Effects of ultrasonic assisted friction stir welding on flow behavior, microstructure and mechanical properties of 7N01-T4 aluminum alloy joints

1. Introduction

Friction stir welding (FSW) is a solid state joining technology, the material is plasticially deformed by the rotating tool, and the advance of the rotating tool moves to achieves the joint connection [1]. Compared to the traditional fusion welding technique, the FSW is considered to be an energy efficient and high-efficiency process, particularly suitable for the welding of aluminum alloys [2]. Numerous studies have shown that defect-free aluminum alloy FSW joints have better performance (mechanical properties) than the fusion joints [[3], [4], [5]]. An adequate flow of materials is the most important factor in obtaining defect-free joints with good mechanical properties. Thus, sufficient heat input is essential to ensure the adequate material flow. In particular, the welding of high-strength materials often requires higher heat input to improve metal flow and to reduce joint defects. In general, the tool rotational speed, welding speed and geometric parameters of the tool (pin profile and shoulder diameter) play a crucial role in material flow, which determines the mechanical properties of the joint [[6], [7], [8], [9], [10]].

Ultrasonic vibration has been integrated in various processing technologies, such as welding, drawing, and machining [11]. The results of various studies show that ultrasonic vibration can reduce flow stress, improve surface quality and increase processing efficiency [12,13]. Park et al. [14] and Rostamiyan et al. [15] applied horizontal ultrasonic vibration from the side of the rotating tool through the bearings during the butt welding and spot welding of the 6061 aluminum alloy, respectively. The direction of the ultrasonic vibration was parallel to the welding direction. Their results indicated that the ultrasonic vibration helped to reduce the chance of the formation of welding defects, and improved the lap shear force of spot-welded joints. Some scholars introduced ultrasonic energy to the one side of the workpiece during the friction stir welding of Al/Steel and Al/Mg dissimilar alloys [16,17]. It was shown that the continuous intermetallic compounds at the interface were broken and the mechanical properties of the joint were improved. These experimental results are attributed to the good material flow induced by mechanical vibration. Recently, Wu et al. [[18], [19], [20], [21], [22], [23], [24]] implemented an ultrasonic vibration transmission directly into the localized area of the workpiece near and ahead of the tool. The material flow, microstructure evolution and welding load aluminum alloy joints were systematically studied by numerical and experimental methods. They found that the superimposing ultrasonic vibration can increase the material flow velocity and strain rate, but the thermal effect of the ultrasonic vibration was relatively small, and the ultrasonic effect was most effective at the NZ center. In order to enhance the material flow in the bottom of joint, Liu et al. [25] placed the sonotrode on the bottom surface of the workpiece to assist the friction stir welding of the 2219-T6 aluminum alloy. They found that the ultrasonic vibration caused the hardness distributions in the thickness direction were more homogeneous, and the tensile strength and elongation of the joint were improved.

As mentioned above, the main function of applying ultrasonic vibration in FSW process is to promote the flow of NZ metal without obvious temperature rise. This effect is significant in eliminating welding defects and improving mechanical properties of joints. However, in most ultrasonic vibration assisted FSW studies, the ultrasonic vibration device is separated from the tool and the ultrasonic vibrations are not directly applied to the weld. With the development of related technology, some researchers have integrated ultrasonic transducer, horn and tool into a composite device. The tool vibrates in the axial direction while rotating, so that the ultrasonic vibration directly acts on the weld to reduce the loss of ultrasonic energy [26,27]. This kind of ultrasonic vibration application method has the advantages of simple structure, high utilization rate of energy and good reliability, thus demonstrating industrial potential. However, the investigation on axial ultrasonic vibration assisted FSW is not systematic enough, and the research content is still limited, mainly on weld forming and mechanical properties.

For the weld, the NZ can be divided into four sub-zones, namely the shoulder-driven zone (SDZ), the transition zone (TZ), the pin-driven zone (PDZ) and the swirl zone (SWZ) [[28], [29], [30]]. In addition to the tunneling defect and kiss bonding in the NZ, loose voids are easily formed at the TZ in the upper part of the joint [31]. These welding defects formed in different sub-zones of the NZ are considered to be related to the metal flow behavior. However, the research results of metal flow behavior of the NZ by applying axial ultrasonic vibration assisted FSW have not been reported.

