Superplasticity of a friction stir processed overaged WE54 magnesium alloy

Oscr A.Runo,Mrt Álvrez-Lel,b,Alberto Orozco-Cbllero,c,Fernndo Crreño,∗

a Physical Metallurgy Department,CENIM-CSIC,Av.Gregorio del Amo 8,28040 Madrid,Spain

b Now at: Department of R&D and programmes,Technology Centre of Metal-Mechanical and Transport (CETEMET).Avda.1° de Mayo,s/n.Linares,Jaén,2370,Spain

c Now at: Department of Mechanical Engineering,Chemistry and Industrial Design,Polytechnic University of Madrid,Ronda de Valencia 3,28012,Madrid,Spain

Abstract The coarse-grained WE54 magnesium alloy was heat treated in order to have minimum hardness minimizing the effects of precipitates and solid solution.Friction stir processing (FSP) was applied in severe conditions to obtain fine equiaxed and highly misoriented grains,with grain sizes even less than 1 μm.The high severity of processing demonstrated to have a strong impact in the microstructure.Consequently,the processed materials exhibited excellent superplasticity at the high strain rate 10−2 s−1,and temperatures between 300 and 400 °C.The maximum tensile superplastic elongation of 756% was achieved at 400 °C thanks to the operation of grain boundary sliding mechanism(GBS).Besides the new data obtained through tensile testing,the paper deals with a transcendental question regarding the large differences in strain rate values at a given stress in the superplastic regime at maximum elongation compared to other magnesium-based alloys.With this is mind,19 magnesium alloys from 22 different investigations were analyzed to give some light to this behavior that never was treated before.It is proposed that this behavior has to be attributed to the accommodation process,necessary for GBS to occur,which is hindered by reinforcing solutes.

Keywords: WE54 magnesium alloy;Friction stir processing;Processing severity;Superplasticity;Grain boundary sliding;Solutes.

1.Introduction

Magnesium alloys have been used for light weighted structural parts for almost a century.However,its use has not become widespread due to its limited mechanical properties,narrowing its applications to non-critical structural parts[1,2].

The advantages of magnesium alloys are their low density,high specifi strength and high castability,making them appropriated for the fabrication of simple geometry components,especially for car seats and wheels in the automotive industry,as well as in the aeronautic industry for lightweight airframes [3].However,these applications have to take into account the disadvantages of the low ductility related to the relative low number of easy slip planes of the hexagonal crystal lattice,and the associated anisotropy caused by the intense texture obtained during forming processes,such as extrusion or sheet forming [1,4].Furthermore,the susceptibility to corrosion limits its applicability [5].

Magnesium creep behavior is an important issue for its application considering its low melting temperature (Tm).For instance,temperatures as low as 100°C,experienced by many components in service,are equivalent to 0.4 Tmwhich may trigger softening creep mechanisms.

The fundamental studies of creep in pure magnesium take for granted dislocation or slip creep where climb is the diffusion-controlled deformation mechanism.However,other mechanisms may control deformation in these alloys,especially when containing high fraction of grain boundaries as in the case of grain boundary sliding,GBS.In the case of slip creep,low creep rate is desired with very low creep strains,whereas for superplasticity,high strain rates and low fl w stresses are desirable.The nature of these creep mechanisms is different but both will compete to control deformation.

Generally,the creep rate or strain rate,˙ε,is influence by the stress,σ,and the temperature,T,through the relation[6,7]:

where f(s)is a function of the microstructure,E is the average unrelaxed polycrystalline Young’s modulus,n is the stress exponent,Q is the activation energy for plastic fl w,and R the universal gas constant.

The function f(s) represents mainly the influenc of grain size,L,but it may represent the subgrain size or the dislocation density,depending on the overall controlling creep mechanism.Generally,over a certain temperature range,Q is constant and is related to each deformation mechanism.

Apart from climb-controlled slip creep,other independent creep mechanisms may control deformation.In fine-graine materials these mechanisms are grain boundary sliding accommodated by slip [8,9] and directional diffusional fl w[10–12].In coarse grain materials containing atoms in solution,which is the case in many magnesium alloys,also solute-drag creep may control deformation[13,14].These four mechanisms are independent of each other,are controlled by atom diffusion,and can be described by Eq.(1) using proper activation energies and stress exponents.The activation energies may be equal to that for lattice self-diffusion,QL(dominant at high temperatures),grain boundary diffusion,Qgb(at intermediate-low temperatures),or solute diffusion,Qs(depending on atoms in solid solution).In the particular case of this fin grained WE54 material,apart from grain boundary sliding,slip creep and solute-drag creep are competing for being rate controlling in the same ranges of temperature and strain rate of the present work.Regarding the stress exponents,plastic deformation by climb-controlled slip creep is associated ton= 5 or higher,grain boundary sliding accommodated by slip byn= 2,diffusional fl w byn= 1 and solute-drag creep byn= 3 [15].

