Laser Welding Aluminium Alloys using Different Laser Sources

J P Weston ^{( i )} , I A Jones ^{( ii )} , E R Wallach ^{( i )}

(i Department of Materials Science, University of Cambridge)
(ii TWI)

Presented at CISFFEL6, 6 ^th International Conference on Welding and Melting by Electron and Laser Beams, Toulon, France, 15-19 June 1998

Abstract

Five different aluminium alloys were welded using four laser sources: CO ₂ , CO and Nd:YAG (pulsed and high powered continuous). The laser welds were studied for their appearance, mechanical properties, hot cracking response and microstructure. It is evident that high quality welds could be produced using all four laser sources.

Introduction

Stricter environmental laws are leading automobile manufacturers to look for ways of reducing fuel consumption and pollutant emissions of their vehicles. One of the main ways to achieve this is by reducing vehicle weight, as fuel consumption and therefore exhaust emissions are directly related to weight. With this in mind, there is keen interest in the use of laser welding to produce lightweight and economical vehicle structures in aluminium. ^[l-3] Laser welding is a flexible high energy density welding process that produces welds of low distortion and a small heat affected zone through the use of easily automated and controlled equipment. ^{[4, 5]}

Currently there are two laser sources capable of delivering the power densities required to weld aluminium alloys in industrial settings; these are the CO ₂ gas laser and the Nd:YAG solid-state laser. In addition, a new, high powered laser is under development, the CO gas laser. These three types of laser differ greatly in their heat source characteristics as a consequence of different means of laser light generation, beam delivery systems and wavelengths.

The aim of the current work was, therefore, to investigate the viability of using different lasers to weld a range of aluminium alloys. Mechanical properties and hot cracking susceptibilities were evaluated.

Experimental procedures

Materials and welding conditions

Five aluminium alloys of standard commercial composition were welded autogenously using CO, CO ₂ and Nd:YAG lasers, and welds were also made with filler wires using a CO ₂ laser. The alloys and their ultimate tensile strengths are shown in Table 1.

Table 1: Properties of materials used in this study

Alloy	Thickness (mm)	Ultimate tensile strength (MPa)	Welding wire	Wire speed (m/min)
2219 T87	3.2	450	2319	3
5083	1.8	314	5556A	2
6061	1.8	308	4047A	2
7475 T761	1.2	480	5556A	2	Clad with Al
8090	1.8	288	4043	2

For each laser, a similar welding set-up was used. Sheets were guillotined to size (250xl00 mm) and degreased with acetone. Samples to be butt welded were lightly abraded on the edges to be joined. No other cleaning or drying procedures were used. The samples were clamped flat on a jig mounted 3-8° off perpendicular to the laser beam. This prevented the reflected beam from re-entering the optics. A summary of alloys welded and welding speeds is provided in Table 2.

Table 2: Welding speeds (m/min)

Alloy	CO 2		Nd:YAG 2 kW			Nd:YAG 5 kW	CO
	Autogenous	CW with filler wire	100Hz pulsed	500 Hz pulsed	CW	CW	CW
2219	2.5	1.0	0.75	NA	NA	7.0	0.4
5083	6.0	3.5	2.5	2.5	NA	11	1.75
6061	7.5 *	5.5 *	2.5	2.5	NA	10	1.25
7475	11	8.0	4.0	3.0	1.25	16	4.0
8090	12	8.0	2.75	2.75	1.25	14	3.5
*Data from J.Y.Yoon. [12]

For the autogenous CO ₂ welding, a Laser Ecosse AF5 fast axial flow DC excited 5 kW laser was used, focused by a KCl lens of 150 mm focal length. Gas shielding was achieved by co-axial and underbead flow of helium. In the wire feed welding system, a Laser Ecosse AF8 8 kW laser was used, with reflective optics. Wire speeds and alloy/wire combinations are shown in Table 1.

A 2 kW Lumonics Multiwave ^TM was used for the medium powered Nd:YAG work. The laser beam was sent down a 1 mm diameter fibre optic. A coverplate was used to prevent weld splatter and fumes from reaching the lens. This coverplate had only 93% transmittance so the laser source power was increased to compensate for this loss. Gas shielding was again achieved by co-axial and underbead flow of helium. For the pulsed Nd:YAG laser work, square pulses were used, with a duty cycle of 40% giving a 5 kW peak from a 2 kW mean power output laser.

