A microstructural investigation into the stationary shoulder corner friction stir welding of aluminium alloys

• Main Author: Gascoyne, Samuel John
• Other Authors: Wynne, Brad ; Prangnell, Phillip
• Published: University of Sheffield 2014
• Subjects: 620
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.659047

A microstructural investigation was performed on a new type of friction joining – stationary shoulder corner friction stir welding (SSCFSW). This technique involves welding at a 45° angle and using filler materials in order to produce rounded corner welds. The filler materials were later coated in copper to ascertain the flow behaviour of the material and to determine what happens to the interface between the vertical and horizontal columns being welded. A new phenomenon – the so called “blades” effect – demonstrated two clearly defined regions within the dynamically recrystallised zone (DRX) which had different chemical compositions from each other. The investigation focused on some of the most commonly used wrought aluminium alloys – 2xxx, 5xxx, 6xxx, and 7xxx series aluminium. These were selected to give a good comparison between heat treated and non-heat treated alloys. Different aluminium series were also cross welded (2xxx to 7xxx) in order to assess feasibility and material flow behaviour. The vast majority of welds were performed in the 6xxx series material; these included the copper tracer samples and a stationary shoulder friction stir butt weld. The first type of analysis was hardness mapping and was initially applied to cross welded alloys AA2014-T4 and AA7075-T6. The hardness maps showed that there were sharp hardness transitions within the DRX. Further analysis with SEM/EDS revealed that the “blades” region - which demonstrated highest hardness within the DRX – was exclusively AA7075-T6 and that the “non-blades” region was exclusively AA2014-T4. Hardness maps were also performed on SSCFSW 6082-T6 and 5083-O. As these were not cross welded materials they didn’t exhibit the same sharp transitions in hardness across the “blade” and “non-blade” regions. However, the hardness profiles did highlight the difference heat treated and non-heat treated alloys, as the 6xxx series saw a drop of hardness across the DRX with some recovery, and the 5xxx series material saw an increase in hardness across the DRX. Both AA7075-T6 and AA2014-T4 are heat treated alloys, so the intense thermo-mechanical process of FSW is expected to cause the hardening precipitates to coarsen and dissolve. Depending on the post weld cooling process the dissolved second phase precipitates may reprecipitate and cause the material to regain some of its hardness. The blade and non-blade regions in the cross welded alloys were segregated on the basis of alloys composition, i.e. the blade region was entirely AA7075 and the non-blade region was entirely AA2014. Thus the post weld cooling phase favoured reprecipitation in AA7075 over AA2014, hence the far higher hardness measurements found in the blade region. For alloys that weren’t cross welded, the difference between the two regions is negligible, indicating that blade and non-blade regions are a primarily a result of material flow. Further tests were carried out on 6082-T6 welds but using copper tracers either between the interfaces of the two aluminium plates to be welded or around the filler wires that would be incorporated into the SSCFSW. A visual inspection showed the distinct presence of the blade and non-blade regions and that copper had preferentially distributed itself into the blade regions. A SEM and EDS was performed on the blade and non-blade regions and confirmed that the blades were copper rich and the non-blades regions were copper free. The filler wires that were coated in copper also saw the copper distributed preferentially into the blade region, however, as the filler wires are only consumed the in the top half of the weld, no copper was found towards the base the weld. This indicated that while material is being segregated during welding it is not massively being dragged down. A copper tracer was also placed between the interface of two plates in a stationary shoulder FSW butt weld, and once again the copper preferentially segregated itself into the blade regions. This indicates that the main mechanism for the blade effect is the stationary shoulder and tool, not the angle of the weld. Crystallographic texture analysis was performed using EBSD for the DRX stretching from the advancing side to the retreating side of weld region. Both regions towards the base and top of the weld region were analysed for a single material SSCFSW of AA6082-T6. The texture had a strong <111> crystal orientation, and was dominated by simple shear torsion texture. The simple shear components of / and C were detected, but instead of a uniform distribution, alternating bands of the B and components, and trace amounts of the C components. For the stationary shoulder butt welds in AA6082-T6 a similar pattern emerged, but with a much stronger detection of the C component, and more in a banded formation. For the other aluminium alloys tested, the prevalence of the <111> crystal texture was also observed, along with the simple shear components of / and C, but the presence of banding was either faint or non-existent. There was no evidence of the blade effect occurring in terms of texture, as EBSD runs were performed across bands that contained both blade and non-blade regions. The blades effect appears to be primarily a result of material flow behaviour and not dynamic recrystallisation mechanisms. The DRX has a consistent grain size throughout, but certain materials during welding preferentially distribute themselves either into the ‘blade’ or ‘non-blade’ regions. This phenomenon is likely to be linked to the threading on the tool used for the SSCFSW process, and the lack of interference of the shoulder.