The hot compressive deformation behavior of cast Mg-Gd-Y-Zn-Zr alloys with and without LPSO phase in their initial microstructures

H.T.Jeong,W.J.Kim

Departmentof Materials Science and Engineering,Hongik University,Mapo-gu,Sangsu-dong 72-1,Seoul 121-791,Republic of Korea

Abstract Samples of Mg-8.2Gd-3.8Y-1.1Zn-0.4Zr alloy with and without an intragranular lamellae-shaped long period stacking ordered(LPSO)phase were prepared through heat treatment and a series of hot compression tests on these materials were conducted to examine and evaluate the influence of LPSO on the hot compressive deformation behavior and deformation mechanisms at a given alloy composition.The values of activation energy for plastic flow(Qc)of the solution treated(without LPSO phase)and annealed alloys(with intragranular LPSO phase)were larger than that for pure Mg,indicating that the presence of a high amount of rare earth(RE)elements and LPSO in the Mg matrix significantly increases Qc.The Qc value of the annealed alloy was larger than that of the solution treated alloy at all the strain levels(223.3 vs.195.5 kJ/mol in average)and the largest difference in Qc between the two alloys was recorded at the smallest strain of 0.1 where precipitation of LPSO during deformation was limited in the solution treated alloy.These observations imply that the formation of LPSO phase out of the RE-rich solid solution matrix during deformation increases Qc,but the increment is not so large.Analysis of the hot compressive data of the alloys with LPSO phase and the alloys with RE-rich solid solution matrix in literatures indicates the similarity of the effect of the LPSO and RE-rich solid solution matrix phases on Qc and high-temperature strength.

Keywords:Magnesium alloys;LPSO;Deformation mechanism;Dynamic recrystallization;Activation energy for plastic flow;Processing maps.?Corresponding author.

1.Introduction

Magnesium alloys have great potential for widespread use in the automotive,aerospace and mobile electronics industries due to their low density,high specific strength,good recyclability and easy machinability[1–4].However,further improvements in strength and formability are required to make Mg alloys more attractive for practical applications,and efforts toward these improvements have been made to date.One of the strategies for improving the mechanical properties of Mg alloys is to optimize the type and content of alloying elements[5–20].Rare earth(RE)elements such as Y and Gd have high solubility in theα-Mg matrix and the solute solubility decreases rapidly with decreasing temperature[8].Thus,Mg-RE alloys often exhibit an excellent age-hardening and precipitation strengthening response[9,10],leading to outstanding mechanical properties at ambient temperature.Mg-Zn-Y(ZW)series alloys[6,11–14,19–23]have generated considerable interest among researchers due to their unique secondary phases:I-phase(Mg3YZn6),W-phase(Mg3Y2Zn3),and long period stacking ordered(LPSO)phase(Mg12ZnY).The LPSO phase has(0001)close-packed planes as the Mg matrix phase does with a hexagonal close-packed structure,but the formation of LPSO phases involves not only stacking sequence ordering but also chemical composition ordering[14,24,25].The dominant plastic deformation mode in the LPSO phase is(0001)＜110＞basal slip[26],which is the same as that in the Mg matrix phase.However,the operation of basal slip and twinning is difficult in the ordered structure of LPSO phase[26,27],and thus LPSO phase can reinforce the Mg matrix[15,28].Kink banding,which is characterizedby the massive generation of basal dislocations in a localized region,is another deformation mode of LPSO phase[26].