The 5 kW continuous Nd:YAG work was carried out at Lumonics Ltd. in Rugby, England using an experimental Multiwave ^TM laser, with a fibre optic delivery system.

The CO laser used at TWI was a converted 5 kW CO ₂ Laser Ecosse laser fitted with chilling and drying systems. It was run in a continuous mode, producing a 2.0 kW beam focused using parabolic mirrors. The beam path was purged with dry air and beam path losses were approximately 15% resulting in 1.7 kW power being delivered to the workpiece.

Lasers were characterised by measurements of spot size and delivered power. The spot sizes for the gas lasers were measured using a Prometec Laserscope UFF1OO rotating needle beam; for the Nd:YAG laser, spot sizes were calculated from fibre diameter. Delivered power measurements were made with a Joule stick power meter. A summary of the results is presented in Table 3.

Table 3: Laser parameters

	CO 2		Nd:YAG 2 kW	Nd:YAG 5 kW	CO
	Autogenous	Filler wire	Pulsed	CW	CW
Power at workpiece (kW)	5.0	5.0 *	2 average 5 peak	4.8	1.7
Spot diameter (mm)	0.26	0.26	0.4	0.4	0.31
Focusing method	150 mm lens	150 mm	lens	lens	mirrors
Irradiance W mm -2	9.4x10 4	9.4x10 4	4.0x10 4	3.8x10 4	2.2x10 4
*6.0 kW was used for welding the thick 2219 material, giving 11.3x10 4 W/mm -2

Bead-on-plate melt runs were carried out to optimise welding parameters. The power delivered to the workpiece was held constant at 5 kW where possible. Weld speed and focus height were adjusted to give the highest quality full penetration welds at the fastest possible welding speeds. For the 2219 wire feed welding, acceptable welds could not be made using 5 kW laser power. The power delivered to the workpiece was increased to 6 kW resulting in an improvement in the quality of welds.

Cracking evaluation

Melt runs, using run-in and run-off tabs, were carried out on tapered specimens to investigate their cracking response. After welding, the samples were cleaned, degreased and then crack lengths determined using a dye penetrant method. The cracking susceptibility was expressed as percentage crack length, i.e. mm of cracked length compared with a sample length of 200 mm.

Examination of welds

Transverse cross-sections were taken from the welds, mounted and polished. For the optical microscopy, samples were etched with Kellers reagent to reveal the second phase regions.

Tensile samples were machined with weld beads at the centre of the tensile sample gauge length, normal to the axis of tensile loading. The weld beads were not machined off before testing. The tensile tests were carried out on a Schenck Trebel electric screw tensile machine, at a cross-head rate of 1 mm min ^-1.

Results and discussion

Metallography

Several of the alloy/laser combinations showed the characteristic 'wine glass' shape often found in laser welding. An example is shown in Figure 1, a Nd:YAG laser weld in 2219 material. Two types of heating take place during laser welding. One is the absorption of laser light from the keyhole, which releases heat throughout the thickness of the material and tends to create a parallel sided weld. The second heat source is the plasma above the weld, which absorbs and re-emits laser energy to create a hemispherical weld centred near the surface of the sheet. The observed weld shapes are affected by the balance between these two heat sources. ^[6]

In Figure 1, the rolled microstructure of the parent plate can be seen as can the very fine microstructure of the weld bead. Small round pores are present and are more common in the lower part of the weld. This autogenous weld shows only a small amount of undercutting due to gap filling and some losses through evaporation and splatter.

Figure 2 shows a plan view of the edge of a weld in 8090 made with the continuous wave Nd:YAG laser. The weld metal, to the top of the image, shows the very fine growth structure in comparison, to the base metal. The growth type in these welds was mainly cellular-dendritic, with dendrite arm spacings, (where visible) of 5 µm or less.

