Beam-ion losses velocity-space distribution under neutral-beam injection on EAST

PAPER • OPEN ACCESS Beam-ion losses velocity-space distribution under neutral-beam injection on EAST To cite this article: S.S. Wang et al 2025 Nucl. Fusion 65 016026 View the article online for updates and enhancements. You may also like EMC3-EIRENE simulations of edge plasma and impurity transport by toroidally localized argon seeding on CFETR Xdivertor T. Xie, H. Li, W. Zhang et al. - Influence of the density gradient on turbulent heat transport at ion-scales: an inter-machine study with the gyrokinetic code stella H. Thienpondt, J.M. García-Regaña, I. Calvo et al. - Experimental characteristics of lost fast negative ions on EAST tokamak Z.X. Zhang, J. Huang, J.F. Chang et al. - This content was downloaded from IP address 193.147.173.203 on 24/04/2025 at 09:01 International Atomic Energy Agency Nuclear Fusion Nucl. Fusion 65 (2025) 016026 (15pp) https://doi.org/10.1088/1741-4326/ad933f Beam-ion losses velocity-space distribution under neutral-beam injection on EAST S.S. Wang1,3,a, Z.X. Zhang1,2,a, J. Huang1,∗, J.F. Chang1,∗, J. Galdon-Quiroga4, L. Sanchis4, W. Gao1, J. Fu1,2, Y.X. Sun1,2, X.H. Wang1,2, C. Shi1,2 and the EAST Team1,b 1Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei 230031, China 2University of Science and Technology of China, Hefei 230026, China 3Qingdao Sifang SRI Intellectual Technology Co., Ltd, Qingdao 266031, China 4Department of Atomic, Molecular and Nuclear Physics, University of Seville, Seville 41012, Spain E-mail: [email protected] and [email protected] Received 21 August 2024, revised 5 November 2024 Accepted for publication 15 November 2024 Published 26 November 2024 Abstract The velocity-space distribution of the fast-ion loss in EAST neutral-beam injection (NBI) heating discharge is obtained both from Scintillator-based fast-ion loss detector (FILD) signals and by ASCOT5 and FILDSIM simulations. The results of simulations are in good agreement with the distribution of beam-ion losses measured with FILD of EAST and the correctness of the fast-ion loss distribution has been demonstrated. Simulations indicate that the beam-ion losses observed by the FILD probe are attributed to the fast ions from both the high-field side (HFS) and the low-field side (LFS). However, the beam-ion losses from the HFS (associated with NBI1L) have not been observed experimentally due to the limited detecting range of the FILD probe. Therefore, an upgrade and modification of the FILD probe was carried out in 2022 to enable the detection of fast-ion loss with smaller pitch angles. Comparative analysis is conducted in neutral-beam injection (NBI2R) discharges after the upgrade, which indicates that the velocity-space distribution of beam-ion losses from the HFS has strong agreement between experimental measurements and simulation results. However, the experimental and simulated results of the velocity-space distribution of beam-ion losses from the LFS shows inconsistencies, primarily because the BBNBI module in the simulation does not consider the contributions of boundary neutral particles to neutral-beam deposition (ionization reactions). These conclusions not only provide valuable references for improving the neutral-beam deposition model but also establish a fundamental basis for further exploring the mechanisms of fast-ion loss under various conditions on the EAST tokamak. aThese authors contributed equally to this work and should be considered co-first authors. bSee Wan et al 2017 (https://doi.org/10.1088/1741-4326/aa7861) for the EAST Team. ∗Authors to whom any correspondence should be addressed. Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. 1741-4326/25/016026+15$33.00 Printed in the UK 1 © 2024 The Author(s). Published by IOP Publishing Ltd on behalf of the IAEA Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Keywords: beam-ion losses, velocity-space distribution, ASCOT, FILDSIM (Some figures may appear in colour only in the online journal) 1. Introduction It is well known that effective control of fast-ion loss is one of the key physical challenges in achieving steady-state and highperformance plasma for future fusion device such as ITER and CFETR. Fast-ion losses in fusion devices can reduce heating and current drive efficiency and may even cause localized damage to the first wall if concentrated [1]. Fast-ion populations created by fusion reactions, by neutral-beam injection (NBI) and by radiofrequency (RF) heating are usually concentrated in the center of the plasma [2]. The velocity-space distribution of lost fast ions is a key physical quantity