springer kostka11

Materials Science and Engineering A 528 (2011) 46304642

Contents lists available at ScienceDirect

Materials Science and Engineering A

journa l homepage: www.e lsev ier .co

Inuen ll-proper um

H. Springa Max-Planck-Ib Helmholtz-Ze ning P

a r t i c l

Article history:Received 10 DReceived in reAccepted 17 FAvailable onlin

Keywords:Friction stir wAluminiumSteelInterdiffusionIntermetallic p

n layetweerfacfter aers o

ineticts wilayepha

oceduyers.

1. Introduction

Dissimilar joining of iron (Fe) to aluminium (Al) alloys is ofeminent tectial engineestrength, gobe combinegood corroApplicationubiquitousin transportection in thfor example

Joiningtheirmechaof the formphase diagr (Al2Fe)Fe-rich phatant role fopresent in sfound, exhib

CorresponMikrostrukturGermany. Fax:

E-mail add

ity ranges. However, not all of those equilibriumphases necessarilyform during joining or subsequent heat treatments [14,15].

Most previous experiments concerning interdiffusion between

0921-5093/$ doi:10.1016/j.hnical interest, as it allows the use of these two essen-ring materials in the same design [1,2]. Thus, the highod creep resistance, and formability of steels [3,4] mayd with the low density, high thermal conductivity andsion resistance of Al alloys [5,6] in one hybrid part.s for such dissimilar joints between Fe and Al alloys areand range from an improved strength-to-weight ratiotation systems [2] to high temperature corrosion pro-e chemical industry [7] or enhanced cooling efciencyin cryogenic technology [8].

of Fe and Al alloys is impeded by the differences innical properties andmelting temperatures andbecauseation of brittle intermetallic phases [9]. The AlFeam [10,11] shows three Al-rich intermetallic phases, (Al5Fe2) and (Al13Fe4) and the two orderedses (AlFe) and (AlFe3). Silicon (Si) plays an impor-r the reactions and intermetallic phases when it isteel/Al alloy joints [12,13]. 11 ternary phases have beeniting relatively largeandcloselypositionedhomogene-

ding author at: Max-Planck-Institut fr Eisenforschung GmbH,physik und Umformtechnik, Max-Planck-Str. 1, D-40237 Dsseldorf,+49 211 6792 333.ress: [email protected] (A. Kostka).

Fe or steel and Al alloys are concerned with phase identicationand growth kinetics. The phase was observed as the dominantreaction product at the interface between solid Fe and Al melts[1618] with an additionally detected thin layer of between and the solidied Al melt via selected area diffraction (SAD) in thetransmission electronmicroscope (TEM) [19]. Interdiffusion exper-imentswhere bothAl and Fe alloys remain in the solid state showedthat thephase retains as themajor fraction of the developed reac-tion layer with a temperature dependent, parabolic growth rate[15,20,21]. The addition of Si to Al has been found to not onlychange the composition of the reaction layer but also to have apronounced, temperature dependent effect on its growth kinet-ics compared to respective experiments with pure Al: numerousresearchers described a deceleration of the reaction layer growth ininterdiffusion experiments with Si containing Al melts [17,2226].However, an accelerated growth was observed for interdiffusionbetween low carbon (C) steel and an AlSi alloy at 600 C, belowthe melting temperature of Al [15].

The inuence of intermetallic phases at the interface betweensteel and Al alloys on the mechanical properties of joints has beeninvestigated for both solid state processes [2729] and solid/liquidjoining procedures, where the Al base material is molten (welded)and the Fe or steel side of the joint remains solid (brazed)[3032]. Achar et al. [30] studied the tensile strength of arc-welded

see front matter 2011 Elsevier B.V. All rights reserved.msea.2011.02.057ce of intermetallic phases and Kirkendaties of joints between steel and alumini

era, A. Kostkaa,, J.F. dos Santosb, D. Raabea

nstitut fr Eisenforschung GmbH, 40237 Dsseldorf, Germanyntrum Geesthacht, Institute of Materials Research, Materials Mechanics, Solid-State Joi

e i n f o

ecember 2010vised form 16 February 2011ebruary 2011e 24 February 2011

elding

hases

a b s t r a c t

The formation of intermetallic reactiotigated in friction stir welded joints bSi. Characterisation of the steel/Al intsamples in the as-welded state and awas performed to obtain reaction layreaction layers grew with parabolic kannealing at 450 C and above. In joinof Kirkendall-porosity at the reactionis controlled by the thickness of the induced by the friction stirwelding prand growth kinetics of the reaction lam/locate /msea

porosity on the mechanicalalloys

rocesses, 21502 Geesthacht, Germany

ers and their inuence on mechanical properties was inves-en a low C steel and both pure Al (99.5wt.%) and Al5wt.%e, tensile tests and fractography analysis were performed onnnealing in the range of 200600 C for 964min. Annealingf distinct thickness and composition. For both Al alloys, thes with the phase (Al5Fe2) as the dominant component afterth pure Al, the tensile strength is governed by the formationr/Al interface. The tensile strength of joints with Al5wt.% Sise (Al5Fe2) layer. The pre-deformation of the base materials,re, was found to have a pronounced effect on the composition

2011 Elsevier B.V. All rights reserved.

H. Springer et al. / Materials Science and Engineering A 528 (2011) 46304642 4631

(solid/liquid) joints after annealing at 150450 C for 1100h.Withgrowing intermetallic layer thickness caused by the heat treat-ment, strongly decreasing joint strength was observed and thesamples failed in the steel/Al interface region rather than the Albase materiplex build-uAgudo et al(EBSD) in tcal TEM tosolid/liquidprocess. It win reasonabiments [31]fabricated bmuch thinntions.Mechfor 80h at 3but they fraallel decreato nucleatiomade to idewhich the ring paramewas claimeenergy disption betweedetected [2

While thmechanicalical thickne10m [30mechanism

2. Objectiv

The objetal mechanmechanicalalloys. Usinwelding (FSintermetallsystematicatesting andtural respotreatmentwere investstrongly pre

3. Materia

Steel samlow C steelcut from 2m99.5wt.% pusheetswere

3.1. Produc

The spewelded alodirection oHelmholtz-parameters5mms1; a13mmand

llustrations of the sample production process: (a) sketch of the friction-stir-basematerials and thedimension and locations of samples for tensile testingd interface characterisation (right). (b) Temperature/time matrix showingrent annealing treatments applied in this study.

