Demonstration of transient CHI startup using a floating biased electrode configuration

Results from the successful solenoid-free plasma startup using the method of transient coaxial helicity injection (transient CHI) in the QUEST spherical tokamak (ST) are reported. Unlike previous applications of CHI on HIT-II and on NSTX which required two toroidal insulating breaks to the vacuum vessel, QUEST uses a first of its kind, floating single biased electrode configuration, which does not use such a vacuum break. Instead, the CHI electrode is simply insulated from the outer lower divertor plate support structure. This configuration is much more suitable for implementation in a fusion reactor than the previous configurations. Transient CHI generated toroidal currents of 135 kA were obtained. The toroidal current during the formation of a closed flux configuration was over 50 kA. These results bode well for the application of transient CHI in a new generation of compact high-field STs and tokamaks in which the space for the central solenoid is very restricted.

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.

Introduction
As a result of its small aspect ratio, defined as the ratio of the major radius to the minor radius of the plasma torus, the spherical tokamak (ST) is capable of simultaneous operation at high beta and high bootstrap current fraction.However, due to the very restricted space available for a central solenoid in an ST, the capability for initiating the plasma discharge without reliance on a central solenoid is an important design consideration [1].Recent design studies for a compact, major radius = 3 m, tokamak configuration also show that some solenoid-free plasma startup capability is necessary for the system to generate net electricity while satisfying the need for tritium self-sufficiency [2].In support of this development, a number of solenoid-free plasma startup methods are/have been studied.These are, Wave current drive on COMPASS-D [3], QUEST [4], TST-2 [5], and LATE [6], Helicity Injection on HIST [7] and PEGASUS [8], Merging Reconnection on START [9] and MAST [10,11], induction from the outer poloidal field coils [12].
Transient coaxial helicity injection (transient CHI) is a promising solenoid-free current start-up method, first developed on the HIT-II [13] experiment at the University of Washington, and then on the NSTX device at PPPL [14].Transient CHI discharges on NSTX have been coupled to subsequent inductive drive resulting in solenoid flux savings [15].As shown in figure 1(a), a transient CHI discharge is generated by driving injector current along magnetic field lines (the injector flux) that connect the inner and outer divertor plates on one end of the ST.On both these devices, the inner and outer vessel walls were electrically insulated from each other by inserting large toroidal ceramic breaks in the top and bottom of the vessel vacuum so that the entire inner and outer walls of the ST could be used as the electrodes.
Installation of large vacuum ceramic insulators may not be possible in a fusion reactor so an alternate configuration is desirable for the transient CHI electrodes [16].In addition, the insulator must be protected from neutron irradiation.The study described in [16] was to develop the insulator concept to address both of these issues.The study showed that in reactor designs such as the ones that rely on advanced divertor concepts, the divertor coils are ideally situated to generate a large magnitude of magnetic flux needed for CHI operation.In addition, it should be possible to use the blanket structure to shield the insulator from neutron damage.In the conceptual design reported in [16], neutronics calculation shows the calculated dose on the insulator to be ∼10 10 Gy at 6 full power years, which is much less than the dose limit of 10 11 Gy for the MgO insulator that would be used for this application.The QUEST CHI effort was to show that transient CHI discharges are compatible with such an insulator concept.
The QUEST device uses a floating single biased (FSB) electrode configuration in which one of the divertor plates is insulated from the rest of the vessel using a toroidally continuous (but segmented) non-vacuum insulator that separates the electrode plate from the outer divertor plate support structure, as shown in figure 1(b) and described in [17].Thus, no toroidal vacuum breaks are required in the vacuum vessel.As a result, the method could also be implemented on any currently operating ST or tokamak, as only one of the divertor plate area needs to be insulated from the divertor plate support structure.A primary objective of the transient CHI experiments on QUEST is to assess if a FSB electrode configuration can generate a transient CHI discharge that is similar to those generated on HIT-II and NSTX.
