当前位置：文档库 › 铝合金搅拌摩擦焊1

铝合金搅拌摩擦焊1

Mode I fracture and microstructure for 2024-T3friction stir welds

Michael A.Sutton *,Anthony P.Reynolds,Bangcheng Yang,Robert Taylor

Department of Mechanical Engineering,Uni v ersity of South Carolina,300Main Street,Columbia,SC 29208,USA

Recei v ed 25September 2001;recei v ed in re v ised form 25January 2002

Abstract

Detailed microstructural studies and mode I fracture experiments ha v e been performed on both base material and three families of friction stir welds (FSWs)in 7mm thick,2024-T3aluminum plate,designated hot,medium and cold due to the le v el of nominal heat input during the joining process.Microstructural studies indicate that the FSW nugget grain structure is relati v ely uniform in all welds,with a banded microstructure existing on horizontal cross-sections tra v ersing the weld region;the spatial wa v elength of the bands corresponds to the tool ad v ance per re v olution.The microstructural bands ha v e ele v ated particle concentrations,with the particles ha v ing the same elemental content as base metal impurities,implying that the FSW process is responsible for the obser v ed particle redistribution and microstructural banding.Furthermore,particle redistribution in all welds resulted in (a-1)particle size and v olume fraction reduction on the ad v ancing side of the weld nugget and (a-2)an attendant increase in particle v olume fraction on the retreating side of the weld nugget.Finally,results indicate that hardness minima are present in the heat affected zone (HAZ)outside of the weld nugget on both the ad v ancing side and retreating side for all welds,with the hot weld ha v ing the lowest o v erall weld hardness.Results from mode I fracture tests indicate that the measured critical COD at a fixed distance behind the crack tip is a v iable fracture parameter for FSW joints that is capable of correlating the obser v ed load-crack extension response for both the base metal and all FSWs.In addition,critical COD measurements indicate that FSW joints ha v e a through-thickness v ariation in fracture resistance.Finally,the obser v ed ductile crack growth path (which remained in the FSW region for all FSW joints)can be correlated with the locations of hardness minima,with microstructure v ariations affecting local fracture processes and the corresponding crack path.

Keywords:Friction stir weld;2024-T3aluminum;Microstructuremode I/stable tearing;Crack opening displacement

1.Introduction

Friction stir welding,FSW,is a solid state joining process de v eloped in the early 1990’s at the Welding Institute in the UK [1].Pre v ious work has shown that FSW can pro v ide v ery high strength joints in aluminum alloys that ha v e pre v iously been considered either marginally weldable or unweldable.For example,typi-cal FSW joint strength for 2024-T3is in the range of 80á85%of the parent metal strength (based on tensile strengths).Similar results ha v e also been demonstrated for 7XXX series aluminum alloys [2].Because of their poor properties in the as-fusion-welded condition,high

strength,aerospace aluminum alloys are not normally considered for applications where assembly by welding is required.

The de v elopment of the FSW process may make it possible to produce high quality,welded airframes economically.One reason for this belief is that FSW is an in situ,solid-state,extrusion and forging operation that results in joint formation by bringing oxide free material on both sides of the joint line into intimate contact.Because the process occurs in the solid state,no solidification structures are created,thereby eliminating the brittle,inter-dendritic,eutectic phases common in fusion welded joints using highly alloyed aerospace aluminum.This fact is demonstrated by se v eral studies performed by other in v estigators (e.g.[3á5]).Another reason for interest in FSW as a replacement for fusion welding is the supposed reduction in the le v el of residual

*Corresponding author.Tel.:'1-803-777-4158;fax:'1-803-777-0106.

E-mail address:sutton@https://www.wendangku.net/doc/6b16761238.html, (M.A.Sutton).Materials Science and Engineering A354(2003)6á

www.else v https://www.wendangku.net/doc/6b16761238.html,/locate/msea

stresses in a FSW relati v e to those in typical fusion welds.Howe v er,literature support for this general supposition remains scanty[6,7].

