Yuan Li,Yong Chae Lim,Jian Chen,Jiheon Jun,Zhili Feng

Materials Science and Technology Division,Oak Ridge National Laboratory,Oak Ridge,TN 37831,USA

Abstract Carbon fibe reinforced polymer(CFRP)and AZ31B Mg alloy were joined by the friction self-piercing riveting(F-SPR)with different steel rivet shank sizes.With the increase of rivet shank size,lap shear fracture load and mechanical interlock distance increased.Ultrafin grains were formed at the joint in AZ31B as a result of dynamic recrystallization,which contributed to the higher hardness.Fatigue life of the CFRP-AZ31B joint was studied at various peak loads of 0.5,1,2,and 3 kN and compared with the resistance spot welded AZ31B-AZ31B from the open literature.The fatigue performance was better at higher peak load(＞2 kN)and comparable to that of resistance spot welding of AZ31B to AZ31B at lower peak loads(＜1 kN).From fractography,the crack initiation for lower peak load(＜1 kN)case was observed at the fretting positions on the top and bottom surfaces of AZ31B sheet.When peak load was increased,fretting between the rivet and the top of AZ31B became more dominant to initiate a crack during fatigue testing.? 2022 Chongqing University.Publishing services provided by Elsevier B.V.on behalf of KeAi Communications Co.Ltd.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/)Peer review under responsibility of Chongqing University

Keywords:Friction self-piercing riveting;Magnesium alloy;Carbon fibe reinforced polymer;Dynamic recrystallization;Fatigue life;Crack initiation.

Notice:This manuscript has been authored by UT-Battelle,LLC,under contract DE-AC05-00OR22725 with the US Department of Energy(DOE).The US government retains and the publisher,by accepting the article for publication,acknowledges that the US government retains a nonexclusive,paid-up,irrevocable,worldwide license to publish or reproduce the published form of this manuscript,or allow others to do so,for US government purposes.DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan(http://energy.gov/downloads/doe-public-access-plan).

1.Introduction

The concept of lightweight multi-materials vehicles has been an important topic for automotive industries to improve the fuel or energy economy and to reduce green gas emission.Replacement of conventional structural steel with higher specifi strength materials including aluminum(Al)alloys,magnesium(Mg)alloys,carbon fibe reinforced polymer(CFRP)and advanced/ultra-high strength steels can effectively reduce the vehicle weight.To enable multi-materials autobody structure,reliable and economical dissimilar materials joining techniques are demanded.Conventional fusion welding techniques,such as resistance spot welding(RSW)and laser welding,can form brittle intermetallic compounds[1-3].In addition,some welding defects(e.g.,cracks and pores)are hard to avoid during fusion welding of Mg alloy due to higher affinit to oxygen and higher vapor pressure of Mg,which is detrimental to the mechanical performance of the joint[4].

Self-piercing riveting(SPR)has been widely used for automotive and transportation sectors as a major mechanical fastening technique.A semi-tubular rivet is punched into the top sheet and flare into the bottom sheet supported on a die[5].It has been applied for various material combinations,such as Al to steel[6-8],Al to Al[9-11]and Al to CFRP[12-14].However,when the bottom sheet is low ductility material,such as Mg alloys,cracks often occur due to poor ductility at room temperature[15].Therefore,to improve ductility of Mg alloy for a sound joint,the thermal-assisted joining techniques must be employed to activate more slip systems of Mg alloy at elevated temperature.For example,Wang et al.[15]compared the joint quality of SPR of AZ31B to AZ31B at ambient temperature,150,180 and 200 °C.The cracks of AZ31B formed at room temperature was eliminated at elevated temperature and the joint strength increased by 17% at 200 °C.Durandet et al.[16]used laser to preheat the AZ31B Mg sheets during SPR of AZ31B to AZ31B and they also concluded that the cracking can be prevented when the preheating temperature is higher than 200 °C.

