Mechanism of enhanced ion temperature by impurity seeding in EAST H-mode plasma

Ion temperature (T i) profiles are commonly observed to increase in peaking, leading to higher central T i, after impurity seeding in the EAST H-mode plasma. Argon can be more efficient at raising T i than neon. Toroidal rotation can also be enhanced in scenarios with NBI heating. A more significant increase in toroidal rotation is brought about by seeding argon than seeding neon. Turbulence is experimentally observed to be suppressed. Extensive modeling using the quasilinear gyrokinetic code QuaLiKiz is performed to explain the above observations. It is found that the enhanced T i can always be explained by the turbulence stabilization. However, the mechanism of turbulence stabilization is related to heating methods and the seeding impurity species. In the pure RF (ECRH + LHW) heating scenarios, where only the trapped electron mode (TEM) exists, argon can stabilize the TEM more significantly than neon due to its higher charge and heavier mass. In scenarios with increasing NBI power, the ion heat flux can be dominated by the ion temperature gradient (ITG), thus the enhanced T i is mainly attributed to ITG stabilization. In these cases, except argon’s ability to more efficiently stabilize TEM, more evident increased toroidal rotation brought about by argon seeding can also be beneficial to stabilize turbulence.

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
Radiative divertor detachment with impurity seeding (nitrogen, neon and argon) offers a promising solution for the control of heat and particle fluxes on the divertor target [1].In addition to the increased power radiation and fuel dilution brought about by impurity seeding, turbulence can also be influenced, leading to a change in ion temperature [2].Besides, achieving high core ion temperatures is one of the main goals for future fusion reactors.Thus, the study of the influence of impurity seeding on turbulence and ion temperature is meaningful to achieve the compatibility of high core performance and divertor protection.
The enhanced ion temperature induced by impurity seeding through the stabilization of turbulence has been observed in several magnetic fusion devices, such as DIII-D [3,4], JET [5][6][7], WEST [8], ASDEX-U [9,10], JT-60U [11], Alcator C-Mod [12], LHD [13,14] and HL-2A [15].EAST has demonstrated a compatible core and edge integration in high β P scenarios sustained by active impurity seeding [16] and a large number of divertor feedback control experiments have also been performed on EAST [17][18][19].In these experiments, the central T i is commonly observed to increase.The relevant physics mechanism needs to be further investigated.Based on the previous study, the ion heat transport is mainly determined by the turbulence [16].Thus, in this paper, the influence of impurity seeding on turbulence is investigated to explain the enhanced central T i .
The trapped electron mode (TEM) and ion temperature gradient (ITG) mode are the main candidate micro-instabilities that can drive substantial heat transport [2].The influence of impurity on turbulence has been modelled in different conditions, such as TEM [20][21][22], ITG [23][24][25], coexistence of TEM and ITG [26,27], etc. Modeling shows that impurity species and their density profiles have impacts on the TEM.In addition, the increased collisionality brought by impurity seeding can also impact turbulence [20,28,29].For the ITG, it is mainly through the dilution of the main ions that influence its threshold [30][31][32].In this paper, the effects of impurity on turbulence were extensively modeled with the quasi-linear gyrokinetic code QuaLiKiz [33] through studying the growth rates of turbulence and the corresponding driven ion heat flux in different situations.The effect of impurity on turbulence is also compared with toroidal rotation, as toroidal rotation increases after impurity seeding, especially when argon is seeded in plasma with NBI heating, which can stabilize turbulence by increasing E × B shear flow.Additionally, ONETWO [34] is applied to evaluate the power deposition.
This paper is organized as follows.In section 2, the experimental observations will be introduced, including the influence of impurity seeding on the T i profile, turbulence, toroidal rotation, etc.The differences between seeding argon and neon are also presented in different heating scenarios.In section 3, modeling tools are described and applied to explain the experimental observations.We mainly focus on the evaluation of the influence of impurity species, impurity density gradient and toroidal rotation on turbulence in different situations.Finally, the conclusions and discussions are given in section 4.

