Measurements and modelling of impurity flows in the TJ-II stellarator

This thesis is concerned with the understanding of plasma velocity fields in three-dimensional toroidal magnetic confinement plasma devices such as the TJ-II stellarator {the magnetic confinement device subject of this thesis. Flow in these magnetic configurations is predominantly tangential to a set of nested toroidal surfaces known as magnetic or flux surfaces. The way in which the main plasma species (generally hydrogen or deuterium in fusion-oriented experiments) moves along these flux surfaces can have important effects on the confinement and stability properties of these devices. As an example, the improved confinement regime, known as H-mode, is associated with a strong variation of the ExB flow component across neighbouring flux surfaces. The main experimental technique used in this work is Charge eXchange Recombination Spectroscopy (CXRS) of fully ionized carbon, C⁶+. Velocity and temperature measurements are obtained from Doppler shift and broadening of a C⁵+ spectral emission line, respectively, after C⁶+ ions undergo a charge exchange reaction with a neutral atom. The sightline arrangement of the TJ-II CXRS system, together with a compact neutral beam injector to stimulate C⁵+ line emission, allows local perpendicular and parallel velocity components to be obtained at different positions within a set of magnetic surfaces. CXRS velocity measurements are compared with theoretical expectations of flow spatial variation and of flow component size in strongly magnetized plasmas. The flow component perpendicular to the magnetic field direction consist of ExB and diamagnetic flows. In toroidal systems such flows are compressible, i.e. they have a non-zero divergence. To conserve the number of particles, the parallel ow component is constrained to vary so that the total (perpendicular plus parallel) flow is incompressible. In this framework, the full incompressible velocity field within a flux surface is determined by two constant factors that multiply different flow components. These constants are directly related to the radial electric field, Eᵣ, and the so-called bootstrap factor, both of which are obtained from standard neoclassical theory in non-axisymmetric configurations. The understanding of the mechanisms that determine these quantities is important. For instance, the radial electric field profile has profound consequences on the confinement of particles, as mentioned before. On the other hand, the differences between the bootstrap components of ions and electrons give rise to the bootstrap current, which can modify some properties of the magnetic structure and affect the stability of the plasma. The results of this comparison are twofold. In low pressure-gradient plasmas of TJ-II good agreement is obtained, both in the incompressible variation of the flow (which CXRS allows to verify) and between the measured and computed values of Eᵣ and bootstrap factor at several magnetic surfaces. A careful treatment of the complicated magnetic geometry has been fundamental to obtain this agreement, that provides confidence in the capability of the neoclassical theory to determine such factors. In medium pressure-gradient plasmas, parallel flow measurements in TJ-II show notable and systematic deviations from an incompressible flow pattern. To interpret this observation it should be recalled that (1) incompressibility of the flow field of a given species relies on its number density being approximately constant on magnetic surfaces and that (2) CXRS measures the ow of C⁶+ ions. Because of their high charge number, carbon impurities can be in a collisional regime that is prone to develop density inhomogeneities on fl ux surfaces. The resulting impurity parallel pressure gradient drives a return flow that departs from incompressibility. The calculation of the impurity density inhomogeneity and parallel return flow requires solving the particle continuity equation and parallel momentum balance consistently. This is done numerically with the aid of magnetic libraries. The results of these calculations for the higher pressure gradient plasma cast an impurity return flow of a size comparable to the measured incompressible flow deviations. However, at the location of the measurements the model generally predicts a small deviation with opposite sign to the observations. This remains true when additional terms like parallel electric fields or inertia are included in the model. To conclude it should be noted that an explanation for such fl ow deviations is of considerable importance as it can modify the ion-impurity friction driven radial impurity transportThis work was partially funded by the FPI grant awarded by CIEMAT (BOE resolution no 171, 24/06/2008) and the Spanish Ministry of Science and Innovation under contract ENE2010- 19676