Grain size dependence of annealing strengthening of an extruded Mg-Gd-Zn alloy subjected to pre-compression deformation

Qinghng Wng,Siyun Chen,Bin Jing,Zhoyng Jin,Lingyu Zho,Junjie He,Dingfei Zhng,Gungsheng Hung,Fusheng Pn

a School of Mechanical Engineering,Yangzhou University,Yangzhou,225127,China

b National Engineering Research Center for Magnesium Alloys,Chongqing University,Chongqing,400044,China

cSchool of Materials and Energy,Yunnan University,Kunming,650599,China

Abstract In this work,pre-strain annealing strengthening(PSAS)effect was investigated in an extruded Mg-1.0Gd-1.5Zn(wt.%)alloy with respect to different grain sizes.The evolution of microstructures was provided by scanning electron microscopy(SEM),electron backscattered diffraction(EBSD),transmission electron microscopy(TEM)and high-angle annular dark-fiel scanning transmission electron microscopy(HAADF-STEM)under the initial state,pre-compression,intermediate annealing and re-compression conditions.The obtained results showed a grain size-dependent PSAS effect in the alloy.The sample with larger grain sizes corresponded to a higher strengthening effect,which mainly resulted from a more remarkable hindrance for the growth of existing twins and a larger proportion of activation for the nucleation of new twins.This was closely associated with the increase of back stress and friction stress for twin boundary motion impeded by the larger restraint of dislocations,the higher stress fiel surrounding solutes and the more Zn segregation.In addition to twinning behavior,Guinier Preston(G.P.)zones on basal〈a〉dislocations were found after intermediate annealing and provided an extra strengthening by inhibiting the motions of gilding pre-existing dislocations and newly formed ones,but it was independent on the grain size.? 2021 Chongqing University.Publishing services provided by Elsevier B.V.on behalf of KeAi Communications Co.Ltd.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/)Peer review under responsibility of Chongqing University

Keywords:Mg-Gd-Zn alloy;Grain size;Pre-compression;Annealing strengthening;Twinning.

1.Introduction

All the time,low absolute strength restricts industrial applications of magnesium(Mg)alloys.Development of high rare earth(RE)containing Mg based alloys to some extent compensates for this shortcoming.However,high cost tends to be not conducive to their large-scale production.Given that Mg alloys are sensitive to Hall-Petch relationship,grain refinemen becomes an effective way to enhance the strength of Mg alloys.Multi-pass severe plastic deformation techniques,such as equal channel angular pressing(ECAP)[1,2]and high pressure torsion(HPT)[3],can significantl reduce grain sizes of Mg alloys from millimeter to micron even nanometer level apparently improving their strengths up to near 400MPa[1,2].However,these deformation processes also greatly increases the manufacturing cost.

Heat-treatable low-alloyed Mg alloys provide a bright prospect to break a“cost-properties trade-off dilemma”.T6 aging heat treatment is a common method to enhance the alloy strength.However,high content alloying elements and longtime heat treatment are necessary for precipitate strengthening,which remarkably violates the principle of low cost.In order to further save cost and more effectively improve the alloy strength,low-alloyed short-time strengthening Mg alloys are proposed.Bian et al.[4]reported that after pretension 2% strain and annealing at 170 °C for 20min,Mg-1.3Al-0.8Zn-0.7Mn-0.5Ca(wt.%)alloy showed a rapid age-hardening response resulting in a significan increase of the fl w stress from 198 to 238MPa[4].Such an strength increment was depended on the co-segregation of Al,Zn and Ca atoms to basal〈a〉dislocations,contributing to the strengthening by pinning dislocation motions.Zeng et al.[5]and Chen et al.[6]entitled this phenomenon as“pre-strain annealing strengthening(PSAS)”effect.By regulating the annealing temperature,the annealing time and the pre-strain,the PSAS effect is able to be maximized.With a pre-tension of 2%strain and subsequent annealing at 200 °C for 1h,Mg-0.3Zn-0.1Ca(at.%)alloy exhibited a remarkable tensile strengthening response[5].Also,when applied to a pre-compressive strain of 7% and subsequent annealing temperature at 150 and 200 °C for different annealing time,Mg-0.9Zn-0.25Ca(wt.%)alloy showed a higher compressive yield stress regarding annealing at 150 °C for 4h compared to annealing 200 °C for 2h[6].In our previous work,we also investigated the effect of pre-compressive strain on the annealing strength of Mg-1.5Al-1.0Zn-0.3Mn(wt.%)alloy.We found that the PSAS effect,with increasing pre-compressive strain,gradually declined even disappeared,since the nucleation rate of new twins was lower than the growth rate of existing twins during re-compression,even though the co-segregation of Al and Zn atoms at twin boundaries(TBs)impeded the TB motion[7].

