Prod. Eng. Res. Devel.

(2017) 11:9–18
DOI 10.1007/s11740-016-0705-4

PRODUCTION PROCESS

Keyhole stability during laser welding—Part II: process pores

and spatters
Jörg Volpp1

Received: 27 April 2016 / Accepted: 18 November 2016 / Published online: 28 November 2016
 German Academic Society for Production Engineering (WGP) 2016

Abstract Spatters and pores that occur during laser deep 1 Introduction
penetration welding are unwanted defects that make post
processing necessary. Therefore, these defects need to be The dynamic behavior of the laser deep penetration process
avoided. The origin of pores and spatters are assumed to be plays an important role in the formation of process pores
in the process dynamics, especially the highly dynamic and spatters. In particular, the keyhole dynamics are
vapor channel. The keyhole is assumed to be responsible assumed to cause the process defects [1]. It is assumed that
for producing these defects. Based on keyhole calculations the origin of pores and spatters are the same, namely the
from Part I, in this paper, calculated keyhole properties are dynamics in the keyhole and melt pool. Pores can be
correlated to pore and spatter formation observed during formed when a bulging of the rear wall happens and col-
experiments. Spatters are recorded using a high speed lapses [2] or when the keyhole collapses and gas in the
camera and a spatter detection routine for measuring lower keyhole is trapped [3]. Spatters are assumed to be
spatter speed, size and number per time. Pore number per formed either from the bulging of the keyhole rear wall that
time and pore sizes are detected using X-ray imaging after affects the melt pool and can lead to melt ejection [4] or by
welding. Temporal spatter and pore characteristics can be material from the keyhole wall that is pushed out of the
correlated to dynamic values of the keyhole. The analytical keyhole due to the vapor flow in the keyhole [5].
model can predict spatter and pore formation depending on Therefore a main goal in keyhole welding is to obtain a
local frequencies and amplitudes in the keyhole. When better understanding of the dynamic behavior of the pro-
using different beam profiles, extremely high local fre- cess in order to find ways of minimizing the unwanted
quencies of the Top Hat beam at smaller keyhole dimen- effects of the dynamics. There are a lot of approaches that
sions seem to lead to an increased spatter and pore help reducing process defects. Spatters can be reduced
formation. The spring coefficient can be used as an indi- when welding through the specimen instead of partial (full
cator to predict spatter and pore characteristics. penetration welding) [6] or when welding in reduced
ambient pressure [7]. Pores that form after a keyhole col-
Keywords Laser deep penetration welding  Keyhole lapse can be reduced by widening the keyhole opening with
dynamics  Pore and spatter formation  Intensity an additional gas flow [8] or by modulation of the laser
distribution power [9]. In addition, pores were found to be reduced at
defocusing the laser beam slightly underneath the surface
of the specimen [1].
In Part I [10] and in the authoŕs earlier works [11] it
could be shown that the spatial laser intensity distribution
also influences the keyhole behavior.
& Jörg Volpp In the present paper the effects of different keyhole
volpp@bias.de
properties on the process defects are observed in order to
1
Bremer Institut für angewandte Strahltechnik GmbH, determine which keyhole properties may decrease the
Klagenfurter Straße 2, 28359 Bremen, Germany occurrence of process defects.

