Metal Transfer during GMAW with
Thin Electrodes and Ar-CO2
Shielding Gas Mixtures
Droplet diameters do not decrease proportionally with electrode diameters as
small as 0.016 in. and repelled transfer mode is dominant above 30% CO2
BY E. J. SODERSTROM AND P. F. MENDEZ
ABSTRACT. Metal transfer modes in gas controlled by several parameters, includ- Previous Research on Metal Transfer
metal arc welding (GMAW) using direct ing current, voltage, polarity, electrode ex-
current electrode positive (DCEP) in a va- tension, shielding gas composition, and The American Welding Society (AWS)
riety of binary Ar-CO2 shielding gas mix- electrode diameter. and the International Institute of Welding
tures and electrodes with diameters as The choice of shielding gas affects (IIW) have classified metal transfer into
small as 0.016 in. (0.41 mm) were investi- welding quality through its influence on different categories (Refs. 1, 2). For
gated. Droplet detachment frequency was metal transfer and also has a direct impact GMAW applications, the two main cate-
determined by analyzing the voltage signal on welding costs. CO2 is more plentiful, gories are short circuiting and free flight.
with Fast Fourier Transform, and the re- widely available, and less expensive than Short circuiting transfer is characterized
sults were verified with high-speed laser argon; however, weld bead quality and de- by the electrode periodically contacting
shadowgraph techniques. A newly de- position rates often decrease with the in- the weld pool. The electrode never con-
signed contact tip was used with the crease of CO2 in a binary Ar-CO2 mixture. tacts the weld pool during free-flight
thinnest electrode that improved process Currently, argon costs two to three times transfer; molten droplets detach from the
stability at feed rates up to 1443 in./min. It as much as CO2. If it were possible to cre- electrode, travel through the arc, and are
was found that the average droplet diam- ate welds of the same quality and deposi- deposited on the base metal. Free-flight
eter in the spray region did not decrease tion rates of high Ar mixtures with less ex- transfer is usually divided into subcate-
proportionally with electrode diameters. pensive mixtures containing significant gories that include globular, spray, rotat-
When using small-diameter electrodes amounts of CO2, the savings in the weld- ing, and repelled (nonaxial globular). Ob-
(<0.035 in.) with shielding gas mixtures ing industry would be substantial. servations of metal transfer during
containing less than 30% CO2, the average This research studied the transition be- GMAW began to be published in the
droplet diameters did not become smaller tween globular and spray transfer in Ar- 1950s and continue to be of interest in re-
than electrode diameters, regardless of CO2 atmospheres. What differentiates search because of its direct application to
the current used. Repelled transfer was this work from previous investigations is industrial conditions.
dominant with shielding gas compositions that behavior of thin electrodes (as small A key aspect of free-flight metal trans-
containing more than 30% CO2 regardless as 0.016 in. diameter) is explored. The mo- fer in GMAW is the existence of a rela-
of electrode diameter. tivation to study these thin electrodes is tively sharp “transition current.” Among
the possibility that they might avoid glob- the first to report on this transition current
Introduction ular and repelled transfer, as is discussed were Muller, Greene, and Rothschild
later. Our experiments show that this is (Ref. 3). They used DCEP polarity and
Gas metal arc welding (GMAW) is cur- not the case with DCEP welding, but that showed that metal transfer is influenced
rently the most widely used arc welding an unexpected metal transfer mechanism by the type of shielding gas, the electrode
process in industry. Benefits such as high occured. To the best of the authors’ knowl- composition, welding current, voltage,
production rates, high weld quality, ease edge, there is no published research on and electrode extension.
of automation, and the ability to weld metal transfer with electrodes smaller Below the transition current, metal is
many metals make it attractive to manu- than 0.030 in. (0.76 mm). transferred in the form of large droplets
facturers. One of the unique characteris- (globular transfer). In this regime, the di-
tics in this process is the way molten metal ameter of the droplets is often larger than
is transferred across the arc. The transfer KEYWORDS the wire diameter, and the frequency of
of metal from the electrode to the work- droplets is relatively low. Above the tran-
piece influences penetration, bead mor- Gas Metal Arc Welding sition current, metal is transferred as small
phology, fume generation, process stabil- Droplet Detachment droplets at a relatively high frequency
ity, and spatter. Metal transfer is DCEP (spray transfer).
CO2 Shielding Gas For example, Lesnewich (Refs. 4, 5) re-
Contact Tip ported a transfer rate of 5 Hz for droplets
E. J. SODERSTROM (firstname.lastname@example.org) is Metal Transfer Modes of 0.16 in. diameter in the globular regime
a graduate student and P. F. MENDEZ with 0.0625-in.- (1.59-mm-) diameter elec-
(email@example.com) is assistant professor,
trode, argon+1%O2 shielding. In the
Colorado School of Mines, Golden, Colo.
spray transfer regime, he reported a de-
124 -s MAY 2008, VOL. 87
Fig. 1 — Effect of shielding gas composition on transition region using Fig. 2 — Schematic showing the effects of increased welding current in
0.0625-in.- (1.59-mm-) diameter electrodes. Increasing amounts of CO2 in- argon. The current density at the anode spot (dotted lines) remains con-
crease transition currents until the transition from globular to spray is re- stant. When current increases, the anode spot area increases.
