Friction self-piercing riveting (F-SPR) of aluminum alloy to magnesium alloy using a flat die

Bingxin Yang, Yunwu Ma, He Shan, Sizhe Niu, Yongbing Li,?

a Shanghai Key Laboratory of Digital Manufacture for Thin-walled Structures, School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, PR China

b State Key Laboratory of Mechanical System and Vibration, School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, PR China

c Joining and Welding Research Institute, Osaka University, Osaka 567-0047, Japan

Abstract

Keywords: Friction self-piercing riveting (F-SPR); Flat die; Aluminum alloy; Magnesium alloy; Mechanical joining; Solid-state bonding.

1.Introduction

With the emphasis on energy-saving and emission reduction in the world, the lightweight design of vehicle bodies has become a trend in the automotive industry [1].High specific strength metals, such as aluminum and magnesium alloys,have been increasingly applied to achieve the mass reduction target while maintaining sufficient rigidity of the body structures.To date, many automotive parts such as steering wheel frames, suspension brackets, and door frames are made of lightweight alloys.According to the US Department of Energy, aluminum and magnesium alloys will account for more than 30% of the automotive body materials by 2035 [2].Due to the large differences in physical and chemical properties between Al and Mg, the joining of these dissimilar metals poses a huge challenge to vehicle body assembly.Fusion welding of Al and Mg is very difficult due to the easy formation of thick and brittle intermetallic compounds (IMCs)[3].Therefore, more attention is paid to solid-state welding and mechanical joining technologies for Al-Mg joining.

Friction stir welding (FSW) is a widely recognized solidstate method for dissimilar material joining.As a variant of FSW, friction spot welding (FSpW) [4]has been developed for spot joining.FSpW process uses a three-part tool com-prising a pin, a sleeve, and a clamping ring to form a lapped joint with a flat surface.However, the inherent hook-shaped defects are created by the back-and-forth movement of the tool and left in the final joint as a low-energy crack path,causing a detrimental effect on the mechanical performance of the joint.Suhuddin et al.[5]found that the hook-shaped defects were caused by the bending of the lap joint during welding.Cao et al.[6]reduced the negative effect of the hook-shaped defects by optimizing the penetration depth of the sleeve and improved the joint mechanical performance.

Self-piercing riveting (SPR) is a mechanical joining technology that has already been widely used in joining aluminum alloys [7]as well as Al-steel [8]dissimilar materials.Mori and Abe [9]reviewed the SPR process of aluminum alloy with high-strength steel sheets, and pointed out that by optimizing the die shape to control the plastic deformation, an effective interlock between the rivet and the sheets can be realized.Haque [10]described the key parameters which affect the SPR joint quality and introduced several assistive technologies for the improvement of the joint quality.However,when the SPR process is used to rivet low ductility materials such as 7 series aluminum alloy or AZ31B magnesium alloy, the joints are prone to cracking due to large deformation during the SPR process.Although crack-free joints can be obtained by preheating the sheets to about 200 °C using an AC current [11]or a laser heat source [12], these preheating SPR methods not only complicate the joining process but also increase the process cost, limiting their use in production.

Li et al.[13]proposed the friction self-piercing riveting(F-SPR) process by introducing the high-speed rotation motion of rivet into the SPR process.Frictional heat is generated during the rivet rotation, which not only softens the low ductility material to inhibit cracking but also promotes the formation of solid-state bonding in the joint, producing a mechanical-solid state hybrid joint.Based on this process,low ductility materials, such as AZ31B magnesium alloy and AA7075-T6 aluminum alloy, were studied, and high-quality joints without cracks were obtained, indicating the feasibility of the F-SPR process.By means of simulation and experiment, the joint formation process [14], the mechanical performance [15], the effectiveness of single-sided process variant[16], and the function of mechanical-solid state hybrid joining on the fracture behavior of the joint [17]were analyzed in detail, which provided a comprehensive understanding of the process.Liu et al.[18]also took the F-SPR process to join AA7075-T6 aluminum alloy and AZ31B magnesium alloy, and Al-Mg IMCs were observed in the obtained joints,the same phenomenon was explained by Li et al.[13].Recently, the F-SPR process was extended to join carbon fiber reinforced plastic (CFRP) and Mg alloy [19].

