Introduction

Manufacturing industries, such as the aerospace, automotive, and shipbuilding industries, often need to join dissimilar metals to create structures that are sufficiently lightweight and mechanically sound. In particular, because aluminum alloys offer a wide range of properties depending on their chemical composition and heat treatments, the joining of dissimilar aluminum alloys facilitates a flexible product design by leveraging the desired properties of each material (Ref 1). The chemical and mechanical incompatibilities of dissimilar metals, however, can detrimentally impact weld quality during traditional fusion welding. Because friction stir welding (FSW) achieves the joining process in solid state and because the heat input during FSW is lower than traditional methods, FSW is an extremely promising method for joining dissimilar metallic materials. Though friction stir welding is now a mature manufacturing technology, research and development programs have primarily focused on the joining of the same metallic material (Ref 2, 3). Therefore, in order to produce high-quality, defect-free welds with the desired mechanical properties in dissimilar metallic materials, the FSW parameters and tool geometry for these configurations still require thorough investigation and optimization.

DebRoy and Bhadeshia (Ref 4), Murr (Ref 5), and Kumar et al. (Ref 6) have reviewed the current state of research regarding dissimilar FSW joints. The primary difficulty in joining dissimilar materials by FSW rests in the diversity of properties (such as melting point, thermal conductivity, strength, density, viscosity) of the welded alloys. The discontinuity in properties across the abutting surfaces creates an asymmetry in both heat generation and material flow during friction stir welding (Ref 6). For this reason, placing the appropriate alloy on the advancing or retreating side is critical to optimizing mixing and to maximizing weld properties during the welding of dissimilar materials (Ref 7-12). For example, Park et al. (Ref 7) evaluated the influence of weld configuration on the properties of friction stir welded joints of aluminum alloys 5052-H32 and 6061-T6. They found that better mixing occurred when the softer alloy (5052) was located on the advancing side. In the reverse configuration, a thinner stir zone and inadequate mixing were observed. Similar findings were reported by Guo et al. (Ref 8), who investigated the friction stir welding of 6061 and 7075 aluminum alloys. Here, the authors also concluded that superior mixing of the alloys was achieved when the softer 6061 alloy was placed on the advancing side.

Many papers, however, present successful investigations where the harder/stronger material is placed on the advancing side in the dissimilar weld, for example, 2024 on the advancing side of the 2024-6056 configuration in (Ref 13), 7075 in the 7075-2017A configuration in (Ref 14), and 2219 in the 2219-5083 configuration in (Ref 15). Reza-E-Rabby et al. (Ref 9) postulated that the alloy placement significantly influences the weldability of dissimilar friction stir welds and showed that placing the stronger material on the advancing side resulted in fewer defects and discontinuities in the weld.

Lee et al. (Ref 10) found that in the case of friction stir welding of cast A356 and wrought 6061 aluminum alloys, the mechanical properties of weld were mainly dependent on the retreating side alloy since the microstructure of the stir zone was primarily composed of this alloy. In yet another research that focused on cast Al 6061welded with wrought Al 6061, the weld configuration and tool rotation speed altered the dominant composition in the stir zone (Ref 16). For tool rotational speeds of 800 rev min−1 and 1000 rev min−1, the material placed on the retreating side dominated the weld zone. When the tool rotational speed was greater than 1000 rev min−1, the material placed on the advancing side primarily comprised the weld zone. Cole et al. (Ref 11) examined the friction stir welds of 6061 and 7075 aluminum alloys and concluded that not only should the material with the lowest solidus temperature be placed on the “cold” side of the weld (retreating side) but also that the material more susceptible to weakening at elevated temperatures should experience a lesser amount of energy input during the weld process. However, Sato et al. (Ref 12) reported that in their research focused on 7075 and 2024 dissimilar welds, the effect of weld configuration on the microstructure and hardness profiles was negligible.

The research presented here characterizes the microstructure and mechanical properties of friction stir welded 7075-T651 and 5083-H111 commercial aluminum alloys. These aluminum alloys are widely used in the aerospace and shipbuilding industries, and, as such, the joining of these alloys is often necessary. Because the 7075 alloy is not recommended for conventional fusion welding, the application of friction stir welding constitutes a reasonable alternative. Studies on the FSW process for dissimilar aluminum 5xxx and 7xxx alloys joints are infrequent, and those that have already been published in the open literature mainly focused on mechanical properties (Ref 17, 18). In the present investigation, in addition to mechanical testing, numerical modeling and microstructural characterization were performed. In particular, the influence of the weld configuration (locations of the 7075 and 5083 alloys alternately on the advancing and retreating sides) on material flow, microstructure, and mechanical properties is thoroughly discussed.

