Metal Transfer Modes and Weld Fume Emission in MIG/MAG Welds

TWI Ltd., Granta Park, Abington, Cambridge, UK.

Eighth European Conference on Joining Technology, Eurojoin 8, Pula, Croatia, 24 - 26 May 2012.

Abstract

Health and safety in welding is always an important topic and with an increasing litigious and informed society, interest will continue to accelerate in the future. Welding is regarded as being a hazardous activity, requiring measures to be in place to minimise the risk. Public awareness of possible ill health arising from inhalation of fume from arc welding processes has increased significantly recently. However, although a vast amount of data is available, there is no clear understanding of the mechanisms involved in fume generation. Measurements have shown that fume emission rates vary significantly with welding parameters, but a detailed examination of the mechanisms involved in fume formation and the relationship between metal transfer mode and fume formation rate has not been undertaken.

In this work, metal transfer modes in MIG/MAG welding have been studies using high speed video (5000 fps) and high sampling rate parameter monitoring. Welding fume emission rates have been measured in accordance with ISO 15011-1:2009. Fume emission rates have been correlated with the MI/MAG metal transfer modes over a range of welding conditions.

Tests have been carried out using a mild steel solid, copper coated, welding wire (G3Si1) of 1.2mmm diameter, with argon shielding gas mixtures containing between 5 and 18% carbon dioxide. The effects of pulsed welding current have also been investigated.

These trials have shown that reduced levels of particulate welding fume can be achieved with the correct combination of shielding gas and welding parameters. Whilst fume emission rate is, in general, proportional to current, it has been shown that the correct voltage is also important, to ensure stable metal transfer and hence low fume emission. In particular, the size of the droplet and transit time, and the position of the arc on the tip of the wire are important for fume emission. In pulsed welding it has been shown that fume emission rate can be lower provided the correct parameters are used to ensure droplet transfer. However, a study of metal transfer modes explains why a low voltage or high current may result in similar fume emission rates to conventional transfer.

1. Introduction

Health and safety in welding is always an important topic and with an increasing litigious and informed society, interest will continue to accelerate in the future. Welding is regarded as being a hazardous activity, requiring measures to be in place to minimise the risk. Welding fume is an important topic that has been studied widely over many years and public awareness of the hazards of fume from arc welding processes has increased significantly recently.

Both experimental studies and mathematical modelling has been carried out to measure and gain a better understanding of fume emission. Many studies have also been carried out on the health effects of welding fume and standardized techniques have been established for laboratory sampling and breathing zone measurements. However, although a vast amount of data is available, there is no clear understanding of the mechanisms involved in fume generation. Measurements have shown that fume emission rates vary significantly with welding parameters, but a detailed examination of the mechanisms involved and the relationship between metal transfer mode and fume has not been undertaken.

Welding fume consists of metal oxide particles that remain suspended in the air and thus inhaled by welders.^{[Voitkevich, 1995]} The chemical composition and particle size of the fume particulates are important parameters in determining the toxicity of the fume. The chemical composition and fume formation rate (FFR) depend on several factors, the welding process, parameters, shielding gas, the filler and base metal composition. ^{[Castles and French, 1995]}

Previous work on mechanisms of fume formation is quite limited. Early work by ^{[Heile and Hill, 1975]} suggested vapourisation, condensation and oxidation of the welding wire and droplets in the arc zone to be the main source of fume. According to Heile and Hill, the factors that cause higher FFR are increased droplet temperature, longer time in the arc zone, lower metal transfer stability and higher arc energy. ^{[Gray et al, 1982]} suggest that fume can be sub divided into two components. For MIG/MAG welding, fume evolves from evaporation from the molten wire, the droplet and the weld pool, within the welding zone protected by the shielding gas. Fume also arises from spatter ejected from the arc zone and oxidised. However, recent work by ^{[Jenkins and Eager, 2005]} disputes that evaporation from spatter contributes significantly to the fume.

^{[Norrish and Richardson, 1988]} describe the three main metal transfer modes, dip (short arc), globular and spray transfer. During dip transfer a small amount of fume occurs from vaporisation of the wire, but explosive metal transfer creates spatter which is ejected from the arc zone, is oxidised and evaporates to form fume. In globular transfer there is more fume formed from the molten droplet and some fume occurs from spatter as the droplet necks and breaks free from the wire tip. ^{[Gray et al, 1982]} estimate that a third of fume originates from spatter and about 12% evolves from the work piece. In spray transfer mode fume is mainly caused by evaporation from the droplets. It is suggested that the higher plasma velocity in spray transfer causes forced convection around the droplets which increases the FFR.

