Comparison of partial and deep energy detachment behaviors with Ar seeding on EAST new corner slot divertor

It is necessary for future fusion reactor to reduce the heat fluxes on the entire divertor target, especially if view of long pulse high performance operation. In recent EAST experiments, partial energy detachment without confinement degradation, and deep energy detachment with protection of the entire divertor target have both been confirmed on EAST corner slot divertor by argon (Ar) seeding, which can provide reference for the divertor protection on future fusion reactors. In the deep energy detachment state, the electron temperature T et along entire lower outer divertor target decreases to less than 10 eV and heat fluxes are also strongly mitigated with peak heat flux reduction of more than 90%. Compared to the attached state, there is a moderate confinement degradation with H 98,y2 from ∼1 to ∼0.9 because of Ar radiation in the core region. This confinement degradation can be avoided in the partial energy detachment state, where the radiative power losses in the core are reduced. The experiment and SOLPS-ITER simulation results show that there is no decrease of particle flux js on the divertor target in the partial energy detachment state because the momentum loss in the SOL region is not strong enough. With increasing Ar seeding, there is a js decrease in the deep energy detachment state. The increases of momentum and power losses in the SOL region, and the decrease of upstream pressure all contribute to the js reduction.

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 future fusion reactors, if unmitigated, the excessively high heat and particle fluxes on the divertor target are higher than the material limit [1].Detachment with impurity seeding (i.e.nitrogen (N 2 ), neon (Ne) or argon (Ar)) has been proved as an effective means for the divertor target protection [2,3].Detachment is accompanied with significant decreases of electron temperature T et , particle and heat fluxes on the divertor target [4,5].In ITER with the power crossing the separatrix (P sep ) of ∼100 MW, it is sufficient to achieve partial detachment characterized by strong mitigation of heat and particle fluxes around the strike point but no detachment in the far SOL region [1].However, in China fusion engineering test reactor (CFETR) with P sep of ∼200 MW, partial detachment is unacceptable to protect the entire divertor target including the far SOL region [6,7].Deeper or complete detachment [8] is needed for the mitigation of heat fluxes on the entire divertor target.Complete detachment is typically characterized by very low and almost flat profiles of j s , T et and q t along the entire divertor target [8].
Impurity seeding, such as N 2 , Ne and Ar, is typically used to access detachment.It is observed that the radiation usually peaks around the divertor target at the start of impurity seeding, then gradually moves to X-point and core regions from partial to complete detachment state [8][9][10].In recent DIII-D N 2 seeding experiment, the complete detachment with high confinement has been achieved with H 98,y2 ∼ 1.5, β N ∼ 3 and β p > 2 on the graphite divertor [11].Without the assistance of intrinsic carbon radiation, the detachment is harder to be achieved with a tungsten divertor [12].On JET tungsten divertor, the complete detachment with Ne seeding has also been achieved with high input power of 29 MW but with a confinement degradation [13], similar to the complete detachment results in ASDEX Upgrade with N 2 seeding.The detachment over 10 cm or 2 power decay widths with the peak divertor power flux less than 2 MW m −2 , which is the state between partial and complete detachment, has also been achieved on tungsten divertor in ASDEX Upgrade with a moderate confinement degradation [14].The above detachment state is an attractive operation mode for future fusion reactors such as CFETR, considering the compatibility of the core plasma performance and the mitigation of heat fluxes on the entire divertor target.
In this work, the combination of increased momentum loss in the SOL region and T et < 10 eV is used to determine the onset of detachment [5].In the previous EAST detachment experiment with Ne or Ar seeding [15][16][17], there is a significant reduction of heat flux around the strike point but no reduction of particle flux, so we call it 'partial energy detachment' [4].The deep energy detachment with strong mitigation of heat fluxes along the entire target and rollover of particle flux j s around strike point has also been achieved by Ar seeding on the new corner slot tungsten divertor in EAST, which is also the state between partial and complete detachment.In the deep energy detachment state on EAST, there is still strong j s of ∼15 A cm −2 in the SOL region, which differs from the complete detachment.There is a moderate confinement degradation with H 98,y2 from ∼1 to ∼0.9 in the deep energy detachment state but no confinement degradation in the partial energy detachment state with H 98,y2 ∼ 1.
The rest of this paper is organized as follows: section 2 describes the experimental setup and results including divertor detachment and plasma performance in the core and pedestal regions.The simulation analysis is shown in section 3, including setup of simulation, momentum and power losses in divertor region, ionization and recombination distribution.The discussion and summary are presented in section 4.