Materials and Methods

Friction Stir Welding

Friction stir welding of 7075-T651 (age-hardenable) and 5083-H111 (work-hardenable) commercial aluminum alloys was performed at the Instytut Spawalnictwa (Polish Welding Centre of Excellence) in Gliwice, Poland, utilizing a typical milling machine specifically adapted for the welding trials. The chemical composition and properties of these alloys are shown in Tables 1 and 2, respectively. The workpieces were in the form of plates 200 mm long, 75 mm wide, and 6 mm thick joined in two, distinct butt weld configurations—one with 5083 situated on the advancing side and 7075 on the retreating side and the other with the alloy positions reversed. Numerous welding experiments were performed to determine the weld parameters, i.e., welding velocity, rotational speed and vertical force, and tool geometry that maximized weld quality. Ultimately, the FSW tool was made of HS6-5-2 high-speed steel with a convex-scrolled shoulder (24 mm in diameter with a 1.27 mm pitch) and a Triflute pin (10 mm in diameter and 5.8 mm in length with a 1.27 mm pitch on the convex surfaces). The tool tilt angle during processing was held constant at 1.5°. The rotational speed, ω, and tool velocity, u w , were 355 rev min−1 and 140 mm min−1, respectively, and the applied force during processing was set to 28.3 kN. St. Węglowski et al. present more details associated with the capabilities of the FSW equipment, such as force and torque measurement in (Ref 19, 20). The temperature profile across the weld zone was recorded using a Vigocam v50 Thermal Imaging Camera (±2% accuracy). All temperature measurements were taken in air atmosphere at a temperature of 20 °C and a relative humidity of 50% in daylight.

Table 1 Chemical composition of the joined aluminum alloys (weight %)
Full size table
Table 2 Properties of the joined aluminum alloys
Full size table

Microscopy

The microstructure of the produced welds was characterized by light microscopy (LM) and scanning electron microscopy (SEM). Metallographic samples were excised from sections perpendicular to the weld direction and then subjected to a standard preparation technique, i.e., initial grinding on abrasive papers (SiC) and mechanical polishing on cloths. Samples for LM were ultimately anodized in a 1.8 ml HBF4 with 100 ml H2O electrolyte solution. Olympus SZX7 and Zeiss Axio Imager M1 m light microscopes were used. SEM analysis was performed on a high-resolution FEI Nova NanoSEM scanning electron microscope equipped with a field emission gun and an EDAX system for chemical analysis.

Mechanical Testing

The microstructural studies were supplemented by hardness measurements (Vickers) taken on the same sections as the metallographic examinations. The measurements were taken on a Wolpert–Wilson Tukon 2500 apparatus. The Vickers results were used to construct a hardness profile along the line of the mid-thickness plane on the weld cross section. The applied load was 1 kg, the indentation time was 10 s, and the distance between test points was 1.0 mm. Tensile samples were excised perpendicular to the friction stir welds such that the weld was centered along the tensile specimen in the reduced section. Tensile tests were performed in accordance with the ASTM E 8 standard (with strain rate of 0.001 mm s−1) using an MTS 810 testing machine. SEM was utilized for the characterization of fracture surfaces from ruptured tensile specimens.