Enhanced central T i by neon seeding
In EAST radiative divertor feedback control experiments, it is commonly observed that seeding impurity can raise the central T i [17,18].One EAST typical high β P discharge (H 98y2 > 1.2, β P ∼ 2.5, β N ∼ 2.0, f bs ∼ 50%, n e /n GW ∼ 0.7, q 95 ∼ 6.7,I p = 400 kA, B t = 2.4 T, P LHW+EC ∼ 3.4 MW and P NBI ∼ 5.4 MW) is shown in figure 1.In this shot, the peak heat flux is reduced by 20%-30% on an ITER-like tungsten divertor by seeding a mixture of 50% neon and 50% D 2 near the upper outer strike point.It is noticeable that central T i increases while T i remains unchanged at ρ = 0.45.Moreover, one can see that T i becomes more peaked from ρ = 0.45-0.2,leading to higher central T i , according to T i profiles before and after neon seeding shown in figure 2, as well as the corresponding R/L Ti , where R/L X is defined as R/L X = −X * R/(dX/dr).As for T e , n e and q, they are marginally influenced, as seen in figures 3(a)-(c), respectively.Meanwhile, energy confinement quality is maintained and plasma performance remains stable, showing good compatibility of core and edge conditions.In addition, the effective charge number (Z eff ) and line emission of NeVIII increase, implying that neon is transported to the core plasma.The increased Z eff will be used to calculate neon content in the modeling analysis.It is also interestingly noted that the tungsten content increases significantly while the line emission of CVI is slightly impacted.It is also worth mentioning that the toroidal rotation in this shot slightly increases not shown in this paper.Furthermore, power deposition analysis using ONETWO before and after neon seeding is performed as illustrated in figure 3(d).We find that Q i is marginally impacted by neon seeding while Q e significantly decreases because of the increased radiation power.

Comparison between neon and argon seeding
In addition to neon, argon is also applied in radiative diverter experiments.In some of these discharges, the central T i can almost double, which is meaningful for raising the EAST central T i .However, no such effect is achieved when neon is seeded.Two typical discharges with neon and argon seeding from the lower outer divertor are compared in figure 4. They have similar plasma parameters: lower single null configuration, I p = 400 kA, n e ∼ 4 × 10 19 m −3 , T e (0) ∼ 4 keV, stored energy W MHD ∼ 170 kJ and B t ∼ 2.4 T, P LHW = 2.8 MW/P NBI = 1.4 MW/P EC = 1.5 MW for discharge #102544 and P LHW = 2 MW/ P NBI = 1.4 MW/P EC = 1.4 MW for discharge #103777.It can be seen that with similar increased Z eff , central T i with argon seeding increases almost by twice while for the neon seeding case central T i only increases by 100-200 eV.Here, the increased Z eff is considered mainly to be brought about by the neon or argon.Combined with quasi-neutrality of the plasma, the contribution of each impurity charge state to Z eff can be expressed as δZ eff = n Z * Z (Z − 1) /n, where Z and n Z represent charge number and density, respectively.As with such T e the averaged charge number of argon, which can be  (a) T i profiles from CXRS (charge exchange recombination spectroscopy) [35] and (b) the corresponding R/L Ti before and after neon seeding for discharge #94437.calculated by STRAHL [38], is much higher than neon, the concentration of argon will be much lower than neon with similar increased Z eff .Thus, it can be concluded that even though the concentration of argon is lower than neon, argon is still more efficient to raise central T i .Comparisons of T i profiles between neon and argon seeding, as well as their corresponding R/L Ti , are illustrated in figure 5.These T i profiles, as well as the following T i profiles in this paper, are measured by tangential x-ray crystal spectrometer (TXCS) [39].It can be seen that compared with neon seeding, a more peaked T i profile by argon seeding leads to higher central T i .
In addition, it is worth stressing that toroidal rotations are observed to increase after both neon and argon seeding at different levels as shown in figure 4, which seems to be related to the degree of enhanced central T i .More significantly raised toroidal rotation can be brought about by argon seeding.The increased toroidal rotation can enhance the E × B shear flow and contribute to the suppression of turbulence.Moreover, we observe that after argon seeding turbulence is suppressed as shown at the bottom of figure 4. Turbulence is monitored by the CO 2 laser collective scattering system [40].In addition, comparisons of impacts on n e and T e between seeding argon and neon are shown in figure 6.It can be seen that similar to discharge #94437, T e profiles are marginally influenced.For n e , in discharges #102544 and #103777, central n e increases by 10%-15%, which may be caused by the ionization of impurity ions in the core region.More comparisons of the influence of plasma between seeding argon and neon can be found in  [ 18,19,27].In these references, it is always found that T i is enhanced while T e and n e are limited impacted after impurity seeding.
Regardless of heating methods, T i can always be enhanced by impurity seeding according to EAST experiments.Enhanced T i in the NBI dominating and RF (LHW + ECRH) dominating heating plasmas are presented above.Here, in cases with pure RF heating discharges as well as in other cases with NBI dominating heating discharges, enhanced T i by different impurities seeding is given in figure 7   are #116411 and #116409 with argon and neon seeding, respectively, and they have the same heating power (P LHW = 1.8 MW/P EC = 0.85 MW/P NBI = 3.7 MW).Similar to the above observations, the enhanced central T i is due to the more peaked T i profiles and argon has a stronger ability to enhance central T i than neon.Moreover, it is worth mentioning that in pure RF scenarios, after impurity seeding, toroidal rotation can decrease, contrary to the scenarios with NBI.The toroidal rotations for discharges #113552 and #85291 decrease by 20 km s −1 and 10 km s −1 , respectively, while for discharges #116411 and #116409, the toroidal rotations increase by 14 km s −1 and 6 km s −1 , respectively.

