Materials Science & Engineering A 639 (2015) 280–287

Contents lists available at ScienceDirect

Materials Science & Engineering A

http://d0921-50

n CorrCarolinaTel.: þ1

E-m

journal homepage: www.elsevier.com/locate/msea

On the effect of microstructure on the torsional responseof AA7050-T7651 at elevated strain rates

Silas Mallon, Behrad Koohbor, Addis Kidane n, Anthony P. ReynoldsDepartment of Mechanical Engineering, University of South Carolina, 300 Main Street, Columbia, SC 29208, United States

a r t i c l e i n f o

Article history:Received 3 January 2015Received in revised form1 May 2015Accepted 5 May 2015Available online 14 May 2015

Keywords:Mechanical characterizationAluminum alloysAA7050Torsional Hopkinson barHigh strain rateFriction-stir weld

x.doi.org/10.1016/j.msea.2015.05.01393/& 2015 Elsevier B.V. All rights reserved.

espondence to: Department of Mechanical En, 300 Main Street, Room A132, Columbia803 777 2502; fax: þ1 803 777 0106.ail address: kidanea@cec.sc.edu (A. Kidane).

a b s t r a c t

A torsional split Hopkinson bar has been implemented to observe strain rate effects in the materialresponse of aluminum alloy 7050-T7651. The material, received in the form of 32 mm thick plates, hasbeen tested in the as received condition and also in the as friction-stir welded (FSW) condition.Microstructural observations have been performed using optical microscopy, revealing a fine, equiaxedstructure of small grains in the FSW nugget material and an elongated and markedly larger grainstructure in the as-received material condition, typical of rolled, un-recrystallized plate. Quasi-statictensile and high rate torsional testing reveal lower yield and flow stresses in the FSW material whencompared to the base metal. Testing further demonstrates the presence of appreciable rate dependencein both material states. The present work is also indicative of an increased work hardening rate in theFSW material, while the base metal exhibits only minimal work hardening. The increased propensity forwork hardening and decreased mechanical strength in the FSW material is found to be the primarydifference between the two material states.

& 2015 Elsevier B.V. All rights reserved.

1. Introduction

Aluminum alloys have traditionally been the engineeringmaterial of choice in aerospace applications, mainly due to theirhigh strength to weight ratio and corrosion resistance [1,2]. Alu-minum alloy 7050 is an aerospace grade of aluminum that incor-porates desirable qualities such as high strength, high resistance tostress corrosion cracking (SCC), and good toughness. The temperstudied here is 7050-T7651, which possesses the greatest strengthand good corrosion resistance, while maintaining an acceptableresistance to stress corrosion cracking [3]. This alloy also exhibitslow quench sensitivity, allowing it to maintain high strength inthicker sections, making the alloy desirable in applicationsrequiring thick plate (75–150 mm) [4], such as spars and bulk-heads. These aerospace components may be vulnerable to impactloading in the event of a collision with flying debris, for example.Therefore, an adequate understanding of material behavior at highstrain rates is desirable.

In general, the open literature contains little study of aluminumalloy 7050. Some investigations regarding micromechanical andfracture response of this material have been performed, mostlyutilizing indentation techniques [5,6]. Han et al. [7] have studied

gineering, University of South, SC 29208, United States.

the effect of solution heat treatment on the mechanical char-acteristics and fracture properties of AA7050. Existing work by Renet al. [8] considers the effect of orientation with respect to materialelongation during the forging process on the tensile and fatigueproperties of AA7050. Zheng et al. [9] characterized the effect oflarge plastic deformation on the microstructure and mechanicalresponse of AA7050. Published work dealing with AA7050 islimited to cases considering quasi-static loading conditions.Although a vast amount of effort has been devoted to character-izing similar, older alloys such as AA7075, it would seem that amore thorough understanding of material behavior exhibited byAA7050 would be worthwhile, particularly when consideringelevated strain rate conditions.

