Introduction to Atomic-Scale Tomography

Tomography is a general term used to describe experimental approaches for determining the internal 3D structure of objects. The term ‘tomography’ was coined by Mayer in 1914 (as noted in [1]) to describe the derivation of 3D information from 2D X-radiographs. Eventually, the TEM was used to obtain tomographic images [2] on a finer scale than X-rays. As TEM image resolution reached sub-0.2 nm, this begged the question whether tomography resolution could reach single atoms. Atomic-Scale Tomography (AST) is a discrete tomography that seeks each atom’s identity and location in a structure. A definition of AST was presented in [3] as: 100% of the atoms should be positioned and identified. The 3D position of each atom should be determined with a precision and accuracy sufficient to determine crystal structure ∼or atomic structure of noncrystalline materials! Deviations from perfect crystal structure, that is, crystal defects, should be apparent. The atomic (and isotopic) identities should be determined with high precision and accuracy. The information should cover a large enough volume to be technologically useful. Based on today’s standards, this will be taken to mean millions to billions of atoms. In addition to the original four points, a later paper added one additional point to define ASAT [4]: Experimental chemical bond state information should be available on the atomic scale throughout the volume in sufficient detail to unequivocally match computed chemical state information. These five points describe an experimental enterprise where a 3D representation of the specimen contains structural, chemical, compositional, and physical properties information at the atomic scale. Atomic-scale tomographs have been made with TEM by Bals et al. [5] of a single element and Yang et al. [6] of a binary alloy where contrast could be used to identify the atom type, Figure 1. These tomographic images consisted of thousands of atoms. 4D STEM [7] enables determination the position of millions of atoms in a specimen with sub-100 pm precision. Impressive as this may be, the atom identities are not known except by image contrast and inference. Atom probe tomography (APT) has also been pursued as a basis for AST [3,4,8]. APT readily produces tomographs with millions of atoms which are isotopically identified in 3D by time-of-flight spectrometry. All isotopes are detected with equal efficacy. However, two principal limitations would need to be addressed if AST is to be achieved by APT alone. 1) Ion detector technology is based on microchannel plates which provide ∼80% detection efficiency while AST seeks detection of all (100%) atoms. Efforts to develop superconducting ion detectors previously at CAMECA and now at Steam Instruments aim to deliver 100% detection. 2) Atom positioning precision is limited in reconstruction algorithms that lack information on the trajectories that ions take from the specimen to the detector. Building an atom probe in the pole piece of a (S)TEM would make it possible to determine the information needed to correct the reconstruction errors. This approach is described in detail in [8] and is being pursued in Project Tomo. Left: Tomographic image of silver atoms in a precipitate in aluminum. From Bals et al [5]. Right: FePt binary allow imaged with STEM. Fe (red), Pt (blue) from Yang et al. [6]. Left: Rendered ball model of atoms in silicon with impurity atoms. Right: Nominal hardware configuration for AST [8].