The effect of impurity seeding on edge toroidal rotation in the ADITYA-U tokamak

Intrinsic toroidal rotation velocity (Vφ ) has been measured from the Doppler shift of C5+ carbon spectral lines (at 529.05 nm) in the edge region of the ADITYA-U tokamak without any auxiliary torque input in an ohmically heated pure hydrogen (H2) plasma as well as in H2 plasmas seeded with medium-Z (neon and argon) impurities . The toroidal rotation in the edge region is observed to reverse its direction from the counter-current to the co-current direction with an increase in plasma current beyond I p ∼ 145–150 kA. Furthermore, a systematic decrease in the co-current Vφ has been observed with the edge density, which tends to decrease to almost zero velocity with an increase in the edge density. The injection of medium-Z (neon and argon) impurities is observed to influence the edge toroidal rotation significantly. In low I p discharges, argon injection leads to a reversal of edge intrinsic rotation from the counter-current to the co-current direction. In high I p discharges, both neon and argon seeding enhance the co-current rotation by about ∼5–10 km s−1, at a constant I p compared to pure H2 discharges. Simultaneous measurements of the edge radial electric field, E r, shows that the E r × Bθ flow seems to be driving the edge toroidal rotation in ADITYA-U. With impurity injection, the E r also gets modified, leading to an observed increase in the edge toroidal rotation.

