Material Flow Visualization of Dissimilar Friction STIR Welding Process Using Nano-Computed Tomography

The growing concerns on energy saving and reduction of CO₂ emission increase the demand for lightweight vehicles. A multimaterial structure is an efficient solution, which accordingly requires reliable and economical dissimilar material joining technologies. One promising process is the friction stir welding (FSW) based on its solid-state nature. There have been several experimental studies regarding the FSW of an aluminum alloy to steel, and the majority focused on microstructural characterization of the Al–Fe interfacial layer. Uzun et al. [1] performed the feasibility study on FSW of 304 stainless steel to Al 6013-T4 and showed a joint efficiency of 70% with the aluminum strength as the baseline. Dehghani et al. [2,] studied the effects of welding parameters including tool traverse speed, plunge depth, tilt angle, and tool pin geometry on the joint quality and intermetallic compounds (IMCs) formation during FSW of a mild steel to Al 5186. The tool is located at the position where a pin edge touches the steel matrix. The plunge depth shows to be a highly sensitive variable for the weld quality. The heat input factor, which is a combination of tool rotational speed, welding speed, and tool radius, is proposed for determining the appropriate process window. They also showed that threaded pins can facilitate the material flow and effectively remove the tunnel defect. Abbasi et al. [3,] applied advanced transmission Kikuchi diffraction (TKD) for a detailed characterization of the Fe-rich fragments that are generally observed in the stirring zone of Al–Fe FSW welds. A layer of orthorhombic Al₅Fe₂ intermetallic is observed to cover these fragments, which are further surrounded by a mixed region of Al nanograins with submicrometer Al₃Fe crystals distributed along the grain boundaries. Inside of these fragments are fine elongated grains of Fe. Ghosh et al. [4 ] applied transmission electron microscopy (TEM) to characterize the interface of pure Al-304 stainless steel FSW welds. Diffusion of Fe, Cr, and Ni elements into the Al matrix is substantial, while the diffusion in the reverse direction is limited. Fe₃Al was observed at the Al/steel interface. Lee et al. [5 ] identified that the Al–Fe interface consists of mixed layer of ultra-fine Al grains and Al₄Fe IMC with a hexagonal close-packed structure and thickness of 250 nm.

Variants of FSW, including friction stir lap welding and friction stir spot welding were also applied to join aluminum to steel. Shen et al. [6,] lap welded Al5754 to DP600 steel. Layered structure of steel, aluminum, and IMCs is observed at the Al–Fe interface in the stirring center. The plunge depth is a decisive parameter for joint strength. When the depth is lower than 0.17 mm, shear fracture occurs along the interface. Lee et al. [7,] lap welded Al 5083 to St-12 mild steel. A 2-μm IMC layer with the composition of Fe₃Al, Fe₄Al₁₃ was observed at the Al–Fe interface. Movahedi et al. [7 ] did lap welding on Al-5083 to St-12 alloy and also observed a swirl-layered structure. They introduced the parameter of welding rate, which is defined as the ratio between the tool rotational and travel speed and stated that the weld zone defects decrease with increasing the welding rate. Bozzi et al. [8 ] did friction stir spot welding between Al 6016 to IF-steel. Tangles of elliptical intermetallic compounds with the compositions of FeAl₃, Fe₂Al₅, or FeAl₂ are observed using TEM. They reported IMC with an appropriate thickness is beneficial for the joint strength.

In our previous works, FSW and friction stir spot welding of aluminum alloy Al6061 to the transformation-induced plasticity (TRIP) steel was systematically studied with both experimental and model analysis [9–12,]. Steel fragments and a piece of steel hook are generally observed to be embedded in the aluminum matrix from cross-sectional view of the welds. However, since these samples were sectioned and prepared with conventional metallurgical grinding and polishing procedures, only two-dimensional (2D) information can be revealed. As the material flow during FSW is a key factor to achieve sound joints, a fully three-dimensional (3D) visualization of the material distribution is highly desirable for understanding this important phenomenon. Developments in computed tomography (CT) imaging have provided unprecedented avenues into nondestructive visualization of arbitrary sectional slices, which enable 3D characterizations. More specifically, nano-computed tomography (nano-CT) was introduced during the past decade. Its high power and nano-focused beam enables a superior resolution of up to 400 nm [13,14,]. Nano-CT has been widely applied in biomedical research, including high resolution imaging and density analysis of tissues, organs, and bones. In addition, various other research fields have employed the capabilities of nano-CT as well, such as examination of devices microstructure in electronic industry [15,], characterization of pore structure of leached cement pastes in civil engineering [16,] and examination of voids and air bubbles for fused deposition modeling process in additive manufacturing [17 ]. Since the CT imaging comes from different X-ray attenuation coefficients of different materials, including the atomic number, sample density, and thickness [18 ], it will be an efficient approach to resolve the internal material distribution in dissimilar FSW welds. So far, no literature has reported relevant results in this regard.

