The effect of Ar and He shielding gas on fibre laser weld shape and microstructure in AA 2024-T3

Joseph Ahn1*, Enguang He2, Li Chen2, John Dear1 and Catrin Davies1

1 Department of Mechanical Engineering, Imperial College London, South Kensington Campus, London, UK SW7 2AZ

2 Science and Technology on Power Beam Lab, Beijing Aeronautical Manufacturing Technology Research Institute, China

ABSTRACT

The effect of using and shielding gas on weld quality, defect formation and microstructure of laser welded aluminium alloy 2024-T3 was investigated. Full penetration autogenous welds were made at a constant laser power of 4.9 kW using a continuous-wave (CW) fibre laser at travel speeds of 3.0-5.0 m/min and focal positions of +4 to -4 mm. To investigate this effect, a comparison was made between Ar and He by examining the weld quality in terms of the face and root weld width, the weld width ratio; and the presence of defects including undercut, underfill, reinforcement, porosity and crack. Optical metallography, energy-dispersive X-ray spectroscopy and micro-hardness indentation testing on weld cross-sections were used to identify how the chemical and physical properties of the shielding gases and the characteristics of the fibre laser affect the overall weld geometry. Based on the results, it was believed that relatively small influence of ionisation on fibre laser induced plume enhanced the welding process stability and lowered the threshold power density for keyhole formation. Both Ar and He shielding gases could therefore, be used effectively to produce good quality welds. However, at the lowest speed and also at the maximum focal position, higher ionisation potential and of helium resulted in an excessive weld width when He was used even though, the overall weld quality was better than that with Ar.

KEYWORDS

Defects; Laser Welding; Laser Processing; Aluminium; Microstructure; Shielding Gas; Argon; Helium; Fibre Laser

*corresponding author: [email protected]; Tel: 020 7594 7035

1 1 INTRODUCTION

Aluminium alloy 2024-T3 is a popular material in the aircraft industry due to its light weight, moderate cost and good mechanical properties including specific stiffness, strength and ductility, and good fracture toughness. It is primarily used in fatigue critical structures such as fuselage skin, frames and bulkheads which are under predominantly tension loading conditions. The 2024- T3 alloy with high damage tolerance, has a combination of high fracture toughness, low cycle fatigue strength and resistance to fatigue crack growth and corrosion, making it suitable for fuselage structures, where good static strength, fatigue and fracture resistance are required [1].

Although initially attempts were made to weld AA 2024-T3 aircraft structures, it was determined unweldable by conventional fusion welding processes due to problems related to liquation and solidification cracking and porosity. As a result, it had very limited use for fusion welding applications especially in any stress environment. Fibre laser welds have become attractive alternatives to existing CO2 and Nd:YAG laser welds due to their higher laser efficiency, superior beam quality, lower maintenance, reduced cooling requirements, smaller footprint and more compact design [2]. While extensive research has been conducted on fibre laser welding of fusion weldable 5xxx, 6xxx and 7xxx series aluminium alloys over the last 20 years, little has been reported on the of AA 2024-T3. The possibility of laser welding AA 2024-T3 would lead to significant cost savings through weight reduction, improved fuel efficiency, lower emissions and manufacturing time compared to riveted structures as well as those welded using the 6000 series alloys currently in use which exhibit lower strength and damage tolerance compared to AA 2024- T3 [3].

Some of the earlier research conducted on laser welding of AA 2024-T3 include partial penetration Nd:YAG laser welding of 13 mm thick AA 2024-T3 in conduction mode by Bardin et al. [4], autogenous welding of 3 mm thick AA 2024 sheets also using Nd:YAG laser by Hu and Richardson [5], bead on plate welding of 2 mm 2024-T3 in conduction mode using diode laser by Sánchez-Amaya et al. [6] and Yb:YAG disc laser welding of 3.2 mm thick AA 2024 by Alfieri et al. [7] and Caiazzo et al. [8]. In the majority of cases, severe keyhole instability was observed which resulted in poor weld seam quality, porosity formation, cracking and incomplete penetration.

There is very limited research addressing the feasibility of fibre laser welding AA 2024-T3. Janasekaran et al. [9] and Enz et al. [10] studied dissimilar T-joint fibre laser welding of AA 2024 and AA 7075 and showed that single-sided T-joints produced using fibre laser had significantly less macro pores and micro pores than Nd:YAG laser welded AA 6082-AA 5082 and CO2 laser welded AA 2198-AA 2196 dissimilar joints. Recently, Ahn et al. [11,12] and Chen et al. [13] investigated the effect of varying feed rate on weld pool chemical composition and the resulting hot crack sensitivity and porosity formation of fibre laser welded 3 mm thick AA 2024- T3. It was found that micro hot cracks and porosities were reduced by diluting the weld pool with

2 less than 0.6% silicon content so that the fraction of Mg2Si, the solidification temperature and total shrinkage during freezing all decreased.

As fibre laser welding of AA 2024-T3 is relatively untested and unproven, investigations still need to be conducted to study the properties and performance of fibre laser welded AA 2024-T3, optimise the welding procedure and processing parameters to consistently produce high quality welds with no welding defects and with good mechanical properties [11].

Welding in atmospheric conditions can introduce contamination to the weld and lead to reduced effective laser power, keyhole instability and undesired levels of defects. Helium (He) and argon (Ar) inert gases are often used to protect the weld pool from the atmosphere, where Ar is usually the preferred choice because of its lower cost. The composition of the shielding gas affects the heat distribution in the weld, and therefore, controls the weld shape and the welding speed. The increase in welding speed can be economically substantial. In general, the relatively narrow cross section of an Ar shielded weld has a higher potential for gas entrapment and porosity than of He. The extra heat potential of the He can reduce gas entrapment and porosity levels by broadening the weld fusion and penetration [14]. Therefore, the extra cost of the He may be offset by improved quality.

