Extent of interlocking and metallurgical bonding in friction riveting of aluminum alloy to steel

In this study, the joining of 6061-T6 aluminum alloy and DP590 steel using a M42 steel rivet via friction riveting technique is investigated. The surface morphology and microstructure characterization reveal the formation of an anchor zone that imparts mechanical interlock as well as the formation of metallurgical bonds at the interface of aluminum and steel. A combination of interlocking and bonding results in the achievement of a high load-carrying capacity of 5.7 kN during lap shear testing at room temperature. A finite element-based computational model was developed which accurately predicted the lap shear response of the joint. The model revealed that the metallurgical bond formed during fric-riveting adds 39% peak load strength to the joint. An extensive microstructural investigation, post-lap-shear fractography, and the modeling results, together provided insights on the joint failure mechanism. This study highlights that friction riveting is a promising method for aluminum-to-steel dissimilar joining, which is important for lighweighing automotive vehicles for energy efficiency.


Introduction
The drive to conserve energy, improve performance, and reduce carbon footprint in the transportation industry has reignited the research focus on the application of lightweight materials [1,2]. Recent efforts are focused on multi-material joining, specifically steel with lightweight aluminum (Al), which could provide an optimum combination of lightweight and strength [3][4][5][6]. However, joining Al with steel is challenging due to the large differences in their thermal, physical, and mechanical properties such as conductivity, melting temperature, coefficient of thermal expansion, Young's modulus, and yield strength. Moreover, the formation of brittle intermetallic compounds (IMCs) makes it difficult to form effective joints via conventional melt-based welding techniques [7,8]. Several recent publications discuss solid phase joining techniques such as lap joints by friction stir welding (FSW) of Al-steel and report the formation of IMCs ranging from a few microns to nanometers across the joint [9][10][11][12]. However, the conclusions on the favorable IMC types, their morphology, and their definitive effects on joint strength do not find a consensus. Additionally, there are reports on adhesive bonding and mechanical fastening to join Al to steel [13]. But these processes have their own shortcomings [13]; using adhesives is time-consuming and they degrade over time, while mechanical fastening could develop stress concentrations and become loose in response to vibrations, moisture, and other hazards.
To overcome the challenges of joining dissimilar materials, a promising friction-based joining technique developed at Helmholtz-Zentrum Geesthacht [14] called friction riveting or fric-riveting is highly suitable. This method, depicted schematically in Fig. 1, has been successful in joining a range of different materials, including polymers and metals. Blaga et al. [15] demonstrated the fric-riveting of Al alloys with glass-fiber reinforced polymer using titanium (Ti) rivets. In their work, on securing the rivet head with a stainless-steel washer and nut, the joint had a load-carrying capacity of 4.27 ± 0.6 kN in lap shear tensile test configurations. In another study, Borba et al. [16] investigated the 1 3 joining of carbon-fiber reinforced polyether-ether-ketone laminate single lap joints using Ti alloy rivets and secured each joint with a nut and washer. They achieved the lap shear force of 7.4 ± 0.6 kN, similar to that obtained via a mechanical fastening. While in a different study of the same joint configuration, the fatigue life was higher by 88% in the fric-riveted joint than in the mechanical fastening [17]. Wang et al [18]. demonstrated joining AA5052 to AA5052 using T10 steel rivets using a complex design and achieved a strong bond strength of the joint. Although several combinations of metals and alloys have been joined via fric-riveting, joints between Al alloys and steel have not been reported. Given the importance of Al-steel joints in the transportation industry, it is imperative to use this technique to join the two using a simple rivet design, enhancing the opportunity for high production and low cost. Additionally, very little is known about the mechanism of bonding that is responsible for attaining a high-strength joint.
This study dealt with the successful demonstration of fric-riveting to join AA6061 with dual-phase steel (DP590) for the first time. A three-dimensional finite element (FE) based model was developed to study the effects of mechanical interlocking and anchor zone formation at the interface on joint strength and the corresponding failure mechanisms during lap shear tests. A detailed multi-modal microstructural characterization was conducted at the interface and in the surrounding area to study the effect of metallurgical bonding during fric-riveting process.

Experiments
Al alloy AA6061-T6 (referred to as AA6061 hereafter) sheets and DP590 steel sheets were components of interest in this study, which were to be joined using rods of M42 high-speed steel employed as the rivet material. Sheet samples, 100 mm long and 25 mm wide, were cut from rolled sheets of AA6061 (3 mm thick) and DP590 steel (1 mm thick). The M42 rivet had a diameter of 6.35 mm, and its length was 10 mm. The chemical compositions and mechanical properties of the materials used are shown in supplementary Table S1 and Table S2, respectively.
A FSW machine from Transformation Technologies, Inc. (TTI) was used to perform the friction riveting (referred to as Fric-rivet hereafter) process to join the AA6061 and DP590 sheets. Pairs of sheets were stacked with an overlap of 25 mm and clamped together, having the AA6061 as the top sheet and DP590 steel as the bottom sheet. The fric-riveting process starts with installing the rivet in the spindle head of the FSW machine. Following this, the rivet is lowered into the sheet assembly at a selected location, while rotating at 1500 rpm (rotational speed). The plunge speed of the rivet is varied from 4 -50 mm/min in four stages while applying varying forces to ensure the penetration of the rivet into the joint assembly. To achieve the joint, the rivet is initially drilled into the assembly, after which it is forged in place. The Al-steel fric-riveted joint is then reclaimed for further experimentation. The process parameters used are listed in Table 1 and discussed in Section 3.
For microstructural characterization of the fric-riveted joints, an Olympus SZX16 microscope has been used for optical microscopy (OM), along with a Thermo Fisher Scientific Quanta 200 scanning electron microscope (SEM) outfitted with an Oxford Instruments energy dispersive x-ray spectroscopy (EDS) system for micrographs coupled with electron backscattered diffraction (EBSD) and energy dispersive x-ray spectroscopy (EDS). The samples for microstructure analysis were prepared by sectioning the joint from the middle through the diameter of the rivet, using a wire electrical discharge machine (EDM), in such a way that the cross-section of the rivet, the AA6061, and the DP590 could be observed. Samples were also prepared on the wire EDM for x-ray computed tomography (XCT) scans. The XCT results enable a non-destructive method of detailed analysis of the rivet and material deformation without cutting through the samples.
The performance of the fric-riveted joints was evaluated by conducting lap shear tests on an Instron 5582 servomechanical test framework at room temperature using an Epsilon optical non-contact extensometer. Four samples were lap shear tested to verify repeatability. The crosshead speed was maintained at 1.27 mm/min for all the tests. To deduce the failure mechanism during the lap shear tests, fractured surfaces were observed using SEM and OM.

Finite element-based computational modeling
A computational model has been developed for the quantitative prediction of maximum joint load, ductility, and capture failure mechanisms for different interfacial configurations. These models can provide information on the process parameters needed to generate interfacial features and characteristics to achieve a strong joint. The threedimensional FE model was based on digital images of the cross-section of fric-riveted joints. The model simulated the lap shear test with different boundary conditions-that is, with and without the assumption that a metallurgical bond exists between the M42 rivet and the DP590 sheet during the joining process. Subsequently, the computational results were compared with experimental observations to ascertain the accuracy of the FE model.