However, all these existing F-SPR processes use pip dies,which required a high rivet-die coaxial accuracy to ensure joint quality.In real production of vehicle bodies, the rivet driver and the die of F-SPR are generally mounted to the two opening ends of C-frame equipment.A large riveting force and torque can cause deflection at the C-frame opening,which creates a radial misalignment condition between the rivet and the die.Thus, the sheet material near the pip die may be penetrated through by the rivet, or the rivet may contact with the pip die,causing unwanted die wear,as shown in Fig.1(a).

In this paper, the F-SPR process using a flat die was first proposed.As shown in Fig.1(c), the flat die in this study is very wide and different from the flat bottom die, also referred to as the DZ-die in the SPR process,which is shown in Fig.1(b).The goal of using a wider flat die surface is to avoid the misalignment problem, whose radial error can be up to 2mm in the pip-die-based F-SPR process.

To better understand the factors influencing joint performance in the flat-die F-SPR process, commonly used AA6061-T6 aluminum alloy was joined to low ductility AZ31B magnesium alloy in the present study.The effects of die distance and rivet feed rate were analyzed and then correlated to the riveting force as well as torque during the F-SPR process,which was then further related to the hardness distribution of the joint, microstructure at dissimilar material interfaces, and joint mechanical performance.Finally, three types of the F-SPR joints were compared by using a pip die,a flat bottom die, and a flat die respectively in the case of a 1.5mm misalignment between the rivet and the die.

2.Experimental procedures

2.1.Materials and rivets

1.27mm-thick AA6061-T6 aluminum alloy and 3mm-thick AZ31B magnesium alloy were used as the upper and lower sheet, respectively.The rivet was made from 35CrMo steel without coating and designed with a patented cover to transmit the driving torque to a semi-hollow body shank.Key dimensions and physical images of the rivets are given in Fig.2.The chemical composition and mechanical properties of the rivet and sheets are listed in Tables 1 and 2, respectively.

Table 1Chemical composition of the sheet materials and rivets (wt.%).

Table 2Mechanical properties of the sheet materials and rivets.

2.2.Flat-die F-SPR process and experimental setup

The flat-die F-SPR process, as shown in Fig.3, consists of the following steps:

(1) Positioning.The rivet is matched with the driver, and the overlapped workpieces are placed between the rivet and the support.The flat die is fixed at a distanceDbelow the bottom workpiece, refer to Fig.3(a).

(2) Softening.The rivet is driven to rotate at a spindle speedωand meanwhile to feed at a rateftowards the workpieces.The workpieces are softened by the generated frictional heat, refer to Fig.3(b).

(3) Deforming.The displaced workpieces contact with the flat die and the rivet leg opens outward due to the feeding resistance, refer to Fig.3(c).

(4) Stopping.The feeding and rotation movements are ceased abruptly once the rivet is fully inserted, i.e., the plunge depth reaches a predefined distancex, refer to Fig.3(d).

Fig.1.Schematic diagram of the F-SPR joint obtained by different dies with misalignment: (a) Pip die, (b) flat bottom die, and (c) flat die.

Fig.2.Rivets: (a) cross-section with key dimensions, and (b) physical appearance.

(5) Releasing.The driver feeds back to its initial position.A mechanical-solid state hybrid joint is formed, refer to Fig.3(e).

As described above, the flat-die F-SPR process has four key parameters, namely die distance (D), rivet rotation speed(ω), feed rate (f), and plunge depth (x).The total energy (E)can be obtained by summing up the energy generated by the riveting force (EF) and the energy generated by the torque(EM):

whereFandMrepresent the instantaneous riveting force and torque during the riveting process, respectively.

In the current research, the effect of die distance on joint formation was investigated firstly by fixing the feed rate and rivet rotation speed, refer to test # 1–3 in Table 3.Then,an optimized die distance was used in the subsequent study.Previous research [15]found that the feed rate (f) was the dominant factor in determining total energy.Therefore, the current research only studied the effect of feed rate on the joints characteristics, refer to test # 2 and 4–7 in Table 3.The plunge depth in all experiments was fixed at 5.5mm.

Table 3Process parameter combinations used in this research.The plunge depth is 5.5mm for all tests.

Fig.3.Schematic diagram of the flat-die F-SPR process.(a) Positioning, (b) softening, (c) deforming, (d) stopping, and (e) releasing.

