Visit http://www.esab.com/ for more information about our products.
First published: Svetsaren 2/2000
Keywords: aluminium, AA 5083, AA 6082, Friction Stir, FSW, microstructure, mechanical properties, fatigue, Volvo
Summary: A report on microstructural observations of joints welded using Friction Stir Welding.
Nowadays, aluminium alloys are used in many applications in which the combination of high strength and low weight is attractive. Shipbuilding is one area in which the low weight can be of significant value. In fact, the first aluminium boat was built in 1891 and the first welded aluminium ship in 1953 .
The two most frequently-used aluminium alloys for shipbuilding are AA 5083 (AlMg4.5Mn) for plates and AA 6082 (AlSi1Mg) for extrusions. The main alloying element in the 5000 series is magnesium. A magnesium content of around 5% provides good strength and high corrosion resistance in sea water. The 6000 series is mainly alloyed with magnesium and silicon, which results in a hardenable alloy. Al is also of interest in many other applications, such as the topside structures of offshore platforms, railway wagons and in the brewing industry.
MIG welding is a flexible and productive method and is therefore widely used for welding aluminium alloys in shipbuilding. However, two disadvantages with MIG welding are the deformation of the base material and a decrease in the strength of the heat affected zone. Other fusion welding techniques like TIG and plasma welding are also widely used. However, these methods have the same weakness as MIG welding. An alternative to other fusion welding methods is the recently introduced Friction Stir Welding technique (FSW).
The components that are going to be joined are placed on a backing plate and clamped using a powerful fixture. A rotating tool, consisting of a specially profiled pin with a shoulder, is forced down into the material until the shoulder meets the surface of the material (see Fig 1). The material is thereby frictionally heated to temperatures at which it is easily plasticised. As the tool is moved forwards, material is forced to flow from the leading face of the pin to the trailing one.
The technique was developed at TWI at the beginning of the 1990s  and processing technologies have been further developed by ESAB AB . The first installation has been used successfully for more than one year for joining long plates and panels, up to 6 m by 16 m, mainly in AA 5083 and AA 6082 aluminium . Although the practical application of the FSW technique has been successful, there is still a lack of design data and understanding of the failure mechanisms.
Furthermore, as has been pointed out in a recent summary, much of the microstructural knowledge of friction stir welds is at an embryonic stage .
The aim of the present paper is to report on microstructural observations and provide information about the mechanical properties of joints welded using FSW.
2. Experimental set-up
2.1 Welding and materials
Friction stir welds were produced in AA 5083 and AA 6082 aluminium using different combinations of plate thickness and welding speed.
Alloy AA 5083 is a non-heat-treatable Al-Mg alloy (4.6%Mg, 0.6%Mn, 0.3%Si) with good corrosion resistance, which is commonly used in seawater applications. AA 6082 aluminium is alloyed with Mg and Si (0.7%Mg, 0.5%Mn, 0.9%Si) and age hardens by the formation of Mg2Si precipitates.
2.2 Microstructural studies
An in-depth microstructural investigation was carried out on the 5 mm AA 6082, welded at 75 cm/min, and on the 6 mm AA 5083, welded at 13.2 cm/min. Overviews of the microstructure were obtained using light optical microscopy (LOM), whereas scanning electron microscopy (SEM) was used for the more detailed studies.
2.3 Hardness measurements and mechanical testing
A detailed hardness assessment was made of the two welds subjected to an in-depth microstructural study. The hardness (HV1) was measured horizontally and vertically through the centre of the nugget, on cross-sections.
Transverse tensile testing of the welds was also performed and the location of the fracture relative to the weld centre was examined. Detailed microhardness measurements (HV25g) were made to clarify whether there were any hardness differences between the ring pattern and locations between rings. The accuracy was estimated at two to three units for HV1 and four to six units for HV25g.
Fatigue tests were conducted at room temperature and with a constant amplitude in tensile testing, according to ASTM E466. The frequency was 140 Hz. The load ratio R (=min stress/max stress) was 0.1 and the stress range was 90 to 150 MPa. The fatigue samples were in the as-welded condition without machining the weld top and root faces.
