Surface tension effects in welding

TWI Bulletin, June 1985

by Mark Rodwell

Mark Rodwell, BSc, is a Research Engineer in the Arc Welding Department.

Molten metal surface tension has been found to have a critical influence on several aspects of welding including TIG penetration profile, MIG metal transfer, weld bead formation and defect occurrence. This article outlines the origin and significance of these effects.

Most people are familiar with the concept of surface tension, that is, the tendency for a liquid to behave as if an elastic membrane were stretched across its surface. It is conventionally demonstrated by carefully laying a needle on the surface of a beaker of water; the needle remains on the surface without sinking, yet it cannot be described as floating since the surface layer has not been broken. Another simple experiment to demonstrate that there is indeed a force acting along the surface involves placing two matches close together in the centre of a beaker of water. If a drop of liquid which has a low surface tension ( e.g. ether) is placed between the matches, they move rapidly apart under the influence of the greater surface tension of the surrounding water.

Surface tension is thought to play a major role in fusion welding and in certain areas complex mathematical treatments have been applied to explain the observed phenomenon. ^[1-3] The aim of this review is to explain qualitatively the origin of surface tension, examine its significance with regard to four major aspects of welding, namely TIG weld penetration, MIG metal transfer, defect formation and high power density welding, and describe practical methods for its measurement.

Origin of surface tension in liquids

Within the bulk of a liquid, the molecules (or atoms) are in a continuous state of random motion. However, because each molecule is surrounded by other molecules on all sides, there is effectively no net force on it. In contrast, a molecule at the surface experiences a higher molecular density on one side than the other ( Fig.1) and this results in an inward force which is predominantly electric in origin. Work must be done to bring molecules from the body of the liquid to the surface against these inwardly acting forces, and the energy required to produce one unit of new area in this way is called the free surface energy or simply surface energy.

Fig.1. Representation of molecular forces, showing that the surface molecules have a net attraction towards the liquid bulk

Because the surface has a tendency to contract, it may be considered to be in a state of tension, and the force acting along a unit length of the surface is termed the surface tension( γ). The geometrical shape with the minimum surface area for a given volume is a sphere, hence in the absence of more dominant forces, liquid surfaces will adopt a spherical geometry in an attempt to reduce their surface area and therefore their surface energy. This effect is best seen in soap bubbles which are effectively spherical since their mass is not sufficient to allow gravitational forces to distort their shape.

The same principle is utilised in the manufacture of lead shot, for example, where molten lead is simply dropped from the top of a tower. In the liquid phase the lead forms into discrete, near spherical pellets which then solidify and arrive at the bottom as lead shot of varying diameter.

A second important concept to establish is that the liquid surface itself can move in the presence of surface tension forces which are not in equilibrium or balance. For example, a strongly alcoholic drink exposed to air in an open glass produces the phenomenon often called 'curtaining' or 'weeping', whereby liquid is continually dragged up the sides of the glass as a film and eventually forms droplets which roll back under gravity. The explanation lies in the fact that alcohol has a lower surface tension than water, but is more volatile. As the alcohol evaporates from the surface in the centre region of the glass, it is easily replenished from the bulk of the liquid. However, in the thin film on the glass wall the evaporating alcohol is not so easily replaced and hence the surface tension increases to a value higher than the surface tension above the main bulk of the liquid. This surface tension gradient exerts a shear stress on the liquid surface causing it to rise up the glass sides ( Fig.2).

Fig.2. Illustration of weeping effect when a mixture of alcohol (low surface tension) and water (high surface tension) is present in a vessel

Variable fusion geometry in TIG welding

The mechanised TIG welding process is occasionally susceptible to major variations in the weld profile when the parent material is derived from more than one cast or heat, despite the use of a constant welding procedure. The problem, commonly referred to as 'cast-to-cast variations, is generally manifested as one of intermittent or lack of penetration when autogenously welding the root of pipe or linear butt joints ( Fig.3). The vast majority of reported cases have occurred when welding stainless steel, although this may simply reflect the predominance of this material for high quality welding using the mechanised TIG process, particularly pipework in nuclear and chemical plant fabrication.