Numerical Simulation

Numerical Model and Slip Factor

A coupled thermal/flow model was developed with Comsol multi-physics software to simulate the joining of aluminum alloys 5083 and 7075 by friction stir welding. Figure 1 displays the geometry and meshing utilized in this simulation (39,044 tetrahedral elements, 8004 triangular elements, 887 edge and 68 vertex elements). Beginning with the methodology of Hamilton et al. (Ref 21), the simulation presented here further subdivides the flow-capable region between the advancing and retreating sides into two additional flow regions that represent the unique materials for each weld configuration. Each flow region, therefore, is characterized by distinct thermal properties, slip factors, and viscosities according to which alloy is placed on the advancing or retreating sides. Unlike the previous model that only captured heat generation at the tool shoulder and pin bottom, this simulation also includes heat generation along the side of the Triflute pin. The temperature-dependent thermal conductivities, k, and the specific heat capacities, c p , of the two aluminum workpieces are taken from (Ref 22, 23) for 5083 and from (Ref 24) for 7075. The detailed calculation of the viscosity for each aluminum alloy and of the strain rate is presented in (Ref 25). The values of Q, A, α, and n used in the viscosity calculations are shown in Table 3 (taken from (Ref 26) for both alloys). For the process parameters utilized in this investigation, the maximum effective strain rate is 43.0 s−1.

Fig. 1
figure 1

Computational solid model of the friction stir welding process

Full size image
Table 3 Material constants for the Sheppard and Wright flow stress equation and Zener–Hollomon parameter
Full size table

Originally, Hamilton et al. (Ref 27) defined the slip factor for like-metal welding based upon the energy per unit length of weld relative to a theoretical maximum weld energy at the solidus temperature of that alloy. For this investigation of dissimilar metal welding, however, the weld energy exceeds the theoretical maximum energy for 5083 as defined in (Ref 27). Therefore, the slip factor, δ, was redefined based on the welding temperature, T, relative to the solidus temperature, T s , of each aluminum alloy:

$$\varvec{\delta}= {\mathbf{exp}}\left( {\varvec{\beta}\frac{\varvec{T}}{{\varvec{T}_{\varvec{s}} }}} \right)$$
(1)

where β is a scaling factor. To determine the value of β, the temperature-based slip factor was introduced into a prior, calibrated simulation of friction stir welded aluminum 7136-T76 presented in (Ref 28). From these data, the appropriate value for β was found to be 4/7. This definition of the slip factor was also utilized and verified in (Ref 29) for the dissimilar metal welding of aluminum alloys 2017A and 7075.

Verification

In order to verify the redefined slip factor and the subsequent temperature predictions of this simulation, weld temperatures were experimentally recorded by thermal imaging. Specifically, thermographic data taken from the weld surface located 3 mm behind the tool shoulder were compared with the simulation temperatures for each weld configuration at this same location. When 5083 is placed on the advancing side and 7075 on the retreating side, the peak temperature recorded by the thermal imaging camera in this region was 670 K, which compares favorably with the 660 K predicted by the simulation. Overall, there is good agreement between the experimental temperatures and the predicted temperatures across the weld zone (±12 mm from the weld centerline). When 7075 is placed on the advancing side and 5083 on the retreating side, the recorded peak temperature in this region for this configuration is 623 K and that predicted by the simulation is 610 K, and again, overall there is good agreement between the experimental and numerical results across the weld zone. Given the correlation between the predicted and experimental temperatures within the flow-capable and tool regions for both weld configurations, the numerical simulation is a suitable tool to help characterize the friction stir welding of 5083 and 7075.

Results and Discussion

Macrostructural Analysis

Macrographs of the produced joints are presented in Fig. 2 and show the overall surface appearance and weld structure. In these images, the surfaces of the welds exhibit good quality and are without defects. Flash appears on the 5083 side of the joint in both weld configurations; however, the amount of flash is smaller for the AS 5083–RS 7075 configuration (Fig. 2b) When observing the macrostructures of the weld cross sections, the AS 7075–RS 5083 configuration (Fig. 2a) contains volumetric defects in the lower part of the weld as indicated in the figure. Also, a distinct “zigzag” boundary develops between the alloys in this weld configuration with the absence of a well-defined nugget or stir zone. In contrast, the AS 5083–RS 7075 weld configuration displays mixing of the workpieces and the formation of a nugget/stir zone (Fig. 2b). The “zigzag” type of boundary between the alloys is similar to that observed by Sato et al. (Ref 12) and more recently by da Silva et al. (Ref 30) for the friction stir welding of 2024 and 7075 at low heat input. Both of these research teams, however, reported satisfactory mixing of the 2024 and 7075 alloys and stir zone formation when the heat input was increased. This issue is further elaborated in the proceeding section.