The FSW joint region is normally di v ided into a thermo-mechanically affected zone(TMZ),and a heat affected zone(HAZ).The part of the TMZ that experiences high strain and undergoes recrystallization, e v en in alloys that are highly resistant to recrystalliza-tion,is known as the nugget.The nugget is approxi-mately symmetric about the weld centerline(although other weld features are not)and is typically similar in diameter to the pin of the FSW tool.As pre v iously noted by Sutton et al.[8],a trans v erse cross-section through a2024-T3FSW shows that a refined,equiaxed grain structure is present in the nugget.Near the edge of the FSW process zone(i.e.one tool shoulder radius from the weld centerline at the crown of the weld and one tool pin radius at the root),there is a rapid transition to the base metal microstructure,with large grains elongated in the rolling direction.Interestingly, metallographic sectioning at different depths in the weld indicate the presence of microstructural bands in the FSW.The bands appear to be associated with the flow of material during the welding process,with frequency with the weld processing parameters.

E v en though the FSW process has many attracti v e features,identifying and v erifying a reliable fracture parameter to characterize the toughness of the hetero-geneous weld microstructure presents significant diffi-culties.In this regard,pre v ious work has shown that crack tip fields are complex and highly three-dimen-sional in character,e v en in nominally homogeneous materials.Thus,it seems clear that two-dimensional fracture parameters(e.g.caustic size,mode I stress intensity factor)will most likely not be adequate to predict the fracture process in a heterogeneous FSW joint[9á11].

Gi v en the potential ad v antages of FSW(it is currently being considered as a ri v et replacement technology for airframe production)it is important that an in-depth understanding of the damage tolerance of FSWs be de v eloped so that the FSW processing parameters can be chosen to pro v ide the optimum mix of toughness, strength,and other critical properties.In this study, three different weld process parameter sets ha v e been chosen in order to facilitate a correlation between FSW processing conditions,weld mechanical response,and microstructure.

In Section2,a basic terminology for describing the location and orientation of microstructural features in FSW joints is presented and the experimental proce-dures used in this study are described.Results from the microstructural examinations and a series of mode I fracture experiments are presented in Section3,with emphasis on the measured v alues of critical crack opening displacement(COD)during stable crack growth.Section4presents a discussion of results with emphasis on integrating both pre v ious work and current results.

2.Experimental procedures and FSW processing

2.1.Geometric description of friction stir welds

Fig.1shows a schematic v iew of an FSW and the associated coordinate system used to describe the FSW in this work.As shown in Fig.1,the tool contacts the crown side of an FSW during processing while the root side is in direct contact with a backing plate and opposite to the crown side.The ad v ancing and retreating sides of the weld correspond to positions of the maximum and minimum relati v e v elocities between the rotating tool and the work-piece.To be consistent with typical fracture mechanics coordinate systems,where X corresponds to the direction of crack growth,the longitudinal(X)axis in this study is chosen to be opposite to the direction of tool tra v el(this choice is opposite to the direction used in pre v ious work[8]).The depth(Z)axis is in the specimen thickness direction and directed towards the weld crown.The trans v erse(Y) axis is perpendicular to the X-direction and is directed towards the ad v ancing side of the weld.In this case,the XáYáZ system is a right-handed coordinate system with the origin located at mid-thickness and at the weld centerline;the longitudinal location of the origin within the specimen is arbitrary since all data used in this work were obtained from steady-state weld locations.

2.2.Friction stir welding procedure

Pieces of7mm thick,2024-T351plate,1.22m long by 0.61m wide,were friction stir welded together along their long sides.The tool used for welding these plates had a23.0mm diameter shoulder and an8.2mm diameter pin.The pin was threaded with a standard pitch machine screw thread.The longitudinal direction of the FSW was perpendicular to the rolling direction of the plate.

The primary process v ariables controlling the energy input to a FSW(assuming sufficient downward force in the Z-direction to achie v e full consolidation within the weld)are the welding speed and the tool rotation rate. Table1presents a summary of the welding parameters for the three different welding parameter sets used in this study.Based on the specific energy input for the three welds(J mm(1),they are henceforward termed the cold,medium,and hot welds.

The specific energy input to the welds was calculated using the equation E0(T v)/v t,where E is the specific energy input in J mm(1,T is the measured torque in N m,v is the angular v elocity of the tool(rad s(1)and

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á167

v t is the tool tra v el speed (mm s (1)during welding.The formula noted abo v e does not account for parasitic energy losses by conduction through the work-piece fixture (e.g.backing plate on root side of weld)or by con v ection from the top surface of the welded plate.2.3.Microstructural characterization

The grain structure in both horizontal and trans v erse cross-sections of the three weld nuggets was character-ized by optical metallography.Specimens were polished by standard metallographic techniques,anodized in Barker’s reagent,and v iewed with polarized light to pro v ide crystallographic orientation contrast.Fig.2shows clearly defined bands on a horizontal cross-section in both the base material and also the FSW near the ad v ancing side.Figs.3and 4show the microstructure on a v ertical trans v erse cross-section through the FSW joint.

