Visualization of phase-space orbit topological boundary using imaging neutral particle analyzer

A newly-developed Imaging Neutral Particle Analyzer (INPA) in the DIII-D tokamak interrogates phase space occupied by fast ions on multiple different orbit topologies, including passing, stagnation, trapped and potato orbits. Depending on plasma parameters and beam injection geometries, this new INPA system is capable of visualizing distributions of fast ions on the selected orbit topology and its associated orbit topological boundaries. More importantly, the system is able to directly visualize the effective pitch angle scattering ν eff in phase space by measuring fast ions that are scattered across the trapped-passing orbit topological boundaries and from counter-passing orbits to co-passing orbits. It also enables visualization of fast ion confined-loss boundaries and resolves the change of the boundary in phase space, as plasma equilibrium evolves. The key goal of this new INPA system is to directly measure ν eff across phase space induced by drift waves and its interaction with Alfvén eigenmodes, i.e. a key issue towards a future fusion power plant.


Introduction
Sufficient confinement of fast ions, such as α particles, is crucial to sustain deuterium-tritium burning plasmas in future fusion reactors.Understanding the birth, thermalization and migration of fast ions requires accurate measurements of the 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.
fast ion population at a broad range of local phase space over multiple orbit topological boundaries.
The Imaging Neutral Particle Analyzer (INPA) diagnostic, which measures fast neutrals from charge exchange reactions between fast ions and actively-injected beam neutrals, was firstly developed and validated in the DIII-D tokamak [1][2][3].The system had great success in precisely measuring confined fast ions on passing orbits and resolving the fast ion distribution via a tomographic inversion technique [4,5].It also resolved the phase space flow, driven by multiple small-amplitude Alfvén Eigenmodes (AE) [6,7] and the observed phase-space flow is well interpreted by a nonlinear kinetic Magnetohydrodynamic (MHD) hybrid simulation [8].More recently, INPA systems have been developed worldwide.INPA is successfully commissioned in ASDEX-upgrade [9] and systems are under development in Large Helical Device (LHD), TCV, EAST and HL-2M.To date, all of the INPA systems focus on measurement of fast ions on passing orbits.
In this paper, the development of the second INPA system at the DIII-D tokamak is reported, which interrogates a small pitch |v ∥ /v| ∼ 0.45 (v ∥ refers to the fast ion velocity parallel to the plasma current).The system provides energy and pitch-resolved, radial density profiles of local fast ions on multiple orbit topologies, including trapped, potato, passing, and stagnation orbits.As with the first INPA, the new instrument provides precise measurement of the confined fast ion distribution but, in addition, the new system is particularly well-suited for visualization of phase space orbit topology and related physics phenomena.Depending on plasma configurations and beam geometries, the system is able to visualize the trapped-passing orbit topological boundary in phase space and monitor fast ion flux across the trapped-passing boundary via effective pitch angle scattering (ν eff ) processes.It should be noted that ν eff is not only attributed to Coulomb scattering, but also possibly induced by drift waves, wave-particle interactions, MHD-instability-induced perturbations, etc.The new system is also capable of visualizing the confined-loss orbit boundary in phase space, as it evolves with plasma current and fast ion energies.
It should be pointed out that these new measurement capabilities are critical to address several long-standing, open questions in energetic particle physics towards future fusion power plant.For example, theory predicts that scattering of resonant fast ions away from phase-space island structures by drift waves would play an important role in nonlinear saturation of AE instabilities and thus, impacts on fast ion transport [10][11][12][13][14][15].The intermittent transport of fast ions is also thought to potentially be the result of thermal plasma damping rate randomness for AEs, induced by background drift wave noise [16].These theories are difficult to validate in experiment, due to the lack of measurement capability.Moreover, the capability to visualize the confined-loss orbit boundary is important for the evaluation of velocity space gradients and the understanding of the drive of MHD instabilities like Energetic particle driven Geodesic Acoustic Modes (EGAM) [17][18][19][20][21].It could also contribute to the understanding of radial electric field formation and plasma rotation near the plasma edge, which often relates to orbit losses and/or redistribution of fast ions towards the first wall [22][23][24].
In section 2, we briefly discuss the principle of the INPA system and the design of the new system for the measurement of the fast ions on multiple orbit topolgies, especially for those on trapped orbits.Section 3 shows the visualization of ν eff across phase space, observed by the INPA2 system.Section 4 demonstrates the capability of the INPA2 system to visualize the confined-loss boundary in phase space.The summary and discussion appear in section 5. Hereafter, the newly-developed second INPA is referred to as 'trapped INPA' or 't-INPA', since this system is unique to interrogate fast ions on trapped/potato orbits in the DIII-D tokamak.

