Materials Science and Engineering A 527 (2010) 29512961

Contents lists available at ScienceDirect

Materials Science and Engineering A

journa l homepage: www.e lsev ier .co

Tensile iorand fric oy

S.M. Cho idaa Department o to, Onb Aerospace Ma uncil Cc Department o enue W

a r t i c l

Article history:Received 17 NReceived in reAccepted 11 Ja

Keywords:AZ31 magnesium alloyDouble-sided arc weldingFriction stir weldingMicrostructureTensile propertiesStrain hardeni

d wo31B-Hgher,ed s

ng exation

twice that of the base metal. The DSAWed samples exhibited stronger strain-hardening capacity due tothe larger grain size coupled with the divorced eutectic structure containing -Mg17Al12 particles in thefusion zone, compared to the FSWed samples and base metal. KocksMecking type plots were used toshow strain-hardening stages. Stage III hardening occurred after yielding in both the base metal and thewelded samples. At lower strains a higher strain-hardening rate was observed in the base metal, but it

1. Introdu

Weightimportantmronment [1alloys, haveandaerospato-weight rcastability [to further exwelding (FSWelding Intial for joinweld defec[1,5]. FSWplastic defoworkabilityOn the othe

CorresponE-mail add

0921-5093/$ doi:10.1016/j.ng decreased rapidlywith increasingnetowstress. Athigher strains the strain-hardening rateof theweldedsamples became higher, because the recrystallized grains in the FSWed and the larger re-solidied grainscoupledwith particles in the DSAWed providedmore space to accommodate dislocationmultiplicationduring plastic deformation. The strain-rate sensitivity evaluated via Lindholms approach was observedto be higher in the base metal than in the welded samples.

2010 Elsevier B.V. All rights reserved.

ction

reduction in ground vehicles and aircraft is one of theeasures to improve fuel economyandprotect the envi-

]. Magnesium alloys, as the lightest metallic structuralbeen and will be increasingly used in the automotivece industries due to their lowdensity [1], high strength-atio [13], environmental friendliness, recyclability and2,3]. However, effective joining techniques are requiredpand the applications ofmagnesiumalloys.Friction stirW), a solid-state joining technique developed by Thestitute of Cambridge, UK, in 1991 [4], has great poten-ing magnesium alloys, since it can signicantly reducets normally associated with fusion welding processeshas also been used to rene the grain size via severermation and recrystallization [69] so as to improve theof Mg alloys and increase the strength of welded joints.r hand, the weldability of magnesium alloys by some

ding author. Tel.: +1 416 979 5000x6487; fax: +1 416 979 5265.ress: dchen@ryerson.ca (D.L. Chen).

arc welding processes such as gas tungsten arc welding (GTAW) isconsidered to be excellent as well [10]. In 1999, Zhang and Zhang[11] developed and patented a novel arc welding process referredto as double-sided arc welding (DSAW). The DSAW process usesone welding power supply and two torches; frequently a plasmaarc welding (PAW) and GTAW torch each connected directly to oneof the power supply terminals. The torches are positioned on oppo-site sides of a work-piece such that the welding current ows fromone torch through the work-piece to the opposite torch. Zhang etal. [1215] have examined the feasibility of using the DSAW pro-cess to make vertical-up, keyhole-mode welds in 612mm thickplain carbon steel, stainless steel or aluminum alloy plates. Morerecently,Weckman and co-workers [16,17] have examined the fea-sibility of using the DSAW process for conduction-mode weldingof 1.2mm thick AA5182-O aluminum sheet for tailor welded blankapplications. It was noted that the opposing welding torches andsquare-wave AC welding current successfully cleaned the oxidefromboth sides of the joint and produced visually acceptableweldsat speeds up to 3.6m/min. Through-thickness heating was moreuniform with DSAW than with other single-sided welding pro-cesses allowing symmetric welds to be produced with minimalangular distortion of the sheets [16,17].

see front matter 2010 Elsevier B.V. All rights reserved.msea.2010.01.031properties and strain-hardening behavtion stir welded AZ31B magnesium all

wdhurya, D.L. Chena,, S.D. Bholea, X. Caob, E. Powf Mechanical and Industrial Engineering, Ryerson University, 350 Victoria Street, Toronnufacturing Technology Centre, Institute for Aerospace Research, National Research Cof Mechanical and Mechatronics Engineering, University of Waterloo, 200 University Av

e i n f o

ovember 2009vised form 8 January 2010nuary 2010

a b s t r a c t

Microstructures, tensile properties anand friction stir welded (FSWed) AZrates. While the yield strength was hithe FSWed samples than in the DSAWin the FSWed samples. Strain-hardeniLudwik equation and a modied equm/locate /msea

of double-sided arc welded

jkoc, D.C. Weckmanc, Y. Zhouc

tario M5B 2K3, Canadaanada, 5145 Decelles Avenue, Montreal, Quebec H3T 2B2, Canadaest, Waterloo, Ontario N2L 3G1, Canada

rk hardening behavior of double-sided arc welded (DSAWed)24 magnesium alloy sheet were studied at different strain

both the ultimate tensile strength and ductility were lower inamples due to welding defects present at the bottom surfaceponents were evaluated using the Hollomon relationship, the. After welding, the strain-hardening exponents were nearly

2952 S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961

Mechanical properties such as strength, ductility, strain-hardening behavior, strain-hardening sensitivity, etc., of all weldsmade inmagnesiumalloypartsused in structural applicationsmustbeevaluated toensure the integrityandsafetyof the joint andstruc-ture. Whilemagnesiumerties of we[18] studiedsile propertreported therties innonalloy. Zhu eon the welddiode laserbeendoneoAZ61 [25] alier results oFSWed magumented (etests on a Mthe values oing strain raAZ31B-H24FSWed thixhardening[32] studiesium alloy sSome authonesium allograin size smaterial [2formation owrought mpasses (i.e.,side and thside of thedouble-sidebeen examicould be usable mechatherefore, wproperties,and FSWed

2. Materia

In the pwas used.2.53.5wt%[37]. Twoemployed tthe butt joimadewith ttion of thetorch wereA detailed dparameterspieces wereon the surfaarea of thewwere madeof 1.4 kW.10mm/s an1.65mmroface oxideswas cleaned

