As a solid-state joining process, the primary sources of heat in Friction Stir Spot Welding (FSSW) are the friction between the shoulder and pin of the tool and the material, which are correlated with the tool rotation speed and the plastic deformation the material undergoes. The calculated energy values for the welds produced with various parameter interactions are presented in Table I.
TABLE I. Total Energy for Different Parameters
Rotation Speed (rpm) | Mixing time (s) | Energy (kJ) |
1435 | 3 | 24.2 |
1600 | 2 | 16.9 |
1600 | 4 | 25.6 |
2000 | 1.6 | 22.7 |
2000 | 3 | 25.5 |
2000 | 4.4 | 27.9 |
2400 | 2 | 27.1 |
2400 | 4 | 29.2 |
2565 | 3 | 34.9 |
The supplied energy was calculated indirectly using Eq. 1, as cited in Eagar (1986) and employed by Khan [19], as a function of the normal force F, tool positioning X, torque τ, angular velocity ω, and process duration time. The data were collected at the output of the frequency inverter of the machine used. Since the tool does not move, the first part of the equation is disregarded, leaving only the relationship between angular velocity, torque, and time.
The data indicates that energy values increased with the increase in rotation and time; however, rotation has a greater influence on the energy input compared to the mixing time. The standardized effects graph of the parameters can be viewed in Fig. 3, generated using Minitab software. The rotation speed was found to be the most significant parameter.
The energy calculations exhibited a normal distribution with an 95% confidence level according to statistical analysis. However, it’s important to note that process losses were not factored in, and the collected energy data from the inverter output might not precisely reflect the actual values. Nonetheless, this consistent method was applied to all samples, ensuring consistency. A response surface showing the probability of generated energy is displayed in Fig. 4, created using Minitab software.
The probability of energy predictability is illustrated in Fig. 5, created using Minitab software.
The values of the welded area and the corresponding shear strength of the welds obtained for various parameter combinations are presented in Table II. The correlation analysis between the energy data and area did not reveal a direct relationship between these variables. Khan, in the experiments with transformable steels, described a similar effect, highlighting the lack of linearity between energy, yield strength, and welded areas, although was observed a larger welded area with the application of higher energy [19].
In this specific study, the welded area varied around a mean value, making it difficult to identify a clear trend. However, Feng observations suggest that increasing the time in experiments resulted in an expanded bonding region [20]. Indeed, the nonlinearity observed in this study aligns with the complexity of the welding process and underscores the importance of considering additional factors. One such factor that may influence the formation of the weld region thickness, but has not been evaluated before, is the pressure exerted by the tool. Xie, citing Santella et al., argues that, for the same material, higher pressure can have a significant impact on the strength limit [12]. The relationship between weld area and shear strength for various rotation speeds and mixing times is depicted in Fig. 6. The lack of a clear correlation between these variables suggests the potential influence of other factors on the shear strength limit.
Remarkably, the weld achieved at a rotation speed of 2000 rpm and 4.4 seconds exhibited unusually the lowest shear strength limit. In this case, the area of the welded region increased compared to the weld obtained at 2000 rpm and 3 seconds. This discrepancy could arise from the formation of different constituents or discontinuities during the process, potentially diminishing both the strength and the effectively welded area. Such conditions might have contributed to the weld’s early failure.
When comparing the images, it becomes evident that there is a lack of a consistent fracture pattern, except for the fact that it seems to have originated at a point along the line formed between the sheets. It is noteworthy in this case that such occurrence indicates a partial union or even absence thereof at the outer end of the joint. This failure process traverses the line formed on the lower sheet until reaching the rupture in the stir zone of the upper sheet. The macrographs, which provide a visualization of the fractured surface aspect of the welds after the test, are presented in Fig. 7 and Fig. 8.
The macrographs of the welded specimens revealed distinctive regions whose size and composition varied depending on the parameters employed affected by process energy. Figure 9 presents a typical weld cross-section, delineating the stir zone, thermomechanically affected zone, thermally affected zone, and base metal, as elucidated by [20] and further classified into sub-regions by [21]. The microstructural examination specifically targeted the central points within these regions to assess the constituent elements.