Fig.4.F-SPR system: (a) overall set-up, (b) close-up view of the experiment platform, and (c) close-up view of the rivet and flat die.

The F-SPR system consists of a control unit and a Cframe, as shown in Fig.4(a).At the lower end of the Cframe opening is the experiment platform, where workpieces are fixed,refer to Fig.4(b).The flat die is an M18 hex-shaped bolt fixing in the center of the 30-mm-diameter hole of the platform, and the vertical distance of the flat die surface is adjustable to achieve variousD, as shown in Fig.4(c).Since the die diameter is much larger than the rivet diameter, there is no need to ensure axial alignment of them.Moreover, the F-SPR system is equipped with a force sensor and a torque sensor for monitoring the riveting force and the driving torque throughout the riveting process.

2.3.Joint quality evaluation

The joints are evaluated in terms of macroscopic morphology, microstructure, microhardness, and lap-shear performance.Macro profiles of the cross-sectioned joints were imaged using a Leica optical microscope after standard grinding and polishing procedures.Two geometrical indexes, i.e., interlock amount and bottom thickness, were used to evaluate the mechanical joining quality, as shown in Fig.5(a).The interlock amount represents the distance of the rivet shank flaring relative to the original size.The larger the interlock amount, the stronger the mechanical interlocking between the rivet and the workpieces.The bottom thickness refers to the vertical distance between the rivet leg tip and the bottom surface of the joint.The smaller the bottom thickness value, the greater risk of cracking at the joint bottom.Finer scale observations of the microstructure were conducted on a VEGA 3 (LaB6) scanning electron microscope (SEM) with an acceleration voltage of 15keV.Hardness mapping of the rivet and the surrounding materials was measured on a WILSON VH1102 Vickers hardness tester using the pressure of 0.3 kgf and 0.05 kgf, respectively.

Fig.5(b) presents the dimensions of the lap-shear specimen.The workpieces were cut into 130×38mm coupons.A single rivet was riveted at the center of the overlapping square area.To decrease the influence of bending stress in lap-shear testing, a pair of spacers with 1.27mm and 3mm in thickness was placed on the gripping end of the Mg and the Al sheets,respectively.The lap-shear test was carried out on a SUNS UTM5504X electronic universal material testing machine at a tensile rate of 3.0mm/min, and three samples were repeated for each condition.

3.Results and discussions

3.1.Joint morphology

Fig.6 presents the cross-section profiles of the flat-die FSPR joints made with different die distance values, refer to test # 1–3 in Table 3.Obviously, the die distance has a significant effect on the rivet deformation and mechanical interlocking.At the die distance of 0.5mm, the deformation and downward movement of the bottom Mg sheet was constrained by the flat die.As a result, the rivet was severely upset and an obvious gap between the two sheets was formed in thejoint, refer to Fig.6(a).Differently, when the die distance was 1.0mm, more magnesium was squeezed to the joint bottom and the rivet leg fully flared,forming a sound mechanical interlock with the bottom sheet, refer to Fig.6(b).However,when the die distance was 1.5mm, the rivet was barely flared due to insufficient feeding resistance,refer to Fig.6(c).Therefore, the die distance of 1.0mm was used in the following study.

Fig.5.Joint quality evaluation standard: (a) geometrical indexes, and (b) lap-shear specimen geometry (Unit: mm).

Fig.6.Cross-section profiles of the joints under different flat-die distances: (a) 0.5mm, (b) 1.0mm, and (c) 1.5mm.

Fig.7 shows the joint morphology at various feed rates,refer to test # 2 and 4–7 in Table 3, and the interlock amount and bottom thickness are summarized in Fig.8.It can be seen that when the feed rate was 2.0–8.0mm/s, the rivet penetrated the lower sheet successfully, and the joint exhibited increased interlock amount as well as the bottom thickness with feed rate.However, the rivet leg was upset and failed to penetrate the Mg sheet at 10mm/s feed rate.This can be explained by the amount of generated heat.In the F-SPR process, heat is mainly generated by the rotation of the rivet.A lower feed rate corresponds to longer process time.Therefore, a larger amount of frictional heat is generated and accumulated surrounding the rivet to soften the workpieces and reduce the feeding resistance.As the feed rate increases, the workpieces are less softened, resulting in a larger feeding resistance to the rivet, which deforms the rivet leg more greatly and forms a larger interlock amount.The over-deformed rivet leg at 10mm/s feed rate indicates an insufficient heat generation.Moreover, a larger interlock amount also leads to a decreased rivet insertion depth in the lower sheet, and consequently, the bottom thickness increases.Higher heat input also caused burrs and inter-sheet gaps, which are particularly noticeable at the feed rate of 2.0mm/s and 4.0mm/s, refer to Fig.7(a) and (b).

