Base material
The work began by procuring AA 5754-H111 aluminium alloy sheets with its quotes chemical composition and mechanical properties shown in Tables 1 and 2, respectively. Four sheets having 0.3 mm thickness were cut in sizes of 45 × 110 mm.
Table 1 Chemical composition of AA 5754-H111 aluminium alloy quoted by the material supplier Table 2 Mechanical properties of AA 5754-H111 aluminum alloy quoted by the material supplier Friction stir spot welding process
The process of FSSW studied in this paper relied on using a force-controlled welding machine EJOWELD C50R FSSW (EJOT GmbH & Co. KG Geschäftsbereich EJOWELD, Tambach-Dietharz, Germany). The maximum rotational speed in this machine can go up to 9000 rpm, with a maximal axial load of 8 kN and a dwell time of 5 s. A detailed description of the process setup has been provided elsewhere [15]. The tool used in this work (see Fig. 1) was made from H13 (X40CrMo51) hot-work steel. The tool had a convex contact surface which provides a gradual increase of the pressure providing FSSW of ultrathin sheets without tearing the sheet with tool edge. Consequently, four ultrathin aluminium sheets were lap joined with this convex probe-less tool.
The experimental parameters used in this study were in accordance with the previous study [10]. Two samples were welded at an axial load of 4 kN, the penetration depth of 0.25 mm, and two representative rotational speeds of 1000 rpm and 4500 rpm.
Macro- and microstructural characterization
The two samples processed by FSSW were prepared for metallographic examination. The samples were ground with abrasive paper (grit 360 to 2500) and polished with diamond suspension (6 and 3 µm). The final step was fine polishing with colloidal silica with an average particle size of 0.05 µm (OP-S). After polishing, electrolytic etching was performed with Barker’s etching agent (5 mL HBF4 + 200 mL H2O) using a Struers LectroPol-5 device (40 V, 2 min). The cross-section of the samples was analysed by Zeiss, AxioScope A1 light microscope with AxioCam ICc3. Cross-polarised light under sensitive tint was used for revealing different colours of the grains having different orientations.
The WFI on both samples was analysed with a 200 kV transmission electron microscope (TEM, Tecnai Osiris, FEI) additionally equipped with a scanning unit (STEM) including a high-angle annular dark-field (HAADF, Fischione Co.) detector and energy dispersive X-ray spectrometer (EDX, Super-X system with 4 Bruker silicon drift detectors, Thermo Fisher Co.).
Electrical properties
The electric circuit used for measuring the electric resistance in samples welded at 1000 and 4500 rpm and base material is shown in Fig. 2. The calculation of specific electric resistance was done by using Ohm’s law [16]:
$$R=\frac{V}{I}=\rho \cdot \frac{l}{w\cdot h}\to \rho =\frac{V}{I}\cdot \frac{w\cdot h}{l}$$
where R is electrical resistance (Ω); V is voltage (here present at 2 V), I is the measured current (A), l is the length between the two contacts (set at 8 mm), and w and h are widths and height of the cross-section measured individually for each specimen by a micrometer (mm).
A Time Electronics 5025 generator was used to supply 2 A current (shown by the red line in the circuit in Fig. 2). The voltage was measured at the pre-set distance of 8 mm (cross-section of the red circle in Fig. 2a), by using the FLUKE 8846A device (Fig. 2b). This is the distance which signifies start of the weld (stir) zone (red circle). This measurement scheme provided specific measurements in the stir zone. In all samples, the measurements were performed 12 times at the bottom side of the welds. The average values with standard deviation are presented in this paper.
Mechanical property testing
To analyse the mechanical properties of the weld joints, microhardness and shear tensile tests were performed. Microhardness test was done by using standard Vickers microhardness test with Struers DuraScan 70 machine with 0.1 kg loading. Figure 3 shows an illustrative scheme that was used to obtain the cross-section hardness profile of welds obtained at 1000 and 4500 rpm. The hardness measurements data at WFI of stir zone is presented as a chart. The tests were repeated three times, and standard deviation was calculated. Furthermore, the results were presented as a plot showing the change of hardness through all zones (base material, heat affected zone, thermomechanically affected zone, and stir zone). Both tests were performed according to relevant standards [17].
A Hegewald & Peschke Inspect Retrofit universal tensile testing machine with a maximal load of 20 kN was used to test tensile shear specimens. Testing speed was set to be 10 mm/min, and measurements were performed according to the relevant standards [18]. The specimens were prepared as shown in Fig. 4. The standard setup was slightly modified to engulf all three WFI. For every test, three samples were used and standard deviation was calculated.