The welded joints perpendicular to the welding direction werecut and cold mounted in order to examine the microstructure ofthe fusion zone (FZ), heat-affected zone (HAZ) and basemetal (BM).The mounted samples were manually ground, polished, and etched

cetic picral (10mL acetic acid (99%), 4.2 g picric acid, 10mL0mL ethanol (95%)) [38]. The microstructure was observedopticalmicroscope equippedwith quantitative image anal-

ftware. Vickers microhardness tests were conducted withuterized Buehler machine across the sectioned weld with

ing of 0.5mm. A load of 100g and dwell time of 15 s wereduring the hardness tests. Sub-sized tensile specimens in

ancewithASTME8M-08 standard [39]weremachined alongling (or longitudinal) direction for both the base metal andjoints, where the weld was positioned at the center of the

area. Tensile testswereperformedusing a computerized ten-ting machine at constant strain rates of 1102, 1103,4 and 1105 s1 at room temperature. At least two sam-ere tested at each strain rate. The fracture surfaces wereed using a scanning electron microscope (SEM) equipped

n energy dispersive X-ray spectroscopy (EDS) system and 3Draphic analysis capacity.

ults and discussion

icros

mic1, wes wbaseby ronealimAZ3welof t

d zoed anparised inken poy wHAZargeallizesent

Typicathere are numerous investigations on the properties ofalloys, only a limited number of studies of the prop-lded magnesium alloy joints are reported. Quan et al.the effects of heat input on microstructure and ten-

ies of laser welded AZ31 Mg alloy. Liu and Dong [19]e effect of microstructural changes on the tensile prop--autogenousgas tungstenarcweldedAZ31magnesiumt al. [20] presented the effect of welding parametersing defects and change of microstructure in CO2 and

welded AZ31 magnesium alloy. Tensile testing has alson friction stirweldedAZ31 [1,5,2124], FSWedwroughtnd a ne-grained laser welded Mg alloy [26]. Some ear-n the microstructural changes and strengths of variousnesium alloys and other alloys have also been well doc-.g., in refs. [2729]). Takuda et al. [30] performed tensileg9Li1Y alloy at room temperature and observed thatf strain-hardening exponents increased with increas-te. Afrin et al. [24] obtained similar results for a FSWedMg alloy. Yu et al. [31] evaluated the tensile strength ofomolded AE42 Mg alloy, but no information on strain-and strain-rate sensitivity was given, while Lee et al.d the formability of friction stir welded AZ31 magne-heet and other alloys experimentally and numerically.rs have studied the strain-hardening behavior of mag-ys with emphasis on the relationships between thetrengthening and dislocation strain hardening of the,24,3335]. While Shen et al. [36] have evaluated thef macropores in double-sided gas tungsten arc weldedagnesium AZ91D alloy plates made with two separatewelded with one partial penetration weld on the topen a separate partial penetration weld on the backplate), the mechanical properties of conduction-moded arc welds made in magnesium alloy sheet have notned. It is unknown if this novel arc welding techniqueed to produce welds in magnesium alloys with accept-nical properties. The aim of the present investigation,as to evaluate and compare the microstructure, tensile

strain-hardening and strain-rate sensitivity of DSAWedAZ31B-H24 Mg alloy sheet.

ls and experimental procedure

resent study, 2mm thick AZ31B-H24 Mg alloy sheetThe nominal chemical composition of this alloy wasAl, 0.71.3wt% Zn, 0.21.0wt% Mn and balance Mg

different welding methods, DSAW and FSW, wereo make autogenous welds between the work-pieces innt conguration. Both DSAWed and FSWed joints werehewelding direction perpendicular to the rolling direc-sheet. In the DSAW process, a PAW torch and a GTAWused with a square-wave AC welding power supply.escription of the welding apparatus and other processusedmay be found in [16,17]. Prior to DSAW, thework-degreased using acetone and then alcohol. The oxidece of the sheets was then mechanically removed in theeld using a stainless steelwire brush. TheDSAWweldsusing a welding speed of 25mm/s and welding powerThe FSW welds were made using a welding speed ofd a right-hand threaded pin tool having a pin length oftating clockwise at a rate of 2000 rpm. Prior to FSW, sur-were removed with a steel brush and then the surfaceusing ethanol as well.

using aH2O, 7with anysis soa compa spacappliedaccordthe rolweldedgaugesile tes110ples wexaminwith afractog

3. Res

3.1. M

Thein Fig.ing sizof thesheettial anabout 5FSWedthe topsectionaffecteequiaxin comappearalso tathe allof thesome lrecrystthe pr

Fig. 1.tructure

rostructure of the AZ31B-H24 Mg base metal is shownhere elongated and pancake-shaped grains with vary-ere observed. The heterogeneity in the grain structuremetal was due to both deformation of the 2mm thicklling and incomplete dynamic recrystallization (par-ng) [1]. The average grain size of the base metal was. The typical macroscopic and microscopic structures of1B-H24 Mg alloys are shown in Fig. 2. Fig. 2(a) showsd bead after FSW and Fig. 2(b) presents a typical cross-he FSWed sample including HAZ, thermomechanicallyne (TMAZ) and stir zone (SZ). As seen in Fig. 2(c), bothd elongated grains were present in the HAZ. However,on to the base metal (Fig. 1), far more equiaxed grainsthe HAZ, indicating that partial recrystallization hadlace during FSW. The recrystallization temperature of

as approximately 205 C. Thus, the temperature in partmay have been above this value. This is conrmed bygrains observed in the HAZ due to grain growth afteration [1]. The grain structure in the TMAZ (Fig. 2(d)) instudy is basically equiaxed and recrystallized, which

l microstructures of the base metal (BM) of the AZ31-H24 Mg alloy.