The demand for a joining process able to produce high strengthbonds containing minimal defects in materials difficult to join withconventional fusion processes, such as 7000 series aluminumalloys, has given rise to the friction stir welding (FSW) process[10,11]. Changes in microstructure and hardness as a result offriction-stir welding have been well documented in AA7050 [12].Tensile strength and hardness in the friction stir weld materialsmade in air and submerged water have also been studied whileinvestigating the effect of quenching rate in the joint and weldresponse parameters. It was observed that, in the range of para-meters tested, submerged welds show improvement in tensilestrength and elongation [13]. However, mechanical characteristicsof friction-stir welded material at elevated strain rates havereceived no attention in published works.

RD

TD

Weld zone Base Metal

Fig. 1. Orientation of specimens extracted from FSW process zone and base metal.

Fig. 2. (a) A typical thin walled tubular specimen with the dimensions shown in(b).

S. Mallon et al. / Materials Science & Engineering A 639 (2015) 280–287 281

The present work characterizes the mechanical behavior of7050-T7651 aluminum alloy in as received and as friction stirwelded conditions at effective strain rates of 300–900 s�1 with theuse of a torsional Hopkinson bar apparatus (TSHB). The TSHB hasproved to be a robust piece of equipment, having been used suc-cessfully to test a wide variety of materials including composites,aluminum, stainless steel, and the super alloy Inconel at elevatedstrain rates [14–21]. The dynamic deformation response of thespecimens in this work was compared with the quasi-staticresults, as well. The main reason behind comparing the quasi-static results with the dynamic results was to investigate whetherthe loading rate has any significant influence on the mechanicalresponse and strain hardening rate of friction stir processedmaterial compared with base metal.

2. Material and specimen geometry

Aluminum alloy 7050-T7651 plate manufactured by KaiserAluminum (California, US) was used as the base metal for thepresent study. The friction stir welding process was used to jointwo 32 mm thick plates with a butt joint, producing a process zoneapproximately 19 mm wide at the root and 28 mm wide at thecrown. The welding process was carried out using an MTS frictionstir welding machine (Eden Prairie, Minnesota, USA) on the USCFSW Process Development System, on which further detail can befound in the literature [11,12]. This allows for specimens to beextracted exclusively from the large process zone, producing idealconditions in which the FSW modified material is the sole con-stituent of the specimen. Specimens were also extracted from thebase metal in similar fashion. All specimens were extracted withthe specimen axis of symmetry aligned with the rolling and welddirections, as depicted in Fig. 1.

Experiments carried out here utilized tubular, thin-walledspecimens, with dimensions shown in Fig. 2. The gage section is2.41 mm in length, with a nominal wall thickness and mean dia-meter of 0.5 mm and 12.97 mm, respectively. In order to reducethe likelihood of distortion and irregularity in the thin-walled gagelength, cutting stresses during manufacturing were minimized bycarefully selecting the order in which machining processes werecarried out in the manner described next. First the bar is turned(machined on a lathe) to the desired outer diameter, and the gagesection grove was made using a sharp cutting tool. Then the innersection was drilled to a diameter large enough to accommodate aboring bar. Finally, the inside diameter was bored to the desireddimension in very small increments. All machining processes wereperformed on CNC equipment. Specialized gage blocks were fab-ricated so that the outer diameter of the thin-walled gage section

could be precisely measured, accounting for any variations inspecimen dimensions.

The grain structure of the base metal and FSW process zonewere observed and analyzed prior to testing. Standard metallo-graphic procedures were followed for each specimen includinggrinding, polishing, and etching. Keller's etchant solution wasutilized to reveal the grain structure of the material, and theaverage grain size was measured by determining the number ofgrains intersecting a given length of a random line [22]. Thefracture surfaces of the failed specimens were also observed usingscanning electron microscopy (SEM).