Table 4Summary of interface microstructures at different locations.

3.2.Joint formation process

To further explore the flat-die F-SPR process, the joint formation process which reveals the material flow pattern at 1mm, 2mm, 3mm, 4mm, 5mm, and 5.5mm plunge depth under the 6mm/s feed rate was presented in Fig.9.At the beginning of the process, the rivet penetrates into the topAl layer, squeezing the Al material into the rivet cavity, at which time the lower sheet and the rivet are substantially free of deformation.When the plunge depth is 2mm, the Al sheet is penetrated through by the rivet, and a gap is formed between the two sheets.When the rivet plunges to 3mm, the sheet materials are heated and displaced upward to fill the rivet cavity, and the lower sheet continues to bend towards the flat die.When the plunge depth continues to 4mm, a large amount of the Mg material is extruded into the flat die,and the rivet leg starts to expand under the increased feeding resistance.With the further plunge of the rivet, the rivet leg opens quickly and reaches its maximum value at the end of the process.Besides, when the rivet head contacts with the upper Al sheet, burrs are created by the high-speed rotation movement.

Fig.7.Cross-section profiles of the joints under different feed rates: (a) 2mm/s, (b) 4mm/s, (c) 6mm/s, (d) 8mm/s, and (e) 10mm/s.The die distance of 1mm was adopted.

Fig.8.The interlock amount and bottom thickness as functions of the feed rate using the 1mm die distance.

Fig.9.Joint formation process under 3000rpm – 6mm/s and plunge depth at (a) 1mm, (b) 2mm, (c) 3mm, (d) 4mm, (e) 5mm, and (f) 5.5mm.The blue arrows show the flow direction of the base material.

As functions of the plunge depth, the riveting force as well as the torque under 2.0–8.0mm/s feed rates are plotted in Fig.10, showing different variation trends respectively.Obviously, the riveting force increases with feed rate due to the decreased amount of heat generation at a higher feed rate.In the F-SPR process, the increase of plunge depth results in a greater contact area between rivet and sheet,which contributes to the increase of rivet feeding resistance.Meanwhile, the work materials are softened by the generated frictional heat,which decreases the rivet feeding resistance.Consequently,the riveting force increases slowly at the initial stage.When the rivet plunges to 3.0mm,the downward movement of work material is constrained by the flat die, producing additional resistance against the rivet feed motion.Therefore, the riveting force starts to increase faster after 3.0mm.The turning point and peak point of the riveting force are summarized in Fig.11(a).Obviously, both the two points increase with the feed rate.This can be explained by the heat input.The increase of the feed rate caused the friction heat generation time at the same plunge depth to decrease, so a lower temperature was reached, and the effect of thermal softening to the material was reduced correspondingly.The torque curve is also the reflection of the competition between the frictional thermal softening effect and the material’s deformation resistance as the increase of plunge depth.Each curve shows an initial peak and a succedent trough, indicating that the aforemen-tioned balance was reached twice in the whole process.The first peak and the first trough torques at different feed rates are summarized in Fig.11(b), where both peak and trough values increase with the feed rate, which is consistent with the interpretation of the riveting force.

Fig.10.Variation of the riveting force and torque with the plunge depth.

Additionally, Fig.11(c) summarizes the corresponding plunge depth at the extreme point in the riveting force and torque curves with the feed rate.It can be seen that the peak riveting force is obtained at the end of the process regardless of the feed rate.However, for the torque curve, the plunge depth at each extreme point increases with the feed rate.The peak point occurs when the thermal softening effect is just stronger than the deformation resistance.The increase in the feed rate requires a larger plunge depth to accumulate the same heat, so the plunge depth at the peak point increases with the feed rate.Conversely, the trough point appears when the thermal softening effect is at a disadvantage.Increasing the feed rate will reduce the time for frictional heat generation at the same plunge depth, and reduce the thermal softening of the material, so the curve declines more gently, and the plunge depth at the trough point increases with the feed rate.What is worth mentioning, since a large amount of heat was accumulated at the feed rate of 2mm/s, leading to the frictional thermal softening effect stronger, a second peak point appeared at about 4.8mm compared to other feed rates.