for studying the control mechanisms of the fast-ion loss, as it enables the resolution of lost fast-ions orbits through trace reversal. Scintillator-based fast-ion loss detector (FILD), which can measure pitch angle and gyroradius simultaneously, is an important diagnostic tool to study fast-ion loss. The FILD diagnostic has been applied in several tokamaks, such as ASDEX Upgrade [3], MAST-U [4], DIII-D [5], JET [6], KSTAR [7] and NSTX [8]. FILD has been installed on Experimental Advanced Superconducting Tokamak (EAST) to investigate the behavior of fast-ion losses. It provides velocity-space distribution during NBI and ion cyclotron resonance heating (ICRH) [9]. In this work, the velocity-space distribution is inferred from FILD measurements and ASCOT simulation in the EAST. The data analysis tools consist of FILDSIM [10] and ASCOT [11]. The model is based on the FILD strike-map and FILDSIM model. FILDSIM is a Monte Carlo orbit-tracing code that can establish connections between losses at the first wall and at FILD scintillator and it has been applied in ASDEX Upgrade and MAST-U [12] amongst others. The velocity-space distribution of beam ion prompt loss at FILD location can be obtained by ASCOT because it can also simulate in scrapeoff layer (SOL) and FILD probe locate at SOL. ASCOT has been applied in ASDEX Upgrade [13], W7-X [14], ITER [15] and JET [16] under NBI [17] and ICRF [18] heating for fastion loss [19], wall load [20] and neutron [21] studies in fusion devices. The work is structured as follows. After the introduction in section 1, the FILD system on EAST is introduced in section 2. Section 3describes the verification and analysis of beamion loss velocity-space distribution by FILD on EAST. The impact of neutral particles on beam-ion losses is discussed in section 4. Section 5summarizes the features of beam-ion loss velocity-space distribution in the EAST. 2. Experimental setups FILD has been installed on EAST to investigate fast-ion loss behavior in high performance plasma [9]. The NBI heating system is shown in figure 1(a) and location and direction of NBI2 has been changed in NBI upgraded at 2020. The reciprocating FILD is installed above the outer midplane on EAST Port J as shown in figure 1(b). FILD diagnostic system is composed of the detector head, the exchange box, the optical splitter, the detection/data acquisition system, and the long shaft system [22]. A high-speed charge-coupled device (CCD) camera system (Phantom V2010) captures images of the scintillator screen, providing information about the pitch angle and gyroradius of lost fast-ions. Additionally, a 25-channel photomultiplier tube (PMT) data acquisition system tracks the evolution of the fast-ion loss signal over time [23]. The main components of the detector head are the ion collimator, which consists of a pinhole, slit and the scintillator screen, as shown in figure 2. The probe geometry was upgraded during experiments conducted in the first half of 2022. The pinhole was relocated from the center of the collimator to its periphery to extend the detection range of the FILD, as illustrated in figure 3(a). The key parameters of probe such as length and width of the collimator pinhole, the distance from collimator pinhole to the scintillator, the length and width of the scintillator determine the position of fast ions striking the scintillator with different energy and pitch angles. The trajectory of fastion in the probe is shown in figure 3(b). The lost fast ions pass sequentially through pinhole and slit and finally strike to the scintillator screen. Before this work, the gyroradius of lost ions was calculated based on the assumption of a circular orbit passing through the center of the pinhole, slit, and the strike point on the scintillator. However, this method is inaccurate due to the limitations imposed by the finite sizes of the pinhole and slit. In fact, fast ions with fixed energy and gyroradius at front aperture will present a distribution when they strike to scintillator instead of a point. Hence, it is necessary to establish a connection between velocity-space distribution at front aperture and velocity-space distribution at scintillator, or even the CCD. In order to characterize the response of the FILD detector, we introduced FILDSIM for trajectory calculation [10]. To verify the distribution of beam-ion loss, ASCOT has been applied to simulate the velocity-space distribution of beam-ion loss at FILD probe location. 