3. Experimental apparatus and procedure

3.1. Apparatus

The experimental apparatus utilized in the current work,depicted in Fig. 3, is a stored torque torsional split Hopkinson bar.The apparatus consists of an incident and a transmitter bar, eachcomprised of a 25.4 mm in diameter 7075-T6 aluminum rod, inaddition to a rotary actuator and a quick release clampingmechanism. A specimen is affixed between the incident andtransmitter bar with the use of a high strength epoxy. Thesecomponents are suspended by Teflon bearings atop an I-beamserving as a supporting frame. Torsional stress waves are gener-ated in the incident bar by the instantaneous release of a torquethat is elastically stored in a segment of the incident bar. Externaltorque is applied with the use of a rotary hydraulic actuator fas-tened to free end of the incident bar, in conjunction with a quick-release hydraulic clamping mechanism. The rotary actuator is usedto impose a predetermined rotation at the end of the incident barwhilst the bar is held stationary by the clamping mechanism, thuselastically straining the portion of the incident bar between the

Fig. 3. (a) Schematic and (b) photograph of the TSHB apparatus used in the presentwork. All dimensions in mm.

S. Mallon et al. / Materials Science & Engineering A 639 (2015) 280–287282

actuator and clamping mechanism, wherein the initial torque is“stored”. The clamp is then instantaneously released, allowing thestored torsional elastic strain energy to propagate along the inci-dent bar in the direction of the specimen and transmitter bar. Theclamping and releasing mechanisms rely on the fracture of anotched bolt fashioned from 2024-T3 aluminum to provide theinstantaneous release of the incident bar that is essential to thestored torque torsional Hopkinson bar. The stored torque propa-gates in the form of a square wave, or pulse, imposing a state ofhigh rate torsional loading upon reaching the specimen.

3.2. Data acquisition

Experimental data is acquired with the use of strain gagesmounted on the incident and transmitter bars in conjunction withan oscilloscope. The oscilloscope is also used to trigger dataacquisition. Signal conditioning amplifiers are employed to max-imize precision in the obtained strain measurements. Two straingages are mounted on each bar, diametrically opposed to eachother, to negate the presence of small amounts of bending duringloading. The strain gages are mounted in the center of each bar toavoid the overlap of waves reflected back from the end of the bars.

3.3. Experimental procedure

Experiments are carried out by first affixing a specimenbetween the incident and transmitter bars with the use of a high-strength epoxy. The epoxy is then allowed to cure for several hoursto ensure an adequate bond. The clamping mechanism is thenreadied by installing an unbroken, notched bolt into the jaws ofthe vise. Clamping pressure is applied to the incident bar, and therotary actuator is then used to apply a predetermined rotation to

the free end of the incident bar; thus elastically storing a torque inthe portion of the incident bar between the rotary actuator andclamping mechanism. This torque is then instantaneously releasedby further applying clamping force, causing the notched bolt tofracture and the clamping jaws to release the incident bar. A tor-sional stress wave then propagates towards the specimen, dyna-mically loading the specimen in torsion. Once the wave reachesthe specimen, some portion of the wave is transmitted throughthe specimen and into the transmitter bar, while the remainder ofthe wave is reflected back to the incident bar. These stress wavescan be recorded using strain gages attached into the incident andtransmitter bars. A number of experiments at different strain rateswere conducted using specimens extracted from both base metaland friction stir weld material. The strain rate was altered byvarying the predetermined rotation angle at the end of the inci-dent bar.

3.4. Data analysis

Detail theoretical background and analysis of the torsional splitHokinson pressure bar technique can be found elsewhere [14–21],but is presented briefly here for completeness.

The average strain rate in the specimen can be expressed interms of stress wave measured in the incident and transmitter baras shown in Eq. (1) [19]:

⎡⎣ ⎤⎦tC DL D

t t t1S

S S

SI R Tγ γ γ γ( ) = ( ) − ( ) − ( )

( )∘

where Ds and Ls are gage mean diameter and gage length of thespecimen, respectively, and Cs is the shear wave velocity of the bar,D is the diameter of the bar.