Sections of the base material and the FSW parallel to the rolling plane and at mid-thickness were examined

using back-scattered electron imaging to determine the distribution of particles within the welds as a function of welding parameters.Particle number and size distribu-tions were determined using an automated

image

Fig.1.Schematic of FSW fracture specimen with coordinate system and ?aw location.

Table 1

Welding parameters for the three FSW types Weld parameters

Cold weld Medium weld Hot weld Welding speed,v t (mm s (1) 4.45

3.29

1.28

Angular v elocity,v (rad s (1)37.7(360rpm)37.7(360rpm)22.5(215rpm)Power (W)

329528962148Energy per unit weld length,E (J mm (1)

741

879

1680

Fig.2.Horizontal cross-sections at centerline of plate (Z 00mm)for (a)base metal and (b,c,d)FSW near ad v ancing side.

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á16

analysis program.The quantitati v e data were obtained using multiple sample areas,0.67)2mm 2in size.The magnification was such that the minimum particle size resolution was just under 2m m;the minimum particle size corresponds to the pixel size in the magnified image.Figs.5á7show (a)the particle number density,(b)mean particle size and (c)particle v olume fraction along the T -direction at the weld centerline in a v ertical trans v erse cross-section.In Figs.5á7,the v ertical dashed lines indicate the pin diameter,the horizontal dashed line is the reference base metal v alue and the weld centerline is at zero on the horizontal axis.The a v erage nugget grain size was determined by calculation of the mean linear intercept distance of boundaries along a test line in the FSW nugget region [12].Table 2presents the nugget grain size data for all three welds,where the data was obtained.

Finally,Knoop micro-hardness measurements were made across the FSW joint.The hardness v alues were obtained using a load of 100g and a dwell time of 20s.Knoop micro-hardness as a function of position relati v e to the weld centerline of each weld is shown in Fig.8.

2.4.Fracture testing

Standard compact tension (CT)specimens ha v ing the geometry shown in Fig.9and a width of W 076mm [13]were used for both the base metal and FSW studies.All specimens were machined with the same orientation relati v e to the plate rolling direction,with the base metal specimens in the L áT orientation.FSW specimens were machined so that the notch plane contained the through-thickness plate direction and the centerline

Fig.3.Microstructure for medium FSW in v ertical trans v erse cross-section at mid-thickness of specimen (Z 00).All dimensions in mm.

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á169

Fig.4.Microstructures as a function of FSW process at two positions near mid-thickness in v ertical trans v erse section (Z 00

mm).

Fig.5.Particle number density v ersus trans v erse position at Z 00for hot medium and cold

welds.Fig.6.Mean particle size v ersus trans v erse position,Z 00for hot,medium and cold welds.

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á16

the weld;starter notches were 6.4mm long as measured from the specimen load-line.

It is noted that all but one of the FSW specimens were machined with the crack growing opposite to the welding direction.In all cases,the crown side of each FSW specimen was machined to remo v e flash and ensure that the sides of the CT specimens were parallel to each other,as a result,the final FSW specimen thickness was 5.5mm.Fatigue pre-cracking was per-formed at a constant D K 07.7MPa am and R 00.1for

both FSW and base metal specimens;the final fatigue pre-crack length was 30.5mm.

2.5.Crack opening displacement measurements Pre v ious work by one of the authors [14á26]has demonstrated that COD at a fixed distance behind the current crack tip is a v alid parameter for characterizing the stable tearing resistance of rolled aluminum alloy plate material undergoing both mode I and mixed mode loading.

In this work,the data acquisition procedure de v el-oped during pre v ious studies [22,23]was used to acquire critical COD v alues during stable tearing.Briefly,the procedure is as follows.First,a random black and white pattern (Fig.10)ha v ing a spot size of approximately 45m m was applied to the specimen surface using commer-cial enamel spray paints.Second,a 7)10mm 2region around the current crack tip location was imaged using a combination lens system and a CCD camera.In this work a Pulnix CCD camera with a 200mm Canon FD lens and two-2)-extenders was used to obtain a typical image magnification of approximately 78pixels mm (1for all specimens.Third,loading was performed in displacement control until surface crack growth was obser v ed on the v ideo monitor.At this point,images of the crack tip region were obtained either (a)as the current crack front began to ad v ance or (b)just after incremental crack growth had occurred.Fig.10shows two of the images used in this work to quantify COD during crack growth.After an image has been acquired,the load-point displacement is increased and the process repeated until unstable crack growth occurred.

