Friction-based riveting technique for AZ31 magnesium alloy

Tianhao Wang, Scott Whalen, Xiaolong Ma, Joshua Silverstein, Hrishikesh Das,Madhusudhan R.Pallaka, Angel Ortiz, Timothy Roosendaal, Piyush Upadhyay,Keerti S Kappagantula

Pacific Northwest National Laboratory, MSIN: K2-03, 902 Battelle Blvd., Richland, WA 99352, United States

Abstract A new friction-based riveting technique, Rotating Hammer Riveting (RHR), is demonstrated to fully form AZ31 Mg rivet heads in a mere 0.23 s.Heat and pressure generated through severe plastic deformation during the process was sufficient to form the Mg rivet head without the need for a pre-heating operation.Due to preliminary twinning and followed by dynamic recrystallization, AZ31 Mg grains in the rivet head were refined during RHR, which enhance the formability of Mg rivets by triggering grain boundary sliding and reducing plastic anisotropy of Mg.In addition, RHR joints showed a metallurgical bond between the rivet head and top AZ31 Mg sheet, which eliminates a significant pathway for corrosion.

Keywords: Rivet; Magnesium; Dissimilar joining.

1.Introduction

Riveting is one of the most common mechanical fastening methods for joining two or more sheets of material.Various riveting techniques have been developed and widely applied such as self-drilling blind riveting [1], cold-driven riveting[2], hot-driven riveting [3]and self-piercing riveting [4,5,6].Common metallic rivets include steel [7], copper [8]and aluminum (Al) alloy [9].Magnesium (Mg) rivets are rarely used, in part, because they cannot be driven at room temperature due to low formability and must be pre-heated to elevated temperature, which is not convenient in practical applications.And Mg is highly reactive in aqueous compared to common engineering materials [10].However, Mg alloys are one of the lightest structural metals with a density of 1.8 g/cm3and tensile strength up to ?290 MPa [11].Using Mg rivets can significantly improve fuel efficiency in the transportation industry since its density is approximately 67% of that of Al but with comparable strength [12].This has been a widespread weight reduction strategy in automotive applications; for example, Mg products in the cast condition are applied in certain automotive components such as steering columns, instrument panels, and other interior components [13].Yet, the poor formability of Mg alloys at room temperature hinders their widespread application for other uses such as riveting [14].The poor formability arises from the intrinsic plastic anisotropy induced by the hexagonal close-packed (hcp) crystal structure [15].

Previous investigations showed that the addition of rare earth elements of Ce, Nd and Y in Mg alloys can increase formability [16].Other than that, addition of Ca and Mn [17],Y and Zr [18,19]in Mg-Zn system have been proved to be effective in enhancing both strength and ductility.However,a more feasible approach is taking advantage of thermomechanical processing, which can randomize and weaken the strong basal texture in both conventional Mg alloys [20]and rare earth element contained Mg alloys [21].For example,altering the microstructure such as grain size and texture viasevere plastic deformation can improve formability of Mg alloys [22].High strain rate sensitivity [15], non-basal slip activity [23]and dynamic recrystallization (DRX) activity[24]are also known to enhance formability.Meanwhile, Mg alloys are readily formable at elevated temperatures [25].Inspired by efforts to improve formability of Mg alloys via thermomechanical processing, a novel friction-based riveting technique, rotational hammer riveting (RHR), was applied for riveting with Mg alloy AZ31 in this study.During the RHR process, AZ31 Mg rivet heads are formed via severe plastic deformation of a solid shank enabled by a temperature rise via heat generated from friction and plastic deformation.Two dissimilar material combinations of AZ31/AA7055-T6 and AZ31/thermoplastic carbon fiber reinforced polymer(CFRP) were selected to verify the RHR technique for joining dissimilar materials in this study.

