On the use of density-based algorithms for the analysis of solute clustering in atom probe tomography data

Abstract

Because atom probe tomography (APT) provides three-dimensional reconstructions of small volumes by resolving atomic chemical identities and positions, it is uniquely suited to analyze solute clustering phenomena in materials. A number of approaches have been developed to extract clustering information from the 3D reconstructed dataset, and numerous reports can be found applying these methods to a wide variety of materials questions. However, results from clustering analyses can differ significantly from one report to another, even when performed on similar microstructures, raising questions about the reliability of APT to quantitatively describe solute clustering. In addition, analysis details are often not provided, preventing independent confirmation of the results. With the number of APT research groups growing quickly, the APT community recognizes the need for educating new users about common methods and artefacts, and for developing analysis and data reporting protocols that address issues of reproducibility, errors, and variability. To this end, a round robin experiment was organized among ten different international institutions. The goal is to provide a consistent framework for the analysis of irradiated stainless steels using APT. Through the development of more reliable and reproducible data analysis and through communication, this project also aims to advance the understanding between irradiated microstructure and materials performance by providing more complete quantitative microstructural input for modeling. The results, methods, and findings of this round robin will also apply to other clustering phenomena studied using APT, beyond the theme of radiation damage.