Sang Won Lee, Sang-Hoon Kim, Sung Hyuk Park
School of Materials Science and Engineering, Kyungpook National University, Daegu 41566, Republic of Korea
Abstract This study investigates the microstructural characteristics of AZ31 Mg alloys rolled at room temperature (RT) and cryogenic temperature(CT) and the variation in their microstructure and hardness during subsequent annealing.Cryorolling induces the formation of more side cracks than does RT rolling, because of the reduction in the ability of the material to accommodate deformation at CT.Numerous {10-11}contraction and {10-11}-{10-12} double twins are formed in both the material rolled at RT and that rolled at CT, because the grains of the initial material are favorably oriented for {10-11} twinning under rolling.The RT-rolled material has a higher dislocation density than the cryorolled material, and more twins are uniformly distributed throughout the former material.As a result, static recrystallization during subsequent annealing is more pronounced in the RT-rolled material, which results in the formation of a highly recrystallized homogeneous microstructure after annealing.In contrast, the formed twins are predominantly present along the shear bands in the cryorolled material, as a result of which this material has an inhomogeneous bimodal structure containing a large amount of coarse unrecrystallized grains after annealing.The hardness of the annealed RT-rolled material is higher than that of the annealed cryorolled material owing to the fine grain structure of the former.
Keywords: Magnesium; Rolling; Cryogenic temperature; Annealing; Microstructure.
In recent decades, great efforts have been made to solve environmental problems associated with global warming; for example, numerous studies have been conducted for reducing the carbon dioxide emissions of vehicles via reduction of their weight [1-3].Mg components have been attracting considerable attention as substitutes for steel and Al components for achieving such weight reduction because of the low density and high specifi strength of Mg.In particular, wrought Mg alloys exhibit much higher mechanical properties than their cast counterparts because metal forming processes such as rolling, extrusion, and forging lead to the elimination of casting defects, grain refinemen via dynamic recrystallization, formation of fin precipitates via dynamic precipitation,and/or work hardening via accumulation of dislocations[4-7]; therefore, weight reduction can be achieved to a greater extent through the application of wrought Mg alloys.
Metal forming can be performed using techniques such as cold working and hot working, which are usually conducted at low temperatures (<0.3 Tm) and high temperatures(>0.6 Tm), respectively [5,8-10].Cryogenic forming at?196 °C using liquid nitrogen has been applied as a metal forming technique in several studies [11-14].Rolling of commercially pure Ti at cryogenic temperature (CT) induces the formation of more deformation twins than does rolling at room temperature (RT), which consequently improves the strength of the rolled material by the grain refinemen effect [11].Wang et al.[12] reported that rolling of pure Cu at liquid-nitrogen temperature and subsequent annealing at 200 °C for 3min imparted extraordinary tensile properties,i.e., high strength and ductility, to the rolled material.Rangaraju et al.[13] reported that rolling of pure Al at CT and subsequent annealing resulted in the formation of an ultrafin grain structure, which consequently endowed the rolled material with both high strength and ductility.Stepanov et al.[14] reported that a CoCrFeNiMn high-entropy alloy rolled at ?196 °C had higher strength than that rolled at 25 °C, because of the promoted activation of twinning at?196 °C.However, few studies have thus far examined the cryorolling of Mg alloys [15-17] and no in-depth studies have yet been conducted on the microstructural differences between RT-rolled Mg alloy and cryorolled counterpart and their microstructural variations during subsequent annealing,which is usually performed after cold rolling.Therefore, in this study, the effects of cryorolling on the microstructure of Mg alloys are investigated by rolling a commercial AZ31 alloy at 25 °C (i.e., RT) and ?196 °C (i.e., CT) and comparing their microstructural characteristics after both rolling and subsequent annealing.
For rolling, two samples, each with dimensions of 80mm(length)×60mm (width)×2mm (thickness), were machined from a commercially hot-rolled AZ31 (Mg-3Al-1Zn, wt%)plate with a thickness of 20mm (Fig.1a); the length, width,and thickness of the samples corresponded to the rolling direction (RD), transverse direction (TD), and normal direction (ND), respectively, of the plate.Before rolling, one of the machined samples was precooled at ?196 °C for 30min in liquid nitrogen and then rolled to a fina sheet thickness of 1.5mm (rolling reduction of 25%) in a single pass.The other machined sample was rolled at RT without any heat treatment prior to rolling under the same rolling condition.Rolling of both the samples was performed at a speed of 5rpm using a two-high mill with a roll diameter of 295mm.For simplicity, the sheets obtained by rolling of the machined samples at 25 °C and ?196 °C are hereafter referred to as rolled RT and CT materials, respectively (Fig.1b).To investigate the static recrystallization behavior of the rolled sheets during subsequent heat treatment, they were annealed at 200°C for 5, 25, and 60min in an electric furnace and then water-quenched.