123
10 Prod. Eng. Res. Devel. (2017) 11:9–18

2 Methodology captured by the camera. Images from the videos are pro-
cessed using Matlab (Version R2009a). More than 500
2.1 Modeling keyhole dynamics frames ([0.25 s process time) are used for analysis from
each video. The routine loads every image separately in
The model used for determination of keyhole properties is order to detect the single spatters (Fig. 2). A transformation
described in detail in Part I [10] based on previous works of the grey scale image to a black/white (B/W) binary
where it could be shown that the model can be used also for image is used by setting a brightness dependent threshold
bigger spot sizes and laser powers [11]. Modeling is done of a grey scale level above which the value is set to be
in two steps. First, the quasi-static keyhole properties are white and recognized as a spatter. The plume is removed
calculated based on the energy balance on cylindrically from the image by blackening the lower part of the images.
assumed keyhole elements of the same height. The energy The spatter size is recorded for each found spatter and the
input is calculated from the actual spatial laser intensity average size of all found spatters in the video is calculated.
distribution of the laser beam. Radii of the individual In addition, the spatter distribution in each image is
keyhole sections and the spring coefficients, determining compared to the spatter distribution in the previous image.
the pressure change at radius deviation, are calculated from Each individual spatter in the current distribution is related
the pressure balance on the keyhole walls. The average to the closest individual spatter in the previous frame. In
spring coefficient is calculated from the single values of all case of a higher spatter amount in the new image, addi-
keyhole sections. tional spatters that could not be related to an individual
Based on the keyhole shape the dynamic model is built spatter in the previous image are determined as newly
using a set of differential equations that consider influences developed spatters. That way, the absolute number of
on the radial keyhole wall movement and keyhole pressure. spatters can be recorded and the number of spatters per
Characteristic oscillations can be found, namely the fre- time can be evaluated. As the time interval between the
quencies and amplitudes of the radial keyhole section wall images and the geometrical distance between the spatters is
movement. The average and maximums of these dynamic known, the projected spatter speed in the image plane can
characteristics as well as the values in the single keyhole be calculated for each spatter. Due to the observation
sections are evaluated. The keyhole is split in 10 sections, method only 2-D-frames can be captured and the contri-
while the keyhole tip is named number 1. bution of the speed in the third dimension cannot be con-
sidered. The value of the calculated spatter speed will be
2.2 Spatter detection underestimated.

Welding experiments are conducted while a high-speed 2.3 Pore detection

camera is installed to observe the spatters above the
specimen (Fig. 1). After welding, the weld seams are analyzed by a subse-
Spatters are detected in high-speed-videos (recording quent X-ray analysis for pore detection. The weld seams
frequency 6 kHz; resolution 256 9 512 pixel) taken from are scanned from the top of the weld (Fig. 3).
the side of the process to observe the region above the
keyhole. Illumination with a pulsed laser (808 nm wave
length) is used in combination with a notch filter on the
camera, so only reflected light of the illumination laser is

Fig. 1 Welding setup with high-speed-camera observation Fig. 2 Recognition procedure for spatter detection

123
Prod. Eng. Res. Devel. (2017) 11:9–18 11

 2
PL  r2rL0
igs ðr Þ ¼ 2
e ð2Þ
p  rL0
 r  r 
PL 1 1 L0
ith ðr Þ ¼ 2
 arctan ð3Þ
p  rL0 2 p k
Table 1 shows the reference parameter set. From these
parameters the laser power (from 800 to 1000 W), welding
speed (from 0.5 to 2 m/min) and focal position (from -2 to
?1 mm) are varied. No shielding gas has been used. The
used base material is aluminum alloy EN AW-1050.
Fig. 3 Pore detection method
Specimens are of the size 40 mm 9 6 mm 9 100 mm.
Density differences in the weld seams are visible in the
pictures and can be analyzed using a Matlab (Version
R2009a) routine that calculates the size and number of the 3 Results
pores. The pore number is used to calculate the pores per
second PPS (Eq. 1). 3.1 Spatter evaluation
welding velocity
PPS ¼ pore number  ð1Þ Spatter characteristics are evaluated at all varied process
seam part length parameters. In this chapter the influence of the laser
power is shown as an example, while the results of the
2.4 Parameters and material varied welding velocity and focal position are shown in
‘‘Appendix’’. Figure 5 shows the results of the spatter
The same parameters as in Part I [10] are used for the evaluation at different laser power. The Top Hat beam
evaluation. A Gaussian beam profile (single-mode fiber leads to more spatters compared to a Gaussian beam at
laser IPG YLR-1000SM) is collimated with a lens of the same parameters. The spatter ejection frequency
160 mm focal length and focused with a lens of 560 mm (spatters per second) show that for a Top Hat beam at
focal length (Fig. 4a). A Top Hat intensity distribution increasing laser power more spatters are produced
(Fig. 4b) is created using a beam shaper (Focal pi-shaper (Fig. 5). For a Gaussian beam there is a minimum
by Adloptica) as collimator. The beam is focused with a spatter ejection frequency at 900 W. At different weld-
lens with a focal length of 200 mm. ing velocities no significant tendency of ejected spatters
Equations 2 (Gaussian, gs) and 3 (Top Hat, th) are used per second can be observed. A slight increase is
for modeling the beam profiles ij including the laser power observed at 1 m/min welding velocity for both beam
PL and the wave length k. profiles (Fig. 18a). Similar to the results from the vari-
ation of the welding speed there is no relevant visible
tendency when varying the focal position. At a focal
position underneath the surface at zf = -1 mm a slight
minimum of the spatters per time are observed for a Top
Hat beam (Fig. 18b).
Spatter sizes are evaluated at different welding
parameters. The average spatter sizes are shown in
Fig. 6 including the minimum and maximum measured
values at different laser power. Maximum spatter sizes at
Gaussian beam welding are in general much smaller than
at Top Hat welds. Besides small spatters, the number of
big spatters increases at a laser power of 1000 W at both
beam profiles.

