## Institut für Physik

### Refine

#### Year of publication

#### Document Type

- Doctoral Thesis (161)
- Article (98)
- Conference Proceeding (17)

#### Has Fulltext

- yes (276)

#### Is part of the Bibliography

- no (276)

#### Keywords

- - (78)
- Plasma (25)
- Plasmaphysik (25)
- Plasmadiagnostik (14)
- Stellarator (13)
- Komplexes Plasma (7)
- Polyelektrolyt (7)
- Wendelstein 7-X (7)
- Cluster (6)
- Kernfusion (6)

#### Institute

- Institut für Physik (276)

#### Publisher

- IOP Publishing (63)
- MDPI (13)
- Copernicus (7)
- AIP Publishing (6)
- Frontiers Media S.A. (4)
- Wiley (4)
- American Physical Society (APS) (3)
- Springer Nature (3)
- Cambridge University Press (2)
- European Geosciences Union (2)

Physics-regularized Machine Learning To Approximate 3D Ideal-MHD Equilibria At Wendelstein 7-X
(2024)

The magnetohydrodynamic (MHD) equilibrium model is one of the fundamental building blocks in the description of a magnetically confined plasma. The computational cost of constructing solutions to the 3D ideal-MHD equilibrium problem is one of the limiting factors in stellarator research and design; in particular, it limits the extent to which we can perform sample-intensive applications, applications which require many samples to be evaluated to yield meaningful results. Sample-intensive applications in stellarator research and design include, for example, equilibrium reconstruction, stellarator optimization, and flight simulators. In this thesis, we investigate how faithfully artificial neural networks (NNs) can quickly approximate ideal-MHD equilibria in stellarator geometries, starting with Wendelstein 7-X (W7-X), the world’s most advanced stellarator. In particular, we investigate (see section 1.7):
RQI: to what extent can NN models approximate the MHD equilibrium solution for different W7-X configurations and plasma profiles? What
is the speed-accuracy trade-off offered by NN models?
RQII: to what degree the NN model faithfully reproduces equilibrium quantities of interest (e. g., MHD stability)? To what extent can NN models meet the requirements of downstream applications (e. g., Bayesian
inference, stellarator optimization) in terms of equilibrium quantities
accuracy?
RQIII: whether we can exploit the implicit representation of a MHD equilibrium, i. e., the equilibrium solution should satisfy the ideal-MHD force
balance equation, to improve the NN approximation’s accuracy;
RQIV: the reconstruction of the full posterior istribution of plasma parameters and equilibrium quantities with self-consistent MHD equilibria; moreover, how does the adoption of MHD equilibria approximated by NN models affect the inferred plasma parameters?
A deep NN model is developed to learn the ideal-MHD solution operator in W7-X operational subspace, yielding 3D equilibria up to six orders of magnitude faster than currently available MHD equilibrium codes. Physics domain knowledge is embeded into the NN model: equilibrium solution symmetries are satisfied by construction, and the MHD force balance regularizes the NN model to satisfy the ideal-MHD equations. The model accurately predicts the equilibrium solution and it faithfully reproduces global equilibrium quantities and proxy functions used in stellarator optimization. Finally, the developed fast NN equilibrium model has been applied in downstream applications to obtain W7-X configurations with improved fast-particle confinement and to infer plasma parameters with self-consistent MHD equilibria at W7-X.

Copper tungsten oxide films are deposited with the help of reactive high power impulse magnetron sputtering (HiPIMS) in an argon/oxygen gas mixture. Two magnetrons, one equipped with a tungsten target and the other with a copper target, are employed. The HiPIMS discharge is operated with a repetition frequency of f=100 Hz. Pulse widths of 100 and 20 µs separated by 25 µs are chosen for the tungsten and copper target, respectively. Films deposited on two different glass substrates [soda lime glass and fluorine doped tin oxide (FTO) coated glass] are characterized by energy dispersive x-ray spectroscopy, x-ray photoelectron spectroscopy, x-ray diffraction, Raman spectroscopy, and ellipsometry. Photoelectrochemical activity was investigated by linear voltammetry. The composition and crystal structure of as-deposited and annealed films are found to depend on the deposition conditions. Annealed films deposited on FTO glass are composed of WO3 and CuWO4 or Cu2WO4 crystal phases. Films deposited on soda lime glass are subject to sodium diffusion into the films during annealing and the formation of Na2W2O7 and Na2W4O13 phases.