placed by the occurrence of repelled transfer (Ref. 11). 1<C1<C2<C3. Transfer mode changes when the anode spot envelops
Fig. 3 — Schematic showing the effects of CO2 on current density and metal Fig. 4 — Schematic drawing showing the effects of decreasing electrode
transfer. 1>K1>K2. The current density becomes higher with increasing diameter. The current density remains constant, but a change in transfer
amounts of CO2, resulting in a smaller anode spot at the same current, and mode occurs due to position of the arc attachment point.
changing the transfer mode from spray to repelled.
tachment frequency of 240 Hz for droplets theory (PIT) and static force balance the- mental values in the transition region
of 0.04 in. diameter. The maximum mea- ory (SFBT). Rhee and Kannatey-Asibu (Ref. 7). A comprehensive model for
sured current for pure globular transfer (Ref. 6) analyzed both and found that metal transfer in GMAW was developed
was 255 A, and the minimum measured SFBT gives good predictions for the glob- by Hu and Tsai (Ref. 8); with it, they were
current for pure spray transfer was 265 A. ular regime, and PIT is better for the spray able to predict droplet size, detachment,
Lesnewich defined the transition current regime. During the transition from globu- and velocity for both constant and pulsed
as the average between these limiting val- lar to spray, neither theory by itself is ac- current conditions.
ues (260 A in this case). In this work, we curate in predicting metal transfer. Re-
use the same definition. cently, computer models have enabled a Previous Research on Carbon Dioxide
The exact mechanism that causes the better understanding of the physics of in Shielding Gas
transition in metal transfer mode is not yet metal transfer. A GMAW simulation
fully understood; however, two main the- model based on computational fluid dy- Cost savings were identified as early as
ories have provided good results in the namics generated predictions of droplet 1956 when Rothschild (Ref. 9) reported
analysis of metal transfer: pinch instability diameters that agreed well with experi- that CO2 shielded arc welding was a feasi-
WELDING JOURNAL 125 -s
namic processes that ling the current is one of the main advan-
occur in a CO2 arc are tages of pulsed power supplies because it
concentrated within a directly affects the EM forces and metal
narrower region transfer mode. Needham and Carter (Ref.
when compared to 18) showed it is possible to have better
argon. The interac- quality welds made at larger deposition
tions between the rates in CO2 when using pulsed current.
plasma and the elec- Nonetheless, this new technology was un-
trode, including able to surpass the deposition rates and
chemical reactions, cleanliness of welds made with argon-
spatter formation, based shielding gases. Later, Matsuda et
and droplet detach- al. (Ref. 19) used an adjustable rectangu-
ment are significantly lar-wave pulse machine to create welds in
changed with addi- CO2. Using a 0.045-in.- (1.14-mm-) diam-
tions of CO2. Nem- eter electrode, mean current of 250 A, and
chinsky (Ref. 14) de- pulsing frequencies of 38 Hz, welds were
veloped a model for made in CO2. This reduced the spatter to
current conduction in 20% of the nonpulsed process.
Fig. 5 — Experimental matrix used in this research. Solid circles represent the near-anode Current state-of-the-art processes for
full data collection parameters and hollow circles represent exploratory ex- plasma layer and GMAW with CO2 do not use free-flight
periments. The transition line marks the area of mixed transfer, where both compared current transfer; instead, they use advanced short
nonaxial and axial detachments are observed. distributions between circuiting metal transfer. Sophisticated
argon, helium, and computer program control of welding pa-
molecular gases. The rameters enables waveforms never before
ble alternative to argon during the joining results agree with observations that arc used in GMAW. In this technology, the
of mild steel. Using uncoated steel elec- constriction and current distribution over droplet is formed with a large current
trodes and nonpulsed power supplies with the anode surface is controlled by the pulse, dipped into the weld pool, and de-
DCEP, Smith (Ref. 10) reported stable, plasma gas. Pires et al. (Ref. 15) recently tached by surface tension. Power source
axial types of free-flight transfer when the proposed a similar model involving arc en- manufacturers offer several variations on
CO2 concentrations are less than 25%. velopment and anode spot contraction this technology, which modulates the
Above 25% CO2, the operating character- that was supported with experiments using power input very quickly in order to re-
istics of the process changed to repelled seven different Ar-O2-CO2 gas mixtures. duce the amount of spatter. Miller Elec-
transfer during free-flight mode; however, Two approaches to improve free-flight tric’s Regulated Metal Deposition
using short circuiting transfer, quality metal transfer with CO2 levels above 25% (RMD™) and Lincoln Electric’s Surface
welds could be made with 100% CO2 at have been proposed. The first approach Tension Transfer (STT®) are two varia-
decreased deposition rates. uses an electrode with dilute coatings, and tions of this technology available for use in
The addition of CO2 in the shielding the second approach uses pulsing power industry. Deposition rates are approaching
gas increases the transition current and supplies. Dilute coatings of alkali and rare those of free-flight transfer, but limiting
decreases the maximum droplet detach- earth metals on the electrode were inves- factors such as stubbing and arc length
ment frequency, as shown in Fig. 1. Re- tigated at Airco by Lesnewich (Ref. 16) changes become prominent at high wire
searchers (Ref. 11) conducted experi- and Cushman (Ref. 17). They developed feed speeds. Recent developments com-
ments using varying compositions of an electrode that operated in 100% CO2 bine the use of pulsing and high-speed re-
shielding gas with 0.0625-in. (1.59-mm) shielding gas that gave stable metal trans- versible wire feeding to address these is-
steel electrodes. These changes are appar- fer and generated much less spatter than sues, as described by Cuiuri et al. (Ref. 20).