A thin white layer is evident at the fusion line. In this region, the temperature gradient is high. As the laser approaches and passes, the parent material melts back and then begins to solidify, giving very low growth rates at the very edge of the weld. The growth rate then increases slowly, depending upon the shape of the back of the weld pool. Thus the temperature gradient/growth rate ratio is very large and this causes the initial stages of growth to be planar. The rejection of solute that occurs as this solidification starts results in an initial layer of weld metal that is free from second phase particles; it is this layer that is visible at the fusion line.

In the base metal, in the lower half of Figure 2, clear evidence of constitutional liquation can be seen. The dark lines along the grain boundaries are areas where melting has occurred. These are only present close to the fusion line where temperatures approach the bulk solidus and are above the local solidus. This evidence of constitutional liquation was seen in all alloys, but most clearly in the 7475 and 8090 alloys. In these parent materials, the composition is not homogeneous on a micron scale, with variations in composition within primary α grains and regions at grain boundaries where larger amounts of alloying elements are present, and so lower the local melting point. ^[7]

 

The behaviour of the liquated regions depends upon the thermal cycle and upon diffision of the solute materials into the grains. The narrow width of the films (<1 µm in this study) means that the diffusion distances involved are short and it has been suggested that diffusion has a controlling effect upon the behaviour and persistence of these films. ^[8]

These liquid films can act as initiation points for cracking. While no liquation cracking was observed in the current study, if the hot cracking was reduced, for example by the use of appropriate filler wires, then liquation cracking may become a difficulty in these alloys.

Hot crack tests

The five different alloys were welded under seven differing laser conditions and the cracking results obtained are presented in Table 4. Note that the crack test used is very severe; no cracking was observed in butt welds made for tensile testing. The measured crack lengths were highly variable and this variation limits, to some extent, the deductions that can be made from the data. The standard deviation within each group of nominally identical samples was on average 15 mm. This variation might come from differences in sample geometry (the stress field around each weld being a function of the geometry of the material around the weld) or from the positioning of the laser beam path which may have not always been exactly central to the sample.

Table 4: Cracking (% crack length) by alloy and welding method.

Alloy	CO 2		Nd:YAG 2 kW			Nd:YAG 5 kW	CO
	CW	CW with filler wire	100 Hz pulsed	500 Hz pulsed	CW	CW	CW
2219	43	86	24	NA	NA	50	NA
5083	18	71	2	0	NA	60	N0
6061	65 *	20-90 *	96	41	NA	84	50
7475	58	93	70	98	98	100	6
8090	7	91	93	83	95	74	21
*Result from previous work by J.Y.Yoon. [12]

Even so, significant differences were found within the data. For example, the 500 Hz Nd:YAG samples in the 6061 and 8090 materials cracked significantly less than the 100 Hz samples while the situation was reversed in the 7475 material. As pulse rate affects energy deposition and thus cooling rate, it is seems probable that cooling rate affects cracking in a manner that is different for the various alloys.

Similarly, significant differences were seen between cracking in pulsed Nd:YAG laser welds and in continuous high powered Nd:YAG laser welds. Several studies, reported in a recent literature review, suggest that while pulsed Nd:YAG laser welds are extremely susceptible to cracking, this is much less common in continuous wave Nd:YAG laser welds. ^[9] The results in the current study do not support this suggestion. Comparing the 5 kW continuous Nd:YAG laser with the 2 kW 100 Hz pulsed Nd:YAG laser, both delivered equal powers to the weld (the 2 kW laser was run on a 40% duty cycle to give pulses of 5 kW). However, for the 2219, 5083 and 7475 alloys, the continuous wave Nd:YAG laser welds cracked significantly more than their counterparts welded with pulsed Nd:YAG lasers. For the 8090 alloy, the cracking of the continuous wave laser welds was also higher than for the pulsed laser welds, but not significantly. Only in the 6061 alloy did the continuous wave laser welds have significantly less cracking than in the pulsed laser welds. However, in the work by Cieslak and Fuerschbach, quoted in the above literature review, three alloys were studied including 6061. ^[10] They found that, for conduction mode, partially penetrating welds, all pulsed laser welds showed cracks while no cracks were seen in welds made with continuous laser sources.