The multi-cell Penning–Malmberg trap concept has been proposed as a way to accumulate and confine unprecedented numbers of antiparticles, an attractive but challenging goal. We report on the commissioning and first results (using electron plasmas) of the World's second prototype of such a trap, which builds and improves on the findings of its predecessor. Reliable alignment of both ‘master’ and ‘storage’ cells with the axial magnetic field has enabled confinement of plasmas, without use of the ‘rotating wall’ (RW) compression technique, for over an hour in the master cell and tens of seconds in the storage cells. In the master cell, attachment to background neutrals is found to be the main source of charge loss, with an overall charge-confinement time of 8.6 h. Transfer to on-axis and off-axis storage cells has been demonstrated, with an off-axis transfer rate of 50% of the initial particles, and confinement times in the storage cells in the tens of seconds (again, without RW compression). This, in turn, has enabled the first simultaneous plasma confinement in two off-axis cells, a milestone for the multi-cell trap concept.

The controlled formation and adjustment of size and density of magnetic skyrmions in Ta/CoFeB/MgO trilayers with low Dzyaloshinskii–Moriya interaction is demonstrated. Close to the out-of-plane to in-plane magnetic spin reorientation transition, we find that small energy contributions enable skyrmion formation in a narrow window of 20 pm in CoFeB thickness. Zero-field stable skyrmions are established with proper magnetic field initialization within a 10 pm CoFeB thickness range. Using magneto-optical imaging with quantitative image processing, variations in skyrmion distribution and diameter are analyzed quantitatively, the latter for sizes well below the optical resolution limit. We demonstrate the controlled merging of individual skyrmions. The overall demonstrated degree of comprehension of skyrmion control aids to the development of envisioned skyrmion based magnetic memory devices.

