Corrosion Resistance Evaluation and Effects of Prior Corrosion and Stress on Fatigue Behavior of Friction Stir Welded AA2024-T3

Friction stir welding can be used to weld high-strength Al alloys; therefore, this technology is now being applied in many industries, e.g., aeronautical engineering. Compared to the riveted-joint approach, friction stir welding procedures are faster and have a lower fabrication cost. For application to aeronautical engineering, however, many properties of friction stir welded (FSW) joints must be evaluated before they can be using in aircraft primary structures. At present, these structures are typically fastened with riveted joints. Previously, Rhodes, et al., evaluated FSW-formed microstructures, showing that recrystallized small grains and smaller precipitates appear in the stir zone (SZ).¹  Further, Sato and Kokawa observed the microstructures of FSW joints for 6063-T5 (UNS A96063⁽¹⁾) Al alloy.²  The strengths of those joints were evaluated; hence, it was reported that the minimum hardness determined the global yield and ultimate tensile strength. In addition, the fatigue properties of FSW joints were evaluated by Lemmen, et al.,³  Uematsu, et al.,⁴  and Basel, et al.⁵  It was observed that the residual stresses and microstructures formed by the welding affect the fatigue initiation location. Further, Biallas, et al., observed that concentric ripples above or roll scratches below the SZ become fracture origins in as-welded FSW joints.⁶  The effect of the kissing bond was evaluated by Dickerson and Przydatek⁷  and Basel, et al.⁸  In addition, Bussu and Irving⁹  and Fratini, et al.,¹⁰  evaluated fatigue crack growth behavior in 2024-T351 (UNS A92024) Al alloy FSW joints. Similarly, the features of Al-Li alloy FSW joints were evaluated by Ma, et al.,^11-12  whereas Lemmen, et al., observed the fracture surfaces of fatigue crack growth specimens in FSW joints.¹³  It was found that the specimen size also affects the residual stress and fatigue crack growth behavior of an FSW joint. In addition, it was noted that not only the residual stress, but also the yield strength on the weld line affect the fatigue crack growth behavior of the FSW joint. In addition, the authors previously evaluated the fatigue properties and fracture origins of FSW joints,¹⁴  along with the fatigue crack growth behavior under a residual stress field.¹⁵ 

It is well known that corrosion damage can affect the structural integrity of aircraft, and considerable research on the corrosion of base materials has been conducted. Previously, both Hoeppner¹⁶  and Kondo¹⁷  investigated corrosion pit growth behavior in a corrosive environment with cyclic loading. They proposed the pit growth model using a formula based on experimental observation. In addition, Birbilis, et al., identified the composition of the constituent intermetallic particles related to the localized corrosion behavior in 2024-T3 Al alloy and they observed that localized corrosion preferentially occurred at intermetallic sites with lower Cu and higher Si content.¹⁸  Clark and Hoeppner evaluated pit growth behavior that would become the origin of fatigue cracks.¹⁹  They found that the maximum pit depths for 0.063 in sheets were well predicted.¹⁹  Further, Chen, et al., investigated the pit growth behavior and transition from pitting to fatigue crack growth occurring in 2024-T3 Al alloy under a corrosive environment.²⁰  The results of those studies indicated that pit growth often involved coalescence of individual particle-nucleated pits and corrosion fatigue cracks typically nucleated from one or two larger pits.²⁰  In addition, Jones, et al., evaluated the effect of prior corrosion on short crack growth behavior,²¹  along with the pit-to-crack transition and fatigue life of a prior-corroded 2024-T3 Al alloy.²²  They identified that the pit with the greatest depth was not necessarily the origin of fatigue fracture. The pit surface area and surrounding pit proximity affected the fatigue life of the corroded specimen and prior corrosion influenced the propagation of short cracks. Further, van der Walde and Hillberry evaluated the effect of pitting damage on the fatigue life of 2024-T3 Al alloy, and the predicted fatigue life, obtained by considering the pits as initial cracks and evaluating their crack growth, was comparable with the experimental results.²³  Sankaran, et al., evaluated the effect of corrosion pits on the fatigue life of 7075-T6 Al alloy (UNS A97075) through experiments and analysis, considering the average pit size, and found that crack growth lives were comparable with the experimental results.²⁴  Arriscorreta and Hoeppner evaluated the effects of prior corrosion and concomitant corrosion fatigue on the fatigue behavior of 7075-T6 Al alloy.²⁵  It was shown that the applied stress was more detrimental compared to the corrosion time. Finally, Ralston, et al., evaluated corrosion behavior on grain-refined 2024-T3 Al alloy and determined that the grain size was one of the parameters controlling the corrosion damage.²⁶ 

