Study on corrosion resistance of HAZ and TMAZ in friction stir welding joint of 7075 aluminum alloy by thermal simulation

It is difficult to characterize the variation of corrosion resistance of the narrow areas in friction stir welding (FSW) joints due to the large temperature gradient. In this paper, the welding thermal simulation was performed to simulate the heat affected zone (HAZ) and thermo-mechanical affected zone (TMAZ) of the FSW 7075-T6 aluminum alloy, and the corrosion resistance and microstructure of the simulated samples were studied. Results show that the corrosion potential changes greatly under different thermal simulation temperatures. The pitting corrosion of the HAZ simulated samples presents two pitting potentials, but for the TMAZ simulated samples, two pitting potentials will gradually evolve to one pitting potential with the increase of the maximum temperature. The electrochemical impedance spectroscopy results show that the corrosion mechanism of the HAZ and TMAZ is completely inconsistent, which is related to the differences in precipitate and grain characteristics.


Introduction
As an advanced join method, friction stir welding (FSW) shows the advantages of low heat input and fewer welding defects, which has been widely used in the welding of aluminum alloy, especially for the aging strengthened aluminum alloy. The strength of the FSW joint is also improved compared to the fusion welding due to the avoiding of various welding defects [1][2][3][4][5]. And different from fusion welding, previous studies have widely shown that the thermo-mechanical affected zone (TMAZ) is the weakest area of mechanical properties for FSW joints of aging precipitated alloy. Because of the dissolution and coarsening of strengthening precipitates, the TMAZ usually exhibits low hardness and poor corrosion resistance [6][7][8][9][10].
There is a large temperature gradient for different positions in the HAZ and TMAZ during FSW thermal cycle, so the microstructure and performance of the joint also have a significant heterogeneity [11]. It is difficult to analyze the variation of performance for a specific point in the HAZ or TMAZ alone. But the welding thermal cycle of a selected point in the HAZ or TMAZ can be obtained through thermal simulation testing according to the heating, cooling, and deformation parameters. Then microstructure and performance of the differently tiny regions in the joint could be predicted by the simulated sample [12][13][14][15][16]. In the past, various studies have been conducted to analyze the thermal process [17,18] or corrosion behaviors [19,20] of FSW Al-Zn-Mg-Cu alloy. However, report related to studying the local corrosion resistance of FSW joint by thermal simulation is rare.
In this paper, the cylindrical samples were thermal cycled to simulate the formation of different regions in HAZ and TMAZ during FSW of 7075-T6 aluminum alloy, then the corrosion resistance of the simulated samples for different positions in the HAZ and TMAZ of the joint was tested, and the relationship between the performance and microstructure was also analyzed.

Materials and thermal simulation
The chemical composition of the 7075 plates used in this study is shown in table 1, which was T6 tempered before FSW and thermal simulation. The thermal simulated samples with the size of Φ 8 × 12 mm were cut from the plate. The simulated welding parameters of FSW include a rotating speed of 600 r min −1 , welding speed of 60 mm min −1 , and tool sinking value of 0.05 mm.
The measured thermal cycle during the FSW process at different positions in the HAZ and TMAZ was simplified for simulation convenience. And figure 1 gives the temperature profiles used for the FSW thermal simulation, where the actual thermal cycle curve was replaced by the straight line between the representative temperature points, and the maximum temperature is varied to simulate the positions with different distances from the weld center. The thermal simulation was performed by the a Gleeble 3500 testing machine. And as shown in table 2, the maximum temperature point for samples simulated HAZ and TMAZ is 332°C ∼ 372°C and 378°C ∼ 428°C, respectively. The accurate simulation of thermo-mechanical effect during FSW is very difficult to achieve, here the thermo-compression with a strain of 40% was applied during heating at 378°C to simulate the thermal deformation in TMAZ, and the strain rate was 0.7 mm s −1 .

