Microstructure and Elongation Anisotropy of Cold Rolled and Solution Treated A356 Alloy Strips Fabricated via High-Speed Twin-Roll Casting
2018 Volume 59 Issue 11 Pages 1777-1783
Details
2018 Volume 59 Issue 11 Pages 1777-1783
A356 alloy strips fabricated via high-speed twin-roll casting were cold rolled at the reductions of 0%, 12%, 30%, 50% and 73% and then solution treated at 793 K for 1 h. Microstructure observations and tensile tests were performed for the processed strips. Upon increasing the reduction from 0% to 50%, an improvement in elongation with significant anisotropy was observed; the elongation along the transverse direction was inferior to that in the rolling direction. However, on further reduction up to 73%, this anisotropy was eliminated and an elongation above 20% was achieved. This behavior is caused by the characteristic changes occurring in the second-phase particles that are located in the mid-thickness region of the strip. To achieve a high elongation without anisotropy, a process for refining the coarse particles in the mid-thickness region of the twin-roll-cast strips and homogeneously dispersing them into the matrix needs to be developed.
The use of aluminum alloys in automotive body panels has increased because of the demand for lighter vehicles. The production costs of wrought aluminum alloys should be reduced to accelerate the replacement of steel by aluminum alloys. Vertical-type high-speed twin-roll casting shows promise as an effective cost-reduction measure.1–3) Since this process produces thin cast strips directly from the melt, homogenization heat treatment and hot rolling, which are necessary in conventional DC processes, can be skipped. Furthermore, as its very high cooling rate can reduce the harmful effects of crystalline compounds,4,5) it is also expected to apply recycled materials or cast alloys to wrought materials using this method.
When fabricating wrought materials from twin-roll-cast strips, additional minimum cold rolling and heat treatments are inevitable to control the thickness and internal quality of the products. In the case of heat treatable aluminum body panels, which are bake hardened after press forming, the solution treatments subsequent to the cold rolling are required. However, the correlation between the microstructural change and the evolution of mechanical properties of the strips throughout the entire process has not been sufficiently studied yet. Moreover, reports on rolling process for cast aluminum alloys are scarce.7,8)
Herein, an A356 cast alloy is applied to a high-speed twin-roll casting and subsequent thermo-mechanical processes. The high-speed twin-roll-cast A356 alloy is cold rolled at various rolling reductions. After the solution treatment, by comparing those materials, the influence of the reduction on the microstructure and mechanical properties, particularly on the elongation which is most crucial for press formability, is studied.
The material used in this study was an A356 alloy which composition is listed in 1. The as-received material was melted in a ceramic crucible at 1023 K. Then, the melt was transferred to a ladle and degassed for 10 min by Ar gas bubbling. After degassing, the molten alloy (898 K) was fed into a vertical-type twin-roll caster comprising a pair of copper rolls with a diameter of 600 mm and a width of 600 mm. These rolls had no internal cooling circuit. The casting speed, initial roll gap and roll-separating force were fixed to 60 m/min, 1 mm, and 10 kN, respectively. The cast strip obtained was 6000 mm-long, 600 mm-wide and 2.6 mm-thick. After cooling the cast strips to room temperature, small samples with the length of 150 mm and width of 100 mm were collected from them. These samples were cold rolled along the casting direction using a small rolling mill (the roll size was 104 mm in diameter and 160 mm in width). The reduction at each pass was set at approximately 10%, and total rolling reductions (r) of 0%, 12%, 30%, 50%, and 73% were achieved by changing the number of the passes. The rolled strips were solution treated at 793 K for 1 h and were immediately quenched in water.
After the solution treatment, the microstructures in the longitudinal (RD) and transverse sections (TD) of the rolled strips were observed without etching using an optical microscope. An image analysis was performed for both sections using an image processor ImageJ to study the character of second-phase particles.9) The particles were fitted to an ellipse to evaluate their length, aspect ratio, orientation, and distribution. A threshold of 0.5 µm was set as the minimum particle size to avoid capturing background noise. The orientation was determined by the angle between the major axis of the particle and the planer direction of the strip. The particle distribution was evaluated using Voronoi diagram composed of perpendicular bisectors of all particles and measuring the cell size and variation.