However,an auxiliary heating equipment and specifi joint designs are required for this thermal-assisted SPR.As a single joint process for low ductility materials,friction self-piercing riveting(F-SPR)was proposed by Li et al.[17].In this process,the rivet is driven by a rotating tool and punched into the stacked sheets.The friction heat can soften the sheets and make them more ductile,thereby eliminating the cracks of the low-ductility bottom plate.Pros of F-SPR process are no external heating equipment required,no-pilot hole required,and possible for multiple stacks.Cons of this process can be adding rivet weight,and galvanic corrosion between the rivet and surrounded dissimilar materials.Li et al.and Xun et al.have successfully joined Al6061-T6 and Al7075-T6 to AZ31B Mg alloy without cracking in AZ31B sheets[17-20].Recently,Huang et al.[21]applied friction fillin stacking joining(FFSJ)process with a composite solid rivet to join Al 6082-T6 and polypropylene.Advantages of their joining technique are relatively simple surface preparation,lightweight rivet material,and no galvanic corrosion issue between the metal and polymer joint interface.However,pre-machined pilot hole on metal is required prior to joining step which can increase additional cost and time.Next,alignment between a composite rivet and a pre-machine pilot hole is critical to achieve a high-quality joint.In addition,this FFSJ process can be limited to a thermoplastic substrate to a thermoplastic rivet to form strong chemical bonding.Therefore,thermoset polymer substrate cannot be easily joined by FFSJ process with a thermoplastic rivet.The authors’recent works[22,23]demonstrated joining of thermoplastic CFRP to AZ31B and thermoset(TS)CFRP to AZ31B Mg alloy by F-SPR.In this paper,we further explored the relationship between rivet shank size and lap shear peak load for F-SPR joint of TS CFRP and AZ31B.Also,microstructure evolutions in the joint were investigated by an optical microscope and Vickers microhardness measurement.Next,the fatigue life of the TS CFRP-AZ31B joint was studied at various peak loads.Fatigue crack initiation and propagation was studied by digital image correlation(DIC)during fatigue and scanning electron microscopy(SEM)was used to study fractography for fatigue tested samples.

2.Materials and experiments

2.1.Materials

Thermoset CFRP(Clearwater Composites,MN)with 1.9 mm thickness is stacked onto the 2.3 mm-thickness AZ31B Mg alloy sheet.The CFRP was made with the G-83 prepreg(T700,Toray)in a 0 deg/90 deg,unidirectional,nineply layup with 60% vol.of continuous carbon fibe.Fig.1a shows the nine-ply layup of CFRP.Fig.1b shows the 0°(A1)and 90°(A2)direction of carbon fibe,respectively.The measured diameter of the carbon fibe is ranged from 7 to 8μm.The top and bottom sheets were all 25.4 mm in width and 101.6 mm in length with overlap area of 25.4×25.4 mm2.The steel rivet was made of Japanese Industrial Standard G3057-2 carbon steel(SWCH18A).The measured Vickers microhardness of the as-received rivet shank is very uniform except the rivet tip(around 200μm).Also,the measured average microhardness of rivet head and rivet shank is 266 HV and 222 HV,which is slightly different potentially due to manufacturing method by the external vendor.The rivet head was designed into hexagonal profil with width of 9.52 mm to be externally driven by the rivet holder during joining.Rivet shank diameter was 5.3 mm while different rivet shank lengths,5,5.5 and 6 mm,were used to study joint formation and mechanical joint strength.For a bottom supporting anvil,a pip die with 1.7 mm cavity depth was applied.The details of the dimension of rivet and pip die can be found in the reference[23,24].

2.2.F-SPR process

Fig.2 illustrates the schematic of F-SPR process.First,the spinning rivet is plunged into the top sheet with controlled plunge and rotation speed.Due to interaction of rotating rivet and surrounded material,frictional heat is generated during joining process,leading to local softening AZ31B bottom sheet.As the plunge depth increases,the bottom sheet is forced to fil up the cavity in the die.Meanwhile,the resistance from the supporting die forces the rivet shank to flar out in the bottom sheet,forming mechanical interlocking.The rivet shank length of 5,5.5 and 6 mm were used to study joint formation and its mechanical joints strength.Because of our machine response with respect to Z-axis plunge depth and the tightness of the joint,the actual plunge depth was deeper than the rivet shank length.The spindle rotation speed was 1500 rpm with downward plunge speed of 2.86 mm/s for all the joints based on the authors’recent works[23].

2.3.Mechanical tests

To evaluate mechanical joint strength of F-SPR specimens,the lap shear tensile tests were carried out using MTS tensile system with constant speed of 10 mm/min at room temperature.Spacers were used to make the grip regions in the same thickness to prevent bending during the testing.

