Electron Backscatter Diffraction Study on Microstructure, Texture, and Strain Evolution in Armco Iron Severely Deformed by the Differential Speed Rolling Method

Commercially pure Fe Armco samples in the form of rectangular plates with dimensions of 25 × 10 × 150 mm (width × thickness × length) were subjected to a preliminary thermomechanical treatment, which included cold rolling to 80 pct of a thickness reduction followed by 1-hour recrystallization annealing at 700 °C. This treatment allows obtainment of fully recrystallized material with a fine-grained structure. Such prepared samples were subsequently processed by main cold rolling to 85 pct of cold work. The final thickness reduction of 85 pct was obtained in two consecutive rolling passes. In the first pass, the reduction was 50 pct (from 2- to 1-mm thickness), and in the second pass, it was 70 pct (from 1 to 0.3 mm). The samples were 180 deg rotated along their normal axis (“back and forth”) between rolling passes. It was previously shown[17],[18] that this kind of alteration of deformation path (due to a change of the shear strain direction between consecutive processing operations) is beneficial in terms of enhancing the grain refinement effect and, thus, improving the mechanical properties of processed material.

The deformation process was carried out on a sexton-type rolling mill equipped with working rolls made of tungsten carbide, with an equal diameter of 85 mm. In the present work, the cold rolling process was conducted in the following variants involving different values of the roll speed differentiation parameter R:

The EBSD system coupled with an FEI Quanta 3D field emission gun–scanning electron microscope was applied in order to analyze the microstructure, the microtexture, and the lattice strain evolution in the cold-deformed Fe samples. Before the examination, all samples were subjected to a careful metallographic preparation routine (grinding with 120- to 4000-SiC papers, polishing with 3- to 0.25-µm diamond suspensions, and final polishing with 0.1- to 0.06-µm silica suspensions) in order to minimize the negative effects of the surface condition on acquired diffraction data. The EBSD investigation was carried out on the middle part of the longitudinal sections of rolled samples (in the rolling direction–normal direction (RD–ND) plane). For each sample, data were acquired from an area that was approximately 150 × 150 µm in size for a step size of 0.6 µm (for a textural analysis) and from an area of 15 × 15 µm in size for a step size of 0.06 µm (for a more detailed microstructural analysis). Several scans were performed for each sample, and then representative examples were chosen for publication. Acquired data were analyzed using TSL OIM Analysis 5 commercial software. Evaluation of a grain boundary distribution was made with an assumption of 2 to 15 deg misorientation angle for low-angle grain boundaries (LAGBs) and above 15 deg for HAGBs. Due to the presence of an “orientation noise” in heavily deformed samples, a software cleaning routine of as-received diffraction data was applied; points with neighbor-to-neighbor confidence index values lower than 0.2 were excluded from the analysis.[19] A quantitative microstructural analysis was conducted according to definitions proposed in the work of Jazaeri and Humphreys,[20] where a subgrain size was defined as an average LAGB spacing measured both in the normal and rolling directions and lamellar boundary (LB) spacing as HAGB intercepts measured in the ND.

The EBSD method was additionally applied to analyze lattice strain introduced into the material during plastic deformation by analyzed rolling techniques. Principles of the EBSD lattice strain assessment have been described elsewhere (e.g., in our previous works[21],[22]). In the present work, diffraction pattern quality deterioration and local misorientation analysis approaches[23],[24] were chosen.

EBSD microtextural analysis was conducted by orientation distribution function (ODF) calculations using the Harmonic Series Expansion method, assuming both cubic crystal symmetry and an orthonormal (for the initial material) or a triclinic (for all rolled specimens) symmetry of the sample. The analysis was conducted in an extended Euler angle space (φ ₁ = 0 ÷ 360 deg, Φ = 0 ÷ 90 deg, and φ ₂ = 0 ÷ 90 deg).

In order to analyze the effect of applied cold rolling variants on the change of mechanical properties of the investigated material, Vickers microhardness measurements were conducted on longitudinal sections of samples using a 100-g load and a 15-seconds loading time for a single measurement (15 measurements were performed for each sample, which were used to calculate average values and standard deviations).

