Effect of Process Parameters on Wavy Interfacial Morphology During Magnetic Pulse Welding

The demand for lightweight structures has been increasing for decades in various industries in order to reduce energy consumption. Selecting the perfect material for every specific component is one means to achieve lightweight structures; however, traditional fusion welding processes cannot be used for such dissimilar materials applications due to differences in material characteristics and the tendency to form brittle intermetallic compounds. Daehn [1] showed that high-speed impact welding is a viable method for joining dissimilar materials, and high speed forming and welding have additional advantages including reduced springback, decreased wrinkling, improved formability, increased dimensional accuracy, absence of heat affected zone, and potentially a stronger weld seam than parent welding partners. In addition, Zhang et al. [2] described surface contaminants and oxides of both welding partners are ejected off during the welding process to prepare clean surfaces for the high pressure contact, i.e., a jetting phenomenon.

Various high-speed, impact welding processes have been investigated by researchers for decades. Laser impact welding (LIW) was proposed by Daehn and Lippold [3]. An intense, pulsed laser irradiates into an ablative layer that absorbs the large amount of energy and is vaporized into a high-pressure plasma, which accelerates the workpiece. Welding between various material combinations, including AA1100/AA1100 [2], AA1100/low carbon steel [4], and aluminum/copper [5], were successfully achieved by LIW. Numerical studies of the LIW process were also conducted by Wang et al. [6].

For large-scale workpieces, explosive welding (EXW) is used where, instead of an intense laser pulse, chemical explosives to drive the workpiece [7]. Numerous studies of EXW have focused on the interfacial morphologies between different metal combinations including Ti/Ti [8], Ti/steel [9], Mo/Cu [10], and Ti/stainless steel [11]. Furthermore, Ege et al. [12] introduced a third material as an interlayer for Al/Ti, Hokamoto et al. [13] used a stainless steel plate as an interlayer for aluminum alloys/stainless steel material couples, and the microstructure-property relationship in EXW was reported by Kacar and Acarer [14].

Vaporizing foil actuator welding (VFAW) was developed by Daehn’s group [15]. Vivek et al. [16] then proposed the weldability window of VFAW for the material combination of CP-Ti and Cu110. An analytical model for VFAW was also presented by Hahn et al. [17] to calculate workpiece velocity based on the gas particle velocity and fully clamped beam bending theory. Effects of welding partner thicknesses and internal stress waves on the interfacial morphology were investigated for Al1100/AISI1018 [18], and CP-Ti/Cu110 material combinations [19]. VFAW has only been used for sheet metal welding thus far. Alternatively, magnetic pulse welding (MPW) is available for both sheet and tubular welding (see schematic in Fig. 1 [20]) using magnetic pressure to drive the workpiece. A large amount of energy (typically on the order of tens of kilojoules) in the form of a sinusoid damped current trace is dissipated from a capacitor bank into a specially designed coil, which produces a magnetic field. Eddy currents are induced in any electrically conductive workpiece in proximity to the coil. Owing to these two opposing currents and magnetic fields, Lorentz forces are generated between the workpiece and coil to accelerate the flyer workpiece to several hundred meters per second. If a stationary target workpiece is impacted at a critical angle and impact velocity, a solid state weld is created. As with other impact welding processes, various process parameters have been investigated for MPW. For example, the effect of flyer kinetics was reported by Lueg-Althoff et al. [21] for an Al6060-C45 material couple. Raoelison et al. [22] studied the welding condition and proposed a weldability window for MPW of the Al6060-Al6060 system [23]. An analytical model was also proposed by Lueg-Althoff et al. [24] to predict the impact velocity of MPW based on the magnetic pressure and verified by experimental observations. Kinsey and Nassiri [25] analytically predicted flyer shape profiles after deformation for tubular parts and sheet parts [26].

With appropriate process parameters, a wavy interfacial morphology is commonly obtained at the interface of the two impact welded partners. Although intact straight weld seams have been reported in the literature, these interfacial waves indicate a sound weld. Figure 2 shows interfacial morphologies for different material systems for various impact welding processes. Some wavy interfacial morphologies are so pronounced that vortices are observed. There is still no consensus with respect to the formation mechanism of such wavy interfacial morphologies among researchers due to the complex interfacial kinematics. Ben-Artzy et al. [30] attributed the wave formation to the elastic stress waves caused by the impact in MPW. Blazynski [31] used the same theory to explain the wave formation in EXW. Ben-Artzy et al. [32] believed that the melting layer and solidification cause intermetallic phases at the interface, which contribute to the wavy morphology.