Principle and design
The principle of the INPA system is systematically reported in [1,2,4].A brief discussion is given here, as follows.As deuterium fast ions traverse an active probe neutral beam, the neutralization of the fast ions may occur through the charge exchange process.The resultant fast neutrals are not confined by magnetic fields and travel in a straight line with the energy and direction they had upon neutralization.The neutrals with trajectories that match the solid angle of the views of the INPA arrive at stripping foils and are possibly ionized by the stripping carbon foil with a thickness of 10 nm, as shown in figure 1(c).The incident ionized neutrals are then deflected by the Lorentz force and are dispersed onto the phosphor before completing a gyroorbit.The strike position is determined by the mass and energy of the neutral, the angle between the neutral trajectory and the stripping foil plane and the local magnetic field direction and strength.After absorbing energy from the incident particles, the scintillator emits photons that are collected by the CCD camera.
As seen from figures 1(a) and (b), the t-INPA, located ∼62 cm below the midplane, views the diagnostic beam (33R beam) upward.Compared to the views of the first INPA (see the green colored region in figure 1(a)), which are more parallel to toroidal magnetic field B t , the views of t-INPA (see the blue colored region) are more perpendicular to B t .Therefore, the t-INPA system measures primarily fast ions on trapped orbits, which are populated efficiently by the beams more perpendicular to magnetic field lines, such as 33R, 30R, 15R and 21L beams, as labeled in figure 2.
The views of t-INPA are in the same toroidal direction with the injection direction of 33R beam, i.e.Counter-Clockwise (CCW) direction from the top view of the DIII-D.This is opposite to the other INPA system, as seen in figure 1, which views the Clockwise (CW) direction.That is, when the plasma current is in the CCW direction (so called normal-Ip configuration for the DIII-D tokamak), t-INPA measures fast ions traveling in the counter-I p direction.When the plasma current is in the CW direction (so called reverse-Ip configuration), t-INPA measures fast ions traveling in the co-I p direction.
Adjusting the I p direction allows the t-INPA system to diagnose different orbit topologies and image different orbit topological boundaries in phase space.