Moreover, the existing investigations on the UAFSW process were mainly focused on 2xxx and 6xxx aluminum alloys. There are less studies addressing the UAFSW of 7xxx aluminum alloys for which the welding is always needed to fabricate the final products. Thus, in the present work, we conducted a thorough microstructural investigation on welds produced by FSW and by an axial ultrasonic vibration assisted friction stir welding device with a cooling system and an ultrasonic output power real-time monitoring system. The experimental material in this paper was a medium strength aluminum alloy 7N01-T4. Due to its better strength and formability, it is often used as roof structure, frame and other important structural parts in rail vehicle carriages. The difference of metal flow behavior in different sub-zones of the NZ of FSW and UAFSW joint was compared by the marker insert technique, and the influence of ultrasonic vibration on the evolution of microstructure characteristics of the NZ along the thickness direction was revealed. Generally, the mechanical properties of the 7xxx aluminum alloy FSW joints change obviously in the long-term natural aging process. Therefore, the mechanical properties of the FSW and the UAFSW joints after post-weld natural aging (PWNA) are also studied, which provides theoretical research data for the engineering application of axial ultrasonic vibration assisted FSW.

2. Experimental

The base material (BM) in this study was a cold rolled 7N01-T4 aluminum alloy plate, its chemical composition is listed in Table 1. Specimens with dimensions of 175 mm × 75 mm × 6 mm (length × width × thickness) were used for welding, the welding direction was parallel to the rolling direction. Fig. 1(a) shows the schematic diagram of UAFSW system. An ultrasonic transducer was intergrated into the rotating tool holder that drove the welding tool to vibrate in the axial direction, which ensured an direct introduction of the the high frequency vibration acts directly onto the weld. The frequency of the ultrasonic vibration was 20 kHz, and the maximum vibration amplitude was 10 μm. The experimental setup of the UAFSW is shown in Fig. 1(b). The diameter of the scroll shoulder was 16 mm, the pin diameter was tapered linearly from 8 mm (root) to 4.8 mm (head) and the pin depth was 5.8 mm. In addition, three flats were machined on the pin surface. For all specimens, the spindle tilt angle of 2.5° and the plunge depth of 0.2 mm were kept constant during the welding. The joints were fabricated at a fixed tool rotational speed of 1200 rpm, and at different welding speeds by conventional FSW and UAFSW. The effects of ultrasonic vibration on the weld formation, axial force, microstructure and mechanical properties of the joints were investigated.

In order to investigate the effect of ultrasonic vibration on the material flow patterns, a 0.2 mm thick pure aluminum foil was used as a marker material and was placed between the plates, as shown in Fig. 2. After the welding was completed, the joints were cut into metallographic samples for the transverse cross-section observation. The macroscopic features were etched with Keller’s reagent (1 ml of hydrofluoric acid, 1.5 ml of hydrochloric acid, 2.5 ml of nitric acid and 95 ml of water) for optical microscopy (OM, OLYMPUS GX71) examinations. The surface topography of the FSW and the UAFSW joints was examined by laser scanning confocal microscopy (LSCM, OLYMPUS 3100). To examine the grain characteristics at different depths in the NZ of the joints, the electron back scattering diffraction (EBSD, TESCAN MAIA3 and Channel 5) measurements was performed with a step size of 0.6 μm. Microhardness measurements were conducted on the middle cross sections of the joints under a load of 100 g and with a dwell time of 15 s. The distance between the adjacent testing points is 1 mm. As the mechanical properties of the 7xxx FSW joints are closely related to the natural aging time [32]. The joints were tested by tensile tests at 120 h, 720 h and 4320 h of PWNA. Tensile specimens with a gauge length of 150 mm were cut perpendicularly to the welding direction, in accordance with the ASTME8-04. The tensile tests were performed on a 100 kN universal test machine with a tensile speed of 5 mm/min at room temperature. The final values for the ultimate tensile/yield strength and elongation are the average from three tested samples from the same joint. After the tensile tests, the fracture surfaces were characterized by scanning electron microscopy (SEM, JOEL 7001 F).

