Microstructures and mechanical properties of non-thinning and penetrating friction stir welded 2219-T6 aluminum alloy

The aim of this paper is to offer a novel friction stir welding technique to achieve non-thinning and lack of kissing bond welds with excellent mechanical properties. In the conventional friction stir welding (CFSW) process, a certain size spindle tilting angle and a shoulder plunge depth were applied in order to obtain high-quality welds. However, what was the reason for the thinning phenomenon of weld and thus the weakening of weld bearing capacity? The reduction of spindle tilting angle and shoulder plunge depth was likely to cause the welding tool not being able to reach the weld root, and thus, it was engendered for the kissing bond or the absence of complete penetration. In the present study, a novel welding tool with large size shoulder and Archimedes spiral groove was designed in order to achieve non-thinning and lack of kissing bond welds. The microstructures and mechanical properties of the joints obtained by the non-thinning and penetrating friction stir welding (NTPFSW) and the penetrating friction stir welding (PFSW) had been analyzed in detail. Compared with PFSW, NTPFSW had a narrow shoulder-affected zone. The thickness of NTPFSW joint is 0.1 mm thicker than that of BM and 0.22 mm thicker than that of PFSW joint. The stir zones (SZ) obtained by the PFSW and NTPFSW were both composed of fine equiaxed grains. The grains in the upper part of NTPFSW joint were larger in size, and the number of precipitates (θ’ phase) was less. NTPFSW completely eliminated the weld thinning phenomenon, and the tensile properties were 365 Mpa, which is 2% higher than that of PFSW.


Introduction
Friction stir welding (FSW) is one of the most successful joining techniques in the twentieth century [1][2][3]. Due to its many advantages such as high energy efficiency, low cost, environmental friendliness, and high welding quality [4,5], FSW has gained considerable scientific and technological attention in many fields, including aerospace, railway, renewable energy, and automobile [6][7][8][9][10]. During the welding, a suitable shoulder plunge depth can provide adequate forging force and heat input [11][12][13], improve the material flow in stirring zone, avoid the formation of internal defects, and finally obtain a sound FSW joint [14]. However, due to the existence of plunge depth, the flash defects and burrs are inevitably formed on the weld edge. Flash defects can cause the loss of material, and the thickness of weld is thinner than that of the base material, which is called weld thinning [15]. The weld thinning has many negative effects on the joint. Firstly, the integrity of external dimension of joint is destroyed. The flash defects induce by weld thinning maybe become a potential security risks during service [16]. Secondly, the effective load bearing area is reduced due to weld thinning [15]. Kumar et al. [17] indicated that when the plunge of the rotating shoulder increased to a certain extent, weld thinning markedly increased, which easily caused joint fracture in the weak region of the weld and drastically deteriorated the mechanical properties. Thirdly, the aluminum alloy structures, such as those use in aerospace and railway, usually beard both static load and dynamic load during service. The thinning of joint can greatly reduce its fatigue properties. In addition, the stress concentration cause by thinning can greatly reduce the stress corrosion resistance of the joint. Li and Liu [18] have welded 2219 aluminum alloy by non-rotational shoulder-assisted friction stir welding (NRSAFSW). Although the weld thinning was eliminated by NRSAFSW, the maximum mechanical property was only 307 MPa, equivalent to 69% of the BM. However, the tensile strength of the conventional FSW (CFSW) joint obtained at the optimal welding parameter with the same base material was 346 MPa [19], which demonstrated that the mechanical properties of joints obtained by NRSAFSW were unsatisfactory. Zhang et al. [20] designed a new type of welding tool to carry out non-thinning friction stir welding (NTFSW). A concave geometrical feature was utilized to the end surface of tool shoulder, and three equally spaced scrolls were machined on the concave shoulder. In such case, the weld thinning phenomenon was avoided. However, the shoulder plunge depth was only 0 mm, and thus, insufficient heat input resulted in the formation of kissing bond at the joint root.
The weld thinning reduces the actual bearing area of the weld and deteriorated the dimensional accuracy of the whole joint. At the same time, the existence of depression also led to stress concentration and reduce the mechanical properties and corrosion resistance of the joints. In the conventional friction stir welding (CFSW), the length of the stirring pin was a bit shorter than the thickness of workpiece [21,22]. Therefore, a kissing bond was formed at the joint root, resulting in the deterioration of mechanical properties. Especially, under the dynamic loading conditions, the kissing bond became a crack source in the joint and significantly reduced the fatigue strength of the joint. In order to completely cancel the kissing bond, Hu et al. [23] designed a slight longer stirring probe than the thickness of workpiece in order to accomplish tilt probe penetrating friction stir welding (PFSW), and the kissing bond at the root was successfully eliminated, but the weld thinning phenomenon was still exist.
All in all, the previous studies have indicated that the weld thinning and kissing bond are difficult to remove from the FSW joints in the same time. To break through the limitations, a non-thinning and penetrating friction stir welding (NTPFSW) is put ward, which is expected to efficiently eliminate the weld thinning phenomenon and to avoid the formation of kissing bond at the joint root in the present study. By using the same process parameters, the microstructure and mechanical properties of NTPFSW and PFSW joints were compared and analyzed, and thus, the advantage of NTPFSW was revealed.

