Assessment of microstructure and mechanical properties of friction stir welded AA2014-O and AA2014-T6 sheets

In this study, friction stir welding of AA2014-O and AA2014-T6 aluminum alloy was performed at various welding speeds to evaluate the influence of temper conditions of base metal (BM) on the properties of the welded joints. The results showed strong influence of BM temper conditions on the microstructural morphologies and mechanical behavior of the welded joints. In the 2014-O joints, different zones of weld joint were diffused into each other and there was no clear interface between them. In 2014-T6 joints, there was a distinct demarcation between the nugget zone (NZ), thermomechanically affected zone TMAZ, heat-affected zone (HAZ), and BM. The welded joints in 2014-O temper condition showed increase in hardness in the vicinity of weld center due to grain refinement whereas, in 2014-T6, softening occurred in the same region by the dissolution of strengthening precipitates. The mechanical properties of 2014-O joints were equivalent to the base metal showing a 100% weld efficiency with fracture located in the base metal, whereas 2014-T6 welds exhibited about 70% weld efficiency with fracture located at the NZ/TMAZ interface. All the samples in mechanical testing fractured at retreating side (RS) which exhibited heterogeneity in the mechanical properties of the welded joints. Scanning electron microscopy fractographic analysis revealed a ductile fracture mode comprising of dimples in both temper conditions. The size and shape of the dimples were strongly dependent on base metal temper condition.


Introduction
The heat treatable aluminum alloys of 2xxx series are the key alloys used in structures where the key design benchmark is high specific strength, good fracture toughness, and damage tolerance such as aerospace, marine, petrochemical, power generation, and transport industries [1][2][3]. The weldability of AA2xxx aluminum alloys, by conventional fusion methods, is poor due to the formation of hot cracks, undesirable microstructure, and porosity. However, newly developed friction stir welding method can join metal alloys that are considered unsuitable for welding via conventional fusion joining methods [4]. The process is being progressively practiced for welding of similar as well as dissimilar metals with ease. This technique also has the benefit of being solid state, which alleviates the need for liquid filler metals that are common with conventional fusion welding techniques [5,6]. Friction stir welding involves a rotating non-consumable tool of harder than base metal material. The tool generates sufficient heat, by friction and plastic strain, to soften the base material without melting. The friction occurs between the tool shoulder and upper surface of the sheets whereas plastic strain takes place by the rotating pin of the tool. These thermo-mechanical conditions result in diverse microstructures in the weld region. The transverse section of the weld joint shows a region of severe deformation and material flow and is known as nugget zone (NZ) [7]. The microstructure of this zone consists of very fine equiaxed grains having high angle grain boundaries. The temperature in the NZ may reach up to 0.95T m [8] (T m is the melting temperature of the material) and thus is high enough to completely or partially dissolve the strengthening precipitates [7,9,10]. The region next to NZ is called thermomechanically affected zone (TMAZ). This zone does not experience the direct action of pin or shoulder. However, it is subjected to severe thermo-mechanical fluctuations due to internal shear stresses. The strengthening precipitates are coarsened in this region due to plastic deformation and high temperature resulting in decrease in hardness [7,11]. The region between the TMAZ and base metal (BM) is known as heat-affected zone (HAZ). Due to the dissolution or coarsening of strengthening precipitates, this region shows a significant decrease in hardness [7] as compared to the BM. Many studies have focused on the mechanical properties of friction stir welded precipitation hardenable aluminum alloys of 2xxx series. Yu et al. [12] observed that tensile and yield strength decreased while ductility was increased in the nugget zone of the friction stir welded 2198-T8 alloy due to softening in the welded region. Meshram et al. [13] studied the effect of tool pin profile and welding speed on the quality of weld joints in AA2014-T6 plates. Aydin et al. [14] observed that welding parameters strongly affect the mechanical and fatigue properties of the friction stir welded AA2014-T6 alloy. The effects of welding parameters on FSW characteristics (such as microstructural morphologies in joints), was studied by Ramanjaneyulu et al. [15] and it was observed that microstructure of friction stir welded AA2014-T6 is governed by the tool pin profile and welding parameters which in turn influences the mechanical properties of the joint. Norman et al. [16] studied the grain structure of friction stir welded joints of 2024-T351 alloy with the help of high-resolution electron backscattered diffractrography (EBSD) technique. They concluded that different local deformation conditions and high strain rates result in different stages of dynamically recrystallized grains structures. Similar observations were made by Jones et al. [17] in friction stir welded 2024-T351 alloy. Zhang et al. [18] found that anisotropy in microstructure