In-situ SEM and EBSD investigation of the deformation behavior of extruded Mg-6Al-1Zn-1.1Sc alloy

The present study subjects the extruded Mg-6Al-1Zn-1.1Sc (wt%) alloy to reveal the deformation during in-situ tensile testing at room temperature by scanning electron microscopy (SEM) and electron backscatter diffraction (EBSD). The results demonstrate that the parallel slip traces are increasingly observed on the surfaces of grains with increasing strains owing to the movement of dislocations inside the grains toward the sample surface, while microcracks are frequently observed at grain boundaries. The slip trace analysis indicate that many basal slips are activated during the deformation. The twinning behaviors of the extruded alloy during tensile testing are dominated by extension twinning. The orientation of grains is demonstrated to have a profound effect on their deformation, where grains with orientations deviating greatest from 〈0001〉//TD exhibit the highest dislocation density after deformation. This can be attributed to the fact that these oriented grains are prone to activate basal slip due to the large Schmid factor (SF). Similarly, the deformation of grains is also found to be highly dependent on their size, where the dislocation density of coarser grains increases more significantly than that of finer grains during deformation because coarser grains have greater space available for accommodating dislocations than finer grains.


Introduction
Magnesium (Mg) alloys are widely used in aerospace, transportation, military, and biomedical fields due to their low density, excellent damping capacity, ease of recycling, and superior biocompatibility [1][2][3]. However, Mg alloys exhibit poor plastic forming ability and low mechanical strengths at room temperature, which significantly hinders the further application of Mg alloys as structural materials [4,5]. The poor plastic deformability of Mg alloys derives from the hexagonal close-packed (HCP) structure of Mg with a c/a ratio of 1.624 [6]. This low symmetry crystal structure results in a much greater critical resolved shear stresses (CRSS) for the non-basal slip systems than that for the basal slip system at ambient temperature, which reduces the likelihood of the non-basal slips being activated during deformation [7,8]. As a result, internal stress is not effectively coordinated along the c-axis of Mg alloys, resulting in poor plasticity [5,8]. Previous researches showed that alloying [9,10] and optimization of processing technology [11,12] could significantly improve the plasticity of Mg alloys. For example, the addition of rare earth elements to pure Mg can weaken the texture [9,13] and reduce the ratios between non-basal slips CRSS and basal slip CRSS, and cross slip barrier. The initiations of non-basal slip systems during deformation can provide more pathways for inducing dislocation slip [14,15], which enhances the plasticity of Mg alloys. Therefore, adding rare earth elements to Mg alloys has received extensive attention. In this respect, Mg alloys containing Sc have been demonstrated to have greater application potential than those with other rare earth elements because Sc has a better grain refining effect, higher solubility in Mg (>25 wt%), and stronger corrosion resistance [8,16].
Grain orientation and size are two important characteristics describing the microstructure of Mg alloys that have critical impacts on the damping properties [17], corrosion properties [18][19][20], tension-compression asymmetry [21,22], strength and plasticity [17,23] of Mg alloys. The most commonly used wrought Mg alloys are mainly produced by extrusion or rolling [21,23,24]. However, wrought Mg alloys usually exhibit a strong texture relative to the extrusion direction (ED) and rolling direction (RD), which includes fiber textures (i.e., (0001)//ED) and plate textures (i.e., (0001)//RD) [21,25]. A considerable amount of work has been done on the mechanical properties and anisotropy of the wrought Mg alloys. For instance, Wang et al [26] found that the tensile properties of an extruded Mg-Gd-Y-Zn-Zr alloy differed substantially in different directions due to texture. Hou et al [23] found that rolled AZ31 alloy plates had a higher work hardening rate under compression loading at temperatures less than 200°C when loaded along the normal direction (ND) due to the difficulty of activating pyramidal slip and contraction twins. By summarizing these studies, it is not difficult to find that the anisotropy in the mechanical properties of wrought Mg alloys with textures can be readily ascribed to the effect of grain orientation on the Schmid factor (SF) of the different deformation modes. Thus, the deformation mechanisms of wrought Mg alloys are different when the alloys are loaded in different directions, which finally reflects differences in mechanical properties. The majority of past studies focused on evaluating the effect of grain orientation on the deformation behavior and mechanical properties of Mg alloys were conducted by sampling Mg alloys with textures from different directions. However, few reports have evaluated the deformation behavior of grains with different orientations in the same sample. Although Zhou et al [27] and Woo et al [28] discussed the deformation mechanisms of grains with different orientations in deformed samples, the contribution of these grains with different orientations to the overall deformation behavior remains unclear.