By integrating Eq. (1) with respect to time, the specimen strainas a function of time can be expressed as

⎡⎣ ⎤⎦tC DL D

t t t dt2

2SS S

S

t

T I R0

∫γ γ γ γ( ) = ( ) − ( ) + ( )( )

the average shear stress present in the gage section of thespecimen can be expressed as

tT t

D t

2

3S

S

S S2

τπ

( ) =( )

( )

where ts is the gage thickness of the specimen, and Ts is theaverage of the torque exerted on the specimen interface with theincident bar, T1, and the torque exerted on the specimen interfacewith the transmitter bar, T2. The torque at the specimen interfacewith the incident bar can be calculated from stress waves in theincident bar, as well as the shear modulus and diameter of the barusing Eq. (4):

⎡⎣ ⎤⎦T tGD

t t16 4I R1

3π γ γ( ) = ( ) + ( ) ( )

similarly, the torque present at the specimen interface with thetransmitter bar is given as

⎡⎣ ⎤⎦T tGD

t16 5T2

3π γ( ) = ( ) ( )

by substituting Eqs. (4) and (5) into Eq. (3), the shear stress inthe specimen can be expressed directly as a function of the strainwaves present in the incident and transmitter bars as

⎡⎣ ⎤⎦tGDD t

t t t16 6

SS S

I R T

3

2τ γ γ γ( ) = ( ) + ( ) + ( )

( )

The torques T1 and T2 can be of additional utility whenattempting to quantify the presence of stress equilibrium in the

Fig. 4. Optical micrograph of (a) friction-stir weld fusion zone and (b) base metal.

S. Mallon et al. / Materials Science & Engineering A 639 (2015) 280–287 283

specimen. The torque ratio, T1/T2, can be observed and interpretedto represent the degree to which stress equilibrium has beenachieved in the specimen, with a ratio of unity representing per-fect equilibrium. As aforementioned, it is often impractical for oneto expect such perfect equilibrium in a typical test. However, atorque ratio as near as possible to unity is desirable. Furtherinformation regarding principles of operation and theory of theTSHB apparatus can be obtained elsewhere in the literature [21].

Fig. 5. Typical strain histories obtained from strain gage pairs on incident andtransmitter bars.

4. Results and discussion

4.1. Microstructural observations

The microstructure of specimens extracted from the base metaland FSW process zone are shown in Fig. 4. The grain structures aredifferent in size and morphology. The FSW material is character-ized by a highly regular structure of small, equiaxed grains with amean grain size of 10 mm. In contrast, the base metal exhibitsirregularity in grain size and structure, with a much larger averagegrain size of 30 mm and a substantial elongation in the rollingdirection. Based on these measurements, approximately 17 grainsexist across the gage thickness of the typical base metal specimen,while approximately 52 grains exist across the gage thickness inthe typical FSW specimen. In either case, the corresponding ratioof grain size to wall thickness is much larger than the 1:7 ratiopresent in Al 1100-0 specimens used successfully by Duffy et al.[15]. Based on these observations, it is clear that a more thanadequate representative volume is present in the gage sections ofthe specimens used in the present work.

4.2. Wave analysis

The typical strain time histories obtained from the gage pairsmounted on the incident and transmitter bars are given in Fig. 5.The obtained wave forms are nearly ideal, characterized by shortrise times, long durations at a constant magnitude, and only smallrounded peaks present in the incident pulse immediately after risetime. The relation given in Eq. (2) is exemplified in the typicalstrain histories in Fig. 5, evident in the prominent increase in themagnitude of the reflected strain pulse, corresponding to theinstance at which the transmitted pulse diminishes.

Typical incident, reflected, and transmitted strain pulses takenfrom the strain histories are shown in Fig. 6a. It is worthwhile tonote that the duration of the transmitted pulse is also the durationof the experiment. This is clear from the fact that once the speci-men has failed incident strain pulses can no longer propagatethrough the specimen into the transmitter bar. The typical pulsesshown in Fig. 6a depict an experimental duration beginning andending at approximately 100 μs and 300 μs, respectively. Theexperimental duration is highly dependent upon strain rate,typically decreasing with increasing strain rate.

The specimen strain rate can then be found from these pulsesby applying Eq. (2). A typical evolution of specimen strain rate isgiven in Fig. 6b. Again, the evolution of specimen strain rateapproaches the ideal, with a nearly constant strain rate present inthe specimen throughout the entire duration of the experiment.

