Erlangen 2026 – wissenschaftliches Programm

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

P: Fachverband Plasmaphysik

P 4: Codes and Modeling I

P 4.1: Hauptvortrag

Montag, 16. März 2026, 16:15–16:45, KH 01.020

Data-integrated simulations and machine learning analysis of plasma processing of SiOx/Cu memristive devices — Tobias Gergs, Rouven Lamprecht, Ole Gronenberg, Sahitya Yarragolla, Hermann Kohlstedt, and •Jan Trieschmann — Kiel University, Kaiserstraße 2, 24143 Kiel, Germany

The characteristics of SiOx/Cu memristive devices [1] deposited by reactive magnetron sputtering are highly sensitive to the obtained material properties, requiring fine control over local physical conditions during plasma deposition. The latter are investigated through machine learning (ML) surrogate modeling, data-integrated physical simulations, and a data-driven analysis of corresponding wafer-level measurements. An ML surrogate model of the reactive plasma-surface interaction during Ar and O2 ion impingement on SiOx is integrated in an axially symmetric 2D particle-in-cell/Monte Carlo collision simulation with dynamic surface conditions. It provides a comprehensive prediction of discharge and surface conditions, e.g., fluxes, energy. Insights from this simulation are correlated with a data-driven classification of measured device-level electrical characteristics. A statistical analysis over a wafer is applied to over 50,000 devices to identify how processing conditions influence device behavior. The analysis reveals distinct device types linked to the local physical conditions during processing, highlighting the importance of plasma process control in determining functional outcomes in nanoscale electronic devices.

[1] Lamprecht et al., Adv. Eng. Mater. 27, 2401824 (2025)

German Research Foundation – Project-ID 434434223 (SFB 1461)

Keywords: Plasma processing; Machine learning; Modeling and simulation