Additionally, the formation of a discontinuity situated between the welded plates and the coalesced material, as illustrated in Fig. 10, was observed in all welds. This phenomenon tends to become more visible when the distance, created by the central pin promoting material mixing, increases. The dimensions of it tend to reduce with material diffusion become more effective by heating, which intensifies with the proximity of the central pin. Rotation speed also influences this, as noted by [12].
Observing the highlighted detail in the figure, can be noted the presence of acute angles on the outer part, formed between the coalesced material that was forced between the plates during the tool pin movement. Under certain processing conditions, as illustrated in the macrograph in Fig. 11, the presence of cracks was observed starting from the edge of the discontinuity. Internal stresses resulting from welding or external forces can propagate it, potentially leading to the total failure of the weld. This phenomenon is commonly observed in welds with high energy in transformable steels, in which cracks propagate through the mixing zone until complete failure [19].
The discontinuity formation, indicating a mixing lacking mainly in the external regions where the maximum temperatures are lower, was observed in all welds produced. However, it is more evident in the weld depicted in Fig. 11, which was produced using 2000 rpm (rotation speed) and 4.4 seconds (mixing time) where a crack appears at the end of the discontinuity located along the mixed and thermomechanically affected zones in the bottom position.
The base metal microstructure is illustrated in Fig. 12 showcasing ferrite, martensite, bainite, and retained austenite even after the initial plastic deformation of 10%. The presence of martensite is not inherent to this steel in as-received condition indicating an effective transformation phenomenon induced by plasticity. The martensite fraction observed results from the TRIP effect, leading to a decrease in the fraction of retained austenite. The remaining fraction suggests that the transformation effect can still manifest in this material under stress and deformation.
The images captured with SEM at a magnification of 3500 times depicted in Fig. 13 represent the mixed zone of the welds generated by using a mixing time of 3 seconds across three distinct rotation speed levels. Generally, a microstructure predominantly comprising martensite and bainite is discernible.
The presence of martensite can be attributed to cooling rates surpassing the critical value during welding cycle, in a region where maximum temperatures reached is higher than those necessary for complete austenitization. Additionally, it is evident that the rise in rotation speed led to an enlargement in the size of the martensite laths. This observation can be linked to the development of larger austenitic grains in processes conducted at higher rotation speeds, which are correlated with increased input energy and, consequently, higher peak temperatures. This effect is consistent with Mazzaferro findings that higher rotation speeds resulted in a decrease in both the quantity and size of bainite grains in similar materials [21]. As the rotation speed increases, martensite plates tend to enlarge while bainite experiences a reduction in quantity and size.
In another hand, images from Fig. 14 obtained using SEM, reveal the impact of mixing time on the constituents at a fixed rotation speed parameter of 2000 rpm. The mixed zone exhibits a refined structure predominantly composed of martensite along with what appears to be bainite. This is consistent with findings in other samples where the presence of martensite is likely attributed to the critical cooling temperature reached after attaining the austenitization temperature. Any austenite that did not undergo martensitic transformation may persist as retained austenite or decompose into bainite.
In the sample with a time of 1.6 seconds, the presence of ferrite was restricted and localized at certain grain boundaries, a phenomenon not observed in other samples. Additionally, the size of martensite plates tends to increase with higher energy input in the process, which, in this case, is directly correlated with time.
Austenite that has not undergone martensitic transformation may persist in the form of retained austenite or decompose into bainite. The observed martensite plates also tend to be larger with higher energy in the process, which, in this case, is associated with the growth of austenitic grain favored by mixing time increasing.
The mixed zone exhibits a refined structure primarily composed of martensite and bainite, which is consistent with earlier observations. The presence of martensite is likely linked to the critical cooling temperature after reaching the austenitization temperature. Austenite that hasn’t undergone martensitic transformation may persist as retained austenite or decompose into bainite. Larger martensite plates are observed with higher energy in the process, likely due to increased growth of austenitic grain facilitated by time and rotation speed.