Markedly more powerful lasers (5 kW versus 600 W) were used in the current work and only keyhole mode fully penetrating welds were made. Thus comparisons between the current and previous work may not be valid. Nonetheless, it is interesting to note that of the five alloys in this study only 6061 showed lower cracking when welded with a continuous laser source rather than with a pulsed laser.

While the above results refer to the comparison of pulsed and continuous lasers of equal incident power, further work was carried out to compare lasers of equal average power, namely 2 kW pulsed against 2 kW continuous wave. However, the 2 kW YAG laser was limited in its ability to weld ~2 mm thick material in CW mode. Thus only two of the alloys in this study, the thin (1.2 mm) 7475 and the easily weldable 8090 aluminium-lithium, could be used to produce directly comparable results between pulsed and CW laser welds. Both alloys produced high levels of cracking, at the limit of the cracking susceptibility resolved by the test. Thus the only conclusion that can be drawn about the effect of continuous versus pulsed laser sources upon cracking in this case is that the crack responses were similar and that the continuous laser welded samples did not have greatly reduced cracking susceptibilities.

In four cases out of five, the 5 kW continuous wave Nd:YAG laser produced significantly more cracked welds than the 5 kW continuous wave CO ₂ laser. Only in the 2219 alloy was this not significant, although the results still show higher cracking for the Nd:YAG. The Nd:YAG laser typically welded 30-50% faster than the CO ₂ laser, despite its lower intensity (focused spot diameter of 0.4 mm rather than 0.26 mm), suggesting that the lower wavelength of the Nd:YAG laser increased the absorption of the laser beam by the workpiece. This has been confirmed in recent work. Thus the heat input from the Nd:YAG laser was higher than that from the CO ₂ . It has been stated that higher heat inputs give more cracking ^[11] ; this was observed in this case.

In general, the CO laser gave significantly lower cracking than the majority of the other lasers for each of the alloys.

It is evident from the results from all laser/alloy combinations that is not a simple relationship between laser parameters, material parameters and cracking response.

Mechanical properties

Ultimate tensile strengths and joint efficiencies are shown in Table 5, from which it is seen that the welds were, with a few exceptions, generally of high quality. To detemine joint efficiencies, seven to nine samples of the base materials were tested. The results were consistent, producing 95% confidence limits generally of ±5 MPa. The strengths of the welds were more widely distributed. At least five samples were taken from each weld run, (at least seven for the 5083 welds), producing 95% confidence limits of ±l5 MPa.

Table 5 As welded ultimate tensile strengths and joint efficiencies (%)

Alloys	Autogeneous CO 2 welds		Pulsed 2 kW Nd:YAG welds		CO 2 + filler welds		CO welds
	MPa	Joint Efficiency	MPa	Joint Efficiency	MPa	Joint Efficiency	MPa	Joint Efficiency
5083	283	90	264	84	297	95	250	80
6061	189 *	58 *	250 *	68 *	200 *	961 *	232	864
7475	233	49	126	26	324	68	301	63
8090	242	84	236	82	208	72	221	77
2219			263	58 (81 )	270	59 (95 )	278	61
* Figures from previous work by J.Y.Yoon, [12] Joint efficiencies after T6 heat treatment

In the welds that gave the highest joint efficiencies, the filler wire welds in 5083, three of the eight tensile specimens fractured in the parent metal, not in or near the weld bead. Failure modes varied with alloy. For example, the 6061 welds all failed in the weld bead, with the fracture running from the fusion line at the top of the bead to the weld centreline at the underside of the bead. Rounded porosity was seen on many of the fracture surfaces.

The high weld strengths are probably a result of the fine microstructures developed during the rapid solidification found during laser welding. As can be seen from Figure 2, the weld microstructure is significantly finer than that of the parent metal. For the non-heat treatable alloy in the study, 5083, this effect was enough to produce very high joint efficiencies. However, for the heat treatable alloys, the joint efficiencies are lower as the weld metal has not received an optimal heat treatment. Clearly, the strength of such weld metal can be increased through the use of an appropriate heat treatment. To test this, welds in the 2219 alloy were heat treated to the T6 condition (solution treated, quenched then peak aged), along with a sample of base metal for comparison. For pulsed Nd:YAG laser welds, joint efficiencies were raised from 58% to 81%, and for filler wire welds made with a CO ₂ laser joint efficiencies were increased from 59% to 95%.