Visualization of ν eff in reverse-I p plasma
Figure 3(a) shows the guiding center motion in a reverse-I p plasma of 80 keV fast ions that are collected by the t-INPA system at different major radii.The typical injection energy of neutral beams in DIII-D is 75-81 keV.The system measures fast ions on two types of orbits: (1) fast ions on co-I p passing orbits in the plasma core (see blue curves in figure 3(a)), which are populated by neutral beams that inject in the same toroidal direction as the plasma current (21R and 21L beams); (2) fast ions on co-I p leg of trapped orbits (see red curves in figure 3(a)), which are populated by all neutral beams, but more efficiently from neutral beams more perpendicular to the magnetic field, such as the 33R, 30R, 15R neutral beams in figure 2. In this case, fast ions are born on the counter-I p leg of trapped orbits, travel radially outward until a charge-exchange reaction occurs near the midplane on the co-I p leg, and at last, escape from the plasma as fast neutrals that are detected by the t-INPA system.
Figure 3(b) shows an orbit topology map of fast ions at 80 keV as the function of energy and major radii on the midplane of the DIII-D tokamak.The interrogated pitch of the t-INPA is illustrated by the white asterisk markers.It is found that the t-INPA measures well-confined fast ions on trapped orbits from R ∼ 2.05 m to ∼2.1 m.At R < 2.05 m, t-INPA interrogates well-confined fast ions on co-I p passing/stagnation orbits.It should be emphasized that the only beams that directly populate the phase space occupied by fast ions on co-I p passing orbits in reverse-I p plasmas are the 21L and 21R beams, as shown in figure 2. That is, if 21L and 21R beams are switched off, only fast ions on trapped orbits contribute to the t-INPA signal.
The t-INPA image without 21L and 21R beams is shown in figure 4(a1).The dominant signal, i.e. camera counts >1500, appears outside R ∼ 2.0 m and a noticeable signal, i.e. camera counts <500, is seen at R < 1.95 m.The weak and strong signals are separated by the orbit topological boundary between trapped and passing orbits, which is indicated by the white dotted lines in figure 4. The orbit topological boundary is identified by computing guiding center motion of fast ions over the interrogated phase space using ASCOT5 code [25], using reconstructed plasma equilibria at the corresponding timings.The uncertainty of the radial locations of the boundary is ∼±3.5 cm, which is attributed to the gyro-orbit size of the fast ions and spatial resolution of the system due to the finite width of the diagnostic beam, pinhole size, etc.
The observed t-INPA image in figure 4(a1) is similar to the neutral flux towards the t-INPA pinhole (integrated over vertical direction), calculated by FIDASIM code [26], as seen from figure 4(a2).This calculation of neutral flux uses fast ion distribution from NUBEAM module of TRANSP code [27] without considering fast ion transport induced by MHD instabilities and drift waves.The synthetic image of the t-INPA system in figure 4(a3) is also computed by FIDASIM code and INPASIM [4].INPA synthetic image for any given fast ion distributions F can be constructed by a convolution integral, S syn = W#F, where W is the instrumental weights.The computation of W is documented in detail in [4].Note that the calibration of the INPA camera counts to absolute value of neutral flux has not been done and therefore, the quantitative comparison between the synthetic images and measured images is not possible at present.Nevertheless, the pattern of calculated synthetic image roughly matches the measured image, except for low-energy region of E < 40 keV.The deviation could be due to the assumption of linear ionoluminescent response on the incident fast ion energy for a given flux in INPASIM [4].It could also be related to EGAM and/or low-frequency broadband fluctuation, which are observed by CO 2 interferometer.The possible connection of this discrepancy to MHD activities and/or drift waves will be addressed via future experiments.
For comparison, the t-INPA image after adding 21R beam is shown in figure 4(b1).It is observed that the signal is significantly enhanced at the full and half energy of the injected fast ions across the trapped-passing orbit topological boundary.Consistently, this phase space near the plasma core is occupied by fast ions on co-I p passing/stagnation orbits, populated only by 21 beams.The sightline-sampled neutral flux and corresponding synthetic images of t-INPA also show the increased fast ion density and fast neutral density near the plasma core, as seen from figures 4(b2) and (b3), respectively.The results are qualitatively consistent with the measurement.
One interesting finding is that a noticeable amount of fast ions are observed near the plasma core in figure 4(a1), i.e. the phase space occupied by fast ions on co-I p passing orbits, even when all neutral beams are injected in the counter-I p direction without 21 beams.To better understand the origin of the signal, a time series of t-INPA images after switch-on of 33R beam is given in figure 5.The waveform of the neutral beam, along with the sampled timing of the INPA images (a)-(e), is shown in figure 5(f ).It is seen from figure 5(a) that the signal at R < 1.95 m, related to fast ions on co-I p passing orbits, is at or below the noise level.On the other hand, the signal related to fast ions on trapped orbits are clearly observed at ∼0.314 s, 14 ms after the switch-on of the 33R beam.As time evolves, the signal near the injection energy is enhanced by continuous fueling of the 33R beam.Interestingly, an enhanced signal firstly appears below ∼40 keV close to the trapped-passing orbit topological boundary, and gradually expands toward the plasma core and higher energy, as the electron temperature increases (see the white arrows in figures 5(b)-(e)).This is consistent with the increasing relative importance of ion drag and pitch angle scattering, as increasing Te raises the critical energy where electron and ion drag are equal.A numerical experiment is conducted to further clarify this effect and the details are given, as follows.
To understand the physics reason of the appearance of the weak signal in plasma core (indicated by white arrows in figure 5), the orbits of a slowing-down fast ion is studied by ASCOT5 code, using measured plasma profiles and equilibrium but modified T e profiles.Figure 6(a) shows the orbit of a fast ion, which is launched at R = 2.1 m, Z = 0 m and slows down from 80 keV to 25 keV in 5.8 ms in a plasma with 10% of measured T e .Due to the reduced critical beam energy, the The waveform of the injected neutral beam is given in f, along with temporal evolution of electron density, measured by CO2 interferometer and electron temperature near the magnetic axis (channel 24), measured by electron cyclotron emission, respectively.The white arrows show the weak signal firstly appears in the phase space below ∼40 keV, and gradually expands toward the plasma core and higher energy, as the electron temperature increases.fast ion is thermalized by electron drag.So negligible pitchangle scattering, caused by Coulomb collisions, occurs.In such a case, both the co-I p leg and counter-I p leg of trapped fast ions moves radially inward, due to a reduced energy and drift orbit width.However, the pitch of the counter-I p leg remains negative during the entire slowing-down process, as seen from the red line in figure 6(c) and thus, it is not detectable by t-INPA system.That is, if ν eff is negligibly small, the orbit change due to the thermalization of fast ions on trapped orbits will not contribute to the t-INPA signal in the phase space occupied by fast ions on co-passing orbits in plasma core.As a comparison, figure 6(b) shows the orbit of a fast ion, which is launched at R = 2.1 m, Z = 0 m and slows down from 80 keV to 51.3 keV in 300 ms in a plasma with ten times of the measured T e .The critical beam energy is well above the injection energy of 80 keV and the fast ion is thermalized by ion drag.So significant pitch-angle scattering by Coulomb collisions occurs.In this case, the orbit sometimes changes from trapped orbit to co-passing orbits.The pitch in the High Field Side (HFS) transiently changes from negative value to positive value.Therefore, the fast ion enters the INPA interrogated pitch, as seen from the black line in figure 6(c).That is, if ν eff is significant, fast ions on trapped orbits can be scattered to co-passing orbits and contribute to the t-INPA signal in plasma core.It is demonstrated that the system is able to directly visualize the strength of ν eff in phase space by measuring fast ions that are scattered across the trapped-passing orbit topological boundaries.
The other source of fast ions, which may also contribute to the t-INPA signal in the core, is the counter-passing fast ions populated by 33R beam.These fast ions on counter-passing orbits could also be scattered from negative pitch to positive pitch through ν eff .It should be noted the change of magnetic configuration, and birth profiles will not vary the traveling direction of fast ions, since it is only related to the injection orientation of neutral beams.
Figure 7 shows the temporal evolution of the measured profiles of fast neutrals, integrated over the low energy range from 30 keV to 40 keV and over the high energy range from 65 keV to 75 keV.The radial profiles are more flattened in the low energy range, compared to that in the high energy range.This could be caused by pitch angle scattering due to Coulomb collision, which, according to neoclassical theory, occurs more frequency for low-energy ions that slow down primarily on thermal ions than for high-energy ions that slow down primarily on electrons.Note that it could also be impacted by drift waves and MHD instabilities, which is sensitive to fast ion energies too.Clarification on the contribution of each physics mechanism on the observed ν eff is left for future dedicated experiments.Overall, the data show a clear inward phase-space diffusion process across the trappedpassing orbit topological boundary via effective scattering processes.

