Corrosion Behavior of Friction Stir Welded 2A97 Al-Cu-Li alloy

For weight saving, welding is used to replace riveting and fastening for the structure parts in the aerospace industry. Friction stir welding (FSW) is of significant interest to the industry as the automated process is capable of producing reproducible, high-quality welds with aluminum alloys.^1-8  During friction stir welding, relatively high-temperature and high-plastic deformation are introduced to the alloy due to the frictional interaction between the welding tool and the work piece, resulting in modified microstructures in the weld, namely, heat affected zone (HAZ), thermomechanically affected zone (TMAZ), and nugget.^1,4  The modified microstructure may affect the corrosion property of the welded alloys.^3,7,9-11  A previous study of friction stir welded AA2024 alloy reveals that HAZ has higher susceptibility to intergranular corrosion than the parent alloy and other welding zones due to the formation of S/S’ phase precipitates along the grain boundaries in HAZ.⁷  By contrast, the edge region of TMAZ in friction stir welded AA7108 alloy is more susceptible to intergranular corrosion than HAZ because of the non-uniform distribution of η/η′ phase precipitates within the edge region of TMAZ.² 

Al-Cu-Li alloys have been increasingly used for structure parts in aircrafts, such as floor beams, fuselage skin, and wing stringers, in order to achieve further weight reduction. However, the presence of lithium may increase the corrosion susceptibility of the alloys. Thus, extensive attention has been attracted to the study of corrosion behavior of the alloys and their FSW welds.^6,10-30  It is reported that the corrosion behavior of the alloys is closely associated with the distribution of the strengthening precipitates, such as T₁ (Al₂CuLi), δ′ (Al₃Li), θ′ (Al₂Cu), T₂ (Al₅CuLi₃), and T_B (Al₇Cu₄Li) phases precipitates.^18-20,29-31  Buchheit, et al.,²³  investigated the electrochemical behavior of T₁ intermetallic compound and found that T₁ is a relatively active phase compared to the alloy matrix. They suggested that T₁ phase tends to form on subgrain boundaries and matrix dislocations in artificially aged AA 2090 alloy and that subgrain boundary attack occurs by preferential dissolution of the T₁ phase when the alloy is exposed to corrosive environment. Li, et al.,²²  found that T₁ and T₂ phases are initially anodic with respect to the alloy matrix, leading to their preferential dissolution. They further suggested that during the dissolution process, lithium is selectively dissolved and noble element Cu enriches, resulting in that the potential of the Cu-rich remnant shifts to positive direction; subsequently, the remnants become cathodic with respect to its periphery, leading to the anodic dissolution of the alloy matrix in the periphery.

Li, et al.,²⁹  also investigated the effect of thermal aging and addition of zinc on the intergranular corrosion (IGC) susceptibility of Al-2.7Cu-1.7Li-0.3Mg alloys in an NaCl solution. They found that IGC susceptibility increases initially then decreases with aging time, with the under-aged temper just prior to peak-aging showing the highest susceptibility and that the IGC susceptibility is associated with the distribution of T₁ and T₂ phase precipitates at the grain boundaries and within grain interior. They also observed that small Zn addition enhances the IGC resistance of the Al-Cu-Li-Mg alloys as Zn is incorporated into grain boundary T₁ precipitate, substituting Zn for Cu in T₁ and forming Al₂(CuZn)Li, therefore changing the electrochemical property of T₁ precipitate.²⁹  Proton, et al.,¹⁸  also investigated the influence of thermal ageing on the corrosion behavior of AA2050 alloy. They found that the alloy in T3 temper is susceptible to intergranular corrosion and that after ageing at 155°C for 30 h, the alloy becomes susceptible to intragranular corrosion.

Considering precipitate volume fractions at both grain boundary and matrix would be expected to increase monotonically with aging, Connolly, et al.,^26-27  suggested that the T₁ phase precipitates dissolution mechanism is not the sole cause for the susceptibility for both under-aged and over-aged tempers, which is more likely due to a combination of different microstructural features.

