basis unit cell setting with a close-packed h1 1  2 0i direction aligned with the x-axis. This is dierent from the convention used by the TSL acquisition software, which aligns a h 1 1 0 0i direction with the x-axis. It is therefore necessary, when using the TSL software, to introduce an additional 30 rotation about u2 to appropriately align the Euler angles and ODFs for comparison with these prior studies. It is also worth mentioning that the Euler angles of shear textures produced during ECAE through an angled die are sometimes [14,16,40] reported in the ECAE reference frame, which introduces an additional 45 rotation in u1 (for a 90 die) relative to the values reported in this paper. This is normally denoted by a h or u subscript associated with those data. Also, many of the shear textures are identical except in the orientation of those textures. It is thus critical to accurately establish the actual shear frame of reference in three dimensions to correctly identify the observed textures. Finally, most EBSD acquisition systems apply a 180 rotation to the data, so extreme care is necessary to determine the three-dimensional orientation of the shear plane produced by the submerged surface of the tool (or the eective surface produced from the combined eects of tool and shoulder). It is best to compare data from at least two regions with dierent shear orientations to ensure consistency among the data and accurate identication of the local shear orientations. Finally, it is critical to recognize the exact sense of shear deformation produced by the rotating tool. Switching between a positive and negative shear direction alters the sign of u1, changing between the corresponding texture components (i.e., D1 and D2). 3. Results The three welds examined in this study were prepared at a constant 234 rpm tool rotation rate but dierent weld speeds of 0.4, 1 and 1.7 mm s1 (1, 2.5 and 4 ipm), designated as hot, moderate, and cold welding parameters (HP, MP, and CP, respectively, Table 2). Thermocouple measurements from equivalent positions in welds from each of these conditions (Fig. 4) conrm this trend the CP weld only briey experienced elevated temperatures while the HP weld had a much more extended exposure. The MP weld had an intermediate exposure to high temperatures. These thermocouple curves correlate to the heat input into the welds per unit length (see Table 2), with higher heat inputs reecting longer times of exposure to elevated temperatures at the slower weld speeds. The peak temperatures measured by these thermocouples were similarly correlated to the weld speed, with the highest peak temperatures for the HP weld, the lowest peak temperatures for the CP weld, and intermediate peak temperatures for the MP weld. The thermal proles shown in Fig. 4 were obtained from the base of the weld approximately 7.6 mm from the centerline towards the retreating side, which lies outside the stir zone, but the observed trends were generally consistent among all 14 thermocouple locations. While the temperatures measured are below the transus temperature of 980 C

6456

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

Fig. 4. Thermal proles from the three welding conditions measured at base of weld 7.6 mm from the weld center. Curves are oset in time for clarity.

[41], microstructural evidence from the weld nuggets indicates that the peak temperature exceeded the beta transus temperature in all three welds. The three friction stir welding conditions produce welds (shown in Fig. 5) similar in appearance to other titanium friction stir welds [3135,42]. They consist of a weld nugget with a very rened grain size surrounded by a very narrow (<100 lm wide) thermo-mechanically aected zone (TMAZ), where the base plate lath microstructure was deformed during welding. The transverse cross-sections in Fig. 5 do not exhibit any systematic variation with the different welding conditions, and no heat aected zone is apparent. The welding conditions and tool geometry resulted in an eective angle of the inclined surface of the