Intrinsic toroidal rotation velocity (V ϕ ) has been measured from the Doppler shift of C 5+ carbon spectral lines (at 529.05 nm) in the edge region of the ADITYA-U tokamak without any auxiliary torque input in an ohmically heated pure hydrogen (H 2 ) plasma as well as in H 2 plasmas seeded with medium-Z (neon and argon) impurities .The toroidal rotation in the edge region is observed to reverse its direction from the counter-current to the co-current direction with an increase in plasma current beyond I p ∼ 145-150 kA.Furthermore, a systematic decrease in the co-current V ϕ has been observed with the edge density, which tends to decrease to almost zero velocity with an increase in the edge density.The injection of medium-Z (neon and argon) impurities is observed to influence the edge toroidal rotation significantly.In low I p discharges, argon injection leads to a reversal of edge intrinsic rotation from the counter-current to the co-current direction.In high I p discharges, both neon and argon seeding enhance the co-current rotation by about ∼5-10 km s −1 , at a constant I p compared to pure H 2 discharges.Simultaneous measurements of the edge radial electric field, E r , shows that the E r × B θ flow seems to be driving the edge toroidal rotation in ADITYA-U.With impurity injection, the E r also gets modified, leading to an observed increase in the edge toroidal rotation.
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
In magnetically confined tokamak plasma, toroidal plasma rotation has become an important area of research as it strongly impacts the confinement and stability of the burning plasma [1,2].Toroidal rotation and its shear play a key role in the L-H transition [3,4], stabilizing the magnetohydrodynamic instabilities (e.g.resistive wall modes) [5] and strengthening the transport barriers [6].In current tokamaks, toroidal rotation is often driven by the auxiliary momentum sources, such as neutral beam injection (NBI) [7].NBI provides external torque to the plasma, which helps to understand the variation in the rotation velocity with diverse plasma parameters [8,9].However, in the next-generation large fusion device like ITER, due to its large machine size and high density, this externally applied torque might not be able to provide sufficient momentum to rotate the burning plasma [9][10][11].In that scenario, another mechanism to generate the torque will be required to drive the plasma [12].
In recent decades, a non-zero toroidal rotation velocity has been observed in several tokamaks worldwide without having any external momentum input [11][12][13][14].This self-driven flow generated by the plasma itself (in the absence of any auxiliary torque input) is known as intrinsic plasma rotation or spontaneous rotation [12,15].In a tokamak, intrinsic rotation can be observed in both toroidal and poloidal directions.Toroidal plasma rotation that is directed with/against the direction of the plasma current is termed as co-current/countercurrent rotation.Conventionally, a positive/negative value of toroidal rotation velocity denotes that the rotation is in the cocurrent/counter-current direction, and the same convention has been followed in this paper.The magnitude and direction of intrinsic toroidal rotation is observed to be different in different regions of tokamak plasma.Furthermore, intrinsic toroidal rotation is also observed to be reversing its rotation direction, depending upon some critical values of the plasma parameters.This phenomenon is referred to as rotation reversal.
Although intrinsic plasma rotation is an interesting and an important phenomenon of tokamak plasmas, the underlying physical mechanisms of its generation and its reversal are still not fully understood and it has attracted several experimental and theoretical studies [7,16,17].In recent decades, significant efforts have been made to understand the source and the sink of this puzzling phenomenon [18,19].Mechanisms behind the generation as well as the determination of the level of intrinsic rotation in the core region might be offered by the residual stress or magnetic ripple, or the radial electric field [7].Importantly, the edge plasma rotation plays a key role in setting the boundary conditions for the rest of the rotation profile [20].It has also been observed that the toroidal rotation tends to propagate from the edge region toward the core plasma [7,19], making the understanding of intrinsic toroidal rotation and its generation in the edge very crucial.
The intrinsic toroidal rotation in the edge region of the ADITYA-U tokamak is measured using Doppler shifts of C 5+ ions.A detailed parametric study has been carried out by varying several plasma parameters, such as the plasma current I p , edge density n e , and the radial electric field E r to obtain the dependency of these parameters on the edge toroidal rotation velocity.The influence of impurity injection on the intrinsic edge toroidal rotation is also studied by puffing neon and argon into the discharge.Neon and argon seeding is well known for enhancing plasma confinement [21,22].These gases are also seeded for radiative divertor cooling and heat-load reduction to the tungsten targets as well as to lower the H-mode threshold in the presence of metallic walls in several tokamaks [23].The edge toroidal velocity is denoted by V ϕ throughout this paper.
It has been observed that V ϕ reverses its direction from the counter-current to the co-current direction with an increase in plasma current beyond I p ∼ 145-150 kA.The co-current V ϕ decreases to almost zero velocity with an increase in the edge density.Importantly, significant effects of medium-Z impurities like neon and argon are observed on V ϕ .Argon injection in low I p discharges (I p ∼ 130-135 kA) leads to a reversal of V ϕ from the counter-current to co-current direction.Whereas both neon and argon seeding increases the co-current rotation by about ∼5-10 km s −1 in high I p discharges, compared to pure H 2 discharges.
In the edge region, the radial electric field E r plays a crucial role in generating the E r × B flows [24,25].The E r required to produce the toroidal rotation is given by the following equation: ( The E r might get generated out of the gradients in the plasma potential and in the temperature [24] which, in turn, give rise to E r × B θ flow.Simultaneous measurements of the radial electric field E r using a Langmuir probe (LP) revealed that the E r × B θ flow seems to be driving the edge toroidal rotation in ADITYA-U.The E r also gets modified with the impurity injection, leading to the observed changes in the edge toroidal rotation.
The edge and scrape-off-layer regions of the plasma are known to be affected by E × B flows and the influence of neutral particles on momentum transport.Although these mechanisms may not be present or significantly reduced in ITER and other future devices, they should be adequately understood to correctly extrapolate to ITER and other future devices.Hence, the effect of E × B flows and impurity neutrals on edge toroidal rotation observed in the ADITYA-U tokamak are important for a better understanding of the physical mechanism of intrinsic edge toroidal plasma rotation and will be useful in predicting the toroidal rotation of ITER and other future devices.
The outline of this paper is as follows: the experimental setup is described in section 2. Section 3 is divided into two subsections; the first subsection consists of the observation made of the reversal of intrinsic toroidal rotation in pure H 2 plasma, while the second section consists of the effect of medium-Z impurity seeding on the intrinsic rotation.Section 4 contains a discussion on the mechanism behind the observed intrinsic toroidal rotation in both pure H 2 plasma and impurityseeded H 2 plasma.Section 5 contains a conclusion.