Even if efforts above being devoted to investigating many factors to influenc the PSAS effect in heat-treatable lowalloyed Mg alloys,there still lacks a systematical analysis on the grain size dependence of the PSAS effect.To the best of author’s knowledge,Dobroˇn et al.[8]pointed out that after pre-compression of 3% strain and heat treatment at 150°C for 16h,the PSAS effect in an extruded Mg-1.0Zn-0.3Ca(wt.%)alloy was observed in the reversed tensile loaded precompressed sample without any change in the Hall-Petch slope throughout the grain size interval.Note that the formation of numerous precipitates on dislocations was a dominant reason for the mentioned results in this alloy due to relatively long-time heat treatment.Whether this findin is applicable to low-alloyed short-time heat-treatable Mg alloys needs to be further explored and verified Therefore,in this work,we compare systematically the PSAS effects of a low-alloyed extruded Mg-Gd-Zn alloy with respect to a varied grain size.Moreover,we also discuss the relationship among the grain size,twinning behavior,dislocation slip and yield fl w stress before and after pre-compression with intermediate annealing.The results show that there is a grain size-related PSAS effect in the Mg-Gd-Zn alloy.With increasing grain size,this PSAS increment is more remarkable.This is mainly due to the more apparent impediment for the growth of existing twins and the larger proportion of activation for the nucleation of new twins in the sample with larger grain sizes than with smaller grain sizes during re-compression with intermediate annealing.

2.Materials and method

The Mg-1.0Gd-1.5Zn(GZ11,wt.%)alloy was prepared by gravity casting and,prior to extrusion,the machined billets were heat-treated at 530 °C for 12h in order to maintain a solid solution condition.The chemical composition of as-cast alloy was measured by inductively coupled plasmaatomic emission spectroscopy(ICP-AES)and the result is shown in Table 1.The extrusion process for receiving a sheet with 3mm in thickness(extrusion ratio of 34:1)was carried out at 430 °C with an extrusion speed of 3mm/s to produce a fine-graine microstructure.The extruded GZ11 alloy was annealed at 450 °C for 2h and 470 °C for 0.5h,respectively,to achieve microstructures with a range of different grain sizes.

Table 1Actual composition of as-cast Mg-1.0Gd-1.5Zn alloy,wt.%.

Compression tests were performed using a CMT6305-300 KN universal testing machine with a strain rate of 2×10-3s-1at room temperature.All samples were machined from the extruded sheet with their loading direction along the extrusion direction(ED).Rectangular samples for compressive loading had a length of 7mm,a width of 5mm and a thickness of 3mm.Conditions for compression-annealing treatment were based on our previous knowledge of the extruded Mg-1.5Al-1.0Zn-0.3Mn(wt.%)alloy[7].In the present work,all samples with a range of different grain sizes were precompressed up to 2.5% of plastic strain in order to introduce deformation twins inside the material.Further,the annealing at 180 °C for 2h followed by water quenching was implemented for the strengthening effect.Using the mentioned deformation-thermal treatment provided a reference basis for studies of the deformation behaviors and compressive yield fl w stresses of samples with respect to different grain sizes during re-compression.

The microstructural observation and texture analysis of all samples were carried out by electron backscattered diffraction(EBSD)technique using JEOL JSM-7800F device.EBSD preparation consisted of grinding on SiC papers of grit sizes 280,400,600,800,1000,and 1200#,washing,blow-drying as well as electro-polishing at a voltage of 20V and an electric current of 0.03 A for 90s at a temperature of-25 °C with a special electrolyte named as AC2.A step size of EBSD scan of each sample in initial state was set as 1μm,while for deformed samples from smaller grain sizes to larger grain sizes the step sizes were set as 0.1,0.3 and 0.5μm,respectively.All EBSD data were analyzed using Channel 5 software.The average grain sizes of samples were measured by Image-pro plus software.Scanning electron microscopy(SEM)was employed to observe the second phases of the samples in the extruded(or subsequently annealed)condition by TESCAN VEGA 3 LMH SEM device with energy dispersive spectrometer(EDS).To further accurately identify the chemical constitutions of these phases,they were also observed by transmission electron microscopy(TEM)using FEI TECNAI G2 F20 device with energy dispersive X-Ray spectroscopy(EDX).TEM observation was also used to analyze the occurrence of precipitates in samples after pre-compression with intermediate annealing.Vickers hardness testing was conducted under a load of 50g and the holding time of 60s.To observe the segregation of solutes into TBs(or dislocations),High-angle annular dark-fiel scanning transmission electron microscopy(HAADF-STEM)observation was performed by using Titan Cubed Themis G2 300 with an operation voltage of 60-300kV.Thin foils for TEM and HAADF-STEM were prepared by mechanical polishing(～40μm)and then ion beam thinning by GATAN,PIPS 691 device.The EDX mapping and line scanning in STEM of TBs(or dislocations)was used to probe the distribution of solute atoms.

Fig.1.Microstructures and orientation maps from longitudinal sections of samples in the extruded-annealed conditions:(a,d)GZ11-10 sample;(b,e)GZ11-31 sample;(c,f)GZ11-65 sample.