3. Results and discussion

3.1. Effect of ultrasonic vibration on weld formation

The cross sections of the FSW and the UAFSW joints welded under a tool rotational speed of 1200 rpm and different welding speeds are shown in Fig. 3 and Fig. 4, respectively. It is seen that at lower welding speeds, the joints are defect-free (Fig. 3(a)-(c) and Fig. 4(a)-(e)), indicating an adequate metal flow during the welding processes. When the welding speed reached 240 mm/min for the FSW, the tunneling defects can be observed at the advancing side (AS) (indicated by the red circle in Fig. 3(d)), and the high magnification image is shown in Fig. 5(a). Lower welding speed resulted in higher heat input, therefore, adequate material flow results in the formation of a defect-free NZ. In general, when the tool is traversed, the plasticized material in the leading edge of the tool flows via retreating side (RS) to the trailing edge, and the rotating action of the tool brings the plasticized material from the RS to AS [33]. Consequently, with the increase of the welding speed, the peak temperature of the thermal cycle decreases and the cooling rate increases. Low heat input results in a lack of stirring, hence higher flow resistance of the material. The plasticized material cannot be refilled to the advancing side, resulting in the formation of the so-called tunneling defects [34]. However, for the UAFSW, when the welding speed reached 340 mm/min, a tunneling defect is found in the UAFSW joint, that is also located on the AS (indicated by the red circle in Fig. 4(f)). A high magnification image is shown in Fig. 5(b). At the rotational speed of 1200 rpm, the welding speed of FSW and UAFSW joints without defects are 200 mm/min and 300 mm/min, respectively. Consequently, ultrasonic vibration allows a 50% increase in the welding speed. Clearly, the welding processing window of the UAFSW is broader than that of the FSW, indicating the potential to improve productivity.

Many investigations on the effect of the ultrasonic vibration on formability during plastic forming process showed that ultrasonic vibration increases the deformability and formation uniformity, resulting in the improvement of product quality. The stress superposition and acoustic softening effect of ultrasonic vibration are considered to be the major effects in the vibration assisted forming process. That can reduce the metal flow resistance and improve the metal plastic flow capability [35,36]. Stress superposition is usually caused by the elasto-plastic property of the metallic materials. One explanation for the acoustic softening, according to the dislocation theory, is that the preferential absorption of acoustic energy by lattice imperfections, such as dislocations or grain boundaries, which tends to reduce the shear stress and significantly softens metal materials without obvious heating [37]. Thus the increased metal flow in the UAFSW process in the present study should result from the two effects of the ultrasonic vibration. Next, the metal flow behavior, microstructure and mechanical properties of the FSW and UAFSW joints with welding parameter of 1200 rpm and 160 mm/min were further investigated.

Surface topography of the FSW and the UAFSW joints using the parameter of 1200 rpm and 160 mm/min was examined by LSCM, and the results are shown in Fig. 6. Clearly, after the ultrasonic vibration was applied, the color of the surface of the joint is darkened due to the change in the flow behavior of the material. That is, in the process of ultrasonic vibration, the magnitude and direction of the resultant sliding velocity vector between the shoulder and the metal changes [38], resulting in the difference in the roughness of the joint surface, as shown in Fig. 6(a). Comparison of the three-dimensional surface topography of the FSW with that of the UAFSW joints (Fig. 6(c) and Fig. 6(d)) reveals that the semicircular marks of the surface texture topography of the two joints are relatively intact. Clearly, the arc surface morphology of the UAFSW joint is more uniform, and the transition between the wave crest and trough is well defined. The surface profile results can be seen in Fig. 6(b). Ultrasonic vibration improves the surface roughness of the joint, not only the arc outline in unit cycle is smoother, but also the maximum distance between wave crest and trough is reduced. This is due to the axial high-frequency vibration of the tool, which causes the shoulder and the softened layer of metal on the trailing edge of the shoulder to be in a "contact-separated" state. The binding force of the shoulder/softening layer of metal is reduced, and the separation of the shoulder and this part of metal is promoted. The reduction of joint surface roughness is beneficial for improving the corrosion resistance and fatigue performance of the joint.