On one hand,in order to increase both low temperature strength and high temperature creep resistance,alloying with transition and rare earth elements such as Ni,Mn,Y,Zn,Zr,and Ce,Dy,Gd,Nd,respectively,is usual [16–21].On the other hand,adequate thermomechanical processing and/or severe plastic deformation also increase room temperature strength and ductility,as well as attaining high superplastic response at high temperatures thanks to grain refinin [16–37].A recent interesting technology able to modify the microstructure of metallic materials is friction stir welding and/or processing (FSW/FSP),which it has been applied successfully to magnesium alloys [19,23–24,28,37].Specificall ,by per-forming friction stir processing,it is possible to attain very fine equiaxed and highly misoriented microstructures,prone to superplastic deformation at high temperatures.

Therefore,in this work,we used FSP to achieve an ultrafin microstructure and superplastic properties in an overaged highly alloyed WE54 magnesium alloy.Our overaged alloy contrasts with the T6 treated initial WE54 alloy [14,38,39]having minimum hardness by minimizing fin precipitation and solid solution hardening.Our aim was to process the alloy with the maximum possible severity,i.e.,with the lowest heat input and maximum cooling rate to assess high strain rate superplasticity in a wide superplastic window.Therefore,a clear processing-microstructure-mechanical behavior correlation is established: severe FSP refine drastically the initial coarse-grained microstructure to obtain a deformation mechanism change from solute-drag creep to superplastic grain boundary sliding (GBS) mechanism.Additionally,we investigated the alloy creep behavior with especial attention to the involved deformation mechanisms,and its comparison with other WE54 alloys and with other superplastic magnesium alloys.With this goal in mind,19 magnesium alloys from 22 different investigations [16–37] have been analyzed to draw homogenizing conclusions on superplastic deformation behavior of magnesium alloys.

2.Material and experimental procedure

Fig.1 shows the experimental-fl w diagram followed in this research.

2.1.Material

The WE54 alloy was received in the form of extruded plates with dimensions 300×80×5 mm3in the T6 temper(solution-heat treating at 525 °C for 8 h,followed by hot water quenching (60 °C) and fina aging at 250 °C for 12 h)[38].The composition of the alloy is given in Table 1.

Table 1 The composition of the magnesium WE54 alloy,in wt.%,balance Mg.

The initial WE54-T6 alloy was subjected to different temperatures during different times until the minimum hardness was obtained.This was reached by a solution-heat treatment at 500 °C for 1 h and slow cooling at 1 °C/min [14].The material subjected to such temper was named as WE54-TT.The mechanical properties of the coarse-grained WE54 alloy are also given at [14].

2.2.Material processing

The material in the TT condition was processed by FSP.The tool was made of a MP159 nickel superalloy with scrolled shoulder 9.5 mm in diameter and a concentric threaded conical pin with flute 4.7-4.1 mm in diameter and 1.8 mm in length [39].The FSP machine was a MTS PDS-4 Intelligent-Stir specially designed to perform FSW/P with fi e degrees of freedom.The tool was tilted 1.5° to perform FSP of the samples.A key aspect of the present research was to process the materials under different severity conditions,aiming ultra-fin grain sizes at the highest severity conditions.In order to reach this goal,the material was processed using low heat input (HI) values through low tool rotation speed (r orω) and high traverse speed (v),being HI∝ω2/v [40–42].Two different backing plates,one made of steel and other one made of copper,were used to obtain different processing temperatures and cooling rates.The copper plate contains a series of cavities where liquid nitrogen fl ws.Before testing,the steel backing plate was at room temperature while the copper one reached a temperature of about −60 °C facilitating a high heat extraction during FSP and limiting grain growth.The processing conditions,combining different traverse speeds and rotational speeds are given in Table 2.The specifi cooling conditions are named refrigerated and nonrefrigerated for the copper and steel backing plate,preceded by the letter R or N,respectively.

Fig.1.Experimental-fl w diagram followed in this research.

Table 2 Nomenclatures and values of the FSP processing conditions.