3.Results

3.1.Initial conditions

Fig.1 shows microstructures and orientation maps from longitudinal sections of the samples in the extruded and subsequently annealed conditions.Obviously,the grain size varies as a result of the high-temperature annealing treatment after extrusion.An increase of the annealing temperature would typically result in grain growth.The grain sizes of the investigated samples alter between～10 and～65μm with measurement error about 2%,as shown in Fig.1a-c.To simplify a description of the different GZ11 alloys,the grain sizes of the varied state are added to the alloy label,i.e.,a designation GZ11-10,GZ11-31 and GZ11-65,are used throughout this work.In case of the finest-graine GZ11-10 sample,the microstructure is homogenous and completely recrystallized.In the two other cases(GZ11-31 and GZ11-65 samples),the sizes of recrystallized grains gradually increase together with abnormal grain growth,especially for the GZ11-65 sample.The extrusion condition results in a texture with primary components tilted～±30° away from the normal direction(ND)to the ED and weak pole intensities distributed along the transverse direction(TD)in the(0001)pole figur of the GZ11-10 sample,and there is no obvious preferred orientation on the(10-10)plane(see Fig.1d).With respect to the two other cases,very similar texture features including type and intensity are observed in the(0001)and(10-10)pole figures as demonstrated in Fig.1e and f.Therefore,the investigated samples are characterized by comparable grain sizes and textures,and they can be distinguished only by the grain size.

To exclude the effect of the second phase,SEM and TEM results of the investigated samples are revealed in Fig.2.A few granular-shaped second phases uniformly distribute in the matrix for the three samples via extrusion process.High-temperature annealing treatment never changes the amounts,sizes and morphology characteristics of these phases in Fig.2a-c.The amount and size are～12% of the total and～460nm for each sample.The detailed data are shown in Fig.2d.The results of SEM-EDS mapping display that these phases contain Mg,Gd and Zn elements,meaning that they may be Mg-Gd-Zn phases(see Fig.2a-c).In the previous study,we have confirme these Mg-Gd-Zn phases as Mg3Gd2Zn3phases(W phases)in the extruded GZ11 sheet[9].Herein the chemical composition of“A”particle in the annealed condition(GZ11-65 sample)is measured by TEMEDX point scanning(see Fig.2e)and selected-area electron diffraction pattern(SADP)(see Fig.2f).The analyzed result indicates that“A”particle still belongs to W phase,and the thermal treatment hardly can alter the chemical constitution of Mg-Gd-Zn phase.Thus,based on the comparison among grain sizes,textures and second phases in the extruded-annealed conditions,the grain size is the sole variation among them.This desired result provides a guarantee for discussing the grain size dependence of the PSAS effect in the investigated samples.

3.2.PSAS effect

Fig.3a-c show true stress-strain curves of the investigated samples after pre/re-compression with intermediate annealing.The corresponding pre/re-compressive fl w stresses with and without intermediate annealing are listed in Table 2.Deformation curves of the GZ10-10,GZ11-31 and GZ11-65 samples exhibit a similar trend throughout the loading direction.After pre/re-compressive yielding,an increase of the slope during strain hardening range is observed up to an inflectio point before the slope decreases again.These“S”-shaped compressive curves are associated with twinningdominated deformation rather than slip-dominated deformation[7,10,11].The 0.2% fl w stresses correspond to～94±2,～77±1 and～70±2MPa for the initial GZ10-10,GZ11-31 and GZ11-65 samples,respectively.In order to achieve the 2.5% of pre-compression the following stress levels have to be reached～104±1,～85±2 and～77±1MPa for them.As seen,both the 0.2% and 2.5% fl w stresses decrease with increasing grain size,and this result is suitable to Hall-Petch relationship,as shown in Fig.3d(black dotted and solid lines,respectively),consistent with the related studies[12-14].There is a similar tendency that the 0.2% fl w stresses of samples in re-compressed state with intermediate annealing are higher than the stresses applied to achieve the 2.5% of strain during pre-compression,as shown in blue dotted box of each compressive stress-strain curve,compared to the cases without intermediate annealing(no any strengthening effect is observed).This is known as the PSAS effect.In order to capture the changes in the PSAS effect with respect to the grain size,the increment of the re-compressive 0.2% fl w stresses with intermediate annealing comparable to the pre-compressive 2.5% fl w stresses was calculated.With increasing grain size,this increment gradually accumulates from～17±2 to～30±3MPa.Moreover,the re-compressive 0.2% fl w stresses also match Hall-Petch relationship,and its slope(～77)of Hall-Petch relationship is lower than these in the cases of the pre-compressive 2.5% fl w stresses(～140)and the re-compressive 0.2% fl w stresses without intermediate annealing(～139)in Fig.3d(red,black and blue solid lines,respectively).Obviously,there is a grain size-dependent PSAS effect in the investigated samples.This findin is more remarkable in the GZ11-65 sample with large grain sizes.

Fig.2.SEM and TEM images of samples under the extruded-annealed conditions:(a)GZ11-10 sample;(b)GZ11-31 sample;(c)GZ11-65 sample;(d)second phase sizes and proportions in the three samples;(e)TEM bright fiel image and selected-area electron diffraction pattern of“A”particle in the GZ11-65 sample;(f)EDX result of“A”particle in(e).