3.2. Effect of ultrasonic vibration on axial force

In order to avoid the error in axial force measurement caused by the differences between the plates (plate thickness tolerances, surface roughness and clamping state) and clamping state and to clarify the effect of vibration on Z-axial force, the FSW and the UAFSW alternately welding test were performed on the same butt welded aluminum plate at a tool rotating speed of 1200 rpm and a welding speed of 160 mm/min. The curve of Z-axis force of the FSW and the UAFSW experiments versus time are shown in Fig. 7. The A-B stage is called the plunge stage. There are two prominent peaks (point A and point B) in the plunge stage. Point A is a result of the tool pin plunging into the solid surface of the workpiece. During this process, the Z-axis force decreases as the plunging depth increases and the degree of thermal softening of the metal around the pin is increased. After that, the shoulder of the tool plunges 0.2 mm into the workpiece, resulting in the second peak (point B). B-C is the dwell stage during which the tool rotates for 20 s. The Z-axis force then decreases gradually due to the softening of the BM as a result of the higher heat generation. As the tool begins to moves to start the welding, the initial value of the Z-axial force increases slightly (C-D). This phenomenon is attributed to the instantaneous increase in deformation resistance when the tool moves from a softened area to a cold material. In the welding stage (D-H), the Z-axial force remained in a relative steady state. The Z-axis force in the D-E stage (FSW) is maintained at around 7200 N. When the ultrasonic vibration is applied at point E, it can be clearly observed that the Z-axis force is decreased to 6600 N during the E-F stage (UAFSW). When the ultrasonic vibration equipment is turned off at point F, the Z-axis force is increased to 7200 N in the F-G stage (FSW). Ultrasonic vibration is applied again at point G, and the Z-axial force is reduced to 6600 N in the G-H stage (UAFSW), then the welding is finished. H-I is the tool retracting stage in which the Z-axial force gradually decreases as the shoulder and the pin are separated from the workpiece. From the above result, one can find that the ultrasound vibration can result in a decrease of the Z-axis force by about 9 %. This ultrasonic softening leads to a better plasticization and flow of metals, which is the main reason for reducing the Z-axial force in the UAFSW compared to the FSW.

3.3. Characteristics of material flow in the transverse cross-section

The cross-sectional macrographs of the FSW and the UAFSW joints welded at a tool rotating speed of 1200 rpm and welding speed of 160 mm/min are shown in Fig. 8. In the FSW joint, the thickness of the TZ along the thickness of the plate is about 800 μm. However, when the ultrasonic vibration was applied, the thickness of the region was compressed to about 450 μm. Similar experimental phenomenon was also observed in a previous study by Huang et al. [39]. They deduced that the enhanced material flow is due to the increase of the tool rotational speed. The upward and downward forces become stronger, resulting in the compression of the layer. This phenomenon further demonstrates that ultrasonic vibration can significantly improve the flow of the metal, especially the upward and downward movement of the metal in the PDZ. The microstructures of the TZ in the FSW and the UAFSW joints are shown in Fig. 9. A large amount of retained deformation microstructure characteristics can be observed in the TZ of the FSW joint (indicated by the arrows in Fig.9 (a)). However, the TZ of the UAFSW joint consists of fine equiaxed grains (Fig. 9(b)), indicating typical dynamic recrystallized microstructure. The microstructure difference in the TZ of the two joints is attributed to the change of metal flow behavior caused by ultrasonic vibration. Tao et al. [28] found that the fracture behavior of 2198-T8 FSW joints fractured in the stirred zone (SZ) rather than in the lowest hardness zone (LHZ). The smallest Taylor factor and lithium segregation at grain boundaries contributed to preferential intergranular fracture in the TZ and subsequently resulted in fracture in the SZ. Based on the foregoing analysis, it is reasonable to believe that the application of axial ultrasonic vibration in the FSW process promotes the dynamic recrystallization of TZ, which is of great significance for further improving the mechanical properties of the joint. In order to understand the effect of ultrasonic vibration on the metal flow behavior of different micro-domains in the NZ, the distribution of deformed aluminum foil of the SDZ, PDZ and SWZ in the FSW and the UAFSW were compared (Fig. 10, Fig. 11, Fig. 12). Fig. 10 shows the morphological distribution of the deformed aluminum foil fragments (in dark orange) in the SDZ. It is noted that the fragments in the SDZ of the FSW and the UAFSW joints are ultimately mainly deposited on the AS. In this zone, the material is supposed to be driven from the RS, through the weld center line, to the AS behind the tool. The aluminum foil fragments are seen to display a continuous distribution in the FSW joint, as shown in Fig. 10(a). The aluminum foil (indicated by the arrow) extends horizontally to the AS, which corresponds to the flow direction of the metal in the SDZ. However, the aluminum foil fragments in the SDZ of the UAFSW joint exhibit dispersed and curved morphology. A preliminary interpretation of this phenomenon is that the aluminum foil is interrupted by the high-frequency vibration of the shoulder, resulting in discontinuous dispersion of the fragments. The magnified images of the interface between the NZ and thermo-mechanically affected zone (TMAZ) of AS are shown in Fig. 11. It is clear that the interface of the UAFSW joint is smoother than that of the FSW joint, and a bulge structure was formed in the FSW joint. The smoother interface indicates that the metal has better flowability. This should be due to the reduction of deformation resistance of the material by the ultrasonic vibration, which improves the uniformity of the metal flow at the interface.