Shielding gas also serves to suppress and blow away laser induced plasma and vapour which partially absorbs, scatters and attenuates the laser energy [15]. Part of the refracted laser beam due to the interaction with vapour plume or plasma prevents the full power density in the incident laser beam from reaching the workpiece and thus, influences the keyhole geometry. Formation of plasma above the keyhole in some aspect, can be considered beneficial as it assists in coupling the beam energy to the weld pool. Heating of the workpiece occurs by significant absorption of beam energy from the keyhole and from the vapour plume above the weld, which releases the absorbed laser energy into the weld near the surface to create a wider face width [16]. The observed weld shapes can therefore, be controlled by the balance between these mechanisms.

The characteristics of the laser induced plasma is related to the wavelength of the laser. Welding aluminium alloys using CO2 laser with a larger wavelength of 10.6 μm induces strongly ionised plasma with temperatures over 16000 K [15] above the keyhole and partially absorb the laser radiation. On the other hand, near infrared shorter wavelength lasers such as Nd:YAG, disc and fibre lasers (1.07 μm) show lower tendency to form plasma but rather forms less ionised plume of metal vapour and shielding gas within the range of 3000-5000 K as illustrated in Figure 1.

3

Figure 1 Plume and plasma behaviour for different types of laser and shielding gas [14] The inverse Bremsstrahlung absorption of the laser beam in the vapour plume is proportional to the of the laser wavelength so the absorption coefficient of fibre laser is almost 100 times less than CO2 laser [17]. Therefore, the vapour plume is either non-ionised or in extreme cases, only weakly ionised in these lasers, whereas, the vapour ionises to form plasma in CO2 laser. The plume is mainly composed of vaporised metal atoms and metal ions that are ejected from the keyhole but not ionised shielding gases. Due to the apparent absence of significant plasma with fibre laser, the plasma shielding effect is small so beam attenuation and scattering, and the widening of the beam intensity distribution is reduced.

Uspenskiy et al. [18] conducted spectral analysis of vapour plumed formed during 10kW high power fibre laser welding 6 mm Ti-VT-23 titanium alloy and detected around 3 cm high vapour plume on the surface of the workpiece. The vapour plume was only weakly ionised and so the absorbed radiation was negligible of less than 1. Kawahito et al. [14] found that the plume induced when fibre laser welding at 10 kW was not significant enough to influence penetration depth. Gao et al. [19] on the other hand, studied the effect of laser power on the characteristics of fibre laser induced plume via emission spectroscopic analysis and found that a strong plasma shielding effect dominated by inverse bremsstrahlung absorption appeared when a laser power greater than 5 kW was used. The relatively small influence of ionisation on fibre laser induced plume was expected to enhance the welding process stability and lower the threshold power density for keyhole formation because of the enhanced Fresnel absorption and reduced beam attenuation and scattering. Katayama et al. [20] investigated the behaviour of laser induced plume or plasma and compared their interaction between CO2 and fibre laser for welding austenitic stainless steel and aluminium alloy. They found that shielding gas had a greater effect on plasma formation in the case of CO2 lasers, whereas, the shielding gas effect was less with Nd:YAG and fibre laser.

In this investigation, a series of laser welding experiments was performed to evaluate the influence of Ar and He on the weldability and morphology of high power, continuous-wave Yb- fibre laser welds. The first set compared the difference between using Ar and He shielding gases as a function of welding speed and the second set as a function of focal position. The resulting welds were characterised using optical metallography, energy-dispersive X-ray spectroscopy (EDX) and micro-hardness indentation.

4 2 MATERIAL AND METHODS

Heat treatable aluminium alloy (Al-Cu-Mg) alloy 2024 sheets in the T3 temper condition (i.e. solution heat-treated, cold worked and naturally aged) of 3 mm thickness were used. The chemical compositions of AA 2024-T3 are listed in Table 1

Table 1 Chemical composition of AA 2024-T3 (Wt. %) Material Al Cu Mg Mn Cr Si AA 2024-T3 92.1 5.9 1.0 0.6 0.1 0.3

Beat on plate welding was performed using a 5 kW continuous wave (CW) ytterbium fibre laser system YLS-5000 from IPG Photonics operating in TEM01* mode at 1070 nm wavelength. The beam had a beam quality factor (M2) of 7.3, divergence half angle of the focused beam of 12.5 mrad and Rayleigh length of 3.1 mm. The focal length of focusing lens was 300 mm, and the diameter of the focusing lens was 50 mm. The focal length of the collimator lens was 100 mm and the diameter of the collimator lens was 50 mm. The beam diameter at focus was 630 μm and a beam parameter product (BPP) of less than 2.5 mm-mrad was formed. A schematic of the welding setup is shown in Figure 2.

Top and side shielding gas supply

Fibre laser supply

Filler wire Supply

Clamping Root Workpiece bars shielding gas supply Figure 2 Schematic of the experimental set-up for fibre laser welding AA 2024

Industrial grade Ar and He with 99.999% purity were used. The shielding gas was supplied to protect both top and underside of the weld. The coaxial shielding gas was delivered via the weld nozzle (either 100% Ar or 100% He) to protect molten pool and the back protecting shield gas (100% Ar) was supplied via the shielding gas path in the copper insert and purged to remove from the vicinity of the weld joint, contaminants and protect the back weld. Both top and bottom surfaces of each specimen were brushed with a stainless steel wire brush and then cleaned using an industrial grade unbuffered 99.9% pure acetone before welding. The flow rate for back shielding was 15 L/min and 20 L/min for coaxial and side-jet shielding as shown in Table 2 and Figure 2. He and Ar have different chemical and physical properties such as density, ionisation potential and thermal conductivity which affect their suitability as shielding gases as shown in Figure 3. Ar gas has an ionisation potential of 15.8 eV, a thermal conductivity of 0.016 W/mK and a density of 1.661 kg/m3. He gas on the other hand, has a higher ionisation potential of 24.6 eV and thermal conductivity of 0.142 W/mK, but a lower density of 0.1664 kg/m3 [21].

