Ultimate Rolling Texture in Pure Aluminum Highly Deformed by Accumulative Roll Bonding: Taylor Orientation Formed beyond Grain Subdivision
2017 Volume 58 Issue 8 Pages 1127-1133
Details
2017 Volume 58 Issue 8 Pages 1127-1133
The texture evolution of an accumulative roll bonding (ARB) processed commercial purity aluminium with lubrication was measured by electron backscattering diffraction with a field emission-type scanning electron microscope. As a result of texture analysis using orientation distribution function maps, the highest intensity texture component was found to be Taylor orientation after ARB 6 cycle. The result suggests that the deformation of ultrafine grains is similar to Taylor model, which occurs in not only high purity aluminium but also commercial purity aluminium.
An accumulative roll bonding (ARB) process was established as one of severe plastic deformation (SPD) processes, which is used for fabricating ultra-fine grained metals having grain size less than a few μm1–3). Although the microstructure evolution during the SPD processes is important, it has not been fully understood since it is essentially impossible to observe or measure the microstructure evolution during the SPD processes including the ARB processes. Therefore, the microstructure evolution has been investigated based on the observation or measurement before and after a SPD process, and, more information is requested.
The texture analysis can provide additional information about the microstructure evolution during the SPD processes including the ARB process. As shown in Fig. 1, the ARB process contains rolling of which texture has been extensively studied4–6). Strictly speaking, the ARB process consists of four steps; cutting a metal sheet into two, surface treatment in order to remove the surface contamination and oxide layers, stacking the two plates, and roll bonding with the thickness reduction of 50% in one pass. Thus, the thickness of the metal sheet is always the same as that before the ARB process. Table 1 shows the relationship between the effective thickness reduction and the number of ARB cycle n.
Schematic diagram of accumulative roll bonding process.
| Number of ARB cycle, n | Effective rolling reduction (%) |
|---|---|
| 0 | 0 |
| 1 | 50 |
| 2 | 75 |
| 3 | 87.5 |
| 4 | 93.75 |
| 5 | 96.875 |
| 6 | 98.438 |
| 7 | 99.219 |
| 8 | 99.609 |
| 9 | 99.805 |
| 10 | 99.902 |
| 11 | 99.951 |
| 12 | 99.976 |
| 13 | 99.988 |
| 14 | 99.994 |
| 15 | 99.997 |
The orientation of a rolled sheet is normally represented as $\{ {\rm h \ k \ l} \} \langle {\rm u \ v \ w} \rangle $, here $\{ {\rm h \ k \ l} \}$ and $\langle {\rm u \ v \ w} \rangle$ represent the Miller index of the crystallographic plane and direction parallel to rolling plane and rolling direction (RD), respectively. Taylor orientation $\{4 \ 4 \ 11 \} \langle 11 \ 11 \ 8 \rangle$ which has been also known as Dillamore orientation and theoretically predicted as the destination of the rolling texture of fcc metals by Dillamore and Kato7). Furthermore, Taylor orientation has also appeared in the simulation5). However, Taylor orientation has never been experimentally observed as the major component of texture in rolled fcc metal sheets, and therefore, it has been the common sense that copper $\{1 \ 1 \ 2 \} \langle1 \ 1 \ 1 \rangle$ orientation is the major component of fcc rolling texture8).
However, the maximum rolling reduction which is defined as the reduced thickness due to the rolling deformation divided by the thickness before rolling was normally up to around 95%, and, up to around 99% as a maximum. Such rolling reduction has been enough for obtaining commercial metal sheets with the thickness of mm order. If the thickness reduction is much higher than above values, metal foils tend to be obtained, but, there is a possibility that the texture of metal foils might be different from that of metal sheets.