S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961 2953

Fig. 2. TypicaAZ31-H24 allo(c) heat-affectstir zone (SZ).

was similarand Afrin etions [41] welongated gbecame notThese chanl macroscopic and microscopic structures of a friction stir weldedy. (a) Top weld bead surface, (b) cross-section of the welded joint,ed zone (HAZ), (d) thermomechanically affected zone (TMAZ), and (e)

to the recent results reported by Cao and Jahazi [1]t al. [5,40], while it was different from earlier observa-here the TMAZ was still characterized by deformed andrains. The grains in the SZ were equiaxed (Fig. 2(e)) andiceably bigger in the center of the stir zone (8m).ges were caused by dynamic recrystallization during

Fig. 3. TypicalAZ31-H24 alloheat-affected

FSW [42]. Aand Jahazi [and Lim et

Fig. 3(a)section ofweld beadthe oxide. Tshown in Fito the effectshows an eqple that waof DSAWedmacroscopic andmicroscopic structures of a double-sided arcweldedy. (a) Top weld bead surface, (b) cross-section of the welded joint, (c)zone (HAZ), and (d) fusion zone (FZ).

larger grain size in the SZ was also reported by Cao1], Afrin et al. [5], Fairman et al. [40], Pareek et al. [43],al. [44].and (b) shows the top weld bead surface and a cross-

the DSAW weld, respectively. It is seen that the topquality appeared excellent with complete cleaning ofhe bottom weld bead surface was similar to the top. Asg. 3(b), there was a slight sagging of the weld pool dues of gravity on the molten pool during welding. Fig. 3(c)uiaxed microstructure of the HAZ in the DSAWed sam-s completely recrystallized. The grain size in the HAZsample (Fig. 3(c)) was larger than that in the HAZ of

2954 S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961

Fig. 4. EDS lin

the FSWedtemperaturfusion boungrain size iwith (Mge scan showing the compositional variation across the particles in (a) base metal, (b) hea

sample (Fig. 2(c)). This was attributed to the highere experienced in the HAZ of the DSAWed sample. Thedary had a slight hour glass shape (Fig. 3(b)) and then the FZ (Fig. 3(d)) became further larger (15m)17Al12) phase particles arising from a divorced eutec-

tic that formthe solidic

EDS anacles/inclusiFig. 4(a). St-affected zone (HAZ), (c) and (d) fusion zone (FZ) of a DSAWed joint.

ed in the interdendritic and intergranular regions ofation microstructure in the FZ.lysis indicated that some MnAl containing parti-ons were present in the base material, as shown inimilar results were observed by Lin and Chen [45].

S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961 2955

F

Fig. 4(b) showithEDSanMnAl contspite of themicrostructdivorced euboth phaMnAl contboundary, aimage. Thetic -Mg anbe seen in Fdue to the faing, since nlayers of taining onlythe coolingcooling). Asmany -Mglaser/arc hymany spotto be the intermetalljoints. Ben-and reportecored -Mgregions whformed to welding andAZ91 and Amanaban eintergranul

3.2. Microh

A typicaalloy is showgradually frto approximjoints. Thistures in thein the FZ inBM (Fig. 1).from the dethe half-harecrystallizfactors led

ypicaly, FSW

bsered at

nsile

6 shoof that ae strhadultimof abboutratecreasr thel becin raM30h theasesencrfacen inage

rate od Fiserventsubsof t

hat sleadig. 5. Microhardness prole across a DSAWed sample.

ws a typical particle in the heat-affected zone. Line scanalysis signied that theseparticleswere similar to thoseaining particle observed in the base metal (Fig. 4(a)), inremainder of the material experiencing a transition ofure in the HAZ (Fig. 3(c)). Fig. 4(c) revealed clearly atectic structure in the FZ of the DSAWed sample, withse particles on the left-hand side and the unmeltedaining particle on the right-hand side along the grainlthough both particles had the same white color on theeutectic-like structure consisting of alternating eutec-d eutectic -Mg17Al12 along the grain boundary couldig. 4(d). The presence of the eutectic-like structure wasst non-equilibrium cooling of theweld pool afterweld-o normal eutectic structure in the form of alternatingand would be possible in the AZ31 Mg alloy con-3wt% Al based on the equilibrium phase diagram if

rate would be innitely slow (i.e., under equilibriumseen in Fig. 4(c) and (d) and Fig. 3(d) there were also

17Al12 particles within the grains. Liu et al. [46] studiedbrid welding behavior on Mg alloy and reported that precipitates formed within the Mg grain were likely-Mg17Al12 phase. Liu et al. [47] also observed MgAlic brittle phase in the FZ of TIG welded Mg/Al dissimilarHamu et al. [48] studied GTA welded AZ31B Mg alloyd that the fusion zone microstructure consisted of amatrix and a divorced eutectic in the interdendritic

ich were originally Mg17Al12 that subsequently trans-phase (Mg32(Al,Zn)49) intermetallics. Electron beamgas tungsten arc welding had been employed to weld

Z31B Mg alloys, respectively, by Su et al. [49] and Pad-t al. [50], and they observed ne equiaxed grains withar -Mg17Al12 precipitates as well.

Fig. 6. Tbase allo

were ooccurr

3.3. Te

Fig.curvestestedboth thsamplehigherciencyonly astrainUTS inrate foand %Eof stra[30], A

Botof the bthe pretomsuas showSEM imstraintion analso obmovemtion ofsurfaceclear twouldardness

l hardness prole across the DSAWed AZ31B-H24 Mgn in Fig. 5. It is seen that the hardness value decreased

om about HV 70 in the half-hardened H24 temper BMately HV 50 at the center of the FZ of the welded

is due to the formation of non-equilibrium cast struc-FZ (Fig. 3(d)), in conjunction with a larger grain sizecomparison with that in the HAZ (Fig. 3(c)) and in theFurthermore, the grain shape had a signicant changeformed and elongated (or pancake-shaped) grains in

rdened H24 condition (Fig. 1) to the fully annealed ored equiaxed grains in the HAZ (Fig. 3(c)). All of theseto the hardness change shown in Fig. 5. Similar results

a signicanmature failand ductilitand (c)), inwith those

It has re(downwardjoint in anunderstandferent toolbutt joint otools werestir regionshand, additdefect-free,engineering stress versus engineering strain curves of theAZ31B-H24ed and DSAWed samples tested at a strain rate of 1105 s1.

ved for the FSWed joints, where the lowest hardnessthe center of stir zone as well.