ent up to 25% CO2, then the process ex- uncoated electrodes. The researchers con-
hibits repelled transfer as the dominant cluded that spray transfer is impossible Behavior of Anode Spot
mode of metal transfer at all currents. when the path of welding current at the tip
Higher amounts of CO2 in the shielding of the electrode is confined to a small The envelopment of the droplet by the
gas leads to constriction of the arc, which high-current-density area, and postulated arc is essential to the metal transfer mode,
results in repelled transfer. Haidar and that by having negative polarity on the therefore it is useful to analyze it in detail.
Lowke reported an increase in anode spot electrode and adding thermionic emissive Through experimental observation, Rhee
current density from 7 × 103 to 3.3 × 104 agents to the surface of the electrode, they and Kannatey-Asibu (Refs. 6, 11) re-
A/cm2 as CO2 concentration increases could control the current density at the tip ported that the globular/spray transition
from 0 to 100% (Ref. 12). Mechev et al. of the electrode. occurs when the arc covers the droplet sur-
(Ref. 13) showed through calculations that The second approach uses pulsing of face and suggested that helium and CO2
the CO2 arcs are more constricted; the dy- the welding current or voltage. Control- shielding gases do not exhibit transition
because the arc does not climb over the
droplet. This evidence suggests that the
Table 1 — Electrode Diameters and Compositions Used in This Research transition from globular to spray occurs
when the arc covers the entire droplet. Arc
Wire Diameter (in.) AWS Classification Composition (wt-%) envelopment of the droplet is expected to
C Mn Si P S Cu
change the distribution of forces upon the
0.045 ER70S-6 0.07 1.40 0.80 droplet, thus directly influencing the
0.035 ER70S-6 to to to 0.025 0.035 0.50 transfer mode.
0.023 ER70S-6 0.15 1.85 1.15 The mechanism of envelopment of the
0.016 ER70S-G 0.13 0.51 0.08 0.010 0.010 0.62 droplet in an argon rich atmosphere is il-
lustrated schematically in Fig. 2. In this
126-s MAY 2008, VOL. 87
Fig. 6 — A — Voltage signal of a weld made with 0.035-in.-diameter electrode in a 90Ar-10CO2 shielding gas mixture; B — Fast Fourier Transform (FFT) of
A. The voltage signal is fairly periodic, leading to a distinguishing frequency peak in the FFT. C — Voltage signal of a weld made with 0.035-in.-diameter elec-
trode in 60Ar-40CO2 that shows repelled metal transfer. Droplet detachment is random and nonaxial; D — FFT of C.
case, the current density at the anode spot servations of arc constriction indicate that arc attachment point climbs up the droplet
is approximately constant, independent of current density depends mainly on shield- because current density remains the same.
current or droplet size (Ref. 12). At low ing gas composition. Figure 3 is a When the electrode diameter becomes
currents, represented on the left side, the schematic that illustrates how increasing small enough, the arc attachment point
anode spot covers only a small portion of amounts of CO2 in the shielding gas alters moves above the droplet and results in a
the droplet’s surface area, A. In the figure, the metal transfer mode. From left to transition from globular to spray.
the anode spot area is represented by the right, the CO2 concentration is increasing Because the current density at the elec-
dotted interface. Globular transfer is the in the shielding gas. When concentrations trode tip is a function of shielding gas, in-
dominant transfer mode, having droplet approach a critical level, a transition oc- creasing the amount of CO2 will increase
diameters larger than the electrode. As curs where metal transfer changes from the current density, such that the anode
the current increases from I to C1I, the axial spray to repelled transfer. Greater spot area will decrease for a given current
anode spot area increases to C1A as well, amounts of CO2 increase the current den- and lead to arc constriction on the droplet.
but does not completely envelop the sity at the electrode. For a constant cur- This work explores whether this effect
droplet. Transition occurs at higher cur- rent, the anode spot becomes smaller with could be counteracted by decreasing elec-
rents, when the current increases to C2I, increasing amounts of CO2. If the anode trode diameter such that the arc attach-
and the relative anode spot area, C2A, be- spot becomes small enough that it is un- ment is forced above the droplet, en-
comes large enough to cover the entire able to cover the droplet, the transfer veloping it, and causing a transition to
droplet. At this point, the arc attachment mode changes from spray to globular. De- spray transfer.
is above the droplet and climbing up the pending on the current and current den-
electrode, climbing even higher as current sity values, the plasma pressure concen- Procedure
increases to C3I. At these high currents, trated in a small area on the droplet can
more of the electrode is in the arc and result in a force large enough to levitate Experimental Matrix
melting occurs radially, creating a taper. the droplet, resulting in repelled transfer.