There are two types of surface condition for the rolling mill for the ARB process; one is without lubrication and the other is with lubrication by machine oil. The difference between the two surface conditions is whether the friction between the surfaces of the metal sheets and the rolls is high or not. The high friction between surfaces gives additional shear strain close to the surface of the metal sheets9). As a result, the ARB process without lubrication provides metal sheets having folded layer formed by the additional shear strain, whereas, the ARB process with lubrication has almost no layer formed by the additional shear strain. It should be pointed out that the initial report of the ARB process was carried out without lubrication in order to obtain ultrafine grains (UFG) with less number of ARB cycle using the additional shear strain2). However, the texture is different between with and without lubrication due to the additional shear strain10–12).
There were several reports about texture of fcc metal sheets processed by ARB without lubrication, such as, A1100 (purity is about 99%) and A5083 (Al-Mg alloy) sheets up to 6 cycles13–15), and A8011 (Al-Li alloy) sheets up to 15 cycles. The reports concluded that Taylor orientation is one of the major components in addition to S component $\{1 \ 2 \ 3 \} \langle 6 \ 3 \ 4 \rangle $ and copper component. There was also a report about texture of Al-0.2%Sc alloy processed by ARB with lubrication up to 10 cycles16), which shows that solution treated Al-0.2%Sc alloy shows relatively strong intensity of Taylor component after the ARB process in addition to copper component.
It should be emphasised again that the texture of ARB without lubrication is strongly affected by the additional shear strain. Actually, the grain refinement is much faster when the additional shear strain exists. Therefore, the observed Taylor component in Al and Al alloys processed by ARB without lubrication might be the results of additional shear strain.
Lately, the formation of dominant Dillamore orientation in aluminium having 99.99% purity (4N-Al) fabricated from multi-layered thin Al sheets by ARB (multi-layered ARB) was reported17). It should be pointed out that such multi-layered ARB process is different from the conventional ARB process, but, it can lead plane strain compression even the surface of rolls are not lubricated by machine oil. The dominant Dillamore orientation in multi-layered 4N-Al is attributed to the deformation of UFG aluminium close to the Taylor model and the movement of Y-type junction. It is well known that Taylor model for the fcc metals has three assumptions18). First, all the grain experiences the same deformation as the polycrystalline object. Second assumption is that at least five slip systems are activated in order to satisfy the first assumption. Finally, the slip systems are selected following the minimum work principle. In other words, if Taylor texture is found in cold rolled polycrystalline metals, the deformation mode is comparable to the Taylor model. However, it is also known that purity of metal affects the formation of UFG metals, such as, the minimum grain size is saturated above certain effective strain and the value depends on the purity. Thus, the effect of purity for the formation of Taylor orientation is still interesting.
The texture of commercial purity aluminium sheet (A1100 alloy) processed by the ARB with lubrication up to 15 cycles was reported19). It should be noted that the effective rolling reduction of ARB 15 cycles is about 99.997% which is strictly speaking 99.99695%, as shown in Table 1. Such extremely high effective rolling reduction corresponds to the thickness reduction from about 32.7 m to 1 mm. The report was concluded that the major component of the texture for ARB processed commercial purity aluminium with lubrication more than 10 cycles seems to consists of copper component based on the analysis with $\{111\}$ pole figures (PFs). However, $\{111\}$ pole figures at higher ARB cycle numbers seem to contain Taylor orientation. Thus, more precise texture analysis can be performed using orientation distribution function (ODF) maps, which can provide more information about texture of ARB processed commercial purity aluminium with lubrication at extremely high effective rolling reduction. The purpose of this study is to provide new information about the formation of UFG using SPD processes such as the ARB process from the texture point of view, especially, whether the concentration of alloying elements is important or not.