Table 1 Reference parameters

Laser power PL 1 kW
Welding velocity v 1 m/min
Fig. 4 Measured intensity profiles: a Gaussian beam profile and Focal position zf -2 mm
b Top Hat beam profile

123
12 Prod. Eng. Res. Devel. (2017) 11:9–18

Fig. 5 Spatters per second at different laser power and beam profiles
(based on reference parameters, Table 1; EN AW-1050)
Fig. 7 Spatter speeds at different laser power and beam profiles
(based on reference parameters, Table 1; EN AW-1050)

In high speed videos different process behaviors could

be observed which lead to different spatter formation.
Typical spatter ejections are shown in Figs. 8 and 9.
Figure 8 shows spatters which are randomly ejected in an
eruptive way during an otherwise stable process. The sur-
face of the melt pool is smooth but lots of small, fast
spatters are ejected from the vicinity of the keyhole. This
effect can be seen in the Gaussian welds at low laser power
(800 W, Fig. 8, and 900 W), at a low welding velocity
(0.5 m/min) or at a focal position of zf = 0 mm. At Top
Hat welding this effect is seen at zf = -1 mm. A high
number of little solidified spatters can be found on the
Fig. 6 Spatter sizes at different laser power and beam profiles (based surface.
on reference parameters, Table 1; EN AW-1050) Figure 9 shows the occurrence of bigger, slow spatters
that are produced from swellings on the melt pool surface
which are ejected by the high acceleration in the melt pool
No clear trend in spatter size is seen when welding at and the metal vapor from the keyhole. These spatters
different welding velocities (Fig. 19a). At a focal position appear continuously. This spatter formation effect occurs at
of zf = -2 mm (underneath the surface) highest maximum high laser power (1000 W) at Gaussian beam welding and
spatter sizes are found for both beam profiles. Smallest mainly at Top Hat welding (e.g. at 0.5 m/min welding
maximum spatter sizes are observed at zf = 0 mm for a velocities, Fig. 9).
Gaussian beam and at zf = -1 mm for a Top Hat beam In general, during Top Hat welding the occurrence of
profile (Fig. 19b). big, slow spatters is increased which is an indicator that
The average spatter speed for all detected spatters is another spatter formation mechanism dominates the pro-
evaluated including the minimum and maximum values cess compared to the Gaussian beam processes where
(Fig. 7). In general, Gaussian beam welds show higher mainly eruptive ejections are observed. Mainly big spatters
spatter speeds compared to the Top Hat welds. Maximum
spatter speed is observed at 800 W laser power at a
Gaussian beam profile while the maximum and average
speed decrease at higher laser power. The spatters pro-
duced with a Top Hat beam show, in general, lower
maximum spatter speeds (Fig. 7).
When varying the welding velocity the spatter speed is
mainly constant with a significant high value at 0.5 m/min
for both beam profiles (Fig. 20a). Lowest spatter speed of
all probes is observed at a focal position of zf = -1 mm
Fig. 8 Spatter formation observation in high speed video pictures at
for both beam profiles at otherwise constant values Gaussian beam profile at 800 W laser power (based on reference
(Fig. 20b). parameters, Table 1; EN AW-1050)

