The linear plasma generator Magnum-PSI

The Dutch Institute for Fundamental Energy Research (DIFFER) has built the new experimental research facility Magnum-PSI. In Magnum-PSI, Plasma Surface Interaction (PSI) research for the nuclear fusion reactor ITER and reactors beyond ITER will be carried out. As such, it is essential that the experiment can access the high-density, lowtemperature regime relevant for the ITER divertor strike zones (electron density ne ~ 1020 – 1021 m-3 and electron temperature Te ~ 1 – 5 eV). Magnum-PSI is a linear plasma generator in which the interaction between a large (up to 10 cm) diameter plasma beam with a target material can be investigated. It will have a steady state high flux of 1023 - 1025 ions m-2 s-1 at a plasma temperature in the eV range confined by a magnetic field of 2.5 T. In this way, molecules and dust particles that come off the surface are confined and remain part of the plasma-surface interaction system. The work described in this thesis forms the basis for the design of the new machine and shows its excellent performance. The design makes sure that the ITER relevant regime of PSI can be accessed while having sufficient diagnostic access to the machine. Key elements are a low neutral gas density in the target region and a continuous magnetic field of sufficient magnitude. The neutral gas contribution to the surface interaction region coming directly from the source should be much lower than the plasma flux, and the magnetic field should be strong enough to guide the ionized particles from the source to the target and to confine particles of a certain size which are ionized after coming off the surface. The demands for a steady state magnetic field without limiting the diagnostic access to the machine have led to the design of a superconducting magnet with a 1.3 m warm bore and 16 radial access ports. This design is the result of a conceptual design study to optimize the following parameters: field strength and homogeneity, diagnostic transparency, inner diameter, ramp-up and down times, safety and costs (capital and running costs). The most cost effective cooling solution turned out to be a recondensing system, where the coils are submerged in liquid helium while the evaporated helium is stored in a gas balloon to be re-condensed by cryocoolers. The stray field of this large magnet is shielded by iron walls surrounding the experimental area. The dimensions and positions of these walls have been determined with a nonlinear finite element solver. The vacuum system must be capable of reaching the required neutral pressure in the target region while the plasma source puts in large quantities of gas to reach the high fluxes. This demand has led to the design of a 3 stage differentially pumped vacuum system, where the vacuum vessel consists of separate chambers divided by skimmers and pumped with their own pumping stations. In a comparative study between roots pumps, vapor boosters, cryopumps and turbo pumps, the roots pumps were identified as being the most reliable, economic and user friendly pump solution for these high gas loads. The pumps have been commissioned and the measured pump speeds are comparable to the calculated values. It was shown that the chosen pump solution is capable of reaching its requirements. Since the flow in Magnum-PSI is mostly in the rarified regime, a direct simulation Monte Carlo (DSMC) method was used to model the flows. Two stage differential pumping was modeled and applied in the linear plasma devices Pilot-PSI and PLEXIS. Both simulations and experiments showed that the optimum skimmer position depends on the position of the shock and therefore shifts for different gas parameters. In Magnum-PSI, a moveable plasma source is implemented which makes it possible to change the distance between source and skimmer for different operating conditions. Using DSMC simulations, the shape of the skimmer has been designed in such a way that it effectively stops neutrals from flowing to the next chamber, while having a minimum impact on the shock structure. During the commissioning phase of Magnum-PSI, the non-magnetized expansion of 5 Pa m3 s-1 (3 slm) argon in the differentially pumped vacuum vessel at low background pressures was studied. The behavior of the neutral component was studied with Rayleigh scattering measurements and DSMC simulations. Thomson scattering and double Langmuir probe measurements were performed on the ionized fraction. It was found that the electrons and neutral particles are not completely coupled in the shock front. The neutral fraction shows clear signs of invasion from hotter background gas, causing the average temperature and density to increase before the shock. This is also evident from the ionization ratio, which has been determined in front of and behind the first skimmer. This study has validated the design of the machine and has helped to understand the behavior of the gas flow. Due to a significant delay in the delivery of the superconducting magnet, Magnum-PSI has started operation as a pulsed machine using conventional electromagnets. To accommodate the smaller diameter coils, one of the vacuum chambers is temporarily removed leading to a two stage differentially pumped vacuum system. This results in a somewhat higher neutral pressure in the target region. The coils provide a magnetic field in the source region of 1.9 T for 6 seconds. Lower field settings with correspondingly longer pulse lengths are available. Due to the expanding magnetic field in the target region, the plasma beam expands a factor of 2.3 in diameter before it reaches the target. First experiments have been carried out on 5 Pa m3 s-1 (3 slm) magnetized hydrogen and deuterium plasma beams. Thomson scattering 25 mm in front of the target has yielded electron densities up to 3.6×1020 m-3 and electron temperatures up to 3.7 eV in a 25 mm (FWHM) diameter beam. Exposure of a tungsten target has led to average heat and particle fluxes well in excess of 4 MW m-2 and 1024 m-2 s-1 respectively. Using pressure and calorimetric measurements, it was shown that the plasma surface interaction is mainly determined by the incoming ion flux and not by neutrals coming directly from the source. The application of a differentially pumped vacuum vessel has led to an unprecedented high ionization efficiency of the plasma source. These first results validated the design of Magnum-PSI and have shown that ITER relevant conditions have been achieved.