2.3.Material characterization

Samples for a microstructural characterization by electron back-scattered diffraction (EBSD) were prepared by mechanical polishing up to a fina 1 μm diamond particles step,followed by chemical polishing for 5 to 10 s using a solution of 4.2 g picric acid,10 mL acetic acid,10 mL H2O and 70 mL ethanol.Color-coded inverse pole figur (IPF) maps in orientation imaging microscopy (OIM) were drawn according to the orientations corresponding to the unit triangle of the inverse pole figure

2.4.Mechanical testing

The mechanical behavior of the alloy was characterized by constant crosshead speed tensile tests (CCST) at initial strain rate of 10−2s−1at different testing temperatures (250–450 °C).Moreover,strain rate changes tensile tests (SRCT)ranging from 10−1to 10−5s−1were performed to characterize the high temperature (300–450 °C) deformation mechanisms.The tests were carried out using a universal Instron 1362 testing machine equipped with a four-lamp ellipsoidal furnace.The samples were heated in 10 min and held 10 min at temperature for the CCST and 20 min for the SRCT tests.Planar dog-bone tensile samples with 6.5 mm gage length × 2 mm width × 1.6 mm thickness gage dimensions were electrodischarge machined so that their longitudinal axis was parallel to the FSP or extrusion direction.

3.Results

3.1.Microstructures

An orientation-imaging map,OIM,of the as-received WE54 Mg alloy is shown in Fig.2.The micrograph was taken in the L plane and reveals coarse equiaxed grains of diameters in the range 50–250 μm.A full description of the microstructure can be found elsewhere [43–45].

Fig.3 shows four orientation-imaging maps of the WE54-TT alloy after FSP in the conditions refrigerated (R) and non-refrigerated (N),14r05v and 10r10v.The micrographs show that the grain sizes of the R10r10v and N10r10v materials,which correspond to the most severe FSP conditions,are fine than the ones observed for the less severe conditions,R14r05v and N14r05v,respectively.Note that the HI values corresponding to the 14r05v conditions are four times higher than those for the 10r10v conditions.It is also worth noting that this last traverse speed,1000 mm/min,is very fast and therefore might be appropriated for industrial processing.A grain size analysis from the EBSD measurements for every FSP processing condition was performed.The equivalent Feret diameters,L,are given in Table 3 including their lognormal confidenc intervals,also plotted in Fig.4.This fig ure presents histograms for the distribution of the logarithm(base 10) of grain sizes (L,in μm) in the R14r05v,N14r05v,R10r10v and N10r10v conditions.The figur shows fin grain sizes,even below 1 μm,that are about two orders of magnitude fine than those before FSP.

Fig.2.Orientation imaging map(OIM)obtained by EBSD of the as-received WE54 Mg alloy.

Fig.3.Orientation imaging maps of the WE54-TT alloy (EBSD inverse pole figur maps) after FSP in the conditions refrigerated and non-refrigerated,14r05v and 10r10v.

Table 3 Grain size,L (μm),from EBSD measurements for the extreme FSP conditions of the WE54-TT alloy together with their confidenc intervals asuming a lognormal distribution (all values in μm).

Fig.4.Histograms for the distribution of grain sizes (in decimal logarithm)of the WE54-TT FSP materials.Grain size,L,is given in μm.

Fig.5.Stress-strain curves of materials procesed by FSP,non-refrigerated(N,left) and refrigerated (R,right),and tested at an initial strain rate of 10−2 s−1 at temperatures from 250 to 450 °C.

3.2.Tensile tests to rupture of WE54 alloy processed by FSP

Fig.5 shows tensile true stress-true strain curves at initial strain rates of 10−2s−1and temperatures 250–450 °C of the WE54-TT alloy processed by FSP under three FSP conditions and two cooling conditions: refrigerated (R) and non-refrigerated (N).

Table 4 provides the mechanical parameters corresponding to the tensile test to failure in the temperature range 250–450 °C for all processing and cooling conditions.The fl w stress values refer to the maximum value of the true stresstrue strain curves obtained.It is remarkable the high elongation values reached at 400 °C for all the refrigerated materials as well as the most severe processing condition,N10r10v,at such high strain rate.The maximum elongation to failure,756%,was reached by the N10r10v condition.Nonetheless,very high values,such as 640 and 703%,are also obtained at 400 °C in refrigerated conditions.It is also interesting to note that very high elongations are also obtained at 350 °C,such as 341,366 and 455%,and 210% at 300 °C for the R10r10v condition.On the other side,at 450 °C there is a clear loss of ductility for all materials,which is attributed to massive grain growth,and it will be discussed in the next section.This behavior is clearly observed in Fig.6.Inversely to the evolution of the elongation to failure,the fl w stress decreases with increasing temperature,except at the highest temperature,450 °C,which is clearly evidenced in Fig.7.Both behaviors,ductility and fl w stress,reflec the changes in deformation mechanisms with increasing temperature,showing that ultrafin microstructures obtained by FSP are stable up to 400 °C.