As a starting material, A1100 commercial purity aluminium sheets of which chemical composition as shown in Table 2 with 1 mm thickness were annealed at 673 K for 1.8 ks, and then subjected to the ARB process with lubrication. Thus, the starting material had a recrystallized microstructure and texture. The purity of starting material was 99.1%, and hereafter it is denoted as 2N-Al.
| Element | Si | Fe | Cu | Mn | Zn | Al |
|---|---|---|---|---|---|---|
| Wt.% | 0.12 | 0.53 | 0.13 | 0.02 | 0.01 | 99.14 |
In this study, the ARB was carried out using a two-high rolling mill with a roll diameter of 310 mm. The ARB process was performed at room temperature (RT) with lubrication in each ARB cycle, and repeated up to 15 cycles. After each roll bonding, the sheets were water quenched. The equivalent strain ($\varepsilon_{\rm eq}$) of 0.8 was applied at each ARB cycle20). Hereafter, the starting material and a metal sheet applied n-cycle ARB process are denoted by ARB 0c and ARB nc, respectively. The sample coordination was defined by normal direction (ND), transverse direction (TD) and RD.
Microstructures and texture of the ARB processed specimens were measured by electron back-scattering diffraction (EBSD) in a scanning electron microscope with a field emission gun (FE-SEM: Philips XL30) operated at 15 kV. The specimens for EBSD were cut from the sheets using an arc discharge wire cutting machine, and then, prepared using mechanical polishing and then electrolytic polishing. The observations were performed on the TD section (ND-RD planes). As shown in Fig. 2, the region chosen for the EBSD observation was a quarter thickness distance from the surface since it is thought to be the least affected region by the friction between the surface of the rolls and the surface of the plates. It should be noted that the friction cannot be zero even ARB with lubrication, but, can be ignored.19) The size of observed area (ND × RD) and the step size of EBSD measurements are 1400 μm × 200 μm and 1 μm step for 0c, 200 μm × 250 μm and 0.5 μm step for 2c, and 30 μm × 30 μm and 0.05 μm step for 4c, respectively. These are 30 μm × 20 μm and 0.5 μm step for ARB 8c, 12c and 15c.
Schematic illustration of the measurement area for electron backscattering diffraction.
Orientation distribution function (ODF) maps were constructed from the EBSD data in order to perform the quantitative texture analysis by OIM analysis (TSL) using harmonic series expansion with series rank of 22 and Gaussian smoothing of 5°. The Bunge's notation for Euler angles was used in this study. It is noted that the symmetry of rolling was assumed when the texture analysis is performed.
Figure 3 shows the grain boundary map of the commercial purity aluminium ARB processed with lubrication between 0c and 15c. The green and red lines represent high-angle grain boundaries (HAGBs) and low-angle grain boundaries (LAGBs), respectively. The misorientation angle of HAGB is defined equal to and higher than 15 degrees, whereas that of LAGB is defined lower than 15 degrees. The grain boundaries having misorientation angle lower than 2 degrees are ignored due to the accuracy of the measurement. The recrystallized-equiaxial grains at ARB 0c are elongated along RD by roll bonding at each ARB process, and grain thickness is continuously reduced with increasing n. At the same time, the fraction of HAGBs seems to increase after ARB 8c.
Grain boundary map of commercial purity aluminium with lubrication; (a) 0 cycle, (b) 4 cycle, (c) 8 cycle, (d) 12 cycle and (e) 15 cycle.
Figure 4 displays the mean separation of HAGB along ND $d_{\rm ND}^{\rm HAGB}$ versus n, which is evaluated from the grain boundary maps using interception method. The initial $d_{\rm ND}^{\rm HAGB}$ is 27.6 μm, and then, $d_{\rm ND}^{\rm HAGB}$ drastically decreases with increasing n up to around 6 cycles. After ARB 6c, $d_{\rm ND}^{\rm HAGB}$ stops decreasing with increasing n and then saturates with the value at around 0.2 μm. This trend of $d_{\rm ND}^{\rm HAGB}$ is the typical grain size evolution of ARB processed 2N-Al21), which is known as “grain subdivision” during SPD processes22).
Mean separation of high-angle grain boundary along normal direction versus number of accumulative roll bonding cycle.