Model setup
A three-dimensional FE model was created to study the overall strength of the joint and the failure mechanisms in a longitudinal lap shear configuration. The model included only one-half of the joint (equivalent to sectioning the joint from the middle, along the length of the rivet) to reduce the computational time, even though the joint geometry was not strictly symmetrical with respect to the axis of rotation of the rivet in the experiments. Figure 2 shows an OM image of the joint cross-section as well as a schematic of the geometry, mesh, and boundary conditions for the FE model. All movements (rotation and translation) were restricted at the fixed end and a longitudinal displacement boundary condition of u = 1.0 mm was applied at the moving edge of the FE model. The top AA6061 sheet is denoted with blue elements while the bottom DP590 sheet is denoted with red  Fig. 2(a) was used to extract the interfacial characteristics of the joint and recreate them in the FE model. Since the FE model used a revolution technique to generate the model, the FE geometry incorporated only the right half of the experimental cross-section. Then, an image digitizer tool [20] was used to select 412 points to define the cross-section of the FE model (see Fig. 2(b)). A large number of points allowed a highfidelity transfer to the FE model of all the interfacial features seen experimentally. The interface geometry was defined using multiple cubic splines with continuous second-order derivatives to avoid any artificial stress concentration areas. Figure 2(b) shows the computationally generated rivet and its interfacial features. Figure 2(c) shows the mesh refinement used near the hook features (anchoring zone region) to increase the accuracy of the finite element simulations.
The elastic properties of the materials used in this model were obtained from the literature and are summarized in Table 2 [21,22]. For the M42 steel, an elastoplastic model with isotropic hardening was assumed for the material behavior and Table 3 shows a correlation between yield stress and plastic strain. For the DP590 and AA6061 materials, a Johnson-Cook (J-C) strength model [23] was used to recreate the elasto-viscoplastic material response. These J-C parameters were taken from the literature [21,22] and are shown in Table 4.
In addition, a J-C damage model was added to the FE simulation [25]. In this fracture model, the fracture strain depends on the stress triaxiality ratio. This relationship was established with the fracture parameters shown in Table 4, which was obtained from the literature [21,24]. Once the damage is initiated (a certain threshold is surpassed), the damage model describes the rate of degradation of the material stiffness until a certain fracture criterion is reached. Once the fracture is achieved in a finite element, this element is deleted from the simulation because its loading capacity is zero. The damage parameters used for the M42 commercial steel are the same as those available for DP590. Multiple trials, variations in process parameters, and clamping, details of which have not been discussed here, were used to identify the optimum process parameters. Figure 3(b) presents the process response (Z force) and parameters (plunge speed and rotational speed) as a function of the plunge depth of the rivet during the joining process. As discussed previously, the fric-riveting process response can be seen to have two distinct portions: (a) drilling and (b) forging. The first two plunge steps are drilling the rivet through the AA6061 alloy sheet, using a slow plunge speed of 4 mm/min., then increasing it to 20 mm/min. and using a rotational speed of 1500 RPM. During the drilling phase, high heat generation is required to facilitate the softening of the sheet's surface. Therefore, a high rivet rotational speed along with a comparatively low plunge speed was selected to enable the rivet to penetrate the top sheet. The second phase is the forging step. Here, while the rotational speed was maintained at the same range as in the previous drilling step, the plunge speed was increased to 50 mm/min. Since the rivet has already been heated up by the previous drilling phase, a high plunge  speed at this step enables the plastic deformation of the rivet tip. The tip is also restricted by the lower DP590 sheet which further provides reactive forces to the rivet tip and causes it to deform. Consequently, the rivet tip flares outward, forming a mushroom-shaped feature referred to as the anchoring zone, as shown in the OM image in Fig. 2(a). The sudden increment of Z-force toward the end of the process as shown in Fig. 3 further drives the anchor zone formation. Such a deformation of the rivet provides an interlocking/anchoring effect that strengthens the joint. It is important to note that the plunge depth reads as 4 mm in Fig. 3(b), whereas the rivet plunge depth measured from the OM image ( Fig. 2(a)), was 3.8 mm. This difference exists due to the anchor zone formation which also expanded the rivet tip diameter from 6.35 mm to 8.4 mm. Similar deformation behavior of the rivet tip during friction riveting that enhanced joint strength has been reported in studies that used different rivet materials, such as grade-2 Ti [15], Ti6Al4V [16,17], and AA2024-T351 [26], to join various combinations of polymers and metallic alloys.