Fig.1.Photographs showing (a) initial sample for rolling and (b) materials rolled at cryogenic temperature (CT) and room temperature (RT), i.e., rolled CT and rolled RT materials, respectively.RD denotes the rolling direction.
The microstructures of the initial plate, rolled sheets,and annealed samples were observed on the RD-TD plane by optical microscopy (OM), electron backscatter diffraction (EBSD), and X-ray diffraction (XRD).For microstructure observations, specimens machined from the initial plate,rolled sheets, and annealed samples were mechanically polished with progressively fine grades of silicon carbide paper(from #120 to #2000) and then finishe with 3 and 1μm diamond pastes.Subsequently, these specimens were etched with acetic-picral acid solution (10ml distilled water+10ml acetic acid+3.0g picric acid+100ml ethanol)for OM observations.EBSD scans were performed on the RD-TD plane with step sizes of 1.3μm and 0.5μm for the initial plate and the 5-min-annealed samples, respectively.The EBSD data were analyzed using the Tex-SEM Laboratories orientation imaging microscopy 7.0 software.XRD measurements were performed using Cu Kαradiation at a scan speed of 2°/min in the 2θrange of 20-80°.Hardnesses of the rolled sheets and annealed samples were measured using a Vickers microhardness tester.For each specimen, hardness was measured at 10 different positions on the RD-TD plane under a load of 0.3N, and the obtained values excluding the maximum and minimum values were averaged.
Fig.2 shows optical micrographs depicting the side cracks formed in the rolled RT and CT materials.When Mg alloys are rolled at temperatures higher than 250°C, their rollability is relatively high because nonbasal slip systems such as prismatic slip and pyramidal slip are activated and softening mechanisms such as dynamic recovery and recrystallization occur [18,19].However, at low temperatures (<150°C),the rollability of Mg alloys decreases greatly because only basal slip—which has just two independent slip systems—is predominantly activated and the recovery process that occurs is not effective.As a result, in both the rolled RT and the rolled CT materials, which are fabricated by rolling at 25°C and ?196°C, respectively, many cracks 2-4mm in size are formed on the side of the material owing to the reduction in the material’s ability to accommodate compressive plastic deformation at low temperatures (Fig.2a and b).Although these side cracks are of similar sizes in both the materials,the number of cracks formed in each material is quite different.The values of the average spacing between the side cracks are 3.5mm and 2.0mm for the rolled RT and CT materials,respectively, and the number densities of cracks in these materials are 2.6/cm and 4.3/cm, respectively (Fig.2c).That is,the number of side cracks formed in the rolled CT material is about 1.7 times that in the rolled RT material,which indicates that the material rolled at CT has higher susceptibility to side cracking.
Fig.2.Optical micrographs showing side cracks formed in rolled (a) RT and (b) CT materials.(c) Average spacing between side cracks and number density of side cracks in rolled materials.TD denotes the transverse direction.
Fig.3.Optical micrographs of rolled (a) RT and (b) CT materials and (c) X-ray diffraction (XRD) patterns of initial material and rolled RT and CT materials as measured on RD-TD plane. I(10-11)/I(0002) denotes the intensity ratio of the (10-11) peak to the (0002) peak.
Optical micrographs of the rolled RT and CT materials as measured on the RD-TD plane are shown in Fig.3a and b,respectively.Numerous deformation twins are observed to have formed in both the materials.The initial samples for rolling,which were machined from the hot-rolled AZ31 alloy plate,have a strong basal texture wherein thec-axes of most grains are oriented parallel to the ND (Fig.4a).Accordingly, rolling conditions wherein compressive strain is imposed in the ND are unfavorable for the formation of{10-12} extension twins.Indeed,the average Schmid factor(SF)for{10-12}extension twinning under compression along the ND is as low as 0.01(Fig.4b).In contrast, the average SF for {10-11} contraction twinning is significantl high, 0.47 (Fig.4c), which means that {10-11} contraction twins can be easily formed during rolling.After the formation of {10-11} contraction twins,{10-12} extension twinning occurs within them by further deformation because the crystallographic orientation of the{10-11} twinned region is favorable for {10-12} twinning;consequently, {10-11}-{10-12} double twins are formed[20-22].Therefore, most deformation twins formed in the rolled RT and CT materials are{10-11}contraction and{10-11}-{10-12} double twins, rather than {10-12} extension twins.Moreover, the formed twins have a narrow width and needle-like shape, which is typical morphology of contraction and double twins.It is notable that the formed twins are more homogeneously distributed in the rolled RT material than in the rolled CT material and that the number of twins is considerably higher in the former material (Fig.3a and b).
