Micro-particle injection experiments in ADITYA-U tokamak using an inductively driven pellet injector

A first-of-its-kind, inductively driven micro-particle (Pellet) accelerator and injector have been developed and operated successfully in ADITYA-U circular plasma operations, which may ably address the critical need for a suitable disruption control mechanism in ITER and future tokamak. The device combines the principles of electromagnetic induction, pulse power technology, impact, and fracture dynamics. It is designed to operate in a variety of environments, including atmospheric pressure and ultra-high vacuum. It can also accommodate a wide range of pellet quantities, sizes, and materials and can adjust the pellets’ velocities over a coarse and fine range. The device has a modular design such that the maximum velocity can be increased by increasing the number of modules. A cluster of lithium titanate/carbonate (Li2TiO3/Li2CO3) impurity particles with variable particle sizes, weighing ∼50–200 mg are injected with velocities of the order of ∼200 m s−1 during the current plateau in ADITYA-U tokamak. This leads to a complete collapse of the plasma current within ∼5–6 ms of triggering the injector. The current quench time is dependent on the amount of impurity injected as well as the compound, with Li2TiO3 injection causing a faster current quench than Li2CO3 injection, as more power is radiated in the case of Li2TiO3. The increase in radiation due to the macro-particle injection starts in the plasma core, while the soft x-ray emission indicates that the entire plasma core collapses at once.