Analyses with respect to intermetallics between aluminum and copper alloys, a popular material combination, were studied by previous researchers. Wu and Shang [33] identified Al/Cu intermetallics with different atomic ratios and discussed the mechanism of MPW. Raoelison et al. [34] investigated the intermetallic formation effect on welded features of Al/Cu joint and compared these results with an Al/Al joint. In addition, the formation of such features was also discussed. Microstructure evolution at the interface between Al and Cu sheets was studied by Itoi et al. [35]. Atomic-scale bonds were formed between the intermediate layer and sheet surfaces. An amorphous alloy was observed at the interface and joint strength was tested as well in their study. Psyk et al. [36] discussed Al and steel tubes joined by MPW and analyzed the effect of inner tube support on the joint quality in terms of the interface microstructure and weld strength. Experimental observations were supported by numerical simulations conducted. Experimental investigation of interfacial microstructure for another material combination, i.e., two stainless steel sheets, was studied by Liu et al. [37]. Shear-induced features such as nanoscale elongated grains and adiabatic shear bands were among their findings. Numerical analysis based on SPH method also supported the experimental results in terms of local temperature and strain distribution predictions.

Analytical modeling has also been conducted to investigate high velocity, impact welding. Chemin and Qingming [38] presented a hydro-plastic model to explain the interfacial wave formation by means of a high shear rate deformation. The most common theory is a Kevin-Helmholtz instability, i.e., Deribas and Zakharenko [39] explained a hydrodynamic instability which occurs between the two surfaces of the welded partners and causes the wavy interface as well as jetting [40]. Hunt [41] also explained the wave formation in EXW by using this theory. Numerical studies based on the Kevin-Helmholtz instability to describe wave formation have also been reported by Mousavi and Al-Hassani [42]. Nassiri et al. [43] also explored a wave formation mechanism based on this theory using temporal stability analysis.

Instead of conducting experimental tests in laboratories or considering analytical models of the process, numerical simulations provide an alternative technique to investigate the high speed, impact welding process for parameters that are difficult to measure experimentally or capture analytically. However, finite element analyses (FEA) for impact welding processes are challenging due to the highly dynamic nature of the process and large local deformation in the vicinity of the welding zone. Therefore, using the traditional Lagrangian FEA method is not feasible due to excessive element distortions at the interface. Alternatively, multiple methods including Arbitrary Lagrangian–Eulerian for Al6061T6/Al6061T6 material system, smoothed particle hydrodynamics (SPH) for Al6061T6/Al6061T6 [29], CP-Ti/Cu110 [44], and Ti/Cu material combinations [45], and Eulerian methods for Al/Al [46], and CP-Ti/Cu110 [44], were used to investigate the interfacial morphology for high speed impact welding.

For the present paper, SPH was applied. It is a meshless method that is able to handle large deformation problems. The SPH method was traditionally used in fluid dynamics modeling. Applications of SPH were extended to solid mechanics [47] after addressing problems with respect to tensile instability, lack of consistency in material deformation, and accuracy of solutions [48] so, e.g., structural crash problems could be simulated effectively. In the SPH method, continuum materials are represented by particles, each of which has a spatial location in the model. The particles interact with each other through an interpolation kernel function with a characteristic length known as the smoothing length [49]. The physical properties of each particle are affected by neighboring ones by summing up their relative characteristics. Nearby particles have more influence on the physical properties than distant ones.

In this study, the influence of the impact velocity, target thickness, and mandrel inclusion on the interfacial morphology during MPW for a material combination of Al6060-T4 and Cu-ETP was investigated. The wavy interface was considered as an indication of an optimum weld strength and metallurgical bonding for successful welds. In addition, numerical simulations using the SPH method were conducted to compare the interfacial morphologies. Multiple process parameters were numerically obtained, and the metal jet composition was studied. The hypothesis tested in this research is that a “well-supported target,” i.e., either a thick target or the support of a mandrel, allows for vortices to be created at the interface during MPW provided that the impact velocity is sufficient.