Fig.4.Electron backscatter diffraction (EBSD) results of initial material used for rolling: (a) inverse pole figur map and (0002) basal pole figur and (b,c) Schmid factor (SF) maps for (b) {10-12} twinning and (c) (10-11} twinning under compression along normal direction (ND). SF{10-12} and SF{10-11}denote the average SF for {10-12} extension twinning and average SF for {10-11} contraction twinning, respectively.
It is difficul to measure the area fraction of twins formed in the two materials through EBSD—which is generally used to distinguish twinned and untwinned regions—because the high dislocation density of the rolled materials causes distortion of the backscatter Kikuchi diffraction patterns.However,since deformation twinning leads to lattice reorientation, the area fraction of the twinned region can be inferred from the variation in the XRD peak intensity of a material after deformation[23-25].In the present study,as the initial material before rolling has a strong ND texture(Fig.4a),the(0002)peak intensity in the XRD pattern measured on the RD-TD plane of the material is significantl high;however,the intensities of the (10-11) and (10-10) peaks of this material are very low(Fig.3c).{10-11} Contraction twinning and {10-11}-{10-12} double twinning cause lattice rotations of 56° and 38°,respectively, from the crystallographic orientation of the original matrix.Accordingly, the formation of the contraction and double twins causes a decrease in the (0002) peak intensity by lattice rotation.In this study, the degree of activation of twinning is measured through the relative (0002) peak intensity,which is expressed as the ratio of the(10-11)peak to the(0002)peak(denoted as(I(10-11)/I(0002)).This intensity ratio of the initial material before rolling is considerably low, 0.009,owing to the absence of twins; however, after rolling, this ratio increases greatly by a factor of 10 or more because of the formation of numerous contraction and double twins(Fig.3c).In addition, the intensity ratio (0.104) of the rolled RT material is 1.17 times that of the rolled CT material (0.089),which indicates that more twins are formed in the rolled RT material.Since twinning is activated through atomic motion,it is strongly related to deformation temperature.Chapuis and Driver [26] conducted plane strain compression tests using pure Mg single crystals at temperatures ranging from 25°C to 450°C and demonstrated that the critical resolved shear stress of{10-11} contraction twinning increased substantially with decreasing deformation temperature.In our study, the formation of twins during CT rolling is less pronounced than that during RT rolling because of the increase in the stress required to activate twinning at CT.This suppression of twinning activity at CT in the present study is consistent with previously reported results.Huo et al.[27] reported that when an extruded AZ31 alloy was compressed along the extrusion direction (ED) at 25°C and ?196°C, the number of {10-12} extension twins formed during deformation at ?196°C was much smaller than that during deformation at 25°C and therefore concluded that twinning was inhibited at CT.This suppression of twinning activity at CT is also observed under deformation conditions in which {10-11} contraction twins are predominantly formed.Zhang et al.[28] conducted tensile tests of an extruded Mg-10Gd-3Y-0.5Zr (wt%) alloy at 25°C and ?196°C along the ED and found that{10-11}contraction twinning was considerably suppressed at CT.In addition to the number of twins formed, the distributions of these twins in the two rolled materials in this study are also different: in the rolled RT material, the twins are evenly distributed throughout the material,whereas in the rolled CT material,the twins are concentrated in the localized areas (Fig.3a and b).As the deformation temperature decreases, the homogeneity of deformation reduces because of the increased plastic instability, which eventually causes localized deformation [29-31].Accordingly, the localization of compressive deformation during rolling is more pronounced in the CT material, which is rolled at a substantially lower temperature than the RT material.As a result, deformation twins are locally formed along the shear bands in the rolled CT material.In contrast, rolling at RT causes the compressive deformation to be imposed relatively evenly throughout the material,which results in the formation of deformation twins uniformly across the rolled RT material.
Fig.5.Optical micrographs of rolled (a-c) RT and (d-f) CT materials subjected to subsequent annealing at 200°C for (a, d) 5min, (b, e) 25min, and (c, f)60min.