Welding conditions affect the fume formation rate and welding current is considered to be the most important parameter. Generally, increasing current increases the FFR, but the relationship is not entirely clear.^{[Castles and French, 1995]} Some authors have noted a decrease in FFR at the transition point between globular and spray transfer ^{[Yates et al, 1992]} but others ^{[Gray, 1982]} found little correlation between current and FFR. It is believed that this result was affected by the wide voltage range used in the tests. The effect of voltage on FFR is not so well defined and many authors claim to use an appropriate voltage for the current. Presumably, this is to give a stable condition. Studies that have reported changing the voltage have found that FFR increases away from the optimum voltage.^{[Carter, 1998 and Willingham, 1986]} The effect of current on FFR has also been modelled ^{[Redding, 2002]} and the reduction in FFR at the spray transition current observed.

Recent studies on welding fume eg ^{[Sterjovski et al]} investigated the effect of parameters on particle size. This work has shown that particle size increases with increasing voltage. ^{[Yamazaki et al, 2010]} have investigated pulsed welding techniques to control droplet size and FFR.

Whilst it has generally been observed that FFR increases with welding current, some authors have noted a reduction in fume at the spray transition current. However, it is believed that this observed reduction in FFR is affected by voltage which will have a significant effect on the MAG process stability. It is concluded that there is still some disagreement about fume mechanisms and further studies are required to gain a better understanding.

Some papers have claimed that pulsed welding techniques reduce FFR.^{[Yamazaki et al, 2010]} Composition of the shielding gas is reported to affect FFR and as a general rule, emission rate tends to increase with higher concentrations of active components (ie oxygen and carbon dioxide).^{[Wiktorowicz R, 2002]} However, ^{[Carter G J (1998)]}, comments that shielding gas composition does not have a major influence on particulate fume emission rate.

2. Experimental programme

Welding trials were carried using an inverter MIG/MAG power source operating in the conventional transfer mode (dip/globular/spray) and pulse transfer mode. Welding conditions were set manually on the power source. A standard manual swan neck welding torch was connected to the power source and clamped by the handle above a moving traverse. The torch nozzle was angled slightly at 15° to the vertical and welding progressed in the 'push' direction. A fume hood as specified in EN 15001-1 was suspended above the welding torch to collect the welding fume. This equipment is shown in Figure 1. Welding was carried out on steel plates, with a mild steel solid, copper coated, welding wire (G3Si1) of 1.2mmm diameter. The plates were clamped in a fixture on the traverse and the weld progressed by moving the plate under a stationary welding torch. Initial trials established that a travel speed of 500mm/min gave an acceptable weld for sufficient duration for the tests.

A Motion Pro X4 high speed video camera manufactured by Redlake, was used to capture close up recording of the metal transfer mode. These recording were made with a high power (8kW) back light, used to balance out the light from the arc. Using this technique, rather than band pass filtering with a laser illumination source, enables some of the arc light to be see, which is helpful to establish where the arc sits on the droplet. Videos were captures at 5000 frames per second, at which speed the detachment of individual drops of molten metal from the wire tip can be observed. A recording comprises a series of individual Tiff images which are then compiled into a video format. The video is annotated with frame numbers so can then be played back and individual tiff images selected for further analysis.

High speed recording of the electrical parameters was also carried out using a Triton Electronics AMV400 recorder. Welding current and arc voltage waveforms were collected at the maximum sample rate of 100 ks/sec. At this sample rate details of changes in the arc can be observed, such as fluctuation in voltage and current as metal transfer occurs. Welding parameters are stored within the instrument and then transferred to a PC for analysis using dedicated software.

Welding trails were carried out using mild steel plates and a solid, copper coated, welding wire (G3Si1) of 1.2mmm diameter. The following shielding gases were evaluated;

• Argon 18% carbon dioxide 2% oxygen.
• Argon 12% carbon dioxide 2% oxygen.
• Argon 5% carbon dioxide 2% oxygen.

In this paper, the shielding gases are referenced to the percentage of carbon dioxide in argon, as oxygen is constant at 2%.