3. Beam-ion losses velocity-space distribution measured by FILD 3.1. Analysis and modelling tools The main analysis and modelling tools in this work are FILDSIM and ASCOT. FILDSIM is a code for trajectory calculations of the ions in the FILD probe. Based on the initial velocity-space distribution of the incoming ions that reach the 2 Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Figure 1. (a) Distribution of the neutral-beam injection (NBI) heating systems and location of FILD (top view). The configuration of NBI2 has been changed during the NBI upgrade in 2020. The black beam line represents NBI2 after 2020, while the blue beam line represents NBI2 before 2020. The NBI1L and NBI2R are more tangential than NBI1R and NBI2L before 2020. The NBI1L and NBI2L are more tangential than NBI1R and NBI2R after 2020. (b) Location of FILD at R-Z space at J port (toroidal angle =22.5◦). Figure 2. The 3D model of detector head. The pinhole of the detector head has been moved from the center to the edge of the collimator. (a) Before 2022. (b) After 2022. pinhole, FILDSIM is employed to generate the corresponding distribution on the scintillator plate [10,24]. FILDSIM model uses a weight function formalism to relate the velocityspace distribution of fast-ion losses reaching the detector pinhole to the scintillator pattern obtained experimentally, which can be understood as a distortion of the velocity-space distribution due to the finite resolution of the system. The weight function consists of efficiency function and probability matrix. The efficiency function is associated with scintillator yield and probability matrix can be calculated by FILDSIM. A detailed description of the FILDSIM model can be found in [10]. In this work, we only considered the probability matrix and set the efficiency function as constant for qualitative analysis. Since we only investigate NBI ions that primarily lose energy close to the NBI injection energy (which is nearly monoenergetic), the use of a constant yield is justified. We will also assess the luminescence efficiency (yield) of the scintillator material (ZnS:Ag) in the future. The resolutions of the FILD on EAST were obtained using FILDSIM after incorporating the geometrical parameters of the FILD into the simulation. The resolution in gyroradius is illustrated in figure 4(a), which shows gyroradius distribution profiles obtained in the scintillator velocity-space along a line of constant pitch angle. The different colors correspond to different values of the particle gyroradius started at the detector pinhole. The strike point distributions can be fairly well modeled as skew Gaussian, similar 3 Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Figure 3. (a) The position of pinhole relative to scintillator. The green line is position of pinhole before 2022 and red lines are positions of pinhole after 2022. (b) Fast-ion orbit which strikes to scintillator in FILD probe. to the FILD on ASDEX Upgrade. The resolution of gyroradius is poorer for larger gyroradius for the FILD. The resolution in pitch-angle illustrated in figure 4(b) indicates the strike-point distributions of pitch-angle fits to Gaussian function. The results suggest that the FILDSIM model is applicable to the FILD on EAST. ASCOT is a code used for simulating the motion of fastions and impurity particles through Monte Carlo particle tracking methods [11]. The information (e.g. location, orbit, momentum, energy, and distribution) of markers in fusion device can be obtained by ASCOT simulation. The velocity distribution of beam ion prompt loss at FILD location can be obtained since FILD is situated in the SOL, and ASCOT also allowed simulations in the SOL region. The input parameters for ASCOT mainly include markers, magnetic field, electric field, plasma profiles, and wall model. The markers were generated by an ASCOT model named BBNBI5 [25], which takes into account the fine structure of the injector, follows the injected neutrals until ionization, and generates a source ensemble of ionized NBI markers for slowing down calculations. Figure 5shows the direction of the magnetic field and plasma current for discharge #85626 in ASCOT, along with the 2 million markers of NBI1L generated by BBNBI5. The magnetic field and plasma current are both directed counterclockwise in the top view of discharge #85626. 