3.3.Microstructure

To identify the solid-state bonding conditions and alloying elements at different interfaces of the F-SPR joint, EDS line scanning analysis was performed on the joint under 6.0mm/s feed rate, refer to Fig.12.The interface characteristics are summarized in Table 4 as well as Fig.13.For the definitions of contact type, “Dynamic” means the two parts in contact have relative motions during the F-SPR process, whereas“Static” corresponds to a static contact state throughout the F-SPR process.

Location A shows the interface between the upper Al layer and the rivet head, where a 3μm wide gap is presented, refer to Fig.12(a).Location B is the interface of the rivet leg and the upper Al layer in the rivet cavity.A sharp slope but no significant plateau between Al and Fe was observed,refer to Fig.12(b), which indicated that a 3μm-thick atom interdiffusion layer was formed.

Location C and D show the interfaces between the rivet leg and the lower Mg layer.As shown in Fig.12(c) and 12(d),a certain amount of Al was squeezed into the lower sheet as the rivet plunged in, forming an Al interlayer between the rivet leg and the Mg layer.The EDS line scan exhibited a sharp slope at the Fe-Al interface for both locations C and D,suggesting the interdiffusion between Al and Fe.Besides, extreme mechanical stirring, plastic deformation, and frictional heating caused interdiffusion of Mg and Al atoms near the Mg side, eventually forming a mechanical mixture area.Among them, the amount of the two atoms changed slowly within a certain range, revealing that an Mg-Al IMC layer was formed[20].Since the temperature in the rivet cavity was higher than that outside, the IMC layer was thicker in location C than location D.

Location E shows the interface between the upper Al layer and the lower Mg layer on the outside of the rivet leg.Asshown in Fig.12(e),an approximately 12μm thick IMC layer was observed at the Al-Mg interface.The proportion of Al atom varied between 40% and 65% in the EDS line scanning result, indicating that the anticipated IMC are Al12Mg17and Al3Mg2.The contact surfaces at location E were always stationary during the riveting process, promoting the diffusion of the atomics and the growth of the thick IMC layer [13].Moreover, the maximum temperature of the F-SPR process was close to the melting point of AA6061-T6 aluminum alloy, i.e., 582°C [21, 22], which promoted the formation of IMC [23, 24].Location F is the interface between the upper Al layer and the lower Mg layer in the rivet cavity, where the mechanical mixture of Al and Mg was observed, as shown in Fig.12(f).To further evaluate the material distribution, the EDS mapping of Al and Mg was performed.As shown in Fig.14, several Al fragments were inserted in the Mg sheet,and a large amount of Al diffused into the Mg base materials, which indicated the intense motion and high friction heat accumulation at this interface.

Fig.11.Key points in (a) riveting force curves and (b) torque curves of Fig.10, and (c) corresponding plunge depth at extreme points.

3.4.Hardness distribution

The hardness of Al sheets, Mg sheets, and rivets in the joints under different feed rates are shown in Fig.15(a1)–(a4),(b1)–(b4), and (c1)–(c4), respectively.As listed in Table 2,the average hardness of the as-received upper Al layer, the lower Mg layer,and the rivet are respectively 110 HV,55 HV,and 252 HV.It can be seen that the Al materials inside the rivet and outside surrounding the rivet are obviously softened.Softening of Al sheets is because the solubility of the alloying elements in the heat treatable 6061-T6 matrix increases as the temperature increases, and the strengthening phase grows and dissolves in the aluminum matrix so that the precipitation strengthening effect becomes lower.Besides, the thermal activation accelerates the dynamic recovery of the aluminum alloy, which leads to a decrease of dislocation density and the elastic strain energy and results in a decrease in deformation resistance.That is, dynamic softening counteracts the effect of work hardening.Moreover, due to the slower heat dissipation inside the rivet, which makes the temperature of the inside Al materials higher, and then causes the precipitation phase to grow and dissolve more seriously.Therefore,the Al material inside the rivet is more softened than that on the outside.As the feed rate increases, the softening zone becomes gradually narrow due to the decrease of total heat input.Conversely, unlike the upper Al alloy, the Mg material surrounding the rivet is more or less hardened.During the riveting process, the Mg alloy is affected by both fine-grain strengthening and work hardening.On the one hand, the heat generated by the rotation of the rivet refines the grain, and grain refinement suppresses local plastic deformation,forming a large number of grain boundaries, resulting in an increase in hardness [25].On the other hand, the extrusion of the rivet causes work hardening of the Mg alloy, which increases the hardness of the material.The increase of feed rate results in both higher riveting force and less heat input.As a result, the magnesium alloy is more work-hardened with the increase of feed rate.Moreover, the Mg alloy in the rivet cavity undergoes larger deformation and thereby exhibits higher hardness compared to the outer material.However, the different feed rates have not caused an evident variation on the rivet hardness, which is different from the base materials.It is worth pointing out that the hardness of both the inside and the topof the rivet is higher, which is mainly caused by the cold heading forming process.