As previously discussed, the ratio of the torques applied toeither side of the specimen is generally of significant interest. Inthe present work, the so-called torque ratio was found to becharacteristic of the apparatus and specimen geometry, with alltests maintaining a nearly constant value of approximately 0.85during the entirety of the experimental duration, regardless ofstrain rates present in the specimen. The torque ratio approachingunity present in all tests discussed in the current work suggeststhe presence of near ideal stress equilibrium in the specimen; avital assumption that must be met in any TSHB test [15].

4.3. Quasi-static stress–strain behavior

Quasi-static tensile tests were performed to evaluate thestress–strain response of both the base and FSW materials. Tensiletests were carried out based on ASTM-E8 on flat specimens of50.8 mm gauge length, 12.7 mm width, 4.1 mm thickness and6.35 mm fillet radius. The tensile specimens were extracted alongthe rolling direction and weld line, for base and FSW specimens,respectively. Stress–strain relations obtained from this testing aregiven in Fig. 7. The alloy is shown to exhibit obvious differences inmechanical behavior between these two states, as one wouldexpect from the different microstructures and processing. Mostnotably, the FSW material state experiences a relatively largeamount of work hardening while also exhibiting a lower yield andmaximum flow stress. In contrast, the base metal exhibits a muchhigher yield stress, with less work hardening. Considering the

Fig. 6. (a) Typical incident, reflected, and transmitted strain pulses and (b) typicalshear strain rate developed in the specimen, as a function of time.

Fig. 7. Stress–strain curves obtained from quasi-static tensile tests for FSW andbase metal.

S. Mallon et al. / Materials Science & Engineering A 639 (2015) 280–287284

substantially smaller grain size present in the FSW specimen,higher yield strength might be expected, following the well-established Hall–Petch relation [23,24]. However, in alloys like

7050, the primary strengthening mechanism is precipitationhardening. The base metal, in the T7651 condition has a largepopulation of η′ precipitates which act as barriers to dislocationmotion. The FSW process, on the other hand, subjects the materialin the weld nugget to the equivalent of a solution heat treatmentalong with complete recrystallization (as evidenced by the refined,equiaxed grain structure). Hence, the precipitate content in theweld nugget is low and approaches zero, although subsequentartificial aging would result in substantial recovery of precipitationhardening [12,13,25,26]. This material condition is similar to thatof a “W” temper. Accordingly, it is evident that in quasi-staticloading conditions, the principal strengthening mechanism for thebase material is the precipitation hardening. The differences inwork hardening between the two conditions also arise from thisdifference in precipitate content [12]. Note that the large popula-tion of η′ precipitates present within the structure of the basemetal can give rise to the saturation of dislocation density at afaster pace compared with the precipitate-free FSW specimen.Such fast saturation of dislocation density shows itself as adecreasing work hardening rate, particularly at larger strainmagnitudes.

4.4. Dynamic stress–strain behavior

Stress–strain curves obtained at elevated strain rates are pre-sented for FSW and as-received material in Fig. 8. These resultingstress–strain curves are given in terms of effective stress, strain,and strain rate using the relations in Eq. (7) in order to facilitatecomparison with quasi-static tensile results.

3 7-aσ τ¯ = ( )

3 7-bε γ¯ =

( )

The dynamic curves are found to lie above those obtained fromquasi-static conditions in all cases, indicating some appreciablestrain rate dependence in both material states. Stress levelsincrease gradually with increasing strain rate. Stress levels indynamic tests on specimens extracted from FSW material appearto be slightly more rate dependent when compared to similar testson base metal. For both material states, the rate of work hardeningobtained from dynamic torsional testing is observed to be lessthan that found in the corresponding quasi-static case. The FSWmaterial continues to exhibit lower yield and plastic flow stresseswhen compared to the base metal at elevated strain rates. Thestress–strain relations obtained experimentally at elevated strainrates for this alloy were fitted using the Holloman equation takingthe form depicted as follows:

K 8nσ ε¯ = ¯ ( )

where K and n are the strength coefficient and strain hardeningexponents, respectively. This type of fitting yields curves havinghigh coefficients of determination with respect to the experi-mentally obtained dynamic data, typically 0.96 and above. Itshould be made clear that the fitted curves shown in Fig. 9, areproposed only to facilitate analysis within the range of parametersexplored experimentally.