properties

ws typical engineering stress versus engineering straine basemetal, FSWed andDSAWedAZ31Mg alloy sheetsstrain rate of 1105 s1. It is seen that after welding,ength and elongation were reduced. While the FSWedhigher yield strength (YS), the DSAWed sample had aate tensile strength (UTS) and elongation. A joint ef-

out 83% was achieved for the DSAWed joints, but it was72% for the FSWed joint. Fig. 7 presents the effect ofon the tensile properties. It was clear that the YS anded and ductility (%El) decreased with increasing strainbase metal, but the effect of strain rate on the YS, UTSame smaller after both types of welding. Similar effectte on the YS and UTS was also reported in Mg9Li1Y[51], cryo-rolled Cu [52] and AZ31B alloys [2,24,53].UTS and%El of theDSAWed joints lay in-between thosemetal and the FSWed joints. The reason behind this wase of a signicant welding defect observed near the bot-of FSWed samples using a right-hand threadedpin tool,Fig. 8. The welding defect could be better seen from the

s taken from a fracture surface after tensile testing at af 1103 s1 shown in Fig. 9(a) at a lower magnica-

g. 9(b) at a higher magnication. Similar defects wereed by Cao and Jahazi [1] who noted that the upwardof the material in the stir zone may cause the forma-urface porosity or even root notches near the bottomhe work-piece when the right-hand pin was used. It isuch defects at the bottom surface in the FSWed jointsto a strong notch effect or stress concentration andhavet inuence on the mechanical properties, causing pre-

ure as shown in Fig. 6. As a consequence, both the UTSy of the FSWed samples were reduced notably (Fig. 7(b)spite of the slightly higher YS (Fig. 7(a)), in comparisonof DSAWed joints.cently been reported that tool proles and axial forceforce)haveasignicanteffecton thedefect-freeFSWed

Al alloy and the subsequent tensile properties [54]. Tothe effect of axial force and tool pin prole, ve dif-

pin proles and three different axial force levels to thef Al6061 Al alloy were examined. Square pin proledobserved to produce defect-free, good quality friction, regardless of the applied axial force levels. On the otherional axial force (7 kN) increased heat input and led togood quality friction stir regions as well, irrespective of

2956 S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961

Fig. 7 , and (

tool pin proright-handeoperation ipin ends dutool. The sufacewerenin the samewelding deFigs. 8 androtated cloc

Fig. 8. An OMspeed of 10min the clockwi

tenth w. Effect of strain rate on (a) yield strength (YS), (b) ultimate tensile strength (UTS)

les [54]. In the present study, a cylindrical tool withd threads was used for the FSWed joints. During FSW

Thestrengt was observed that metal owed up and stuck to thee to the use of right-handed threads in the rotatingbsurface porosity and notch defect at the bottom sur-ot observed in the FSWwhenusing left-handedpin toolsclockwise rotation [5]. This suggests that the observed

fects at the bottom surface of FSWed joints as shown in9 were due to the use of a right-hand threaded pin toolkwise.

micrograph of a FSWed sample near the bottom surface at a weldingm/s and rotational rate of 2000 rpm using a right-hand threaded pinse rotation.

speed (v) twelding dewas still eqwelding spein the preseliterature [effect on threduced dution aroundof these destrain of thinternal toin the gaugsignicantenhanced tmoted the fit is necessa

3.4. Strain-

The harratio of they [24,57].hardening c

Hc =UTS

y

The obtwelded samcapacity wac) ductility of the base metal, FSWed and DSAWed samples.

sile test data are summarized in Fig. 10, where theas presented as a function of the ratio of the welding

o the rotational rate (). Despite the presence of thefects at the bottom surface of the FSWed joints, the UTSuivalent to those reported in the literature at variouseds and tool rotational rates [5,43,44]. The YS obtainednt study was indeed higher than those reported in the

5,43,44]. The fact that the pores did not have a stronge YS but the UTS could be reducedwas simply due to thectility. That is, the sample failed due to strain localiza-the defects before the UTS was reached. The number

fects appeared to have a dominant effect on the truee welded joint at localization. These voids, which werethe material, resulted in a reduced cross-sectional areae area of the material [55]. Some authors also observedinteractions between the surface defect and pores thathe formation of macroscopic shear bands, which pro-ailure [56]. Therefore, to increase the UTS and ductilityry to minimize or avoid the weld defects during FSW.

hardening behavior

dening capacity of a material may be considered as aultimate tensile strength UTS, to the yield strength

Afrin et al. [24] re-dened a normalized parameter ofapacity, Hc, as follows:

y = UTSy

1 (1)

ained hardening capacity of the base metal and theples is listed in Table 1. It is seen that the hardenings enhanced after FSW, similar to the results reported

S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961 2957

Fig. 9. SEM imof 1103 s1defects at a hig

by Afrin et awas furthereffect of thethehardenithat was fuof the mateYS accordindislocationity. A decreresistance b

Fig. 10. A comspeed (v) to thein the literatur

Table 1Hardening capacityof thebase alloy, FSWedandDSAWedsamples testedatdifferentstrain rates.

Specimen Strain rate (s1) Hardening capacity

Base metal

Friction stir

Double-side

reduced thDSAWed sain Figs. 3(dcapacity init should bstrong textverse direc

er, akeneof t

tionthe (d shld stengted asentages of fracture surfaces of a FSWed sample tested at a strain rate. (a) Overall view at a lower magnication, and (b) details of welding

Howevbe weachangediffracalongresolvethe yiesile strincreasthe preher magnication.

l. [24]. The hardening capacity in the DSAWed sampleshigher than that in the FSWed samples. There was littlestrain rate on the hardening capacity. Based on Eq. (1)

ng capacity of amaterialwas related to its yield strengthrther associated with the microstructure and texturerial. An increase in the grain size would decrease theg to the HallPetch relationship [5860] and increasestorage capacity, leading to thehigher hardening capac-ase in the grain size reduced the difference of the owetween the grain boundary and interior, which in turn

parison of the YS and UTS, as a function of the ratio of the weldingrotational rate (), obtained in thepresent studywith those reportede.