Experiment setup
In 2020-2021, the EAST lower tungsten (W) divertor has been upgraded with steady-state heat flux exhaust ability of 10 MW m −2 [7,18].As shown in figure 1, the lower outer (LO) divertor with corner slot geometry is more closed than the upper divertor.With the lower closed divertor and LO strike point near the divertor corner, the line-averaged midplane density threshold for detachment by increasing density is ∼10% lower than that with upper open divertor [19], which is consistent with the results in DIII-D [20] and TCV [21].In addition, with the upper open divertor, the partial energy detachment with Ar seeding is usually accompanied with confinement degradation of ∼10% [16], and the deep energy detachment has not been achieved.In this work, we perform the deep energy detachment experiment in the lower divertor.The lower single null (LSN) configurations of shots #102528 and #102531 are shown in figure 1(b).In EAST, N 2 is not used because of daily wall conditioning of the lithium (Li) coating.According to previous radiative divertor experiment, Ar is more efficient for divertor target cooling than Ne because of the radiation characteristic [16].Therefore, Ar is chosen to achieve deep energy detachment.In addition, we also find in the previous experiments that it is more efficient for detachment onset with gas puff location in the near SOL region [22,23].Therefore, we seed Ar from LO near SOL region, as shown in figure 1(b).Based on the previous experiments with pure Ar seeding, the minimal seeding rate of control valve is still too large to cause confinement degradation even disruption.To avoid this, mixed gas with a volume ratio of Ar: D 2 = 50%: 50% is chosen in this work to reduce the Ar seeding rate.
The main diagnostics used in this work are shown in figure 1(a).The ion flux densities j s , divertor electron temperature T et , static electron pressure p et,static and q t on the new lower corner slot divertor targets are measured by Langmuir probes (LP) consisting of 32 triple probe arrays [24].The q t is computed by: where θ is the grazing angle between the incident magnetic field and the divertor target surface, and γ is the sheath transmission factor, assumed to be 7. Considering the q t measurement uncertainty of probe, the absolute values of q t are not very accurate.The total bulk plasma radiation P rad,bulk and radiation distribution are estimated by the absolute extreme ultraviolet (AXUV) system [25].The chord-integrated Ar XVI 23.507 Å line emission around the mid-plane is measured by extreme ultraviolet (EUV) spectrometer in a wavelength range of 10-130 Å, with a time resolution of 5 ms [26].The temperatures of LO and lower inner (LI) divertor target surfaces are measured by infrared (IR) cameras with a time resolution of 10 ms [27].The Thomson scattering system provides the plasma electron temperature in the core and pedestal regions with a spatial resolution of 1-2 cm [28].The profiles of ion temperature (T i ) in the core region are measured by an x-ray crystal spectrometer system by analyzing Ar spectra [29].The electron densities in the core and pedestal regions are measured by a polarimeter/interferometer system with 11 channels and a reflectometer system, separately [30].Additional NBI blips are to generate the beam emission spectrum for the motional stark effect diagnostics, which is not used in this work.