123
Prod. Eng. Res. Devel. (2017) 11:9–18 13

Fig. 9 Spatter formation observation in high speed video frames at

0.5 m/min welding velocity (Top Hat beam, based on reference
parameters, Table 1; EN AW-1050)

from melt pool swellings are visible but also smaller

spatters are found.
Fig. 11 Pore sizes at different laser power and beam profiles (based
3.2 Pore evaluation on reference parameters, Table 1; EN AW-1050)

Pores per second are evaluated while, in general, a Gaus- for a Top Hat beam at otherwise low pore sizes (Fig. 22a).
sian beam weld shows fewer pores per second than a Top Minimum pore size is found at zf = 0 mm for a Gaussian
Hat weld (Fig. 21). For a Gaussian beam the pores per beam and at zf = -1 mm for a Top Hat beam (Fig. 22b).
second decrease at increasing laser power (Fig. 10). Eval-
uation of the Top Hat beam shows values in the same range
with a minimum at 900 W (Fig. 10). 4 Discussion
When welding at different welding velocities the
Gaussian welds show a local maximum at 1 m/min and a 4.1 Spatters and porosity
drop to very low porosity at higher welding velocities. The
Top Hat welds show an increase of porosity at higher In this paper a correlation between keyhole properties and
welding velocities (Fig. 21a). At varied focal position local the occurrence of pores and spatters is investigated. The
minima of porosity are detected at zf = 0 mm for a occurrence of pores and spatters can be summarized as
Gaussian and zf = -1 mm for a Top Hat beam (Fig. 21b). follows:
Comparing the pore sizes at different beam profiles no
clear trend is seen. In general, higher maximum pore sizes • At Gaussian beam welding small, fast spatters occur in
are found in Gaussian welds. Maximum and average pore an eruptive way.
sizes increase at higher laser power for a Gaussian profile, • When using a Top Hat beam profile, the spatter number
while for a Top Hat profile a maximum is found at 900 W increases, they become bigger and slower.
(Fig. 11). At 1 m/min welding velocity the maximum pore • When using a Top Hat beam, a tendency to higher pore
sizes are detected for a Gaussian beam and at 0.5 m/min number is visible compared to a Gaussian beam.
According to literature, it can be assumed that the
keyhole dynamics are the origin of process defects. This
means that spatters and pores should have the same origin.
In this work, it seems that different effects lead to the pore
and spatter formation as tendencies in its occurrence vary.
This leads to the conclusion that the formation mechanisms
are different or different kinds of spatters and pores exist.
Kaplan and Powell [12] found that spatters can arise from
the keyhole side or front wall as ejections due to the vapor
flow or by melt pool movement leading to swellings on the
surface and to a detachment of bigger melt parts. Pores can
either be formed due to a bulging of the keyhole rear wall
or due to a keyhole collapse. In this work, the results of the
modeling (frequencies, amplitudes and spring coefficients)
Fig. 10 Pores per second at different laser power and beam profiles are correlated to the pore and spatter formation
(based on reference parameters, Table 1; EN AW-1050) characteristics.