6.5mm. The Al samples were placed at the retreating sidee steel samples on the advancing side of the rotating tool,the bottom edge of the pin touched the bottom line of theample.m the bonded plates, samples for tensile testing and cross-s for microstructural analysis were cut by spark erosiona)). For both types of joints, the samples were cut perpen-to the welding direction and the steel/Al interfaces were

ned in the middle of the specimens. The cut samples werennealed (air atmosphere) in the range of 200600 C forin. Fig. 1(b) displays the different time (t)/temperature (T)s applied in this study. Both Al alloys remain solid in thetemperature range applied here; only at 600 C the AlSi5onsists of about 50% solid phase and 50% melt, accordingbinary AlSi phase diagram [33]. Despite being in this semi-wophase state, theAlSi5 samples remain sufciently viscousC. Heating of the samples from room temperature to exper-l temperatures took about 4min; cooling was performede the furnace after the thermal exposure was complete.al. The developed reaction layers exhibited a very com-p due to a bronze buffer layer on the steel sheet [30].. [32] applied electron backscatter diffraction analysishe scanning electron microscope (SEM) and analyti-characterise the intermetallic phase layers formed injoints produced by a gas shielded metal arc weldingas shown that the build-up of the reaction layer was

le agreement to those formed in interdiffusion exper-. Other researchers [27,28] investigated steel/Al jointsy solid-state friction welding, which typically results iner reaction layers than solid/liquid interaction condi-sner andKlock [27] found that tensile samples annealed50 C and above no longer failed in the Al base materialcturedbrittle in the interface region.Alongwith thepar-sing strength of the joints, this behaviour was ascribedn and growth of intermetallic phases, but no effort wasntify them[27]. Yilmazet al. [28] tested steel/Al joints ineaction layer thickness was varied via differing weld-ters. The intermetallic phase present at the interfaced to be Al3Fe (corresponds to the phase) based onersive X-ray (EDX) measurements, but no clear rela-n reaction layer thickness and joint strength could be8].e detrimental effect of growing reaction layers on theproperties of steel/Al joints is well-known and the crit-ss for intermetallic reaction layers was found to be], only little information exists on the microstructurals governing this embrittlement-process.

e

ctive of the present work is to elicit the fundamen-isms how intermetallic reaction layers inuence theproperties of dissimilar joints between steel and Alg samples of steel sheets pre-bonded by friction stirW) to two different Al alloys (with and without Si),

ic layers of distinct thickness and composition werelly produced by heat treatments, followed by tensileinterface characterisation.Additionally, themicrostruc-nse of the friction-stirred base materials to the heatand the growth kinetics of the intermetallic layersigated in these interdiffusion experiments startingwith-deformed base materials.

ls and methods

ples were cut from 2mm thick sheets of unalloyed(0.12wt.% C; European grade: DC01). Al samples werem thick sheets of either commercially available Al ofrity (Al99.5) or Al containing 5wt.% Si (AlSi5). TheAlSi5hot rolled at 450 C to 2mmthickness from cast blocks.

tion of samples

cimen pairs (Al/steel, AlSi/steel) were friction stirng their 300mm base line (parallel to the rollingf the sheets) in a butt joint conguration at theZentrum Geesthacht, Germany. The following processwere applied: rotation speed, 400 rpm; travel speed,xial force, 10kN. The tool had a shoulder diameter ofa non-threaded, 2mm long conical pinwith a top diam-

Fig. 1. Iwelded(left) anthe diffe

eter ofand thwhilesteel s

Frosection(Fig. 1(dicularpositiothen a964msettingentirealloy cto thesolid, tat 600imentaoutsid

4632 H. Springer et al. / Materials Science and Engineering A 528 (2011) 46304642

Fig. 2. Overview of the interfacial region between steel (right) and Al alloy (left) asa montage of SEM micrographs.

3.2. Characterisation

After annealing, the cross-sectional areas were prepared bygrinding and polishing with standard metallographic techniques.The cross-sections were investigated using optical microscopy(OM), SEM, and TEM. Thicknesses of the reaction layers >1mwere measured using a Leica DM4000M OM, which was equippedwith an Image Access 9 image analysis system. Average values ofreaction layer thicknesses were determined following same pro-cedures as in [15,21,34]: the cross-sectional area of the reactionlayer was measured over a distance of 150m; dividing this areaby the length of the base line yielded the average thickness. Forlayers


Fig. 4. Examp(d) EDX line scred).

face betweein Fig. 2) is ainvestigatioin this regiocan be obsearrows in Fthe Al alloy(not shownfailed to ful

4.1.1. InterfIn the as

intermetallthe interfacwere prepashows a STEside) and ahighlightedreaction layTransformaon the righrespondingles for SEManalysis of the reaction layers formed between steel andAl99.5: (ac) SEMmicran result obtained along the red arrow in (c). (e) Colour coded EBSD phase map after an

n steel and Al alloy (highlighted by a black dashed ovalngleddue to the conical shape of the tool pin. All furtherns concerning the interfacial reactions were conductedn of the samples. Additionally, detached steel particlesrved in theAl side of the cross-section (marked bywhiteig. 2) as a consequence of the complex material ow ofduring FSW [38]. Pores in the stir zone of the Al alloyhere) rarely occurred on an irregular basis if the Al alloyly remerge after the tool had passed through.

