Nanoscale voxel spectroscopy by simultaneous EELS and EDS tomography

Abstract

Extending the capabilities of electron tomography with advanced imaging techniques and novel data processing methods, can augment the information content in three-dimensional (3D) reconstructions from projections taken in the transmission electron microscope (TEM). In this work we present the application of simultaneous electron energy-loss spectroscopy (EELS) and energy-dispersive X-ray spectroscopy (EDS) to scanning TEM tomography. Various tools, including refined tilt alignment procedures, multivariate statistical analysis and total-variation minimization enable the 3D reconstruction of analytical tomograms, providing 3D analytical metrics of materials science samples at the nanometer scale. This includes volumetric elemental maps, and reconstructions of EDS, low-loss and core-loss EELS spectra as four-dimensional spectrum volumes containing 3D local voxel spectra. From these spectra, compositional, 3D localized elemental analysis becomes possible opening the pathway to 3D nanoscale elemental quantification.

Linear chemically sensitive electron tomography using DualEELS and dictionary-based compressed sensing

“…Compressed sensing has been recently applied to ET [13,[31][32][33][34] and shows promise for reducing missing wedge artefacts. The use of image sparsity, as a priori knowledge to enhance the accuracy of reconstruction, can reduce the number of projections than what is required by conventional techniques.…”

“…The multiple scattering can be dealt with by using either deconvolution [10] to remove it or modeling [11,12] to take account of it in the quantification. This has recently been used by Haberfehlner et al [13] to perform chemically sensitive 3D reconstruction of precipitates in an Al-Si alloy containing Yb.…”

“…For a constant density material, N will have a linear scaling with thickness and thus, in order to achieve a truly linear relationship between integrated, background-subtracted core-less signals and thickness, the signal I should be normalised by the intensity present in the zero loss peak I 0 . This normalisation is implicit in the model based approach used by Haberfehlner et al [13] but needs to be performed explicitly if deconvolution, background subtraction and numerical integration of edge intensity is to be used to produce maps for three-dimensional reconstruction.…”

Three dimensional mapping of Fe dopants in ceria nanocrystals using direct spectroscopic electron tomography

“…An alternative is to reconstruct each energy loss separately yielding a 4D data cube where an EELS spectrum can be extracted from each 3D voxel. This method was for example recently used for the characterization of an Al-5 wt% Si alloy and for the 3D investigation of variations of the valence state of Ce in ceria nanocrystals [3,12].…”

Quantitative Aspects of 3D Chemical Tomography in the Scanning Transmission Electron Microscope

“…In ET, the 3-D structure of a specimen is mathematically calculated from a series of 2-D projections acquired over a range of specimen orientations. Recent advances in specimen preparation, automation of data collection, and improved reconstruction methods have also made it feasible to couple tomographic data acquisition with X-ray energy dispersive and electron energy-loss spectroscopies (XEDS and EELS, respectively) in order to carry out 3-D elemental and chemical analysis at similar length scales [1][2][3][4][5][6][7].…”