Fig.5 shows the optical micrographs of the rolled RT and CT materials subjected to subsequent annealing at 200°C (hereafter referred to as annealed RT and CT samples, respectively) for various durations.The annealed RT samples have a homogeneous microstructure in which most regions consist of fin recrystallized grains, and the variation in the microstructure with annealing time is insignifican(Fig.5a-c).In contrast, the 5-min-annealed CT sample has a bimodal structure consisting of fin recrystallized grains and coarse unrecrystallized grains; this microstructure remains unchanged even after annealing for 60min (Fig.5d-f).Inverse pole figur maps of the 5-min-annealed RT and CT samples reveal that the area fraction of recrystallized grains of the RT sample (72.5%) is considerably higher than that of the CT sample (51.3%) (Fig.6).This indicates that the activation of static recrystallization during subsequent annealing is more pronounced in the rolled RT material than in the rolled CT material.It has been reported that when Al,Cu, Ni, and Ti alloys are rolled at CT, numerous dislocations are accumulated in the material due to the suppression of dynamic recovery; consequently, static recrystallization is promoted during subsequent annealing, thereby forming a homogeneous grain structure [11,13,32-34].In contrast, in this study, the rolled CT material exhibits suppressed recrystallization behavior during annealing compared with the rolled RT material.This difference in the extent of recrystallization between the two materials is a result of the difference in their stored energy, which is a driving force for recrystallization.
To compare the stored strain energy of the rolled RT material with that of the rolled CT material, their dislocation densities were calculated from the XRD results.Factors such as the crystallite size, microstrain, and type of instrument may cause broadening of XRD peaks [35].Here, the crystallite size (t) is calculated ast=0.9λ/Bcosθ, whereλis the wavelength of Cu Kαradiation (0.15406nm),Bis the broadening width of the peak (i.e., the full width at half-maximum,FWHM), andθis the Bragg angle [36].In this study, the calculated crystallite sizes (32.5-129.5nm) of the materials are considerably smaller than their grain sizes (6.7-38μm),which suggests that the crystallite size is the subgrain size,which is a domain surrounded by dislocation walls, rather than the grain size.The dislocation density (ρ) can be calculated from its relationship with the crystallite size, i.e., asρ=1/t2[35].The dislocation densities of the initial material, rolled RT and CT materials, and 5-min-annealed RT and CT samples calculated using these equations are shown in Fig.7.The dislocation density of the initial material is relatively low owing to the prior homogenization treatment.The dislocation density increases significantl after both RT and CT rolling because the recovery process, which causes annihilation and rearrangement of dislocations, is ineffective under low-temperature deformation owing to the suppression of dislocation climb and cross-slip phenomena.Moreover, the dislocation density of the rolled RT material is ~11% higher than that of the rolled CT material (Fig.7), which is nearly consistent with the ~17% higherI(10-11)/I(0002)value of the rolled RT material (Fig.3c).The {10-11} contraction and{10-11}-{10-12} double twinned regions have “soft” orientations that are favorable for dislocation slip, which results in considerable accumulation of dislocations within the twins during deformation [37].Consequently, the higher dislocation density of the rolled RT material is attributed mainly to the presence of more numerous contraction and double twins in the material.Furthermore,the extremely low CT provides insufficien activation energy for dislocation generation,which causes suppression of the accumulation of dislocations in the material [26,38].Therefore, during CT rolling, fewer contraction and double twins are formed and the activity of dislocations is inhibited, which results in lower stored strain energy of the rolled CT material; consequently, during annealing, the rolled CT material undergoes recrystallization to a smaller extent compared with the rolled RT material.It is known that contraction and double twins preferentially act as recrystallization sites during subsequent heat treatment owing to their high internal strain energy [39,40].Since numerous contraction and double twins are uniformly distributed in the rolled RT material, static recrystallization occurs throughout the material during its subsequent annealing, which results in the formation of a highly recrystallized homogeneous microstructure.The localized distribution of contraction and double twins in the rolled CT material induces partial recrystallization concentrated in the highly twinned regions during annealing; this partial recrystallization eventually leads to the formation of an inhomogeneous microstructure containing a large amount of coarse unrecrystallized grains.
Fig.6.EBSD inverse pole figur maps of 5-min-annealed (a-c) RT and (d-f) CT samples: (a, d) total, (b, e) recrystallized, and (c, f) unrecrystallized regions.dtotal, dRXed, and dunRXed denote the average grain sizes of the total, recrystallized, and unrecrystallized regions, respectively. fRXed and funRXed denote the area fractions of the recrystallized and unrecrystallized regions, respectively.
Fig.7.Dislocation densities of initial material, as-rolled materials, and 5-min-annealed samples, as calculated from XRD results.
Fig.8.Variation in hardness of rolled RT and CT materials with annealing time.
Fig.9.(a, c) Grain orientation spread (GOS) and (b, d) kernel average misorientation (KAM) maps of 5-min-annealed (a, b) RT and (c, d) CT samples.GOSavg and KAMavg denote the average GOS and KAM values, respectively.