The distribution of the deformed aluminum foil in the PDZ and SWZ of the joints is shown in Fig. 12, where the site of the original aluminum foil is represented by the white dotted line. In the PDZ and the SWZ, the original aluminum foil in the butt plates is mainly sheared and extruded by the pin. Due to the stir action of the tool, a large centrifugal force is generated to extrude the aluminum foil away from the original position. The periodicity of material flows from the RS to the AS, the tool drives deformed and stretched fragments are driven by the tool and are ultimately deposited on the RS and extended to the AS. The height of the PDZ of the UAFSW is significantly larger than that of the FSW, indicating that the ultrasonic vibration enhances the flow of metal along the thickness direction and increases the volume of the PDZ. In addition, it is further illustrated that the TZ is pressed by the metal in the SDZ and PDZ, leading to a decrease in the thickness of the TZ. It is worth noting that the offset distance of aluminum foil is small in the SWZ, indicating an inadequate material flow at the bottom. The angle between the deformed aluminum foil and the vertical line is around 63.3° in the UAFSW, which is larger than 55.6° of the FSW. The effect of ultrasonic vibration on the metal flow at the bottom of the joint is still effective.

3.4. Microstructure evolution of NZ

The EBSD grain boundary maps at the different depths (1, 3 and 5 mm from the top surface) in the NZ of the FSW and the UAFSW joints welded at 1200 rpm and 160 mm/min are shown in Fig. 13. The grain boundaries were divided into three types of boundaries, one is high-angle grain boundary (HAGBS, θ >15°, θ is the disorientation angle between the adjacent grains), which is indicated by black lines. The other two are low-angle grain boundaries (LAGBs, 5° < θ < 15°) indicated with the red lines, and one low angle boundary (2° < θ < 5°), indicated with the green lines, the latter is considered as sub-grain boundary. Dynamic recrystallization results in the formation of fine and equiaxed grains due to the maximum strain and peak temperatures in the NZ [40]. Due to the high stacking fault energy (SFE) of aluminum, the grain refinement mechanism is usually continuous dynamic recrystallization (CDRX). During the hot deformation, dislocations are introduced continuously in the subgrain boundary regions, and LAGBs are transformed to HAGBs to form a fine-grained structure. In some highly strained regions of the NZ, the accumulation of dislocations may cause nucleation of recrystallization, thereby forming ultrafine grains [41]. The average grain size (AGS) and the proportion of different types of grains at the different depths in the NZ of the FSW and UAFSW joints are listed in Table 2. EBSD analysis revealed that the proportion of recrystallization in the UAFSW is lower than that in the FSW. This indicates that the ultrasonic vibration may hinder recrystallization. Furthermore, it has been found that the proportion of deformed and substructured grains in the UAFSW sample is slightly higher than those at the same position in the FSW sample. Fig. 14 shows the magnified view of the rectangle in Fig. 13(a) and (b). Compared with FSW (Fig. 14(a)), there are more low-angle boundaries in the selected area of the UAFSW joint. In addition, there is a phenomenon that the grain boundaries of the multiple colors indicates that the subgrain boundaries gradually transform to the low-angle boundaries (blue arrows in Fig. 14(a) and (b)), further suggesting that CDRX dominated microstructural evolution in the NZ.

