Porosity in aluminium welds is caused by gases that are absorbed in the molten weld pool, and are then not able to escape before the pool solidifies. The primary gas of concern for aluminium is hydrogen, which originates from a contaminant, either a hydrocarbon (grease, skin oils) or moisture, both of which can break down in the arc plasma to produce atomic hydrogen.
Hydrogen has a very high solubility in molten aluminium, and can be easily absorbed into the weld pool, but it has a much lower solubility in solid aluminium (approximate 20x lower), and so as the aluminium solidifies, hydrogen gas is evolved and forms gas pores which then are frozen within the weld bead.
There a number of possible ways these contaminants can be introduced to the weld pool. The first is directly, if they are present on eg the parent material being welded, or the filler wire used to make the joint. They may also be on tools or welders' hands, and can be introduced by touching the surface.
An important element of the parent and filler material characteristics to note is that aluminium forms a thick oxide on its surface (which gives it its good corrosion resistance properties). It is possible for this oxide to become "hydrated" ie it absorbs moisture simply from the surrounding air, and if this oxide is introduced into the weld pool, it can release this moisture into the weld pool and act as a source of hydrogen.
Secondly, moisture may be present in the atmosphere, for example in either a damp cold environment, or a hot humid one. If the shielding of the weld pool by an inert gas is insufficiently good (or is interrupted in some manner), moisture may be drawn into the arc and therefore absorbed into the weld pool. Atmospheric moisture may also be drawn in from the root of partial penetration or fillet welds, where it may not have been possible to fully clean existing hidden surfaces.
The flow of shielding gas itself can act as a source of moisture, either if there is contamination in the shielding gas source (eg a cylinder) or if the transmission of shielding gas from source to welding torch allows for moisture pickup. For example, gas transmission lines which are porous or contain leaks may allow environmental moisture to encroach into the otherwise dry shielding gas, or in a cold enviroment, when shielding gas is not flowing (perhaps overnight), condensation can allow a build-up of moisture within the gas lines.
Finally hydrogen may come from the originally solid material itself, particularly in cases of cast or sintered productes.
The primary avoidance of porosity in aluminium is therefore achieved by reducing the potential for hydrogen to be absorbed into the weld pool. This is done by accounting for all of the possible sources above, as well as process features of the different welding methods.
The most important element is generally considered to be cleanliness. Any possible contaminants need to be removed from the parent and filler material, both in the form of direct contamination, but also by removal of potentially hydrated oxide. Prior to welding, any surface to be welded (and a portion of the parent material outside of this) needs to be thoroughly degreased, typically by a solvent cleaner such as acetone or similar. Solvent cleaners also contain hydrocarbons and/or water and need to be removed prior to welding, but typically are volatile and will evaporate quickly. However, once a joint is assembled, it is very difficult to remove any cleaning solution, so cleaning should always precede joint assembly.
After cleaning, a mechanical operation, often stainless steel brushing, is required to remove the potentially hydrated surface oxides. This should be performed with dedicated tools which are kept in conditions to minimise any possible contamination.
More aggressive chemical cleaning solutions can be used, such as sodium hydroxide or nitric acid which will remove the oxide film directly, but these naturally come with more stringent health and safety requirements.
When parent material and filler wire have been cleaned, welding needs to proceed as quickly as possible before significant oxide can reform and re-absorb moisture on the surface of the material. A maximum of four hours may be recommended between cleaning and welding. Materials should be handled carefully after this point, with clean gloves worn and wire stored in clean conditions (eg not on a contaminated workbench).
To avoid introduction of contaminants through the shielding gas, a number of good practices are recommended. The distance the gas has to travel from source to torch should be minimised, and piping materials with a low permeation characteristic should be used (stainless steel being the most expensive but most effective when dealing with long distances). Mechanical joints are generally considered more prone to leaks than brazed or welded joints. Gas systems should be checked for leaks, and the gas quality should be checked at the torch, to achieve a dew point preferably below -50degC. Where systems are left idle for some time, pre-purging gas through them may remove condensation and reduce porosity in the first welds to be produced.
Introduction of enviromental moisture into the weld pool can be eliminated or reduced by avoiding disruption of the arc or shielding gas flow. The welding area should be set up to avoid draughts, and the wire feed should be smooth and consistent to reduce the risk of turbulence in the weld pool or arc plasma. The wire should contain no kinks, and be fed through a contact tip of the correct size and good condition, by drive rolls which are correctly adjusted. A new wire feed liner, or re-positioning of a poorly placed torch cable can reduce some of these problems.
Welding parameters can also have an impact on turbulence and porosity. A long arc gap can give molten droplets more time to pick up enviromental moisture and introduce these to the weld pool, and a higher welding current can increase the temperature of the weld pool, increasing the hydrogen solubility, and allowing it to absorb more hydrogen which is then ejected on solidification.
However, an increase in welding current, as well as a reduction in travel speed, can also potentially improve porosity by giving hydrogen more time to "bubble out" of the weld pool, if the rate of gas evolution is higher than the rate of gas absorption. This can also be aided by preheat, heating the backing bar if one is used, or by using an Ar-He shielding gas, rather than purely argon.
During the welding process, oxide is also removed due to the nature of the welding process. In TIG welding, an AC welding current is used, which rapidly flips the polarity of the welding current between having the tungsten electrode negative and the workpiece positive, and the tungsten electrode positive and the workpiece negative.
When the electrode is positive, electrons are stripped from the workpiece and it is bombarded with ions, and this breaks up and disperses the oxide film. However, this results in significant heating of the tungsten electrode and so cannot be used all the time, resulting in the use of AC as a compromise position, with the electrode positive generally 20-40% of the time, with this value being set for a particular welding application. In MIG welding, the electrode is always positive and so cleans by default.
Finally, where possible aluminium welding should be kept separate from other material fabrication, to avoid cross-contamination via fume, metal powders or tool use.