Friction-Stir Welding Effects on Microstructure and Fatigue of Aluminum Alloy 7050-T7451 K.V. JATA, K.K. SANKARAN, and J.J. RUSCHAU Aluminum alloy 7050 was friction-stir welded (FSW) in a T7451 temper to investigate the effects on the microstructure and mechanical properties. Results are discussed for the as-welded condition (as-FSW) and for a postweld heat-treated condition consisting of 121 8C for 24 hours (as-FSW 1 T6). Optical microscopy and transmission electron microscopy (TEM) examination of the weld-nugget region show that the FS welding process transforms the initial millimeter-sized pancake-shaped grains in the parent material to fine 1 to 5 mm dynamically recrystallized grains; also, the FS welding process redissolves the strengthening precipitates in the weld-nugget region. In the heat-affected zone (HAZ), the initial grain size is retained, while the size of the strengthening precipitates and of the precipitatefree zone (PFZ) is coarsened by a factor of 5. Tensile specimens tested transverse to the weld show that there is a 25 to 30 pct reduction in the strength level, a 60 pct reduction in the elongation in the as-FSW condition, and that the fracture path is in the HAZ. The postweld heat treatment of 121 8C for 24 hours did not result in an improvement either in the strength or the ductility of the welded material. Comparison of fatigue-crack growth rates (FCGRs) between the parent T7451 material and the as-FSW 1 T6 condition, at a stress ratio of R 5 0.33, shows that the FCG resistance of the weldnugget region is decreased, while the FCG resistance of the HAZ is increased. Differences in FCGRs, however, are substantially reduced at a stress ratio of R 5 0.70. Analysis of residual stresses, fatiguecrack closure, and fatigue fracture surfaces suggests that decrease in fatigue crack growth resistance in the weld-nugget region is due to an intergranular failure mechanism; in the HAZ region, residual stresses are more dominant than the microstructure improving the fatigue crack growth resistance. I. INTRODUCTION COMPARED to many of the fusion-welding processes that are routinely used for joining structural alloys, frictionstir (FS) welding is an emerging solid-state joining process[1–12] in which the material that is being welded does not melt and recast. Therefore, when alloys are friction-stir welded (FSW), phase transformations that occur during the cool down of the weld are of a solid-state type. Due to the absence of parent-metal melting, the new FS welding process is observed to offer several advantages over fusion welding. The benefits that stand out most are welding of difficult-toweld aluminum alloys such as the 7xxx series, better retention of baseline material properties, fewer weld defects, low residual stresses, and better dimensional stability of the welded structure. Also, FS welding is an environmentally cleaner process, due to the absence of a need for the various gases that normally accompany fusion welding. A schematic of the FS welding assembly is shown in Figure 1. The FS welding process uses a nonconsumable pin made from a high-strength material that extends from a cylindrical shoulder. The shoulder and the pin rotate at several hundred revolutions per minute. The work pieces that are to be joined are firmly clamped to the work table, and the pin is plunged into the work pieces where the weld bond K.V. JATA, Senior Research Materials Scientist, is with the Materials and Manufacturing Directorate, Air Force Research Laboratory, AFRL/ MLLM, Wright-Patterson Air Force Base, OH 45433. K.K. SANKARAN, Technical Fellow, is with the Boeing Company, St. Louis, MO 63166. J.J. RUSCHAU, Research Engineer, is with the Materials Engineering Division, University of Dayton Research Institute, Dayton, OH 45469-0136. Manuscript submitted September 22, 1999. METALLURGICAL AND MATERIALS TRANSACTIONS A line is desired. The height of the pin is slightly smaller than the thickness of the alloy plates that are being joined, so the penetration of the pin into the work pieces stops as soon as the shoulder of the cylinder makes contact with the surface of the work piece. The rotating pin (extending from the cylindrical shoulder) produces the stirring action in the material along the bond line and produces the required thermomechanical deformation. Frictional heating is produced from the interaction of the cylinder shoulder with the work piece and the downward applied forging pressure. To produce a longitudinal weld, the work piece assembly is translated relative to the shoulder and pin assembly. To produce an ideal defect-free weld, the revolutions per minute of the cylinder shoulder-pin assembly, travel speed, downward forging force, and pin tool design have to be optimized. Although the development of FS welding technology to make complex welds is proceeding at an extremely rapid pace, primarily due to the efforts of the industry, understanding of the microstructural transformations that occur during the welding process and of the postweld mechanical properties[5–13] has been slow. There has also been significant activity in addressing the much needed modeling research, in order to understand the weld- and parent-metal constitutive behavior,[14,15] heat transfer, and thermomechanical analysis of FS welding.