3.3. The Impact of Corrosion on the Mechanical Performance
The mechanical performance of SPR and RSW joints before and after corrosion was evaluated by lap-shear testing with regard to the strength, axial stiffness, the energy absorbed at failure, and fracture modes.
The force and displacement (F-D) curves of SPR and RSW are shown in
Figure 16a,b, respectively. For SPR joints, the F-D curves tend to be taller and slimmer as the corrosion duration increases, indicating the fracture mode turns ductile into a more brittle fracture. “Hills” were also observed in the curve after 26 cycles, and as the corrosion duration increased, the “hill” became taller and steeper. This was because the brittle corrosion products became thicker and tightened the joints. When they break under loading, the force drops significantly. Similarly, the F-D curves of RSW become steeper after seven cycles, implying the fracture mode turns to become more brittle.
The strength of non-corroded and corroded Al–steel SPR and RSW joints are plotted in
Figure 17a. The average strength of both SPR and RSW joints increases slightly as the corrosion proceeds instead of decreasing. In SPR joints, galvanic corrosion attacks AA6022 at the Al–rivet head interface and the interlock is not affected until 104 cycles. In fact, the rivet leg is only exposed and corroded in one sample of 104-cycle joints. As a result, the strength does not decrease. However, the corrosion products are not stable and easily become loosened and removed from the joints during the service, which will lead to a significant decrease in strength. For RSW joints, the average strength decreases significantly since the corrosion proceeds into the Al–steel faying interface within the joint, and SCC occurs in the IMC layer after 48 cycles. Once the whole layer of IMCs cracks, the joint fails to bear any loading, and the strength drops to 0 (104 cycles). According to the above analysis, the SPR joint performs much better in retaining strength than the RSW joint does when exposed to a corrosive environment.
The strength, stiffness, and energy absorbed at failure were extracted from the F-D curve. The stiffness is obtained by fitting the data within the range of 0~2 KN, and the energy absorbed at failure was calculated by (), where and are the force and displacement of the th set of experimental data, respectively.
The energy to failure is calculated based on Equation (1):
where
and
denote the force and displacement of the
th set of experimental data, respectively. The energy to failure of non-corroded and corroded Al–steel SPR and RSW joints are plotted in
Figure 17c. For SPR joints, the energy does not show too much difference as corrosion proceeds. This is because though the shape of the curves becomes slim, the elastic slope decreases, and the maximum load increases due to the corrosion products tightening the joint; therefore, the energy absorbed till failure does not change too much. For RSW joints, the energy increases slightly at one cycle and drops drastically from one cycle to seven cycles. Since then, it has not changed significantly and finally drops to 0 at 104 cycles. This is in accordance with the shape of the F-D curves. In one cycle, the stiffness and the extension do not change significantly, but strength increases due to the tightening effect of the corrosion products. From 7 to 14 cycles, the stiffness decreases, but the extension does not change; hence, the energy drops. The stiffness continues to decrease, but the extension increases a lot during 14~72 cycles. At some specific time between 72 cycles and 104 cycles, joints were debonded due to SCC propagating to the whole layer of the IMCs.
Therefore, SPR joints perform better than RSW joints regarding the energy to failure, especially under long-term corrosion.
The comparison of the strength, stiffness, and energy of SPR and RSW joints before and after corrosion is shown in
Figure 17a–c, respectively.
According to
Figure 17a, the average strength of SPR joints increases slightly as the corrosion proceeds instead of decreasing. This is because the corrosion products generated tightened the mechanical interlocking [
36]. Some bonding marks were observed in the Al–steel overlapping area after fracture; the lap-shear test also helped to prove this, as shown in
Figure 5. According to
Figure 8, galvanic corrosion attacks AA 6022 at the Al–rivet head interface, and the interlock was not affected until 104 cycles (only in one 104-cycle sample was the rivet observed to be exposed and corroded). As a result, the strength did not decrease. However, in the foreseeable future, as corrosion continues, the interlock will decrease, and hence the strength will be reduced. For RSW joints, the average strength increases slightly from 1 cycle to 26 cycles as well. However, after 48 cycles of corrosion, the strength decreases significantly since corrosion proceeds into the IMC layer and causes SSC. Once the whole layer of IMCs cracks, the joints fail to bear any loading, and strength drops to 0 (104 cycles).
The stiffness changes of SPR and RSW show a similar trend, which can be divided into 3 phases, as shown in
Figure 17b. Phase 1 is from one cycle to seven cycles, and in this phase, stiffness decreases. Phase 2 is from 7 cycles to 26 cycles, in which stiffness increases. After 26 cycles, the stiffness decreases significantly, which is phase 3.
In phase 1, very slight corrosion occurs on the surface, as shown in
Figure 5 and
Figure 12. It is supposed not to affect the mechanical performance significantly. The decreasing stiffness might be due to the residual stress relaxation under the elevated temperature in the Humid Stage and Dry-off Stage in the cyclic corrosion test. To validate the hypothesis, four as-received SPR joints were heated up, following the elevated temperature history in one cycle (49 °C for 8 h and 60 °C for another 8 h), and then a lap-shear test was performed. The stiffness of the four heated joints is 6.602 KN/mm, 6.876 KN/mm, 6.518 KN/mm, and 6.823 KN/mm, respectively. Compared with the stiffness of the as-received joints, which is 7.812 ± 0.173 KN/mm, the stiffness of the heated joints decreases significantly and is comparable with that of one cycle joints, 6.332 ± 0.496 KN/mm. Therefore, the stiffness decreases because residual stress is released under elevated temperature. According to
Figure 17b, the stiffness continues to decrease from one cycle to seven cycles, indicating that the residual stress was not released completely until seven cycles.