Performance testing and microstructure characterization
The Vickers hardness of samples was measured by a DHV-1000 digital hardness tester under a load of 1.96 N and a holding time of 15 s. The cyclic polarization curves of the polished samples were tested by the electrochemical workstation (CHI-660E, Shanghai Chenhua Instrument Co., Ltd. China) in 3.5% NaCl solution after immersing for 4000 s. The scanning range of the polarization curve is from −1.2 V to −0.4 V (versus the reference potential), the scanning speed is 1 mV s −1 , and the number of scanning segments is 2. The standard threeelectrode system is adopted, in which the reference electrode is a saturated calomel electrode (SCE) and the counter electrode is an Ag electrode. The characteristic potentials on the cyclic polarization curve-E corr , E pit and E rp were measured. The corrosion pits after cyclic polarization were further observed by the scanning electron microscope (SEM, JSM-7800F, Japan). Electron backscattered diffraction (EBSD) characterization was performed to evaluate the grain size of samples after different treatments. The samples were electro-polished in a mixture solution of 10% perchloric acid and 90% ethanol with a voltage of 16 V and a temperature of 4 ∼ 8°C for 1 min. EBSD data were acquired via the JSM-7800F SEM attached the Oxford NordlysMax3 EBSD system. And the data was analyzed by Channel 5 commercial software. Transmission electron microscope (TEM) slices were first mechanically thinned to 120 μm. Then the foils with diameter of 3 mm were punched from the slices and electro-polished to 30 μm at a voltage of 16 V in a solution of 10% perchloric acid and 90% ethanol. Finally, the TEM samples were obtained by ion beam thinner. TEM observation was performed on a JEM-F200 apparatus at an acceleration voltage of 200 kV. Figure 2 shows the cyclic polarization curves of the HAZ simulated samples. Based on the cyclic polarization curve, the corresponding self-corrosion potential (E corr ), pitting potential (E pit ) and repair potential (E rp ) can be obtained, and the results are listed in table 3. It can be seen two pitting potentials exist for the HAZ simulated samples with different maximum temperatures. With the increase of the maximum temperature, the self- corrosion potential gradually shifts to a lower potential in general. But the first pitting potential increases first and then decreases, and the second pitting potential also moves in a more negative direction. Figure 3 shows the corrosion morphology of HAZ simulated samples after different thermal cycle temperatures. It can be seen that a small number of numerous corrosion pits are formed after cyclic polarization corrosion at low temperatures (such as 332°C), As the temperature increases (332°C), the number of pitting pits increases, some of them even connects into flakes, but the depth of corrosion pits is generally shallow. With the increase of maximum temperature, the corrosion pit shows an expanding trend, such as the corrosion morphology at 352°C. When the temperature reaches 372°C, the size of the corrosion pit becomes smaller, but also develops deeply. This is closely related to the phase transformation in the thermal cycle. Figure 4 shows the cyclic polarization curves of the TMAZ simulated samples, the corresponding characteristic potentials are also listed in table 4. The self-corrosion potential moves to the high potential direction with the increase of maximum temperature. The first pitting potential is approximate to the HAZ, which changes from −0.76 V to −0.77 V, corresponding to the rupture of oxide film. While the second pitting potential moves to a more negative direction, and with the increase of maximum temperature, the second pitting potential gradually becomes less obvious. Compared with the HAZ, the repair potential also shifts in a more negative direction. Figure 5 shows the corrosion morphology of TMAZ simulated samples after cyclic polarization, which is quite different from that of the HAZ. For 378°C, 388°C, and 398°C samples with lower temperatures, the number of pitting pits decreased significantly. While for 418°C and 428°C samples, the number of pitting pits  begins to increase, and there were obvious differences in pit characteristics, the pit tends to develop in the direction of depth. Figure 6 shows the electrochemical impedance spectroscopy (EIS) analysis results of the selected samples and their corresponding fitting circuit. The fitting values of each component are also summarized in tables 5 and 6, where R s , R f, and R ct represent solution resistance, membrane resistance and transfer resistance, respectively, and Q f and Q dl represent the capacitance model of constant phase element (CPE) with surface facial mask and double-layer capacitance, respectively. W is the Weber impedance introduced by considering the ion diffusion in the surface facial mask, and L is the inductance characteristic promoted by the weakening of the aluminum oxide layer caused by anodic dissolution. Due to the role of corrosive anions in the solution, the surface oxide layer has very poor protection, resulting in pitting corrosion on the surface. The existence of oxide film leads to the surface of HAZ-322 and HAZ-352 samples undergoing relatively uniform corrosion, producing corrosion products, and forming a new diffusion layer. However, the TMAZ-408 sample has a large inhomogeneity in the surface structure, the surface oxide film has a weak position, and the protection is very poor, so the obvious pitting phenomenon on the surface results in the phenomenon of inductance.   