The influence of the ratio between the wire and process velocities on the wetting process during laser brazing

Laser brazing is used in automotive production for the joining of car body parts with customer-visible seams, whereby the highest optical seam quality is demanded. During the wetting process, a constriction formation can occur within the molten brazing material, which subsequently collapses; this induces a stepwise progression of the wetting front, thereby decreasing the optical seam quality. The aim of this study is to investigate the influence of the wire and process velocities on this constriction formation. For this purpose, high-speed recordings of the process zone are performed during laser brazing and then evaluated concerning the wetting process. The results show that the constriction formation depends on the relationship between the process velocity and the process direction component of the wire velocity. The constriction formation only occurs when the process velocity exceeds the parallel-oriented component of the wire velocity. This constriction formation is unstable and collapses repeatedly. In this study, increasing velocity differences stabilizes the constriction existence. This stabilization correlates with a uniform wetting process and a reduced frequency of the occurrence of blowout events, probably due to improved outgassing of zinc bubbles from the melt pool.


Introduction
Laser material processing is a technology that is used for many applications in the production process of the automotive industry [1]. Here, laser brazing is used for the joining of car body parts [2], which are by default zinc-coated for corrosion protection [3]. Standard applications are a connection between the two-part trunk lid [4] or the connection of the roof and side panel [5]. The brazed seams remain directly visible (except for painting) to the customers [6], and hence, the highest optical quality standards are demanded for these seams [7]. This includes, for example, straight seam edges, a smooth seam surface, and the absence of adherent spatter or pores and holes [8].
To ensure optical seam quality, the wetting process in laser brazing is of major importance [9]. As an example, the mechanism of pore formation is described by Frank et al. as the formation of constrictions in the area of the wetting front as a consequence of a temporary interruption of the wire feed [10]. However, such a constriction formation is not only observed during disturbed process behavior. Grimm and Schmidt first reported these constrictions for a sound brazing process with constant process parameters, occurring at a frequency of approx. 30 Hz and inducing a stepwise progression of the wetting process [11]. They assume that the formation of constrictions depends on effects that are connected to the process velocity. This stepwise progression of the wetting front was also reported by Reimann, who found a wetting frequency of 39 Hz [12], and by Heitmanek et al., who determined a frequency of 42 Hz [13]. Further evaluation of the formation of constrictions was carried out by Heuberger et al. [14], who detected the constriction formation in a constant brazing process using high-speed recordings from the top view. Here, the length of the constrictions increased with increasing process velocity. At a critical length, the constrictions spontaneously collapsed and induced an abrupt propagation of the wetting process.

3
This abrupt collapse of the constrictions has been found to be a reason for defects concerning the optical seam quality, such as the formation of notches at the seam edges. Furthermore the collapse of constrictions was found as a cause for wavy seam edges due to an change in seam width during the abrupt progress of the wetting fronts [15].
Another factor influencing the wetting process and the optical seam quality is the zinc coating of the base material. Gatzen et al. showed that the molten zinc layer can merge with the melt pool during the brazing process [16]. The zinc, which has a comparatively low boiling temperature of 907 °C [17], can hereby evaporate during the process and create melt pool surface instabilities [14] and openings in the melt pool in combination with spatter formation [18].

Aim
As mentioned above, a constriction formation with subsequent collapse during laser brazing can induce an unsteadily progressing wetting process and lead to seam defects like pores, notches or wavy seam edges. However, it is currently not known what dominant mechanisms are responsible for this constriction formation. A possible explanation is that it originates from an unsuitable relationship between the feed rate of the filler wire and the process velocity. Therefore, this study examines the hypothesis that the formation of constrictions is induced by a mismatch in the relative velocity between the process movement and the velocity component of the wire feed oriented parallel to the process direction. This velocity difference is illustrated in Fig. 1.

Materials
The base material for the brazing experiments was the electrolytically galvanized steel DC06 + ZE75/75. The dimensions of the base material sheets were 150 mm in length, 50 mm in width, and 0.75 mm in thickness. The filler wire had a diameter of 1.2 mm and the material was CuSi3Mn1, which has a melting range of 965°-1035 °C and whose composition is shown in Table 1. The materials used are common in the industrial application of laser brazing in automotive body construction.