Visualization of confined-loss boundary in normal-I p plasma
Figure 8(a) shows the guiding center motion in a normal-I p plasma of 80 keV fast ions that are collected by the t-INPA system at different major radii.The system collects fast neutrals that travel in the direction opposite to the plasma current, including: (1) fast ions on counter-I p passing orbits that are populated by neutral beams that are injected opposite to the plasma current direction, so called counter-I p beams (21R and 21L beams); (2) fast ions on the counter-I p leg of trapped orbits and on prompt loss orbits that connect to the first wall.Fast ions that are populated by the 33R, 30R, 15R neutral beams are born on the co-I p leg of trapped orbits, then travel radially inward until a charge-exchange reaction occurs near the midplane on the counter-I p leg of the orbit, then escape as fast neutrals towards the t-INPA system.In contrast, fast ions that are populated by the 21R and 21L neutral beams are born on the counter-I p leg of trapped orbits and escape immediately after ionization via a charge-exchange reaction on the counter-I p leg near the midplane.
Figure 8(b) shows an orbit topology map of fast ions at 80 keV as a function of pitch and major radii on the midplane of the DIII-D tokamak in a high-q min plasma (q min : minimum safety factor ), overlaid with interrogated pitch of the t-INPA (see the white asterisk markers).In this specific case, for R > 1.9 m, t-INPA measures a portion of phase space that is occupied by fast ions on loss orbits.The signal from fast ions on loss orbits is negligibly small, because (1) fast ions, populated by 33R, 30R, 15R neutral beams, will hit the wall immediately after the ionization without crossing the t-INPA views and (2) even though fast ions, populated by 21R and 21L beams, are born near the t-INPA views, they cannot circulate along the torus for sufficient time to make a charge-exchange reaction likely.
Deep into the plasma core, the system measures wellconfined fast ions on potato orbits and trapped orbits at R < 1.9 m and measures well-confined counter-I p passing fast ions on the HFS.
Figures 9(a)-(c) show three t-INPA images at the q min ∼ 3.2, ∼2 and ∼1, respectively.The images are obtained during the plasma current ramp-up and q min drops from ∼3.2 at 0.39 s to ∼1 at 1.05 s, measured by Motional Stark Effect (MSE) diagnostics [28], as seen from figure 9(e).In this experiment, the multiple right beams, i.e. 33R, 15R and 21R beams, are injected at a reduced energy of 35 keV and the 30L beam is injected at 81 keV to probe the confined-loss boundary in phase space, as shown in figure 9(f ).
It is found that, although the injection energy of the 30L beam is up to ∼81 keV, the t-INPA does not observe the signal near the injection energy at 0.39 s, as seen in figure 9(a).On the other hand, at ∼0.6 s and ∼1.0 s with comparable n e , the signals near the injection energy are clearly observed in figures 9(b) and (c).In addition, a clear 'gap' (see the region labeled as 'Lost' in figures 9(a)-(c)) is noticeable near the plasma major radii of R ∼ 2.0 m.The radial width of the gap varies with energy.That is, the higher the energy is, the wider the gaps are.This could be explained by the fact that higher energy E FI fast ions have a wider drift orbit width ∆, i.e. ∆ ∝ q √ E FI /B t and thus, a deeper loss cone in phase space.Note that such gaps are not observed in the t-INPA images obtained in reverse-I p plasma.
Full orbits of fast ions over the interrogated phase space are computed by ASCOT5 code, using reconstructed plasma equilibrium.The confined-loss boundary in INPA-interrogated phase space is marked as the dashed lines in figures 9(a)-(c).The result suggests that fast ions at the injection energy of 81 keV are not confined at q min ∼ 3.2 at 0.39 s at the interrogated pitch.
As q min decreases to 2, fast ion confinement is significantly improved near the injection energy.The confinedloss orbit topological boundary moves radially outward and to higher energy.The predicted loss cone in phase space is well aligned with the observed gap structure in figure 9(b).
As q min is further reduced to 1, figure 9(c) shows that the fast ion loss region further shrinks in the t-INPA interrogated phase space.The observed fast ion loss boundary as measured by the t-INPA agrees well with the simulation.
It is noted that that the INPA image does not align with the confined-loss boundary well at ∼388 ms in figure 9(a).The reason is not fully understood.The charge exchange reaction between fast ions populated by 30L beam and edge cold neutrals, so called passive signal, is not responsible for the observed discrepancy.Figure 9(d) shows the passive signal from left beams (15L and 33L beams without 33R beam) is quite small, which is measured at 0.293 s.One possibility is the inaccuracy of the reconstructed safety factor at the very early period of the discharge, hinting at a possible constraint of equilibrium reconstruction using t-INPA data.