Experimental tests were conducted on a Poynting SMU 0612 FS machine with an eight-turn compression coil and field shaper. The coil featured an inner diameter of 97 mm and an axial length of the active area of 90 mm. The field shaper, which was made of CuCr1Zr, had an inner diameter of 41 mm and the length of its pressure concentration zone was 10 mm. Figure 3 shows the experimental area and setup in the laboratory at TU Dortmund University, Germany. A detailed description of the experimental conditions can be found in Ref. [50]. Photon Doppler velocimetry (PDV) was used to measure flyer velocities on the outer surface of the tube, so a combination of holes and pockets was drilled into the field shaper to provide a line of sight for the PDV laser probes [51]. The working length l_w on the flyer was approximately 6 mm. The flyer tube material was Al6060T4 with an outer diameter of 40 mm and thicknesses of 1.5 and 2 mm for various tests. The target tubes were Cu-ETP with thicknesses of 2, 4, and 6 mm. The gap distance between the flyer and target were 1.5 mm for specific tests in group 4, and 2 mm for other tests (see Table 1). Figure 4 gives a schematic of the experimental setup focusing on the pressure concentration zone of the field shaper. To investigate the effect of a third inner component, i.e., mandrel, on interfacial morphology, additional experiments were conducted. The mandrel was made of polyurethane (PUR) with a Shore hardness of 92A and pre-stressed via a washer and a nut for specific tests.

Various tests were conducted and divided into five comparison groups to investigate the influence of process parameters on interfacial morphology. Table 1 gives the basic information of each test including wall thicknesses of both tubes, impact velocities, and mandrel inclusion. In group 1, flyer and target tubes wall thicknesses remained constant, mandrel tube was not included, and charging energies were 4, 4.75, and 5.25 kJ which correspond to impact velocities of 290, 306, and 340 m/s, respectively. The target thickness effect was tested in group 2, while the same charging energy was used resulting in an impact velocity of 250 m/s and still no mandrel was included. In group 3, the target thickness effect was tested with a higher impact velocity. Groups 4 and 5 investigated the influence of a mandrel, which was included into the target tube to reduce the target’s radial deformation during the impact process.

Numerical simulations were conducted in the commercial software package ls-dyna. There are multiple choices of formulations for SPH approximation in ls-dyna to improve the result accuracy and reduce the computational time including 2D and axisymmetric assumptions. For this research, a simplified numerical model was set up in LS-PREPOST (see Fig. 5(b)). Two or three parts were modeled depending on whether a mandrel was used. The initial impact angle was fixed between the flyer and target at 12 deg based on the geometry, i.e., gap and working length, of the experimental setup, but this is a dynamic angle which will change slightly during the numerical simulations. But, only a small portion of the interface was considered in the numerical analyses, so this was deemed acceptable. Others have reported a larger change in angle, which this is process specific [53]. Experimentally, the impact angle was created instantaneously as the flyer collided with the target (see Fig. 5). The length for all of the parts in the model was 10 mm. Various thicknesses and initial velocities were defined in order to investigate their effects. Materials for the tubes were Al6060T4 and Cu-ETP for the flyer and target, respectively. As per the experiments, the mandrel material was PUR with a Shore hardness of 92A. The left edges of the target (and mandrel in groups 4 and 5) were fixed as boundary conditions to imitate the experimental setup. The effect of this boundary condition as opposed to other options, e.g., fixing the bottom of the target or mandrel, was confirmed to not affect the simulation results substantively. A fine SPH particle size of 5 µm was defined in the model in order to capture the interfacial morphology. Interactions through a normal interpolation method were possible between SPH parts since the material densities have the same order of magnitude [54]. The total number of particles in the model ranged from 300,000 to 1,400,000 depending on the tube thickness and mandrel inclusion.

The cross section of a welded interface is shown in Fig. 6 for a selected test from group 3 with an impact velocity of 340 m/s. Flyer and target thicknesses were 1.5 and 2 mm, respectively. At the beginning of the impact, a small portion of flat interface is observed. Then, the interfacial instability initiated and caused a stable wavy morphology to evolve in the welding direction. In a portion of the wavy interface, a relatively consistent wavelength was obtained (see Figs. 6 and 7). An amplified view of Fig. 6 is shown in Fig. 7(a). With the same impact velocity and flyer thickness, vortices occurred at the interface for a larger target thickness of 6 mm (see Fig. 7(b)). Thus, target thickness affects the wavy interfacial morphology as predicted in the hypothesis. More details are discussed below for the various experimental cases conducted. Voids and intermetallic phases were also observed owing to the elevated temperature at the interface caused by local, large plastic deformations. This phenomenon was correctly predicted in numerical simulations as shown in Figs. 7(c) and 7(d). The increased temperature only surpassed the melting point of the flyer, which is consistent with the experimental results where seemingly voids and intermetallic phases are present on the Al6060T4 side of the weld.

