Introduction

There is currently great interest in increasing the mass-efficiency of automotive structures by the introduction of multimaterial designs, involving aluminum joined to cheaper materials such as zinc-coated steel sheet.1 To date, resistance spot welding (RSW) has been the principle joining process in the automotive sector, as a result of its simplicity, speed, and low cost of operation. However, RSW of aluminum is problematic because of high energy costs, unstable weld quality, and short electrode life.2,3 Further, it is difficult to apply this process to dissimilar joints, as the kinetics of forming brittle intermetallic reaction layers is high in the liquid phase.3

Ultrasonic spot welding (USW) is an interesting method for joining dissimilar alloys that can potentially avoid many of the issues associated with fusion processes, including rapid intermetallic formation.4 Until recently, USW was primarily used for welding thin gauge materials in the electronics industry, requiring low power. The method was first applied in 1938, and first results were reported by Willrich in 1950.5 An important advantage of this process is that it can be used to weld dissimilar material combinations as diverse as metal/ceramic, glass/metal, Al/Cu, and Al/steel.6 The technique also requires significantly less energy than other processes such as RSW.7–9 Most research on low-power USW assumes that bonding occurs at relatively low temperatures (<300°C) and is dominated by contact mechanics.8 Sliding across the interface breaks the oxide between the two surfaces at asperities, forming microwelds, which increase in density and spread over the area affected by the vibration of the sonotrode tips. The weld strength is thus primarily related to the microbond area.9 Under these conditions, very little interfacial reaction is reported between most material combinations (e.g., Al-Cu, Al-Au,6,10 Al-Fe4,11). So far, only a few studies have been reported on joining aluminum using high-power USW of thicker gauge sheet materials for automotive applications6,12 and even fewer on joints between dissimilar metals.4,9,11 In one of the few studies published on Al to steel, Haddadi et al.11 showed that high weld strengths of ~2.8 kN can be achieved when welding Al to uncoated steel, but failure loads were limited by the formation of an interfacial reaction layer with excessive weld times. Watanabe et al.4 also obtained failure loads of 0.6 kN, between steel and A5052 Al sheet. Although a very thin intermetallic compound (IMC) interfacial reaction layer has been observed when welding uncoated steel to aluminum,4,11 there is currently little information on the role of zinc coatings on the weld performance and microstructure evolution at the weld interface in USW.

In the present investigation we studied USW of a typical aluminum automotive alloy (6111) to two different zinc-coated steel sheets, with soft (hot-dipped) and hard (galv-annealed) coatings using a 2.5-kW dual-reed ultrasonic welder. The objective of this work is to gain better understanding of the dominant factors determining joint performance, with particular emphasis on the role of the interface microstructural evolution.

Experimental Procedures

The USW spot welds in this research were performed by welding aluminum 6111-T4 to either DX56-Z (hot dipped zinc coating) or DX53-ZF (galv-annealed) steel in 1-mm gauge sheet. The results were compared with the performance of Al-Al2 and Al-uncoated (DC04) steel welds.11 Welding was carried out using a Sonobond dual-reed system operating at 20.5 kHz, under constant pressure. The welds were performed on 25 mm × 100 mm strips at the center of a 25-mm overlap following standard sample geometry.13 Ridged sonotrode tips were used for the Al and steel sides of the welds. While the aluminum tip was flat, the tip used for the steel side of the joint had a slight radius. The power was kept constant at 2.5 kW, and the welding time was varied between 0.25 s and 4.0 s under axial pressure of 2.3 kN. During welding, the thermal field was characterized by thermal imaging and the interface temperature at the edge of the weld was measured using 0.5-mm K-type thermocouples placed as close to the weld position as possible. Lap shear tensile tests were carried out for each welding condition under a constant displacement rate of 1 mm min−1. The welds were sectioned through their center and imaged using FEI Sirion and Philips XL30 scanning electron microscopes (SEMs), fitted with an HKL™ electron backscattered diffraction (EBSD) system.