Table 2Flow stresses(in MPa)of each investigated sample in initial state(0.2%),precompressed state(2.5%),re-compressed state without intermediate annealing(0.2%)and re-compressed state with intermediate annealing(0.2%).

Fig.3.True stress-strain curves of samples with different grain sizes under pre-compressed and re-compressed with intermediate annealing conditions:(a)GZ11-10 sample;(b)GZ11-31 sample;(c)GZ11-65 sample;(d)Hall-Petch relationships for samples with different states.

3.3.Microstructure evolution

To better understand the grain size dependence of the PSAS effect,an analysis of microstructures of samples after pre/re-compression with intermediate annealing are provided by EBSD,TEM and HAADF-STEM in Figs.4-9.Particularly,quasi-in-situEBSD images are used to systematically investigate twinning behaviors of samples with different grain sizes before and after re-compression with intermediate annealing.By comparing the nucleation of new twins and the growth of existing twins taken from same area before and after loading,the modes of twinning behavior are revealed.

3.3.1.Microstructures of samples after pre-compression

Fig.4a-c show the EBSD images of samples after precompression without intermediate annealing.It is seen that the applied pre-compressive strain leads to the formation of lenticular twins insides some grains.At room temperature,when applied to compressive strains,those grains with caxis perpendicular to loading direction are easier to activate{10-12}extension twins rather than other twin types[10].{10-12}extension TBs with a reorientation of the original lattice by 86.3°(±5°)are labeled in black color in Fig.4a-c.As seen in Fig.4d,massive{10-12}extension twins are formed in these highlighted grain orientations with c-axis closely perpendicular to the ED in red dotted-lines during pre-compression along the ED.Since large grains provide more spaces of twin nucleation and growth[15,16],the GZ11-65 sample with larger grain sizes exhibits a higher twin area fraction(～21±3%)than other two samples after pre-compression,even the more remarkable intersection of twins,in Fig.4c and f.Besides twinning behavior,dislocation slips also simultaneously occur insides some grains,especially for these with ED double-peak texture orientations because they have high Schmid factors for basal slip during pre-compression along the ED(see Fig.4e).The average kernel misorientation(KAM)value can reflec the strain level(or dislocation density)induced by deformation.This is exemplifie by the variation in KAM fraction(see Fig.4g).The number fraction of KAM about 0.47 in the GZ11-10 sample is very similar with these in the GZ11-31(～0.51)and GZ11-65(～0.48)samples after pre-compression.This result indicates that the effect of grain size on the storage of dislocation is not obvious under the compression condition in the investigated alloys.

Fig.4.Microstructural analysis of samples after pre-compression:(a-c)EBSD maps of GZ11-10 sample;GZ11-31 and GZ11-65,respectively;(d,e)schematic diagrams of{10-12}extension twinning and basal dislocation slip,respectively;(f)twin area fractions;(g)KAM values.

3.3.2.Microstructures of samples after pre-compression with intermediate annealing

After pre-compression,2h@180 °C intermediate annealing is performed to achieve the PSAS effect.Fig.5 shows the comparison of microstructures in the selected GZ11-31 sample after pre-compression with and without intermediate annealing byquasi-in-situEBSD,KAM map and average KAM value.We can observe before intermediate annealing some of grain boundaries(GBs)and{10-12}extension TBs concentrate on higher KAM values(green regions)than the ones without boundaries in Fig.5a and b.After intermediate annealing,these{10-12}extension twins never vary significantl,and the distribution of dislocation density is also similar with that before intermediate annealing,as shown in Fig.5d and e.From the average KAM values(see Fig.5c and f),there is a sole reduction from～0.49 to～0.47,indicating that intermediate annealing hardly can eliminate the residual strains induced by pre-compression.Even though it has been revealed that dislocation densities gradually decreased by enhancing the annealing temperature and prolonging the annealing time[6,17],the low-temperature and short-time annealing treatment fails to meet the mentioned conditions,thereby making microstructures of samples after pre-compression nearly no different after intermediate annealing.

In previous studies,Mg-Gd-Zn alloys could exhibit an agehardening response.The experimental observations made by TEM indicated that the precipitation process during isothermal aging at 200 and 250 °C of a Mg-6Gd-1Zn-0.6Zr(wt.%)alloy involved the formation of metastableγ′′andγ′phases and that the peak hardness was associated with theγ′′precipitates[18].In this work,for further exquisitely observing the microstructural variation,TEM,even HAADF-STEM observation is carried out.Fig.6a-c demonstrates TEM bright fiel images showing the distribution of precipitates on GBs,TBs and dislocations insides grain in the GZ11-65 sample after pre-compression with intermediate annealing.We cannot observe the occurrence of precipitates on GBs,TBs and dislocations insides grain.The age-hardening response level is also measured for samples after pre-compression with intermediate annealing at 180 °C for 0～10h by hardness index in Fig.6d.As seen,the hardness value of sample with larger grain sizes is lower than with smaller grain sizes after precompression.With prolonging annealing time,the hardness value of each sample fluctuate within a very small range.This result indicates that there is no obvious age-hardening response within the allowed error range.Therefore,2h@180°C intermediate annealing never leads to the precipitation.