The authors’ recent investigation¹⁹  on the corrosion behavior of 2A97 Al-Cu-Li alloy in T3 temper during immersion in 3.5 wt% NaCl revealed that localized corrosion occurs preferentially along the sub-grain boundaries. The local plastic deformation in the periphery of the subgrain boundaries plays a more decisive role in corrosion propagation along the boundaries compared to T₁ phase precipitates at the boundaries. It is also revealed that compared with the grain boundaries decorated by T₁ phase precipitates, the grain boundaries decorated by T_B phase precipitates shows a relatively low susceptibility to corrosion. Ma, et al.,²⁸  investigated the influence of thermomechanical history on the corrosion behavior of AA2099 alloy and found that the alloy in T8 temper has higher susceptibility to severe localized corrosion than T3, T4, and T6 tempers, related to the localized plastic deformation introduced during pre-aging cold working and heterogeneous precipitation of T₁ (Al₂CuLi) phase during subsequent aging. Further, they found that the difference in the distribution of alloying elements in the alloy influences the mechanism of localized corrosion propagation. For the alloy in T3 and T4 conditions, the alloying elements are largely in the solid solution, resulting in crystallographic corrosion; while the alloy in T6 and T8 conditions exhibit preferential dissolution of T₁ precipitates at the grain boundaries and within the grain interior.

However, as described above, despite the extensive research on correlating the corrosion behavior of the Al-Cu-Li alloy with the thermosmechanical history and the consequent microstructure, the role of the microstructural features, such as precipitate distribution, in corrosion initiation and propagation in the alloys still remains inadequately understood. Further, as revealed by Decreus, et al.,³²  the formation of strengthening precipitates in AA2198 Al-Cu-Li alloy is strongly sensitive to cold working performed before thermal ageing. As the various weld zones are subjected to different degrees of thermal transient and plastic deformation during friction stir welding, the distribution of the precipitates in the various weld zones could be significantly complicated and, consequently, their corrosion behavior may be significantly different from the parent alloy.^2-3,10-11 

Concerning the corrosion behavior of FSW Al-Cu-Li alloys, there exist observations that are different from each other, perhaps depending on the welding parameters, the composition of the parent alloys and their thermomechanical history. Corral, et al.,³³  suggest the corrosion behavior of the various zones of the welded AA2195 alloy in 3.5% NaCl is the same from a macroscopic point of view. The recent work by Proton, et al.,^10-11  on friction stir welded 2050 alloy reveals that the HAZ and the nugget are susceptible to intergranular corrosion due to the formation of T₁ phase precipitates at the grain boundaries during FSW. In the present study, the microstructure and the corrosion behavior of friction stir welded 2A97 Al-Cu-Li alloy are investigated in order to advance the understanding of microstructure modification in the various welding zones and its influence on the corrosion mechanism of this new Al-Cu-Li alloy.

2A97-T3 Al-Cu-Li alloy (Li 1.5 wt%; Cu 3.7 wt%; Zn 0.5 wt%; Mg 0.3 wt%; Mn 0.3 wt%; Zr 0.1 wt%; Al rem.) sheets of 1-mm thickness was welded with a typical friction stir welding process. The weld specimens were successively ground with silicon carbide papers to 4000 grit and polished sequentially using 3-μm and 1-μm diamond pastes. The specimens were then cleaned ultrasonically in an acetone bath and dried in a cool air stream. For microstructure characterization, the mechanically polished specimens were electropolished using a mixture of 800-ml ethanol and 200-ml perchloric acid at the temperature of 7°C. For optical microscopy of the weld, the electropolished specimens were further anodized with Barker’s reagent (1.8 vol% fluoroboric acid) at 20 V for 2 min to reveal the grain structure.