Additionally, previous researches showed that grain size significantly affected the activation of deformation modes during Mg alloys deformation. Therefore, grain size also exhibits a key influence on the strength and plasticity of Mg alloys. For example, Wei et al [29] reported that the yield strength and uniform elongation of coarse-grained pure Mg (with average grain sizes d ̅ of 125 μm) under tensile loading were 67.7 MPa and 5.3%, respectively. In comparison, the yield strength and uniform elongation of fine-grained pure Mg (d ̅ = 5.5 μm) were 93.8 MPa and 18.3%, respectively. The researchers attributed the simultaneous enhancement of strength and plasticity observed for the fine-grained pure Mg to the activation of non-basal slip and the dissociation of unstable 〈c+a〉 dislocations. Similarly, Luo et al [30] reported that the strength and plasticity of Mg-3Gd alloy also increased significantly with the decreasing of grain sizes. Here, transmission electron microscopy (TEM) demonstrated that the deformation of the grains with grain sizes greater than 10 μm was dominated by 〈a〉 dislocations and extension twins. In comparison, the deformation of grains with grain sizes less than 5 μm was dominated by 〈a〉 and 〈c+a〉 dislocations. Yu et al [31] also reported a strong effect of grain size on the applied stress required to produce deformation twinning in an HCP titanium alloy single crystal. Accordingly, these studies and others [32,33] have demonstrated that grain size predominantly affects the mechanical properties of Mg alloys by affecting the relative frequency with which different deformation modes are activated. However, existing investigations have taken only the average grain sizes into account during the analysis, while the actual sizes of grains in an alloy are always distributed over a range. Therefore, the deformation mechanisms in the same sample of varied sized grains and their relationship between the overall deformation behavior remain poorly understood. Therefore, the present work addresses the above-discussed issues by subjecting extruded Mg-6Al-1Zn-1.1Sc (AZ61-1.1Sc) alloy, and SEM and EBSD are employed to investigate its microstructure evolution and deformation behaviors during in-situ tensile testing at room temperature. The results presented herein clarify the influences of grain orientations and size on the deformation behaviors of Mg alloys and thus support the further application of Mg alloys.

Experimental procedures
The AZ61-1.1Sc alloy ingots were prepared by casting a mixture of pure Mg (99.99 wt%), pure Al (99.99 wt%), pure Zn (99.99 wt%), and Mg-20Sc (wt%) master alloys. The surfaces of the pure Mg ingots were first ground to remove the oxide layer. Then, the raw materials were held at a temperature of 300°C for 3 h to ensure that moisture was fully ejected. Alloy ingots were smelted in an electric resistance furnace under an atmosphere composed of Ar and SF 6 in a ratio of 99:1 by volume. First, the Mg ingots were melted at a temperature of 750°C, and the Mg-20Sc alloy, Al, and Zn were added successively. Next, the melted metals were mechanically stirred evenly, and the scum arising on the surface of the melt was removed. Then, the temperature was reduced to 730°C, and the refining agent was added to further refine the melt. After refining and holding the melt at 730°C for 30 min, the melt was cast into a steel mold preheated at 300°C to obtain alloy ingots. The ingots were allowed to cool naturally to room temperature in air, and were then cut into cylindrical billets with dimensions of f46 mm ×70 mm. The billets were homogenized, which then served as extrusion samples. Before extrusion, the surface of the extrusion die was coated with graphite for lubrication. Then, the extrusion die with an extrusion ratio of 13:1, the homogenized billet, and the extrusion head were preheated to the extrusion temperature of 380°C and held at that temperature for 90 min. The preheated billet was extruded through the extrusion die using an IM-Y300 four-column hydraulic press. The extruded alloy sample with a diameter of 12.8 mm was aircooled to room temperature.
