Improving 4D-STEM Phase Mapping with Correlated EDX Signals

Structural Ambiguity in Similar Phases

In many advanced materials, phases may differ only subtly in their lattice structure but significantly in their properties and function. In such cases, electron diffraction alone may not resolve these differences, leading to high sensitivity to the noise in acquired data and inaccurate phase analysis.

This presents a challenge for researchers working on alloys, thin films, or ceramics where multiple phases coexist with near-identical crystallographic structures. An incorrect phase map may affect downstream analysis, including process control or functional predictions.

A Unified Multimodal Workflow

The study used a polycrystalline aluminum film doped with gold nanoparticles, measured on the Tescan TENSOR system at 100 kV. Two data sets were collected under identical conditions:

• Beam convergence: 1.5 mrad
• Beam current: 1.0 nA
• Precession:  0.8°
• Dwell time: 10 ms

In one dataset, phase mapping was performed using 4D-STEM diffraction data only. In the other, the algorithm incorporated simultaneously acquired EDX signals to assist template matching of the 4D-STEM data. The Explore^TM software interface enabled straightforward inclusion of the correlated EDX data in the 4D-STEM data acquisition and the phase map calculation.

Improving Phase Differentiation with Elemental Composition

The comparison revealed that using EDX data in the mapping process eliminated incorrect phase assignments. Regions of pure aluminum were no longer labeled as gold, and gold nanoparticles were correctly segmented.

This correction is not just visual; it also impacts quantitative analysis. Without EDX, phase maps can fragment continuous grains and skew statistical metrics. The correlative 4D‑STEM/EDX method eliminates the influence of different orientations in different grains, supports accurate phase analysis, and reflects the true material composition.

The approach scales to materials with three or more crystalline phases, providing reliable separation when neither EDX nor diffraction alone delivers robust and accurate results.