A compact solid state neutral particle analyzer (SSNPA) diagnostic, previously installed at NSTX-U, has been moved to MAST-U and successfully operated in the first physics campaign (MU01). The SSNPA operates by detecting the flux of fast neutral particles produced by charge exchange (CX) reactions to diagnose the fast ion distribution. The diagnostic consists of three 16-channel sensors, which provide a radial view of the plasma and have a sightline intersection with the South–South neutral beam line. From this radial geometry, active CX signals from mostly trapped particles are observed. Each channel has a spatial resolution of 3–4 cm, a temporal resolution of 200 kHz, and an average pitch angle resolution of a few degrees. The three-sensor configuration allows for coarse energy resolution of the CX signals; each sensor sees similar sightlines but different filter thicknesses alter the energy cutoffs by known amounts. Experimental data show that all channels are collecting data as intended. The signal to noise ratio is typically around 15. Preliminary data analysis shows a correlation between the SSNPA signal and magnetohydrodynamic activity in the plasma as expected.
I. HARDWARE SETUP AND MAIN FEATURES
A. NSTX-U history
In this diagnostic’s installation on NSTX-U,1 three separate subsystems were used, allowing for radial and tangential intersections with neutral beams (NB), and a passive view, which did not intersect any beams. This diagnostic uses an AXUV16ELG array type photodiode with custom filtering discussed in the original work1 and later in this work. Custom developed first and second stage amplifiers feed this signal to a D-Tacq ACQ196PCI-96-500 for digitization. One of the major challenges for operations on NSTX-U was the electrical noise associated with using the vacuum vessel for grounding. Since the vessel ground can fluctuate through the course of a shot, this made noise suppression for the entire sensor, amplifier, and digitization stack difficult.
B. MAST-U updates
The installation on MAST-U is a single unit of these subsystems, mounted in Sector 5, at the midplane of the device (Fig. 1). This port gives the set of three 16-channel sensors a radial view of the plasma intersection with the South–South NB. Due to the issues associated with grounding in the 2016 campaign on NSTX-U, the decision was made to electrically isolate the entire device from the vessel, and in doing so, use a diagnostic ground for the experiment. A vespel gasket was used to space the flange off the vessel, and vespel sleeves and washers were used to ensure the isolation of any connecting screws. While this does create a capacitive coupling between the solid state neutral particle analyzer (SSNPA) flange and vessel, noise from this should be low compared to the noise due to fluctuations in vessel ground experienced on NSTX-U.
The quality of experimental data from MAST-U has so far confirmed that this technique is effective in noise reduction.
Other differences in the installation on MAST-U compared to NSTX-U come down to the sight lines. In the three array configuration, the sight line for each array is determined by the aperture placement. In the case of the original SSNPA design, the upper and lower apertures are placed such that the upper and lower arrays have sight lines that converge toward the middle sight line at 1.05°. This design was originally intended to bring the three sight lines to convergence on a neutral beam of NSTX-U, but now provides slight vertical spatial resolution at the intersection with the MAST-U beam. Specifically, since the higher number of channels for each array intersects the SS neutral beam closer to the beam entrance, the sight lines are vertically separated by 1.5 cm. As the intersection with the SS neutral beam moves radially inward, the sight lines converge, and then cross and diverge. Since the spread of the neutral beam is large compared to this difference, we do not expect that this has a measurable effect.
In this new installation on MAST-U, the spatial and pitch resolution of the SSNPA need to be calculated. For the spatial resolution, the individual detector line of sight (LOS) polyhedra and their intersection with the SS neutral beam are defined. The average distance between intersection center lines is then used to establish a spatial resolution. For the pitch measurement, an equilibrium reconstruction is used. The resulting B field equilibrium from shot No. 45424 is used to calculate the pitch angle in a beam density weighted average along the line of sight of each detector. These pitch angles ranged from 76.2° to 113.0°, with an average channel to channel difference of 2.5° along the center of the sight line. The pitch resolution is determined by considering the weighted average pitch measurement along the edge of each detector LOS and comparing this value to the weighted average pitch measure on the other LOS edge for the same detector. It should be noted that this pitch resolution varies across the channels due to the different radial location of the sight line intersection with the neutral beam. For CH 1, this resolution is 7.8° and for CH 15, this is only 3.1°, with the average over all channels being 4.7°.
The arrays have different tungsten foil thicknesses to provide coarse energy resolution. In order to better understand the results of the tungsten foil on the incoming neutral particles, the SRIM2 code was used to simulate the ionized neutral scattering through the tungsten and other layers into the sensor. What is clear from this simulation (Fig. 2) is that for sufficiently thick foil, all neutrals of a set energy will fail to reach the sensor behind. This sets an effective cut-off energy, after which the foil+sensor response is roughly linear in energy. This agrees with the results from NSTX as published by Shinohara et al.3
II. EXPERIMENTAL RESULTS AND DISCUSSION
A. Ohmic shot validation
Due to the fact that this instrument was already validated on NSTX-U in 2016,1 less validation was required. Due to the changes in grounding mentioned in the hardware section, much lower noise levels were observed on the blind channels as well as in the signal itself. Figure 3 shows an example of the low noise observed on one of the blind channels.