The morphology of the taper and relative The increase in anode spot density with Figure 5 shows the experimental ma-
size of the droplets will determine whether CO2 is not well known, and it is not neces- trix followed in this research. The circles
the spray transfer is projected or stream- sarily linear. shown on the graph represent particular
ing spray. In projected spray transfer, the For typical electrode diameters, their welding parameters used. The full circles
electrode has a short taper with droplets size also influences the metal transfer are the focus of this research, while the
slightly smaller than the diameter of the mode. Figure 4 shows the effects that di- hollow circles represent parameters where
electrode. Streaming spray has an elec- ameter has on the transition from globu- repelled transfer occurred and prevented
trode with a long taper and the droplets lar to spray transfer. At a given current further measurements. The dotted line
are much smaller than the electrode. and shielding gas composition, the arc at- shows the transition region where the
Shielding gas composition can change tachment point is located under the transfer was mixed between nonaxial (re-
the metal transfer mode with all other droplet for the largest electrode. As the di- pelled) and axial types of detachments. On
variables held constant. Experimental ob- ameter of the electrode is decreased, the the graph, the double cross-hatched shad-
WELDING JOURNAL 127-s
Fig. 7 — Schematic of the newly designed contact tip (A) and the fabricated part (B). The elec-
trical contact point is restricted to the point shown. An insulating alumina tube lines the hole in
the main body.
ing on the right shows regions where pub- where dd is the droplet diameter in inches,
lished data were found. On the left, the WFS is the wire feed speed in
single cross-hatched region shows unex- inches/minute, de is electrode diameter in
plored parameters for electrode diame- inches, and fd is the droplet detachment
ters and shielding gas compositions. From frequency in Hertz. This calculation as- Fig. 8 — A detailed view of the modified wire feed-
ing unit. The distance from the end drive rolls to
the conditions that are shown on the sumes that losses due to evaporation are the welding gun has been minimized to prevent
graph, testing revealed that repelled trans- small, and requires a measurement of the buckling and allow smooth feeding of small-
fer begins to dominate when concentra- droplet detachment frequency. diameter electrodes.
tions of CO2 are at or above 30% for elec-
trode diameters 0.023 in. (0.58 mm) and Frequency Measurement
larger. For the 0.016-in.- (0.41-mm-) di-
ameter electrode, repelled transfer be- Droplet detachment frequency was
In this work, two methods were used to
comes dominant at 20% CO2. also measured using high-speed laser
measure droplet detachment frequency:
For a given electrode diameter and shadowgraphs. In this case, images were
Fast Fourier Transform (FFT) of the volt-
shielding gas composition, the transition taken at 3000 frames per s., and detach-
age signal, and high-speed laser shadow-
current was determined using Lesnewich’s ments were counted for three separate
graph. FFT of the current signal was also
definition described previously. The tests 0.1-s intervals of the run. Because of its
performed, but was of inferior quality than
began at low wire feed speeds (WFS) that simplicity of use, most of the frequencies
the FFT from voltage.
corresponded to globular transfer mode reported in this work correspond to the
Fast Fourier Transform (FFT) is simple
and relatively low currents. The WFS was peak frequency of the FFT spectrum.
to implement and accurate for the stable
increased until a transition was observed When the FFT spectrum did not show a
transfer cases, in which a peak in the FFT
and spray transfer mode becomes domi- sharp peak, the frequency reported corre-
spectrum is clearly discernible. In our
nant. Voltages were adjusted in order to sponds to the droplet counting technique
case, the FFT spectra were generated by
keep the electrode extension constant. using high-speed video. The agreement
sampling the voltage signals at 5 kHz for 5
Contact tip-to-workpiece distance between the two frequency measurement
s. Figure 6 shows the voltage waveforms
(CTWD) remained constant for each elec- techniques was tested for stable globu-
and the FFT plots for both stable and re-
trode diameter tested. The three largest lar/spray transfer modes. In this case, both
pelled metal transfer. During stable metal
electrodes were tested with 1.0-in. (25.4- measurements are within 10% of each
transfer, the voltage waveform shows a
mm) CTWD and 0.5-in. (12.7-mm) arc other for all electrode diameters, consis-
relatively stable frequency and amplitude.
length. The 0.016-in.- (0.41-mm-) diame- tent with previous observations by other
The sharp peak in a narrow frequency
ter electrode was tested using 0.75-in. researchers (Refs. 21–23).
range in Fig. 6B indicates that detachment
(19.1-mm) CTWD and 0.38-in. (9.7-mm) frequency is relatively constant, and con- Materials
arc length for improved process stability. strained within a narrow range. In this
case, the peak in frequency distribution Flat position bead-on-plate welds were
Droplet Diameter Measurement begins at approximately 300 Hz, has a made on 0.375-in.- (9.5-mm-) thick ASTM
maximum value at 340 Hz, and ends at ap- A36 bars using four different electrode di-
Comparing droplet size to wire diameter proximately 375 Hz. This spread in fre- ameters ranging from 0.045 to 0.016 in.
is useful, since the traditional criterion of quencies corresponds to a spread in (1.14–0.41 mm). Table 1 gives the chemi-
globular transfer is stated in those terms. droplet diameter between 0.037 and 0.039 cal compositions of the different elec-
For the parameters tested, high-speed laser in. (an error of approximately 5%). trodes. The three largest electrodes were
shadowgraphs indicated that the detaching When the transfer mode approaches commercially available ER70S-6 welding
droplets have a shape close to spherical for repelled transfer, droplet detachment be- electrodes manufactured by Hobart.
nonrepelled transfer. Droplet diameters comes erratic and the FFT spectrum be- Smaller electrode diameters are not com-
can be calculated using a volume balance re- comes broad, without a clearly identifiable mercially available, so a special 0.016-in.-
sulting in the following expression: peak representative of the detachment diameter electrode was manufactured by
1 frequency. The distinctly different wave- California Fine Wire Co. Industrial-grade
⎛ WFS ⋅ d 2 ⎞ 3 form and FFT are shown in Fig. 6D. The argon and CO2 were used as shielding
d =⎜ e ⎟ broad spectra distributed over a large fre- gases. Mixtures of 100Ar, 90Ar-10CO2,
40 ⋅ f ⎟ quency range represent the inconsistent
⎝ d ⎠ 80Ar-20CO2, and 70Ar-30CO2 were used
timing of detachment events. for the four different electrode sizes.