After acquiring images of the crack at v arious stages of growth,two subsets were selected at approximately 1mm behind the current crack tip and along the crack line (Fig.10).By comparing the subset locations in

two

Fig.7.Particle v olume fraction on horizontal weld cross sections at Z 00.

Table 2

A v erage FSW nugget grain size FSW weld type Nugget grain size (microns)Cold FSW 7.3Medium FSW 6.4Hot FSW

5.5

Fig.8.Knoop hardness as a function of position relati v e to the weld centerline at Z 0

Fig.9.Geometry of CT specimen.

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á1611

images using custom image analysis software,the relati v e displacements between the two subsets can be quantified and COD measured.In this work,the two-dimensional digital image correlation [24á30]software package,VIC-2D [31],was used to post-process the images and quantify COD throughout the crack growth process in the FSW and base metal specimens.As in pre v ious work,COD is defined as the relati v e crack surface displacement at 1mm behind the current crack tip and can be written in the form

COD 0f (u 2(u 1)2'(v 2(v 1)2g

1=2

where (u i ,v i )are components of crack surface displace-ment transformed into the local x and y directions along the current crack direction (see Fig.10b)for subsets 1and 2.

3.Results

3.1.Microstructural characterization

As shown in Figs.3and 4,the nugget region has a grain structure on a trans v erse v ertical cross-section that is nearly uniform,with transitions to the base metal microstructure on both the ad v ancing and retreating sides.In all cases,the transition is more abrupt on the ad v ancing side of the weld,consistent with current,qualitati v e,ideas regarding the strain and strain rate distributions in FSWs [32].

As shown in Table 2,the mean grain size in the weld nugget decreases with increasing heat input.Based on a v ailable processing data,it is likely that the post-welding nugget grain size v ariation shown in Table 2is the result of a complex combination of static and dynamic re-crystallization and reco v ery processes.Thermo-mechanical history data,most likely obtained through detailed numerical simulation studies in the FSW process,will impro v e our ability to predict the measured distribution in nugget grain size within the weld nugget.

The optical micrographs in Fig.2highlight the constituent particle distributions in the base metal and FSW’s at Z 00,with the welding direction horizontal,right to left.In the base metal v iew,the rolling direction is v ertical.Visible in the weld micrographs are particle rich and particle poor bands.Energy dispersi v e X-ray spectroscopy (EDX)studies re v ealed that all of the particles v isible in Fig.2(both base material and FSW zone)ha v e the same composition,and one that is typical of 2XXX alloy series constituent particles:high in copper,magnesium,iron,and manganese.The arcs described by the bands in the FSW region correspond to the primary material transport path during friction stir welding [33].Along the weld centerline,the band spacing is 0.74mm,0.54mm and 0.35mm for the cold,medium and hot welds,respecti v ely.Furthermore,the band spacing corresponds to the tool ad v ance per re v olution for each particular weld.

One conjecture regarding the processing source for the banding is that the heating and intense straining of material during the rotational extrusion process results in segregation of the hard particles along a ‘flow front’,resulting into well-defined,particle rich bands.Though the precise reason for particle banding is not fully understood,the presence of regular,meso-structural features ha v ing a one-to-one relationship with the tool ad v ance per re v olution appears to be a fundamental result of the FSW process [34].

Quantitati v e information regarding the particle dis-tribution in longitudinal cross-sections of each weld is shown in Figs.5á7.Fig.5shows that for the cold and hot welds,there is a general increase in the number density of particles within the nugget region

(approxi-

Fig.10.Process used for measurement of COD at 1mm behind current crack tip.