Fig.3. Variations in tube butt weld profile. Both welds were made with identical welding parameters

The problem has been the subject of intensive international research for many years which has attributed the variability to small differences in the level of impurity elements in the steel. ^[4] A full explanation has not been forthcoming and researchers remain divided over the relative importance of arc and weld pool effects.

Nevertheless, a number of interesting theories have been proposed to account for the behaviour, based on the surface tension of the weld pool.

A simple model suggested that variations in weld pool surface tension could result in inconsistent weld penetration by controlling the degree of distortion of the pool beneath the arc. ^[5] Electromagnetic, gravitational and arc forces all act to depress the centre region of the weld pool, but the surface tension inhibits depression formation since this involves creating new surface area which raises the total surface energy, ( Fig.4). The depression effectively reduces the plate thickness, allowing heat from the arc to be directed deeper into the weld pool and hence serves to increase the depth of fusion achieved. An analytical model was used to examine the importance of surface tension and concluded that it was a major factor in determining the weld profile, although no explanation was offered for its dependence on minor element concentrations.

Fig.4. Arc and Lorentz forces create a depression on the weld pool surface which is opposed by the weld pool surface tension

A more recent mathematical and experimental examination of the surface depression phenomenon revealed that a depression 4-5mm deep could result from currents in excess of 250A, and proposed that the major cause of deep penetration under these conditions was a vortex caused by circumferential rotation of the weld pool. ^[6] However, it was not thought that the surface depression played an important role when operating at lower currents.

A more sophisticated explanation for cast-to-cast variations in weld shape has been suggested, based not on differences in the absolute value of weld pool surface tension, but rather on variations in the temperature dependence of surface tension. ^[7] The mechanism is based on the fact that some elements in group VI of the periodic table are 'surface active' in iron based alloys and can have a dramatic effect on liquid surface tension, even when present in very low concentrations. When in solution, these elements (S,O,Se, Te) tend to migrate preferentially to the weld pool boundary where they are responsible for a significant lowering of the surface tension.

For a steel having only a low concentration of surface active impurities, the surface tension of the melt is expected to behave in a fashion similar to that exhibited by pure metals, i.e. the surface tension decreases with increasing temperature ( Fig.5). Casts of material containing a substantial content of surface active impurities (primarily S and O) will have a greatly reduced surface tension near the melting point as the impurities migrate to the surface layers. However, as the temperature is raised, the level of molecular agitation within the liquid is increased, thereby promoting a greater involvement of the surface active impurities with the liquid bulk, i.e. their concentration at the surface is reduced.

The net result is that the surface tension rises with increasing temperature until the impurities cannot maintain an effective presence at the surface, at which point the surface tension - temperature behaviour reverts to that of the pure metal ( Fig.5).

Fig.5. Temperature dependence of surface tension for iron-based alloys showing effect of surface active impurities [7]

The temperature distribution over the weld pool surface is such that the temperature is hottest in the centre region beneath the arc, and coolest at the fusion boundary. The model proposes that, depending on the level of surface active impurities present, two opposite types of convective flow patterns may be set up by movement of the surface layers from regions of low surface tension to regions of high surface tension, ( Fig.6). Materials possessing the flow pattern which carries hot material towards the weld toe (low S,O) would be expected to increase the weld width, whereas those exhibiting the radially inwards flow direction (high S,O) should promote a deeper penetration profile.

Fig.6. Surface tension driven mechanism to account for variable TIG weld fusion profile: [7] 

a) Impure alloy high sulphur, oxygen content;

b) Pure alloy low sulphur, oxygen content

This suggested model is difficult to prove conclusively, but there are a number of supporting observations:

1. Casts of material which are low in sulphur do tend to display poorer penetration characteristics. ^[8]
2. Additions of Ce and Al tend to result in wider, shallower welds. ^[9,l0] This is expected because Al reacts with oxygen in steels and Ce reacts with both oxygen and sulphur to form stable compounds which reduce the concentration of the surface active elements in solution.
3. Variations in fusion profile were observed by doping laser and electron beam welds with selenium, ^[11] suggesting that convective flow in the weld pool occurs even in the absence of arc effects.