Recent experimental campaigns in the Wendelstein 7-X stellarator, a
plasma-confining device designed to investigate the Magnetic Confinement Fusion
(MCF) approach to generating electrical power, have shown that the injection of
fuelling pellets had an unexpected and considerable impact on the performance of
the plasma. Rather than simply refuelling the device and `diluting' the plasma
energy, pellet injection is followed by a significant increase in the ratio of
the ion temperature to the electron temperature. It has been suggested that this
is not merely due to the improved confinement following the reduction of
turbulent transport after the pellet material has homogenised with the bulk
plasma, but also due to a direct transfer of energy from electrons to ions. The
proposed mechanism for this energy transfer is the ambipolar expansion of the
pellet plasmoid, the localised plasma structure produced by the
ionisation of ablated pellet material, along magnetic field lines.
Early work on pellet plasmoid expansion predicted that half the heating power
deposited in plasmoid electrons by collisions with hot ambient electrons is
transferred to plasmoid ions in the form of flow velocity as the plasmoid
expands. The complicated nature of the system of the pellet plasmoid embedded in
the ambient plasma, particularly the behaviour of electrons, which experience
many collisional and collisionless phenomena on multiple disparate timescales,
means that early models of the expansion were not wholly self-consistent, but
rather made use of strong approximations that apply in some regions of the
plasmoid but not in others. For example, only electrons and ions associated with
the plasmoid were rigorously treated, meaning that the framework was one of
`expansion into vacuum'. Combined with the assumption of Maxwellian electrons,
this led to an electric potential that was unbounded at infinity. Naturally, the
validity of the conclusions of such a model are called into question because the
approximations lose their validity far from the plasmoid and as time advances,
yet predictions about the final state of the plasma are desired. A deeper
investigation is required: careful consideration of the phenomena in question
and the timescales (and lengthscales) on which they act must be made in order to
rigorously construct a model that is valid throughout the entire expansion.
The first two papers presented in this thesis iterate on the model established
in the paper that first predicted the electron-to-ion energy transfer; their aim
was to find out how the character of the expansion changes with a more
sophisticated and accurate description of various phenomena, while remaining
within the existing framework of expansion into vacuum. Ultimately, we find that
the qualitative character is unchanged, and that approximately half the heating
power deposited in plasmoid electrons is transferred to ions.
Two other papers in this thesis address the limitations of the original model.
This is achieved by properly considering the electron kinetic problem in a
plasmoid. One paper considers the electron kinetic problem when electrons are
highly isotropised. In this case the kinetic equation can be integrated to
remove all but two independent variables, which is the maximum possible
reduction considering it is a time-dependent problem. The full nonlinear
integro-differential Landau self-collision operator is integrated exactly and
few approximations are made, leading to a rather general kinetic equation.
However, for fuelling pellets some anisotropy in the electron distribution is
expected. Another paper considers the electron kinetic problem (and the entire
plasmoid expansion) allowing for electron anisotropy. Careful consideration of
the ordering of timescales of electron phenomena in a pellet plasmoid leads to a
steady-state kinetic problem that we call collisional quasi-equilibrium (QE). QE
appears in many ways similar to the collisional steady-state characterising a
true thermal equilibrium. It was found that the time-dependent kinetic problem
of the earlier paper, with isotropic electrons, produces the QE distribution
function, corroborating the existence of the QE state. We then take moments of
the electron kinetic equation that is valid on the expansion timescale, assuming
that the electron distribution is that given as the solution to the QE kinetic
problem. This is completely analogous to what is done to obtain the Braginskii
equations or any Chapman-Enskog theory. The result is a set of equations for the
long-term evolution of the macroscopic quantities that describe the distribution
function existing in a quasi-steady-state at each point in time. It is from this
point that one may feasibly describe the plasmoid expansion with an accurate
picture of the electron kinetics and finally obtain the electron-to-ion energy
transfer so desired in a rigorous model of the expansion.
From a broader point of view, the two frameworks provided by these rigorous
investigations of the electron kinetic problem serve as a basis for the future
study of plasmoids. Such a `first-principles' approach to plasmoid dynamics is
novel and interesting in its own right, but it will be demonstrated that such an
approach is essential for pellet plasmoids owing to the fact that they are
poorly described by the `standard tools' of plasma physics.
Using the QE framework it was found that, once more, about half the heating
power experienced by plasmoid electrons is transferred to plasmoid ions. The
incredible robustness of the prediction of such an energy transfer is, in the
author's opinion, the result of the self-similar nature of the expansion found
as a solution to the original model. As a rule, the profiles of self-similar
solutions tend to be attractors for the `real', more complicated, system, and
the qualitative predictions involving no parameters, of which the
electron-to-ion energy transfer is one, tend to be very sturdy.
Aside from fuelling pellets, composed of hydrogen or deuterium, one paper in
this thesis investigates the physics of high-Z pellets that are designed to
terminate the plasma safely in the event of a `disruption', where much of the
magnetic field energy is channelled into a runaway electron beam with
potentially disastrous consequences if the beam encounters a plasma-facing
component. The paper draws on the work carried out in the paper concerning the
kinetic problem of isotropised electrons in a plasmoid.
This thesis is `cumulative'; the vast majority of the work carried out is
described within a set of Papers, labelled A-E, placed at the back of the text.
There is a preceding `wrapper text' (given in numbered Sections) tasked with
introducing the reader to the topic, guiding the reader through the papers, and
expounding some of their main results. Some amount of material not present in
the papers is also provided in the wrapper text. Naturally, the wrapper text
mainly focusses on the results of the papers which are under my first
authorship. In the course of publishing papers over an extended period of time
the nomenclature is bound to vary. Although it is mostly consistent between the
papers, a few difference do arise, and the section `Common symbols and
subscripts' is provided in the frontmatter to alleviate confusion. Particular
care should be taken with the symbols x and z; both can refer to the
coordinate parallel to the magnetic field line, but in papers where z is used
for this purpose x tends to have another definition. In the wrapper text the
choice of symbols is generally chosen to reflect those in the corresponding
paper.