Experimental setup
The ADITYA-U tokamak is an upgraded version of the ADITYA tokamak [26], designed for operation in both diverter and limiter configurations.It is a medium-sized tokamak with minor radius (a) = 25 cm and major radius (R) = 75 cm.All the intrinsic rotation measurements presented in this paper are performed in ohmic circular plasmas with a limiter configuration.In the analyzed and reported discharges, the toroidal magnetic field, seen from the top, is in the anti-clockwise direction, while the plasma current is in the clockwise direction.The discharge parameters are as follows: plasma current ∼120-170 (kA), B T = 1.28 T (at R = 75 cm), reservoir pressure = 2.2 Bar, discharge duration ∼100-300 (ms), central chord-averaged electron density n core ∼ 1-3 × 10 19 (m −3 ), edge electron density n edge ∼ 0.5-3.0 × 10 18 (m −3 ) and central electron temperature 500 eV-1 keV.The chord-averaged electron density and temperature are measured using a microwave interferometer and soft-ray diagnostics, respectively [27].Edge density and floating potential measurements are performed using LPs.A rake LP is installed at the radial port on the low-field side (LFS) of the tokamak at the midplane (i.e. at z = 0 cm), which consists of six LPs, out of which the first LP is always kept inside the last closed flux surface (LCFS) for density measurements while the other LPs are utilized to measure the floating potential.These floating potential measurements are used to calculate the radial electric field (E r ) in the edge region.
The toroidal rotation velocity is estimated from the Doppler shift measurement of C 5+ (at 529.05 nm) spectral line emission.Note here that carbon is the main intrinsic impurity present in ADITYA-U tokamak plasma, originating from the graphite limiters.The passive charge exchange (PCX) line emission from C 5+ ions (at 529.05 nm) from the edge plasma region is monitored using a 1.0 m Czerny-Turner configured visible-range spectrometer from Princeton Instruments (AM 510, Acton Research Corporation).The spectrometer is equipped with a grating with 1800 grooves mm −1 (blazed at A fiber whose LOS is passing through ρ ∼ 0.96, i.e. r = 24 cm, is used in this paper to calculate the intrinsic rotation in the edge region.Here, ρ is defined by r/a, where r = 24 cm is the radial location and a = 25 cm is the minor radius of ADITYA-U tokamak plasma.The Czerny-Turner spectrometer used in this experiment is a multi-track spectrometer with a fiber bundle which can accommodate up to ten optical fibers at a time.Out of these ten tracks, one is used for the edge rotation measurement, while two tracks are utilized to calculate the rest wavelength (explained in the next subsection).At the exit slit, we use a sCMOS camera with 2560 × 2160 pixels (with pixel size ∼6.5 µm), to detect the line emission spectra.The exposure time for acquiring the data is kept at 30 ms, while the entrance slit is kept at 50 µm in all the measurements.The resolution of the spectrometer is 0.003 nm pixel −1 .The alignment of the fiber bundle at the entrance slit and that of the detector at the exit slit is performed in such a way that the binned spectra of the emission coming from the known source differ only by ∼0.5 pixel in the topmost and the lower-most track.Since 1 pixel shift corresponds to a velocity of 1.66 km s −1 , this misalignment of 0.5 pixel will give an error of ∼0.8 km s −1 in the measurements.Since, this misalignment does not change with time, and gives a fixed error in rotation velocity measurement, we adjust the obtained spectra while analyzing the spectral data to compensate this error.Therefore, the above-mentioned error is minimized.