ace in joints with Al99.5welded state and in specimens annealed up to 400 C,

ic reaction layers could not be observed in the SEM ate between Al99.5 and steel. Therefore, TEM samplesred from corresponding cross-sections (Fig. 3). Fig. 3(a)MBF picture of the interface in the as-welded state (lefthigh resolution TEM micrograph (right side) of the areaon the left by a white dashed square. No intermetallicers can be observed at the interface. The Fast-Fourier-tion of the high resolution pictureshown as an insertt image in Fig. 3(a) exhibits diffraction reexes cor-only to the base materials. Fig. 3(b) displays STEM BF

micrographcontinuousbetweenAlimage on thin the left iiments in tlayer consis

After aninterface poand selectemicrograph500 and 600of reactionhighlightedsteel (downlower layerincreasingtion layer greaction layface. Additiadjacent baface of the uographs after annealing for 9min at 450, 500 and 600 C, respectively.nealing at 450 C for 64min (Al green, blue, yellow, steel

s of the interface after annealing at 400 C for 9min. Areaction layer of about 50nmthickness canbeobservedand steel and is highlighted bywhite dashed lines in thee right, which corresponds to the white dashed squaremage taken at lower magnication. Diffraction exper-he TEM (not shown here) suggested that the reactionts of the phase.nealing at 450 C and above, the reaction layers at thessessed a sufcient thickness for SEM analysis (>2m)d results are compiled in Fig. 4. Fig. 4(ac) shows SEMs of the reaction zones after annealing for 9min at 450,C, respectively, at different magnications. Two types

layers with differing grey level can be observed and arebywhitedashed lines, namely, a lower layer adjacent to/right) and an upper layer adjacent to Al (top/left). Thetakes up the larger fraction of the reaction layer. With

annealing temperature the total thickness of the reac-rows from 1.9m at 450 C to 13.6m at 600 C. Theers are continuous but vary in thickness along the inter-onally, the morphology of the interfaces towards thesematerials changeswith temperature:while the inter-pper layer towards Al remains irregularly fringed on a


Fig. 5. STEM Hzone between

ner scale,and steel dedescribed inline scan recolour codefor 64min areaction layphase (coloto steel is mand red in F

In ordertion layers awere prepaFig. 5 showtions from thighlighted(not shownobtained bypores can bAl and phAADF images taken at different magnications, showing the reactionsteel and Al99.5 after annealing at 500 C for 9min.

the smooth interface at 450 C between the lower layervelops larger scale wavy features at 500 and 600 C, asprevious interdiffusion experiments [15,21]. The EDX

sult (Fig. 4(d)) along the red arrow in Fig. 4(c) and thed EBSD phase map (Fig. 4(e)) after annealing at 450 Cllow to identify the two types of phases present in theers. The upper layer adjacent to Al consists of the ured blue in Fig. 4(e)), whereas the lower layer adjacentade up of phase (yellow). Al and steel appear greenig. 4(e), respectively.to conrm the SEM results and to investigate the reac-nd their interfaceswithhigher resolution, TEMsamplesred from a specimen annealed at 500 C for 9min.s STEM HAADF micrographs of increasing magnica-op to bottom; the corresponding areas of the images areby white dashed rectangles. Diffraction experimentshere) conrmed the phase sequence of AlsteelSEM measurements (Fig. 4). A seam of interconnectede observed at large fractions of the interface betweenase, as marked by the white arrow in Fig. 5(b). Fig. 5(c)

Fig. 6. TEM inshowing STEMregions (right500 C for 9m

shows poreto the partdark featurand phasnano-sized

4.1.2. InterIn samp

to 400 C,characteristigations restate and dabout 30nm

In contrat 450 C foreaction laytion using Tvestigation of the interface region in joints between steel and AlSi5,HAADF micrographs (left side) and SAD patterns of single phase

side): (a) after annealing at 450 C for 64min. (b) After annealing atin.

s of about 200nm diameter in the phase region closely disconnected Al/ interface. Additionally, a trail ofes leading to a pore through both phase (right side)e (left side) can be observed; which seem to consist ofvoids.

face in joints with AlSi5les in the as-welded state and after annealing at upthe interface between AlSi5 and steel shows similartics as in sampleswithAl99.5 (Section4.1.1): TEM inves-vealed the absence of reaction layers in the as-weldediscontinuous islands of intermetallic compounds ofthickness after annealing at 400 C for 9min.

ast to experiments with Al99.5, even after annealingr up to 64min and 500 C for short annealing times,ers remained so thin (


Fig. 7. Exampand at 600 C fgreen, 6 vio

annealing aside) two coobserved widentied bmaterial anthe two comin Fig. 6(a)right side. Tfor 9min aron the left s(patterns githe base mbetween th

After anthe developcharacterisain Fig. 7. SEand600 C flayers can bFig. 7(c). Tobut roughlylayer. The 4les for SEM analysis of the reaction layers formed between steel and AlSi5: (ac) SEM mor 9min, respectively. (d) EDX line scan result obtained along the red arrow in (b). (e) Colet, 5 orange, blue, yellow, steel red).

t 450 C for 64min. In theHAADF STEMmicrograph (leftmponents in the now continuous reaction layer can behich differ in grey level. These two components werey SAD as 6 phase (Al4.5FeSi) adjacent to the AlSi5 based phase adjacent to steel. The respective locations ofponents are marked in the micrograph on the left side

; examples of indexed SAD patterns are given on theEM observations of samples after annealing at 500 Ce displayed in Fig. 6(b). In the HAADF STEM micrographide, three components can now be differentiated. SADven on the right side) identied the phases adjacent toaterial as 6 and phase, the third component locatedese two phases was identied as phase.nealing at 500 C for more than 9min and at 600 C,ed reaction layers reached thickness values allowingtion by SEM analysis; selected results are presented

M micrographs of reaction zones after annealing at 500or different times are shown in Fig. 7(ac). Four reactione differentiated as highlighted by white dashed lines inwards the AlSi5 base material, three layers of varyingequal thickness together take up about 1/3 of the entireth layer adjacent to steel accounts for the remaining

2/3 of the lAlSi5 remathe interfacsampleswialong the rmap (Fig. 7of phases padjacent toin Fig. 7(e))phase (blue(yellow). Th and (mbe Si-rich;shown in p

4.2. Micros

The appand subseqface (Sectioof the baseand after aicrographs after annealing at 500 C for 16min, at 500 C for 64minlour coded EBSD phase map after annealing at 600 C for 16min (Al

ayer. While the interface of the reaction layer towardsins smoothly irregular with increasing layer thickness,e towards steel develops wavy features as observed inthAl99.5 (Fig. 4). FromtheEDX line scan result (Fig. 7(d))ed arrow in Fig. 7(b) and the colour coded EBSD phase(e)) after annealing at 600 C for 16min, the four typesresent in the reaction layers can be identied. Directlythe AlSi5 base material is the 6 phase (violet coloured, followed by layers of 5 phase (Al8Fe2Si; orange) and ). The dominant component adjacent to steel is phasee white particles within and at the interface betweenarked with a white arrow in Fig. 7(c)) were found tomost probably consisting of the 1 phase (Al2Fe3Si3) asrevious interdiffusion experiments [15,39].