In phase 2, from 7 cycles to 26 cycles, HDG was corroded away, and in 14 cycles, steel started to be exposed and was completely exposed in 26 cycles, according to
Figure 5 and
Figure 14. The corrosion products (zinc oxide mainly) and a great amount of corrosive solution solids collect at the overlap ends. For one thing, galvanic corrosion at the overlap starts here; for another thing, the joints were leaned on the holder in the corrosion chamber, and corrosion products and corrosive solution flowed to and then collected here. Under lap-shear loading, these solids acted as an “adhesive” and provided extra resistance to shearing and peeling. The schematics of the joints without/with corrosion products, which are collected at the overlapping area in the lap-shear test, are shown in
Figure 18a,b, respectively. Because of the eccentricity of the forces on the fixed end and moving end, moment will be generated in both as received and corroded joints. The moment acted on the steel and Al is shown in Equations (2) and (3), respectively:
where
is the clamping force before loading,
is the thickness of steel, and
is the thickness of Al.
Under this moment
and
, steel and Al start to bend, as shown in
Figure 18a,b. Because the corrosion products collect at the overlap ends, the material at the overlap area gains extra strength to resist deformation, and the material will bend more slightly in the corroded joint than in the as-received joint. During the lap-shear testing, the movement of the crosshead is controlled by the displacement (moving speed is 2 mm/min), therefore after some certain time
, the displacement
of as-received and corroded joints are equivalent. For as-received joints,
where
and
are the length of steel from the fixed end to the center of the joint at the beginning and after a certain time t, respectively,
and
are the length of Al from the moving end to the center of the joint at the beginning and after a certain time t, respectively, and
is the rotation angle of material from the fixed end to the moving end.
When the force is aligned,
has a maximum value that:
therefore,
is smaller than 0.92° and
is small enough to be neglected. In this case, for as-received joints,
can be reduced into
Similarly, for corroded joints
where
and
are the length of steel from the fixed end to the left overlap end before loading and after some certain time
when corrosion products collect (including corrosive solution solids) at the overlap end, and
and
are the length of Al from the moving end to the right overlap end before loading and after some certain time
when corrosion products collect at the overlap end. According to Equation (7), the deformation of the material in the overlap area does not contribute to
; it is because corrosion products collect at two ends, which act as “adhesive”; the material in the overlap area gains extra strength to resist deformation.
To obtain a specific displacement
, the force applied to the as-received joint
and the corroded one
are shown in Equations (8) and (9), respectively:
where
and
are the cross-section area of steel and Al, respectively, and
and
are the Young’s modulus of steel and Al, respectively.
Since the displacement of the as-received joint and corroded joint after some certain time is equivalent, hence . For simplicity, the portion of the deformation of Al and steel are considered as the same in two conditions, that is , . Substituting this relationship in Equations (8) and (9), since , . Therefore, the stiffness of the joint in which corrosion solids collect at the overlap ends is greater than that in the as-received joint ().
It is also noted that the stiffness of SPR joints increased faster than that of RSW joints from 7 to 14 cycles and then kept almost the same from 14 to 26 cycles, while the stiffness of RSW joints continues to increase. This is because the smaller gap of SPR joints at the Al–steel interface prevents the corrosion products from flowing into the crevice, resulting in more corrosion products collecting at the overlap end. It costs RSW joints more time to collect the same amount of corrosion products since a larger gap exists along the Al–steel interface.
In phase 3 (26~104 cycles), HDG has been corroded away. Without the protection of HDG, steel starts to corrode significantly. Meanwhile, AA 6022 was under severe galvanic corrosion at the overlapping area. AA 6022 became highly porous at the overlapping ends, as shown in
Figure 5 and
Figure 14. These large amounts of material loss, especially the large number of pits generated in AA 6022 at overlap ends, contribute to the decrease in stiffness. According to
Figure 17b, the decreasing rate of the stiffness in RSW joints is much higher than that in SPR joints, which implies more severe galvanic corrosion occurs at the overlap area.
Figure 17c shows the energy absorbed at the failure of SPR and RSW joints before and after corrosion. For SPR joints, the energy does not show too much difference as corrosion proceeds. This is because though the shape of the curve becomes slim and stiffness decreases, the peak load increases due to the corrosion products tightening the joints a little bit; therefore, the energy absorbed does not change too much. For RSW joints, the energy increased slightly at one cycle and dropped drastically from one cycle to seven cycles. Since then, it has not changed too much and finally drops to 0 at 104 cycles. It is in accordance with the shape of the F-D curves. At one cycle, the stiffness and the extension do not change significantly, but strength increases due to the tightening effect of the corrosion products. For 7~14 cycles, the stiffness decreases, but the extension does not change; hence, the energy drops. After 14 cycles to 72 cycles, the stiffness continues to decrease, but the extension increases a lot. For 104 cycles, joints were debonded due to SSC and lost any load-bearing capacity.
Compared to the corrosion behavior and the mechanical performance of corroded SPR and RSW joints, SPR joints retain the mechanical properties much better than RSW joints, including the strength, stiffness, and energy absorbed at failure.