2.Materials and methods

In the present work, AZ31, CFRP, AA7055-T6 sheets with thicknesses of 2.4, 3.1 and 2.6 mm, respectively, were used.The CFRP sheets obtained from BASF Corporation, commercially referred to as Ultramid Advanced N XA-3454, were comprised of PA9T thermoplastic reinforced polymer with a 40 vol.% short carbon fiber.Relevant properties of the sheet and rivet materials are listed in Table 1.In this study, AZ31 sheets were placed on top of CFRP or AA7055-T6 sheets to make AZ31/CFRP and AZ31/AA7055-T6 joints respectively.Holes with diameter of 5 mm and center to center spacing of 15 mm were predrilled in the sheets and aligned in a lap arrangement as shown in Fig.1(b).Solid shank AZ31 Mg rivets with a matching diameter of 5 mm were inserted through the aligned holes in the joint assembly and extended 5 mm above the surface of the top sheet to allow sufficient material availability to form a rivet.Detailed dimensions of base sheet and rivet materials are presented in the Supplementary section 1(See Figure S1).The starting microstructure of the AZ31 rivet is presented in Fig.1(a) which was machined from extruded barstock.Overall, grains are equiaxed, and have a grain size of 18 ± 10 μm and inherit a strong texture from the original extrudate.The RHR tool employed in this study was made of H13 steel with a diameter of 12 mm and a concave angle of 10° (Fig.1(c)).During the process, temperature was measured using a type-K thermocouple spot welded at the center of the tool face (See Figure S2 in Supplementary section 2).Axial force was measured via load cells within the spindle assembly of the Transformational Technologies, Inc.(TTI) LS2-2.5 friction stir welding machine (See Figure S2 in Supplementary section 2).Fig.1(d) shows the schematic of the RHR process.A range of process parameters were explored in order to determine the optimum conditions.Tool rotation rate of 1000 - 2000 rpm and tool plunge speed in the range of 50 - 800 mm/min were examined in this study for making AZ31/CFRP and AZ31/AA7055-T6 joints.Rivet heads obtained with a shiny appearance and without surface defects were deemed optimal and were further explored for microstructural and mechanical performance.

Table 1Mechanical properties of base sheet materials.

RHR joints were cut along the centerline of AZ31 rivets to obtain cross-sectional samples for microstructural characterization.Specimens for microstructural analysis were mounted in epoxy and polished to a final surface finish of 0.05 μm using colloidal silica.Optical microscopy was performed using an Olympus SZX16 microscope.In addition, scanning electron microscopy along with electron backscatter diffraction (EBSD) mapping was carried out to characterize the grain size and texture evolution in the AZ31 Mg rivets using a JEOL 7001F field emission scanning electron microscope equipped with a Bruker Quantax e-Flash EBSD detector.Hardness values of the AZ31 rivets and the AZ31 and AA7055-T6 sheets were measured using a Vickers microhardness tester at 200 g load with a dwell time of 12 s and an indentation spacing of 0.3 mm.Lap shear tensile tests were performed at room temperature using an MTS 22 kip test frame at an extension rate of 1.27 mm/min using 15 mm wide AZ31/CFRP and AZ31/AA7055 assembly samples prepared with the AZ31 Mg rivet located at the center of the sample.

3.Results and discussion

3.1.Parameter selection

During the RHR process, friction between the rotating RHR tool and Mg rivet shank produces heat, which softens the Mg thereby reducing flow stress.Under the combined effect of heat and pressure from the RHR tool, the Mg rivet head is formed.The edges of the rivet head are mixed with the top sheet in the joint assembly which forms a metallurgical bond at the interface between the Mg rivet head and AZ31 sheet.The association between riveting parameters (rotation rate & plunge speed) and rivet appearance (surface & crosssection) for riveted AZ31/CFRP is displayed in Fig.2.A rivetable parameter window was obtained for the AZ31/CFRP joints.There are two conditions (cold run & hot run) outside the rivetable parameter window, as shown in Fig.2.At a relatively high tool rotation rate of 1950 rpm and tool plunge speed of 240 mm/min, it was seen that a portion of the AZ31 rivet head remained adhered to the concave surface of the tool due to excessive heat input.On the other hand, rivet heads tend to crack with low rotation rate (500 rpm) and high plunge speed (780 mm/min) plausibly owing to the heat generated during the thermomechanical process being insufficient to fully form the Mg rivet head.Owing to the short cycle times facilitated by the use of high plunge speed as observed from the AZ31/CFRP trials, AZ31/AA7055-T6 joints were made varying tool rotation rate while keeping the plungespeed at 780 mm/min (FSW machine limit).As shown in Fig.3, the AZ31 Mg is rivetable in the range of 500 to 1000 rpm.However, the rivet head adheres to the tool with rotation rate of 1500 rpm, while cracks with rotation rate of 250 rpm.The variance in rivetable parameter window comes from the difference of bottom sheet, which will be discucssed in the Section 3.2 and 3.3.Furthermore, rivet height varies for AZ31/CFRP and AZ31/AA7055 joints due to their different overall thickness.(see Figure S1).A constant plunge depth of 5 mm is applied for both AZ31/CFRP and AZ31/AA7055 joints since the protruding height of rivet above top sheet is 5 mm for both dissimilar combinations.