tool of approximately 35 at the plate mid-thickness of each weld (see Fig. 5). The curvature apparent towards the top of the stir zone indicates an inuence from the tool shoulder in the top 1/4 to 1/3 of the weld. That shoulder inuence is expected to be negligible at the mid-plane location examined in this paper. Plan-view cross-sections of the exit holes at the plate mid-thickness (Fig. 6a) reveal that the tool is surrounded by a $1 mm thick band of grain-rened material that extends continuously into the deposited weld. The abrupt transition between this grain-rened material and the base plate, which is not visible at the resolution in Fig. 6a, was revealed by EBSD to be approximately 100 lm wide for the three weld speeds examined. The deposited weld microstructures were examined by optical microscopy for both the transverse and plan-view cross-sections, and in plan-view with BSE imaging and EBSD scans near the weld centerline at the plate mid-thickness (see indicated region in Fig. 6a). All three welds contained small, equiaxed prior-b grains, often delineated by grain boundary a, that were subdivided by a laths. The prior-b grains conrm that the deposited welds exceeded the b transus during welding and the relatively narrow thickness of the grain boundary a indicates that the cooling rate from the b phase eld was relatively fast. The BSE images (Fig. 6bd) illustrate that the prior-b grain size is similar in the MP and HP welds and larger in the CP weld, in agreement with optical microscopy and EBSD observations. These images also show a variation from the predominantly cellular a lath structure in the HP weld to the more acicular Widmanstatten appearance of the a laths in the CP weld, reecting an increase in the cooling rate of these welds. In parallel to this, there also appears to be a coarsening of the a lath structure from the HP to the CP weld conditions. The crystallographic textures exhibited by these three welds are displayed in Fig. 7. As with the grain size determinations, the pole gures of the HP and MP welds are similar, but dier somewhat from the pole gures from the CP weld. All three welds exhibit textures dominated by a 0001 pole tilted $35 from ND in the direction away from the tool and alignment of h1 1  2 0ia with the presumed shear direction along the tool tangent. There are other minor texture components, which become more pronounced for the CP weld. 4. Discussion Analysis of the deposited weld textures in Fig. 7 requires that they be rotated into the shear deformation frame of reference. That rotation was accomplished using a similar approach as in previous reports [30,3335]. For a plan-view scan at the weld centerline, the shear direction should correspond to the tool tangent (approximately vertical in Fig. 7) and the shear plane normal should correspond to the normal of the eective inclined surface of the tool, which considers both the actual surface orientation of the

Fig. 5. Transverse cross-sections of the three welding conditions with superimposed tool prole. The boundary between the ne-grained weld nugget and base plate is approximately 35 at the plate mid-thickness.

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

6457

Fig. 6. (a) Plan-view cross-section of the exit hole of the MP weld at plate mid-thickness with indicated rotation direction, weld direction, negative shear sense, and approximate EBSD scan region in deposited weld. (bd) BSE images of the microstructures in the deposited HP, MP, and CP welds.

tool and any deviations from that orientation due to tool shoulder or end eects. The shear direction was estimated as the tangent to the tool exit hole directly upstream of the scan area, and is readily determined to correspond to

the observed displacements of the primary 0001 peak from the horizontal (weld) direction in the 0001 pole gures in Fig. 7. Using this orientation and the $35 eective angle of the inclined surface of the tool, as determined from Fig. 5, the pole gures were rotated into an approximate shear frame of reference. Then, using the similarities between the resultant orientations, and to the orientations reported by Knipling and Fonda [34], additional small rotations were applied to align the predominant 0001 peak at the pole gure center and a 11 20 peak along the shear direction, as shown in Fig. 8. The similarity between the HP and MP welding conditions is even more pronounced in this gure, and the small but systematic dierences to the CP welding condition become more apparent. While the observed pole gures shown in Fig. 8 resemble those reported by Knipling and Fonda [34], they also exhibit some important dierences. The general appearance of these pole gures indicates that both the bcc D1 shear texture (after transformation to hcp according to the Burgers OR) and the hcp P1 shear texture are present. However, the CP weld exhibits a weaker D1 texture than was observed by Knipling and Fonda, and the other welding conditions show even weaker contributions from that texture. The HP and MP welds that were prepared with a hotter welding condition (slower welding speed) exhibited more of the hcp P1 shear texture, seemingly indicating a colder temperature during shear deformation, while the CP weld prepared with a colder welding condition (faster welding speed) exhibited more of the bcc D1 shear texture, seemingly indicating a hotter temperature during shear deformation. It is important to note that it was dicult to fabricate defect-free welds using the CP weld parameters due to occasional lack-of-consolidation behind the tool. This indicates that, at faster weld speeds, the rotating tool may retreat from the deposited weld too quickly to maintain a good coupling between the tool and the weld. Thus, material deposited in the wake of the tool at faster weld speeds would be subject to less shear deformation during cooling than at slower weld speeds. As that material cools below the b transus, each bcc crystal orientation can form into 12 variants according to the Burgers OR. The amount of each variant formed can vary in response to such factors as applied stress, temperature gradient, crystallographic texture, and microstructure [43,44]. Thus, if the deposited weld material is under shear stress at the time of that transformation, it is likely to preferentially form the variant(s) that are best oriented to accommodate that shear. Since one of these variants corresponds to the hcp P1 shear orientation, that variant is expected to be the preferred orientation. However, poor coupling between the tool and the deposited weld at fast welding speeds would allow this transformation to occur with little intervariant selection due to the reduced shear stress, giving rise to the observed D1 shear texture. This also explains the larger grain size observed in the CP weld. The better coupling maintained between the tool and deposited weld at slower weld speeds would allow deformation of the grain structure to continue as