The Armco iron in its initial state (after preliminary thermomechanical treatment) has a fully recrystallized microstructure with an average grain size of ~6 µm (Figure 1). Cold rolling by all considered variants leads to an intensive microstructural evolution. Figures 2 and 3 show EBSD inverse pole figure maps of Fe samples in their initial state and after cold rolling with different R parameter values. Typical for cold-rolled metals, lamellar structures consisting of elongated boundaries and short interconnecting boundaries are observed (Figure 2). Moreover, extremely fine subgrains and grains are found for samples deformed with a high ratio of roll speed differentiation (i.e., R = 3 and R = 4; Figures 3(d) and (e), respectively). Since the size of these grains is much lower (below 1 µm) than that of the initial grains, it is reasonable to assume that some processes of strain-assisted grain refinement take place upon these rolling variants.

It was shown by Hughes and Hansen[25] that a strain-induced structure evolution upon cold rolling involves subdivision of initial grains on “microvolumes” separated by LAGBs or HAGBs. With the rise in deformation degree, average spacing (a distance) between adjacent boundaries of both types decreases with a simultaneously observed increase of the misorientation angle. Results of a transmission electron microscopy evaluation show that the deformed, subdivided structure of metals is composed of small dislocation-free regions (cells) arranged in blocks and surrounded by dislocation walls. Upon increasing the imposed strain, the structure undergoes further transformation, which includes internal rotations in deformation bands leading to the formation of new grains. On the other hand, an alternative mechanism of structure refinement during DSR processing of dual-phase low carbon steel was proposed in the recent work of Hamad et al.[26] The authors concluded that the formation of fine grains located along deformation band boundaries (a so-called “necklace” structure) is an effect of a continuous dynamic recrystallization process. This process is based on the strain-activated coalescence of subgrains inside deformed elongated grains, which leads to a transfiguration of LAGBs into HAGBs. The fraction of grains with size below 1 µm was also found to increase with the increasing number of DSR operations.

The EBSD images presented in Figure 3 confirm a decrease of boundary spacing with the rising value of the R parameter; thus, it may be concluded that, for all DSR variants of rolling (R = 2 ÷ 4), a higher strain was introduced into the material than in the case of rolling with equal speed of both rolls (R = 1).

The results of the quantitative microstructural analysis are presented in Figure 4. It is found (Figure 4(a)) that the average distance between adjacent lamellar boundaries is nearly two times (for the R = 2) and 3 times (for the R = 3 and the R = 4) smaller than for the sample rolled with an equal speed of both rolls. Values of the LB spacing are far below 1 μm for all DSR-processed samples. A similar trend was also observed for the subgrain size calculated as an average LAGB spacing. The results of grain size diameter distribution presented in Figure 4(b) confirm a structural refinement during the DSR processing of iron; the fraction of grains with the submicron size increases upon raising the R value, reaching ~70 pct for the highest considered value of the R parameter. These results are similar to those presented by Hamad et al.[14] However, since some fraction of these grains (~40 pct) is also found in the sample cold deformed with an equal speed of both rolls, the obtained results imply that the differentiation of the roll speed results with a triggering of some additional phenomenon. According to the results shown by Megantoro et al.,[27] an application of DSR processing leads to a prominent temperature rise in an IF steel sample. It was shown that the temperature rise increases with an increase in either the thickness reduction or the roll speed ratio, approaching a maximum value of 373 K (100 °C) under the thickness reduction of 40 pct and a roll speed ratio of 1:4. Since, in the present study, the reduction per pass was even higher (50 and 70 pct in the first and second rolling passes, respectively), it may be assumed that the generated heat was also higher. However, in the present case of cold rolling of Armco iron, the temperature rise of 100 °C (or even higher) is too low to activate a dynamic recrystallization process. Nevertheless, the results of microstructural examination revealed that an application of the high roll speed mismatch (the R = 4) leads to both a decrease of boundary spacing and a formation of a well-developed subgrain structure inside the deformation bands. These findings are similar to those presented by Valiev et al.,[28] who concluded that high plastic straining leads to an increase in the number of dislocations that are known to concentrate in cell walls. The decrease of dislocation wall thickness, and thus the increase of cell size, is usually attributed to the recovery process, which permits segment rearrangements. Thus, it seems to be more reasonable to call the observed structure evolution a subgrain structure evolution through accumulated shear deformation, which may be related to the dynamic recovery process. Since the observed structural changes take place inside deformation bands, and do not lead directly to the formation of a fully equiaxed grain structure, it may be stated that a high fraction of submicron grains in the material subjected to the high ratio DSR method is an effect of both the strain-induced grain subdivision and the temperature rise, which supports activation of the continuous dynamic recovery and shear-assisted substructure rearrangement.

