**4. Discussion**

#### *4.1. Influence of Kinetic Energy Input on Weld Interface and Welding Window*

For the comparison of the results of the model test rig and the MPW setup, it is important to recall that the collision conditions continuously change during the MPW process [18–20]. Hence, the impact starts at a small collision angle close to 0◦ and at the maximum impact velocity. This results in a high collision point velocity. Subsequently, the collision angle increases and the impact velocity decreases until the weldable region of the welding window is entered and then, after further progression of the collision conditions, left again. Depending on the process parameters, the path through the material-specific welding window differs. In the test rig, in contrast, the collision angle and the impact velocity stay constant during the collision. Therefore, each experiment represents a single point in the welding window. Performing a series of experiments with different collision conditions allows defining welding windows for different process and material parameters. This in turn provides an extended understanding of the governing welding mechanisms and delivers additional information for the design of the MPW process with industrial relevance.

Considering the determined welding windows for di fferent thicknesses in the test rig (Series 1.2), it was possible to analyze the formation of the weld interface in the MPW setup (Series 2.1, 2.2, 2.3); compare Figures 6 and 9. The progression of the collision angle was similar for di fferent thicknesses, until the start of the bond formation. In addition, higher thicknesses resulted in longer weld interfaces. For MPW it was not possible to identify, whether the bond formation stopped (i.e. the collision kinetics left the weldable region of the welding window) due to an increase of the collision angle or due to a combined change of collision angle and impact velocity. The latter depends on the di fferent thicknesses, the resulting di fferent sti ffness values and thus a varied forming behavior of the flyer during the rolling movement on the target.

When the acceleration distance was increased, the location of the weld interface was shifted closer to the initial point of collision. This was due to the higher progression rate of the collision angle leading to a flyer rolling movement that the lower boundary angle was exceeded earlier and the upper boundary angle was reached faster. This is in good agreemen<sup>t</sup> with the findings of Sarvari et al. [21] who investigated di fferent acceleration distances in an MPW setup.

To understand the influence of the mass induced kinetic energy on the bond formation, the results of the test rig experiments with varied thickness (Series 1.1) and varied impact velocity (Series 1.2) are plotted together in the classical β-*v*c-welding window in Figure 12. It can be observed that the increase in energy input by changing both the flyer mass and a higher impact velocity led to an expansion of the weldable region towards lower collision point velocities. The change of the impact velocity to achieve a certain kinetic energy also changed the collision conditions while they remained equal when the flyer mass was increased. Moreover, the weldable area of the welding window was affected by the increase in energy. Similar findings were reported by Lysak and Kuzmin in [26] for explosion welding. To explain the solid-state welding mechanism, they related the impact velocity, the collision point velocity and the involved mass to three physical parameters, the pressure at the PoC, the duration of the applied pressure and the temperature in the weld interface zone. In the next section, it is explained how these parameters influenced the phenomena in the joining gap and the governing bond mechanisms in the case of changed energy input.

**Figure 12.** Welding window by collision angle β over collision point velocity *v*c for di fferent flyer thicknesses *s* and impact velocities *<sup>v</sup>*imp measured in the test rig; symbols in shaded graphs represent experiments with bond, symbols outside of the graphs represent experiments with no bond. The different flyer thicknesses at *<sup>v</sup>*imp = 262 m/s are plotted at slightly different impact velocities to improve the visibility. Arrow indicates the influence of the increasing value on the upper left boundary.

#### *4.2. Influence of the Collision Kinetics on the Phenomena and the Bond Mechanisms*

The findings of the test rig experiments (Series 1.1 and 1.2) and the analysis of the weld interface revealed that the acting phenomena, especially the role of the CoP, at di fferent points of the welding window have a pronounced influence on the occurring bond formation mechanisms. This has not been taken into account previously. Hence, due to the transient propagation through the welding window during MPW, it can be assumed that di fferent mechanisms of bond formation apply in certain sections depending on the process parameters and the formed CoP.