of friction stir welded 2024-T3 was responsible for difference in mechanical properties. Material flow behavior in the welds has also been addressed by many researchers such as Huaxia et al. [19] who studied the material flow behavior of friction stir welded 2024-T351 with the help of thin copper strips. They observed that at different welding parameters, the material fragmentation and dispersion are decreased at high heat inputs and vice versa. It was explained in terms of base material softening and shear stresses. Li et al. [20] studied the material flow behavior during refill friction stir spot welding of alclad 2A12-T4 alloy and found that plasticized material flow was due to extrusion and shearing actions. Qin et al. [21] studied experimentally and numerically the material flow behavior in high-speed friction stir welded joints of 2024 sheets. They concluded that at high rotation speeds, flow velocities were high and they decrease with increasing distance from the weld center. Residual stress analysis in the friction stir welded joints has also been a subject of many investigations. Li et al. [22] carried out experimental and numerical investigation of residual stresses in friction stir welded 2024-T4 alloy and observed that due to the tool force, the longitudinal residual tensile stresses became smaller and were non-uniformly distributed at different sides of the weld center. Fratini et al. [23], in their work of effect of longitudinal residual stresses on fatigue crack growth in friction stir welded 2024-T351, concluded that fatigue crack growth rate was controlled by residual stresses outside the weld zone and microstructure and hardness are the controlling factors within the weld zone. Delijaicov et al. [24] showed that tool rotation speed and welding speed were the controlling factors for residual stresses during the friction stir welding of dissimilar butt joint of AA2024-T3 and AA7475-T761. Although there are a few studies concerning the effects of base material temper conditions on the properties of friction stir welding of 2xxx series aluminum alloys, they are mainly focused on more commercial, AA2024 and AA2219, alloys. The research work related to friction stir welding of AA 2014 alloy is mostly focused on T6-tempered condition and the comparative studies of weld characteristics and joint performance in different heat treatment conditions of base metal are not available. In the present investigation, AA2014 aluminum alloy sheets were friction stir welded in two different heat treatment states, i.e., annealed "O" and age-hardened "T6" conditions, to evaluate the influence of different temper conditions of the base metal on microstructural morphologies, strength, and fracture characteristics. A comprehensive assessment of microstructural development, strength, and hardness in the weld zone of the joints is also made between the "O" and "T6" temper conditions.

Materials and experimental method
The material used in this study was commercially available 2.5-mm-thick, rolled sheets of AA2014 alloy. The chemical composition and mechanical properties of the materials used in this study are given in Tables 1 and 2, respectively.
The blanks of 250 × 80 × 2.5 mm dimension were cut from the sheet by a shear cutter. A set of pieces was then subjected to T6 treatment by solutionizing at 510°C followed by warm water quenching. Artificial aging was then performed at 170°C for 16 h in an electric oven. In order to get smooth and parallel faying edges, the sheared edges were milled on a milling machine. The edges were rubbed by emery paper just before welding to remove surface oxides followed by cleaning with a suitable solvent to wipe out dirt, debris, and oil remnants. The friction stir welding was carried out on a computer-controlled vertical milling machine. The tool used was made of hardened and tempered H13 tool steel having an eccentric grooved shoulder of ϕ10mm and a threaded taper pin of ϕ4mm with a length of 2.25mm. The welding parameters used in this study are given in Table 3. The tool rotation speed and tool tilt angle were kept the same for welds and only the tool travel speed was varied. All the welds were in butt joint configuration. The welding, of equal lengths of 200 mm, was done in a single pass, perpendicular to the rolling direction of the sheet along the long edges of the blanks. A physical picture and schematic diagram [25] of the welding setup is shown in Fig. 1a and b. After welding, the samples for metallographic analysis and hardness measurement were extracted from the joints by cutting across the welding direction. Proper care was taken to avoid any artifacts during the cutting process.
For microstructural analyses, the cross-sections of the joints were mounted and then polished according to the standard sample preparation procedure and then etched with Keller's reagent [26] to reveal the microstructure. The samples were then observed under an optical microscope in the bright field mode.
Microhardness profiles of the joints were determined with a Vickers micro-hardness testing machine at 100g load applied for 10 s. The measurements were made at mid-thickness, transverse to the welding direction, keeping a distance of 0.25 mm between the adjacent indents. The measurements were made starting from the base material (BM) to the heataffected zone (HAZ), the thermomechanically affected zone (TMAZ), and the nugget zone (NZ) from retreating side (RS) up to the advancing side (AS).
