Grain boundary segregation is often reported as an average composition. Look more closely, however, and that average can hide substantial chemical variation across the interface. This raises a practical question for atom probe tomography: how much grain boundary area must be analyzed before the measured composition is truly representative?
Jacob B. Mack and Christopher A. Schuh explored that question in nanocrystalline Ni-2.5Pd by comparing APT measurements with atomistic simulations. They tracked segregation from small interfacial patches to entire boundaries and tested the methods used to extract those measurements.
The results point to an important distinction. In this system, palladium concentration varied far more within the same random boundary than it did from one measured boundary to another. The area included in the analysis therefore mattered as much as the boundaries selected.
“Why APT? The problem was inherently spatial. APT provided three-dimensional atomic positions and chemical identities across large analysis volumes. This allowed the researchers to locate boundaries, measure local interfacial excess, map palladium across thousands of square nanometers, and see how the sampled area changed the result.”
The experimental material was nanocrystalline Ni-2.5Pd powder equilibrated at 600 °C, with an average grain size of 110 nm. Researchers analyzed APT specimens using a CAMECA LEAP 5000 XS at the Northwestern University Center for Atom Probe Tomography (NUCAPT). Typical runs collected more than 10 million ions, and a total of eight grain boundaries were captured.
The boundaries became visible through an increased apparent density of palladium atoms, reflecting both chemical segregation and local magnification effects. The researchers traced each oriented boundary through successive slices, constructed a three-dimensional mesh, and assigned atoms to equal-area trajectories normal to the interface.
Ladder diagrams then provided the Gibbsian interfacial excess for each part of the mesh, showing how palladium segregation changed across the boundary.
The analysis workflow is shown in Figure 1: Overview of extracting local grain boundary concentrations from Atom Probe Tomography (a-d) and Atomistic Simulation (e-h) datasets. In order; (a,e) raw data, highlighting positions of solute atoms, from Atom Probe and Atomistics, (b,f) Ellipsoidal binning to label the grain boundary, (c,g) 3D mesh created from labeled points, (d,h) atom assignments for each mesh element, leading to the creation of ladder diagrams.Overview of extracting local grain boundary concentrations from Atom Probe Tomography (a-d) and Atomistic Simulation (e-h) datasets. In order; (a,e) raw data, highlighting positions of solute atoms, from Atom Probe and Atomistics, (b,f) Ellipsoidal binning to label the grain boundary, (c,g) 3D mesh created from labeled points, (d,h) atom assignments for each mesh element, leading to the creation of ladder diagrams.
At a patch size of about 4 nm², the APT maps showed local palladium concentrations ranging from 0 to 30 at.% within one boundary. That boundary averaged 9.8 at.%, but many smaller regions departed sharply from the mean. The variation remained apparent even at 100 nm², where local values differed from the boundary average by approximately ±4 at.%.
Measurement uncertainty could not account for the full spread. At 4 nm², the estimated measurement standard deviation was 2 at.%, compared with an observed standard deviation of 5.2 at.% across the combined dataset. The range narrowed as more of the interface was included, approaching a stable boundary average.
The relationship between sampled area and measured concentration is shown in Figure 2.
Jacob Mack is currently a Ph.D. student in the Department of Materials Science and Engineering (DMSE) at Massachusetts Institute of Technology. His research is focused on the chemistry of grain boundaries in metallic alloys, and how it can be manipulated to alter the functional properties of polycrystalline materials. This work spans a range of techniques, including Atom Probe Tomography (APT), TEM, X-ray Diffraction, and atomistic simulations.
Christopher Schuh is the Dean of the Robert R. McCormick School of Engineering at Northwestern University, and holds the title of John G. Searle Professor of Materials Science and Engineering. His research is broadly focused on structural materials including metals and ceramics. Dean Schuh has worked on the topic of grain boundaries and nanocrystalline structures for over twenty years, including the commercialization of various nanocrystalline alloys.
Note that the study reports sulfur isotope analyses performed by secondary ion mass spectrometry (SIMS). The current-generation SIMS platform available from CAMECA is the IMS 7f-Auto.
Read the scientific article: Dupeyron, J., Pasquier, V., Guibourdenche, L., Busigny, V., Cartigny, P., Jézéquel, D., Bernasconi, S.M., Marin Carbonne, J. (2025) Decadal, sediment-driven sulfur isotope evolution in Lake Cadagno. Geochemical Perspectives Letters 38, 17–22. https://doi.org/10.7185/geochemlet.2550.
