How TIG welding procedure affects penetration and slag island formation - part 2

TWI Bulletin, January/February 1989

David Harvey, MA, is a Senior Research Engineer in the Arc Welding Department at The Welding Institute.

Effects of welding procedure on penetration and slag island formation were examined in part 1. Part 2 concludes by investigating dispersal of these islands and obtaining uniform penetration profiles.

Phase III: Slag island dispersion

Two techniques were applied in an attempt to disperse the slag islands and reduce the erratic penetration associated with their appearance:

1. Arc oscillation using an external electromagnetic field;
2. Pulsing the welding current.

These were used with a shielding gas/travel speed/arc energy combination known to produce slag islands.

Magnetic arc oscillation

When used with welding conditions known to produce slag islands, the oscillating magnetic field dispersed slag into smaller pieces ( Fig.15). Although slag was still formed, arc disturbances were eliminated and penetration was consistent.

Pulsing the welding current

A typical slag generating condition was taken and simple pulsed welding current values determined giving the same mean welding current. Although slag was still formed, it was dispersed by the pulsing action so that arc disturbances were eliminated and the previously erratic penetration was improved ( Fig.16, 17).

Phase IV: Materials analysis

Inspection of cast 3D56 took the form of metallographic examination, qualitative scanning electron microscope (SEM) chemical analysis and bulk chemical analysis.

Metallographic examination

Cast 3D56 displayed a macrostructure typical of a type 316 austenitic stainless steel, with banded ferrite and no gross inclusions. However, inspection at magnification of X1000, indicated the presence of widely spaced and randomly located bands of inclusions up to 2µm across and 100µm in length, parallel to the rolling direction ( Fig.18).

Table 1 Chemical analyses of two casts of Type 316 stainless steel, wt%*

Table 1 Chemical analyses of two casts of Type 316 stainless steel, wt%* - continued

Table 2 Calcium concentration in type 316 stainless steels, wt%

SEM analysis

A qualitative SEM analysis performed on slag free areas of weld metal indicated the presence of those elements usually found in a type 316 stainless steel, i.e. iron, chromium, nickel and molybdenum ( Fig.19).

Examination of slag islands revealed strong indications of two additional elements, calcium and silicon ( Fig.20).

Examination of the fine bands of inclusions in cast 3D56 again indicated elements usually present in type 316 stainless steel, plus calcium and silicon, and four other elements, magnesium, aluminium, titanium and copper, not found in the slag islands ( Fig.21).

Bulk analysis

Bulk analysis of casts 3D56 and 3D55 by X-ray fluorescence (XRF) and inert gas fusion (IGF) indicated that their chemical compositions lay within the specification for type 316 stainless steel, Table 1. To obtain a value for calcium content, optical emission spectroscopy (OES) was employed, Table 2. Four other casts were analysed in an attempt to establish a more general picture for the concentration of calcium in type 316 stainless steel. With the exception of one cast (1C135), cast 3D56 contained about two to three times as much calcium as the other sample casts. Within the limited range of welding parameters applied to cast 1C135 slag islands were not generated.

Discussion

The cast under investigation in this study suffered from two distinct and apparently unrelated problems; poor penetration characteristics and slag island formation.

Variation in cast weldability as a result of poor penetration is an increasingly familiar problem during mechanised TIG operations. The mechanism most frequently proposed to explain cast to cast variation in weld penetration suggests that the concentration of minor elements, such as sulphur and oxygen, critically affects the surface tension of stainless steel. ^[6] The temperature dependence of surface tension is affected in such a way as to produce two possible types of convection flow in the molten pool ( Fig.22). The flow is driven by differences in temperature between the edge and centre of the molten pool. Choice of shielding gas has a major influence on the molten pool surface temperature distribution, e.g. helium is more thermally conductive than argon, resulting in a lesser thermal gradient across the surface of the molten pool. Welding current, travel speed and arc voltage define arc energy input, which determines size, shape and temperature of the molten pool and the period during which it remains molten at any point along the centreline. Factors such as maximum pool temperature and reaction time available may significantly alter minor element concentrations on which penetration characteristics appear to depend. Despite tentative general agreement over factors influencing molten pool behaviour, the complex reactions involved are not well understood.

Formation of slag islands is a less widely documented problem. Although slag generation is frequently observed, amounts produced rarely affect process operation, weld quality or cosmetic appearance to the extent of the problematic cast under investigation in this study. Reactive elements such as calcium and magnesium are known to promote arc instability as a result of formation of stable refractory oxides on the molten pool surface. ^[7] Scanning electron microscope analysis of the slag islands indicated a strong presence of calcium and silicon, whilst adjacent slag free weld metal and parent material indicated no measurable calcium and a greatly reduced concentration of silicon. There is evidence linking slag island formation during TIG welding of type 316 stainless steels to the presence of calcium bearing inclusions in the metal. ^[8] It is suggested that inclusions float to the surface of the molten pool where they combine to form calcium silicate bearing slag. Widely spaced and randomly located calcium and silicon containing inclusions were observed in the sectioned parent material.

Welding trials indicate that slag islands were formed with certain combinations only of the parameters shielding gas composition, welding current and travel speed. It can be seen that adding either a high proportion of helium to the shielding gas or 5% hydrogen greatly reduces the range of welding parameters producing slag islands. Addition of either of these gases increases arc voltage and consequently arc temperature, and may result in greater stirring of the molten pool, inhibiting coalescence of inclusions. Furthermore, hydrogen is often added to shielding gas to provide a chemically reducing atmosphere and should reduce the incidence of slag formation.

Whilst modification of welding parameters can be used to eliminate slag formation, two simple techniques have been established which increase the process tolerance to slag formation (without its eradication); either pulsing the welding current or oscillating the arc using an externally applied magnetic field. Most mechanised TIG welding equipments have pulsed welding current facilities, hence in most instances this is the simplest modification to the welding procedure. Where pulsed welding current is not available magnetic arc oscillation is a relatively cheap alternative. Weld penetration can be maintained by oscillating the arc in the welding direction.

Summary

A particularly problematic cast of type 316 stainless steel displayed both poor penetration characteristics and slag island formation. There was no apparent relationship between the two phenomena, although both could be overcome by careful selection of welding parameters. In particular, addition of hydrogen and use of the helium rich shielding gas (80%helium-20%argon) proved beneficial across the greatest range of welding current and travel speed combinations.

Tolerance of the TIG welding process to slag island formation could also be improved by either pulsing the welding current or oscillating the arc under the influence of an external magnetic field.

References