A 4D-STEM Tomographic Framework Assisted by Object Tracking for Nanoparticle Structure Determination

Three-dimensional electron diffraction (3D ED) has emerged as a powerful method for solving the structure of sub-micron-sized particles down to nanoparticles. However, it faces technical challenges when applied to beam-sensitive samples or agglomerated nanoparticles. This study presents a novel approach that combines 4D-STEM tomography with object tracking and segmentation algorithms to overcome these limitations, and achieve single-crystalline 3D ED datasets from nanopowder samples. The method and data quality are assessed on brookite TiO₂ nanorods and beam-sensitive CsPbBr₃ nanoparticles. To finely sample the reciprocal space, the data acquisition was automated to acquire hundreds of 4D-STEM scans at fine tilt steps using a slightly convergent beam (0.6 to 1 mrad). Compared to conventional 3D ED methodologies, the proposed method provides enhanced signal-to-noise ratio, low illumination time for reducing beam damage, and the ability to analyze multiple particles from a single tomographic dataset. The procedure is optimized to be feasible using commercially available desktops and detectors. This extends the method applicability to systems and samples that were very challenging for conventional 3D ED methods, by eliminating several technical challenges for the data acquisition.