Clustering characteristic diffraction vectors in 4-D STEM data sets from overlapping structures in nanocrystalline...

Area of Science:

• Materials Science
• Electron Microscopy
• Crystallography

Background:

• Analyzing diffraction patterns in electron microscopy is crucial for material characterization.
• Overlapping structures in projection can complicate the interpretation of diffraction data.
• Distinguishing true symmetries from artifacts caused by overlapping structures is challenging.

Purpose of the Study:

• To develop a robust method for identifying and clustering diffraction vectors in 4-D scanning transmission electron microscopy (STEM) data.
• To accurately determine characteristic diffraction patterns from overlapping structures in projection.
• To differentiate between single, rotationally symmetric structures and apparent symmetries arising from overlapping structures.

Main Methods:

• Convolution of 4-D STEM data with a 4-D kernel.
• Identification and clustering of diffraction vectors using density-based clustering.
• Application of a metric emphasizing rotational symmetries for vector clustering.

Main Results:

• The method successfully identifies and clusters diffraction vectors in both crystalline and amorphous samples under various experimental conditions (high/low dose).
• Performance metrics were established using simulated data of overlapping aluminum nanocrystals, showing robustness against Poisson noise.
• Experimental data from aluminum nanocrystals and amorphous Pd77.5Cu6Si16.5 thin films confirmed the method's efficacy, revealing 4- and 6-fold symmetry structures.

Conclusions:

• The developed method effectively resolves diffraction patterns from overlapping structures in 4-D STEM data.
• Quantifying background diffraction from overlapping structures aids in accurate symmetry analysis.
• This technique provides valuable insights into the structural characteristics of complex materials, including glassy structures.