[9,16] The present research is aimed at understanding the microstructural and mechanical-property changes brought about by FS welding of a high-strength Al-Zn-Mg-Cu alloy, 7050T7451. T7451 is an overaged temper, specially developed by the aluminum industry to optimize corrosion resistance and mechanical properties, and is widely used by aircraft manufacturers. U.S. GOVERNMENT WORK NOT PROTECTED BY U.S. COPYRIGHT VOLUME 31A, SEPTEMBER 2000—2181 Fig. 1—Schematic illustration of the FS welding assembly. Table I. Chemical Composition of the 7050-T7451 Alloy in Weight Percent Cu Mg Zn Zr Fe Si 2.23 2.25 6.2 0.12 ,0.12 ,0.15 Fig. 2—ESE(T) specimen used in the FCGR tests. All dimensions are in millimeters. II. EXPERIMENTAL PROCEDURES The nominal chemical composition of the alloy 7050T7451 is shown in Table I. Several 6.35-mm-thick alloy 7050-T7451 plates were butt-welded using the new FS welding process. The tool rotation speed was 6.6 revolutions s21, and the tool traverse speed was 1.7 mm s21. Material was either tested in the as-FSW condition or in the postweld heattreated condition, which consisted of heating the material in an air furnace at 121 8C for 24 hours. The heat-treated condition is designated “as-FSW 1 T6.” Kellers etch was used for revealing the optical microstructure of the weld zone. Optical microscopy was used to study the grain-size and grain-flow variation in the entire weld zone. After optically identifying the weld-nugget region, the thermomechanically affected zone (TMAZ), and the heat-affected zone (HAZ), thin foils were prepared for transmission electron microscopy (TEM) observations to study the precipitate microstructure from these regions in the as-FSW and the as-FSW 1 T6 conditions. Hardness measurements were performed to study the variation in hardness with distance from the centerline of the weld. Tensile properties were evaluated in a transverse direction to the weld, with the specimen test section comprising the weld and HAZ regions. The gage length and the thickness of the tensile specimens were 51 and 6.35 mm, respectively. The fatigue-crack growth rate (FCGR) was evaluated only in the as-FSW 1 T6 treated material to understand the effects of the postweld heat treatment. In evaluating the FCGR behavior, eccentrically loaded single-edge-tension (ESE(T)) specimens (previously known as extended compact-tension specimens), per ASTM E647-99 (Figure 2), were prepared. The ESE(T) specimen design is well suited for evaluating the fatigue growth rates in the weld, as the specimen is less prone to out-of-plane cracking than the compact-tension 2182—VOLUME 31A, SEPTEMBER 2000 Fig. 3—Schematic illustration showing the specimen layout to evaluate FCGR of the weld nugget and the HAZ. specimen. Specimens were tested to obtain near-threshold FCGRs in the weld nugget and in the HAZ. The FCGRs in the weld nugget were obtained by machining the initial notch along the centerline of the weld. For HAZ evaluation, notches were machined 7 mm away from the centerline of weld. This ensured the starting notch to be in the HAZ region and that the subsequent FCGR test indeed evaluated the HAZ region of the weld. A diagram showing the layout of the specimens from the FSW plate is shown in Figure 3. Specimens were tested for the FCG threshold at room temperature in laboratory air (33 pct relative humidity) at two load-ratio (R) levels of 0.3 and 0.7. The test frequency employed for all tests was 30 Hz. The stress-intensity (K ) geometrical correction factor (FESE(T)) equations[17] used are shown subsequently. K 5 (P/BW1/2) FESE(T) FESE(T) 5 a1/2 (1.4 1 a) (1 2 a)23.2 G METALLURGICAL AND MATERIALS TRANSACTIONS A Fig. 4—(a) Optical macrophotograph showing the weld nugget and the TMAZ in the FSW weld of 7050-T7451. Higher magnification optical micrographs: (b) fine grain size in the weld nugget and (c) bent grains in the TMAZ. G 5 3.97 2 10.88a 1 26.25a 2 2 38.9a 3 1 30.15a 4 2 9.27a 5 a 5 a/W Crack lengths were measured using the compliance technique, with a crack-mouth-opening gage at the front face. The expression used is as follows:[17] V0 BE/P 5 (15.52a 2 26.38a 2 1 49.7a 3 2 40.74a 4 1 14.44a 5)/(1 2 a)2 where B and W are the thickness and the width of the ESE(T) specimen, respectively; P is the load applied; E is the elastic modulus of the material; V0 is the crack-mouth-opening displacement at the front face; and a is the crack length. III. RESULTS A. Microstructural Observations A portion of the weld in the thickness section of the welded plate (containing the long- and short-transverse directions of the 7050 plate) is shown in the low-magnification optical photograph (Figure 4(a)). Two regions that are distinctly evident in this micrograph are the weld-nugget zone (dynamically recrystallized zone) and the TMAZ. The weld-nugget METALLURGICAL AND MATERIALS TRANSACTIONS A zone and the TMAZ zone are further shown at high magnifications in the optical micrographs in Figures 4(b) and (c), respectively. Figures 4(a) through (c) clearly show that FS welding converts the initial flat grains in the parent metal to fine equiaxed grains in the dynamically recrystallized