Fig.7. Balance of forces during droplet growth on electrode tip: 

i) Initial stage of tip fusion;

ii) Retention forces dominate - detachment inhibited;

iii) Critical droplet radius - detachment imminent;

iv) Retention forces overcome - detachment occurs

Metal transfer in MIG welding

The MIG welding process supports three major natural modes of metal transfer i.e. spray, globular and short-circuiting or dip transfer. The high current spray mode of operation is characterised by a stream of fine molten droplets which are detached from the wire tip and travel across the arc to the weld pool. Globular metal transfer occurs with mid-range currents in conjunction with a relatively high voltage, and involves a lower frequency transfer of coarser droplets whose diameter is typically two or three times that of the wire. The low current, low voltage operating regime is characterised by periodic short-circuits of the wire with the molten pool, transfer of filler material being effected during this phase of the repetitive cycle.

An important variation on these three basic modes is realised in the pulsed spray type arc in which discrete metal droplets are induced to detach by the application of a current pulse. The superimposition of these current pulses upon a lower background current results in axial spray transfer and a steady open arc at a mean current level well below that required for conventional spray arc welding.

During spray type metal transfer, whether induced by direct or pulsed current, the resulting weld appearance and quality are critically dependent on a regular and smooth detachment of droplets from the wire. Indeed, most reported incidents of MIG arc instability are associated with spray transfer, where non-axial or irregular detachment can result in excessive spatter and violent fluctuations in arc length. ^[12] Investigations into these problems showed that the common feature was a breakdown in the regularity of the size, shape and transfer frequency of droplets, and these variations have been attributed directly to the wire composition rather than to any malfunction of the welding equipment

A metal droplet forming on the wire tip is urged to separate by the forces of gravity, aerodynamic effects from the gas shielding and an electromagnetic pinch arising from the high local current density. Opposing these forces are those of surface tension and the back reaction caused by metal vapourisation at the active spot of the electrode tip ^[13] , both of which promote the formation of a large pendant droplet. The droplet detaches at some critical radius where the forces favouring separation exceed those for retention, ( Fig.7). Hence the surface tension of the molten filler wire plays an important part in governing the size of the droplets which travel across the arc.

The commonly found surface active impurities (sulphur and oxygen) when present in significant levels will result in a lower surface tension and consequently will encourage a higher frequency transfer of finer droplets. The elements Ce, Mg, C and Si are also known to affect the surface tension. ^[13] Ce and Mg form sulphides and oxides which lower the free sulphur and oxygen levels and hence raise the surface tension, whereas C and Si result in a lower surface tension of the melt. It is thought that these latter elements may influence surface tension directly by the formation of a surface active complex such as SiC. ^[14]

Major variations in arcing behaviour have been observed for austenitic wires conforming to the same specification, but which originate from different casts or heats. ^[15] Coarse, irregular metal transfer was found to be associated with erratic arc behaviour, whilst fine droplet transfer resulted in satisfactory arcing. Analysis of a number of such wires revealed a correlation between the mode of droplet transfer and the type of oxide inclusions present. In the case of fine droplet transfer the oxygen was predominantly combined with Fe, Mn and Cr, whereas Al ₂ O ₃ and MgO inclusions were dominant in wires exhibiting large metal droplets. These findings were interpreted in terms of the surface tension forces on the droplet, which must be overcome for detachment to occur.

During melting of the wire tip, oxygen will be produced more rapidly from the decomposition of Fe, Mn and Cr oxides than from the high melting point oxide Al ₂ O ₃ . Since oxygen reduces the surface tension of iron melts, wires containing the lower melting point oxide inclusions would be expected to promote regular transfer of small droplets. Conversely, a molten droplet starved of oxygen would have a higher surface tension and hence a tendency to grow on the wire tip. Even the use of shielding gases containing up to 5% oxygen did not counteract the adverse effect of Al ₂ O ₃ inclusions.

