The Correlation between Grain Boundary Character and Intergranular Corrosion Susceptibility of 2124 Aluminum Alloy

The effects of grain boundary character on the intergranular corrosion susceptibility of 2124 aluminum alloy were examined. In the study, the alloy was heat treated at 540oC and corrosion tested according to ASTM G110 standards. After obtaining grain orientations from the automated Electron Back-Scatter Diffraction (EBSD), both grain boundary character and grain boundary plane distributions were analyzed. Results show that low-angle boundaries and boundaries with a sigma-3 or sigma-7 coincident site lattice relationship tend to have a higher corrosion resistance than other random boundaries. INTRODUCTION Grain boundary dependent properties such as fracture and intergranular corrosion have been found to be strongly dependent on the crystallographic nature of the grain boundaries [1]. The concept of “Grain Boundary Design and Control”, otherwise known as “Grain Boundary Engineering” (GBE), was first introduced by Watanabe [1]. GBE asserts that by manipulating the Grain Boundary Character Distribution (GBCD) to generate a higher fraction of low-energy Coincident Site Lattice (CSL) boundaries, the crack or corrosion resistance of the material can be improved [1]. GBE has been shown to be successful in numerous studies on nickel-based 600 alloy [2, 3], and 304 austenitic stainless steels [4-6]. In most cases, the corrosion resistance of the material increased with an increasing fraction of low CSL boundaries (Σ≤29), and especially with Σ3 boundaries being the most corrosion-resistant. Other studies however, have mentioned that not all Σ≤29 boundaries are corrosion resistant. For example, Gertsman et al. [7] and Henrie et al. [8] have both observed that not all Σ3 boundaries are resistant to corrosion. In this study, the effects of grain boundary character on the intergranular corrosion resistance of the 2124 aluminum alloy will be examined. Figure 1. (a) SEM image of the corroded surface, where only the corroded boundaries are visible. (b) Image Quality map of the OIM scanned image, which shows the microstructure of the alloy. MATERIAL PROCESSING AND CORROSION TESTING The material used in this study was 2124 aluminum alloy; having a nominal chemical composition (wt %) of Cu: 3.8, Mg: 1.2, Mn: 0.48, Fe: 0.09, Si: 0.04, Zn: 0.04, Ti: 0.02, and balance Al. Plates of 6.3mm thickness first underwent a heat treatment at 350°C for 120 minutes followed by air cooling to remove any residual strain before it was subsequently cold rolled to 82% reduction with 23 passes through a constant speed rolling mill. After cold working, samples received a solutionizing heat treatment of 540°C for 2h followed by quenching in room temperature water immediately after removal from the box furnace. The water quenching ensures that the precipitates remain evenly distributed throughout the material. The microstructure of the alloy can be seen in the image quality map presented in Figure 1b. As can be seen in the figure, the grains are equiaxed with a non-uniform grain size distribution. All annealed samples were subjected to ASTM G110 corrosion testing, where 2.5cm x 2.5cm samples were constantly immersed in a sodium chloride and hydrogen peroxide solution for 24 h. Figure 2. Inverse pole figure maps showing (a) the EBSD data as obtained from the OIM system, (b) the EBSD data that has been cleaned up with the TSL software and the grain boundary segments traced-in, and (c) the grain map with the corroded grain boundaries manually traced back onto the cleaned map. EBSD DATA COLLECTION The cross-sections perpendicular to the rolling plane of the corroded samples, namely the ND-RD plane, were mechanically polished for microstructural characterization through Electron Backscatter Diffraction (EBSD). All observations of grain boundary corrosion and images shown in this study were obtained on such ND-RD planes after mechanical polishing. The grain boundary microstructures were examined with a Phillips XL40 FEG instrument equipped with an EBSD indexing system provided by TexSEM Laboratories, Inc. The variables used in collecting diffraction data include: an accelerating voltage of 20kV, a working distance of approximately 17mm, and a step size of 1μm scanning on a hexagonal sampling grid. The Orientation Imaging Microscopy (OIM) software package is used to identify CSL boundaries using Brandon’s criterion [9], generate inverse pole figure maps, and extract boundary line traces that were used to determine the grain boundary plane distribution. Figure 2 shows one of the seven EBSD data sets portrayed with an inverse pole figure map, and demonstrates the method used to process the EBSD data. As a cleanup procedure in the OIM software, the scan data was grain dilated with a grain being defined as any group of pixels having more than 15 pixels with grain boundaries having a misorientation angle greater than 5°. The cleaned EBSD data was then converted into a square grid to allow the inverse pole figure map to be exported such that each pixel on the image represents one scanning step size. The corroded boundaries are traced by hand back onto the cleaned OIM image. A computer program written in C++ (see pseudo code) was then used to read in the image and categorize the boundary data into corroded and non-corroded boundary groups. Pseudo Code: Separating corroded and non-corroded boundaries. Input: exported .ang file and reconstructed boundary file from TSL software Output: corroded boundary file and non-corroded boundary file scan input files for number of grain boundary segments read in and store confidence index (CI) values for all points from .ang file for each boundary segment do open elliptical scanning window with a=1 and c=3 pixel size align scanning window to have the long axis be perpendicular to and centered on boundary segment move scanning window along boundary segment in 1 pixel step size count up the number of black points (CI value = -1) within the elliptical window at each step of the moving window sort data and find median number of black points for the boundary segment if median=0 copy boundary segment data into non-corroded boundary file else copy boundary segment data into corroded boundary file end for ______________________________________________________________________________ RESULTS AND DISCUSSION EBSD data results show that the area-weighted average grain size is 66.7μm, while the number-weighted average grain size is 41.8μm. Pole figures of the corroded samples show that the material has the expected recrystallization texture of face-centered cubic (FCC) materials, the cube texture, as shown in Figure 3. Volume percents of the texture components that are commonly found in FCC materials are given in Table I. The volume percents of the texture components indicate that the material generally has a random texture with the exception of having a higher volume of cube texture and lower volume of the brass texture. The probability density plot for the corroded and non-corroded boundary groups are plotted in Figure 4. The probability density of the corroded grain boundaries closely follows the McKenzie distribution for a random sample, which seems to indicate that corrosion occurs randomly on all boundaries. However, there is also a very high fraction of low-angle grain boundaries that is associated with the non-corroded boundary group. Such results show that low angle grain boundaries clearly have a higher resistance to corrosion, which is consistent with studies performed on other aluminum alloys [10, 11]. Figure 3. Example of a pole figure for one of the EBSD data sets showing cube texture. Table I. Volume percents of typical FCC texture components as observed in the EBSD data sets compared to values expected in random texture. Texture Components (φ1,Φ,φ2) Experimentally Observed (Volume Percent) Expected In Random Texture (Volume Percent) Cube (0°, 0°, 0°) 9.44 2.45 Brass (35°, 45°, 0°) 0.61 3.23 Copper (90°, 35°, 45°) 3.90 4.29 S (50°, 35°, 70°) 7.76 7.39 Goss (0°, 45°, 0°) 1.27 2.64