Rest wavelength calibration
The setup shown in figure 2 exhibits the arrangement used to determine the rest wavelength of the C 5+ ions from the PCX line emission at 529.05 nm.It is a crucial step toward the determination of toroidal rotation velocity [28].Two optical fibers are mounted on the two opposite ports (toroidally separated by 90 • ) of the ADITYA-U tokamak.One of the optical fibers in this setup belongs to the set of eight fibers mounted on the re-entrant port described in the previous section.
The spectra obtained using these two fibers will provide the redshifted (λ red ) and the blueshifted (λ blue ) wavelength as they are seeing the plasma from opposite sides.By taking the mean value of these two wavelengths, i.e. λ red and λ blue , the rest wavelength can be calculated.The collection optics used to measure the toroidal rotation velocity and the rest wavelength are the same.From this setup, we get the simultaneous measurements of the rest wavelength (λ rest ) along with the observed Doppler's shifted wavelength (λ obs ) from that one fiber viewing the edge plasma.The edge rotation is now measured using the following equation, where λ rest = 0.5 × (λ red + λ blue ) is the rest wavelength.
The error arising due to rest wavelength measurements is approximately ∼0.5-0.7 pixels (seen in various discharges statistically), giving rise to an error of ∼1 km s −1 in the toroidal rotation measurements.

In H 2 plasma (without impurity seeding)
Previous measurements on ADITYA-U were performed in ohmic limiter plasma with low I p ∼ 100-120 kA.A direction reversal of the core plasma rotation from the co-current to counter-current direction was observed above a density threshold [11].In this subsection, experimental observation of edge rotation V ϕ is shown.The experiments are performed in the ohmic limiter H 2 plasma (without any impurity seeding) with I p ∼ 120-170 kA.All the data shown in figure 3 belong to 44 different plasma discharges, with the same B T (1.28 T) and the same reservoir pressure of 2.2 Bar.The horizontal plasma positions in all these discharges remain almost the same within a margin of ∼1 cm.Measurements are made during the plasma current flat-top phase.Discharges that have good signal to noise ratios in the sCMOS detector are considered for the analysis.All the discharges that did not fulfill the above-mentioned criteria were discarded and were not considered for the analysis.
The intrinsic rotation measurements from Doppler shifts of C 5+ ions suggest that the V ϕ remains in the counter current direction in discharges that have low I p ∼ 120-125 kA, which tends to decrease systematically with an increase in the plasma current, as shown in figure 3.In the discharges with I p ∼ 145 kA, the V ϕ becomes almost zero.Upon increasing the I p further, the direction reversal of the intrinsic toroidal rotation is observed in the edge region.Discharges have I p greater than ∼150 kA, and the edge plasma rotates in the co-current direction with its velocity increasing with the increase in the plasma current.
Further analysis of edge rotation velocity suggests that it is dependent on the edge density.It is observed that the intrinsic edge toroidal rotation in ADITYA-U remains in the co-current direction at the lower value of the edge density, as shown in figure 4.An increase in this edge density tends to damp the edge toroidal rotation, as shown in figure 4 in a fixed range of plasma current with I P variation <5%.Interestingly, this observation is consistent for different plasma current ranges (I p ∼ 148 ± 6 kA, 157 ± 3 kA and 165 ± 5 kA), as shown in figure 4 with different colors.Due to the unavailability of LP density measurements in discharges with low I P (where the rotation is in the counter-current direction), we are not able to show the dependence of rotation velocity on the edge density with I p < 140 kA.
The density measurements are performed inside the LCFS using the LP.Both plasma rotation and the edge density measurements are performed on the LFS at the vertical midplane.Discharges with almost the same plasma positions were analyzed.Since the exposure time for the spectroscopic measurement is ∼30 ms, the LP density data are averaged over the respective 30 ms of spectroscopic measurements, to study the dependence of rotation velocity on the edge density.Bottom: time evolution of intrinsic rotation velocity is shown for these shots with the same respective colors.Black and blue dashed lines at ∼100 ms show the application of argon and neon puff-pulses, respectively.