Recently,Mg-Gd-Y-Zn-Zr alloys have received attention due to their outstanding mechanical strength at room temperature and excellent creep resistance[9,16–18,29].The hot deformation behavior and microstructural evolution of Mg-Gd-Y-Zn-Zr alloys have been studied by several investigators[30–42].There are two different morphology types in LPSO phase:(1)interdendritic LPSO on theα-Mg grain boundaries[30,31,34,36,39,42]and(2)intragranular LPSO withinα-Mg grains[30–37,39–42].Zhang et al[30].investigated the hot deformation behavior of a homogenized Mg-13.5Gd-3.2Y-2.3Zn-0.5Zr alloy with interdendritic LPSO phases via hot compression tests in the temperature range between 623 and 773 K.The processing maps showed that high power dissipation domains appeared in the temperature/strain rate range where dynamic recrystallization(DRX)actively occurred.Zhang et al[31].studied the effects of different morphologies and distributions of LPSO phases(intragranular and interdendritic LPSO phases)on the DRX behaviors of a homogenized Mg-6Gd-3Y-1Zn-0.4Zr(wt.%)alloy with a coarse grain size.They observed that intragranular laminar LPSO phases suppressed DRX,while the interdendritic LPSO phases promoted DRX.Many researchers have reported high activation energies for plastic flow(Qc,235–359 kJ/mol)of Mg-Gd-Y-Zn-Zr alloys with interdendritic or intergranular LPSO phase[30–42].The measuredQcvalues were significantly larger than that of pure Mg(=135 kJ/mol[43]).Zhou et al[34].performed hot compression tests on cast Mg-4.9Gd-3.2Y-1.1Zn-0.5Zr(wt.%)with interdendritic LPSO phase at temperatures of 623–773 K and reported a high activation energy for plastic flow of 285 kJ/mol.Xue et al[35].conducted hot deformation on cast Mg-7.5Gd-2.5Y-1.5Zn-0.5Zr(wt.%)alloy with intragranular LPSO phase at temperatures between 623 K and 723 K and strain rates between 10-3and 1 s-1.Under high strain rate conditions,the LPSO phase distorted and formed kink bands.TheQcvalue of the alloy was 234.6 kJ/mol.Zhou et al[36].attributed this high activation energy to the presence of LPSO phase that hinders thermally activated motion of dislocations.However,the Mg-Gd-Y-Zr alloys without LPSO(having the solid solution matrix phase only)also exhibited high activation energies for plastic flow(202.4–209 kJ/mol)[44–47].For example,Li and Zhang[46]studied the hot deformation of the homogenized Mg-9Gd-4Y-0.6Zr alloy and reported the highQcvalue of 209 kJ/mol.The highQcof the solid solution Mg-Gd-Y-Zr alloys was attributed to the segregation effect of the large amount RE solutes around moving dislocations[37].Because of this similarity in theQcvalue between the alloys with and without LPSO phase,it is not clear whether the formation of LPSO phase out of the RE-rich solid solution matrix increasesQc.For better understanding of the effect of LPSO formation on the hot deformation behavior and mechanism of Mg-Gd-Y-Zn-Zr alloys,therefore,it is important to compare the activation energy for plastic flow,flow stress and microstructure evolution in the same chemical composition alloys with and without LPSO phase.However,there are few such studies available.In the present work,we prepared the samples of a Mg-8.2Gd-3.8Y-1.1Zn-0.4Zr alloy with and without intragranular LPSO phase through heat treatment.The effect of the LPSO phase on the deformation behavior,processing maps,and microstructural evolution was investigated by conducting a series of hot compression tests at various temperatures and strain rates.Additionally,the obtained results were compared with those of the pure Mg and other RE-containing Mg alloys(Mg-Y-Zn,Mg-Y-Zn-Zr and Mg-Gd-Y-Zr)with and without LPSO studied by other investigators for understanding of the role of LPSO on flow stress,activation energy for plastic flow and hot workability(processing maps)of Mg alloys.

2.Experimental procedures

An ingot of Mg-8.2Gd-3.8Y-1.1Zn-0.4Zr(wt.%)alloy,the chemical composition of which was determined by using inductively coupled plasma analysis,was melted and poured into a preheated steel mold.The ingot was homogenized at 773 K for 30 h under an argon atmosphere and then subjected to water quenching(WQ).Some of the homogenized Mg alloy was annealed at 673 K for 20 h under an argon atmosphere and then water quenched.The material homogenized at 773 K and then water quenched and the material homogenized,annealed at 673 K and then water quenched will be referred to as the solution treated alloy and the annealed alloy hereafter,respectively.The heat treatment conditions for the solution treated and annealed alloys were determined based on thermodynamic calculations by JMatPro(JMatPro 7.0)simulation software.JMatPro software was also used to determine and calculate the amount of major secondary phases in Mg alloys with different chemical compositions,the hightemperature deformation properties of which were compared with those of the present alloy.

Hot compression tests were carried out for the solution treated and the annealed alloys using a Gleeble3500 thermomechanical simulator,and compression samples 10×12 mm in size(diameter×height)were extracted from both materials.The specimens were compressed at 623,673,723 and 773 K at strain rates of 10-3,10-2,10-1,and 1 s-1,but at 773 K,hot cracking occurred in both alloys during compression.This hot cracking most likely occurred by partial melting due to adiabatic temperature rise during deformation because incipient melting took place around 790 K according to the endothermic curve for the solution treated alloy in temperature-rise measurement by a differential scanning calorimeter(not shown here).K-type thermocouples were used to control and monitor the temperature of the sample during hot compression.Before compression,the specimens were heated to the target temperature at a heating rate of 10 K/s and held for 180 s to homogenize the sample temperature.To minimize friction during compression at high temperatures,graphite foils with nickel paste were applied between anvils and the top and bottom surfaces of the samples.The samples were compressed to a true strain of 0.8 and then cooled.The true stress-true strain curves were drawn based on the load-displacement data obtained by using an L-gauge extensometer(a linear variable differential transformer designed to interface with the Gleeble)and load cell under constant strain rate conditions.An adiabatic heating correction was made by using the linear interpolation between logσand 1/T[48],whereσis the flow stress andTis the absolute temperature.