Results and discussions
Macro- and microstructure analysis of the weld zone
Figure 5a and b show cross-sections of etched macrographs obtained at rotational speeds 1000 and 4500 rpm, respectively. In the two presented macrographs, a difference in the taper angle of stir zones can be observed. The increase of taper angle from 47° in sample welded at 1000 rpm to 73° in sample welded at 4500 rpm was quite prominent. This event is related to the material’s increasing resistance [19]. Sample welded at 1000 rpm possess high viscosity due to a strain-hardening effect as explained earlier [15, 19]. This phenomenon causes stress to be oriented towards the bottom and a less conical shape of the stir zone. This way the size of the effectively welded area in all three weld faying interfaces remains similar (see Fig. 6). However, in sample welded at 4500 rpm, local thermal softening causes a high drop in the viscosity which pushes the material upwards, towards the tool periphery, and causes reduced pressure in the III WFI. Therefore, a high deviation of effective welded surfaces by weld faying interfaces can be seen. Namely, the I WFI has effective welded surfaces approximately three times smaller compared to III WFI, which can be also observed from Fig. 5b.
The strain hardening effect in the sample welded at 1000 rpm causes the grain refinement (Fig. 5c) compared to the thermal softened sample welded at 4500 rpm (Fig. 5c) and base material (Fig. 5c). The influence of all aforementioned effects on mechanical properties will be discussed in a later section.
To fully understand the influence of rotational speed on WFI characteristics, transmission electron microscope (TEM) was used. The TEM lamella was cut from II WFI in both samples at the same place as shown in Fig. 7a. The lamellae were cut at the beginning of the stir zone because in the middle of the stir zone, there is a large scattering of WFI due to the convex shape of the tool. In Fig. 7b and c, the TEM lamellae prepared from samples welded at 1000 and 4500 rpm, respectively, are shown. Samples welded at 1000 rpm showed a wider and more complex WFI. The measured thickness of WFI in the samples welded at 1000 rpm was 2.19 ± 0.21 µm. The WFI was found to be composed of an oxide layer in the middle (dashed blue line in Fig. 7a) which is 0.22 ± 0.08 µm thick and complex zone surrounding oxide layer composed of globular and needle-like precipitates Al2Mg3 (Fig. 8a and b). Furthermore, it can be noticed that the grains in WFI of welds obtained at 1000 rpm were fibrous and significantly smaller compared to recrystallised stir zone (RSZ) which surrounds WFI. This shape and size of the grains can be a consequence of cyclic microplastic deformation of the protrusions from the rough sheet surface before the welding [10]. This can be explained by the so-called dislocation-glide-assisted-subgrain-rotation model proposed by Jata et al. [20]. In presence of high-level deformation, dislocation glide gives rise to a gradual relative rotation of adjacent subgrains. As such, the shear stress in the FSSW process is dominated by the grains, and subgrains are rotated and elongated in direction of shear load.
In the samples welded at 4500 rpm, the WFI was composed of only an oxide layer 0.54 ± 0.21 µm thick (Fig. 8c and d). This oxide layer was, however, thicker and more compact than that obtained from samples welded at 1000 rpm. Furthermore, there were larger pits than those in the WFI of samples obtained at 1000 rpm. It seems that the WFI of the sample welded at 4500 rpm, due to a lack of stress transferred to lower interfaces, was not stretched as the II WFI of the sample welded at 1000 rpm. Therefore, the oxide layer did not allow diffusion to occur between fragmented oxide particles. This effect was also observed by Sato et al. [11] during friction stir welding of aluminium AA 1050. The entrapped oxide layer affected the mechanical properties of the welds. However, it was observed that only root bend properties of weld were affected where the oxide layer was continuous. No effect of the oxide layer was found in the middle where the oxide layer was fragmented.
The chemical composition of the precipitates and oxide layer was determined by using energy-dispersive X-ray spectroscopy (EDX analysis), and the spatial distribution of the elements can be seen from Fig. 9 for the sample welded at 1000 rpm and in Fig. 10 for the sample welded at 4500 rpm. This analysis revealed that the oxide layer is composed of magnesium and oxygen. It indicated that in aluminium alloy AA 5754 with 3% of magnesium, Al2O3, on the surface layer of the alloy, reacts with magnesium during the FSSW process to form MgO [10]. This phenomenon was first identified by Shirzadi et al. [21] during solid-state diffusion bonding. They noted that when aluminium alloy contains active elements such as Mg, amorphous Al2O3 reacts with active elements and creates crystalline MgO. Needle-like intermetallics are also visible in elemental mapping in Fig. 9. Those intermetallics were composed of magnesium and aluminium. On the other hand, Fig. 10 shows another observation of the intermetallics composed of aluminium, iron, and manganese. According to Choi et al. [22], those are Al6(FeMn) and Al3Mg2 intermetallics, typical for 5xxx alloy. It was observed that the WFI obtained at a low rpm of 1000 contained a much larger volume of those intermetallics. According to Labus Zlatanovic et al. [10], a high local microplastic deformation caused dynamical precipitation responsible for a higher volume of intermetallics in WFI. According to D’Urso et al. [23], an increase in the rotational speed increases the local temperature which in turn leads to thermal softening of WFI (such as in the case of 4500 rpm) leading to a lower viscosity and dynamic recovery as also confirmed by Soare et al. [24].