Figures 7(a) and (b) shows the average hardness of the HAZ and TMAZ thermal simulation samples with different maximum temperature, respectively. For HAZ simulated samples, it can be seen the hardness of the sample at 352°C is the lowest, which verifies that it has the worst precipitation strengthening effect at this  temperature. For TMAZ simulated samples, the sample with a peak temperature of 388°C shows the lowest average hardness. This temperature corresponds to the initial dissolution temperature of η phase, that is, the precipitate strengthening effect reaches a relatively low level in this condition. With the peak temperature further increases, accompanied by the influence of deformation, more alloy elements are dissolved into the matrix and the precipitates may form again in the cooling stage, which improves the hardness of samples. Figures 8 and 9 show the TEM images and EDS elements distribution of the HAZ simulated sample with the maximum temperature of 352°C. The main strengthening precipitate in 7 series aluminum alloy is the η' phase. According to the precipitation sequence of Al-Zn-Mg-Cu alloy, the η' phase is precipitated from the supersaturated solid solution, and with the increase of aging temperature, η' phase will gradually be transformed into η phase. The strengthening effect of η phase is obviously weakened compared to the η' phase. Besides, due to the addition of Cu, the precipitation of θ' (Al 2 Cu) phase is also possible. Previous studies have shown that the formation temperature of the η' phase is about 120°C and its dissolution temperature is about 250°C, the precipitation temperature of the η phase is 200°C ∼ 250°C and its dissolution temperature is about 370°C. Therefore, for the thermal cycle of the HAZ simulated samples, the temperature reaches the dissolution temperature of the η' phase, which also corresponds to the temperature of the formation of η phase. As shown in figures 8 and 9, the coarse η phase and grain boundary precipitates can clearly be observed. TEM images of the TMAZ simulated sample with the maximum temperature of 408°C are shown in figure 10. It can be seen the number of η phase has an evident increase compared to the HAZ simulated sample. The peak temperature of the thermal cycle in the TMAZ is higher than that in the HAZ, so it can be inferred that the re-dissolution of precipitates in TMAZ is more intensified. In addition, a deformation process was also contained in thermal cycle of TMAZ, which will introduce dislocations and then affect the diffusion of solute elements. Thus, the further increase of peak temperature and addition of thermo-mechanical effect accelerates the solid solution of aging precipitates, especially for the precipitates containing Cu element. The self-corrosion potential of Cu is higher, which can improve the self-corrosion potential of the matrix. At the same time, with       the increase of peak temperature, there was more solid solution, and new secondary precipitation occurred in the cooling process. Comparing with the HAZ and the TMAZ, it can be found that in addition to the influence of temperature, deformation should also have an important influence on the performance of the joint. Figures 11 and 12 are the EBSD IPF maps of the HAZ and TMAZ simulated samples after different thermal cycles, respectively, where the low angle grain boundary (LAGB) refers to the misorientation angles of two adjacent grains is between 2°∼ 10°, and the high angle grain boundary (HAGB) refers to the misorientation angles of two adjacent grains is more than 10°. The energy of LAGB is lower than that of HAGB, so the diffusion of solute atoms in lower-mobility LAGB is more difficult than that in HAGB [21,22]. Therefore, the solute atoms are easy to diffuse in samples with a relatively more proportion of the HAGBs, which means that the diffusion of solute atoms along the grain boundaries of the sample with a small proportion of LAGB is less difficult [23]. Therefore, the redissolution of η' phase and the precipitation of η phase are easier in samples with a lower percentage of LAGBs, so it shows a low hardness. The concentration of solute elements also affects the corrosion potential of aluminum alloys, and the number and distribution of precipitates affect the pitting performance.