In this study, more subgrains and deformed grains can be found in the UAFSW joint. The results can be explained as follows: ultrasonic vibration produces a large strain gradient, and the multiplication speed of dislocation increases with the increase of the dislocation movement rate. The dislocation density of some grains is high and it increases with the increase of strain. These dislocation rearrangement produces dislocation walls, which eventually transform into low or high-angle grain boundaries and then form fine grains. Siu et al. [42] also observed the phenomenon that the ultrasonic irradiation enhances the formation of subgrain during deformation. They concluded that the superposition of ultrasonic stress fields may promote the formation of subgrain.

The effect of ultrasonic vibration on grain size of NZ is also noted. Grain size characteristics at different depths in the NZ of FSW and UAFSW joints are the same, and the grain size decreases gradually along the thickness direction. The grain structure is characterized by the largest grain size at the top, which is mainly due to the large amount of heat generated at the shoulder [43]. The temperature decreases gradually from the top to the bottom of the NZ, and a large temperature gradient distribution is exhibited [44]. Consequently, the grain size decreases with increasing distance from the top surface. At the bottom, the grain size is the smallest, which is related to the rapid cooling rate of the backing plate and the severe stirring effect at the end of the pin [45].

For the same position below the weld surface the AGS of the UAFSW sample is smaller than that of the FSW sample. The size of dynamically recrystallized grain was calculated using Eq. (1) and Eq. (2) [46]:

d^-1=a+bln(Z) (1)

where d is the grain diameter, a and b are constants, and Z is the Zener-Hollomon parameter which is expressed as:

Z=$\dot{\varepsilon} exp (\frac{Q}{RT})$ (2)

where $\dot{ε}$ is the strain rate, Q is the activation energy, R is the gas constant and T is the thermodynamic temperature. The size of a dynamically recrystallized grain decreases with increasing Z-parameter. As the thermal effect of ultrasonic vibration is relatively small [20], only the strain rate of the NZ is considered. Ultrasonic vibration can increase the dislocation multiplication and promote the mobility of dislocation, which results in the increase of dislocation density and strain rate [47,48]. This may lead to the formation of smaller grain size in the UAFSW joint. Similarly, the high frequency vibration leads to intense plastic deformation of NZ along the thickness direction of the plate, resulting in more high-strain regions in the NZ, which promotes dynamic nucleation.

Grain size distribution of NZ at various depths of FSW and UAFSW joints are shown in Fig. 15. The main distribution range of grain size at the top of the NZ was 2-10 μm, while the proportion of grain with 2-5 μm is larger at the middle and bottom of the NZ. Obvious difference exists in the grain size distribution at the top surfaces of the FSW and of the UAFSW joints. The proportion of grains with a size less than 5 μm in the UAFSW is significantly larger than in the FSW, but with the increase of the distance from top surface, the distribution characteristics of grain size are basically the same. Padhy et al. [24] have studied the subgrain formation in UVeFSW of 6061 aluminum alloy, and the decrease of the grain size due to the high degree of plastic deformation and recrystallization of UVeFSW was analysed. However, the present study shows that the degree of recrystallization may be reduced due to the ultrasonic vibration. This is inconsistent with the results of Padhy, which may be related to the difference in the way that the ultrasonic vibration is applied during the welding process. Additionally, the difference in alloy composition of experimental materials is also one of the factors worth considering.

The above results show that the ultrasonic effect is most pronounced on the top surface of the NZ, and decreases gradually along the central axis of the NZ. The main reason for this may be that the high-frequency vibration of the shoulder acts directly on the top surface of weld. Thus, the ultrasonic vibration gradually weakens along the direction of the plate thickness during the propagation, thereby reducing the effect of the ultrasonic action.

3.5. Mechanical properties

Fig. 16 shows the microhardness profiles along the mid-thickness of FSW and UAFSW joints for different natural aging times. In Fig. 16(a), the hardness profiles of the FSW and UAFSW joints are basically the same after 120 h of PWNA. The hardness profiles show a plateau in the central part of the weld with two insignificant minimums on both sides. It is also noted that the hardness in the central part of the weld are lower than that of the BM (100-110 HV). The finer grains may cause the hardness of the UAFSW joint to be slightly higher than that of the FSW joint. The minimum hardness positions of the FSW and UAFSW joints are located at 15 mm and 25 mm from the center of the weld, respectively.