Specimen for in-situ tensile testing with the shape and dimensions illustrated in the inset of figure 1 was cut from the center of the extruded bar by wire-electrode cutting. Hence, the gauge of the specimen is 5.5 mm (L) × 2.0 mm (W) × 1.5 mm (T). The metallographic etchant was a picric acid solution. A Leica DMI3000I inverted microscope was employed to observe the metallography. After mechanical polishing of the tensile specimen, 7 vol% perchloric acid solution was used for electrolytic polishing at −37°C for 60 s for in-situ characterization. The polishing voltage was 20V, and the polishing current was 0.20A. The tensile tests were conducted within the vacuum chamber of a Zeiss Merlin compact field emission SEM equipped with an EBSD probe to enable the microstructure of the specimens to be characterized by SEM and EBSD. The tensile rate was 0.1 mm min −1 , and the loading direction was parallel to the ED, as indicated in the inset of figure 1. A 20 kV working voltage was applied during EBSD data acquisition, and the scanning step was 0.4 μm. The EBSD data was processed using Oxford Instruments Aztec Crystal software. Figure 1 shows the in-situ tensile engineering stress-strain curve of extruded AZ61-1.1Sc alloy. The yield strength, ultimate tensile strength, and elongation to fracture of the alloy under in-situ tensile loading are 201.1 MPa, 308.1 MPa, and 25.9%, respectively. As can be seen from the stress-strain curve, the tensile loading was held when the strains (ε) were 0.0%, 2.5%, 6.5%, 10.3%, and 16.3% for conducting microstructural characterization by SEM and EBSD. Of note, only the SEM characterization was conducted when the engineering strain was 16.3%, owing to the deterioration of the EBSD resolution due to sample deformation.

Initial microstructure
show the optical and SEM images, band contrast map, and grain size distribution histogram of the extruded alloy, respectively. The microstructure of the alloy is composed of equiaxed grains of different sizes distributed from 0 to 45 μm, and the average grain size of the alloy was calculated to be 13.96 μm. Additionally, two types of secondary phases are observed in figure 2(b), where a previous report [34] has demonstrated that the coarser blocky secondary phase is Al 3 Sc, and the finer secondary phase distributed along the extrusion zone is mainly Mg 17 Al 12 .  Figure 3 presents the SEM images of the alloy at different strains. As can be seen, the specimen surface is relatively flat prior to loading, and the specimen surface exhibits slight relief at a strain of 2.5%, illustrating that tensile loading causes the grains to rotate. Meanwhile, the areas outlined by the yellow ellipses in figure 3(b) exhibit the presence of parallel slip traces on the surfaces of some grains, which are observed as straight lines resulting from the intersection of active slip planes with the sample surface. These traces are caused by the movement of dislocations inside the grains toward the specimen surface, indicating that many dislocations are activated at this relatively low level of strain. The surface relief becomes increasingly evident as strain increases to  6.5%, and some micro cracks are observable at the grain boundaries, as indicated by the yellow arrows in figure 3(c). At a strain of 10.3%, there are more obvious slip traces in the grains, and the different directions of these traces are highlighted by the red lines in figure 3(d). It can also be seen that the slip traces in the same grains always lie parallel to each other, and follow a single direction. This demonstrates that the dislocation slip in each grain is mainly dominated by a single slip mode [35]. These slip traces are analyzed in detail in section 4.1. Meanwhile, the increasing slip trace densities observed in the grains with increasing strains suggest that the movement of dislocations inside the grains also becomes more active. In addition, the number and length of microcracks also increase with increasing strains, which indicates that an increasing level of applied stress causes the intensity of incompatible deformations within some of the intergranular regions to increase under the effect of stress, causing microcracks to initiate and propagate at the grain boundaries [4,36,37]. Finally, the SEM image shown in figure 3(e) presents a seriously uneven surface at a strain of 16.3% due to the considerable degree of plastic deformation.