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
The global energy crisis and challenges of reducing carbon footprint are being addressed by actively pursuing nuclear fusion research through magnetic confinement of hightemperature plasmas.International efforts to build commercial fusion reactors are currently plagued by many unresolved issues related to the deleterious effects of mechanical load (due to electromagnetic effects), heat load, energetic particles generated from the plasma disruption, and sudden termination of plasma [1,2].Research and development are underway globally to develop robust mitigation systems against disruptive events by quenching a burning fusion plasma using pellet injectors.These injectors mostly inject gas [3] or frozen-gas (cryogenic) pellets [4][5][6], the latter often shattered before injection (shattered pellet injection or SPI) [7][8][9][10].Massive gas injection (MGI) techniques are inherently limited by complex quench requirements, and high energy barriers at the edge of the plasma which further get aggravated when large-volume plasmas must be accessed [11][12][13].SPI systems remain much superior and are chosen as the baseline technology for ITER disruption mitigation (DMS) [10].While it was initially thought that a large fraction may get vaporized upon shattering thereby reducing the chances of deep fragmentation, however, these concerns are now well addressed by Gebhart et al [14].Though inherently limited by the slow response time of mechanical valves present in the gas-feed system and/or issues connected with situating the valves near the reactor vessels, a response time of SPI is now well within ∼20 ms outlined by the ITER DMS specifications [14][15][16].The experiments on DIII-D have experimentally demonstrated the advantages of SPI over the MGI in all regards achieving pellet velocities of ∼200 m s −1 , before fragmentation [7,8,17].
Despite the good progress made with Shattered Neon Pellet Injection, presently considered the most promising candidate for ITER DMS, challenging scenarios can be envisaged during disruption with conflicting requirements that demands simultaneous reduction of induced vessel forces, conducted heat loads, and runaway electrons.In recent times pneumatically driven Solid Shell Pellets [18] have also been injected to achieve discharge shutdown.The radiative material that is typically held in a shell can accomplish deep penetration without ablation and is directly deposited in the desired region where it is the most effective.The resulting inward-outward thermal quench (TQ) lowers the conducted heat load as is predicted for high-Z shutdowns, whereas prolonged current quench and reduced induced vessel forces are expected as a characteristic of lower Z-shutdowns.The tracer-encapsulated solid pellets of ∼900 µm are injected into the Large Helical Device with a velocity of 300-500 m s −1 [19].However, due to the higher masses of shell or solid particle injection approaches based on well-established pneumatic drives may fall well short of reaching the desired injection velocity.Due to the modest thermal velocity of the propellant gas molecules, velocities are expected to be restricted to 300-400 m s −1 [16], this increases the time required to travel the few meter lengths before it reaches the plasma edge.Furthermore, even though it is anticipated that most ITER disruptions will have a warning time of at least 20 ms, due to the involvement of mechanical valves, such acceleration techniques may not be appropriate in case of a shorter warning time scale in the incident of shutting down the plasma discharge with a large plasma volume.Apart from DMS dissipation of postdisruption runaway electron plateaus by SPI in DIII-D is also reported [20].
An electromagnetic pellet injector is projected to readily overcome these restrictions since higher velocities can be attained and such a device can accomplish short warning timelines [21]-both of which are critical for large-sized fusion-grade reactors.Acceleration is due to Lorentz forces experienced by current-carrying conductors in a magnetic field.In one such proposal being actively pursued by Raman et al [21,22], the concept of a Railgun has been adapted to accelerate a solid pellet.An external energy source has been used to drive a primary current through two current-carrying rails.A conducting capsule placed across the rails helps to complete the primary circuit and is also configured to act as a solid pellet to be released into the fusion reactor.Depositing the radiative material directly in the runaway current channel formation region both the TQ and formation of runaway electrons could be suppressed [22].
In this paper, a first-of-its-kind inductively driven microparticle accelerator and injector (IPI) is presented, that employs an alternate method of electromagnetic acceleration with several distinct advantages over other pneumatic and electromagnetic drivers proposed for fusion applications.The IPI is a modular electromagnetic accelerator where each module uses a pulsed electromagnet powered by a capacitor bank.It uses transformer action with induced secondary currents on the cartridge to accelerate a metallic (electrically conducting) cartridge through electromagnetic (Lorentz) forces, to the desired velocity.The cartridge is filled with pellets that are off-loaded into the reactor vessel while the spent cartridge is itself retained at the muzzle of the accelerator.The tuning of such an induction-base driver for accelerating micro-particles in a lightweight capsule involves a complex correlation of multiple parameters such as driver energy, frequency of drive, masses involved, and time synchronization of multiple modules.Optimization of multiple parameters obtained numerically through particle swarm optimization is a major inventive step carried out toward the design and development of IPI.The cartridge design is ingenuous: it carries the pellets as a payload and releases them through a novel stop-and-rupture mechanism.
IPI overcomes the drawbacks of conventional pneumatic injectors by achieving high speed, fast response time, and avoiding any gas feed.It allows contactless acceleration through electromagnetic induction and therefore is not prone to damage due to erosion, arcing, etc.With IPI, it is possible to inject micro-granular pellets of any shape, size, and material, of calculated volume/mass and at different speeds, all of which can be varied over a wide range.Directly injecting microparticles obviates the need to shatter pellets.It, therefore, brings to the table a certain kind of versatility not commonly attributed to contemporary injectors and has demonstrated immense potential for application in next-generation fusion reactors like ITER.A comparison of IPI vis-à-vis other accelerators (pneumatic and electromagnetic) and the possibility of scaling this method to ITER-like devices is discussed later in the paper.
A suitably configured IPI system has been integrated and commissioned in a magnetic confinement device, namely ADITYA-U [23,24] at Institute for Plasma Research, India.A cluster of particles of lithium titanate/carbonate (Li 2 TiO 3 /Li 2 CO 3 ), weighing ∼50-200 mg with individual particles having dimensions of 50-100 µm, is injected with velocities ∼200 m s −1 during the current plateau phase of ADITYA-U tokamak discharge.While the end objective would be to quench the plasma in the event of a disruption, prediction, and mitigation of an actual disruptive instability is beyond the scope of this work.What is demonstrated is that in the event of any disruption prediction, IPI can respond adequately fast to a trigger and inject micro-particles at desired mean velocities leading to disruption shutdown of plasma discharge within ∼5-6 ms of triggering the injector.A substantial increase in radiation due to the injected impurities indicates radiative loss of the plasma stored energy leading to thermal and current quench.The radiative dissipation of the plasma stored energy during a major disruption in ITER by fast injection of massive pellets of low Z impurities, such as Li and Be pellet injection has been numerically modelled by Lukash et al [25].
To the best of our knowledge, this is the first instance in which an electromagnetic pellet injector has been integrated with a tokamak.The system successfully demonstrates the quick response time, the ability to directly inject microparticles at desired velocities, vacuum operation, and the ability to work in a tokamak's challenging environment.The present device while being upgraded for large-scale tokamaks, is presently ready and well-suited for carrying out experiments with different solid pellets for disruption control in any magnetically confined fusion device.
Section 2 describes the working principle and design of the IPI system and the system characterization is presented in section 3. Section 4 describes the results of experiments carried out in ADITYA-U tokamak using IPI.Discussion and summary are presented in section 5.

Single stage operation
Figure 1 depicts the working principle of a single module of an IPI.A pulsed power supply, usually a capacitor bank, is used to drive a damped oscillating current, I p , through a solenoid, which generates a time-varying axial magnetic field (B Z ).The changing flux induces an electric potential in a contactless way on a conducting cartridge (of mass m), placed within the solenoid.An induced (secondary) current (I θ ) flowing circumferentially on the conducting cartridge and the radial magnetic field (B r ) of the electromagnet generate an electromagnetic force, F z , (Lorentz force) that accelerates the cartridge (loaded with pellets) axially.In a multi-module system, more than one electromagnet is placed linearly to incrementally accelerate the cartridge to higher velocities.Each electromagnet of inductance L has its customized capacitor bank of capacitance, C charged to a voltage V.The V and C together decide the electrical energy input ( 1 2 CV 2 ) to the cartridge, a fraction of which is converted into its kinetic energy ( 1 2 mv 2 ).