Note that the work hardening rates (θ) are extracted from thefitted stress–strain curves based on Eq. (9) and are presented inFig. 10.

dd 9θ σ

ε= ¯

¯ ( )

In all instances, the FSW material is seen to exhibit sub-stantially higher work hardening than the base metal. This again

363 (s-1)

5×10-4 (s-1)

639 (s-1)841 (s-1)

FSW

5×10-4 (s-1)

605 (s-1) 718 (s-1)

326 (s-1)

Base Metal

Fig. 8. Stress–strain curves experimentally obtained from dynamic torsion tests.

FSW

841 (s-1)

639 (s-1)

363 (s-1)

Base Metal

718 (s-1)

605 (s-1)

326 (s-1)

Fig. 9. Fitted stress–strain curves obtained using experimental data.

S. Mallon et al. / Materials Science & Engineering A 639 (2015) 280–287 285

can be explained through the role of precipitates existing withinthe microstructure of the base metal. The nanometer-sizedcoherent precipitates in the base metal can create barriers to thedislocation motion. Such barriers can act as dislocation generatingsources, increasing the dislocation density at a fast pace. Theincrease in the dislocation density is followed by a saturation,which is revealed as steady–state flow stress (see Fig. 9). On theother hand, the precipitate-free friction stir processed specimensshow a more noticeable strain hardening rate which can be due tothe relatively delayed achievement of dislocation density satura-tion. Furthermore, increasing strain rates appear to lower the workhardening rate in the FSW material, where work hardening ratesare seen to converge in all high strain rate tests to a value that is

approximately 60% of that observed in the quasi-static case. Thebase metal does not follow this trend, with some increase in workhardening rate occurring at elevated strain rates when comparedwith the quasi-static case. The dynamic tests exhibit some con-vergence at only slightly elevated work hardening rates for thebase metal. All work hardening rates are seen to decrease withincreasing strain. The higher strain hardening rates in the FSWmaterial are consistent with typical observations for W vs. T7 or T4vs. T6 type tempers and are due to different strengtheningmechanisms. For example, in the W-temper, the main strength-ening mechanisms are grain boundary strengthening, work hard-ening via production of sessile forest dislocations which act asbarriers to subsequent dislocation motion, and solid solution

FSW

(s-1)

(s-1)

(s-1)

Base Metal

(s-1)

(s-1)

(s-1)

Fig. 10. Work hardening rates as a function of effective strain (extracted using fitted stress–strain curves).

S. Mallon et al. / Materials Science & Engineering A 639 (2015) 280–287286

strengthening. In TXXX type tempers, the primary strengtheningmechanism is precipitation hardening as the distance betweenprecipitates forming dislocation barriers is lower than for theother boundaries and much of the solute has been depleted inorder to form the precipitates.

In explaining the decrease in work hardening observed in theFSW material as strain rate increases, the dislocation annihilationphenomenon must be considered. This phenomenon occurs par-ticularly in metals with high stacking fault energy (SFE), such asaluminum alloys, for which recovery is the principal restorationmechanism [27,28]. At high strain rates, dislocation velocity ishigher, thus the dislocation density will increase at a higher rate,particularly at grain boundaries. With greater dislocation densitiesat grain boundaries, the chance of dislocation annihilation at thesedislocation pile-up regions will increase. This might decrease theoverall dislocation density in the material. As a consequence ofthis lower overall dislocation density, work hardening would beseen to decrease at elevated strain rates. Such a mechanism isactually evident in the experimental data, wherein dynamicexperiments tend to achieve a plateau in the stress–strain curve atstrain levels much lower than those at which plateaus areachieved in similar quasi-static tests.

The overall outcome of the study can be summarized fromthree main known material strengthening mechanisms, grain sizecontrol, precipitation hardening and strain rate hardening.