5 Table 2 Physical properties of Ar and He shielding gases [22] Density at Density at First ionization Thermal Threshold for laser Gas RT 1600°C potential conductivity breakdown (g/L) (g/L) (kJ/mole) (W/m-K) (GW/cm2) Ar 1.78 0.283 1520 0.016 113 He 0.179 0.028 2450 0.138 -

5 0.5

4 0.4

) 3 3 0.3

2 He Density 0.2

Ar Density Density (kg/m Density He Therm Cond

1 Ar Therm cond 0.1 Thermal conductivity (W/mK) conductivity Thermal

0 0.0 0 500 1000 1500 Temperature (K)

Figure 3 Temperature dependent density and thermal conductivity of Ar and He shielding gases

The microstructural constituents of the weld were revealed by using Keller’s reagent, a mixture of 95% distilled water, 2.5% HNO3, 1.5% HCl and 1.0% HF, immersed for 30 seconds fresh. Microscopic inspection of etched transverse weld cross-sections was done using Zeiss Axio A1 optical microscope.

Experiments were conducted in accordance with the requirements specified in BS EN ISO 15614- 11 and BS EN ISO 13919-2. These standards provide guidance on levels of imperfections in electron and laser beam welded joints in metallic materials. The relevant examination and tests for welds in accordance with acceptance level D. Visual and metallographic examination of welded specimens were done.

Since a minimum of one section extracted from the weld was required for metallographic examination, it was necessary to check the conformity of the weld seams as shown in Figure 5 and Figure 8. A single weld cross-section per welding parameter set could not guarantee the uniformity of weld seam geometry over the whole weld seam. In order to obtain representative trends in variations of the weld seam geometry for varying welding parameters, the conformity of weld seam geometry had to be checked.

The possible methods of checking the conformity of weld seam geometry over the whole weld include extracting sections parallel to the welding direction or making a large number of samples with the same parameters. However, since a large number of parameter variations were investigated in this study, it was not practical to follow these methods which consume a lot of time, effort and material costs. Instead, the shape and the quality of the top and bottom weld seam surfaces were used as an indicator for the conformity of the weld seam geometry. Random inspections of weld seams showed a positive relationship between the quality of the top and bottom weld seam surfaces and the uniformity of the weld geometry. Three cross-sections were

6 extracted; one at the centre, one near the weld start and the other close to the weld end positions. The shape of the weld seams remained fairly constant and only small variations were observed on the top and bottom weld seam surfaces On the other hand, for welding parameter sets which produced poor weld seam quality, the source of weld imperfections was readily identifiable from the top and bottom weld seam surface due to the fact that the welded sheets were thin and in most cases fully penetrated. Therefore, the shape of the top and bottom weld seam surfaces were used to evaluate the conformity of the whole weld, making it possible to assume that the chosen weld cross-section is representative for the whole weld.

The main criteria assessed per AWS D17.1, BS EN ISO 13919-2 and BS EN 4678 were the top and bottom weld widths, the ratio of root to face width, Rw, the depth of undercut and underfill, the size of weld porosity and the height of reinforcement. The Rw was used to evaluate the processing stability of full penetration welding and a value of 0.6 as determined by Chen et al. [23] from Nd:YAG laser welding 3 mm thick 5A90 Al-Li was used.

Table 3 Weld quality assessment criteria for 3 mm thick AA2024-T3 as defined by AWS D17.1, BS EN ISO 13919-2 and BS EN 4678 [24–26] Face Root Porosity Undercut Underfill Reinforcement Standard Level width width (mm) (mm) (mm) (mm) (mm) (mm) AWS D17.1 Class A N/A N/A ≤ 0.99 ≤ 0.05 ≤ 0.13 ≤ 0.99 BS EN ISO 13919-2 stringent B N/A N/A ≤ 0.90 ≤ 0.15 ≤ 0.15 ≤ 0.65 BS EN 4678 AA ≤ 4.00 ≤ 2.50 ≤ 0.90 ≤ 0.15 ≤ 0.30 ≤ 0.55

The influence of shielding gas type on microstructural transformations and variations of local hardness profiles of AA 2024-T3 welds was evaluated by measuring micro-hardness in the direction perpendicular to the transverse weld cross-sections along mid-thickness, with a 1 mm gap between indentations. Micro-hardness measurements were performed using a Zwick Roell Z2.5 (ZHU 0.2) hardness testing machine at a load of 100 g and a dwell period of 15 seconds at a speed of 60 μms-1 to characterise the whole hardness profile across the weld seams up to the base metal. To avoid work hardening contributions from the adjacent indents, indentations were separated by 200 μm and the lines of indentation were separated by 500 μm. Energy dispersive X-ray spectroscopy (EDX) in an environmental scanning electron microscope (SEM), Hitachi S- 3400N VPSEM was used to determine chemical compositions of the specimens at an accelerating voltage of 15 kV, an emission current of 76 μA, a working distance of 6.8 mm, an elevation of 35° and a live time of 50 seconds.

Table 4 summarises the laser weld and beam parameters for welds made at different travel speeds and focal positions relative to the top surface (0 mm) of the specimens. Sufficiently high laser power and low welding speed were used ensure full penetration [11]. The values were based on optimised parameters as identified in the investigation by Ahn et al. [11].

7 Table 4 Fibre laser welding parameters used in this investigation Weld ID Laser power Welding speed Focal position Shielding gas (kW) (m/min) (mm) 1 4.9 3.0 +4 Ar 2 4.9 4.0 +4 Ar 3 4.9 5.0 +4 Ar 4 4.9 3.0 -2 Ar 5 4.9 3.0 -4 Ar 6 4.9 3.0 +2 Ar 7 4.9 3.0 0 Ar 8 4.9 3.0 +4 He 9 4.9 4.0 +4 He 10 4.9 5.0 +4 He 11 4.9 3.0 +2 He 12 4.9 3.0 0 He 13 4.9 3.0 -2 He 14 4.9 3.0 -4 He