Cold rolling of bcc metals and alloys results in a formation of deformation texture that is composed of few orientation fibers, among which the most important are the α-fiber and the γ-fiber, both located in the φ ₂ = 45 deg section of the Euler angle space (Figure 5). The former fiber contains orientations that have 〈110〉 directions parallel to the RD (and is also called an RD fiber). The latter one connects orientations with {111} planes parallel to the rolling plane (and for a cubic crystal may be described also as an ND fiber).[29]

Figure 6 shows φ ₂ = 45 deg sections of the Euler space obtained for the material in its initial state and after cold rolling to 85 pct thickness reduction with different values of the R parameter. It is found (Figure 6(a)) that material in its initial state exhibits a weak texture with the {100}〈100〉 component and some residual (after the previous thermomechanical treatment) trace of the γ-fiber. Cold rolling to 85 pct thickness reduction by all proposed variants leads to a formation of rolling texture that is composed of the aforementioned fibers.

The material deformed without roll speed differentiation (R = 1, Figure 6(b)) shows a typical bcc metal rolling texture characterized by a well-developed γ-fiber (with a special emphasis on a {111}〈112〉 orientation) and an α-fiber dominated by a {100}〈110〉 component. By analyzing the results presented in Figures 6(c) through (e), it may be concluded that the DSR processing of iron results in an overall texture weakening effect and a displacement of both the α-fiber and γ-fiber. The effect of the aforementioned fiber deterioration due to an introduction of shear deformation into the rolling process was also noted in the recent work by Tóth et al.[30] The authors, using the viscoplastic slip and the Taylor model-based calculations, found that both the RD and the ND fibers are strongly affected by deformation mode and are completely unstable for simple shear. In this attempt, the lack of texture component (or a fiber) stabilization means that, under an imposed deformation gradient, its rotation rate is nonzero and may move to some different, more stable orientation. It was previously shown[31] that the deformation gradient in the DSR method is approximated by a superposition of a plane strain (which is specific for the normal rolling process) and a simple shear in the rolling direction (Eq. [1]). Additionally, the impact of the simple shear component (ε _xz ) rises with an increase in the R parameter, leading to an accumulation of a higher strain.

This effect of a progressive shifting of the main orientation fibers from their nominal positions with the increasing R value is confirmed by fiber plots calculated for their “ideal course” (Figure 7). It may be found that, due to the displacement of fibers in the Euler space, calculated ODF values in the exact positions of each fiber are lowered with the rise in R parameter values. This fact is prominently visible in the run of the γ-fiber (Figure 7(b)) for the R = 4 sample, which exhibits only marginal ODF values that are even lower than those calculated for the material before rolling. However, samples cold rolled with high roll speed mismatch (R = 3 and R = 4), despite overall lower values of calculated ODFs, exhibit strong peaks near {111}〈1-21〉 and {111}〈-211〉 locations, respectively, indicating that these positions are preferably stable under imposed shear deformation. The obtained results showing texture fibers “blurring” are also in good agreement with those presented by Hamad et al.,[14] where a crystallographic texture of the DSR-processed IF steel was characterized as typical for cold-rolled bcc metal orientations without traces of any additional shear components. Thus, it is reasonable to state that the DSR processing of iron leads to a shifting of normal rolling texture components, rather than to a formation of completely new ones.

An evaluation of strain introduced during cold rolling was carried out by means of microhardness measurements and EBSD lattice strain analysis. The results of microhardness measurements are shown in Figure 8(a). It is found that cold rolling to 85 pct thickness reduction by each considered deformation method leads to a significant strain hardening effect. However, the DSR processed samples exhibited a distinctly higher hardness than the normally rolled sample (R = 1). Moreover, a decrease of calculated standard deviation values upon raising the roll speed mismatch confirms a more homogeneous course of deformation with the presence of additional shear strain. This finding is in line with the results of previously reported works on various DSR-processed materials (e.g., Al[32] and Cu[11]) and may be related to a decrease in surface frictional stresses (which are cited as the main reason for the nonuniformity of deformation upon rolling[33]) with a rise in roll speed mismatch.