In this study, FSW of aluminum alloy 6061 to TRIP steel are performed under various welding conditions with two sets of tools. Temperature of the process is measured with thermocouples. The 3D internal microstructure of the welded samples is reconstructed using nano-CT, which combined with conventional optical and scanning electron microscopic (SEM) analysis will reveal the material flow behavior and welding mechanism during the dissimilar material joining process. Mechanical properties of the joints are evaluated through tensile tests. Relationships between the processing condition, weld macro/micro structure and joint strength are established.

The experimental setup is shown schematically in Fig. 1. Al6061-T6511 sheets and TRIP 780/800 steel sheets with the thickness of 1.5 mm are butt-to-butt configured. The aluminum sheet has an ultimate tensile strength of 283 MPa [9]. In all the weld trials, steel is placed in the advancing side, where the tool rotating direction aligns with the tool traverse direction. Relative position of the FSW tool with regard to the two materials is quantified with the parameter of tool offset, which is the distance between the tool axis and original interface of the two sheets. During all experiments, the tool is shifted to the aluminum side, and a larger tool offset means more fraction of aluminum is stirred in the nugget. Four thermocouples are placed symmetrically to the weld interface at the bottom surface of the workpiece. Detailed distance and positions are provided in Fig. 1. Two FSW tools are investigated, which have the same shape, i.e., a flat shoulder and a conical pin with the taper angle of 20 deg and pin length of 1.2 mm. The size of the tool varies, where the bigger tool has a shoulder diameter D_s of 15.88 mm and pin diameter D_p of 4.03 mm, while those for the small tool are 9.53 mm and 2.42 mm, respectively. For each tool, two levels of rotating speed, welding speed, and tool offset are studied. All the experimental conditions along with the corresponding tensile tests results are summarized in Table 1. It should be noted that the 1.65 mm tool offset condition is not applicable to the small tool, since the pin would be entirely immersed in the aluminum side without effectively joining the two materials together. In addition, the volume fraction of steel in the enclosed volume of pin for the small tool with 1 mm tool offset is equal to that for the big tool with 1.65 mm offset condition.

Dog bone tensile test specimens are prepared according to the ASTM E8 standard. Relative positions of dogbone specimens and nano-CT samples with regard to the weld are illustrated in Fig. 2.

For nano-CT analysis, weld specimens were scanned with the GE Nanotom S nano-CT system (GE Measurement & Control, Wunstorf, Germany). The sample is placed on a rotation platform positioned in between an X-ray source and an X-ray detector. As the X-rays pass through the sample, they will be absorbed to different degrees according to the material atomic numbers, density, and thickness. A higher density material has a higher absorption rate. The two-dimensional projected absorption images are collected by the X-ray detector. As the sample rotates 360 deg, these collected absorption images can be synthesized and reconstructed into a complete 3D representation of the internal material distribution inside of the weld. Scanning parameters were optimized to maximize spatial resolution and image contrast. In this study, these parameters include 120 kV voltage, 110 μA current, and 750 ms exposure time and a voxel size of 3.5 μm. For visualization and post analysis of the CT images, the imagej software with BoneJ program is applied [19].

Weld cross sections were further characterized using optical microscopy (Nikon H600L (Nikon Instruments Inc, Tokyo, Japan)) and SEM for a higher magnified of the weld interface (FEI Nova Nanolab 200 (FEI, Hillsboro, OR)).

The top view of a typical normal Al–Fe FSW weld is provided in Fig. 3(a), where a relatively smooth surface with minimal amount of flash is observed. In the condition of a big tool with a smaller tool offset, acceptable joints can only be achieved under the combined scenario of higher level of rotating speed and lower level of welding speed. The remaining three cases would yield visible groove defect on the surface, as shown in Fig. 3(b). This indicates an ineffective stirring action and poor material flow, which is due to the insufficient heat generation to deform a higher fraction of steel in the weld zone with the larger pin. This will be further confirmed with the temperature measurement results in Sec. 3.2.