Table 4 Yield stress (σ0.2),fl w stress (σ),uniform elongation (eu) and elongation to failure (eF) at different test temperatures of the alloy WE54-TT processed by FSP and tensile tested at 10−2 s−1.

Fig.6.The elongation to failure as a function of temperature for the FSP WE54-TT alloy tested at 10−2 s−1 from 250 to 450 °C.

Fig.7.The fl w stress as a function of temperature for the severe FSP WE54-TT alloy tested at 10−2 s−1 from 250 to 450 °C.

For the sake of comparison,Figs.6 and 7 incorporate also the elongation to failure and fl w stress of the WE54-TT material before FSP,respectively.The ductility increases steadily with temperature and a clear maximum is observed at 400 °C for all the FSP materials.It is interesting to note that the maximum ductility values at 300 and 350 °C are obtained for the R10r10v condition,the most severe with the lowest heat input,and refrigerated.Additionally,regarding the ductility improvement with respect to the original material,it can be concluded that it is not worth to process on the standard non-refrigerated backing plate except for the most severe condition.This observation points to the fact that this magnesium alloy tends to undergo rapid grain growth at high temperature,which is the reason behind the difficult of obtaining average grain sizes below 1 μm.We were able to diminish such high temperature microstructural instability by using the refrigerated backing anvil and low HI during FSP.Moreover,a lower than expected ductility for some fine-graine superplastic FSP samples (the least severe FSP conditions N14r05v and N10r05v) respect to the overaged non-FSP alloy may be taken as an indication that the accommodation process for GBS is hindered somehow,probably by solutes in solid solution.

At 250 °C most processing conditions show higher fl w stresses than the initial material (Fig.7).Nevertheless,at higher temperatures,the stress strongly decreases for all FSP conditions.The highest stress drop is observed at 400 °C for the most severe processed conditions,which are the N10r10v condition and the refrigerated ones.However,at the highest temperature,450 °C,these alloys experiment a “contrenature” increase of fl w stress towards the value similar to the one observed for the initial WE54-TT.As mentioned for the elongation to failure,this evidences the changein the deformation mechanisms.

Fig.8.SEM micrographs of samples processed at different FSP conditions and tensile tested at 10−2 s−1 and three temperatures.

3.3.Microstructure of samples after the constant strain rate tests

The microstructures obtained after CCST at 10−2s−1are shown in Fig.8 for the 14r05v and 10r10v WE54-TT materials at 350,400 and 450 °C.It should be noted that the aspect of the micrographs is influence by the presence of an oxide layer formed during high temperature testing.Grain coarsening is observed,up to about 4 μm,at 350 and 400 °C.In the case of 450 °C very large grains exceeding 20 μm can be found.This shows that the obtained fin microstructures are relatively stable up to 400 °C and able to attain very large elongations,in excess of 700% (Fig.6).Nevertheless,at 450 °C the microstructure becomes unstable and tends to grow quickly [39].

3.4.Strain rate change tests

˙ε−σdata pairs obtained from the SRCT performed at temperatures ranging 300 to 450 °C and at strain rates in the range 10−1to 10−5s−1are shown in Fig.9.The slope of the curves corresponds to the apparent stress exponent,napor just n along the text,which is indicative of the controlling deformation mechanism.Low n values,close to 2,are observed at low strain rates.However,n values increase in most samples at high strain rates.The materials with values closest ton= 2 are the refrigerated ones,and the N10r10v,processed at the most severe condition.The lowest stress values for a given strain rate,as well as the lowest n values for all materials,are observed at 400 °C.This contrasts with the anomalous behavior observed at 450 °C,where the stress

Fig.9.(a-f) Strain rate as a function of stress data obtained by strain rate change tests for the various FSP WE54-TT alloys,non-refrigerated (N,left)and refrigerated (R,right),and temperatures (300–450 °C).

values are higher than those at 400 °C.Regarding the activation energy values,Q,it is difficul to calculate trustworthy since SRCT data are separated 50 °C in temperature and there are continuous changes on the microstructure and deformation mechanism,leading to a remarkable dispersion of values.Additionally,the influenc of other deformation mechanisms makes the strain rate vs stress curves,at the different temperatures,non-perfectly parallel,and thus,a dispersion of activation energy values arises.Despite such inconveniences,average values around 180 kJ/mol are obtained between 300 and 400 °C,which are somewhat higher than QL.In order to obtain more precise values it is necessary to perform a much larger number of tests and microstructural observations at intermediate temperatures.

