H.Peng, D.L.Chen, X.F.Bi, P.Q.Wng, D.Y.Li, X.Q.Jing

a College of Engineering and Technology, Southwest University, Tiansheng Road 2, Beibei District, Chongqing 400715, China

b Department of Mechanical and Industrial Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario M5B 2K3, Canada

c School of Materials and Energy, Southwest University, Tiansheng Road 2, Beibei District, Chongqing 400715, China

d Department of Chemical and Materials Engineering, University of Alberta, Edmonton, Alberta T6G 1H9, Canada

Abstract Lightweight ZEK100-O Mg alloy and Al6022-T43 Al alloy with an Ag interlayer were joined via ultrasonic spot welding(USW),focusing on the microstructural change and tensile lap shear strength of the welded joints in relation to welding energy.Mg/Al interface was superseded by Mg/Ag and Al/Ag interfaces, and unfavorable Mg17Al12 intermetallic compound was eliminated.Ag foil was observed to be intact in the nugget center, while it was broken or dissolved at the nugget edge at high welding energy levels.The diffusion layer at the Mg/Ag interface consisted of two distinctive sub-layers: Mg3Ag intermetallic compound adjoining Ag foil, and Mg3Ag+Mg eutectic structure adjacent to Mg.Only a thin diffusion layer consisting mainly of Ag3Al occurred at the Al/Ag interface.The tensile lap shear strength firs increased,reached its peak value, and then decreased with increasing welding energy.The shear strength achieved in the present study was ～31%higher than that of the joint without interlayer.Interfacial failure occurred at all energy levels, with Ag foil particles or fragments being stuck on both Mg and Al sides due to its intense interaction with Mg and Al via accelerated diffusion during USW.The results obtained pave the way for the challenging dissimilar welding between Mg and Al alloys.

Keywords: Magnesium alloy; Aluminum alloy; Ultrasonic spot welding; Ag interlayer; Microstructure; Tensile lap shear strength.

1.Introduction

With the global crisis in energy source and environmental protection [1-5], lightweight materials and multi-material design approach have been adopted extensively by the transportation industry to manufacture structural parts [6].Magnesium and aluminum alloys, which have good characters such as low density, high specifi strength, good machinability and recyclability, are being employed to substitute heavier mild steels for improving fuel efficien y and reducing environmental damage [7-11].Therefore, hybrid structures with magnesium and aluminum alloys have become essential[12-19];this inevitably demands a suitable method for welding magnesium and aluminum alloys.A number of welding techniques, such as resistance spot welding (RSW), friction stir spot welding(FSSW) and ultrasonic spot welding (USW), have been studied [12-24].USW is an environment-friendly solid-state joining technique.It has been reported to attain defect-free joints of lightweight magnesium and aluminum alloys with superior strength compared with fusion welding processes[20-24].However, a layer of brittle intermetallic compounds (IMCs)between magnesium and aluminum alloys still occurred at the weld interface quickly even using USW [20-25].It has been reported that the thickness of IMCs was a key factor for the bonding strength and the thick IMCs would significantl reduce the joining strength [12,14,20-24,26].Therefore, it is a big challenge to control the growth of IMCs to increase the welding strength between magnesium and aluminum alloys.

Fig.1.(a) Schematic diagram of a ZEK100-Al6022 dissimilar lap joint with an Ag interlayer, and (b) a 2.5kW dual-wedge reed Sonobond MH2016 HP USW system.