123
14 Prod. Eng. Res. Devel. (2017) 11:9–18

4.2 Keyhole dynamics and spatter formation

It is assumed that certain keyhole properties have to be

present for a certain dynamic behavior leading to process
defects like pores and spatters.
In Part I [10] it could be concluded that besides local
effects the spring coefficient on the keyhole wall influences
the keyhole dynamics. A high spring coefficient is an
indicator for a high back pressure to the equilibrium state at
radius deviation. High local spring coefficients could be
related to high frequencies at low amplitudes in that key-
hole section. When a Top Hat beam is considered, the
frequencies are calculated to very high local values com-
pared to a Gaussian beam but the keyhole diameters
decrease. In this work, global and local calculated keyhole
properties are correlated to the pore and spatter formation.
Figure 12 shows the significant values from the model
that can predict the ejected spatters per second. Small
amplitudes in the top element (material surface, Fig. 12a)
and high maximum frequencies (occurring in the middle
Fig. 13 Spatter size correlated to a maximum amplitudes and
elements of the keyhole, Fig. 12b) correlate to a high
b average frequencies (EN AW-1050 base material; varied laser
number of ejected spatters. It can be assumed that the high power, welding speed and focal position based on reference
frequencies lead to fast occurring little bumps on the parameters, Table 1)
keyhole wall that can be detached by the impulse of the
melt and can be ejected by the shear stress of the metal impulse is high enough to directly eject spatters from the
vapor. keyhole wall surface, while at low frequencies and high
The small bumps (small amplitudes) on the keyhole wall amplitudes smaller spatter sizes arise (Fig. 13a, b). These
occurring at high frequencies seem to result in comparably occur in an eruptive way (Fig. 8). It can be assumed that
big spatter sizes (Fig. 13a). However, it seems that the the impulse is not high enough to eject a big amount of
material but the vapor is able to detach small amounts of
melt [13] when the melt oscillates in the keyhole. It has
been recently found that the keyhole surface shows a cer-
tain micro-roughness [14] that would offer points of attack
to the vapor for detachment of micro-ejections. This can
explain the eruptive character of the ejections in the
otherwise stable process.
The high spatter speed of the eruptive ejections can be
correlated to low (absolute) spring coefficients (Fig. 14).
On the other hand, this means that a stiff keyhole (high

Fig. 12 Spatters per second correlated to a amplitudes (top keyhole

element) and b maximum frequencies in the keyhole (EN AW-1050 Fig. 14 Spatter speed correlated to the average spring coefficient
base material; varied laser power, welding speed and focal position (EN AW-1050 base material; varied laser power, welding speed and
based on reference parameters, Table 1) focal position based on reference parameters, Table 1)

123
Prod. Eng. Res. Devel. (2017) 11:9–18 15

absolute spring coefficients) leads to big keyhole wall Fig. 15a) seem to support the entry of small amounts of
ejections at comparably lower speed. ambient gas which produce more pores per second.
The observed bulging of the melt pool on the surface Melt pool wave frequencies were observed to oscillate at
(Fig. 9) cannot be directly described by the model as melt 100 to 600 Hz [16]. Therefore, also the big waves of the
pool oscillations are not calculated. However, it can be melt pool oscillation seem to be not the main initiation of a
assumed that the high frequencies occurring at high (ab- situation that leads to a trapping of ambient gas. It can be
solute) spring coefficients lead to a superposition of assumed that an interaction of oscillations of different
oscillations that can be transported to the material surface amplitudes and frequencies from the keyhole and from the
where it can produce the bulging when hitting the melt pool melt pool lead to situations that enable gas suction and gas
waves. trapping in the keyhole.
When correlating the results of the model and the pore
4.3 Keyhole dynamics and pore formation size it can be found that a stiffer keyhole (higher absolute
spring coefficient) lead to a smaller pore size compared to
Calculated keyhole properties are correlated to pore lower (absolute) spring coefficients (Fig. 16). A stiffer
characteristics. keyhole is more resistant against keyhole diameter chan-
At high occurring maximum frequencies in the keyhole ges, which results in smaller bulging. A higher (absolute)
also a high number of pores per second is ejected (Fig. 15). spring coefficient necessitates smaller keyhole diameters
The highest frequencies usually occur in one of the sections [10] which can be the reason for a collapsing keyhole
in the middle of the keyhole [10]. However, the frequencies although the keyhole is stiffer. Due to the smaller volume
in the keyhole are much higher than the pore occurrence of the keyhole, smaller pores are formed. The smaller pores
frequencies. It is assumed that not all keyhole diameter appear more often due to the high occurring frequencies
reductions in the collapse endangered sections in the (Fig. 15a).
middle of the keyhole [15] lead to a collapse of the keyhole
and not all collapses lead to porosity. In order to produce a 4.4 Beam profiles
pore, ambient gas needs to be trapped in the keyhole [2]
which seems not to happen at every keyhole closure pro- Based on the comparisons of calculated process properties
cess. The amplitudes of the surface near section seem to and spatter and pore occurrence, a process characteristic
correlate to the occurrence of pores per second (Fig. 15b). diagram is developed that shows the occurrence of process
Small amplitudes on the surface (at high frequencies, defects depending on the process properties (Fig. 17). At
values close to the center of the web the occurrence of the
influenced spatter or pore characteristic is low. The further
away from the center the value is calculated the higher the
occurrence of the characteristic.
In general, the beam profiles show different keyhole
properties and pore and spatter formation characteristics.
When using a Top Hat beam the local frequencies are
calculated to higher and the amplitudes to lower values
compared to a Gaussian beam while the spring coefficients
are much higher. For the Top Hat beam welds this results

