**1. Introduction**

One of the biggest challenges today is climate change and its impact on the environment and human society. Driven by politics and self-motivation, the industry is increasingly striving for sustainable products and manufacturing processes, e.g., by introducing clean production methods without toxic

components and with less energy and raw material consumption. Furthermore, new products have to be more environmentally friendly. For example, the consistent lightweight design in the transport sector is an important factor to reduce emissions. Reliable joining techniques are key to implement load-adapted material usage and to fulfill further operational functions. Conventional fusion-based joining processes, however, reach their technological limits when it comes to metallurgical joining between dissimilar metals. In contrast, solid-state welding techniques like magnetic pulse welding (MPW) can lead to advantageous properties like high bond strengths, no heat-affected zones and low electrical resistance, even between metals with differing thermomechanical and chemical properties [1,2]

MPW is based on the oblique collision between two joining partners at high relative velocities [3], thus belonging to the category of collision welding processes like explosion welding or laser impact welding. Usually, one of the joining partners, called flyer, is accelerated up to several hundred meters per second and collides with a stationary so-called target at an impact velocity *<sup>v</sup>*imp under a collision angle β (see Figure 1). Due to this angle, a collision front (or in the two-dimensional case a point of collision (PoC)) moves along the colliding surfaces characterized by the collision point velocity *v*c. High strain rates of up to 10<sup>6</sup> 1/s and high pressures of up to several GPa occur at the collision point [4,5]. When the dynamic elastic limit of the material is exceeded, material flow results from the plastic deformation of the contact surfaces and a stream of material is pushed ahead of the collision point, see detail in Figure 1 [6–8]. This phenomenon is called jetting and is, besides other criteria, regarded as a necessary condition for bond formation. The jet can remain as a cumulative stream or can disperse in particles. Furthermore, the extensive local strains at the point of the collision lead to the removal of brittle oxide layers and surface contaminations from the surfaces, which are ejected either as a compact stream or as a dispersed cloud of particles (CoP; see Figure 1) [9]. Depending on the collision conditions, the CoP either results of the dispersed material stream, the spalled surface contamination and oxide layers or both phenomena, whereas the cumulative jet can be partly or completely hidden by the CoP [9,10]. This ejection is typically accompanied by a process glare in the form of a bright light or flash emission [11]. Due to the high pressure and temperature at the PoC, the clean surfaces are forced into intimate contact, which ultimately triggers the bonding mechanism. Cui et al. [12] identified three different joining mechanisms when joining aluminum and titanium by MPW, depending on the prevalent kinetic and thermal energies: solid-phase metallurgical bonding by diffusion, liquid-state bonding by melting and solid-liquid coexistence state bonding.

**Figure 1.** Schematic illustration of symmetric sheet magnetic pulse welding (MPW) with the formation of the jet as material flow at the point of collision according to [7,10] in detail. The cloud of particles (CoP) is either formed by dispersed jet particles, spalled contamination and oxide layers or both phenomena.

During collision welding, the process-related acceleration of the joining partners determines the provided kinetic energy. In the case of MPW, the mobile joining partner is accelerated by an induced electromagnetic pressure (see Figure 1) that is generated by closing an electrical circuit that consists of a charged capacitor bank and a coil actuator [3]. The energy input can be adjusted to the welding task by variation of the charging energy. Different coil designs allow welding of overlap joints of profiles, tubes and sheets, respectively. The high repetition rate and the ability to integrate the system into production lines are two main advantages for the usage of MPW in mass production [13]. However, the joining process is not ye<sup>t</sup> widely applied. One reason for this is the necessary certification of the joining properties, which, in addition to characterization of joint strength, also requires the verification of fatigue strength, corrosion behavior or gas tightness, for instance. Recently, these issues have been objects of concentrated research, and some promising results have already been obtained [14–18].

In addition, a deeper understanding of the joint behavior under di fferent loading scenarios during service is essential for an adequate design of the joining partners and the joint. Ideally, this would cover the prediction of the weld interface position, its shape and area as a function of the material and process parameters. Therefore, a wider comprehension of the bond formation and its influencing parameters is necessary. However, the collision parameters change along the propagating collision front due to the transient behavior during MPW and, thus, cannot be simply calculated or measured [19–21]. In addition, the nature of the process hinders a separate investigation of the influence of the velocity, the mass and the resulting e ffective energy, which result from the selected acceleration distance and the charging energy of the MPW setup [22,23]. Thus, no direct correlation between the energy input and the welding result can be drawn. Therefore, a purely mechanical model test rig (described in Section 2.1) was designed that allows an independent adjustment of impact velocity and collision angle [24]. A study with copper-copper joints on weld interface formation depending on these parameters showed that after reaching a minimum velocity, the shape and size of the weld interface was influenced particularly by the collision angle [25]. Higher impact velocities increase the region where large weld interfaces can be produced and shift this region towards larger collision angles [25].