6458

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

Fig. 7. Pole gures obtained from large scan regions of the deposited HP, MP, and CP welds. Course contours show increments of one times random.

the weld cools, resulting in the ner grain sizes observed at slower weld speeds. Determining the texture components present in the observed pole gures (Fig. 8) is dicult due to both the similarity of many of the hcp and bcc shear components and the large number of potential variants resulting from the transformation of the bcc constituents to the low-temperature hcp phase. Orientation distribution functions (ODFs) are more eective than pole gures for discriminat-

ing between the various texture components. An ODF of the hcp shear components, shown in Fig. 9, is readily constructed from the Euler angles listed in Table 1 and the stable orientations along those bers shown in Fig. 3. The ODF of the bcc ideal shear components that would be generated in this alloy at the elevated welding temperatures are, however, complicated by the bcc-to-hcp transformation that occurs during cooling. For these orientations, it is necessary to consider not only the Euler angles of the

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

6459

Fig. 8. Pole gures obtained from large scan regions of the deposited welds, rotated into the shear frame of reference for the observed negative shear. Ideal orientations of the hcp P1 shear texture and the hcp Burgers variants of the bcc D1 shear texture are indicated. Course contours show increments of one times random.

bcc shear orientations as shown in Table 1, but also all 12 hcp Burgers variants associated with each bcc orientation. The ODFs of the expected orientations of the bcc ideal shear components after transformation to the low-temperature hcp phase according to the Burgers OR are shown in Fig. 10.

The experimental ODFs from each welding condition, shown in Fig. 11, can be compared to the ODFs of the ideal hcp shear texture (Fig. 9) and of the ideal bcc shear texture after transformation to hcp (Fig. 10) to determine the texture components present within the deposited welds. The most prominent feature in these ODFs is located every

6460

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

Fig. 9. Orientation distribution function plots showing the important ideal orientations associated with simple shear deformation of hcp materials.

u1 = 60 at U = 0 and 180, which could contain contributions from the P1, D1, D2, and/or F shear textures. It is not possible to isolate the contributions from the hcp P1 shear texture and the bcc D1 shear texture, since one of the hcp Burgers variants of the D1 texture (the D1(1) variant) has the same orientation as the P1 shear texture. This also indicates that the presence of an hcp P1 shear texture may be the result of preferential variant selection during the bccto-hcp transformation from a parent D1 shear texture due to the shear stress transferred from the rotating tool, rather than the shear deformation of that material below the b transus temperature. The presence of a bcc D1 shear texture in these welds is revealed through the ve interior peaks in the u2 = 30 ODF sections. These peaks have a moderate strength in the CP weld, but are weaker in the MP and HP welds, which is consistent with the pole gures (Figs. 7 and 8). Analysis of the dierent hcp variants resulting from transformation of the D1 texture is shown in Table 3. The variant populations are only approximate, since a 20 angular tolerance was used to discriminate among these orientations and some orientations dier only by 10.5. (Data

acquired with reduced angular tolerances are consistent with these values, but become less precise with the reduced sampling.) Table 3 demonstrates that the D1(1) variant (corresponding to the hcp P1 shear texture) is the dominant orientation for all welding conditions, and accounts for approximately 25% of all the variants in the HP and MP welds. Variants 2, 5, and 11 also exhibit larger populations than the other hcp variants of the D1 texture. The three weak peaks at U = 90 in the u2 = 0 ODF sections suggest the presence of a weak J/J texture of similar intensity in all three welds. This is supported by the elongation of the intense peaks at U = 0 and 180 towards the J/J texture orientations located at U = 35.3 and 144.7. There is also evidence of a very weak F texture component in the u2 = 30 ODF sections of each weld. The strong P1 and D1 textures observed in these near-a titanium friction stir welds are in agreement with much of the previous literature on FSW in bcc and hcp materials. The hcp P1 (1 0  1 0)[1 1  2 0] shear texture that was reported as the sole texture component in friction stir welds of a commercial purity a titanium alloy [35] is consistent with the strong P1 texture component observed in the present study.