Fig. 15 Pores per second correlated to a maximum frequencies and

b amplitudes in the top keyhole element (EN AW-1050 base material; Fig. 16 Pore size correlated to the average spring coefficient (EN
varied laser power, welding speed and focal position based on AW-1050 base material; varied laser power, welding speed and focal
reference parameters, Table 1) position based on reference parameters, Table 1)

123
16 Prod. Eng. Res. Devel. (2017) 11:9–18

2. High frequencies at small amplitudes result in

spatter detachment directly from the keyhole wall
due to the high impulse on the keyhole wall. This
effect is mainly observed at Top Hat beam welding.
These spatters show bigger sizes and smaller speed
compared to the eruptive ejected spatters at Gaus-
sian welding.
• High frequencies at small amplitudes in local keyhole
sections seem to support the gas entrapment into the
keyhole leading to a high amount of pores produced per
second. However, the high frequencies due to high
(absolute) spring coefficients result in a stiffer keyhole
which leads to a small pore size. This effect is mainly
observed at Top Hat beam welding.

Acknowledgements This work was accomplished within the Center

Fig. 17 Process characteristics at Gaussian and Top Hat beam of Competence for Welding of Aluminum Alloys (Centr-Al). Funding
profiles (at reference parameters, Table 1) by the DFG—Deutsche Forschungsgemeinschaft (VO 530/52-2) is
gratefully acknowledged. The ‘‘BIAS ID’’ numbers are part of the
figures and allow the retraceability of the results with respect to
mandatory documentation required by the funding organization.
in more spatters and more pores per second and in bigger
spatter sizes, smaller spatter speeds at smaller pore sizes.
Different kinds of spatters are produced for the different Appendix
beam profiles (Figs. 8, 9). Due to the different keyhole
properties it is assumed that different kinds of spatters See Figs. 18, 19, 20, 21 and 22.
occur for the different beam profiles. In general, the Top
Hat welds show more spatters per second (Fig. 5) at bigger
size (Fig. 6) but lower speed (Fig. 7). This is an indicator
that more swellings are produced that lead to the occur-
rence of bigger but slower spatters at Top Hat welding
compared to Gaussian beam welding. The high occurring
maximum frequencies at Top Hat welding seem to lead to a
higher impulse on the melt and result in the permanently
ejected, slow, big spatters. The eruptive ejections are
mainly observed at Gaussian welding.

5 Conclusions

• The calculated keyhole properties can be used to pre-

dict the stability of the keyhole against spatters and
pores.
• Two mechanisms of spatter ejections are observed:
1. Eruptive spatter ejections appear in an otherwise
stable process which can be correlated to big
amplitudes and low frequencies in the keyhole. A
detachment of micro-ejections from exiting vapor Fig. 18 Spatters per second at different a welding velocities and
is assumed. This effect is mainly observed at b focal positions (based on reference parameters, Table 1; EN AW-
1050)
Gaussian beam welding.