The role of the energy input for bond formation is, however, still unclear. This includes additional effects which occur when the kinetic energy is varied either by changing the accelerated mass or the impact velocity. Former investigations mostly varied the kinetic energy of the flyer via the impact velocity at a constant flyer mass. However, this also changes the collision conditions. For explosion welding, Lysak et al. [26] related the collision conditions to the energy part for the metal's plastic deformation by adding the averaged mass to the classical welding window as a third dimension. Thereby the hydrodynamic processes of the collision are linked to the metallophysical processes of bond formation.

To transfer these findings to the MPW process at comparatively low energy input, the influence of the energy input is examined in this paper separately from the collision conditions by changing the flyer mass and keeping the impact velocity constant. For this purpose, several experimental series with flyers of three di fferent thicknesses are carried out using two di fferent experimental setups. The experiments on the mechanical model test rig serve to change the particular parameters individually and to determine their influence on the weld interface, they are supplemented by experiments with di fferent impact velocities. For this purpose, the di fferent phenomena occurring at collision welding can be related to the formation of the weld interface for di fferent points in the welding window. Subsequently, the results are validated by experiments on an MPW setup with di fferent flyer thicknesses, for which the necessary impact velocity parameter settings are determined with an adapted measurement setup. This allows the determination of the influence of di fferent energy inputs at otherwise comparable collision conditions. Based on the experimental results, the following questions are addressed in the present paper:


## **2. Materials and Methods**

## *2.1. Description of the Series of Experiments*

Three series of experiments, performed in two different setups, a model test rig (see Section 2.2) and MPW setup (see Section 2.3), were carried out with aluminum sheets (EN AW-1050A Hx4, yield strength: 99 MPa, tensile strength: 105 MPa) with an initial thickness of *s* = 2 mm for the target as well as the flyer. The specimens for the test rig were produced by laser cutting, while specimens for the MPW setup were cut to size (40 mm × 100 mm) by plate shears. To study the influence of the flyer thickness while keeping the material properties constant, the sheet thickness for both setups was reduced by milling to 1 and 1.5 mm, respectively. The main experiments were carried out with an impact velocity of 262 m/s (see Table 1, Series 1.1 and 2.1., 2.2, 2.3). Furthermore, a second series of complementary experiments were carried out in the test rig with a target and flyer thickness of 2 mm and varied impact velocities of 220 m/s and 240 m/s, respectively (see Table 1, Series 1.2).


**Table 1.** Summary of series of experiments in applied setups with varied and constant parameters.

In the test rig, the collision angle was varied to define the weldable region of the welding window. From earlier investigations, it was known that collision angles that led to welding are in the range of 3◦ to 9◦ for 2 mm thick joining partners of aluminum at an impact velocity of 262 m/s. Based on this, the weldable range was determined for each flyer thickness and impact velocity. The joints are considered welded if they cannot be separated manually after the experiments.

The initial impact velocity was adjusted close to 262 m/s using the MPW setup by means of photonic Doppler velocimetry (see Section 2.4.3 and Table 1). As mentioned above, the further progression of impact velocity and collision angle was unsteady and difficult to measure in this setup. However, the initial collision angle and its rate of change along the propagating collision front was varied by different acceleration distances of 1.5 mm, 2 mm and 2.5 mm.

## *2.2. Model Test Rig*

Besides the individual and precise adjustment of the collision parameters at stationary process conditions, the model test rig was built up (by the PtU Institute, Darmstadt, Germany) with the intention to provide good observability, which was realized by a purely mechanical concept (see Figure 2a). The main components are two rotors with a diameter of 500 mm, each one driven by a synchronous motor. As a joining partner, specimens with a collision area of 12 mm × 12.5 mm were mounted and prebent with a certain angle at one end of each rotor (see Figure 2b,c). In order to start the collision welding operation, both rotors rotated in the same turning direction but with a phase offset of 45◦. As the rotational speed reached half of the desired impact velocity, the phase offset was compensated within one revolution. Thus, the two specimens collided with high accuracy and repeatability in the center between the two turning points. After the collision and the accompanying welding process, the specimens were torn off at the predetermined breaking point since the rotors could not be stopped instantaneously; see Figure 2d. The applied configuration of the test rig for these experiments led to a maximum absolute impact velocity *<sup>v</sup>*imp of 262 m/s [27].