Fig.1.(a)As-received CFRP,(b)A1:high magnificatio of 0° direction carbon fibe,A2:high magnificatio of 90° direction carbon fibe.

Fig.2.Schematic of F-SPR process.The top sheet is CFRP while the bottom one is AZ31B Mg alloy.

The lap shear tensile specimens were also used for fatigue testing.Fatigue testing was performed at frequency of 20 Hz with R ratio of 0.1 at various peak loads of 0.5,1,2 and 3 kN.DIC system was used to visually study crack initiation and propagation during fatigue testing.Random speckle pattern was painted on the backside of AZ31B sheet in the joint for DIC measurement[25].Due to longer fatigue testing times,image acquisition rate was 2 images per minute.Then,VIC-2D software was used to calculate the strain in the joint as a postprocessing.

2.4.Characterizations

For metallographic characterizations of F-SPR joints,some lap joint samples were cut through the middle of the rivet using diamond saw and then mounted in epoxy.The crosssectioned samples were ground by silicon carbide(SiC)papers with 600,800,1200 grits and then polished with diamond suspension of 3,1,and 0.5μm using Struers polishing machine.An acetic-picric solution(5 mL acetic acid,1 g picric acid,10 mL water,100 mL ethanol)was used to etch AZ31B Mg alloy.Zeiss Axio microscope was used for optical characterization.

To correlate the microstructural changes and local mechanical properties at the joint,Vickers microhardness was measured with a load of 0.2 kgf and a dwell time of 15 s using LECO AMH43 Automatic Micro/Macro-indentation Hardness Testing System.The interspacing between each indent was at least 100μm away horizontally and vertically.

The fracture surfaces of the fatigue tested samples were characterized using SEM(TESCAN MIRA3)with accelerating voltage of 20 kV.

3.Results

3.1.Lap shear failure load of different rivet shank lengths

Fig.3a shows the F-SPR machine for this study and Fig.3b is the Z-axis plunge force and plunge depth as a function of time during F-SPR.First,the downward Z-axis plunge force immediately increased when the rivet was engaged the top surface on the CFRP sheet.The firs peak force appeared around 0.9-1 kN at weld time of 0.9 s.Next,when the rivet plunged into the top surface of the bottom sheet,Z-axial force increased again exhibiting the second peak(～2.5 kN).This rapid increase of plunge force is because the bottom AZ31B sheet was in full contact with the back supporting pip die.Then,the Z-axis peak plunge force dropped due to frictional heat generation during joining process before fina“flaring stage.This is due to softening of the AZ31B sheet by frictional heat.Finally,the plunge force reaches the maximum(～6 kN)during the rivet flaring Similar axial force and displacement trend was found by Ma et al.for joining of AA7075-T6 by F-SPR[26].

Fig.4a presents the cross-sectional view of F-SPR joint with different rivet shank length of 5,5.5 and 6 mm.It shows that the interlock distance increased as the rivet shank length increased.Fig.4b presents the lap shear failure load and interlock distance as a function of rivet shank size.By increasing the rivet shank length,both the lap shear failure load and interlock distance increased.For instance,the lap shear failure loads for rivet shank lengths of 5,5.5 and 6 mm are 3.54 kN,4.17 kN,and 5.07 kN,respectively and interlock distance are 0.19 mm,0.31 mm,and 0.46 mm,respectively.Table 1 summarizes the actual axial plunge depth & force,mechanical interlocking distance,averaged lap shear failure load and failure mode for each rivet.Actual axial rivet plunge depth was deeper than the rivet shank length because of machine response and tightness of joint.For 5 mm rivet leg length,axial plunge was much deeper(0.84 mm)than the rivet shank size,so there was some damage of top surface on CFRP.For 5.5 mm rivet leg length,axial plunge depth was still deeper(0.62 mm)than the rivet shank length.For this reason,axial reaction force is the resistance from CFRP to the rivet head due to the much deeper plunge depth than the rivet leg length.Although Z-axial force is relatively high ranged from 4.6 to 5.5 kN,rivet leg length(i.e.,5 and 5.5 mm)is not long enough to flar out for mechanical interlocking compared with 6 mm rivet length.Therefore,the joints by shorter rivets have smaller mechanical interlock.For this reason,5 and 5.5 mm rivet length specimen showed the rivet pullout,while 6 mm rivet length case showed bottom AZ31B pullout.For CFRP to AA2024-T6 joint by SPR with the same coupon dimensions,the tensile shear peak load ranges from 3 to 3.7 kN based on the different oil pressures[13].Compared to the reported result,the CFPR to Mg alloy joints by F-SPR in this work achieved higher lap shear strength.According to Duan et.al,F-SPR joints of AZ31B to AZ31B exhibited better lap shear strength and fatigue life than those by SPR[27].The shear strength of F-SPR of AZ31B to AZ31B ranges from 4.48 kN to 5.23 kN depending on the punch pressure,which is comparable to this study.