128-s MAY 2008, VOL. 87
No commercially produced contact
tubes were available for the 0.016-in.-
diameter electrode, so two different types
were fabricated. The first design followed
the conventional tube-type contact tip
configuration. Meltback events increased
as electrode diameters became smaller be-
cause of the sensitivity of the system.
Small changes in current have larger ef-
fects on electrode extension with smaller
electrodes. Also, the point of electrical
contact is unknown due to the variability
of contact points within the tube. Waszink
and Van Den Heuvel have estimated this Fig. 9 — High-speed shadowgraphs of repelled transfer during welding using a 0.035-in. electrode in a
point to shift as much as 0.050 in. (1.25 60Ar-40CO2 shielding atmosphere. The contact tip shadow is clearly visible at the top of the screen, with
mm) during the welding operation (Ref. the electrode shown in the middle. The droplet first moves upward before detaching from the electrode.
24). As the electrical contact points vary,
so does the effective electrode extension.
A new design was needed as electrode di- An Omega gas proportioning rotame- Results and Discussion
ameters became smaller and process sta- ter was used for varying the composition
bility became more sensitive to small of the shielding gas. A calibrated flow- Welding Current Effects
changes in welding parameters. chart was supplied from the manufacturer
Figure 7 shows the new design for the for the binary Ar-CO2 gas mixtures. Con- Figures 10–13 show the relationship
contact tip used with the 0.016-in.-diame- stant flow rates of 40 ft3/h were used between average droplet diameter and
ter electrode. The function of the device throughout the entire study. The mixer welding current for the four electrode di-
relies on the stiffness of the electrode to was received from the factory calibrated ameters tested with different shielding gas
establish electrical contact with the tip. with an accuracy of ±2% in composition. mixtures. Several distinguishable trends
The point at which the electrode intersects In this work, laser shadowgraph tech- can be noted. Shielding gas composition
the U-wire is the exact place of electrical niques and a high-speed digital camera seems to have little influence on droplet
transfer and was used to measure the were used to image metal transfer. The size for currents above the transition.
CTWD. This ensures that the electrode system used in this research was similar to Shielding gas compositions containing
extension is kept constant, unlike tradi- that of Allemand’s (Ref. 26). The laser more than 30% CO2 exhibited substantial
tional tube-style contact tips. This feature source was a helium-neon laser manufac- amounts of repelled transfer and are not
reduces the amount of meltbacks when tured by Melles-Groit with a maximum included in these plots for the 0.045-,
compared to the conventional design and output of 30 mW at a wavelength of 632.8 0.035-, and 0.023-in.-diameter electrodes.
enables a fast change when they do occur. nm. The beam passed through a spatial fil- Shielding gas compositions only up to
This new design is well suited for thin elec- ter, collimator, aiming mirror, and then 20% CO2 are shown in Fig. 13 because re-
trodes (0.016 in.) because it avoids the the arc. On the other side of the arc, the pelled transfer occurred above this gas
tight manufacturing tolerances of the beam traveled through a bandpass inter- composition. It is unclear whether this ef-
small hole of a conventional contact tip. ference filter, allowing light in the range of fect is due to the extremely small electrode
The improved performance of this new 632.8 ± 0.5 nm to pass. The shadow of the diameter or to the small differences in
contact tip suggests that process stability contact tip, electrode, droplets, and base chemical composition of that electrode.
can be improved by precisely controlling metal were projected onto a piece of
the point of electrical contact between the frosted glass and filmed with high-speed Droplet Diameter
electrode and the contact tip. A patent is video. A Kodak Ektapro EM digital high-
pending for the design of the new contact speed camera was used to record the weld- Figure 14 shows the relationship be-
tip (Ref. 25). ing process. Figure 9 shows a compilation tween average droplet diameter and cur-
of screenshots taken from the same ex- rent for the different diameter electrodes
Equipment periment as Fig. 6C, D. The frames are 10 tested in a 90Ar-10CO2 atmosphere. All
ms apart, during repelled transfer using a electrodes exhibit a transition from large
The power source used in this research 0.035-in. electrode in a 60Ar-40CO2 droplet diameters (low detachment fre-
was a Miller Electric Maxtron 450 CC/CV shielding gas mixture. quency) to small droplet diameters (high
power supply operated in CV mode. No The second method for determining detachment frequency). After the transi-
pulsing or waveform programming was metal transfer mode used voltage and cur- tion, a lower shelf for droplet diameters
implemented in the experiments. The wire rent analysis. Both the current and voltage exists for all electrodes. For the 0.045-in.
