Lap shear strength and fatigue behavior of friction stir spot welded dissimilar magnesium-to-aluminu

Lap shear strength and fatigue behavior of friction stir spot welded dissimilar magnesium-to-aluminum joints with adhesiveS.H.Chowdhury a,D.L.Chen a,n,S.D.Bhole a,X.Cao b,P.Wanjara ba Department of Mechanical and Industrial Engineering,Ryerson University,350Victoria Street,Toronto,Ontario M5B2K3,Canadab National Research Council Canada Aerospace,5145Decelles Avenue,Montreal,Quebec H3T2B2,Canadaa r t i c l e i n f oArticle history:Received1September2012Received in revised form11November2012Accepted12November2012Available online19November2012Keywords:Magnesium alloyAluminum alloyAdhesiveFriction stir spot weldingLap shear strengthFatigue propertiesa b s t r a c tLightweighting is currently considered as an effective way in improving fuel efficiency and reducinganthropogenic greenhouse gas emissions.The structural applications of lightweight magnesium andaluminum alloys in the aerospace and automotive sectors unavoidably involve welding and joiningwhile guaranteeing the safety and durability of motor vehicles.The objective of this study was toevaluate the lap shear strength and fatigue properties of friction stir spot welded(FSSWed)dissimilarAZ31B-H24Mg alloy and Al alloy(AA)5754-O in three combinations,i.e.,(top)Al/Mg(bottom),Al/Mgwith an adhesive interlayer,and Mg/Al with an adhesive interlayer.For all the dissimilar Mg-to-Al weldcombinations,FSSW induced an interfacial layer in the stir zone(SZ)that was composed ofintermetallic compounds of Al3Mg2and Al12Mg17,which led to an increase in hardness.Both Mg/Aland Al/Mg dissimilar adhesive welds had significantly higher lap shear strength,failure energy andfatigue life than the Al/Mg dissimilar weld without adhesive.Two different types of fatigue failuremodes were observed.In the Al/Mg adhesive weld,at high cyclic loads nugget pull-out failure occurreddue to fatigue crack propagation circumferentially around the nugget.At low cyclic loads,fatiguefailure occurred in the bottom Mg sheet due to the stress concentration of the keyhole leading to crackinitiation followed by propagation perpendicular to the loading direction.In the Mg/Al adhesive weld,nugget pull-out failure mode was primarily observed at both high and low cyclic loads.&2012Elsevier B.V.All rights reserved.1.IntroductionSustainable growth in the transportation sector for the nextgeneration of motor vehicles will be increasingly contingent onthe development of energy efficient products that economicallyuse and conserve global resources whilst protecting the environ-ment through reduced emissions[1,2].The main challenge interms of manufacturing in the aerospace and ground transporta-tion sectors is to reduce the weight of the vehicle body viaapplying advanced lightweight materials without compromisingreliability and safety.This has recently prompted the develop-ment and application of ultra-lightweight magnesium alloys dueto their low density,high strength-to-weight ratio and superiordamping capacity[3–7].On the other hand,design and manu-facturing with aluminum alloys are longstanding and increasinglysignificant in aerospace and automotive structures to take advan-tage of the light weight and proper combination of strength withductility[8,9].Co-existent applications of magnesium and alumi-num alloys afford design freedom and manufacturingflexibility,but require the development of advanced joining processes,espe-cially with due consideration of the challenges related to theconventional fusion welding to join magnesium-to-aluminum alloys.Friction stir welding(FSW),a solid-state joining technology,developed by The Welding Institute of Cambridge,UK,in1991,shows a considerable potential for assembly of sheet/plate mate-rials as the low heat input and process temperatures(absence ofmelting)enable good retention of the baseline mechanical prop-erties and high weld quality with relatively few defects.As aderivative of the FSW