8 3 RESULTS AND DISCUSSION

3.1 Microstructure of AA 2024-T3 laser welds The effect of using Ar and He on the weld microstructure was examined. Figure 4 shows the typical microstructures of a good quality fibre laser beam welded joint produced using a laser power of 4.9 kW, a welding speed of 3.0 m/min, a focal position of +4 mm and Ar. The weld exhibited an hourglass shaped transverse cross-section as shown in Figure 4 a) and consisted of fusion zone (FZ), heat affected zone (HAZ) and parent material (PM).

a) b) PM HAZ FZ

c) d) HAZ FZ

Figure 4 Optical images of the AA 2024-T3 microstructure in the a) weld at 50x b) HAZ at 100x, c) HAZ at 500x and d) FZ at 500x magnification welded using 4.9 kW, 3.0 m/min, +4 mm and Ar only

The equilibrium precipitate phases of the PM were dominantly the semi-coherent CuMgAl2, CuAl2 and CuAl2–CuMgAl2. The dark precipitates observed on the left of Figure 4 b) were the intermetallic compounds, varying in size, shape and chemical composition, formed during the solution heat treatment and natural ageing in T3 temper before welding, some of which included

CuMgAl2, CuAl2, Al20Cu2Mn3, Al7Cu2Fe, Al10CuMg and Al3CuFeMn.

The HAZ was the region in between the BM and the FZ on the right of Figure 4 b) and c), which experienced peak temperatures below the melting point of the PM but high enough to exceed the solvus curves, and affect its microstructure. The HAZ adjacent to the FZ showed partially melted zone with dissolved strengthening precipitates, whereas, the region adjacent to the PM showed over-ageing by coarsening of CuMgAl2 and then transformation from semi-coherent to incoherent

9 CuMgAl2 [11]. The width of the HAZ was very narrow with limited metallurgical modifications, less than 100 µm with only few grains across its width, due to the steep thermal gradient, low heat input and fast cooling rate.

The FZ as shown in Figure 4 d), mainly consisted of α -Al dendritic phase with a surrounding eutectic CuMgAl2 inter-dendritic phase. The initial solidification occurred epitaxially near the outer FZ boundary with a fine grain zone of planar front growth from the PM. The planar front solidification switched to dendritic grain growth of elongated columnar dendrites towards the centre due to constitutional super-cooling. The fast welding speed and also high thermal gradients at the FZ boundary led to an elongated weld pool which promoted formation of columnar dendritic structures. A small fraction of equiaxed dendrites were found at the centre of the FZ

3.2 Relative comparison of Ar and He with speed In order to study the effect of shielding gas type on weld quality and microstructure using Ar and He shielding gas, laser welds were made using several welding speed sets that were design to compare Ar and He. In these welds, a laser power of 4.9 kW and focal position of +4 mm were kept constant in order to concentrate only on the effect of changing the shielding gas type. The results are shown in Figure 5 for both Ar and He. Visual examination of the weld seams on the top and bottom surfaces indicated that under identical welding conditions, using He produced more stable and higher quality weld seams than Ar. A noticeable underfill defects highlighted in red circles at 3m/min were observed both on the top and bottom surfaces in Ar, whereas, they were absent in He at the same welding speed. The depth of the underfill at all welding speed was quantified as displayed in Figure 6 b). In addition, greater amount of spatter was observed in Ar compared to He especially at 3 and 4 m/min. However, spatter was not assessed according to the weld quality assessment criteria as it does not contribute much towards the weld properties. No other visible surface defects were observed at 4 and 5 m/min in both Ar and He.

v=3 m/min Underfill

v=4 m/min Ar, Top v=5 m/min

v=3 m/min

v=4 m/min He, Top v=5 m/min

v=3 m/min Underfill v=4 m/min Ar, Bottom v=5 m/min

v=3 m/min v=4 m/min He, Bottom v=5 m/min Figure 5 Top and bottom surface appearances of fully penetrated bead on plate autogenous fibre laser weld beads in 3 mm thick AA 2024-T3 at 4.9 kW, +4 mm at a range of welding speeds

10 The difference in top weld widths with welding speed among those welded in Ar and those welded in He as shown in Figure 6 a) was negligible at the lower welding speed of 3 m/min, by less than around 0.02 mm, whereas, a larger difference of around 0.2 mm was observed for the bottom weld widths at 3 m/min. The top weld width was initially wider in Ar at 3 m/min than in He. However, it decreased at a faster rate with increasing welding speed in Ar than in He, which indicated that the extra heat potential and higher thermal conductivity of He increased the energy input to the workpiece at higher welding speeds of 4 and 5 m/min more effectively than Ar. The bottom weld width at 3 m/min in He exceeded the maximum tolerable root width of 2.5 mm specified in BS EN 4678 as highlighted in Figure 6 a), whereas, it was within the safe limit in Ar. No maximum top (face) width was specified in any of the standards so there was no problem associated with the type of shielding gas in terms of top weld width.

While the heat input, as indicated by the bottom weld width, was excessive at 3 m/min in He. , the measured bottom widths at 4 and 5 m/min were acceptable. Therefore, as long as the welding speed is high enough, above 3 m/min in this case, both Ar and He can be used. Of course, at lower welding speeds, using Ar regarding the bottom weld width. Wide weld widths which are comparable to the thickness of the welded sheet is not desired as it can have detrimental effects on the weld joint strength.

The weld width ratio (Rw) as shown in Figure 6 b) hardly changed with increasing welding speed in He, with values in the range between 0.94 and 0.96. On the other hand, the Rw displayed a clear trend where it increased with increasing welding speed in Ar. Nevertheless, the Rw values were well all above the minimum limit of 0.6 required for keyhole stability both in Ar and He.

Figure 6 b) also shows that the specimens welded in Ar were more greatly affected by variations in welding speed than those in He. The size of reinforcement measured was acceptable by a wide margin at all welding speeds in both Ar and He compared to the most stringent limit of 0.55 mm as listed in Table 3. No definite trend was observed in terms of reinforcement but in most cases decreased with increasing welding speed.