tructure evolution of the Al-side

lied heat treatment does not only result in interdiffusionuent intermetallic phase formation at the steel/Al inter-n 4.1), but is also expected to alter the microstructurematerials. The Al-side of samples in the as-welded statennealing for 9min at 200, 400 and 600 C was investi-


Fig. 8. Inverseand after anne

gated forboboth outsidcent to theexemplaryas-weldedsamples wipictures, blaat the bottoimproperly

In the aaverage graAlSi5 basea considerawithin laterner microalloys as a rFSW procesgrains size u

After annanAGS of 35and a morebi-modal grcesses [40]remain at a

A furtheshown in Fmaterials wthe interfacpole gure maps obtained by EBSD analysis of the Al side of joints between steel and Alaling at 400 C for 9min (bottom) for joints with Al99.5 (a) and AlSi5 (b), respectively.

thAl alloys. EBSDscanswereperformed foreachsamplee the weld region (base material) and directly adja-steel/Al interface (interface). Fig. 8(a) and (b) showsresults as inverse pole gure maps from samples in thestate (top) and after annealing at 400 C (bottom) forth Al99.5 and AlSi5, respectively. Within the interfaceck and/or ne grained features within the Al alloy and

mright edge represent steel particles andbasematerial,indexed by the chosen EBSD settings.s-welded state, the Al99.5 base material exhibits anin size (AGS; approximate values) of 20m and thematerial an AGS of 120m, while both alloys containble amount of low-angle grain boundaries (sub-grains)ally elongated grains. At the interface region, a much

structure with an AGS of 3m can be observed in bothesult of the severe plastic deformation induced by thes [38]. Annealing at 200 C (not shown in Fig. 8) left thenchanged compared to the as-welded state.ealingat400 Cthebasematerial ofbothalloysexhibitsmeach, a signicantly reduced number of sub-grainsequiaxed microstructure. At the interface region, a

ain growth typical for secondary recrystallisation pro-can be observed for both Al alloys: while some grainssize of 3m, others grow to 80m in diameter.r increase of the annealing temperature to 600 C (notig. 8) resulted in a slightly increasing AGS of the baseith 40m for Al99.5 and 50m for AlSi5. The AGS ate region reached 30m for both alloys with a much

more homoannealing a

4.3. Growthbetween ste

The averiments at 4function of

d = k

t

The thicFig. 9(a) as cas squares.lled symbannealing tof the differ4.1 the temperatur

Thegrowsampleswi0.176 for 45even morethe tted livalues obta

ResultsFig. 9(b) dples with Aalloys: maps from the two scan locations in the as-welded state (top)

genous grain size distribution than as observed aftert 400 C.

of the intermetallic reaction layers at the interfaceel and Al alloys

age thickness data (d inm) from the annealing exper-50, 500 and 600 C are presented in Fig. 9(a) and (b) assquare root of time in min:

(1)

kness values from samples with Al99.5 are shown inircles and thevalues fromsampleswithAlSi5 in Fig. 9(b)Empty symbols represent experiments at 450 C, semi-ols stand for 500 C and solid lled symbols for 600 Cemperature. No effort was made to measure the growthent components of the layers, but as shown in Sectionphase takes up the largest fraction of the layers in thise range.th rates (k inms1/2) derived fromthekineticdata for

thAl99.5 showsimilar values for all three temperatures:0 C, 0.141 for 500 C and 0.242 for 600 C. At 500 and

pronounced at 450 C, however, the deviation betweenne and the measured values is not as small as for theined at 600 C.for samples with AlSi5 plotted at a different scale ineviate signicantly from the values measured in sam-l99.5: at 450 C, almost no growth of the reaction layer


Fig. 9. Growth(a) layer thicknthickness in jo

takes placefor 500 C ithe growthlarger than

4.4. Tensile

In orderinvestigatedmaterials (nsamples ofthe sheets aFig. 10 showalloys. Thefracture elorespective vAl99.5.

4.4.1. TensiThe valu

tensile teststhe respectias-welded jciency com

Stress/strain curves from tensile tests of the three base materials from this

welded state as well as after annealing at temperatures upC failed in the Al weld region. In this temperature rangeies by about 10MPa due to minimal porosity in Al stirnd slightly differing depth of the groove left by the pass-lding tool shoulder (Section 4.1). After annealing at 300 C,he joints drops to about 55MPa and remains constant aftering at 400 and 450 C. The value of B, however, increases.6% in the as-welded state to values up to 9.3% after anneal-ween 200 and 450 C. After annealing at 500 C the samplesFig. 10.study.

the as-to 450Rm varzone aing weRm of tannealfrom 3ing betkinetics of the reaction layers formed between steel and Al alloys:ess as function of square root of time for joints with Al99.5. (b) Layerints with AlSi5.

, resulting in a rate of 0.01. While the k value of 0.201s quite similar to the respective value shown Fig. 9(a),rate at 600 C is with a value of 2.303 almost 10 timeswith Al99.5 at the same temperature.

testing of joints between steel and Al alloys

to have a reference point for the strength of thejoints, tensile tests were performed for all three base

ot annealed). The samples were prepared as the tensilethe joints; i.e. perpendicular to the rolling direction ofnd having the same dimensions as sketched in Fig. 1(a).s typical stress/strain curves for the steel and both Al

steel exhibited a tensile strength Rm of 330MPa and angation B of 28%. The Al alloys reached much loweralues of 150MPa/9.5% for AlSi5 and 120MPa/6.5% for

le testing of joints with Al99.5es of Rm, B and the location of failure obtained for allin joints with Al99.5 are listed in Table 1, together withve T/t cycle and the total reaction layer thickness d. Theoints possess a tensile strength of 71.6MPa (60% ef-pared to the Al99.5 base material). All samples from

begin to faigrowing abSamples aning mountifrom SEM inin joints wmeasuremecation, whon the over(Fig. 11(b))differentiatcompositiocompositiosurfaces wh