zone. The width of the weld-nugget zone is almost equal to the diameter of the FS welding pin, which in this case is 7.6 mm. The bending of the grains in the TMAZ region suggests that the stirring action of the FS welding tool causes the flat grains of the parent metal to be drawn into the weldnugget zone. The microstructure in the parent alloy is partially recrystallized. The unrecrystallized regions contain 1 mm pancakeshaped grains with 1 to 10 mm subgrains within these grains. The TEM micrograph in Figure 5(a) shows the subgrains in the unrecrystallized regions of the parent alloy. Figure 5(b) is a high-magnification bright-field TEM micrograph that shows the strengthening precipitates in the grain interior and grain-boundary precipitates along the boundaries. Since the parent alloy was in a commercial temper, no specific attempt was made to discern the chemistry and the structure of the precipitates. The strengthening precipitates in the T7451 temper are well established to be h8 Mg (Zn,Cu,Al)2 and the high-angle grain-boundary precipitates to be h MgZn2 and/or Mg3Zn3Al2. A bright-field TEM micrograph (Figure 6(a)) shows the microstructure of the weld-nugget VOLUME 31A, SEPTEMBER 2000—2183 (b) (a) Fig. 5—Bright-field transmission electron micrographs: (a) subgrain structure and (b) precipitate microstructure in the grain interior and along grain boundaries in the parent material, 7050-T7451. (a) (b) (c) Fig. 6—(a) Low-magnification bright-field transmission electron micrograph showing subgrain structure, (b) dislocations pinned by second-phase particles, and (c) high-magnification bright-field transmission electron micrograph indicating absence of strengthening phases and presence of second-phase particles. All pictures correspond to the as-FSW condition of the alloy 7050-T7451. region at a low magnification. Compared to the parentmaterial microstructure that contains nonequiaxed subgrains, fine equiaxed grains ranging from 1 to 5 mm are observed in the entire weld-nugget region. These grains are recrystallized, and the grain boundaries are of the high-angle type. The TEM investigation of the weld-nugget region showed some grains to contain a high dislocation density (Figure 6(b)), with the dislocations pinned by the Al3Zr dispersoids. Some of the dislocations are also pinned by Al7Cu2Fe inclusions and second-phase particles. The small-size particles seen in the micrograph (Figure 6(b)) arise due to the breaking up of the large-size particles from the stirring action of the 2184—VOLUME 31A, SEPTEMBER 2000 FS welding tool. As shown in Figure 6(c), the strengthening phases are not present in the weld-nugget region, suggesting that the temperature in the weld nugget during FS welding reached the solution heat-treatment temperature. Unlike the strengthening precipitates, which have a lower dissolution temperature, the dispersoids have a much higher dissolution temperature. The presence of the dispersoids indicates that the temperature in the weld nugget did not reach the dispersoid dissolution temperature. Figure 7(a) is a TEM micrograph of the HAZ of the asFSW condition. This micrograph shows that the FS welding process has relatively little effect on the size of the subgrain METALLURGICAL AND MATERIALS TRANSACTIONS A heat treating the as-FSW alloy for 24 hours at 120 8C. Figure 8(a) shows that the T6 heat treatment does not result in a change in the subgrain size. Very fine precipitation, most likely Guinier–Preston (GP) zones, is observed to occur in the weld-nugget region during the postweld T6 heat treatment (Figure 8(b)). However, the T6 heat treatment did not produce an appreciable change either in the strengthening precipitate size or the grain-boundary precipitate size in the HAZ (Figure 8(c)). Also, as seen in the TEM micrograph of Figure 8(c), there is no significant widening of the PFZ in the HAZ. B. Hardness and Tensile Results (a) (b) Fig. 7—Bright-field transmission electron micrographs: (a) subgrain structure and (b) coarsened precipitates in the grain interior and along the grain boundaries and PFZ in the HAZ of the weld of as-FSW 7050-T7451. in the HAZ. Higher-magnification TEM micrographs show that, in contrast to the weld-nugget region, the strengthening phases are still present in the HAZ region. A representative micrograph (Figure 7(b)) shows the precipitates in the grain interior and along a grain boundary in the HAZ. A comparison of the HAZ microstructure in Figure 7(b) to the parentmaterial microstructure (Figure 5(b)) shows that the strengthening phases and the precipitate-free zone (PFZ) are coarsened by a factor of 5. Changes in the microstructure of the as-FSW1T6 condition are shown in Figure 8. This condition corresponds to METALLURGICAL AND MATERIALS TRANSACTIONS A Hardness data between the top and back sides of the aswelded plate are shown in Figure 9(a). On the root side, the entire weld zone (nugget 1 TMAZ 1 HAZ) extends about 25 mm on either side of the center of the weld, with the minimum in the hardness occurring at about 12 mm on either side of the weld center. Clearly, there is a difference in the hardness between the top and back sides of the weld. The top side consistently shows lower hardness numbers. This is due to the fact that this is the side of the plate that is in full contact with the FS welding tool shoulder and, thus, experiences direct heat from the rapidly rotating tool shoulder. The back side, on the other hand, is in direct contact with a back plate that acts as a heat sink and rapidly draws away heat. Such differences in the hardness profiles between and top and back sides of a FS welding plate have also been observed for a FSW Al-Li-Cu alloy.