Local edge parameters on the ASDEX Upgrade tokamak are investigated at the L-mode to H-mode transition, during phases with various types of edge-localized modes (ELMs), and at the density limit. A scaling law for the boundary electron temperature,e,b B 0.8 t I 0.5 p , is found which describes the H-mode threshold for deuterium-puffed discharges with favourable ion ∇B-drift direction. The region of stable operation is bounded by type I ELMs near the ideal ballooning limit and by a minimum temperature necessary to avoid thermal instability of the plasma edge. Stationary operation with type III ELMs imposes an upper limit on the edge temperature. Within the entire range of boundary densities investigated (n e,b 8 × 10 19 m −3 ), both L-mode and H-mode are found to be accessible. During type I ELMy H-mode, a relation of global confinement with the edge pressure gradient is found which is connected with a loss of the favourable density dependence predicted by the ITER-92P and ITER-93H ELMy H-mode scalings. At high density, better confinement is achieved in H-modes with an edge pressure gradient below the ideal ballooning limit, e.g. during type III ELMy H-mode with impurity-seeded radiation.

The paper describes the effect of the isotopic mass on plasma parameters as observed in the ASDEX tokamak. The paper comprises Ohmic as well as L mode, H mode and H* mode scenarios. The measurements reveal that the ion mass is a substantial and robust parameter, which affects all the confinement times (energy, particle and momentum) in the whole operational window. Both core properties such as the sawtooth repetition time and edge properties such as the separatrix density change with the isotopic mass. Specific emphasis is given to the edge parameters and changes of the edge plasma due to different types of wall conditioning, such as carbonization and boronization. The pronounced isotope dependences of the edge and divertor parameters are explained by the secondary effect of different power fluxes into the scrape-off layer plasma and onto the divertor plates. Finally, the observations serve to test different transport theories. With respect to the ion temperature gradient driven turbulence, the isotope effect is also studied in pellet refuelled discharges with peaked density profiles. The results from ASDEX are compared with the results from other experiments

Global and local H-mode threshold analyses are presented and discussed. The density window was extended up to the density limit towards which the power threshold exhibits a dramatic increase. First results with the new divertor are compared to the previous threshold scaling. Local edge measurements at the L-H transition show that the temperature decreases slightly with density and increases with magnetic field, and that the ion collisionality at the plasma edge is always clearly above unity, between 5 and 15. Experiments at high power threshold exhibit improved L-mode confinement and suggest that two mechanisms are required for the L-H transition.

The elements of transport into and across the scrape-off layer in the poloidal divertor tokamak ASDEX Upgrade are analysed for different operational regimes with emphasis on enhanced confinement regimes with an edge barrier. Utilizing the existing set of edge diagnostics, especially the highresolution multi-pulse edge Thomson scattering system, in combination with long discharge plateaus, radial sweeps and advanced averaging techniques, detailed radial mid-plane profiles of diverted plasmas are obtained. Profiles are smooth across the separatrix, indicating strong radial correlation, and there is no remarkable variation across the second separatrix either. Together with measured input, recycling, pumping and bypass fluxes, a corrected separatrix position is determined and transport characteristics are derived in the different radial zones generally identified in the profile structure. Transport in the steep gradient region inside and across the separatrix shows typical ballooning-type critical electron pressure gradient scaling and, in parallel, even a clear correlation between radial electron density and temperature decay lengths (e.g. η e = d(ln T )/d(ln n) ∼ 2 for type-I ELMy H-modes). These findings indicate the importance of stiff profiles in this region, while diffusion coefficients are secondary parameters, determined essentially by the source distribution. The outer scrape-off layer wing exhibits a more filamentary structure with preferential outward drift especially in high-performance discharges, with formal diffusion coefficients far above the Bohm value in agreement with results on the old ASDEX experiment. A basic mechanism involved there seems to be partial loss of equilibrium and fast curvaturedriven outward acceleration, in principle well known from theory, investigated decades ago in pinch experiments and utilized recently in high-field-side pellet fuelling.