Fric-rivet processing
While the rivet expands to form the anchor zone at the interface of the sheets, a second mushroom-like formation was observed in the rivet close to the Al sheet, pronounced with a crack. This crack is undesirable and its influence on the strength of the joint was investigated using the FE model as discussed in Sect. 3.3. This crack forms probably because of the rivet tip plasticization by the frictional heat generated during fric-riveting process, resulting in its bonding with the bottom sheet which causes the rivet failure to initiate at the boundary of the non-plasticized and plasticized region.
The XCT results are shown in Fig. 4(a) and (b) which further clarify the rivet tip deformation. While Fig. 4(a) shows a cross-sectional view of the sample, Fig. 4(b) presents an isometric view of the same. It is evident that the rivet completely penetrates the top Al sheet, but only partial penetration occurs in the bottom steel sheet and deforms at the end, resulting in a hook formation.

Microstructural analysis
The SEM and EDS results of the as-received DP590, AA6061, and M42 are shown in Fig. 5(a)-(c), respectively. Figure 5(a) shows the microstructure of the DP590 steel, with an average grain size of 4.1 ± 0.6 µm. A hard martensite phase (bright region) was dispersed in a ductile ferrite matrix (dark region); this would provide a combination of strength and ductility to the sheet [27,28]. Figure 5(b) shows the microstructure of AA6061, which has elongated grains, as a consequence of the rolling process, with an average grain size of 31 ± 11 µm. AA6061 typically contains magnesium, silicon, iron, chromium, and copper as alloying elements. The EDS maps show homogeneous distribution of the alloying elements, with sporadic high concentrations of Cr, Fe, and Si. These Table 4 Johnson-Cook mechanical properties [21,22] and damage parameters [21,24] Material  . 3 a Top and front views of AA6061-DP590 fric-riveting joint sample using an M42 steel rivet. b The fric-riveting process control parameters (plunge speed and rotational speed) and Z-force response as a function of plunge depth high-concentration regions show the presence of secondary phase particles such as Mg 2 Si and AlFeSi. The formation of such secondary phases in AA6061 has also been observed in the literature [29,30]. The presence of Fe as an impurity is primarily because of the increased use of recycled Al to manufacture different parts [31]. Fe, having low solubility in Al matrix, is observed as secondary particles. Figure 5(c) shows the microstructure of M42 steel, where the bright spots are rich in Mo and W carbides. M42, being high-speed steel (HSS), differs from DP590 or other types of steel in that HSS contains high percentages of carbides, providing the steel with high wear resistance and excellent hot hardness. This enabled the M42 rivet to completely penetrate the AA6061 sheet and partially penetrate the DP590 steel sheet. M42 steel's microstructure also contains lath martensite that provides it the high hardness necessary for riveting operations [32]. The evolution of microstructure in both sheets, the rivet, and the interfaces after fric-riveting process is depicted in Figs. 6 and 7. The OM images of the cross sections shown in Fig. 6(a) and Fig. 7(a) are outlined with different colored squares depicting the regions from which SEM images have been obtained. Microstructural characterization is performed from the following regions: A) DP590 sheet immediately below the rivet. The microstructure of DP590 sheet directly below the rivet is shown in Fig. 6(b). The localized thermomechanical  deformation of DP590 during fric-riveting and the subsequent natural atmospheric cooling result in the formation of deformed martensite from the initial equiaxed microstructure. Hence, local variations in the mechanical properties of DP590 could occur after fric-riveting. The variation in mechanical properties of DP steels due to different microstructural morphology and volume fraction of martensite and ferrite phases have been extensively investigated in the literature [27,28]. However, in the present study, this change in microstructure has not been the limiting factor in the joint strength achieved.