3.5.Processing severity

Clearly,as it could be evidenced in previous results,an important aspect that influence the creep behavior of the FSP materials is the severity of their processing,since is the most important factor for microstructural refinemen [39,46–51].Fig.10 shows strain rate vs.stress curves at 400 °C for the refrigerated materials with different processing conditions compared with the initial WE54-TT alloy.Effectively,the severe processing results in a reduction of one order of magnitude on the stress values or two orders of magnitude on the strain rates higher than for the initial alloy.The coarsegrained initial WE54-TT stress exponent (slope in Fig.10) is aboutn= 3,whereas it isn= 2 for the processed materials.This is also an evidence of deformation mechanism changes at high temperatures as result of the extensive refinin produced by the severe FSP processing.

Fig.10.The strain rate as a function of stress at 400 °C for the WE54-TT before FSP and the three refrigerated FSP materials showing the processing influence

4.Discussion

In this work,the initial high hardness WE54-T6 alloy was firstl overaged to minimum hardness (WE54-TT) minimizing fin precipitation and solid solution hardening.With this in mind,easier processing and richer discussion about the influenc of grain size and/or solutes were expected to follow.Effectively,the overaged WE54-TT alloy was successfully refine by FSP attaining ultrafin grains following a strategy consisting in lowering the heat input and increasing cooling rate.Six processing conditions were analyzed using three heat indexes (HI) and two cooling rates as shown in Table 2.The influenc of processing severity on the microstructure and mechanical properties is clear comparing,for instance,the two most extreme conditions analyzed: N14r05v and R10r10v.The last condition is the most severe,having four times lower HI,at a rapid traverse speed,1000 mm/min,and refrigerated on a liquid nitrogen copper anvil.As a result,the R10r10v condition attained fine grains (∼1 μm,Fig.3),more homogeneous grain size histogram (Fig.4) and higher ductility at all temperatures (Fig.5,Table 4).Additionally,the tensile tests to fracture of the six processed conditions are compared with the non-processed coarse-grained one (Figs.6 and 7) showing the differences: i) a pronounced valley at 400 °C in stress values (Fig.7),ii) pronounced ductility peaks at 400 °C (Fig.6) and iii) higher differences the higher the processing severity.This behavior,which contrasts with the unprocessed alloy,pointed to a change in deformation mechanism associated to grain size (Fig.8).Having this idea in mind,SRCT tests were performed in order to determine the operative deformation mechanisms (Fig.9).The figur clearly shows that for the FSP conditions maximum softening is obtained at 400 °C,exhibitingn= 2.However,at 450 °C the stress increases,as well as the stress exponent(towardsn= 3),in most cases,i.e.,towards the unprocessed alloy value.Fig.10 shows a comparison of the creep behavior among the fin vs.coarse-grained (FSP vs.non-processed conditions) at 400 °C,which evidences two distinct deformation mechanisms.The obtained results for the present WE54-TT alloy do not differ much from the FSP WE54-T6 alloy[39] except for the scarce superplastic ductility values of the least severe FSP conditions of the WE54-TT alloy.This result led us to think that a difficul accommodation process may be occurring,probably due to the reinforcing solutes and the influenc of solid solution as it will be developed later.

4.1.Deformation mechanisms

The dependence of the strain rate with the stress over a wide range of temperatures for fine-graine materials presents a stress exponent of 2.This dependence has led to attribute GBS as the principal deformation mechanism.Moreover,GBS depends on the accommodation process,and thus,generally,on lattice diffusion,DL,or grain boundary diffusion,DGB.The GBS constitutive equations are the following [6,7,52]:

Generally,lattice diffusion is dominant at high temperature,whereas grain boundary diffusion usually operates at intermediate to low temperatures.In order to elucidate such dependence,it is necessary to have accurate values of the activation energy over a significan range of strain rates,stresses and temperatures.For reliable measurements,it is important to check that the microstructure is the same at the various temperatures.Grain boundary sliding (GBS) mechanism usually operates at a certain range of temperatures and strain rates,at the so-called “superplastic window”.This window is usually evidenced by large elongations,low stress and low stress exponent values and microstructurally by equiaxed grains and randomization of texture after stretching.Additionally,the maximum superplastic behavior is obtained at higher strain rates,lower stresses and temperatures,the fine the grain size.