The microhardness measurements were also supplemented by the EBSD strain assessment conducted by both image quality (IQ) analysis and local misorientation analysis represented by a grain orientation spread (GOS) parameter. One of the most important conditions of receiving valuable results from the IQ analysis is a careful and repeatable surface preparation of analyzed samples. Thus, we have paid special attention to the sample preparation process in order to exclude its possible effect on the obtained results. To the best of our knowledge, the results presented in this article are free from disruptions from surface defects. In this analysis, the quality of the obtained diffraction pattern is assumed to be significantly affected by the crystal lattice quality (e.g., the density of lattice defects). It was shown that the lattice strain introduced during plastic deformation leads to a higher fuzziness of the Kikuchi diffraction patterns (and, thus, shifting the IQ histogram to lower values). The basis of the method is a comparative analysis of the IQ histograms of two samples: one sample is fully recrystallized, and the other sample is deformed to some strain value (the principle of the method is graphically presented in Figure 8(b)). This comparative study allows for an extraction of these points, which belong to deformed or undeformed volumes of the material, respectively. To associate the experimental IQ values with the introduced strain, the normalized IQ histograms are separated into deformed (region A in Figure 8(b)) and undeformed (region B in Figure 8(b)) material volumes. However, some points included in the histograms (region C in Figure 8(b)), which are common to both distributions, could not be unequivocally assigned to either the deformed or undeformed material. Thus, a minimum deformed fraction X _min and not a total deformed fraction of the analyzed sample could be estimated. The experimental reduced IQ histograms that were obtained for iron samples after different variants of deformation (Figure 8(c)) were used to calculate the minimum deformed fraction X _min (Figure 8(d)).

The second approach to the EBSD strain assessment involves an analysis of local misorienation changes. The principles of this method are based on an assumption of strain-assisted internal rotations of microvolumes (cells or subgrains) within grains, which means that a transmission and accommodation of a plastic strain result in an increase of internal GOS. Therefore, grains that undergo deformation (and, thus, are extensively defected) exhibit high GOS value, and reversely, those that are undeformed (defect free) have very low GOS values, close to zero.[34]

Interestingly, despite the observed higher hardness of DSR specimens, calculated changes of the X _min plotted as a function of the R parameter exhibited an inverse, negative, and almost linear trend (Figure 8(d)). Since this parameter is closely related to lattice defect density, it may be stated that, during rolling with an additional presence of shear strain (R = 2, 3, or 4), some part of the stored energy of deformation is released due to the activation of dynamic structure restoration processes (e.g., a recovery). These results seem to be confirmed by the EBSD maps presented in Figure 4, which show the formation of cell-like substructural features for the sample rolled with the highest considered value of the roll speed ratio (R = 4, Figure 3(e)).

Results of the GOS-based analysis are shown in Figure 9. In our previous research, devoted to the EBSD study on recrystallization of cold-rolled Ni₃Al-based alloy,[22] we showed that the GOS-based analysis allows distinguishing grains that belong to the deformed matrix and those that undergo static recovery or recrystallization. GOS values below 1 deg and in the range of 1 to 3 deg were designated as newly formed (recrystallized) and recovered grains, respectively. In the present case, it was found that, since Armco iron in the initial condition, in its fully recrystallized state (Fig. 9a), is characterized by extremely low GOS values (mostly below 1 deg), cold deformation leads to a substantial increase and scattering of the orientation spread. Nevertheless, maps presented in Figure 9 reveal that some fine grains observed in the microstructure of cold-deformed samples also exhibit GOS values below 1 deg and up to 3 deg, which may be related to defect-free volumes, as previously stated. Moreover, a fraction of these grains as well as a fraction of HAGBs increases with the rise in roll speed mismatch (Table I). These results also support the possible activation of structure restoration phenomena upon DSR processing.