Microstructural evolution during hot compression was examined by optical microscopy(OM)and scanning electron microscopy(SEM)equipped with electron backscatter diffraction(EBSD).The hot-compressed specimens were sectioned parallel to the compression axis along the centerline and mechanically ground down to 4000 grit.Then,specimens were mechanically polished to 1 μm with a diamond suspension and finalized via ion milling.

Microstructure observation was carried out at a fixed position that was 1/3 the distance away from the surface toward the center of the specimens.The analysis of the EBSD results was performed with TSL-OIM analysis software,and the data were acquired using an accelerating voltage of 20 kV and a step size of 0.3 μm.The solution treated sample deformed at 673 K-10-3s-1was scanned four times over different areas for accuracy of determination of grain size and fraction of dynamically recrystallized grains.The tolerance angle was 5°.To distinguish between the recrystallized and unrecrystallized grains,a grain orientation spread(GOS)map that describes internal misorientation was used,and the criterion misorientation was set to 2[49].Grains with a GOS value below 2°were considered dynamically recrystallized(DRXed)grains.

Phase identification was carried out using X-ray diffraction(XRD)with a Cu target.The sample was scanned through 2θfrom 20 to 100° with a scanning rate of 2°/min.Thermal properties of the solution treated alloy was analyzed using a differential scanning calorimeter(DSC;HCT-2)at a heating rate of 10 K/min.

3.Results and discussion

3.1.Initial microstructures

The variation in the weight percent of various equilibrium phases with increasing temperature for the current composition alloy,which was calculated by using the JMatPro simulation software,is shown in Fig.1.According to the calculation result,the homogenization treatment at 773 K leads to full dissolution of LPSO in the Mg matrix,while the annealing treatment at 673 K after the homogenization treatment results in nearly full precipitation of LPSO phase.

Fig.1.The variation in the weight percent of various equilibrium phases in the Mg-8.2Gd-3.8Y-1.1Zn-0.4Zr alloy with increasing temperature,which was calculated by using JMatPro simulation software.

Fig.2(a)and(b)show optical images of the solution treated and annealed alloys.The solution treated and annealed alloys have similar grain sizes of～96 μm.The insets in the figures show the backscattered SEM micrographs of the two materials,and the element atomic percentages obtained from EDS analysis at different positions are given in the inserted table.LPSO phase was not observed in the solution treated alloy and only small particles in the cluster were sparsely distributed on grain boundaries ofα-Mg matrix.In contrast,high-density intragranular lamellae-shaped phases filled most of theα-Mg matrix grains in the annealed alloy.The laminar phases were arrayed in one direction within each grain.Fig.3 shows the XRD patterns of the solution treated and annealed alloys.In the solution treated alloy the presence of theWphase(Mg5(Gd,Y))was confirmed but the unique peaks identifying the LPSO phase were not found.In the annealed alloy,on the other hand,the unique peaks for LPSO phase(Mg12(Gd,Y)Zn)were found,but those for theWphase were absent.This XRD result is in good agreement with the prediction of the JMatPro software and the optical and SEM observations with the EDS analysis.Zhou et al[50].and Zhang et al[51].reported that LPSO phases exhibited 14H stacking structures in theα-Mg matrices of Mg-6.9Gd-3.2Y-1.5Zn-0.5Zr and Mg-6Gd-3Y-1Zn-0.4Zr alloys,of which compositions are similar to that of the present alloy.

3.2.Hot compressive behaviors

Fig.4(a)–(f)show the true stress-true strain curves for the solution treated and annealed alloys at different temperatures and strain rates obtained from the series of hot compression tests.The flow stresses were corrected,reflecting the adiabatic temperature rise during deformation.In both materials,failure occurred by cracking during compressive deformation at high strain rates at the lowest temperature of 623 K.This cracking failure occurred at a lower strain rate in the annealed alloy than in the solution treated alloy,implying that the former was less ductile.For every curve,rapid strain hardening was observed to take place during the initial stage of deformation.The degree of hardening was higher in the annealed alloy than in the solution treated alloy,and this was more distinct at a lower temperature.This observation implies that strain hardening was more pronounced when the LPSO phase was present in the matrix.Kink-band hardening in LPSO,which hinders the motion of basal dislocations[52],may be responsible for the higher strain hardening in the annealed alloy.Strain hardening in the annealed alloy notably decreasesas temperature increases and this may be because kink-band hardening decreases as the temperature increases due to the operation of non-basal slip at increased temperatures[53].The flow stress continuously decreases after attaining a peak value toward a steady state-like regime where a stable flow stress characterized by nearly equal rates of strain hardening and strain softening is attained.DRX is known to play an important role in the softening behavior of magnesium alloys[54–57].

Fig.2.Optical photographs of the(a)solution treated and(b)annealed alloys.The insets in(a)and(b)show the backscattered SEM micrographs of the two materials.The element atomic percentages obtained from EDS analysis at different locations are given in the inserted table.