Fig.3.(a)The automated tool used for F-SPR process,(b)Z-axis axial plunge(downward)force and plunge depth as a function of time during F-SPR with 6 mm rivet shank length.

Fig.4.(a)Cross-sectional view of F-SPR joint with different rivet shank length:L=5.0,5.5 and 6.0 mm,(b)summary of lap shear strength and interlock distance as a function of rivet shank length.

Table 1Summary of actual plunge depth,peak Z-axial plunge force,interlocking distance,averaged peak lap shear failure load,and failure mode for different rivet shank lengths.

3.2.Microstructure evolutions of AZ31B after F-SPR

When the rivet shank length was 6 mm,the highest lap shear tensile strength was achieved.For this reason,we used F-SPR joint with 6 mm rivet shank length for further characterizations.Fig.5a shows the macrograph of F-SPR joint in cross-sectional view,covering microstructures of AZ31B Mg alloy at different locations in the F-SPR joint(i.e.,near AZ31-CFRP-rivet joint interface and base material of AZ31B).As shown in Fig.5b,the grain size ranged from submicron to 10μm along the distance between a location in AZ31B matrix and the rivet-AZ31B joint interface as indicated in Region P1.Fig.5c shows the grains of the base material(Region P2)with average grain size of 10μm.A lot of twins(in needle shape)formed during rolling were observed.In Fig.5d,the grain size change in Region P11 can be clearly seen.In Fig.5e,the region P12 closer to the joint interface showed refine grain size ranging from 0.3 to 1μm in a SEM image.

F-SPR is feasible for the joining of low-ductility materials at room temperature due to the friction heat generated during joining.Rising temperature can improve the local ductility of the material,and it can produce a sound joint.However,it is difficul to measure the actual temperature in the material during F-SPR due to complex joining motions(e.g.,simultaneous plunging and rotation).Therefore,an indirect approach was used to estimate the peak temperature during F-SPR.

According to the relationship between dynamic recrystallized(DRXed)grain size of AZ31B and Zener-Hollomon parameter(Z,unit:s-1)[28]:

in which d is DRXed grain size,and Z is related with the strain rate(˙ε)and temperature(T):

in which Q is the self-diffusion activation energy of Mg,135 kJ/mol[29].

The strain rate(˙ε)of materials during friction stir processing can be calculated by following equation[28]:

in whichωis rotation speed(unit:rotation per second(rps)),randLare the effective radius and depth of the DRX zone,respectively.During F-SPR,the DRX zone depth is increasing linearly with time.For this reason,the effective depth of DRX zone L can be estimated as a half of the depth the shank plunges into AZ31B,namely L～1.85 mm.The rotational speed is 25 rps,andr～2.65 mm is the outer radius of the rivet shank.Therefore,the strain rate(˙ε)is calculated as 112.6 s-1.With measured DRX grain size ranged from 0.3 to 1μm in Fig.5e,the estimated peak temperature can be ranged from 248 to 294 °C.Ma et al.[30]simulated F-SPR process for joining of Al6061-AZ31B with low carbon steel rivet.The peak temperature of large deformation region at the bottom AZ31B was predicted as 298 °C at rotation speed of 1450 rpm.From the study of the thermal-assisted[15]and laser-assisted SPR[16],the bottom AZ31B plate should reach above 200 °C to eliminate the cracking issue.The estimated temperature ranges during F-SPR are comparable to the previously reported temperature that minimizes the cracking of Mg alloys[15,16,30].

