Any task in simulation consists of two fundamental problems: knowing the physical laws governing the process and knowing the initial and boundary conditions of the simulation. In the quest to create 'digital materials', both problems are key to creating a comprehensive solution.
For continuum problems, a number of 3D characterisation methods are available to either determine the initial state of a material before an event or the final state. With adequate data treatment, these datasets can then be used as initial conditions for simulation. To achieve something equivalent for atomistic simulations, the knowledge of the position of each atom within a certain volume is required, ideally with the ability to identify the atom's chemical identity. While the former has been achieved in limited volumes using transmission electron microscopy  and field ion microscopy , the latter is yet to be done. The possibility to create such data however has moved closer to reality in recent years with detectors in atom probe tomography approaching 90% detection efficiency (percentage of atoms captured out of atoms field evaporated) and detectors with effectively 100% efficiency on the horizon for the coming years.
In this talk we will present our strategies and initial experimental results on the creation of atomistically simulatable 3D data from metallic materials. This work in this case is based on 3D field ion microscopy, an atomic scale 'serial sectioning' method where tip shaped samples are field evaporated slowly while an atomic resolution image of the surface is acquired. Using computer reconstruction algorithms, a 3D dataset, in which all atoms are imaged within a limited volume is then produced. This dataset faithfully reproduces the atomic topology of the crystal lattice as well as crystal defects, however, global distortions may still be present. These reconstructions also lay the groundwork for atom probe tomography reconstructions from high efficiency detectors, where instead of thousands of individual ion impacts, each atom is only registered once, upon its impact.
 Van Aert et al., Nature 470 (2011) 374-377
 Vurpillot et al., Microscopy & Microanalysis 23 (2017) 210-220