Achieving low-porosity laser welds in aerospace aluminium alloy

TWI

S Barnes
Warwick Manufacturing Group

Paper presented at 2003 Aerospace Manufacturing Technology Conference (AMTC), 8-12 September 2003, Montreal, Canada.

Abstract

Aluminium is currently the preferred material and riveting the preferred joining method for the manufacture of thin-gauge airframe structures. Although the potential of laser welding as a low-distortion alternative for such applications is recognised, questions are still being raised about the weld quality, and in particular the porosity levels, that can be achieved in aluminium. This paper focuses on the cleaning of parent material and filler material prior to welding, the use of a twin-spot energy profile in the laser beam focus and the use of a low-moisture shielding gas and shielding gas delivery, and their individual and combined influence on the presence of weld metal porosity for Nd:YAG laser welds in 3.2mm thickness 2024 aluminium alloy. The paper describes how, through careful selection of processing conditions and aforementioned factors, fully penetrating, square-edge butt welds were achievable with levels of weld metal porosity lower than those specified in the stringent weld quality class of standards relevant to the aerospace industry, including the European BS EN ISO 13919-2:2001 and the American AWS D17.1.

Introduction

Increased international competition has encouraged aerospace companies to investigate new approaches to aircraft design and manufacturing methods to provide reliable and competitively priced products. ^[1] The work described in this paper was carried out as part of an initiative in this field, a programme named CEMWAM (Cost Effective Manufacture: Welding of Aerospace Materials), initiated by a number of leading UK industrial companies, Research and Technology Organisations (RTOs) and Universities.

Currently, the preferred manufacturing route for aircraft fuselage structures is riveting and the principal material for these structures is aluminium. Recent analyses, however, have indicated that a move from riveted to welded airframe structures could lead to manufacturing cost savings in the region of 30%. ^[1] Laser welding is one of the processes currently being considered for this, because of the high processing speeds, low heat input, low distortion, good weld quality and the overall flexibility that the process offers. ^[1-3] Over the past few years, there has been a great effort in the automotive sector to introduce laser welding technology onto the shop floor. A great deal has been learnt from these experiences, but further scrutiny of welding procedures is required to ensure that the weld quality needed for aerospace applications can be achieved reliably, in particular for the laser welding of aluminium.

Although possible, the laser welding of aluminium is generally perceived to be difficult because of the initial high surface reflectivity and the high thermal conductivity of aluminium, both of which contribute to the risk of weld imperfections such as lack of penetration or cracking, in certain alloys. Weld metal porosity is also frequently associated with the laser welding of aluminium and this weld imperfection is the subject of the work described in this paper. Weld metal porosity is always an issue when fusion welding aluminium ^[4,5] and laser welding is no exception. What causes porosity and, in particular, how it is formed in laser welds are issues still being debated, but not considered in detail here. Some suggest that volatilisation of low boiling point constituents in some of the aluminium alloys causes keyhole instabilities, others believe it is simply entrapment of shielding gas in the solidifying weld pool, whilst others attribute porosity to hydrogen entrapment during weld pool cooling and solidification. ^[1,6,7] Irrespective of the cause however, porosity is generally categorised as either fine or coarse, typically differentiated at an average pore diameter of 0.5mm. Fine porosity appears as a distribution of spherical pores and is generally understood to originate from hydrogen or from the rejection of dissolved shielding gases on solidification. Coarse porosity is characterised by larger, more irregularly shaped voids, randomly distributed throughout the weld bead. These are generally considered to be the result of keyhole instabilities and are typically present in partially penetrating welds. ^[1,7] Coarse porosity can have a detrimental effect on a welded joint's mechanical performance. ^[8]