To undethe strain-hhardening eharden; theing for a gihave propoexponent. H

= Kn,wheren is tcoefcient,To better qutted their

= y + K1where n1 iscoefcientwhardeningtion using nyielding:

= y + K

where n*,true stress,rial, respecincrement( y) = 1.Fig. 11. Eq.tic deformaonly the datUTS are usein Fig. 11 w1102 0.371103 0.361104 0.361105 0.41

welded sample 1102 0.491103 0.671104 0.661105 0.57

d arc welded sample 1102 0.841103 0.861104 0.841105 0.84

e hardening capacity [57]. Since the grain size of themples was larger than that of FSWed sample, as shown) and 2(e), respectively, the higher value of hardeningthe DSAWed samples would be anticipated. In addition,e noted that for rolled Mg plates there was usually aure with the (0002) plane perpendicular to the trans-tion (TD) and normal direction (ND) of the plate [61].fter FSW and DSAW the texture would be expected tod. Yang et al. [61] reported that after FSW a signicantexture was observed with a high intensity of (0 002)around the TMAZ, and the FSWed joint prone to slip0002) with an orientation of 45 at a lower criticalear stress exhibited an approximately one third drop inrength but only a modest decrease in the ultimate ten-h. This implied that the hardening capacity after FSWs well, in good agreement with the results observed instudy.

rstand strain-hardening behavior it is better to examineardening exponent of different materials. The strain-xponent is a measure of the ability of a metal to strainlarger its magnitude, the greater the strain harden-

ven amount of plastic strain [58]. Several researcherssed different equations to evaluate the strain-hardeningollomon [62] gave the following expression:

(2)

he strain (orwork)hardeningexponent,K is the strength is the true stress and is the true strain [60,6264].antify the strain-hardening response, Chen and Lu [65]tensile curves using the Ludwik equation [64,66]:

n1 (3)

the strain-hardening exponent and K1 is the strengthhich represents the increment in strengthdue to strain

at =1. Afrin et al. [24] proposed the following equa-et ow stress and net plastic strain of materials after( y)n

(4)

, , y and y are the strain-hardening exponent,true strain, yield strength and yield strain of a mate-

tively. K* is the strength coefcient which reects thein strength due to strain hardening corresponding toThe above three equations could be better illustrated in(2) has the origin positioned at O (including the elas-tion stage where the Hookes law holds true), althougha in the uniform deformation stage between the YS andd. Eq. (3) corresponds to a shift of the origin fromO toO1here the yield stress is excluded, while Eq. (4) proposed

2958 S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961

Fig. 11. Schematic illustration of a true stress versus true strain curve.

recently by Afrin et al. [24] represents a further shift of the originfromO1 toO*. It implies thatboth theyield stress andyield strainareprecluded. Fig. 12 presents the evaluated strain-hardening expo-nents (n, n1, n*) as a function of strain rate for the base metal,FSWed and DSAWed samples. Almost no effect of strain rate onthe strain-hardening exponents was seen in the base metal. But forthe DSAWed and FSWed samples, the strain-hardening exponentsincreased with increasing strain rate. The strain-hardening expo-nents evaluated according to all the above three equations wereobviously higher after welding, and the DSAW resulted in some-what higher strain-hardening exponents than the FSW, as shownin Fig. 12. It is also seen that the n values were the smallest and n1values were the highest with n* lying in-between the two. Similarresults were also reported by Afrin et al. [24].

One of themost important contributions to the strain hardeningis related to the formation and multiplication of dislocations. In theplastic deformation stage, the net ow stress due to dislocationdensity could be expressed as [24,33,34]:

y (5)where is the dislocation density. The net ow stress necessaryto continue deformation of a material is proportional to the squareroot of the dislocation density. The dislocation density in a metalincreases with deformation or cold work due to dislocation multi-plication or the formation of new dislocations which decreases thespacing among dislocations and their interactions become repul-sive. The net result would be that the motion of a dislocation isimpeded by other dislocations. As the dislocation density increases,the resistance to dislocation motion by other dislocations becomesmore pronounced. Thus, a higher stress is necessary to deform ametal [58].

Fig. 13 shows a typical KocksMecking plot of strain-hardeningrate ( =d/d) versus net ow stress ( y) at different strainrates from 1102 s1 to 1104 s1 for the base metal. It isseen that no stage I hardening (or easy glide) which dependsstrongly on the orientation of the crystal and stage II linear harden-ing where the strain-hardening rate should be constant occurred.Srinivasan and Stoebe [67] reported that the presence of stage IIstrain hardening could be due to the interactions of the disloca-tions in the primary slip system with those in an intersecting slipsystem. Stage III hardening is characterized by a hardening ratethat decreases monotonically with increasing ow stress leadingto the much repeated term parabolic hardening on the stressstrain curve [68]. This stage is very sensitive to the temperatureand rate of deformation [33]. In this present study test temper-ature remained constant but strain rates were changed. As seenfrom Fig. 13, stage III was somewhat strain-rate sensitive, i.e., theFig. 12. Effect of strain rate on (a) n-value, (b) n1-value and (c) n*-value in the base metal, FSWed and DSAWed samples.

S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961 2959

Fig. 13. strain-hardening rate () versus net ow stress ( y) of the base metaltested at different strain rates.

strain-hardening rate increased with increasing strain rate. Simi-lar result was observed by Lin and Chen [2] in AZ31B extruded Mgalloy as well. This could be explained on the basis of the followingequation [2

= 0 Rd1

where isq is an expand is rathetemperaturthe presentremained chardeningstress.

As stageappeared. Dto track thewith geom[33]. Howevextended toto the disorhardening rstage IV instrain-hard

Fig. 14 srate versussamples tesnet ow stthe base mestrainsornecouldbeun

Fig. 14. strainFSWed and DS

dislocation strain hardening [24,34,35]:

= 0 + HP + d, (7)

where 0 is the frictional contribution, HP = kd1/2 is theHallPetch contribution and d =MGb1/2 is the Taylor disloca-tion contribution (where G is the shear modulus, b is the Burgersvector, M is the Taylor factor and is constant). Sinclair et al. [70]and Kovacs et al. [71] reported that at lower strains the grain sizehad a strong contribution to the strain hardening and the inu-ence of the grain size on the strain hardening vanished at higherstrains due todislocation screening anddynamic recovery effects atgrain boundaries. Since the basemetal had a smaller grain size cou-pled with its deformed grain structure, the HallPetch contribution(HP) would be stronger at lower strains, leading to a higher strain-hardening rate in comparison with the welded joints. On the otherhand, at higher strains or later stage of deformation strain harden-ing was higher in the welded joints. This could be due to the largergrain size and the presence of particles in the fusion zone of theDSAWed samples. It was stated that the presence of precipitatesincreased the dislocation density [72] and larger grain size pro-

ore space to accommodate dislocations [24]. As the grainthe Das obetalhereAZ3

ce ofiled psamthermatlm [lowin

() +

Linde ploLindbasepectich ww stonshws:

= k2

d,33]:

/q

(6)

the strain-hardening rate in stage III, 0 is a constant,erimental stress exponent varying with temperaturesr large namely about n=8 at higher and n=35 at loweres, Rd is also a temperature dependent parameter. Inexperiments Rd and q were constant as temperatureonstant. It is evident from Eq. (6) that the strain-rate increased with increasing strain rate at a given

III approached a saturation level, hardening stage IVue to the low hardening level of stage IV, it is difculthardening mechanisms because they must compete

etrical and structural instabilities including damageer, Pantleon [69] reported that stage III hardening wasstage IV by incorporation of excess dislocations relatedientations. He further stated that in stage IV the workate often remains constant. It is seen from Fig. 14 thatthe base metal remained almost constant with a lowening rate.hows a typical KocksMecking plot of strain-hardeningnet ow stress in the base metal, FSWed and DSAWedted at a strain rate of 1102 s1. At lower strains orress levels after yielding the strain-hardening rate oftal was higher than that of the welded joints. At highertowstress levels itwas reversed. The strainhardeningderstoodon thebasis of thegrain size strengtheningand

vided msize ofrate wbase mtions wsizes inpresenples faFSWed

Anoior ofLindhothe fol

= 0

ThemL. Thwherefor the3.5, resing, whand oa relatias follo

ln -hardening rate () versus net ow stress ( y) of the base metal,AWed samples tested at a strain rate of 1102 s1.

Fig. 15. A typivalue of 2.5% vsamples.SAWed samples was larger, a strong strain-hardeningserved after DSAW in comparison with the FSW and the(Fig. 14). This is also in agreement with other observa-the hardening rate decreasing with decreasing grain

1 alloy was presented [24,34,73]. However, due to theweld defects at the bottom surface, the FSWed sam-rematurely, thus the net ow stress up to failure in theple was the lowest (Fig. 14).important parameter involving thedeformationbehav-erials was the strain-rate sensitivity (SRS), m. The74] approach was used to evaluate the SRS based ong equation:

1() log (8)

holm SRS is 1() in the equation commonly termed ast of versus log at 2.5% true strain is shown in Fig. 15,holm SRS is represented by the slope, mL. The mL valuesmetal, DSAWed joins and FSWed joints were 7, 4 and

ively. It follows that the SRS became lower after weld-ould be related to the difference in the microstructure

ress. del Valle and Ruano [35] have recently presentedip between stress, SRS and grain size for the AZ31 alloy

1/2(Mc 2Mcg) + Mcg, (9)

cal plot used to evaluate the strain-rate sensitivity,mL , at a true strainia the Lindholms approach for the base metal, DSAWed and FSWed

2960 S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961

Fig. 16. Typicstrain rate of 1

where k isstress, Mcg =resents a ligrain size iwith decreasize and higwith that ofreported by[75].

3.5. Fractog

Fig. 16base metal,1103 s1fracture apdenoted du

sions [59]. Some inclusions were present on the fracture surfaceof AZ31B rolled magnesium alloy. Similar fracture surface char-acteristics were reported in [2,5]. As shown in Fig. 16(b), somecleavage-like at facets in conjunction with dimples and river

g coby t

lograg led

thesam

he aret a

at thAWeets whe sps an

clusmarkincausedcrystalnectinwell toFSWedfrom tby Limdefectthe DSlike facfrom tsample

4. Conal SEM images showing the fracture surfaces of samples tested at a103 s1. (a) Basemetal, (b) FSWedsample, and (c)DSAWedsample.

Boltzmann constant, d is the grain size, is the ow lncg/ ln and Mc = ln c/ ln . This equation rep-near correlation between the SRS, ow stress andn the form of d1/2. It means that the SRS increasedsing grain size. Since the base metal had a smaller grainher ow stress, its SRS would be higher in comparisonthe welded joints. This is in agreement with the resultsdel Valle and Ruano [35] and Prasad and Armstrong

raphy

shows typical images of the fracture surfaces of theFSWed and DSAWed samples tested at a strain rate of. It is seen from Fig. 16(a) that dimple-like elongatedpeared more apparent. This type of fracture surfacectile fracture thatwas characterized by cup-like depres-

1. The micrsmall elowhich wFSW resuSZ and TMDSAW, fsize werture conintergran

2. While thFSWedsaprematuthe FSWethe causUTS in thsamplesliteraturspeed lo

3. The straobservedhardeninThe straevaluatetimes hig

4. A higherwas obsmanydiswith strograin sizwelded stallizatiowith p

5. The YS,rate incrstrain rabase mein compa

Acknowled

The autEngineeringwork of CeninvestigatioResearch Pruld be seen in the FSWed sample. The rivermarkingwashe crack moving through the grain along a number ofphic planes which formed a series of plateaus and con-ges [59]. The fractographic observations correspondedrelatively low percentage elongation of 4% in the

ples.While tensile fracture initiation could have startedea between the weld nugget and the TMAZ as reportedl. [44], the crack initiation occurred from the weldinge bottom surface in the present study (Figs. 8 and 9). Ind sample, more dimples together with fewer cleavage-ere observed, as shown in Fig. 16(c). The crack initiatedecimen surface or near surface defects in the DSAWedd in the base metal as well.

ions

ostructure of the as-received AZ31B-H24 consisted ofngated grains with some MnAl containing inclusionsere still present in different zones after welding. Thelted in recrystallized and relatively small grains in theAZ, and partially recrystallized grains in theHAZ. After

ully recrystallized grains with a relatively large graine observed in the HAZ, and the divorced eutectic struc-taining -Mg17Al12 particles in the interdendritic andular regions appeared in the fusion zone.e YS was higher, the UTS and ductility were lower in themples than in theDSAWedsamples. Thiswasdue to there fracture caused by the presence of welding defects ind samples at the bottom surface. The defects were also

e for the small net ow stress from yielding up to thee FSWed samples. However, the strength of the FSWedwas still similar to or higher than those reported in thee due to the smaller grain sizes arising from the highw temperature FSW.in-hardening capacity of the DSAWed samples wasto be twice that of the base metal, with the strain-

g capacity of FSWed samples lying in-between them.in-hardening exponents after both types of welding,d via three different approaches, were all about twoher than those of base metal.strain-hardening rate of the base metal at lower strainserved due to smaller and pre-deformed grains wherelocationshadbeengenerated in thebasemetal, couplednger HallPetch contribution stemming from smalleres. At higher strains, the strain-hardening rate of theamples became higher due to the occurrence of recrys-n in the FSWed samples, and the larger grains togetherarticles in the fusion zone in the DSAWed samples.UTS, strain-hardening exponent and strain-hardeningeased slightly, and ductility decreased with increasingte. Stronger strain-rate sensitivity was observed in thetal due to the smaller grain size and higher ow stressrison with the welded joints.