HAADF-STEM is usually employed to trace the tracks of heavy atoms based on the principle of atomic contrast imaging.Taking the GZ11-65 sample after pre-compression with intermediate annealing as an example,Figs.7 and 8 show the distribution of Gd and Zn solutes into TBs and on dislocations after intermediate annealing,respectively.It can be seen that the matrix is divided into three regions(matrix-twin-matrix)by two clear{10-12}extension TBs(see Fig.7a and b).HAADF-STEM mapping result reflect that massive Zn solutes,rather than Gd,segregates remarkably into a TB shown in Fig,7c-d.In addition,we also pay attention to the segregation of solutes on dislocations.The low magnificatio view of TEM bright fiel image can fin a large number of residual basal〈a〉dislocations on the(0001)basal plane when viewed along[11-20]zone axis,and these are straight black lines magnifie and labeled by red arrows,as shown in Fig 8a and b.In HAADDF-STEM image(see Fig.8c)corresponded to Fig.8b,local elongated bright discs overlap the black lines(dislocations)indicating a fact that solutes distribute on basal〈a〉dislocations.From the analysis of the morphology,these elongated bright discs can be deemed as Guinier Preston(G.P.)zones[5,19].And they are likely to be two-layer atomic G.P.zones on basal〈a〉dislocations by high magnificatio view in Fig.8d.

Fig.5.Microstructures of the GZ11-31 sample after pre-compression before and after intermediate annealing:(a,d)EBSD maps;(b,e)KAM maps;(c,f)KAM distributions.

Fig.6.TEM bright fiel images showing(a)GBs,(b)TBs and(c)dislocations without any precipitates in the GZ11-65 sample after pre-compression with intermediate annealing;(d)hardening values of all pre-compressed samples as a function of annealing time.

Fig.7.(a)HAADF image;(b)selected-area electron diffraction pattern(SADP)map;(c)high magnificatio view of(a);(d,e)STEM maps of Zn and Gd elements for the GZ11-65 sample after pre-compression with intermediate annealing.

Fig.8.(a,b)TEM bright fiel image;(c)HAADF-STEM image showing dislocations formed on(0001)basal plane;(d)high magnificatio view of dislocations marked by red arrows shown in(c)for the GZ11-65 sample after pre-compression with intermediate annealing.(For interpretation of the references to colour in this figur legend,the reader is referred to the web version of this article.)

Table 3Twin area fractions(in%)of each pre-compressed sample with intermediate annealing before and after re-compression.

As mentioned above,2h@180 °C intermediate annealing hardly induces the obvious precipitation and the variations in twin morphology and dislocation density,while the partial segregation of solutes into TBs and the existence of G.P.zones on basal〈a〉dislocations can be clearly observed for all samples after pre-compression with intermediate annealing.Combining the analysis of twin area fraction in Section 3.3.1,we can speculate that twin area fractions,dislocation densities around TBs,solute segregations into TBs and G.P.zones on basal〈a〉dislocations may be closely associated with the grain size-dependent PSAS effect.

Table 4Contributions(in%)of new twin nucleation and existing twin growth of each pre-compressed sample with intermediate annealing to twin area fraction increment after re-compression.

3.3.3.Microstructures of samples after re-compression

To evaluate the grain size-dependent PSAS effect,the twin volume fraction increment as a result of further loading of samples after pre-compression with intermediate annealing can be used as a representative numerical parameter.Twin nucleation and growth strongly affect the fl w stress of the alloy.In general,the critical stress of new twin nucleation is higher than that of existing twin growth during further loading at room temperature[20-22].If the volume fraction of new nucleated twins is higher than that of existing grown twins,the fl w stress increases.Otherwise,it decreases.The development of the twin area fraction before and after recompression of samples is depicted byquasi-in-situEBSD in Fig.9,and the corresponding data are listed in Table 3.EBSD maps in Fig.9a-c are a result of the pre-compression with intermediate annealing.The twin area fractions of～8±1,～20±1 and～26±1% are observed in the GZ11-10,GZ11-31 and GZ11-65 samples after pre-compression with intermediate annealing,respectively(see Table 3),almost similar with these in their pre-compressive microstructures because intermediate annealing hardly can change twin morphology.The applied re-compression provides a certain extent of new nucleated twins and existing grown twins in the microstructures of all re-compressed samples,resulting in the increased twin area fraction in Fig.9d-f.All observed twins still are consistent with{10-12}extension twins.The twin area fraction is～18±1,～28±1 and～31±1% corresponded to the GZ11-10,GZ11-31 and GZ11-65 samples after re-compression,respectively(see Table 3).From Fig.9g,there is a decreased tendency for the twin area fraction increment:～10±2(GZ11-10),～8±2(GZ11-31)and～5±2%(GZ11-65).A detailed statistics shows the contributions of new twin nucleation and existing twin growth to the twin area fraction increment for each sample in Fig.9h.With increasing grain size,the existing twin growth dramatically decreases from～92±3 to～16±2%(or the new twin nucleation remarkably increases from～8±3 to～84±2%)reflecte in Table 4.To further intuitively understand twinning behaviors of samples with respect to different grain sizes before and after re-compression,some highlighted grains are extracted from each sample pre-compression with intermediate annealing and corresponding sample after re-compression shown in Fig.9i.For the GZ11-10 sample with smaller grain sizes,pre-compression leads to the formation of one or two twin embryos,and these twins propagate towards the grain during re-compression.With the increase of grain size,some twin variants are occurred in the GZ11-31 sample after precompression.During re-compression,these twins grow accompanied by the formation of a few new twins.In the case of the GZ11-65 sample with larger grain sizes,multiple twin variants are formed insides grains after pre-compression,and these intersected twins hardly grow when re-compressed.However,a larger proportion of new twins are activated in the GZ11-65 sample after re-compression.These results suggest that during re-compression,the mobility of TBs is more easily impeded giving rise to a decrease in existing twin growth increment(or new twins are more obviously activated to increase the contribution of twin nucleation to twin area fraction increment)in the sample with larger grain sizes.Therefore,compared to the samples with smaller grain sizes(GZ11-10 and GZ11-31),the sample with larger grain sizes(GZ11-65)need to apply a higher stress to accommodate the plastic deformation by promoting the TB mobility and activating the new twins during re-compression,resulting in the grain sizedependent PSAS effect.