Summary and discussion
A new INPA system, which is capable of measuring multiple orbit topologies, has been successfully developed in DIII-D tokamak.Like the first INPA developed in DIII-D tokamak [1,2], the t-INPA is able to precisely measure confined fast ions, especially for those on trapped orbits.
Previous studies of pitch angle scattering, using conventional NPAs [29], often monitor a very small phase volume (single pitch, radial location and energy), which is barely perturbed by a distant source of neutral beam.The detector observes a small amount of the neutral flux from this specific local phase-space volume after the slowing down of fast ions [30].The other study in LHD investigates the pitch angle scattering via the decay rate of neutral flux, which is from charge exchange reaction with cold neutrals in plasma peripheral region, using a beam blip technique [31].Even though t-INPA measures a nearly constant pitch, the system is able to directly visualize ν eff in the plasma core over a broad range of phase space volumes, showing a significant advance in the measurement capability.This is achieved by populating fast ions on confined trapped orbits and counter-passing orbits, and measuring their migration towards co-passing orbits in plasma core in reverse-I p plasma configuration of the DIII-D tokamak.It is an inward diffusion across phase space possibly induced by coulomb collision, drift waves, and MHD instabilities.
In the normal-I p plasma configuration, the t-INPA directly measures the time evolution of confined-loss boundary of fast ions in phase space during the plasma current ramp-up with time resolution of 6.25 ms.It should be noted that the confinedloss boundary is also measured by conventional NPA system [32,33], which is achieved by installing a horizontal or vertical scanning system [34].For exmaple, the typical scanning rate of conventional NPA in LHD is 0.03 • s −1 [34].Therefore, it is not able to resolve the fast evolution of confined-loss boundary in the time scale of MHD events and current diffusion time.
These new measurement capabilities, provided by the t-INPA, will play a key role in addressing critical issues in energetic particle physics towards fusion power plant in future DIII-D experiments.For example, the t-INPA system has been used to directly measure ν eff across orbit topological boundaries for different amplitudes of drift waves.The result will be reported in a separated paper.A strict comparison with Coulomb scattering based on neoclassical theory will help distinguish the contribution to effective scattering in phase space from drift waves or other sources, if any.The system will also be used to measure AE-driven phase-space flow in trapped orbit phase space, using the beam modulation technique [6,7,35].A well-diagnosed confined-loss boundary in phase space could also be a powerful tool to diagnose q profiles, similar with the previous estimation of q by the measurement of drift orbit widths reported in [36,37].The design of an optimized INPA system for the measurement of q profiles deserves future study.