Multistage operation: synchronisation of modules
The cartridge, under the action of electromagnetic forces, accelerates through a continuous flyway tube that runs seamlessly through one or more electromagnets.In a multi-module system, the switching of capacitor banks is synchronized with the movement of the cartridge (figure 2).The cartridge must reach an optimal position inside the respective electromagnets when the switch is activated; this optimal position for each electromagnet, dependent on the velocity and mass of the cartridge is known a priori from computer simulations carried out  Switching synchronization with the movement of the cartridge using velocity feedback with the aid of a microcontroller: the optimal position, POn, within each electromagnet (n being the nth electromagnet) is known a priori from computer simulations carried out for the system.The time delay, Dn, required to reach the optimum position inside each electromagnet is estimated in real-time from the velocity, Vn, of the moving cartridge (at the exit of the previous module) and the distance, Xn, needed to travel to reach its optimal position (in the next module).
for the system.The position and velocity of the cartridge as it travels down the flyway tube are ascertained by a 'time-offlight' velocity measurement system.Collimated light beam sources (lasers) paired with fibre-coupled fast optical sensors (photodiodes) are placed along the flyway tube, and transverse to the direction of movement of the cartridge.As the cartridge, occludes the optical signal, the length of the cartridge upon the duration of interruption (sensed by the photodiode) gives the velocity (v) of the cartridge at that location (averaged throughout interruption).The measurement error is related to the beam width of the laser and is less than ±2% for a cartridge length of ∼25 mm.The real-time measurement of cartridge velocity is repeated after every module.A microcontroller-based Master Control Hardware (MCH) is programmed to act on the realtime feedback from the velocity measurement system, locate the cartridge at discreet locations, estimate its velocity, and estimate the time of arrival of the cartridge at the optimal location inside each subsequent electromagnet.With this realtime feedback, the MCH carries out a synchronized, sequential switching of capacitor banks associated with each electromagnet.The MCH may be initiated by a manual trigger or one that is generated from the fusion device.

Design of IPI
A suit of computational tools has been developed and used for designing the IPI.Coil and capacitor bank are modelled as a series RLC circuit.The cartridge is treated as a multisegmented RL (short-circuited) armature inside a coil, magnetically coupled to it.Model equations are obtained by applying Kirchhoff's Voltage Law for the coil & cartridge equivalent circuits.A full 2D-magnetohydrodynamic code that solves the circuit equations with appropriate material model (for the cartridge material) has been used to self-consistently solve for the coil and armature currents, for a given location of the armature.Force, acceleration, velocity, and temperature of the armature are estimated at every time step.Computations for multi-module operation has been used to predict the final muzzle velocity and validated against experimental results.The design analysis for accelerating a cartridge of given dimensions and mass requires complex correlation of multiple parameters of Capacitor Bank and Coil with those of Cartridge.The developed code has been used as a driver for a particle-swarm-optimisation (PSO) algorithm.The PSO code is used to find out the optimum values such as capacitance, capacitor charging voltage, optimal armature positions,  and coil parameters for intended velocities of cartridge.FEM analysis is used to separately assess the forces on the cartridge and design it accordingly.
The overall arrangement of the IPI system (figure 3) can be divided into two parts: (1) Core-System and (2) Peripheral-Systems.The core system comprising of the Electromagnets, Flyway tubes, Cartridge, Stopper, and their respective supporting structures are made of vacuum-compatible materials, bakeable up to 100 • C, and are all enclosed in an electropolished stainless steel (SS) vacuum chamber.
The peripheral systems comprising Capacitor banks, Charging Power Supplies, Velocity measurement, and Control Hardware are all located outside the vacuum assembly and interface with the core system through suitable vacuum ports/feed-throughs.The vacuum chamber, housing the Core system can be pumped down to 10 −8 mbar.It has diagnostic ports, pumping ports, and viewing ports with vacuum sealing in a manner that ensures (a) interfacing with peripheral systems (like high voltage power sources and velocity diagnostics) placed outside the vacuum, (b) electrical isolation with reactor vessel and its sub-systems, (c) coupling with a reactor vessel under vacuum and (d) vacuum isolation so that the system can be isolated from the reactor vessel and independently vented for placement and retrieval of cartridges.

Design optimization of cartridge
At the core of the novel device is a Cartridge (figure 4) configured to carry the pellets within it to their final velocity and further off-load the pellets at the muzzle through a well-optimized separation mechanism induced by impact onto a Stopper.It is a closed, cylindrical shell made of a lightweight, conducting material, aluminium.The outer diameter is approximately equal to the bore of the flyway tube for maximum flux linkage.The shell thickness is chosen a few times the 'skin depth' (∼2.3 mm) of magnetic-field diffusion calculated for the frequency of oscillating current (2.5 kHz) and conductivity of Al-6061.The rear end is closed with a threaded plug.The plug has a hollow central axial tube that is filled with desired pellets.The cartridge face is designed to flair open upon impact with a Stopper, releasing the pellets through a centrally located hole in the Stopper.The impact face of the cartridge is flat and of appropriate thickness and has two cross-slits that help in initiating the rupture.The impact force generated is primarily dependent on the impact velocity.It also depends on the dimensions of the Cartridge and Stopper and the strength of the material chosen for both.

