TY - STD
AU - Schmitz, Alexander
PY - 2023
DA - 2023/07/17
TI - HERMiNe: A neural network for kinetic Mie polarimetry - particle size diagnostics in nanodusty plasmas
KW - neural network, machine learning, nano particle, refractive index, nanodusty, complex plasma, reactive plasma, plasma diagnostics, Mie scattering, polarimetry
AB - HERMiNe is a deep neural network for solving the kinetic Mie problem of light, scattered by nanoparticles. The polarization state of the scattered light is usable as in-situ size diagnostics for the size and the complex refractive index of particles. For the kinetic Mie polarimetry to work, it is necessary for the particles to undergo a change in size (eg. via growth processes or etching), while the refractive index is assumed to be sufficiently constant in time. Given a time series input of the ellipsometric angles Psi and Delta, the network predicts the best-fitting refractive index, from which the particle radii can be calculated via Mie theory. A detailed treatise on the theory, network properties and uncertainties can be found in: A neural network approach to kinetic Mie polarimetry for particle size diagnostics in nanodusty plasmas by Schmitz et al (DOI 10.1088/1361-6463/aceb71). The network itself is provided as: MATALB DAGNetwork format. Requires the MATLAB Deep Learning Toolbox (version 2022a or newer) [tested]. ONNX format for use with other, open frameworks, such as Pytorch etc [not tested]. For the MATLAB framework, a script is provided to facilitate operation. It takes the measurement data as input and returns the refractive index as vector elements. The documentation is accessible via the command help predictN When using another framework or custom readout functions, please note, that the network output for the imaginary part of the refractive index is increased by a factor of ten compared to the physical value, due to numerical reasons.
L1 - https://opendata.uni-kiel.de/servlets/MCRZipServlet/fdr_derivate_00000011
UR - https://opendata.uni-kiel.de/receive/fdr_mods_00000010
UR - https://doi.org/10.57892/100-10
DO - 10.57892/100-10
LA - en
ID - fdr_mods_00000010
ER -