Results and Discussion

As-Received Materials

Cross-sections of the coatings on the galvanized steel sheets are shown in Fig. 1, along with the phases identified. The soft hot dipped coating (DX56-Z) had a thin layer of Fe5Al5−x Zn x at the steel-zinc interface, to inhibit reaction between the zinc bath and iron, and mostly comprised a Zn solid solution. In comparison, the galv-annealed (DX53-ZF) was made up of a complex sequence of Fe-Zn intermetallic phases. In galv-annealed coatings, the sequential nucleation of Fe-Zn phases begins with zeta, followed by delta phase, and then after some time, the gamma phase.14 As seen in Fig. 3a, the gamma (Γ) layer is very thin and contains both Γ and Γ1 phases. The delta phase has a columnar appearance due to preferential growth along the {0001} basal plan of the hexagonal close-packed (HCP) structure.15

Fig. 1
figure 1

Cross-sections of the zinc coatings on the two steel sheets before welding: (a) DX56-Z and (b) DX53-ZF. The phases identified in the coatings are listed in the accompanying table

Full size image

Lap Shear Strength Performance with Weld Time

Figure 2 compares example tensile lap shear test results for Al (6111) to DX56-Z and DX53-ZF welds performed with a 2 s welding time and shows a summary of the average peak failure loads plotted against welding time (or energy) for constant power of 2.5 kW. Figure 2b clearly shows that the bimetallic welds required far higher weld energies (>3× more), with an optimum welding time of at least 1 s, compared with 0.4 s for the Al-Al weld. However, the optimum weld time was even longer for the hard zinc (DX53-ZF) and far longer for the soft zinc-coated (DX56-Z) steel weld. Further, while the hard zinc-coated and uncoated steel reached a maximum strength of around 2.7 kN, the failure load for the soft zinc-coated steel-Al welds continued to increase even after 3 s and eventually approached a level close to that of the optimum Al-Al welds9 of 3.5 kN. In comparison, Watanabe et al.4 only obtained a 0.6 kN shear strength for a welding time of 2.5 s under 0.6 kN axial pressure. The USW process can also be compared with friction stir spot welding,2 using which Fukumoto et al.16 obtained a 3.5 kN lap shear strength with a longer welding time of 5 s. Therefore, the current results suggest that USW offers a very favorable solution for dissimilar metal joining.

Fig. 2
figure 2

(a) Typical displacement-load curves from lap shear test for welding time of 2 s. (b) Effect of welding time on shear strength for coated steel-Al sheet welds with axial pressure of 2.4 kN compared with uncoated Al-steel (DC04)11 and Al-Al welds9 (Previous data for Al-Al and Al-uncoated steel welds obtained at lower pressure of 1.9 kN)

Full size image

In prior work it has been shown that, at optimum weld energies, Al-Al welds fail by weld nugget pullout, and for longer weld times the strength is limited by excessive thinning of the weld area.2 It has further been found that, when welding Al-uncoated steel, the weld strength is limited by the development of an intermetallic reaction layer, comprising mainly Fe2Al5 phase, which increases in thickness with weld time.11 Hence, interfacial reaction can have an important effect on determining the weld strength, and this point will be returned to below.

Figure 3 shows fracture surfaces of the steel sheets from Al6111-T4 to DX56-Z and DX53-ZF welds for increasing welding time. It can be observed that longer welding times lead to an increase in weld area. In the case of the hot-dipped zinc coating, after 1 s, at the center of the weld the coating has lifted off the steel surface and evidence of melting of the zinc can be seen extending outside the weld footprint. In comparison, with the galv-annealed zinc welds, only limited melting of the coating can be seen outside of the weld area, even after the 2.5 s sample, as a result of the higher melting temperature of the intermetallic phases present in the galv-annealed coat.

Fig. 3
figure 3

The steel surface of the fractured welds from lap shear tests, as a function of welding time: for the (a) DX56-Z and (b) DX56-ZF steel. (c) Temperature field as seen by thermal imaging for two weld times

Full size image

Thermal Measurements

Figure 3b depicts the temperature field seen at different weld times obtained with a thermal imaging camera focused on the steel surface. The thermograms reveal that the temperature distribution rises steeply towards the edge of the sonotrode tip and is affected by proximity to the edge of the steel sheet, which explains the asymmetric melted regions seen in Fig. 3a. Peak temperatures recorded by a thermocouple, in contact with the weld interface at the edge of the sonotrode tip, are shown in Fig. 4. The peak temperature for the coated steel welds is consistently lower than for the uncoated steel welds by about 100°C, but still reaches over 380°C by a weld time of 3 s. In addition, the harder ZF steel coat results in temperatures 10–60°C higher than for the softer, Z-coated, steel. These differences can be related to different friction conditions between the steel and aluminum sheets which, in particular, would be expected to greatly change if the zinc coating melts. Finite-element (FE) modeling suggests that the interface temperature could be ~60°C higher than in the current measurement position.13 Therefore, the weld temperatures at the join line would be expected to greatly exceed the Al-Zn eutectic melting point of 382°C17 at long welding times, but would probably be below this temperature for weld times of <1.5 s.