Operating parameters
The present prototype of IPI has been configured to deploy it on a medium-sized Indian Tokamak (ADITYA-U) [23] and therefore intended to accelerate cartridges to velocities in the range between 100 m s −1 and 250 m s −1 .The system and operating parameters are arrived at and optimized using system simulations.The velocity can be set at any value within the range, by varying a set of operating parameters that allow coarse and fine control.Some of the operating parameters that are adjustable to achieve the desired velocity are (i) several operational stages (ii) the charging voltages of individual stages (iii) switching delays of each stage.The present prototype is a 2-module system having a total length (breech to muzzle) of 219 mm.Each electromagnet is 52 mm in length, having an inductance of 7 µH.Each is driven by a capacitor bank of 370 µF with a charging voltage of each bank,  Peak currents of 34 kA (Coil 1, V1 = 6 kV) and 33 kA (Coil 2, V2 = 6 kV) are generated.The slight difference in currents results due to the difference in cable lengths and is inconsequential as far as the velocity of the cartridge is concerned.A muzzle velocity of ∼250 m s −1 is achieved in this case.
The prototype is tested offline for cartridge performance and velocity measurements.In the present case, the cartridge is optimally designed to rupture after impacting the stopper, at impact velocities, ranging from 100 m s −1 to 250 m s −1 .The cartridge is made of non-tempered Al-6061 and the Stopper of hardened steel.The cartridge mass (with pellets) is 17 gm with 25 mm length, 23.6 mm outer diameter, and 0.5 mm thickness of impacting face.The design is optimized for the given range of velocities, such that the impact forces lead to a 'successful' opening-up of the cartridge, to a diameter of ∼5 mm (or more), in less than 50-60 µs.Quick opening up of the cartridge ensures that the mean velocity of pellets or micro-particles remains close to the cartridge velocity.Figure 6 shows snapshots of the cartridge during its rupture as the impact with Stopper forces it open.

The velocity of cartridge and pellets
Cartridge velocities have been set and achieved between 100 m s −1 to 250 m s −1 with a resolution of 25 m s −1 in the prototype and have been measured with an accuracy of ±5 m s −1 using the time-of-flight technique at the muzzle.Shot-to-shot reproducibility is within ±10 m s −1 and is most sensitive to jitters in the activation instant of the electromagnets.During offline testing, Li 2 TiO 3 pellets of diameter 75-100 µm have been accelerated in the atmosphere.Velocities of pellets after release (in the atmosphere) have been estimated with a fast video camera and correlated with the cartridge velocity.Observations have been carried out at a frame rate of 12 500 frames per second (shutter speed: 5.6 µs) with a resolution 1024 × 304 pixels (0.9032 mm/pixel).Image processing of the acquired videos gives the velocity distribution of the pellets.Snapshots of pellets after injection (figure 7; see supplementary video) and the velocity distribution of pellets obtained from image processing are shown in figure 8 for a cartridge velocity of 200 m s −1 .An optimally designed cartridge leads to the successful opening and release of pellets with the average velocity distribution of the pellets close to the muzzle velocity of the cartridge.Higher velocities result in a more collimated jet.Pellets can also be shattered further to act as a dispersive load of pellets by using a 'breaking plate'.Appropriate steering plates can also be used to steer/bounce off the pellets at desired angles.