[18] Despite the relatively small thickness of the 7050-T7451 plate, the effects of the heating and thermal transients during FS welding are not uniform through the thickness. These differences can be attributed to the through-thickness variation in the extent of precipitate dissolution in the nugget region and also the subsequent postweld room-temperature aging. The latter is the room-temperature aging process that occurs in alloy 7050 between the completion of the welding and the time at which hardness measurements are performed. The hardness of the nugget region in the as-welded condition results from contributions of the fine grain size, natural aging, and the dislocation density. A significant decrease in hardness is observed in the HAZ. This is due to the coarse precipitates in the HAZ. As discussed before, precipitates in the HAZ are about 5 times coarser than in the parent material. The effect of the postweld heat treatment on the hardness is shown in Figures 9(b) and (c). As expected, aging of the FSW alloy does not alter the grain structure in the nugget region. Aging to the T6 treatment reduces the dislocation density in the nugget region and results in fine precipitates. This fine precipitation more than compensates for any loss in hardness due to the reduction of dislocation density and raises the hardness of the nugget region. However, the hardness and microstructure are not truly representative of the T6 condition, because of the lack of solution heat treatment and quenching. Tensile properties of the baseline 7050-T7451 material, as-FSW, as-FSW1T6, and an additional heat-treated condition termed as-FSW1T7 (aging treatment consisting of 121 8C for 8 hours 1 175 8C for 8 hours) are shown in Table II. The additional heat treatment is currently being investigated in more detail. Friction-stir welding significantly decreases the overall ductility of this alloy. Also, as shown VOLUME 31A, SEPTEMBER 2000—2185 (a) (b) (c) Fig. 8—Bright-field transmission electron micrographs of “as-FSW1T6” condition, of alloy 7050-T7451: (a) subgrain structure, (b) fine precipitates in the weld nugget region, and (c) precipitates in the grain interior and along grain boundaries in the HAZ. “as-FSW1T6” corresponds to friction stir welding followed by a postweld heat treatment of 120 8C for 24 h. There is no solution heat treatment of the weld involved. in Table II, simple heat treatments of the weld do not restore the ductility of the as-FSW condition. Such a reduction in ductility or percent elongation has been observed for other alloy systems as well and is attributed to strain localization in the soft regions of the weld.[18] C. Fatigue-Crack Growth The FCGRs in the near-threshold region were obtained for the microstructurally stabilized weld condition, i.e., asFSW1T6 condition, and compared to the FCGRs in the 2186—VOLUME 31A, SEPTEMBER 2000 parent material, alloy 7050-T7451. For the as-FSW1T6 condition, the FCG resistance of the weld was evaluated for two different locations in the weld. In one case, the starter notch was placed along the weld centerline in the nugget region; in the second case, the starter notch was placed 7 mm away from the weld centerline and in the HAZ. Results for both these conditions are shown in Figure 10 and compared to the FCGR for the parent (baseline) material. In the weld-nugget region, the near-threshold FCG resistance along the weld centerline decreases in comparison to the parent material for the stress ratio of 0.33; the near-fatigue threshold METALLURGICAL AND MATERIALS TRANSACTIONS A (b) (a) (c) Fig. 9—Comparison of hardness profiles between (a) weld and root side of the as-FSW plate, (b) weld side of as-FSW and as-FSW1T6, and (c) root side of the as-FSW and as-FSW1T6. “as-FSW1T6” corresponds to a postweld heat treatment of 24 h at 120 8C of the weld without a solution heat treatment step. Table II. Tensile Properties of the 7050-T7451 Alloy before and after FS Welding Condition Parent material as-FSW as-FSW 1 T6 as-FSW 1 T7 YS (MPa) UTS (MPa) Pct Elongation 489 304 291 287 555 429 417 371 16.7 6 3.8 2.4 decreases from 1.83 MPa!m in the parent material to 1.04 MPa!m. In the HAZ, however, a significant increase in the near-threshold FCG resistance is observed. At R 5 0.33, the fatigue threshold increases from 1.83 to 3.24 MPa!m. At the higher stress ratio of R 5 0.7, differences in the FCGRs of the baseline material, weld centerline, and HAZ are almost negligible. Fatigue-crack closure was monitored in all the experiments throughout the FCG test, using the compliance METALLURGICAL AND MATERIALS TRANSACTIONS A Fig. 10—Comparison of FCGRs between the weld nugget and HAZ, at R 5 0.33 and R 5 0.7. FCGRs were evaluated in the as-FSW1T6 