Simulation setup
To research the underlying mechanism for the enhanced T i , the latest updated QuaLiKiz [33] is applied to perform instability analysis and ONETWO [34] is applied to perform the   power balance analysis, taking into account power deposition by auxiliary heating, collisional equipartition and radiated power.In QuaLiKiz, TEM and ITG are included with the poloidal wave number spectrum k θ ρ s in the range of 0.1-2, with ρ s = m i v th,i / (eB), and the radial wave number is taken at k x ρ s = 0.The quasilinear weight is given by γ/ < k 2 ⊥ >, with the operator < > being an average over the mode structure along the field line.Electron, deuterium (main ions), hydrogen, carbon and tungsten are included in the modeling.The ratio of the hydrogen content to the sum content of hydrogen and deuterium is set to be 7% according to the measurement of the optical spectroscopic multichannel analysis system [41].Carbon content is calculated from Z eff before impurity seeding.Tungsten's content is set low enough to satisfy the trace limit.The seeded impurities (argon and neon) are included separately to study their own influence on turbulence.The form of the impurities density profiles is assumed to be the same as the electron density.In addition, it is worth mentioning that an arbitrary number of traces or active ions can be accounted for in QuaLiKiz, making it possible to study the influence of impurities on turbulence.In ONETWO, GENRAY-CQL3D, TORAY-GENRAY and NUBEAM are invoked to calculate the power depositions of LHCD, ECRH and NBI, respectively [34].