Table 1Reaction layerthe failure locparameters ap

Joints with A

T/C

As-welded200300400450450450450500500500500600600600600

a a =Al-weldb Not measuc Sample prl at the steel/Al interface. With reaction layer thicknessove 7m, Rm and B of the joints drastically decrease.nealed at 600 C for 36min or longer even failed dur-ng in the testing machine. Fig. 11 shows typical resultsvestigations of fracture surfaces of an interfacial failure

ith Al99.5. Chemical compositions obtained from EDXnts are included in the inserts taken at higher magni-ose locations are highlighted by white dashed squaresview pictures of both Al side (Fig. 11(a)) and steel sideof the fracture surfaces. Three types of fractures can beed: (1) bright, smooth cleavage surfaceswith a chemicaln of 73 at.% Al and 27at.% Fe on both sides, matching then of and/or phase (Fig. 4(d)). (2) Fine grained, porousich appear bright on the steel side (73 at.% Al) and dark

thickness d, ultimate tensile strength Rm , fracture elongation B andation of joints with Al99.5 together with the respective annealingplied. The data represent average values from three tensile tests.

l99.5

t/min Rm/MPa B/% d/m Failure ina

71.6 3.6 0 a

9 74 4.2


Fig. 11. SEM i(b) Micrographsquares on the(at.%).

on the Al siure (100at.(2). Fracturesample surftion layer thquantity ofsurfaces of

4.4.2. TensiTable 2

together withickness. Ithe resultsonly at a higwelded stat300 and 450500 C but aon, the ducwith jointsa plateau ointerfacial flogue as deson both steeimate chemmatching tAl-rich ridgnvestigations of the fracture surfaces from a tensile test of a joint between steel and Al99s of the steel side. The pictures in the middle of this gure show details of the surfacesouter pictures. Chemical concentrations of different features of the surfaces, as highligh

de (100at.% Al). (3) Dark ridges resembling ductile fail-% Al on both sides) located on the ne grained surfacetypes (2) and (3) take up the largest area fraction of the

aces exhibiting interfacial failure.With increasing reac-ickness and subsequently decreasing joint strength, theductile ridges (3) decreases until only the ne grainedtype (2) prevail.

le testing of joints with AlSi5lists the results of all tensile tests of joints with AlSi5,th respective annealing parameters and reaction layern the as-welded state and after annealing up to 450 C,exhibit similar characteristics as for joints with Al99.5,her level of strength: Rm drops from 115MPa in the as-e (77% efciency) to 100MPa after annealing betweenC. Interfacial failure of joints sets in after annealing att reaction layer thicknesses of just 1.6m. From theretility decreases with growing reaction layers to 1% aswith Al99.5. The strength of joints, however, remains atf 30MPa. Fig. 12 displays fractography results for anailure in joints with AlSi5 (SEM and EDX analysis ana-cribed in Fig. 11). The dominant fracture type observedl andAl side is a smooth cleavage exhibiting an approx-ical composition of 72 at.% Al, 26 at.% Fe and 2% Si,

he composition of phase (Fig. 7(d)). Very few dark,es with ductile fracture behaviour appear on both sur-

Table 2Reaction layerthe failure locparameters ap

Joints with A

T/C

As-welded200300400450450450450500500500500600600600600

a a =Al-weldb Not measu

faces. On thobserved, wlayer thickn.5 after annealing at 500 C for 9min. (a) Micrographs of the Al side.at higher magnication, as indicated by the respective white dashedted by circles on the respective micrographs, was measured by EDX

thickness d, ultimate tensile strength Rm , fracture elongation B andation of joints with AlSi5 together with the respective annealingplied. The data represent average values from three tensile tests.

lSi5

t/min Rm/MPa B/% d/nm Failure ina

115 4.4 0 a9 115.3 5.3


Fig. 12. SEM i(b) Micrograpsquares on the(at.%).

5. Discussi

5.1. Formatbetween ste

Despiteresolution ters could bfrom this dbe observeAl99.5 andperature ratotal layerous resultsgrowth alonarrangemen

After anof Alappearancestate interdtion of Fe abe explainerials inducehigher amotem of this phase evenvestigations of the fracture surfaces from a tensile test of a joint between steel and AlShs of the steel side. The pictures in the middle of this gure show details of the surfacesouter pictures. Chemical concentrations of different features of the surfaces, as highligh

on

ion of intermetallic reaction layers at the interfaceel and Al alloys

the investigation of the steel/Al interface using highechniques such as TEM, no intermetallic reaction lay-e detected in the as-welded state (Fig. 3(a)). Startingirect bonding, reaction layer thicknesses >1m couldd after annealing at 450 C and above for joints withof 500 C for joints with AlSi5 (Section 4). In this tem-nge, the phase takes up the largest fraction of thethickness for both Al alloys in agreement with previ-[15,21]. This prevalence of was attributed to its rapidg its c-axis facilitated [18] by a highly open structuralt of atoms along this crystallographic direction [41].

nealing in the range of 450600 C, the phase sequencesteel can be observed in joints with Al99.5 (Fig. 4). Theof between and Al is contradictory to previous solidiffusion experiments [15,21] and typical for the reac-lloys with Al melts [15,18,19,42]. This new result cand by the severe plastic deformation of the base mate-d by the FSW process prior to the heat treatment: theunt of stored energy available in the interdiffusion sys-study appears to facilitate nucleation and growth of then at temperatures below the melting point of Al. This

difference ishort anneathin bandselsewhere [

The pha450 C are500 and 60(Figs. 6 andof the reacprevious neutectic phbelow the e

5.2. Growthbetween ste

For bothapproximat(Fig. 9(a) abase matertion layersLayer thicking experimreaction tim

