71 72 STRAIN-INDUCED TEXTURE DEVELOPMENT IN THE MAGNESIUM ALLOY AZ31 Shiyao Huang, Dayong Li and Yinghong Peng Shanghai Jiaotong University, Shanghai, China. John Allison University of Michigan, Ann Arbor, Michigan ABSTRACT Magnesium alloys are receiving increasing attention as a structural material in vehicles to reduce weight and increase fuel economy. Understanding the formability of lightweight alloys is one key to their successful introduction in vehicles. We present results of X-Ray diffraction and electron back-scatter diffraction texture measurements in as-cast and uniaxially compressed AZ31 for a variety of strains, strain rates and temperatures (0.2 to 1.0, 0.01s -1 to 1.0s -1, 623K to 673K, respectively). Pole figures show that all the samples are dominated by large grains, even those that have undergone very large deformations. In order to understand the trends in texture, it was necessary to average results from multiple sample sections to overcome the poor statistics of the coarse grained samples. Our results indicate that the expected basal texture develops at very low strains (<0.4) and remains essentially unchanged at higher strains (0.4-1.0). INTRODUCTION The continuous push to higher vehicle fuel economy will require improvements in both powertrain technology, aerodynamics and weight reduction. Magnesium alloys offer an opportunity to achieve significant weight reductions when compared to steel or aluminum, while maintaining high strength compared to polymers. To enable reliable engineering choices that maintain materials properties, a predictive tool is needed. At Ford, we are developing a Integrated Computational Materials Engineering (ICME) tool that incorporates materials chemistry and thermodynamics with processing parameters (casting conditions, forming operations, heat treatments) to predict material strength, fatigue resistance and ductility. To develop the tool, empirical measurements of processed materials, as well as computational data from first-principle calculations (e.g., Density functional theory estimations of formation enthalpies) and finite-element simulations are used to refine weights in the engineering decision trees. There are various techniques which can measure grain orientation texture, including Electron Backscatter Diffraction (EBSD), neutron diffraction and X-ray diffraction (XRD). EBSD provides micron-scale lateral spatial resolution but is only sensitive to nanometer depths. This feature presents a significant challenge to studying alkali and alkaline-earth metals due to their rapid oxidation, where surface oxides less than 10nm thick can obscure the electron diffraction arcs. In addition, for large grained materials, EBSD can be very cumbersome, since large areas need to be scanned to develop a statistically meaningful average of the volume-weighted orientation distribution function, resulting in long acquisition times. Neutron diffraction solves the problems of sample preparation and sampling by using diffraction from a large volume (cm 3 )
This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website www.dxcicdd.com ICDD Website - www.icdd.com
72 73 in each measurement without any surface preparation. However, neutron facilities are very limited (reactor or spallation sources), and requiring long lead-times for access. Intermediate between these two alternatives is texture measurements by laboratory-based XRD, where sample preparation requirements are minimal (smooth polish) and volumes are of the order ~0.01mm 3 can be sampled in a few hours of acquisition time. Previous studies of texture development in Mg have used all three techniques cited above. Yi, et al. (2006) investigated texture changes in AZ31 extruded rods using high-energy X-ray texture measurements during quasistatic deformation in-situ. For uniaxial compressive strains along the extrusion direction spanning 0-0.09, they found significant texture changes: grains with their basal planes initially parallel to the compression direction were reoriented such that their basal planes became perpendicular to the compression direction. The effect largely saturated above a total strain of 0.06. Jiang et al. (2008) examined the effects of the initial orientation on sheets of AZ31 using EBSD. Although they examined areas 0.5mm 0.5mm, some maps clearly indicated that large grains were present that represented a significant proportion of the volume examined and the resulting pole figures suffered from the effects of low sampling described above. Choi, et al.(2007) used neutron diffraction to characterize texture in rolled sheets of AZ31 and demonstrated strong basal (0001) texture with excellent fidelity, clearly demonstrating the ability of this technique to sample very large volumes. To implement the ICME approach to design with Mg alloys, a basic understanding of the texture development for a variety of strains at different temperatures and strain-rates. We describe here our attempts to characterize the changes in texture of cast AZ31 in uniaxial compressive strain experiments for several temperatures and strain rates using ex-situ laboratory-based XRD measurements. METHODS AND MATERIALS Samples of AZ31 alloy were obtained from a single, cylindrical cast billet produced at a casting speed of 120mm/min from a 680 C melt into a mold cooled with water at 25 C. The billet was produced at Shanghai Light Alloy Net-Forming National Engineering Research Center Co., LTD. The finished billet (100mm diameter, 1000mm long) is shown in Figure 1. Slices cut from the billet were cored to extract samples at two different radii (~10mm and ~40mm) (Figure 2). In addition to as-cast samples, multiple samples were plastically deformed to strains 0.2, 0.4, 0.6, 0.8 and 1.0 and at strain rates of 0.01, 0.1 and 1.0 s -1 at 473K, 523K, 573K, 623K and 673K. Samples deformed at elevated temperatures were indirectly heated with a cylindrical electrical element while loaded in a Gleeble 3500 hydraulic press. Figure 1. As-cast AZ31 billet. Figure 2. Disc cut from billet before and after coring. Individual samples are also show after compression.