Fig.3.XRD curves for the solution treated and annealed alloys.

The hot working deformation of materials involves a thermal activation process.Garofalo[58]showed that a hyperbolic sine function can be used to characterize the relationship of the minimum creep rate and the applied stress at high temperatures,which satisfies conditions for both low and high stresses where power law creep to power law breakdown(PLB)occurs,respectively.Sellars and McTegart[59]applied this function to hot working deformation of metals.Considering the elastic modulus dependence on temperature,Garofalo’s hyperbolic sine function can be modified in terms of the flow stress compensated by the elastic modulus:

wherengis the stress exponent,which is equal tonat sufficiently low stresses[60]but is actually measured to beslightly different depending on the number of datum points available for analysis.At high stresses where PLB occurs,Eq.(1)reduces to an exponential relationship:

Fig.4.The true stress-true strain curves for the(a)(b)(c)solution treated and(d)(e)(f)annealed alloys at different temperatures and strain rates obtained from the series of hot compression tests.

whereβ=αng.

Based on Eq.(1),the activation energy for plastic flow was calculated at different strains(ε=0.1–0.6).The values ofngandβat each temperature were determined by linear fitting according to the relationship of lnε˙-lnand lncurves(whereE=42.9(1–5.3×10-4(T-300))GPa for pure Mg[61])by measuring the slope of the regression line fitted to the data at the corresponding temperature in Fig.5(a)and(b),respectively.Then,αcould be calculated from the relation ofα=β/ng.In Table 1,the average values ofng,αandβfor the solution treated and annealed alloys measured at different temperatures at a given strain of 0.6 are listed.By taking logarithms on both sides of Eq.(1),the relation of lnε˙=lnA+nln[]-Qc/RTis obtained.According to the relationship of lnlnε˙-ln[sinh]and ln[sinh]-1/T,the slopes obtained by linear fitting to the data at each temperature and strain rate in Fig.5(c)and(d)are represented byandrespectively.Then,Qcat a given strain can be determined as follows:

Table 1The average ng,α,β and n values at different temperatures for the solution treated and annealed alloys measured at a strain of 0.6.

whereNandSare the average of thenvalues at different temperatures and the average of thesvalues at different strain rates,respectively.TheQcvalues for the solution treated and annealed alloys measured at different strains are plotted in Fig.6.Overall,theQcvalues of the annealed alloy(211.9–260.3 kJ/mol)are 22–46 kJ/mol larger than those of the solution treated alloy(189–214.1 kJ/mol).As the strain increases from 0.1(near peak stresses)to 0.3,the difference inQcbetween the two alloys decreases from 46.2 to 22.1 kJ/mol and then varies slightly with further straining.TheQcvalues obtained in the present study are considerably larger thanQc(135 kJ/mol)for the self-diffusion of pure Mg.They are also larger than activation energy(126.7 kJ/mol)for impurity diffusion of Y in Mg[62],activation energy(125 kJ/mol)for impurity diffusion of Zn in Mg[62],and activation energy(127.8 kJ/mol)for impurity diffusion of Gd in Mg[63].

3.3.Processing maps

Hot workability can be estimated by the processing map,which provides information about the efficiency of power dissipation(η,η=2(1-)),which is consumed for microstructural evolution,and the processing conditions where unstable flow,such as cracks or shear bands,develops during deformation[64].Recently,Kim and Jeong[65]suggestedan empirical equation that describes theηvalues of metallic materials as a function ofngonly:

Fig.5.The plots for(a)ln˙ε-lnσ,(b)ln˙ε-σ,(c)ln˙ε-ln[sinh(ασ/E)]and(d)ln[sinh(ασ)]-1000/T constructed at ε=0.6 for the solution treated and annealed alloys.

Fig.6.The Qc values for the solution treated and annealed alloys measured at different strains.

whereare the efficiency of power dissipation derived based on the constitutive equations for power law creep and PLB,respectively.Kim and Jeong[61,65]demonstrated that Eq.(5)depicts well theηvalues for many metallic materials despite the difference in crystal structure,alloy composition,and strain level.

The instability parameter(ξ),which characterizes the occurrence of unstable flow in the workpiece during hot deformation,has been proposed based on the extremum principles of irreversible thermodynamics[64].The proposed instability criterion is depicted as[64]

When the empirical expression forηgiven by Eq.(5)was put into Eq.(6)and then the relation ofξvs.ngwas plotted for many metals tested under hot compression,it was found thatξbecame negative atng＞7[65].This result was interpreted that unstable flow occurs at the onset of the transition from power law creep to PLB[65].

Fig.7(a)and(b)show the processing maps for the solution treated and annealed alloys atε=0.6.The solution treatedalloy shows higherηvalues,especially at high temperatures above 673 K and at low strain rates below 10-1s-1.The best hot workability is obtained near 723 K and 10-3s-1from both materials.Below 673 K,unstable flow develops at high strain rates due to the occurrence of fracture cracking during deformation.