GBS deformation mechanism contrasts with typical slip creep mechanisms,which depend on dislocation movement,either glide,or climb.Its constitutive equations follow Eq.(1) with the appropriate stress exponent,n.Generally,climb of dislocations over obstacles is the controlling mechanism,which depends also on the self-diffusion coefficient Its stress exponent associated isn=5,independent of grain size,but characterized for generating a stress-dependent subgrain size [13,53].The associated activation energy for creep is related to lattice diffusion,QL,or pipe diffusion,Qp,at high or intermediate temperatures,respectively.The latter case is usually accompanied by an additional increase of stress exponent of 2.On the other hand,when glide of dislocations is the controlling deformation process,typical of solid solution alloys,the stress exponent isn= 3 and,again,it does not depend on grain size.Its associated activation energy value,Qs,depends on the responsible solute.This mechanism is solutedrag creep [13] and has been proved to operate in materials with atoms in solid solution,as is the case of the coarsegrained WE54 alloy [14].This mechanism may give rise to large elongations of 300% or even more,although lower than those of GBS,and may control deformation of coarsegrained alloys at about the strain rates and temperatures where GBS may operate in the refine alloys.Microstructurally,the solute-drag mechanism is characterized by the presence of elongated grains after deformation.

We can now state that the determination of the high temperature deformation behavior of the WE54 alloy processed by FSP is not trivial,since different creep mechanisms are active depending on the testing conditions,the initial microstructure and its evolution with time and temperature.Clearly,a very distinct behavior between fin and coarse grain WE54 alloy is found,as depicted in Figs.6,7 and 10.Nevertheless,at 250 °C about similar stress and ductility values are found for all conditions.This is a very low temperature for noting any GBS with these microstructures.However,increasing temperature up to 400 °C induces large changes in creep behavior attaining larger elongations(up to about 700%)and much lower stresses,the fine the microstructures.Additionally,the stress exponent (Fig.10) changes fromn= 3(coarse grain) ton= 2 (fin grain),i.e.,from solute-drag creep [14] to GBS mechanism.The deformation mechanism changes progresively towards GBS with increasing temperature up to 400 °C.At 450 °C the microstructures coarsen excesively (Fig.8)and a change back to the original solute-drag mechanism is taking place.Effectively,from 400 to 450 °C,an evident increase of fl w stress and decrease in ductility is observed for most conditions.This means that deformation by grain boundary sliding,that depends strongly of grain size,is vanishing due to the rapid grain growth developing at 450 °C.The curves of Fig.9 show that the optimal temperature presenting the lowest stresses and stress exponents in a wide range of strain rates is 400 °C.On the contrary,at 450 °C,the stressses and stress exponents increase with increasing severity (Fig.9d and f).Additionally,there is a hint of high stress exponents at very low strain rates under some conditions that should be attributed to the change in microstructure as a consequence of the large time that the sample is exposed at temperature during the creep test.This evident grain growth at high temperature makes difficul to refin this alloy by FSP below 1 μm.In this regard,we have made a remarkable effort in this research to reduce the heat input during processing,by incressing traverse speed and lowering rotational rate,as well as increasing cooling rate by using a liquid nitrogen refrigerated backing anvil [49–51].The most promising microstructures and mechanical properties were those obtained with the lowest HI and the refrigerated backing anvil,thus indicating that the deformed material should be rapidly cooled down right after the tool pass.In fact,the non-refrigerated FSP conditions,as an average,attain lower ductilities (Fig.6) and some higher stresses (Fig.7) than refrigerated ones,showing the alloy resistance to behave superplastically,despite the important refinement This may be a hint that the underlying influenc of solutes is important during creep behavior,and that the different thermomechanical treatments may alter the distribution of solutes,whether in precipitates or as in solid solution.

Fig.11.Data at 400 °C for various Mg-Y-Nd-RE alloys,a) strain rate as a function of stress and b) strain rate compensated by grain size as a function of stress.

As mentioned before,it is challenging to obtain precise values of activation energy when having continuous changes in microstructure and deformation mechanisms in a temperature interval as small as 150 °C.However,a convenient comparison with other superplastic magnesium-rare earth alloys can be made at a constant temperature so that the activation energies are irrelevant.Fig.11a shows data from the literature for Mg-Y-Nd-RE materials with strain rate as a function of stress at 400 °C,which is the temperature for optimum superplasticity for all these alloys.The main difference between WE54 and WE43 alloys is the amount of alloying elements,which is larger for the WE54 alloy.