condition. Crack growth rates are also compared to that in the parent material, 7050T7451. Tests were conducted in laboratory air. VOLUME 31A, SEPTEMBER 2000—2187 (a) (b) (c) Fig. 11—Load-crack opening displacement traces near the FCG threshold region: (a) for the parent material 7050-T7451, and (b) and (c) crack propagating in the weld-nugget region and the and HAZ region, respectively. Alloy condition for (b) and (c) is “as-FSW1T6.” method. If crack closure is present, the force-displacement trace develops a nonlinearity. The force–crack opening displacement traces in the near-threshold regime are shown in Figures 11(a) through (c). Both the parent-material specimen and the specimen with the crack propagating along the weld centerline show linear traces suggesting the absence of crack closure near the fatigue threshold, for a stress ratio of R 5 0.33. In comparison, crack closure is very predominant and significant in the specimen in which the crack propagates in the HAZ. The force at which nonlinearity occurs in the force-displacement trace is designated as Popen in Figure 11(c). This implies that the crack remains closed until the applied force increases from Pmin to Popen. Thus, the crack is effectively driven by the crack driving force, equal to 2188—VOLUME 31A, SEPTEMBER 2000 Pmax 2 Popen, instead of the entire applied force (Pmax 2 Pmin). Therefore, the effective stress-intensity range driving the crack at R 5 0.33 in the HAZ specimen is less than the applied stress-intensity range, resulting in an apparent fatigue threshold. At the higher R value of 0.7, crack closure is not observed, and the FCGRs and fatigue thresholds for the three conditions (parent material, weld centerline, and HAZ) are similar. IV. DISCUSSION The alloy was FSW in a single pass using parameters very similar to the ones used for most aluminum alloys, i.e., the tool was rotating 300 to 500 rpm at a 100 to 120 mm/ METALLURGICAL AND MATERIALS TRANSACTIONS A min traverse speed. As shown in the optical micrographs of the weld section (Figure 4), there are three distinct regions: the weld nugget, TMAZ, and HAZ. The two regions that have been paid most attention to in this work are the weldnugget region and the HAZ. The weld-nugget region, as shown in the TEM micrographs, clearly contains fine dynamically recrystallized grains with the absence of strengthening precipitates. The heating and thermal transients experienced in this region are high enough to dissolve the strengthening phases. Arbegast and Hartley[3] measured the temperature rise by placing thermocouples ahead of the rotating pin and showed that the temperatures can reach 410 8C to 450 8C in high-strength aluminum alloys such as Al-Li alloys. Measurements of temperature right underneath the rotating pin tool have not yet been made, but it is assumed that the temperature right underneath the pin would even be slightly higher. The formation of fine recrystallized grains in the weldnugget region suggests that the strain rate, temperature, and work input into the volume underneath the pin are sufficient enough to cause dynamic recrystallization. Strain measurements in the weld-nugget region during FS welding are still underway.[14] However, one could gain insight into the recrystallization phenomenon by considering the mechanisms proposed in the literature for the dependence of recrystallized grain size on the stress imposed during thermomechanical work. For example, Derby[19] has discussed the dependence of dynamically recrystallized grain size on the stress imposed during thermomechanical work for several metallic and ceramic materials. Both mechanisms by which dynamic recrystallization can occur were included, i.e., (1) a gradual increase in the misorientation angle between the walls of a stable dislocation-cell structure until they become high-angle grain boundaries, and (2) migration of preexisting high-angle grain boundaries. Derby developed simple expressions for predicting the stable recrystallized grain size as a function of applied stress and shear modulus. The expression, shown subsequently, is for the mechanism where a balance between the rate of formation of the dislocation substructure and mean velocity for recrystallized grain boundaries is considered to occur. 1 , s/G (DR /b)0.66 , 10 Here, DR is the recrystallized grain size and b is the Burgers vector. The terms G and s are the shear modulus and imposed stress, respectively. Thermal measurements made by several investigators during FS welding have shown that the temperatures can reach 410 8C to 450 8C, depending on process parameters such as the number of revolutions per minute and travel speed. At these temperatures, the flow stress and the shear modulus for the 7050 alloy are in the range of 52 MPa and 21.5 GPa, respectively, for a strain rate of 1 s21. (It is to be noted that the strain rate of 1 s21 has been chosen arbitrarily). Using a Burgers vector of 0.25 nm, Derby’s expression predicts a grain size of 2.1 mm as a lower-bound value. The experimentally observed grain size from Figure 3(b) is in the range from 1 to 1.5 mm. Recently, Jata and Semiatin[20] showed that the fine grains in the weld-nugget region form by a continuous recrystallization