A series of trials were carried out varying the wire feed speed between 7 to 10m/min and over a range of voltage of 26 to 32V with FFR and electrical parameter recording.

Trials were also carried out using pulsed welding in a shielding gas of argon 12%CO₂. In pulsed welding, parameters are chosen to control the wire burn-off to, in theory, produce one droplet per pulse of current. The pulse current and duration is adjusted to melt one droplet from the wire. To maintain the burn-off relationship, the frequency (background time) is proportional to wire feed speed and the background current is adjusted in proportion.

Welding trials were carried out between 7 and 10m/min. The pulse parameters were pre-programmed in the power source. The pulse current and duration remained essentially constant at 436-448A and 2ms respectively, whilst the frequency increased from 174 to 242Hz and the background current increased from 78 to155A as the wire feed speed changed from 7 to 10m/min. The voltage was determined by the programme and for pulsed welding the setting is the voltage during the pulse, because this is used to control arc length. The actual (measured) mean voltage is significantly less and both values were measured.

3. Results

With a shielding gas of argon 12%CO₂, welding conditions varied from an irregular dip transfer arc at 7m/min 26V to a spray transfer arc at 10m/min 32V. The lowest FFR of 3.8mg/s was obtained for a droplet transfer condition with a drop size approximately one wire diameter. This occurred at a wire feed speed of 8m/min at 28V (Figure 2). Due to the short arc length the droplet is quickly transferred into the weld pool, with a slight bridging action.

At a lower wire feed speed of 7m/min and at 26V, an irregular dip transfer occurred, resulting in a FFR of 5.0mg/s (Figure 3).

At a wire feed speed of 9m/min, FFR were higher than for 8m/min. The lowest FFR of 4.2mg/s occurred at 30V (Figure 4), which corresponded to a drop transfer condition. However, due to the long arc length, the time for a droplet to transfer to the weld pool is much longer than for the transfer shown in Figure 3. At lower and higher voltages (28 and 32V) corresponding to irregular dip transfer and spray transfer with a long arc length, FFRs were higher. The metal transfer at 32V was found to be unstable. The long arc length resulted in variable droplet transfer and oscillations in the welding current.

A further increase in wire feed speed to 10m/min at 30 and 32V (Figure 5) led to a reduction in FFR (4.3 and 4.2mg/s). This is believed to be due to the drop transfer mode and stability of the arc. As the wire feed speed is higher the arc length is shorter at the same voltage. The high speed videos of Figures 4 and 5 which gave the lowest FFRs are remarkably similar, despite the different welding parameters (9m/min 30V and 10m/min 32V).

With a shielding gas of argon 18%CO₂ at a wire feed speed of 8 and 9m/min a minimum FFR occurred for a metal transfer mode in which a droplet was formed on the end of the wire, of about one wire diameter. With this welding condition (8m/min 32V), which produced a FFR of 4.2 mg/s, the droplet was stable and detached in free flight transfer. For the same wire feed speed of 8m/min at lower voltages, a larger droplet formed. At 28V, the arc length was short and the droplet was observed to touch the weld pool resulting in spikes in the voltage waveform An increase in voltage to 30V produced an open arc condition, but a large droplet formed on the wire tip, producing a FFR of 5.7mg/s. At a wire feed speed of 9m/min FFR increased with a maximum of 6.3mg/s at 28V. At this condition, a large droplet formed on the wire, regularly dipping into the weld pool. At higher voltages, FFRs were slightly lower. Overall, compared to the results for a shielding gas of argon 12%CO₂, the FFRs for argon 18%CO₂ were higher as shown in Figure 6.

Trials were also carried out for a shielding gas of argon 5%CO₂ at 9m/min. At 28V the short arc length and dip transfer condition, resulted in a FFR of 5.7mg/s, but at higher voltages the FFR was significantly reduced (2.7mg/s at 30V and 3.9mg/s at 32V).

These FFRs are some of the lowest measured. Analysis of the high speed videos shows a streaming metal transfer, Figure 7. Due to the pinch effect (Lorenz force), the metal is transferred in a narrow thread like stream of molten metal. Although the arc length is long, the surface area of molten metal emitting vapour is small, which may account for the low FFR. These results are compared to the other shielding gases in Figure 8.