Undercut defect was not present at all welding speeds in Ar as well as in He except one specimen welded at 4 m/min. An undercut almost as deep as 0.15 mm was measured at this welding speed, which was considerable larger than the maximum limit of 0.05 specified in AWS D17.1 and close to the 0.15 limit specified in the other two standards as listed in Table 3. Underfill defect in general was larger in He than in Ar as expected due to the larger heat input supplied to the workpiece. However, at 3 m/min, the depth of underfill in Ar of 0.12 mm was similar to that in He of 0.13 mm, being close to the maximum limit of 0.13 mm specified in AWS D17.1. In fact, the underfill was not considerably greater in He because the extra energy from He was used to produce much wider excessive root penetration instead. At 4 and 5 m/min, the undercut and underfill defects were no longer identified in Ar, but a rather large undercut at 4 m/min and underfill at 5 m/min were found in He. In this case, more energy was transferred into the weld to form these defects.

11 In addition, another reason why the underfill defects were formed was because fast cooling rate at these welding speeds failed to provide enough time for the weld to solidify with a sufficient liquid metal flow. Underfill can act as a crack initiation point and reduce the cross-sectional thickness, which has a negative impact on the weld mechanical properties. Still, according to the requirements of the two less stringent standards, the higher thermal conductivity of He is useful for increasing the energy input to the workpiece and so a faster welding speed can be used in He to produce acceptable welds and make up for its higher cost.

a)

3.0 3.0 Greater than the limiting root width for 2.8 2.8 helium shielding at 3 m/min.

2.6 2.6 max root width BS EN 4678 max root width BS EN 4678 2.4 2.4

2.2 2.2

2.0 2.0

Weld width (mm) width Weld Weld width (mm) width Weld

1.8 1.8 Top (He) Bot (He) 1.6 1.6 Top (Ar) Bot (Ar) 1.4 1.4 2 3 4 5 6 2 3 4 5 6 Speed (m/min) Speed (m/min) b) 1.2 0.40 Rw (He) Undercut (He) Rw (Ar) Underfill (He) 0.35 Reinforcement (He) 1.1 Undercut (Ar) Underfill (Ar) Reinforcement (Ar) 0.30 max underfill limit BS EN 4678------1.0

0.25 0.9 0.20 0.8

0.15 max underfill limit BS EN ISO 13919-2, max undercut limit BS EN ISO 13919-2 and 4678------Defect size (mm) size Defect max underfill limit AWS D17.1------0.7 Weld face to root width ratio width root to face Weld 0.10

min Rw 0.6 0.05 max undercut limit AWS D17.1------

0.5 0.00 2 3 4 5 6 3 4 5 Speed (m/min) Speed (m/min) Figure 6 a) Relationship between weld width and welding speed at a laser power of 4.9 kW, no defocus and with either argon or helium shielding gas, and b) the resultant Rw, undercut, underfill and reinforcement defects Metallographic sections were made through the welds as shown in Figure 7, which compares the welds made under Ar to those of He. The overall weld shape as shown in Figure 7 was an hourglass both for Ar and He. The weld profile improved with He because the plume effect was reduced owing to He’s higher ionisation potential. The fusion boundaries were more parallel through thickness with less reinforcement compared to Ar. The weld cross-section at 5 m/min with Ar was narrower and so a micro pore with 0.064 mm diameter was found. The size of the pore, however, was much smaller than the maximum tolerable pore size of 0.90 mm. On the other hand, no porosity was observed in the microstructure of the same welds made in He. The higher heat produced in the He shielded welds led to hotter weld pool and longer solidification time which

12 permitted gas bubbles to escape and as a result, reduced the risk of gas entrapment. It was concluded that the risk of porosity formation was reduced in He.

Porosity 1.0 mm 1.0 mm 1.0 mm

2 mm 2 mm 2 mm

3.0 m/min 4.0 m/min 5.0 m/min Argon

1.0 mm 1.0 mm

2 mm 2 mm 2 mm 3.0 m/min 4.0 m/min 5.0 m/min Helium Figure 7 Transverse sections of welds and weld top surface bead profiles produced with different welding speeds at 4.9 kW, +4 mm, in Ar and He

13 3.3 Relative comparison of Ar and He with focal position The effect of Ar and He on weld quality and microstructure was studied at a range of focal positions. A constant laser power of 4.9 kW and welding speed of 3 m/min were used. The weld surface quality is shown in Figure 8 for both Ar and He. Visual examination of the weld seams on the top and bottom surfaces indicated that under identical welding conditions, using He produced more stable, smoother and higher quality weld seams than Ar. Underfill defects visible by eyes highlighted in red circles were identified at -2 mm on the top surface and at +2 mm on the bottom surfaces all in Ar, whereas, no underfill defect was found in He. The depth of the underfill was quantified as displayed in Figure 9 b). An overlap defect was found at -4 mm in Ar on the top surface at the weld toe, caused by metal flowing on to the surface of the parent material without completely fusing to it, most likely as a result of the large weld pool size. However, since the length of the overlap is short, its presence can be accepted. No visible pores or cracks were found.

f=-4 mm Overlap

f=-2 mm Underfill Ar, Top f=0 mm f=+2 mm f=+4 mm

f=-4 mm

f=-2 mm He, Top

f=0 mm f=+2 mm f=+4 mm

f=-4 mm Ar, Bottom f=-2 mm f=0 mm

f=+2 mm Underfill f=+4 mm

f=-4 mm He, Bottom f=-2 mm

f=0 mm f=+2 mm f=+4 mm Figure 8 Top and bottom surface appearances of fully penetrated bead on plate autogenous fibre laser weld beads in 3 mm thick AA 2024-T3 at 4.9 kW, 3 m/min and various focal positions for Ar and He As shown in Figure 9 a), the difference in the weld widths with changing focal position between Ar and He was negligible except at a focal position of +4 mm, where the bottom weld width in He of around 2.6 mm exceeded the maximum acceptable root width of 2.5 mm according to BS EN 4678. The rate of change of top weld width with focal position was similar for both Ar and He. In contrast, the rate of change of bottom weld width was greater towards the positive and negative

14 ends for He than Ar. The higher heat input associated with He was found to be more effective at higher positive focal positions because the effect of reduced specific energy and lower average temperature in the weld pool due to positive defocusing was compensated. On the other hand, this effect was less significant at negative focal positions because the threshold power density for keyhole formation was lower than at positive focal positions.