QUEST is a mid-size ST device with major radius R 0 = 0.64 m, minor radius a = 0.4 m, and toroidal field (TF) B t0 ≤ 0.25 T at R 0 [4].The first application of CHI on QUEST used a configuration referred to as low field side injection in which the CHI electrode was biased with respect to the outer vessel wall of QUEST [17].In this configuration, as the CHI plasma filled the vessel, it was found that the injector flux footprint locations on the electrodes moved away from each other [18].As described below, this condition is not suitable for the formation of closed flux surfaces.Biasing the CHI electrode with respect to the inner wall in a configuration referred to as high field side (HFS) injection was a potential way to resolve this issue.However, as the required modifications to the QUEST divertor region were substantial, to assess its potential, in a subsequent study, a simpler version of the HFS configuration was tested [18].While this configuration demonstrated discharge evolution into the vessel, and suggested the potential for closed flux formation, it was plagued by spurious arcs.Nevertheless, it provided confidence that a redesign of the lower divertor region for an improved HFS configuration should work.Figure 1(b) shows the details of the resulting improved CHI injector design used in the experiments described here.
An annular metal plate referred to as a bias electrode is installed on top of the baseplate used to support the original divertor plate.A metallic cylinder is attached to the bias electrode as shown in figure 1(b).This cylindrical metallic electrode is used as the outer cylindrical electrode.This cylindrical electrode, which acts as the cathode, is biased with respect to another cylindrical electrode that is adjacent to the center stack and maintained at ground potential (the anode).The bias electrode is insulated from the baseplate (and from the QUEST vacuum vessel) by sandwiching a ceramic plate between the electrode (which is on top) and the baseplate (which is below the ceramic plate).Another cylindrical ceramic insulator separates the outer cylindrical electrode from the vessel components below the baseplate.The ceramic plate and the cylindrical ceramic insulator shown in figure 1(b) are the primary CHI insulators.A gas manifold located between these two cylindrical electrodes is used to inject fuel gas in-between the two cylindrical electrodes, which is the CHI injector region.The PF5-1 coil located below the injector region generates the injector flux that connects the metallic cylindrical electrodes that are at opposite potential with respect to each other.The PF5-2 coil is used to shape the injector flux so that most of the injector flux connects the two cylindrical electrodes.For these experiments, an insulator plate is added to the top of the CHI electrode to ensure that the CHI discharge is primarily confined to the region near the cylindrical electrodes.In a future variation, this region could be covered by a floating metallic plate, as described in [19].
In a CHI discharge as the toroidal field is increased, the increased spiraling action of the field line increases its connection length.As described in [20,21], the initial magnetic field line configuration for a CHI discharge results in field lines with a short connection length in the injector region and much longer connection lengths in other parts of the vessel.Paschen's law for gas breakdown requires that the combination of the magnetic field line length and gas pressure satisfy a minimum threshold level at a given applied voltage.Thus, a higher gas pressure is needed in the injector region than elsewhere in the vessel.If the gas breakdown requirements are not adequately satisfied in the injector region, the injected gas dispersing into the vessel can initiate a discharge elsewhere in the vessel, as these regions require much less gas pressure.To avoid this issue, the gas injection system requires a gas manifold (shown in figure 1) that can delay the injected gas dispersing into the vessel for a sufficiently long duration to permit the discharge in the injector region to take place.An advantage of transient CHI is that the applied high voltage pulse duration is very short, about 0.5 ms.Beyond this time, the applied voltage drops to a low level.Even if the gas were to disperse into other parts of the vessel, such as in the outboard region, it would not be able to generate a substantial arc that can drain energy from the primary CHI plasma discharge, as much of the energy in the suitably sized capacitor is drained during the primary current pulse in the injector region.These are described in more detail in [19].