Figure 5 (a)–(h) show the {111} PFs constructed from EBSD data of ARB 0c – ARB 15c, and, Fig. 5 (i) shows the ideal orientations frequently appeared as major component of the rolling texture; S, brass, copper and cube. It should be noted that some {111} PFs in Fig. 5 were used in the literature19). In this study, the precise texture analysis with the ODF maps during ARB process with lubrication will be performed in addition to the analysis with {111} PFs.
(111) pole figure constructed from electron back scattering diffraction data of accumulative roll bonding processed commercial purity aluminium; (a) 0 cycle, (b) 2 cycle, (c) 4 cycle, (d) 6 cycle, (e) 8 cycle, (f) 10 cycle, (g) 12 cycle, (h) 15 cycle, and (i) ideal orientations of the fcc rolling texture along β-fibre. It should be noted that some of the data can be found elsewhere17).
ARB 0c has strong poles coincide to cube $\{0 \ 0 \ 1 \} \langle 1 \ 0 \ 0 \rangle $ orientation since it is the typical component of rolled recrystallization texture of fcc metals as shown in Fig. 5. The poles shift from the cube orientation at ARB 0c to the typical rolling texture appeared at ARB 6c - ARB 10c (see Fig. 5 (d)–(f)) through the transit stage at ARB 2c and ARB 4c (see Fig. 5 (b) and (c)). As shown in Table 1, the corresponding effective rolling reduction is between about 98.4% and about 99.9% at the transit stage such as ARB 6c, ARB 8c, and ARB 10c, where after the typical rolling texture appears.
However, the $\{ 1 \ 1 \ 1\}$ PF of ARB 12c and ARB 15c (see Fig. 5 (g) and (h)) are obviously different from the typical rolling texture such as ARB 6c – ARB 10c (see Fig. 5 (d)–(f)). First, two poles around each RD appeared between ARB 6c and ARB 10c seem to shift, and then, combined to one strong peak at each RD at ARB 12c and ARB 15c. Secondly, poles appeared at around TD between ARB 4c and ARB 8c of which intensity is relatively weak, and seem to disappear between ARB 12c and ARB 15c. Furthermore, four strong poles seem to newly appear at around copper orientation.
Although the change of textures during the ARB process is intuitive on PFs as shown in Fig. 5, PFs are not suitable for the precise analysis of texture since the PFs are the projection of the three dimensional information toward the two dimensional plane. Thus, the Euler angle space $( \varphi_1, \varPhi, \varphi_2)$ represented as ODF maps has been used for the precise analysis on texture4–6). An ODF map is the set of the sliced planes (In general, every 5° along $\varphi_2$ axis) in three dimensional Euler angle space. The ODF map of ARB 0c, ARB 8c, and ARB 15c are shown in Fig. 6 (a), (b), and (c), respectively. In an ODF map, the ODF intensity is used to express the relative frequency of the orientation compared with the random orientation which is defined as unity.
The orientation distribution function maps commercial purity aluminium processed by accumulative roll bonding with lubrication for (a) 0 cycle, (b) 8 cycle, (c) 15 cycle, and (d) ideal positions maps of typical fcc rolling texture components.
It is well known that the typical rolling texture appears in Euler angle space as a high intensity region like a fibre shape, so-called β-fibre4), which connects the typical rolling texture components, brass $(35^{\circ}, 45^{\circ}, 0^{\circ})$ and copper $(90^{\circ}, 35^{\circ}, 45^{\circ})$ through S $(59^{\circ}, 37^{\circ}, 63^{\circ})$. Thus, the analysis of the rolling texture in Euler angle space has been performed along the β-fibre4–6), and the actual procedure is as follows. First, the $\varphi_2$ sections ($\varphi_1 - \varPhi $ planes) with every five degrees between 45° and 90° are chosen, and then, the highest intensity position of β-fibre on each $\varphi_2$ section is searched. Consequently, the highest intensity and its location (Φ and ϕ1) as a function of $\varphi_2$ sections are plotted as shown in Fig. 7.