Tensile test samples, for the determination of mechanical properties, with 25-mm gage length, were cut across the welding direction with the help of EDM-wire cutting machine according to ASTM E08 [27] according to the dimensions shown in Fig. 2. All the samples were uniaxially tested, at a universal tensile testing machine equipped with highresolution extensometers. Tests were performed at room temperature and at a crosshead speed of 1 mm/min. At least three samples were tested for each condition to maintain the reliability of the tensile test results. After the tensile test, fracture surfaces of the samples were secured for SEM fractographic examination.

Microstructural morphologies
The macrographs of 2014-O and 2014-T6 welds, at different welding parameters, are shown in Figs. 3 and 4, respectively. All the macrographs are taken at ×20 magnification. The advancing and retreating sides of the joint are indicated by "AS" and "RS" respectively. The typical four regions of the friction stir welded joint are also labeled as NZ, TMAZ, HAZ, and BM. From the figures, it is clearly observed that the joints in both temper conditions have very different microstructural morphologies. Figure 3 shows 2014-O joints welded at different welding speeds. It is seen that the weld nuggets have partial "onion rings" features in all the three welds. The fraction of these features reduces as the welding speed is increased. On the other hand, in 2014-T6 joints (Fig. 4), the "onion rings" features are present in greater amount over the entire width of the NZ. The presence of "onion rings" like features shows that the joints experienced sufficient plastic flow during welding process and the plasticized material was deposited in multiple layers [28]. It was observed by Sutton et al. that "onion ring" like features are due to the banded microstructure. These bands are formed due to uneven distribution of hard particles [29]. In 2014-O base material, majority of the precipitates exist in stable state and a relatively less fraction of metastable precipitates is present in the material in the form of Al 2 Cu, AlCuMn, and AlCuMg [30]. At low welding speed, the peak temperature during FSW is high enough to dissolve and solutionize some fraction of the metastable particles into the aluminum matrix. However, the large amount of stable precipitates remains unaffected and is accumulated in high-strain regions giving rise to formation of bands of low and high particle density [29]. These bands give rise to partial "onion rings" morphology in the NZ of 2014-O on advancing side. It is worth noting that, as the welding speed in increased, the peak temperature is reduced, and hence, the proportion of "onion rings" features also reduces (Fig. 3c).
For 2014-T6 base material, the majority of the precipitates exist in the metastable state and only a small fraction of stable precipitates is present in the material. During FSW process, an  appreciable amount of these metastable precipitates is dissolved in the aluminum matrix. As discussed above, the stable precipitates remain unaffected and are segregated in high strain regions, resulting in the formation of high and low particle density bands in the NZ. Such bands appear as "onion rings" features in the NZ of 2014-T6 joints.  [33]. This may also be ascribed to the difference in hardness of both the materials. As 2014-O has much lower hardness than 2014-T6 samples, the mechanical agitation and subsequent material flow and temperature rise will be more severe in 2014-O than in 2014-T6 leading to coarsening of recrystallized grains during cooling from weld temperature to room temperature in the former than in the latter. The microstructure in both temper states consists of recrystallized fineequiaxed grains in the NZ. This can be explained in terms of rotation and linear motion of tool during FSW process which causes severe plastic deformation by tool pin and high thermal effects by tool shoulder, resulting in the semi-solid material in this region. This semi-solid material forcefully flows along the tool pin surface of the rotating tool causing severe shear strains. Due to severe plastic strains at elevated temperature, dynamically recrystallized grains are nucleated at the grain boundaries of initially large and elongated grains in the NZ [34,35]. The same trend was found in 2014-O and 2014-T6 samples welded at different welding speeds.