Microstructure evolution during in-situ tensile
The IPF maps and (0001) pole figures of the alloy at different strains are shown in figure 4. After the first step of deformation, the characteristics of the IPF map change little. However, apparent color gradients appear in some of the grains when strain increases to 6.5%. This indicates that some of the grains undergo uneven plastic deformation. Moreover, a comparison of figures 4(c) and (d) indicates that the emergence of color gradients increases with increasing strain. Furthermore, before loading, the extruded AZ61-1.1Sc alloy exhibits typical fiber texture characteristics (i.e., (0001)//ED). Figure 4(e) presents that tensile loading has a little observable effect on the texture type and intensity. The slight difference in the texture intensity may be ascribed to the activation frequency of different deformation modes or the slight distinction in the scanning area [37,38]. Therefore, the influence of tensile deformation on the texture of the alloy could be ignored.
Figures 5(a)-(d) shows the Kernel average misorientation (KAM) maps of the alloy at different strains. The dislocation density distribution, that is, the geometrically necessary dislocation (GND) density in the deformed microstructure, can be directly reflected by the KAM map [8,29]. The KAM values observed in figure 5(b) at a strain of 2.5% are generally similar to those in figure 5(a) before loading, except that green highlights indicative of high KAM values tend to accumulate near the grain boundaries. This trend continues under increasing strain, as shown in figures 5(c) and (d). This is because the grain boundaries can effectively block the movement of dislocations to neighboring grains, which causes dislocations to accumulate at these boundaries. The equation applied for calculating the GND density ( GND r ) is given as follows: where θ denotes the degree of local misorientation obtained from the KAM map, μ is the scanning step size  as a function of strain. Accordingly, it can be concluded that the dislocation density in the alloy increases significantly with increasing deformation. Figures 5(e)-(h) show the grain boundary maps of the alloy during in-situ tensile. As shown in the bottom right-hand corner of figure 5, the black line represents high-angle grain boundaries (HAGBs) with a degree of local misorientation that is greater than 10°, the green line indicates low-angle grain boundaries (LAGBs), and the line marked in red is extension twin boundaries (ETBs). It is noted that the LAGBs in the grain boundary maps are mainly distributed near initial HAGBs. This is because the dislocations accumulating at the grain boundaries become entangled to form substructures, and the LAGBs serve as the boundaries between the substructures and the original grains. The corresponding distribution maps of misorientation angle at different strains are plotted in figures 5(i)-(l). The results demonstrate that the proportion of HAGBs gradually decreases and the proportion of LAGBs increases progressively with increasing strain. Furthermore, the increasing proportion of ETBs with increasing deformation illustrates that these twins are constantly nucleating and propagating during deformation. The results in figures 5(e)-(l) were further analyzed to provide the total proportions of HAGBs, LAGBs, and ETBs, which are plotted in figure 5(n) as a function of strain. In particular, it is noted that the ETBs increases prominently at a strain of 6.5% to coordinate the plastic deformation.
To further investigate the twinning behaviors of the extruded alloy during in-situ tensile. Figure 6(a) shows the twinning evolution of four grains in the alloy. Among these, twins exist in grains A, B, and C at a strain of 6.5%, while twin appears in grain D when the strain is 10.3%. Figure 5(n) also shows that the twin content increases substantially at a strain of 6.5%. That can be analyzed from figure 1, which indicates that the first stage of deformation of the alloy is mostly elastic deformation, and the plastic deformation during this stage is very limited. Therefore, the content of deformation twins in the alloy does not increase dramatically until the deformation becomes dominated by plastic deformation. Additionally, the SFs of extension twinning in grains A-D were calculated, and the values are presented in figure 6(b), which shows the orientations of the parent grains and their twins in the (0001) pole figures. Although the CRSS of the extension twin is relatively low and easily activated [3], the corresponding SFs calculated here are small. Therefore, the activation of these twins requires a relatively large load. The misorientation angles along arrows 1-4 marked in figure 6(a) are plotted in figure 6(c). The misorientation angles are all about 86°at the twin boundaries, which further indicates that these are extension twins. Meanwhile, few grains are observed to include any other type of twins than extension twins. Therefore, the twinning behaviors of the extruded AZ61-1.1Sc alloy during tensile are dominated by extension twinning.