Texture analysis along $\beta $ -fibre on ϕ2 sections for every 5° in Euler space for accumulative roll bonding processed commercial purity aluminium between 2 cycle and 15 cycle; (a) intensity at the peak on a ϕ2 section, and (b) Φ and (c) ϕ1 of the peak position on the ϕ2 section.
In Fig. 7 (a), the intensity of β -fibre less than $\varphi_2$ = 70° continuously increases with increasing n up to eight. Furthermore, the $\varphi_2$ section of 45° seems to have the largest intensity along β-fibre more than n = 8. Whereas, the intensity of the β-fibre more than $\varphi_2$ = 70° initially increases with n up to 10, but, it starts decreasing after n = 10. Here, it should be noted that the typical rolling texture component, copper and brass are on the section at $\varphi_2$ of 45° and 90°, respectively. Another component S exists at around a section of $\varphi_2$ = 65°.
The trend of the intensity of β-fibre of ARB processed 2N-Al with lubrication seems to be different from that of typical rolling texture of Cu up to 99%4) and ARB processed multi-layered 4N-Al17). Generally speaking, the intensity of β-fibre continuously increases with increasing the rolling reduction up to about 97% in the case of pure Cu, and the rolling reduction of 97% is equivalent to ARB 5c as shown in Table 1. When the rolling reduction reaches to 99% in the case of pure Cu, the intensity of β-fibre with $\varphi_2$ section of 45° and 50° starts decreasing, whereas, that more than 50° still increases, and, the highest intensity along β-fibre locates at the section of 65° and 70° which corresponds to S. It should be noted that the thickness reduction of 99% is equivalent to ARB 9c as shown in Table 1.
The trend of the intensity for β-fibre of ARB processed 2N-Al with lubrication seems to be also different from that of ARB processed Al without lubrication13,14). The highest intensity locates close to $\varphi_2$ section of 65° in the case of ARB with lubrication, whereas, the strong S orientation can be associated with ARB without lubrication since the friction between the surfaces of the metal sheet and the rolls gives additional shear strain close to the surface of the sheets at each roll bonding during the ARB process. It should be noted that the texture of ARB processed Al-Sc alloy with lubrication seems to be similar to that of 2N-Al when the $\{ 1 \ 1 \ 1 \}$ pole figures are compared16). However, it is unknown where the highest intensity along β-fibre is, since such analysis was not performed in the report. Nevertheless, the highest intensity locates at $\varphi_2$ = 45° section where copper orientation exists in the case of the ARB processed 2N-Al with lubrication. From the comparison of the case of ARB processed multi-layered 4N-Al, similar trend is observed, but, development of β-fiber texture of 2N-Al seems to be slower than that of ARB processed multi-layered 4N-Al17).
Figure 8 shows the enlarged $\varphi_2$ = 45° section of the ODF map for ARB 2c – ARB 15c. Only cube component can be seen at ARB 2c, and, the peaks start to shift toward copper and Taylor $(90^{\circ}, 27^{\circ}, 45^{\circ} )$ orientations. The shift of the peak toward copper and Taylor orientation progresses at ARB 4c, and, a component close to copper and Taylor becomes stronger and stronger after n = 4. It should be noted that the distance between these two orientations in Euler angle space are 8 degrees, and therefore, the precise location of the peak should be discussed carefully.
ϕ2 = 45° section for (a) accumulative roll bonding (ARB) 2 cycle (c), (b) ARB 4c, (c) ARB 6c, (d) ARB 8c, (e) ARB 10c, (f) ARB 12c, (g) ARB 15c, and (h) ideal location of cube, Taylor and copper orientations.
Figure 9 shows the position of the β-fibre on the section of $\varphi_2$ = 45° between ARB 2c – ARB 15c constructed from Fig. 7 (b) and (c). Although the location of the peak for ARB 2c and ARB 4c seem to be coincided with copper, that after ARB 6c does not coincide with copper. It is noteworthy that the peak position at ARB 12c and ARB 15c are almost at the ideal Taylor orientation. Consequently, the highest intensity component of ARB processed 2N-Al with lubrication seems to be rather Taylor (or Dillamore) orientation after ARB 6c.