In most metals, grain size refinement plays an influential role instrengthening mechanism of the material. However, merely thegrain size refinement might not result in the desired strengtheningeffect. In our study, in the case of both quasi-static and dynamicloading conditions, though the FSW specimen has smaller grainstructure than the base material, its yield strength was lowercompared with the base metal. This behavior can be related to themore substantial role of the η′ precipitates present within the basemetal, hindering the dislocation motion and consequentlyincreasing the strength of the base metal in the as-received state.

The work hardening rate of the FSW processed specimens inthis work was found to be 100% higher than the base metal. Thiscould be due to the fact that the FSW process dissolves the pre-cipitates, diminishing the influence of precipitation hardening

mechanism in the material. This phenomenon not only reducesthe overall strength of the weld metal compared with the basemetal, but also influences the rate of dislocation density satura-tion, such that the strain hardening rate in the precipitate-freeFSW specimen would be significantly greater than that of the basemetal.

4.5. Fractography

Fracture surfaces present on FSW and base metal specimenstested at the highest strain rates achieved were examined with theuse of a scanning electron microscope (SEM). Images of thesefracture surfaces are presented in Fig. 11. Voids are apparent in thefracture surface of both specimen types, indicative of a ductilefailure mechanism [16,29]. This observation is bolstered by thelarge amounts of strain achieved prior to failure in all tests. Voidsize and spacing is likely correlated with constituent particlespacing. In the FSW material, these constituents are typicallybroken up, such that there are a larger number of smaller con-stituent particles present in the material when compared to thebase metal. The shape of a typical void appears to be elongated inthe direction of the applied shear stress in both specimen types, asis illustrated in the images with the use of arrows.

5. Conclusions

A TSHB apparatus has been used to perform high strain ratetesting on aluminum alloy 7050T-6571 in both the as-received andfriction-stir welded conditions. Quasi-static tensile tests have beenperformed for reference on both material states. The micro-structures of each state are observed to be different: the FSW statecontaining small, equiaxed grain structure while the base metalconsists of an irregular and much larger grain structure. Althoughthe FSW material consists of a significantly smaller grain structure,the overall strength of the base metal, which incorporates fine η′precipitates, is significantly higher in both quasi-static anddynamic loading conditions. This indicated the substantial influ-ence of precipitation hardening mechanism in this material. In

Fig. 11. (a) A typical fractured specimen, indicating that the failure has taken place from the gage section, with high magnification SEM images of fracture surfaces of (b) abase metal specimen and (c) a FSW material specimen tested at high strain loading.

S. Mallon et al. / Materials Science & Engineering A 639 (2015) 280–287 287

addition, the role of precipitates was also highlighted in the workhardening response of the two material states. The distribution offine η′ precipitates within the as-received base metal can attributeto the early saturation of dislocation density in the material,indicated by steady-state flow stress and lower work hardeningrates when compared with precipitate-free friction stir weldedspecimens. The FSW material exhibits a decreased propensity forwork hardening at elevated strain rates. Such response can berelated to the greater chance of dislocation annihilation at elevatedstrain rates, in which increased dislocation velocity can result indislocation pile-ups to be more rapidly formed in the grainboundaries followed by dynamic recovery. SEM observations offracture surfaces present on specimens tested at elevated strainrates depict the presence of voids in both material states, havingdiameters commensurate with grain sized. Elongation is alsoapparent in these voids, occurring in the direction of applied shearstress. The presence of voids suggests the material remains ductilein both the FSW and base metal states at dynamic strain rates.

Acknowledgment

The financial support from the University of South Carolina'sResearch Office, the College of Engineering and Computing and theDepartment of Mechanical Engineering is acknowledged.

References

[1] A. Heinz, A. Haszler, C. Keidel, S. Moldenhauer, Mater. Sci. Eng. A 280 (2000)102–107.

[2] J.P. Immarigeon, R.T. Holt, A.K. Koul, L. Zhao, W. Wallace, J.C. Beddoes, Mater.Charact. 35 (1995) 41–46.