feed machine was a constant-speed Miller transducers were manufactured by LEM, electrode, calculations indicate that the
Electric S-70 with high-speed motor op- with the signal conditioning units made in- average droplet diameter becomes
tion, capable of feed rates up to 1443 house. A National Instruments data acqui- smaller than the electrode diameter. This
in./min. Several modifications were made sition system interfaced to a computer, observation is consistent with several
to the unit to improve the performance where National Instruments Labview pro- other researchers, and has been the tradi-
during this research, shown in Fig. 8. The gram was used as the control software. The tional definition of spray transfer.
distance between the contact tip and the sensors were calibrated with a Fluke multi- The 0.035-, 0.023-, and 0.016-in.-diam-
drive rolls was minimized from 72 to 8 in. meter having both voltage and current eter electrodes still exhibit the same type
for better feeding of thin electrodes. An measuring capabilities. For all experi- of transition from large to small droplet
adapter was fabricated that repositioned ments, the voltage and current signals were diameters with increasing current. Past
the welding gun immediately adjacent to sampled at 5000 Hz for approximately 5 s, the transition, the transfer mode would
the wire drive assembly. and then analyzed using MATLAB. appear and sound like spray transfer to the
WELDING JOURNAL 129 -s
Fig. 10 — The average calculated detached droplet diameter as a func- Fig. 11 — The average calculated detached droplet diameter as a function
tion of current for the 0.045-in. electrode with different shielding gas of current for the 0.035-in. electrode with different shielding gas
Fig. 12 — The average calculated detached droplet diameter as a func- Fig. 13 — The average calculated detached droplet diameter as a function
tion of current for the 0.023-in. electrode with different shielding gas of current for the 0.016-in. electrode with different shielding gas
senses of a skilled welder; however, it only changed from 0.042 to 0.038 in. These in Fig. 16. The transition current was de-
would not fit to the traditional definition small variations of spray transfer droplet termined using Lesnewich’s definition de-
of spray transfer. Analysis shows that the diameter with wire diameter may be con- scribed above. The smallest (0.016-in.-
average droplet diameters do not become trolled by surface tension; however, initial diameter) electrode is not included in this
smaller than the electrode diameters. The calculations indicate that additional fac- graph because an upper limit in the spray
traditional spray definition that droplet tors not yet identified must be at play. region was not found, since it occurs at
diameters are smaller than electrode di- wire feed speeds beyond the capability of
ameters is not applicable to small-diame- Shielding Gas Effects the equipment.
ter electrodes. Figure 14 shows the results A transition from large-diameter
for a 90Ar-10CO2 shielding gas mixture, Free flight mode begins to become re- droplets to small-diameter droplets occurs
and the same trends are evident with the pelled with concentrations of CO2 be- for all-diameter electrodes in shielding gas
100Ar and 80Ar-20CO2 mixtures. tween 20 and 30%. This is consistent with concentrations of up to 30% CO2 as the
The change in minimum droplet diam- several other researchers’ observations current is increased.
eter is smaller than the change in their as- and can now be extended to electrodes The effect of shielding gas on transi-
sociated electrode diameters. Figure 15 with diameters as small as 0.016 in. The tion current is a relatively gradual increase
shows the average droplet size just after addition of CO2 does not have a propor- for the 0.045-in. wire, while it shows a lit-
the transition as a function of electrode di- tional influence on metal transfer. In this tle effect, even a slight decrease, for the
ameter for 90Ar-10CO2 gas mixtures. As research, there is little effect until CO2 0.023- and 0.035-in. wires. The behavior of
the electrode diameter changes from concentrations reach 30% and then re- the transition current for the largest elec-
0.045 to 0.035 in., the average droplet size pelled transfer begins to appear, as shown trode (0.045-in.) is consistent with that de-
130-s MAY 2008, VOL. 87
Fig. 14 — Characteristics between droplet diameter and current for four different electrode diameters. The droplet diameters never become smaller than elec-
trode diameters in the three smallest electrodes.
scribed in the schematic of Fig. 3, in which droplet. The direction of the plasma pres- diameter. The arc was unable to climb
a more constricted anode spot requires sure is upward and acts as an attaching over the droplet, therefore no transition
larger currents to envelop the droplet and force on the droplet. The droplet must occurred. While operating, the process
create the transition. The effect of CO2 on then grow to a larger size in order to de- sounded and appeared to be spray trans-
the thinner wires does not follow this pat- tach. In all cases, increasing the amounts fer mode; however, detailed inspection of
tern, suggesting that the behavior of the of CO2 above a critical level leads to the laser shadowgraphs did not show the
anode spot in thin electrodes involves fac- plasma pressure becoming large enough envelopment of the droplet by the arc,
tors that are not relevant for larger elec- to suspend the droplet and cause erratic characteristic of spray transfer. Pulsing
trodes. This effect is currently being detachment. Other researchers (Refs. 13, could potentially overcome this problem
investigated. 14) have noted similar findings. by detaching the droplets before they
At shielding gas concentrations of By using electrode diameters as thin as grow, and it is a current area of research.