Fig.12.SEM images and EDS line scanning analysis at specified locations: (a)–(f) correspond to locations A-F, respectively.

Fig.13.Solid-state bonding characteristics in the joint.

Fig.14.EDS map scanning of major elements: (a) the distribution map of Al, and (b) the distribution map of Mg.

3.5.Mechanical performance

Fig.16.(a) Lap-shear load-displacement curves under different feed rates,and (b) relationship between peak load & energy absorption and the feed rate.

Fig.16(a) shows the lap-shear load-displacement curves of the flat-die F-SPR joints of test # 2 and 4–6 in Table 3, and Fig.16(b) summarizes their peak load and energy absorption.The joints exhibited three fracture modes, i.e., the lower sheet fracture(LSF),the rivet pull-out fracture(RPF),and the upper sheet fracture (USF).

As shown, the lap-shear peak load decreases with feed rate from 2.0mm/s to 8.0mm/s.However, the lowest energy absorption occurs at a 4.0mm/s feed rate.Mechanical joining performs a big role on the lap-shear performance of the flat-die F-SPR joint.As important indicators for evaluating the mechanical joining effect, both the interlock amount and bottom thickness increase with the feed rate.When the feed rate is 2mm/s, the bottom thickness is too thin and becomes the weakest area so that the failure occurs at thethinnest portion of the bottom sheet, i.e., Position 1, refer to Fig.17(a).When the feed rate is 4mm/s,the bottom thickness becomes thicker so that the bottom strength becomes stronger.As shown in Fig.17(b), the joint failed by pulling the rivet out from the bottom sheet, i.e., RPF, which indicates that the bottom strength is higher than the strength of the mechanical joining, and the mechanical joining becomes the weakest area.Compared to the feed rate of 2mm/s, the rivet leg at the feed rate of 4mm/s inserts the lower plate shallower and has poorer ability to resist tensile deformation, therefore showing lower strength as well as energy absorption.When the feed rate is 6mm/s or 8mm/s, the bottom thickness and interlock amount are further increased, leading to the bottom strength and the mechanical joining further increase, so that the upper Al sheet becomes the weakest area of the joint.Finally, the joint failed in the hardness transition zone of the upper sheet, i.e., USF, and the joint strength is determined by the upper Al sheet.Compared with the feed rate of 2mm/s or 4mm/s, the hardness transition zone of the upper sheet under 6mm/s or 8mm/s is thinner and under greater stress concentration, making the area 2 in Fig.17(c) weaker.Therefore, when the lap-shear force is applied, the fracture occurs on the upper sheet and is accompanied by a lower strength.However, the USF mode exhibits larger displacement by tearing the upper sheet compared to the other two modes, refer to Fig.16(a),and results in much higher energy absorption,refer to Fig.16(b).

Fig.15.Joints hardness in (a1)–(a4) aluminum alloy, (b1)–(b4) magnesium alloy, and (c1)–(c4) rivet under the feed rates of 2mm/s, 4mm/s, 6mm/s, and 8mm/s, respectively.

Fig.17.Lap-shear failure modes of the flat die F-SPR joints: (a) the lower sheet fracture (LSF), (b) the rivet pull-out fracture (RPF), and (c) the upper sheet fracture (USF).