process,friction stir spot welding(FSSW)was later proposed and successfully applied to manufacture hoodand rear doors of a sport vehicle[10].For aluminum andmagnesium alloys,conventional assembly by resistance spotwelding presents some technical concerns,including weld porosity,electrode wear,high energy consumption,low productionefficiency,and inconsistency in failure modes[11–15].Thusalternative and novel technologies are of avid interest in theautomotive industry.Recent research and development thrustsin joining aluminum-to-magnesium sheet materials are toexplore methods such as structural adhesives,rivets and toggle-locks.The FSSW process is an emergent technology that may offera potentially viable solution for dissimilar metal assembly with-out adding extra weight whilst enabling joint strength and weldContents lists available at SciVerse ScienceDirectjournal homepage:/locate/mseaMaterials Science&Engineering A0921-5093/$-see front matter&2012Elsevier B.V.All rights reserved./10.1016/j.msea.2012.11.039n Corresponding author.Tel.:þ14169795000x6487;fax:þ14169795265.E-mail address:dchen@ryerson.ca(D.L.Chen).Materials Science&Engineering A562(2013)53–60quality with a reasonable operational cost[11].Although a number of papers on the joining of Mg and Al alloys using FSSW have recently been reported[8,10,16,17],only limited studies have been done on lap shear strength and especially fatigue resistance of the dissimilar Al-to-Mg FSSW joints,a topic of vital importance for ensuring safe and reliable applications.Lin et al.[18]observed differences in the fracture path under quasi-static loading compared with cyclic loading for Al6111-T4alloy.The FSSWed joints of dissimilar Al-to-Mg alloys were observed to have different failure modes compared with the similar welds (i.e.,identical alloys for the top and bottom sheets)[8,16,19]. Recently,some new kinds of hybrid welding techniques have been used to join dissimilar Al-to-Mg alloys,such as laser adhesive welding[20–23].Liu and Ren[20]indicated that the lap shear strength increased in the MIG spot adhesive welded AZ31B-to-Al6061joints,while Wang et al.[23]observed that the addition of the adhesive also increased the weld penetration depth in the Al alloy side for Nd-YAG laser adhesive welds of dissimilar Mg-to-Al alloys.In the present study,friction stir spot adhesive welding,an innovative derivative to FSSW and adhesive bonding,was envi-saged for assembly of dissimilar Al-to-Mg alloys.While adhesives provide excellent uniform stress distribution over a large bonding area,the weld itself improves the peel resistance of the adhesive joints.However,the integrity and durability of FSSWed dissimilar Al/Mg and Mg/Al adhesive welds especially under cyclic loading have not been studied yet,since any lap shear strength and fatigue behavior of such adhesive welds in relation to the failure mode changes have not been reported in the literature.It is also unknown how large is the difference between the FSSWed dissimilar Al/Mg welds with and without the use of the adhesive. The main objective of the present study was,therefore,to identify the lap shear strength and fatigue behavior of the friction stir spot adhesive welded AZ31B-H24alloy to AA5754-O alloy and com-pare the results with those for the FSSWed joints.2.Experimental procedureCommercial AZ31B-H24Mg and AA5754-O Al alloy sheets with a thickness of2mm were selected for FSSW.The nominal chemical composition was Mg-3Al-1Zn-0.6Mn-0.005Ni-0.005Fe for the AZ31B-H24and Al-3.42Mg-0.23Sc-0.22Zr for the AA5754-O.The Mg and Al alloy sheets were cut into small coupons of35mmÂ100mm with the loading axis along the rolling direction.Two coupons were overlaid over an area of 35mmÂ35mm between which Terokal5089adhesive had been applied and cured at a temperature of1701C for20min.FSSW was then performed at the center of the overlapped area.Three types of dissimilar material combinations,i.e.,(top)Al/Mg(bot-tom),(top)Al/Mg(bottom)with adhesive,and(top)Mg/Al (bottom)alloys with adhesive,were used.The FSSW tool fabri-cated from H13tool steel consisted of a scrolled shoulder with a diameter of13mm and a left-hand threaded pin with a diameter of5mm.FSSW was performed using a MTS I-STIR FSW machine with the process parameters that consisted of a pin length of 2.8mm,tool rotational rate of2000rpm,tool plunge rate of 3mm/s,tool removal rate of15mm/s,shoulder plunge depth of 0.2mm and dwell time of2s.The welded samples were carefully sectioned using a low speed diamond blade to extract specimens for microstructural examination and weld integrity analysis. Metallographic preparation of the specimens involved cold mounting followed by grinding with SiC papers to1200grit, polishing to a0.05m mfinish and etching the Mg alloy side with acetic picral(10ml acetic acid(99%),4.2g picric acid,10ml H2O, 70ml ethanol(95%))and the Al alloy side with Keller’s reagent (2.5ml nitric acid,1.5ml hydrochloric acid,1ml hydrofluoric acid and100ml distilled water).The microstructure was observed with an optical microscope equipped with Clemex quantitative image analysis software.Vickers microhardness measurements across the weld were obtained using a computerized Buehler microhardness tester operated at a load of100g with a dwell time of15s.It is noteworthy that according to ASTM:E384-11e1, the spacing between two adjacent indentations must be at least three times the diagonal length of the indentation.Hence,the spacing between two adjacent indentations during microhardness testing was suitably selected to avoid any potential effect of the strainfields caused by adjacent indentations.To evaluate the mechanical strength of the joints,lap shear tensile tests of the welds were conducted in air at room temperature(RT)using a fully computerized United tensile testing machine at a crosshead displacement speed of10mm/min.Fatigue tests were carried out using a fully computerized Instron8801servo-hydraulic testing system under load control at RT and different load amplitudes.A load ratio of R(P min/P max) equal to0.2,sinusoidal waveform,and a frequency of50Hz were applied for all the tests.At least two samples were tested at each load level.The fracture surfaces of the FSSWed joints after fatigue testing were examined using a JSM-6380LV scanning electron microscope(SEM)equipped with Oxford energy disper-sive X-ray spectroscopy(EDS)system and three-dimensional (3D)fractographic analysis capacity.Additionally,a multi-functional PANalytical X-ray diffractometer was used to identify the formation of potential intermetallic compounds in the dis-similar Mg-to-Al adhesive welds from the fracture surface(Mg side)after fatigue testing.X-ray diffraction(XRD)was conducted using CuK a radiation(wavelength l¼0.15406nm)at45kV and 40mA.The diffraction angle(2y)at which the X-rays hit the sample varied from201to1101with a step size of0.051and3s in each step.3.Results and discussion3.1.MicrostructureTypical base metal(BM)microstructures for the AZ31B-H24 Mg alloy and the AA5754-O Al alloy are shown in Fig.1(a)and(b), respectively.As shown in Fig.1(a),elongated and pancake-shaped grains with varying sizes were observed in the BM of the Mg alloy due to both the deformation of the2mm thick sheet by rolling and incomplete dynamic recrystallization(partial annealing) [24–26].Typical optical images at the interface of the dissimilar Al/Mg and Mg/Al adhesive welds are shown in Fig.1(c)and(d), respectively.For the Al/Mg weld,the interface was marked by the presence of an interfacial layer which covered most of the boundary but the thickness of the interfacial layer varied.How-ever,it was different for the Mg/Al adhesive weld where the interface was marked by both the adhesive and interfacial layer. Fig.2(a)shows an SEM image of the interface taken from the cross-section of an Al/Mg weld.A distinguishable interlayer with a thickness up to$20m m was observed at the interface between the Mg(at the left side in Fig.2(a))and the Al(at the