[3] Y. Wu, F. Froes, A. Alvarez, C. Li, J. Liu, Mater. Des. 18 (1997) 211–215.[4] M.R. Hill, T.L. Panontin, Eng. Fract. Mech. 69 (2002) 2163–2186.

[5] J. Li, F. Li, F. Xue, J. Cai, B. Chen, Mater. Des. 37 (2012) 491–499.[6] J. Li, F. Li, M. He, F. Xue, M. Zhang, C. Wang, Mater. Des. 40 (2012) 176–184.[7] N.M. Han, X.M. Zhang, S.D. Liu, D.G. He, R. Zhang, J. Alloy. Compd. 509 (2011)

4138–4145.[8] X. Ren, L. Zhang, Y. Chen, F. Sun, W. Zhang, L. Lin, Y. Wang, Procedia Eng. 27

(2012) 914–922.[9] L.J. Zheng, H.X. Li, M.F. Hashmi, C.Q. Chen, Y. Zhang, M.G. Zeng, J. Mater. Pro-

cess. Technol. 171 (2006) 100–107.[10] N. Mendes, A. Loureiro, C. Martins, P. Neto, J.N. Pires, Mater. Des. 58 (2014)

457–464.[11] W.M. Thomas, E.D. Nicholas, Mater. Des. 18 (1997) 269–273.[12] C.T. Canaday, M.A. Moore, W. Tang, A.P. Reynolds, Mater. Sci. Eng. A 559 (2013)

678–682.[13] P. Upadhyay, A.P. Reynolds, Mater. Sci. Eng. A 527 (2010) 1537–1543.[14] N.K. Naik, A. Asmelash, V.R. Kavala, V. Ch, Mech. Mater. 39 (2007) 1043–1052.[15] J. Duffy, J.D. Campbell, R.H. Hawley, J. Appl. Mech. 38 (1971) 83–91.[16] W.S. Lee, J.I. Cheng, C.F. Lin, Mater. Sci. Eng. A 381 (2004) 206–215.[17] T. Kobayashi, J.W. Simons, C.S. Brown, D.A. Shockey, Int. J. Impact Eng. 35

(2008) 389–396.[18] A. Kidane, J.A. Wang, A new method for dynamic fracture toughness deter-

mination using torsion Hopkinson pressure bar, dynamic behavior of materi-als, in: Proceedings of Society for Experimental Mechanics Series, vol. 1, 2014,pp. 307–312.

[19] J.A. Wang, A. Kidane, A new approach to determine the quasi-static anddynamic fracture toughness of engineering materials, dynamic behavior ofmaterials, in: Proceedings of Society for Experimental Mechanics Series, vol. 1,2013, pp. 545–551.

[20] A. Gilat, C.S. Cheng, Int. J. Plast. 18 (2002) 787–799.[21] A. Gilat, Torsional Kolsky bar testing, in: H. Kuhn, D. Medlin (Eds.), ASM

Handbook Volume 8: Mechanical Testing and Evaluation, ASM, Ohio, 2000,pp. 505–515.

[22] R.L. Higginson, C.M. Sellars, Worked Examples in Quantitative Metallography,Maney Publishing, UK, 2003.

[23] Y.S. Sato, M. Urata, H. Kokawa, K. Ikeda, Mater. Sci. Eng. A 354 (2003) 298–305.[24] N. Hansen, Scr. Mater. 51 (2004) 801–806.[25] N. Kamp, A.P. Reynolds, J.D. Robson, Sci. Technol. Weld. Join. 14 (7) (2009)

589–596.[26] J. Yan, A.P. Reynolds, Sci. Technol. Weld. Join. 14 (4) (2009) 582–587.[27] F. Barlat, M.V. Glazov, J.C. Brem, J.C. Lege, Int. J. Plast. 18 (2002) 919–939.[28] F.J. Humphreys, M. Hatherly, Recrystallization and Related Annealing Phe-

nomena, 2nd ed., Elsevier, UK, 2004.[29] D. Hull, Fractography – Observing, Measuring and Interpreting Fracture Sur-

face Topography, Cambridge University Press, UK, 1999.