Fig.1.(a) Starting microstructure in AZ31 rivet: EBSD inverse pole figure map (upper) and pole figure (lower).(b) Schematic of RHR setup, (c) RHR tool and (d) schematic of RHR process.

Fig.2.Parameter window for AZ31 riveted AZ31/CFRP sheets.Note that the machine limit including rotation rate limit of 1950 rpm and plunge speed limit of 780 mm/min defines the overall parameter range.

Fig.3.Parameter window for AZ31 riveted AZ31/AA7055 sheets.

Fig.4.Cross sections of riveted AZ31/CFRP joints with (a) rotation rate of 1000 rpm and plunge speed of 60 mm/min and (b) rotation rate of 1950 rpm and plunge speed of 780 mm/min.Note that deflection of bottom CFRP sheets are labeled by white arrows.

Load and temperature experienced by the RHR tool during AZ31/CFRP riveting were measured using load sensor and thermocouple embedded in the concave vertex of the tool, as displayed in Figure S2(a) and (b).With a constant plunge speed of 780 mm/min and increasing rotation rate from 400 to 1950 rpm, the peak temperature experienced by the RHR tool increased from about 200 to 320 °C while peak load on RHR tool reduced from about 30 to 8 kN, as shown in Figure S2(c).

3.2.Microstructural analysis

Among defect-free riveted AZ31/CFRP joints,two samples at extremes of the rivetable zone were selected for optical microscopy and are shown in Figs.4(a)and(b)corresponding to tool rotation rate/plunge speed of 1000 rpm/60 mm/min and 1950 rpm/780 mm/min respectively.For both runs, the rivet head and AZ31 sheet were mixed, forming a fully bonded interface.The mixing region was larger for the joint manufactured at a tool rotation rate of 1000 rpm and plunge speed of 60 mm/min as can be seen from Fig.4(a), compared to the joint manufactured at a rotation rate of 1950 rpm and plunge speed of 780 mm/min (Fig.4(b)).It implies that effect of plunge speed increasing from 60 to 780 mm/min overweighs the effect of rotation rate increment from 1000 to 1950 rpm on heat generation and resultant plasticized deformation zone.In addition, deflection of the CFRP reduced from ?0.65 mm to ?0.37 mm with the increase in plunge speed from 60 mm/min to 780 mm/min as evident from Figs.4(a)and(b).As the RHR plunge speed was increased from 60 mm/min to 780 mm/min, the process time decreased from 3 s to 0.23 s.The shorter duration associated with 780 mm/min implied that even though higher temperatures were recorded for this sample at the rivet head, there may not have been sufficient time for the higher heat generated at 1950 rpm to penetrate the joint, thereby resulting in a smaller mixed region at the AZ31 sheet and rivet-head interface, as well as smaller CFRP deflection.

While process time is an important consideration, it must be noted that tool rotation rate has a large effect on extent of mixing at the interface of the rivet and the top sheet in the joint assembly.To demonstrate this, optical microscopy images of two samples manufactured at the extremes of the parameter window for riveted AZ31/AA7055 are displayed in Fig.5(a) and (b).Both the joints were processed at a tool plunge speed of 780 mm/min; however, the sample in Fig.5(a) was processed at 500 rpm while the sample in Fig.5(b) was made at 1000 rpm.The rivet head and AZ31sheet were mixed well for both samples.The mixing region increased in size with the increase in rotation rate.This confirms that higher rotation rate leads to higher heat generation and resulted in a correspondingly larger plasticized deformation zone under the rotating RHR tool.

Fig.5.Cross sections of riveted AZ31/AA7055 joints with (a) rotation rate of 500 rpm and plunge speed of 780 mm/min and (b) rotation rate of 1000 rpm and plunge speed of 780 mm/min.