Figure 15 shows the thermomechanically affect zone constituents of the welds produced at rotation speeds of 1435 rpm, 2000 rpm, and 2565 rpm, with a welding time of 3 seconds. The sample with a rotation speed of 1435 rpm exhibits a structure closer to that of the base metal. It is possible that this region did not reach temperatures to promote significant changes in constituents, and the martensite likely resulted from the initial 10% deformation. In the sample produced using 2000 rpm, there was a decrease in the amount of austenite, accompanied by the growth of ferrite grains and an increase in bainite, which is expected from austenite decomposition. The occurrence of retained austenite is also minimal, and existing martensite likely must be related to the initial deformation. The sample with a parameter of 2565 rpm exhibited a martensitic matrix. Alongside martensite, there is an expected bainite constituent, with ferrite occurring in small amounts and the presence of retained austenite is minimal.
In these instances, the formation of higher hardness constituents in the thermomechanically affected zone correlated with increased rotation speed. This could be attributed to the higher peak temperatures reached, as noted by Mazzaferro (2008), potentially leading to the attainment of the austenitization temperature.
Weld microstructure produced with fixed rotation speed (2000 rpm) but varying in mixing time is presented in Fig. 16. The weld made with a time of 1.6 seconds demonstrates a reduction in retained austenite within the thermomechanically affected zone compared to the base metal. The presence of martensite is likely due to the initial 10% deformation, with small occurrences of austenite and bainite. In contrast, the weld with 3 seconds mixing time showcases an increase in ferrite grain size and a higher amount of bainite resulting from austenite decomposition when compared to the weld with a 1.6 second.
The weld with a mixing time of 4.4 seconds exhibited a martensitic matrix accompanied by what seems to be bainite in smaller proportions. Among all samples, this one displayed the highest martensite content in the thermomechanically affected zone, suggesting extensive austenitization with a longer mixing time. This, coupled with critical cooling rates, facilitated the formation of significant quantities of martensite. The development of internal stresses may have contributed to the crack observed earlier in Fig. 11. In these instances, the formation of higher hardness constituents in the thermomechanically affected zone was observed with an increase in mixing time.
Various parameters influenced the behavior of the weld, leading to alterations in the dimensions of the formed zones and the resulting microstructures. Hardness profiles were obtained through tests performed using a Vickers Hardness tester, and indentations were made to analyze the hardness of the constituents formed in different regions. Although the hardness values in the mixed zone are similar, the size of martensite plates increased with the rotation speed. Additionally, the thermomechanically affected zone exhibited a composition with a higher fraction of microconstituents, such as martensite and bainite, at higher rotation speed values. Lower values allowed the presence of constituents like ferrite and retained austenite, remaining from the base metal.
The hardness profile conducted on samples with a base time of 3 seconds and rotation speeds of 1435, 2000, and 2565 rpm is depicted in Fig. 17, revealing consistent values with the founded mixed zone microstructure. As the rotation speed increased from 1435 to 2565 rpm, hardness values exhibited a slight reduction, a phenomenon also noted by Mazzaferro (2010) [22]. This reduction could be attributed to the larger size of martensite plates. However, the values, when considered alongside the standard deviation, overlap, hindering a definitive conclusion about the trend. Furthermore, hardness values decrease as one moves away from the mixed zone, a pattern observed by both Feng (2005) and Mazzaferro (2010) [20, 22]. It is hypothesized that the transformation of martensite in the mixed zone is linked to the critical cooling rate after reaching temperatures AC1 and AC3 with a high degree of austenitization.
The hardness profile depicted in Fig. 18 for samples with a fixed rotation speed of 2000 rpm and varied times (1.6 s, 3 s, and 4.4 s) showed typical hardness expected considering the founded mixed zone microstructure. Comparing the average hardness of mixed zone is observed a slight reduction with a mixed time increasing but the variations are within standard deviation. This behavior must be related to the thermal cycle and the resulted microstructure which presents differences in the martensite laths.
In all samples, the thermomechanically affected zone serves as a transition area where hardness values decrease as one moves away from the mixed zone to base metal, as observed in the hardness profile. This can be clearly understood by the microstructure formed in this region. The microstructural variations resulted by rotation speed and mixed time changes also must be observed in the hardness profile. In turn, as expected, the thermally affected zone tends to presents hardness close to the base metal.