The approaches used to improve the strength of magnesium-to-aluminum dissimilar welded joints include the application of shorter welding time, lower temperature and fille material addition.One of the advantages of USW is its short welding time (～1s or shorter per spot weld) along with the relative low interfacial temperature (lower than ～500°C)[20,27,28],which can be used to better control the IMC thickness during dissimilar welding.The other effective means is to add a fille material, which can influenc the type, amount and distribution of IMCs at the weld interface.Several studies on the USW of magnesium alloys such as AZ31 to aluminum alloys have been done and found that the joining strength of the Mg/Al dissimilar welded joints could be improved by adding an interlayer [20,22,29,30].Panteli et al.[29] studied the microstructure and mechanical properties of USWed AZ31/Al6111 dissimilar joints with Al coating, and observed that the thickness of IMCs was significantl reduced, and the fracture energy was twice higher than that without coating.Patel et al.[20] studied the microstructure and mechanical properties of USWed AZ31/Al5754 without and with a Sn interlayer.They found that Mg17Al12was substituted by an Mg-Mg2Sn eutectic layer at the interface of Mg and Al alloys, and the strength of AZ31/Al5754 dissimilar joints with a Sn interlayer was higher than that without Sn interlayer.Dai et al.[22] studied arc assisted ultrasonic seam welding of AZ31B/Al6061 joints with a Zn interlayer,and also showed that the joint strength with a Zn interlayer was higher than that without Zn interlayer due to the fact that the IMC phases of Al12Mg17and Al3Mg2were successfully replaced by MgZn2-containing eutectic.Dai et al.[30] also studied arc assisted ultrasonic seam welding of Mg/Al joints with Sn/Zn composite interlayer, and found that the joining strength of AZ31B/Al6061 with Sn/Zn composite interlayer was ～150% as high as that without Sn/Zn composite interlayer.It should be noted that the Sn or Zn interlayer added during USW has a melting point below that of aluminum and magnesium alloys.Also, most researchers selected the common AZ31 magnesium alloy sheet in the Mg-to-Al dissimilar welding via USW.Only limited studies on the USW of a relatively newer ZEK100 Mg alloy (without aluminum and with a minimal addition of RE element, i.e., only 0.2wt%Nd) to aluminum alloys [21,23,28], although it exhibited superior room temperature formability [31] and fatigue resistance [32].Peng et al.[28] studied the microstructure and tensile lap shear strength of USWed ZEK100/Al6022 with a Cu interlayer,which has a melting point(1085°C)higher than that of aluminum and magnesium alloys.The results demonstrated that the Al12Mg17was substituted by the Al2Cu at the Al side and Mg+Mg2Cu at the Mg side, and the strength of ZEK100/Al6022 with a Cu interlayer was ～135% that without a Cu interlayer.This confirme that the addition of a higher melting point fille material is capable of generating favorable results during USW.Based on the binary phase diagrams of Mg/Ag and Al/Ag, Ag as an interlayer is also possible.Wang et al.[33]studied microstructure and mechanical properties of diffusion-bonded Mg-to-Al joints using silver (Ag) fil as an interlayer, they found that the strength of pure magnesium to pure aluminum with silver interlayer was also higher than that without silver interlayer.However, to the authors’knowledge,no information about the dissimilar USW of Mg-to-Al alloys with a silver interlayer has been reported in the open literature.In this study, Al6022 alloy with lower alloying elements, better formability and corrosion resistance than 6016 and 6111 aluminum alloys was chosen, which was frequently used as interior and exterior sheets in the automotive industry[34].Therefore,it is necessary to study the weldability of ZEK100/Al6022 dissimilar joints with an addition of Ag interlayer using USW, focusing on the microstructural evolution at both Mg/Ag and Ag/Al interfaces and mechanical properties in relation to welding energy.

2.Materials and experimental procedure

As shown in Fig.1(a), commercially available 2.0mm thick sheets of ZEK100-O Mg alloy and 1.3mm thick sheets of Al6022-T43 alloy cut into strips with a dimension of 80mm in length and 15mm in width were used as base metals in this study.The chemical composition of both alloys was shown in Table 1.Before welding, the test coupons was ground with 120 grit sand paper to roughen the surface,cleaned with ethanol and dried using compressed air to keep surface consistency and avoid potential oxides on the interface.During USW, the ZEK100 Mg strip was put on the top and the Al6022 strip at the bottom, and a 90μm thick pure Ag foil (interlayer) of 99.9% purity with a rectangular shape of 20mm in length and in 15mm width was placed inbetween the magnesium and aluminum strips, after the similar surface treatment to the test coupons.The dissimilar joints of USWed ZEK100/Al6022 with an Ag interlayer were made using a 2.5kW dual-wedge reed Sonobond-MH2016 HP ultrasonic spot welder in an energy-control mode at a frequency of 20kHz as shown in Fig.1(b).Both upper and lower sonotrode tips, having a dimension of 8mm×5mm with nine parallel teeth to keep good grasping of the top and bottom strips,were positioned at the center of a 20mm long overlapped section with the vibration direction perpendicular to the rolling direction as shown in the inset in Fig.1(b) enclosed by a red box.The welding parameters selected in this study were tabulated in Table 2.The welding energy (Q), welding power (P) and welding time (t) follow such a relationship:Q≈P×t.For example, a welding power is 2.0kW and a welding time of 1.5s correspond to a welding energy of 3000J.

Table 1Chemical composition (wt.%) of ZEK100-O Mg alloy and 6022-T43 Al alloy.

Table 2Welding parameters selected in the present study.

Four samples were welded in each welding condition.One of them was used for the microstructural examinations, and the other three were used for the tensile lap shear tests.The selected samples welded at 750J, 1250J and 2000J for observations via scanning electron microscopy (SEM) were sectioned along their center and parallel to the vibration direction with a slow diamond cutter, and then prepared by cold-mounting with epoxy, followed by mechanical grounding using abrasive papers (grit #320, #600, #1000 and #2000)and polishing with diamond paste (6μm, 3μm and 1μm)and colloidal silica (0.5μm).The tensile lap shear tests were performed at room temperature using a fully computerized United tensile machine at a constant crosshead speed of 1mm/min.To avoid the specimen bending moment and rotation, two spacers of 15mm×35mm (width×length) with a thickness corresponding to that of the other end of the USWed specimen were added at both specimen ends during mechanical tests.The selected typical tensile failed samples were examined via a JSM-6380LV SEM equipped with Oxford energy dispersive X-ray spectroscopy (EDS), electron backscatter diffraction (EBSD) and 3D surface/fractographic analysis capacity.To identify the phases generated during USW, X-ray diffraction (XRD) was performed on both matching fracture surfaces after the lap shear tensile test with Cu Kαradiation having a wavelength of 1.5406 °A at 45kV and 40mA.The diffraction angle (2θ) at which the X-rays were incident on the samples varied from 20° to 100°, with a step size of 0.05°and 2.0s in each step.