To evaluate the corrosion behavior of the weld, the mechanically polished specimens were immersed in a 3.5 wt% NaCl solution at ambient temperature for 22 h. After the immersion, the specimens were rinsed with deionized water and dried in a cool air stream. The open circuit potentials (OCP) of the different weld zones were also measured in a 3.5 wt% NaCl solution. As the weld zones are narrow bands parallel to the welding line, the samples for OCP measurement were masked using beeswax to define a working area of 2 × 20 mm² with the length parallel to the welding line. Five measurements were performed for each of the zones to ensure reproducibility.

A diamond knife on a Leica Ultracut ultramicrotome was used to generate cross sections at specific corrosion sites for scanning electron microscopy (SEM). Further, focused ion beam (FIB) was used to produce electron transparent foils from the as-welded alloy and corrosion tested welds, which were then examined by transmission Kikuchi diffraction (TKD) and transmission electron microscopy (TEM) operating at 200 kV along with energy dispersive x-ray (EDX) analysis.

Figure 1 shows the optical micrographs of friction stir welded 2A97 Al-Cu-Li alloy surface after anodizing in Barker’s reagent, revealing four distinctive zones, namely, parent alloy (PA), HAZ, TMAZ, and nugget. The parent alloy exhibits elongated grains as expected for the rolled thin sheet (Figure 1[b]). Within some of the grains, subgrains are illustrated by the slight variation in color and brightness (as anodizing in Barker’s reagent enhances the variation of optical appearance caused by the variation in grain orientation) in the optical image, indicating a partially recrystallized grain structure. This was confirmed by electron backscatter diffraction (EBSD) in a previous work.¹⁹  Next to PA, HAZ (Figure 1[c]) shows grains with decreased number of subgrains compared with the initial grain structure of the parent alloy. Figure 1(d) reveals a severely deformed grain structure within TMAZ, consisting of grains with typical sizes of hundreds of micrometers and fine subgrains with typical sizes of micrometers. Finally, the nugget region (Figure 1[e]) displays equiaxed fine grains with typical dimensions ranging from 3 μm to 10 μm, significantly finer than that of the grains in the parent alloy, indicating that dynamic recrystallization has occurred in the nugget during the welding process.

The optical micrographs of the cross section of the welded alloy are displayed in Figure 2. As marked in Figure 2(a), similar to the weld surface (Figure 1), four distinctive zones are evident, including nugget, TMAZ, HAZ, and PA. The TMAZ widens from the bottom surface to the top surface in a slightly asymmetric shape (as marked with white-dashed lines), which is closely associated with the difference in the shear strains between the advancing and the retreating sides of the weld² . The parent alloy, as shown in Figure 2(b), exhibits elongated, partially recrystallized grain structure. Further, the interface between TMAZ and HAZ is displayed in Figure 2(c), revealing significant difference in grain structure between the two zones, consistent with that shown in Figure 1. Figure 2(d) displays the cross section of nugget, exhibiting relatively fine, equiaxed grains with typical dimensions of approximately 3 μm to 10 μm, consistent with that on the top surface (Figure 1).