In H 2 plasma (with neon and argon seeding)
The injection of medium-Z impurities, e.g.neon and argon, has shown confinement improvement in many tokamaks [22,29,30].Neon and argon are deliberately injected in ADITYA-U to carry out various experiments [31][32][33].A significant change is observed in the intrinsic rotation velocity in the edge after neon and argon injection in ADITYA-U plasmas.Figure 5 exhibits the time evolution of intrinsic toroidal rotation in the edge region in a pure H 2 plasma along with those measured in the argon and neon seeded plasma.The rotation measurements are performed every 30 ms during the current flat-top and plasma current fall phase.The impurity gas is seeded for ∼1.6 ms, starting at ∼100 ms, into the discharge from the bottom port of ADITYA-U.The observed rotation velocity after impurity seeding is compared with the discharges with no impurity injection in figure 6.
Figure 6 depicts that the intrinsic rotation of the C 5+ ions gets enhanced significantly in the edge region (at the same value of plasma current) due to impurity injection.This suggests that impurity injection provides a torque to the plasma in the edge region, helping it to rotate very fast in the co-current direction.Further, the rotation velocity is compared with the edge density for the impurity-seeded discharge (as shown in figure 7).

In pure H 2 plasma (without impurity seeding)
Simultaneous measurement of the radial electric field, E r , along with the rotation velocity has been performed in the edge region to explore its contribution toward the observed intrinsic rotation velocity.This E r is estimated by measuring the floating potential at two different locations (r = 25.8 cm and r = 26.6 cm) using two LPs.This provides an average E r between the two probe locations.Since the probes are kept outside the LCFS, the E r is estimated by assuming a negligible temperature gradient at the probe locations.Another LP is placed at r = 24.5 cm (just inside the LCFS) to monitor the edge density simultaneously.Figure 8 shows the variations in the average radial electric field (E r ) between r = 25.8 cm and r = 26.6 cm with the edge density.
From the measurement of the E r , the E r × B θ velocity is estimated.The poloidal magnetic field, B θ is calculated using the following relation: where I P is the plasma current, µ o is the vacuum magnetic permeability and 'a' is the plasma radius.The edge rotation velocity determined using Doppler shifts is compared with the E r × B θ flow velocity measured in the edge region in different discharges for two plasma current ranges (I P = 142-154 kA and I P = 154-160 kA) and is shown in figure 9. Simultaneous measurement of V ϕ and E r × B θ flow velocity in a discharge is represented by the same symbols in red and blue colors, respectively, for the current range of I P = 142-154 kA and with gray and cyan, respectively, for I P = 154-160 kA.Different symbols represent data from different discharges.It can be clearly seen from figure 9 that the measured rotation velocity in the edge region compares well with the estimated E r × B θ flow in a fairly wide edge density range.

In H 2 plasma (with neon and argon seeding)
In neon-seeded discharge, in which the neon was injected at time t = 104 ms, a significant change in E r has been observed.The value of E r changes from approximately −0.4 kV m −1 before neon injection (50-80 ms) to 0 kV m −1 after neon injection (110-140 ms).During these two time intervals, the measured rotation velocity using Doppler shifts shows that during 50-80 ms, when the E r ∼ −0.4 kV m −1 , the toroidal rotation is ∼0 km s −1 .But during 110-140 ms, when the E r ∼ 0 kV m −1 , the toroidal rotation increased to ∼12 km s −1 .Therefore, this rise in intrinsic rotation suggests its dependence on the radial electric field.In another set of neon injected discharges (#34561 and #34554), LPs were placed at r = 24.5 cm and r = 25.3 cm.In these shots, as shown in figure 10, it can be clearly seen that there is a significant rise of ∼1-2 kV m −1 in the radial electric field, which has the capability to enhance E r /B θ flow by 10-20 km s −1 in the co-current direction.The E r in these discharges before neon injection is negative (i.e.radially inwards), which is observed to flip its sign to radially outward after neon seeding.This might be the reason behind the intrinsic rotation velocity enhancement observed during impurity seeding.Unfortunately, spectroscopic measurements were not available in these shots.
In shot #36440, a rotation reversal has been observed just after the argon injection, which changed the velocity of counter-rotating C 5+ ions from −9 km s −1 to +8 km s −1 (in the co-current direction).