**Figure 2.** (**a**) Test rig assembly [28], (**b**) shown in detail: mounted joining partners and the resulting process parameters at the moment of the initial impact [28]. (**c**) Geometry of the specimen in the test rig [24] and (**d**) welded specimens [28]. (**e**) A long-term exposure of process glare (2 s exposure time) and (**f**) high-speed image [24] of the collision welding process with 20 ns exposure time were recorded. (a,b,d) are reproduced from [28], with permission from Elsevier, 2017; (c,f) are reproduced from [24], with permission from John Wiley and Sons, 2019. The entire figure is also published in the companion paper [10].

## *2.3. MPW Setup*

The MPW experiments were carried out with the pulse generator BlueWave PS48–16 from PSTproducts GmbH, Alzenau, Germany in combination with the sheet welding tool coil B80/10. The pulse generator provides a maximum charging energy of 48 kJ and a maximum charging voltage of 16 kV. The effective part of the tool coil had a width of 10 mm, a length of 80 mm, and a thickness of 5 mm. It can be operated up to a maximum peak current of 500 kA. During the welding experiments, the flyer and target sheets were positioned in the center above the coil with an overlap of 30 mm and were fixed by a steel backing plate (see Figure 3).

**Figure 3.** MPW setup for sheet welding: the acceleration gap *g* and the flyer sheet thickness *s* were varied. Section A-A shows the photonic Doppler velocimetry (PDV) measurement configuration at the initial state of the welding process (no deformation of the flyer). The oscilloscope recorded the signals of the PDV system, Rogowski coil and measured intensities of the process glare.

## *2.4. Methods of Process Observation*

## 2.4.1. Process Observation in the Model Test Rig

Two observation methods were implemented in the test rig. First, the collision welding process was observed by an image intensifier camera hsfc pro (by PCO, Kelheim, Germany) with a long-distance microscope lens. This system allowed taking up to eight images per experiment. Due to the high-velocity collision an exposure time of 20 ns was applied (Figure 2f). During the experiments, a CAVILUX Smart lighting laser (by Cavitar, Tampere, Finland) with a power of 400 W and a wavelength of 640 nm provided sufficient brightness. In combination with an optical bandpass filter placed in front of the camera lens, the bright process glare was suppressed, which would otherwise outshine the phenomena in the closing gap. A script in MATLAB (by MathWorks, Natick, Massachusetts, MA, USA) was used to measure the collision angle β by edge detection in each high-speed image, as shown exemplarily in Figure 2f [24,28,29].

Second, a qualitative examination of the process glare was realized by long-term exposures (Figure 2e), with a single-lens reflex camera 5D (by Canon, Ota, Tokio, Japan) and a 100 mm macro lens, ¯ which was positioned at an angle of about 45◦ to the rotor center axis. The image acquisition settings were set to an exposure time of 2 s, an aperture of F13 and a light sensitivity of ISO 100. The results are shown in Appendix A.

## 2.4.2. Rogowski Coil

The recording of the discharge current was performed via a Rogowski pickup coil type *CWR* 3000 B (by Power Electronic Measurements Ltd., Long Eaton, Nottingham, UK) in combination with a high-resolution oscilloscope (see Figure 3). Rogowski coils are especially suitable for the measurement of oscillating currents with high amplitudes and frequencies like those occurring during MPW. The discharge current curve *I*(*t*) contains important information for the evaluation of the temporal evolution of the MPW process and is shown in Figure 4, together with the recorded light intensity.

**Figure 4.** Signals of the Rogowski pickup coil and the flash measurement system showing the time-dependent evolution of the tool coil current and of the light emission, respectively, in the setup with the pulse generator BlueWave PS48-16.

## 2.4.3. Photonic Doppler Velocimetry

An accurate, quantitative determination of the flyer velocity at the moment of impact is crucial for the setting of similar impact velocities at the different experimental setups. Compared to the test rig, the accessibility of the collision zone during the MPW process is very limited for cameras due to the

small acceleration distance and the progression of the collision front. The initial impact velocity at a fixed acceleration distance can be adjusted via the charging energy, but the analytical or numerical determination is elaborate and the MPW process is very sensitive to small disturbances like variations of material properties or flyer dimensions. Therefore, photonic Doppler velocimetry was applied for the measurement of the impact velocity *<sup>v</sup>*imp during the MPW experiments. This robust and accurate method, developed by Strand et al [30], is based on the laser Doppler e ffect. It is an established measurement technology in the field of high-velocity forming and joining [31]. With the applied PDV system, velocities of up to 1.1 km/s can be measured using three parallel channels. The characteristics of the applied system are described by Lueg-Altho ff in [32].