Divertor detachment behavior
In shots #102528 and #102531, the peak Ar seeding rate is fixed at 1.1 × 10 20 p s −1 and 1.2 × 10 20 p s −1 , separately.The two shots both work in the pulse width modulation mode [31], which means in each control cycle, the working voltage of the valve is constant, i.e., fixed puffing rate.The feedback controller modulates the pulse width and duty cycle.In shot #102528, the feedback control target T et,target around the LO strike point is 8 eV and constant in the whole shot.The control phase starts at 3 s.At the beginning of the control duration, because of the large control error between the true value and the T et,target , Ar is injected with the longest pulse width until the T et decreases to less than 8 eV.Considering the strong effect of radiation increment caused by Ar, the maximum single Ar pulse length used in the feedback control is set as 12 ms to avoid the confinement degradation.However, because of this single pulse limitation, the feedback control system spends relatively long time (>1 s) to lower T et with 'full load', i.e. using the longest injecting pulse.In shot#102531, the T et,target is set as a slope from 30 to 8 eV (dashed line in figure 2(g)) to make the impurity seeding milder at the beginning.It could avoid the accumulation of too many Ar ions in the core region at the beginning of Ar seeding.In addition, compared to shot #102528, a higher integrated gain is used in the feedback controller in shot #102531 so that Ar is still injected for ∼0.8 s after the target electron temperature around the LO strike point decreases to 8 eV.Therefore, the deep energy detachment state is achieved.
It should be pointed out that Ar injection has little effect on plasma at the start of both shots.One reason is that there is a 100 ms delay between the valve opening and the gas injection in the vessel.Secondly, the radiation caused by Ar at the start is too small to be measured by AXUV and EUV with short pulse (<12 ms) seeding mode.It takes a relatively long time for the accumulation of Ar.As shown in figure 2, P rad,bulk and Ar XVI line emission increase gradually with Ar seeding, while T et drops suddenly.It is similar as the detachment cliff observed on DIII-D [32], which is driven by the interdependence of the E × B drift fluxes, divertor electric potential structure, and divertor conditions.In the future, the analysis of the T et cliff with impurity seeding will be performed in details on EAST.
As shown in figure 3, in shot #102528, the target electron temperature around the LO strike point decreases from ∼50 eV to less than 5 eV while there is no significant decrease of T et in the far SOL region with Ar seeding, as well as the q t on the LO divertor target.It is the typical partial energy detachment phenomenon.The peak surface temperature of the LO divertor target T div,LO decreases by ∼60%, from ∼600 • C in the attached state to ∼250 • C in the partial energy detachment state, as shown in figure 4. In addition, the q t around strike point decreases by more than 90% in the partial energy detachment state, compared to the attached state, as shown in figure 3. The above results show that the heat flux around the LO strike point is strongly mitigated.There is also a slight decrease, by ∼20%, of the peak surface temperature on the LI divertor target T div,LI from ∼400 • C in the attached state to still more than 300 • C in the partial energy detachment state.
In the partial energy detachment state in shot #102528, there is no significant decrease of j s , which is a common phenomenon in the radiative divertor experiments on EAST.As shown in figure 3(c), p et,static around the LO strike point decreases by ∼70% with Ar seeding.Figure 5(d) in section 2.3 shows that upstream pressure p e around separatrix in outer mid-plane (OMP) only decreases by ∼20%.According to the modified two point model [33,34] and formula (2) in section 3, the pressure changes in OMP and target show that there is a momentum loss increase in the SOL region with Ar seeding, compared to the attached state without Ar seeding.The phenomenon is different from the high recycling state characterized by no significant momentum loss.The above result reveals the onset of detachment.However, the momentum loss increase is not strong enough to drive the decrease of j s , which is also observed in JET detachment experiments [13].The detailed analysis will be shown in section 3 through the combination of experiment and simulation results.In this work, we call it 'partial energy detachment' with a significant reduction of heat flux around the LO strike point but no reduction of particle flux.
In the deep energy detachment state in shot #102531, the T et along the entire LO divertor target, ∼20 cm, decreases to less than 10 eV with Ar seeding, as shown in figure 3(e).
Compared to the attached state, the peak q t around the LO strike point decreases by more than 90% and the q t along the entire LO target is close to a flat profile in the deep energy detachment state.As shown in figure 4(d), the peak T div,LO around the LO strike point decreases by more than 60%, from ∼600 • C in the attached state to less than 200 • C in the deep energy detachment state.There is also a significant decrease of T div,LO in the far SOL region.The T div,LO profile along the entire LO divertor target, ∼40 cm, is close to a flat profile.The above results show that the heat fluxes along the entire LO target are strongly mitigated in the deep energy detachment state.In addition, there is a decrease of j s around the LO strike point in the deep energy detachment discharge #102531.It is mainly attributed to the increases of momentum  and power losses, and the decrease of upstream pressure.The detailed analysis will be shown in section 3. On the LI divertor target, the peak T div,LI around strike point decreases by ∼50% with Ar seeding from LO divertor, from ∼400 • C to ∼200 • C.
The above results show that only the heat flux around the LO strike point can be mitigated in the partial energy detachment state.However, heat fluxes along the entire LO corner slot divertor target can be strongly mitigated in the deep energy detachment state.In the following section, we will compare the plasma behavior in the core and pedestal regions in partial and deep energy detachment states.

Plasma behavior in the core and pedestal regions
In the partial energy detachment state with Ar seeding in shot #102528, there is a significant increase of Ar XVI around mid-plane measured by EUV, as shown in figure 2(d).P rad,bulk  measured by AXUV increases to ∼0.9 MW and effective charge Z eff increases to ∼1.9.However, there is no significant change in W MHD and H 98,y2 (∼1) in the partial energy detachment state with Ar seeding, compared to the attached state.As shown in figure 5(a), there are decreases of T e and p e in the core region because of Ar radiation (shown in figure 6) in the partial energy detachment state.The detailed radiation distribution analysis based on results of experiment measurements and SOLPS-ITER simulations will be shown in section 3.There is a significant increase of T i in the core region, which may be attributed to the dilution effects and stabilization of the ion temperature gradient mode [13,35].The increases of n e and T i in the core region could partly compensate the power loss caused by Ar radiation so that there is no significant confinement degradation.
In the deep energy detachment state in shot #102531, there is a larger amount of Ar injection and higher Ar XVI intensity around mid-plane, compared to the partial energy detachment state, as shown in figure 2(h).P rad,bulk increases to ∼1.2 MW and Z eff increases to ∼2. Figure 7 shows that there are significant decreases of T e and p e in the core and pedestal regions in the deep energy detachment state, compared to the attached state, which is due to the increase of the radiated power caused by Ar seeding, as shown in figure 6.The changes are consistent with the moderate decreases of W MHD and H 98,y2 , as shown in figure 2(f ).W MHD decreases by ∼13%, from ∼150 kJ in the attached state to ∼130 kJ in the deep energy detachment state, and H 98,y2 decreases from ∼1 to ∼0.9.As shown in figure 7, there are increases of n e and T i in the core region in the deep energy detachment state.