right side in Fig.2(a))alloys.The EDS line analysis revealed the mutual presence of Al and Mg across the reaction layer,suggesting that the intermetallic compound(IMC)layer formed at the Al/Mg interface during FSSW.As shown by the corresponding EDS line analysis of the image in Fig.2(b),the presence of the adhesive in the Mg/Al adhesive weld interface decreased the thickness of the IMC layer.An X-ray diffraction pattern obtained from the fracture surface of the Mg/Al adhesive weld(Mg alloy side)is shown inS.H.Chowdhury et al./Materials Science&Engineering A562(2013)53–60 54Fig.3.Besides Mg,Al 3Mg 2and Al 12Mg 17were detected on the fracture surface of the Mg alloy side.Similar results were also reported by Sato et al.[16]and Choi et al.[27]for the FSSW of AZ31to AA 5083and AA 6K21alloys,and Firouzdor and Kou for the lap and butt FSW of AZ31B to 6061Al alloys [28,29].The formation of those intermetallic compounds in the Al/Mg dissim-ilar welds was due to the constitutive liquation during FSSW [16,27,30].Gerlich et al.[31]conducted FSSW of AA6111to AZ91alloys and reported that the peak temperature in the vicinity of the tool rapidly reached the a -Mg and Al 12Mg 17eutectic tem-perature of 4371C,at which the eutectic microstructure formed.According to the Al–Mg phase diagram [32],a wide range in the chemical composition exhibited a low melting temperature in the Al–Mg system which most likely assisted in the formation of the inhomogeneous interfacial layer in the SZ of the Al/Mg as well as the Al/Mg and Mg/Al adhesive welds.3.2.MicrohardnessThe hardness distribution across the Al/Mg and Mg/Al adhe-sive welds obtained from the upper and lower sheets is shown in Fig.4(a)and (b),respectively.There was only a small difference in50 μm 100 μm50 μmIntermetallicsAZ31B-H24 Mg 5754-O Al100 μmIntermetallics5754-O AlAZ31B-H24 MgFig.1.Typical microstructures of a FSS weld of AZ31B-H24Mg alloy to AA5754-O Al alloy:(a)base metal of AZ31B-H24Mg alloy,(b)base metal of AA5754-O Al alloy,(c)presence of intermetallics at the interface of a dissimilar Al/Mg weld,and (d)presence of intermetallics at the interface of a dissimilar Mg/Al adhesiveweld.010020030020406080I n t e n s i t y , C o u n t sPosition, μm 0100200300400020406080I n t e n s i t y , C o u n t sPosition, μmFig.2.SEM micrographs of the Al–Mg phase at the interface of dissimilar (a)Al/Mg weld,(b)Mg/Al adhesive weld and the corresponding EDS line scans across the interfaces showing the compositional variations of Al and Mg.S.H.Chowdhury et al./Materials Science &Engineering A 562(2013)53–6055hardness between the upper and lower sheets and the hardness increased progressively close to the keyhole periphery.As shown in Fig.4(a),the hardness distribution of the Al/Mg adhesive weld was observed to be nearly symmetric with respect to the center of the keyhole.As shown in Fig.4(a)and (b),the SZ in the Al/Mg andMg/Al adhesive welds had much higher hardness values lying in-between HV 90and 125.This higher hardness was predominantly due to the presence of hard and brittle IMCs (Al 3Mg 2and Al 12Mg 17phases),as identified by XRD (Fig.3).SimilarresultsFig. 3.X-ray diffraction pattern obtained from the fracture surface of the dissimilar Mg/Al adhesive weld on the Mg side.V i c k e r s h a r d n e s s , H VV i c k e r s h a r d n e s s , H V04080120160200-15-10-5051015Distance from weld center, mmFSSW Al/Mg adhesive Bottom (Mg)BMBMHAZ HAZ T M AZ T MA Z SZ 5 mm04080120160200-15-10-5051015Distance from weld center, mmFig.4.Typical microhardness profiles of the dissimilar (a)Al/Mg,and (b)Mg/Al adhesive welds at differentlocations.02468101201234L oa d , k NDisplacement, mm36912Al-Mg Al-Mg (adhesive)Mg-Al (adhesive)M a x i m u m l o a d , k N51015202530Al-MgAl-Mg(adhesive)Mg-Al(adhesive)F a i l u r e e n e r g y , J5p shear properties of the dissimilar Al/Mg weld,Al/Mg adhesive weld,and Mg/Al adhesive weld tested at a crosshead speed of 10mm/min:(a)the load vs.displacement curves,(b)the maximum load,and (c)the failure energy.S.H.Chowdhury et al./Materials Science &Engineering A 562(2013)53–6056were also reported by other researchers [16,27,33–38].Therefore,the microhardness test results provided a further corroboration for the occurrence of hard IMCs in the Al/Mg and Mg/Al adhesivewelds,which may become an easy fracture path to be discussed in the sections p shear strengthTypical load vs.displacement curves for the Al/Mg weld as well as the Al/Mg and Mg/Al adhesive welds are shown in Fig.5(a).It is seen that both the Al/Mg and Mg/Al adhesive welds were much stronger than the Al/Mg weld,and the Mg/Al adhesive weld was in turn stronger than the Al/Mg adhesive weld.The obtained maximum lap shear load is shown in Fig.5(b).It is clear that the maximum lap shear load of the Mg/Al adhesive weld was higher than that of the Al/Mg adhesive weld,which was in turn higher than that of the Al/Mg weld.The considerably lower maximum lap shear load in the Al/Mg dissimilar weld was obviously attributed to the formation of the brittle IMC layer at the interface (Fig.1(c)and (b)).Decreasing the extent of IMC formation at the interface by using an adhesive in the Mg/Al adhesive weld resulted in a higher lap shear load.A similar observation of the maximum lap shear load was also reported by Liu and Ren [20]for 1.5mm thick MIG spot adhesive welded AZ31to AA 6061sheets;they reported that the maximum lap shear load increased by three times compared to that of the MIG welded joint without adhesive.Higher lap shear loads for laser adhesive welding of AZ31to AA 6061alloy was also reported in [21,22].To characterize more completely the lap shear character-istics of the welds,the failure energy was estimated,where the02468101214161E+11E+21E+31E+41E+51E+61E+71E+8M a x i m u m l o a d , k NNumber of cycles to failure, N fFig.6.S –N curves of the dissimilar Al/Mg weld,Al/Mg adhesive weld and Mg/Al adhesive weld tested at R ¼0.2,50Hz and room temperature,where solid symbols indicates the nugget pull-out failure and empty symbols indicates the failure perpendicular to the loading direction.13wt%Mg 87wt%AlCrack initiationCrack initiationFig.7.Typical macroscopic and SEM images of the fatigue fracture surface of the dissimilar Al/Mg adhesive weld:(a)failure pattern at a higher maximum load of P max ¼8kN,(b)failure pattern at a lower maximum load of P max ¼3kN,(c)normal tensile fracture surface at a lower P max ¼3kN,(d)a magnified image showing the crack initiation site at a weld nugget at a lower P max ¼3kN (as indicated in (c)),(e)fatigue striations in the crack propagation zone at a higher magnification,and (f)nugget pull-out fracture surface at a higher P max ¼8kN.S.H.Chowdhury et al./Materials Science &Engineering A 562(2013)53–6057failure energy was defined as the area below the load–displacement curve shown in Fig.5(a).When the tensile lap shear remained continuous after reaching the maximum load without failure (i.e.,in the case of Al/Mg and Mg/Al adhesive welds in the present study),the failure energy was only calculated correspond-ing to the maximum load in a more conservative manner.The failure energy results are plotted in Fig.5(c).The failure energy of the Mg/Al adhesive weld ($26.5J)was greater than that of the Al/Mg adhesive weld ($9.5J)and both of these values were substantially greater than that of the Al/Mg weld ($0.34J).Hence the application of an adhesive during FSSW influences markedly the joint strength or stability through the creation of a large lap bonded area.Also,since there exists a direct correlation between the failure energy in the lap shear tensile tests and impact tests,it is expected that the Mg/Al adhesive weld would have a higher impact resistance than that of the Al/Mg adhesive weld.Further-more,nugget pull-out failure was observed in the Al/Mg and Mg/Al adhesive welds while nugget debonding failure was observed in the Al/Mg weld during lap shear tensile tests (Fig.5(a)).Therefore,the presence of adhesive provided a further