Fig. 2
figure 2

Surface macrographs of welds: (a) AS 7075–RS 5083 and (b) AS 5083–RS 7075

Full size image

Microstructural Analysis

Figures 3 and 4 present the microstructures observed by light microscopy on the weld cross sections (the samples were anodized to enhance the microstructural details). Toward the bottom area of the AS 7075–RS 5083 weld in Fig. 3 and as also observed in Fig. 2(a), wormhole-type defects and microvoids are easily discernable. These weld defects are even more recognizable in the SEM image shown in Fig. 5. In contrast, however, the AS 5083–RS 7075 weld configuration in Fig. 4 is defect-free. These results, therefore, suggest that superior weld quality is obtained when the softer alloy (5083 in this case) is placed on the advancing side and the harder alloy (7075) is placed on the retreating side. According to (Ref 9), however, defect-free welds are produced when the stronger alloy (2050 in their case) is located on the advancing side.

Fig. 3
figure 3

Microstructure of AS 7075–RS 5083 weld

Full size image
Fig. 4
figure 4

Microstructure of AS 5083–RS 7075 weld

Full size image
Fig. 5
figure 5

SEM images of (a) tunnel defect and (b) microvoids in the bottom part of AS 7075–RS 5083 weld (marked by arrows)

Full size image

Perhaps the question as to the appropriate alloy to place on the advancing or retreating side to maximize weld quality should not be framed in the context of room-temperature mechanical properties. Instead, the question should be framed in the context of the flow stress of each alloy and the potential temperatures generated during friction stir welding for a given tool geometry. The temperature distribution during friction stir welding is asymmetric, and in general, the advancing side temperatures are hotter than those on the retreating side (Ref 21). As processing temperatures increase during FSW, the flow stresses of the workpieces are reduced, thereby lowering the viscosities of the alloys and permitting better material flow. Due to the higher processing temperatures on the advancing side, therefore, the potential to reduce the flow stress and the viscosity of the alloy placed in this location are greater than that on the retreating side. If, during the friction stir welding of dissimilar alloys, there is a significant difference in the flow stress between the two materials, placing the higher flow stress alloy on the advancing side could enhance material flow and, therefore, maximize weld quality. Similarly, placing the higher flow stress material on the retreating side could inhibit material flow due to the lower processing temperatures on this side of the weld. In addition, at low heat inputs, the difference in flow stress between the workpieces would be exaggerated by the lower weld temperatures, and, consequently, the need to place the higher flow stress material on the advancing side would become more critical to weld quality. At high heat inputs, however, the difference in flow stress between the alloys would be mitigated by the higher weld temperatures, and the weld configuration would become less critical to weld quality.

Sheppard and Wright (Ref 31) proposed the familiar formulation for the flow stress utilizing the Zener–Hollomon parameter. In the current investigation of joining 5083 and 7075, even though 7075-T651 has greater room-temperature mechanical properties than 5083-H111, the flow stress of 5083 is greater than that of 7075 based on calculations using the constants for the Sheppard–Wright formulation, i.e., Q, A, α, and n, from either (Ref 26) or (Ref 32) for 5083 and from (Ref 26) for 7075. The proposed hypothesis on weld configuration, therefore, suggests that superior weld quality should be obtained when 5083 is placed on the advancing side of weld, and this is, indeed, experimentally demonstrated. As to the work of (Ref 9) in relation to 6061-T651 and 2050-T4, aluminum 2050 has a higher flow stress than 6061 based on calculations using the Sheppard–Wright formulation constants for 2050 from (Ref 33) and from (Ref 26) for 6061. As previously noted, the authors in (Ref 9) concluded that placing the 2050 alloy on the advancing side produced better weld quality, as suggested by the proposed hypothesis. Now, rather than there being conflict between their results and those presented here, there is agreement. In the studies presented by (Ref 11) for welding 6061-7075, by (Ref 13) for welding 2024-6056, and by (Ref 14) for welding 7075-2017A, superior weld quality was obtained when the higher flow stress material was placed on the advancing side, thus supporting the hypothesis. It should be noted, however, that in (Ref 12), Sato concluded that alloy placement did not ultimately influence the weld quality and in (Ref 8), Guo clearly obtained better weld quality when the lower flow stress material of 6061 was placed on the advancing side and the higher flow stress 7075 was placed on the retreating side. The constants for the Sheppard–Wright formulation of flow stress for the different alloys referenced in this discussion are summarized in Table 3.