Slip trace analysis
Slip trace analysis is an effective and widely reported method used in conjunction with SEM to determine the modes of slip for individual grains during in-situ deformation [3,9,15,38,39]. The Euler angles representing the orientation of each grain can be obtained through EBSD and then inputting the Euler angles into a MATLAB code for a calculation to obtain the theoretical slip trace direction of each slip system of the grain [3,15,40].  [41,42]. Therefore, these results verify the results obtained herein by applying the slip trace analysis. The same process was applied to the collection of slip traces marked with red lines in figure 3(d). The results demonstrate that these slip traces are also highly consistent with the basal slip. Hence, basal slip appears to be activated in a high proportion of grains during the tensile deformation of the AZ61-1.1Sc alloy.

Effect of grain orientation on deformation
The influence of grain orientation on the deformation behavior during tensile is evaluated by dividing the microstructure into three categories according to the angle by which a grain deviates from the 〈0001〉//TD.   (2) are presented in figures 9(g)-(i). As can be seen, the R KAM values obtained for grains deviating from the 〈0001〉//TD within angle ranges of 0°-30°, 30°-60°, and 60°-90°are 81.3%, 91.5%, and 104.2%, respectively. These results clearly demonstrate that grain orientations have a significant impact on the degree of grain deformation under tensile loading, and the contribution to the overall plastic deformation is different as well. Moreover, this effect is most remarkable for grains that deviate most from the 〈0001〉//TD. It is well known that the SF obtained under a given loading direction depends on the grain orientation [43,44]. Hence, the deformation mode responses of grains with different orientations are generally different under the same loading conditions. Therefore, subjecting the microstructure to SF analysis prior to deformation helps to determine the deformation mode initiated during deformation. The SF maps with SF distributions in the insets corresponding to the texture component maps in figures 8(a)-(c) are respectively shown in figures 10(a), (d), and (g) for the basal 〈a〉 {0001}〈1120 ̅ 〉 slip system, while those obtained for the prismatic 〈a〉 {101 0 ̅ }〈1120 ̅ 〉 slip system are presented in figures 10(b), (e), and (h), and those obtained for the pyramidal 〈c +a〉 {1122 ̅ }〈1123 ̅ 〉 slip system are plotted in figures 10(c), (f), and (i). It can be seen from the SF distributions in the insets that the SFs of the basal slip are most affected by the deviation angle relative to the 〈0001〉//TD, where the SFs of the basal slip increase dramatically with increasing deviation angle. In contrast, the SFs of the other slip systems are relatively insensitive to the deviation angle. Therefore, the deviation angles relative to the 〈0001〉// TD can affect the activation of the basal slip. Moreover, the low CRSS makes basal slip easily activated [3], and the higher SFs associated with basal slip in the grains with the greatest deviation from the 〈0001〉//TD can trigger basal slip in a high proportion.
In-grain misorientation axes (IGMA) analysis based on EBSD proposed by Chun et al [45,46] is widely used to determine the main deformation mode in deformed grains. Previous reports indicated that the lattice rotation axes corresponding to basal slip, prismatic slip, and pyramidal 〈c+a〉 slip are 〈1 1 00 ̅ 〉, 〈0001〉, and 〈1 100 ̅ 〉 respectively [1,47,48]. The results of IGMA analysis obtained during in-situ tensile for all grains and for only those grains with deviations from the 〈0001〉//TD of 0°-30°, 30°-60°, and 60°-90°are presented in figure 11. As can be seen, the IGMA distributions of all grains gradually evolve from random distributions before loading to 〈uvt0〉 distributions at 10.3% strain. This illustrates that the overall deformation is dominated by basal 〈a〉 and pyramidal 〈c+a〉 slip systems. However, pyramidal 〈c+a〉 slip is not generally activated during deformation at room temperature due to its high CRSS [49]. Therefore, combining these results with the results of slip trace analysis, it can be determined that the extruded AZ61-1.1Sc alloy deformation modes are dominated by basal slip. Moreover, the same trend is basically observed for the grains with different orientations.
