On correction of translational misalignments between section planes in 3D EBSD

Summary Three‐dimensional electron backscatter diffraction allows obtaining the 3D image of a material from the stack of 2D sections. This is achieved by repeated application of two different beams; electron beam for electron backscatter diffraction mapping of the surface and focused ion beam for removing a thin layer of material from the surface. In most of these systems with two beams, the experiment requires stage movements for correct positioning of the sample to the respective beams. However, imperfections in this positioning are difficult to avoid, which yield small translational misalignments between the sections in the output data. In this work, we deal with an important task of correcting these misalignments between the sections such that the 3D image is recovered properly. On a simple example, we demonstrate that commonly used methods fail in case there is a structural anisotropy in the material under consideration. We propose an improved alignment algorithm which can neglect this behaviour with the use of external support information on a systematic trend in the translational misalignments. Efficiency of the algorithm is proven on a number of simulated data with different kinds of anisotropy. Application to a real data sample of a fine grained aluminium alloy is also given. The algorithm is available in an open‐source library. Lay Description In serial sectioning experiments, a 3D image is recovered from a stack of 2D sections. This is utilized by repeated analysis of the surface and removing a thin layer from the surface of the material. We mainly focus on one of these techniques, three‐dimensional electron backscatter diffraction (3D EBSD), which can be used to observe grain microstructure in polycrystalline materials. Due to positioning of the sample in different stages of the 3D EBSD experiment, small translational misalignments are usually present in the produced EBSD maps. It is thus an important post‐processing task to align the maps such that the 3D image is recovered properly. This is usually done by comparing selected microstructural features in the data. However, common alignment methods can be inaccurate if these features have some preferred spatial orientations. A typical example, often taking place in deformed or wrought materials, is a grain elongation in one direction. We propose an improved alignment algorithm which can account for these spatial arrangements and provide a correct reconstruction of the 3D image. Our method uses e