Fig. 10. Orientation distribution function plots of the hcp Burgers variants resulting from transformation of the important ideal orientations associated with simple shear deformation of bcc materials.

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

6461

Fig. 11. Calculated orientation distribution function plots of the textures of the deposited HP, MP, and CP welds (symbols identied in Figs. 9 and 10).

As mentioned above, the textures produced during FSW of the metastable bcc b-21S titanium alloy [31] were later shown [34] to match the D1( 1 1 2)[1 1 1] shear texture (in Li et al.s [14] notation). Although Mironov et al. [30] observed a D2( 1 1 2)[1 1 1] shear texture during friction stir

processing of pure iron, this identication was based on comparisons between Li et als ideal orientations for negative simple shear [16] and Mironov et als results represented in positive shear. The texture observed by Mironov et al. thus corresponds to the same D1( 1 1 2)[1 1 1] shear texture

6462

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463

Table 3 Crystallographic orientation of the 12 hcp a variants derived from the bcc D1 shear texture through the Burgers OR, and their populations (in%) within the three welds. Note that the hcp D1(1) variant corresponds to the same orientation as the P1 shear component. U u2 HP MP CP hcp Variant {0 0 0 1} h1 1  2 0i || u1 D1(1) D1(2) D1(3) D1(4) D1(5) D1(6) D1(7) D1(8) D1(9) D1(10) D1(11) D1(12) {1  1 0} h1 1 1i {1  1 0} h1 1  1i {1 1 0} h1  1 1i {1 1 0} h1  1 1i {0  1 1} h1 1  1i {0  1 1} h1  1 1i {0  1 1} h 1 1 1i {0  1 1} h1  1 1i {1 0  1} h1 1 1i {1 0  1} h1  1 1i {1 0 1} h 1 1 1i {1 0 1} h1  1 1i 90 19.5 144.7 144.7 19.5 19.5 270 270 90 90 199.5 199.5 0 0 90 90 60 60 60 60 60 60 60 60 30 30 84.7 155.3 30 139.5 30 100.5 30 100.5 30 139.5 25 13 7 8 8 4 6 5 4 6 9 5 26 11 9 6 9 4 6 6 4 4 9 5 19 11 7 7 10 6 7 6 6 6 11 5

(in negative simple shear). In a later FSW study on Ti6Al 4V, Mironov et al. [33] reported that the very limited number of orientation measurements from the retained b could be interpreted as a bcc J(1 1 0)[1  1 2] ideal simple shear texture. If those results were also presented in the positive shear reference frame, then this texture corresponds just as well to the D1( 1 1 2)[1 1 1] shear texture as to the reported J texture. The D1 texture was also observed as the dominant texture in thick (12 mm, 0.5 in.) friction stir welds in Ti5111 by Knipling and Fonda [34]. 5. Conclusions The microstructures and crystallographic textures produced during friction stir welding of a near-a titanium alloy, Ti-5111, were studied for three weld travel speeds, corresponding to three weld heat inputs. The stir-zone microstructures were studied using electron back-scattered diraction (EBSD) and the crystallographic textures were correlated with those predicted from bcc and hcp simple shear textures. The observed textures were also compared with prior EBSD and equal channel angular extrusion (ECAE) studies on bcc and hcp materials. The main results are as follows: 1. Simple shear textures were calculated for hcp and bcc crystals (after transformation to the low-temperature hcp a phase according to the Burgers orientation relationship) for both pole gure (Figs. 2 and 3) and orientation distribution function (Figs. 9 and 10) representations. 2. The deposited welds consisted of small, equiaxed prior-b grains delineated by grain boundary a and containing a laths, indicating weld temperatures exceeded the b transus in all three welds. Similar grain structures were produced for all welding conditions, although the fastest welding speed resulted in a larger grain size than the slower weld speeds. 3. Pole gures of these welds revealed similar textures, dominated by the hcp P1 and bcc D1 shear textures.