SOLPS-ITER simulation analysis
The underlying mechanism of detachment is difficult to be uncovered with just the experimental measurements.Therefore, in this section, SOLPS-ITER simulations are performed to shed more light on the detachment mechanisms at play in these experiments.The power and momentum losses are analyzed through the combination of simulations and experiments in the partial and deep energy detachment states.

Setup of simulations
SOLPS-ITER [36,37] consists of the multi-fluid transport code B2.5 and the Monte Carlo neutral tracer EIRENE [38], which is utilized here to reproduce the experimental measurements and analyze the relevant detachment physical processes.The simulation cases are based on the similar basic background plasma parameters in shots #102528 and #102531.For the attached state without Ar seeding, the time t = 3.3 s in shot #102528 is chosen.For the partial energy detachment state with Ar seeding, the time t = 5.1 s in shot #102528 is chosen.As shown in figure 3, the LO strike point in the deep energy detachment state in shot #102531 shifts towards the high field side (HFS) by ∼0.9 cm, compared to the attached state.Compared to the partial energy detachment state in shot #102528, the LO strike point in the deep energy detachment state in shot #102531 shifts towards the HFS by ∼0.4 cm.Based on previous experiment, the influence of the change of strike point location of <1 cm on the divertor condition is relatively modest.However, in order to exclude its effect and study the detachment evolution progress more precisely in the modeling, we directly increase the Ar seeding rate without changing the magnetic equilibrium to extend from partial energy detachment to deep energy detachment.
Figure 8(a) shows the grid in simulations with a resolution of 96 poloidal cells and 36 radial cells.The plasma species D + , D 2 + , Ar + , …, Ar 18+ and the neutral species D, D 2 , Ar are included in simulations.The atomic and molecular volumetric processes in SOLPS-ITER simulations are from the ADAS database [39].The total input power into the numerical grid is fixed at 1.76 MW with the assumption that 20% of P abs is radiated in the core region.In addition, the input power is assumed to be shared equally between ions and electrons.Pure D 2 fueling is seeded from the OMP for the feedback control of the electron density at the OMP separatrix n e,sep , n e,sep = 1.05 × 10 19 m −3 without Ar seeding and n e,sep = 1.04 × 10 19 m −3 with Ar seeding.Figure 8(a) shows that the mixed impurity gas (50% Ar & 50% D 2 ) is seeded from LO divertor target, similar to the experiment setup.According to the experiment setup, the surface albedo of 0.998 representing a pumping rate of 13.2 m 3 s −1 and 0.98 representing a pumping rate of 45.6 m 3 s −1 are set for the upper pumping port-end and cryopump, respectively.For the lower pumping port-end and cryopump, the surface albedo of 0.994 representing a pumping rate of 3.6 m 3 s −1 and 0.96 representing a pumping rate of 30.0 m 3 s −1 are set.Apart from that, the incident particles are assumed to be completely recycled from the material surfaces.The drift effects are not included in this work.
Figures 8(b) and (c) show the radial diffusion coefficients D ⊥ and heat conductivities χ i,e .In the divertor region, D ⊥ = 3.0 m 2 s −1 and χ i,e = 0.1 m 2 s −1 are set for the case without Ar seeding.D ⊥ = 2.0 m 2 s −1 and χ i,e = 1.0 m 2 s −1 are set for the case with Ar seeding.These values are chosen to achieve that the radial profiles of T e and n e at the OMP in the SOLPS-ITER simulations match well with the experimental measurements in the attached state without Ar seeding and the partial energy detachment state with Ar seeding, as shown in figure 9.As shown in figure 10, there is also a good match between the simulation and experiment results of radial profiles of T et and poloidal ion flux j s,p on the LO divertor target in the attached and partial energy detachment states.For the partial energy detachment case, v Ar = 1.0 × 10 19 p s −1 and the D 2 seeding rate on the LO divertor target v D2 = 1.0 × 10 19 p s −1 are set.Compared to the attached state, the target electron temperature around the LO strike point decreases to less than 5 eV but j s,p does not decrease in the partial energy detachment state, which are similar to the experiment results in shot #102528.By increasing Ar seeding rate, the deep energy detachment is also achieved with v Ar = 2.6 × 10 19 p s −1 & v D2 = 2.6 × 10 19 p s −1 .The T et with distance from the LO strike point less than 12 cm are less than 5 eV and j s,p around the LO strike point decreases significantly, which are both similar to the experiment results in shot #102531.The deep energy detachments in simulations and experiments are both far from the complete detachment since j s does not decrease in the mid and far SOL regions.