Another interesting phenomenon is that compared to the pip-die F-SPR process [26], the Al-Mg interface in the rivet cavity under the three fracture modes did not break during the lap-shear process.This indicates the solid-state bonding and mechanical mixture near the Al-Mg interface in the rivet cavity impart a certain strength to resist fracture and deformation.This phenomenon also reflects the superiority of the flat-die F-SPR process.

To sum up,the feed rate of 6mm/s and 8mm/s was recommended.Firstly, it is possible to avoid burrs and gap defects caused by excessive heat generation, and can also avoid upset problems caused by insufficient heat generation.Secondly,the joints obtained at these two feed rates exhibit relatively larger lap-shear strength and energy absorption.Finally, the processing time under these two feed rates is relatively short,both within 1s, which meets the requirements of the fast-beat car body production.

3.6.Comparison of different die types

Finally, three F-SPR joints were obtained by using a pip die, a flat bottom die, and a flat die respectively in the case of a 1.5mm misalignment between the rivet and the die.The three joints were produced under the same process parameters as Test # 2 in Table 3, and then compared in terms of cross-sectional morphology, bottom morphology, and lapshear performance.As shown in Fig.18, the joints all have a large interlock amount.However, when using the pip dieor the flat bottom die, due to the misalignment of the rivet and die, the material unevenly flowed and deformed, and was pierced by the rivet leg, resulting in cracks at the joint bottom, as shown in Fig.18(a) and (b).However, as shown in Fig.18(c),the above problems can be effectively solved when the flat die was used.

Fig.18.Cross-sectional morphology and bottom morphology of the F-SPR joints obtained under (a) pip die, (b) flat bottom die, and (c) flat die.

Fig.19.(a) Lap-shear load-displacement curves, and (b) comparison of the peak load and energy absorption under three types of the die.

Fig.19 summarized the lap-shear performance of the FSPR joints obtained by the three types of dies.Compared with the joints under the pip die and the flat bottom die,the joints obtained by using the flat die have the greater lapshear force and energy absorption, which further shows the advantages of the flat die.

Conclusions

Aiming at solving the misalignment issue in the pip-die FSPR process, a new flat-die F-SPR process was proposed to join AA6061-T6 Al alloy and AZ31B Mg alloy sheets.Based on the above investigations, the following conclusions can be obtained:

(1) Die distance has a great influence on joint quality.A too-small die distance results in severe rivet upsetting,while a too-large die distance results in insufficient feeding resistance to deform the rivet.The 1.0mm die distance was found suitable to join the studied stack-up.

(2) In the joint formation process, the riveting force was always increasing and rose at a higher rate after the lower material contacted the flat-die surface.The magnitude of riveting force increases with feed rate, leading to the larger interlock amount and bottom thickness.

(3) Solid-state bonding was formed in the flat-die F-SPR joint.Al-Mg IMC was found at the Al/Mg interface on the outside of the rivet as well as the interface between the rivet leg and the base material.However, only Al-Fe atom interdiffusion was found at the interface between the rivet and the Al sheet.Besides, there was a large mechanical mixture zone at the Al/Mg interface inside the rivet cavity.

(4) After the flat-die F-SPR process, the upper Al layer was softened, and the lower Mg layer was hardened,while the rivet hardness was not affected obviously.An increase in the feed rate results in a decrease in heat input, which decreases the softening of the upper aluminum sheet and increases the hardening of the lower magnesium sheet but had no obvious effect on rivet hardness.

(5) With the increase of feed rate, the weakest location of the flat-die F-SPR joint under lap-shear loading changes from the thinnest portion of the bottom sheet at 2.0mm/s feed rate to the interlocking at 4.0mm/s and further to the hardness transition zone of the upper sheet near rivet head at 6.0 and 8.0mm/s feed rate, producing three different failure modes.The lower sheet fracture mode corresponds to the highest peak load, whereas the upper sheet fracture mode exhibits the highest energy absorption.

Author contribution section

Bingxin Yang: Design and performed experiments, analyzed data, and co-wrote the paper

Yunwu Ma: Designed the experiments and co-wrote the paper

He Shan: Performed experiments and processed the data

Sizhe Niu: Analyzed the data

Yongbing Li: Designed the experiments and co-wrote the paper

Declaration of Competing Interests

No conflict of interests in regards to the presented work can be declared.

Acknowledgments

The authors would like to acknowledge the financial support of the National Natural Science Foundation of China(Grant Nos.52025058 and U1764251).