Fig.7.The processing maps for the(a)solution treated and(b)annealed alloys.

Fig.8.Optical photographs and SEM micrographs of the(a)(b)solution treated and(c)(d)annealed alloys after hot compression at 723 K and 10-2 s-1.The optical photograph and SEM micrograph of the(e)(f)solution treated alloy interrupted at a strain of 0.1 during hot compression at 723 K and 10-2 s-1.

3.4.Microstructure evolution

Fig.8(a)–(d)show optical photographs and SEM micrographs of the solution treated and annealed alloys after hot compression at 723 K and 10-2s-1.DRX preferentially occurred at the grain boundaries in both materials.Consequently,a bimodal microstructure consisting of coarse deformed original grains and fine dynamically recrystallized grains developed.The fraction of dynamically recrystallized grains is higher in the solution treated alloy,while the size of dynamically recrystallized grains is smaller in the annealed alloy.For the solution treated alloy,LPSO phase formed along deformation bands during the course of hot deformation(Fig.8(a)).Generation of dislocations by hot deformation might have increased solute atom transport through dislocation cores(by pipe diffusion),leading to the preferential precipitation of LPSO phase in the deformation bands.Theamount of the LPSO phase in the deformed solution treated alloy is,however,apparently smaller than that observed in the annealed alloy.Fig.8(e)and(f)show the optical photograph and SEM micrograph of the solution treated alloy interrupted at a strain of 0.1 during hot compression at 723 K and 10-2s-1.At this stage of deformation,limited precipitation of LPSO phase occurred.

Fig.9.EBDS inverse pole figure(IPF)maps and GB maps for the(a)-(e)solution treated and(f)-(j)annealed alloys deformed under different temperature and strain rate conditions:(a)(f)673 K-10-3 s-1,(b)(g)673 K-10-2 s-1,(c)(h)673 K-10-1 s-1,(d)(i)723 K-10-2 s-1,and(e)(j)723 K-10-1 s-1.In the GB map,low angle grain boundaries(LAGBs,2–5°)are in blue,intermediate angle grain boundaries(MAGBs,5–15°)are in yellow,high angle grain boundaries(HAGBs,＞15°)are in red and twin boundaries(TBs)are in green(For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.).

Fig.9(a)–(j)show the EBDS inverse pole figure(IPF)maps and GB maps for the solution treated and annealed alloys compressed under different temperature and strain rate conditions.The size of dynamically recrystallized grains and the fraction of dynamically recrystallized grains measured by analyzing the EBSD data of the deformed samples are plotted in Fig.10(a)and(b),respectively.At 673 K-10-1s-1,deformation in both materials was mainly concentrated around the initial grain boundaries,and limited DRX occurred.When the strain rate decreased to 10-2s-1,necklaces of dynamically recrystallized grains formed along the grain boundaries in both materials.As the strain rate further decreased to 10-3s-1,the degree of DRX increased in both materials.The fractions of dynamically recrystallized grains are 64.1% and 9% at 10-3and 10-2s-1,respectively,in the annealed alloy,while they are 80.6% and 26.9% in the solution treated alloy.These observations indicate that as the strain rate decreases,expansion of DRX into the grain interior from grain boundaries occurs by the formation of subsequent layers of necklaces in both materials,but the expansion rate of DRX into the grain interior is slower in the annealed alloy than in the solution treatedalloy.The dynamically recrystallized grain size,on the other hand,is considerably smaller in the annealed alloy than in the solution treated alloy,indicating suppressed grain growth after DRX in the former.Compared with that at 673 K,the fraction of dynamically recrystallized grains is higher at 723 K in both materials.This result can be attributed to the decrease of incubation time for the nucleation and growth of dynamically recrystallized grains at an increased temperature.For example,at 723 K and 10-2s-1,the DRX fractions are 60.8%and 20.3% in the solution treated alloy and the annealed alloy,respectively,which are higher than those measured at 673 K and 10-2s-1(26.9% and 9%).The size of the dynamically recrystallized grains,on the other hand,changed only marginally as the temperature increased from 673 to 723 K in the annealed alloy,while it increased noticeably in the solution treated alloy.The above observations,showing the lower degree of DRX and smaller grain growth after DRX in the annealed alloy,indicate that the intragranular LPSO phase delays the dynamic recrystallization kinetics and suppresses grain growth of the Mg matrix during hot compression.Zhou et al.[36]also reported that during compression at 753 K,the intragranular LPSO phase in the Mg-6.9Gd-3.2Y-1.5Zn-0.5Zr alloy reduced the degree of DRX by hindering propagation of the dynamically recrystallized grains.

Fig.10.(a)The fraction of dynamically recrystallized grains and(b)the size of dynamically recrystallized grains measured from the EBSD data for the samples deformed under various testing conditions.