Fig.11a shows a dispersion of values of about one and a half order of magnitude in strain rate among the different bibliographic data.This dispersion could be attributed to the different grain sizes of the respective materials.Assuming that Eq.(2) represents the creep behavior of these materials,we can incorporate an additional representation where the strain rate is compensated by the grain size,Fig.11b.The strain rate of three of the alloys represented in Fig.11b are close to each other except the WE43 of Vávra et al.[16],which is about one order of magnitude higher.This is difficul to explain since in this case of constant temperature,400 °C,Eq.(2),there are no variables left to explain the differences in strain rates,and the different processing techniques used,FSP,this work,ECAP,Vávra et al.[16],TMP,Watanabe et al.[17],MAF,Kandalam et al.[18],are considered important only to refin the microstructure.However,as we pointed out previously,there may be an influenc of processing via distribution of precipitates and/or elements in solid solution.In fact,comparing the different processing methods,ECAP is the one which can be performed at lower temperatures.

Fig.12.Lattice-diffusion-and-grain size-compensated strain rate as a function of modulus-compensated stress for various magnesium alloys that showed superplasticity.Four regions are also depicted by colored ellipses and respective numbers associated to: (1) “light” solute elements with “two phase” type microstructures,(2) “light” solute elements,(3) “heavy” solute elements,and(4) “heavy” and solute elements prone to form LPSO phases.

4.2.Superplasticity of Mg alloys

In order to have a global picture of the behavior in magnesium alloys,19 magnesium alloys from 22 different investigations have been considered [16–37] and represented in Fig.12 as strain rate compensated by the grain size and DLversus the fl w stress compensated by the Young modulus.Only data corresponding to the maximum ductility have been considered.Therefore,the data recorded for each material consisted of one strain rate,one stress,one temperature and one grain size.This is a novel procedure to compare data from different superplastic alloys and draw conclusions on the effect of different parameters on the deformation behavior.

There is some uncertainty for the temperature dependence of the strain rate since,although most of the investigations fin activation energies for creep close to that of lattice diffusion,some of them give values close to grain boundary diffusion.This may have an influenc on the data distribution but it should not be relevant since the temperature range is narrow,between 300 and 400 °C in most of the cases.Moreover,usually,the higher the temperature,the higher the activation energy.The solid blue line in Fig.12 represents the best fi of all the data assumingn= 2.The obtained pre-exponential constant,A= 5 × 108,is 4 times lower than that given in Eq.(2),corresponding to the dashed orange line.This means that these Mg materials are,in average,stronger than expected,compared with other alloys.

An effect that needs to be clarifie is the evident scatter in Fig.12 of more than three orders of magnitude in ˙εL2/DL.Assuming that the preexponential parameter,A,should be the same for all materials,and L and the temperature have been considered when plotting Fig.12,these variables cannot explain such large dispersion of data.It is worth noting that magnesium alloys grains tend to coarse rapidly at high temperature,especially with low alloy content,which,on one hand,hinders grain refinin when processing,and,on the other hand,allows evident coarsening during superplastic deformation.However,even for the highly alloyed magnesium alloys grain growth occurs during superplastic deformation.Therefore,some scatter of superplastic data may arise from the uncertainty in the deforming grain size values assumed in the literature.However,we do not attribute a scatter in ˙εL2/DLof more than an order of magnitude to this concept.

As a firs approximation to explain the low A average value,it may be worth noting that the number of slip systems available for magnesium,as compared to aluminum,for instance,increases with temperature.This means that if temperature is not very high,there would be less slip systems to accommodate superplastic deformation,and this fact should be compensated by a lower grain size to obtain similar strain rates.In fact,this is the main reason we fin to explain the acute ductility peak at 400 °C for the WE54 alloys compared to Aluminum 7075 processed similarly by FSP,for which ductility presents a broader peak in temperature [51].

Nonetheless,some trends are observed in Fig.12.Specifi cally,four broad groups of alloys are found,with some interpenetration,as depicted in Fig.12 with colored ellipses and numbers.The group 1,corresponding to the much weaker materials,shows low optimum superplastic temperatures (200 to 300 °C): W) Uoya,Mg-Li-Ga,300 °C [34],P) Mabuchi,PM AZ91,250 °C [29],Z) Zhou,Mg-Li-Zn,200 °C [37].These also show the lowest tendency to grain growth.These alloys are characterized by large amounts of second phase and/or not very reinforcing solutes,which have low atomic numbers(“light elements”) and high diffusion rates in Mg.They behave as two-phase,duplex or eutectic microstructures,having lower tendency for coarsening.The group 2 corresponds to the alloys with a behavior close to the average,having solute elements without special influence such as Al,Ga,Zn[25–33,36],which are light elements showing fast diffusion in the Mg matrix,as well as along the grain boundaries.The group 3 comprises the magnesium alloys incorporating transition and rare earth metals in solid solution,such as the four Mg-Y-Nd-RE materials (A to D) including the present WE54 alloy [16–18,22–24].These elements are usually heavy,with large atomic numbers,and therefore,large atomic masses,that generally show much slower diffusion rates in Mg than those corresponding to“light”solutes.Finally,the group 4,the most creep resistant alloys,with an optimum superplastic temperature of about 400 °C,which incorporate heavy elements such as Y,Gd,Ni,RE,and solutes prone to form LPSO phases(long period stacking-order) (E to H) [19–21].In this case,synergic interaction among the different solutes and dislocations brings an extra difficult to dislocation glide at high temperatures,thus influencin decisively the accommodation process for grain boundary sliding (GBS).