80Ar-20CO2, the dominant transfer is sta- 0.016 in., the traditionally defined spray
ble axial, whether within the globular or transfer mode was not achieved in CO2 Conclusions
spray regime. For shielding gas mixtures concentrations greater than 30%. It was
containing 60Ar-40CO2, repelled transfer expected that smaller electrode diameters This investigation extended the range
becomes dominant at all currents. The would produce smaller droplet diameters of droplet detachment analysis in GMAW
mixture at 70Ar-30CO2 shows a mixed and cause the arc to climb above the to electrodes as thin as 0.016 in. in binary
mode. At this level, repelled transfer be- droplet forcing a transition from globular Ar-CO2 gas mixtures operated with
gins to become apparent not only through to spray, but this was not the case. Droplet DCEP. Voltage signal analysis and high-
high-speed video but also through in- diameters after the transition did not de- speed laser shadowgraphs were used to
creased spatter on the base metal. The crease proportionally with the decrease in determine transfer mode, droplet detach-
0.016-in.-diameter electrode showed electrode diameter, as shown schemati- ment frequency, and average droplet
more repelled transfer at 80Ar-20CO2 cally in Fig. 17. This is a key result of this diameters.
than the larger electrodes. research. Instead of droplet diameters be- A newly designed contact tip was used
The appearance of repelled transfer is coming smaller with decreasing electrode with the 0.016-in. electrode that improved
consistent with an increase in current den- diameters, the droplets remained rela- process stability. By precisely controlling
sity on the electrode and a subsequent in- tively equal in size. The anode spot area is the electrical contact point between the
crease in plasma pressure exerted on the not significantly changed by the electrode electrode and contact tip, the number of
WELDING JOURNAL 131-s
Fig. 15 — Droplet diameter in the spray region plotted against the electrode Fig. 16 — Transition currents for various electrode diameters in different
diameter. composition shielding gases. The transition current here is defined as the av-
erage between the upper and lower shelves of droplet diameters. Above 30%
CO2, repelled transfer began to dominate. The 0.016-in.-diameter electrode
is not included because the upper shelf was not found.
ameters when CO2 help of Bruce Albrecht from Miller Elec-
concentrations in the tric and Tom Siewert and Tim Quinn of
shielding gas are lower the National Institute of Standards and
than 30%. The aver- Technology (NIST) in Boulder, Colo.
age current used to
define the transition is References
not much affected by
the amount of CO2 for 1. Welding Handbook, 7th ed., Vol. 1, Fun-
small electrode diam- damentals of Welding. 1981. Ed. C. Weisman.
Miami, Fla.: American Welding Society.
eters (0.023 and 0.035
2. IIW, Classification of metal transfer.
in.) and increases 1977. Welding in the World 15(5/6): 113–116.
slightly for 0.045 in. 3. Muller, A., Greene, W. J., and Rothschild,
This supports previ- G. R. 1951. Characteristics of inert-gas-
ous findings as well as shielded metal arcs. Welding Journal 31(8):
extends the results to 717–727.
the unexplored range 4. Lesnewich, A. 1958. Control of melting
of 0.016-in.-diameter rate and metal transfer in gas-shielded metal-
electrodes. arc welding: Part I — Control of electrode melt-
The traditional clas- ing rate. Welding Journal 37(8): 343-s to 353-s.
Fig. 17 — The arc is unable to climb up and envelop the droplet because sification of spray and 5. Lesnewich, A. 1958. Control of melting
droplet sizes do not decrease as electrode diameter decreases. Figure is not globular transfer re- rate and metal transfer in gas-shielded metal-
to scale. arc welding: Part II — Control of metal trans-
lies on comparing the fer. Welding Journal 37(9): 418-s to 425-s.
diameter of the 6. Rhee, S., and Kannatey-Asibu, E. 1991.
droplets to that of the Analysis of arc pressure effect on metal trans-
meltback events decreased when com- electrode. However, the findings of this fer in gas-metal arc-welding. Journal of Applied
pared to conventional tube-type designs. research indicate that this definition is not Physics 70(9): 5068–5075.
Binary Ar-CO2 shielding gas composi- useful for 0.035-in. steel electrodes and 7. Wang, G., Huang, P. G., and Zhang, Y. M.
tions were tested with CO2 amounts up to smaller. All electrodes exhibited an in- 2003. Numerical analysis of metal transfer in
50%. At 30% CO2, the process began to gas metal arc welding. Metallurgical and Materi-
crease in droplet detachment frequency
als Transactions B 34B (June): 345–353.
exhibit repelled transfer for the 0.045-, with current as well as a transition from 8. Hu, J., and Tsai, H. L. 2006. Effects of cur-
0.035-, and 0.023-in. electrodes and at large- to small-diameter droplets. How- rent on droplet generation and arc plasma in
20% CO2 for the 0.016-in.-diameter elec- ever, the average droplet size is still larger gas metal arc welding. Journal of Applied
trode. During repelled transfer, the aver- than the electrode diameter (up to about Physics, 100 (Article No. 053304).
age droplet size grows larger than the elec- twice as large), regardless of current. 9. Rothschild, G. R. 1956. Carbon-dioxide-
trode and observations from high-speed shielded consumable-electrode arc welding.
shadowgraphs suggest that the constricted Acknowledgments Welding Journal 35(1): 19–29.
arc might cause an upward force that lev- 10. Smith, A. A. 1971. CO2 Welding of Steel,
itates the droplet. The authors would like to gratefully 3rd ed. Cambridge, UK: The Welding Institute.