Experimental material and procedures
The base material (BM) used in this study was 5-mm-thick 2219-T6 aluminum alloy in T6 condition with the length of 300 mm and width of 80 mm, which was a type of aviation aluminum-alloy belonging to Al-Cu series. Its chemical composition is shown in Table 1, all in wt%. The tensile strength of BM was 436 MPa, and the elongation was 11.2%.
In order to produce NTPFSW joints without kissing bond and weld thinning, the welding tool should be properly designed. The tool was designed to split style, which was divided into three parts, i.e., clamping part, shoulder, and stirring pin. Each part was fixed by thread, and Archimedes spiral groove was machined on the concave shoulder. The spiral groove was 2 mm in width, 0.75 mm in depth, and the spiral equation was ρ = 2/π, which started from the edge of shoulder central hole and anticlockwise extended to the outer edge of the shoulder. The shoulder diameter and the pin length were 24 mm and 6 mm. The physical NTPFSW tool is displayed in Fig. 1a-d. The tool tilt angle was 0.5° and the shoulder plunge depth was 0.1 mm during the NPTFSW. The contrastive analyses of an NTPFSW joint and a PFSW joint were conducted. As displayed in Fig. 2a, a tool with a shoulder 16 mm in diameter and a pin of 6 mm in length is used in PFSW. PFSW was proved superior to CFSW in mechanical properties at the welding speed of 300 mm/min and a rotation speed of 800 rpm [23]. In order to compare with NTPFSW and PFSW, both welded joints were obtained with the same process parameters at the welding speed of 300 mm/min and a rotation speed of 800 rpm. In NTPFSW and PFSW, the traditional backing plate is replaced by stationary shoulder in which a blind hole of 8 mm in diameter and 4 mm in depth was machined in the stationary shoulder center, as shown in Fig. 2b. The function of blind hole was to accommodate the penetrated part of the stirring probe because the length of probe was longer than the thickness of workpiece. Both NTPFSW and PFSW are carried out in the same welding system, as shown in Fig. 2c.
The microstructure analysis methods mainly included optical microscope (OM), electron backscattered diffraction (EBSD), and transmission electron microscope (TEM). The OM, EBSD, and TEM samples of the welds were cross-sectioned by electrical discharge machining. Different granularity abrasive papers were used to grind samples. 2.5 μm and 1 μm polishing paste were used for mechanical polishing and then followed by Krolls reagent (92 ml distilled water + 6 ml HNO 3 + 2 ml HF). The macrostructure and microstructure of the joints were observed by OM (Olympus VHX-1000). The samples were prepared by Argon ion polishing and used for SUPRA55 field emission scanning electron microscope EBSD analysis. The tensile test samples were obtained with a size of 150 mm × 15 mm × 5 mm. Three tensile test samples with the same size were taken from NTPFSW joint and PFSW joint and were subjected to iso-strain test at room temperature with a loading speed of 1 mm/min by a computer-controlled tensile testing machine (Instron5569). The final tensile properties of the joints were the average of the three values. Figure 3 shows the weld cross section of the PFSW and NTPFSW joints, which are obtained at the welding speed of 300 mm/min and rotation speed of 800 rpm. No kissing bond and welding defects are found in the joints obtained from either welding methods. Each joint is divided into four zones, i.e., stir zone (SZ), thermal mechanically affected zone (TMAZ), heat-affected zone (HAZ), and base material (BM). The SZ is divided into shoulder-affected zone (SAZ) and pin-affected zone (PAZ). It can be seen that the depth of SAZ in PFSW is deeper than that of NTPFSW. This illustrates that under the same nominal inserting depth, the extrusion effect of NTPFSW shoulder on the upper surface of the joint is reduced, and the material flow behavior is controlled. Figure 3 shows the weld appearances of joints, which are obtained be PFSW and NTPFSW with the same welding parameters. As Fig. 3a exhibits, the welding thinning occurs in PFSW joint. Compared with BM, it can be clearly found in Fig. 4a that PFSW joint is depressed and there are obvious  Fig. 4b. The arc corrugation is remarkable for PFSW, whereas the upper surface is flat for NTPFSW. In the FSW process, in order to ensure that the materials have sufficient thermal plasticization and forging forming effects, the shoulder of welding tool needs to be pressed into the workpiece to a sufficient depth. The extruded materials become the flashes of the weld, and the joint shows a significant depression relative to base metal, leading to the weld thinning phenomenon. As shown in Fig. 5, the value of weld thickness (ΔT) is defined as the thickness of the base material (ΔT BM ) minus the thickness of the joint (ΔT weld ). △T of the PFSW joint is − 0.12 mm, which is 2.4% of the base metal thickness. In comparison, ΔT of the NTPFSW joint is 0.1 mm which means the thickness of NTPFSW joints is not thinner than the BM. The tool is subject to the axial reaction force from BM, while it performs forging action in the FSW process. If the axial reaction force is large enough, the tool can be lifted from its original position [24,25]. For the welding speed of 300 mm/min, the interaction force between the tool and BM is strong. The ability of axis shoulder to converge materials together is improved by the Archimedes spiral groove on the NTPFSW shoulder; hence, the NTPFSW tool also is given more axial reaction force. Therefore, the upper surface of NTPFSW joint is slightly higher relative to BM. This result indicates that the weld thinning phenomenon has been completely eliminated, and nonthinning friction stir welding can be achieved by NTPFSW.