During test rig experiments with small collision angles, maximum flyer mass and high impact velocity, the CoP had a strong impact on the bond formation. The SEM images in Figure 8a (1–3) for *s* = 2 mm indicate that at first, due to the small angle and resulting high flow resistance, the CoP could not be ejected out of the gap, was entrapped and hindered the contact of the activated base materials. It was found by the analysis of the associated process glare that the CoP can reach temperatures of several thousand Kelvin (see the accompanying paper [10] and compare it with Figure A1). This might also result in the nonwelded regions due to excessive melting of the surfaces (4). Later on, during the collision front propagation towards the free end of the flyer, the resistance to eject the CoP was low enough with the result that the interfaces came into contact and formed a bond (5). In this case, the mechanism of bond formation was attributed to fusion-like bonding, regarding the estimated high-temperature development and no visible signs of a jet in terms of a metal stream. This hypothesis is in good agreemen<sup>t</sup> with the findings of Bellmann et al. [35], who observed no deformation of the surface-near layers in the form of a jet at small collision angles but melting close to and in the weld interface.

In contrast, the formation of a jet occurred at larger collision angles (Figure 8b). In the case of large welded areas (ratio: 0.82 to the total overlapping area) in the middle of the estimated welding window for *s* = 2 mm (Figure 8b), most parts of the weld interface exhibited only a few defects (2) and were almost not distinguishable from the base material in the SEM. A determination of the governing bond mechanism was thus not possible. The absence of porous structures at the weld interface may indicate solid-state welding. At this point, it should be noted that the present investigation focused on joints between similar metals. No additional aspects were considered that would occur during the welding of dissimilar metals, like the formation of intermetallic phases.

The weld interface close to the upper boundary angle for *s* = 2 mm exhibited a less distinct jet and several areas without bond (see Figure 8c). This indicates that the deformation of the surfaces was considerably smaller and less adjacent base material was deformed compared to smaller collision angles. Two approaches provide possible explanations for these findings: First, according to Manikandan et al. [36], the depth of deformation of the adjacent base material by the colliding surfaces depends on the induced kinetic energy. Therefore, it can be argued that for large collision angles less induced energy was available for the microscopic deformation close to the point of collision. This can be explained by the fact that at larger collision angles more work was needed to close the joining gap by the continuous bending of the flyer. A second explanation may be the di fferent interactions of the CoP with the surfaces in front of the point of collision. As mentioned above, the CoP can reach very high temperatures. Even if it did not melt the surfaces, the induced heat could cause a considerable reduction of the flow stress close to the point of collision, allowing more material to flow in the area of the contact surfaces, which was already described by Khaustov et al. [37] for explosion welding. At larger collision angles the temperature was lower due to the lower compression and less heat was transferred to the surfaces. Thus, the potential plastic deformation in the point of collision was reduced at increasing collision angles, which also explains the reduced intensity of the process glare (Figure A1). Moreover, the interaction of both phenomena is conceivable to explain the di fferences between the cases in Figure 8b,c.

Considering the influence of di fferent flyer thicknesses, for *s* = 1.5 mm a comparable maximum welded area was obtained (ratio: 0.77), whereas for *s* = 1.0 mm the maximum value is smaller (ratio: 0.64). In both cases, the joint was produced in a region close to the lower collision angle boundary,

where for *s* = 2.0 mm the bond formation (ratio: 0.62) has not reached its maximum value. Although some pores and melted interlayers were determined in sections of the weld interfaces of smaller flyer thicknesses, which likely resulted from the interaction with the hot CoP, large inclusions of the CoP could not be detected. Since the impact velocity, the collision angle and thus the air flow in the joining gap were comparable, this could not be attributed to a better ejection condition of the CoP. A possible explanation could be that due to the higher energy input more deformation around the point of collision occurred and, thus, also a larger jet was formed. As shown in Figure 8b (2), the jet can spall in particles. Furthermore, Pabst [38] determined that during the collision welding of aluminum, additional heat was induced into the CoP by an exothermic reaction of chipped particles of the base material. If more base material particles were chipped out by the impact, this in turn might cause more heating. The influence was lower for smaller flyer thicknesses and resulted in less spalled particles, heating and process glare (see Figure A1).

Concerning the governing bond formation mechanisms and considering the SEM images, the largest welded area for *s* = 2 mm was likely produced in the test rig experiments by solid-state welding, whereas excessive melting prevented welding at smaller collision angles. In contrast, the largest welded areas for the smaller flyer thicknesses were formed at least partly by fusion-like bonding. Both processes were strongly influenced by local interaction of the surfaces with the CoP.