Fig. 4
figure 4

Peak temperatures recorded for Al to uncoated steel (DC04) and zinc-coated steel welds (DX56-Z and DX53-ZF)

Full size image

Weld Interface Microstructures and Weld Formation

From Fig. 5 it is evident that, at the initiation of welding, there is limited indentation of the sonotrode tips into the Al sheet surface. Rapid penetration of the tips then occurs as the material softens due to the rising temperature. However, compared with the extensive plastic deformation seen throughout the weld area in Al-Al welds,9 with Al-steel welds it has been found that deformation due to ultrasonic vibration, as opposed to that caused by the forging force, is localized to the weld interface region and takes place almost entirely in the softer aluminum material.11 In USW of Al-uncoated steel using the same setup, it has been previously found that welding occurs by local breakdown of oxide films at asperities on the contacting surfaces, which allows interdiffusion of aluminum and iron to occur with simultaneous formation of intermetallic reaction products, comprising mainly Fe2Al5 phase. In the case of the zinc-coated steel welds, discussed here, four typical interface structures have been observed as a function of position, coating type, and increasing weld time (selected examples of which are shown in Fig. 6), including:

Fig. 5
figure 5

Macro weld sections showing penetration of the sonotrode tip teeth into the sheet surfaces with weld time for Al to DX53-ZF welds

Full size image
Fig. 6
figure 6

Different weld interface microstructures found at the edge of the weld area, depicting (a) granulation and melting of the soft zinc coat (DX56-Z; 1.0 s weld time), (b) granulation of the ZF coating (DX53-ZF; 1 s weld time), and (c) fully melted, squeezed-out, and resolidified material from the DX56-Z steel weld at the edge of the weld area (2.5 s weld time). (d, e) Examples of the dispersed zinc coating on the DX56-Z steel at the center of the weld (2.5 s weld time). (f–h) The effect of weld time on the progressive melting behavior of the galv-annealed coating on the DX53-ZF steel, again at the weld center, with weld times of 0.25 s, 1.5 s, and 2.5 s, respectively

Full size image
  1. (i)

    Unbonded regions containing granulated particles from the zinc coat (Fig. 6a, b)

  2. (ii)

    Areas effectively brazed by thick resolidified eutectic Al-Zn liquid (Fig. 6c)

  3. (iii)

    Regions where partial dispersal of the zinc coat occurs by formation of a liquid film and incorporation of zinc into the aluminum sheet by grain boundary penetration (Fig. 6f, g)

  4. (iv)

    Areas where near-full dispersal of the zinc coat takes place by formation and ejection of eutectic liquid and incorporation of zinc into the aluminum sheet with extensive penetration along flow features and grain boundaries (Fig. 6d–h).

Of these, microstructures (i) and (ii) occur near the edge of the weld with increasing weld time. Initially, wear debris is found concentrated at the edge of the weld area (Fig. 6a). However, an Al-Zn eutectic film develops with increasing temperature, and at longer weld times this is replaced by resolidified eutectic liquid, which is also squeezed out of the joint (Fig. 6c). In comparison, structures (iii) and (iv) are seen at the weld center with increasing weld time (Fig. 6d–h). Further, the zinc coating from the hot-dipped steel melts much more quickly, so that the wear debris contains a substantial volume of liquid (Fig. 6a), and at the weld center stage (iv) is reached much more rapidly. In contrast, the harder and more brittle Fe-Zn intermetallic coating on the galv-annealed steel melts more slowly and progressively as the temperature rises. In this case, unmelted coating fragments can be found near the edge of the weld for short weld times (Fig. 6b), and it takes longer for the coating to become dispersed at the center of the weld (Fig. 6f–h).