The interface between the rivet and the bottom sheet is shown in Fig. 6(c) and (d). As the rivet plunges into the Al sheet, it softens the Al due to frictional heat, and thus a thin layer of Al bonds to the rivet tip. This further interacts with the bottom DP590 sheet resulting in the formation of a mixed layer between the bottom sheet and the rivet interface as seen in Fig. 6(c) and (d). This is confirmed by the EDS results (see Fig. 6(c1)). It is possible that the IMCs layer also forms in the mixed layer, but it is quite thin to play any role in the joint strength. The influence of IMCs on joint strengths of Al and steel has been observed in the selfpiercing riveting of AA6061 and DP590 conducted by Lou et al. [10] and in the joining of Al with zinc-coated steel by using a modified metal inert gas welding-brazing process by Zhang et al. [7]. In both studies, it was found that thin IMC formation did not affect joint strength. Figure 6(c) and (d) also show the formation of a metallurgical bond between Al and steel. The plasticized Al acts as a filler between the M42 rivet and DP590 sheet and helps in forming a metallurgical bond which provides additional strength to the joint apart from the strength achieved by mechanical interlocking of the rivet with the sheets.
The microstructure of the mechanically interlocked hook region in the anchoring zone is shown in Fig. 7(b), while Fig. 7(c) is a high-resolution image of that hook region and shows intermixing of the DP590, M42, and AA6061 alloys. The EDS results again confirm the presence of an Al layer between DP590 and M42. The shear-induced mixing during fric-riveting forms the observed swirly pattern and nanofilaments of Al and steel. The hook region shows the excellent bond formation and intermixing of the sheets and the rivet and highlights an atomic scale compositional joining instead of just a mechanical hook joint and is responsible for the high joint performance. The AA6061 grains show significant refinement by localized heating near the hook region, with an average grain size of 1.11 µm.
The microstructure evolution in the AA6061 sheet near the rivet is shown in Fig. 7(d). Crystallographic texture measurement showed a difference in grain orientation and grain refinement compared to the as-received AA6061. The grain size near the rivet is 5.36 µm, whereas it is 31 ± 11 µm in the rest of the material. The changes in microstructure result from the significant plastic deformation caused by shearing forces and frictional heat generated due to the rivet rotation during the fric-riveting process. The M42 rivet showed limited variation in microstructure observed under the magnification of SEM as it has excellent hardness and wear resistance. The dark region between the AA6061 sheet and M42 rivet (refer to Fig. 7(d)) is the gap between the top sheet and the vertical portion of the rivet. Figure 8 shows the experimental and simulated load vs. displacement curves for the lap shear test. The average maximum load obtained experimentally was 5.7 ± 0.16 kN. The joint strength obtained in the present study is comparable to that found in the literature, as shown in Table 5. It is evident that the thicknesses of bottom sheets have a significant effect on the lap shear strength obtained. A thicker bottom sheet provides more volume for the rivet to hold, hence resulting in higher strength. Kim et al. [33] describe work with a high average lap shear load of 9.6 kN, using 3 mm thick bottom sheets made of various Al alloys. When the bottom sheet thickness was lower (1.2 mm), the average lap shear strength obtained was 3.8 kN, and the failure occurred when the fastener was pulled out as it could no longer hold the bottom plate. Similar effects of bottom sheet thickness on lap shear load are seen in other joining methods, as listed in Table 5. Although it must be noted that too thick a lower sheet is detrimental to the joint performance as it increases the mechanical constraint of the joint and reduces secondary bending during testing. Secondary bending causes a "peel" loading condition which has a negative influence on the joint performance. Zhang et al. [34] reported joining a combination of AA6061-T6 and DP590 sheets via metallic bump-assisted resistance spot welding and resistance spot welding, where the lap shear load was significantly lower than that obtained in the present study. The high lap shear joint loads obtained in the fric-riveted joints investigated here are attributed to the following reasons:

Lap shear tests
(a) Formation of an anchoring zone at the rivet tip results in hook formation at the interface of the three materials. This provides mechanical interlocking, imparting strength to the joints. Such behavior has also been observed in other friction riveting studies conducted by dos Santos et al. [15][16][17]26] for other material combinations. (b) The fric-riveting process forms a metallurgical bond between the bottom portion of the rivet and the sheets, which enhances joint strength.
