Neutral beam driven hydrogen spectroscopy in fusion plasmas

Abstract

ITER, the next generation fusion reactor, is currently being constructed with the challenge of demonstrating the technical feasibility of a nuclear fusion driven power plant. In ITER, the generated fusion power will exceed the external heating by a factor 10, generating a large population of fusion-born helium in the plasma that needs to be measured and eventually controlled to prevent fuel dilution. The only method capable of measuring the local density of fully ionized helium in a fusion plasma is charge exchange spectroscopy (CXS), which relies on populating excited levels of plasma ions by charge transfer collisions with hydrogen atoms, injected by a powerful neutral hydrogen beam. The work described in this thesis is focussed on 2 open questions: 'Can the emission by the neutral beam itself be reliably combined with CXS to obtain absolute helium concentrations?' and 'Can CXS be used to measure density profiles of non-thermal ions?'. The quantitative interpretation of beam emission (BES), i.e. relating the observed excited population to the ground state density, has been the subject of investigation by several groups over the last 20 years, leading to different results. In this thesis, collisional-radiative models from different modellers have been compared and inconsistencies, both in the models and in the cross sections that were used, have been solved. The resulting emission rates have been used to compare expected beam densities with measured densities using Balmer-alpha and Balmer-beta beam emission spectra from the JET tokamak. Excellent agreement between the Balmer-alpha and -beta spectra, and reasonable agreement with the expected ground state density was found. The Stark line intensities within the beam emission multiplet have been compared with a newly developed sublevel resolved collisional-radiative model, showing good agreement, and providing an explanation to earlier failed experiments of using the line ratios to obtain information about the magnetic pitch angle. As a side result, beam emission spectroscopy has been used to correct and validate the determination of the fractional energy components in the neutral beams. A setup combining CXS and BES, similar to the one on ITER, has been used on the TEXTOR tokamak to validate the helium concentration measurements during strong helium gas puffs against the increase in electron density. The TEXTOR helium concentration data as well as the beam densities obtained on JET give confidence that the proposed CXS analysis scheme on ITER combining beam emission and charge exchange data will not induce systematic uncertainties larger than 20-30%. Purely statistical errors from fitting the helium and MSE spectrum on ITER are obtained from modelling the spectra and are expected to be lower than 10 over the entire radial range. Measuring fast ions in tokamak plasmas with CXS combines challenges on both the hardware and analysis side to measure and interprete a faint, broad and anisotropic spectrum, polluted by parastic emission and distorted by cross section effects. A high-resolution, high-throughput spectrometer has been commissioned on TEXTOR and the capability to detect the slowing down spectrum of beam injected ions has been demonstrated. The measured spectral shape agrees well with a slowing down model for the beam ions, but strong passive charge exchange emission was detected. Faster detectors and beam modulation is proposed for the TEXTOR diagnostic. Preliminary fast ion density profiles are obtained by sustracting the first frame after the NB switch off, showing a flattening of the core fast ion profile. Fast beam ion spectra have also been obtained on JET, but the signal to noise ratio is rather small due to the limited optical throughput of the diagnostic. Fast helium CX spectra on ITER are modelled and the feasibility to extract the fast ion density profiles is discussed as a function of the optical throughput that could be achieved on the diagnostic. Prospects are limited. Only with coarse wavelength binning and very low contamination with impurity lines could a reasonable signal to noise ratio be obtained.