The application of PDV measurements requires the direct accessibility of the surface of the moving object, i.e., the flyer part in MPW. Therefore, three focuser probes were positioned normal to the setup (see Figure 3). Small boreholes were drilled into dummy target parts and the backing plate in order to allow the laser beams to directly illuminate the moving flyer surface. This allows measuring the temporal evolution of the flyer velocity from the beginning of the movement until the impact, and to adjust the impact velocity *<sup>v</sup>*imp for all three flyer thicknesses and acceleration distances by modifying the charging energy of the pulse generator and the corresponding current flow in the tool coil, respectively. In Table 2, the determined charging energies are listed for several combinations of flyer thickness and acceleration gap (Experimental Series 2.1, 2.2, 2.3), as well as the averaged measurements of maximum current, discharge frequency and impact velocity (including minimum and maximum values). The boreholes in the target plate inhibited welding between the flyer and the dummy target parts. Nevertheless, the evolution of the flyer impact velocity was not a ffected by the presence of the boreholes until the collision because the magnetic field is completely shielded by the conductive flyer part. Therefore, it was assumed that the normal impact velocity *<sup>v</sup>*imp determined at the three measuring points was identical in the experiments with a solid target for the real welding experiments.


**Table 2.** Determined charging energies for combinations of flyer thickness and acceleration gap and averaged measurements of maximum current, discharge frequency and impact velocity.

## 2.4.4. Flash Detection

High-speed collision processes are accompanied by a characteristic flash, which is called impact flash [33]. During the MPW experiments, the time-dependent evolution of the light emission was measured with the flash measurement system explained in [34]. Figure 4 shows an example of the time-resolved light intensity as well as the derived characteristic values—the starting time of the flash, its duration and maximum intensity. According to previous studies, the flash duration was defined as the duration between the initial increase of the light intensity and its steep decrease [34]. These parameters were taken with two independent sensors at the same distance to the welding zone of approximately 15 mm and were then averaged for each series of experiments with a specific flyer thickness and acceleration gap, respectively. The results are shown in Appendix A.

## *2.5. Analysis of the Weld Interface*

A nondestructive analysis method of the welded area was carried out using a 2D ultrasonic measurement with the MiniScanner (by Amsterdam Technology, Zwinderen, The Netherlands) for experiments on the model test rig. This device scans an area of 12 mm × 25 mm during a single scan run in pulse-echo mode with a local resolution of approximately 0.1 mm × 0.1 mm spots. For each spot, a so-called A-scan echo was recorded with its two-dimensional coordinates. This scan information was analyzed using a MathWorks MATLAB script. Depending on the signal, a differentiation was made between "bond", "no bond" or "no information" for each scanned point. This approach delivered both a qualitative and quantitative result in terms of the welded area [24]. Since the joining partners tear off after their collision in the model test rig, subsequent collisions with the still rotating rotors and the housing can occur. These joints were post-treated by flattening prior to ultrasonic examination. In addition, scratches and burrs were removed from the surfaces by grinding to improve the signal quality. If the deformation exceeds a certain limit, the evaluation of the joint was affected and hindered locally or completely.

The joints produced at the MPW setup were tested for their bond strength by tensile shear testing in a Z100 testing machine (by Zwick, Ulm, Germany) with three repetitions for each parameter set at a testing velocity of 10 mm/min. Additionally, for one joint of each series, a cross-section was prepared parallel to the central plane in the welding direction as shown in Figure 5. Due to the symmetric collision of the flyer with the target, there were two symmetric points of collision that moved in opposite directions and formed two weld interfaces; see also Figure 1. To characterize the welding result, the widths of the two welding interfaces and the gap between them were measured in the cross-section using an optical microscope (OM) DM2700 (by Leica Microsystems, Wetzlar, Germany).

**Figure 5.** (**a**) Schematic sketch of a welded MPW joint (**b**) with a fracture image of the welding interface with typical elliptical ring-shape after tensile shear test. The location of the cross-section is marked by the dot-dashed line (A–A) and was analyzed by OM and SEM. An exemplary OM-image of a cross-section (A–A) is shown in (**c**).

The microstructures of the weld interfaces of joints made with both setups were further analyzed with a Ultra Plus (by Zeiss, Jena, Germany) scanning electron microscope (SEM), using the secondary electron detector to validate the results of the ultrasonic examination and also to determine how the different weld interface types have been formed.