Brazing setup
A Trumpf TruDisk 12002 disk laser with a wavelength of 1030 nm was used as a laser beam source. The beam was transferred by a fiber with a core diameter of 0.6 mm to a Scansonic ALO3 brazing optic, which is commonly used for brazing in automotive production. In the brazing optic, an imaging ratio of 4:1 was realized, resulting in a nominal focus diameter of 2.4 mm. The focal plane of the laser beam was set on the surface of the workpiece in order to achieve a top-hat intensity profile at this position. The movement of the additional axes of the brazing optic, which are used in industrial production for tactile seam tracking, was blocked during this study. The wire nozzle, which is part of the brazing optic, had an inclination angle of 45° relative to the horizontal axis of the optic. To avoid back reflections from the  workpiece, the brazing optic was mounted to the processing station with an angle of 5° to the vertical, meaning the total wire angle relative to the process direction was 40°. The wire stick-out and the impinging point of the filler wire relative to the laser spot were calibrated using a setup spike provided by the optic manufacturer. The stick-out was set to 8 mm and the impinging point of the filler wire was positioned with a shift of 0.5 mm to the spot center in the direction of the wire nozzle. For wire feeding, a High-Dynamic-Drive (HDD) system with a push-pull configuration by Dinse was used. In contrast to the standard in industrial application, the process movement was not realized by moving the brazing optic but rather by moving the base material in the opposite direction; cf. Fig. 1. This enabled a permanent process monitoring of the wetting process by a stationary camera, whereby the position of the wetting process relative to the camera did not change. Using this setup, bead-on-plate experiments with seam lengths of 120 mm were performed. The experimental setup is illustrated in Fig. 2.

Process monitoring
The wetting process during the brazing experiments was recorded using a stationary high-speed camera (Vision Research Phantom VEO 410L), which was positioned orthogonal to the process direction and with a vertical inclination of 5° (cf. Fig. 2). A macro-objective in combination with an extension bellows was used to achieve a high magnification of the process zone. In order to eliminate any disturbing influences of the brazing process radiation, a Cavitar Cavilux HF illumination laser system was used, which illuminated the process zone with radiation of a central wavelength of 810 nm. By mounting an 810 ± 12 nm FWHM (Full Width at Half Maximum) bandpass filter in front of the camera, only the reflected radiation of the illumination laser was recorded by the camera. The parameters used for the recordings are given in Table 2.

Variation of the velocity difference Δv p
The experiments of this study included a variation of the velocity difference Δv p referring to the exemplary brazing setup given in Fig. 1. According to Eq. 1, where v p is the process velocity, v w is the wire velocity, and α is the constant inclination angle of the wire relative to the process direction, the velocity difference can be adjusted by changing the ratio between the process and wire velocity. Here, a positive value of Δv p means a higher process velocity compared to the component of the wire velocity oriented parallel to the process direction.
In the first instance, the wire velocity was set to a constant value of 50 mm/s and the velocity difference Δv p was varied from − 20 up to 90 mm/s. Here it has to be mentioned, that a change in Δv p also means a change in the seam volume due to a change in amount of filler wire per brazing length. Furthermore, experiments with Δv p = 0 for varied values of the wire velocity, starting from 25 up to 100 mm/s, were conducted. In all experiments, the energy per wire volume  was kept constant with a value of 40 J/mm 3 . Each experiment was carried out three times.

Evaluation of the wetting front (and wire) movement
The experiments described in the previous section were evaluated concerning the movement of the wetting fronts.
To visualize the wetting front behavior, streak images were created from the high-speed recordings. For the creation of streak images, one row in the recording was selected and extracted from each frame of the video, and then each was assembled one below the other in chronological order. When choosing a row in which the wetting process was visible, the corresponding streak image revealed the time-dependent movement of the wetting front. Due to the illumination by the illumination-laser system, the recorded image depended on the reflection of the radiation in the process zone. Hereby, the wetting fronts directly reflected the radiation into the camera, resulting in a local brightness maximum in the recordings. By tracing the right border of the brightness maximum, the wetting front movement could be extracted from the streak images (cf. Fig. 3). 3000 frames (0.3 s) were selected from the middle of the recording per experiment to eliminate side effects at the process start and end. Considering the movement of the wetting front as an oscillating process (as repeatedly assumed in the literature), the amplitude of the wetting front movement was determined. For this, each streak image was divided into ten equally sized parts and the distance between the first and the second visible local maxima was measured for each part. Values for the amplitude of the oscillation are calculated by dividing the measured values by two. The generation of streak images is also used to illustrate the movement of the filler wire over time. For this, a row where the lower border of the filler wire is visible and the wetting process is not visible (to prevent the lower edge from being covered by the wetting process) is selected (cf. Fig. 3).