Fig.2.Typical SEM macroscopic images of the cross-section of USWed ZEK100/Ag/Al6022 dissimilar joints made at an energy of (a) 750J, (b)1250J, and (c) 2000J.

3.Results and discussion

3.1.Microstructural evolution

Typical SEM macrographs of the cross-section of ZEK100/Ag/Al6022 welded joints at a welding energy of 750J, 1250J and 2000J are shown in Fig.2.It is seen that a joined and unbroken Ag foil was observed in the nugget center.However, as indicated by the yellow circles, the Ag foil was broken or dissolved at the nugget edge; in the case of 2000J the complete dissolution of Ag foil occurred.This was mainly associated with the difference of deformation resistance between base metal and welding zone, which led to a higher stress concentration, Ag foil bending and breaking with fast rubbing at the nugget edge.These observations were in agreement with the results reported in[20,23,35-39],where the more severe plastic deformation and stress concentration happened at the nugget edge.The higher temperature induced by further more severe plastic deformation and longer welding time at a higher welding energy of 2000J resulted in the dissolution of Ag foil.This was likely as a consequence of the occurrence of eutectic reaction between Ag and Mg and between Ag and Al at an equilibrium eutectic transformation temperatures of 472°C (Ag-Mg phase diagram) and 567°C (Ag-Al phase diagram), respectively, not to mention the presence of melting-point depression which occurs due to either the reduced particle size or under high strain-rate and high-frequency dynamic deformation [40-42].

Fig.3.Typical SEM images at the Mg/Ag interface of ZEK100-Ag-Al6022 dissimilar joints along with EDS line scan and point analysis (in at.%) at a welding energy of (a,b) 750J, (c,d) 1250J, and (e,f) 2000J.

Typical SEM micrographs of the cross-section of ZEK100/Ag/Al6022 welded joints at an energy of 750J,1250J and 2000J are shown in Figs.3 and 4, alone with the relevant results of EDS line scan and point analyses.Due to the implantation of Ag foil, the Mg/Al interface was superseded by two interfaces of Mg/Ag and Al/Ag, where the interfacial defects such as cracks, voids and pores were not observed as shown in Figs.2-4.However, due to the ultrafast or accelerated diffusion during USW, the interfacial microstructures at Mg/Ag interface and Al/Ag interface were different, which are discussed as follows.

3.1.1.Mg/Ag interface microstructure

The SEM images of the typical Mg/Ag interface of ZEK100/Ag/Al6022 joint welded at an energy of 750J,1250J and 2000J are shown in Fig.3(a), (c) and (e), respectively.Fig.3(b), (d) and (f) show the EDS line scan results corresponding to the location of the red dashed lines in Fig.3(a),(c) and (e).It is seen that a thin and non-uniform Mg/Ag diffusion layer, with an average thickness of ～2μm appeared at a high magnificatio of 5000×at a low energy of 750J in Fig.3(a).This was related to the short welding time (0.37s),the relatively low temperature and non-uniform distribution of heat.In this case, the inter-diffusion between Ag and Mg atoms was limited compared with the situation between Mg and Al.Ag atoms diffused to Mg base below 390°C(Do(Ag-to?Mg)=3.65×10?4m2/s)was far lower than Al to Mg(Do(Al-to?Mg)=1.20×10?3m2/s), which led to the thickness of Mg/Ag interface being thinner than Mg/Al interface at the same energy [23,33,43].Meanwhile, due to the micro-scale uneven surface of welding strips by the roughening of grinding, the frictional heat firs occurred at the initially contacted asperities between the two strips,where the microscopic bulge emerged, which would undergo a relatively longer rubbing time and higher temperature so called “hot spots”.Therefore,somewhat non-uniform distribution of diffusion layer at the interface occurred.To identify the composition distribution,EDS point analyses were carried out in the Z1 and Z2 regions, as indicated by the red boxes in Fig.3(a), where an approximate composition of 78.9 at% Mg, 20.1 at% Ag and 1.0 at% Zn, and 49.8 at% Mg, 48.7 at% Ag and 1.5 at% Zn was detected in the Z1 and Z2 regions, respectively.Based on the Mg-Ag binary phase diagram, the results would indicate the onset for the formation of Mg3Ag intermetallic compound[44].The Mg3Ag intermetallic compound was also detected during diffusion bonding at 380°C [33].According to the binary phase diagram, Mg and Al were not easy to form solid solution, while the solid solution would be formed at the Mg/Ag interlayer, as reported in [33,43].

