III. RESULTS AND DISCUSSION

Figure 2 shows the surface SEM image and the corresponding orientation image maps of the loading direction (LD) and normal direction (ND) before applying load. The red, green, and blue colors represent the <001>, <110>, and <111> crystallographic directions, respectively. Figures 2(a) and 2(b) show that the specimen is perfectly a single crystal with the initial orientation of {001}<110> before tensile loading. We also checked the microstructure and orientations of other regions of the specimen, showing the same result as Fig. 2. Therefore, by using aluminum single crystals, we could disregard the effect of pristine grain boundaries on DRX interrupting dislocation movement.

FIG. 2.

SEM image showing the surface of the specimen before applying load and corresponding orientation maps of (a) LD and (b) ND.

The specimen was uniaxially elongated using the in situ tension device at room temperature. Figure 3 exhibits the SEM images at each deformation step showing the relative displacements of the marked indents. It reveals that the minor strain was approximately 1.2%, whereas the major strain was 30%, representing that the plane strain condition was well satisfied. Necking of the specimen took place at 23% strain and a crack started propagating at 27% strain. EBSD analysis was performed along the crack propagation direction at 30% strain. The dynamically recrystallized grain structure was fully developed around the crack propagation route. It should be noted that the strain of 30% is quite smaller than that for inducing DRX in other previous researches, which may disregard the self-heating effect on DRX. Figure 4(a) shows the measured orientation image maps of the LD and ND in the necking/crack region of the specimen. It can be seen that the blue and pink colors, which represent the <111> and <112> directions, are dominant in the LD and ND maps, respectively. This means that the initial {001}<110> orientation rotates toward the stable orientation of {112}<111> during the plane strain tension.^{Reference Han, Jee and Oh26,Reference Kim, Kim, Kim, Rollett, Oh and Han27} The left side of the crack area represented by blue color in the ND map might be due to the surface inclination by necking.

FIG. 3.

In situ SEM images at strains of (a) 0%, (b) 10%, (c) 20%, and (d) 30% showing the relative displacements of the marked indents.

FIG. 4.

(a) Orientation maps of dynamic recrystallized grains along the necking area at a strain of 30% and (b) inverse pole figures showing the orientation distribution of the dynamically recrystallized grains.

In addition to the changes in crystallographic orientation, interestingly, newly created grains with sizes varying from 100 to 300 μm were observed along the crack propagation route and had orientations that differed from the surrounding orientation. The corresponding orientation images reveal that the DRX could take place at room temperature during plastic deformation in regions severely deformed locally, such as crack propagation routes or necking areas. Also, another specimen with the initial high-index orientation of {114}<252> showed the same DRX behavior along the crack propagation route (not shown here). The recrystallized grains shown in Fig. 4(a) have directionality, such that the recrystallized grains grow along the directions with angles less than 25° from the crack propagation direction. Furthermore, recrystallized grains were observed not only around the crack but also at the front of the crack tip shown in Fig. 4(a). It could be considered that new grains with HAGBs formed and grew preferentially at the necking region, after which a crack propagated across the recrystallized grains. Figure 4(b) shows inverse pole figures with the orientation distribution of the recrystallized grains to LD and ND. The {001}<100> cube texture seems to be slightly dominant, but it is more reasonable that the overall texture is relatively random. The details about the directional grain growth and random texture evolution during DRX will be discussed in the future.