These findings are transferable to the MPW interface (Series 2.1, 2.2, 2.3). A certain angle had to be exceeded to allow the ejection of the CoP and to initiate bond formation. The weld interface started with a partially melted but bonded interface where parts of the CoP were entrapped which is also supported by the findings in [6,39]. The further transient rolling movement of the flyer facilitated the ejection of the CoP, whereas the formation of the jet by plastic deformation occurred more intensively at first and then weakened again. This resulted in a jet behavior similar to that at large collision angles in the test rig (see Figure 9) (6). These sections were not clearly distinguishable in the SEM images. Considering the locations of weld defects by melting or CoP-entrapment, it is assumed that in the case of 1.5 mm, most of the weld interface was a result of fusion-like bonding, whereas the latter sections were bonded by solid-state welding due to the occurrence of the jet at the end. In the case of 2 mm flyer thickness, solid-state bonding dominated the weld interface due to the absence of such weld failures in the interface. On the other hand, a jet at the end of the weld interface was not clearly visible (Figure 9) (4).

Despite these di fferent bond mechanisms and resulting weld interfaces, no significant dependence of the joint strength on bond mechanisms was identified (see Figure 11). The di fferences in the determined tensile forces mostly depended on the weld interface width and the question of whether a complete ring-elliptical weld interface was formed or not, which directly influences the size of the welded area (compare with Figure 11).

#### *4.3. Prediction and Control of the Weld Interface's Formation and Properties*

The results in Appendix A show that the process glare detected during the MPW experiments was brighter for an increased flyer thickness (Series 2.1, 2.2, 2.3). This is in good agreemen<sup>t</sup> with the experiments performed in the test rig where the kinetic collision conditions were kept constant. It is likely that the increased kinetic flyer energy led to an increased formation of CoP (or jet) as also described by Eichhorn [33]. Figure A2 shows the tendency of an increased flash intensity for smaller acceleration distances, which points to the conclusion that the collision angles were smaller and, thus, the compression and light emission of the CoP (or jet) were intensified. This finding can also be related to higher temperatures, as reported by Bellmann et al. [35].

The correlation of process glare parameters with the welding results revealed that the longest weld seams and the highest bond strengths were achieved in experiments with the longest and brightest impact flashes. This was the case for the highest flyer thickness and energy input, respectively, as well as for the smallest acceleration distance. Although the impact flash is only a *necessary* but not a su fficient welding criterion, this observation underlines the importance of a su fficient surface activation prior to

surface contact, which comes along with a bright impact flash. At this point, it should be noted once more that only the impact velocity was adjusted directly during the MPW experiments, but not the collision angle. Thus, the reason for the increased weld length and strength obtained by using thicker flyers and smaller acceleration distances might also be attributed to the di fferences in the collision kinetics, especially the reduced collision angle. Starting with a small collision angle and assuming similar progression rates, a larger portion of the flyer collides under weldable conditions with the target compared to a high initial collision angle. This led to longer weld seams and brighter impact flashes. Hence, the correlation between the impact flash and the weld strength reported here is valid and appears to be a powerful tool for process development and quality assurance during MPW processes.

Considering the largest ratios of welded to the overlapping area at the test rig (Figure 7), it can be derived that there is a collision angle region leading to optimum collision welding conditions. This region increases and is shifted to larger angles when the kinetic energy input is increased. This kinetic energy input can be varied via the impact velocity [24] or the flyer thickness. The resulting adjustment of the progression rate of the collision angle allows controlling location and length of the weld interface during MPW.

## **5. Conclusions and Outlook**

The collision kinetics influence both the CoP formation and its temperature (measured by analyzing the process glare in the companion paper [10]). It, therefore, determines the governing bond mechanism and thus, the reachable amount of welded area. The latter is, however, also influenced by other parameters like (initial) collision angle, its progression and the rolling movement.

Depending on the process conditions, the CoP can be useful for, or harmful to, the bond formation, it furthermore determines the predominant bonding mechanism.