The effects of corrosion on the FSW joint properties must also be well understood before FSW can be used in safety-critical structures. Thus, Yun, et al., previously evaluated the corrosion behavior in the cross sections of FSW joints perpendicular to the FSW direction and found that the increase in Cu concentration at the thermomechanically affected zone (TMAZ)/heat-affected zone (HAZ) boundaries resulted in a decrease in corrosion resistance.²⁷  Finally, Ghidini, et al., evaluated fatigue life in prior-corroded 2024-T3 Al alloy, assuming the initial crack as an ellipse connecting the corrosion pits that became fracture origins and concluded that the predicted fatigue life was comparable with the experimental result.²⁸  However, despite these efforts, research into the corrosion resistance of FSW joint still seems insufficient. For example, the exposed surface of an FSW panel can also be affected by corrosion damage; thus, evaluation of its corrosion resistance is required.

In this paper, the corrosion test results of an FSW 2024-T3 Al alloy joint are presented. The effect of prior-corroded damage on the fatigue life of the joint is also evaluated.

In this study, two types of specimen were prepared, i.e., a 2024-T3 Al alloy base material and FSW joint specimens cut from a single FSW panel made from the base material. Note that the specimens were cut from the panel at least 1 month after the welding, so that the joint mechanical properties were stabilized. The mechanical properties of the base material and FSW joint specimens are listed in Table 1. The specimens were subjected to corrosion and then inspected. These same corroded specimens were then subjected to fatigue testing to evaluate the effect of corrosion damage, with the results also being compared with those from fatigue tests of uncorroded specimens. The specimen preparation and test procedures are described in the following sections.

The FSW panel used in this study was fabricated from the base material by an aircraft manufacturer, with a ratio of FSW tool travel speed to rotation speed of 2.0. The welding direction was perpendicular to the loading direction and the weld line was located at the center of each FSW joint specimen. The rolling direction of the base material was parallel to the welding direction; thus, it was perpendicular to the loading direction. The base material did not have a clad layer, because of the welding process. The sizes of the pin and shoulder in the FSW section were measured as approximately 4 mm and 10 mm, respectively.

Because surface concavities and convexities in the tool mark and burr may induce corrosive agent concentration and generate corrosion damage in addition to that caused by simple corrosive exposure, the surfaces of the FSW joint specimens were ground out using the surface treatment described in ASTM E466. The same surface treatment was applied to the base material specimens so that corrosion damage was induced for the same surface conditions. The base material and FSW joint specimens were 2.0 mm and 1.6 mm thick before and after the surface treatment, respectively.

A 6-mm-wide area was designated in the center of each specimen surface in order to exclude edge effects. The surfaces outside this area were coated with nail varnish to protect them from corrosion. The corrosive medium used in this study was a 3.0 wt% sodium chloride solution at 60°C. Four exposure durations of 24, 48, 72, and 96 h were used to evaluate the effect of the exposure period on the corrosion damage to the base material and FSW joint. Four specimens were exposed to the corrosive medium for each exposure period. However, problem with the temperature control during exposure occurred for one base material specimen exposed for 96 h; thus, the number of base material samples for that case was reduced to three.