Microstructure characteristics of welded joints
The EBSD inverse pole figures (IPF) of the NTPFSW joint and PFSW joint are shown in Fig. 6. The grain size decreases from upper to bottom. The SZ of joints obtained by both welding processes consists of equiaxed grains, which are mixed by a certain number of small and large grains. Therefore, the SZ of joint undergoes complete dynamic recovery and dynamic recrystallization during the friction heat generation and intense plastic deformation. The vast majority of the grains are trended to < 111 > orientation (blue). The grain elongation in the upper part of NTPFSW joint is more obvious. In Fig. 6, black lines correspond to high angle boundaries (HABs) with misorientation angles exceeding 15°, and the white lines correspond to low angle boundaries (LABs) with misorientation angles ranging from 2° to 5°. Obviously, the HAB content of HABs in upper part of NTPFSW joint reaches to 75.5%, which is higher than that in PFSW joint. Figure 7 shows the number and area fraction of grains with varied sizes. As shown in Fig. 7a, the average grain size in the upper part of NTPFSW joint and PFSW joint are, respectively, 7.86 μm and 7.26 μm. The proportion of grains with the size greater than 15 μm is significantly high in the NTPFSW joint. As can be seen from Fig. 8a, c, no significant difference in HAZ grain size is observed between the two joints. HAZ is far from SZ, which is little affected by the welding thermal cycle. Therefore, the grain size is larger and close to that of BM. Figure 8b and d exhibit that the effect of grain elongation in TMAZ of NTPFSW joint is more obvious than that of PFSW joint. This is due to the fact that the shoulder of NTPFSW is larger than that of PFSW and more welding heat is produced at the same parameters. During welding process, more welding heat is produced in NTPFSW, leading to better thermoplastic of material. The grain elongation is more obvious under the thermal mechanical effects. The grain size of HAZ and TMAZ is analyzed quantitatively in Fig. 9. As exhibited in Fig. 9a, b, the average grain size of HAZ in NTPFSW is 21.55 μm which is basically the same size as PFSW (22.96 μm). The grain size of NTPFSW joint in TMAZ is larger than that of PFSW ( Fig. 9c). Figure 9d illustrates that the grain size less than 10 mm is more in PFSW joint.
In conclusion, the diameter of NTPFSW shoulder is larger than that of PFSW, and the peak temperature is higher, and the high temperature residence time is longer, and the grain can fully grow. The fine equiaxed crystals in the joint are mostly surrounded by HABs, which does not contain substructures. Their growth conforms to the definition of discontinuous dynamic recrystallization [26]. The grain size of the NTPFSW and PFSW joints decreases gradually from the upper part to the lower part of joints, as commonly observed in other published literature for the FSW joints [27]. With respect to the NPTFSW joint, the magnitude of grain size change in three layers is smaller than that in PFSW. It is