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á16

mately94mm relati v e to the weld centerline)as compared to the base metal.The trend obser v ed in the hot and cold weld are different in the medium weld since it exhibits an increased particle number density on the retreating side(left of weld center)and a decrease relati v e to the base metal on the ad v ancing side.As shown in Fig.6,all three welds exhibit the same trends in particle size v s.position relati v e to the weld center-line;particle size diminishes gradually from the base metal le v el on the retreating side of the weld to a minimum on the ad v ancing side.There is an abrupt change from the minimum on the ad v ancing side to the base metal v alue just outside of the weld.As shown in Fig.7,the v olume fraction trend is the same for all welds and similar to the distribution of a v erage particle sizes; the v olume fraction approaches the base metal v alue at the retreating side of the weld and is at a minimum on the ad v ancing side.In this case,howe v er,the medium weld exhibits a much lower minimum than either the hot or cold welds.The low v alue of v olume fraction at the ad v ancing side of the medium weld correlates with the data from Figs.5and6(v olume fraction will correlate with the product of the number density and the a v erage size).

The Knoop micro-hardness data shown in Fig.8 demonstrates se v eral points.All of the welds ha v e hardness minima(local or absolute)on both the ad v ancing and retreating sides.The trans v erse locations of the hardness minima at the plate centerline(Z00) are in the HAZ,outside of the nugget and independent of any particle redistribution(for example,compare to Fig.7).The minima are located at trans v erse locations that are approximately on a line drawn from the shoulder radius on the crown to the pin radius at the root of the weld.All of the welds exhibit nugget hardness less than the base metal hardness.Both cold and medium FSW joints ha v e nearly constant hardness v alues(KHN$130)within the nugget region and deeper minima on the retreating side than on the ad v ancing side.In contrast,the hot FSW nugget region exhibits a gradual decrease from a high of125on the ad v ancing side of the nugget to110on the retreating side of the nugget.The hot weld also has a deeper minimum on the ad v ancing side HAZ(KHN096)than on the retreating side(KHN0102).

With the exception of two outlying points in the retreating side HAZ’s,the hot weld exhibits generally lower hardness compared to the cold and medium welds: this is expected based on the higher energy input and may arise from a reduced le v el of residual dislocation density in the hot weld nugget.

Based on the data presented in Figs.5á8,the hot, medium and cold welds exhibit se v eral similar trends in particle size distribution,particle v olume fraction and in hardness.It is clear that the v olume fraction of detectable particles within each of the welds is being reduced,particularly on the ad v ancing side.Since no increase in v olume fraction abo v e the base metal le v el was obser v ed,it is likely that the particles are either being dissol v ed or their size is being reduced to such an extent that they are no longer detectable at the magnification used for analysis or both.The medium weld exhibits a significant difference in magnitude of particle v olume fraction reduction within the weld nugget;howe v er,all of the welds exhibit generally similar trends.

3.2.Mode I fracture experiments

Table3presents the measured v alues of COD and maximum load for the hot FSW,medium FSW,cold FSW and base material specimens.Fig.11presents typical load-crack extension data for fatigue pre-cracked CT specimens fabricated from the base material,the hot FSW,the medium FSW and the cold FSW.Data for the three welds fall into a v ery tight band slightly below the base metal data.The lower loads required for stable tearing in the weld specimens as compared to the base metal are also reflected in the COD data presented in Table3.While the COD data accurately reflect the general trend represented by the load-crack extension data,the heterogeneous nature of the welds makes it difficult to assign a single COD to the welds.For example,different COD v alues are measured on the root and crown sides of each weld and COD may v ary with crack extension due to the heterogeneous nature of the joining process.In the base metal,once stable tearing conditions are established,COD remains constant. Howe v er,it appears that the crack must grow to a steady-state condition within the weld(e.g.the micro-structure sampled by the crack front is in v ariant)before a stable COD v alue is obser v ed.The v ariability in COD was most ob v ious in the hot weld,which ne v er appeared to reach a steady state v alue on the root side.Further-more,e v en after the steady state condition is obtained, the root and crown sides do not ha v e the same COD in any of the welds.

Perhaps more striking than the differences in load-crack extension and COD exhibited by the three welds and the base metal are the differences in fracture paths. The base metal,hot FSW and cold FSW exhibit relati v ely typical macroscopic fracture surface slant angles during mode I tearing;the steady state slant angles for the cold weld and the base metal are approximately458,while the hot weld has a somewhat larger angle of approximately558.The medium FSW exhibits a macroscopically flat fracture(08).Crack slant angle as a function of crack extension is plotted for all four conditions in Fig.12.