In addition, scanning electron microscope (SEM) was used for observation and a detailed investigation of the morphological changes occurring in the second-phase particles. For the three dimensional observations, the samples were etched for 15 min using 5% of NaOH aqueous solution.
2.3 Tensile testsThe relationship between rolling reduction and mechanical properties was evaluated by means of tensile tests. Six dog-bone shaped tensile test specimens were machined from each as rolled strip at 0°(RD) and 90°(TD) relative to the rolling direction. Test piece size was adjusted in proportion to the thickness of the strips. The machined specimens were solution treated at 793 K for 1 h, followed by water quenching. To avoid the instability caused by the natural aging, specimens were left at room temperature for more than 24 hours from solution treatment. After that, the tensile tests were conducted at room temperature and at a fixed strain rate of 0.1 s−1 for all samples.
2.4 FractographyTo study the influence of rolling reduction on the fracture mode, fracture surface observations were performed on the tensile test specimens using an SEM. Both RD and TD fracture specimens were observed at an acceleration voltage of 15 kV.
The microstructures observed in the TD section are shown in Fig. 1. Without cold rolling, the microstructure strongly depended on the distance from the surface. Very fine second phase particles were homogeneously distributed in the surface region. In contrast, coarse plate-like particles larger than 20 µm in size were observed in the inner region, which is located at about a quarter thickness from the surface. In the center of the strip, a mixture of small and coarse particles was inhomogeneously distributed. Although most of these second phase particles are considered to be eutectic Si, light grey crystalline compounds are also observed in the inner region. This microstructural variation in the thickness direction can be attributed to the solidification manner of vertical-type high-speed twin-roll castings.5,6) Since solidified shells grow from the roll surface toward the center of the strip accompanied with the cooling speed gradient, fine particles crystallize in the surface and coarse particles appear in the growth front, i.e., the quarter thickness region. In contrast, in the center of the strip, the mushy layer consisting of floating crystals, broken dendrite branches and enriched liquid is solidified by sandwiched between two shells.
Microstructure of the transverse section of solution treated strips rolled at different rolling reduction.
With regard to evolution associated with increasing reduction, while there was little change in particle morphology and distribution in the surface region, those characteristics changed in the internal region as a function of the reduction. With increasing reduction, the size of the coarse plate-like particles present in the quarter-thickness region decreased. At the same time, the variations in particle size or shape in the center region also decreased gradually. When the reduction was increased to 73%, the particle morphology and distribution in the internal region became almost the same as those in the surface region.
3.1.2 Morphological change of second-phase particlesFigure 2 shows the SEM images of second-phase particles in deep-etched as cast and as rolled samples. Three positions (namely, surface, quarter-thickness and center) were observed in the TD section as in the case of the optical microscope observation (Fig. 1). In spite of the rolling reduction, the crystallized fine particles in the surface region remained nearly unchanged and were homogeneously distributed. In the quarter thickness region, coarse crystalline plate-like particles cracked because of cold rolling at low reductions. These cracks followed a direction perpendicular to the rolling direction. Increasing the reduction facilitated particle cracking; at a reduction of 73%, small particles with diameters between 2–3 µm were arranged along the rolling direction. In the center region, coarse coralloidal particles were observed in the as cast state. Upon cold rolling, these particles were crushed and elongated along the rolling direction with cracks. When the reduction reached 73%, the coralloidal shape disappeared, and small cracked particles with diameters of about 5 µm were linearly arranged along the rolling direction.
SEM images of second-phase particles in as cast and as rolled strips: (a) As cast, (b) r = 30%, (c) r = 73%. The upper section shows the center of the thickness, the middle shows the quarter-thickness, and the lower shows the surface region.