The typical FSW microstructural zones are easily distinguished in Fig. 3 and 4: the stir zone (SZ) or nugget in the weld center, the thermomechanically affected zone (TMAZ), and the heat affected zone (HAZ) near the base materials. For both weld configurations, an asymmetry in the microstructure occurs not only in the stir zone but also in the TMAZ on the advancing (AS) and retreating sides (RS). It has been well established that in FSW joints between alloys of the same kind, the SZ/TMAZ boundary is very sharp and well defined on the advancing side, but more diffuse on the retreating side (Ref 34, 35). The same phenomenon is observed in the present study for the dissimilar welds; however, it was additionally discovered that the thermomechanical zones differ in width depending on the weld side and weld configuration. On the advancing side, the TMAZ is much narrower than on the retreating side. This is illustrated in Fig. 3 (region C) as well as in Fig. 4 (region C). Moreover, in the TMAZ, on the 5083 alloy side, the grains were less elongated.

The stir zone is the most complex part of the weld, in terms of the microstructure that is formed during welding, since it is the location where mixing and mutual interactions between the dissimilar alloys take place. The stir zone in the examined welds is characterized by a vortex-like structure, but it is different for each weld configuration. The stir zone area in the AS 5083–RS 7075 weld is much wider at the bottom, and its microstructure is more complex than that in the AS 7075–RS 5083 configuration. In the case of the AS 7075–RS 5083 configuration, the microstructure resembles the “onion rings” pattern that is typical in friction stir welds between alloys of the same kind (Ref 3, 34). These patterns occur prominently toward the weld bottom on the advancing side (see region E in Fig. 3). EDS analysis in these areas shows that the 7075 alloy is the dominant material, i.e., the alloy placed on the advancing side as was reported by Priya et al. (Ref 36). It may be that the “onion rings” pattern is associated with the unique temperature histories of the mixed materials during the FSW process. Such a hypothesis is based on an earlier work of Hamilton et al. (Ref 21) who showed that during the FSW process, “hotter” surface material flows into the weld nugget where it interleaves with “cooler” in situ material, giving rise to the formation of onion rings in aluminum 7042-T6. The “onion rings” structure in dissimilar welds could also be associated with a mixing process that is incomplete for dissimilar alloys. This incompleteness could also result in the formation of a microstructure composed of interleaving bands from the different alloys. However, the non-uniform temperature distribution within the weld undoubtedly influences the complex microstructure forming in the stir zone.

According to the numerical model and confirmed by experimental measurements, when the higher flow stress material (5083) is placed on the advancing side, the maximum weld temperature (~773 K) is higher than that in the reverse configuration (~723 K). Reza-E-Rabby (Ref 9) similarly noted that higher weld temperatures were achieved when the higher flow stress aluminum 2050 was placed on the advancing side with the lower flow stress aluminum 6061 on the retreating side. The authors theorized that the weld temperature may be related to the contact area between the tool and workpiece or due to the discrepancy in thermal conductivity between the alloys. They noted, however, that their observations of temperature held even when the weld power was sometimes less when the 2050 alloy was placed on the advancing side. In the present research, when the 5083 was placed on the advancing side, the measured torque was 63.3 Nm as opposed to 58.4 Nm when the 7075 was placed on the advancing side, suggesting a higher heat input for AS 5083–RS 7075 arrangement. With all other joining conditions held constant, the amount of heat generated during FSW and the temperatures achieved, therefore, are unequivocally dependent upon the alloy placement in weld. The larger amount of heat in the AS 5083–RS 7075 configuration produced a wider stir zone with more effective mixing of the joined materials.

SEM observations and EDS analysis provided additional characteristics of the observed bands. The micrographs produced by backscattered electrons (Fig. 6) revealed lighter and darker bands in the stir zone that were composed of fine, equiaxed grains but with different grain sizes. The grains of lighter bands are 2–4 µm in size while those of the darker bands are 5–8 µm in size. EDS results (Table 4) also demonstrate that the bands differ in content of the main constituent elements (magnesium, zinc, copper, and copper). The brighter areas contain, besides aluminum, about 6% zinc, 3.2% magnesium, and 2% copper. The darker areas predominantly contain aluminum and magnesium. These EDS results suggest that the 7075 alloy comprises the lighter bands and the 5083 alloy the darker ones. However, in the AS 5083–RS 7075 weld configuration, in addition to these bands, additional bands (gray in color) are observed. The gray bands contained less magnesium and more zinc and copper than the darker ones. This elemental composition suggests that the gray bands are formed by both alloy workpieces. Hence, the higher weld temperatures realized for this configuration led to the development of a more complex microstructure, i.e., creating bands with element content comprised from both base materials.