Recent research into MIG welding with a transistorised power source has revealed two distinct types of spray metal transfer, termed drop spray and stream spray. ^[16] Using a 1.2mm diameter mild steel wire in conjunction with Ar-5%CO ₂ shielding gas, globular transfer was observed below 25oA and conventional (stream) spray transfer occurred above 270A. However, between 250 and 270A, regular detachment of spherical drops took place and the arc was observed to change markedly. Furthermore, when operating within this narrow current range, the process was characterised by a high melting efficiency, low spatter levels and low fume formation.

The transition from drop to stream spray transfer is explained by considering the balance of opposing electromagnetic and surface tension forces acting on the electrode tip. It was proposed that the mean droplet temperature rose sharply at 270A, resulting in boiling at the surface which significantly reduced the surface tension, so the tip became more tapered and the metal transfer changed to the more conventional stream of finer droplets.

The globular and short circuiting modes of operation effectively cover the entire current range with CO ₂ shielding gas because spray transfer is inhibited by the arc roots which push the forming droplets sideways prior to their erratic detachment. In both modes, titanium has been found to significantly influence the welding characteristics of steel electrode wires. ^[17] Additions of Ti result in a softer arc and less spatter during globular transfer, yet Ti-bearing wires are not generally suitable for short-circuiting welding because of the tendency for the wire tip to grow excessively. Although spatter was reduced, the short-circuit frequency also fell sharply which is an undesirable feature for positional welding. The high affinity of titanium for oxygen explains why Ti-bearing wires suffer from enlarged tips, since any reduction in the 'free' oxygen concentration will increase the drop surface tension and hence encourage growth prior to detachment.

Bead shape and defect formation

Everyday experiences of surface tension effects lead us to suspect that surface tension plays a major part in determining the shape of a solidified weld bead. For example, a droplet of mercury placed on a piece of ordinary glass tends to remain as a well-defined droplet standing proud of the glass because its high surface tension acts to minimise the surface area/volume ratio and inhibits the mercury from 'wetting' the glass. In contrast, a water droplet has a much lower surface tension and consequently the droplet spreads out under gravitational forces to form a thin layer covering a wide area, i.e. the degree of wetting is considerably improved.

Ideally, weld bead surfaces should normally be wide and flat ( i.e. large radius of curvature) and exhibit a smooth transition of profile at the weld toes. These characteristics are associated with a low liquid metal surface tension which promotes good wetting of the solid surface. However, occasionally welds are produced which are narrow, humped or peaky in appearance, or even display undercutting in extreme cases, ( Fig.8). These undesirable properties have been largely explained in terms of the surface tension of the molten weld pool prior to solidification. ^[18-21]

Fig.8. Variations in MIG weld profile with two different casts of stainless steel wire, using identical welding conditions: [20] 

a) Surface appearance;

b) Flat profile - low surface tension, good wetting;

c) Peaky profile - high surface tension, poor wetting

Undercutting is normally caused by excessive welding speed which causes the weld metal to solidify before surface depressions created by the arc forces can be completely filled. ^[18] Surface tension forces control the degree of wetting at the edges of the weld pool, which is improved if oxides are present on the molten metal. A low weld pool surface tension promotes good wetting characteristics which encourage the molten metal to climb the sides of the arc gouge and refill the groove before solidification occurs.

Poor weld bead profiles have been encountered when stainless steel MIG wires contained intentionally added rare earth metal (REM) additions. ^[19] It is thought that the REM treated wires form a viscous, refractory-type slag film composed of high melting point oxides of rare earths. This slag substantially increases the weld pool surface tension causing narrow peaky weld profiles and undercutting. As described in the previous sections, oxygen has a significant lowering effect on surface tension and this is further illustrated by the occasionally high sensitivity of stainless steel MIG weld profiles to the concentration of oxygen in the argon shielding gas, ^[20] ( Fig.9). Increasing the oxygen percentage from 1 to 5% was found to alter appreciably the contact angle of the weld metal with the parent material ( i.e. improved wetting at the weld toes) provided that the wire had low levels of soluble sulphur and oxygen. A weld pool high in these impurities would have little affinity for additional oxygen absorbed from the shielding gas.