Fig.4.Typical SEM images at the Al/Ag interface of ZEK100-Ag-Al6022 dissimilar joints along with EDS line scan and point analysis (in at.%) at a welding energy of (a,b) 750J, (c,d) 1250J, and (e,f) 2000J.

With the energy increasing to 1250J, a continuous and thicker diffusion layer with an average thickness of ～10μm and two distinct zones named Z3 and Z4 were observed in Fig.3(c).This was associated with the longer welding/rubbing time (0.63s) and more severe plastic deformation, which led to a higher temperature and thus a larger extent of diffusion between Ag and Mg atoms.Z3 zone near the Mg side with a higher Mg content (93.4 at% Mg, 6.0 at% Ag and 0.6 at%Zn) would reflec a localized liquid or mush region during USW due to the formation of eutectic structure consisting of Mg3Ag+Mg stemming from the occurrence of eutectic reaction which would occur at an equilibrium eutectic temperature of 472°C [44].Such a non-equilibrium eutectic structure formed at high cooling rates is sometimes called as pseudoeutectic structure (also referred to as divorced eutectic structure [45-47]).Z4 zone adjacent to Ag side in Fig.3(c) with a composition of 72.9 at% Mg, 26.2 at% Ag, and 0.9 at% Zn would just correspond to Mg3Ag intermetallic compound.It is worth noting that due to the presence of “melting-point depression”phenomenon caused by the high-frequency and high strain-rate rubbing during USW [29,38,40-42,48,49], the temperature for the formation of intermetallic compound and/or eutectic structure would be expected to be lower than the equilibrium eutectic temperature of 472°C.Recent study on the dissimilar Al-Cu USW via molecular dynamics simulations of atomic diffusion revealed that the diffusion during USW was a dynamic and unsteady process with a diffusion coefficien much higher than that of a steady-state diffusion process despite the lower interfacial temperature due to the shear plastic deformation and high strain rate at the weld interface at the interface [50].EDS line scan results in Fig.3(d)also show that the content of Ag element across the Mg/Ag interface near the Mg side firs increased gradually in layer I, then formed a plateau where both Ag and Mg elements co-existed in a certain proportion in layer II, and then the content of Ag element near the Ag side increased sharply to the pure Ag level and the content of Mg element decreased rapidly to zero.This also indicated that two different regions in the diffusion layer of Mg/Ag interface were present.This result was different from that of USWed ZEK100/Al6022 dissimilar joints reported earlier in [23], where there was a layer including Mg17Al12intermetallic compound in the diffusion layer of Mg/Al interface, and no sub-layer was observed.When the welding energy was increased to 2000J, a further thicker interfacial diffusion layer with a thickness of up to～30μm, could be observed in Fig.3(e).Compared with the Mg/Ag interface of the USWed joint made at the energy of 1250J, both thickness and morphology of the diffusion layer have significantl changed, with a much wider distribution of Mg3Ag+Mg eutectic structure in Z5 zone, where a composition of 94.9 at% Mg, 4.7 at% Ag and 0.4 at% Zn was obtained.In the meantime,Mg3Ag intermetallic compound region adjoining to Ag foil (Z6 with a composition of 77.2 at%Mg, 22.2 at% Ag and 0.6 at% Zn) became also wider but in the form of porous structure, as if it was partly swallowed by the eutectic structure (Z5).This would provide evidence for the formation of deformation-induced vacancies due to further higher strain rate and more severe plastic deformation with increasing energy and longer welding time (1.0s), which also led to stronger interatomic diffusion along with a longer diffusion distance.EDS line scanning results shown in layers III and IV in Fig.3(f) also revealed a non-uniform distribution of Ag elements across the Mg/Ag interface especially adjacent to the Ag foil as a result of the presence of porous intermetallic compound.Such an asymmetric non-equilibrium diffusion during USW was associated with the lower melting point of Mg in comparison with Ag, the difference in the atomic radius (a smaller value of Ag atom (0.144nm) vs.Mg atom (0.160nm)), a big difference of thermal conductivity between Mg and Ag (the thermal conductivity of Ag(429W/m·K) versus ZEK100 Mg alloy (96W/m·K)), and the much lower specifi heat capacity of Ag (232J/kg K) compared with that of Mg alloy (1024J/kg K) [25,28,51].It follows that the heat generated by high-frequency friction and deformation during USW would mainly sustain at the Mg side instead of Ag side.A similar phenomenon was also observed during the USW of ZEK100/Cu/Al6022 [28].From the above observations shown in Fig.3(a)-(f) it could be concluded that: i) the overall diffusion layer across the Mg/Ag interface consisted of two distinctive sub-layers: Mg3Ag intermetallic compound adjacent to Ag foil, and the mixed Mg3Ag+Mg eutectic structure adjacent to Mg; ii) the Mg3Ag intermetallic compound appeared first and then the Mg3Ag+Mg eutectic structure was formed by fragmenting Mg3Ag intermetallic compound as a consequence of high-frequency rubbing and the related accelerated and asymmetric diffusion during the USW of dissimilar materials; iii) while the thickness of both sub-layers increased with increasing welding energy, the eutectic structure grew faster and the Mg3Ag intermetallic compound became a porous structure;iv)there was no very brittle Mg17Al12intermetallic compound detected.Mg17Al12had a relative lower melting point than Mg3Ag, and thus would be easier to form cracks and pores [33,52].