Discussion
From the point of the pitting corrosion potential, the first pitting corrosion potentials are generally similar, which means that they may not be related to the phase transition. Based on the previous investigation of Meng et al [9,24] on the polarization curves of the 7050, 7075, 7029, 7039 and 7004 aluminum alloys in 0.5 mol l −1 NaCl, there are two pitting potentials in the corresponding polarization curves, and the corrosion in the first pitting corrosion potential will not cause the large pitting pits, instead, the homogeneously distributed and thin layer of Al(OH) 3 will form on the surface (the thickness is about 120 nm). This thin layer can prevent the further development of the pitting pit. The second pitting potential is the stable growing potential for the pitting pits, which will not only result in intergranular corrosion, but also cause great damage to the crystal (possibly owing to the large secondary phase) [25]. Therefore, in this study, it can be considered that the first pitting potential corresponds to the destruction of the surface oxide film. Due to the uniform and similar consistency of all samples, the first pitting potential also shows a close value. The second pitting potential obviously shifts to the lower potential with the increase of thermal simulation temperature, which is likely to be related to the precipitation of η phase. With the precipitation and coarsening of the η phase, the difference in potential between the η phase and the matrix increases, which accelerates the progress of pitting corrosion. At the same time, the self-corrosion potential first decreases and then increases, which can be understood as with the increase of the thermal simulation temperature, the number of the dissolved η phase increases at the beginning, so the  content of Zn and other elements in the matrix also has an increase. These low-potential alloy elements can reduce the overall self-corrosion potential. With the further increase of the thermal simulation temperature, the precipitation of η phase is enhanced, so the content of alloying elements in the aluminum matrix decreases, and the self-corrosion potential of the entire matrix rise. It is particularly noted that the 352°C samples have the lowest self-corrosion potential, while the repair potential exceeds the self-corrosion potential. This means that at this temperature, the pitting pits formed on the surface can be well repaired to avoid its deep development. This is also well proved by the pitting morphology as figure 3 showed. When the peak temperature of the thermal cycle is low, the distribution of precipitates is relatively uniform, so the number of pits is large and densely connected into pieces. With the peak temperature of the thermal cycle increasing, the η' phase dissolves and the η phase precipitates. Large-size precipitates reduce the number of galvanic corrosion pitting, resulting in the appearance of large-size and relatively few numbers of corrosion pits. With a further increase in temperature, the η phase also reaches the re-dissolution temperature. However, due to the short heating time of the thermal cycle, only a short peak temperature is not enough to promote all solid solutions, and only the enrichment of alloy elements centered on the η phase. The alloy elements are often aggregated around the grain boundaries, so the corrosion will develop along the grain boundaries to the deep position, as shown in the morphology of corrosion pits at large magnification. The large size of the precipitates reduces the number of galvanic corrosion pits, so the large corrosion pits show a relatively small number. As the maximum temperature continues to rise, the grain boundary precipitates were further coarsened, and when corrosion occurs, they will develop to the depths along the grain boundaries.

Conclusion
In this paper, the welding thermal simulations were performed to simulate the HAZ and TMAZ of the FSW 7075-T6 aluminum alloy, and the corrosion resistance and microstructure of the simulated samples were analyzed. From the results and discussion, it can be concluded that the hardness of different regions in TMAZ varies in a wide range, but the variation of hardness in the HAZ is relatively small. The corrosion potential of the HAZ simulated samples fluctuates greatly with the variation of the maximum temperature, which may correspond to the dissolution and coarsening of precipitates under different thermal cycles. The TMAZ simulated sample shows the lowest hardness when the maximum temperature of the thermal simulation cycle is 388°C. With the maximum temperature increasing, the corrosion potential of the TMAZ simulated sample decreases first and then increases, and the pitting corrosion also becomes more serious. The corrosion mechanism of EIS further indicates that the corrosion mechanism of the HAZ and TMAZ is completely inconsistent, which is mainly related to the differences in precipitate and grain characteristics.