Beschrijving
This thesis investigates the feasibility of developing a non-invasive method to infer thetwo-dimensional electron density distribution in helium glow discharge plasmas within the
TOroidally MAgnetised System (TOMAS) using optical emission from visible-spectrum
cameras. Building on prior research that observed a local linear correlation between light
intensity and Langmuir probe – measured electron density without theoretical backing or
2D extension, this work establishes a physical foundation through analysis of photon emission coefficients (PECs) and neutral helium dynamics in the corona equilibrium regime. In
TOMAS’ parameter space (electron densities 1014−1017 m−3
, temperatures 1−10 eV), He-I
PECs for lines such as 668 nm and 728 nm are effectively density-independent, supporting the emission proportionality to electron density under stable temperatures. However,
this holds only in low-density conditions where recombination, ionisation, and metastable
effects are minimal; unmodelled metastables may cause emission enhancements, leading
to potential density underestimations of 40 − 50%.
A workflow was developed integrating high-resolution imaging, Calcam geometric calibration, and Tikhonov-regularised SART tomography. Single-view reconstructions proved
ill-posed, yielding artefacts like edge brightening and ray-streaks due to underconstrained
matrices and null-space issues. Two opposing views reduced these, producing reliable
2D emissivity maps for 668 nm and 728 nm lines (502 nm discarded due to an order-ofmagnitude discrepancy). Direct density calculations and Bayesian line analysis (668 and
728 nm) generated 2D electron density inference maps. The profiles of the estimate and
Langmuir probe reference correlated strongly (r ≈ 0.87 for 668 nm), featuring central
plateaus and edge drops, but the absolute values were underestimated by 45-50%, with
Bayesian credible intervals up to 15% of the mean of the posterior distribution owing to
prior degeneracies, inversion uncertainties, and neglected processes like metastable states
and reflections.
Addressing the research questions critically: Optical measurements enable a 2D density
inference under constant temperature conditions, but are methodologically challenging
because of limited views. The incompleteness of the theoretical model compromises precision, making the inference qualitative rather than quantitative without probe calibration.
This technique innovates TOMAS with its first grounded camera-based 2D estimates for
the electron density, foundational for wall conditioning studies. Limitations like spatial
coverage at low currents (<3 A) and emission discrepancies highlight the need for multiview expansions, metastable-inclusive models, and reflection corrections to enhance the
reliability in wall conditioning-relevant plasmas
Bijkomende beschrijving
Miquel Hillen is the master student| Periode | 1 jan. 2025 → 28 jan. 2026 |
|---|---|
| Examen gehouden op |
|
| Mate van erkenning | Internationaal |