Evaluation of the wetting front length
The value for the length of the wetting front is defined in this study as the distance between the right border of the wetting front and the impinging point of the lower border of the filler wire on the base material in the process recordings. This method allows the measurement independent of whether a constriction in the area of the wetting fronts is present or not. The rightmost measured values for each of the ten parts from the streak images (cf. previous section) were used to determine the position of the right border from the process recordings. The impinging point of the filler wire was determined from ten equally distributed frames from each process recording. Here, the intersection between the lower wire border with a line, defined by the projection of the wire axis on the base material, was considered, as illustrated in Fig. 4.

Evaluation of the blowout event density
Blowout events in the region of the wetting fronts, characterized by a disturbance of the melt pool surface and an explosive melt ejection, were visible in the streak images as horizontal lines because they only appeared in a short time frame (cf. Fig. 5). To determine a value for the density of these blowout events for the different experiments, the horizontal lines in the streak image were counted and this number was divided by the respective process velocity. Thereby, one value for the blowout event density was calculated for each experiment.

Results
In Fig. 6 the wetting process is visualized by the streak images for different values of Δv p , starting with a velocity difference of 0 mm/s and decreasing the value, resulting in a higher wire velocity in the process direction relative to the process velocity. The results only show small fluctuations in the movement of the wetting progress for varied Δv p , except for the lowest investigated value of − 20 mm/s, where irregular variations of the wetting front position occur.
Considering the same type of presentation for velocity differences from 0 to 90 mm/s, whereby the process velocity was higher than the component of the wire velocity in the process direction, major changes in the wetting process could be observed (cf. Fig. 7). Compared to the initial state without a velocity difference, the variation of the wetting front position increased in its amplitude up to Δv p = 30 mm/s. At a further increase in the velocity difference, the amplitude of the variation of the wetting front position decreased again. Besides the decrease in amplitude, an increase in the frequency of the wetting front movement could also be observed.
The measurements of the amplitude of the wetting front movement are displayed in Fig. 8. At Δv p = 0 mm/s, a wetting front amplitude of 0.06 ± 0.02 mm was obtained. With increasing velocity difference, the amplitude also increased up to a value of 0.13 ± 0.03 mm for Δv p = 30 mm/s. The lowest average value of 0.05 ± 0.03 mm was determined for a velocity difference of − 10 mm/s. Comparing the measuring results of the wetting front amplitude for Δv p = 0 mm/s with the highest investigated values for Δv p (≥ 60 mm/s), the average values were at the same level.
Besides the aforementioned movement of the wetting process, the shape of the wetting fronts was also further analyzed, as shown in Fig. 9. Without a velocity difference and with negative values for Δv p , the wetting fronts showed a steep angle, connecting the melt pool with the base material. A change in shape was detected when a positive velocity difference was present. Here, the formation of a constriction between the melting filler wire and the wetted base material was present. A further increase in the velocity difference resulted in an increased length of these constrictions.
Streak images, which illustrate the position of the filler wire during the brazing process for varied values of Δv p, are given in Fig. 10. For all investigated experiments, no displacement of the wire position during the process nor an oscillating or discontinuous behavior were detected.
The measuring results of the constriction length are presented in Fig. 11. The constriction length increased with increasing Δv p until a value of Δv p = 30 mm/s. For higher values for Δv p , no further steady increase can be observed. The highest constriction length was found for the highest investigated velocity difference of 90 mm/s with a value of 0.60 mm ± 0.05 mm. Even though no constriction formation could be detected for values of Δv p ≤ 0, a shortening of the wetting front at decreasing velocity differences was determined.
The results of the evaluation concerning the blowout event density from the area of the wetting fronts are presented in Fig. 12; the median value and the maximum and minimum deviations are given. The density tends to decrease with increasing velocity difference Δv p . At the highest investigated Δv p , the value of 0.08 + 0.03 mm −1 indicates that nearly no blowout events occurred.
The streak images for experiments with different wire and process velocities are presented in Fig. 13. The process velocity v p was adapted for four different wire velocities v w to obtain the constant relation of Δv p = 0 between v w and Δv p . Here, it can be seen that the variation of the wetting fronts was at the same level for the wire velocities of 50 mm/s, 75 mm/s, and 100 mm/s and that no significant differences in the wetting process could be observed. An exception was the experiment with the lowest investigated wire velocity of 25 mm/s, as here, apart from an otherwise constant wetting process, occasional changes of the wetting front position could be observed.
Experiments with different values for Δv p and v w were compared concerning the process velocity to separate the effects of a velocity difference Δv p from the effects of the absolute process velocity v p (cf. Fig. 14). Here, it can be seen that the lowest amplitude was obtained for Δv p = 0 mm/s (purple circles).
A direct comparison of the process recordings of the abovementioned experiments with comparable absolute process velocities in the range of 80 mm/s for different values of Δv p is presented in Fig. 15. Here, it can be seen that in the case of Δv p = 45 mm/s, a constriction formation

