Influence of zinc on intermetallic compounds formed in friction stir welding of AA5754 aluminium alloy to galvanised ultra-high strength steel

Mechanical properties of friction stir welded joints

Figure 2 shows the influence of translational speed on the tensile shear failure load of the strip samples for the three surface coating conditions. All types of steels showed maximum shear strength at the moderate translational speed (80 mm min⁻¹) with similar variability. Fully strong joints can be obtained when AA5754 and 14-µm-galvanised steel are produced at a translational speed of 80 mm min⁻¹, two of which fractured in the aluminium base material sheet. As seen from Figure 2, the 14-µm-galvanised joints have obviously higher failure loads at the translational speed of 40 and 80 mm min⁻¹ than the other coating conditions. The failure loads of both galvanised steel samples become more similar as the translational speed increases. In contrast, the failure load of the uncoated steel is significantly reduced as the translational speed increases. The differences in failure load between the uncoated and 8-µm-galvanised steels are very narrow in the range of translational speed between 40 and 80 mm min⁻¹ as can be seen from the overlapping error bars. It is noted that the joint with uncoated steel usually performs a poor performance especially without abrasive cleaning of the steel surface.

OPEN IN VIEWER

Figure 2

Maximum tensile force of FSW joints between AA5754 and DP1000 for different translational speeds and coating conditions.

Microstructure of AA5754/DP1000 joint

Figure 3 shows a typical cross-section of a joint welded at the translational speed of 80 mm min⁻¹. It is seen that Al and steel can be successfully welded by FSW without tool penetration into the steel. Figure 4(a)–(c) presents typical microstructures in the stir zone (SZ) of joints welded at 80 mm min⁻¹ (position C in Figure 3). A reaction layer can be seen at the interface between aluminium and steel for all three surface coating conditions. The EDS analysis confirmed that the chemical compositions of this reaction layer consist of the elements Al and Fe. Two intermetallic phases are observed which measurements revealed to be Al₁₃Fe₄ on the Al side and Al₅Fe₂ on the steel side in all three cases. The major part of the reaction layer is Al₁₃Fe₄ for both galvanised steels as indicated in Figure 4(a,b). Our previous study, also on galvanised steel, indicated that the thickness of the Al₁₃Fe₄ layer is highly dependent on translational speed (as result of heat input) while the thickness of the Al₅Fe₂ layer is rather constant [12]. In addition, this study found that the Al₁₃Fe₄ and Al₅Fe₂ layers are of similar thickness on the uncoated steel (see Figure 4(c)). It is noted that the results appear to contradict previous studies in which the Al₅Fe₂ phase is the major constituent of the reaction layer [6, 13, 14]. The discrepancy can be attributed to a detection of the tongue-like interface between the Al₅Fe₂ phase and the steel in literature compared to the smooth interface in this study (see Figure 4(b)). The smooth interface into the steel means the growth of the intermetallic layer only took place towards the aluminium alloy. This can correspond to different temperature history and shorter reaction time relative to the literature.

OPEN IN VIEWER

Figure 3

Typical cross-section of joint between AA5754 and DP1000.

OPEN IN VIEWER

Figure 4

Back scattered electron images of cross-sections between AA5754 and different steels at various translational speeds (a) 8-µm-galvanised steel at 80 mm min⁻¹ (b) 14-µm-galvanised steel at 80 mm min⁻¹ (c) uncoated steel at 80 mm min⁻¹ (d) 8-µm-galvanised steel at 160 mm min⁻¹ (e) 14-µm-galvanised steel at 160 mm min⁻¹ (f) 14-µm-galvanised steel at 3 mm from the weld line on advancing side at 160 mm min⁻¹ (g) uncoated steel at 160 mm min⁻¹.

EDS measurements of cross-sections of joints between AA5754 and the two galvanised steels are shown in Table 3. Presence of zinc in the aluminium matrix with 8-µm-galvanised steel was detected as 0.73 and 0.86 wt-% (Point 1 and 4) at translational speeds of 80 and 160 mm min⁻¹, respectively, while the aluminium matrix with 14-µm-galvanised steel contained 5.16 and 1.16 wt-% Zn for the corresponding conditions. It can be seen that the amount of zinc that can be dissolved into the aluminium matrix in the case of HDG steel having the 8-µm-galvanised layer is rather constant, regardless of translational speed, and is also lower than for the 14-µm-galvanised layer. In contrast, the level of dissolved zinc for the thicker galvanised layer (14 µm) is obviously dependent on translational speed.

