A new method for quantitative XEDS tomography of complex hetero‐nanostructures

Abstract

Over the last decades, electron tomography based on HAADF‐STEM has evolved into a standard technique to investigate the morphology and inner structure of nanomaterials. The HAADF‐STEM intensity depends on sample thickness but also scales with the atomic number Z and therefore, chemical compositions can be studied from these three‐dimensional (3D) reconstructions.[1] Nevertheless, it is not straightforward to interpret the gray levels in a 3D HAADF‐STEM reconstruction, when mixing of elements is expected or elements with atomic number Z close to each other are present. In an increasing number of recent studies, X‐ray Energy Dispersive Spectroscopy (XEDS) has been combined with tomography to understand complex nanostructure morphology and composition in 3D.[2] These studies rely on newly developed XEDS detectors such as the Super‐X detection system, which consists of four individual detectors, symmetrically arranged around the TEM sample.[3] By using the Super‐X detector, one is able to overcome problems that were previously related to extreme shadowing of the XEDS signal caused by the sample‐detector configuration. Although this problem can be largely overcome, some shadowing effects remain, as illustrated in Figure 1. Since such shadowing effects vary for different tilt angles, the XEDS will also depend on the tilt angle and the projection principle for electron tomography is no longer fulfilled. Because of this problem and the low signal‐to‐noise ratio, typical of XEDS mapping, it remains challenging to obtain quantitative information by 3D XEDS and further progress is required. Here, we propose an alternative approach to optimize the reconstruction of an XEDS tomography series by minimizing the impact of shadowing effects and improving the spatial resolution. The method is based on the synergistic combination of HAADF‐STEM tomography and XEDS quantitative mapping.[4‐5] HAADF‐STEM yields a relatively high signal‐to‐noise ratio, enabling an accurate reconstruction of the morphology. XEDS, on the other hand, yields chemical information, but the limited amount of data that can be usually collected, hampers a good morphological reconstruction. As a proof of principle, we apply our methodology to a nanostructure containing a mix of Au and Ag atoms. It should be mentioned that the approach we propose here enables quantitative 3D chemical characterization of a broad variety of nanostructures.