3.1 Macrostructure of the weld zone
The structure of FSW welds contains features that are not found in fusion welds. In a cross-section of a welded joint, the central part has a shape of a “nugget“ (often asymmetrical), in contrast to the well-defined beads of a MIG weld. Larsson et al.  compared FSW welds with MIG welds and noted the presence of the “annual rings“ (or onion ring structure, as defined by Threadgill ) in the FSW weld area which typically consists of concentric ovals.
Immediately adjacent to the nugget is the plastically-deformed and heat-affected so-called “thermomechanically-affected zone“ , which has only been affected by the heat flow.
Macrographs in Figures 1 and 2 show one weld in AA 5083 and one in AA 6082 respectively, in three perpendicular sections. The cross-sections (Figs. 1a and 2a) show that the overall shape of the nugget is very variable, depending on the alloy used and the precise process conditions. However, one common feature is the central “ring structure“ and the more well-defined nugget boundary on the side (to the right in Figs. 1a and 2a) where the tool travel and rotation direction coincide. Appendages to the nugget, extending to the edge of the tool shoulder on the upper surface, can also be seen on this side of the nugget. The complex shape of the nugget and the “ring structure“ is also evident in sections parallel to the top surface (Figs. 1b and 2b) and in longitudinal sections parallel to the original joint face (Figs. 1c and 2c). However, it should be borne in mind that the appearance of the “ring structure“ in these sections is dependent on the precise location of the section.
3.2 Microstructure of base material and weld zone
The microstructure of the 6 mm AA 5083 material was homogeneous with grains elongated in the rolling direction. Typical grain sizes were 30 m along the rolling direction and 20 m across. In AA 6082, the grain size was non-homogeneous (Fig. 3a), due to partial recrystallization. Recrystallized grains were about 10 m in size, whereas the size of non-recrystallized grains typically varied between 50 m and 150 m.
The recrystallization of the nugget zone during friction stir welding effectively wiped out any trace of the previous grain structure. The nugget zone in both materials consisted of fine, equiaxed grains with a grain size of about 10 m (Fig. 3b). The transition between the nugget and the thermomechanically-affected zone was clearly visible in the AA 6082 alloy, as shown in Fig. 3c. This figure also illustrates that the “ring contrast“ is not due to grain size differences. The contrast instead appears to be related to variations in grain orientation and possibly to the degree of relative disorientation between adjacent grains.
The hardness of unaffected base material was approximately 75 HV1 and was practically constant across the weld, both horizontally and vertically, in AA 5083.
The horizontal hardness profile across the weld in AA 6082 had a significantly different appearance. Unaffected base material is harder (about 110 HV1) and there is a decrease in hardness towards the weld, with a minimum of about 60-65 HV1 in the thermomechanically-affected zone. The hardness of the nugget zone itself is typically 70-75 HV1. A slight tendency towards decreasing hardness towards the top surface of the weld was noted at the centre of the weld. The location of isohardness curves, corresponding to 85 HV1, approximately corresponds to the width of the tool shoulder at the top side and becomes narrower towards the root side.
Microhardness measurements (HV25g) across the nugget zone in AA 6082 did not show any systematic variations that could be correlated to the ring pattern (see also ). Nor did measurements in rings and between rings reveal any differences in hardness.
3.4 Tensile properties
There is a considerable difference in the transverse strength of friction stir welds in the two alloys. The transverse strength was between 303 and 344 MPa for AA 5083, while AA 6082 had a transverse strength in the range of 226 to 254 MPa.
An interesting pattern was found when examining the location of the fracture. For welds in AA 5083, the fracture in most cases was close to the centre of the weld and the fracture surface was generally inclined about 45 degrees. The fracture surface was close to the centre of the weld on the root side, but the original joint line never appeared to be the initiation point. In AA 6082, the fracture was with few exceptions close to where the outer edge of the tool shoulder had touched the top side. The fracture surface was inclined, with the fracture surface closer to the weld centre at the root side but still displaced a few mm.
3.5 Fatigue properties
The fatigue properties of FSW weldments are compared with design curves . In most samples, the fracture was initiated in the base material, or in the centre of the weld. In only a few samples, the fractures started in the weld metal/base material transition region. All the tested samples showed very good fatigue behaviour. There are some indications that a lower welding speed results in higher resistance in the weld, although this has to be confirmed by further testing. Welds in base material AA 5083 showed better fatigue properties, compared with welds in AA 6082. The scatter was also larger for welds in AA 6082 than for welds in AA 5083.