3.1.2.Al/Ag interface microstructure

Typical SEM images of Al/Ag interface diffusion layer of ZEK100/Ag/Al6022 joint welded at an energy of 750J,1250J and 2000J are shown in Fig.4(a), (c) and (e), respectively.Fig.4(b), (d) and (f) show the EDS line scan results corresponding to the red dashed lines in Fig.4(a), (c) and (e),respectively.The irregular interface, suggesting the mechanical interlocking being one of the main bonding mechanisms,could be clearly seen at the Al/Ag interface in Fig.4(a), (c)and (e) due to the non-uniform severe plastic deformation caused by the different deformation resistances of Ag and Al and local “hot spots” as mentioned above.A similar result was observed during USW of ZEK100/Cu/Al6022 and ZEK100/Zn/Al6061 [22,28].As shown in Fig.4(a), the diffusion layer of the dissimilar joint made at 750J was very thin with a thickness of ～0.5μm, which could be observed only at a high magnificatio of 6000×.This was related to the insufficien diffusion between Ag and Al atoms due to the short welding time(0.37s)and low interface temperature.The EDS line scan in Fig.4(b) also showed slight interdiffusion of Ag atoms and Al atoms, reflectin the inception of a diffusion layer to form at the Al/Ag interface.With increasing welding energy and welding time to 1250J and 0.63s, respectively, a diffusion layer with a thickness of ～1μm was observed in Fig.4(c) due to the higher temperature and more time of diffusion between Ag and Al atoms.When the welding energy and welding time increased to 2000J and 1.0s, respectively,the atomic kinetic energy and inter-diffusion were intensified leading to a longer diffusion distance and thus a thicker diffusion layer of ～1.5μm, as seen from Fig.4(e).EDS point analysis was performed in the middle of the layer, showing a composition (in at.%) of 65.1Ag, 33.7 Al, 0.7Zn and 0.5 Si,respectively, which indicated the formation of a new phase between Ag and Al.

Fig.5.X-ray diffraction patterns obtained from both matching fracture surfaces of (a) Mg side and (b) Al side after the tensile lap shear test of a welded joint made at a welding energy of 1250J.

It is worthwhile noting that local melting might not occur at the Al/Ag interface,since the eutectic temperature is nearly 100°C higher for the Al-Ag system(567°C)than for the Mg-Ag system (472°C).In order to identify the phases occurred at the interface, XRD was conducted on the both matching fracture surfaces of the welded joint made at an energy of 1250J after the tensile lap shear test, and the obtained results are shown in Fig.5.It is seen that in addition to the peaks of Mg, Al, and Ag, the phases of Mg3Ag and Ag3Al could be observed at both Mg and Al sides of the joint.This suggested the presence of Mg3Ag at the Mg/Ag interface (either in the form of thin and porous intermetallic compound or within the eutectic structure as shown in Fig.3) and Ag3Al diffusion layer at the Al/Ag interface as shown in Fig.4.The XRD results corresponded nicely to the above EDS analyses and the relevant phase diagrams.

3.2.Tensile lap shear strength

The tensile lap shear strength and failure energy of the joints with and without an Ag interlayer made at varying energy levels from 500J to 2000J were shown in Fig.6(a)and (b), respectively.The tensile lap shear strength and failure energy firs increased, reached a peak value and then decreased with increasing welding energy in both cases.While the ZEK100-Al6022 dissimilar joints without an Ag interlayer reached the peak value earlier at 750J, the peak tensile lap shear strength was lower due to the formation of brittle Al12Mg17intermetallic compound [20,21,23,24,42,53,54].With the addition of Ag foil, both peak values of the tensile lap shear strength and failure energy became higher in spite of a higher welding energy of 1250J.This is understandable: the higher strength was achieved due to the elimination or reduction of Al12Mg17intermetallic compound, and the higher welding energy required for the peak value was attributed to the higher melting point of Ag (～962°C) relative to the ZEK100 Mg alloy and Al6022 Al alloy.This was consistent with our previous results when a higher melting point (～1085°C) copper interlayer was implanted in-between ZEK100 Mg alloy to Al6022 Al alloy during USW [28].It is worthwhile to mention that unlike the present situation, when the added metal interlayer has a lower melting point with respect to two dissimilar sheets to be welded, such as tin(～232°C) and zinc (～420°C), the welding energy required to arrive at the peak tensile lap shear strength was lower, as reported in[20,21,53].It should be noted that the standard deviations of the lap shear stresses at different welding energy levels were different, ranging from 3.0 to 5.6MPa.

Fig.6.(a)Tensile lap shear strength and(b)total failure energy as a function of welding energy of ZEK100-Al6022 dissimilar joints with and without an Ag interlayer.