Disclaimer
This report was prepared as an account of work sponsored by an agency of the United States Government.Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights.Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof.The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof.

Figure 1 .
Figure 1.(a) 3D views of the t-INPA and 'old' INPA, which monitors fast ions on passing orbits in the DIII-D tokamak, along with diagnostic beam (33R).(b) The view of the t-INPA at its lower port inside the vacuum vessel.(c) The layout of phosphor and carbon foils of the t-INPA system.(d) The t-INPA diagnostics before the installation.

Figure 2 .
Figure 2. Injection geometry of neutral beams, along with t-INPA views (black line overlaid on 33R) and definition of the direction of plasma current in DIII-D tokamak.Here, 'R' and 'L' refer to 'right' and 'left', respectively, as viewed towards the neutral sources.

Figure 3 .
Figure 3. (a) The orbits of fast ions at 80 keV, which are collected by t-INPA system in a reverse-Ip plasma.Red curves represent fast ions on trapped orbits and blue curves represent fast ions on co-Ip passing orbits.The equilibrium is reconstructed from shot 192693 at 1 s.(b) Orbit topology of fast ions at 80 keV, which are launched from pitch −1 to 1 and major radii from 1.3 m to 2.3 m on the midplane in reverse Ip plasma configuration.The plasma equilibrium is reconstructed for shot 192693 at t = 880 ms.The pitch interrogated by the t-INPA system is labeled by white crosses in (b).