After the corrosion treatment, the test specimens were cleaned in an ultrasonic bath of acetone and the overall condition of the corroded surface was recorded by a charge-coupled device camera. The surface was also inspected using a Keyence digital laser microscope KS-1100^† with an LT-9010^† sensor to measure the corrosion morphology, including the corrosion pit size. The resolution of this instrument in the depth direction is 0.01 μm with a spot size of 0.2 μm. Based on the measured results, the in-plane pit geometry was identified. The ellipse circumscribing the in-plane pit geometry was determined and the long and short axes of the ellipse were obtained. In this paper, the ratio of the axis of the ellipse circumscribing the in-plane pit geometry to the pit depth is defined as the aspect ratio of the corrosion pits. The pit volume was calculated as the sum of the multiples of the obtained pit depth at each location and the square of the observed point.

Following the corrosion test, the corroded specimens were fatigue tested to determine the effect of the corrosion pits on the fatigue strength. The dimensions of the prior-corroded fatigue specimens and the fatigue test procedure were in accordance with ASTM E466. The specimen dimensions and the area of prior-corrosion damage are shown in Figure 1. An Instron 8800^† servohydraulic testing system with a 25 kN load range was used for the fatigue test, which was conducted under load control. A sinusoidal waveform was applied, while the stress ratio and test speed were 0.1 Hz and 5 Hz, respectively. The maximum stresses σ_max used for this test were 200 MPa and 250 MPa, respectively. For each fatigue test condition, two out of the four specimens exposed to the corrosive medium for the same duration were used.

A HITACHI S-4700^† field emission scanning electron microscope was used for the fracture surface observation. Transmission electron microscopy (TEM) analysis was also conducted using a JEM-2100F^† TEM. A focused ion beam was used to cut 2 mm² squares from the specimens for observation. In addition, a JED-2300^† energy dispersive x-ray spectrometer (EDS) was used for EDS evaluation, after the composition evaluation location was fixed based on the TEM observation.

The corroded surfaces of the base material samples exposed for 24, 48, 72, and 96 h are shown in Figure 2. For each exposure period, the corrosion pits are randomly distributed over the exposed surface. The surfaces of the FSW joint samples exposed for the same periods are shown in Figure 3, having different characteristics to the base material. For 24 h exposure, the area around the edge of the weld line seems to have more damage than the base material case. For longer-term exposure, the difference between the base material and FSW joint specimens is more notable; the corrosion pits in the surface of the FSW joint are deeper and wider than in the base material. Further, the corrosion pits are concentrated around the weld edge and near the center of the weld line. Based on metallographic observation,¹⁴  these areas are identified as the SZ and TMAZ.

Fatigue tests were conducted to evaluate the effect of the corrosion damage. The test results are listed in Table 2 and plotted in Figure 4. Note that the results for a benign FSW specimen with the same surface treatment are also shown in Figure 4. For σ_max = 250 MPa, all specimens with corrosion damage fractured during the fatigue test. Further, the fatigue life of an uncorroded FSW joint was found to be more than 10 times than that of a corroded joint. Thus, it is apparent that corrosion damage has a severely detrimental effect on fatigue life. As shown in the figure, the ranges of the fatigue lives of the prior-corroded base material and FSW specimens are similar, although one base material specimen with 24 h exposure exhibited a longer fatigue life, at approximately 4.2 × 10⁵ cycles. For σ_max = 200 MPa, several specimens exposed for 24 h and 48 h did not fracture for 5 × 10⁶ cycles; these were treated as run-out cases. The test results also show that the specimens exposed for a longer period fractured even if the maximum stress was 200 MPa. For longer exposure, the effect of the maximum stress on the fatigue life became less noticeable.

Next, fracture surface observation was conducted in order to identify the fracture origins. For all specimens, the corrosion pits were found to be the fracture origins. Table 2 also reports the numbers of corrosion pits that became fracture origins. In FSW joints, multiple corrosion pits tend to become fracture origins (as shown in Figure 5), in contrast to the base material. This feature becomes more prominent for longer exposure to the corrosive medium.