Special attention was paid to the annual rings seen in the nugget zone. A similar pattern has previously been observed in aluminium welded using pulsed TIG. This ring pattern is due to varying grain size caused by a periodic change in cooling rate. However, no difference in grain size was noted between rings and areas between rings in friction stir welds (Fig.3c). Nor was any difference in particle distribution detected . The absence of hardness differences between rings and areas between rings also supports the assumption that the ring structure is not associated with precipitation. A likely explanation is therefore that the movement of the rotating, profiled pin-tool through the material results in periodic variations in strain. This produce variations in grain orientation, or in the relative orientation of adjacent grains, resulting in differences in etching response. However, further investigations are needed to verify this hypothesis.
4.3 Mechanical properties
Welded AA 5083 had a tensile strength close to that of material in annealed condition, whereas the weld strength of AA 6082 was between that typical of cold-aged and heat-treated material. From measured strength levels it would therefore be expected that fracture would occur in the “most annealed region“ in AA 5083 welds. This is in good agreement with the fracture usually taking place in the fully-annealed, equiaxed microstructure in the centre of the nugget zone. It was difficult to predict where to expect fracture in AA 6082 from strength comparisons. However, there was a clear correlation between fracture path and the measured line of lowest hardness.
4.4 Fatigue properties
The fatigue resistance of a weld structure is largely controlled by the geometry and quality of the weldments. The fatigue stresses on the structure in a ship come from two main sources, externally applied loads from its progress through the water and initially generated loads from the machinery. Studies of fatigue resistance have generated a large amount of data which are summarised as design curves in the European Recommendations for Aluminium Alloy Structures Fatigue Design (ECCS)  and British Standard 8118 “Structural Use of Aluminium“ . The quality of workmanship can greatly affect the durability of structures and for many years the welding standards have specified quality levels for weldments.
This investigation shows good fatigue properties for FSW samples with values above the design curves. The absence of any weld reinforcement, which results in minimal stress concentrations, is probably a major factor contributing to the good fatigue resistance. There are some differences between the fatigue properties of FSW welds in AA 5083 and AA 6082 base material. Welds in AA 5083 have a higher fatigue range, compared with welds in AA 6082 which showed a somewhat lower strength and large scatter.
5. Conclusions• The microstructure and hardness in rings inside the nugget zone of friction stir welds did not differ from that between rings.
• Most probably the nugget zone ring pattern is an effect of periodic differences in the crystallographic orientation of grains, or varying relative orientation in adjacent grains.
• The hardness only varied a little across the weld in AA 5083, whereas a marked minimum was seen in the thermomechanically-affected zone in AA 6082.
• Fracture in tensile specimens coincided with the line of lowest hardness for AA 6082 and was located in the nugget zone for AA 5083. The fracture path was not related to the ring pattern.
 W.M. Thomas: Int. Patent Application No PCT/GB92/02203, 10 June 1993.
 K.-E. Knipström, and B. Pekkari: Svetsaren Vol. 52, No 1-2, 1997, pp 49-52.
 P. Threadgill: TWI Bulletin, March/April, 1997, pp 30-33.
 H. Larsson, L.-E. Svensson, and L. Karlsson: Proc. Welding and Joining Science and Technology, Madrid, Spain, ASM, 10-12 March 1997.
 J. Karlsson, B. Karlsson, H. Larsson, L. Karlsson, and L.-E. Svensson: Proc. INALCO 98, 7th Int. Conf. On Joints in Aluminium, Cambridge, UK, 15-17 April 1998.
 L. Karlsson, L.-E. Svensson and H. Larsson: Proc. 5th Int. Conf. on trends in welding research, Pine Mountains, GA, USA, 1-5 June 1998.
 Å. Andersson, A. Norlin, and J. Backlund: Proc. International Engineering Conference “Advanced Technologies & Processes“ Stuttgart, Germany, 30 Sept-2 Oct 1997.
 European convention for structural steelwork. European Recommendations for Aluminium Alloy Structures Fatigue Design. ECCS 68, 1992.
 British Standard BS8118, part 1 - Structural Use of Aluminium.