Fig.5.(a)Macroscopic cross-sectional view of F-SPR joint,(b)microstructure of AZ31B Mg alloy at the F-SPR joint interface(Region P1),(c)microstructure of base material(Region P2),the microstructure in(d)magnifie SEM image at Region P11 and(e)magnifie SEM image at Region P12.

The estimated peak temperature during joining process is higher than the glass temperature(Tg)of the thermoset CFRP(130-140°C),but it is lower than the decomposition temperature(above 400°C)based on Differential Scanning Calorimetry and Thermogravimetric Analysis by the authors.Thermal degradation of CFRP near the joint interface is possible,but this degradation will be very limited to the CFRP joint region because of very low thermal conductivity of CFRP compared with metals.In addition,no fume and decomposition gases from CFRP were observed during joining process.Hence,the inversely calculated peak temperature during F-SPR is reasonable.Lim et al.[31]investigated thermal degradation of CFRP joined by friction bit joining between the CFRP and dual phase 980 steel using fourier-transform infrared spectroscopy and X-ray diffraction technique,but they didn’t fin significan thermal degradation of the CFRP at near the joint.

Fig.6.(a)Hardness mapping of steel rivet and AZ31B sheet;(b)hardness line scan of steel rivet following the vertical line in Fig.6a;and(c)hardness line scan of AZ31B following the horizontal line in Fig.6a.

3.3.Vickers microhardness of AZ31B and rivet

Fig.6a is the Vickers microhardness mapping of steel rivet and AZ31B Mg alloy in the CFRP-Mg alloy joint.Fig.6b and 6c show the line scan results of steel rivet and AZ31B Mg alloy,respectively.The averaged rivet head and shank hardness of steel rivet is 270 and 227 HV0.20,respectively.Compared with the as-received rivet(rivet head:266 HV and rivet leg:222 HV),the hardness for rivet head and shank after joining process does not significantl change.However,the flare the rivet tip in the F-SPR joint shows a larger hardening region due to work hardening under compressed axial force.Niuklin et al.[32]studied the high-temperature mechanical properties of 1022 low carbon steel and showed that the yield strength and ultimate tensile strength begin decreasing at 400 °C.Our estimated peak temperature during F-SPR is 248-294 °C.Therefore,it is reasonable to expect that our low carbon steel rivet(0.18 wt%)[23]hardness doesn’t deteriorate after joining.Furthermore,it demonstrates that our estimation of peak temperature during joining is reasonable.

The Vickers microhardness of base material AZ31B is estimated 72.5±2.7 HV0.20.As shown in Fig.6c,the microhardness of AZ31B adjacent to the steel rivet-AZ31B joint interface is 98.6 HV0.20(32%higher than that of base AZ31B),which is due to the ultrafin grain size present in this region due to dynamic recrystallization during the joining process.However,the hardness of AZ31B Mg alloy away from joint interface did not show significan change.

The hardness of AZ31B Mg alloy near at F-SPR joint increased from 73 HV to 99 HV when grain size was refine from 10μm to 0.3-1μm.Chang et al.[28]systematically studied the grain size refinemen in AZ31B using friction stir processing.The hardness increased from 50 HV to 83 HV as the grain size was refine from 75μm to 3μm.Fitting using H(HV)～d(μm)-1/2,the Hall-Petch relationship can be obtained as depicted in Fig.7:

Fig.7.Plot of Hall-Petch relationship of grain size of present study and literature[28].

The R2coefficien for the liner regression fittin is 0.91.Our result shows reasonable agreement with the reported results by friction stir processed AZ31B[28].

3.4.Crack initiation and growth during fatigue tests with DIC measurement

Because most of available literatures for CFRP-metal joints provide static mechanical joint properties,very limited information is available for fatigue life of CFRP-metal.Also,fatigue performances are greatly affected by specimen dimension,joint strength,testing peak failure load,and material properties.Therefore,it may not be fair to directly compare fatigue life of CFRP-metal joints from different joining techniques.Since RSW is a popular joining technique in automotive industry,the fatigue test results from our present study were compared with the available literature for resistance spot welded AZ31B and AZ31B specimens[33].This is because the previous work used the same coupons dimensions and