Microparticle injection experiments in ADITYA-U
The inductively driven pellet injector (IPI) has since been commissioned successfully in ADITYA-U, a medium-sized (R 0 = 75 cm, a = 25 cm) Ohmically heated, air core Tokamak with toroidal-belt limiter, capable of producing circular as well as shaped plasma with single and double null open divertor configuration.The IPI system is mounted on one of the radial ports of the ADITYA-U vacuum vessel, as illustrated in figure 9, after assuring vacuum and electrical isolation of the system through a vacuum gate valve and a ceramic ring.Once the filled cartridge containing Li 2 TiO 3 particles is loaded, the vacuum vessel enclosing the core system of IPI is sealed and evacuated.The experiments are conducted during circular plasma operations.The plasma parameters are toroidal magnetic field ∼ 1.The MCH responds in ∼µs and activates the first module.It takes ∼1.5 ms for the cartridge to reach a muzzle velocity of 210 m s −1 over a length of 219 mm after the trigger.At the muzzle, as the cartridge strikes the stopper, the pellets are released with a mean velocity of ∼200 m s −1 .At these velocities, the bulk of the particles takes about ∼4 ms after their release (∼5.5 ms after IPI is triggered) from the cartridge to reach the plasma core (0.875 m away from the muzzle).
Standard magnetic diagnostics [24] are used for the measurement of plasma parameters, such as plasma current, loop voltage, Mirnov oscillations, etc. with and without the pellet injection to obtain the characteristics of pellet-induced disruption in ADITYA-U discharges.The soft x-ray (SXR) emission intensity is measured using a SXR tomography camera [26] which consists of an array of 8 channels of AXUV photodiodes having a beryllium filter of thickness of 10 µm.Additionally, integrated SXR emission intensity is measured by one surface barrier detector collimated to view emission from core plasma.Central chord-averaged density is measured using a 100 GHz heterodyne microwave interferometer [27].Temporal evolution of different plasma species Hα (656.2 nm), C III (464.7 nm), and Li I (670.8nm) is measured with an optical setup and photomultiplier tube (PMT) based diagnostic systems.The Hα and C III light is collected through the chords passing through the plasma centre viewing the plasma column horizontally from the high-field side to the low-field side, whereas the lithium light is collected by the chord viewing the plasma vertically from the top port and passing through the plasma centre.The chords are shown in figure 10(a).For obtaining the spatial variation of lithium radiation during the particle injection, spectral lines of Li I (670.8nm) have been monitored using a half-meter and a 1 m spectrometer [28][29][30].The plasma is viewed along the major radius from the low-field side horizontally for the halfmeter spectrometer.Light collected from seven radial locations encompassing the entire outer plasma radius (edge to centre) using seven toroidal line-off-sights is fed into a 1 m  spectrometer for obtaining the radial profile of emissions [28,29].The light collection chords for the spectrometer are shown in figure 10 (b) [30].Before performing the particle injection experiment in ADITYA-U, the machine has been prepared for the experiment.The standard reference discharges of I p ∼ 120 kA, 100-120 ms have been established.Later, lithium-titanate (Li 2 TiO 3 ) particles have been injected during the plasma current flat-top at ∼52 ms in shot # 33317.The typical size of a lithium-titanate particle of ∼50-100 µm has been chosen for the study.
The time evolution of two consecutive discharges ADITYA-U, black curve (#33317) with pellet injection and red curve (#33318) without pellet injection are shown in figure 11.After the IPI is triggered at ∼52 ms in discharge #33317, the plasma current terminates at ∼58 ms indicating the impurity particles reach the ADITYA-U core plasma within ∼5-6 ms and causing fast termination of plasma current.The chord averaged electron density and electron temperature, terminates very rapidly (figures 11(b) and (c) respectively), due to the increase in the plasma radiation (figure 11(d)) after the impurity particle injections.The TQ occurs before the current quench as evident from figures 11(a) and (c).The total radiated power measurement indicated that more than 2/3 of the input Ohmic power is radiated away after the particle injection.
The emission intensity of the Li I spectral line at 670.8 nm increase after t ∼ 52 ms as shown in figure 12, confirms the signature of Li 2 TiO 3 injection.The spatial variation of the lithium   spectral line has been monitored to obtain the profile of the deposition of the pellet inside the plasma.Figure 13 shows the spatial variation of Li spectral line intensity from different radial locations.It has been observed that Li I emission intensity is high near the edge while it is relatively less in the central chords.However, further investigation revealed that the less intensity near the core is due to Li self-absorption and the density is maximum there [31] as explained later in the text.Note here that no aluminium spectral lines are observed, suggesting no aluminium particles from the sabot enters the plasma.