Fig.9.Microstructural analysis of samples before and after re-compression with intermediate annealing:EBSD maps of(a,d)GZ11-10 sample;(b,e)GZ11-31 sample and(c,f)GZ11-65 sample;(g)twin area fractions;(h)contributions of new twin nucleation and existing twin growth to twin area fraction increment;(i)highlighted grains of each sample with respect to new twin nucleation and existing twin growth.

4.Discussion

We have investigated the PSAS effect of the investigated alloy with respect to different grain sizes subjected to pre/recompression with intermediate annealing,and demonstrated that it can be quantitatively evaluated by the increment between the pre-compressive 2.5% fl w stresses and the recompressive 0.2% fl w stresses,which depends strongly on twinning behavior(new twin nucleation and existing twin growth).During deformation,these two twinning modes compete with each other.In this work,we mainly discuss the effect of grain size on the TB mobility,an indicator of grain growth,of samples before and after re-compression with intermediate annealing to reveal the origin of the grain sizedependent PSAS effect.

4.1.Back stress and friction stress for TB motion

We introduce back stress and friction stress to represent the intrinsic characteristic of TBs.Back stress is generated due to the non-uniform distribution of stress on both sides of TBs,and friction stress represents the stress required for TB motion[23,24].The relationship between back stress and friction stress for TB motion with the corresponding re-compressive yield fl w stress can be expressed by Cui et al.[24]as follows:

whereσre-comp.,τbackandτfricrepresent re-compressive yield fl w stress,back stress and friction stress,respectively.It indicates that the increments of back stress and friction stress for TB motion promote the enhanced re-compressive yield fl w stress.

4.2.The effect of grain size on back stress for TB motion

According to the work of Cui et al.[23],back stress for TB motion can be expressed by:

whereC,mt,fandstrepresent intrinsic elastic stiffness tensor of twin,Schmid tensor for twin system,twin area fraction and twin shear,respectively.In an alloy system,C,mtandstkeep almost constants.Therefore,back stress for TB motion strongly depends on the twin area fraction of alloy system.As shown in Table 3 and Fig.9g,with increasing grain size,there is an apparent upward trend for the twin area fraction.This indicates the GZ11-65 sample with larger grain sizes bears a higher back stress for TB motion during re-compression compared to two other samples with smaller grain sizes.

To better understand the effect of grain size on back stress for TB motion during re-compression with intermediate annealing,we present in Fig.10a schematic illustration at various stages of compression.It is well known that the formation of twin derives from a quick shear deformation of parent grain[25,26].Twin propagation and growth tend to be restricted by adjacent GBs,leading to stress field between twins and them(i.e.,adjacent GBs,dislocations,TBs,solutes),especially nearby twin tips.These stress field(back stresses)inhibit twin propagation and growth.Compared to two other samples,the GZ11-65 sample shows massive twins(including numerous intersected twins)after pre-compression due to larger grain sizes.These twins sustain higher back stresses for TB motion.After intermediate annealing,dislocation densities slightly reduce(highlighted by black arrows),thereby making those back stresses almost never eliminate.During re-compression,owing to a higher back stress,the GZ11-65 sample demonstrates a lower TB motion,especially for these intersected twins[27].As shown in Fig.10c,when twin“A”is struck by nearby twins“B”and“C”(especially for twin tips),stress field(back stresses)are generated in areas where twins intersect.These stress field apparently impede the TB motion of twin“A”during re-compression.Therefore,a higher increment of re-compressive fl w stress can be obtained because of an increased back stress.

