Towards Qualification of Friction Stir Welding to AA5083-O and AA5052-O Aluminum Alloys

Friction Stir Welding (FSW) is an excellent alternative for joining dissimilar and similar materials in comparison to conventional welding processes. In this sense, this work aims at qualifying FSW to similar AA5083-O and AA5052-O aluminum alloys, with 6.35 mm thickness, in a CNC machining center. Therefore, four welding experiments were undertaken. At first, two types of plate surface finish were considered before joining. Afterward, two tool probe geometries, and changes in the base material position were evaluated. Hence, the welds were mainly analyzed by metallography, tensile testing, and bending tests. Results showed that the machining step to oxides removal and plates alignment is beneficial for the weld processing and the factor that has significant influence to achieve suitable joints is the tool probe geometry. Finally, by obtaining an average ultimate tensile strength (UTS) of around 191 MPa, enhanced microhardness, and bent angle of approximately 150°, the best joint was chosen.


Introduction
One of the main focuses of the metal industry is weight reduction and, in this scenario, aluminum has a density about three times smaller than steel as well as excellent mechanical properties 1 . Welding of aluminum alloys can be relatively difficult due to several factors that may hinder the achievement of a high-quality joint 2 . Therefore, the formation of defects such as cracks, porosities, and deterioration in mechanical properties resulting from traditional welding processes is due, in large part, to the higher heat input imposed, making the use of these techniques sometimes unattractive to industry 3,4 . As a consequence, different joining methods are used, such as riveted, screwed joints, amongst others 5 . Another noticeable difficulty encountered by traditional fusion joining processes is to weld materials with differences in their chemical compositions, hardness, UTS, melting point and which are commonly known as dissimilar materials.
As an alternative for dissimilar welding, FSW has been drawing attention, a process which achieves lower heat input. This thermomechanical processing employs a nonconsumable tool to promote the material's mixing in the hot plasticized state to produce the sound joint. Moreover, the joint is formed as a response to frictional heat and forging force between the tool and the plates to be welded, where the temperatures reached are usually below the material's melting point 2 . Therefore, this method is carried out in a solid-state manner 6 , i.e., the material reaches temperatures sufficient for recrystallization to occur theoretically without melting. Furthermore, FSW attractive characteristics have expanded its application to world-renowned companies such as NASA 7 , BMW 8 , Ford 8 , Volvo 8 , Audi 9 , Airbus 10 , Apple 11 , among others. Recently, authors 6 have suggested that there is an emergent potential of applying this technique to oil and gas pipes. Still, various alloys can be welded, among them, it can be mentioned steel 12 and nonferrous ones such as Al 13,14,15,16,17 , Ti 18 and Ni 6,19,20 . Finally, this process can be a solution for joining dissimilar materials such Ti and Al 21 , Al and Mg 22 , Al and Cu 23 , which proves a unique welding perspective by this technology.
The present work focuses on FSW application for similar Al alloys and the subsequent qualification of the welds processed. Hence, four welding experiments were performed based on process parameters, pre-finishing of the aluminum plates, tool probe geometry, and base materials position (advancing side (AS) or retreating side (RS)).

