**3. Results**

## *3.1. Model Test Rig*

Figure 6 shows the welding windows for the series of Experiments 1.1 and 1.2 performed in the test rig. Although in both cases, the lower boundary collision angle, below which bond formation is inhibited, did not differ strongly, the upper boundary collision angle increased both with higher flyer thickness and increased impact velocity.

**Figure 6.** Welding windows (**a**) for different flyer thicknesses and collision angles and (**b**) for different impact velocities and collision angles, each point represents one experiment. The upper boundary angle increased in both cases, while the lower boundary varied only slightly.

For Series 1.2, the findings correspond to those already obtained in welding window investigations for another batch of the same material. In this context, the lower boundary angle was related to the suppressed ejection of the CoP, which inhibits bond formation by the reinclusion of the CoP particles at the PoC. In contrast, the upper boundary angle defines the process parameters up to which jet formation can be initiated and sustained [23].

The results of the ultrasonic analysis of the weld interface of Series 1.1 are shown in Figure 7. A similar curve of the ratio between the welded and overlapping areas over the collision angle was achieved for all three flyer thicknesses (see Figure 7a). Just a small amount of the overlapping area was welded close to the lower and upper boundary angles. In between, there is a region where large area welds can be formed. However, none of the specimens was completely welded. The range of this region, its maximum value and the corresponding collision angle increased with increasing flyer thickness and was shifted to larger collision angles. This behavior had also been observed in experiments where copper was welded at different impact velocities [24] and was further validated by the results of Experimental Series 1.2 with lower impact velocities using a flyer thickness of 2.0 mm. The types of weld interfaces described in [25] can also be found here and support the described phenomena at the upper and lower boundary angle. For small collision angles and without inhibiting the bond formation, the CoP could only escape sufficiently in the lateral regions and at the end of the closing gap. At large collision angles, the jet formation was initiated after an entry region, but then broke down while the flyer continued to deform on top of the target (see Figure 7b). However, this weld interface type was not as pronounced regarding the termination of the weld interface formation as in the previous experiments with copper. Furthermore, no completely welded interface could be obtained, which might be due to the fact that the investigations were carried out close to the lower limit of the welding process with respect to the energy input.

**Figure 7.** (**a**) Progression of welded to the overlapping area over different collision angles for the three flyer thicknesses (target thickness: 2 mm, *<sup>v</sup>*imp = 262 m/s); (**b**) two-dimensional representation of the weld interface obtained from the ultrasonic analysis.

Figure 8 summarizes the SEM analysis of the interfaces of Experimental Series 1.1. The welded interfaces are mostly straight and only single instances of wavy patterns can be found. Furthermore, the findings of the nonwelded interfaces and the transition regions support the hypothesis regarding the boundaries by collision angle and the related mechanisms. In Figure 8a the collision at an angle close to the lower boundary angle for 2 mm flyer thickness started without visible interaction of the surfaces (1). Shortly afterwards, the surfaces were contaminated by the enclosed CoP (2) whose amount increased along the joining gap (3). At a certain stage, the conditions in the gap changed in a way that local melting and resolidification occurred at the surfaces and the surfaces go<sup>t</sup> continuously closer, until the formation of the weld interface began (4, 5). At the end of the weld interface a continuous melted and resolidified interlayer was found (6).

**Figure 8.** *Cont*.

**Figure 8.** SEM analysis of test rig joints at different locations of the weld interface. Blue coloring indicates bond, red indicates no bond. (**<sup>a</sup>**–**<sup>c</sup>**): *s* = 2 mm: (**a**) Close to the lower boundary collision angle (4.6◦); (**b**) in the region with a large welded interface (5.6◦), (**c**) close to the upper boundary collision angle (7.3◦); (**d**) largest welded area (5.0◦) for *s* = 1.5 mm; (**e**) largest welded area (4.5◦) for *s* = 1 mm (note that different magnifications are used in the SEM micrographs to highlight relevant features).

In the region with a large welded area for 2 mm flyer thickness, the weld interface could hardly be recognized in the SEM micrograph (see Figure 8b) (1); however, nonwelded areas could be clearly identified. Only at the beginning of the weld interface, some melted structures and a porous interface were observed. Later on, the few nonwelded regions in the center did not contain porous material from the CoP but obviously parts of the jet stream which were torn off and rolled over by the PoC and hindered the bond formation (2). Moreover, at the end of the weld interface, the spilled jet was clearly visible (3—arrow).

In joints with 2 mm flyer thickness, produced close to the upper boundary angle, the welded regions were not properly formed and contained several imperfections (Figure 8c) (1, 2). The jet at the end of the weld interface was significantly thinner (3) than in the region described above.