3.2. Results of experiments and simulations To research the velocity-space distribution of beam-ion loss in EAST, we chose three time points (t1=3.1 s, t2=4.5 s, t3=6.5 s) in discharge #85626 for analysis. Discharge #85626 is a H-mode discharge before FILD probe geometry was upgraded in 2022 and beam geometry was upgraded in 2020. As shown in figure 6, red dashed line indicates the three time points and each time point corresponds to only one NBI beamline. The energies of three NBI beamlines are all about 50 keV. The plasma current is 400 kA and Bt =2.40 T. The plasma profiles of discharge #85626 are shown in figure 7, the electron density profiles are derived from reflectometry system [26] and Polarimeter-Interferometer (POINT) system [27], while the electron temperature and ion temperature profiles are respectively acquired through TS (Thomson Scattering) Diagnostic System [28] and CXRS (Charge Exchange Recombination Spectroscopy) [29]. Covered the strike map grid to the FILD signal of discharge 85626 at 3.1 s, 4.5 s and 6.5 s, the velocity-space distribution of lost fast-ion has been shown in figure 8. FILD can detect fast-ion with pitch angles ranging from 50◦to 140◦ based on the strike map grid, and we selected 1 cm to 12 cm in the grid because the resolution of gyroradius is poorer for larger gyroradius. More intuitive velocity-space distribution has been acquired through FILD signal by FILDSIM remapping as shown in figure 9. Figure 9(a) showed a bright spot with gyroradius about 2.1 cm and pitch angle about 57◦, which corresponding to the fast-ion loss from NBI1L injection. In the same way the gyroradius and pitch-angle of beam ion prompt loss of NBI2L are respectively about 2.1 cm and 66◦from figure 9(b). After remapping FILD signal at 6.5 s, the gyroradius and pitch-angle of beam ion prompt loss of NBI2R are respectively about 2.1 cm and 57◦from figure 9(c). Considering the ion species (deuterium) and the magnetic field at the probe location (∼1.9 T), the energy of lost beam ions can be calculated based on the gyroradius. Figures 9(d)–(f) indicate that the energy of the lost fast-ions produced by the three beamlines is approximately 45 keV. As shown in figure 1, 4 Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Figure 4. (a) Gyroradius distribution with pitch angle held constant in the scintillator velocity space; (b) pitch angle distribution with gyroradius held constant in the scintillator velocity space. Figure 5. Direction of the magnetic field and plasma current and markers from NBI1L. The blue arrow indicates the direction of the magnetic field, while the red arrow indicates the direction of the current. The black dots represent the test particles. the NBI2L and NBI2R beam lines originate from the same source but have different directions, resulting in different pitch angles in their velocity distributions. Meanwhile, the NBI1L and NBI2R beam lines have different sources but similar beam line directions, which leads to similar pitch angles in their velocity distributions, as the pitch angle of the velocity-space distribution is related to the beam line direction. The wall load data have been obtained through ASCOT5 simulation of discharge 85626 at 3.1 s. Each marker was tracked along its full orbit for up to 10 ms, until it impacted the first wall and the tracking was terminated. A distinct bright spot is visible on the wall at the FILD location, with a toroidal angle of approximately 22.5◦and a poloidal angle of around 25◦, as shown in figure 10. The four regions of beam-ion loss are the FILD, the Lower Hybrid Wave limiter, the main limiter, and the divertor. The majority of lost beam ions are struck on the main limiter. Two million markers were simulated over 10 ms (using 160 cores on the ShenMa High-Performance Computing Cluster with a runtime of approximately 70 h) to increase the number of markers collected at the FILD probe Figure 6. Time traces of main parameters for EAST #85626. (a) Ip. (b) Density. (c) Dαsignal. (d) NBI energy for NBI1L, NBI2L and NBI2R. The red dashed line indicates the three selected moments. location. The velocity-space distribution of beam-ion loss with different beam injectors have been obtained by ASCOT5, and the results are shown in figure 11. The distribution consists of two areas with only NBI1L injection as shown in figure 11(a) and initial position of markers for two areas are different as shown in figure 12. The pitch angle of markers near the highfield side (HFS) ranges from 20◦to 40◦. For markers near the low-field side (LFS), the pitch angle is approximately 57◦, and this observation is consistent with the distribution measured by FILD. Another distribution area cannot be measured by FILD because the FILD pitch-angle measuring range is 50◦– 140◦. Figures 11(b) and (c) show velocity-space distributions of beam-ion loss with NBI2L and NBI2R. The synthetic velocity-space distribution simulated by ASCOT is not at FILD scintillator but rather at the pinhole. Therefore, we should use FILDSIM model to obtain synthetic