Despite these general trends, it is further noticed that a major difference in this trend is found at 10.3% strain for the grains deviating 0°-30°from the 〈0001〉//TD, where a concentrated distribution of IGMA is also observed on the 〈0001〉 axis, as shown by the dotted ellipse in figure 11. This concentrated distribution demonstrates that the prismatic slip is activated in these grains, as indicated in a past study [47]. According to the SF calculation results presented in figure 10(b), the mean SF of the prismatic slip in grains with deviation angles of 0°-30°is 0.46, while the corresponding mean SF of the basal slip is only 0.16. Therefore, these grains with low deviation angle are conducive to activating prismatic 〈a〉 dislocations. Moreover, while the CRSS of the prismatic slip is also much greater than that of the basal slip, the applied stress is as great as ∼270 MPa when the strain is 10.3%. This issue can be clarified based on Schmid law, which defines the minimum applied stress ( ) s required to activate a specific slip system as follows [50]: where m represents the SF of the corresponding slip system. According to the previous reports, the CRSS of prismatic 〈a〉 slip in Mg alloy matrix is 39-100MPa [15,[51][52][53]. Thus, it can be estimated that the applied stress required to activate prismatic slip is about 84.8-217.4 MPa. Therefore, the applied stress at 10.3% strain (∼270 MPa) exceeds the critical stress required for activating prismatic slip in grains with deviation angles of 0°-30°from 〈0001〉//TD.

Effect of grain size on deformation
As discussed above, the primary deformation mechanisms of Mg alloys include twinning and dislocation slips. The effect of grain size on twinning is relatively simple, where twinning is more likely to occur during coarsegrained alloy deformation [54,55]. However, the effect of grain size on the activation of the dislocation slip mechanism has been shown to be more complex [33]. Some studies have reported that fine grains are more conducive to activating 〈c+a〉 dislocations [29,30], while other investigations have demonstrated the opposite conclusion [56]. Additionally, several studies have carefully analyzed the synergistic effect of coarse and fine grains on the deformation behaviors of Mg alloys, and have observed that these grains play different roles, such as enhancing the plasticity by coarse grains and improving the strength by fine grains [7,57,58].
To further explore the relationship between grain size and deformation, figure 12 shows the relationship between average grain orientation spread (GOS) and grain size at different strains during in situ tensile. GOS can be used to determine the strain level of grains, which can be obtained by the following equation [59]: where n is the number of pixels in grain Gi, ij w is the misorientation between point i and adjacent point j based on EBSD data. In general, high GOS represents high dislocation density or more substructure in grains, because an increasing local misorientation is indicative of increasing strain in alloys [60,61]. Figure 12 shows that the average GOS of each grain in the microstructure generally increases substantially with the increased strain. Moreover, the average GOS of coarse grains is observed to increase more dramatically than that of fine grains.  The impact of grain size on the average GOS during in-situ tensile can be evaluated by applying linear fitting to the results in the individual plots in figure 12 and comparing the slopes. From this, it is found that the slope of the linear fitting to the results in figure 12(a) obtained before deformation was only 0.057, but this slope increased to 0.098 at 10.3% strain in figure 12(d). Accordingly, these results suggest that the degree of deformation in coarse grains is greater than that in fine grains, and the stress level of these coarse grains increases more significantly with increasing strain.
Previous studies have shown that coarse grain deformation is mainly dominated by basal slip and extension twinning [29,30,62]. Moreover, the CRSS of these modes is extremely low at ambient temperature [49,51]. Therefore, these deformation modes are readily activated, such that the internal strain level of coarse grains is greater than finer grains. Zhu et al [33] simulated the plastic deformation of Mg-3Al-3Sn (wt%) alloy with different grain sizes based on an EVPSC model. The results demonstrated that relatively coarse grains were more conducive than finer grains to the nucleation of dislocations, such that more intragranular stress variation occurred within coarse grains. During the in-situ tensile of Mg-0.47Ca (wt%) alloy, Zhu et al [63] found that coarse grains underwent reduced grain rotation than finer grains. This was attributed to the activation of multiple slip systems in different regions of the coarse grains, which could partially nullify each other and thereby reduces the overall rotation of the whole grain.