The work described in this paper details Nd:YAG laser welding trials, carried out on 3.2mm thickness 2024 aerospace aluminium alloy, aimed at reducing both the fine and coarse porosity in aluminium weldments. Particular efforts were on reducing the fine porosity, and especially fine porosity resulting from hydrogen-entrapment. Hydrogen dissolves very rapidly into the aluminium weld pool ^[4] but has a very low solubility in solid aluminium, as is shown in Figure 1. With a high-speed process such as laser welding, the time available for diffusion is sufficiently low that a certain amount of hydrogen can become entrapped in the solidifying weld pool. ^[6] Hydrogen can originate from the parent material, the filler wire or the shielding gas and these sources were all investigated. In addition, the performance of twin-spot Nd:YAG laser welding was assessed to establish whether this technique resulted in a reduced level of porosity, because of the resultant elongated weld pool found when using this technique. ^[9]

Fig. 1. Solubility of hydrogen in aluminium

Scope of work

The welding trials were completed using a high-power flashlamp-pumped continuous wave (CW) Nd:YAG laser delivering 3kW of laser power to the workpiece through a 0.6mm diameter step-index fibre optic. Initial trials used a single spot beam focus 1:1 imaging lens, producing a 0.6mm minimum focus spot diameter. Square-edge butt welds were produced in 3.2mm thickness 2024-T3 aluminium alloy (and some in 3.2mm thickness 6056-T4 aluminium alloy) by traversing the laser output housing, mounted on an articulated arm robot, over a stationary welding jig, holding the samples in position during welding in the flat (PA) orientation ( Figure 2). The material samples were 150mm wide and 300mm long with the cold band sawn (long) edges dry machined to ensure a good joint fit-up. The samples were degreased with acetone immediately after the dry machining operation. Industrial grade helium (purity grade 5.0 or 99.999% pure) conforming to BS EN 439:1994, ^[10] typically containing around 5ppm moisture, was used for shielding both top and bottom of the weld pool. The 1.2mm diameter 2319 and 4047 filler wire for the 2024 and 6056 aluminium alloy respectively, was introduced into the leading edge of the weld pool. ^[11]

Fig. 2. Set-up for welding trials

Process parameters such as welding speed, laser focus position (in relation to the material surface), filler wire position (in relation to the laser-workpiece impingement point) and shielding gas flow rates were varied to obtain fully penetrating, visually acceptable, square-edge butt welds, in accordance with BS EN ISO 13919-2:2001. ^[12] The absence of cracks in the weld metal and heat-affected zone (HAZ) was confirmed through radiographic and macro-metallurgical examination. The presence of micro-cracks in weld metal or HAZ was not examined. Subsequent welds were produced using the process parameters established earlier, but with different conditions for parent material cleanliness, filler wire cleanliness and condition and delivery of process shielding gas. These factors, widely considered to be the main causes of hydrogen-induced weld metal porosity, were varied in a controlled fashion. In addition, the effect of a twin-spot versus a single-spot laser energy distribution on the level of weld metal porosity was also investigated.

A selection of the welds was radiographed and a pore count carried out over a 100mm longitudinal section of the weld, representative of the entire weld length, to quantify the levels of weld metal porosity. These levels and the equivalent pore length/area per given weld length/area calculated from these levels, were compared with the acceptance levels of the stringent weld quality class defined in two international standards, BS EN ISO 13919-2:2001 ^[12] and AWS D17.1:2001 ^[13] , and one company internal standard ABP 2-4102. ^[14] The European standard was selected because it is specific to laser welding of aluminium, whereas the American standard and the internal standard were chosen because they are specific to fusion welding for aerospace applications. Table 1 summarises the acceptance criteria that need to be fulfilled to achieve a given weld quality category according to all three standards. These criteria relate to the diameter of the largest pore, the minimum distance between adjacent pores and the equivalent projected pore area (for the European standard) or pore length (for the American and internal standard) per given length of weld. In this work, the second criterion was not considered as it was fulfilled for all welds.

