Mechanical behavior associated with metallurgical aspects of friction stir welded Al–Li alloy exposed to exfoliation corrosion test

This research aims to investigate the effect of the exfoliation corrosion exposure time on the mechanical properties, the strength and elongation, of friction stir welded Al–Li alloy type 2195-T8. The exfoliation corrosion test was performed using the exfoliation corrosion (EXCO) solution, based on ASTM G34. The samples were exposed to different exposure times 24, 48, 96, 192 and 384 h. The results showed that both the strength and elongation of the welded specimens—exposed to the exfoliation corrosion tests—were reduced. For example, the samples that were exposed to 384 h, their initial tensile strength and elongation were reduced by 13% and 17% respectively. The degradation process due to the exfoliation corrosion on the tensile strength was divided into three stages: fast (0–96 h: 443.7 MPa-416.3 MPa, the degraded rate was 0.29 MPa/h), steady (96–192 h: 416.3 MPa-413.4 MPa, the degraded rate was 0.03 MPa h−1) and medium rate (192–384 h: 413.4 MPa-386.7 MPa, the degraded rate was 0.14 MPa h−1). For the elongation, in general, the reduction was similar to the style of strength, but with different rates. The TEM images showed that this degradation was due to the dissolution of T1 (Al2CuLi) and S′ (Al2CuMg) phase. Also, the corrosion products and their role of adherence on the surface of the tested specimens were investigated. An empirical equation p-t (mechanical properties-exposure time) was established to calculate the effect of exposure corrosion time on the performance of welded specimens.


Introduction
In the last two decades, the Al-Li alloys have become widely used due to their good properties, reducing the weight structure by 10%-15% and increasing the stiffness to 15%-20% compared with traditional Al alloys, thereby becoming a candidate with great promise to replace other Al alloys [1][2][3][4][5]. Joining of aluminum alloys is a major process in many engineering applications, for example in aerospace. However, welding in general, and the fusion welding-based processes in particular, to some extends will limit the performance of Al-Li alloys, as it causes defects such as hot cracks, porosity, softening, difficultly removed oxide film, etc.
The friction stir welding FSW process noticeably reduces the defects that associate with the fusion weldingbased methods. FSW is a solid-state, autogenously process without melting [6][7][8] and is extensively applied in joining Al-Li alloys [9,10]. In general, the welded parts by using FSW method are divided into a base material (BM), heat affected zone (HAZ), thermal-mechanical affect zone (TMAZ) and weld nugget zone (WNZ) [10], thus each zone shows different microstructure [11][12][13]. The precipitates in the BM primarily consist of T 1 (Al 2 CuLi) and θ´(Al 2 Cu) phases [11,14,15], while in HAZ the phases are δ′ (Al 3 Li) [14,[16][17][18][19], while there are reports show S′ (Al 2 CuMg) phase in HAZ [10,[20][21][22]. In terms of WNZ, which is affected by both thermomechanical and stirring, both δ′ and β′ (Al 3 Zr) are the major strengthening phases [14,16,23,24]. Due to these microstructural changes, many researchers have studied the corrosion performance of welded alloys [25][26][27][28]. Liu et al [22] found that the high proportion of T 1 phase reduced the corrosion sensitivity of the 2198 Al-Li alloy. Ma et al [29] reported that the dissolution of T 1 phase in FSW joint made the corrosion degree lower Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. than BM in 2099 Al-Li alloy. Others [20,30] found that the corrosion potential of θ´phase was higher than the matrix and was often acted as a cathode to promote the corrosion. Ren et al [20] revealed two corrosion mechanisms of S′ phase, they found that pitting occurs around the S′ phase. Bousquet et al [21] found out that HAZ was more sensitive to intergranular corrosion than the BM and other welding areas due to S/S′ phase precipitation along HAZ grain boundaries in welded alloys. Yi et al [31] showed that the effect of δ′ and β′ phase on local corrosion was relatively small, while the uniform corrosion was the main phenomenon. The microstructural characterization and mechanical properties of Al-Li alloy with FSW joints were studied by many researchers [32][33][34], however, the major concerns remained was about the corrosion behavior of the alloy. Li et al [35] and Huang et al [36] studied the intergranular corrosion behaviors of 2195 Al-Li alloy and found that the intergranular corrosion of the alloy was related to T 1 phase. Luo et al [37] and Li et al [38] studied the corrosion behaviors of Al-Li alloys in sodium chloride solution, they found that general corrosion and localized corrosion of intergranular and pitting corrosion in the alloy.