OPEN IN VIEWER

Table 3

EDS analysis results showing chemical compositions at different positions.

	Element (wt-%)
Position	Al	Fe	Zn	Mg	Possible phases
1	95.65	0.70	0.73	2.93	Al+Zn
2	80.06	11.36	5.59	2.99	Al+Zn
3	90.88	0.97	5.16	3.00	Al+Zn
4	95.43	0.85	0.86	2.86	Al+Zn
5	95.07	0.86	1.16	2.91	Al+Zn

Microstructures of different regions of joints between AA5754 and galvanised steels are presented in Figure 5(a–d). It is seen that at position B, the microstructure consists primarily of Zn–Al solid solution (white zone) and a notable amount of α + η phase (grey zone). This solidified structure is a result of the lateral extrusion of molten Al–Zn alloy due to the forging pressure of the shoulder. A comparison between Figure 5(a) and (c) demonstrates that melting of galvanised layer for the 14-µm-galvanised steel occurred at a smaller extent of the surface area than that of the 8-µm-galvanised steel. This is because the thicker galvanised layer requires a higher amount of heat generated during welding for melting of the zinc layer. The joining processes can be regarded as so-called transient liquid phase bonding which are divided into four stages [15]: (i) diffusion of Al/Zn in a solid state (ii) melting of zinc layer (iii) dissolution of Al into molten zinc or molten Zn–Al eutectic and (iv) isothermal solidification. The high volume fraction of η phase at position B suggests the zinc layer liquefies through melting of pure zinc. This is followed by the dissolution of Al into the liquid phase and possibly solidified by hypoeutectic solidification. In addition, diffusion of Zn into the aluminium matrix can be observed in Figure 5(b). One possible explanation is that the liquid phase has isothermally solidified which allows a short time for diffusion of zinc into the aluminium.

OPEN IN VIEWER

Figure 5

Back scattered electron images of cross-sections along interface at translational speed of 80 mm min⁻¹ (a) 8-µm-galvanised steel at point A (b) 8-µm-galvanised steel at point B (c) 14-µm-galvanised steel at point A (d) 14-µm-galvanised steel at point B.

At a translational speed of 80 mm min⁻¹, porosities can be seen near the interface in the case of the 8-µm-galvanised layer (see Figure 4(a)) but not for the 14-µm-galvanised layer (see Figure 4(b)). The elongated shape of porosities and their location close to the IMC layer can degrade the joint strength in tensile shear loading where both shear and peel loads are involved. The zinc content inside the porosity (point 2) in Figure 4(a) is detected to be 5.59 wt-% which is higher than its presence in the aluminium matrix in joint with the 8-µm-galvanised layer (0.73 wt-%). This indicates that the porosity can be a result of the partial liquid phase in the aluminium matrix with 8-µm-galvanised steel which has a larger extent of the liquid phase compared to the 14-µm-galvanised steel. However, when the translational speed increases to 160 mm min⁻¹, the tensile failure load of the 14-µm-galvanised steel seems to be lower than that of 8-µm-galvanised steel as seen in Figure 2. The microstructure also shows that the joint with the 14-µm-galvanised layer has a high level of voids compared to the 8-µm-galvanised layer, as shown in Figure 4(d,e). One indication is that the solubility of zinc in the aluminium matrix goes down from 5.16 wt-% at 80 mm min⁻¹ to 1.16 wt-% at 160 mm min⁻¹ for the 14-µm-galvanised steel, while the percentage of dissolved zinc for the 8-µm-galvanised steel is rather the same. This can be ascribed to an accumulation of molten zinc on the advancing side (3 mm from the weld line) in the 14-µm-galvanised steel joined at 160 mm min⁻¹, as seen in Figure 4(f). The molten zinc seems to be the result of a stagnant flow of material which may occur on the advancing side due to counter flow line by actions of tool translation and rotation [16]. When the translational speed is high, it is not possible to dissolve a large amount of zinc into the aluminium matrix due to the shorter processing time [17], with lower heat input and less material circulation cycle/translational distance. Therefore, it can be inferred that the thickness of the galvanised layer plays a crucial role in percentage of dissolved zinc in the aluminium matrix depending on translational speed. However, the characteristics of the intermetallic layers are not different between two galvanised steels.