Fig.6 shows inverse pole figure (IPF) mapping and the pole figure results from a representative AZ31/CFRP joint,manufactured with 1000 rpm and 60 mm/min.The RHR process generated a refined, with an average grain size of 7.2 ± 4.6 μm, and strongly textured microstructure within the AZ31 Mg rivet head (Location 1 in Fig.6).Conversely,the rivet bottom remained coarse-grained (average grain size of 15 ± 5 μm) consistent with the starting microstructure(Fig.1(a)) but with extensive twins produced by the compressive strain during RHR (Location 2 in Fig.6).The grain refinement could be attributed to two primary factors: (i) severe plastic strain, which facilitates the grain subdivision process by dislocation slip and twinning [26,27]and (ii) temperature rise, which assists in the dynamic recrystallization to form fine grains during the short processing period.The altered and intensified texture at the rivet head (see multiples of uniform density (MUD) values in Fig.6- a higher MUD value suggests a stronger texture) also results from the severe strain that was imposed into the rivet head.Notably, the AZ31 rivet used to join AZ31/AA7055 assembly also has a nearly identical microstructure to that in Fig.6 (see Figure S3 in Supplementary section 3).

The formation of fine grains along with the temperature rise has important implications for the success of RHR for magnesium alloys in general.Mg has few slip systems due to the low symmetry of hcp crystal structure, and the critical resolved shear stress (CRSS) for those slip systems (basal and non-basal slips) vary widely [28].Grain refinement alleviates this plastic anisotropy by narrowing the gap of CRSS for different slip systems as well as activating grain boundary sliding as additional deformation modes [29,30].Therefore,fine grains become more ductile and formable even at room temperature [31,32].Similarly, the temperature rise also reduces the difference of CRSS between basal and non-basal slip[15,33].It also promotes the diffusion-based grain boundary sliding process.Collectively,the fine grains along with the temperature rise enables a large extent of plastic deformation of the Mg alloy without failure, which otherwise is unattainable in conventional linear impact riveting techniques.

Fig.7.Schematic displaying evolution of grain structure and texture during RHR.

Texture evolution during RHR is also discussed in the current study.As displayed in Fig.7, the base rivet is made from extrusion, with basal poles perpendicular to the extrusion direction (ED) which is aligned parallel to the long axis of the shank.Presumably, deformation twinning occurs at the intial stage of RHR process when a compressive strain is imposed while the rotation-shear-induced temperature rise is relatively limited.A significant amount of grains are reoriented with ?86° as a result of twinning.As RHR tool plunges deeper,temperature spikes above 200 °C (Figure S2 in Supplementary section 2), faciliating the DRX in the microstructure,especially those nucleated at the newly-formed twin boundaries [34].Driven by grain boundary energy reduction [35],the DRX transforms the low-angle grain boundaries between c-axes to the high-angle grain boundaries in the basal plane,further strengthening the basal texture perpendicular to the flow direction (FD).

3.3.Mechanical performance characterization

Figs.8(a), (b) and (c) show hardness maps of the crosssections of riveted AZ31/CFRP and AZ31/AA7055 joints.For AZ31/CFRP joint, hardness indents were only applied on the AZ31 sheet and rivet (i.e.hardness of CFRP was not measured) while for the AZ31/AA7055, they were applied for all the metallic surfaces.The AZ31 rivet in the AZ31/CFRP joint showed a slight hardness increase in the rivet head and deformed region of the shank, which can be attributed to the grain refinement observed in Fig.6.For AZ31/AA7055 joints,hardness increase in the rivet head and shank was more pronounced (Figs.8(b) and (c)).Two hardness maps with different ranges are provided for Figs.8(b) and (c).The hardness map ranging from 45 to 80 HV displays the hardness variance in AZ31, while the other hardness map ranging from 45 to 190 HV displays the hardness variance in both AZ31 and AA7055.There are two major reasons behind the hardness variation.Firstly, rotation rate and resultant heat generation of AZ31/AA7055 at 500 and 1000 rpm is lower than that of AZ31/CFRP at 1950 rpm.Secondly, given that the thermal conductivity of AA7055 is much higher than that of CFRP,heat dissipates heat much faster comparatively.Both these phenomena contribute to reducing the effects of over aging of the AZ31 rivet within the AZ31/AA7055 joint leading to smaller grains in its microstructure resulting in correspondingly higher hardness.

