SurfaceScience21 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
O: Fachverband Oberflächenphysik
O 55: Poster Session IV: Poster to Mini-Symposium: Machine learning applications in surface science II
O 55.6: Poster
Dienstag, 2. März 2021, 13:30–15:30, P
Inverse problem to AFM imaging with iterative correction loop — •Prokop Hapala1, Lauri Kurki2, Niko Oinonen2, Fedor Urtev2, Filippo Federici Canova2, Juho Kannala2, and Adam S. Foster2 — 1Department of Condensed Matter Theory, FZÚ AV ČR, v.v.i. — 2Department of Applied Physics, Aalto University Espoo, Finland
In the last year we pioneered machine-learning methods for reconstruction of molecular structure from high-resolution AFM images of non-planar organic molecules [1], which opens the way to broader application of this experimental technique for single-molecule analysis [2] e.g. in the pharmaceutical industry. Nevertheless, the robustness of one-shot scheme relying on general-purpose convolutional neural networks (CNN) seems limited as it discards physical insight. We attempt to improve our method by integrating the CNN module together with an image simulation module and interatomic force-field into an iterative feedback loop, which gradually improves the match between reference and simulated image. Such a scheme, with a machine-learned model providing educated trial-move within a global optimization algorithm, can be possibly useful also for solving other difficult inverse problems. [1] Alldritt B., et al., Sci. Adv., vol. 6, no. 9, p. Eaay6913. (2020) [2] Schuler, B., et.al. JACS, 137(31), 9870-9876. (2015)