Quantified particle size, shape, and distribution are shown in Fig. 3, 4, and 5, respectively. The median values are plotted in the diagram along with the D15 and D85 values of the measurements given by the scatter bars. The particles in the surface region exhibited a fine size, spherical shape, and homogeneous distribution; these particle characteristics depend on neither the reduction nor observation direction (Fig. 3(a), 4(a), 5(a)). In contrast, the particle characteristics in the quarter-thickness and center regions strongly depend on the rolling reduction. Furthermore, the changes also depend on the observed direction. When the particles in the quarter-thickness region were observed from RD, their size, aspect ratio, and the dispersion of the Voronoi cell area monotonically decreased with an increase in the rolling reduction. In contrast, when these parameters were measured from TD, they remained nearly unchanged, even if the reduction increased from 12% to 50% (Fig. 3(b), 4(b), and 5(b)). Although differences based on the observation direction were also confirmed for the center region, they were more pronounced in the quarter-thickness region. Interestingly, these differences were almost eliminated by increasing the reduction up to 73%. In addition, the size, aspect ratio, and the Voronoi cell area of the 73% rolled strip became nearly the same as those of the particles in the surface region.
Comparison of the sizes of second-phase particles in the RD and TD sections at different rolling reductions: (a) Surface, (b) Quarter-thickness, (c) Center of the thickness.
Comparison of the aspect ratios of second-phase particles in the RD and TD sections at different rolling reductions: (a) Surface, (b) Quarter-thickness, (c) Center of the thickness.
Comparison of the Voronoi cell area of second-phase particles in the RD and TD sections at different rolling reductions: (a) Surface, (b) Quarter-thickness, (c) Center of the thickness.
Figure 6 shows the changes in orientation distribution for particles in the quarter-thickness region. The particles in the 0% rolled strip had random orientations regardless of the observed section (Fig. 6(a)). Upon increasing the reduction to more than 30%, a particle orientation toward the rolling direction was clearly recognized in the TD section.
Orientation of second-phase particles in the quarter-thickness region of the RD and TD sections at different rolling reductions: (a) r = 0%, (b) r = 12%, (c) r = 30%, (d) r = 50%, and (e) r = 73%.
The relation between the rolling reduction and tensile properties is given in Fig. 7. The plotted values for each rolling reduction are the average of six test results. Although the proof stress and tensile strength did not depend on the rolling reduction, the elongation greatly improved with increasing reduction. In terms of the difference based on the tensile directions, the anisotropy of tensile strength and elongation was confirmed at the reductions ranging from 12% to 50%. In particular, the elongation exhibited significant anisotropy, which was 5%–10% lower along the transverse direction (TD) compared to the rolling direction (RD). However, when the rolling reduction was increased to 73%, the elongation along TD was improved to the same level of that along RD and the anisotropy was almost eliminated.
Effect of rolling reduction on tensile properties along RD and TD.
Figure 8 shows the comparison of representative engineering stress-strain curves obtained from strips rolled at different reductions. Regardless of the reduction and tensile direction, all stress-strain curves followed a similar trend. However, there was a clear difference in the fracture points. At the reductions of 0% and 12%, both specimens strained along RD and TD fractured without necking around the 10% strain. When the reduction increased to either 30% or 50% and the elongation along RD improved to show necking, the specimens strained along TD fractured rapidly as soon as the stress reached its ultimate value. At a reduction of 73%, not only did the specimens strained along RD but also those strained along TD fractured with necking and both achieved a high elongation above 20%. According to these results, it can be assumed that the difference in fracture time associated with the rolling reduction and tensile direction leads to evolution of elongation accompanied by a significant anisotropy.
Representative stress-strain curves of strips rolled at different rolling reductions along: (a) RD and (b) TD.