The simulated load vs. displacement curves overlapping the experimental results in Fig. 8 are plotted for two conditions: (1) no bond forms between the rivet and the sheets (shown by the blue dotted line in the figure); (2) bond formation does occur (shown by the black dotted line in the figure). Comparing the load-displacement curves, the presence of the metallurgical bond results in a higher peak load of the joint than in the case without a bond. Figure 8 also allows a comparison of the computational load-displacement curves for cases with and without a metallurgical bond against experimental results. There is good agreement with the experimental results for the FE case with metallurgical bond. A possible reason for the small discrepancy in the initial stage of the curve in Fig. 8 is that the model assumed a symmetrical sample while the real sample may not be symmetrical. These results validate the assumption of a perfectly bonded interface between the rivet and the bottom sheet. In addition, a comparison of the two computational curves indicates that the performance increase due to the presence of the metallurgical bond can be computed. The metallurgical bond adds a total of 39% peak load strength to the joint.
The fractography results and EDS analysis of a lap shear tested sample are shown in Fig. 9. Figure 9(a)-(d) shows the SEM images taken from the DP590 sheet post lap shear testing. Figure 9(a) and (b) show the fractured surface morphology at the center of the riveted region, whereas Fig. 9(c) and (d) show the same from the edges of the riveted area. The microstructure shows a riverine pattern in the center that indicates cleavage fracture, whereas both riverine and dimple morphology can be observed near the edge (anchor zone circumference), evidencing both ductile and brittle failure of the joint. Dimples, which indicate ductile failure, arising from (1) the pre-existing crack that formed during the rivet forging phase of fric-riveting where the temperature is high; (2) failure during the lap shear test. The EDS results of the failure surface are shown in Fig. 9 (d1-d12). It is interesting to note that bonded Al is found near the edges. The elemental analysis also shows that part of the rivet is bonded to the DP590 sheet. This indicates that failure occurred by shearing the rivet. The formation of cracks in the rivet during fric-riveting process helps with the initiation of failure during the lap shear test, and the presence of inclusions in the M42 rivet results in brittle failure of the material. Failure within the rivet could also result from a strong metallurgical bond formed between the DP590 and the rivet. Similar fracture morphology has been reported in friction stir riveting [36] and friction element welding [37] of Al alloys and steel. Results of OM morphological analyses are shown in Fig. 9(e) and (f). The distance between the deepest point in the bottom sheet observed by OM and the fractured surface is a result of the M42 rivet being attached to the bottom sheet. This provides further evidence of a strong metallurgical bond between the rivet and DP590 sheet. In Fig. 9(e) and (f), the DP590 sheet is taken as the base reference layer and the measured peak heights are with respect to this reference layer. The DP590 sheet bent slightly during lap shear testing, which is evident in the height (color) difference in Fig. 9(f). Here the maximum height of the rivet remnant attached to the bottom DP590 sheet is 833 µm. The deepest point is − 333.94 µm. This value is smaller than the rivet penetration depth in the bottom sheet, which is 650 µm, as measured from the OM in Fig. 2. Fig. 10(a) and (b) for illustrative purposes. As can be observed in Region 2, the interface between the rivet and the bottom sheet is completely separated. That is because the model assumes no metallurgical bond, and the bending of the sheets forces the surface separation. Then, all the interfacial strength of the joint is caused by the presence of the left hook (Region 1). This interfacial feature creates a mechanical interlocking between the sheets and the rivet. From both figures, there is noticeable stress concentration and severe plasticity development at the center of the rivet, where there are two crack tips. However, the plastic deformation observed in Fig. 10(b) is greater in the left hook (Region 1) than at the center of In the second case study, the model assumes that the interface between the rivet and the DP590 is a perfect bond. Figure 2(b) shows the rivet surface to be bonded highlighted in red. This was achieved in