On QUEST, a transient CHI discharge is initiated as follows.In the presence of an external toroidal field (TF), the PF5-1 coil is energized to generate a poloidal magnetic field that connects the inner and outer electrodes in the injector region.This is the injector flux.Currents are driven in the PF5-2 coil to ensure that most of the injector flux intersects the vertical section of the outer floating electrode.Gas is then injected through the gas manifold, and voltage is applied to the electrodes by discharging a 5 mF capacitor bank charged to about 2000 V.This causes the gas in the injector region to be ionized, which results in currents flowing along the injector field lines.This externally driven current is the injector current.If the injector current is above a threshold level, then the forces from the interaction of j pol × B tor , exceed a condition known as the bubble burst condition [22].This would cause the injector flux to stretch into the vacuum vessel.The bubble burst condition is derived by balancing the magnetic field line tension of the injector flux with the j pol × B tor force.It describes the relationship between injector flux and the injector current I inj required to overcome the restraining magnetic field line tension of the injector flux.The bubble burst current requirement is given by the relation Here ψ inj is the injector flux, I TF is the current through the center stack of the ST, µ 0 is the vacuum magnetic permeability, and 'd' is the separation distance between the injector flux footprint locations shown in figure 1.During the transient CHI process, the injector current supplied by the external power supply is quickly reduced.If by this time, the injector flux has filled most of the vessel and the separation distance between the injector flux footprints is sufficiently small, then the proximity of the oppositely directed field lines in the narrow region (figure 1) would make it easier for them to reconnect in the region marked with a star in figure 1(a).During this reconnection process, the magnetic field injected into the vessel can disconnect from the main injector flux generating a closed magnetic field flux configuration inside the vessel.
In this letter, we report on the first successful demonstration of this process, which generates a closed flux configuration, using, a first of its kind, FSB electrode configuration.Section 2 describes the experimental results.Section 3 is a summary of the main results.

Experimental results
Using the improved HFS configuration shown in figure 1(b), stable CHI discharges were achieved under high injector flux conditions without the generation of spurious arcs observed in the previsions HFS configurations.Figure 2(a) shows current traces from a transient CHI discharge with 23 mWb of injector flux.0.72 Pa m 3 of hydrogen fuel gas was injected through the gas manifold.The toroidal field at the machine major axis was 0.25 T. The central solenoid was not used during these experiments.A 5 mF capacitor charged to 1925 V was discharged across the injector electrodes.This caused the injected gas to breakdown, forming a plasma and causing currents to flow along the injector flux field lines.The combination of the toroidal and poloidal magnetic fields causes the injected current to develop a toroidal component.This is the initial phase of toroidal current formation during which currents flow solely on open field lines in the injector region.During the initial short spike in the injector current, and as a consequence of the bubble burst condition being satisfied on some of the injector flux field lines, the magnetic flux begins to the injected into the vessel.The bubble burst current requirement equation for these discharges requires that the effective parameter 'd' to be about 20 cm to inject 23 mWb of poloidal flux at an injector current of 20 kA.The distance between injector flux footprints can vary from 10.8 cm, which is the distance between the cylindrical electrodes, to over 20 cm at the top of the injector above the cylindrical electrodes.23 mWb is the total injector flux over the entire outer electrode.The injector flux starting from the part of the outer electrode closest to the horizontal plate on the gas manifold is 18 mWb, which requires a flux foot separation distance of 16 cm.The part of the injector flux below the horizontal plate on the gas manifold connects to the gas manifolds itself, inside the gas manifold cavity, with a much shorter flux separation distance and it is likely that this injector flux is not contributing to the useful injector flux.This suggests only a part of the injector flux that is closest to the vacuum vessel is probably being injected in these experiments.It should also be noted that the parameter 'd' is not a precisely estimable parameter, as it depends on which field lines carry the injector current and the bubble burst current requirements for those field lines.It is used as primarily as a guide to sizing the capacitor bank requirements for a given CHI system and needs further studies for this specific injector configuration.