The change of the tensile lap shear strength and failure energy with welding energy directly reflect the microstructural change at the Mg/Ag and Ag/Al interfaces of ZEK100-Ag-Al6022 dissimilar joints as shown in Figs. 3 and 4. At the energy levels of 500J and 750J, due to the relatively low interdiffusion rate of Al and Ag below 390°C like the condition reported in [33,43], i.e., Do(Ag-to-Al)=1.18×10?5m2/s,Do(Al-to-Ag)=1.30×10?5m2/s, and a relatively high eutectic reaction temperature between Al and Ag compared with that between Mg and Ag, inadequate bonding happened with a low density of micro-joints at the Al/Ag interface. Failure happened at the Al/Ag interface instead of Mg/Ag interface at the lower energy level of 750J, as seen from the inset in Fig. 6(a), leading to a lower value of lap shear strength and failure energy, which was even lower than that of ZEK100-Al6022 USWed joints without any interlayer. At the energy of 1000J and 1250J, the severer plastic deformation, longer welding time and higher temperature resulted in a more intimate contact, more microwelds and more intense diffusion at both Mg/Ag and Al/Ag interfaces. The moderate thickness of Mg3Ag diffusion layer and the existence of α-Mg+Mg3Ag eutectic structure could improve the joining strength at the Mg/Ag interface, and the thin Ag3Al diffusion layer along with solid solution of Al and Ag could also improve the bonding strength at the Al/Ag interface [23,27,28,55]. In this case, failure could occur at both Mg/Ag interface and Al/Ag interface as shown in the inset in Fig. 6(a) and also corroborated by the XRD results shown in Fig. 5, and the tensile lap shear strength and failure energy reached their maximum average values of ～62MPa and 2J, respectively, being ～31%and ～26% higher than those of the USWed ZEK100-Al6022 joints without any interlayer.

When the welding energy further increased, e.g., to 2000J,the diffusion layer consisting of two-sublayers at the Mg/Ag interface increased considerably, becoming excessively thick(Fig.3(e)), although the diffusion layer at the Al/Ag interface remained still fairly thin (Fig.4(e)).In this case, the overly-thick diffusion layer could deteriorate tensile lap shear strength and failure energy, thus giving rise to low average values of ～46.0MPa and 0.9J, respectively, at the energy of 2000J.However,they were still higher than those of ZEK100-Al6022 joints without any interlayer made at the same energy of 2000J.This suggests that the elimination of Mg17Al12intermetallic compound in welding magnesium alloys to aluminum alloys via an interlayer addition is beneficial even in the case of the formed diffusion layer exceeding its desirable thickness.

A comparison of the average peak tensile lap shear strengths of the dissimilar joints made with different Mg and Al alloys with different interlayers via USW is illustrated in Fig.7 [18,20,22,29,30,56].In this study, the strength of the dissimilar joints, with the addition of Ag interlayer with a higher melting point than Al and Mg alloys, was higher than that of the dissimilar joints with the addition of other interlayers such as pure Al cold spray coatings, Zn, Sn and Sn/Zn layers with a lower or similar melting point compared with Al and Mg alloys.It should be noted that the tensile lap shear strength of USWed joints was assessed by the failure load dividing by the welding tip area, which was used by other researchers as well [35,39,57-62].This could be understood from the following points: (1) The brittle IMCs like Mg17Al12and Mg2Al3could be effectively eliminated due to the presence of Ag interlayer with a higher melting point,which could not be easily melted completely during USW.(2) According to the binary phase diagram, a Ag-based solid solution layer of Mg and Al could improve the bonding and strength.(3) The unique double-layer structure like Mg3Ag intermetallic compound adjoining Ag foil and Mg3Ag+Mg eutectic structure adjacent to Mg would lead the IMCs relatively more dispersed to improve the joining strength.(4) As reported by [21,23,32], ZEK100 magnesium alloy containing Nd and Zn and having superior ductility and formability could result in a better bonding and a higher strength with an Al alloy via USW, compared with AZ31 magnesium alloy.Indeed, the tensile lap shear strength achieved in the present study was much (about four-fold) higher than that of the dissimilar Mg-Al joints via diffusion bonding also with an Ag interlayer, where the maximum shear strength was reported to be 14.5MPa at a bonding temperature of 390°C [33,43].

Fig.7.A comparison of the average maximum tensile lap shear strengths of the dissimilar joints made with different Mg and Al alloys with different interlayers via USW.