Abaqus/Explicit by applying a tie constraint between the two surfaces in the model [19]. This constraint ties two surfaces together so that there is no relative motion between them. This modeling assumption comes from postmortem examination of experimental samples in which the bottom part of the rivet was still bonded to the DP590 plate after the lap-shear test; see Fig. 9. Figure 10(c) and (d) show the von Mises stress and equivalent plastic strain (PEEQ) distributions, respectively, for a cross-section of the bonded area obtained using the 3-D FE model. In this case, there are three regions of interest. Region 1 corresponds to the interface between the rivet and the DP590 plate. Region 2 is the center of the rivet, where there are two crack tips, and Region 3 denotes the left hook. As in the previous case, the differential bending between sheets and the shear load causes the interface to fail. However, with a bond between the rivet and the bottom sheet, the failure mode is different. There are two failure modes competing against each other: rivet fracture and left hook failure. As shown in Fig. 10(c) and (d), the center rivet material that connects the top and bottom mushroomed rivets has highstress concentration and plastic deformation. The damage of this location governs the lap shear failure mode of the entire joint, which agrees well with the experimental observation in Fig. 9 (e) and (f). Numerical results reveal that most part of the sheared rivet has no plastic deformation, indicating a brittle fracture is likely to occur along the sheared rivet interface, which is also consistent with the experiment (see Fig. 9 (a) and (b)). Once the rivet is fractured, the left hook can still carry the load until bending causes the AA6061 and DP590 to separate. The remnant hook morphology predicted by the FEM model can be observed from the surface morphology image given in Fig. 9 (f). Severe plastic deformation is predicted by the model in the hook area, indicating ductile fracture is likely to take place at this location. This explains the dimple morphology observed at the edge of the riveted region in Fig. 9. (c). Moreover, a fric-riveting process modeling in combination with the consequent lap shear simulation can help better understand the relations between process parameters and joint strength in order to help optimize the fric-riveting process. However, this process modeling needs advanced numerical methods [38,39] to handle the complicated physics and large material deformation, which is our ongoing work.

Conclusion
This work presented a successful demonstration of the fric-riveting technique to join AA6061-T6 with DP590 steel sheet using an M42 steel rivet. This method has the Fig. 10 Von Mises stress and equivalent plastic strain distributions at peak load for the case without bonding a and b and the case with a bonded interface c and d potential to drastically enhance the mechanical properties of dissimilar joints. The microstructure and joint performance were investigated, and finite element modeling supported the experimental results well. The main findings are summarized as follows: 1. Fric-riveting is a promising technique for joining dissimilar metals, which are challenging to join owing to their vastly different thermophysical properties. 2. High strength joint between AA6061-T6 and DP590 steel was achieved. The formation of the anchoring zone, metallurgical bond, mechanical interlock, and mixing of Al and steel are the main reasons for high strength and providing fric-riveting technique an advantage over other metal joining processes. 3. The computational model matches the behavior of experimentally observed lap joints when the model assumes a perfect bond between the rivet and the bottom sheet. The peak load of the joint increased by 39% with a metallurgical bond at the rivet/bottom-plate interface.
In addition, the model can capture failure mechanisms for different interfacial configurations. If no bonding is assumed, the interface completely separates. If a metallurgical bond is present between the rivet and the bottom sheet, fracture occurs in the rivet. 4. Based on the results from the 3D finite element model, a combination of parameters can affect whether the lap joint fails at the weld interface or in the material. Several different simulations indicate a fine balance between the hook morphology and interfacial parameters that can result in one failure mode or the other. Future work on three-dimensional computational modeling of fric-riveting may provide a better prediction of the behavior of the welded joints.