scintillator velocity-space distribution. Figure 13(a) shows the synthetic pinhole velocity-space distribution by 5 Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Figure 7. Plasma profiles of EAST 85626 at t=3.1 s (NBI1L), t=4.5 s (NBI2R) and t=6.5 s (NBI2L) are shown in (a) electron density. (b) Electron temperature, and (c) ion temperature. FILDSIM with ASCOT distribution (LFS: pitch-angle ∼57◦) as input. The synthetic scintillator velocity-space distribution of beam-ion loss can be gained by FILDSIM forward model which considered probability matrix. From figures 9and 13, the gyroradius of beam ion prompt loss of NBI1L and NBI2R injection is about 2.1 cm and the pitch-angle is about 57◦. At the same time, the gyroradius of the beam ion prompt loss from NBI2L injection is approximately 2.1 cm, with a pitch angle of about 66◦. The beam lines NBI2L and NBI2R are counterIPbeams, demonstrating significantly larger losses compared to the co-IPbeam (NBI1L). The results also indicate that the more tangential beamlines (NBI1L, NBI2R) exhibit lower prompt losses compared to less tangential beams (NBI2L), which is consistent with prior studies [30,31]. By comparing the gyroradius profiles of beam-ion loss distributions obtained from ASCOT simulations and FILD measurements, while keeping the pitch angle constant (at the center of the bright spot) across different NBI injections, we find good agreement Figure 8. Velocity-space distribution of beam-ion loss of #85626 by FILD CCD mapping with different NBI injection: (a) NBI1L, (b) NBI2L, and (c) NBI2R. between the ASCOT simulations and FILD measurements for all three NBI injections, as shown in figure 14. 3.3. Orbit analysis of beam-ion losses In order to analyze the loss mechanism of different pitch-angle areas in figure 10(a), the orbit of lost fast-ions in two area have been calculated by ASCOT5. Figures 15(a) and (b) respectively show orbit of HFS and LFS lost beam ions with NBI1L. The orbit of lost fast-ions whose initial position near HFS is passing lost orbit while the orbit of lost fast-ions whose initial position near LFS is trapped orbit. Figure 16 shows the orbit classification and topological boundaries of these two areas. The orbits of lost beam ions whose initial position near HFS are all co-passing lost orbits and orbits of lost beam ions whose initial position near LFS are nearly all trapped-lost orbits. The results need short loss time because the boundary here based on conservation of energy, magnetic moment and toroidal canonical angular momentum. Therefore, we can distinguish between trapped particles and passing particle losses by pitch angle of beam-ion loss velocity-space distribution during NBI1L injection. 4. The impact of neutral particles on beam-ion losses 4.1. Beam-ion losses with modified FILD probe head In order to further verify the simulated velocity-space distribution, as shown in figure 17, the FILD probe was upgraded 6 Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Figure 9. Velocity-space Gyroradius-Pitch angle distribution of beam-ion loss of #85626 by FILD CCD remap with different NBI injections: (a) NBI1L, (b) NBI2L, and (c) NBI2R. Velocity-space Energy-Pitch angle distribution of beam-ion loss of #85626 by FILD CCD remap with different NBI injections: (d) NBI1L, (e) NBI2L, and (f) NBI2R. Figure 10. EAST #85626 at 3.1 s wall load with NBI1L injection obtained by ASCOT5. in the first half of 2022, allowing it to receive beam ions with pitch angles ranging from approximately 30◦–100◦. Discharge #115236 was conducted with only NBI2R injection after the FILD probe and NBI were upgraded on EAST. The time traces of main parameters for EAST #115236 are shown in figure 18. The energy of NBI2R beamline is all about 60 keV. The 7 Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Figure 11. The velocity-space distributions of beam-ion loss at FILD probe location by ASCOT with different NBI. (a) NBI1L. (b) NBI2L. (c) NBI2R. The orange dotted lines indicate the probe detection range. The velocity-space distributions of markers from different beam lines are used as input conditions for FILDSIM, with the resulting output shown in figure 13. Figure 12. The initial positions of the test particles collected at the FILD probe location shown in figure 11(a). The green dot and the blue dot respectively represent test particles with a smaller pitch angle and a larger pitch angle in the fast-ion loss velocity-space distributions. plasma current is 500 kA and Bt=1.64 T. NBI2R injection is a verticality beam line after NBI upgrade. The plasma profiles of #115236 at 4 s are shown in figure 19(a), where