Figure 6 shows a corrosion pit that acted as a fracture origin in a base material specimen exposed for 96 h. The fracture surface observation conducted in this study revealed that the corrosion pits that acted as fracture origins in the base material did not grow under the surface, as shown in Figure 6. The features of the corrosion pits acting as fracture origins are also summarized in Table 2. Identically to the corrosion pits formed in the base material, those formed in the TMAZs of the FSW joint specimen did not exhibit tunneling, as shown in Figure 5. On the other hand, Figure 7 shows a case in which a corrosion pit formed in the SZ grew under the surface and became a fracture origin, for an FSW joint specimen exposed for 96 h. From Table 2, it is apparent that pits of this type were observed for four FSW specimens subjected to different exposure periods.

Previously, Yun, et al., observed that apparent corrosion pits formed around the TMAZ/HAZ of FSW 7075-T6 Al alloy joints.²⁷  Hence, these researchers concluded that enrichment of the Cu concentration at the TMAZ/HAZ boundaries resulted in a decrease in the corrosion resistance.²⁷  In this study, TEM inspection was used to reveal the microstructural features of each weld location compared to the base material. The results are shown in Figure 8. Note that the observations conducted in this study yielded somewhat different features than the results of Yun, et al.²⁷  Here, TEM inspection of the base material indicated needle-shaped constituent particles randomly distributed within the grains. The number of constituent particles in the HAZ was small compared to the base material, and many were located around the grain boundaries. In addition, many of the constituent particles in the HAZ had granular shapes, and only a small proportion were needle shaped. On the other hand, the constituent particles in the TMAZ and SZ had granular shapes, with many of the constituent particles being located around the grain boundaries (Figure 8).

EDS was also used to determine the constituent particle compositions in each area. As shown in Figure 9, EDS of the constituent particles in the base material indicated the presence of Al, Cu, and Mn. Constituent particles composed of Al, Cu, and Mg were observed; however, those particles comprised less than one tenth of the particles. The same characteristics were observed in the HAZ (Figure 10). On the other hand, constituent particles composed of Al, Cu, and Mn, along with those composed of Al, Cu, and Mg were found in the TMAZ (Figure 11). In addition, constituent particles containing Mg were located at the grain boundaries. These two types of constituent particles were also observed in the SZ (Figure 12), where the constituent particles containing Mg were also located at the grain boundaries. Mg is more susceptible to corrosion than Cu; thus, the grain boundaries in the TMAZ and SZ, having constituent particles containing Mg, are more corrosion-sensitive than those of the HAZ and base material. The EDS observation also indicated that the precipitates at the grain boundaries in the SZ had irregular shapes. Further, the precipitates were approximately 600 nm in size and larger than the constituent particles; this size difference possibly caused a reduced fatigue life.

As described in Bussu and Irving,⁹  the grains in the specimen SZs had equiaxial shapes with sizes of 2 μm. As described in the previous paragraph, the grain boundaries in the SZ and TMAZ are considered more susceptible to corrosion than those of the base material and HAZ. For the TMAZ, some grains are stirred by the FSW process and stretched in the thickness direction; consequently, corrosion may grow more easily in the thickness direction.

Table 3 lists the number of corrosion pits observed for each exposure period. For the FSW joint specimens, the corrosion pits in the SZ were excluded from the count. Further, Table 4 lists the average and variant coefficients of the pit depths for each exposure period. The average pit depth increased nonlinearly for both the base material and FSW joints with increased exposure time. In addition, the average pit depth for the base material was approximately 10 μm smaller than that for the FSW joints. The variant coefficient for the FSW joints was equal to or larger than that for the base material. As shown in Figures 9 through 12, the metallurgical properties of the FSW joints differ perpendicular to the weld line; thus, the corrosion resistance may differ in that direction. This would cause larger variation in the corrosion damage compared to that of the base material.