Fig.11(a)-(d)show the orientation gradients marked along lines 1–4(near grain boundaries or from one grain boundary to the interior of the grain)in Fig.9(d)and(i).Lines 1–4 show that the accumulated misorientation(point-to-origin misorientation)progressively increases up to 25–35°,indicating the development of a large misorientation gradient within grains during hot deformation.Meanwhile,the corresponding local misorientation varies from 1 to 3° These features characterize a continuous dynamic recrystallization(CDRX)mechanism where progressive rotation of subgrains with lowangle grain boundaries forms new(recrystallized)grains with high-angle grain boundaries,resulting in refinement of coarse grains into smaller grains with different orientations.Lowangle and intermediate grain boundaries are connected to form high-angle grain boundaries in many places,as shown in the GB maps in Fig.9,supporting the occurrence of the CDRX process by the evolution of low-angle to high-angle grain boundaries in deformed grains.Similar point-to-origin misorientation profiles have been reported for the Mg-Gd-Y-Zn-Zr alloy extruded at 733 K[66]and AZ31 alloy compressed at 623 K[67].

4.Discussion

4.1.Recrystallization behavior

The necklaces of new dynamically recrystallized grains formed along the boundaries of deformed grains,which are observed in Fig.9,are a typical characteristic of discontinuous DRX(DDRX)[55,56],while the large misorientation gradient within grains represents the characteristics of CDRX.It has been reported that CDRX and DDRX patterns can coexist in hot deformed Mg alloys[42,66,68].Zhou et al[42].reported that in the Mg-5.5Gd-4.4Y-1.1Zn-0.5Zr(wt.%)alloy,at the early stage of deformation,DDRX occurred near the interdendritic LPSO phases via the particle-stimulated nucleation mechanism,but as the accumulated strain increased,the CDRX mechanism became more pronounced.It has been shown,however,that CDRX also produces a necklace-like microstructure when a rapid increase in strain gradients close to the initial grain boundaries develops[55,56].The microstructure analysis after hot deformation(Fig.10(b))indicates that CDRX kinetics is delayed in the annealed alloy.This is because LPSO delays the CDRX mechanism by inhibiting dislocation slip and climb motions,which is essential for forming subgrains.Kink-aided dynamic recrystallization(KDRX)[69,70]was hardly be detected in the present study.This is because KDRX has been reported to occur when the LPSO phase is fairly fragmented[69,70],but in this work,the LPSO lamellae were kinked but not fragmented.Tension twins were observed within the magnesium matrix in both alloys at 673 K below 10-2s-1(Fig.9),but there is no evidence that twins served as starting points for twin-induced DRX[71,72].

Fig.11.The misorientation gradients marked along lines 1–4 in Fig.9(d)and(i).

4.2.Activation energy for plastic flow

Since LPSO phase is precipitated out during hightemperature compression,theQcvalue measured from the solution treated alloy may not truly represent theQcfor the solid solution matrix alloy without LPSO phase.However,the largest difference inQcbetween the two alloys(46 kJ/mol)is obtained at the smallest strain of 0.1 where precipitation of LPSO is limited in the solution treated alloy(Fig.8(e)and(f))and as LPSO precipitating occurs in the solution treated alloy,the difference inQcbetween the two alloys tends to decrease with straining.These observations clearly indicate that theQcof the alloy with LPSO phase is higher than that of the alloy with solid solution matrix and the formation of LPSO phase out of the solid solution matrix increasesQc.Fig.12(a)shows the values ofQcfor various Mg alloys(with intermediate and large grain sizes)studied by many investigators.TheQcvalue for each alloy was calculated following the same procedures used in this study(considering the dependence of the elastic modulus on temperature)by analyzing the stress-strain curves(available in the references)at peak stresses or in the steady-state regimes.The calculatedQcvalues are listed in Table 2.The type of major phases identified at room temperature by the authors of the references and the type and amount of major phase at the hot compression temperature predicted by using the JMatPro software are also listed.For the pure Mg[64,73],the activation energy for plastic flow is measured to be 134.6±8.1 kJ/mol,which is very close to the activation energy for self-diffusion(135 kJ/mol[43]).For the ZK(Mg-Zn-Zr)alloys[74–77],the activation energy for plastic flow is virtually unchanged(137.4±25.1 kJ/mol)compared to that for pure Mg.For the ZW(Mg-Zn-Y)alloys withWandIphases[78–82],the activation energy for plastic flow only slightly increases or is almost unchanged compared to that for the pure Mg,but for the ZW and ZWK alloys with large amounts of LPSO[83–86],the activation energy greatly increases(240.8±39.2 kJ/mol).For the GWZK(Mg-Gd-Y-Zn-Zr)alloys with a mixture of LPSO andWphases[30,35,40],the activation energy is 193.9±12.1 kJ/mol,and the GWZK alloys with a major phase of LPSO[34,37,39,41](including the present data)have the activation energy of 217.6±21.5 kJ/mol.The GWK(Mg-Gd-Y-Zr)alloys with RE-rich solid solution matrix phase(containing only a small amount of secondary phases at high temperatures)[44–47]also have the high activation energy for plastic flow(203.8±12.1 kJ/mol).The comparison of the activation energies of the GWK alloys and the GWZK alloys with LPSO+W and the GWZK alloys with LPSO indicates that the high value ofQcof the Mg-Gd-Y based alloys originates from either the presence of a high solute amount of RE in the Mg matrix or the presence of LPSO phase,but the contribution of LPSO to highQcis slightly higher.It is worthwhile to note that the values ofQcfor the solution treated alloy is similar to those of the GWK alloys with RErich solid solution matrix and theQcvalue for the annealed alloy is similar to those of the GWZK alloys with the LPSO phase.Fig.12(b)shows theQcvalues of the alloys belonging to the alloy groups classified in Fig.12(a)plotted as a function of wt.% of LPSO phase calculated by the JMatPro software at 673 K.In the RE-rich solid solution state without forming LPSO,Qccan increase up to 220 kJ/mol and as the LPSO phase starts to form,it remains little changed up to a LPSO amount of 5 wt.%.TheQcvalue then gradually increases with increasing the amount of LPSO phase,reaching as high as 250 kJ/mol at the LPSO amount of 15 wt.%.