Simulation results
As shown in figure 11(c), the momentum loss from the OMP to the LO divertor target mainly occurs in the divertor region.In addition, energy input due to cross-field transport from the core region would influence the power loss analysis from the OMP to X-point [40,41].Therefore, in SOLPS-ITER simulations of this work, we perform the momentum and power balance analyses in the LO divertor region, from X-point to LO divertor target.The entrance to the LO divertor is chosen as the upstream (u), and the downstream is the LO divertor target (t), as shown in figure 11(e).The momentum and power losses are integrated along the flux tubes from the upstream to the LO divertor target, which can be decomposed into various source terms [40].
The momentum loss factor f mom-loss and the power loss factor f power-loss in the SOL region are defined as follows [5]:   where p tot,u is the total upstream pressure, p tot,t is the total pressure on the divertor target; q ∥u is the total upstream energy flux, q ∥t is the total energy flux on the divertor target; A ∥u and A ∥t are the surface area of the mesh cell normal to the parallel direction.

Momentum losses.
Figure 12 shows the analysis of the momentum balance in the outer divertor chamber from the separatrix to the far SOL region.In the attached state without Ar seeding, there is a momentum loss near the separatrix, f mom-loss ∼ 0.4, which is mainly driven by the radial momentum transport and the radial viscosity.The former is due to the radial particle transport [41], and the latter is caused by the strong radial velocity shear [40].The contribution of the plasma-neutral interactions is relatively smaller than above two effects in this case.In the far SOL region, f mom-loss < 0, which means that there are momentum gains, as shown by the blue dotted line in figure 12(a).It can be ascribed to the poloidal viscosity effect, since the Bohm sheath boundary condition forces the parallel velocity to match the sound speed near target and thus causes a momentum gain [42,43].In this simulation, the parallel velocity is set at the sound speed on the target.
As the divertor plasma moves from the attached state to the partial energy detachment state with Ar seeding, f mom-loss near the separatrix increases to 0.7 but there is no momentum loss in the far SOL region, as shown in figure 12(b).The increase of momentum loss in simulation is consistent with the reduction of p et,static measured by LP in experiment.It is also similar to the detachment results in JET, ASDEX Upgrade and KSTAR with impurity seeding [5,42,44].However, the momentum loss is not strong enough to drive the decrease of j s in the partial energy detachment state on EAST.There is a significant decrease of the target electron temperature around the LO strike point, from ∼40 eV to ∼5 eV.The above experiment and simulation results on EAST confirm the viewpoint in [33,34] that momentum loss can not drive j s down if 1−f mom-loss decreases slower than T et 1/2 according to following formula: In the partial energy detachment state, the plasmaneutral interactions become much stronger and dominate the momentum loss in the near SOL region, as shown in figure 12(b), which mainly include charge exchange and    decrease of j s,p , as shown in figure 10.In addition, there is also a momentum loss in the mid SOL region but no momentum loss in the far SOL region with distance from the separatrix of ∼11 cm although T et decreases to less than 10 eV.There are also no decreases of p et,static and j s in the far SOL region measured by LP in the deep energy detachment state, compared to the attached state, as shown in figure 3. The above results show that detachment gradually extends from around the LO strike point to the SOL region but the complete detachment along the entire target has not been achieved.The increases of momentum loss near the separatrix and in the mid SOL region are both mainly caused by strengthen of plasma-neutral interactions, as shown in figure 12(c).In addition, the decrease of p tot,u and the increase of power loss also contribute to the decrease of j s in the deep energy detachment state.As shown in formula (4), the decrease of j s can be directly influenced by the decrease of p tot,u .The influence of power loss in divertor region will be discussed in the following two sections.