Injection of cryogenic deuterium pellets has been successfully applied in ASDEX Upgrade for external edge localized mode (ELM) frequency control in type-I ELMy H-mode discharge scenarios. A pellet velocity of 560 m s −1 and a size of about 6×10 19 D-atoms was selected for technical reasons, although even lower masses were found sufficient to trigger ELMs. A moderate repetition rate close to 20 Hz was chosen to avoid over-fuelling of the core plasma. Pellet sequences of up to 4 s duration were injected into discharges close to the L-H threshold, intrinsically developing large compound ELMs at a rate of 3 Hz. With pellet injection, these large ELMs were completely replaced by smaller type-I ELMs at the much higher pellet frequency, accompanied by a slight increase of density and even of stored energy. This external ELM control could be repeatedly switched on and off by just interrupting the pellet train. ELMs were triggered in less than 200 µs after pellet arrival at the plasma edge, at which time only a fraction of the pellet has been ablated, forming a rather localized, three-dimensional plasmoid, which drives the edge unstable well before the deposited mass is spread toroidally. The pellet controlled case has also been compared with a discharge at a somewhat lower density, but with otherwise rather similar data, developing spontaneous 20 Hz type-I ELMs. Despite the different trigger mechanisms, the general ELM features turn out to be qualitatively similar, possibly because of the similarity of the two cases in terms of ELM relevant parameters. The scaling with background plasma, heating power, pellet launch parameters, etc over a larger range still remains to be investigated.

At the central column of ASDEX Upgrade, an area of 5.5 m 2 of graphite tiles was replaced by tungsten-coated tiles representing about two-thirds of the total area of the central column. No negative influence on the plasma performance was found, except for internal transport barrier limiter discharges. The tungsten influx W stayed below the detection limit only during direct plasma wall contact or for reduced clearance in divertor discharges spectroscopic evidence for W could be found. From these observations a penetration factor of the order of 1% and effective sputtering yields of about 10 −3 could be derived, pointing to a strong contribution by light intrinsic impurities to the total W-sputtering. The tungsten concentrations ranged from below 10 −6 up to a few times 10 −5 . Generally, in discharges with increased density peaking, a tendency for increased central tungsten concentrations or even accumulation was observed. Central heating (mostly) by ECRH led to a strong reduction of the central impurity content, accompanied by a very benign reduction of the energy confinement. The observations suggest that the W-source strength plays only an inferior role for the central W-content compared to the transport, since in the discharges with increased W-concentration neither an increase in the W-influx nor a change in the edge parameters was observed. In contrast, there is strong experimental evidence, that the central impurity concentration can be controlled externally by central heating.

A new 10 channel Thomson scattering (TS) system was installed on the ASDEX Upgrade tokamak to measure radial profiles of electron density and temperature at the plasma edge with high radial resolution. Together with the already existing TS system, which is now used for the core plasma, electron density and temperature profiles extending from the edge to the core are now obtained in a single discharge. The TS systems are relatively calibrated by an optical parametric oscillator

Density limit investigations on ASDEX have been performed under a variety of conditions: ohmically heated and neutral injection heated plasmas in H2, D2 and He have been studied in different divertor configurations, after various wall coating procedures, with gas puff and pellet fuelling, and in different confinement regimes with their characteristically different density profiles. A detailed description of the parametric dependence of the density limit, which in all cases is a disruptive limit, is given. This limit is shown to be a limit to the density at the plasma edge. Therefore, the highest densities corresponding to neRqa/Bt>30*1019 m-2.T-1 are obtained with centrally peaked ne profiles. Radiation from the main plasma at the density limit is always significantly below the total input power. The plasma disruption is due to an m=2 instability which for medium and high qa is preceded by one or more minor disruptions. In this range of qa, the disruptive instability is initiated by the occurrence of a Marfe on the high field side as a consequence of strong plasma cooling in this region. The duration of the Marfe increases with increasing distance between the plasma edge and the q=2 surface. After penetrating onto closed flux surfaces the Marfe leads to a current contraction and a subsequent destabilization of the m = 2 mode. In helium plasmas a strongly radiating, poloidally symmetric shell is observed before the density limit instead of a Marfe. An instantaneous destabilization of this mode is observed at low qa. Detailed measurements of plasma edge and divertor parameters close to the density limit indicate the development of a cold, dense divertor plasma before the disruption. Models describing the scrape-off layer and the divertor region predict an upper limit to the edge density at low divertor temperatures according to power balance considerations. Their relations to the experimental findings, especially the low field side cooling, ar