Linear chemically sensitive electron tomography using DualEELS and compressed sensing

Abstract

Electron tomography (ET) is now increasingly important for recovering the three-dimensional (3D) morphology of nanostructured materials in the physical and life sciences. ET typically involves the acquisition of a set of two-dimensional projection images at different tilts using (scanning) transmission electron microscopy ([S]TEM), followed by alignment and reconstruction using established algorithms to reconstruct a 3D volume that represent the physical morphology or 3D distribution of some other property of the specimen under investigation. In principle, the methodology is independent of the nature of the images and is applicable to any imaging technique that fulfils the projection requirement [1] such that the signal should change monotonically with the physical property of the sample. This condition is approximately fulfilled for mass-thickness contrast in bright field TEM of amorphous biological specimens, and high angle annular dark field (HAADF) STEM imaging of thin specimens. Consequently, both imaging techniques have been widely used in ET. Recently, ET has been performed using spectroscopic signals, including X-ray spectroscopy, energyfiltered TEM (EFTEM), and electron energy loss spectroscopy (EELS) in the STEM, to achieve a chemically sensitive 3D reconstruction. Whilst X-rays and EFTEM mainly allow the mapping of elemental contents, EELS offers additional possibilities for studying detailed chemistry. Early studies have already shown the feasibility of EELS-STEM tomography [2,3]. Nevertheless, EELS suffers significantly from multiple scattering, especially for thicker specimens, and this makes the backgroundsubtracted edge signal a non-linear function of thickness, which leads to reconstruction artifacts.

Electron tomography image reconstruction using data-driven adaptive compressed sensing

“…In this article, we extend our work in (AlAfeef et al ., ,,b) and explore the strengths of patch‐based adaptive redundant dictionaries for electron tomography. The main contribution of this article is the development of a fast and robust numerical algorithm, which we have named DLET, for ET reconstruction and the application on experimental dataset.…”