In addition to the modification to the grain structure, the distribution of precipitates in the various weld zones is also significantly affected by FSW, as shown in Figure 3. The backscattered electron micrographs of parent alloy are displayed in Figures 3(a) through (c). The channeling contrast between grains of different orientations reveals its partially recrystallized grain structure (Figure 3[a]), which is consistent with that shown in Figure 1. It is evident that most grain boundaries remain clean without precipitates (Figure 3[b]). Typical grain boundary precipitates in PA are marked with the white arrows in Figure 3(c). The needle-shaped precipitates with typical dimensions of hundreds of nanometers that are discretely distributed along the grain boundary in parent alloy are either T₁ (Al₂CuLi) or T_B (Al₇Cu₄Li) phases, which were characterized using high-resolution TEM and electron energy loss spectroscopy previously.¹⁹  Figures 3(d) through (f) display backscattered electron micrographs of HAZ. Partially recrystallized grain structure is still evident. Scrutiny of Figure 3(d) reveals that, within the area consisting of subgrains, two relatively coarse grains (marked with white arrows) are present, suggesting that further recrystallization has occurred in the HAZ during FSW. Compared to PA, an increased population density of grain boundary precipitates is observed (as indicated by the white arrows in Figures 3[e] through [f]), suggesting that the high temperature thermal transient introduced by FSW process within HAZ stimulates the formation of the precipitates at the grain boundaries. By contrast, significantly different grain structure is exhibited in TMAZ, which displays severely deformed grains, as shown in Figure 3(g). The non-uniform distribution of grain boundary precipitates in TMAZ is evident (Figures 3[h] through [i]). Some grain boundaries are decorated with almost continuously distributed precipitates as shown in Figure 3(i). A high-population density of fine needle-shaped precipitates is also evident within the grain interior (Figure 3[i]). Further, due to the selective dissolution of the precipitates during specimen preparation using electropolishing, some precipitates were preferentially removed from the alloy surface, resulting in cavities that appear as dark spots (Figures 3[h] through [i]). In the nugget region, the fine, equiaxed grains are observed (Figures 3[j] through [l]), consistent with the optical micrographs shown in Figures 1 and 2. Compared to TMAZ, the population densities of the precipitates in the grain interior and at the grain boundaries are significantly reduced.

During FSW, the weld experiences significant shear strain and a short-term, high-temperature thermal transient associated with frictional heating. Consequently, the microstructure within the various weld zones has been modified significantly, as shown in Figures 1 through 3. As the central nugget experiences the highest temperature that can be over 400°C¹  and the most severe plastic deformation, dynamic recrystallization occurs, leading to the formation of fine equiaxed grains in the nugget. Next to the nugget, the combined effect of relatively high temperature thermal transient and high shear strain results in a heavily deformed grain structure characterized by grains that are elongated in the alloy flowing direction during FSW. By contrast, HAZ, which experiences relatively small temperature rise, displays slightly coarser grains in comparison with parent alloy. In addition to the changes of grain structure, the distribution of precipitates also exhibits significant difference within the various weld zones. The fine needle-shaped precipitates observed in the grain interior and at the grain boundaries within the TMAZ are initially absent in the parent alloy, suggesting that the significant plastic deformation, high temperature thermal transient and the relatively slow cooling rate associated with the air cooling of the weld during FSW has promoted the formation of the precipitates in the TMAZ.

In order to identify the precipitates in TMAZ, electron transparent foils generated from TMAZ by FIB was examined using TEM. Figures 4(a) through (b) display a bright field TEM micrograph and the corresponding selected area diffraction pattern in [011]_Al zone axis. High-population densities of needle-shaped and spherical fine precipitates are revealed in the bright field TEM micrograph. Three sets of diffraction spots are evident in Figure 4(b), including that from aluminum matrix, the superlattice spots (marked with solid-line circles) from δ′ (Al₃Li) phase, and the superlattice spots at 1/3 (001)_Al (marked with dashed-line circles) from T₁ phase.^20-21  As T₁ phase has {111}_Al as the habit plane, the needle-shaped precipitates, indicated with dashed-line arrows in Figure 4(a), are T₁ phase. Thus, the spherical precipitates, marked with solid-line arrows in Figure 4(a), are δ′ phase.

A typical high angle grain boundary (HAGB, θ > 15°) in TMAZ is displayed in the high angle annular dark field (HAADF) micrograph of Figure 4(c). Evidently, the HAGB is decorated by almost continuously distributed precipitates (indicated with the solid-line arrow). By contrast, the two low angle grain boundaries (LAGB, 1° < θ < 15°) revealed in Figure 4(c) are free of precipitates (indicated with dashed-line arrows). In Figure 4(d), it is evident that some LAGBs (indicated with dashed-line arrows) are decorated by discretely-distributed fine precipitates. A large number of HAGBs/LAGBs were examined, indicating that some LAGBs are decorated by discretely-distributed fine precipitates while precipitates are absent from other LAGBs; by contrast, HAGBs are decorated by almost continuously distributed precipitates. Thus, Figure 4 shows the typical distribution of the precipitates in the TMAZ.