This classificatio brings two implications: First,it is shown that there is an important influenc of composition on the superplastic response,especially related to solute elements hindering dislocation glide.Obviously,the relative influenc of solutes may be modifie by previous processing and by concentration during superplastic deformation,and should not be taken as well-define rigid groups.Secondly,as superplasticity is governed by GBS,the solute influenc must operate through the accommodation process.This accommodation process is able to fil the gap smoothly at triple joints thanks to vast mass transport,both through lattice and/or grain boundary dislocations,aided by both lattice and grain boundary diffusion,depending on grain size and temperature[8,9].Therefore,the various solute elements present in the Mg alloys alter their high temperature deformation behavior through their different diffusion rates,hindering dislocation movement.In the case of fin superplastic alloys,the different diffusivities influenc the accommodation process for GBS.It is apparent an inverse correlation among the atomic numbers of solutes and their diffusion rates,including synergic effects among solutes.In our opinion,it will be worth investigating these relationships to obtain optimized properties with minimum amount of alloying elements,usually scarce and expensive.

Therefore,although some scatter of superplastic data may arise from some uncertainty in the deforming grain size due to concomitant grain growth,three orders of magnitude in strain rate should be related to other effects.Having investigated alloys with high concentration of strengthening solutes[14,16–24,39],it is our contention that the high temperature creep behavior is affected by solutes even if the alloy becomes superplastic after proper thermomechanical processing.A correlation between “light” and “heavy” elements and “fast” and“slow” diffusivities in magnesium may explain the different superplastic behavior of the various alloys through the accommodation process for GBS,which depends on diffusion rates.The Mg alloys prone to form LPSO phases,due to their additional interaction among solutes [54],are the most resistant superplastic magnesium alloys.This is attributed to the increased resistance of dislocations to accommodate the large mass transport at triple junctions.In case of Mg-Y-Nd-RE-Zr alloys,for instance,as shown in Fig.10,the coarse-grained,before FSP,WE54-TT alloy behaves under solute drag creep mechanism,withn= 3 [14],and shows an order of magnitude higher stresses and two order of magnitude lower strain rates than the superplastic FSP materials.Hence,it is expected that the larger strengthening effect of the solutes,the larger effect on the accommodation process for GBS affecting the fl w stresses and superplastic elongations.

5.Conclusions

A thermally treated WE54 magnesium alloy (WE54-TT)was subjected to friction stir processing (FSP) under various severe conditions.All the materials were refine being the grain size proportional to the severity of the process.Grain sizes of about 0.9 μm were observed under the most severe condition.

The processed materials exhibited excellent superplasticity at high strain rate,10−2s−1,and temperatures between 300 and 400 °C.

Large elongations,with a maximum of 756%,were achieved at 400 °C and 10−2s−1.This was the highest temperature before the microstructure strongly coarsens causing a decrease in the elongation to failure at temperatures higher than 400 °C.

Grain boundary sliding (GBS,n= 2) is the mechanism controlling deformation in these fine-graine materials in a wide range of temperatures and strain rates.Solute drag creep(n= 3) is the mechanism competing and controlling deformation at high temperatures and coarser grain sizes.

The superplastic behavior of the FSP WE54-TT alloy is compared with a large number of magnesium superplastic materials.A new way of representing the data allow exact comparison of their superplastic behavior.

Differences of three orders of magnitude in ˙εL2/DLfor a given stress are observed among the materials.We found an important influenc of reinforcing elements (usually with high atomic number and low diffusion rates) in solid solution during superplastic deformation.This is attributed to the accommodation process necessary for GBS,which is hindered by reinforcing solutes.

Acknowledgments

Financial support from MINECO (Spain),Project MAT2015–68919-C3–1-R (MINECO/FEDER) is gratefully acknowledged.A.O–C.also thanks CENIM,CSIC,for a contract funded by the aforementioned project and O.A.R.for a Professor ad honorem position.Pilar Rey at AIMEN(Spain) is acknowledged for assisting with friction stir processing.M.A-L.thanks MINECO for a FPI fellowship,number BES2013–063963 (MINECO/FEDER/ESF).