Back to overview

Lecture

WEB Pushing TEM data analysis beyond the limits of proprietary software



In recent years, TEM has become an increasingly versatile and sophisticated instrument due to advancements in in-situ experiment capabilities, novel STEM imaging and spectroscopy modalities, and fast detectors. The multidimensional datasets that are produced are rich in quantitative information about the sample, but it remains a challenge to extract, clean and interpret this data. The multitude of proprietary data formats developed by instrument and detector manufacturers hinder easy access to the raw data. This confines most ordinary users to a limited number of black-box analysis routines available in the proprietary software of the manufacturer. In the spirit of FAIR principles, our lab is developing a number of STEM data analysis workflows centered around open source software which enable regular users to take their analysis beyond the limits of proprietary software. In this work we will discuss two recent achievements. Firstly, we have developed a workflow to collect precession electron diffraction (PED) datasets using a TVIPS CMOS camera, significantly improving the data quality compared to regular ASTAR datasets where a phosphorescent screen is filmed to collect the diffraction patterns. We have built a GUI program that converts this dataset to the .blo format which can then be analyzed with Nanomegas software. A comparison of ASTAR and TVIPS datasets from a nanocrystalline film will be shown. We also discuss our work in progress on augmenting the indexation in order to analyze complex datasets with overlapping phases. Secondly, we have implemented a workflow to correct scan noise from STEM datasets based on non-rigid registration [1]. This strongly improves the quality of our images and also shows potential for improving spectral maps. Various examples will be shown, including from high entropy alloys and atomically resolved grain boundaries. Finally, we will discuss our strategies for moving away from proprietary data formats entirely.

Speaker:
Dr. Niels Cautaerts
Max-Planck-Institut für Eisenforschung GmbH
Additional Authors:
  • Dr. Jiwon Jeong
    Max-Planck-Institut für Eisenforschung GmbH
  • Dr. Wenqi Guo
    Max-Planck-Institut für Eisenforschung GmbH
  • Prof. Dr. Benjamin Berkels
    University of Aachen
  • Dr. Christian Liebscher
    Max-Planck-Institut für Eisenforschung GmbH