Regensburg 2025 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

O: Fachverband Oberflächenphysik

O 24: Poster Scanning Probe Techniques: Method Development

O 24.7: Poster

Montag, 17. März 2025, 18:00–20:00, P2

Detection and Localization of Atoms and Molecules on Different Surfaces Using Computer Vision — •Lovis Hardeweg, Johannes Schwenk, Wantong Huang, Kwan-Ho Au-Yeung, Máté Stark, Paul Greule, Christoph Sügers, Wolfgang Wernsdorfer, and Philip Willke — Physikalisches Institut (PHI), Karlsruhe Institute of Technology, Karlsruhe, Germany

Scanning Probe Microscopy (SPM) methods are unparalleled in their ability to image and manipulate structures on the atomic scale. In combination with machine learning techniques, this allowed to automate processes such as removing a molecule from a thin layer [1] or moving an adsorbed molecule to a specific position [2]. However, this often relies on prior human interaction to identify and localize objects of interest, like a thin film or a single adsorbate. Here, we discuss methods that automate several steps in SPM experiments, with the goal of advancing single atomic and molecular spin detection experiments. For that, we employ computer vision techniques to STM topography data and are able to extract information, such as the location of single atoms and molecules or the presence of different sample surfaces, for instance ultra-thin MgO films grown on Ag(001). We believe that these abilities, once sufficiently developed, can lead to a significant reduction in the need for human intervention in the automated use of high-resolution low temperature SPM. [1] P. Leinen et al. Sci. Adv., vol. 6, no. 36, p. eabb6987, Sep. 2020, doi: 10.1126/sciadv.abb6987. [2] B. Ramsauer et al. J. Phys. Chem. A, vol. 127, no. 8, pp. 2041-2050, Mar. 2023, doi: 10.1021/acs.jpca.2c08696.

Keywords: scanning probe methods; computer vision; object detection; machine learning