Table 1 Acceptance criteria for weld metal porosity

Results and discussion

Welding conditions versus porosity

A single spot beam focus position on or 1mm below the material surface helped ensure full penetration in 3.2mm thickness 2024 aluminium alloy with bead profiles, shown in Figure 3, acceptable in accordance with BS EN ISO 13919-2:2001. Focusing the laser beam more than 1mm below the material surface resulted in partial absorption of the laser beam and partial penetration. Notwithstanding the acceptable visual appearance of the welds, some coarse porosity was noticeable along the length of the weld ( Figure 3d). Changing standard welding conditions such as laser power, welding speed, shielding gas flow rate and focus position, within their permitted working envelopes of achieving fully penetrating, visually acceptable welds, proved unsuccessful in eliminating the coarse porosity completely. Subsequent trials investigating the influence of parent material cleanliness on coarse porosity produced better results, as detailed below.

Parent material cleanliness versus porosity

In addition to the initial dry machining (and acetone degrease) of the sample edges after cutting, the samples were subjected, just prior to welding, to a secondary cleaning operation. The cleaning methods examined comprised linishing, abrading, scraping, machining or chemical etching (or chemi-etching), each followed by an acetone degrease with a lint-free cloth to remove residual dirt, moisture or lubricant. The acetone degreasing was also investigated on its own as a secondary cleaning operation.

All cleaning operations, except machining and chemi-etching, were carried out manually and, with the exception of acetone degreasing, performed to remove the porous oxide layer, a potential source of contaminants and moisture, and thus porosity. ^[15] A belt sander with a silicon carbide paper (P120 grit size), an abrasive pad and a single-cut edge, high-strength steel scraping tool were used for the linishing, abrading and scraping respectively. Machining was carried out without cutting fluid using a single-cut edge, high-speed steel cutting tool. All these mechanical cleaning methods were carried out until the as-received aluminium oxide surface was completely removed. For the chemi-etch cleaning, the samples were soaked, at least six times, for two minutes, in a nitric and hydrofluoric acid solution, followed by a water rinsing.

Fig. 3. Weld produced with 3.1kW laser power at workpiece, Ø0.6mm focus spot, -1mm focus position, 1.4m/min welding speed, 1.25m/min wire feed speed (Ø1.2mm ER2319):

3a) Top weld bead profile

3b) Bottom weld bead profile

3c) Cross section

3d) Radiograph (total weld length: 300mm)

Fig. 4. Weld produced after samples were linished, using 3.0kW laser power at workpiece, Ø0.6mm focus spot, 0mm focus position, 1.25m/min welding speed, 1.4m/min wire feed speed (Ø1.2mm ER2319)

4a) Top weld bead profile

4b) Bottom weld bead profile

4c) Cross section

4d) Radiograph (total weld length: 270mm)

Fig. 5. Pore counts for different parent material cleaning methods

Table 2 The effect of parent material and filler wire cleaning on weld metal porosity

Table 2 continued

Figure 6 differs from Figure 5 in that the welds depicted were carried out with a 1:0.75 imaging lens, creating a 0.45mm minimum focus spot diameter instead of the previously used 0.6mm spot distribution. Pores 0.1mm in diameter and smaller are depicted in Figure 6.

Fig. 6. Weld metal porosity for cleaned (parent material) samples, cleaned filler wire and twin-spot energy distribution

As with the earlier results (depicted in Figure 5), little difference was noticeable between the linished and the chemi-etched samples. If anything, the chemi-etched samples exhibited less porosity than the linished samples, because of the uniformity and non-manual operation of the chemi-etching, and, contrary to earlier trials, because the chemi-etch operation was completed much earlier prior to welding, i.e. within 24 hours, through the use of smaller batches. Nevertheless, some coarse porosity was present in these welds, which was not present in earlier welds carried out using similar (cleaning) conditions or in any of the welds carried out subsequently. A contributing factor to the coarse porosity was the 40% increase in welding speed, i.e. 1.75m/min instead of 1.25m/min used before, necessary to compensate for the higher power density resulting from the 25% reduction in focus spot diameter, i.e. 0.45mm instead of 0.6mm used in earlier trials.