There are some researchers have studied the influence of corrosive environment on the properties of aluminum alloys. Qiu et al [39] studied the effect of corrosion on the mechanical parameters of aluminum alloy type 2024-T4, using ASTM G34 [40]. The results showed that the tensile strength and fatigue life are exponentially related to the average value of the corrosion rate. Liao et al [41] found slightly different results using the fitting equation. Wang [42] indicated that the performance of 7075 alloys declined rapidly after the first two cycles (one cycle was 48 h) against the marine atmospheric corrosion, while the decline rate of performance was relatively gentle with the extension of the test time. Khoshnaw [43] found that the exfoliation corrosion degree for 2024-T3 increased with increasing the aging time while decreased for 7075-T6. The results were attributed to the type of precipitants in both alloys. Ma et al [44] demonstrated that the pre-corrosion damage and saline environment had significant effects on the detailed fatigue performance of 7xxx series aluminum alloy, which was consistent with the research of Lu et al [45]. Su [46] compared the corrosion behavior and mechanical properties of 2D12 aluminum alloy and 5A90 Al-Li alloy by using marine environmental exposure. The results showed that the corrosion process of the 5A90 Al-Li alloy did not follow the law of power function, and the mechanical damage caused by corrosion was significantly greater than that of 2D12 alloy.
Since the mechanical and corrosion properties of alloys strongly depend on the microstructure, the type, size and distribution of the precipitates. Therefore, further studies on the precipitated phases will help to explain the influence of corrosion on the microstructure of Al-Li alloys. The phases, such as T 1 , as the main strengthening phase of Al-Li alloy, might strongly affect the corrosion performance of the alloy. Li et al [35] and Huang et al [36], both showed that T 1 was closely related to the generation of intergranular corrosion, and Su et al [46] confirmed the same outcomes. Luo et al [37] found that discontinuous localized corrosion in Al-Li alloy is associated with corroded θ′ (Al 2 Cu) particles, as well as pits that are formed by particle fall-out due to dissolution of surrounding Al substrate. The exfoliation corrosion leads to the delamination of thin metal layers, thus the strength and ductility of the welded alloys would be affected, consequently posed a serious threat to the safety of the structural parts [40].
Although there are studies have investigated the effect of corrosion, in general, on different Al alloys. There are still gaps in the effect of specific types of corrosion individually on the mechanical properties of welded Al-Li alloys. Therefore, this research aims to investigate the influence of the exfoliation corrosion on the mechanical properties of friction stir welded Al-Li alloy type 2195-T8.

Materials and experimental works
Plates, 8 mm thick, of Al-Li alloys type 2195-T8, were used in this study. The chemical composition analysis of the alloy was provided by Southwest Aluminum Corporation (Group) Co., Ltd. Table 1 shows the chemical composition of the alloy. The plates were cut into 100×650 mm, and every two pieces of the cut plates were welded by a numerical control Gantry type friction stir welding machine. The rotation speed and the welding speed were 800 rpm and 50 mm min −1 respectively.
The tensile test specimens, with specific dimensions, see figure 1, were prepared, from the welded plates. The loading direction on the tensile specimens was perpendicular to the welded line, based on standard GB/ 228.1-2010. Exfoliation corrosion (EXCO) was carried out based on standard ASTM G34 [40], using EXCO solution, which consists of 234 g NaCl+50 g KNO 3 +6.3 mL HNO 3 /L. The standard ASTM G34 is suitable to be used for as received and welded specimen and widely used to study exfoliation corrosion of Al alloy [35,39,40,43]. This standard provides a prediction exfoliation corrosion behavior of alloys in various types of outdoor environments, especially in marine and industrial environments [40].
The ratio of the volume of the corrosive solution (V m ) to the exposed specimen surface area (S s ) is V m : S s =15 mL cm −2 . After rinsing and drying, the specimens were placed in EXCO solution at 25±3°C. The welded specimens were exposed to exfoliation corrosion for 24, 48, 96, 192 and 384 h. To enhance the reliability of results, each test was repeated three times. Moreover, the corrosive solution was replaced every 96 h to minimize the corrosion byproducts effects. Figure 2 shows a welded specimen, the gauge length completely immerses in the EXCO solution, one end fixed in the middle of the bottom surface of the container and another end vertically towards the top.
The specimens were exposed to exfoliation tests, rinsed and dried. Then the samples were exposed to tension stresses using universal testing machine type WDJ-3008, the extension speed was 1 mm min −1 .
