AutoDisk Demostration

AutoDisk is an automatic convergent beam electron diffraction (CBED) pattern analysis method to detect diffraction disks in each CBED pattern from a four-dimensional scanning transmission electron microscopy (4D-STEM) dataset, refining disk positions and extracting lattice parameters if a 2D crystal lattice is found. This method can perform lattice parameter estimation and strain mapping of a nanometer to mirometer sampling area, which can be the basis for high throughput phase, symmetry, orientation, and crystallographic analysis in general.

In this demostration, we would show an example of AutoDisk being used to map strain of a core-shell Pd@Pt nanoparticle. The Jupyter notebook "AutoDisk_Demo.ipynb" goes through each step illustrating how the 4D-STEM data is read in and pre-processed, how the disks are detected, how lattice parameters are estimated, and finally how strain maps are generated. To run the code, please download the utilities file "autodisk.py" and the demostration file "AutoDisk_Demo.ipynb". Also put the data file "pdpt_x64_y64.raw", a 4D-STEM dataset with 64 * 64 probe positions of 128 * 128 pixel CBED patterns from a square scanning area, in the same folder, and run the notebook. You are welcome to tune the parameters and try the method on your own data.

In the current version, the input file needs to be in the .raw format generated from EMPAD. We are going to implement more compatible data formats in the later version, and add more functions such as pattern clustering and strain mapping of amorphus area, and also accelerate the analysis with code optimazation.

For details of the method, please refer to the paper "AutoDisk: Automated Diffraction Processing and Strain Mapping in 4D-STEM" by Sihan Wang, Tim Eldred, Jacob Smith and Wenpei Gao (https://arxiv.org/abs/2201.00297).