Conclusion
Intrinsic toroidal rotation has been studied in the edge region using a visible spectroscopy system with a time resolution of 30 ms.In pure H 2 plasma, the direction reversal of the C 5+ ion's toroidal rotation is observed above a certain plasma current value in ADITYA-U.At lower I p , a counter-current rotation is observed as soon as the I p goes beyond 145-150 kA, the rotation becomes co-current rotation.Also, if impurities like neon and argon are injected in low I p discharges, the rotation direction reverses.Whereas, in high I p discharges, the impurity injection leads to high co-current rotation velocities compared to those in pure H 2 discharges.Furthermore, it has been found that the rotation velocity is inversely proportional to the edge density in different ranges of I P .With the impurity injection, the edge electron density is reduced as electrons are consumed in the ionization of impurities, and hence the rotation velocity increases.The estimated E r × B θ flow obtained using simultaneous measurements of the edge radial electric field E r using LPs matches quite well with the measured rotation velocities and seems to be driving the edge toroidal rotation in ADITYA-U.With impurity injection, the E r gets modified, leading to an observed increase in the edge toroidal rotation.

Figure 1 .
Figure 1.The LOS coming from eight backlit fibers installed at the re-entrant port is shown.
518 nm).The collection optics are shown in figure1.Eight optical fibers with a core diameter of 400 µm and numerical aperture of 0.22 are attached to the identical collimators.This setup is appropriately fixed on a specially designed re-entrant port (with 43 mm diameter width to accommodate eight optical fibers along with their collimators) installed at the tangential port (at the midplane) of the tokamak.The backlit image of these eight fibers is captured on graph paper (during the machine opening in ADITYA-U) to obtain the lines of sight (LOS) experimentally.The experimentally determined LOS of the eight fibers are observed to pass through the tangency radii: r = −4, 0, 4, 8, 12, 16, 20 and 24 cm.Theoretically expected values of the tangency radii are verified with the above experimentally obtained values.Both these values matched within an error bar of ∼0.5-1.0 cm.Backlit images of the fibers have a spot size of 30 mm at the tangency radii and they almost fall at the midplane with their centers between z = −1 cm and z = +1 cm.These backlit images are captured on graph paper, keeping the graph paper at a distance of ∼80-90 cm from the re-entrant port.The same exercise is repeated in the next machine opening without disturbing the optical setup to sense any change in the LOS.The changes observed were of the order of ∼1-2 mm, which gives the error bar in the LOS measurements.

Figure 2 .
Figure 2. The LOS of two backlit fibers installed at two tangential ports to collect the blueshifted and redshifted spectra.

Figure 3 .
Figure 3. Variation of edge toroidal rotation velocity with plasma current.Direction reversal of the edge intrinsic toroidal rotation is observed at Ip ∼ 145-150 kA.

Figure 4 .
Figure 4. Variation of edge toroidal rotation velocity with edge density for different plasma current ranges (black square: I P = 142-154 kA, red circle: I P = 154-160 kA, and blue downward triangle: I P = 160-170 kA).

Figure 5 .
Figure 5. Top: plasma current for shot #37588 (red: pure H 2 ), 37425 (black: argon seeded) and 37599 (blue: neon seeded).Bottom: time evolution of intrinsic rotation velocity is shown for these shots with the same respective colors.Black and blue dashed lines at ∼100 ms show the application of argon and neon puff-pulses, respectively.

Figure 6 .
Figure 6.Variation of intrinsic toroidal rotation velocity with Ip in H 2 plasma (red: without impurity seeding, black: argon seeded, and blue: neon seeded).

Figure 7 .
Figure 7. Variation of intrinsic rotation with edge density in H 2 plasma (red: without impurity seeding, black: argon seeded, and blue: neon seeded).

Figure 8 .
Figure 8. Variation in the radial electric field Er in the edge region with edge density.

Figure 9 .
Figure 9. Variation of Er/B θ flow velocity and edge toroidal rotation velocity as a function of the edge density.Simultaneous measurement of V ϕ and Er × B θ flow velocity in a discharge is represented by the same symbols in red and blue colors, respectively, for the current range of I P = 142-154 kA and with gray and cyan, respectively, for I P = 154-160 kA.Different symbols represent data from different discharges.