The corrosion morphology was observed by SU8020 scanning electron microscope (SEM). The microstructural changes of the joints of different welded zones were characterized by transmission electron microscopy (TEM), type FEI Tecnai G2 F20, operated at 200 kV. The specimens were prepared by twin-jet electro-polishing using a mixed solution of 30% HNO 3 +70% CH 3 OH at −30°C, as it was carried out in the liquid nitrogen, which helps to control the polishing speed and preventing oxidation.   Figure 3 shows the corrosion morphology under SEM which would be helpful to investigate the EXCO corrosion process. Figure 3(a) shows the surface morphology of the unexposed to corrosion sample, where a large number of white precipitates were observed. Energy Dispersive Spectroscopy-EDS-results indicated as a point in the image associated with a graph, showed that the precipitation phase containing Al and Cu. Figure 3(b) shows that there are many corrosion pits after 24 h EXCO corrosion, which indicates that the corrosion phenomenon originates from pitting, even in EXCO tests in strong acid solutions. Tang et al [47] and Ding et al [48], reported that the corrosion of aluminum alloy in EXCO solution was mainly initiated by pitting, and finally developed into exfoliation corrosion. With time, the corrosion products accumulate continuously, as shown in figures 3(c) and (d). After removing the corrosion products by an external force, it was found, see figure 3(e), that the pits were expanding and forming connections. Figure 3(f) shows a surface morphology of the sample exposed to corrosion for 96 h, as a lot of 'white filiform products' are shown and distributed at the bottom of the pits, caused by the continuous reaction of exposed new surfaces during the corrosion process [43]. Figures 3(g) and (h), show more corrosion products accumulated on the surface after 192 h corrosion, showing a tendency of self-peeling. After removing the corrosion products, it was found that the corrosion was still ongoing under the corrosion pit covered by the corrosion products, and a large number of 'white filiform product' corrosion products adhere to the bottom of the corrosion pits, see figure 3(i). Figure 4 shows the stress-strain diagram of the friction stir welded specimens exposed to exfoliation corrosion for different times. The results showed that with increasing the exposure time, the mechanical properties gradually decreased. For example, with the exposure time equals 24 h, the ultimate tensile strength UTS was 430.9 MPa and the elongation 11.08%. With increasing the exposure time to 384 h, the UTS reduced to 386.7 MPa and the elongation to 9.31%.

Mechanical properties after corrosion
To identify the effect of the exposure time of the exfoliation corrosion on the mechanical properties of the welded specimens, the intensity ratio was calculated, which was defined as the ratio of the UTS of the tested specimens (σ test ) to the original, or unexposed to corrosion sample, UTS (σ o ), denoted as σ test /σ o . Thus, the strength loss rate v sr was defined as the rate at which the intensity ratio σ test /σ o changed over time, t, using equation (1): Similarly, the intensity ratio was calculated for elongation, as a ratio of elongation of the tested specimen (El test ) to the initial elongation (El o ), denoted as El test /El o . The elongation loss rate v er was defined as the rate at which the elongation ratio changed over time, using equation (2): Table 2 shows that both the tensile strength (σ o ) and the elongation (El o ) of the joint deteriorates of different exposure time. Figure 5 shows the changes of the UTS and elongation with the corrosion exposure time. The figure has illustrated these changes in three stages. Figure 5(a) showed that the corrosion had the highest impact on the tensile strength of the joint at stage I (t 96 h), the loss rate of the tensile strength equals 0.29 MPa h −1 . In stage II of corrosion (96 < t 192 h), the loss rate equal to 0.03 MPa h −1 , which can be attributed to the formation of the corrosion products [46], were stuck to the surface of the joint and lowered the strength values. However, in  stage III of corrosion (192 < t 384 h), the tensile strength reduction rate was increased rapidly, and the loss rate was 0.14 MPa h −1 . Figure 5(b) shows the influence of the corrosion exposure time on the elongation. The figure shows, similar to UTS, has divided into three stages, obviously with different rates. In stage I (t 48 h), the joint ductility decreased with the occurrence of corrosion reaction, it was a slow process. In stage II of corrosion (48 < t 96 h), the elongation decreased sharply and the slope of the curve in stage II was the highest, which means that the elongation dropped sharply. In stage III (96 < t 384 h), the elongation of the joint decreased slowly. Figure 6 shows a diagram correlates between mechanical properties loss rate and corrosion exposure time of the joints. The highest UTS reduction rate was equal to 1.21×10 -3 /h at an exposure time of 24 h, then gradually decreased for longer exposure times, the loss rate reduced to 0.33×10 -3 /h for 384 h of corrosion. In terms of the elongation loss rate, the value dropped sharply in the latter part of the early stage of corrosion. In the interval from 48 h to 96 h, the elongation loss rate dropped from 6.67×10 -3 /h to 1.59×10 -3 /h, with a decrease of 319%.