Table 2(continued)

Table 2The Qc values(considering the dependence of the elastic modulus on temperature)for various Mg alloys by analyzing their raw data(from the series of hot compression tests at peak stresses and in the steady-state regimes from the references of 33–35,37–40,58,66,68–78,80,81–86).Information regarding the type and amount of major phase at the testing temperature was provided by using JMatPro software.

Fig.12.(a)The values of Qc for various Mg alloy groups.(b)The Qc values of the alloys belonging to the alloy groups classified in(a)plotted as a function of wt.% of the LPSO phase(in the alloys)calculated by using the JMatpro simulation software.

Fig.13 shows the plot of the Zener-Hollomon parameter(Z)value as a function of the elastic modulus-compensated flow stress for various Mg alloy groups with their ownQcvalues presented in Fig.12(a).Good correlation was observed in each Mg alloy group despite the difference in the amount of major phase in each group.From the plots,it is recognized that power law creep associated withng=5 dominates plastic flow at the low and intermediate stresses,while PLB associated withng＞7 occurs at high flow stresses.The material parameter for dislocation climb creep,A＇in Eq.(2),can be calculated from the fitting result to the data withng=5.The values are given in Table 3.

Table 3The material parameter for dislocation climb creep,A＇in Eq.(2)calculated from the fitting result to the data with ng=5 in Fig.13.

Fig.14(a)and(b)show the plot of log˙εvs logσfor various Mg alloys at two different temperatures(623 and 723 K).Compared with the pure Mg,the alloys exhibit higher flow stresses.The ZK alloys show a higher flow stress than the pure Mg,and the ZW alloys withWandIphases show higher flow stresses than the ZK alloys.The ZW alloys with LPSO,GWK solid solution alloys,GWZK alloys with W+LPSO and GWZK alloys with LPSO show the highest flow stresses,which are 8–10 times higher than those of the pure Mg.The flow stresses of these alloys are not much different at all the strain rates.

Fig.13.Plot of the Zener-Hollomon parameter(Z)value as a function of the elastic modulus-compensated flow stress for various Mg alloy groups with their individual Qc values indicated in Fig.12(a).

Fig.15 shows the plot of theZvalue as a function of the elastic modulus-compensated flow stress for the ZWK alloys with LPSO,the GWK solid solution alloys,the GWZK alloys with W+LPSO and the GWZK alloys with LPSO.As theQcvalues of the four alloy groups are not much different,their average value(=214 kJ/mol)was used for plotting.Good correlation into a single curve is obtained.The four groups of alloys show very similar deformation mechanism behavior(characterized byng)as a function of flow stress and similar flow-stress levels despite differences in the amount of LPSO phase.The similarity inQcand flow stress in the four alloygroups clearly shows that the effect of the LPSO and RE-rich solid solution phases onQcand high-temperature strength is similar.This result suggests the possibility of transition of order to disorder in the LPSO phase at high temperatures during deformation.Tanaka and Yuge[87]calculated the thermal stability of LPSO in Mg-Y-Zn alloy using first principles and concluded that the LPSO in Mg-Y-Zn alloy remains as ordered intermetallic compounds up to a very high temperature.However,this transition temperature may be lowered when strain-induced order-disorder transition occurs during deformation.Su et al[88].observed the formation of a region with supersaturated solute atoms within the grain interior where LPSO structure partially disappeared during hot extrusion of Mg-Gd-Zn-Mn alloy.They suggested that the accumulated deformation at high temperatures destroyed ordered LPSO structure.