Fig. 6
figure 6

SEM microstructure of bands in the stirred zone: (a) AS 7075–RS 5083 and (b) AS 5083–RS 7075 weld

Full size image
Table 4 Content of the main elements in the bands of the SZ (weight %)
Full size table

The small size and equiaxed shape of the grains in the stir zone suggest that recrystallization occurs in this region during friction stir welding. However, the differences in grain size between particular bands (bands of the 7075 and 5083 alloys) may indicate various recrystallization rates or other phenomena such as dissolution and/or re-precipitation of secondary phases occurring in these regions. This, in turn, can be correlated with the dissimilarity of the joined alloys, i.e., the different chemical composition, properties, and initial microstructure. The nucleation and growth of new grains mainly depend on strain rate, annealing conditions, initial grain size, chemical composition, size, and distribution of particles (Ref 37). According to the literature (Ref 38, 39), the mechanisms of recrystallization differ between the alloys. It is believed that the 7075 alloy was subjected to continuous or discontinuous dynamic recrystallization, whereas the 5083 alloy might be recrystallized by geometric dynamic recrystallization. Geometric dynamic recrystallization (GDRX) is a process by which a new grain structure is formed as a result of a change in grain geometry coupled with deformation (Ref 39). Similar phenomena were investigated by Etter et al. (Ref 40) who found that during friction stir welding of workpieces of the same aluminum alloy (5251) but subjected to different treatments (work hardened and annealed), continuous dynamic recrystallization occurred in the cold rolled alloy (5251-H14) while geometric dynamic recrystallization took place in the initially annealed samples (5251-O). Thus, it is quite likely that the recrystallization mechanism in particular metals forming a dissimilar weld is related to their respective strengthening mechanisms.

Mechanical Characterization

Figure 7 presents the hardness distributions and hardness profiles of the weld cross sections from the two weld configurations. The hardness of the baseline 7075-T651 alloy was about 160 HV1 while the hardness of the 5083-H111 alloy was approximately 80 HV1. The changes in hardness, demonstrated clearly on the hardness maps, reflect the microstructural observations. By analyzing the hardness profiles, it is possible to distinguish the specific FSW zones and their width (determined on the mid-thickness of the weld), as marked in Fig. 7. For the AS 7075–RS 5083 configuration, the stir zone is about 8 mm wide and is narrower than for the AS 5083–RS 7075 configuration (approximately 13 mm). Hardness fluctuations in the SZ, especially in the AS 5083–RS 7075 weld, were observed. These fluctuations may be associated with the unique hardness character of the individual 5083 and 7075 bands as similarly noted by Khodir and Shibayanagi (Ref 41) in the welding of 2024 and 7075. Near the weld center, on the AS side, the hardness decreased to 80 HV1 (i.e., the hardness of the baseline 5083 alloy), suggesting that 5083 dominated in this region. Hardness changes in the stir zone in the AS 7075–RS 5083 configuration are not as pronounced, but the minimum hardness occurs in an area close to the weld center on the 5083 alloy side. The hardness of the weld center attained a value of 145 HV1 for the AS 7075–RS 5083 weld and 85 HV1 for the AS 5083–RS 7075 configuration. However, the highest value of hardness (about 150 HV1) was found in the SZ, on 7075 side (i.e., on the advancing side for the AS 7075–RS 5083 configuration and on the retreating side for the AS 5083–RS 7075 weld).

Fig. 7
figure 7

Hardness profiles along mid-plane for (a) AS 7075–RS 5083 and (b) AS 5083–RS 7075

Full size image

For both weld configurations, the hardness in the TMAZ region decreases. On the 7075 side of the weld, the TMAZ is 4 mm wide, and the hardness decreases by approximately 25 HV1. The TMAZ on the 5083 side is more narrow (1–2 mm), and consequently, the drop in hardness is quite abrupt. The hardness of the heat affected zone is different depending on the weld side and configuration. The HAZ on the 5083 side is characterized by the hardness of the 5083 alloy itself, i.e., 80 HV1, while the hardness in the HAZ on the 7075 side increases from 120 HV1 to 160 HV1 (the 7075 baseline value). The shape of the hardness profile on the 7075 side of the weld, from the weld center to the HAZ, corresponds to half of the typical W-shape hardness profile observed in friction stir welded 7xxx series Al alloys (Ref 3).