gements

hors would like to thank the Natural Sciences andResearch Council of Canada (NSERC) and AUTO21 Net-ters of Excellence for providing nancial support. Thisn involves part of CanadaChinaUSA Collaborativeoject on the Magnesium Front End Research and Devel-

S.M. Chowdhury et al. / Materials Science and Engineering A 527 (2010) 29512961 2961

opment (MFERD). The authors also thank General Motors Researchand Development Center for the supply of test materials. One ofthe authors (D.L. Chen) is grateful for the nancial support by thePremiers Research Excellence Award (PREA), Canada Foundationfor Innovation (CFI), and Ryerson Research Chair (RRC) program.The authors would like to thank Q. Li, A. Machin, J. Amankrah, D.Ostrom and R. Churaman for their assistance in the experiments.The authors also thank Dr. S. Xu, Dr. K. Sadayappan, Dr. J. Jackman,Professor N. Atalla, Professor S. Lambert, Professor H. Jahed, Profes-sor Y.S. Yang, Professor B. Jordon, Dr. A.A. Luo, Mr. R. Osborne, Dr.X.M. Su, and Mr. L. Zhang for helpful discussion.

References

[1] X. Cao, M. Jahazi, Mater. Des. 30 (2009) 20332042.[2] X.Z. Lin, D.L. Chen, J. Mater. Eng. Perform. 17 (2008) 894901.[3] A. Tharumarajah, P. Koltun, J. Clean. Prod. 15 (2007) 10071013.[4] W.M. Thomas, E.D. Nicholas, J.C. Needham, M.G. Church, P. Templesmith, C.J.

Dawes, GB Patent Application No. 9125978.9 (December 1991).[5] N. Afrin, D.L. Chen, X. Cao, M. Jahazi, Mater. Sci. Eng. A 472 (2008) 179186.[6] H.W. Lee, T.S. Lui, L.H. Chen, J. Alloys Compd. 475 (2009) 139144.[7] H. Watanabe, H. Tsutsui, T. Mukai, K. Ishikawa, Y. Okanda, M. Kohzu, K. Higashi,

Mater. Trans. A 42 (2001) 12001205.[8] T.Mukai,M. Yamanoi, H.Watanabe, H. Higashi, ScriptaMater. 45 (2001) 8994.[9] A. Bussiba, A.B. Artzy, A. Shtechman, S. Ifergan, M. Kupiec, Mater. Sci. Eng. A

302 (2001[10] ASM Met

1990 ASM[11] Y.M. Zha

torches, U[12] Y.M. Zhan[13] Y.M. Zhan[14] Y.M. Zhan[15] Y.M. Zhan[16] Y. Kwon,[17] J.A. Moult[18] Y.J. Quan,[19] L. Liu, C. D[20] J. Zhu, L. L[21] W. Woo,

1701171[22] X.H. Wan[23] L. Commi[24] N. Afrin, D[25] P.B. Srini

389396.[26] L. Yu, K. N[27] R.S. Mish[28] R.Nandan[29] S.H. Park,[30] H. Takud

2266227[31] L. Yu, K. N[32] W. Lee, K.

Int. J. Plas[33] U.F. Kock[34] J.A. del Va[35] J.A. del Va

[36] J. Shen, G.Q. You, S.Y. Long, F.S. Pan, Mater. Charact. 59 (2008) 10591065.[37] ASM Handbook, 10th Edition, vol. 2, Properties and Selection: Nonferrous

Alloys and Special Purpose Materials, 1990, ASM International, Materials Park,OH.

[38] G.F. Vander Voort, Metallography Principles and Practice, ASM International,Materials Park, OH, USA, 1999.

[39] ASTM Standard E8/E8M, 2008, Standard Test Methods for Tension Test-ing of Metallic Materials, ASTM International, West Conshohohocken, PA,2008.

[40] M. Fairman, N. Afrin, D.L. Chen, X.J. Cao, M. Jahazi, Can. Metall. Quart. 46 (2007)425432.

[41] W.B. Lee, Y.M. Yeon, S.K. Kim, Y.J. Kim, S.B. Jung, in:H.I. Kaplan (Ed.),MagnesiumTechnology 2002, TMS, 2002, pp. 309312.

[42] J.A. Esparza, W.C. Davis, E.A. Trillo, L.E. Murr, J. Mater. Sci. Lett. 21 (2002)917920.

[43] M. Pareek, A. Polar, F. Rumiche, J.E. Indacochea, J.Mater. Eng. Perform. 16 (2007)655662.

[44] S. Lim, S. Kim, C.G. Lee, C.D. Yim, S.J. Kim, Metall. Mater. Trans. A 36 (2005)16091612.