4.3.The effect of grain size on friction stress for TB motion

Besides the back stress,friction stress also plays a significan role on TB motion.Previous researches have highlighted that friction stress was strongly influence by dislocation density in the matrix,TB coherency,solute concentration,precipitates and segregation of solute atoms in the TBs[7,17,23,24,27-29].Friction stress can be roughly expressed as[23]:

As analyzed Section 3.3,residual strains mainly distribute around GBs and TBs in three samples after pre-compression,and are independent on the grain size.During low-temperature and short-time intermediate annealing,interestingly,these dislocations are hardly eliminated shown in Fig.5.Therefore,there is the similarfor all samples to provide a friction stress induced by dislocations for the decreased TB motion during re-compression.

The reports of Liu et al.[30]and Shi et al.[31]have revealed that the necessary stress of incoherent TB motion was higher than that of coherent TB motion.In an AZ31 alloy,Cui et al.[29]found that the areas adjacent to TBs were severely distorted after pre-compression by high-resolution TEM observation,indicating the incoherence of TBs,as also found by Shin et al.[32].After intermediate annealing,these distortion areas around TBs were alleviated significantl and TBs became more coherent,because of the reduction of dislocation density surrounding TBs.In this work,however,those incoherent TBs induced by pre-compression hardly can be translated into coherent TBs by intermediate annealing,since intermediate annealing never eliminate those deformation strains containing dislocations around TBs,thereby increasing the stress of TB motion during re-compression.The larger the grain size,the more the area fraction of incoherent TBs.This gives rise to the higherfor impeding the TB motion in the GZ11-65 sample during re-compression.

In addition to forming a few secondary phases,vast majority of Gd and Zn atoms are solid dissolved into Mg lattice to produce an interstitial solid solution due to their high solid solubility.The atomic radii of Gd,Zn,and Mg are 0.180,0.133,and 0.160nm,respectively.Substituting a Mg atom with a Gd atom leads to a large negative misfi(-0.125)and compression strain,and substitution with a Zn atom causes a slightly larger but positive misfi(0.169)and extension strain[33].Gd and Zn atoms distribute uniformly in the matrix,and form the severe distortion energy(stress field around each atom.When TBs move around these atoms,stress field surrounding them strongly inhibit the TB motion.Apparently,the GZ11-65 sample with larger grain sizes bears the large resistancefor TB motion during re-compression because of the higher twin area fraction.

Fig.10.Schematic illustrations of back stress for TBM for samples at various stages(pre-compression,intermediate annealing and re-compression):(a)GZ11-10 sample;(b)GZ11-31 sample;(c)GZ11-65 sample.

As early as 2013,Nie et al.[33]has revealed that Gd and Zn atoms could to some extent segregate/co-segregate into TBs in a pre-compressed Mg-Gd-Zn-Zr alloy during intermediate annealing.They pointed out that the co-segregation of Gd and Zn into TB benefite to reduce its elastic strain for TB stability.This phenomenon of alloying element segregation into TB has been also reported in other alloys[6,7,23,24].In this work,we also fin the segregation of Zn solute into TB,as shown in Fig.7,but no Gd segregation into TB.It seems to be inconsistent with the observation of Nie et al.[33].We speculate that the compositions of Zn and Gd may influenc the solute segregation level.Recently,Hoseini-Athar et al.[34]observed with increasing Zn content in a Mg-2Gd(wt.%)alloy,partial solute Gd atoms precipitated out of the matrix as the W phase,leading to a depletion of Gd atoms in the matrix.As a result,no discernible segregation of Gd atoms was observed in the alloys containing high Zn content.Such a composition-dependent segregation has been also reported for the Mg-Nd-Zn alloys previously[35].Thus,it is very possible that the Gd segregation vanishes deriving from the formation of W phases,when the Zn content is higher than that of Gd element in the present alloy.Although only Zn segregates into TBs,it still can effectively impede the TB motion due to theeffect.The number of TB determines the level of solute segregation into TBs.Based on the data in Fig.9a-c,we calculate the number fraction of TB for each sample after pre-compression,and the GZ11-65 sample is the highest(～22±2%).Two other samples are～17±2 and～7±1%,respectively.Therefore,the lowest TB motion occurs in the GZ11-65 sample than two other samples during re-compression.

It is widely believed that precipitates can obviously strengthen alloys by restricting the motion of dislocations via long-time low-temperature aging.However,Dobroˇn et al.[28]and Cui et al.[23,24]pointed out that precipitate strengthening also can be achieved by relative short-time thermal treatment after introducing pre-deformation in Mg-Zn-Ca and AZ91 alloys,respectively.These precipitates tend to be preferentially formed in GBs,TBs and dislocations.Nevertheless,no any precipitates can be found under the condition of this work shown in Fig.6.This result indicates that effect of precipitate on the TB motion can be negligible.As a result,noeffect restrains the TB motion in three samples during re-compression.