Another interesting feature of the weld crack paths,as compared to the base metal,is illustrated in Fig.13.In this figure,the position of the crack surface centerline

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á1613

relati v e to the weld centerline (coincident with the initial fatigue crack plane)is plotted as a function of crack extension.In the base metal specimen,the amount of crack path de v iation increases rapidly after approxi-mately 7mm of crack extension.This is common for long cracks in bend specimens,such as the CT type,due to v ariations in the crack tip fields.In contrast,cracks in tensile-loaded weld specimens remain more or less in plane at crack extensions up to 30mm.

4.Discussion

The results presented in the pre v ious section ha v e shown that all of the welds ha v e fracture performance that is somewhat inferior to that of the base metal (i.e.COD is lower).This inferiority is expected based on the fact that the weld microstructure has been modified from a form that was de v eloped in order to maximize the fracture toughness of the material (2024-T3).

Measured critical COD v alues indicate that 2024-T3FSW’s with higher heat input typically ha v e lower measured fracture toughness.Based on limited data,it is conjectured that features that are qualitati v ely similar in all of the welds go v ern macroscopic fracture beha v ior of the welds;features that are significantly different do not appear to be important.Furthermore,differences in

fracture resistance cannot be consistently and directly correlated with obser v ed microstructural features.For example,it could be argued that the medium weld has the lowest fracture resistance of the three welds,though examination of the microstructure data shows only minimal quantitati v e differences between the v arious welds.On the other hand,careful inspection of the hardness,v olume fraction and fracture surface data suggests that (a)the lower nugget hardness for the hot

Table 3

COD and maximum load data for base metal and FSWs Specimen and FSW process

COD (mm)P max (kN)

Root side

Crown side A v erage Base metal (CT)0.09090.0140.09090.0140.09014.81Cold (CT)0.08390.0160.07790.0100.08013.61Medium (CT)0.07590.011

0.07090.0140.07313.14Hot (CT)

0.065á0.100(a v g)0.08490.018

0.08190.016

0.0825

12.76Fig.11.Typical load-crack extension data for FSW-CT

specimens.

Fig.12.Measured slant angle and crack surface pro?les during stable tearing tests of CT specimens.

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á16

weld and(b)the large reductions in detectable particle v olume fraction for the medium weld may be primary contributors to the initially low critical COD for the hot weld and the striking surface features on the relati v ely flat fracture surface of the medium weld,respecti v ely. As stated abo v e,all of the welds used in this study are under-matched.That is,the weld metal is lower strength than the base metal.This is illustrated in a qualitati v e way by the hardness data presented in Fig.8.In addition,pre v ious work by Lockwood and Reynolds has shown that the weld metal itself is quite ductile and has,on a v erage,a yield strength approximately75%of the base metal[35].Also,the recent work by Hao et al.

[36]pro v ides some https://www.wendangku.net/doc/6b16761238.html,ing their terminology and data from the work by Lockwood and Reynolds,the FSWs used in this work are‘slightly under-matched’(a weld metal to base metal yield strength ratio,M 0.5 and B1).In addition,the ratio of specimen width minus crack length to the height of the softer FSW zone,is approximately three at the onset of stable tearing.From the analysis of Hao et al.,under-matched FSW zones of this size will(a)experience an ele v ated le v el of constraint at the crack tip during the early stages of growth and(b) tend to respond in a non-homogenous manner.

Using this work for guidance,the following explana-tion of the obser v ed FSW crack growth beha v ior is pro v ided.Since the weld zone acts non-homogeneously relati v e to the o v erall specimen,and Fig.8demonstrates that the weld zone is softer than the surrounding material,localization of plastic deformation and da-mage will occur in the weld zone,ahead of the crack tip. As damage accumulates in the FSW zone,channeling of stable crack extension along the FSW due to localiza-tion of damage in this region will result.For both the hot and cold welds,the damage process resulted in slant crack growth between the two hardness minima without appreciable microstructural interaction.For the medium weld,which had the most significant particle redistribu-tion in the nugget region,the damage process in v ol v ed both local de-lamination along the particle band inter-faces and normal v oid growth processes.