process. They measured the high-angle grain-boundary METALLURGICAL AND MATERIALS TRANSACTIONS A misorientation in the weld-nugget region (or, the dynamically recrystallized region) and showed that the original lowangle grain boundaries in the alloy are replaced by highangle boundaries by a continuous rotation of the original low-angle boundaries during the FS welding process. In contrast to the partially recrystallized microstucture of the mill-processed parent alloy, the weld-nugget region possesses a fine-grained, dynamically recrystallized microstructure with the absence of strengthening precipitates. Thus, the weld nugget could be compared to a 7050 alloy in a near-“W” temper, where W represents a solution heat treatment followed by natural aging at room temperature. Studies by Staley et al.[21] indicate that when a 7050 alloy is left in a W temper, the weld nugget will continue to age indefinitely at room temperature, which will deteriorate the corrosion resistance of the alloy. Thus, in this work, a heat treatment consisting of 24 hours at 121 8C (referred to here as as-FSW1T6) was used for microstructure stabilization. It should be noted that a separate solution heat-treatment step was not used prior to the 121 8C heat treatment, as this would be a difficult and an impractical step to conduct on a structural component in commercial practice. The heat treatment employed here targeted two purposes; one was to stabilize the microstructure and the second was to provide an opportunity to examine the mechanical properties through a simple heat treatment. A number of factors will influence FCGRs and fatigue thresholds in aluminum alloys, and these have been well discussed in the literature. For Al alloys, precipitate size and coherency have been found to be important factors. In underaged alloys where strengthening precipitates are coherent (the weld nugget in the as-FSW1T6 condition has fine GP zones, as in Figure 8(b)), slip reversibility and crack deflection have been suggested as major mechanisms that give rise to higher fatigue thresholds compared to overaged alloys. In underaged alloys, slip reversibility during fatigue cycling is higher because of the ability of dislocations to shear the coherent particles. The shearing action lets the cyclic plasticity be more reversible, and the damage accumulated is lower than in the case for the overaged alloys. The higher slip reversibility is invariably accompanied by a more tortuous crack path (or highly deflected crack), resulting in a higher stress needed to drive the crack. In an overaged Al alloy (such as the HAZ condition here), it is generally observed that the slip reversibility is low. The large incoherent precipitates loop the dislocations and do not allow the dislocations to reverse back to the crack tip, resulting in a larger accumulation of the reverse plasticity. Crack closure has also been found to be high for the underaged alloys compared to the overaged alloys. The results obtained here show that the weld nugget, in spite of having coherent precipitates, has a lower fatigue threshold, faster growth rates, and a lower crack closure than the HAZ alloy. Therefore, to understand the results obtained here, factors other than microstructure also need to be considered. Although it is generally agreed that FS welding produces very little residual stresses, they were measured to consider their effect on FCG for a proper interpretation of the FCGR behavior in the weld zone. In the present work, residual stresses were measured on the ESET FCG specimen after machining the starter notch (Figure 12(a)). This was considered more appropriate than measuring the residual stresses VOLUME 31A, SEPTEMBER 2000—2189 (a) (b) (c) Fig. 12—(a) Residual stresses were measured on the ESE(T) fatigue crack growth sample with a notch. Schematic illustration shows nomenclature for residual stresses and the points along which longitudinal and transverse components were measured; (b) longitudinal residual stress component parallel to the notch plane; and (c) transverse residual stress component perpendicular to the notch plane. 2190—VOLUME 31A, SEPTEMBER 2000 on the welded plate, since machining can partially or entirely relieve residual stresses. Figures 12(b) and (c) show the respective longitudinal and transverse residual-stress component as a function of distance from the center of weld in the ESET FCG specimen with a notch. Also, residual stresses were measured on the top side (weld side) as well as the back side (root side) of the weld plate and are shown in Figure 12. The transverse component is defined as the one that is perpendicular to the fracture plane, while the longitudinal stress component is defined as the one that is along the original notch plane. The data obtained here suggest that the transverse residual stress is negligible. However, the data show that the longitudinal stress component is compressive for both situations, i.e., when the crack is propagating along the weld centerline or the HAZ. Both residual-stress components can influence crack growth resistance.[22,23] The longitudinal component can produce either a clamping or opening moment, depending on whether the stress is compressive or tensile, and will influence the crack-driving stress-intensity range.