From the above,we summaries schematic illustrations of the friction stress for TB motion in three samples in Fig.11.Besides the intrinsic friction stress,the friction stress for impeding the TB motion depends on residual dislocations,stress field induced by solutes and solute segregation.Among three samples,the TB motion for the GZ11-65 sample is lowest than these for two other samples during re-compression.Those obstacles(i.e.,dislocations and solutes)are abundant near TBs.The more TBs there are,the stronger the resistances for TB motion will be.High twin area fraction and more intersected twins are formed in the GZ11-65 sample with larger grain sizes after pre-compression,thereby increasing the friction stress for inhibiting the TB motion during re-compression.

Fig.11.Schematic illustrations of friction stress for TB motion for samples after pre-compression with intermediate annealing:(a)GZ11-10 sample;(b)GZ11-31 sample;(c)GZ11-65 sample.

4.4.The effect of G.P.zone on PSAS effect

In Sections 4.1-4.3,we have revealed the grain sizedependent of the PSAS effect by increasing the back stress and friction stress for impeding the TB motion for the sample with larger grain sizes.In addition to controlling the twinning behavior,pinning mobile dislocations is also considered to impact the PSAS effect.Because of initial texture features shown in Fig.1d-f,some grains with high Schmid factor for basal slip(especially for non-basal oriented grains)tend to be activated basal slip rather than twinning when they are precompressed along the ED.These dislocations,still being preserved after intermediate annealing(see Fig.5),play an important role on two aspects during re-compression as follows:1)inhibiting the mobility of newly formed dislocations.Continuous compression makes slip-oriented grains ceaseless produce new dislocations.When these new dislocations interact with remaining ones,their motions are restrained and stress field with large distortion energies are generated around them giving rise to an increased fl w stress;2)impeding the movement of pre-existing dislocations.It has been reported by Bian et al.[4]and Zeng et al.[5]that deformation-annealing treatment could induce solute co-segregation or G.P.zones,as barriers,on pre-existing dislocations for effectively preventing them to move resulting in the increase of fl w stress.As shown in Fig.8,two-layer G.P.zones are found to locate on basal〈a〉dislocations after intermediate annealing.These G.P.zones not only greatly restrain the further gliding pre-existing dislocations induced by pre-compression,but also can effectively inhibit the motion of newly formed dislocations triggered by re-compression,leading to a strong strengthening effect.

Pre-existing dislocations,annealing temperature and annealing time are crucial parameters to impact the PSAS effect.At a low pre-strain,there are low dislocation densities.The low level of G.P.zones is not enough to inhibit the motion of these dislocations.With increasing pre-strain,the enhanced G.P.zones apparently pin the gliding pre-existing dislocations and show a remarkable strengthening.In addition,if annealing temperature is too low or annealing time is too short,solute atoms are difficul to gather to dislocations to form G.P.zones resulting in a weak pinning effect.However,if intermediate annealing temperature is too high or annealing time is too long,excessive dislocations and work hardening they cause will be easily eliminated,so that softening occurs.In this work,for samples with different grain sizes,they are given the similar pre-existing dislocation densities under deformation-thermal condition,giving rise to the almost same level PSAS increment herein from the role of G.P.zones.

5.Conclusions

In this work,we investigated systematically pre-strain annealing strengthening(PSAS)effect in an extruded Mg-1.0Gd-1.5Zn(GZ11,wt.%)alloy regarding different grain sizes.The main conclusions were summarized as follows:

1.The investigated samples with different grain sizes(～10,～31 and～65μm corresponded to GZ11-10,GZ11-31 and GZ11-65 samples,respectively)were fabricated by various annealing after extrusion.In addition to the grain size,there was no difference(textures and second phases)among three samples.Testing results indicated there was a grain size-dependent PSAS effect that with increasing grain size,the PSAS increment gradually enhanced from～17±2 to～30±3MPa.

2.Two twinning modes strongly influence the grain sizedependent PSAS effect.For the GZ11-10 sample with smaller grain sizes,the growth of existing twins was primary twinning mode leading to the lower PSAS effect;while in the case of the GZ11-65 sample with larger grain sizes the nucleation of new twins dominated twinning deformation inducing the higher PSAS effect during re-compression.

3.Twin boundary(TB)motion was more easily impeded in the GZ11-65 sample with larger grain sizes during recompression with intermediate annealing by increasing the back stress and friction stress for TB motion than in two other samples with smaller grain sizes.This was attributed to the larger restraint of dislocations,the higher stress fiel surrounding solutes and the more Zn segregation in TBs after pre-compression with intermediate annealing.

4.In addition to impeding the TB motion,Guinier Preston(G.P.)zones formed on basal〈a〉dislocations during intermediate annealing inhibited the motions of gliding preexisting dislocations and newly formed ones,resulting in an apparent strengthening.Note that for samples with different grain sizes,similar pre-existing dislocation densities under deformation-thermal condition gave rise to the almost same level of PSAS increment herein from the role of G.P.zones.

Declaration of interest statement

The authors declare that they have no known competing financia interests or personal relationships that could have appeared to influenc the work reported in this paper.

Acknowledgements

The authors are grateful for the financia supports from National Natural Science Foundation of China(U1764253),National Natural Science Foundation of China(51901202)and National Natural Science Foundation of China(51901204).