The transient CHI discharge evolution has three phases.Phase 1 is the gas breakdown and the bubble burst phase.During this phase, the injector current is flowing in the injector region on open field lines.This phase corresponds to the initial short spike in the injector current.After that, during phase 2, as the injector flux stretches into the vessel, the plasma inductance increases causing the injector current to drop, and it remains at about 10 kA as the plasma fills the vessel.By about 19.25 ms, the energy in the capacitor has sufficiently depleted to a level that does not permit the injector current to be sustained at the 10 kA level, and it begins to decrease rapidly.During phase 3, which begins near time index t1, the injector current has dropped well below the level needed to satisfy the force balance relation in the injector region.This can result in the plasma within the vessel responding in two different ways.If the injector flux footprint width, the parameter d, becomes too wide, it would be easier for the plasma in the vessel to simply pull back into the injector region.However, if the injector flux footprint width is kept sufficiently narrow, then it is easier for the plasma within the vessel, through a process of magnetic reconnection in the region indicated by the arrows in figure 1(b), to disconnect from the main injector flux and form a closed flux configuration inside the vessel.
In these discharges, fast camera image frames show the resulting plasma to be a center stack limited discharge, as seen in frames corresponding to time indices t1 to t3 in figure 2(b).
In phase 1 the current multiplication factor, defined as the ratio of the toroidal current to the injector current, is below 10.During the injector current sustained phase 2, it increases to about 10 and the toroidal current increases to 135 kA.In phase 3, the current decay phase, the injector current decreases much more rapidly than the toroidal current which is seen in the current multiplication trace.At the beginning of the phase 3, the current multiplication factor increases because the injector current very rapidly decreases due to the depletion of the energy in the capacitor, after which the plasma within the QUEST vessel begins to disconnect from the injector region and begins to from closed flux surfaces at a time corresponding to t1.This boundary is marked with a red dashed line in the first camera image corresponding to time t1.Even when the slope of the decreasing injector current becomes gentle after the time t1, the current multiplication factor still increases because the toroidal current confined in the closed flux surfaces persists for longer time.The camera images corresponding to times t2 and t3 show this much more clearly as the plasma within the vessel has now moved much farther away from the injector region.The appearance of a clear D-shaped structure corresponding to a center stack limited plasma discharge, such as that normally observed during the inductive central solenoidinitiated plasma startup, is seen in these transient CHI imitated discharges.At the time t3, 52.6 kA of toroidal current is still present, while the injector current decreases to just 0.37 kA making the current multiplication factor well over 100.
In figure 3(a), the camera images corresponding to time indices t4 to t7 are also shown.The center stack limited plasma discharge decreases in size through resistive decay as the plasma current decreases.The reconstructed poloidal flux plots, based on a Grad-Sharfranov solver [23], are shown in figure 3(b).These use the measured toroidal plasma current, the poloidal field coil currents, and a central pressure constraint of 40 Pa.The present CHI discharge is generally low beta so the plasma current not plasma pressure plays the dominant role in the equilibrium reconstruction.While the pressure term is not important, we used the Thomson scattering measured values at the closest time point in figure 4 (to be described later) to be consistent.These are approximately similar to the observed camera images in that they also show the plasma resting on the center stack and are close to the size of the dshaped camera images.From the reconstructions, the plasma inductance L = 0.21µH.For the measured electron temperature of 6 eV in figure 4 (to be described later), and for a circular plasma, the plasma resistance, R, based on Spitzer resistivity, is 1.4 mΩ for the closed flux configuration at t4.The resulting L/R current decay time is roughly estimated to be in the range of 0.2 ms, which is approximately consistent with three e-folding times for the current to decay to zero.
In figure 4 we show Thomson scattering measured profiles on the mid-plane for a continuous sequence of nine similar discharges.An important observation is that these discharges are reproducible without the presence of the spurious arcs seen in previous CHI experiments on QUEST, and their reproducibility is similar to the observations on HIT-II and NSTX.The central electron density, temperature, and pressure were n e ∼ 3 × 10 19 m −3 , T e ∼ 8 eV, and p e ∼ 40 Pa, respectively.The change of the outer peak of the p e profile, 0.75 → 0.63 → 0.48 m corresponding to times t1, t2 and t3 is due to the shrinking plasma size at these times as seen in the fast camera images.