11. Rhee, S., and Kannatey-Asibu, E. 1992.
When operating in the stable metal thank the American Welding Society
Observation of metal transfer during gas metal
transfer regime, a change in transfer mode Foundation for funding this research with arc welding. Welding Journal 71(11): 381–386.
with increasing current from large-diame- a fellowship award. This work is also sup- 12. Haidar, J., and Lowke, J. J. 1997. Effect
ter droplets at low detachment frequency ported by the National Science Founda- of CO2 shielding gas on metal droplet forma-
to small-diameter droplets at high detach- tion CAREER award DMI-0547649. The tion in arc welding. IEEE Transactions on
ment frequency is seen in all electrode di- authors also gratefully acknowledge the Plasma Science 25(5): 931–936.
132 -s MAY 2008, VOL. 87
13. Mechev, V. S., et al. 1982. The thermal Welding Journal 41(1): 14-s to 21-s. on droplet transfer of GMA welds. Proceedings
and physical-properties of gaseous carbon- 18. Needham, J. C., and Carter, A. W. 1967. International Conference on Trends in Welding
dioxide and their effects on the welding arc. Au- Arc and transfer characteristics of the Research. Gatlinburg, Tenn.: ASM
tomatic Welding USSR 35(4): 24–29. steel/CO2 welding process. British Welding Jour- International.
14. Nemchinsky, V. A. 1996. The effect of nal (10): 533–549. 23. Liu, S., and Siewert, T. A. 1989. Metal
the type of plasma gas on current constriction 19. Matsuda, F., Ushio, M., Nishikawa, H., transfer in gas metal arc welding droplet rate.
at the molten tip of an arc electrode. J. Phys. D: and Yokoo, T. 1985. Pulsed GMAW — Spatter- Welding Journal 68(2): 52-s to 58-s.
Appl. Phys. 29: 1202–1208. ing in pulsed CO2 welding. Transactions of JWRI 24. Waszink, J. H., and Van Den Heuvel, G.
15. Pires, I., Quintino, L., and Miranda, R. 14(1): 13–19. J. P. M. 1982. Heat-generation and heat-flow in
M. 2007. Analysis of the influence of shielding 20. Cuiuri, D., Norrish, J., and Cook, C. the filler metal in GMA welding. Welding Jour-
gas mixtures on the gas metal arc welding metal 2002. New approaches to controlling unstable nal 61(8): 269-s to 282-s.
transfer modes and fume formation rate. Mate- gas metal arc welding. Australasian Welding 25. Mendez, P. F., and Soderstrom, E. 2005.
rials & Design 28(5): 1623–1631. Journal 47(3): 39–47. Gas metal arc welding methods and apparatus,
16. Lesnewich, A. 1955. Electrode activa- 21. Kim, Y. S., and Eagar, T. W. 1993. Analy- U.S. Patent Office application # 2006/0237411
tion for inert-gas-shielded metal-arc welding. sis of metal transfer in gas metal arc-welding. A1, Editor.
Welding Journal 35(12): 1167–1178. Welding Journal 72(6): 269-s to 275-s. 26. Allemand, C. D., et al. 1985. A method
17. Cushman, E. 1961. Electrode for spat- 22. Kohn, G., and Siewert, T. A. 1986. The of filming metal transfer in welding arcs. Weld-
ter-free welding of steel in carbon dioxide. effect of power supply response characteristics ing Journal 64(1): 45–47.
CAN WE TALK?
The Welding Journal staff encourages an exchange of ideas with you, our readers. If you’d like to ask a question, share an idea or
voice an opinion, you can call, write, e-mail or fax. Staff e-mail addresses are listed below, along with a guide to help you interact with
the right person.
Publisher/Editor New Products Advertising Production Manager
Andrew Cullison News of the Industry Frank Wilson
firstname.lastname@example.org, Extension 249 email@example.com, Extension 465
Article Submissions Production Manager Advertising Production
Senior Editor firstname.lastname@example.org, Extension 265 Peer Review Coordinator
Mary Ruth Johnsen Design and Production Erin Adams
email@example.com, Extension 238 firstname.lastname@example.org, Extension 275
Feature Articles Advertising Sales Director Peer Review of Research Papers
Associate Editor email@example.com, Extension 243
Howard Woodward Advertising Sales Welding Journal Dept.
firstname.lastname@example.org, Extension 244 550 N.W. LeJeune Rd.
Society News Advertising Sales & Miami, FL 33126
Personnel Promotion Coordinator (800) 443-9353
Lea Garrigan Badwy FAX (305) 443-7404
Assistant Editor email@example.com, Extension 220
Kristin Campbell Production and Promotion
firstname.lastname@example.org, Extension 257
J. E. Ramirez, the author of the paper titled “Characterization of High-Strength Steel Weld
Metals: Chemical Composition, Microstructure, and Nonmetallic Inclusions” published in the
March 2008 issue of the Welding Journal (pages 65-s to 75-s), would like to report the omission of the
following acknowledgment statement.
“The author would like to acknowledge the contribution of Dr. A. J. Ramirez, Mr. J. W. Sowards, and
Dr. J. C. Lippold from The Ohio State University Welding & Joining Metallurgy Group, and Sue Fiore
from EWI in the development of a database for microstructures of high-strength steel weld metals
through an EWI-Cooperative Research Program (CRP). The author also would like to acknowledge
all the member companies of EWI for funding the development of the database through the cooperative
research program. The pictures associated with microstructures and nonmetallic inclusions in this
publication are included in the developed database.”
WELDING JOURNAL 133 -s