In Fig. 7 the ingress of zinc into the aluminum sheet is highlighted. EBSD analysis of the boundary misorientations and EDS mapping confirms that zinc can penetrate distances as high as 100 μm into the aluminum sheet along high misorientation boundaries. Given the short weld cycle, this depth of infiltration is too large for a solid-state diffusion process and is caused by penetration of Al-Zn eutectic liquid. In lower-power USW, it has also been noted that Zn can penetrate high-angle boundaries (>15°), although this was attributed purely to diffusion.18 Further, it is possible that the high vacancy flux generated by the high dynamic strain rate (~103) seen in USW can increase diffusion rates, as well as lower the melting point.18

Fig. 7
figure 7

(a) SEM image and (b) EDS elemental map showing the penetration of Zn along grain boundaries in the aluminum sheet. The boundary misorientations measured by EBSD are presented in (a)

Full size image

It is clear from the above observations that the zinc coating is partially abraded and damaged during the early stages of welding and then melts and is both squeezed out and dispersed into the aluminum sheet by diffusion and penetration of Al-Zn eutectic liquid along high-angle grain boundaries. This provides a clean surface for metallurgical bonding. However, the presence of a low-melting-point liquid film suggests that welding is facilitated partly by brazing. It is interesting that the hot-dipped soft zinc coat melts much more rapidly than the galv-annealed coating. This behavior is simply related to the structure of the coatings (Fig. 1), with the former predominantly made up of a dilute Zn solid solution and the second, several layers of different Fe-Zn intermetallic compounds. In the ZF coating each layer is dominated by a different phase and will thus melt at different temperatures (Fig. 6f–h), making it a slower and more energetically difficult process to fully disperse the coating.

Finally, it is important to correlate the above observations to the weld’s mechanical performance (Fig. 2). For the hot dip-coated steel welds, it is apparent that rapid melting of the zinc coating reduces friction, and dissipates energy, delaying weld formation relative to an uncoated steel weld. The welds progressively improve in strength as the zinc coating melts, and is squeezed out and dispersed within the aluminum sheet. An image showing the fracture path is presented in Fig. 8a, which shows that the aluminum separates at the steel-Al interface where the Al-Zn eutectic film is concentrated and the fracture may be associated with the very thin Fe5Al5−x Zn x layer at the base of the coating. In comparison, with the galv-annealed coating, a lower peak strength is achieved, which decays at high weld energies. For this material combination, the fracture path again follows the steel coating interface, but in this case the remaining unmelted intermetallic layer is much thicker. There is also evidence of cracking of the grain boundaries within the steel substrate. Although we do not currently have direct evidence to explain the loss of weld strength at greater weld energies with this coating, it is possible that the welding process damages this complex intermetallic layer, facilitating crack propagation.

Fig. 8
figure 8

Fracture path for optimized welds produced between Al-6111 and (a) DX56-Z steel and (b) DX53-ZF

Full size image

Conclusions

The weldability of galvanized automotive steel sheet to aluminum with soft (hot-dipped) and hard (galv-annealed) coating has been explored, using a 2.5-kW USW welder. Results show that soft hot-dipped zinc (DX56-Z)-coated steel results in better weld performance than hard (galv-annealed) zinc coatings (DX53-ZF), but for longer weld times. For Al to hard galv-annealed-coated steel welds, lap shear strengths reached a maximum of ~80% of the strength of an Al-Al joint after a 1.0 s welding time. In comparison, welds between Al6111-T4 and hot-dipped soft zinc-coated steel took longer to achieve the same maximum strength, but matched the Al-Al joint properties.

Both coatings were found to melt and disperse during welding. However, in the early stages of joining, granulated wear debris was found within the joint, particularly with the harder galv-annealed coating, which contains a range of Zn-Fe intermetallic compounds and is thus more brittle. The soft hot-dipped coating melts very rapidly owing to the low-melting-point Al-Zn eutectic reaction, which reduces the energy dispersed at the weld interface, and is squeezed out of the joint as well as infiltrating into the aluminum sheet along high-angle boundaries. In comparison, the galv-annealed coating melts in stages at higher temperatures because it comprises layers of intermetallic phases with different individual melting points.