Dresden 2026 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
O: Fachverband Oberflächenphysik
O 8: Nanostructures at surfaces:1D, 2D, networks I
O 8.4: Vortrag
Montag, 9. März 2026, 11:30–11:45, WILL/A317
Autonomous Nanoworld: Building Nanostructure Without Human Intervention — •Bernhard Ramsauer1, Qigang Zhong2, Stefan Pranger3, Bettina Könghofer3, and Oliver T. Hofmann1 — 1nstitute of Solid State Physics, NAWI Graz, Graz University of Technology, Graz, 8010, Austria — 2Institute of Functional Nano & Soft Materials, Soochow University, Suzhou, 215006, China — 3Institute of Applied Information Processing and Communications, Graz University of Technology, Graz, 8010, Austria
In this contribution, we present atomically precise molecular nanostructures fabricated without human intervention. We demonstrate that simultaneous control of molecular position and orientation is attainable - both are prerequisites for constructing covalently bonded nanostructures in the future. We employ deep reinforcement learning agents to learn the complex tip-molecule-surface interactions by determining the optimal manipulation parameters for moving and rotating molecules. Because exhaustive sampling of all possible manipulation parameters is impossible, we generate a simulation based on selected experiments. This simulation allows us to train robust agents that even generalize to unseen adsorption conditions and arbitrary nanostructures. With this strategy, we autonomously assemble molecular building blocks into arbitrary geometries, establishing a reliable platform for automated molecular assembly.
Keywords: Nanostructures; STM (Scanning Tunneling Microscopy); AI (Artificial intelligence); DRL (Deep reinforcement learning); Neural network