similar lap shear strength to the authors’result.At least,we can narrow down the factors that affect the variation of fatigue life.This comparison can provide qualitative information where our joining technology for CFRP-metal is now.For this reason,we only compare our results with resistance spot welded AZ31B and AZ31B with the same coupons dimension and similar lap shear failure load.Fig.8 summarizes the fatigue testing results of F-SPR samples with the rivet length of 6 mm and of RSW of AZ31B-AZ31B using the same sample dimensions from open literature[33].The relationship between fatigue life N(cycle)and the maximum load F(kN)islogN=5.33-2.38logF.Different nugget sizes show different tensile shear failure loads(i.e.,the larger nugget size is,the higher the failure load is)for resistance spot welded AZ31B-AZ31B.As seen in Fig.8,when the peak loads for F-SPR AZ31B-CFRP joint were greater than 2 kN,higher fatigue life was achieved compared with that of RSW AZ31B-AZ31B joints with the nugget size of 8.2,9.5 and 10.4 mm.In particular,for the joint samples with similar average lap shear tensile strength,the fatigue life of F-SPR AZ31B-CFPR is twice of that of RSW AZ31B-AZ31B(nugget size of 8.2 mm)at a maximum load of 2 kN.In addition,the superior fatigue life of F-SPR joint becomes more evident at higher maximum load(＞2 kN).At lower peak loads(e.g.,0.5 and 1 kN),the fatigue results of both joining techniques are comparable.In other words,F-SPR joint shows better fatigue performance for low-cycle fatigue region and comparable for high-cycle fatigue region compared with RSW joint.

Fig.8.Comparison of fatigue life of F-SPR AZ31B-CFRP at different loads with RSW AZ31B to AZ31B joint[33].The numbers in unit of mm are different nugget size of RSW while those in unit of kN indicate the average lap-shear tensile strength.

Fig.10.Failure mode of fatigue test at maximum load of 0.5,1 and 2 kN.All samples show failure in the bottom AZ31B sheet.

Fig.9 shows the series of DIC snapshot images(i.e.,strain in longitudinal direction,eyy)of AZ31B at the joint(backside view)with different number of fatigue cycles for F-SPR joint at maximum load of 0.5 kN.AtN=1,052,371 in Fig.9a,crack initiation close to the rivet bottom perpendicular to the longitudinal direction can be seen.AtN=1,059,771 in Fig.9b,crack reached to the left end of the joint while the crack at the right side was still propagating towards the edge of the AZ31B sample.The cracks fully propagated to the edges of the joint atN=1,070,971 as indicated in Fig.9c.

Fig.10 shows that the fracture mode of fatigue tests of F-SPR joint at maximum loads of 0.5,1 and 2 kN is predominantly AZ31B coupon failure.It can be seen that the fatigue failure modes at different maximum loads of 0.5,1 and 2 kN are all tear failure in AZ31B.Tear failure mode from fatigue testing of SPR joints of Al-Al[34-36],Al-CFRP[37,38]and Al-Fe[39]has been commonly found.

Fig.11 presents fractography of fatigue failure F-SPR samples at maximum loads of 0.5 kN.Fig.11a shows the fretting positions between the rivet and AZ31B Mg alloy.Regions 1 and 2 were in the top fretting region between the rivet and AZ31B,while Regions 3 and 4 were in the bottom fretting region.Fig.11b shows SEM images of fracture surfaces for 0.5 kN maximum load.Crack initiation is evident at the bottom fretting region(location 3 and 4).At the top fretting region of AZ31B sheet(location 1 and 2),a crack nucleated at the top of AZ31B sheet based on the river lines,which are associated with crack initiation.The crack propagation front lines originated from the top and bottom fretting regions of AZ31B and intersected as denoted by the yellow dash lines.For this reason,it is believed that the crack initiation at the top and bottom of AZ31B was due to fretting wear between the rivet and AZ31B.

Fretting wear during fatigue was reported for metal-metal[34,39-44],metal-CFRP[45]joint by SPR.The fretting occurred at the interface between the rivet and the plates.At the joint interface between CFPR,AZ31B and the rivet,the extent of fretting between the rivet and AZ31B is much greater than that of CFRP.It is because of the much higher wear rate of Mg alloy(10-3mm3/(N m)[46])than that of CFRP(10-8mm3/(N m)[47]),which explains the fatigue fracture observed on AZ31B.