For acceptable joints with no surface defects, the observed failure modes generally fall into four categories. In the first group, the joint fractured in the aluminum side away from the stirring zone, as shown in Fig. 4(a). This generally corresponds to a high joint strength of over 230 MPa, i.e., 81% of the base aluminum alloy. In the second category, a continuous steel strip with a hook structure is embedded in the aluminum matrix. The fracture occurs along the outside boundary of this hook structure, as shown in Fig. 4(b). Corresponding joint strength is around 180–200 MPa, i.e., 63–70% of that of the base aluminum. In the third type of joint, which is generated only in the combined condition of big tool and smaller tool offset, large steel fragments were observed in the stirring zone of aluminum. Corresponding welds would fracture along these fragments and result in a relatively low strength of around 140 MPa, i.e., 50% of base Al alloy, as shown in Fig. 4(c). In the fourth type, the fracture path follows the Al/steel interface at the bulk steel side, as in Fig. 4(d). This also corresponds to a low joint strength in the range of 140–160 MPa, i.e., 50–56% of the base Al alloy.

In order to further understand the joint tensile failure mechanisms, fractographic analyses are performed. For type I welds that failed in the aluminum side, presence of dimple structures in Fig. 5 indicates a ductile fracture behavior. On the other hand, some relatively large cavities can be observed, which are possibly generated by some finely dispersed steel inclusions or coarsened precipitates in the aluminum 6061 matrix related to the thermal cycle during the welding process. This fractured surface corresponds to the aluminum part of the failed tensile specimen.

For type III welds that failed along the large steel fragments embedded in the aluminum matrix, a typical fractured surface corresponding to the aluminum part of the failed tensile specimen is presented in Fig. 6. It contains two regions. A flat cleavage feature is observed in the steel fragment region, indicating a brittle behavior. This is due to the insufficient bonding between steel fragments and the aluminum matrix, which will be shown in more details in the weld microstructure analysis session. Accordingly, this corresponds to a low joint strength. In the region adjacent to the steel fragment, dimple structure is observed, which is related to the tearing of the aluminum matrix.

In this section, the peak temperature recorded by the thermocouples at the four positions during the process are plotted and compared under different conditions.

Effects of rotating and welding speeds are shown in Fig. 7. This specifically corresponds to the big tool with smaller tool offset condition. It can be observed that a higher rotating speed and slower welding speed greatly raise the overall temperature profile. Besides, the temperature distribution is more uniform. This explains the elimination of large surface groove defects. For the small tool, temperature under different rotating and welding speeds are compared in Fig. 8. The same trend is observed. A lower rotating speed and higher welding speed reduce the temperature and increase the temperature distribution gradient. This helps explain the presence of crack along the Al–Fe interface at the bottom of the welds obtained in these conditions, which will be shown in Sec. 3.3.

Figure 9 compares the effects of tool size on temperature distribution. This specifically corresponds to the condition of rotating speed of 1800 rpm, 60 mm/min, and 1.00 mm tool offset. Although the peak temperature in the aluminum side near the weld interface is around the same for the two tools, the overall temperature profile is higher using a big tool. Moreover, the distribution is also more uniform. This is because a larger tool directly increases the frictional contact area. Besides, at the same rotating speed, a higher linear velocity is achieved at the outside region of the shoulder with the larger tool radius. Both will generate more frictional heat. Furthermore, a larger volume of material is stirred with a big tool and therefore increases the amount of heat generation from plastic deformation.

Figure 10 compares the effects of tool offsets on temperature distribution, specifically corresponds to the condition of big tool, rotating speed of 1800 rpm, and welding speed of 60 mm/min. By reducing the tool offset and stirring more steel in the nugget, the overall temperature is lower except the 5 mm point in the steel side, since the tool is located closer to this position. As the steel has a much higher deformation resistance, the overall lower temperature indicates that in order to achieve an effective stirring action, other welding parameters need to be adjusted to increase the heat input. However, overheating of the workpiece will also be harmful to the joint performance. Therefore, tool offset is one of the most critical parameters to obtain successful FSW joints between steel and aluminum.

Aluminum to steel FSW welding mechanisms are further revealed with weld microstructure analysis and nano-CT 3D visualization of the material distribution. The weld in Fig. 11 corresponds to the type I weld that failed in the aluminum side. The specific welding condition includes the big tool, large tool offset of 1.65 mm, higher rotating speed of 1800 rpm, and lower welding speed of 60 mm/min. Figure 11(a) shows the 2D cross-sectional view, where the contour of the tool pin, the tool axis, and the original interface between the two sheets are marked out. A continuous steel strip is extruded along the bottom region of the workpiece and embedded in the aluminum matrix. The initial volume of steel located between the left edge of the pin and original faying surface of two materials is expected to be equal to the total volume of the deformed steel in the stirring zone after the weld. These deformed steel material consists of two parts, one is the continuously extruded steel located in the bottom region of the weld and the other is the dispersed steel fragments. Figure 11(b) presents the 3D view of the steel distribution inside of the aluminum matrix. The extruded steel strip observed in 2D actually almost spread an entire flat surface at the bottom of the weld. Steel fragments are dispersed inside of the stirring zone with various sizes. Moreover, these steel fragments can be observed to follow the pattern of the tool rotational and translational movement. Distribution of these steel fragments is not random. The top view shows them align along a series of circular curves. Radius of these curves correlates with that of the FSW tool, and the distance between the consecutive curves is related to the ratio between the tool rotating and moving speed. Most of these steel fragments are located below the middle layer of the weld.