Slowing down the persisting current decay time by generating a higher electron temperature plasma requires that the plasma electron density be further reduced.The electron density in these discharges is much higher than those reported for the transient CHI discharges on NSTX [24].This believed to be due to higher levels of fuel gas injection used in these experiments (0.72 Pa m 3 on QUEST vs about 0.27 Pa m 3 on NSTX).Some additional improvements to the gas injection system are needed to reduce the amount of injected gas by about a factor of two, which should increase the electron temperature and reduce the toroidal current decay rate.

Conclusions
Experimental results from the QUEST ST show that solenoidfree plasma current startup using the transient CHI method is compatible with a FSB electrode configuration.In this configuration, only the CHI electrode plate is insulated from a divertor plate support structure.Unlike the earlier configurations used on HIT-II and NSTX, a toroidal break to the vacuum vessel is not needed.This configuration is much more compatible with a reactor neutron environment.In addition, it could also be deployed on currently operating STs and tokamaks to develop noninductive current startup and ramp up scenarios, to support emerging compact reactor designs.
An important aspect in regard to increasing the generated plasma current is the need for higher toroidal field [16].The benefit of the toroidal field is that, at a given injector flux, it reduces the amount of injector current needed to satisfy the bubble burst current.The generated plasma current is directly proportional to the injector flux [16].From a technology perspective, reducing the injector current is quite important as the increased current density on the electrode surfaces can lead to more influx of impurities.This is the primary reason transient CHI was successful for plasma startup applications, as the short current pulses were able to sufficiently reduce impurity influx.Arc damage to electrodes is not as issue, as the injector pulse duration and energy deposition are small compared to the divertor heat loads involved during normal long pulse tokamak operation.In fact, the post operation examination revealed no visible damages to the electrodes.The total capacitor bank energy in the CHI system in these experiments was just 10 kJ.These successful results were made possible due to a significant modification to the lower divertor structure on QUEST.Peak toroidal currents of 135 kA and persisting closed flux current of over 50 kA have been achieved in these experiments without the spurious arcs that plagued all previous CHI efforts on QUEST.The transient CHI plasma startup in these QUEST discharges is similar to those on NSTX and HIT-II.Future studies on QUEST should aim for further reducing the electron density and using the transient CHI target to develop non inductive current ramp up scenarios using electron cyclotron heating and current drive.

Figure 1 .
Figure 1.(a) Cartoon showing the transient CHI configuration used on NSTX.Toroidal ceramic insulators at the top and the bottom gap of the vacuum vessel electrically separate the entire inner and outer vacuum vessel regions.The solid blue line shows an initial injector flux field line.The dashed blue line shows a cartoon of a field line that has filled the vessel due to the interaction force between the poloidal current, j pol and toroidal field, Btor.The oppositely directed field lines in the injector region that would reconnect to generate a closed flux plasma configuration.(b) The improved high field side CHI injector configuration in QUEST.In this configuration, only the electrode above the outer lower divertor plate support structure is insulated from the rest of the vessel structure.Shown is the CHI injector region consisting of the inner and outer cylindrical electrodes, the gas manifold, and the initial injector flux.The vessel and the cylindrical inner electrode are at ground potential.During CHI operations, the bias electrode and the cylindrical outer electrode are at high voltage (HV).The distance 'd' shown in (a) and (b) is a measure of the initial footprint width on the electrodes that is related to the bubble burst condition.The parameter 'd' is discussed in this section and at the beginning of next section.

Figure 2 .
Figure 2. (a) Time traces of the injector current (I inj ), toroidal current (Itor), and the current multiplication factor (Itor/I inj ) for the transient CHI discharge with 23 mWb of injector flux.P1-3 indicate the discharge phases 1-3.(b) Fast camera images corresponding to times t1, t2 and t3 in figure (a).

Figure 3 .
Figure 3.The comparison between (a) the closed flux configuration in the camera images and (b) the calculated equilibrium reconstructions based on the persisting toroidal currents at the times, t4-t7 in figure 2.