Fig.11.(a)The fretting positions between rivet and AZ31B,(b)fracture surfaces of samples at maximum load of 0.5 kN.

When the maximum load is increased to 1 kN,the crack initiation in the fretting region of AZ31B plate can be clearly observed on the left side of the fracture surface,as shown in Fig.12a,b.The boundary between fatigue growth and fast fracture regions was observed and is designated with a yellow dash line.Another crack initiation was seen at the bottom of the AZ31B plate,but the crack propagation front line was not shown in the fatigue zone,indicating the crack initiation was formed in the late stage of the fatigue.However,on the right side,a combination of two crack propagation was observed,similar to that at the maximum load of 0.5 kN.

At the maximum load of 2 kN in Fig.13a,b,crack growth front line can be seen on the top fretting regions of AZ31B.However,no cracks occurred at the bottom fretting regions,indicating the fretting at the top fretting regions was dominant.With the increase of the maximum load,the fretting between the rivet and the top regions of AZ31B was more severe.

Fig.14 schematically describe the location of crack initiation during fatigue testing at different maximum loads.It can be seen that the crack initiation position changes from the bottom fretting regions to top fretting regions of the AZ31B plate as the maximum load increases.At lower maximum load,the movement amplitude of the rivet is small.The movement of the rivet is resisted by the bottom of the rive leg.Therefore,the fretting at the bottom regions is prominent,causing the crack initiation in the regions.However,at higher maximum load,the movement amplitude of the rivet becomes large,which will loosen the joint interface at the bottom of the rivet.In this way,the top regions endure most of the fretting,causing the crack initiation.

Fig.12.(a)The fretting positions between rivet and AZ31B,(b)fracture surfaces of samples at maximum load of 1 kN.

4.Conclusion

In this study,the lap shear strength of F-SPR joint of CFRP and AZ31B Mg alloy for different rivet shank length was investigated.Then,the fatigue life of down-selected F-SPR joint was studied at various maximum loads.The dynamic recrystallization of AZ31B Mg alloy and the fractography of the fatigue failure samples were discussed.The following conclusion can be drawn:

(1)The lap shear strength increased with the increase of the rivet shank size associated with greater interlock distances.The shear strength can reach to 5.07 kN with the shank size of 6 mm.

(2)Grain size gradient in the AZ31B was observed around the rivet shank at the joint due to the different degrees of dynamic recrystallization.A 100μm layer of submicron grains of AZ31B was formed adjacent to the rivet shank.The peak temperature and strain rate were calculated as 248-294 °C and 112.6 s-1based on Zener-Hollomon relationship.Estimated temperature for AZ31B at the joint is high enough to prevent crack formation during F-SPR.

Fig.13.(a)The fretting positions between rivet and AZ31B,(b)fracture surfaces of samples at maximum load of 2 kN.

Fig.14.The schematic of crack initiation sites,designated in red dots,in fatigue failure of(a)0.5 kN,(b)1 kN and(c)2 kN.

(3)The hardness of AZ31B near the rivet shank at joint interface was as high as 98 HV0.20,higher than 72.7 HV0.20 of the base material(i.e.,AZ31B)due to the fine grain size in the region.

(4)The fatigue life of F-SPR CFRP and AZ31B joints increased as the maximum load decreased.The fatigue life of F-SPR joints was longer than that of RSW of AZ31B to AZ31B at a high maximum load(＞2 kN).However,at a low maximum load(＜1 kN),the fatigue life was similar between the F-SPR and RSW AZ31BAZ31B joints.Under similar average lap shear tensile strength,the fatigue life of F-SPR is higher than that of RSW at 2 kN.Crack initiation occurred at the top and bottom fretting regions of AZ31B at a low maximum load(＜1 kN)due to fretting between the rivet and AZ31B.With the increase of the maximum load,the crack initiation at the top fretting regions of the AZ31B plate became more dominant.

Declaration of Competing Interest

The authors declare no conflict of interest.

Acknowledgment

This research was financiall sponsored by the US Department Energy Vehicle Technologies Office as part of the Joining Core Program.Oak Ridge National Laboratory(ORNL)is managed by UT-Battelle LLC for the US Department of Energy under Contract DE-AC05-00OR22725.The authors would like to thank Donald Erdman III and Rick R.Lowden for their help in the mechanical testing laboratory.