73 74 As-cast and strained samples were initially characterized by optical microscopy and EBSD. Samples for microscopy were prepared by first mechanically polishing on a progression of SiC papers, followed by fine polishing on Al 2 O 3 slurry. Samples for EBSD were further electropolished in a nitric acid/ methanol mixture at 20V. Polished samples were loaded into a scanning electron microscope equipped with a backscatter diffraction stage. Diffraction data from each point was used to automatically determine the crystallite orientation distribution and produce a pole figure. Total data acquisition time for each EBSD pole figure was 18-24h. Samples for XRD characterization were prepared in a fashion similar to the initial preparation used for microscopy, with mechanical polishing to a 1 m surface finish used as the final preparation. Layer removal through polishing was also used to obtain a better estimation of the true average texture. In this case, each successive sample was prepared by removing a ~200 m thick layer using the methods described above to reveal fresh surface for XRD analysis. XRD pole figures were obtained using a Scintag XDS2000 four-circle diffractometer equipped with a copper tube and a Si(Li) detector. Pole intensities for the (0002) and (10-11) reflections were obtained for 0 < < 80 degrees and 0 < < 355 degrees in steps of 5 degrees. Background intensities were obtained by interpolating from points on either side of each reflection and net intensities were normalized to intensities obtained from a powder sample of Mg (Fig. 3a, 3b). To avoid introducing experimental noise and small deviations from the true random powder intensity, the pole intensities from the powder sample were first azimuthally averaged to produce a better estimation of the true defocusing and geometric effects (Fig. 3c). To avoid distortion of the pole intensities from irregularly shaped samples, all measurements (including from the random Mg powder) were confined to a 5mm diameter region on the sample face using a mask made from Ag paint. Pole figures were plotted using the software program popla (Kallend, et al., 1991). Intensity Intensity 3000 2500 2000 1500 1000 500 (0002) 0 0 50 100 150 200 250 300 350 600 500 400 300 200 100 0 0 50 100 150 200 250 300 350 Azimuth angle (10-11) Average Intensity (cps) 10 0 20 40 60 80 Sample Tilt Angle ( ) (deg.) Figure 3. (a), (b): Intensity scans obtained for a sample of Mg powder at the (0002) and (10-11) reflections. (c): Azimuthally averaged intensities (log scale) for several reflections plotted versus tilt ( angle. To avoid introducing noise and geometric errors, averaged values were used to determine the "defocusing" normalization, rather than using raw measured data. 10 4 1000 100 Azimuthally Averaged Mg Powder Data (101) (002) (102) (100) (110)
74 75 RESULTS AND DISCUSSION Samples removed from the inner and outer sampling locations on each disc were compared by measuring their stress-strain response at different strain-rates and temperatures. Results of these measurements show that there was no observable difference between the two sampling locations and will not be discussed further. Representative optical microscopy images from samples of AZ31 strained at 623K and a strain rate of 0.1 are shown in Figure 4. Over the range of Figure 4. Optical images of AZ31 compressed to total strains of 0.2, 0.4, 0.6, 0.8 and 1.0 at a strain rate of 1.0 and temperature of 623K. The scale shown in each is 50 m. Figure 5. (0002) pole figure determined using EBSD over an area of 1.0 0.8 mm. compressive strains (0.2 < 1.0), large grains are dominant, with fine grains appearing at the periphery of the large grains in the so-called "necklace structure". In addition, the larger grains appear to accumulate significant damage in the form of twins and slip dislocations at larger strains. An EBSD (0001) pole figure determined from a 1mm 0.8mm area is shown in Figure 5. The "grainy" texture apparent in the EBSD pole figure is consistent with the large grain size observed in the optical microscopy images and indicates fewer than ~20 distinct grains/mm 2 contribute. This result demonstrates the difficulty in obtaining meaningful data for engineering models for this material by EBSD and is consistent with the generally poor quality of the pole figures previously reported by this technique (Jiang et al., 2008). Although newer, more sophisticated instruments are now available, there is little reason to hope that meaningful data can be obtained with sufficient grain averages by this technique. Four separate specimens strained to =0.2 were prepared and (0002) pole figures obtained by XRD, shown in Figure 6. Despite sampling over ~16 the area and ~10 the depth as EBSD (1/e depth for Cu radiation in Mg at 20 degree incidence is ~51 m), XRD pole figures still show significant graininess, characteristic of poor sampling statistics. As a result, any correspondence with the expected tendency towards an increase in basal texture as a function of strain is difficult to detect. To overcome the poor sampling available from a single sample preparation, multiple XRD pole figures were prepared from strained specimens by successive layer removal using polishing.