Howeveis also consi5 after annealing at 600 C for 9min. (a) Micrographs of the Al side.at higher magnication, as indicated by the respective white dashedted by circles on the respective micrographs, was measured by EDX

n nucleation conditions together with the comparablyling timeschosenheremight alsoexplain theabsenceofof and phase (AlFe3C) between and steel reported15].ses observed in joints with AlSi5 after annealing at6 adjacent the AlSi alloy and towards steel. At0 C the phases and 5 appear in between 6 and 7). This time and temperature dependent compositiontion layers in joints with AlSi5 is expected in view ofdings [15,39]. It is further noteworthy that the ternary

ase 6 could be detected after annealing at 450 C;muchutectic reaction temperature of 573 C [12].

kinetics of the reaction layers at the interfaceel and Al alloys

Al alloys the growth of the reaction layers can beed by parabolic rate laws in the range of 450600 Cnd (b)). The energy stored in the strongly deformedials does not only inuence the composition of the reac-(as shown above) but also facilitates their growth [18].nesses which require hours to grow in diffusion bond-ents [15,20,21] can be observed after only few minutese.

r, the energy available for intermetallic phase formationumed by recrystallisation processes in the adjacent base


materials (Fig. 8), the amount of which depends on the annealingtemperature. High temperatures apparently remediate the pre-deformation faster. Hence, for both Al alloys the best agreementof the kinetic data with a parabolic form (Eq. (1)) is obtained forannealing aagreementof this studlayer growtat 600 C cotion of phon reactionthe effect ofespecially foand thetteannealing tning of the irepresent ostants obtailower and tapplies fordifferent scparison betat 450 C isincreasinglyferently to t

5.3. Relatiomechanical

In orderon mechanare plottedfrom samplsamples wiin the Al wsteel/Al intetioned failuvery similarall results lmetallic phinherently bmental condare integralon joint streboth alloysferentmechthe joint str

Up to a la constant sThe drop toincreasing dby softeninsation proclayers (Fig. 8drasticallysteel/Al intAlbright [43deformatioray diffractthat the failand the Al bconsideratiphenomenotion layer/ASmigelkas [element (Al

(a) Plot of the tensile strength Rm of joints between steel and Al99.5 (cir-AlSi5 (squares) versus the respective reaction layer thicknesses d (empty failure in the Al weld region, solid lled symbols interfacial failure,ed symbols both interfacial and failure and in theAlweld region). (b and c)rographs of the reaction layer formed between steel and AlSi5 after anneal-00 C for 16min at different magnications. The sample fractured duringraphic preparation.

ection of the dominant diffusive ux, condensate into voids.sults of this study suggest a similar behaviour: despite thenown high hardness of the reaction layers compared to thent base materials [9], joints with Al99.5 which exhibit inter-ailure rupturedmainly between the intermetallic phases andbase material, rather than within the reaction layer (Fig. 11).pography of the respective fracture surfaces on steel anddoes not fully match, similar as it was shown by Albrightdditionally, the area fraction of the parallel observed duc-ridges on both sides decreases with declining joint strength).reby it can be concluded that the tensile strength of jointsl99.5 is mainly governed by the formation of Kirkendall-ty: growing intermetallic layers indirectly weaken the jointincreasing formation of pores at the reaction layer/Al

ce. Such a pseudo-embrittlement (microscopic ductilet 600 C and the respective rate constants are in goodwith previously published data [15]. The experimentsy thus corroborate the strong acceleration of reactionh for interdiffusion between steel and semi-solid AlSi5mpared to the reaction with pure Al [15]. The forma-ase in joints with Al99.5 apparently has no drastic effectlayer growth. At lower temperatures (450 and 500 C)deformation on layer growth becomes more apparent,r joints with Al99.5: the deviations between the valuesd line (Fig. 9(a)) indicate a slower growth rate for longerimes, following an accelerated growth at the begin-nterdiffusion process. Thus, the calculated growth ratesnly an approximate value. The true respective rate con-nable after longer thermal exposure are expected to beo decrease with temperature. The same phenomenonsamples with AlSi5 after annealing at 500 C, only thealing in Fig. 9(b) precludes its visibility. A direct com-ween the greatly differing growth rates for both alloysdifcult, as deformation of the base materials becomesimportant and both Al alloys might have reacted dif-

he applied identical FSW parameters.

nship between interfacial microstructure andproperties of the joints

to discuss the inuence of intermetallic reaction layersical properties of the joints, Rm data for both Al alloysversus their respective d values in Fig. 13(a). Valueses with Al99.5 are shown as circles and the values fromth AlSi5 as squares. Empty symbols represent failureeld region and lled symbols stand for failure at therface. Semi-lled symbols indicate that both aforemen-re locations could be observed. As some Rm/d values are especially for very thin reaction layers (


behaviour of the remaining Al-ridges while the entire joint failsat low strain) can also be observed in some Al alloys causedby precipitation-free-zones along the grain boundaries [46]. TheTEM results presented in Fig. 5 might indicate the evolution ofthe observenano-sizedmentioned

While joincreasing rences compbe observedwelded stainterface selevel of stredecreases uat 12m). Fdrop furtheof respectivmainly withtion layer (Sas themostshowing thdifferent mphase mapphase layerseen that thcuts througface (markecolumnar tion of Fe-rarea fractiojoint streng

Thus it cAlSi5 is conincreasingpreserving)becomes leof 25MPa rcorroborateilar relationexperimentjoints withexplained:cial separatstress requipores, on tinvestigatedings [15], wphase andafter longerreaction laypossibly simthe reactionon the formThe reasonenter the ajoints clearlthe compleof joints in

The factjoints withnant compobe observedof 10mtaking the ethe thickne

reaction layer and the constitution of interfaces and adjacent basematerials just aswell have an impact on the properties of dissimilarsteel/Al joints.