3.3.Fractography

All the tensile lap shear joints of ZEK100/Ag/Al6022 failed in the same mode of interfacial failure, which indicated that the imposed stress surpassed the critical shear stress during the tensile lap shear test.In order to understand why the peak value of tensile lap shear strength occurred at a welding energy of 1250J, the matching fracture surfaces of the USWed ZEK100/Ag/Al6022 dissimilar joint made at a welding energy of 1250J after the tensile lap shear test were examined via SEM, with typical images shown in Fig.8.The overall view of the entire fracture surface on the Mg and Al sides is shown in Fig.8(a) and (b).Two zones could be seen on the Mg side in Fig.8(a).One was a white zone enclosed by a dashed red curve,where the fractured Ag foil was stuck on the fracture surface of Mg side, and the other was a grey zone, where the selected red-boxed area “c” was magnifie in Fig.8(c).To identify the elemental distribution in the grey zone, EDS analysis was conducted in the yellow-boxed area labelled M1 in Fig.8(c), and showed a composition (in at.%) of 80.3 Mg, 19.1 Ag, and 0.6Zn.This revealed the occurrence of intense interactions between Ag foil and Mg matrix via accelerated diffusion during USW, leading to the formation of Mg+Mg3Ag eutectic structure and the subsequent failure at the Mg/Ag interface during the tensile lap shear test.As shown in Fig.8(b), three zones could be seen at the Al side,where the selected red-boxed zone“d”,“e”and“f” are magnifie in Fig.8(d), (e) and (f), respectively.EDS analysis performed at M2 in Fig.8(d), which was located on the Ag foil being flippe at the Al/Ag interface (Fig.8(b)),showed a composition (in at.%) of 94.3 Al, 5.1 Ag and 0.6 Mg, suggesting that the Al matrix was pulled out to stick on the Ag foil, and thus leading to a higher tensile lap shear strength.EDS point analyses carried out at M3 and M4 in Fig.8(e) gave a composition (in at.%) of 57.8 Mg, 41.8 Ag and 0.4Zn; 98.1 Mg, 1.5 Ag and 0.4Zn, indicating the existence of some local MgAg intermetallic compound and Mg matrix being pulled out to stick on the Al side at the edge of weld nugget (Fig.8(b)).Meanwhile, the non-uniform distribution of Ag stuck on the Al side was observed in the nugget zone in Fig.8(f) being a magnifie image of area “f”Fig.8(b).While some larger white Ag particles were stuck on the Al side, many small greyish particles or fragments could be seen in Fig.8(f), as revealed by the EDS analysis at M5 with a composition (in at.%) of 93.8 Al, 5.7 Ag and 0.5 Mg.The black area in Fig.8(f) was basically Al, as demonstrated by the EDS analysis at M6.Compared Fig.8(a, c)with Fig.8(b, e and f), it is clear that Ag foil was present on both Mg and Al sides in the form of white and greyish particles/fragments under the ultrasonic high-frequency vibration.Such a mixed failure characteristic suggested robust bonding between Mg and Al alloys with Ag foil as an interlayer,and thus provide a superior tensile lap shear strength in the present study, as discussed above.

Fig.8.Typical SEM images of tensile lap shear fracture surfaces of an USWed ZEK100-Ag-Al6022 dissimilar joint made at a welding energy of 1250J, (a) overall view on the Mg side; (b) overall view on the Al side;(c) the magnifie image from the red-boxed area “c” in (a); (d), (e) and(f) the magnifie images from the red-boxed areas “d”, “e” and “f” in (b),respectively.(For interpretation of the references to color in this figur legend,the reader is referred to the web version of this article.)

To further understand why the peak value of the tensile lap shear strength occurred at a welding energy of 1250J instead of 750J, the matching fracture surfaces of the USWed ZEK100/Ag/Al6022 dissimilar joint made at a welding energy of 750J after the tensile lap shear test were examined via SEM, with typical images shown in Fig.9.It is easy to observe the Ag being stuck on the Mg side instead of the Al side from the overall view of the entire fracture surface on the Mg and Al sides being shown in Fig.9(a) and (b).The fracture mainly occurred at the Al/Ag interface.The selected red-boxed zones “c” and “e” in the Fig.9(a) are magnifie in Fig.9(c)and(e).EDS analyses were conducted in the yellowboxed areas labelled M7 in Fig.9(c)and M10 in Fig.9(e),and the results also verifie the very thin Ag/Al interface present at this energy due to its lower diffusion rate between Ag and Al atoms as discussed above at a short welding time (0.37s)and a low interface temperature, as shown in Fig.4(a).The EDS point analysis was also conducted at M11 in the dark zone circled by a green dashed line, which was located at the edge of weld zone experiencing a stress concentration as described earlier.It is seen that the amount of Ag (14.9 at%Ag) in this dark zone was much lower than that in the outside area (e.g., 47.4 at% Ag in M10), meaning that the fracture happened at the Mg side in this dark zone.This corresponded well to the EDS results of the yellow-boxed areas labelled M8 and M9 in Fig.9(d).Few Ag particles could be seen in Fig.9(f), indicating that the lap shear strength was relative weak.