The SEM images of the fractured surfaces of 12%, 50%, and 73% rolled strips are shown in Fig. 9 to compare the surface characteristics for different tensile directions. The 12% rolled strips had a rough fracture surface with many defects (Fig. 9(a)) regardless of the tensile direction. Since dendrites were observed in the magnified image of the defects (Fig. 9(d) and 9(e)), these defects were considered to be the shrinkage cavity formed during the solidification. With regard to the differences based on the tensile direction, many elongated cavities or cracks were observed along the RD only in the TD fracture surface. Upon increasing the reduction to 50%, the roughness of the fracture surface decreased not only for RD but also for TD. However, in the center region of the TD fracture surface, some cracks were grew along the RD. Detailed observation confirmed the presence of dendrites at the edges of the crack (Fig. 9(f)), which suggests that the shrinkage cavities cannot be completely bonded even after 50% cold rolling. No cracks or dendrites were observed on the fracture surface of the 73% rolled strips (Fig. 9(c)).
SEM images of tensile samples obtained from: (a) r = 12, (b) r = 50%, (c) r = 73%. (d), (e), and (f) are the shrinkage defects in (a) and (b) respectively.
As shown in Fig. 7, we found that the elongation of the A356 high-speed twin-roll-cast strips was greatly improved with significant anisotropy via thermo-mechanical processing. First, we consider the reasons for the dependency of improved elongation with increasing rolling reduction. The first reason is a morphological change in the second-phase particles. In previous studies, it has been reported that the elongation of the A356 alloy increases with decrease in the particle size and aspect ratio.10–12) Since the high-speed twin-roll-cast alloy contains coarse particles in the inner region (Fig. 2), these particles may limit the elongation of the strip. Therefore, the effect of cold rolling, which refines and spheroidizes the coarse particles (Fig. 1–4), leads to an improvement of elongation that is proportional to the rolling reduction. The second reason is a decrease in volume and the number of shrinkage cavities. Upon increasing the reduction, the number of shrinkage cavities on the fracture surface decreases, as shown in Fig. 9. This trend is also expected to have an influence on the high elongation at high rolling reductions.
We further discuss the reason for the elongation anisotropy observed at reductions ranging from 12% to 50%. This anisotropy results from differences in fracture points between tensile directions (Fig. 8). Wang et al. explained the fracture mode of A356 in terms of microcrack formation at the eutectic Si and the linkage of those cracks.11) They reported that the particle size and aspect ratio affected the particle cracking, and that the spacing between cracked particles influenced the linkage. Based on this theory, differences in the particle morphology, distribution, and orientation between the observation directions (Fig. 3–6) are expected to cause elongation anisotropy as described below. The high-speed twin-roll cast strip exhibited no difference in particle morphology and distribution between observed sections. In addition, these particles are randomly orientated. Therefore, the strip has isotropic mechanical properties. Even in the 12% rolled strip, the particle orientation in both sections is nearly random. However, since the shrinkage cavities are elongated toward RD, as shown in Fig. 9(a), the elongation along TD is below than that of RD. Further cold rolling, up to 50%, expands the differences in particle morphology and distribution between the RD and TD sections, particularly in the quarter-thickness region, as shown in Fig. 3(b), 4(b), and 5(b). Also, the number of particles oriented toward RD clearly increased in the TD section (Fig. 6). As these features accelerate the particle cracking and linkage in the TD section, the elongation anisotropy is expanded. When the reduction is increased to 73%, although the particle orientation is escalated (Fig. 6), the differences in particle morphology and distribution between the RD and TD sections are almost eliminated. Since particles in the mid-thickness region are sufficiently refined and spheroidized to the same level as the particles in the surface region (Fig. 3, 4), the effect of particle orientation become negligible. These changes improve the elongation along TD enough to catch up with that along RD, eliminating the elongation anisotropy.
This study shows that it is necessary to develop a process for achieving high elongation without anisotropy to refine coarse particles in the mid-thickness region of the twin-roll-cast strips and homogeneously disperse them into the matrix.