Generally, the strengthening effect of the 7xxx aluminum alloy is mainly determined by microstructural facters, such as precipitates and the grain size [49]. The main strengthening phase of the 7N01-T4 aluminum alloy is the GP zones. Near the NZ severely plastic deformation and high heat input may cause most of the GP zones to dissolve into the aluminum matrix, leading to decrease in the strengthening effect of the precipitates [50]. The differences between the hardness distributions of the FSW and UAFSW joints are small. This indicates that the effect of ultrasonic vibration on the dissolution and re-precipitation of precipitates is not significant. During the welding process, the precipitates are dissolved, coarsened or transformed. In the subsequent cooling stage, a large amount of GP zones are reprecipitated from the supersaturated solid solution. The number and type of precipitates are related to the thermal cycle of the weld. Ultrasonic vibration introduces additional heat input, but the "contact-separation" state of the shoulder and the surface of the plate reduces frictional heat generation, ultimately resulting in a similar thermal cycle of UAFSW to that of FSW. Therefore, the hardness value of the UAFSW joint is not higher than the FSW joint.

Both joints show increased hardness after 720 h of PWNA (Fig. 16(b)). The minimum hardness value in the weld zone increases from 85 to 100 HV, which is an increase of approximately 18 %. Kalemba et al. [51] reported that the hardness recovery of the 7136 aluminum alloy FSW joint during natural aging, and they suggested that the formation of GP(I) zones is responsible for the increased mechanical properties. A large recovery of the hardness in the NZ is mainly due to re-precipitation of the precipitates. Formation of supersaturated solid solution may be caused by high welding temperature. Besides, excess vacancies in the NZ is supposed to be a result of high cooling rate. A high concentration in solid solution and excess vacancies provide a high diffusion path that promotes rapid nucleation and growth of the GP zones, which may shorten the natural aging time [52]. Compared to the joints at 720 h of PWNA, the microhardness of joints increased slightly after 4320 h of PWNA (Fig. 16(c)). Microhardness values of the FSW and the UAFSW joints are similar to that of the BM, and the aging is nearly completed after 4320 h of PWNA.

The ultimate tensile strength (UTS), yield strength (YS) and elongation (EL) of the BM and the joints at different natural aging times are shown in Fig. 17. The average UTS, YS and EL of the FSW joints after 120 h of PWNA are 350.2 MPa, 221.7 MPa and 15.3 %, respectively. These are comparable to the results of the UAFSW joints (352.4 MPa, 225.7 MPa and 12.9 %, respectively). The fracture locations of the FSW and UAFSW joints after 120 h of PWNA and their fractures are shown in Fig. 18 and Fig. 19, respectively. It is noted that the fractures of the FSW joints are located at the NZ, while the fractures of the UAFSW joints are located at the RS-HAZ, as shown in Fig. 18 (a) and (b), respectively. The fractures of the FSW joints are located at the NZ, which may be caused by weld thinning problem of the joints, whereas the fractures of the UAFSW joints are not located at the NZ, which indicates that the weld thinning problem is not the main cause. The kissing bond can be observed at the bottom of the FSW joint due to insufficient plastic flow, as shown in Fig. 18(c). In addition, the SEM observations of the fracture surfaces indicated some micro-cavities in the TZ of the FSW joint (Fig. 19(a) and (b)). The micro-cavities are aligned in the direction parallel to the joint surface. These two regions (kissing bond and TZ) may be the sources of crack initiation, which leads to fracture in the NZ of the FSW joint during the tensile test. As seen from the Fig. 18(b), the fractures in the UAFSW joints are located at the RS-HAZ, which indicates that the NZ is not the weakest part in the UAFSW joint. The ultrasonic vibration may improve the property of NZ due to three reasons: first, ultrasonic vibration reduces the thickness of the TZ, and the microstructure of the TZ becomes denser without micro-cavities; second, the oxide layer changes from continuous arrangement to discrete particles due to the ultrasonic vibration; last but not the least, the arc surface of the UAFSW joint is more uniform and smooth, which reduces local stress concentration. The fracture paths of the FSW and UAFSW tensile samples display a shear fracture mode and the angle between the fracture surface and the loading direction is about 45° (Fig. 18(c) and (d)). The fracture surfaces of the two joints were characterized by transgranular features with a large number of dimples, which is a typical ductile fracture characteristic (Fig. 19(b) and (d)). In Fig. 18(d), obvious plastic deformation can be observed in the NZ of the UAFSW specimens, and the boundary between the NZ and the TMAZ is very clear. Thus, it can be inferred that the deformation of the NZ also occurred before failure due to relatively low hardness of the NZ.