Fig.14.Plot of the log strain rate vs log flow stress for various Mg alloys at the two different temperatures of(a)623 K and(b)723 K.

4.3.Processing maps

In the present study,the processing map shows that the solution treated alloy has a larger efficiency of power dissipation than the annealed alloy due to the occurrence of higher DRX activity in the former.The processing maps were constructed using the three temperatures only due to the occurrence of partial melting beyond 723 K and fracture below 623 K,but the difference in microstructure(the grain size and the degree of DRX)of the two alloys observed after compressive deformation well reflects the difference of power dissipation efficiency in their processing maps.Recently,the current authors[89]proposed the method of constructing the processing maps by coupling the kinetics of dynamic recrystallization and high-temperature deformation mechanism equations,and according to the study,as the degree of DRX increases andthe grain size decreases,the power dissipation efficiency increases,agreeing with the observations in the present study.

Fig.15.Plot of the Z value as a function of the elastic modulus-compensated flow stress for the ZWK alloys with LPSO,the GWK solid solution alloys,the GWZK alloys with W+LPSO and the GWZK alloys with LPSO using the same Qc of 214 kJ/mol.

Fig.16 shows the plot ofηcalculated(using Eq.(5))as a function of strain rate for the pure Mg,the ZW and ZWK alloys with LPSO and the GWZK alloys with LPSO phases at different temperatures based on the measurement ofngvalues from the curves as a function of strain rate in Fig.14.The red dotted horizontal line represents the criterion for the onset of flow instability according to Eq.(6).The following are observed.At 623 K,overall,theηvalues of the pure Mg are greater than those of the alloys with LPSO phases in almost the whole range of strain rate.Flow instability occurs earlier in the alloys with LPSO phases.As the temperature increases to 723 K,the GWZK alloys with LPSO phase show largerηvalues than the pure Mg at low strain rates.The ZWK alloys with LPSO phase,having a large amount of LPSO phase(16–31 wt.%),however,show smallerηvalues than the pure Mg.These observations indicate that LPSO provokes DRX beyond 673 K,but when the amount is too large,it suppresses DRX.Furthermore,the increasing LPSO amount decreases the strain rates associated with the transition from power law creep to PLB;thus,optimum hot working strain rate decreases as the amount of LPSO increases.

5.Conclusions

In the present study,we prepared the samples of a Mg-8.2Gd-3.8Y-1.1Zn-0.4Zr alloy with and without an intragranular lamellae-shaped LPSO phase through heat treatment to examine the effect of the LPSO formation on the hot compressive deformation behavior,hot workability,and microstructural evolution in a given alloy composition.The following results were obtained.

1.TheQcvalue for the annealed alloy with intragranular LPSO was larger than that for the solution treated alloy(without LPSO in its initial microstructure)(223.3 vs.195.5 kJ/mol),implying that the formation of LPSO phase out of the RE-rich Mg matrix increaseQc,but the increase is not so large.

2.TheQcvalues of the solution treated and annealed alloys are significantly larger than that for the pure Mg(135 kJ/mol),indicating that the presence of a high amount of RE elements in the Mg matrix or LPSO phase significantly increases the activation energy for plastic flow.

Fig.16.The plot of η calculated as a function of strain rate for the pure Mg,the ZW and ZWK alloys with LPSO and the GWZK alloys with LPSO at two different temperatures:(a)673 K and(b)723 K,using Eq.(5)based on the measurement of ng values from the curves in Fig.14.

3.CDRX is the main DRX mechanism for both materials.The degree of DRX is lower and the dynamically recrystallized grain size is small in the annealed alloy,indicating that the LPSO phase delays the dynamic recrystallization kinetics and suppresses grain growth of the Mg matrix during hot compression.

4.The processing map shows that the solution treated alloy exhibits largerηvalues than the annealed alloy at temperatures higher than 673 K and at strain rates lower than 3×10-3s-1.This is due to the occurrence of a higher degree of DRX in the solution treated alloy.

5 Analysis of the hot compressive data of the alloys with LPSO and the alloys with RE-rich solid solution matrix from the literatures together with the current data indicates the similarity inQcand flow-stress level desipte the difference in the amount of LPSO phase,suggesting the similarity of the effect of the LPSO and RE-rich solid solution matrix phases onQcand high-temperature strength.

Data availability

The raw/processed data required to reproduce these findings cannot be shared at this time as the data also forms part of an ongoing study.

Declaration of Competing Interest

None.

Acknowledgments

This research was financially supported by the Mid-Career Researcher Program through the National Research Foundation of Korea funded by the Ministry of Education,Science and Technology(NRF-2020R1A2C1008105).I appreciate Mr.K.W.Park’s help in extracting the raw rata from the stressstrain curves provided in the references used in this study.