Fig.9. Effect of shielding gas oxygen concentration on MIG weld profile, ^[20] with two types of wire (Wire A - low sulphur, oxygen. Wire B-high sulphur, oxygen):

a) Wire A, Ar-1%O ₂ ;
b) Wire A, Ar-2%O ₂ ;
c) Wire A, Ar-5%O ₂ ;
d) Wire B, Ar1%O ₂

In the case of manual metal arc welding, the situation is further complicated by the presence of many compounds in the electrode coating which promote fluxing and slag formation, improve the arc characteristics, and provide binding. Large variations in weld shape were produced during investigations with a wide range of electrode coating compounds. ^[21] Peaky welds were found to arise when the coating ingredients resulted in an increase in the measured molten steel surface tension, although the precise effect of each compound could not be predicted in advance. This lack of certainty regarding the mechanism governing these observations is perhaps not surprising when the complex nature of slag-metal-gas reactions is considered.

An important aspect of full penetration TIG welding of butt joints is control of both the top and penetration bead profiles. The latter is particularly important in process tube welds where an overly convex bead results in a lower fluid transmission efficiency, and a slightly concave profile is regarded as an unacceptable reduction in wall thickness.

When using an orbital welding technique, the balance of arc, gravitational and surface tension forces is continually changing making it difficult to achieve a consistent weld profile around the tube circumference ( Fig.10). A mathematical treatment has been produced, based on surface tension and heat flow theory, attempting to predict fusion zone geometries for a range of welding positions and parameters with mild steel, stainless steel and aluminium alloy. ^[22] A satisfactory correlation between the fusion profile and the welding parameters was indeed obtained, confirming the importance of weld pool surface tension and providing a basis for the prediction of TIG weld geometries directly from the welding conditions.

Fig.10. Changes in the balance of major forces on the weld pool during orbital TIG welding

Laser surface melting of metals and alloys is a technique used to eliminate surface defects on powder metal parts and castings, or to alter the surface composition of a material by alloying. However, one problem encountered is solidification rippling which results in a rough surface finish. The phenomenon has been explained in terms of the negative surface tension-temperature gradient exhibited by most metals, that is, the tendency for the surface tension to decrease with increasing temperature. ^[23]

Under the beam centre the liquid temperature is at its highest, hence the surface tension is a minimum in this region. As the temperature of the liquid decreases away from the centre, the surface tension rises pulling material radially outwards, resulting in a central surface depression and a heightening of the surface elsewhere, ( Fig.11). This build up continues until the rising pressure head causes an opposing counterflow of equal magnitude. Because of the large temperature gradients and rapid solidification rates associated with the process, this steady state distortion of the surface becomes frozen into the solid as an undesirable ripple pattern. A practical solution to this problem is to maintain the laser beam sweep velocity above a critical value, such that the surface is rapidly melted and frozen before surface tension induced rippling can occur. ^[23]

Fig.11. Surface distortion caused by surface tension forces during laser surface melting resulting in an undesirable ripple pattern ^[23]

Surface tension effects also play a significant part in the occurrence of weld metal porosity. ^[24,25] MMA welding of austenitic steels with high efficiency electrodes is characterised by an increased susceptibility to the formation of hydrogen pores, arising from the high hydrogen concentration of the iron powder present in the electrode coating (typically 100-200cm ³ /100g). ^[24] Gas bubbles are inevitably formed, but porosity is avoided provided that they float to the weld pool surface prior to solidification. Because the flotation rate of even the smallest bubbles is higher than the crystal growth rate, the greatest likelihood of porosity formation is during the delayed solidification process itself, at which time the gas bubble must grow to some critical volume to enable separation from the substrate. Experiments have shown that this critical volume, and hence the tendency for porosity formation, increases with increasing liquid surface tension and with an increase in the wetting angle at the boundary between solid and liquid phases. The electrode coating composition has been found to be an important factor in the control of weld pool surface tension and therefore the level of porosity experienced.

On a microscopic scale, the successful nucleation of gas bubbles on non-metallic inclusions has been attributed to weld pool wetting behaviour. ^[24,25]

Three forms of pore nucleation are shown in Fig.12 their occurrence depending to some extent on the size of the non-metallic inclusions, but mainly on the degree to which they are wetted by the molten weld pool. Liquids with a high surface tension have poor wetting characteristics and do not encourage the early separation of bubbles from their host inclusions. This separation ahead of the advancing solidification front must occur if weld metal porosity is to be avoided.