Electrical resistance
According to Raeisinia et al. [25], the value of electrical resistivity is dependent on various microstructural parameters. The difference in specific electrical resistance of the base material and samples welded at 1000 and 4500 rpm is presented in Fig. 11.
The samples welded at 1000 rpm showed higher specific electrical resistance, followed by the sample welded at 4500 rpm. The lowest specific resistance was for the base material. This effect can be correlated with two microstructural observations. In the samples welded at 1000 rpm, strain hardening is predominant, and this causes the accumulation of dislocations. Raeisinia et al. [25] investigated the influence of dislocation on the electrical resistance of AA 5754 alloy. They found that an increase in dislocation density causes an increase in the specific resistance. Furthermore, Lee et al. [26] investigated the influence of the intermetallic compound layer thickness on the electrical and mechanical properties of Cu/Al joint during annealing. The influence of intermetallic on the electrical properties was high. It was found that the thicker intermetallic layer degraded the electrical resistance. Furthermore, Mubiayi et al. [27] investigated the influence of rotational speed at electrical resistivity during FSSW of Al/Cu dissimilar alloys. An increase in rotational speed caused a decrease in the electrical resistance. Also, the resistance of welds was higher compared to that of the base material, which was correlated with a higher number of intermetallics present in the weld.
In the samples welded at 1000 rpm, there were three WFI regions with a higher amount of intermetallics. Furthermore, there is also an oxide layer in the middle of every WFI with a very high electrical resistance of about 1016 Ω·m [28] combined with nano pits. All those parameters combined will influence the electrical resistance of the weld, and this influence in weld obtained at 1000 rpm compared to the base material was estimated to be as much as 42% during this investigation. In the samples welded at 4500 rpm, the influence was lower (14%) which can be correlated with the fact that only three thicker oxide layers were present in the stir zone at the weld interface and no additional intermetallics were found.
Mechanical properties
To understand the influence of size and composition of WFI on weld properties, shear-tensile and microhardness tests were done. The shear-tensile properties and microhardness of all three WFI are shown in Fig. 12.
It was observed that the rotational speed during FSSW influences the mechanical properties in two ways:
-
(i)
Thermal softening of stir zone in the samples welded at 4500 rpm causes shear failure load to reduce (Fig. 12a) and microhardness to drop below the values of base material (Fig. 13). Also, the strain hardening of the stir zone in the samples welded at 1000 rpm significantly improves the mechanical properties of the weld (see Fig. 13) which reveals the microhardness profile at the cross-section of the weld.
-
(ii)
The rotational speed also influences the size of the effective welded surface. The difference in the size of the effective welded surface is most pronounced in I and III WFI, while II WFI in both samples (welded at 1000 and 4500 rpm) are approximately the same. However, even the size of effective welded surface in I WFI for sample welded at 4500 rpm is significantly higher compared to that of 1000 rpm (Fig. 6), the shear failure load was almost the same (Fig. 12a). This was caused by higher mechanical properties of the strain hardened stir zone of the weld obtained at 1000 rpm compared to the thermal softened stir zone of the sample welded at 4500 rpm (Fig. 12b). Furthermore, the effective welded surface of the I WFI is larger than that of II WFI, but the shear failure load is higher in II WFI. The significant thinning of the first sheet due to penetration of the tool was responsible for the drop of the shear failure load in I WFI. The size of the III WFI was the smallest of all three WFI (Fig. 6), which caused the shear load to be the lowest. Also, since, III WFI in the sample welded at 1000 rpm had a larger effective welded surface (Fig. 6) and higher microhardness (Fig. 12b), the shear failure load in this case was also higher.
Figure 14 shows the evolution of tensile shear load versus the displacement. The samples welded at 1000 rpm showed a higher tensile shear load capacity for the II and III WFI, while the tensile shear load for the I WFI was almost the same. In the samples welded at 1000 rpm, the values of the tensile shear load between I and III layers were similar compared to the samples welded at 4500 rpm, where a high difference in tensile shear load and elongation was observed. This effect can be explained by the observation made by Labus Zlatanovic et al. [19] that samples welded at higher rotational speed (4000 rpm) under similar conditions revealed a number of delamination circles in the III WFI. Those delamination circles are a consequence of low stress transferred to the bottom of the weld zone due to low viscosity. Most material tends to go upwards which is why the I WFI in the sample welded at 4500 rpm has a higher effective welded surface and III WFI has the smallest one (Fig. 6). The II WFI in the samples welded at 4500 rpm showed unstable behaviour near the shear fracture load, which can also be a consequence of a number of micro and nano pits found in the II WFI during the TEM investigation (shown earlier in Fig. 8). Even when there were also some pits found in the sample welded at 1000 rpm, the samples welded at low rpm did not rupture in the middle of the WFI [10], due to a high microstrain hardening of the WFI (small grains and high volume of dislocation). Instead, the rupture occurs on the border between WFI and recrystallised stir zone, on intermetallics Al2Mg3 which seeds the nucleation for rupture.