Clearly, laser welding need not have a detrimental effect upon material properties.

Conclusions

High quality welds could be made in a number of aluminium alloys using a variety of laser sources and processes. Joint efficiencies of 95% could be reached when welding a non-heat treatable alloy. After heat treatment, joint efficiencies could also reach 95% for welds in heat treatable alloys.

Cracking results have been found that extend current knowledge in an interesting way. For four of the aluminium alloys in this study, the pulsed Nd:YAG welds showed less cracking than those made with the continuous Nd:YAG laser. However, the current work confirmed statements ^[10] that for 6061 alloy only, the use of continuous Nd:YAG lasers reduces cracking, in comparison with pulsed Nd:YAG lasers. Moreover, there was not a consistent difference in cracking behaviour in the five aluminium alloys when using the CO ₂ laser compared with the welds made using the pulsed Nd:YAG laser; the continuous Nd:YAG laser produced more cracking than the CO ₂ laser. Clearly the interaction of laser source and alloy parameters in determining cracking susceptibilities is complex.

Note that less (or no) cracking would be expected in many real applications as no cracking was seen when making the butt welds for tensile testing; the constraints introduced by the crack test are particularly severe.

The two new types of lasers used here, CO and high powered continuous wave Nd:YAG lasers, could both be used to produce high quality welds. While the CO laser used was of lower power than others in this study, preventing a direct comparison, there was a tendency to lower cracking for welds produced with this laser compared with welds using other lasers. In comparison, the continuous wave Nd:YAG laser produced more highly cracked welds.

Acknowledgements

The authors wish to acknowledge both TWI and Lumonics Ltd for their generous provision of both lasers and materials, and would in particular like to thank the technical staff at TWI whose skill and patience made this study possible. This study was funded by the EPSRC. The authors would also like to thank Professor Alan Windle for provision of laboratory facilities in Cambridge.

References

1.	Metzbower, E.A.	'Laser beam welding - thermal profiles and HAZ hardness'. in Recent Trends in Welding Science and Technology. 1989. Gadinburg, Tennessee: ASM International.
2.	Peritz, A.	'Can the tree-huggers' car come of age?, in Professional Engineering'. 1995. p. p24.
3.	Kazmier, R.J.	'Aluminium takes to the road. Advanced Materials and Processes', 1994. 141 (1): p. 42-44.
4.	Irving, B.	'Will the YAG laser revolutionise automotive manufacturing?' Welding Journal, 1993. 72(7): p. 52-56.
5.	Jones, I.A. & S.T. Riches	'Laser welding aluminium alloys'. TWI Bulletin, 1992. 4(July/Aug.): p. 76-78.
6.	Akhter, R., et al.	'A method for calculating the fused zone profile of laser keyhole welds'. Journal of Physics D, 1989. 22(l): p. 23-28.
7.	Grong, Ø.	'Metallurgical Modelling of Welding'. Materials Modelling Series, ed. H.K.D.H. Bhadeshia. 1994: The Institute of Materials. pp. 541-546.
8.	Radhakrishnan, B. and R.G. Thompson	'Model for the formation and solidification of grain boundary liquid in the heat affected zone (HAS) of welds' ' Metallurgical Transactions A, 1992. 23A(June): p. 1783-1799.
9.	Çam, G., J.F. dos Santos, and M. Koçak	'Laser and electron beam welding of Al-alloys: Literature review.' 1997: GKSS-Foschungszentram.
10.	Cieslak, M.J. and P.W. Fuerschbach	'On the weldability, composition and hardness of pulsed and continuous Nd:YAG laser welds in aluminium alloys 6061, 5456 and 5086'. Metallurgical Transactions B, 1988. 19B(4): p. 319-329.
11.	Ogawa, T.	Prevention of weld hot cracking. Welding International, 1991.5(12):p.931-935.
12.	Yoon, J.W.	Laser welding of aluminium alloys, PhD thesis, Department of Materials Science and Metallurgy. 1996, University of Cambridge.