The average wavelength and amplitude in both experimental and numerical results for all the tests conducted are shown in Table 4. Wavelengths, λ, were averaged by measuring four waves as shown in Fig. 7. Every process parameter tested in this study affected the wavy interfacial morphology in terms of wavelength, amplitude, and vortices formation.

With other parameters kept constant, a higher impact velocity led to a larger wavelength. This is consistent with past results in the literature. Also, a thicker target tube and the inclusion of a mandrel also resulted in an increase in the wavelength and amplitude. Furthermore, the vortices occurred at the interface for cases with a well-supported target, i.e., larger target thickness or thin target with mandrel inclusion, when a high impact velocity was applied. The amplitude of waves did not change with an increase of impact velocity in experimental observations. However, numerical simulations predicted the amplitude to remain the same (10 μm) for the two cases with lower velocities, but increased to 15 μm for the simulation with an impact velocity of 340 m/s. The amplitude did not change for a lower impact velocity (250 m/s) when the target thickness was increased from 2 to 4 mm for both experiments and simulations. But in group 3 with a higher impact velocity, when the target thickness changed from 2 to 6 mm, the amplitude increased and vortices were formed for the thicker target case. In groups 4 and 5, a similar trend was observed. With a lower impact velocity, there was little change in amplitude with or without mandrel inclusion. But a larger amplitude was obtained and vortices were formed when a mandrel was included and a high impact velocity was provided. Again, the results support the hypothesis that a “well-supported target,” i.e., either a thick target or the support of a mandrel, allows for vortices to be created at the interface during MPW provided that the impact velocity is sufficient.

The wavy interfacial morphology is believed to result from a Kelvin–Helmholtz mechanism, which is caused by a velocity difference between the two impact partners, i.e., the flyer and target. In addition, shock waves, caused by the impact, travel in both metals away from the impact surface and reflect back off surface interfaces. Therefore, the wavelength is affected by the target thickness, as the propagation distance of reflected shock waves increases with target thickness. Mandrel inclusion provides a similar effect as target thickness increase, since shock waves travel and reflect from the mandrel surface.

Other researchers found that the vortices formation was caused by large shear plastic deformation at the interface and interface instability, if a high-level collision pressure was maintained for a sufficient period of time [42]. Finally, a well-supported target contributed to the difference of the velocity and shear between the joining materials. As the interface evolved, the local target material depressed at the interface when the difference of velocity and shear was sufficient. A vortex was formed at the interface and continued evolving as the depressed region became larger [61]. Thus, the vortices occurred in the cases with a well-supported target for the same impact velocity.

Figure 8 shows the interfacial wavelength trend with respect to the impact velocity and target stability in 3D graphic diagrams. Well-supported targets were achieved by target thickness and mandrel inclusion in Figs. 8(a) and 8(b), respectively. As shown in Fig. 8(a), higher wavelength was obtained with a higher impact velocity and a well-supported target, i.e., higher target thickness. In Fig. 8(b), a high wavelength was also observed with a well-supported target by mandrel inclusion, but now with a lower impact velocity. The reason for this effect could be the difference in flyer thicknesses, as the flyer with lower thickness was used in the case of 310 m/s impact velocity. Thus, stress waves traveled a shorter distance in the flyer.

The relatively small impact velocity difference in group 1 could be the reason that amplitudes changed only slightly in the simulations and remained constant in the experiments. The simulations provide a more idealistic representation of such process variations and thus may demonstrate trends not observed experimentally. In groups 2 and 4 for varying thicknesses and mandrel inclusion, respectively, vortices were not observed and amplitudes were constant as the impact velocities were relatively low. Thus, these process variations only slightly changed the impact conditions and final wavy interfacial morphologies. Changes in the wavelength though were observed perhaps simply due to the scale of the values (on the order of 100 μm) compared with those of amplitude (on the order of 10 μm). In contrast, for groups 3 and 5, the same type of process variations produced vortices and increases in both wavelength and amplitude when a higher velocity (>300 m/s) was used as will be discussed in Sec. 4.2.

Note that there is some subjectivity in the measurements of wavelength and amplitude and defining whether vortices occurred or not. Also, the quantified values for the wavelength and amplitude from the experiments and the numerical simulations did not match exactly. The trends though were evident as seen in Fig. 8, which provides confidence in the numerical results obtained.