Precipitate evolution in the SZ
Precipitates in the SZ of FSW joints are illustrated in Fig. 10. The θ' phases are indicated by the red arrow. Compared with the high-density needle-like θ' phase in the BM (Fig. 10a), the density of θ' phase in the SZ is significantly reduced because a lot of θ' phases are dissolved into the BM under the influence of welding thermal cycle. The plate-like θ' phase existing in SZ, as indicated by the red arrow in Fig. 10b, c, is re-precipitated during the cooling process. Comparing Fig. 10b with c, it can be concluded that the density and size of θ' phase in NTPFSW joint are smaller than those of PFSW joint. This is due to the fact that the shoulder area of NTPFSW is larger than that of PFSW. With the same welding parameters, the welding thermal cycle of NTPFSW has a longer elevated temperature holding time and a slower cooling rate, which makes it possible to re-precipitate θ' phase. PSFW produces less heat than NTPFSW, resulting in shorter elevated temperature holding time and less re-precipitate θ' phase into SZ. The size of precipitate θ' is smaller than that in NTPFSW joint owing to the lack of time to grow up. The selected area diffraction (SAD) patterns along [0 1 1] zone axis of aluminum matrix is illustrated in Fig. 10d. It can be seen that in addition to a set of diffraction spot of the matrix, there is another set of diffraction spot with weak brightness. After calibration, it is the diffraction pattern of θ' phase (Al 2 Cu) and the corresponding crystallographic plane index that has been marked on the picture. The SAED result identifies that the light gray plate-like precipitates are θ' phase and the re-precipitation of θ' phase during cooling process is confirmed. Figure 11 shows the element mapping of SZ in NTPFSW joint. Clearly, apart from the re-precipitated θ' phase, the large bright phases are high melting point intermetallic, which contains Mn, Fe, Si, Zr, and other elements. These elements may form precipitating phases of Al 3 Zr and Cu 2 Mn 3 Al 20 , which are also prone to appear in 2219 aluminum alloy [28,29].

Mechanical properties of welded joints
In order to compare the joint properties, a strength factor is defined as the joint tensile strength divided by the BM tensile strength. The tensile properties of joints are shown in Fig. 12. It can be seen from Fig. 12 that the tensile strength of NTPFSW joint is 365 MPa, and the strength factor is 83.7%, which is 2% higher than that of PFSW joint (355 MPa, 81.4%). Moreover, the elongation of joint is increased from 7.5 to 9.3%.
All in all, by means of non-thinning welding method, the tensile properties of the joint are not reduced and are slightly To further clarify the effects of joint microstructure on the tensile properties, the fracture positions and surfaces of the joints were examined. As can be observed from Fig. 13a, b, at the welding process parameters of 800 mm/ min and 300 rpm, the crack propagates from the SZ periphery of weld bottom and extends along TMAZ. The necking phenomenon can be obviously observed in NTPFSW joint (Fig. 13a). In the mass, both joints are composed of a large number of dimples and partial tearing edge, which shows a typical ductile fracture. The size of dimple in the middle of the joint is larger than that in the upper part. By comparing Fig. 13c-e, f-h, it can be found that the number of tearing edges is more in PFSW joint. The necking phenomenon and the number of tearing edges illustrate that the toughness of NTPFSW joint is better, which is consistent with the previous test results. As displayed in Fig. 13c, f, the dimple size of the upper part of NTPFSW joint is smaller than that of PFSW joint. The reason for this effect is that the size of NTPFSW is large, so its linear velocity is higher at the same rotation speed. In addition, the shoulder contains Archimedes spiral groove, so the strain rate of the upper part is higher, correspondingly, the size of dimples is small [30].

1.
A novel FSW tool with large size shoulder and Archimedes spiral groove is designed, and the weld thinning phenomenon is successfully eliminated. Compared with PFSW, the weld thickness of NTPFSW increases by 4.4%. 2. The upper part of NTPFSW joint is composed of fine equiaxed grains, and the grain size is larger than that of PFSW joint as well. The variation range of grain size is smaller along the direction of plate thickness. 3. The density of θ' phase in the SZ is significantly reduced because a lot of θ' phase is dissolved into the BM, and the density and size of θ' phase in the NTPFSW joint are smaller than those of PFSW joint. PFSW produces less heat than NTPFSW and less re-precipitate θ' phase into BM. The size of precipitate θ' is smaller than that in NTPFSW joint owing to the lack of time to grow up. 4. The tensile strength of NTPFSW joint is up to 365 MPa, the strength factor is 83.7%, and the fracture elongation is 9.3%. Such a result proves that the elimination of weld thinning produces an improvement to tensile properties.

Data availability
The raw/processed data required to reproduce these findings cannot be shared at this time as the data also forms part of an ongoing study.

Competing interests
The authors declare no competing interests.