In NSTX-U,1 despite the tungsten filters, x-ray emission contributed significant noise to the SSNPA signal. In the prior work,1 the authors were able to subtract some of the x-ray signal by using the soft x-ray diagnostic on NSTX-U. Unfortunately, there was no soft x-ray diagnostic available during this first science campaign on MAST-U, so we must assume that significant x-ray noise exists. To confirm this, we can look at a purely ohmic shot (Fig. 3). In such a shot, we do not expect a large fast ion population, nor are we using a neutral beam to diagnose the fast ion population. As on NSTX-U, the observed signal on the SSNPA channels for this shot is not due to an actively diagnosed fast ion population but to x-ray emission. In future experiments with a soft-x ray diagnostic present, this emission can be accounted for to improve the signal to noise ratio.
B. Neutral beam heated shots
Moving on to shots that are heated by neutral beams, we examine shot No. 45424. Looking at Fig. 4, we also note that CH 1 of the SSNPA sees a greatly increased signal compared to the other channels. This particular sight line passes directly by a gas fueling port on the high field side of the device. This fueling is the currently accepted mechanism for this increase in signal. The additional channels see a peak in signal at channel 4, which intersects the SS Neutral beam at a radius of 95 ± 3 cm.
Figure 5 provides us with additional insight into the shot, but fails to show us that there is a locked mode. Regardless, the combination of both neutral beams provides a fast-ion population for our SS neutral beam to diagnose.
By taking the radial positions at which the SSNPA sight lines intersect the SS neutral beam, we can plot a radial profile of the signal (Fig. 6) for a set time. We note strong peaking of the signal starting with channel 3, which intersects the neutral beam at 92.6 ± 3 cm. The magnetic axis for this particular shot is at 92.8 cm, so the stronger soft x-ray emission associated with higher electron temperature Te likely contributes to the peaking of the SSNPA signal at this radius. In Fig. 6, we investigate the active SSNPA signal. By plotting the observed signal right after the SS beam turns off, we are no longer observing active charge exchange (CX) with the beam but a passive signal only. Since this measurement follows so closely after the beam switches off, other plasma parameters such as Te have not yet changed significantly. Thus, we can subtract these data points to produce the third plot in Fig. 6, a representation of the active neutral signal for the SSNPA. Despite the soft-x ray background, the SSNPA still shows good correlation with active and passive signals, as well as magnetohydrodynamic (MHD) modes. Figure 5 shows that the SSNPA signal tracks the start of the SS beam. Unfortunately, the presence of a locked MHD mode does not provide a stable fast ion distribution at the beginning of the SS beam, but the shutdown of the SW beam does result in a reduction of the fast ion population. On a shot with prominent MHD features (Fig. 7), the fishbone early in the shot and the sawteeth later both correlate with the fast ion population crashes in the SSNPA signal. These events also cause sudden drops in fast-ion D-alpha (FIDA)4 emission.
III. SUMMARY AND FUTURE WORK
This paper describes the successful initial operation of the SSNPA system on MAST-U. Correlation between MHD modes and fast-ion losses is present, with the FIDA diagnostic providing additional validation. Much lower electronic noise has been achieved through clean grounding. In the next experimental campaign, soft x-ray diagnostics will be used to remove soft x-ray backgrounds from the signal, providing more accurate measurements of the fast-ion flux.
Simulation of synthetic signals can provide additional validation of the instrument. For this purpose, the TRANSP5 NUBEAM6 module can calculate the fast ion distribution function in the absence of fast-ion transport by instabilities. The plasma parameters and distribution functions provide input to the synthetic diagnostic code FIDASIM,7,8 with the NPA output options enabled. The response of the tungsten filters can then be convolved with the neutral flux calculated by FIDASIM, so an accurate simulation of expected signals can be achieved. Preliminary simulations of this type have yielded profiles for the active neutral signal that are broader than the net experimental profile shown in Fig. 6(b). We note, however, that MAST-U shot No. 45424 has a locked MHD mode that may degrade fast-ion confinement.
Additional simulations for the tungsten filters can also be performed in order to more accurately describe the response function. The completion and validation of this response function will allow for deconvolution, providing rough energy spectra for the observed fast ions.
Future experiments with MHD quiescent shots for proper validation of simulated results are planned. With the completion of the modeling of the energy response function for the tungsten filters and the addition of a soft x-ray diagnostic, future SSNPA analysis will have access to cleaner and more accurate data.
The authors would like to acknowledge Wataru H. J. Hayashi and Jeff Lestz for their assistance. This work was supported at UCI by DOE Grant No. DE-SC0019253 and at UKAEA by RCUK Energy Program No. EP/T012250/1.
Conflict of Interest
The authors have no conflicts to disclose.
Garrett Prechel: Data curation (equal); Formal analysis (equal); Visualization (equal); Writing – original draft (equal); Writing – review & editing (equal). Nicolas Fil: Data curation (equal); Formal analysis (equal); Visualization (equal). Deyong Liu: Conceptualization (equal); Methodology (equal); Supervision (equal). W. W. Heidbrink: Funding acquisition (equal); Investigation (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Writing – review & editing (equal). Clive Michael: Data curation (equal); Resources (supporting); Supervision (supporting). Andrew Robert Jackson: Investigation (supporting); Software (supporting); Validation (supporting); Visualization (supporting).
Data is available upon request and permission from UKAEA.