Compared to the joints after a 120 h of PWNA, the UTS, YS and EL of the FSW joints were further increased with the increase of the natural aging time (720 and 4320 h), reaching 396.1 MPa, 257.7 MPa and 17.1% respectively (720 h). After 4320 h of PWNA, the UTS, YS and EL of the FSW joints are 416.3 MPa, 271.4 MPa and 16.9 %, respectively. Similarly, the UAFSW joints exhibit the strong strengthening effect of the natural aging, the UTS, YS and EL of the UAFSW joints are 399.8 MPa, 261.6 MPa and 18.2 %, respectively (720 h). After 4320 h of PWNA, the UTS, YS and EL of the UAFSW joints are 419.5 MPa, 276.6 MPa and 18.4 %, respectively. In summary, the mechanical properties of the FSW and UAFSW joints increase clearly after the PWNA. However, the difference in the mechanical properties of the two joints is not significant. After 4320 h of PWNA, the increase in EL by ultrasonic vibration is significant (8.8 %).

The fracture locations and fracture surfaces in the FSW and UAFSW joints after 720 and 4320 h of PWNA are shown in Fig. 20. It is seen that all the fractures are located at the NZ, which may be mainly due to weld thinning. The dimples on the fracture surface were fine and equiaxed in the FSW and UAFSW joints (Fig. 20(e) and (f)). Compared with the fractured surface of the FSW, the dimples on the fracture surface of the UAFSW joint are deeper. Grain refinement in the NZ is the main reason for the better ductility of the UAFSW joint [53].

4. Conclusions

In this work, the ultrasonic assisted friction stir welding (UAFSW) was applied to the 6 mm thick 7N01-T4 aluminum alloy. Axial force, weld formation and plastic flow behavior were compared between the joints produced by the UAFSW and the conventional FSW. In addition, the influence of ultrasonic vibration on the microstructure evolution and mechanical properties of the joints after natural aging were also studied. The main conclusions can be summarized as follows:

(1) Ultrasonic vibration expands the welding process window. With the tool rotation speed of 1200 rpm, ultrasonic vibration can increase the welding speed of the defect-free joint from 200 mm/min to 300 mm/min. At the processing parameter of 1200 rpm and 160 mm/min, ultrasonic vibration may improve the surface quality of the weld, and the axial force is reduced by about 9 %.

(2) Stress superposition and acoustic softening caused by ultrasonic vibration can significantly improve the flow of metal, and enlarge the volume of the pin-driven zone (PDZ), which results in a thinner transition zone (TZ) between the PDZ and the shoulder-driven zone (SDZ).

(3) Grain refinement in the UAFSW joint is attributed to the increase of strain rate and strain gradient of the metal in the NZ due to ultrasonic vibration, and more subgrains and deformed grains are formed. The influence of ultrasonic vibration on the microstructure evolution of the NZ is gradually reduced from the top part to the bottom part of the joint.

(4) The tensile properties of the FSW and UAFSW joints were significantly improved after the prolonged natural aging. After 4320 h of PWNA, the UTS and EL of the FSW joints are 416.3 MPa and 16.9 % respectively, and the UTS and EL of the UAFSW joints are 419.5 MPa and 18.4 %, respectively. The EL of UAFSW is 8.8 % higher than that of FSW, which is attributed to the finer grains.

Acknowledgement

This work was funded by the National Key Research and Development Program of China (No.2016YFB1200506-12).and Project of Promoting Talents in Liaoning Province (No. XLYC1808038).

---