Table 5 summarizes the average pit depth to short axis aspect ratio for each exposure period and for each specimen type. The variant coefficients are also shown. The average aspect ratio decreased with increased exposure time for both the base material and FSW joints. These results also show that the aspect ratio for the FSW joints was up to 20% larger than that for the base material for longer exposure. Table 4 indicates that the corrosion pit depths for the FSW joints were deeper than those for the base material. Table 5 indicated that the average short axes of the corrosion pits in the FSW joints were larger than those for the base material. As noted above, the corrosion pits in the FSW joints were primarily observed in the SZ and TMAZ, and the pits in the SZ were excluded from this evaluation. In the TMAZ, some grains were stretched in the thickness direction. In the horizontal direction, those grains were short compared to those in the base material; thus, the corrosion damage spread more rapidly in the FSW joints. Further, the variant coefficients of the aspect ratios in the FSW joints were larger than those in the base material.

Table 6 lists the average corrosion pit volume and variant coefficient for each exposure time and for both specimen types. The average corrosion pit volumes in the FSW joints increased sharply with increased exposure period. This is because the maximum depths and aspect ratios in the FSW joint specimens were larger than those in the base material specimens.

Figure 13 shows the pit-depth distribution in the base material after a 96 h exposure. Seventy-six corrosion pits were observed, approximately half of which were between 71 μm and 80 μm deep. Similarly, Figure 14 shows the pit-depth distribution in an FSW specimen after a 96 h exposure. The corrosion pits on the SZ were not considered, because some exhibited tunneling (Figure 7). Here, 50 corrosion pits were evaluated. The maximum depth had a broad distribution compared to the base material. The median corrosion pit depth was between 81 μm and 90 μm; however, the second-deepest corrosion pits had depths of 151 μm to 160 μm, whereas the greatest pit depths were 231 μm to 240 μm.

The distribution of the pit depth to short axis aspect ratio for base material specimens exposed for 96 h is shown in Figure 15, whereas that for the FSW specimens (excluding the SZ pits) is shown in Figure 16. As shown in the figures, 49 and 24 corrosion pits with aspect ratios of between 2.4 and 3.2 were found for the base material and FSW specimens, respectively. Note that, for the FSW specimens, the aspect ratios for four corrosion pits were beyond 5.6.

Figure 17 shows the volume distribution of the corrosion pits that become fracture origins in the base material specimens exposed for 96 h. In the graph, the mode is in the range of 1.05 × 10⁵ μm³ to 1.2 × 10⁵ μm³ and the maximum volume is in the range of 2.40 × 10⁵ μm³ to 2.55 × 10⁵ μm³. Figure 18 shows the distribution for the FSW specimens exposed for 96 h, omitting the pits in the SZ. The range on the horizontal axis is the same as for the base material figure. The volumes of the corrosion pits on the two specimen types are of the order of 10⁷ μm³ and the distribution is more widely spread than in the base material.

The depths, aspect ratios, and volumes of the corrosion pits identified as fracture origins are indicated by arrows in the various figures. The results presented in these figures show that the deepest pit did not necessarily become a fracture origin in the base material or FSW specimens. In particular, Figures 15 and 16 show that the aspect ratios of the corrosion pits identified as fracture origins are between 2.2 and 3.6 in both cases. Deep corrosion pits do not necessarily become fracture origins. The volumes of the corrosion pits identified as fracture origins in the base material and FSW joint specimens are shown in Figures 17 and 18. For the base material, the corrosion pit with the maximum volume is not a fracture origin (Figure 17). However, for the FSW joint case, the corrosion pit with the maximum volume did become a fracture origin. However, for specimen FSW-96-3, the volumes of the corrosion pits that became the fracture origins are 1.50 × 10⁶ μm³, 3.48 × 10⁶ μm³, and 3.51 × 10⁶ μm³, but the maximum pit volume is 6.68 × 10⁶ μm³. Thus, the corrosion pits with the greatest volume did not necessarily become fracture origins for either specimen type.