Empirical equation
The results in table 2, UTS and elongation versus exposure time, were integrated -using MATLAB-to establish equations   Sum of Squared Errors-SSE value could judge the degree of data fitting, the smaller of SSE, the better of fitting degree. In statistics, R 2 is the determining coefficient ranges between 0 and 1. Based on the fitting equation of σ-Exposure time-assuming the joint failed when the strength of the joint reaches a specific value after corrosion-the service life of the joint in a solution similar to EXCO solution can be estimated. When R 2 is equal to or close to 1, it indicates that the fitting results are relatively accurate. Based on table 3, the SSE and R 2 in the fitting results of the σ-Exposure time were 3.279 and 0.9979, while in the fitting results of El-Exposure time were 3.972 and 0.9224, which means that the fitting degree of σ-Exposure time was more sufficient. The fitting results of R 2 in the equations (3) and (4) were both close to 1, and the confidence level was both 95%, indicating that the regression fitting result was satisfied.
The equations (3) and (4) were combined to cubic functions and could be expressed as equation (5): were a, b and c are the coefficients related to the equation, and d was a constant of the equation. Equation (5) can be used to calculate the effect of the exfoliation corrosion time on the mechanical properties of welded FSW Al-Li alloy type 2195-T8. Figures 7(a)-(c) show the TEM images of the base metal. The figures show that there are laminar-like T 1 phases arranged in a specific direction in the BM sample, which has been indicated by other researchers [11,14,15]. In addition to the T 1 phase, some θ′ phase was observed. However, T 1 phase, with high electrochemical activity, precipitates preferentially at grain boundaries, sub-grain boundaries and dislocations, are the main places of corrosion selectivity [22,29]. . This is an indication that during the welding process, parts of T 1 precipitates have dissolved and accumulated, while growth of θ′ precipitates occurs in specific regions within HAZ. Besides, the S′ (Al 2 CuMg) phase was observed in HAZ, see figure 7(e), which can be attributed to the S′ has precipitated under the action of heating during welding. The S′ phase, as a vulnerable phase [20,21], acts as anode and preferential corrosion. Both θ′ and S′ phases can be observed in intergranular precipitates, see figures 7(e) and (f), which lead to a further reduction in the corrosion resistance of HAZ.

Discussion
Aluminum, magnesium and lithium are sensitive to highly acidic solutions due to their high electrochemical activity [22,46], they all have negative electrochemical force values, which is also called corrosion potential [31]. Hydrogen ions in the acidic environment (EXCO solution) receive electrons consequently hydrogen released, which acts as a driving force for corrosion occurrence. Figure 4 showed the mechanical properties of the welded Al-Li alloy were negatively interrelated with the exposure time. This could be attributed to the fact that since corrosion induces surface pits with time while it is in contact with a corrosive solution, and it has a profound effect on property degradation of materials, thus the longer the exposure time, the lowest the mechanical properties of the joint [39,44]. Similarly, the elongation decreased with the increase of corrosion time, which was related to a transverse pit/crack defect caused by the applied stress and the decrease of the cross-sectional area of samples. These results are similar to other researches, Qiu [39], Nikolaos et al [49] and Leslie et al [50], reported that such degradation process was mainly due to the reaction between the solution and the surface areas where the phases, T 1 and S′ phase, have existed, which are acting to reduce the effective cross-sectional area, i.e. coherency, and the surface pits caused by corrosion would induce the formation of transverse cracks, consequently, the degradation occurs.  Figure 5 showed the changes of the UTS and elongation with the corrosion exposure time. The figure showed that both mechanical properties, UTS and elongation, were divided into three stages. In correlation with the microstructure analysis in figure 7 and based on other researches studied the influence of microstructure on the corrosion behavior of Al-Li alloy [17][18][19][20][21], the three stages behavior could be explained in a way that, T 1 and S′ were consumed continuously due to the priority of corrosion during the corrosion process [14,15], the continuous corrosion leads to a reduction in UTS values, see figure 5(a). While the tensile strength of the joints becomes lower due to the corrosion phenomenon of the preferential phases, the corrosion byproducts generate, gathered, adhered to the surfaces, thus the early elastic drop appears. The protection of corrosion products for the decline of elongation in the middle stage of corrosion makes stage III of elongation decline tend to be steady.