ma

pren lay0.12ce chin th00 Csions

the aobsemed9.5 aaturrea

haseointsinter5wtctionappSi alm be600

alloyckered toa sugn proven

er mth deon oerimhers stuls duh rechantheys a

nts wreac

n layjointjointase;

ate tenthe, theteelroachase

erfacrystals conan octiona rapd porosity (highlighted in Fig. 5(b)), as the trail ofvoids shown in Fig. 5(c) could be linked to the aboveux of vacancies.ints with AlSi5 also show a decrease in strength witheaction layer thickness (Fig. 13(a)), remarkable differ-ared to the respective data for joints with Al99.5 can: the initial drop at low thickness values from the as-

te is not as pronounced and failure at the steel/AlSits in at much lower thickness values and at a higherngth. With further growing reaction layers the strengthntil similar values are obtained aswithAl99.5 (40MParom thereon, however, the strength of joints does notr but remains at about 30MPa. The SEM investigationse fracture surfaces (Fig. 12) show that failure occurredin the phase as the dominant component of the reac-ection4.1.2). Thismore commonbehaviour (failure inbrittle component [9]) is illustrated in Fig. 13(b) and (c),e cross-section of a joint annealed at 600 C for 16min atagnications. The SEM images correspond to the EBSDshown in Fig. 7(e). The sample fractured within the during metallographic preparation. It can be clearlye crack propagating in the brittle intermetallic phaseh remains of steel at the irregular, wavy steel/ inter-d by a white dashed oval in Fig. 13(c)) caused by thephase growth. These ndings correspond to the detec-ich particles on the fracture surfaces (Fig. 12), whosen decreases with growing reaction layers and decliningth.an be concluded that the tensile strength of joints withtrolled by the thickness of the phase layer: with

thickness the energy-consuming (and thus strength-crack-interception at the irregular /steel interface

ss frequent and the joint strength approaches the valueeported as the tensile strength of the phase [9]. Thiss the assumption of Achar et al. [30], who found a sim-between strength and reaction layer thickness in theirs. Moreover, the failure within the Al-weld region ofAl99.5 at thinner reaction layers (8h) and subsequently much thickerers (>150m) as applied/found in this study. Despite ailar diffusive ux of vacancies, the different build-up oflayer in joints with AlSi5 seems to have a strong effectation of Kirkendall-porosity (nucleation conditions).

why the crack propagating through the phase did notdjacent intermetallic layers towards the AlSi side ofy requires further work, but it cannot be ruled out thatx geometry of the steel/Al interface in the investigatedthis study (Fig. 2) also plays a role.that different failure mechanisms can be observed inboth Al alloys despite the phase as the domi-nent in both casesand that interfacial failure couldat much thinner reaction layers than the critical limit

given by Achar et al. [30], highlights the importance ofntire joint into account rather than focussing only onss or hardness of the reaction layer: the build-up of the

6. Sum

Thereactiolow C (Interfajoints2006conclu

(1) InbeforAl9perthe pin jforAlreatheAlfor

(2) AtAlthipardatsioconaft

(3) Boatiexphigthiria

(4) HigmeingallojoithetioofinphandThbyers/sapp pintrecria

(5) Asreabyry and conclusions

sent study investigated the formation of intermetallicers which form at the interface of FSW joints betweenwt.% C) steel and both pure Al (99.5%) and Al5wt.% Si.aracterisation and tensile tests were performed withe as-welded state and after annealing in the range offor 964min. From the obtained results the followingcan be drawn:

s-welded state, no intermetallic reaction layers couldrved in TEM investigations. Layer thicknesses >1mduring annealing at 450 C and above in joints withnd at 500 C and above with Al5wt.% Si. In this tem-e range, the phase is the dominant component ofction layer for both Al alloys. Additionally, a layer ofcan be observed between and the Al base materialwith pure Al, which has been reported previously onlydiffusion reactions of steelswithAlmelts. In jointswith.% Si, a time and temperature dependent build-up of thelayer could be observed. Annealing at 450 C results in

earance of the ternary eutectic phase 6 between theloy and . At higher temperatures the phases and 5tween 6 and .C, the growth of the reaction layers in joints with boths is in good agreement with a parabolic form and muchreaction layers can be observed with Al5wt.% Si com-joints with pure Al. For temperatures below 600 C ourgest a faster growth at the beginning of the interdiffu-cess. In general, similar reaction layer thicknesses as in

tional diffusion bonding experiments could be observeduch shorter annealing times.viations in the phase sequence and a growth acceler-f reaction layers compared to previous interdiffusionents at similar temperatures can be explained by the

amount of stored energy in the system investigated indy, induced by the strong deformation of the basemate-ring FSW prior to the interdiffusion reactions.solution interface characterisation complimented withical testing and fractography analysis allow explain-role of intermetallic phases in joints between both Alnd steel: growing intermetallic layers indirectlyweakenith pure Al by the formation of Kirkendall-porosity attion layer/Al interface. Interfacial failure sets in at reac-er thicknesses of 7m and both strength and ductilitys with pure Al drastically decreases. Interfacial failures with Al5wt.% Si takes place as cleavage within the starting from reaction layer thicknesses of just 1.6ma higher level of strength than in joints with pure Al.sile strength of joints with Al5wt.% Si is governedthickness of the phase. With growing phase lay-energy-consuming crack interception at the irregularinterface becomes less frequent and the joint strengthhes the reported value for the tensile strength of the. At reaction layer thicknesses below the starting ofial failure (temperatures of 200450 C), recovery andllisation processes in the pre-deformed Al base mate-trolled the strength of joints with both Al alloys.utlook, the strength of the bonding in interdiffusions between steel and pure Al melts might be increasedid transit from the melting temperature of Al (660 C)


to about 400 C during cooling of the joint from the reactiontemperature. Thus, the formation of Kirkendall-porosity mightbe minimised and could be further reduced by a compressiveloading of the joint during this temperature range of solid stateinterdiffusion.

Acknowledgements

The authors would like to acknowledge the support from theVirtual Institute for Improving Performance and Productivity ofIntegral Structures through Fundamental Understanding of Met-allurgical Reactions in Metallic Joints (VI-IPSUS). The VI-IPSUSis an initiative of the Helmholtz Association coordinated by theHelmholtz-Zentrum Geesthacht.

References

[1] M.F. Ashby, Materials Selection in Mechanical Design, Butterworth-Heinemann, Burlington, MA, USA, 2005.