Previously, Sankaran, et al., assumed that the corrosion pit size corresponds to the initial crack size and evaluated the fatigue crack growth life from pit to fracture in a specimen.²⁴  Those researchers concluded that the fatigue crack growth life calculated using the average pit size coincided well with the fatigue life obtained through experiment. Figures 19 through 22 show stress vs. the number of cycles to failure (S-N) data for exposures of 24, 48, 72, and 96 h, respectively. In addition, the fatigue crack growth life required for the crack to span the specimen thickness was calculated using NASGRO^† commercial software,²⁹  based on the assumption of correspondence between the average pit size and that of the initial crack; these results are also shown in Figures 19 through 22. The average pit size perpendicular to the loading direction was used for the analysis, with the data being listed in Table 7. It was assumed that the corrosion pit was located at the specimen center and the effects of adjacent cracks were neglected. As shown in these figures, the fatigue life obtained through the crack growth analysis for a maximum stress of 250 MPa is relatively close to that obtained through experiment. On the other hand, considerable differences were observed between the crack growth analysis fatigue lives for a maximum stress of 200 MPa and the experimental data, especially for shorter exposure times. Note that the evaluated fatigue crack growth life did not account for the period between the pit to crack transition and the effect of short crack growth on the crack growth rate; these were larger for the smaller corrosion pits obtained through shorter-term exposure. These differences contributed to the obtained discrepancies between the different results, and have also been reported in Sankaran, et al.²⁴  Further, because the difference in corrosion pit size between the base material and the FSW joint was small, the features observed for both specimen types were similar.

As is apparent from Table 4, the average pit depths of the FSW joint specimens were approximately 10 μm larger than those for the base material. The variation coefficients for the base material indicate that the standard deviations in the base material case were approximately 10 μm. Further, the sum of the average and standard deviation of the base material value is close to the average of the FSW joint value. The variation coefficients of the FSW joint specimens were up to 2.5 times larger than those of the base material specimen. The FSW specimen corrosion pits larger than the average pit depth contributed to shorter fatigue lives compared to the base material specimens. Previously, Jones, et al., evaluated the fatigue behavior of prior-corroded 2024-T3 Al alloy, and also found that the deepest pits did not necessarily become fatigue fracture origins.²¹  In addition, Merati evaluated both the distributions of constituent particles in 2024-T3 Al alloy and their fracture origins under constant fatigue loading, reporting that the largest constituent particles did not necessarily become fracture origins.³⁰  The results presented in this paper show that corrosion pits in FSW joints also exhibit the same behavior.

Because the corrosion pits on an FSW joint tend to concentrate at the weld center and around the weld edges, their density is greater than those in the base material. Such concentrations of corrosion pits on FSW specimens may link more easily, thereby reducing the joint’s fatigue life more drastically than in the base material. This characteristic should be increasingly apparent in wider specimens.

• •

  The corrosion resistance of an FSW joint fabricated from 2024-T3 Al alloy was evaluated. The fatigue lives of corroded samples were also evaluated. The corrosion tests indicated that corrosion pits on an FSW joint tend to concentrate around the edges and center of the weld line, while those on the base material are distributed randomly. In addition, TEM observation of typical weld locations revealed that the constituent particles in the TMAZ and SZ are of granular shape and are located inside of the grain boundaries. In addition, the constituent particles in the base material and HAZ are needle shaped and primarily located inside the grain boundaries. EDS analysis of each location showed that in the TMAZ and SZ, some of the particles contain Mg and are distributed at the grain boundaries, while the particles in the base material and HAZ do not generally contain Mg.

• •

  The depths and volumes of the corrosion pits are larger in the FSW joints than in the base material. While the aspect ratios of the corrosion pits in the base material and FSW joint specimens are similar for a shorter exposure period to a corrosive material, those in the FSW joints become larger for longer exposure periods. Corrosion pits in the FSW joint specimens having larger than average depth contribute to a shorter fatigue life than those of the base material specimens.

• •

  Fracture surface observation after fatigue testing showed that corrosion pits in the SZ may cause tunneling. Note that, in this test case, the corrosion pits in other areas did not grow under the surface. In addition, it was found that the corrosion pits with the greatest depth and greatest volume do not necessarily become fracture origins.

The authors wish to offer sincere thanks to K. Iwasaki, S. Kishishita, and H. Tanaka for their assistance regarding the fatigue test setup and pit measurement after corrosion testing.