Figure 4 .
Figure 4. (a1) The image of t-INPA for shot 192693 at 976 ms without the injection of the co-Ip beam (21R); (a2) Neutral flux along the sightlines towards t-INPA pinhole, for shot 192693 at 976 ms.The flux is integrated over vertical direction.The distribution is calculated by NUBEAM module of TRANSP code and FIDASIM code.(a3) The synthetic image of the t-INPA using calculated neutral flux in (a2).(b1) The image of t-INPA for shot 192693 at 1050 ms with the injection of co-Ip beams; (b2) corresponding neutral flux towards t-INPA system.(b3) The synthetic images of the t-INPA using FIDASIM-calculated neutral flux in (b2).(c) Temporal evolution of density, measured by CO2 interferometer, electron temperature near the magnetic axis (channel 24), measured by electron cyclotron emission and the waveform of the voltage of injected neutral beams.The timings of the t-INPA images (a1) and (b1) are labeled as green vertical lines in (c).

Figure 5 .
Figure 5.The t-INPA images of shot 192693 at 0.314 s, 0.326 s, 0.34 s, 0.351 s, 0.364 s in (a)-(e), respectively.The boundary of trapped-passing orbits are overlaid in black dotted lines, which are estimated by computing guiding center motion of fast ion orbit in INPA interrogated phase space, using ASCOT5 code.The corresponding timings of the images are also indicated by vertical lines in (f ).(f )The waveform of the injected neutral beam is given in f, along with temporal evolution of electron density, measured by CO2 interferometer and electron temperature near the magnetic axis (channel 24), measured by electron cyclotron emission, respectively.The white arrows show the weak signal firstly appears in the phase space below ∼40 keV, and gradually expands toward the plasma core and higher energy, as the electron temperature increases.

Figure 6 .
Figure 6.The orbit of a slowing-down fast ion, which is launched at R = 2.1 m, Z = 0 m, energy of 80 keV and pitch of 0.44, in a plasma with 10% of measured Te (a) and in a plasma with 10 × measured Te (b).The fast ion in (a) is slowed down from 80 keV to 25 keV in 5.8 ms.The fast ion in (b) is slowed down from 80 keV to keV in 300 ms.The pitch of fast ion in (a) and (b) is recorded on the midplane for both high field side (HFS) and low field side (LFS), and the time evolution of the recorded value is given as the red and black curves in (c), corresponding the fast ion in (a) and (b), respectively.

Figure 7 .
Figure 7. Temporal evolution of the fast neutral profiles over major radii, integrated over energy from 30 keV to 40 keV in (a) and from 65 keV to 75 keV in (b).The shaded regions in (a) and (b) indicate the trapped-passing orbit topological boundary estimated by ASCOT5 code.

Figure 8 .
Figure 8.(a) The orbits of fast ions at 80 keV, which are collected by t-INPA system in normal-Ip plasma, using the reconstructed equilibrium at 400 ms in shot 194784.Red curves represent fast ions on trapped orbits and blue curves represent fast ions on counter-Ip passing orbits.(b) Orbit topology of fast ions at 80 keV, which are launched from pitch −1 to 1 and major radii from 1.3 m to 2.3 m on the midplane in normal Ip plasma configuration.The plasma equilibrium is from shot 194784 at t = 400 ms.

Figure 9 .
Figure 9. (a)-(c) A time series of the t-INPA images in shot 194782 (normal Ip) during the decrease of the minimum safety factor.The confined-loss boundary, which is calculated by ASCOT5 code, is indicated by the white dashed lines.(d) The t-INPA image at 0.293 s is from the charge-exchange reaction with cold edge neutrals, when the 33L beam and 15L beam are switched on and the diagnostic beam 33R beam is off.The saturated 'red' color, appeared as the straight lines in the images (a), (c) and (d), are the noise.The timing of images (a)-(c) is indicated by vertical lines in (e), along with the time evolution of electron density, measured by CO 2 interferometer and q min , measured by MSE.(f ) Waveform of the injected neutral beam voltage.