[2] U. Dilthey, L. Stein, Sci. Technol. Welding Joining 11 (2006) 135.[3] R.W.K. Honeycombe, Steels Microstructure and Properties, Arnold, London,

1981.[4] W.C. Leslie, E. Hornbogen, in: R.W. Cahn, P. Haasen (Eds.), Physical Metallurgy,

4th Revised Edition, Elsevier Science, Amsterdam, 1996.[5] D. Altenpohl, Aluminium Viewed from Within, Aluminium-Verlag, Dsseldorf,

1982.[6] I.J. Polmear, Metallurgy of the Light Metals, Arnold, London, 1996.[7] H. Hartwig, Aluminium 57 (1981) 615.[8] S. Fukumoto, T. Inuki, H. Tsubakino, K. Okita, M. Aritoshi, T. Tomita, Mater. Sci.

Technol. 13 (1997) 679.[9] V.R. Ryabov, Aluminizing of Steel, Oxonian Press, New Delhi, 1985.

[10] O. Kubaschewski, Iron Binary Phase Diagrams, Springer, Berlin, 1982.[11] U.R. Kattner, Binary Alloy Phase Diagrams, ASM International, Materials Park,

OH, USA, 1990, p. 147.[12] G. Gosh, Ternary Alloys 5, VCH Verlagsgesellschaft, Weinheim, 1995, p. 394.[13] N. Krendelsberger, F. Weitzer, J.C. Schuster, Metall. Mater. Trans. A 38 (2007)

1631.[14] G.V. Kidson, J. Nucl. Mater. 3 (1961) 21.

[15] H. Springer, A. Kostka, E.J. Payton, D. Raabe, A. Kaysser-Pyzalla, G. Eggeler, ActaMater. 59 (2010) 1586.

[16] W. Seith, C. Ochsenfarth, Z. Metall. 12 (1943) 242.[17] E. Gebhardt, W. Obrowski, Z. Metall. 44 (1953) 154.[18] T. Heumann, S. Dittrich, Z. Metall. 50 (1959) 617.[19] G. Eggeler, H. Vogel, J. Freidrich, H. Kaesche, Prak. Metallogr. 22 (1985) 163.[20] K. Shibata, S. Morozumi, S. Kosa, J. Jpn. Inst. Met. 30 (1966) 382.[21] D. Naoi, M. Kajihara, Mater. Sci. Eng. A 459 (2007) 375.[22] J.E. Nicholls, Corros. Technol. 11 (1964) 16.[23] N. Komatsu, M. Nakamura, H. Fujita, J. Jpn. Inst. Light Met. 18 (1968) 467.[24] A. Knauscher, Neue Htte 19 (1974) 398.[25] S.G. Denner, R.D. Jones, R.J. Thomas, J. Iron Steel Inst. 48 (1975) 241.[26] G. Eggeler, W. Auer, H. Kaesche, J. Mater. Sci. 21 (1986) 3348.[27] K. Mechsner, H. Klock, Aluminium 59 (1983) 850.[28] M. Yilmaz, M. Cl, M. Acet, Mater. Charact. 49 (2003) 421.[29] A. Kobayashi, M. Machida, S. Hukaya, M. Suzuki, JSME Int. J. A 46 (2003) 452.[30] D.R.G. Achar, J. Ruge, S. Sundaresan, Aluminium 56 (1980) 220.[31] C.R. Radscheit Laserstrahlfgen von Aluminium mit Stahl. Ph.D. Thesis, Univer-

sitt Bremen, Bremen, Germany, 1996.[32] L. Agudo, D. Eyidi, C.H. Schmaranzer, E. Arenholz, N. Jank, J. Bruckner, A.R.

Pyzalla, J. Mater. Sci. 42 (2007) 4205.[33] J.L. Murray, A.J. McAllister, Bull. Alloy Phase Diagr. 5 (1984) 74.[34] G. Eggeler, Ph.D. Thesis, Untersuchungen zum Feueraluminieren von

niedriglegiertem Stahl, Friedrich Alexander Universitt, Erlangen, 1985.[35] A.J. Schwartz, M. Kumar, B.L. Adams (Eds.), Electron Backscatter Diffraction in

Materials Science, Kluwer Academic, New York, 2000.[36] N.D. Browning, M.F. Chisolm, S.J. Pennycook, Nature 366 (1993) 143.[37] D.B. Williams, C.B. Carter, Transmission Electron Microscopy: A Textbook for

Materials Science, Plenum Press, New York, 1996.[38] R.S. Coelho, A. Kostka, J.F. dos Santos, A.R. Pyzalla, Adv. Eng. Mater. 10 (2008)

1127.[39] T. Maitra, S.P. Gupta, Mater. Charact. 49 (2003) 293.[40] G. Gottstein, Physical Foundations of Materials Science, Springer Verlag, Berlin,

2004.[41] K. Schubert, U. Rosler, K. Anderko, L. Harle, Naturwiss 16 (1953) 437.[42] A. Bouayad, C. Gerometta, A. Belkebir, A. Ambari, Mater. Sci. Eng. A 363 (2003)

53.[43] C.E. Albright, Welding J., Res. Suppl. (1981) 207.[44] A.D. Smigelkas, E.O. Kirkendall, Trans. AIME 171 (1947) 130.[45] V.R. Ryabov, Welding of Aluminium Alloys to Steels, OPA Amsterdam B.V.,

Netherlands, 1998.[46] Y.S. Sato, M. Urata, H. Kokawa, Met. Mater. Trans. A 33 (2002) 625.

Influence of intermetallic phases and Kirkendall-porosity on the mechanical properties of joints between steel and alumini...IntroductionObjectiveMaterials and methodsProduction of samplesCharacterisationTensile testing and fractography

ResultsCharacterisation of the interface regions between steel and Al alloysInterface in joints with Al99.5Interface in joints with AlSi5

Microstructure evolution of the Al-sideGrowth of the intermetallic reaction layers at the interface between steel and Al alloysTensile testing of joints between steel and Al alloysTensile testing of joints with Al99.5Tensile testing of joints with AlSi5

DiscussionFormation of intermetallic reaction layers at the interface between steel and Al alloysGrowth kinetics of the reaction layers at the interface between steel and Al alloysRelationship between interfacial microstructure and mechanical properties of the joints

Summary and conclusionsAcknowledgementsReferences

springer kostka11

Documents

temperature of al

pure al

al melts1618

aluminium al alloys

fe alloys

alrich intermetallic

al alloys areand range

solidied al melt