Power losses.
The process of detachment with impurity seeding is always accompanied by the increase of power loss in the SOL region.Figure 13 shows the f power-loss and the decompositions of contribution to the power loss in the outer divertor chamber from the separatrix to the far SOL region.In the attached state without Ar seeding, the power loss near the separatrix mainly originates from the radial transport effect due to the strong radial particle transport into the PFR.In the partial energy detachment state with Ar seeding, the power loss near the separatrix increases to ∼0.9, which is caused by the enhancement of the Ar radiation and the plasmaneutral interactions.In the deep energy detachment state, the f power-loss increases to larger than 0.95 near the separatrix and larger than 0.7 in the entire LO divertor region.There is a significant increase of the Ar radiation loss in the entire LO divertor region, which dominates in the power loss.
The increase of radiated power loss in SOL and divertor regions with Ar seeding is mainly from the Ar line radiation power loss [45] and the 2D distribution of it is shown in figure 14.In the partial energy detachment state, the total Ar line radiation power loss in the simulation region increases to ∼0.5 MW and maximizes around the LO divertor target.It seems that there is no significant radiation peak in divertor region measured by AXUV in the partial energy detachment state, as shown in figure 6, which is mainly because parts of LO divertor region, especially near the target, are not covered by the viewing chords of AXUV.There are also increases of Ar line radiation power loss in the core and X-point regions in SOLPS-ITER simulation, which is consistent with the AXUV measurement, as shown in figures 6 and 14.Since most of the bulk plasma region is not included in the SOLPS-ITER simulation, as shown in figure 8, the comparison of absolute radiation power loss value between simulations and experiments is not performed in this work.Unfortunately, 2D reconstruction of the radiation profile measurements is not available in EAST, preventing a detailed comparison of the radiation distribution between the experiments and the simulations.In the deep energy detachment state, the total Ar line radiation loss in the simulation region increases to ∼0.8 MW and the peak radiation region gradually moves away from the divertor target towards the X-point, as shown in figure 14(b).The Ar line radiation in the core region is also significantly enhanced as more Ar transports into the core region in the deep energy detachment state, compared to the partial energy detachment state.According to the AXUV measurement in shot #102531 in figure 6, it seems that there is a radiation peak above the X-point in the deep energy detachment state, which is mainly because the AXUV viewing chords above the X-point are also across the LI divertor region.In SOLPS-ITER simulation, there is also an increase of Ar line radiation power loss in the LI divertor region.

Ionization and recombination.
The power and momentum losses are both closely associated with the ionization and recombination of impurity particles.In the partial energy detachment state, Ar + ionization sources are mainly in the divertor region near the target, as shown in figure 15(a).In the deep energy detachment state, the intensity of Ar + ionization source increases sharply, which would contribute to the increase of radiated power loss caused by Ar.The Ar + ionization source gradually extends to the X-point along the separatrix with increasing Ar seeding rate.In order to confirm the movement of ionization, we show the distribution of the ionization of Ar 6+ in figures 15(b) and (d).The ionization of Ar 6+ also gradually moves away from divertor region in the partial energy detachment state to above X-point in the deep energy detachment state.The movement of ionization region towards upstream contributes to the transport of Ar ions towards the core region leading to higher core radiated power in the deep energy detachment state, compared to the partial energy detachment state [46,47].
Figure 16 shows the intensities of D + ionization source and D recombination source in the LO divertor region.In the partial energy detachment state, there is a movement of D + ionization source from around the strike point to the mid SOL region, and slightly away from divertor target.It is because there is a significant decrease of the target electron temperature around the strike point but T e in the mid SOL region is still high enough for ionization of D. In the deep energy detachment state, there are significant decrease of the intensity of D + ionization source and increase of the intensity of D recombination source near the entire LO divertor target, compared to attached and partial energy detachment states.The decrease of D + ionization source is driven by the reduction of the power available for the ionization, named as 'power starvation' [48], which can contribute to the reduction of j s .The intensities of ionization and recombination sources around LO strike point are both in the level of 10 20 -10 21 s −1 m −3 in the deep energy detachment state with T et ∼ 2 eV.However, the low T et (<1 eV) state with recombination over ionization [5] has not been achieved in the deep energy detachment state.It should be pointed out that considering the compatibility of divertor protection and core plasma confinement, T et < 1 eV with much impurity seeding is not necessary and T et < 4 eV may be sufficient [49].Combining the above analysis in section 3.2, the increases of power and momentum losses, and the decrease of p tot,u are all important for the j s reduction in the deep energy detachment state [50].
It should be pointed out that there is an increase of j s in the SOL region in the deep energy detachment state in shot #102531, compared to the attached state, which is not captured by the SOLPS-ITER simulations.It may be because of the effect of E θ × B drift.The E θ × B drift in the SOL region directs towards PFR with Fav.B t (B × ∇B ↓, LSN).In the deep energy detachment state, the E θ × B drift becomes weaker because E θ ∝ ∂Te ∂s ∥ [32] and ∂Te ∂s ∥ near the divertor target becomes pretty small in the deep energy detachment state, as shown in figure 11.Therefore, the particle flux influenced by the E θ × B drift in the SOL region would return to SOL region rather than drift to PFR region.The simulation including drift effect will be performed in the future to uncover the mechanism.