Almost no difference was found between Ar and He in terms of the Rw. The Rw values were all above 0.6 as shown in Figure 9 b) indicating that keyhole stability was sufficiently high at all focal positions both in Ar and He. The size of reinforcement was below 0.55 mm at all focal positions both in Ar and He. Even the greatest height of reinforcement measured at 0 mm in Ar of around 0.32 mm was safely below the maximum limit.

a) 3.0 3.0 Greater than the limiting root width for

2.8 2.8 helium shielding at +4 mm

2.6 2.6 max root width BS EN 4678

max root width BS EN 4678 2.4 2.4

Weld width (mm) Weld 2.2

Weld width (mm) Weld 2.2

Top (He) Bottom (He) 2.0 2.0 Top (Ar) Bottom (Ar)

1.8 1.8 -5 -4 -3 -2 -1 0 1 2 3 4 5 -5 -4 -3 -2 -1 0 1 2 3 4 5 Focal position (mm) Focal position (mm) b) 1.2 0.35 Rw (He) Undercut (He) Underfill (He) Rw (Ar) Reinforcement (He) 1.1 0.30 max underfill limit BS EN 4678------Undercut (Ar) Underfill (Ar) Reinforcement (Ar)

1.0 0.25

0.9 0.20

0.8 0.15 max underfill limit BS EN ISO 13919-2, max undercut limit BS EN ISO 13919-2 and 4678------

max underfill limit AWS D17.1------Defect size (mm) size Defect

0.7 0.10 Weld face to root width ratio width root to face Weld

min Rw 0.6 0.05 max undercut limit AWS D17.1------

0.5 0.00 -6 -4 -2 0 2 4 6 -4 -2 0 2 4 Focal position (mm) Focal position (mm) Figure 9 a) Relationship between weld width and focal position at a laser power of 4.9 kW, a welding speed of 3 m/min and with either argon or helium shielding gas, and b) the resultant weld width ratio, undercut and reinforcement Underfill defect was observed only at 0 mm both in He and Ar, which was close to the most stringent 0.13 mm limit specified in AWS D17.1, with a value of around 0.13 mm in He and 0.12 mm in Ar. Defocusing the beam in either positive or negative position removed the underfill defect completely by reducing the incident laser power density. While shifting the focal position from the 0 mm position eliminated the underfill defects, it caused undercut defects to form instead. Large undercuts were produced in He at all four focal positions of -4, -2, +2 and +4 mm as shown in

15 Figure 9 b). Similarly, undercut defects were found at -4, +2 and +4 mm in Ar but not at -2 mm. The depths of undercut in He were all greater than the most stringent limit of 0.05 mm and even greater than the less stringent limit of 0.15 mm limit at -2 and +2 mm. The undercut defect in Ar at -4 mm was around 0.04 mm which was below 0.05 mm limit. However, those at +2 and +4 mm were greater than the 0.05 mm limit and also considerably greater than the 0.15 mm limit at +4 mm, with a value of around 0.21 mm. At +4 mm, the undercut defect in He of around 0.06 mm was much smaller than that in Ar. This meant that the higher heat input associated with He was adequate as a result of reduced specific energy and lower average temperature in the weld pool due to positive defocusing as mentioned previously.

The welds were free of crack and porosity at all focal positions in both Ar and He as shown in Figure 10. The weld shape was an hourglass with only a small difference between Ar and He. However, the overall weld shape obtained using Ar was more uniform except at +4 mm, where large undercuts were found both on the top and the bottom surfaces. Meanwhile, the weld quality of the weld at +4 mm in He was superior to that in Ar as well as at lower focal positions. Therefore, it was highly probable that the use of He was more suitable at higher focal positions where the effective power density of the incident laser beam was lower.

1.0 mm 1.0 mm 1.0 mm 1.0 mm

2 mm 2 mm 2 mm 2 mm 2 mm

-4 mm -2 mm 0 mm +2 mm +4 mm Argon

1.0 mm 1.0 mm 1.0 mm 1.0 mm

2 mm 2 mm 2 mm 2 mm 2 mm -4 mm -2 mm 0 mm +2 mm +4 mm Helium Figure 10 Transverse sections of welds and weld top surface bead profiles produced with different focal positions at a laser power of 4.9 kW, at a welding speed of 3 m/min and with either argon or helium shielding gas

16 3.4 Micro-hardness of AA 2024-T3 laser welds in Ar and He Figure 11 shows that all welds were under-matched with the lowest hardness in the FZ. The hardness in the BM was the highest as expected. It was also found that the hardness in the HAZ was greater than in the FZ but lower than in the BM. Micro-hardness increased as a function of the distance from the weld centre in which the FZ had a hardness of around 85-95 HV, the HAZ hardness in the range of 100-120 HV and the BM hardness in the range of 130-140 HV. The HAZ adjacent to the FZ showed hardness values close to that in the FZ whereas, the HAZ adjacent to the BM showed a hardness close to that in the BM. Since the extent of the FZ and the HAZ was very small, the resulting hardness gradient was very steep. The hardness distribution showed small variations across the FZ. The differences in the mid-thickness width of the welds made in Ar and He were small but the measured widths were slightly wider in He due to the reasons explained previously related to He.