Looking at the joints with 1.5 mm and 1 mm flyer thickness, large welded areas were found. While for *s* = 1.5 mm the sound weld interface was mostly hardly visible in the SEM micrographs (Figure 8d) (2), other parts contained locally melted and resolidified interlayers (1). The weld interface for s = 1.0 mm exhibited partially porous regions, melting structures and cracks (see Figure 8e), which were partly declared as sound weld by the ultrasonic analysis (2). Nonwelded regions showed larger melting defects (1).

## *3.2. MPW Setup*

Figure 9 shows the results of the weld interface formation in the MPW setup for the different flyer thicknesses and acceleration distances at selected positions of the weld interfaces (compare Table 1, Series 2.1). Considering the parameters separately, the start of the weld interface was not influenced by the flyer thickness. The end position of the weld interface increased with increasing flyer thickness. When the acceleration distance was enlarged, weld interface formation started earlier, but also ended earlier.

**Figure 9.** Microsections along the central plane parallel to the welding direction of weld interfaces in the MPW setup for an acceleration gap of *g* = 1.5 mm (Series 2.1): While weld interface began at the same position (dashed line), it ends later with increasing flyer thickness (dot-dashed line). Below, the results of the SEM investigation are shown in detail (1–9).

The summed width of both weld interfaces tended to decrease with increasing acceleration distance, especially for smaller flyer thicknesses, see also Figure 10. Furthermore, the flyer thickness of 1 mm exhibited an asymmetric image of the weld interfaces for 1.5 mm and 2.5 mm acceleration distances (Series 2.2, 2.3). The latter was only welded on one side in the sectioned joint (see cross-section in Figure 10), which is visible in the diagram by the smaller total weld interface width and thus, no gap width. This was either a result of the comparably low energy input or of an asymmetric rolling movement of the flyer due to the clamping situation in the weld setup.

**Figure 10.** Summed width of both magnetic pulse welded interfaces and width of the gap in between for different flyer thicknesses and acceleration gaps at constant impact velocity of 262 m/s. All configurations show the same trend: while the gap width varied slightly for the acceleration distances *g*, the summed width of the weld interface increased with increased flyer thickness *s*. Due to the asymmetric weld formation (see cross-section), no gap width could be determined for the configuration *s* = 1.0 mm and *g* = 2.5 mm.

The SEM analysis of the weld interface in Figure 9 revealed similarities with the results that were produced in the test rig, see Section 3.1. Pores and partly melted and resolidified structures were found in front of (1) and at the beginning of the weld interfaces (2, 5, 8). Such weld defects were related to the heating and/or entrapment of the CoP and indicated a collision angle close to the lower boundary. Similar defects were also located in further sections along the weld interfaces (6). The ends of the weld interfaces (4, 7, 9) were similar to the ones produced in the test rig that were welded at a collision angle close to the upper boundary. A thin jet at the end of the weld was identified for *s* = 1.5 mm (7).

The bearable tensile forces of the different weld configurations (Series 2.1, 2.2, 2.3) are shown in Figure 11. The comparison of the tensile forces revealed that for an acceleration distance of 1.5 mm all joints achieved the bearable tensile force calculated from the tensile strength of the base material and thus, all failed in the base material except for two joints. These two parts showed a nonuniform weld interface formation in the fracture pattern. For joints with a flyer thickness of 2 mm and an acceleration distance of 2 mm, failure occurred in the base material, which was also apparent in the achieved tensile force. In this case, the fracture occurred in the neck region of the flyer close to the welded area, where the cross-sectional area was reduced due to plastic deformation during the welding process.

All other joints failed in the weld interface. Relating the tensile force to the total weld interface width resulted in a ratio of approximately 1.5 kN/mm for all configurations. Only the configurations of 2 mm and 2.5 mm acceleration distance with 1 mm flyer thickness varied to lower values due to the incomplete global weld interface formation. The fracture images of these samples revealed that the weld interface was characterized as two parallel lines instead of a complete elliptical ring (see fracture images in Figure 11). Furthermore, all fracture surfaces showed a symmetric weld interface in contrast to the cross-section in Figure 10. Therefore, the width value of this configuration was corrected by the multiplication by a factor of two to calculate the ratio of tensile force to the width in Figure 11 to represent a symmetric weld interface.

**Figure 11.** Averaged tensile force with minimal and maximal deviation for different flyer thicknesses *s* and acceleration distances *g* (left axis, vertical columns). The dashed lines represent the theoretical bearable tensile force value for the particular flyer thickness calculated by the tensile strength of the base material. The ratio of the tensile force to total weld interface width is represented by rectangles and grey line and the right axis scale. The images of the different fracture surface types in top view are shown to explain the variation of the ratio tensile force to total weld interface width.