[45] X.Z. Lin, D.L. Chen, Mater. Sci. Eng. A 496 (2008) 106113.[46] L.M. Liu, G. Song, M.L. Zhu, Metall. Mater. Trans. A 39 (2008) 17021711.[47] P. Liu, Y.J. Li, H.R. Geng, J.A. Wang, Mater. Lett. 61 (2007) 12881291.[48] G. Ben-Hamu, D. Eliezer, C.E. Cross, Th. Bllinghaus, Mater. Sci. Eng. A 452453

(2007) 210218.[49] S.F. Su, J.C. Huang,H.K. Lin, N.J. Ho,Metall.Mater. Trans. A 33 (2002) 14611473.[50] G. Padmanaban, V. Balasubramanian, J.K. Sarin Sundar, JMPEG, in press.[51] L. Jiang, J.J. Jonas, A.A. Luo, A.K. Sachdev, S. Godet, in: A.A. Luo, N.R. Neelameg-

gham, R.S. Beals (Eds.), Magnesium Technology, TMS, 2006, pp. 233238.[52] Y.M. Wang, E. Ma, Mater. Sci. Eng. A 375377 (2004) 4652.

ivescuegghalango(2008. Dav

ns. A 3Bayleo, Z. M. Cal

tion, J. Diete. HosYork

ang, BHollo. HertEditioatke,earch4274. Chenudwilin, 19Sriniv. RollePantle. Sinc

ovacs,del Va. Bar351. Lindh.R.K. P) 5662.als Handbook 10th Edition, vol. 6, Welding, Brazing, and Soldering,International, Materials Park, OH.

ng, S.B. Zhang, Method of arc welding using dual serial opposedS Patent No. 5990446 (November 1999).g, C. Pan, A.T. Male, Metall. Mater. Trans. A 31 (2000) 25372543.g, C. Pan, A.T. Male, J. Mater. Sci. Lett. 19 (2000) 831833.g, S.B. Zhang, Weld. J. 78 (1999) 202206.g, S.B. Zhang, Weld. J. 77 (1998) 5761.D.C. Weckman, Sci. Technol. Weld. Join. 13 (2008) 485495.on, D.C. Weckman, Weld. J. 89 (2010) 11s23s.Z.H. Chen, X.S. Gong, Z.H. Yu, Mater. Charact. 59 (2008) 14911497.ong, Mater. Lett. 60 (2006) 21942197.i, Z. Liu, Appl. Surf. Sci. 247 (2005) 300306.H. Choo, M.B. Prime, Z. Feng, B. Clausen, Acta Mater. 56 (2008)1.g, K.S. Wang, Mater. Sci. Eng. A 431 (2006) 114117.n,M.Dumont, J.-E.Masse, L. Barrallier, ActaMater. 57 (2009) 326334..L. Chen, X. Cao, M. Jahazi, Scripta Mater. 57 (2007) 10041007.

vasan, R. Zettler, C. Blawert, W. Dietzel, Mater. Charact. 60 (2009)

akata, N. Yamamoto, J. Liao, Mater. Lett. 63 (2009) 870872.ra, Z.Y. Ma, Mater. Sci. Eng. R 50 (2005) 178., T. DebRoy,H.K.D.H. Bhadeshia, Prog.Mater. Sci. 53 (2008) 9801023.Y.S. Sato, H. Kokawa, Scripta Mater. 49 (2003) 161166.a, S. Kikuchi, N. Yoshida, H. Okahara, Mater. Trans. A 44 (2003)0.akataa, J. Liao, J. Alloys Compd. 480 (2009) 340346.-H. Chung, D. Kim, J. Kim, C. Kim, K. Okamoto, R.H.Wagoner, K. Chung,t. 25 (2009) 16261654.s, H. Mecking, Prog. Mater. Sci. 48 (2003) 171273.lle, F. Carreno, O.A. Ruano, Acta Mater. 54 (2006) 42474259.lle, O.A. Ruano, Scripta Mater. 55 (2006) 775778.

[53] V. Llam

[54] K. E38

[55] R.WTra

[56] C.J.[57] J. Lu[58] W.D

Edi[59] G.E[60] W.F

New[61] J. Y[62] J.H.[63] R.W

4th[64] L. R

Respp.

[65] X.H[66] P. L

Ber[67] M.[68] A.D[69] W.[70] C.W[71] I. K[72] J.A.[73] M.R

509[74] U.S[75] Y.V, C.M. Cady, E.K. Cerreta, B.L. Henrie, G.T. Gray II, in: A.A. Luo, N.R. Nee-m, R.S. Beals (Eds.), Magnesium Technology, TMS, 2006, pp. 153158.van, V. Balasubramanian, M. Valliappan, Int. J. Adv. Manuf. Technol.) 285295.ies, M.T. Smith, H.E. Oliver, M.A. Khaleel, S.G. Pitman, Metall. Mater.1 (2000) 27552763.

y, A.K. Pilkey, Mater. Sci. Eng. A 403 (2005) 110.ei, W. Tian, Z. Wang, Mater. Sci. Eng. A 441 (2006) 282290.

lister Jr., Materials Science and Engineering An Introduction, 8thohn Wiley & Sons, Inc., New York, USA, 2009.r, Mechanical Metallurgy, 3rd Edition, McGraw-Hill, MA, USA, 1986.ford, Mechanical Behavior of Materials, Cambridge University Press,, USA, 2005..L. Xiao, D. Wang, Z.Y. Ma, Mater. Sci. Eng. A 527 (2010) 708714.mon, Trans. AIME 162 (1945) 268289.zberg, Deformation and FractureMechanics of EngineeringMaterials,n, John Wiley & Sons, New York, USA, 1996.P.I. Welch, 13th Biennial Congress International Deep DrawingGroup: Efciency in SheetMetal Forming,Melbourne, Australia, 1984,35., L. Lu, Scripta Mater. 57 (2007) 133136.k, Elemente der Technologischen Mechanik, Springer-Verlag OHG,09, p. 32.asan, T.G. Stoebe, J. Mater. Sci. 9 (1974) 121128.tt, U.F. Kocks, Sol. Stat. Phenom. 3536 (1994) 118.on, Mater. Sci. Eng. A 387389 (2004) 257261.lair, W.J. Poole, Y. Brechet, Scripta Mater. 55 (2006) 739742.N.Q. Chinh, E. Kovacs-Csetenyi, Phys. Status Solidi A 194 (2002) 318.lle, A.C. Picasso, R. Romero, Acta Mater. 51 (2003) 64436452.

nett, Z. Keshavarz, A.G. Beer, D. Atwell, Acta Mater. 52 (2004)03.olm, J. Mech. Phys. Solids 12 (1964) 317335.rasad, R.W. Armstrong, Philos. Mag. 29 (1974) 14211425.

Tensile properties and strain-hardening behavior of double-sided arc welded and friction stir welded AZ31B magnesium alloyIntroductionMaterials and experimental procedureResults and discussionMicrostructureMicrohardnessTensile propertiesStrain-hardening behaviorFractography

ConclusionsAcknowledgementsReferences