the electron density and temperature, as well as the ion temperature, are obtained through diagnostic measurements and fitting. These data serve as inputs for the plasma parameters in ASCOT. The velocity-space distribution of beam-ion losses at the FILD probe location, originating from NBI2R and simulated by ASCOT, is shown in figure 19(b), with the orange dotted line indicating the probe detection range. In order to validate the consistency between experimental measurements and simulation results, the velocity-space distribution of fast-ion losses at the FILD probe aperture, obtained from ASCOT, was used as an input for FILDSIM. Figure 20 presents the comparison of the velocity-space distributions between the discharge #115236 simulation and the experimental measurements. Both the simulated and experimental beam-ion loss signals exhibit two pitch angle regions. The lost beam ions exhibit a small pitch angle of approximately 45◦–50◦, with a gyroradius range of 3–4 cm. Both simulated and experimental data show good agreement in this range. However, discrepancies are observed between the simulated and experimental results for beam ions with larger pitch angles. The simulated pitch-angle range is approximately 60◦–70◦, corresponding to a gyroradius range of 1.8 cm to 3.8 cm. In contrast, the experimental measurements reveal a pitch-angle range of about 70◦–75◦and a gyroradius range of approximately 3 cm–5 cm. There are two factors contributing to the discrepancies between the simulated and experimental velocity-space distributions of beam-ion losses at larger pitch angles. Firstly, discrepancies arise from errors in aligning the experimental beam-ion loss signals with the strike-map derived from FILDSIM, and inaccuracies in the magnetic field at the FILD probe location can also affect the simulated velocity-space distribution of beam-ion losses at the probe. Secondly, the simulation results indicate that the average time for beamion loss from the HFS to the probe (τHFS_loss ∼6.6 ×10−4s) is significantly shorter than that from the LFS to the probe (τLFS_loss ∼6.4×10−3s). The latter may constitute the predominant factor. The beam ions originating from the LFS undergo a longer period of slowing down before being lost 8 Nucl. Fusion 65 (2025) 016026 S.S. Wang et al Martin R., McClements K.G., Sanchis L. and Zoletnik S. 2019 J. Instrum. 14 C09015 [13] Sanchis L. et al 2019 Plasma Phys. Control. Fusion 61 014038 [14] Kontula J., Koschinsky J.P., Äkäslompolo S. and Kurki-Suonio T. (W7-X Team) 2021 Plasma Phys. Control. Fusion 63 035022 [15] Sanchis L. et al 2021 Nucl. Fusion 61 046006 [16] Sirén P., Varje J., Weisen H. and Koskela T. (JET Contributors) 2017 J. Instrum. 12 C09010 [17] Äkäslompolo S. et al 2019 J. Instrum. 14 C10012 [18] Sipilä S. et al 2021 Nucl. Fusion 61 086026 [19] Särkimäki K. 2020 Nucl. Fusion 60 036002 [20] Kurki-Suonio T. et al 2017 Plasma Phys. Control. Fusion 59 014013 [21] Sirén P., Varje J., Äkäslompolo S., Asunta O., Giroud C., Kurki-Suonio T. and Weisen H. (JET Contributors) 2018 Nucl. Fusion 58 016023 [22] Wu C.R. et al 2018 Rev. Sci. Instrum. 89 10I144 [23] Jin Z., Chang J.F., Huang J., Wu C.R., Xu Z., Chu N., Lin S.Y., Hu L.Q., Isobe M. and Ogawa K. 2017 Fusion Eng. Des. 125 160 [24] Rodriguez-Ramos M. et al 2017 Plasma Phys. Control. Fusion 59 105009 [25] Asunta O., Govenius J., Budny R., Gorelenkova M., Tardini G., Kurki-Suonio T., Salmi A. and Sipilä S. (The ASDEX Upgrade Team and The JET EFDA Contributors) 2015 Comput. Phys. Commun. 188 33–46 [26] Qu H. et al 2015 Plasma Sci. Technol. 17 985 [27] Liu H.Q. et al 2014 Rev. Sci. Instrum. 85 11D405 [28] Yang L., Baonian W., Junyu Z., Qingsheng H., Yanqing J., Xiaoqi X., Xiaofeng H. and Qing Z. 2010 Plasma Sci. Technol. 12 284 [29] Li Y.Y. et al 2016 Rev. Sci. Instrum. 87 11E501 [30] Huang J. et al 2020 Nucl. Fusion 60 016002 [31] Fu J. et al 2022 Plasma Phys. Control. Fusion 64 095006 [32] Gao W. et al 2011 J. Nucl. Mater. 415 S391–4 [33] Liang L., Wei J. and Hu C. 2015 J. Nucl. Sci. Technol. 52 481–9 [34] Chen Y. 2011 Phys. Plasmas 18 062506 [35] Zhang L. et al 2011 Plasma Sci. Technol. 13 431 [36] Ollus P., Akers R., Colling B., El-Haroun H., Keeling D., Kurki-Suonio T., Sharma R., Snicker A. and Varje J. 2022 Plasma Phys. Control. Fusion 64 035014 [37] Suzuki S., Shirai T., Nemoto M., Tobita K., Kubo H., Sugie T., Sakasai A. and Kusama Y. 1998 Plasma Phys. Control. Fusion 40 2097 [38] Kim J.Y., Rhee T., Kim J., Yoon S.W., Park B.H., Isobe M., Ogawa K. and Ko W.H. 2016 AIP Adv. 6105013 [39] Vianello N. et al 2020 Nucl. Fusion 60 016001 [40] Redl A. et al 2024 Nucl. Fusion 64 086064 15