Discussion
Regarding the working hypothesis of this study, it was shown that a constriction formation occurs when the process velocity is higher than the component of the wire velocity in the process direction (cf. Fig. 9). To validate  this statement, the process velocity was varied in order to show that it is not the increase in the process velocity that causes the constriction formation. Here even with increased process velocity a prevention of constriction formation could be achieved by eliminating the velocity difference Δv p by adjusting the wire velocity (cf. Fig. 15 or also the online resources 1 and 2). Furthermore, the effect of a changing wire position (due to a disturbed melting behavior of the filler wire) was assumed as a possible reason for the constriction formation, but it can be proven that no change in wire position occurred during the experiments (cf. Fig. 10). Therefore, the resulting relative velocity in the process zone Δv p is identified as the main cause of the constriction formation. The constriction formation does not occur when the wire velocity in the process direction becomes higher than the process velocity (cf. Fig. 9). The effect that is responsible for the formation of constrictions is the consequently pulling forces from the workpiece on the melt pool of the braze seam and from that on the melting wire due to a too high process velocity or a too low wire velocity, respectively.  If the process velocity slightly exceeds the wire velocity in the process direction, a constriction formation occurs that is not stable (cf. Fig. 7 and online resource 1). These constrictions can collapse at irregular intervals, resulting in an increased variation of the wetting front position (cf. Fig. 8) and thus an abrupt and stepwise progressing wetting process. Furthermore, according to Heuberger et al., these constriction collapses can affect the optical seam quality and induce seam edge defects, such as notches [14]. In the experiments carried out, this regime, with unstable constriction formation and subsequent collapses, was identified between Δv p = 15 mm/s and Δv p = 45 mm/s (cf. Fig. 8). With a further increase in the velocity difference, the constriction formation stabilizes so that no further collapses of the constrictions occur (cf. Fig. 7 and online resource 3). This results in a more uniform wetting process (cf. Fig. 8). In this second regime, with a low wetting front fluctuation due to the stable constriction formation, the density of blowout   15 Direct comparison of the process recordings of two experiments that differed in Δv p but had comparable values for the process velocity v p events is, in addition, significantly reduced (cf. Fig. 12). The reason for these blowout events is considered to be the evaporation of low-boiling zinc through the melt pool, as described by Kimura et al. [18]. It is therefore plausible that the permanently open (cf. online resource 3) and elongated (cf. Fig. 11) constrictions at high values of Δv p support the outgassing of evaporating zinc so that the zinc vapor no longer has to escape through the melt pool or by blowout events.
With regard to a suitable process control for the industrial application of laser brazing, two strategies can be drawn from the presented findings: 1. In order to avoid the formation of constrictions, the relationship between the process velocity and the wire velocity component oriented parallel to the process direction can be adjusted to 0 mm/s, which has been proved to be effective for different process velocities from 19 to 77 mm/s (cf. Figs. 13 and 14). It must be taken into account that the wire volume per seam length then changes. This could possibly be compensated by the wire diameter, although the influences on the wetting behavior and the occurrence of blowout events were not considered in this study. 2. With a focus on achieving high process velocities, and therefore low cycle times in the production process, the objective in process control can be to braze in the supercritical area at a high velocity difference, whereby the constriction formation stabilizes. Hereby, a constant wetting progress as well as the absence of blowout events can be achieved.

Conclusion
The following conclusions can be drawn from the investigations: 1. The formation of constrictions in the wetting process during laser brazing is determined by the difference between the process velocity and the component of the wire velocity in the process direction. 2. If the process velocity exceeds the component of the wire velocity in the process direction, the formation of constrictions is observed.
a. For slightly higher process velocities, this constriction formation is not stable and the constrictions repetitively collapse resulting in an abrupt propagation of the wetting process. b. With further increases in the velocity difference, the constriction formation is stabilized and the constrictions do not collapse but remain open. This homog-enizes the wetting progress and reduces the occurrence of blowout events in the area of the wetting front.