Filler wire cleaning

In combination with the parent material cleaning trials detailed above, the influence of filler material cleaning was also investigated by comparing porosity levels of samples welded with as-received filler wire (pore count results on left hand side for each of the parent material cleaning conditions in Figure 5) and samples welded with filler wire cleaned with an abrasive pad followed by acetone degrease (result on right hand side for each of the parent material cleaning conditions in Figure 5). No clear conclusions can be drawn from it, or from Table 2, on the influence of filler material cleanliness on weld metal porosity, because the results are masked by the influence of parent material cleanliness. The fact that, for example, for the chemi-etched samples, less porosity was observed for the as-received wire compared with the cleaned wire, could be explained by the difference in time elapsed between cleaning and welding, rather than the effect of filler wire cleaning. To assess the individual effects of parent material and filler material cleaning on weld metal porosity, subsequent trials, detailed in Figure 6 and Table 3, were all carried out with the 1:0.75 imaging lens using samples that were chemi-etched less than four hours prior to welding. Welding was carried out with as-received filler wire, as well as with filler wire that was chemi-etched prior to welding, instead of the abrasive pad wipe and acetone degreasing used earlier. Figure 6 shows that the difference between welding with as-received filler wire and cleaned filler wire is now much more pronounced. This can also be seen in Table 3, with the third criterion of the European and internal company standard met only for those welds produced with chemi-etched parent material and chemi-etched filler wire (for a single-spot laser energy profile). Notwithstanding a clear drop in equivalent pore length/area per given weld length/area when chemi-etching the filler wire, only one of the samples with chemi-etched parent material and filler passed the third criterion of the American standard, despite the absence of large pores in this sample.

Twin spot energy distribution versus porosity

Laser welding is a high-speed process. As such, time for hydrogen bubbles to escape from the solidifying aluminium weld pool is limited, which contributes to the incidence of weld metal porosity in aluminium laser welds. Subsequent trials were carried out to examine if an elongated weld pool created by a twin-spot laser energy profile obtained using a process head with an effective focal length of 150mm, would delay this rapid solidification and thus allow more time for hydrogen to escape. ^[9] Welds were produced in both 2024 and 6056 aluminium alloy using a twin-spot energy profile with a 0.27mm spot separation and a 50/50 energy distribution between the spots. These conditions were based on previous TWI work. ^[16,17]

Figure 6 and Table 3 show that the levels of coarse porosity for welds produced with a twin-spot energy profile are lower when compared to those in welds produced (previously) with a single-spot energy profile. However, not only was the twin-spot technique applied, but the welding was also carried out at lower travel speeds, i.e. between 0.8 and 1.0m/min for the twin-spot energy profile and 1.75m/min for the single-spot energy profile. It was not possible to separate the twin-spot energy profile from this lower travel speed, to investigate their individual effects on the reduction of weld metal porosity, because the low travel speed was inherent to the twin-spot technique. With this technique, the total laser power is shared between the two spots, creating a lower power density in each of the spots, necessitating a lower travel speed, when compared with the single-spot technique, to produce fully penetrating welds in a given material thickness. Furthermore, it was not possible to produce fully penetrating butt welds using the single-spot technique at this lower travel speed either, as such welds had excessive penetration and in most cases local burn-through.

Table 3 The effect of parent material and filler wire cleaning, a twin-spot laser energy profile and the use of a low-moisture shielding gas and delivery on weld metal porosity, for a series of welds produced with 2.85kW of laser workpiece power, a 0.45mm diameter focus spot size positioned on the material surface and helium shielding of top and bottom of weld pool.