Figures 14(a) and (b) show the temporal evolution of different plasma parameters with particle injection.Zooming the time duration after the pellet injection, as shown in figure 14(b), the causality of events is observed as follows: the injector has been triggered at ∼52 ms and the particles reach the plasma boundary in ∼2-2.5 ms covering a distance of ∼40-50 cm with a velocity ∼200 m s −1 .The increase in C III spectral line intensity at ∼54.5 ms along with an increase in the MHD activity confirms the arrival of the particles at the plasma boundary.The radiation power starts increasing at ∼56 ms and the SXR emission starts falling ∼56.75 ms indicating a TQ and the plasma current starts collapsing at ∼57.25 ms, within ∼3 ms of the particles reaching the plasma edge.Overall, the plasma current terminates within ∼5-6 ms of triggering the particle injector.
Further experiments are carried out with varying amounts of impurity clusters filled in the cartridge.Figure 15 shows four disrupted discharges with injected particles.In two out of those four discharges, the amount of injected impurity is 200 mg whereas in the other two 50 mg of impurity is injected.It can be seen from the figure that with a 200 mg injection, the plasma current quenches faster than those discharges with a 50 mg of injection.Further, note that in all these shots, the particle injector is triggered at the same time and the observed current quench too starts at the same time indicating both 50 and 200 mg of particles reach the plasma core with the same velocity.
The temporal evolution of the radial radiation profile is obtained using a radial bolometer array.Figure 16 shows the temporal evolution of radiation from different radial chords.
It has been observed that after the particle injection, the radiation in the central channels increases first, indicating the particles reach the core without radiating significantly in the edge.The particles seem to be disintegrating near the core before filling up the whole volume.The Li density selfabsorption described later in the text also supports the same.Furthermore, the radial profile of SXR emission intensity measured from an array looking from the top of the machine also indicated the particles influencing the plasma core directly as shown in figure 17.The central chords start decreasing first, whereas the outer chords show an increase before decreasing.The increase in the SXR emission intensity in the outer chords may be due to an increase in the local density in this region indicating a collapsing profile with radial pivot point around r ∼ 7 cm.
In another set of experiments the compound of the injected particle is changed from lithium titanate (Li 2 TiO 3 ) to lithium carbonate (Li 2 CO 3 ).It has been observed that by keeping the injected amount same of these compounds, the current quench time differs significantly as shown in figure 18.The current quenched faster with the lithium titanate (Li 2 TiO 3 ) injection as compared to the lithium carbonate (Li 2 CO 3 ) injection.
This result is on the expected lines as Ti radiates more due to its high atomic number and hence quenches the current faster as observed in all the spectral emission signals.The higher radiation in the case of Li 2 TiO 3 is evident from the comparison of total radiative power as shown in figures 18(h) and (i).Furthermore, as mentioned earlier and shown in figure 13, the observed Li I emission intensity is relatively less in the central chords as compared to the edge chords.Zooming the spectral line shapes of Li I at 670.8 nm (Li I-2s 2S-2p 2P) emission collected through the chords viewing the near-core plasma regions, a dip in the line-centre has been observed in the spectral line profile.The dip in the line profile in the spectrum recorded from the chord present at 12 cm from the plasma centre is shown in figure 19 and found to be related to the opacity effect due to the high density of lithium neutral present inside the plasma after the particle injection.By fitting the observed spectral line shape with a central dip, an opacity of ∼1.55 is derived.From a quantitative analysis using the radiation transport modelling, the absorption coefficient, opacity, and density of lithium are obtained and radiative loss is estimated.Approximately, 60% of input power seems to be radiated away due to particle injection, and the plasma is disrupted due to the radiative cooling [31].
Figure 20 shows the comparison of flux to the Langmuir probe at the limiter location with and without particle injection.The total radiated power is also shown in the figure and the vertical dash-dot-line shows the trigger to the injector.
It can be seen from the figure 20 that as the total radiation starts increasing due to the particle injection at around ∼55 ms, the flux to the Langmuir probe located at the limiter location, starts decreasing indicating a pinching of the plasma column due to particle injection just after the TQ.The flux to the probe increases again at the time of the current quench.Lastly, an attempt has been made to compare the current quench time between the particle-injected disruptions and that with MGI in similar discharges.The comparison is shown in figure 21.It can be seen from figure 21(b) that the current decays in ∼10 ms with the gas injection, which is much slower than the current quench time ∼<2 ms, observed with particle injection.Figure 21(c) shows that the radiated power  is highest with 200 mg particle injection.The time of initiation of particle injection is shown by the vertical dotted line whereas the gas-puff pulse is shown in the bottom view graph.