Discussion and summary
Partial energy detachment without confinement degradation, and deep energy detachment with a moderate confinement degradation have both been confirmed on EAST corner slot divertor with Ar seeding.In the partial energy detachment state, the heat flux around the LO strike point can be strongly mitigated without significant confinement degradation.The maintaining of high core plasma confinement is mainly because that the decrease of T e in the core region on the plasma performance is compensated by the increases of n e and T i .In the deep energy detachment state, T et along the entire target decreases to less than 10 eV.The peak q t around the LO strike point decreases by more than 90%, compared to the attached state, and the q t along the entire LO target is close to a flat profile.There are also significant decreases of T div on the entire LO and LI divertor targets with peak T div of only ∼200 • C.There is a moderate confinement degradation in the deep energy detachment state with H 98,y2 from ∼1 to ∼0.9.
The experiment and SOLPS-ITER simulation results have shown that there is not strong enough momentum loss in the SOL region in the partial energy detachment state to drive the decrease of j s .In the deep energy detachment state, the increases of momentum and power losses in the SOL region, and the decrease of upstream pressure all contribute to the decrease of j s .In the detachment process, plasma-neutral interactions gradually dominate the momentum loss and the increase of Ar radiation plays an important role in the power loss in the LO divertor region.In the partial energy detachment state, the radiation power peaks near the divertor target.With increasing Ar seeding, the peak radiation region moves towards X-point and core regions in the deep energy detachment state.The Ar + ionization sources also gradually move from divertor towards X-point, which would strengthen the transport of Ar ions to the core region.
The new corner slot divertor on EAST has been proved to be beneficial for the deep energy detachment operation thanks to its closed geometry.The deep energy detachment with a moderate confinement degradation is an attractive operation mode for the protection of entire divertor target on future fusion reactors.To achieve the deep energy detachment with a better core plasma confinement, impurity seeding from divertor region combined with additional D 2 puff from upstream, named as puff and pump strategy [51], will be performed in the future EAST campaign.It has been confirmed on DIII-D [52] that the additional D 2 puff from upstream is efficient to strengthen the SOL flux to limit the transport of impurity towards core region.The electromagnetic drifts would affect the divertor dissipation [53,54] and impurity transport [55].In the future, we plan to further explore the role of drifts on the detachment process in EAST.

Figure 2
Figure2shows the two discharges investigated in this paper, for partial and deep energy detachment with Ar seeding.The scenario is a LSN H-mode plasma in two shots with plasma current I p = 400 kA, line-averaged electron density at the mid-plane n el ∼ 4.0 × 10 19 m −3 , toroidal magnetic field B t = 2.5 T, favorable B t direction (B × ∇B ↓), and poloidal beta β p ∼ 1.6.In these two discharges, the absorbed power is around 2.2 MW, including 0.4 MW 2.45 GHz low hybrid wave heating, 1.1 MW electron cyclotron resonance heating, 0.6 MW co-current neutral beam injection (NBI) and 0.08 MW ohmic heating.It should be pointed out that there are periodic perturbations in the signal of the stored energy (W MHD ) in shot #102531 because of additional periodic NBI blips, as shown in figure2(f ), which do not influence the analysis in this work.Additional NBI blips are to generate the beam emission spectrum for the motional stark effect diagnostics, which is not used in this work.In shots #102528 and #102531, the peak Ar seeding rate is fixed at 1.1 × 10 20 p s −1 and 1.2 × 10 20 p s −1 , separately.The two shots both work in the pulse width modulation mode[31], which means in each control cycle, the working voltage of the valve is constant, i.e., fixed puffing rate.The feedback controller modulates the pulse width and duty cycle.In shot #102528, the feedback control target T et,target around the LO strike point is 8 eV and constant in the whole shot.The control phase starts at 3 s.At the beginning of the control duration, because of the large control error between the true value and the T et,target , Ar is injected with the longest pulse width until the T et decreases to less than 8 eV.Considering the strong effect of radiation increment caused by Ar, the maximum single Ar pulse length used in the feedback control is set as 12 ms to avoid the confinement degradation.However, because of this single pulse limitation, the feedback control system spends relatively long time (>1 s) to lower T et with 'full load', i.e. using the longest injecting pulse.In shot#102531, the T et,target is set as a slope from 30 to

Figure 2 .
Figure 2. Time evolution of the experimental parameters for partial energy detachment discharge #102528 and deep energy detachment discharge #102531 with Ar seeding.(a) and (e) line-averaged electron density at the mid-plane (n el ) and Z eff .(b) and (f ) plasma stored energy (W MHD ) and H 98,y2 factor.(c) and (g) Tet around the LO strike point, the detachment control target Tet,target and radiation power in bulk plasma measured by AXUV (P rad,bulk ).(d) and (h) Ar seeding rate v Ar and Ar XVI line emission around mid-plane measured by EUV.