75 76 Figure 6. XRD (0002) pole figures for four specimens strained to =0.2 at a strain rate of 0.1s -1 at 673K. (a) = 0.2, 26 scans (b) = 0.4, 18 scans (c) = 0.6, 12 scans (d) = 0.8, 15 scans Figure 7. Multiple layer pole figure averages for samples strained at the indicated total strains and at a strain rate of 0.1s -1 at 673K. Intensity (arb. units) 0.1 0.01 (0002) 0.2 0.4 0.6 0.8 Intensity (arb. units) (10-11) s=0.2 s=0.4 s=0.6 s=0.8 0.001 0 20 40 60 80 Tilt Angle (, deg.) 0 20 40 60 80 Tilt (, deg.) Figure 8. Azimuthally averaged (0002) and (10-11) pole figures determined from the multi-layer average of pole figures shown above. Note the log intensity scale for the (0002) averages and linear intensity scale for the (10-11) averages.
76 77 Up to 26 individual pole figures were averaged to produce the composite (0002) and (1011) pole figures show in Figure 7. From the averaged pole figures shown, it is apparent that even modest strains ( =0.2) produces significant basal (0002) texture. However, differences in the degree of texturing for different strains are still not obvious. Similar to the approach for obtaining a distortion-free powder pole figure for normalization described above, azimuthal averaging was applied to the composite pole figures shown to extract the trends in texture development on the assumption that uniaxial compression would lead to basal pole texture. The results of averaging are shown in Figure 8 and indicate little change in texture beyond strains of 0.2. CONCLUSIONS Uniaxially compressed specimens of AZ31 were examined using optical microscopy, EBSD and XRD. Optical microscopy shows a tendency for erosion of large grains at their periphery and an increase in defects during compression. Using EBSD to determine texture proved largely impractical due to the presence of large grains. XRD pole figures from a single 5 mm diameter surface also showed coarse-grain features that largely obscured any trends in texture. To overcome this limitation, we averaged 12-26 pole figures from successive layers exposed by polishing to produce pole figures that show the development of basal (0002) texture at low ( <0.4) strains. To further clarify the differences in each strained sample, pole figures were azimuthally averaged and plotted versus the tilt angle. Basal texture development appears to be nearly complete at strains as low as =0.2, with little increase in texture apparent at higher strains. Although XRD has many advantages over EBSD for texture characterization, in the case of AZ31, there is still the need to increase the number of sampled grains to reliably reveal trends. Using layer removal and averaging, larger volumes can be accessed and included in pole figures and where azimuthal symmetry is expected, further averaging is possible to define subtle changes in texture, although this approach is not applicable in all situations. ACKNOWLEDGEMENTS The authors acknowledge the assistance of A. Drews, M. Li and J. Hangas at Ford Motor Company in Dearborn, Michigan for their assistance with this work. REFERENCES Choi, S.-H., J.K. Kim, B.J. Kim and Y.B. Park (2007) The Effect of Grain Size Distribution on the Shape of Flow Stress Curves of Mg 3Al 1Zn Under Uniaxial Compression, Acta Mater. 55, 4181. Jiang, J., A. Godfrey, W. Liu, and Q. Liu (2008), Microtexture Evolution Via Deformation Twinning and Slip During Compression of Magnesium Alloy AZ31, Mater Sci Eng A 483-484, 576. Kallend, J.S., U.F.Kocks, A.D.Rollet, H.R.Wenk, (1991), popla - An Integrated Software System for Texture Analysis, Mater Sci Eng A. 132, pp. 1-11. See also: J.S.Kallend, U.F.Kocks, A.D.Rollet, H.R.Wenk, (1991) Textures and Microstructures, Volume 14 18, 1203-1208. Yi, S.B., R.E. Bolmaro, H.G. Brokmeier, C.H.J. Davies, J. Homeyer and K.U. Kainer, (2006), Deformation and Texture Evolution in AZ31 Magnesium Alloy During Uniaxial Loading, Acta Mater. 54, 549.