The open circuit potential of the various weld zones in 3.5 wt% NaCl solution are displayed in Figure 5 (the locations of measurement are indicated with the numbers 1-5 in Figure 1[a]). Five measurements were performed for each of the zones to ensure reproducibility. The slight variation between the five measurements for each point is indicated with the error bar in Figure 5. The TMAZ has the most negative OCP at −680 mV_SCE. The nugget shows an OCP at around −650 mV (SCE). By contrast, the OCP of parent alloy is approximately −600 mV (SCE). However, as the HAZ is very narrow, it was difficult to locate precisely the HAZ for OCP measurement. Therefore, OCP measurement of HAZ was not attempted. It is also worth noting that both locations 1 and 5 are parent alloy, and therefore, should have the same OCP value. The slightly different OCP values for locations 1 and 5 are within the measurement variation range. Further, locations 2 and 4 also exhibit slightly different OCP values, which is again within the measurement variation range, but might also be due to the slightly different microstructure between the advancing and the retreating sides of TMAZ as the shear strains introduced in the advancing and the retreating sides of the weld are different.^2,4  As revealed in Figures 3 and 4, a high-population density of T₁ phase precipitates is present in the TMAZ, within the grain interior and at the grain boundaries. As T₁ phase has a more negative electrode potential than aluminum,^22-23  the TMAZ shows the most negative potential compared with other weld zones. Hence, it is expected that the TMAZ is the most susceptible to corrosion compared with other welding zones and the parent alloy. This is confirmed by the corrosion testing results shown in Figure 6.

The optical micrograph of the welded alloy surface after immersion in a 3.5 wt% NaCl solution for 22 h is displayed in Figure 6. Evidently, the TMAZ edges (indicated by the dashed lines) adjacent to HAZ are severely attacked, indicating its relatively high susceptibility to corrosion, consistent with the OCP profile of the weld (Figure 5). The tested alloy surface was gently polished with 1-μm diamond paste in order to remove the corrosion product to allow SEM examination of the corrosion morphology beneath the corrosion product. Figure 7(a) displays the SEM micrograph of a severe localized corrosion site in the edge region of the TMAZ. An open pit of approximately 60-μm width and 150-μm length is revealed. Two cross sections were obtained using ultramicrotomy from the positions indicated as A-A and B-B in Figure 7(a). Figure 7(b) displays the scanning electron micrograph of the cross section along A-A, exhibiting a pit that had developed to a depth of approximately 80 μm beneath the weld surface. Interestingly, the propagation of the pit follows the direction indicated by the dashed-line arrow, suggesting that the localized corrosion propagated along the TMAZ edges adjacent to HAZ (as indicated by the dashed line in Figure 2[a]). The framed area in Figure 7(b) is displayed in Figure 7(c) at increased magnification. Narrow dark lines, namely attacked grain boundaries, are evident, indicating the high intergranular corrosion susceptibility of the TMAZ region. Figure 7(d) illustrates the framed area in Figure 7(c) at increased magnification. Consistent with Figures 3 and 4, a high-population density of fine needle-shaped precipitates is evident within the grain interior. Further, residual precipitates are also evident along the attacked grain boundaries, as indicated with solid-line arrows. Ahead of attacked grain boundaries, intact precipitates are evident (marked with dashed-line arrows), suggesting that corrosion propagated preferentially along the grain boundaries decorated by the almost continuously distributed T₁ phase precipitates.