Fig.8 shows the average hardnesses of the rolled materials and subsequently annealed samples.The hardness of the rolled RT material (79.0 Hv) is higher than that of the rolled CT material (75.5 Hv) because of the higher dislocation density of the former.After annealing for a short period of 5min,the hardnesses of both the materials decrease significantly the hardness of the RT material decreases from 79.0 Hv to 64.5 Hv and that of the CT material decreases from 75.5 Hv to 61.6 Hv.This decrease in hardness is attributed to the formation of new strain-free grains via static recrystallization and is consistent with the significan reduction in the dislocation density of the rolled materials after annealing(Fig.7).All the annealed samples of each material have similar hardness values; that is, the hardness is independent of the annealing time because additional recrystallization rarely occurs with further annealing.The hardnesses of the annealed RT samples are slightly higher than those of the annealed CT samples.The reason for this difference can be confirme from the grain orientation spread(GOS)and kernel average misorientation (KAM) maps of the 5-min-annealed RT and CT samples, as shown in Fig.9.The average GOS value of the annealed CT sample (0.78) is higher than that of the annealed RT sample (0.68), and the average KAM value of the former (1.03) is also higher than that of the latter(0.82).This means that the annealed CT sample has a higher residual strain energy than the annealed RT sample; this higher residual strain energy is attributed to a larger amount of unrecrystallized grains—which have a substantially higher dislocation density than recrystallized grains—in the former.This result agrees well with the fact that the dislocation density of the annealed CT sample (6.92×1013mm?2) is higher than that of the annealed RT sample (6.51×1013mm?2)(Fig.7).A higher residual strain energy of a material induces a stronger strain hardening effect during plastic deformation,thereby improving the hardness and strength of the material[41,42]; however, despite the higher residual strain energy of the annealed CT sample, it has lower hardness than the annealed RT sample.This hardness difference between the two annealed samples can be attributed to the difference in their grain size.The average size of recrystallized grains is similar in the two annealed samples (5.0μm and 5.8μm for the RT and CT samples, respectively; see Fig.6b and e).However, both the average size and the area fraction of unrecrystallized grains of the annealed RT sample (12.4μm and 23.5%, respectively) are nearly half those of unrecrystallized grains of the annealed CT sample (24.5μm and 46.0%,respectively), as shown in Fig.6c and f.As a result, the average grain size of the annealed RT sample (6.7μm) is considerably smaller than that of the annealed CT sample(14.4μm) (Fig.6a and d).Therefore, the hardness of the annealed RT sample is higher than that of the annealed CT sample because of the enhanced Hall-Petch hardening effect induced by the fine grain structure of the former sample despite the weaker strain hardening effect.
In this study, the microstructural characteristics of AZ31 Mg alloys rolled at 25 °C (i.e., RT) and ?196 °C (i.e., CT)are compared and the variations in their microstructures and hardnesses during subsequent annealing are investigated.Side cracking occurs during both RT rolling and CT rolling, but more side cracks are formed during the latter, which indicates that the rollability of the material deteriorates at CT.Numerous {10-11} contraction and {10-11}-{10-12} double twins are formed in both the rolled RT material and the rolled CT material because the grains of the initial material are favorably oriented for{10-11}twinning under rolling.The formed twins are more homogeneously distributed in the rolled RT material than in the rolled CT material and the number of twins is considerably higher in the former.This lower distribution homogeneity of twins in the rolled CT material is attributed to the larger extent of plastic deformation instability during rolling at CT, which results in the concentrated formation of twins along the shear bands.The static recrystallization induced during subsequent annealing is more pronounced in the rolled RT material because of its higher stored strain energy, which consequently leads to the formation of a fine and more homogeneous microstructure after annealing.The annealed CT samples have an obvious bimodal structure consisting of fin recrystallized grains and coarse unrecrystallized grains,which is a result of the localized recrystallization in the highly twinned regions.The hardness of the annealed RT samples is higher than that of the annealed CT samples owing to the fine grain structure of the former.These results demonstrate that rolling at RT is advantageous over rolling at CT in terms of the rollability of the material during rolling,the microstructural homogeneity of the rolled material, and the grain size and hardeness of the rolled material after subsequent annealing.
.Prime Novelty Statement
Declaration of Competing Interest
The authors declare that they have no conflic of interest.
Acknowledgments
This work was supported by National Research Foundation of Korea (NRF) grants funded by the Korean government(MSIP, South Korea) (No.2019R1A2C1085272).
Journal of Magnesium and Alloys2020年2期