Meanwhile, a reduction in hardness was observed in the AA7055 sheet around the rivet shank with rotation rate of 1000 rpm (Fig.8(b)), possibly owing to its over aging during the RHR process.The RHR process provides a combination of mechanical and thermal cycles on the riveted assembly which govern the evolution of hardness.Plastic deformation (mechanical cycle) on the rivet increasesthe local hardness while the friction-generated heat (thermal cycle) dissolves precipitates and coarsens grains in the matrix of the undeformed region of the shank, which reduces the local hardness.The hardness reduction on AA7055 was mitigated with a rotation rate of 500 rpm (Fig.8(c))compared to the joint manufactured at the same plunge speed but higher rotation rate (1000 rpm), further bolstering the importance of process parameters on joint performance.Furthermore, to unveil the correlation between grain structure evolution and resultant hardness change in AZ31 rivet, four typical regions with different grain structure and hardness are plotted in the supplementary section 4 (see Figure S4).

Fig.8.Optical image of hardness indent locations and microhardness mapping on cross-section of riveted joints: (a) AZ31/CFRP (rotation rate of 1950 rpm and plunge speed of 780 mm/min), (b) AZ31/AA7055 (rotation rate of 1000 rpm and plunge speed of 780 mm/min), and (c) AZ31/AA7055 joint (rotation rate of 500 rpm and plunge speed of 780 mm/min).Note that the black dots displayed in the hardness maps represent the indent locations.

Lap shear tensile tests were conducted on the riveted AZ31/CFRP and AZ31/AA7055 joints.Results show that load at failure for the AZ31/CFRP joints ranged from ?1.2-2.2 kN,while the extension ranged from ?0.5-1.1 mm as shown in Fig.9(a).All joints fractured in a brittle manner through the CFRP as shown in Fig.9(b) with the fractures possibly initiated from defects in CFRP resulting from drilling the holes prior to the joining process.Combined with the observation in Figs.4(a) and (b), it is observed that the variation in load and extension is strongly associated with the extent of deflection generated in the CFRP sheet as a result of the RHR process.Results show that the joints where CFRP demonstrated lower deflection(i.e.manufactured at higher rotation rate and plunge speed) led to higher load bearing capacity and extension at failure.While for the AZ31/AA7055 joints, the load was seen to be constant even with varying plunge speed and ranged from ?2.5-2.6 kN while extension ranged narrowly from ?1-1.3 mm as shown in Fig.10(a).All the AZ31/AA7055 lap shear tensile test samples failed by shear through the shank of the AZ31 rivet as shown in Figs.10(b).This implies that the effect of various riveting parameters on joint strength of riveted AZ31/AA7055 is dominated by cross sectional area of the AZ31 rivet shank.This is because the riveting process does not substantially influence the shear strength of the AZ31 shank, which determines the joint strength of riveted AZ31/AA7055.In our previous investigation, finite element analysis was used to display the stress concentration and resultant fracture on the shank during lap shear tensile testing[36].Calculations regarding load capacity of AZ31/CFRP and AZ31/AA7055 joints are presented in supplementary section 5.

Fig.9.(a) Load and extension at failure of AZ31 riveted AZ31/CFRP joints with different parameters, (b) fractured sample after lap shear tensile testing.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig.10.Load and extension of AZ31 riveted AZ31/AA7055 joints with different parameters, (b) fractured sample after lap shear tensile testing.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Conclusion

A friction-based riveting technique has been applied for joining of dissimilar materials using solid shank Mg rivets without a pre-heating operation.Underside of the rivet head is mixed with the underlying metal sheet, thereby forming a continuous metallurgical bond.The combination of elevated temperature and severe plastic strain enable riveting of AZ31 over a wide range of process conditions up to a rotation rate of 1950 rpm and plunge speed of 780 mm/min,which requires just 0.23 s from start to finish which is faster than traditional impact riveting.Joint load capacity reached 2.2 kN and 2.6 kN for riveted AZ31/CFRP and AZ31/AA7055,respectively.Current work presents a new approach to riveting Mg to metal and metal/polymer assemblies, enabling lightweighting of the dissimilar joints.

Acknowledgments

The authors acknowledge the support of the U.S.Department of Energy Vehicle Technologies Office (DOE/VTO)Joining Core Program.The authors are grateful to Daniel Graff in assisting with sample synthesis, Anthony Guzman for preparation of specimens for analysis.The Pacific Northwest National Laboratory is operated by the Battelle Memorial Institute for the United States Department of Energy under contract DE-AC06-76LO1830.

Supplementary materials

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.jma.2021.06.004.