123
Prod. Eng. Res. Devel. (2017) 11:9–18 17

Fig. 21 Pores per second at different a welding velocities and b focal

Fig. 19 Spatter sizes at different a welding velocities and b focal positions (based on reference parameters, Table 1; EN AW-1050)
positions (based on reference parameters, Table 1; EN AW-1050)

Fig. 20 Spatter speeds at different a welding velocities and b focal Fig. 22 Pore sizes at different a welding velocities and b focal
positions (based on reference parameters, Table 1; EN AW-1050) positions (based on reference parameters, Table 1; EN AW-1050)

123
18 Prod. Eng. Res. Devel. (2017) 11:9–18

References and electro-optics (ICALEO), LIA congress proceeding. Paper

1604
1. Mizutani M, Katayama S (2003) Keyhole behavior and pressure 8. Fabbro R (2009) Limiting process for keyhole propagation during
distribution during laser irradiation on molten metal. In: Pro- deep penetration laser welding. In: Proceeding of 5th interna-
tional congress on laser advanced materials processing (LAMP)
ceedings of the 22nd international congress on applications of
lasers and electro-optics (ICALEO), LIA congress proceeding. 9. Kawaguchi I, Tsukamoto S, Honda H, Arakane G (2003) Power
Paper 1004 modulation in deep penetration laser welding—optimization of
2. Berger P, Hügel H, Graf T (2011) Understanding pore formation frequency and waveform to prevent the porosity. In: Proceedings
of 23rd international congress on applications of lasers and
in laser beam welding. In: Schmidt M, Zaeh MF, Graf T,
Ostendorf A (eds) Lasers in manufacturing (LIM), Physics Pro- electro-optics (ICALEO), LIA congress proceeding. Paper 1006
cedia, vol 12. Elsevier, Amsterdam, pp 241–247 10. Volpp J, Vollertsen F (2016) Keyhole stability during laser
3. Tsukamoto S, Kawaguchi I, Arakane G, Honda H (2002) Keyhole welding—Part I: modeling and evaluation. Prod Eng Res Devel
behavior in high power laser welding. In: Proceedings of the 10:443–457
international congress on laser advanced materials processing 11. Volpp J, Vollertsen F (2015) Modeling keyhole oscillations
(LAMP). International Society for Optics and Photonics during laser deep penetration welding at different spatial laser
4. Fabbro R, Slimani S, Doudet I, Coste F, Briand F (2006) intensity distributions. Prod Eng Res Dev 9(2):167–178
Experimental study of the dynamical coupling between the 12. Kaplan AFH, Powell J (2011) Spatter in laser welding. J Laser
induced vapour plume and the melt pool for Nd–Yag CW laser Appl 23(4):032005-1–032005-7
welding. J Phys D Appl Phys 39:394–400 13. Berger P, Schuster R, Hügel H, Graf T (2010) Moving humps at
the capillary front in laser welding. In: Proceedings of the 30th
5. Zhang M, Chen G, Zhou Y, Li S (2013) Direct observation of
keyhole characteristics in deep penetration laser welding with a international congress on applications of lasers and electro-optics
10 kW fiber laser. Opt Express 21(17):19997–20004 (ICALEO), LIA congress proceeding. Paper 106
6. Seto N, Katayama S, Matsunawa A (2000) Porosity formation 14. Kaplan AFH (2015) Local flashing events at the keyhole front in
mechanism and suppression procedure in laser welding of alu- laser welding. Opt Lasers Eng 68:35–41
minium alloys. Q J Jpn Weld Soc 18:243–255 15. Boley M, Weber R, Graf T (2015) Online detection of pore
7. Börner C, Dilger K, Rominger V, Harrer T, Krüssel T, Löwer T formation during laser deep-penetration welding. In: Proceeding
(2011) Influence of ambient pressure on spattering and weld seam of lasers in manufacturing congress (LIM)
quality in laser beam welding with the solid-state laser. In: Pro- 16. Geiger M, Kägeler C, Schmidt M (2008) High-power laser
ceedings of 31st international congress on applications of lasers welding of contaminated steel sheets. Prod Eng Res Dev
2:235–240

123