The hcp P1 texture was stronger for the slower weld speeds (hotter welding conditions) while the bcc D1 texture was more pronounced for the faster welding speed (colder welding condition). Furthermore, the correspondence between the hcp P1 shear texture and one of the hcp variants of the bcc D1 shear texture suggests that the P1 texture may be the result of preferential variant selection from exposure to shear stress during the bccto-hcp transformation rather than the shear deformation of that material below the beta transus temperature. 4. At faster weld speeds, the larger grain size and predominance of the bcc D1 shear texture suggest that the rotating tool may advance too quickly to maintain a good coupling between it and the deposited weld. With good coupling, material behind the tool is subjected to continued shear as it cools, whereas poor coupling results in a decreased transmission of shear deformation into the deposited weld. The occasional lack-of-consolidation defects observed at the fastest welding speed further supports this hypothesis of poor coupling at fast welding speeds. 5. Orientation distribution functions revealed the presence of weak contributions from the J/J and F textures in all welding conditions. 6. Prior studies of the textures produced during friction stir welding of bcc and hcp materials can all be interpreted as being consistent with this study, with the hcp P1 and bcc D1 shear textures dominating and their relative strengths varying as a function of alloy and welding condition.

Acknowledgements The authors would like to acknowledge funding for this research through the Oce of Naval Research, as well as the numerous helpful discussions with Dr. Saiyi Li and Dr. David Rowenhorst. References
[1] Thomas WM, Nicholas ED, Needham JC, Murch MG, TempleSmith P, Dawes CJ. GB Patent Application No. 9125,978.8, 1991: US Patent No. 5460,317, 1995. [2] Mishra RS, Ma ZY. Friction stir welding and processing. Mater Sci Eng R 2005;50:178. [3] Mishra RS. Preface to the viewpoint set on friction stir processing. Scripta Mater 2008;58:3256. [4] McNelley TR, Swaminathan S, Su JQ. Recrystallization mechanisms during friction stir welding/processing of aluminum alloys. Scripta Mater 2008;58:34954. [5] Backofen WA. The torsion texture of copper. Trans AIME 1950;188:14549. [6] Williams RO. Shear textures in copper, brass, aluminum, iron, and zirconium. Trans Metall Soc AIME 1962;224:12940. [7] Canova GR, Kocks UF, Jonas JJ. Theory of torsion texture development. Acta Metall 1984;32:21126. [8] Montheillet F, Cohen M, Jonas JJ. Axial stresses and texture development during the torsion testing of Al, Cu and a-Fe. Acta Metall 1984;32:207789.

R.W. Fonda, K.E. Knipling / Acta Materialia 58 (2010) 64526463 [9] Toth LS, Neale KW, Jonas JJ. Stress response and persistence characteristics of the ideal orientations of shear textures. Acta Metall 1989;37:2197210. [10] Barnett MR, Montheillet F. The generation of new high-angle boundaries in aluminium during hot torsion. Acta Mater 2002;50:228596. [11] Pettersen T, Nes E. On the origin of strain softening during deformation of aluminum in torsion to large strains. Metall Mater Trans A 2003;34A:272736. [12] Backofen WA, Hundy BB. Torsion texture of 70-30 brass and Armco iron. Trans AIME 1953;197:612. [13] Baczynski J, Jonas JJ. Texture development during the torsion testing of a-iron and two IF steels. Acta Mater 1996;44:427388. [14] Li S, Beyerlein IJ, Bourke MAM. Texture formation during equal channel angular extrusion of fcc and bcc materials: comparison with simple shear. Mater Sci Eng A 2005;394:6677. [15] Beausir B, Toth LS, Neale KW. Ideal orientations and persistence characteristics of hexagonal close packed crystals in simple shear. Acta Mater 2007;55:2695705. [16] Li S. Orientation stability in equal channel angular extrusion. Part II: Hexagonal close-packed materials. Acta Mater 2008;56:103143. [17] Jata KV, Semiatin SL. Continuous dynamic recrystallization during friction stir welding of high strength aluminum alloys. Scripta Mater 2000;43:7439. [18] Sato YS, Kokawa H, Ikeda K, Enomoto M, et al. Microtexture in the friction-stir weld of an aluminum alloy. Metall Mater Trans A 2001;32:9418. [19] Fonda RW, Bingert JF, Colligan KJ. Development of grain structure during friction stir welding. Scripta Mater 2004;51:2438. [20] Schneider JA, Nunes AC. Characterization of plastic ow and resulting microtextures in a friction stir weld. Metall Mater Trans B 2004;35:77783. [21] Kang SH, Bang WH, Cho JH, Han HN, et al. Microtexture analysis of friction stir welded Al 6061-T651 plates. Mater Sci Forum 2005;495497:9016. [22] Prangnell PB, Heason CP. Grain structure formation during friction stir welding observed by the stop action technique. Acta Mater 2005;53:317992. [23] Fonda RW, Bingert JF. Precipitation and grain renement in a 2195 Al friction stir weld. Metall Mater Trans A 2006;37A:3593604. [24] Fonda RW, Wert JA, Reynolds AP, Tang W. Friction stir welding of single crystal aluminium. Sci Technol Weld Join 2007;12:30410. [25] Fonda RW, Bingert JF. Texture variations in an aluminum friction stir weld. Scripta Mater 2007;57:10525. [26] Ahmed MMZ, Wynne BP, Rainforth WM, Threadgill PL. Quantifying crystallographic texture in the probe-dominated region of thick-