Role of the zinc coating layer on the interfacial layer

It is seen from Figure 4(b,c) that the overall thickness of the reaction layer does not differ significantly between uncoated and galvanised steel. However, the relative phase fractions of the reaction layers are different. That is to say, a greater extent of Al₅Fe₂ phase can be detected in the joint with the uncoated steel than in the joint with the galvanised steel. One possible reason given by Springer et al. in the case of zinc additions is that the interdiffusion of Al and Fe through the Al₅Fe₂ could be facilitated by an interaction of Zn atoms with structural vacancies of the Al₅Fe₂ phase [14]. However, more work is necessary to investigate this mechanism.

Discontinuities of the reaction layer can be observed at the interface of aluminium to uncoated steel at 160 mm min⁻¹ in Figure 4(g). This indicates that the frictional heat input is not sufficient to activate the atomic bonding between Al and Fe at the high translational speed (low heat input). The small bonding area can be detected on the fracture surface as seen in Figure 6(a). In contrast, both galvanised steels can generate the continuous atomic bonding as indicated by the reaction layer (see Figure 4(d,e)). Fracture surfaces of joints with both galvanised steels are shown in Figure 6(b,c). The joint with 14-µm-galvanised layer contains a large area of solidified zinc as indicated in Figure 6(b). The fracture surface shows its brittleness at the zinc-rich area which might be the reason for the reduction in failure load of the joint with 14-µm-galvanised steel compared with 8-µm-galvanised steel as the translational speed increases.

OPEN IN VIEWER

Figure 6

Secondary electrons images of fracture surfaces of joints with different steel type at 80 mm min⁻¹ (a) uncoated steel (b) 14-µm-galvanised steel (c) 8-µm-galvanised steel.

Several research works have indicated that the presence of zinc does not affect the stability of the Al–Fe intermetallic phase formation [14, 18, 19]. Also, the discrete reaction layer in uncoated steel indicates limitations in nucleation as being the main reason for the partial absence of IMC phases. It is well accepted that zinc can create a more intimate contact between the aluminium matrix and steel due to the lower melting temperature of the Al–Zn solid solution [9]. This can result in an increased possibility to initiate intermetallic nucleation at low temperature which is not always possible without zinc due to impediments such as surface contaminations and residual oxide films.

XRD analysis and fracture path

Figure 7(a–d) presents XRD spectra of fracture surfaces of dissimilar joints at translational speed of 80 mm min⁻¹. Two Al–Fe intermetallic phases can be identified for both uncoated and hot-dipped galvanised steels including Al₅Fe₂ and Al₁₃Fe₄. No ternary Fe–Al–Zn phases were observed in any conditions. The results are in a good agreement with the EDS analyses of the cross-sections mentioned previously. In addition, Fe₂Zn₁₀ can be found on the fracture surfaces of joint with hot-dipped galvanised steels which might be a residual product from the hot-dipped galvanising process [20]. It is seen from Figure 7(a,b) that the Al₅Fe₂ phase are detected on both the aluminium and steel sides in the case of uncoated steel. This is in contrast to the Al₁₃Fe₄ phase which could only be found on the aluminium side, as seen in Figure 7(b), indicating that the fracture path seems to propagate though the Al₅Fe₂ phase in the joint with uncoated steel. This is, however, different from hot-dipped galvanised steels as both Al₅Fe₂ and Al₁₃Fe₄ can be detected on both aluminium and steel sides (see Figure 7(c,d)). The latter results indicate that the crack propagates along the interface between Al₅Fe₂ and Al₁₃Fe₄ or propagates tortuously in both these phases. This is to say as one of the IMCs thickens relative to the other, the fracture path tends to shift into that phase and propagate through it. It is noted that both Al₅Fe₂ and Al₁₃Fe₄ are Al-rich intermetallics which are typically brittle with hardness levels reaching 1100 and 820 HV, respectively [8].

OPEN IN VIEWER

Figure 7

XRD spectra of fracture surfaces between AA5754 and DP1000 under different coating conditions at 80 mm min⁻¹ (a) uncoated steel on iron side (b) uncoated steel on aluminium side (c) 8-µm-galvanised steel on iron side (d) 8-µm-galvanised steel on aluminium side.