145 f = -4 mm 145 f = -2 mm

135 135

125 125

115 115

105 105 Vicker's hardness (HV0.1) 95 Vicker's hardness (HV0.1) 95

He 85 He 85 Ar Ar 75 75 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 Distance from weld centre (mm) Distance from weld centre (mm)

145 f = 0 mm 145 f = +2 mm

135 135

125 125

115 115

105 105

95 Vicker's hardness (HV0.1) 95 Vicker's hardness (HV0.1) He 85 85 He Ar Ar 75 75 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 Distance from weld centre (mm) Distance from weld centre (mm)

145 f = +4 mm

135

125

115

105

95 Vicker's hardness (HV0.1)

85 He Ar 75 -3 -2 -1 0 1 2 3 Distance from weld centre (mm) Figure 11 Micro-indentation hardness distributions of fibre laser welded AA 2024-T3 welds as a function of focal position for Ar and He Hardness was controlled by solidification microstructure and it was a function of heat input, peak temperatures and cooling rate. Since the extent of softening in the FZ and the HAZ was dependent on the heat input, the welds in He showed slightly lower or similar hardness to those in Ar because higher heat input was achieved in He. As changing the focal position has only small

17 influence on the net heat input, the lower hardness trend exhibited by the welds made in He compared to those made in Ar can be explained by the difference between the two types of shielding gases, including their physical properties and interaction with the laser beam and the workpiece. The reduction in hardness with He was attributed to the increased the dendrite cell and grain sizes and the coarse secondary phase in the FZ and the HAZ due to the relatively slower cooling rate. In comparison, the welds made in Ar which experienced lower net heat input and consequently, a faster cooling rate and shorter duration at high temperatures arising from a smaller weld pool volume, showed higher hardness values in the FZ and the HAZ because of the finer grain size.

Softening in the FZ was caused by microstructural changes as a result of temperatures reaching above melting point in the FZ and the associated rapid heating and cooling rates during welding. The heating action of the laser led to segregation of the strengthening elements, magnesium and copper, and their hybrids (intermetallic compounds), formation and growth of non-strengthening coarse precipitates, dissolution of strengthening precipitates and uniform re-distribution of precipitating elements during heating which then froze due to fast cooling rates as shown by the weight percentage of Cu and Mg measured using EDX in Figure 12 [11,27]. In addition, softening can also be attributed to violent vaporization of low boiling point Mg by affecting the weld pool chemistry as indicated by the lower Cu and Mg contents in the weld compared to the parent material. The hardening effect from T3 temper was therefore, removed and the mechanical properties of the weld degraded during welding.

a) b) 90 P=4.9 kW, V=3.0 m/min, f=+4 mm, Ar 80 P=4.9 kW, V=4.0 m/min, f=+4 mm, Ar 70 P=4.9 kW, V=5.0 m/min, f=+4 mm, Ar P=4.9 kW, V=3.0 m/min, f=+4 mm, He 60 P=4.9 kW, V=4.0 m/min, f=+4 mm, He P=4.9 kW, V=5.0 m/min, f=+4 mm, He 50 PM

40 Weight (%) Weight 30

20 Mg content (%) 0.63% 0.65% 0.70% 0.76% 0.80% 0.87% 1.26% 10

0 Al Cu Mg O Figure 12 Weight percentage (%) of main alloying elements in the weld compared to the parent metal obtained using energy dispersive X-ray spectroscopy (EDX)

The weld metal chemical composition at different welding speeds in either Ar or He as shown in Figure 12 obtained using EDX, clearly indicated a drop in magnesium content with decreasing welding speed. A slower welding speed led to an increased heat input and resulted in more material being melted due to more exposure of the laser beam on the workpiece. The vaporisation of the alloying element magnesium during welding, which has a relatively low boiling point of

18 1091⁰ C compared to aluminium of 2470⁰ C and copper of 2562⁰ C, was therefore, controlled by the welding speed. Increased vaporisation of Mg had the effect of increasing the tendency to form macro sized welding defects by affecting the keyhole pressure, so as previously shown Figure 6 a), the keyhole stability increased with increasing welding speed. Under the same welding speeds, the measured magnesium content was consistently higher in He than in Ar which meant that the higher ionisation potential and thermal conductivity of He had a positive effect of reducing the extent of low boiling point alloying element segregation. As the critical level of Mg for crack sensitivity is around 1.5%, the lower Mg content produced in Ar of around 0.6% compared to around 0.8% in He, decreased the crack sensitivity by shifting away from the crack sensitive ranges and also the risk of inducing coarse Mg2Si precipitates [11].

19 4 CONCLUSIONS

The lower ionisation potential and thermal conductivity of Ar was not a critical issue when fibre laser welding AA 2024-T3 due the short wavelength of fibre laser that lowered its sensitivity to ionisation even at 4.9 kW. Fully penetrated welds were produced in AA 2024-T3, which had a considerable amount of volatile alloying elements such as Mg and Zn, even in Ar. In addition, the keyhole stability was high for welds made in both Ar and He with values above 0.6.Therefore, both Ar and He shielding gases were found to be effective.

The weld quality improved with He at lower welding speeds because the plume effect was reduced due to the higher ionisation potential of He, which minimised porosity formation. The higher heat produced in the He shielded welds led to hotter weld pool and longer solidification time which permitted gas bubbles to escape and as a result, reduced the risk of gas entrapment.

Increased energy input when using He allowed faster welding speeds to be used and thus, justifying its higher cost. On the other hand, excessive root width at lower welding speed in He was not observed in Ar.

The use of He was found to be more suitable at higher focal positions where the effective power density of the incident laser beam is lower. The effect of reduced specific energy and lower average temperature in the weld pool due to positive defocusing was compensated.

Specimens welded in He showed slightly lower or similar hardness to those in Ar because higher heat input was achieved in He.

Loss of volatile alloying element magnesium was greater in Ar than in He. However, as the magnesium content was shifted further away from the critical level for crack sensitivity in Ar than in He, using Ar effectively reduced the crack sensitivity compared to when using He under the same welding conditions.

ACKNOWLEDGEMENT

The support from the Aviation Industry Corporation of China (AVIC) and Beijing Aeronautical Manufacturing Technology Research Institute (BAMTRI) for this funded research is appreciated. The research was performed at the AVIC Centre for Structural Design and Manufacture at Imperial College London. Dr C M Davies acknowledges the support of EPSRC under grant number EP/I004351/1.

20 REFERENCES

1. Ion JC. of wrought aluminium alloys. Science and Technology of Welding and Joining. IOM Communications for the Institute of Materials, 2000;5(5): 265– 276.