To understand the microstructural evolution in the nucleation stage of the new grains, the most severely deformed region marked by the white circle in Fig. 4(a) was selected and analyzed using a TEM. Figures 5(a) and 5(b) show the cross-sectional TEM images of the selected region. It can be seen in the images that the pure aluminum single crystal was divided into small subgrains, some of which have distinct grain boundaries. Also, a dislocation network with the appearance of grain boundaries was discovered, as indicated by white arrows in Fig. 5(b). This can be considered significant evidence showing the formation of subgrain boundaries by dislocation accumulation. Figure 5(c) shows the orientation map and misorientation line scan graph of the selected area of Fig. 5(a) (red square), which was analyzed by a high-resolution TEM-EBSD system to accurately measure the misorientation between subgrains. From the misorientation line scan, it was seen that the misorientation varied from 1 to 14° indicating the existence of at least two different types of subgrain boundaries, which were possibly evolved by different dislocation accumulation mechanisms. These boundaries have been termed incidental dislocation boundaries (IDBs) and geometrically necessary boundaries (GNBs), in which IDBs form by networking of dislocations into low-energy configurations and GNBs form between regions of different strain patterns to accommodate the accompanying difference in lattice rotation.^{Reference Hughes, Hansen and Bammann28–Reference Mariani, Mecklenburgh, Wheeler, Prior and Heidelbach30} Hughes et al.^{Reference Hughes, Hansen and Bammann28} suggested that average misorientation angle increases more rapidly with strain for GNBs than IDBs in Al and Ni polycrystals. So, we can propose that the very LAGB (θ < 5°) shown in Fig. 5 may be comparable to IDBs, although the LAGB (5° < θ < 15°) may be GNBs formed to accommodate large lattice rotation. As strain increases, dislocations forming the GNBs can develop into HAGBs, which are equivalent to ordinary grain boundaries. CDRX can be another possible mechanism for the formation of HAGBs because it is described by the accumulation of dislocations within the crystallites, the generation of LAGBs, the subsequent increase of their misorientation angles, and their possible transformation into HAGBs. However, McQueen and Kassner^{Reference McQueen and Kassner31} suggested that CDRX is not a viable concept for deformation in aluminum single crystal because CDRX invokes continuously increasing misorientation of dislocation walls, which is inconsistent with TEM observations. Furthermore, several researchers reported that CDRX barely take place if purity of aluminum is higher than 99.999%.^{Reference Yamagata32,Reference Chovet, Gourdet and Montheillet33} Therefore, it can be considered that the subgrains with GNBs were transformed into grains with HAGBs with increasing strain, and the small grains with HAGBs could have acted as nuclei for DDRX.

FIG. 5.

TEM images and orientation map of the severely deformed region marked by a white circle in Fig. 3(a). (a) Bright-field image of the severely deformed region, (b) high-angle annular dark-field image showing dislocation network and LAGBs, and (c) orientation map and misorientation line scan graph indicating coexistence of IDBs and GNBs.

To investigate the DRX procedure after the nucleation of grains with HAGB, the region highly deformed by 30% indicated by a white square in Fig. 4(a) was selected. The region was deformed additionally by 3% using the in situ tension device, and then its microstructure evolution was examined using an EBSD. Figures 6(a) and 6(b) show the ND orientation maps of the area deformed by 30% and an additional 3%, respectively. The black, green, and white lines in the figure indicate very LAGB (θ < 5°), LAGB (5° < θ < 15°), and HAGB (θ >15°), respectively. As shown in Fig. 6, several interesting and significant behaviors can be observed. Grains 1 and 2 were separated at a strain of 30%, but they were connected during additional deformation, which is due to the highly deformed region between two grains (grains 1 and 2) being transformed into a strain-free region by grain growth via the movement of HAGBs. In addition, the small grain 3 was nucleated on the grain boundary of grain 1 at 30% strain and grew by additional deformation. The grain boundary of grain 3 consists of two parts of LAGB and HAGB, and the average misorientation between grain 3 and the surrounding deformed region is approximately 13°. The two results represent that the dynamically nucleated grains at the severely deformed region grew during additional deformation, which is in agreement with the mechanism of DDRX. The fast movement of HAGBs during DDRX, as shown in Fig. 6, must be caused by the high purity of the aluminum.^{Reference Montheillet and Le Coze9,Reference Yamagata17,Reference Yamagata32–Reference Zhang, Weertman and Eastman34} In addition to the grain growth, the generation of new very LAGBs inside the recrystallized grains was observed, representing the creation of new subgrains during further deformation after DRX. In general, it has been known that the dynamic restoration process changes from DDRX to CDRX as the deformation temperature decreases^{Reference Chovet, Gourdet and Montheillet33} and that it is very difficult for DDRX to occur in aluminum at room temperature.^{Reference Yamagata, Ohuchida, Saito and Otsuka35} However, these experimental results consequently show the possibility of DDRX in aluminum at room temperature, which is considered to be due to the stress localization and high purity (99.9999%) of aluminum.

FIG. 6.

Orientation maps of the region marked by a white square in Fig. 3(a) showing the grain growth of dynamically recrystallized grains as increasing strain. (a) 30% strain and (b) 33% strain.