SMuK 2023 – wissenschaftliches Programm

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

P: Fachverband Plasmaphysik

P 11: Poster I

P 11.14: Poster

Mittwoch, 22. März 2023, 14:00–15:30, HSZ EG

High-efficiency machine learning approach for nanoparticle 2D size characterization via kinetic Mie polarimetry — •Alexander Schmitz, Andreas Petersen, and Franko Greiner — IEAP, Kiel University, 24118 Kiel, Germany

In a nanodusty plasma, the determination of the size of the nanoparticles is crucial to their diagnostics. In the Mie regime, in situ polarization measurements of light scattered by the particles (polarimetry), have proven to be an effective, non-invasive technique.

This method holds a number of challenges. The polarization state depends not only on the particle size, but also on its complex refractive index. Furthermore, the inverse mapping from the measured polarization state to the time dependent particle size and refractive index in a reactive plasma exhibits a strongly non-linear relationship. To resolve this, a customized kinetic fitting algorithm has been introduced in the past [1]. However, that method, based on Least-Square Fits, is highly sensitive to the time series length and requires considerable computing time.

We present a new deep-learning approach to the mapping problem via our High-Efficiency Refractive index MappIng NEural network (HERMINE). With this, the error rate of automated data evaluation, as well as computing time was significantly reduced. This paves the path for future data-intensive, real-time imaging of the particle's growth dynamics in nanodusty plasmas [2].

[1] S Groth et al, J. Phys. D: Appl. Phys., 2015.

[2] S Groth et al, Plasma Sources Sci Technol, 28 (11), 2019.