Precession electron diffraction dataset from nanocrystaline Cu-Ag (FCC) alloy collected on pixelated TVIPS detector

Summary

This is a 4D scanning transmission electron microscopy (4D STEM) dataset collected in near-parallel beam + precession mode (PED) of a nanocrystaline Cu-Ag sample, collected on a high quality pixelated detector inside a transmission electron microscope (TEM). The dataset is represented by a 4D array, comprising a 2D grid of scan points, with each scan point mapping to an electron diffraction spot pattern. From this kind of dataset it is possible to derive local crystal orientations and strains. The dataset is in the .hspy format, the native hdf5 format of the HyperSpy library.

Material and sample preparation

The sample was prepared from a nanocrystalline Cu-Ag thin film. The details are described in the following publication:

Oellers, Tobias, et al. "Thin-Film Microtensile-Test Structures for High-Throughput Characterization of Mechanical Properties." ACS combinatorial science 22.3 (2020): 142-149.

The sample was prepared by punching a 3 mm diameter disc out, gluing this to a Cu single hole grid, and subsequent Ar+ ion milling until perforation at 2.5 kV using a PIPS II system (Gatan). The rough milling was followed with a 0.5 kV cleaning.

Microscopy parameters and data collection

PED was performed in a JEM-2200FS TEM (JEOL) operating at 200 kV. The microscope was operated in nanobeam diffraction mode with the smallest spot size (Spot 5) and a condenser aperture size of 10 μm. The probe diameter was ~ 1 nm with a convergence angle of ~2 mrad. A precession frequency of 100 Hz and a precession angle of 0.5° were applied during the nanobeam scanning. Data was collected on a TemCam-XF416 pixelated CMOS
detector (TVIPS). The camera length as indicated in the operating software was 15 cm, and collected images were 2k by 2k in size (hardware binning of 2). The dataset comprises 150x150 scan points and pixel depth is 2 bytes (unsigned 16 bit integers).

Data processing

The raw data was collected in the .tvips format. The original dataset was about 100 GB in size and is no longer available. This dataset was converted to the .hspy format using the TVIPSconverter tool. In the conversion, the images were binned by an additional factor of 4 to a final size of 512x512. A median filter was also applied to the data.

Data characteristics

Scan shape: 150 x 150 pixels

Image shape: 512 x 512 pixels

Pixel dtype: uint16

Scan pixel size: about 1 nm, scan dimensions were never calibrated

Image pixel size: 0.01155 Angstrom-1 / pixel

Note that scale factors are not stored in the dataset! The dataset can be read with HyperSpy using the load function (please see the HyperSpy documentation). It is highly recommended to have a working installation of Pyxem as well to process the data.

Additional notes

Data was collected with the TVIPS scan generator which can be quite buggy. A large number of the scan points are worthless. In addition, the detector background was not properly subtracted in the image, resulting in striped artifacts in the images.