The influence of pulsed welding parameters was also investigated over a range of wire feed speed from 7 to 10m/min, with a shielding gas of argon 12%CO₂. The voltage was pre-set by the programme in the welding power source. The metal transfer was observed to be a droplet with a diameter about that of the wire for wire fed speeds of 7, 8 and 9m/min (Figure 9).

This resulted in very low FFRs, less than half of the values obtained with this shielding gas using conventional transfer. The mean values of FFR were 2.3, 1.4 and 2.1mg/s for wire feed speeds of 7, 8 and 9m/min. Only at 10m/min did the FFR increase significantly to 4.1mg/s, approximately the same value of FFR (4.2mg/s) as obtained with conventional transfer at these conditions. A comparison of the high speed videos shows a similar welding condition. At 10m/min the mean welding current is too high for pulsed transfer so the welding condition reverts to spray, with a more continuous stream of droplet detachment.

A comparison of the FFR for pulsed transfer and conventional transfer in a shielding gas of argon 12%CO₂ is shown in Figure 10.

4. Discussion

There have been many studies on fume formation in MIG/MAG welding, with conflicting results and conclusions. Some authors have observed a reduction in fume emission at the transition between globular and spray transfer and others have not. In previous trials at TWI (Melton, 2011) this reduction in fume emission at the transition point was observed and attributed to a metal transfer mode in which a uniform droplet was seen to transfer smoothly from the wire electrode, being totally enveloped in the arc and without generating significant spatter. The significance of arc voltage was also established. Too low or high a voltage and consequently arc length, results in higher levels of fume than for the optimum voltage. These findings have been confirmed in this study and it has been observed that low fume emission will result from a combination of parameters that produces the droplet transfer described above.

Higher voltages resulting in a longer arc length, under certain conditions, produced larger droplets with an increase in droplet transfer time, resulting in a higher FFR. Additionally, if the arc rooted locally on the droplet, an agitation of the droplet was observed which is believed to be due to higher surface temperatures, leading to increased evaporation and hence a higher FFR.

There is also some debate on the influence of spatter on fume emission. Some authors believe that up to one third of fume originates from spatter, however other believe that spatter has little influence on fume formation. In this study, FFR was found to increase appreciably for low voltages (short arc lengths) generating spatter. A series of highly disruptive, momentary shorting events could be seen to occur, with spatter ejected from the molten wire tip and from the weld pool. As the voltage increased, FFR in general decreased initially. An increase in 2V from a dip transfer condition resulted in a decrease in FFR by up to 40% suggesting that spatter and particles ejected from the weld pool are a significant source of welding fume.

The composition of the shielding gas is reported to affect FFR and as a general rule, it is believed that emission rate tends to increase with higher concentrations of active components (ie oxygen and carbon dioxide). However, some authors comment that shielding gas composition does not have a major influence on particulate fume emission rate. In this study, it was found that a reduction in carbon dioxide in an argon shielding gas mixture resulted in less fume emission. However, a reduction in carbon dioxide from 18 to 12% did not significantly affect the metal transfer modes, so FFR was reduced only by a relatively small amount. A further reduction in carbon dioxide to 5% resulted in a lower FFR at higher voltages, but again the reduction was small. At lower voltages in dip transfer mode, no reductions in FFR were observed. An increase in voltage to produce globular transfer gave the lowest FFR and at high voltages, although the transfer mode was spray, the wire was observed to melt in streaming metal transfer. This resulted in a narrow stream of molten metal, with a low surface area which is believed to be the reason for the low fume emission rate. A carbon dioxide shielding gas was not investigated, because this composition is rarely used in Europe, however, it is likely that this gas would produce significantly higher levels of welding fume.

This study showed that low FFRs can be obtained for a range of different welding conditions. Both wire feed speed (welding current) and arc voltage affect FFR. An optimum combination of parameters, producing a regular drop transfer, in an open arc condition generally produces the lowest fume. Conflicting results from other studies may be due to a lack of parameter optimisation and an understanding of transfer modes.

Droplet transfer can also be obtained using pulsed welding conditions and this study has shown that the correct combination of pulse parameters can produce very low levels of welding fume. Fume formation rates of less than half of those obtained by conventional transfer, were achieved with pulse transfer. But FFR was found to increase significantly if the voltage was too low and at high wire feed speeds; pulse transfer reverted to spray with a consistent FFR.

These results show that high speed video coupled with high sampling rate data acquisition of electrical parameters (voltage and current) are powerful tools for gaining a better understanding of the formation of welding fume. The recordings of current and voltage clearly show arc instability and short circuiting events, which result in higher levels of fume emission. However, although the metal transfer mode could be clearly seen in the video recordings, it was harder to discern differences in the voltage and current waveforms, in the globular to spray transition range. However, it is believed that with further analysis of the waveforms, using algorithms to determine transfer mode, it may be possible to identify these variations in transfer mode. Also, the analysis carried out so far, has identified the effect of different transfer modes on fume emission. Further image analysis could be used to estimate droplet transfer times, current density and hence surface temperatures.

5. Conclusions

From a study of fume emissions from a solid mild steel welding wire in a range of argon shielding gas mixtures, containing 5-18% carbon dioxide and 2% oxygen, the following conclusions have been drawn:

1. It is generally accepted that FFR increases with welding current but, it has been confirmed that in the globular to spray transition range, a minimum in FFR occurs for an optimum voltage at a given wire feed speed. This minimum in FFR occurs for a regular droplet transfer, with no spatter and with the arc enveloping the droplet resulting in low current density and relatively low droplet temperatures.
2. Higher FFRs were observed for an increase or decrease in arc length, with larger droplets and an increase in droplet transfer time or momentary contacts with the weld pool creating spatter, respectively.
3. A reduction in carbon dioxide content in argon based shielding gases results in a lower FFR, provided the optimum voltage is set. However, this reduction is quite small and will not be observed if the voltage is not optimised. For argon shielding gas mixtures containing only 5% carbon dioxide, a low FFR is produced at high voltages from a thread like, steaming metal transfer condition.
4. Pulsed welding conditions were found to produce the lowest FFR, by producing a similar droplet transfer condition as that obtained in the globular to spray transfer region.
5. Measurement and observation of metal transfer modes provides an insight into the causes of welding fume emission. Low fume conditions are produced by a combination of parameters that can be understood by carrying out these measurements. These measurements can be used to explain discrepancies in results obtained in other studies.

6. References

Carter G J, 1998: 'Shielding gas formulation - its effect on arc welding fume emissions'. TWI Bulletin.

Castles J and French I E, 1995: 'Influence of consumable type, power source characteristics and welding parameters on arc welding fume'. Australian Welding Research, CRC No 3.

EN ISO 15011-1:2009: 'Health and safety in welding and allied processes - laboratory method for sampling fume and gases. Part 1 - determination of fume emission rate during arc welding and collection of fume for analysis'.

Gray C N, Hewitt P J and Dare P R M, 1982: 'New approach would help to control fumes at source'. Welding and Metal Fabrication, Vol. 50, No 8.

Heile R F and Hill D C, 1975: 'Particulate fume generation in arc welding processes'. Welding Journal Research Supplement, Vol 54, No 7.

Jenkins N T and Egar T W, 2005: 'Fume formation from spatter oxidation during arc welding'. Science and Technology of Welding and Joining, Vol 10, No 5.

Norrish J and Richardson I, 1988: 'Metal transfer mechanisms'. Welding and Metal Fabrication, Vol. 56.

Redding C J, 2002: 'Fume model for gas metal arc welding'. Welding Journal, Vol 81, No 6.

Sterjovski Z, Norrish J and Monaghan B J, 2010: 'The effect of voltage and metal transfer mode on particulate fume size during GM (MIG/MAG welding) of plain carbon steel'. Welding in the World, Vol. 54.

Voitkevich V, 1995: 'Welding fumes: formation, properties and biological effects'. Abington Publishing, Cambridge.

Wiktorowicz R, 2002: 'Air Products focus on fume'.

Willingham D C and Hilton D E, 1986: 'Some aspects of fume emissions from MIG welding of stainless steel'. Welding and Metal Fabrication, Vol 54, No 5.

Yamazaki K, Suzuki R, Shimizu H and Koshiishi F, 2010: 'Spatter and fume reduction in CO₂ gas shielded arc welding by regulated drop transfer'. IIW XII-1993-10.

Yates D, Plumridge P N, Hilton D E and Medforth J, 1992: 'Fume generation in GMA and GTA welding'. PACRIMWELDCON 92, WTIA.