Shift of position of peak on ϕ2 = 45° section in Euler space with increasing ARB cycles. Ideal location of Taylor and copper orientation is also displayed.
Taylor orientation is predicted by Dillamore assuming the plane-strain compression with Taylor model for the texture development during rolling deformation7). This texture should appear in the FCC metals with high stacking fault energy (SFE) such as Al, if the deformation can be considered as Taylor model. However, Taylor orientation has not been the highest intensity component in rolled fcc metals when the recrystallized pure metals having coarse grains are used as the starting materials. Only simulation or special initial microstructure can create this texture, and it is written even in a book, such as “ ‘Taylor’ texture component, which is known to differ from the experimental rolling texture.”8). However, Taylor orientation was the dominant orientation of ARB processed multi-layered 4N-Al17).
It should be noted that Sun et al. insisted that the movement of Y-type junction23) is important for the formation of Taylor orientation in addition to the above three assumptions17). The present study (2N-Al) and the ARB processed multi-layered 4N-Al show the uniform microstructure including texture along thickness direction. It is associated with the low friction between the surfaces of the rolls and the sheet due to the lubrication in the case of the present study9,24), whereas, multi-layered ARB allows plane strain compression even the ARB without lubrication is used17). Jamaati insists that Al shows continuous recrystallization due the its high stacking fault energy14), which is associated to the Y-triple junction.
Such Taylor orientation becomes the highest intensity component after ARB 6c with lubrication as shown in Fig. 9. It is interesting that such region for the texture coincides with the saturation region in terms of $d_{\rm ND}^{\rm HAGB}$ as shown in Fig. 4. It is thought that the saturation of the reduction in grain size is strongly related to the movement of triple junction17,23). In other words, continuous recrystallization is thought to be requested for the formation of Taylor orientation. Nevertheless, it is possible to say that the crystal rotation occurs toward only one direction for most of the grains during such saturation region. Thus, the major component becomes Taylor orientation from copper orientation after ARB 6c with lubrication. Although it is just a rephrase of grain subdivision, it can be also said that the crystal rotation toward different orientations occur inside most of the grains before the saturation region.
If a grain rotates toward only one orientation, the combination of the activated slip systems should be almost the same in the grain. Thus, it can be said that the second and third conditions of Taylor model mentioned above can be satisfied, and therefore, the deformation of the UFG may be similar to the Taylor model. It should be pointed out that the first condition cannot be perfectly satisfied since saturation in $d_{\rm ND}^{\rm HAGB}$ occurs. We could say that the highest intensity component can be Taylor orientation only for such a case, and, the purity of Al down to 99% does not affect the formation of Taylor orientation.
This hypothesis agrees with the fact that Taylor model cannot predict the texture evolution during rolling deformation up to about 95% where grain subdivision occurs5). If grain subdivision occurs inside a grain, combination of the slip systems cannot be unity inside the grain anymore, and therefore, deformation is not similar to Taylor model anymore. Thus, the actual texture evolution up to 99% of the rolling reduction coincided with the relaxed constrained model.
The texture evolution of ARB processed commercial purity aluminium with lubrication was measured and the highest intensity texture component was Taylor orientation after ARB 6c. The ARB cycle number where Taylor orientation appears coincided with the start of the saturation region for the reduction of the grain size, where the continuous recrystallization is thought to occur. It was revealed that Al having the purity of 99% also shows Taylor orientation. These results suggest that the deformation of UFG is similar to Taylor model.
This study was financially supported by the Grant-in-Aid for Scientific Research on Innovative Area, “Bulk Nanostructured Metals (area No. 2201) and Grant-in-Aid for Young Scientists (B) (No. 22760556), and the supports are gratefully appreciated.