The cross section at the position B-B indicated in Figure 7(a) is displayed in Figure 8(a), exhibiting corrosion morphology similar with that at the position A-A. The framed area in Figure 8(a) is shown in Figure 8(b) at increased magnification, again, exhibiting attacked grain boundaries in the corrosion front area. Figure 8(c) shows the framed area in Figure 8(b) at increased magnification. Interestingly, needle-shaped pits with typical length of hundreds of nanometers and width of a few nanometers are present in the periphery of attacked grain boundary (marked with white arrows). Comparing the morphology of needle-shaped pits with the T₁ phase precipitates in their periphery, it is evident that the pits are resulted from the dissolution of the needle-shaped T₁ phase precipitates. Li, et al.,²²  observed that during the dissolution process of T₁ phase, lithium is preferentially dissolved, resulting in Cu-rich remnant. The remnant is cathodic with respect to its periphery and, which results in anodic dissolution of the alloy matrix in the periphery and, subsequently, the loss of remnant to testing solution. Thus, corrosion develops from intergranular corrosion to the dissolution of the grain interior. Further, it is evident that a high-population density of T₁ phase precipitates is present in the grain interior. Consequently, dissolution of grain interior leads to the removal of individual grains and the open pit shown in Figure 8(a).

To further analyze the intergranular corrosion development in TMAZ, electron transparent foils are obtained from the cross section of the corrosion front of a localized corrosion site for scanning transmission electron microscopy (STEM) and TKD analysis. Figure 9(a) displays a HAADF micrograph of a localized corrosion site, revealing the attacked grain boundaries. The corresponding grain orientation map in inverse pole figure coloring is displayed in Figure 9(b), with the numbers indicating the corresponding grain boundary misorientations. Comparing the corrosion morphology (Figure 9[a]) with the grain orientation distribution (Figure 9[b]), it is evident that most LAGBs remain intact while HAGBs have been preferentially attacked. Hence, the high angle grain boundaries are more susceptible to corrosion than the low angle grain boundaries.

Further, the framed area in Figure 9(a) is shown in Figure 9(c) at an increased magnification. Three intact LAGBs are indicated with solid-line arrows, where precipitate is absent. In contrast, along the intact parts of the HAGBs, as indicated with dashed-line arrows, precipitates are evident. Hence, along HAGBs, the micro-coupling between the anodically active T₁ phase precipitates and the peripheral alloy matrix leads to the preferential dissolution of T₁ phases precipitates and, consequently, intergranular corrosion in TMAZ. This observation has advanced the authors’ recent work¹⁹  on the corrosion behavior of 2A97 Al-Cu-Li alloy in T3 temper during immersion in 3.5 wt% NaCl. It was revealed that when only a low-population density of discrete T₁ pricipitates is present at grain boundaries of the alloy in T3 temper, the dense dislocation walls associated with the subgrain boundaries play a more decisive role in corrosion propagation along the boundaries compared to T₁ phase precipitates at the boundaries. However, as revealed here, localized corrosion preferentially occurs at the HAGBs in the TMAZ due to the presence of a high-population density of T₁ phase precipitates.

• During friction stir welding of 2A97-T3 Al-Cu-Li alloy, the weld is sensitized through the formation of a high-population density of T₁ phase (Al₂CuLi) precipitates in TMAZ, due to the significant plastic deformation, high temperature thermal transient, and the relatively slow cooling rate associated with air cooling of the weld during FSW.

• The TMAZ edges adjacent to HAZ have the most negative open circuit potential compared with other welding zones and are susceptible to localized corrosion whereas other welding zones are more resistant to corrosion.

• Localized corrosion in TMAZ initially occurs at HAGBs due to the presence of T₁ phase precipitates, whereas the LAGBs remain intact due to the absence of T₁ phase precipitate. Subsequently, corrosion develops into grain interior through preferential dissolution of T₁ phase precipitates, where a high-population density of T₁ phase precipitates is also formed during FSW.

The authors wish to thank the UK Engineering and Physical Sciences Research Council for support of the LATEST2 Programme Grant (EP/H020047/1). China Scholarship Council is also thanked for provision of financial support.

All research data supporting this publication are directly available within this publication.