Dresden 2020 – scientific programme

The DPG Spring Meeting in Dresden had to be cancelled! Read more ...

Parts | Days | Selection | Search | Updates | Downloads | Help

DS: Fachverband Dünne Schichten

DS 40: Thin Film Properties: Structure, Morphology and Composition I

DS 40.2: Talk

Thursday, March 19, 2020, 15:15–15:30, CHE 89

Evaluation of atomically-resolved high-resolution TEM images of Di-and Tri- Re molecules @ SWNT with convolutional neural networks — •Christopher Leist, Kecheng Cao, and Ute Kaiser — Central Facility of Electron Microscopy Materials Science, Ulm University, 89081 Ulm, Germany

Single-walled carbon nanotubes (SWNT) containing transition metal Di- and Tri- Re molecules are investigated using atomic resolution transmission electron microscopy (TEM).The images are taken at 80kV with the Cc/Cs-corrected SALVE instrument operated at 80kV, where the nanotube can be stable and sub-Angstrom resolution can be achieved. Detailed analysis of the Re atom distances is not only time consuming, also many interesting image features are close to the scale of the pixel error, and the manual evaluations are prone to user bias. Here we use deep learning routines which have the potential of both speeding up the evaluation process considerable as well as reducing the accompanied bias. The neural networks are trained on simulated TEM images. Here we present our progress in training the neural network and thereby automating the image investigation.