3D reconstruction of the spatial distribution of dislocation loops using an automated stereo-imaging approach

• Published in: Ultramicroscopy Vol. 195; no. C; pp. 58 - 68
• Main Authors: Yu, Hongbing, Yi, Xiaoou, Hofmann, Felix
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 01-12-2018; Elsevier
• Subjects: Dislocation loops; Stereo-imaging; TEM, 3d spatial distribution; Stereo-imaging; TEM, 3d spatial distribution; Dislocation loops
• ISSN: 0304-3991; 1879-2723
• DOI: 10.1016/j.ultramic.2018.08.018

•Automated stereo-imaging for reconstructing 3D spatial distribution of small dislocation loops from 2D TEM micrographs.•Less sensitive to experimental errors. Fewer projections and smaller angular range needed than for conventional techniques.•Analysis of 3D defect population in self-ion-implanted tungsten reveals depth-dependent variation of the damage formed.
We propose an automated stereoscopic imaging approach for reconstructing the 3D spatial distribution of small dislocation loops (DLs) from 2D TEM micrographs. This method is demonstrated for small DLs in tungsten, formed by low-dose ion-implantation, that appear as circular spots in diffraction contrast TEM images. To extract the 3D position of specific DLs, their 2D position in multiple weak-beam dark-field TEM micrographs, recorded at different tilt angles, is fitted. From this fit the geometric centre and size of each DL in each micrograph can be extracted. To identify each specific DL in all the 2D projections, an automated forward prediction approach is used. A system of linear equations can then be setup, linking the 3D position of each DL to its 2D position in each projection, and solved using least-squares fitting. This approach is initially tested on synthetic data. For low projected loop densities (<20  ×  1015 m−2) only 3 projections are required for perfect recovery of the defect microstructure. More projections are required when the projected number density increases or realistic errors are included. 3D reconstruction of experimental data from the low-dose self-ion implanted tungsten sample reveals a damage microstructure in good agreement with the depth-dependent damage profile predicted by SRIM. A comparison with weighed back-projection shows that the stereo-imaging-approach requires fewer projections, is less sensitive to the angular range spanned, and is more resilient to spurious variations in local contrast. It also allows a more straightforward retrieval of quantitative information such as size and position of each loop.