corrobora-tion on the higher lap shear strength for the Al/Mg and Mg/Al adhesive welds which led to different failure modes under cyclic loading to be seen in the following section.Regarding the observation that the Mg/Al adhesive weld exhibited a higher strength and failure energy than the Al/Mg adhesive weld as shown in Fig.5,the following two potential reasons could be present.First,as seen in the insert of Fig.5(a),while the residual adhesive remained stuck on both Mg and Al sides of the fracture surfaces in both cases in a mix of cohesive and adhesive failure,it appeared that somewhat more cohesive type of failure occurred in the Mg/Al adhesive weld,thus giving rise to a higher strength and failure energy.This was due to the fact that the cohesive type of failure was more desirable as it assured the use of maximum strain energy by the weaker part of the joint [39–41].It should be noted that the cohesive failure means that the failure occurred within the adhesive layer,whereas the adhesive failure denotes that the failure occurred at the interface between the adhesive and the substrate (i.e.,either Al sheet or Mg sheet in the present study).Second,since Mg alloy has a slightly lower melting point than Al alloy,it would be easier for Mg alloy to become more softened and thus penetrate more deeply into the stir zone and interlock with the bottom Al alloy,when Mg alloy was positioned on the top and the same FSSW processing parameters were used.Similar results were also observed by Firouzdor and Kou [28]who reported that in an Al-to-Mg butt FSW,Mg on the advancing side could provide a better chance for AZ31B-H24Mg alloy to penetrate deeper into the stir zone and interlock with 6061-T6Al alloy to increase the jointstrength.Fig.8.Typical macroscopic and SEM images of the fatigue fracture surface of the dissimilar Mg/Al adhesive weld:(a)failure pattern at a lower maximum load of P max ¼4kN,(b)failure pattern at a higher maximum load of P max ¼8kN,(c)nugget pull-out fracture surface at a lower P max ¼4kN,(d)nugget pull-out fracture surface at a higher P max ¼8kN,(e)fatigue striations in the crack propagation zone at a higher magnification,and (f)secondary cracks at a higher magnification as indicated by an arrow at a higher P max ¼8kN.S.H.Chowdhury et al./Materials Science &Engineering A 562(2013)53–60583.4.Fatigue behavior and failure modeFatigue test results of the Al/Mg weld,as well as the Al/Mg and Mg/Al welds with an adhesive obtained at RT,R¼0.2,and50Hz are plotted in Fig.6,where solid symbols indicates the nugget pull-out failure and empty symbols indicates the failure perpen-dicular to the loading direction.Resembling the maximum load and failure energy(Fig.5(b)and(c)),the fatigue life of the Mg/Al adhesive weld was in general longer than that of the Al/Mg adhesive weld.The fatigue life of the Al/Mg dissimilar weld was much lower and more scattered than that of both the Mg/Al and Al/Mg adhesive welds especially at a maximum cyclic load higher than1kN.The substantially improved fatigue resistance of the Mg/Al and Al/Mg adhesive welds was primarily attributed to the elimination of stress concentration surrounding the weld nugget as a result of large lap bonded area which may lessen the formation of IMC layer at the interface and lead to a uniform stress distribution in the lap joint during fatigue testing.Similarly, a longer fatigue life for the resistance spot adhesive welded AZ31-to-HSLA steel sheets was also observed in comparison with that of joints assembled without an adhesive under the same cyclic loading condition(R¼0.2,f¼50Hz)[42].As indicated in Fig.6,the fatigued samples of the Al/Mg dissimilar weld failed in the mode of nugget pull-out at all cyclic stress levels.Also,the samples of the Mg/Al and Al/Mg adhesive welds fatigued at the higher cyclic stress levels failed in this mode.Fig.7(a)shows the onion ring shaped weld nugget,with a mix of both Al and Mg as determined by EDS,which was generated and pulled out