5.Concluding remarks

Our results suggest that surface COD is a v iable parameter for quantitati v ely determining the stable tearing resistance of FSW joints under mode I loading, e v en when ductile crack growth occurs along a through-thickness slant surface.Since COD can be measured when crack extension occurs under mixed mode loading conditions,these results suggest that a generalized COD incorporating three-dimensional crack tip deformations may be a v iable fracture parameter for FSW joints. Under-matching of the2024-T3FSW nugget and obser v ed crack growth response suggest that mode I loading and stable tearing of longitudinal flaws within the nugget region may result in flaw‘channeling’along the FSW joint due to a combination of(a)localized yielding in the FSW zone and damage accumulation in the FSW region and(b)complex,local particle redis-tribution effects.The latter is most likely the primary cause for the relati v ely flat fracture surface exhibited by the medium weld specimen,though the precise interac-tion between the deformation fields and the local microstructure is not well understood.Additional sup-port for this statement recently was pro v ided by an initial mixed mode I/II test on the medium FSW joint, where initial crack kinking preferentially occurred along the particle bands[8]until crack growth was nearly parallel to the applied loading.

Finally,though it is possible that the fracture resistance of the welded specimens is reduced by the microstructural modifications that take place in the weld region;data presented in this work indicates that crack growth in under-matched FSW joints can be predicted using macroscopic concepts(e.g.COD,yield strength) that do not require additional microstructural informa-tion.

Acknowledgements

The authors gratefully acknowledge the?nancial support pro v ided by(a)Dr K.V.Jata at the Air Force Research Laboratory through Contract F33615-96-D-5835,(b)Dr Julius Dasch at NASA HQ through grant NCC5-174and(c)Dr Delcie Durham,NSF/DMI program manager through grant DMI-9978611.In addition,the technical assistance of Dr Jata during this work is noted and greatly

appreciated.

References

[1]W.M.Thomas,E.D.Nicholas,J.C.Needham,M.G.Church,P.

Templesmith,C.J.Dawes,International Patent Application No.

PCT/GB92/02203.

[2]A.P.Reynolds,W.D.Lockwood,T.U.Seidel,Processing-prop-

erty correlation in friction stir welds,in:Materials Science Forum, vol.331á337,Trans Tech Publications,Switzerland,2000,pp.

1719á1724.

[3]C.G.Rhodes,W.M.Mahoney,W.H.Bingel,R.A.Spurling,C.C.

Bampton,Scripta Metallurgica36(1)(1997)69á75.

[4]K.V.Jata,S.L.Semiatin,Scripta Materialia43(2000)743á749.

[5]K.V.Jata,K.K.Sankaran,J.J.Ruschau,Metallurgical and

Materials Transactions A31A(2000)2181á2192.

[6]M.A.Sutton,A.P.Reynolds,D.Q.Wang,C.R.Hubbard,ASME,

Journal for Engineering Materials and Technology124(2)(April 2002)215á221.

[7]M.James,M.Mahoney,D.Waldron,Residual Stress Measure-

ments in Friction Stir Welded Aluminum Alloys,Proceedings of the First International Symposium on Friction Stir Welding, Thousand Oaks,CA,USA,June1999.

[8]M.A.Sutton, A.P.Reynolds, B.Yang,R.Taylor,Materials

Science and Engineering,A A323(2002)160á166.

[9]R.V.Mahajan,K.Ra v i-Chandar,Experimental Mechanics29(1)

(1989)6á12.

[10]A.T.Zehnder,A.J.Rosakis,Three-Dimensional Effects Near a

Crack Tip in a Ductile Three Point Bend Specimen,Part I:A Numerical In v estigation,GALCIT SM Report88-6,Cal Tech, 1988.

[11]R.Dodds,International Journal of Fracture33(1987)R7áR15.

[12]ASTM Standard E112-88,Standard test method for determining

a v erage grain size,Annual Book of ASTM Standards,v ol.3.01,

ASTM,1916Race Street,Philadelphia,PA,1991.

[13]ASTM Standard E-399-90,Standard test method for plane-strain

fracture toughness of metallic materials,Annual Book of ASTM Standards,v ol.3.01,ASTM,1916Race Street,Philadelphia,PA, 1991.

[14]M.A.Sutton,J.D.Helm,M.L.Boone,International Journal of

Fracture109(2001)285á301.

[15]J.D.Helm,M.A.Sutton,M.L.Boone,Characterizing crack

growth in thin aluminum panels under tension-torsion loading with three-dimensional digital image correlation,in:G.F.Lucas, P.C.McKeighan,J.S.Ransom,(Eds.),Nontraditional Methods of Sensing Stress,Strain and Damage in Materials and Structure, ASTM STP1323.

[16]M.A.Sutton,M.L.Boone, F.Ma,J.D.Helm,Engineering

Fracture Mechanics66(2000)171á185.