6463

[27]

[28] [29]

[30] [31] [32]

[33]

[34] [35]

[36] [37]

[38] [39] [40] [41] [42] [43] [44]

section friction-stir-welded aluminium. Scripta Mater 2008;59:50710. Shibayanagi T, Gerlich AP, Kashihara K, North TH. Textures in single-crystal aluminum friction stir spot welds. Metall Mater Trans A 2009;40A:92031. Sato YS, Nelson TW, Sterling CJ. Recrystallization in type 304L stainless steel during friction stirring. Acta Mater 2005;53:63745. Cho JH, Boyce DE, Dawson PR. Modeling strain hardening and texture evolution in friction stir welding of stainless steel. Mater Sci Eng A 2005;398:14663. Mironov S, Sato YS, Kokawa H. Microstructural evolution during friction stir-processing of pure iron. Acta Mater 2008;56:260214. Reynolds AP, Hood E, Tang W. Texture in friction stir welds of Ti metal 21S. Scripta Mater 2005;52:4914. Lee WB, Lee CY, Chang WS, Yeon YM, Jung SB. Microstructural investigation of friction stir welded pure titanium. Mater Lett 2005;59:33158. Mironov S, Zhang Y, Sato YS, Kokawa H. Development of grain structure in b-phase eld during friction stir welding of Ti6Al4V alloy. Scripta Mater 2008;59:2730. Knipling KE, Fonda RW. Texture development in the stir zone of near-a titanium friction stir welds. Scripta Mater 2009;60:1097100. Mironov S, Sato YS, Kokawa H. Development of grain structure during friction stir welding of pure titanium. Acta Mater 2009;57:451928. Rollett AD, Wright SI. In: Kocks UF, Tome SN, Wenk HR, editors. Texture and anisotropy. Cambridge University Press; 1998. p. 202. Burgers WG. On the process of transition of the cubic-body-centered modication into the hexagonal-close-packed modication of zirconium. Physica 1934;1:56186. Been J, Faller K. Using Ti-5111 for marine fastener applications. JOM 1999;51:214. Colligan K. Material ow behavior during friction stir welding of aluminum. Weld J 1999;78:229S37S. Beyerlein IJ, Toth LS. Texture evolution in equal-channel angular extrusion. Prog Mater Sci 2009;54:427510. Timetal 5111 Product Data Sheet, TMC-170. Titanium Metals Corporation, Denver, CO; 2000. Ramirez AJ, Juhas MC. Microstructural evolution in Ti6Al4V friction stir welds. Mater Sci Forum 2003;426432:29993004. Ray RK, Jonas JJ. Transformation textures in steels. Int Met Rev 1990;35:136. Humbert M, Gey N, Germain L. Study and modelling of some variant selections in BCC to HCP phase transformations. Mater Sci Forum 2005;495497:111120.