Materials and Methods
Aluminum plates (AA5052-O and AA5083-O alloys) of distinct chemical compositions were joined in the current work. These alloys belong to 5xxx series, are considered not hardened by heat treatment (precipitation) and were called similar materials since they contain essentially the same alloying elements, but in different quantities. In this context, Table 1 shows the chemical composition of these alloys. Furthermore, the mechanical properties of AA 5052 and AA 5083 alloys are presented in Table 2.  Friction stir welds were produced by a Computer Numerical Control (CNC) machining center at URI-Erechim. This SKYBULL 600 equipment has three axes of movement (x, y, z) and was made by Diplomat ( Figure 1). The friction stir processing parameters were tool rotational speed of 1450 rpm and welding speed of 20 mm/min. The four welding attempts are summarized in Table 3, and their main characteristics were: • Machining of plates surfaces to align the joint and removal of oxides and impurities. This step was carried out in the same FSW machine, after fixing the base material on the Processing parameters such as welding speed, tool rotational, and shoulder penetration did not change in the welding experiments. Still, the tilt angle was 0º, and the axial force was not monitored. It should also be mentioned that this work emphasis on the qualification of the welds, as suggested by ISO 25239-4 26 .
The tool had a shoulder with 19 mm diameter and 2x45º bevel, with spiral striations to facilitate the materials to be forged to the center of the joint as well as allowing a more homogeneous mixing. Two probe variations were employed, as shown in Figure 2. Therefore, probe 1 ( Figure 2a) is a three-sided threaded cam pin, 6.6 mm diameter at the base, 3.5 mm diameter at the tip and an effective length of 6.6 mm, containing an M6 right-hand thread profile throughout its length. On the other hand, probe 2, shown in Figure 2b, is a threaded-tapered probe with 6 mm diameter at the base, 4 mm diameter at the tip and an effective length of 6.2 mm, containing an M6 right-hand thread profile throughout its extension. The material of the shoulder and probes was AISI VC 131 steel. The plate dimensions were 310 mm length, 100 mm width, and 6.35 mm thickness. Therefore, two plates were used for producing the joints. From the 310 mm length, the initial 10 mm and the final 20 mm of the welds were eliminated. From the remaining 280 mm length, 13 specimens were obtained, as presented in Figure 3. These samples were used to the analyzes mentioned below (macrostructure, tensile testing, and bending tests). Before the sample's sectioning, a surface machining process was carried out to remove 0.1 mm from the weld top surface region and, consequently, to eliminate the flash. of welding conditions. Weld G3 showed a sharp flash in comparison to the previous joints obtained (G1 and G2). For the macrostructural analyzes, the samples were prepared according to the basic metallographic practices and etched by aqua regia (distilled water, nitric acid, and hydrochloric acid). These macrostructural observations were done in all specimens.
Tensile tests were performed, considering CP03, CP06, CP09, and CP12 specimens for each welding condition, where the main objective was to obtain the weld UTS and further compare with the base material mechanical properties. The samples were made according to ABNT NBR7549 27, and the tests were carried out on a SHIMADZU brand machine, with a maximum of 300 kN capacity.
Vickers microhardness profile was measured in the CP07 specimen, which was taken from the best-welded joint (G4). The test parameters considered were: load 0.3 Kg and distance between indentations of 0.3 mm. For the microstructural investigation, the sample was also prepared (sanding and polishing) and subsequently etched by Keller reagent (1 ml of hydrofluoric acid, 1.5 ml of hydrochloric acid, 2.5 ml of nitric acid and 95 ml of distilled water).
Finally, samples from each condition were submitted to bending tests. Face bending was performed on CP04 and CP10, and root bending on CP05 and CP11 specimens. The purpose was to bend the samples up to 150° without cracks or fractures.  After this welding experiment (G3), it was verified that the probe crossed the welded plates, a fact that can explain the power oscillation and the flash occasioned. At last, the weld G4 presented the highest flash formed, however, in contrast to the other attempts, it has reached the best top surface finish, resulting in an almost polished shape in the final 80 mm of the weld length. It should also be mentioned that the non-monitoring of axial force may have influenced the flash formation since it becomes a parameter without control and can affect the weld production.

Macrostructure
In general, as shown in Figure 5, the weld macrostructures presented a good mixing within the stirred zone, and no voids or cracks were found in the samples located in the center of the welded length (a more stable region). However, even in visual assessments, some differences between the joints were noted. Macrostructural analyzes of the weld G1 displayed an irregularity in the CP01, a fact that may have occurred due to cutting proximity of this sample to the dive site of the tool (Figure 4a). Regarding the weld G2 (Figure 4b), a slight improvement was obtained in a visual evaluation (compared to the weld G1). In weld G3, the presence of iron particles in CP13 (Figure 4c)) was observed, possibly due to the contact between the probe and the base material support. Finally, small voids formation was noted in some specimens of the weld G4 (Figure 4d). These defects may have occurred due to some inclusions or impurities in the joint 27 .
A study of the base material position effect on friction stir welds of dissimilar aluminum alloys showed that the material's mixing patterns in the joints might be different 27 depending on the alloy location. Therefore, in the present work, which has considered the welding of similar aluminum alloys, when the lower strength alloy (AA5052-O) was located on the AS, and the alloy with higher mechanical properties on the RS (AA5083-O), the materials were better mixed. It can be further verified in Figure 5 by an improved average UTS (from the G3 to G4 experiment). Figure 6 illustrates the average UTS for all the welds. The weld G1 samples exhibited lower UTS than expected, and the rupture occurred in the weld region, which indicates unsuitable welding conditions. On the other hand, the weld G2 obtained higher average UTS in comparison to that of the weld G1, but the results remained unsatisfactory to the weld qualification. Therefore, the samples after tensile tests can be observed in Figure 7.

