PNNL

Abstract

While application of clustering algorithms to atom probe tomography data have enabled quantification of solute clusters in terms of number density, size, and subcomposition there exist other properties (e.g. volume, surface area, and composition) that are better determined by defining an interface between the cluster and the surrounding matrix. Namely volume, surface area, and total composition. The limitation in composition results from a pre-filtering step where the expected matrix ion types are omitted from the cluster search to enhance the contrast between the matrix and cluster and to reduce the complexity of the search. Previously composition determination within solute clusters has utilized a secondary envelopment and erosion step on top of conventional methods such as maximum separation. In this work we present a novel and automated method that combines the particle identification fidelity of a conventional clustering algorithm with the analytical flexibility of mesh-based approaches through the generation of alpha shapes for each identified cluster. The corresponding mesh accounts for concave components of the clusters and determines the volume and surface area of the clusters, additionally the mesh boundary is utilized to update the total composition according to the internal ions.