The weld in Fig. 12 corresponds to the condition where a large surface groove defect exists, as shown in Fig. 3(b). The specific welding condition includes the big tool, small tool offset of 1.00 mm, lower rotating speed of 1200 rpm, and higher welding speed of 120 mm/min. Since a big tool and small tool offset is applied, a larger fraction of steel is involved in the stirring zone. Accordingly, the volume fraction of steel fragments in Fig. 12 can be observed to be much higher than that in Fig. 11. Besides, the sizes of these fragments are generally much larger. This could be due to the combined effects of lower rotating speed and higher welding speed. The amount of heat generation from lower rotating speed is small. In addition, the higher welding speed tends to deform the material at a higher strain rate, which fractures the insufficiently softened steel. As a result, the ineffectively stirred steel are broken into large fragments and scattered in the weld.

On the other hand, for the same big tool and smaller tool offset condition, the steel flow behavior can be greatly improved with a higher rotating speed (1800 rpm) and lower welding speed (60 mm/min), as shown in Fig. 13. The hook feature or large steel fragments can be observed at different cross sections of the same weld. Despite some small particles, overall the steel material is in a more continuous bulk form compared with that in Fig. 12. As greater amount of heat is provided with a higher rotational speed and lower welding speed, the steel deformation resistance is reduced and its flowability is improved. The 2D cross section in Fig. 9(a) is at a particular position along the weldline. A large piece of isolated steel fragment is embedded in the aluminum matrix, which initiated crack during tensile tests.

The weld in Fig. 14 corresponds to the failure type IV where the fracture occurs along the Al–Fe interface near the bulk steel side. This condition is usually associated with a small tool and lower rotating speed. A 2D optical microscopic view is provided in Fig. 14(a). Different from other welding conditions, the Al–Fe interfacial line is almost vertical at the bottom of the weld. A higher magnified view with SEM shows a continuous crack layer with the width of approximate 10 μm in this region. This is consistent with the joint fracture behavior observed in Fig. 4(d), where the Al and Fe materials are cleanly separated in the bottom while a small portion of aluminum sticks to the bulk steel near the weld top region. The combined condition of small tool and lower rotating speed directly reduces the amount of heat generation and increases the temperature gradient along the workpiece thickness direction. Since majority of the frictional heat is generated at the tool shoulder, top part of the weld has a higher temperature and enables certain degree of bonding of aluminum to steel. On the other hand, bottom part of the weld is relatively cold without adequate amount of heat conducted downward. The stirring action of the pin only deforms the materials without creation of actual bonding at this low temperature. Therefore, a crack is left at the interface in the bottom of the weld. In addition to the crack, some large pieces of steel fragments are observed from both the 2D view of the cross section and 3D view of the weld in Fig. 14(b). This is similar to the low heat input conditions with the big tool and smaller tool offset in Fig. 12. The relatively low temperature impairs the flowability of steel and results in limited ductility. Under the stirring action of the pin, the steel materials are broken into large fragments and scattered around in the weld nugget.

In this study, FSW of aluminum alloy 6061 to TRIP 780 steel are performed under various process conditions. Two FSW tools with different sizes are used. Four types of failure modes are observed where the fracture occurs at different positions, including in the aluminum side, along the outside boundary of steel hook, along the interface between the embedded large steel particles and the surrounding aluminum matrix as well as along the original interface between the bulk aluminum and steel sheet. The first failure mode corresponds to the highest joint strength, where the dimple structure is observed on the fractured surface, indicating a ductile behavior. The second and third failure modes are associated with a combined ductile and brittle fracture behavior. The third failure mode has the lowest joint strength. A higher magnification SEM view reveals the existence of a long crack at the interface in the bottom region of this type of welds. The nano-CT enables 3D visualization of the steel materials distribution in the weld. With insufficient amount of heat input, large steel fragments are scattered in the weld zone. Higher heat input improves the material flowability. A continuous steel strip is generated, and the steel fragments are much finer. Depending on the tool position, this steel strip can be either in a flat shape near the bottom or form a hook structure.

Special thanks to the United States Steel Corporation for providing the TRIP 780 steel for research propose, and to Clark Andrea for assistance in the nano-CT scanning.

• National Science Foundation (Grant No. 1537582, Joining of Dissimilar Materials through a Novel Hybrid Friction Stir Resistance Spot Welding Process).