Abstract. Runaway electrons represent a serious problem for the reliable operation of the future experimental tokamak ITER. Due to the multiplication factor of exp(50) in the avalanche even a few seed runaway electrons will result in a beam of high energetic electrons that is able to damage the machine. Thus suppression of runaway electrons is a task of high importance, for which reason we present here a systematic study of runaway electrons following massive gas injection in TEXTOR. Argon injection can cause generation of runaways carrying up to 30% of the initial plasma current, while disruptions triggered by injection of helium or of mixtures of argon (5, 10, 20%) with deuterium are runaway free. Disruptions caused by argon injection finally become runaway free for very large amounts of injected atoms. The appearance/absence of runaway electrons is related to the fraction of atoms delivered to the plasma center. This so called mixing efficiency is deduced from a 0D model of the current quench. The estimated mixing efficiency is: 3% for argon, 15% for an argon/deuterium mixture and about 40% for helium. A low mixing efficiency of high-Z impurities can have a strong implication for the design of the disruption mitigation system for ITER. However, a quantitative prediction requires a better understanding of the mixing mechanism.

The radiative improved (RI) mode is a tokamak regime offering many attractive reactor features. In the article, the RI mode of TEXTOR-94 is shown to follow the same scaling as the linear ohmic confinement regime and is thus identified as one of the most fundamental tokamak operational regimes. The current understanding derived from experiments and modelling of the conditions necessary for sustaining the mode is reviewed, as are the mechanisms leading to L-RI mode transition. The article discusses the compatibility of high impurity seeding with the low central power density of a burning reactor, as well as RI mode properties at and beyond the Greenwald density.

No abstract

The first results of the Dynamic Ergodic Divertor in TEXTOR, when operating in the m=n 3=1 mode configuration, are presented. The deeply penetrating external magnetic field perturbation of this configuration increases the toroidal plasma rotation. Staying below the excitation threshold for the m=n 2=1 tearing mode, this toroidal rotation is always in the direction of the plasma current, even if the toroidal projection of the rotating magnetic field perturbation is in the opposite direction. The observed toroidal rotation direction is consistent with a radial electric field, generated by an enhanced electron transport in the ergodic layers near the resonances of the perturbation. This is an effect different from theoretical predictions, which assume a direct coupling between rotating perturbation and plasma to be the dominant effect of momentum transfer. Helical magnetic field perturbations are introduced in tokamak plasmas to study, on the one hand, the ergodic divertor concept [1,2] and, on the other hand, the interaction of such perturbations with the magnetohydrodynamics (MHD) stability of the plasma [3,4]. Recent experiments, for instance, suggest a control method to mitigate edge localized modes while maintaining the pedestal pressure and thus plasma confinement [5][6][7]. However, open questions remain, in particular, with regard to the influence on the momentum transport of the plasma. Indeed, one motivation to equip the tokamak TEXTOR with the Dynamic Ergodic Divertor (DED) [8] was to be able to study the interaction between helical magnetic field perturbations and plasma transport and stability.The DED consists of 16 magnetic perturbation coils (four quadruples), plus two additional coils for the compensation of the magnetic field imperfections at the feeder regions of the coils. The coils wind helically around the inner side of the torus (major radius: R 1:75 m; minor radius of the circular plasma cross section typically a 0:47 m) with a pitch corresponding to the magnetic field lines of the magnetic flux surface with a safety factor of q 3. Depending on the choice of coil connections to the power supplies, base modes with different poloidal and toroidal mode numbers can be produced. For the DED these are m=n 12=4, 6=2, and 3=1. The penetration depth into the plasma strongly depends on the mode numbers: While the m=n 12=4 affects the edge plasma only, the m=n 3=1 mode reaches into the plasma center (the maximum radial magnetic field component achievable by the DED at the q 2 surface is 10 ÿ3 of the total magnetic field).In this Letter we present results obtained by the m=n 3=1 mode operation. Covering about one-third of the poloidal cross section of the torus, the mode spectrum of the DED does not contain many sidebands. For the m=n 3=1 configuration the three dominant resonant components inside the plasma are m 1, 2, and 3. In Fig. 1 their strengths at the respective resonances are PRL 94, 015003 (2005) P H Y S I C A L

(2008). Temporal evolution of confined fast-ion velocity distributions measured by collective Thomson scattering in TEXTOR. Physical Review E, 77 (1) Fast ions created in the fusion processes will provide up to 70% of the heating in ITER. To optimize heating and current drive in magnetically confined plasmas insight into fast-ion dynamics is important. First measurements of such dynamics by collective Thomson scattering ͑CTS͒ were recently reported ͓Bindslev et al., Phys. Rev. Lett. 97, 205005 2006͔. Here we extend the discussion of these results which were obtained at the TEXTOR tokamak. The fast ions are generated by neutral-beam injection and ion-cyclotron resonance heating. The CTS system uses 100-150 kW of 110-GHz gyrotron probing radiation which scatters off the collective plasma fluctuations driven by the fast-ion motion. The technique measures the projected one-dimensional velocity distribution of confined fast ions in the scattering volume where the probe and receiver beams cross. By shifting the scattering volume a number of scattering locations and different resolved velocity components can be measured. The temporal resolution is 4 ms while the spatial resolution is ϳ10 cm depending on the scattering geometry. Fast-ion velocity distributions in a variety of scenarios are measured, including the evolution of the velocity distribution after turnoff of the ion heating. These results are in close agreement with numerical simulations.