[17]M.A.Sutton,F.Ma,X.Deng,International Journal for Solids

and Structures37(2000)3591á3618.

[18]M.A.Sutton,M.L.Boone, F.Ma,J.D.Helm,Engineering

Fracture Mechanics66(2000)171á185.

[19]M.A.Sutton,W.Zhao,S.R.McNeill,J.D.Helm,W.T.Riddell,

R.S.Piascik,Local crack closure measurements:de v elopment and application of a measurement system using computer v ision and a

far-?eld microscope,in:K.Smith,J.McDowell,(Eds.),ASTM STP1343on Crack Closure Measurements,in press,145á156.

[20]W.T.Riddell,R.S.Piascik,M.A.Sutton,W.Zhao,S.R.McNeill,

J.D.Helm,Determining fatigue crack opening loads from near-crack-tip measurements,in:K.Smith,J.McDowell,(Eds.), ASTM STP1343on Crack Closure Measurements,in press, 157á174.

[21]F.Ma,X.Deng,M.A.Sutton,J.C.Newman,Jr.,A CTOD-Based

Mixed Mode Fracture Criterion,ASTM STP1359on Mixed Mode Crack Beha v ior,86á110(1999).

[22]B.E.Amstutz,M.A.Sutton,D.S.Dawicke,M.L.Boone,ASTM

STP129627(1997)105á125.

[23]B.E.Amstutz,M.A.Sutton,D.S.Dawicke,Experimental Study

of Mixed Mode I/II Stable Crack Growth in Thin2024-T3 Aluminum,Fatigue and Fracture,26;ASTM STP1256,(1995) 256á273.

[24]D.S.Dawicke,J.C.Newman III,M.A.Sutton, C.Bigelow,

Orientation Effects on the Measurement and Analysis of Critical CTOA in an Aluminum Alloy,Fatigue and Fracture,ASTM STP 1256,(1995)243á256.

[25]M.A.Sutton,S.R.McNeill,J.D.Helm,H.Schreier,Computer

v ision allied to shape and deformation measurement,in:P.K.

Rastogi,D.Inaudi(Eds.),Trends in Optical Non-Destructi v e Testing and Inspection,,Else v ier,2000,pp.571á591.

[26]M.A.Sutton,S.R.McNeill,J.D.Helm,Y.J.Chao,Ad v ances in

2D and3D computer v ision for shape and deformation measure-ments,in:P.K Rastogi(Ed.),Topics in Allied Physics:Photo-mechanics,vol.77,Springer,2000,pp.323á372.

[27]H.Schreier,J.Braasch,M.A.Sutton,Optical Engineering39(11)

(2000)2915á2921.

[28]H.A.Bruck,S.R.McNeill,M.A.Sutton,W.H.Peters,III,

Experimental Mechanics29(3)(1989)261á267.

[29]M.A.Sutton,S.R.McNeill,J.Jang,M.Babai,Optical Engineer-

ing27(3)(1988)173á175.

[30]T.C.Chu,W.F.Ranson,M.A.Sutton,W.H.Peters,III,Experi-

mental Mechanics25(3)(1985)232á245.

[31]VIC-2D Image Analysis software,Correlated Solutions,Incorpo-

rated,West Columbia,SC29169.

[32]S.Xu,X.Deng, A.P.Reynolds,T.U.Seidel,Science and

Technology of Welding and Joining6(3)(2001)191á193. [33]T.U.Seidel,A.P.Reynolds,Visualization of Material Flow in

AA2195Friction Stir Welds Using a Marker Insert Technique, Metallurgical and Materials Transactions,A32A(2001)2879á2884.

[34]L.Karlsson,L.E.S v ensson,https://www.wendangku.net/doc/6b16761238.html,rsson,Characteristics of FSW

Aluminum Alloys,Proceedings of the Trends in Welding Re-search Conference,Callaway Gardens,Georgia,USA,1á5June 1998,574á579.

[35]W.D.Lockwood,A.P.Reynolds,Mechanical Response of Fric-

tion Stir Welded AA2024:Experiment and Modeling,Materials Science and Engineering,A A323(1á2)(2002)349á354. [36]S.Hao,A.Cornec,K.-H.Schwalbe,International Journal of

Solids and Structures34(3)(1997)297á326.

M.A.Sutton et al./Materials Science and Engineering A354(2003)6á16 16