Grain boundary networks in nanocrystalline alloys from atom probe tomography quantization and autocorrelation mapping

Abstract

A local spatial autocorrelation-based modelling method is developed to reconstruct nanoscale grain structures in nanocrystalline materials from atom probe tomography (APT) data, which provide atomic positions and species, with minimal noise. Using a nanocrystalline alloy with an average grain size of 16 nm as a model material, we reconstruct the three-dimensional grain boundary network by carrying out two series of APT data quantization using ellipsoidal binning, the first probing the anisotropy in the apparent local atomic density and the second quantifying the local spatial autocorrelation. This approach enables automatic and efficient quantification and visualization of grain structure in a large volume and at the finest nanoscale grain sizes, and provides a means for correlating local chemistry with grain boundaries or triple junctions in nanocrystalline materials.Nanoscale grain boundary networks are reconstructed from atom probe tomography data, which provide atomic positions and species for a fraction of atoms within a nanocrystalline material with an average grain size of 16 nm, using a quantization and local spatial autocorrelation-based approach.

Introducing a Crystallography-Mediated Reconstruction (CMR) Approach to Atom Probe Tomography

“…Extending this manual process to map grain boundaries at multiple slices throughout the evaporation sequence would be an arduous task, and prone to user error. However, it may be possible to apply a method, such as that described by Chen & Schuh (2015), to automatically detect and map grain boundaries networks to isolate vertices within individual grains. This represents an avenue for future work.…”

“…This is particularly true for studies based on critical distances where the accuracy of inter-feature distances is crucial for explaining materials science phenomena. Examples include grain boundary mapping (Yao et al, 2013; Chen & Schuh, 2015) and interfacial excess mapping (Liddicoat et al, 2010; Sha et al, 2011), or for the analyses of crystallographic relationships (Breen et al, 2017).…”