2. Ahn J, Chen L, Davies CM, Dear JP. Parametric optimisation and microstructural analysis on high power Yb-fibre laser welding of Ti–6Al–4V. Optics and Lasers in Engineering. 2016;86: 156–171.

3. Chaturvedi MC. New welding techniques for aerospace engineering. Welding and joining of aerospace materials. 1st ed. Cambridge: Woodhead Publishing Limited; 2012. p. 3–9.

4. Bardin F, Mcbride R, Moore A, Morgan S, Williams S, Jones JDC, et al. Real time temperature measurement for process monitoring of laser conduction welding. Proceedings of the 23rd International Congress on Applications of Lasers and Electro- Optics. San Francisco, CA: Laser Institute of America; 2004. p. 1–24.

5. Hu B, Richardson IM, Introduction I. Autogenous laser keyhole welding of aluminum alloy 2024. 2005;17(2): 70–80.

6. Sanchez-amaya JM, Boukha Z, Amaya-vazques MR, Botana FJ. Weldability of Aluminum Alloys with High-Power Diode Laser. Welding Research. 2012;91: 155–161.

7. Alfieri V, Caiazzo F, Cardaropoli F, Sergi V. Issues in Disk-laser Key-hole Welding of 2024 Aluminum Alloy. Frontiers in Aerospace Engineering. 2012;1(1): 36–44.

8. Caiazzo F, Alfieri V, Cardaropoli F, Sergi V. Bead characterization of disk-laser butt welding of thin AA 2024 sheets. In: Beyer E, Morris T (eds.) Proceedings of SPIE Photonics West 2012-LASE. San Francisco, CA: International Society for Optics and Photonics; 2012. p. 82390E–82390E–12.

9. Janasekaran S, Jamaludin MF, Muhamad MR, Yusof F, Abdul Shukor MH. Autogenous double-sided T-joint welding on aluminum alloys using low power fiber laser. The International Journal of Advanced Manufacturing Technology. Springer London, 2016;90(9–12): 3497–3505.

10. Enz J, Khomenko V, Riekehr S, Ventzke V, Huber N, Kashaev N. Single-sided laser beam welding of a dissimilar AA2024–AA7050 T-joint. Materials & Design. 2015;76: 110–116.

11. Ahn J, Chen L, He E, Davies CM, Dear JP. Effect of filler metal feed rate and composition on microstructure and mechanical properties of fibre laser welded AA 2024-T3. Journal of Manufacturing Processes. 2017;25: 26–36.

12. Ahn J, Chen L, Davies CM, Dear JP. Digital Image Correlation for Determination of Local Constitutive Properties of Fibre Laser Welding Joints in AA2024-T3. Proceedings of the

21 16th International Conference on Experimental Mechanics. Cambridge: University of Cambridge; 2014. p. 1–2.

13. Chen L, He E, Ahn J, Dear J. Parametric optimization and joint heterogeneity characterization of fiber laser welding of AA2024-T3. Proceedings of the 67th Annual Assembly of the International Institute of Welding. Seoul: International Institute of Welding; 2014. p. 1–9.

14. Katayama S, Kawahito Y, Mizutani M. Elucidation of laser welding phenomena and factors affecting weld penetration and welding defects. Physics Procedia. Elsevier, 2010;5: 9–17.

15. Xiao R, Zhang X. Problems and issues in laser beam welding of aluminum–lithium alloys. Journal of Manufacturing Processes. 2014;16(2): 166–175.

16. Akhter R, Davis M, Dowden J, Kapadia P, Ley M, Steen WM. A method for calculating the fused zone profile of laser keyhole welds. Journal of Physics D: Applied Physics. 1989;22(1): 23–28.

17. Kim KR, Farson DF. CO2 laser–plume interaction in materials processing. Journal of Applied Physics. 2001;89(1): 681.

18. Uspenskiy SA, Petrovskiy VN, Bykovskiy DP, Mironov VD, Prokopova NM, Tret’yakov E V. Spectral diagnostics of a vapor-plasma plume produced during welding titanium with a high-power ytterbium fiber laser. Journal of Physics: Conference Series. 2015;594(1): 12033.

19. Gao M, Chen C, Hu M, Guo L, Wang Z, Zeng X. Characteristics of plasma plume in fiber laser welding of aluminum alloy. Applied Surface Science. 2015;326: 181–186.

20. Katayama S, Kawahito Y, Mizutani M. Elucidation of laser welding phenomena and factors affecting weld penetration and welding defects. Physics Procedia. 2010;5: 9–17.

21. Xiao R, Zuo T, Volz R. Experimental research on the influence of laser-induced plasma on the beam focusing during high-power CO2 laser materials processing. Chinese Journal of Lasers B. 1998;B7(6): 550–554.

22. Elmer J, Vaja J, Carlton H, Pong R. The Effect of Ar and N-2 Shielding Gas on Laser Weld Porosity in Steel, Stainless Steels, and Nickel. Welding Journal. 2015;94(10): 313s–325s.

23. Chen L, Shuili G, Yang J. Study on Full Penetration Stability of Light Alloys Sheet Laser Welding. In: Hinduja S, Li L (eds.) Proceedings of the 37th International MATADOR Conference. Manchester: University of Manchester; 2012. p. 315–318.

24. AWS Standard D17.1:2001, ‘Specification for Fusion Welding for Aerospace Applications’, AWS, 2001.

25. ISO Standard 13919-2, 2011, ‘Welding — Electron and laser-beam welded joints —

22 Guidance on quality levels for imperfections — Part 2: Aluminium and its weldable alloys’, ISO, 2011, www.iso.org.

26. BS Standard EN 4678:2011, ‘Aerospace series. Weldments and brazements for aerospace structures. Joints of metallic materials by laser beam welding. Quality of weldments’, BSI, 2011.

27. Cao X, Wallace W, Immarigeon J-PP, Poon C. Research and Progress in Laser Welding of Wrought Aluminum Alloys. II. Metallurgical Microstructures, Defects, and Mechanical Properties. Materials and Manufacturing Processes. 2003;18(1): 23–49.

23