Principal image decomposition for multi-detector backscatter electron topography reconstruction

Jan Neggers (1) , Eva Héripré (1) , Marc Bonnet (2) , Denis Boivin (3) , Alexandre Tanguy (4) , Simon Hallais (4) , Fabrice Gaslain (5) , Elodie Rouesne (6) , Stéphane Roux (2)

1 MSSMat - Laboratoire de mécanique des sols, structures et matériaux
2 LMT - Laboratoire de mécanique et technologie
3 DMAS, ONERA, Université Paris Saclay [Châtillon]
4 LMS - Laboratoire de mécanique des solides
5 CDM - Centre des Matériaux
6 SRMA - Service des Recherches Métallurgiques Appliquées

Scanning Electron Microscopes (SEMs) often generate images with a shaded appearance which gives a natural 3D impression. Ergo, quite a few methods to reconstruct the 3D surface topography from these using shape-from-shading methods are available in the literature. Here, a novel approach is discussed which uses BackScatter Electron (BSE) images from multiple detectors to reconstruct the topography. Classically, algorithms exist which resort to a quad-BSE detector setup. However, other detector configurations are often found in SEMs. A set of images of these non-conforming detectors still contains enough information to allow for reconstruction, but requires a more general algorithm to do so. This article discusses a method based on a modal decomposition of the principal image components. The resulting method is shown to be efficient and independent of the number of detectors or their orientation. In fact, the orientation is identified as part of the algorithm and thus requires very little calibration.

Keywords: Scanning Electron Microscopy; Topography reconstruction; Back Scatter Electron Supprimer; Principal Component Analysis

Domains: Engineering Sciences [physics]

Jan Neggers, Eva Héripré, Marc Bonnet, Denis Boivin, Alexandre Tanguy, et al.. Principal image decomposition for multi-detector backscatter electron topography reconstruction. Ultramicroscopy, 2021, 227, pp.113200. ⟨10.1016/j.ultramic.2020.113200⟩. ⟨hal-03261200⟩