in-between the top and bottom sheets of the Al/Mg adhesive welds at a maximum cyclic load of8kN. The nugget pull-out failure occurred as a result of the crack propagating circumferentially around the nugget(Fig.7(f)).The shear stress in the remaining area of the nugget increased with increasing crack length as the crack advanced.After the crack had propagated around half of the nugget with a keyhole,the shear stresses acting in the area were such that the remaining cross section could not sustain the shear overload,then the nugget pull-out failure happened.In contrast,at a lower maximum cyclic load of3kN the failure of the fatigued samples of the Al/Mg adhesive weld occurred by crack initiation at the border of weld nugget(as indicated in Fig.7(b))and crack propagation nearly at a rightangle to the loading direction.The fractographic image of such a normal tensile fracture surface is shown in Fig.7(c),and the magnified crack initiation site on the left side of Fig.7(c)is shown in Fig.7(d),where the crack initiation was caused by the stress concentration of the keyhole present in the bottom Mg sheet. Fatigue crack propagation with its advancing direction indicated by the dashed arrows in Fig.7(d)was mainly characterized by the formation of fatigue striations,as seen in Fig.7(e).Nugget pull-out failure mode was predominantly observed in the Mg/Al adhesive weld at both high and low cyclic loads(Fig.8(a)–(d)).Specially, except one sample tested at a maximum cyclic load of6kN(Fig.6), all of other fatigued samples of the Mg/Al adhesive weld failed in the mode of nugget pull-out.For the fatigue crack propagation similar fatigue striation characteristics were observed(Fig.8(e))in conjunction with some secondary cracks(Fig.8(f)),which appeared usually perpendicular to the crack propagation direction.Fatigue striations usually occurred by a repeated plastic blunting-sharpening process in the face-centered cubic materials,arising from the glide of dislocations on the slip plane along the slip direction within the plastic zone ahead of a fatigue crack tip[43]. The formation of fatigue striations in the magnesium alloys with a hexagonal close-packed crystal structure was expected to be additionally associated with the occurrence of twinning–detwinning process[3,44–49],since there exist a lower number of slip systems in magnesium alloys at room temperature.Finally,it should be noted that the nugget pull-out or debonding failure mode was also associated with the presence of interme-tallic layers of Al3Mg2and Al12Mg17phases,as observed via SEM/ EDS(Fig.9)and identified via XRD(Fig.3)on the fracture surface of the Mg/Al adhesive weld(Mg side),which would result in a decrease in the fatigue life of the Mg/Al adhesive weld.Further studies are needed to minimize or eliminate the formation of the intermetallic compounds in-between the Mg and Al alloys during FSSW,e.g.,by adding tin or zinc interlayer similar to the ultra-sonic spot welding of Mg and Al alloys[38].4.Conclusions1.A distinctive interfacial layer consisting of Al12Mg17andAl3Mg2intermetallic compounds was observed in the friction stir spot welded dissimilar Al/Mg and Mg/Al adhesive joints.In comparison with the Al/Mg weld without adhesive,the extent of forming the intermetallic compounds decreased in the dissimilar adhesive joints.2.A much higher hardness with values in-between HV90and125was observed in the stir zone of Al/Mg and Mg/Al adhesive welds due to the presence of intermetallic compound layer.3.Both Mg/Al and Al/Mg adhesive welds had significantly higherlap shear strength and failure energy than the Al/Mgdissimilar 020040060080010000102030405060 Intensity,CountsPosition, μmFig.9.(a)A typical SEM micrograph of the Al–Mg phase present on the fracture surface of dissimilar Mg/Al adhesive weld(Mg side),and(b)EDS line scan across the intermetallics indicated in(a)showing the Al and Mg compositional variations.S.H.Chowdhury et al./Materials Science&Engineering A562(2013)53–6059。