The matching fracture surfaces of the USWed ZEK100/Ag/Al6022 dissimilar joint made at a welding energy of 2000J after the tensile lap shear test were examined via SEM, with typical images shown in Fig.10.The overall view of the entire fracture surface made at the welding energy of 2000J on the Mg and Al sides was shown in Fig.10(a) and (b), respectively.Almost the entire fracture surface was grey on the Mg side in Fig.10(a).The area of blue-circled localized mush(or melting) zone in Fig.10(a) could be seen.To see more details of fracture morphology and elemental distribution, the red-boxed zone “c” in Fig.10(a) was further magnifie in Fig.10(c), where the sub-micron particles could be clearly observed.As discussed above, during welding a part of Ag foil would break into fin particles and dissolve due to the fast repeated friction, high-rate deformation and stress concentration at the edge of welding zone (Fig.2), and the Mg alloy would be melted at the initially contacted asperities.The rapid solidificatio of the melted alloys would lead to the formation of sub-micron particles, which were revealed via an EDS analysis of yellow box M14 in Fig.10(c) to be Al-containingε-Mg3(Ag, Al) particles.This was also reported in Mg-Ag-Al joints during diffusion bonding [63].The EDS analyses were conducted in the yellow-boxed areas labelled M17 and M18 in Fig.10(e) with a similar result, indicating the existence of Mg+Mg3Ag eutectic structure and reflectin the occurrence of failure at the Mg/Ag interface.A crack indicated by the blue arrow was also seen in Fig.10(e), which suggested a lower bonding strength.The scrubbing lines along rolling direction were observed in the center of weld zone on the Al side (Fig.10(d)).To better see the “river valley”-like patterns and the elevation and depression on the fracture surface, a three-dimensional image was taken from the red-boxed zone“d” in Fig.10(b), which is shown as an inset in Fig.10(d).As seen from Fig.10(f), the Al matrix would be pulled out in zone M19.Such a mixed failure would corroborate the higher bonding strength at the Al/Ag interface.

Fig.9.Typical SEM images of tensile lap shear fracture surfaces of an USWed ZEK100-Ag-Al6022 dissimilar joint made at a welding energy of 750J, (a)overall view on the Mg side; (b) overall view on the Al side; (c) and (e) the magnifie images from the red-boxed areas “c” and “e” in (a); (d) and (f) the magnifie images from the red-boxed areas “d” and “f” in (b), respectively.(For interpretation of the references to color in this figur legend, the reader is referred to the web version of this article.)

Fig.10.Typical SEM images of tensile lap shear fracture surfaces of an USWed ZEK100-Ag-Al6022 dissimilar joint made at a welding energy of 2000J, (a) overall view on the Mg side; (b) overall view on the Al side; (c)and (e) the magnifie images from the red-boxed areas “c” and “e” in (a);(d) and (f) the magnifie images from the red-boxed areas “d” and “f” in(b), respectively.(For interpretation of the references to color in this figur legend, the reader is referred to the web version of this article.)

4.Conclusions

USWed ZEK100/Ag/Al6022 Al dissimilar joints were successfully achieved to eliminate the unfavorable Mg17Al12intermetallic compound at varying welding energy levels.The microstructures at the Mg/Ag and Al/Ag interfaces and the tensile lap shear strength were characterized.The conclusions can be drawn as follows.

(1) An incorporated and unbroken Ag foil was observed in the nugget center, while the Ag foil was broken or dissolved at the nugget edge with the extent increasing with increasing welding energy.Complete dissolution of Ag foil occurred at a welding energy of 2000J.

(2) The diffusion layer across the Mg/Ag interface consisted of two distinctive sub-layers: Mg3Ag intermetallic compound adjacent to Ag foil, and the mixed Mg3Ag+Mg eutectic structure adjacent to Mg.While the thickness of both sub-layers increased with increasing welding energy, the eutectic structure grew faster and the Mg3Ag intermetallic compound became a porous structure.

(3) The tensile lap shear strength and failure energy firs increased with increasing welding energy up to 1250J,where they reached their peak values of ～62MPa and 2J, respectively, which were ～31% and ～26% higher than those of the USWed ZEK100/Al6022 joints without interlayer.With further increasing welding energy,the tensile lap shear strength and failure energy decreased.

(4) Interfacial failure occurred at varying energy levels.Ag foil was observed to stick on both Mg and Al sides.Such a mixed failure characteristic indicated robust bonding between Mg and Al alloys with an Ag interlayer, thus exhibiting a superior tensile lap shear strength.

Acknowledgments

The authors would like to thank the National Natural Science Foundation of China (Grant No.51971183), Natural Sciences and Engineering Research Council of Canada(NSERC), Fundamental Research Funds for the Central Universities (XDJK2018B108, SWU119065) and Venture and Innovation Support Program for Chongqing Overseas Returnees(CX2018082) in the form of international research collaboration.The authors thank Professor A.A.Luo, Ohio State University (formerly General Motors Research and Development Center) and Dr.T.Skszek, Magna International Inc., for the supply of test materials.H.Peng is grateful for China Scholarship Council for providing a PhD student scholarship.The authors would also like to thank F.Mokdad, A.Macwan,Q.Li, A.Machin, J.Amankrah and R.Churaman for easy access to the laboratory facilities of Ryerson University and their assistance in the experiments.