Fig.12. Formation of gas nuclei on solid non-metallic inclusions: ^[25]

a) On small inclusions;
b) On large inclusions not effectively wetted by the liquid metal;
c) On large inclusions well wetted by the liquid metal

The final aspect of weld bead formation that will be considered concerns the submerged arc process, where, as with MMA welding, analysis of interfacial tension is complicated by the presence of a molten flux over the liquid pool and solid base metal. The interfacial tension between the flux and solid metal ( γ ^fm ), and the flux and liquid metal ( γ ^fl ), have both been shown to influence weld profile, particularly penetration. ^[26] As γ ^fm decreases, wetting of the base metal surface is improved, inhibiting lateral expansion of the weld pool. An increase in γ ^fl encourages the flux and liquid pool to minimise their surface area by forming a more rounded, humped bead. Both phenomena result in the heat input being confined to a narrower region and the penetration increases as a direct consequence, ( Fig.13).

Fig.13. Influence of flux-metal ( γ ^fm ) and flux-liquid ( γ ^fl ) interfacial tensions on submerged arc weld penetration ^[26]

High power density welding

Three major welding processes can be classified as high power density processes (> 10 ⁹ W/m ² ), namely plasma-arc, laser and electron beam. This high power density produces a crater depression in the molten pool surface which may extend through the entire plate thickness to form a cylindrical hole, generally referred to as a 'keyhole.' Movement of the torch or gun relative to the workpiece causes metal to melt in front of the advancing keyhole, travel around the hole sides, and solidify to form the weld bead at the rear. The characteristically high depth to width ratio of the resulting welds leads to fast joint completion speeds with minimal distortion.

Five forces should be considered when analysing the establishment and stability of a keyhole. ^[27] Three act to form and maintain the keyhole:

c) Pressure from process momentum. The pressure arising from impinging photons and electrons in laser and electron beam welding respectively is negligible. However for plasma arc welding, the momentum of the gas stream plays an important part in maintaining the keyhole.

Two further forces act to close keyhole:

e) Surface tension forces. The internal surface of the keyhole acts like a cylindrical elastic membrane which attempts to contract and will collapse the keyhole if the incident energy density falls below a critical value.

For non-fully penetrating welds with deep narrow keyholes, surface tension forces oppose both the deepening and widening of the keyhole since this would require an increase in the internal keyhole surface area and hence the total surface energy. The balance of forces acting within such a keyhole is shown in Fig.14, the significance of the surface tension pressure depending largely on the radius of the keyhole. The relatively large radii associated with plasma-arc keyholes (typically 3mm) result in a surface tension pressure which is small compared to that present in electron beam and laser welding keyholes (typically 0.5mm radius). For electron beam and laser welding, surface tension plays a major part in determining keyhole penetration and stability characteristics.

Fig.14. Balance of forces at the bottom and sides of a non-fully penetrating keyhole. ^[27] The magnitude of Ps increases as the keyhole radius decreases; Ps (bottom) = 2 γ/r, Ps (sides) = γ/r

Surface tension forces can appear to act in the opposite sense to that described above in the case of a fully penetrating weld where the material thickness is less than the weld width. Under these conditions surface tension attempts to reduce the surface area of the molten material and effectively pulls back liquid at the upper and lower plate surfaces. The natural inclination for surface tension to close the keyhole is overcome and a permanent cut is formed behind the advancing beam or arc ( Fig.15).

Fig.15. Fully penetrating weld in thin material. Under these conditions surface tension forces pull the rear of the keyhole apart resulting in a permanent cut ^[27]

Measuring surface tension

Having described the importance of surface tension at some length it is appropriate to include details of two experimental techniques for its measurement. Attempts to measure pure liquid iron surface tension date back to 1955; before this date results were erratic because it was not known that surface active elements in impure materials could dramatically affect the surface tension.