The microstructural variation from NZ to TMAZ is presented in Figs  respectively. In 2014-O joints, the boundary between NZ and TMAZ is not clear on the RS (Fig. 7a) whereas a narrow band of transition of grains from NZ to TMAZ was observed on AS (Fig. 7b). The same trend was also observed in other 2014-O samples which were welded at travel speeds of 200 and 280 mm/min. However, 2014-T6 joints showed a very distinct transition interface between the NZ and TMAZ on RS as well as AS irrespective of the travel speed. The same phenomena have been observed by other workers [8,36,37] and is explained in terms of the difference in plastic flow of material on advancing and retreating sides of a weld. The interface is more distinct on the side having higher shear forces and plastic strains. Since on AS, the directions of tool travel and plastic deformation are in the same direction, the shear forces and plastic strains are higher on this side, resulting in clear interfaces between NZ and TMAZ. The region of TMAZ experiences both high temperature and plastic deformation to lesser extent than that of the NZ. The combined effect of deformation and temperature results in a microstructure which consists of elongated and rotated grains. The recrystallization mainly begins at NZ/TMAZ interface. The grain size of TMAZ is much larger than that of the NZ. Due to the thermal effects, the strengthening precipitates grow and get overaged, thus causing drop in hardness in this region [34].   In both cases, there is no discernable boundary between HAZ and the base metal because the microstructural transition is not significant. HAZ experiences only thermal fluctuations and no plastic deformation takes place in this region. As a result, the grains are not recrystallized and the grain size is similar to or slightly larger than the base metal. Due to the thermal effects, hardening precipitates can dissolve or coarsen resulting in reduction of hardness. The extent of precipitate dissolution and coarsening depends on the base material condition.  [38]. From the hardness profiles in Fig. 8, it can be observed that hardness increases in all 2014-O joints, irrespective of the welding parameters used. Also the increase in hardness is almost the same in all cases. Same observations were also made by [39] in different friction stir welded aluminum alloys. This increase in hardness in "O" temper condition is due to the fact that the grain refinement took place in the weld region especially in the NZ, thus giving rise to hardness in this region. Since the temperature is high enough, there may be some precipitation of strengthening particles during cooling from high temperature to room temperature [40,41]. It was also observed that the rise in hardness did not vary significantly with change in welding speed. It was due to the fact that high temperature and severe plastic strains dissolved the hardening precipitates back into the α-Al matrix of NZ. The samples in 2014-T6 condition showed a typical "W" shaped profile across the welded region (Fig. 11). The partial recovery in hardness of NZ is due to grain refinement and re-precipitation of hardening phases during cooling   from welding temperature to room temperature [40,41]. The loss in hardness was maximum in the TMAZ and HAZ which was due to dissolution and coarsening of hardening precipitates and grain growth [42,43]. The hardness loss was not same for all the samples of 2014-T6 condition at different welding parameters. The samples welded at low travel speed (120 mm/min) showed a maximum loss in hardness by giving a value of minimum hardness of 85 HV, whereas the samples welded at high travel speed (280 mm/min) showed a minimum hardness up to 92 HV. This may be caused by high thermal fluxes at low travel speed resulting in greater volume of precipitate dissolution than at high travel speed where low thermal effects dissolved lesser amount of the hardening precipitates. Same interpretations were made by other researchers in FSW of different aluminum alloys [33,34].

Mechanical properties of the joints
The mechanical properties of the 2014-O and 2014-T6 joints, welded at different welding parameters, are presented graphically in Figs. 12 and 13, respectively, along with the BM properties. The yield strength, ultimate tensile strength, percentage elongation and percentage of joint efficiencies, and the BM properties are also summarized in Table 4 for comparison.
From Fig. 12, it is evident that the yield strength and ultimate tensile strength of 2014-O joints were almost same as that of BM; however, the joints showed about 15% drop in elongation. Thus, the joint efficiency was 100% as shown in Table 4. The decrease in elongation is due to the high hardness in the NZ (Fig. 10). During tensile test, the load concentrates more in the low hardness region of the joint and final fracture takes place in this region [42]. For 2014-O joints, the low hardness region is base metal (see Fig. 10). Therefore, NZ offered greater resistance to plastic deformation during tensile test, thus reducing the overall elongation. Similar observations have been made by other researchers in friction stir welded 6061-O and 7075-O alloys [42,43].