The three stages in figure 5(a) of the influence of corrosion on the tensile strength of the joint could be summarized as the following three steps: In stage I (t 96 h), while the corrosive solution is in contact with the substrate, corrosion reaction occurs, it first reacted with the corrosive phases, such as T 1 and S′ [16,17] and this affected the tensile strength of the joint, thus the UTS decreased significantly.
In stage II (96 < t 192 h) the reduction rate of UTS was slowed down, which was related to the formation of corrosion products, adhere to surfaces, thus the contact between the corrosive environment and the substrate become limited. Similar approaches have been mentioned by other researchers, Su et al [46], Sun et al [51] and João et al [52] the protective effect of corrosion products on the matrix was reported, the existence of this isolation effect would slow down the corrosion reaction, thus reducing the degradation rate.
In stage III (192 < t 384 h), after a long exposure time, a large amount of corrosion products has been generated and accumulated, but because the amount of insoluble corrosion products was larger than the amount of the corroded metal, this was resulting in 'wedge effect' [20,47,48], see figures 3(g) and (h), which supported by the uncorroded metal in the upper layer and caused layer peeling [39,49], the decreased of the effective cross-sectional area made the tensile strength decline rapidly.
For the changes in elongation, shown in figure 5(b), although it was also divided into three stages, the reduction rates were different than the UTS. This is because the increase of joint ductility by corrosion was mainly related to the corrosion consumption T 1 and S´, and the generation, adhesion, and shedding of corrosion products also played a certain role in the change of joint.
In stage I (t 48 h), the T 1 phase was corroded first and the pits were small, therefore the elongation does not decline very quickly. However, when the corrosion continues to stage II (48 < t 96 h), a large amount of T 1 , surrounded by aluminum matrix were dissolved, then exposed S´phase begins to corrode, and the corrosion holes were interconnected and aggregated to form bigger weak areas. The common influence of the T 1 and Sṕ hases, and the formation of weak areas [39], so the elongation decreases rapidly. In the stage III (96<t384 h), a large number of corrosion products were generated and adhered, filled up some gaps and pits on the surface, which compensated the loss of the cross-sectional area due to previous stages of corrosion [39], leading to a reduction in the rate of the change. Even after the spalling of the specimen, the cross-sectional area decreased, and the new surface was exposed, but the presence of many uncorroded and strengthened phases T 1 , see figure 7, caused a decline of elongation not significant. Figure 7 showed θ′ phase as the main strengthening phase of Al-Li alloy. Because Cu has a high electrochemical force which makes this phase acts as cathode compared to Al substrate and the T 1 phase, leading to the dissolution of the surface of the Al alloy [20,30]. Moreover, S′ phase is prone to corrosion, its quantity was far less than that of T 1 phase, and its corrosion resistance was slightly higher than that of T 1 phase [31]. These are the reasons behind the fact that the reduction of strength and elongation in the corrosion process was not completely linear. On the other hand, both δ′ and β′ phases have low electrochemical activity, i.e. high corrosion potential, but they were both the strengthening phases of Al-Li alloy [23,34]. Therefore, during the corrosion process, the joint performance did not show a completely linear relationship with the corrosion, see Figure 6, due to the consumption of T 1 and S′.

Conclusions
• Exfoliation corrosion acted to reduce the UTS and El% of FSW Al-Li type 2195-T8. The UTS values were decreased at different rates with the corrosion exposure times, 97% within 24 h, 95% in 48 h, 94% for 96 h, 93% for 192 h and 87% in 384 h, compared with the initial UTS equals to 443.7 MPa. While, for the same exposure times, El% was reduced by 98%, 97%, 85%, 84% and 83% comparing with the initial El, was equal to 11.27%.
• The degradation of welded specimens due to the exfoliation corrosion was interrelated with the microstructure, mainly to the dissolution of T 1 phase and S′, generation corrosion byproducts, adhesion with the surfaces and shedding the surfaces affected the change of the joint performance, mainly in stage II and stage III.
• A mathematical equation has been derived to predict the effect of the exfoliation corrosion exposure time on the mechanical properties of the FS welded Al-Li alloy type 2195-T8.