The tensile test results for the welded specimens are listed in Table 5, and representative stress–strain curves are provided in Fig. 8. The welded samples exhibited higher mechanical properties compared to the properties of the baseline 5083 alloy (Table 2). The weld configuration did not influence the tensile properties as the tensile strengths for the AS 7075–RS 5083 and AS 5083–RS 7075 configurations were 367 MPa and 365 MPa, respectively. The values of yield strength and elongation for both weld configurations were also similar. The joint efficiencies, which are measured as a ratio of the tensile strength of the weld to the tensile strength of the softer alloy, is above 105% for both weld configurations. In other studies, e.g., on AS 2219–RS 5083 (Ref 15) and AS 6056–RS 2024 welds (Ref 13), the highest joint efficiency, calculated in the same manner, was about 90%. The fracture of samples during the tensile tests always occurred in the baseline 5083 alloy (about 30 mm from the weld center); see Fig. 9. Fracture was not detected at the weld center in the AS 7075–RS 5083 configuration despite the presence of large defects in this area. Also, fracture did not occur at the weld center in the AS 5083–RS 7075 configuration where the hardness values corresponded to that of the 5083 alloy. In all samples, the fracture surfaces exhibited ductile fracture character (Fig. 9).

Table 5 Mechanical properties of the welded specimens (sample standard deviations in parentheses)
Full size table
Fig. 8
figure 8

Representative stress–strain curves for AS 7075–RS 5083 and AS 5083–RS 7075 weld configurations

Full size image
Fig. 9
figure 9

Tensile specimens after fracture and fracture surfaces: (a) AS 7075–RS 5083 weld and (b) AS 5083–RS 7075 weld

Full size image

Conclusions

This investigation demonstrated that with the selected FSW parameters and tool geometry, welds of satisfactory quality can be produced between dissimilar 7075-T651 and 5083-H111 aluminum alloys. The following conclusions are drawn from the investigation:

  1. 1.

    The weld configuration (alloy placement on the advancing or retreating sides) influences the heat generation during the FSW process, though differences in thermal properties of the joined alloys were negligible.

  2. 2.

    The weld produced in the AS 5083–RS 7075 configuration was almost defect-free while the inverse configuration brought about some defects in the form of voids and discontinuities. However, these defects did not ultimately affect the tensile properties of the produced welds.

  3. 3.

    It is hypothesized that better weld quality is achieved in dissimilar FSW welds when the higher flow stress material is placed on the advancing side of the weld. Since the temperature distribution during FSW is asymmetric and the advancing side achieves higher weld temperatures than the retreating side, there is greater potential to reduce the flow stress and the viscosity of the advancing side alloy. If there is a significant difference in the flow stress between the two workpieces, placing the higher flow stress alloy on the advancing side, therefore, could enhance material flow.

  4. 4.

    The microstructure in the stir zone was a clear visualization of the material flow and was characterized by a vortex-like structure with characteristic bands. The microstructure became more complex under the influence of a larger heat input (higher temperatures). The most uniform material flow was observed in the AS 5083–RS 7075 weld.

  5. 5.

    EDS analysis showed different degrees of material mixing. The stir zone microstructure is composed of alternating bands of 5083 and 7075 that also differ in grain size. In the 7075 alloy bands, the grains were finer than that in the 5083 bands. For the AS 5083–RS 7075 weld, the temperatures were sufficient to produce additional bands that consisted of elements from both workpiece alloys.

  6. 6.

    The weld configuration has an influence on material flow, but did not influence the tensile properties. The yield strength, tensile strength, elongation, and maximum hardness in the stir zone were very similar. All tensile specimens fractured on the softer material side, i.e., the 5083 alloy.

  7. 7.

    The joint efficiencies, calculated as a ratio of the tensile strength of the weld and tensile strength of the softer alloy (5083), were above 100% for both weld configurations.