In future fusion reactors disruptions must be avoided at all costs. Disruptions due to the density limit (DL) are typically described by the power-independent Greenwald scaling. Recently, a power dependence of the disruptive DL was predicted by several authors (Zanca et al 2019 Nucl. Fusion 59 126011; Giacomin et al 2022 Phys. Rev. Lett. 128 185003; Singh and Diamond 2022 Plasma Phys. Control. Fusion 64 084004; Stroth et al 2022 Nucl. Fusion 62 076008; Brown and Goldston 2021 Nucl. Mater. Energy 27 101002). It is investigated whether this increases the operational range of the tokamak. Increasing the heating power in the L-mode can induce an L-H transition, and therefore a power-dependent DL and the L-H transition cannot be considered independently. The different models are tested on a data base for separatrix parameters at the separatrix of ASDEX Upgrade and compared with the concept (SepOS) presented in Eich and Manz (2021 Nucl. Fusion 61 086017). The disruptive separatrix density scales with the power ne ∝ P0.38±0.08 in good agreement to all models. Also the back transition from high to low (H-L) confinement shows an approximately Greenwald scaling with an additional power dependence ne ∝ P0.4 according to the SepOS concept. For future devices operating at much higher heating power such a power scaling may allow operation at much higher separatrix densities than are common in H-mode operation. Preconditions to extrapolation for future devices are discussed.

The pulse length dependence of a reactive high power impulse magnetron sputtering (HiPIMS) discharge with a tungsten cathode in an argon+oxygen gas mixture gas was investigated. The HiPIMS discharge is operated with a variable pulse length of 20–500 µs. Discharge current measurements, optical emission spectroscopy of neutral Ar, O, and W lines, and energy-resolved ion mass spectrometry are employed. A pronounced dependence of the discharge current on pulse length is noted while the initial discharge voltage is maintained constant. Energy-resolved mass spectrometry shows that the oxygen-to-tungsten (O+/W+) and the tungsten oxide-to-tungsten (WO+/W+) ion ratio decreases with pulse length due to target cleaning. Simulation results employing the SDTrimSP program show the formation of a non-stoichiometric sub-surface compound layer of oxygen which depends on the impinging ion composition and thus on the pulse length.

Properties of self-excited dust acoustic waves under the influence of active compression of the dust particle system were experimentally studied in the laboratory and under microgravity conditions (parabolic flight). Ground based laboratory experiments clearly show that wave properties can be manipulated by changing the discharge volume, its aspect ratio, and thus the dust particle density. Complementary experiments under microgravity conditions, performed to exclude the effects of gravity inflicted sedimentation and anisotropic behavior, were less conclusive due to residual fluctuations in the planes acceleration indicating the need for a better microgravity environment. A theoretical model, using plasma parameters obtained from particle-in-cell simulations as input, supports the experimental findings. It shows that the waves can be described as a new observation of the dust acoustic mode, which demonstrates their generic character.

Cationic and anionic clusters of the group-14 elements carbon, silicon, germanium, tin, and lead are produced by high-vacuum laser ablation and studied with a multi-reflection time-of-flight mass spectrometer. In-trap photodissociation is performed for cluster species in the size range n=2–10. The clusters’ production rates as well as their dissociation pathways are used to probe the nonmetal–metal transition throughout the group. Carbon clusters show neutral-trimer break-off, while those of the other elements evaporate neutral monomers and, in some cases, form specific charged fragment sizes.

Magnetooptical properties of one-dimensional aperiodic structures formed by stacking together magnetic and nonmagnetic layers according to the Kolakoski self-generation scheme are studied theoretically using the 4x4 transfer matrix method. The effect of the generation stage of the sequence, and the helicity and direction of light propagation through the magneto-photonic crystals on the transmission/reflection spectra as well as Faraday and ellipticity rotations, have been investigated. Our results reveal that this kind of aperiodic magneto-photonic crystals can be used for the fabrication of multifrequency laser cavities, and optical filters/sensors.