Table 3 continued

Fully penetrating, square-edge butt welds were also produced in the 6056 aluminium alloy using the same single-spot and twin-spot conditions as those used for the 2024 alloy. A similar, but smaller influence of the twin-spot technique and the inherent lower travel speed (compared with the single-spot technique) on reducing the level of coarse weld metal porosity was observed for this alloy ( Figure 7). For the welds in the 6056 aluminium alloy, produced using the single-spot energy profile however, no pores larger than 0.7mm in diameter were found, indicating that besides energy beam profile, welding speed and spot size, the material grade or alloy composition can have an effect on the level of coarse porosity.

Shielding gas moisture content versus porosity

To examine how moisture in the shielding gas could contribute to the presence of fine weld metal porosity, welds were also produced using a high purity, low dew-point research grade helium shielding gas and a 'modified' shielding gas delivery system, comprising the shortest (maximum 2m) polyamide tubing lengths possible between gas cylinder and coaxial shielding nozzle. Considering the hygroscopic nature of polyamide, the tubes were 'acclimatised' several hours prior to welding and purged at the onset of the welding trials. The gas delivery system was purged for at least two minutes between subsequent welds. The welds were carried out on chemi-etched samples, with chemi-etched filler wire and using a twin-spot energy profile, based on the thus far accumulated experience.

It can be seen from Figure 6 that the welds carried out using the research grade (low-moisture content) helium shielding gas and the 'modified' shielding gas delivery system, clearly demonstrate a considerable reduction in pore numbers and equivalent pore length/area compared with the welds made under exactly the same conditions but using industrial grade, instead of research grade, helium shielding gas. Moreover, the equivalent pore lengths/areas summarised in Table 3, demonstrate that the welds produced under these shielding gas conditions resulted in a level of weld metal porosity considerably lower than any of those achieved before. In fact, from all welds produced in this work, these were the only set that passed all three porosity criteria for each of the standards, even the third criterion (total pore length per given weld length) of the most stringent standard, AWS D17.1.

Conclusion

A welding procedure was developed for producing fully penetrating, square-edge butt joints in 3.2mm thickness 2024 aluminium alloy using 3kW CW flashlamp-pumped Nd:YAG laser power. By controlling the process conditions, it was possible to achieve a level of weld metal porosity lower than that defined for the stringent quality class in BS EN 13919-1:1997, a typical aerospace industry standard, and even the most rigorous of standards considered, i.e. standard AWS D17.1:2001.

• A focus position on or 1mm below the material surface helps achieve full penetration welds in 3.2mm thickness 2024 aluminium alloy with weld profiles that conform to BS EN ISO 13919-2:2001.
• A high-purity, low dew-point 'research-grade' helium shielding gas, delivered through a moisture and/or condensation-free shielding gas delivery system, should be used, as this produces less weld metal porosity compared with industrial grade helium gas.
• Removing the porous oxide layer prior to welding contributes to reducing the weld metal porosity in laser welded 2024 aluminium alloy. Linishing, scraping, machining or chemical etching can be used for this purpose, but the elapsed time between material preparation and subsequent welding needs to be as short as possible (less than 24 hours recommended) to avoid atmospheric moisture pick-up.
• A further reduction in weld metal porosity can be achieved by cleaning the filler wire, for instance with a chemical etching cleaning operation.
• The use of a twin-spot laser energy profile with a 0.27mm spot separation and a 50/50 energy distribution helps eliminate coarse porosity in 3.2mm thickness 2024 aluminium welds. The technique has less of an effect on pores smaller than 0.4-0.5mm diameter.

At this point, parameter development could be carried out for a given component, to provide a robust and reliable material preparation and welding procedure, to Nd:YAG laser weld typical thin-gauge aerospace aluminium alloys. In addition to porosity-reducing procedures common to all aluminium fusion welding processes, it is possible to mitigate the weld metal porosity to levels acceptable to current aerospace welding standards. However, considering the preparation required, this will be at a cost and it is recommended, that, prior to developing a welding procedure, the maximum allowable levels of weld metal porosity should be established, according to the application and required joint performance. It is unlikely for instance, that some of the smaller, non-surface breaking porosity typically found in laser-welded aluminium will affect static or fatigue performance.