The majority of early determinations were carried out with the static 'sessile-drop' method. ^[28,29] The apparatus consisted of a horizontal silica tube capable of evacuation to pressures below 10 ^-5 torr. A surrounding high frequency induction coil generated heat inside a tantalum susceptor which in turn radiated heat to the test specimen (mass approximately 3g) placed on a recrystalised alumina base. Once molten, the droplet temperature was measured with an optical pyrometer and photographs were taken to record the droplet shape. Considerable care was exercised during the handling of the samples to prevent contamination prior to melting, and the furnace tube was thoroughly degassed before inert argon was introduced for the duration of the fusion cycle. The surface tension could be deduced mathematically from the resulting droplet photographs which depicted the surface profile and contact angle with the base plate. Two major problems with the technique were the calculation of the droplet density which was required for calculation of the surface tension, and contamination of the molten sample by the surface of the base plate.

The technique has undergone some modification recently with the use of a quartz furnace tube, zirconia and yttria base plates which allowed higher temperature measurements above 2000°C, a two colour infra-red pyrometer and tungsten-rhenium thermocouples measuring temperature to ±1% accuracy, and improvements to the optical system. ^[30] A further modern modification was the use of a computer program to calculate surface tension from numerous measurements of the drop perimeter. ^[29,30] The program was also capable of calculating the droplet volume, from which the density could be derived after sample weighing.

The second major experimental technique is the dynamic 'levitated-drop' method based on the relationship between the natural frequency of an oscillating liquid drop and the liquid surface tension, first established by Lord Rayleigh in 1879. A molten droplet weighing approximately 1g was levitated by a high frequency (450kHz) electromagnetic field in a stream of inert gas, and the fundamental vibrational frequencies of the droplet were monitored with a high speed cine camera. ^[31] The advantage of this method over the sessile-drop method described previously is that there is no possible contamination or interaction with a base material and the surface tension is calculated from a knowledge of only the droplet mass and the fundamental oscillation frequency. The droplet temperature was measured with a two colour optical pyrometer and varied by adjusting the power supplied to the levitating coil, or by varying the He/Ar ratio of the inert gas stream. ^[32]

Modifications to the technique in recent years have concentrated on eliminating the laborious and expensive procedure of filming and subsequently analysing droplet oscillations. By allowing the droplet image to fall on a photo-resistive detector, an electrical signal was generated and fed to an oscilloscope simultaneously with a signal of known frequency, ( Fig.16). The two traces were photographed and the oscillation frequency determined by comparative measurements of the two signals. ^[32] A further development was the introduction of a Fourier Analyser linked directly to the photo-detector, completely eliminating the optical apparatus. ^[33] The analyser rapidly resolved the source signal into its component frequencies, hence the sample needed to be maintained at a high temperature for only a few seconds, limiting the chance of oxygen contamination and material loss by vapourisation.

Fig.16. Droplet levitation apparatus used to determine the surface tension of molten samples ^[31-33]

Both the sessile drop and levitated drop methods are currently being employed to investigate the problems of variability in both TIG weld fusion geometry ^[8,30] and MIG weld bead profile, ^[20] although the latter method is now generally regarded as the more accurate.

Summary

This discussion of surface tension phenomena has attempted to explain the origin and significance of surface tensions effects with regard to several aspects of welding. It has illustrated in qualitative terms that molten metal surface tension has a critical influence on TIG penetration profile, MIG metal transfer, weld bead formation, incidence of defects, and keyhole formation with high energy density welding processes. A considerable quantity of literature has been published covering surface tension effects related to welding, yet many papers contain either a largely mathematical treatment, or an overly simplistic approach often involving other parameters such as viscosity and fluidity which need not be directly related to surface tension.

Both approaches are valid but have probably not encouraged a wider understanding of the basic principles involved:- This situation is worsened by the fact that the measurement of surface tension is not a routine task, but involves elaborate apparatus and a well controlled experimental procedure. Consequently it is only recently that investigators have been able to back up theories concerning surface tension with accurate experimental determinations. It seems certain, though, that careful attention to surface tension effects will be of increasing importance in the continued understanding of welding processes and the future development of parent material and consumable specifications.

References