Dirk Wunderlich

The predictive capabilities of population models, such as collisional-radiative and corona models, for atomic and molecular hydrogen benefit from the recent advances in the availability of reaction probabilities, particularly due to the quantum mechanical convergent close-coupling (CCC) and molecular convergent close-coupling (MCCC) methods. The CCC method provides, for the first time, an extensive set of cross sections for the electron impact excitation of atomic hydrogen that are accurate at the excitation threshold, thus enabling precise model predictions at low electron temperatures. The MCCC method provides consistent sets of cross sections for electron impact (de-)excitation of molecular hydrogen with different ro-vibrational resolution, thus enabling for the first time the development of a consistent set of population models with different detail level. This consistent set of population models allows investigations with different detail level of the plasma kinetics based on the same data basis. This work guides through the (differently resolved) population models based on the Yacora solver for atomic (Yacora H) and molecular hydrogen (Yacora H ₂ , Yacora H ₂ (X ¹ Σ _g ⁺ ,v), Yacora H ₂ (v), Yacora H ₂ (v,N) Fulcher) applying state-of-the-art reaction probabilities and demonstrates the respective individual application possibilities

The plasma in the negative hydrogen ion source for ITER NBI is generated in eight cylindrical RF drivers by inductive coupling. Four horizontal pairs of drivers are powered in series by one RF generator each. The plasma expands and overlaps into a rectangular vessel in which also caesium from several caesium ovens is evaporated. During operation of the full-size test facility SPIDER the question arose, how representative operation using a reduced number of drivers is for the performance of the full ion source. In order to answer this question, systematic investigations on the overlap of plasma properties and the extracted as well as accelerated currents have been conducted at the half-size (four RF drivers) test facility ELISE at IPP Garching. While the homogeneity of the extracted negative ion current does not differ between operation of the top or bottom RF generator and both generators, summing the extracted negative ion current from individual generator operation gives a higher result than that from the operation of both generators. This demonstrates that at least one step in the process chain connecting the precursor particles for negative ion production, the relevant plasma parameters, the production and extraction of negative ions shows a non-linear behaviour.

The neutral beam injection system in the upcoming fusion reactor ITER will use radio-frequency (RF), low-pressure, and high-power negative hydrogen/deuterium ion plasma sources. The extraction of the negative ions from the plasma is always accompanied by co-extraction of undesirable electrons, which can limit the source performance. This paper aims to investigate the effect of biased surfaces in the vicinity of the extraction on the plasma parameters towards the reduction of the co-extracted electrons. For this purpose, a 2D fluid plasma model is employed on a prototype plasma ion source where the biases of two different surfaces are varied independently. Since the self-consistent RF fluid plasma models can be computationally heavy and time-demanding, the study of the effect of the biased surfaces is used a simplified model where the RF heating is replaced with a power deposition profile, set as an input parameter. This procedure reduces the degree of freedom with more than one order of magnitude and, respectively, the computational time from tens of hours to several for one run of the model. To justify the method, the simplified model is first compared with a benchmarked RF self-consistent model on a smaller geometry. Both models produce similar results for the plasma parameters in the region of interest, namely, close to the extraction, which gives the confidence to use the simplified model for the purpose of the study. Applying a positive bias to big surfaces in the vicinity of the extraction region provides a higher electron density and plasma potential as well as a change in the electron flux orientation. However, the difference between the surface and the plasma potentials decreases more than 2 times and in combination with the reversed orientation of the flux, the electron current collected by the biased electrodes increases. The modelling results for the currents collected by the biased surfaces indicate similar behaviour as the experimental ones.

Presently, SOLPS-ITER does not replicate the levels of plasma-molecular interactions seen during deep detachment on the Super-X divertors of MAST-Upgrade and TCV. At low temperatures ( T e < 1 eV ) relevant to deep detachment, the AMJUEL reaction database contains discrepancies for detachment relevant molecular processes. In order to correct this the vibrationally resolved ground state Yacora model for H ₂ (Yacora H 2 ( X 1 , v ) ) has been used to calculate the vibrational distribution and thus from the cross-sections of the Yacora H 2 ( X 1 , v ) model effective rate coefficients for these molecular processes. Reduced and extended implementations of Yacora H 2 ( X 1 , v ) were constructed to form EIRENE-style databases for use in SOLPS-ITER. The reduced implementation utilises the same processes as EIRENE, but with updated cross-sectional data, whilst the extended implementation adds additional processes. The improved cross-sectional data lowers the threshold for molecular charge exchange (MCX) compared to AMJUEL, whilst the inclusion of the additional processes boosts the vibrational excitation in the molecular ground state providing the additional excitation energy required for MCX to be significant at low temperatures ( T e < 5 eV ). When applied to a SOLPS-ITER isolated divertor leg setup, both implementations produce greater levels of MCX that result in elevated molecular ion densities ( D 2 + ) than the default rate setup. The higher levels of D 2 + contribute to detachment via increased levels of Molecular Activated Recombination and Molecular Activated Dissociation. These plasma-molecular interactions elevate the neutral population. During detachment this has a strong impact on power and momentum losses, as well as elevated levels of Balmer α line emission and an earlier onset of the ion target flux rollover. These effects are strongest when the extended database is used.

The ELISE test facility at IPP Garching hosts an RF-driven negative hydrogen ion source and extraction and acceleration system. Its target is to demonstrate the performance foreseen for the ITER neutral beam injection (NBI) system in terms of extracted current density (329 A m⁻² in hydrogen, 286 A m⁻² in deuterium), electron-ion-ratio ( < 1) and pulse length (up to 3600 s in deuterium and 1000 s in hydrogen). NBI systems are compromised by a process commonly known as stripping losses (neutralization of negative ions before they achieve full acceleration). According to theoretical estimations, for a source filling pressure of p fill = 0.3 Pa, 29% of the extracted H − / D − ions are lost by stripping in the ITER full-scale NBI system. Experimentally, stripping losses are monitored using Beam Emission Spectroscopy (BES). Previous experimental measurements of the stripping losses at ELISE yielded values around 1.7%, which were much lower than the 10% predicted for the pre-accelerator system. The notable difference has sparked concern about the conventional interpretation of the BES experimental measurements used at IPP. The Bavarian Beam Code for Negative Ions (BBCNI) is a full 3D particle tracking code that can also produce synthetic BES spectra. BBCNI is used to evaluate how accurately different analysis techniques determine the stripping losses from the BES. This is achieved by comparing the results with the total simulated count of stripping losses. The analysis method that produces the most accurate results is identified and then systematically applied to evaluate ELISE’s data from various experimental campaigns. For a nominal source filling pressure p fill = 0.3 Pa an average of 5.8% is measured experimentally when evaluating the data with this method. Although higher than previously determined with BES, stripping losses are still smaller than predicted values. This may open the option to increase the filling pressure in the source as a means to facilitate achieving the targeted ITER conditions.

The neutral beam heating system for the future international fusion experiment ITER will be based on radio-frequency driven ion sources delivering a large (≈1 × 2 m²) and homogeneous negative hydrogen or deuterium ion beam of several ten Amperes for several hundred seconds. The size scaling experiment ELISE (Extraction from a Large Ion Source Experiment) is an integral part of the R&D road-map towards the ITER neutral beam heating system. Recently, 90% of the ITER target for the extracted current density was achieved in hydrogen for 600 s, increasing the pulse length over which such current densities can be achieved by a factor of more than ten. For ten second beam pulses the ITER target current density was achieved. These breakthrough results are made possible by using a steady-state capable high-voltage power supply together with an improved version of internal potential rods and a modified topology of the magnetic filter field.