[22] Thus, a longitudinal compressive-stress component should be beneficial to FCGRs by lowering the appliedstress-intensity range for both the weld centerline and HAZ specimens. The results obtained here suggest that the crack closure observed in the HAZ specimen at R 5 0.33 may indeed be due to the compressive residual-stress effect. The lower CGRs observed in the HAZ specimen at R 5 0.33 may be attributed to the compressive longitudinal residual stresses and the associated crack closure. In the case where the crack is propagating along the weld centerline, crack growth rates are slightly faster than the baseline at R 5 0.33, with no associated crack closure. The compressive residual stresses that are present in the specimen (which do not give have a beneficial effect on crack growth rates, as observed for the HAZ specimen) are being overridden by some other factor and result in faster growth rates. Figure 13 shows the fractography of the fractured FCGR specimens at near-threshold. In both the parent material and in the HAZ, the FCG mechanism remains similar, as shown in scanning electron micrographs (Figures 13(a) and (b)). Both fracture surfaces exhibit predominantly transgranular failure. Transgranular crack growth in the unrecrystallized regions appear as bands in the fracture plane, and crack growth in the recrystallized regions appears in the form of smaller facets. In the HAZ sample, the fracture facet size is much wider compared to the baseline. The fracture facet size for the HAZ sample is much wider, because the plastic2 ) for the same stress-intensity range is zone size (Rp a K 2/s YS larger for the HAZ than for the baseline material. Therefore, fatigue failure occurs in a much wider plastic-zone size which includes failure of several grains, for the case of the HAZ. The failure of several grains during fatigue-crack extension gives rise to a wider fracture facet size. As discussed before, both the parent material and the HAZ have a similar grain size, but the precipitate size for the HAZ material is approximately 5 times coarser. So, the HAZ material is in a much more overaged state than the parent material. The hardness, yield strength, and ductility of the HAZ are also lower than the parent material. The HAZ fracture-surface features are shown for R 5 0.33. This is because, for R 5 0.7, the facets are even bigger than for METALLURGICAL AND MATERIALS TRANSACTIONS A (a) (b) (c) (d ) Fig. 13—SEM fractographs showing fracture features in the near-threshold FCG regime: (a) parent material 7050-T7451 at R 5 0.7, (b) HAZ at R 5 0.3, and (c) and (d ) low- and high-magnification pictures of the weld centerline at R 5 0.3. the R 5 0.33 case. As discussed before, the FCGR in aluminum alloys worsens in an overaged microstructural condition or a coarser precipitate size.[23–27] Results here, on the other hand, show that the FCG resistance of the HAZ material improves at R 5 0.33 and then (nearly) converges with that of the parent material at R 5 0.7. This is due to the presence of compressive residual stresses that induce significant crack closure at R 5 0.33, as shown in Figure 11(c). Also, it is to be noted that this crack closure is not due to surface roughness or crack deflection. The fracture was extremely flat for this alloy for all conditions tested in the ESET specimen configuration. The main factor that is contributing to crack closure in this alloy, for this HAZ region, is due to the longitudinal residual-stress factor, which overrides the microstructural factors and helps the HAZ maintain a FCG resistance that is higher, in spite of an overaged microstructure, relative to the baseline. In the case of the weld-nugget region, the compressive residual-stress argument would suggest that the FCG resistance and threshold should be higher than that of the parent material as well as the HAZ specimen. Also, there should be a significant crack closure present in this condition due to the presence of the compressive residual stresses. However, results indicate otherwise. Fractography in Figures 13(c) and METALLURGICAL AND MATERIALS TRANSACTIONS A (d) shows that the fracture facet size is extremely fine and much smaller than that of the parent material. The fine fracture facet size, in fact, corresponds to the dynamically recrystallized grain size present in the weld nugget. Furthermore, the fatigue fracture in the weld nugget is likely to be intergranular, although further confirmation through an analysis of the fracture facet orientation may be required. The FCG in highly recrystallized aluminum alloys such as alloy 7050 has been shown to propagate through intergranular failure.