This issue is further examined by dividing the microstructure into three parts according to grain sizes ranging from 0 μm to 15 μm (18.1% of the grains), 15 μm to 30 μm (70.8% of the grains), and 30 μm to 45 μm According to figures 13(j)-(l), the R KAM values of the grains with grain sizes ranging from 0-15 μm, 15-30 μm, and 30-45 μm are 90.2%, 95.8%, and 93.8%, respectively. These results further suggest that the dislocation density of coarse grains increases more greatly than that of finer grains during deformation, which is consistent with the GOS analysis results in figure 12. However, Guo et al [57] determined that the dislocation density and degree of deformation of coarse grains in extruded Mg-Sn-Bi-Mn alloy were less than those of finer grains. Nonetheless, these authors relied on a comparison of coarse grains and fine grains in the KAM maps of fracture samples examined ex-situ, and the results therefore cannot accurately reveal changes in the dislocation density before and after deformation. Zhang et al [58] analyzed the microstructural evolution of an Mg-8Al-2Sn-1Zn alloy with bimodal structure at different tensile strains. In this case, the dislocation density of coarse grains increased slowly at the initial stage of tensile, and then increased significantly at the later stage of deformation, while the fine grains followed the opposite trend. The extent to which GND r increased in coarse grains at 14% strain was lower than that observed in finer grains. Li et al [7] obtained similar results when analyzing the microstructure of an extruded Mg-5Li-3Al-2Zn alloy with heterostructure at different tensile strains. However, the component of the microstructure is not entirely consistent for different samples, and the slight difference in the sampling locations also generated differences in the final results.
In the current study, the extent to which GND r increased is greater for relatively coarse grains than for finer grains at a strain of 10.3%. Moreover, the results in figures 13(g)-(i) illustrate that the KAM values in the fine grains are relatively uniform, indicating that dislocation is close to saturation in these fine grains. In fact, the rapid saturation of dislocations within fine grains would be expected with increasing strain, while the greater area of coarse grains would have more space to accommodate newly generated dislocations, such that dislocation could continue to increase within coarse grains with increasing strain [58]. This accommodation of newly generated dislocations would be further facilitated by the fact that the high KAM values in coarse grains are mainly distributed near the grain boundaries. Thus, coarse grains have enough area available for accommodating dislocations. Therefore, the dislocation density in coarse grains would continuously increase with increasing strain, while that in fine grains would be difficult to increase due to saturation. The ability of the coarse grains to accommodate more dislocations during deformation suggests that coarse grains would mainly enhance plasticity for a microstructure composed of coexisting coarse and fine grains. Figures 14(a)-(c) show the SF distribution of microstructures with grain sizes ranging from 0-15 μm, 15-30 μm, and 30-45 μm, respectively. It can be seen that the SF distributions and mean SF values of the three slip systems are practically independent on the grain size, where the differences observed for the grain size range of 30-45 μm are primarily due to the relatively small number of grains available for analysis within the grain size range. Therefore, an analysis of the influence of SF on R KAM is excluded here, and the major factor responsible for the differences in R KAM observed in figures 13(j)-(l) is the grain size itself. In addition, the IGMA analysis results obtained during in-situ tensile for grains with grain sizes ranging from 0-15 μm, 15-30 μm, and 30-45 μm are presented in figure 14(d). As can be seen, the IGMA evolution observed here is consistent with the total IGMA evolution at the top of figure 11. Therefore, the overall deformation mechanism of the microstructures with different grain sizes exhibits no obvious differences.

Conclusions
In this study, the extruded AZ61-1.1Sc alloy was studied by SEM and EBSD during in-situ tensile. The main conclusions of the research can be given as follows.
(1) The surface relief of the sample increases with increasing strain due to the rotation of grains, and the movement of dislocations inside the grains toward the sample surface results in the formation of parallel slip traces. Microcracks begin initiating at the grain boundaries on the sample surface at a strain of 6.5%, and the density of microcracks and their extent of propagation increase with further increasing strain.
(2) The results of slip trace analysis demonstrate that basal slip is frequently activated during deformation. Further evidence associated with the IGMA analysis of grains with different grain orientations and size ranges illustrate that the deformation of the alloy is dominated by basal slip.
(3) The microstructure with grain orientations that deviate from the 〈0001〉//TD of the extruded alloy by a deviation angle of 60°-90°exhibits the highest dislocation density after deformation because these grains have a higher SF of basal slip. Moreover, the rate of increase in the mean KAM value of these grains from 0% strain to 10.3% strain is 104.2%, which is much higher than that of 93.8% observed for the overall microstructure.
(4) The dislocation activity is greater in coarser grains than in finer grains, and the dislocation density of these coarser grains increases more significantly with increasing strain than that of finer grains.