Summary and discussions
While SPI of cryogenic pellets is proposed as the baseline design for DMS in ITER and has garnered most of the attention yielding significant results and progress in recent years, we believe there is scope for exploring alternate mitigation techniques especially using solid pellets.Not much research has been carried out in this field after initial success with Shell pellet injection in DIII-D.This paper pertains to the development of a novel IPI that may be used to accelerate and inject pellets that are solid at room temperatures, with a desired velocity, using electromagnetic forces.The device may be considered as an optimised pulsed version of a linear induction motor with several innovations.At the core of the invention is a novel cartridge that enables it to carry solid pellets of desired material, size, volume and mass.Upon reaching final velocity the cartridge is designed to separate from the pellets through an impact induced fracture mechanism, optimised in a manner, such that the mean velocity of pellets is close to the muzzle velocity of the cartridge.The spent cartridge, left behind in the barrel, can be discarded, before reloading a new cartridge.Presently it is a single injection system though a magazine can be designed to load several cartridges that may be fired without interrupting the vacuum.
The IPI system it can be seen to have has several distinct advantages over other pneumatic and proposed electromagnetic drives some of which are summarised below: (i) IPI allows rapid response times as there is no involvement of mechanical valves.Due to electromagnetic forces, the speeds achieved can be much faster as compared to pneumatically driven pellet injectors.(ii) In Railgun-based electromagnetic accelerators (proposed in [21]), a capsule carrying primary current serves as a moving electrical contact between current-carrying conducting rails.In IPI the forces are generated using secondary currents induced on the cartridge in a noncontact (inductive) way.As the cartridge is not in physical (electrical) contact with the driver coil, electrical arcing, and damage/erosion of the cartridge or the pellets inside are therefore not of concern.(iii) The pellets in the cartridge do not enter the target device, such as a tokamak plasma, but instead, only the pellets are released at the speed of the cartridge.This unique mechanism allows for the injection of only the desired material chosen as payload, while the cartridge can be made from a material suitable for electromagnetic induction but not suitable for injection into the fusion core.(iv) As demonstrated, it is possible to maintain and operate the pellet injector either at atmospheric pressure or under vacuum.It is therefore possible to couple it target system under vacuum without adding any substantial gas load or vent the system independently to atmospheric pressure by isolating it from the target chamber.(v) The possibility of electromagnetic fields of IPI interfering with the magnetic fields of a Fusion Reactor is nil or negligible, as the IPI B fields, are mostly limited within the solenoidal coils (except for an insignificant fringing field outside).It may be noted that the magnetic fields used to accelerate the cartridge are relatively quite high ∼16-25 Tesla.In comparison the external fields of the reactor are expected to be far too weak at the location of the injector (outer port) to cause any major problem.Further, the accelerating fields being pulsed in nature, can be further shielded within the appropriately designed metal vacuum enclosure.(vi) As the design of the device is modular, the number of modules (electromagnets) can be added to increase the maximum limit of the velocity of the pellets.The velocity can be varied within the limit; a set of settable parameters can be adjusted to coarse and fine control the velocity.(vii) The ability of IPI to accelerate and inject a large number of nano/micro/millimetre size pellets, as demonstrated successfully, obviate the need to shatter them for easy dispersion and ablation.This is an attribute not presently associated with other pellet injectors.The size, shape, and material of pellets can be varied as desired.As the amount of pellets being injected is accurately measurable, it allows well-controlled experiments.
A prototype design of IPI has been commissioned on ADITYA-U, a medium-sized Tokamak.Performance characterization of the accelerator and cartridge has been carried out over a velocity range of 100-250 m s −1 .The cartridge is loaded with lithium titanate (Li 2 TiO 3 ) impurity payload weighing ∼50-200 mg.The injector is activated by a trigger from ADITYA-U and responds in less than a µs in shooting off the cartridge.Pellets with a mean velocity of ∼202 m s −1 reach the ADITYA-U core plasma within ∼5-6 ms causing fast termination of plasma current with a sharp fall in plasma temperature and density due to the increased plasma radiation.Using the same system, experiments have been carried out using different compounds of lithium of varying quantities.The distinct effect of quantity and different compounds provide interesting data for comparison, with 250 gm Li 2 TiO 3 being more effective than an equal amount of Li 2 CO 3 or 50 gm of Li 2 TiO 3 .But in every case, the radiative collapse seems to be triggered from the core of the plasma, indicating that the injection with IPI enables the particles to reach the core easily at these velocities.To our knowledge, this is the first demonstration of a discharge shutdown using an electromagnetic pellet injector on a tokamak.
IPI has been presently designed and reported for experimenting with solid impurities in mid-sized Tokamak like ADITYA-U and therefore optimized for injecting such pellets at a velocity of 100-250 m s −1 .In the present 2-stage system, with a 17 gm cartridge, it takes ∼1.5 ms (post trigger), for the cartridge to reach a muzzle velocity of 210 m s −1 over a length of 219 mm.With a 6-stage system, a muzzle velocity of 500 m s −1 is reached over 680 mm travel in 2.5 ms.Going by our calculations, using an optimised design acceleration time of the order of 10 ms can be attained for 2 km s −1 .Though the present range of velocities is suitable to carry out a wide range of experiments with direct dust injection, upgrading IPI to achieve higher velocities so that it can be adapted for large sized tokamaks and reactor grade machines, requires a few challenges to be addressed, as discussed below.
The velocity of pellets essential for large-sized and reactorgrade tokamaks may well be in the range of a few hundred meters to a km s −1 if the pellets are to travel to the core of the plasma in less than ∼10 ms.The IPI system is modular and higher velocities can be achieved in principle with additional electromagnets and/or by increasing the current in each electromagnet.Coils that can withstand more than ∼100 kA (pulsed) have been developed and tested and a 6-stage prototype has already been successfully tested up to 500 m s −1 .Multistage launchers with 6-12 stages are being designed to achieve up to a km s −1 .The parameters of each of the stages, in particular, the L and C of each module are adjusted, especially for later stages such that the rise time of the driving current matches the increasing velocity of the cartridge.Designing and optimizing such a system is not challenging given the electromagnetic simulation capabilities validated through the present design.
One of the challenges is certainly to design the cartridge to ensure opening at high velocities.Impact separation is seen to work well up to 500 m s −1 , with a reinforced cartridge made of Al-7075-T6 (UTS ∼570 MPa, compared to 240 MPa of Al-6061).Admittedly, damage to the cartridge is considerably more at these impact energies.The design of the stopper has been crucially modified to capture most of the sheared material, although the apprehension of some fragments of the spent cartridge being injected into the tokamak is real.Another solution is to use beryllium as a cartridge material as it is considered to be 'reactor-friendly'.It is also suitable for its high UTS (370 MPa) compared to Al-6061.At even higher velocities, the cartridge can be retarded through non-contact, decelerating forces ensuring a 'soft-capture'.A self-rupturing diaphragm under the action of decelerating forces that flair open releasing the pellets from the cartridge also remains a possibility.The detailed design and calculations are being deferred for a future publication.
With the addition of several stages and with energy being added incrementally at each stage, the resulting cartridge heating (due to eddy currents) can result in temperatures close to melting temperatures of aluminium.The addition of stages, therefore, seems to have diminishing returns, unless the heating problem is addressed.As the heating is mostly adiabatic, active cooling cannot be achieved over these time scales.Increasing the diameter of the cartridge and the choice of material, operating frequency and thickness of the cartridge can play a crucial role in deciding the skin depth over which the pulsed current flows and can help to mitigate temperature.Beryllium with higher thermal conductivity (1825 J (kg•K) −1 ) and melting temperature (∼1273 • C) once again scores favourably over aluminium alloys (715 J (kg•K) −1 , melting point ∼700 • C).The marginally higher electrical resistivity allows for higher skin depth.Increasing the thickness only helps in lending more strength to the cartridge (against radial forces); the increase in mass that may result is adequately compensated by the lower density of beryllium (1.83 gm/cc compared to 2.7 gm/cc of Al).The design trade-offs are being studied and optimized through a full magneto-hydrodynamic simulation of the circuit-cartridge combination that includes temperature and material effects, the results of which will be shared in a future publication.In effect upgrading IPI and achieving higher velocities of the order of km s −1 or more with IPI is technologically feasible.
The successful development and operation of IPI we believe is a significant step towards exploring use of solid pellets towards disruption control.Presently the device allows a diverse range of experiments to be performed and controlled with relative ease to explore the mass, material, and velocity of pellets to be used for heat and energetic particle mitigation during disruption.Its usage and utility can be extended to carry out fundamental investigations of micro-dust motion, ablation, and interaction with high-temperature plasmas thereby helping us with a better understanding of plasma-material interaction physics, especially near the reactor walls, better assessment of impurity control and plasma core cooling (due to dust produced in-situ), explore techniques for edge cooling, etc [32].