[28] The low fatigue threshold (and faster CGRs) in the weld nugget is due to the intergranular fatigue failure mechanism (which overrides the compressive stresses and the slip reversibility). Experiments on different specimen geometries are currently being investigated, to isolate the specimen-geometry effect from the microstructural and failure-mechanism effects on FCGRs in FSW aluminum alloys. V. CONCLUSIONS Friction stir welding of alloy 7050-T7451 resulted in a dynamically recrystallized zone, TMAZ, and HAZ. The dynamically recrystallized zone contains 1.5-mm-sized grains, and the HAZ has a grain size and shape similar to that of the parent material. In the as-FSW condition, VOLUME 31A, SEPTEMBER 2000—2191 strengthening precipitates present prior to the FS welding process are dissolved in the weld-nugget region. In the HAZ, precipitates and the the PFZ are coarsened by a factor of 5 compared to the parent material. Heat treatment of the welded plate at 120 8C for 24 hours, to stabilize the as-FSW microstructure, produced very fine strengthening precipitates in weld-nugget region. However, the size of the precipitates and the PFZ in the HAZ were unaffected by the heat treatment. The FCGR results, at a load ratio of 0.33, show that fatigue-crack propagation in the weld-nugget region is inferior, while the FCG in the HAZ is superior to that of alloy 7050-T7451. Analysis of crack closure, fractography, microstructure, and residual stresses suggests that, at a load ratio of 0.33, compressive residual stresses dominate FCG in the HAZ. For the weld-nugget region, microstructure and intergranular failure mechanism dominate FCG. The specimen-geometry effect on the FCGR is important for FSW alloys. Further work is being conducted on various FCG specimen geometries. ACKNOWLEDGMENTS KVJ expresses his appreciation to Dr. William Bozich, Senior Technical Fellow of the Boeing Company, for providing the 7050 FSW plate. The experimental assistance of Mr. Robert Lewis and Ms. Luann Piazza, AFRL/MLLM, in conducting the optical metallography and scanning electron microscopy is greatly appreciated. The numerous discussions with Dr. S. Lee Semiatin are gratefully acknowledged. REFERENCES 1. W.M. Thomas, E.D. Nicholas, J.C. Needham, M.G. Church, P. Templesmith, and C.J. Dawes: International Patent Application No. PCT/ GB92/02203 and GB Patent Application No. 9125978.8, 1991. 2. O.T. Midling: Proc. 4th Int. Conf. Aluminum Alloys, Atlanta, GA, 1994, pp. 451-58. 3. W.J. Arbegast and Paula J. Hartley: 5th Int. Conf. on Trends in Welding Research, Pine Mountain, GA, June 1998, J.M. Vitek and Stan David, eds., ASM International, Materials Park, OH, 1999, pp. 1-7. 4. Karl-Erik Knipstrom and Bertil Pekkari: Weld. J., 1997, Sept., pp. 55-57. 5. M.W. Mahoney, C.G. Rhodes, and W.H. Bingel: Metall. Mater. Trans. A, 1998, vol. 24A, pp. 1955-64. 2192—VOLUME 31A, SEPTEMBER 2000 6. G. Liu, L.E. Murr, C.-S. Niou, J.C. McClure, and F.R. Vega: Scripta Mater., 1997, vol. 37 (3), pp. 355-61. 7. L.E. Murr, G. Liu, and J.C. McClure: J. Mater. Sci. Lett., 1997, No. 16, pp. 1801-03. 8. C.G. Rhodes, M.W. Mahoney, W.H. Bingel, R.A. Spurling, and C.C. Bampton: Scripta Mater., 1997, vol. 36 (1), pp. 69-65. 9. Z. Feng, J.E Gould, and T.J. Linert: in Hot Deformation of Al Alloys— II, T.R. Bieler, L.A. Lalli, and S.R. MacEwen, eds., TMS, Warrendale, PA, 1998, pp. 149-58. 10. T.J. Linert, R.J. Gryllis, J.E. Gould, and H. L. Fraser: in Hot Deformation of Al Alloys—II, T. R. Bieler, L.A. Lalli, and S.R. MacEwen, eds., TMS, Warrendale, PA, 1998, pp. 149-58. 11. G.S. Frankel and Z. Xia: J. Sci. Eng. Corr., 1999, vol. 55 (2), pp. 139-50. 12. Beate Heinz, Birgit Skrotzki, and Gunther Eggeler: Proc. 7th Int. Conf. ICAA7, Charlottesville, VA, Apr 9–14, 2000: Mater. Sci. Forum, 2000, vols. 331–337, pp. 1757-62. 13. A.F. Norman, I. Brough, and P.B. Prangnell: Proc. 7th Int. Conf. ICAA7, Charlottesville, VA, Apr 9–14, 2000: Mater. Sci. Forum, 2000, vols. 331–337, pp. 1713-18. 14. A.P. Reynolds, W.D. Lockwood, and T.U. Siedel: Proc. 7th Int. Conf. ICAA7, Charlottesville, VA, Apr. 9–14, 2000: Mater. Sci. Forum, 2000, vols. 331–337, pp. 1719-24. 15. E.L. Bernstein and A.C. Nunes, Jr.: Weld. J., in press, 2000. 16. Yuh J. Chao and Xinhai Qi: Paper presented at the Int. Symp. of Friction Stir Welding, Thousand Oaks, CA, June 1999. 17. R.S. Piascik, J.C. Newman, Jr., and J.H. Underwood: Fat. Eng. Mater. Struct., 1997, vol. 20 (4), pp. 559-63. 18. K.V. Jata* and R.J. Rioja: *Air Force Research Laboratory, Materials and Manufacturing Directorate, AFRL/MLLM, Wright-Patterson Air Force Base, OH, unpublished research, 1998. 19. B. Derby: Acta Metall., 1991, vol. 39 (5), pp. 955-62. 20. K.V. Jata and S.L. Semiatin: Scripta Mater., in press, 2000. 21. J.T. Staley, S.C Byrne, E.L. Colvin, and K.P. Konnear: J. Mater. Sci. Forum, 1996, vols. 217–222, pp. 1587-92. 22. R.J. Bucci: Fracture Mechanics, 13th Conf., Richard Roberts, ed., ASTM, Philadelphia, PA, 1981, pp. 28-47. 23. L.R. Link: Fatigue and Fracture Testing of Weldments, ASTM STP 1058, H.I. McHenry and J.M. Potter, eds. ASTM, Philadelphia, PA, 1990, pp. 16-33. 24. S. Suresh, A.K. Vasudevan, and P.E. Bretz: Metall. Trans. A, 1984, vol. 15A, pp. 369-79. 25. J. Lindigkeit, G. Terlinde, A. Gysler, and G. Lutjering: Acta Metall., 1979, vol. 27, pp. 1717-26. 26. F.S. Lin and E.A. Starke, Jr.: Mater. Sci. Eng., 1980, No. 43, pp. 65-76. 27. A.K. Vasudevan and R.D. Doherty: Treatise Mater. Sci. Technol., 1989, vol. 31, pp. 445-62. 28. F.D.S. Marquis: in Light Weight Alloys for Aerospace Applications IV, E.W. Lee, W.E. Frazier, N.J. Kim, and K.V. Jata, eds., TMS, Warrendale, PA, 1997, pp. 61-72. METALLURGICAL AND MATERIALS TRANSACTIONS A