Answers are grounded in this paper and cite their sources.
‘Random’ solid solutions often possess chemical short range order (SRO), which can dramatically influence material properties and offers a pathway for SRO-engineering. A central challenge, however, is the accurate experimental characterization of SRO. Here, we use wurtzite ScxAl1-xN (ScAlN) – a semiconductor with attractive functional properties known to be sensitive to nanoscale Sc clustering [1,2] – to perform a comparative study of SRO characterization via four (scanning) transmission electron microscopy ((S)TEM) methods. We compare reciprocal space diffuse scattering, extended energy loss fine structure (EXELFS) analysis, real-space STEM-EELS chemical mapping, and real-space structural imaging via multi-slice electron ptychography (MEP). This comparative analysis leads to several key conclusions for microscopists: 1) no single technique provides a complete picture of SRO; 2) correlating different methods greatly enhances understanding; 3) quantitative analysis is critical; and 4) ptychography is the most informative of the techniques studied here. Reciprocal-space diffuse scattering is a common probe of SRO. With selected area electron diffraction measurements of an MBE-grown Sc0.3Al0.7N film, we observe diffuse scattering around all Bragg peaks (Figure 1a). This result is suggestive of Sc clustering; however, the diffuse scattering could also arise from alternative forms of clustering, e.g. N vacancy clustering or polar nanodomains. To aid interpretation, we studied Sc clustering in ScAlN with DFT and Monte Carlo (MC) simulations. Then, using the MC-derived ScAlN supercells, we performed multi-slice electron diffraction simulations. The simulations reproduce the experimental data, directly linking the measured diffuse scattering to Sc clustering. The diffuse scattering appears as a ‘butterfly’ pattern, which is related to the cluster shape anisotropy. Extended energy loss fine structure (EXELFS) – which is analogous to extended X-ray absorption fine structure (EXAFS) – is sensitive to the local chemical environment of the excited element. We perform Al K-edge EXELFS measurements, and find that Al is preferentially bonded with Al (as opposed to Sc) (Figure 1b,c). This result implies that the system is clustered. From our EXELFS fitting, we are able to extract the average local Sc environment, which is 56±12% Sc (note that the global Sc concentration is 30% Sc). In other words, the Sc clusters are nominally ∼50% Sc. The reliability of the EXELFS fit is largely dependent upon baseline measurements of AlN, which we use to determine the appropriate model hyper-parameters. Real-space STEM-EELS mapping of Sc provides a more direct measure of SRO. However, visual inspection of concentration maps in disordered systems is generally unreliable. Measurement of a statistically significant volume (∼105 atoms [3]) and statistical quantification is necessary (Figure 1d). We calculate the Sc auto-correlation, ρ(R), and we find a positive ρ out to 3 nm (see inset), indicative of Sc clusters with lengths up to ∼3 nm. A drawback of this method is that through-thickness averaging reduces the apparent chemical fluctuations. Real-space structural imaging with MEP offers yet another method to probe SRO (Figure 2) [4]. Since the reconstructed phase is linear with the atomic number, the intensity of metal columns can be used to determine the local composition (Figure 2b). Because MEP provides a depth resolution of several nm, the aforementioned through-thickness averaging is greatly reduced, and MEP yields local chemical fluctuations that are (in some instances) more accurate than EELS (or EDS) chemical mapping. A further benefit of MEP is that the local composition can be correlated in real-space with other structural features, e.g. strain and polar displacements (Figure 2c,d). This work showcases various (S)TEM methods for SRO determination. For comprehensive and reliable SRO quantification, we advocate the use of as many complementary experimental probes as possible. The system studied here – Sc-clustered ScAlN – shows potential for SRO-engineering to tune the dielectric response [5]. a) Selected area electron diffraction of an MBE-grown Sc0.3Al0.7N film, with ‘butterfly’ diffuse scattering indicative of anisotropic clustering. b) EELS of the Al K-edge. c) Real-space representation of the Al K-edge EXELFS data, with the fit indicative of local clusters with ∼50% Sc. d) Atomic-resolution STEM-EELS map of the Sc concentration. The inset shows the Sc auto-correlation, showing Sc clusters on the ∼2-3 nm length scale. a) MEP phase image of the MBE-grown Sc0.3Al0.7N film. The inset shows a zoomed-in view of the metal columns (brighter) and the nitrogen columns (darker). b) Local Sc concentration, extracted based on the relative intensity of the metal columns from the MEP phase image. c) The local polar distortion, quantified by the internal parameter u. d) In-plane strain from the MEP phase image.
No takes yet. Share an insight, caveat, or question.
Hart et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: