X-Ray Absorption-Corrected STEM-EDS Tomography for Absolute Quantification Across Interfaces in a Mg-Al-Ca Alloy

Achieving absolute elemental quantification (atoms/nm3) in three dimensions is essential for understanding the microstructural origins of mechanical and functional properties of materials. Tomographic approaches, such as tilt-series tomography using X-ray energy dispersive spectroscopy in the scanning transmission electron microscope (STEM-EDS), are needed to interrogate nanoscale compositional heterogeneity at interfaces. Inclined and three-dimensional interfaces obscure interface abruptness in two-dimensional analysis. Moreover, disentangling contributions from the sample surface in transmission requires three-dimensional elemental tomography. STEM-EDS tilt-series tomography, however, is complicated by detector shadowing [1] as well as sample thickness (often constrained by the dimensions of the microstructural features of interest) due to X-ray absorption within the sample along distinct three-dimensional and tilt-dependent trajectories to the EDS detectors [2, 3]. Here, we report on the development of absorption-corrected STEM-EDS tomography with absolute quantification. We take a Mg-Al-Ca alloy as a materials system that poses specific challenges for EDS quantification: All Kα emission lines are below 5 keV with Mg and Al Kα below 2 keV, the Al Kα emission line falls just past the onset of X-ray absorption K edge for Mg, and susceptibility to surface oxidation means samples in electron microscopy include oxygen contributions that do not correspond to the bulk alloy composition. Mg-Al-Ca alloys are of significant interest for automobile and aerospace sectors for the development of lightweight alloys that retain suitable creep-resistance within the higher operating temperatures sought for Mg-alloys (125-200°C) [4]. Such Mg-Al-Ca alloys exhibit a characteristic microstructure comprising soft α-Mg within an intermetallic skeleton. The interface between the skeleton and α-Mg plays a key role in cracking and failure [4]. We have consequently sought to develop STEM-EDS tomography for quantification of composition across these critical interfaces. We extracted a sample from such a Mg-Al-Ca alloy across an interface between α-Mg and the intermetallic skeleton in the form of a needle geometry by focused ion beam (FIB) milling. The needle geometry was selected to ensure the sample volume remained consistent throughout the tilt-series. STEM-EDS data were then collected on an FEI Titan G2 80-200 at the Ernst Ruska-Centre, Forschungszentrum Jülich. A STEM-EDS spectrum image tilt-series was collected, taken every 10° between -60o and +60o (13 spectrum images). For quantification, we implemented a partial cross-sections approach using standards-based calibration [5]. MgO and CaCO3 powder samples and an Al wedge sample were selected as calibration standards. For MgO and CaCO3, a single STEM-EDS spectrum image was acquired at 0o tilt followed by a STEM tomography tilt-series to enable cross-section determination from particle standards [6]. Electron energy loss spectroscopy was used for thickness determination alongside STEM-EDS spectrum imaging for cross-section determination from the wedge sample [7]. X-ray absorption and shadowing corrections were then incorporated within the tomographic reconstruction process. For X-ray absorption correction, our approach differs to previous tomographic methods for absorption correction using Cliff-Lorimer quantification and rotation of the reconstruction volume for X-ray trajectory determination3 by now using absolute quantification for the determination of the three-dimensional density and rotated X-ray trajectories within a fixed reconstruction volume [3]. For shadowing correction, the total number of atoms was used for normalisation, updated iteratively in parallel with iterative calculations of X-ray absorption. We have carried out these reconstructions using compressed sensing algorithms based on second order total variation (CS-TV2) to support reconstructions of the Mg-Al-Ca needle with piece-wise linear variation in composition. Figure 1 presents a volume rendering of the STEM-EDS reconstruction using CS-TV2 as well as a map highlighting the oxidation of the sample surface. To benchmark these results, we have also compared the absolute quantification using CS-TV2 reconstructions with STEM-EDS tomography using the simultaneous iterative reconstruction technique (SIRT) together with quantification parameters (scattering and ionization cross-sections) from the Monte Carlo programme MC X-ray for relative composition quantification [8, 9]. Three main compositional bands were identified in the needle (Figure 1a). The uppermost, labelled band 1, was identified as the α-Mg. Two further bands exhibited mixed compositions, denoted bands 2-3. Band 2 contained Mg as a major fraction while Band 3 was majority Al and contained the greatest Ca fraction. Given the position of the Al Kα emission line within the Mg K edge absorption, bands 2-3 were most susceptible to inaccuracies in compositional quantification with absorption-correction. Figure 2a-d illustrates the changes in Al content after absorption-correction in band 3. Final relative quantification results for both reconstruction approaches were similar, though CS-TV2 reconstructions exhibited lower sensitivity to absorption effects. Three-dimensional analysis also revealed heterogeneity within these bands and their interfaces (Figure 2e). Although the interfaces between bands were inclined relative to the electron beam, line profiles could be extracted perpendicular to these interfaces within the reconstruction volume. The dashed line in Figure 2e at ii) marks the interface between bands 2 and 3, with an increase in magnesium concentration at i) just before the interface. The increase in concentration of Al is flagged by iii) and shows small band of a different composition just after the band 2/3 interface. These findings establish significant finescale compositional heterogeneity can occur at interfaces in Mg-Al-Ca alloys and demonstrates the power of absorption-corrected STEM-EDS tomography [10]. a) three-dimensional compressed sensing reconstruction of the metal alloy needle. b) EDS map of oxygen in metal alloy needle. a,b) Elemental composition of band 3 after absorption correction for a) CS-TV2 and b) SIRT reconstructions. c,d) Al concentration in band 3 before and after absorption correction for c) CS-TV2 and d) SIRT reconstructions. e) Line profile extracted perpendicular to the inclined interface from within the 3D volume.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.