Figure 1 .
Figure 1.Working principle of single module.

Figure 2 .
Figure 2.Switching synchronization with the movement of the cartridge using velocity feedback with the aid of a microcontroller: the optimal position, POn, within each electromagnet (n being the nth electromagnet) is known a priori from computer simulations carried out for the system.The time delay, Dn, required to reach the optimum position inside each electromagnet is estimated in real-time from the velocity, Vn, of the moving cartridge (at the exit of the previous module) and the distance, Xn, needed to travel to reach its optimal position (in the next module).

Figure 3 .
Figure 3. (a) 3D-schematic of vacuum chamber enclosing the core system consisting of electromagnets, flyway tube, and supporting structures (b) core system showing electromagnets, flyway tube, and stopper (at the muzzle).

Figure 4 .
Figure 4. (a) Cartridge with threaded axial plug with pellets (b) stopper (c) cartridge stopped at the stopper (d) spent cartridge.

Figure 5 .
Figure 5.Time sequence of coil current and photodiode signals for a particular case.

Figure 7 .
Figure 7. Snapshot of pellets released from the cartridge when it impacts stopper (see supplementary video).

Figure 8 .
Figure 8. Histogram of the velocity of pellets.
3 T; plasma current ∼ 120-130 kA; Chord-averaged density ∼ 1.5-2.0× 10 19 m −3 and central electron temperature ∼ 200 eV.After the discharge current pulse reaches its steady state, a TTL pulse from the Tokamak Control System is sent to MCH to trigger the IPI.

Figure 9 .
Figure 9. IPI system mounted at one of the radial ports of ADITYA U vessel.

Figure 10 .
Figure 10.(a) Poloidal cross-section of the machine: showing a light collection for PMT-based diagnostic system (b) top view of the machine: showing a light collection for spectrograph-based systems.

Figure 12 .
Figure 12.Temporal evolution of Li I spectral line for shot #33317, measured with 0.5 m spectrometer.

Figure 13 .
Figure 13.Spatial evolution of Li I spectral line intensity for shot #3331, measured with a 1.0 m spectrometer.

Figure 14 .
Figure 14.Temporal profile of plasma discharge parameters for shot #33317 (a) plasma current (kA); emission intensities of (b) hydrogen-Hα (656.28 nm), (c) oxygen-OII (441.9 nm), (d) carbon-CIII (464.7 nm), (e) lithium-Li (670.8 nm) and (f ) visible continuum (536 nm); (g) plasma density (h) soft x-ray emission intensity, (i) total radiated power and (j) Mirnov oscillations.The first from the right vertical dotted line in the upper figure indicates the trigger time of the IPI system.The bottom figure is the time-zoom (55 ms to 61 ms) of the upper figure shown by the dotted lines.

Figure 16 .
Figure 16.Radial variation of radiated power, from the bottom −0.25 cm to the top 0.18 cm of radial distance from the plasma centre.

Figure 17 .
Figure 17.Radial variation of SXR emission intensity from high field side at r = −6.1 cm to low field side at r = 9.9 cm.

Figure 19 .
Figure 19.Self-absorbed Li I spectral line at 670.8 nm.

Figure 20 .
Figure 20.Comparison of flux to the Langmuir probe located at the limiter location with (solid-blue line) and without (dotted-green line) particle injection.The total radiated power is shown with a black-dash line.The vertical line shows the trigger to the injector.

Figure 21 .
Figure 21.Comparison of pellet-driven and massive gas puff-driven current quench times.The massive gas puff pulse is shown in blue in the bottom-most plot.