Tensile testing
Changing the tool probe geometry promoted variations in tensile tests findings, and the weld G3 was able to withstand a mean UTS of 188.6 MPa. Therefore, the tool  probe characteristics are fundamental for obtaining adequate joints. Also, all the specimens fractured in the AA5052-O alloy, outside of the stir zone. In weld G4 case, the average UTS was even higher, approximately of 191 MPa, and all the ruptures occurred on the opposite side, but likewise on the AA 5052-O alloy (due to the base material location inversion), facts that corroborate to the best weld quality. Moreover, as expected, the fractures were in the weak material 28 . Still, the average UTS reached by the welded joints G3 and G4 was above the minimum UTS of AA5052-O alloy (170 MPa-refer to Table 1). Finally, as can also be noted in Figure 7, the majority of these specimens (weld G3 and G4) broke at the base material and achieved the requirements as suggested in 29 .

Microhardness and microstructure
Macrostructure and microhardness profile of the best joint (G4) are presented together, as can be noted in Figure  8b). Regarding the base materials, distinct alloys were etched in a different manner, i.e., distinguishable colors, as can be observed in Figures 8b), c), and d). Non-uniform material flow can also be recognized near the top surface, which is probably related to the CNC machine limitations (tool tilt angle of 0º).
Furthermore, microstructural features of selected regions are displayed in Figures 8a), c), and d). In this sense, finer grains were verified in the stir zone ( Figure  8a)). Therefore, the FSW application to similar aluminum alloys (AA5083-O and AA5052-O) was effective, promoting grain refinement, increased microhardness, and improved average UTS. Hence, the highest microhardness value (~78 Hv) was identified in the weld zone.

Bending tests
As can be seen in Figure 9, the G1 and G2 samples reached root bent angle of 150º, without any cracking. However, in face bending, the samples of these welding experiments broke at the beginning of the test. Further, an improvement in weld G2 was observed in relation to the weld G1, showing that the preliminary machining step is beneficial. In weld G3 and G4 specimens, a bent angle of around 150º was reached, both face and root, indicating that the tool probe geometry is of great importance for achieving high-quality joints.
All the above results reported attempts to the qualification of friction stir welding to similar AA5083-O and AA5052-O aluminum alloys. There were important aspects toward the weld qualification and to achieve a satisfactory joint condition such as plate surface finish, tool probe geometry, and base material location (AS or RS). Furthermore, the weld processed in G4 experiment can be considered qualified and was chosen as the best joint, based on its suitable top surface quality, macrostructure, enhanced microhardness and UTS, and approval in bending tests.

Conclusions
FSW was successfully applied to similar AA5083-O and AA5052-O aluminum alloys. The results of the present work can be understood as follows: * Using surface machining practice on the plates, before welding, resulted in improvement in welds G1 and G2 results, due to the elimination of oxides and impurities in the aluminum surfaces and also to prior plates alignment.
* Tool probe geometry is essential for producing good weld quality; * In initial welding experiments (G1 and G2), most of the specimens subjected to tensile testing broke at the weld centerline, with an average UTS of 156 and 167 MPa, respectively. Still, these welds were not approved in face bending tests, as they did not reach a bent angle of 150°; * Welds G3 and G4 obtained similar results in tensile testing. As an average, the UTS of these experiments was para 188 and 190 MPa, respectively. Moreover, regarding bending tests, the joints G3 and G4 achieved satisfactory results, both face, and root bending. Finally, the joint G4 presented the highest average UTS, may be considered qualified and was selected as the best weld.