The tensile samples of 2014-T6 condition showed different mechanical behaviors as compared to 2014-O samples. The mechanical properties of the 2014-T6 joints were lower than those of the base metal as illustrated in     Fig. 13. The mechanical properties of the friction stir welded joints are determined by welding defects and the hardness distribution across the weld joint [34]. In a defect-free joint, hardness profile across the weld mainly controls the mechanical properties. It is also known that the friction stir welded joint consists of a non-uniform microstructure with different zones along with their interfaces, all of which have different mechanical properties [8,44]. From Fig. 11, it can be noted that 2014-T6 joints have low hardness regions, such as NZ, TMAZs, and HAZs, as compared to BM. The mechanical properties of these regions are lower than those of the base metal. As discussed earlier, the stress is concentrated in these regions during the tensile test and the joints show lower mechanical properties than the BM. The 2014-T6 joints also showed a remarkable loss in ductility (Table 4). This may be due to the excessive loss of hardness in the weld region. During tensile test, the load is confined to this region and major portion of plastic deformation takes place within weld region while BM remains almost intact due to higher hardness. Similar observations were also made in other friction stir welded aluminum alloys [34,42,43]. The yield strength efficiency (YSE), ultimate tensile strength efficiency (UTSE), and elongation efficiency (EE) for 2014-O and 2014-T6 joints are summarized in Table 4. The YSE and UTSE and EE are the ratios of the average yield strength, ultimate tensile strength, and elongation of the joints, respectively, to those of the BMs [28]. It can be observed that efficiencies of different mechanical properties strongly depend on BM condition. The 2014-O BM has a complete stable condition and the YS and UTS efficiencies of 2014-O joints are almost same as those of BM irrespective of the welding parameter used. However, the EE is lower than BM, being 90% at low welding speed and 85% at high welding speed. In 2014-T6, the joints showed much lower weld efficiencies than the BM. The YSE remained around 50% of BM, whereas UTSE was around 70% of the BM. The EE in heat treated condition reduced significantly, being 50% at low welding speed and 32% at high welding speed. In general, it was observed that the strength and elongation efficiencies are decreased when the stability for precipitation of the BM decreases [34].

Fracture locations and fracture surfaces of the joints
The analysis of fracture locations and fracture surfaces was carried out to understand the mechanical behavior of 2014-O and 2014-T6 welded joints. The fracture location represents the weakest region of the joint. Analysis of fractured tensile samples provides useful information about structure-property relationship [32]. From Table 5, which shows the representative fracture locations of 200 samples, it can be seen that 2014-O and  (Fig. 10), all the stress concentration occurred in BM.
In 2014-T6 samples, the fracture occurred on RS at NZ/ TMAZ interface because it was the weakest region of the weld with minimum hardness (Fig. 11). At the same time, there was drastic variation of microstructure at NZ/TMAZ interface. The TMAZ in these samples has twisted coarse-grained microstructure as compared to NZ and HAZ with partially dissolved and coarsened strengthening particles. This significant difference in the internal structure gave rise to the weakest region in the weld joint. Due to low hardness and distinct microstructure, the stress was concentrated in this region during the tensile test.
SEM fractographic analysis has been carried out to understand the mechanism of fracture in the samples. The SEM fractograph of the fracture surface of the 2014-O tensile sample is shown in Fig. 14. Since all the samples, welded at different welding parameters, fractured from BM, only a representative fractograph of 200 samples is shown here. The fracture surface of 2014-O joints revealed a typical ductile fracture consisting of deep dimples, of varying sizes and shapes, which result from microvoid nucleation and coalescence. The size of the dimples shows that appreciable plastic deformation took place in the BM which led to deep dimples and high ductility. The notable necking in the vicinity of fracture also supports this observation. Within the dimples, different particles were also observed. Hence, it can be stated that samples in "O" condition fractured by the microvoid coalescence at coarse constituent particles [45]. Figure 15 shows the SEM fractographs of fractured samples in 2014-T6 condition at different welding parameters. Since fracture in "T6" condition always occurs in weld zone, all the fractographs show a different morphology than the BM. The fracture surfaces show a fine grain size that becomes finer at high welding speed (Fig. 15c). Although the fracture surfaces represent a ductile fracture morphology, yet the dimple size was much finer and shallower than the 2014-O samples which shows less plastic deformation during loading and a low ductility [34]. This observation is in agreement with the elongation data given in Table 4.

Conclusions
The following conclusions can be drawn from the analysis of results of the friction stir welded 2014 alloy in different heat treatment conditions: & The microstructural morphologies of the friction stir welded joints of AA2014 strongly depend on BM temper conditions. & The hardness of the 2014-O joints increased than that of BM due to grain refinement strengthening, whereas it decreased in 2014-T6 joints due to dissolution of strengthening precipitates. & The mechanical properties of friction stir welded joints were compatible to those of BM in 2014-O temper.  The elongation efficiencies showed that ductility was reduced in all cases; the loss was more prominent in T6 temper condition. & SEM fractographic analysis of fracture surfaces revealed dimples of varying sizes representing ductile mode of failure in both temper conditions. The dimples were representative of ductility of the joint being deep in "O" condition and shallow in "T6" condition.