Acknowledgments

This work was funded jointly by Industrial Members of TWI (through its Core Research Programme) and by the UK's Engineering Physics Sciences Research Council (EPSRC) (through the CEMWAM project). The author would like to thank the CEMWAM partners for their help, assistance and provision of materials, and the EPSRC for funding the EngD programme. The assistance of Mr F A S Nolan, R Lombardi and A S Spencer, who carried out the processing trials is gratefully acknowledged. Special thanks also to Dr N C Sekhar for his assistance throughout the work programme.

References

1. Dunkerton S B: 'Welding for the 21^st century'. Materials World, vol.10, no.3, Mar 2002, pp18-19.
2. Ion J C: 'Laser beam welding of wrought aluminium alloys'. Science and Technology of Welding and Joining, Vol.5, No.5, 2000, pp.265-275.
3. Clarke J A: 'Laser welding aluminium alloys'. Industrial Laser Solutions, vol.15, no.6. June 2000, pp17-18.
4. Gingell A B D and Gooch T G: 'Review of factors influencing porosity in aluminium arc welds'. TWI members report No. 625/1997, Oct 1997.
5. Mathers E: 'Arc welding aluminium - a guide to best practice'. Technical knowledge section of TWI website www.twi.co.uk, Sept 2002.
6. Sekhar N C: 'Review of laser welding of aerospace aluminium alloys'. CEMWAM report, CEMWAM II/ PT1B/TWI/038, July 2000.
7. Seto N, Katayama S and Matsunawa S: 'Porosity formation mechanism and suppression procedure in laser welding of aluminium alloys'. Welding International 2001 vol.15, pp191-202.
8. Katayama S: 'Evaluation of mechanical properties of laser welded aluminium alloy'. Proceedings of ICALEO 1997, 83 (part 2), G93-G102.
9. Iwase T, Shibata K, Sakamoto H, Dausinger F, Hohenberger F, Muller M, Matsunawa A and Seto N: 'Real Time X-Ray Observation of Dual Focus Beam Welding of Aluminium Alloys'. Proceedings of ICALEO 2000,Section C, pp26-34.
10. BS EN 439:1994 'Welding consumables - shielding gases for arc welding and cutting: Welding consumables - shielding gases for arc welding and cutting'.
11. Gittos M F and Scott M H: 'Selection of filler materials for arc welding aluminium alloys'. The Welding Institute Research Bulletin, vol.29, Feb 1988.
12. BS EN ISO 13919-2:2001 'Welding - Electron and laser beam welded joints - Guidance on quality levels for imperfections'.
13. AWS D17.1:2001 'Specification for fusion welding for aerospace applications'.
14. ABP 2-4102 (Appendix C) 'Acceptance standards for aluminium and aluminium alloys'. Internal British Aerospace Plc standard, issue 3, Jan 1996.
15. 'Welding Aluminium: Theory and Practice'. The Aluminium Association, Third edition, Nov 1997, ISBN 89-080539.
16. Lugan A: 'Twin-spot laser welding of 2024 aluminium alloy'. TWI report 765/2003, Mar 2003.
17. Hohenberger B, Chang C L, Schinzel C, Dausinger F and Hugel H: 'Laser welding with Nd:YAG- multi-beam technique'. Institut fur Strahlwerkzeuge (IFSW), University of Stuttgart, Proceedings of ICALEO 1999conference, Section D, pp167-176.

Contact

Geert is a Mechanical Engineer and European Welding Engineer who started his career in Belgium, working for steel producer Sidmar (part of Arcelor). He joined TWI Ltd (The Welding Institute) in 1996 where he currently works as a Senior Project Leader in the Laser and Sheet Processes Group. He has particular involvement with projects for the road transport and aerospace industry sector.