Figure 3 .
Figure 3. Profiles of js, Tet, p et,static and qt on the LO divertor target measured by LP in the attached state (3.30-3.40s in shots #102528 and #102531), the partial energy detachment state (5.10-5.20 s in shot #102528) and the deep energy detachment state (7.10-7.20 s in shot #102531).The qt in the SOL region in partial and deep energy detachment states are shown in the insets in (d) and (h).Every point is the averaged value of 10 ms.These dashed lines are corresponding to the positions of the LO strike point.The positive and negative positions are corresponding to the LO vertical and horizontal targets, separately.Unfortunately, Langmuir probes measurements of Tet, p et,static and qt are not available in the private flux region (PFR).

Figure 4 .
Figure 4.The time evolution of surface temperature T div on the LI and LO divertor targets in shots #102528 and #102531.The magenta lines are corresponding to LI and LO strike points.

Figure 5 .
Figure 5. Profiles of Te (a), T i (b), ne (c) and pe (d) around the mid-plane in the stable attached state (3.3 s) and the partial energy detachment state (5.1 s) with Ar seeding in shot #102528.Insets in all: pedestal and SOL region profiles in the range of 0.9 < ρ < 1.02.

Figure 6 .
Figure 6.The chord-integrated radiation power intensity measured by AXUV in shots #102528 and #102531.The vertical coordinate in (b) corresponds to the height (value of Z) on EAST at R = 1.85 m as the red dashed line shows in (a).The blue dashed line in (b) corresponds to the position of the lower X-point.

Figure 7 .
Figure 7. Profiles of Te (a), T i (b), ne (c) and pe (d) around the mid-plane in the stable attached state (3.3 s) and the deep energy detachment state (7.1 s) with Ar seeding in shot #102531.Insets in all: pedestal and SOL region profiles in the range of 0.9 < ρ < 1.02.

Figure 8 .
Figure 8.(a) SOLPS-ITER computational grid based on the magnetic equilibrium of shot #102528.The black triangles are the Eirene grid.The radial transport coefficients at the OMP without (b) and with (c) Ar seeding.

Figure 9 .
Figure 9.The comparison of the radial profiles of Te and ne at the OMP between simulation and experiment results in the attached (a), (b) and partial energy detachment (c), (d) states.

Figure 10 .
Figure 10.The radial profiles of Tet (a), (d), js,p (b), (e) and net (c), (f ) on the LO divertor target in the attached, partial energy detachment and deep energy detachment states.The circles are the experiment results and the lines are the simulation results.

Figure 11 .
Figure 11.The poloidal distributions of ne (a), Te (b), ptot (c) and poloidal power flux qp (d) from the OMP to the LO divertor target along the third flux tube, where the heat flux in the attached state without Ar seeding peaks.(e) Portion of the flux tube used to calculate the momentum and energy balances.

Figure 12 .
Figure 12.Momentum losses along the flux tubes between the upstream and LO divertor target normalized to ptot,u as a function of the distance from the separatrix, and the decompositions of the total poloidal integrated momentum source.(a) The attached case.(b) The partial energy detachment case.(c) The deep energy detachment case.

Figure 12 (
Figure 12(c) shows that, with increasing Ar seeding rate, f mom-loss near the separatrix increases to larger than 0.8 in the deep energy detachment state, which can contribute to the

Figure 13 .
Figure 13.Power losses along the flux tubes between the upstream and divertor target normalized to the q ||u A ||u as a function of the distance from the separatrix, and the decompositions of the total poloidally-integrated power source.(a) The attached case.(b) The partial energy detachment case.(c) The deep energy detachment case.

Figure 14 .
Figure 14.The 2D distributions of Ar line radiation power loss including Ar 0 -Ar 18+ .(a) The partial energy detachment case.(b) The deep energy detachment case.

Figure 15 .
Figure 15.The 2D distribution of ionization of Ar 0 and Ar 6+ for partial energy detachment case (a), (c) and deep energy detachment case (b), (d).

Figure 16 .
Figure 16.The intensities of D + ionization source and D recombination source in flux tubes in the separatrix (a), (d), mid SOL region with ∼4.5 cm to the separatrix (b), (e) and far SOL region with ∼11.4 cm to the separatrix (c), (f ) for the cases of attachment, partial and deep energy detachment.