The results of the test rig experiments confirm that the width of the weld interface can be increased by a smaller gradient of the collision angle, when the weldable area of the welding window is reached. Therefore, it could be useful to prepare the flyer geometry to influence its rolling movement on top of the target during MPW and, thus, to improve the weld interface formation. Together with the acquired knowledge about the di fferent ways to increase the energy input, it is possible to adjust the size and location of the weld interface by setting the process parameters. In addition, monitoring of the process glare potentially enables quality assurance of high-strength joints.

Considering that the amount of the CoP increases continuously with the length of the colliding joining partners, the question arises in which areas it is still possible to consider a stationary welding process despite constant kinetic conditions. Therefore, it is of interest for further investigations if and how the local bond properties change and if the welding window can be extended by considering, as an additional factor, the quality of the weld interface.

**Author Contributions:** Conceptualization, B.N., E.S., J.L.-A., J.B. and M.B.; methodology, data analysis, B.N., E.S., J.L.-A. and J.B.; design of experiments, investigation, validation, B.N. (test rig experiments, ultrasonic investigation, long time exposures, SEM analysis), E.S. (MPW experiments, tensile shear tests, optical microscope and SEM analysis), J.L.-A. (PDV-measurement, flash measurement), J.B. (flash measurement) and M.B. (characterization of material's properties); writing—original draft preparation, visualization, B.N., E.S. (Sections 2.3, 2.5, 3.2, 4 and 5), J.L.-A. (Sections 2.3, 2.4.2, 2.4.3, 4 and 5) and J.B. (Sections 2.4.4, 4 and 5); writing—review and editing, S.B., A.E.T., C.L., M.F.-X.W. and P.G.; resources, supervision, project administration, funding acquisition, S.B., A.E.T., E.B., M.F.-X.W., and P.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), gran<sup>t</sup> number BE 1875/30-3, TE 508/39-3, GR 1818/49-3, WA 2602/5-3, BO 1980/23-1 and is based on the results of the working group "high-speed joining" of the priority program 1640 ("joining by plastic deformation"). The working group consists of the subprojects A1, A5, A8 and A9. We also acknowledge support by the German Research Foundation and the Open Access Publishing Fund of the Technical University of Darmstadt for providing this publication as open access.

**Acknowledgments:** The experiments at the model test rig were conducted at the Bachelor Thesis of Yabing Wang under the supervision by Benedikt Niessen. Moreover, we would like to thank Walter Tutsch of PCO AG for the support in setting up the image intensifier camera. The authors also greatly appreciate the help of Stephan Ditscher of Baumüller who supported the programming of the system control of the test rig. We thank Ammar Ahsan for his e fforts in proofreading the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

## **Appendix A Analysis of the Process Glare**

The qualitative examination of the process glare during test rig experiments is represented in Figure A1 for flyer thicknesses of 1 mm and 1.5 mm at di fferent collision angles. It should be noted that the majority of process glare at the images was visible after leaving the joining gap and no temporal resolution is shown. Nevertheless, it can be seen that the intensity increases with the thickness of the flyer and the glare's color changes from red-orange to white-blue. Furthermore, the collision angle had a significant influence on the intensity, color and amount of the process glare. For angles lower than the lower boundary angle, the glare in the gap was not visible. Therefore, the glare was visible at the sides of the colliding joining partners with only weak intensity. With increasing collision angle its intensity and shape also increased up to a maximum. At larger collision angles the shape remained at the same level, but the intensity decreased, especially when the upper boundary angle was exceeded.

**Figure A1.** Comparison of the process glare for 1 mm and 1.5 mm flyer thickness at di fferent collision angles, recorded by longtime exposure.

As mentioned in Section 2.4.4, the weld formation was accompanied by a bright flash when impact welding is performed in ambient atmosphere. The average values of the impact flash duration and its maximum intensities were evaluated for each parameter set (see Figure A2). The flash duration as well as the maximum intensity increased with higher flyer thickness. The influence of the acceleration distance on the light emission was not that clear due to the data spread. There is a slight tendency showing a decrease in the flash duration and intensity for increased acceleration distances.

**Figure A2.** (**a**) Increasing flash durations and (**b**) increasing intensities for increasing flyer thicknesses (1.0 mm, 1.5 mm and 2.0 mm) at different acceleration distances *g*. Each point represents four or five experiments with the corresponding standard deviation.