Vortices formation was only observed in two cases (see Table 4 and Fig. 7(b)). Both cases were conducted with a relatively high impact velocity of over 300 m/s. But a 6 mm target thickness was used in group 3, and a mandrel was included inside a 2 mm thick target in group 5. In both tests, a comparably well-supported target was provided, and vortices were formed with the appropriate impact velocities for this material combination. Figure 9 shows comparisons of target deflections at the same time increment for both groups 3 and 5. The boundary condition in the numerical analyses was consistent with the experimental setup. The left edge of the target was fixed, and the other edges of the target were free to move during the impact process. Thus, the impact between the weld partners led to the target deflection. Smaller target deflections were obtained for cases where vortices formed, resulting in a more consistent impact angle during the impact process, 12.54 deg and 12.11 deg for group 3 and 5 with well-supported targets, respectively (see Figs. 9(c) and 9(d)), compared with tests with larger target deflections where the impact angle tends to become smaller, 10.97 deg and 11.10 deg with 2 mm targets for group 3 and 5, respectively (see Figs. 9(a) and 9(b)). The interfacial wave morphology occurred if the appropriate impact velocity and angle were present. Note that the interface was always flat at the beginning of the impact due to the small impact angle. In groups 3 and 5 with well-supported targets, larger impact angles were observed in these cases with small deflections, which contributed to the wavelength increase and vortices formation. The numerical observations of deflections and changes in impact angle provide some physical rationale for the variations in interfacial morphologies observed as MPW is known to be sensitive to this key process parameter.

The jetting phenomenon was captured well by using the SPH method. Figure 10 shows the numerical result for the case with the impact velocity of 340 m/s and 2 mm target thickness. The velocity of the metal jet was approximately 2 km/s. This value was numerically found to be dependent on impact velocity, but independent of the target thickness. Increasing the impact velocity led to a higher velocity of the metal jet for the same given impact angle (group 1), since the metal jet velocity can be estimated by the contact pressure [45], which is correlated to the impact velocity. Further, the composition of the metal jet was dependent on material properties. The ejected material was mostly composed of the welding partner with the lower density, in this case, Al6060T4 (see Fig. 10 and Table 2). This numerical finding was consistent with the results using an Eulerian formulation method for the numerical simulations [44].

To verify the numerical result with respect to the ejected material composition being the material with the lowest density, another material with lower density compared with Cu-ETP, i.e., CP-Titanium, was used in the simulations for either flyer or target material. The results also showed that the ejected materials were mainly CP-Titanium for the material combination of CP-Titanium and Cu-ETP, no matter whether CP-Titanium was used as the flyer or target (see Fig. 11). Experimental observations of metal jet composition, by Kakizaki et al. [62] were also in a good agreement with these simulation results.

In this study, experimental tests and numerical simulations were conducted for tubular MPW of Al6060T4 (flyer) to Cu-ETP (target). Effects of process parameters including impact velocity, target thickness, and mandrel inclusion on wavy interfacial morphology were investigated. The hypothesis tested in this research is that a “well-supported target,” i.e., either a thick target or the support of a mandrel, allows for vortices to be created at the interface during MPW provided that the impact velocity is sufficient. Numerical analyses using SPH were in a good agreement with experimental observations. Both increasing impact velocity and target thickness led to an increase in wavelength. Similarly, inserting a mandrel into a thin target to minimize undesired deflection also resulted in wavelength growth. The influence of these parameters on amplitude showed a similar trend, but only for high impact velocities (over 300 m/s). The amplitude changed little for low-velocity tests. Furthermore, vortices can be obtained for this material combination with appropriate impact velocity (>300 m/s) and well-supported target, i.e., either a thick target (6 mm) or a 2 mm target with the support of a mandrel. The results obtained support the hypothesis. Furthermore, SPH was shown to be able to predict the jetting phenomenon. The ejected materials were mostly composed of the metal tube with lower density, and the velocity of metal jets correlated to the impact velocity. Although a functional, strong weld can be achieved with a flat interface for certain material combinations, a wave interface was considered as the preferred weld structure due to an increase in the contact surface between the weld partners, and vortices formation was a sign of material interlocking for this material couple. Thus, these results provide guidance, i.e., use a high impact velocity (>300 m/s) and a well-supported target, either through inclusion of a mandrel or increased target thickness, when designing a MPW process if a wavy morphology at the interface is desired as an indication of weld strength.

Funding from the U.S. National Science Foundation (CMII-1537471) and the German Research Foundation (DFG, project TE 508/39-3) is gratefully acknowledged.

There are no conflicts of interest.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request. The authors attest that all data for this study are included in the paper.