Instability analysis
In figure 2, T i starts to become more peaked at ρ = 0.4 and this position is also located at the foot of eITB, where ITG most likely exists [42].The following analysis will focus on this position.The main input data and plasma parameters are listed in table 1.The ratio of electron to ion temperature T e /T i is from figures 2(a) and (b).R/L ne , R/L Te and R/L Ti are evaluated from the measured relevant profiles.R/L Ti has a large uncertainty because NeXI line emission is applied to measure T i instead of the routinely used CVII line emission which is polluted by other neon line emission.Safety factor (q) and magnetic shear (ŝ) are from figure 3(c).The measured line averaged Z eff is applied.Electron collisionality ν * is high enough to be nonnegligible in modeling.Q i and Q e are calculated by ONETWO and can be considered as the experimental measured data, which will be compared with those driven by turbulence modeled by QuaLiKiz.
In QuaLiKiz, the impact of collisionality on TEM turbulence stabilization is overestimated because of the reduced Krook operator [8].The latest version has a new collision operator which works well in the moderate and low collision region but still needs to be improved in the high collision region [43].Thus, the collision multiplier (collmult) of QuaLiKiz is benchmarked by the gyrokinetic code GKW.GKW has a more accurate collision operator but needs more time.So QuaLiKiz will be applied to perform extensive modeling after tuning the collmult.
With the main input data in table 1 and same k θ ρ s in GKW and QuaLiKiz, it is found that when collmult = 0.21, modeling results of QuaLiKiz match well with GKW as shown in figure 8.It can be noted that the growth rates of TEM by QuaLiKiz are close to the GKW modeling results for all k θ ρ s .There is a small difference for the frequencies from 0.1 to 0.5 but when k θ ρ s > 0.5 frequencies are almost overlapped.Thus, in the following modeling, collmult is set to be 0.21.Besides, it is worth mentioning that QuaLiKiz can get the three most unstable modes for each k θ ρ s , which is meaningful for the situation of coexistence of TEM and ITG to evaluate their own Q i .In figure 8, by QuaLiKiz, ITG can be found from k θ ρ s = 0.6-1.6 but not found by GKW as GKW only obtains the most unstable mode.
To match well with heat deposition evaluated by ONETWO, two D scanning of R/L Ti , R/L ne within their possible error bars with different R/L Te are performed by QuaLiKiz.When R/L Te in table 1 is applied, no region can be found to satisfy the experimental Q e because of very strong Q e driven by R/L Te .Thus, R/L Te is adjusted to be smaller within its uncertainty.It is found that when R/L Te = 9.5, within the uncertainties of R/L Ti and R/L ne , region both satisfy experimental Q i and Q e can be found as shown in figure 9(a), and the corresponding R/L ne and R/L Ti are 2.2 and 5.3, respectively.Thus, to obtain a more convincing instability analysis, such modified R/L Te , R/L ne and R/L Ti will be applied in the following analysis.With these modified data, as well as other input sets described above, it is found that ITG and TEM coexist by QuaLiKiz as shown in figures 9(b) and (c).The ITG can be found at low k θ ρ s from 0.25 to 0.5 and is the most unstable mode from k θ ρ s = 0.3-0.5.TEM can be found from k θ ρ s = 0.2-1.2,overlapping ITG regions, and its largest growth rate is larger than ITG.

Modeling for neon seeding
Due to the ability of QuaLiKiz to separately output Q i driven by TEM and ITG, evaluation of their own contributions to Q i becomes possible.In figure 10(a) Q i separately driven by TEM and ITG versus impurity content is shown.Here, the impurity contribution to Z eff (δZ eff ) is applied as a proxy for the impurity content.It can be seen that, when no neon (δZ eff = 0) is included, Q i driven by ITG is about three times larger than TEM, even though the largest growth rate of TEM is larger than ITG as shown in figure 9.In addition, the sum of Q i driven by TEM and ITG well with the measured one.With increasing neon content, Q i driven by both ITG and TEM decreases but the former declines more quickly than the latter.The normalized growth rate and frequency spectra are compared for cases with and without neon in figures 10(b) and (c), respectively.For the case with neon, where δZ eff is set to be 0.2, the growth rates of both ITG and TEM decreases, corresponding to the decreasing Q i in figure 10(a).High k θ ρ s from 0.7 to 1.2 of TEM is even totally stabilized.In addition, the frequency spectra do not change except for the totally stabilized k θ ρ s .Thus, it can be concluded that both stabilization of TEM and ITG contribute to the enhanced T i .As Q i is mainly driven by ITG and it declines more quickly with increasing neon content than TEM, it can be concluded that the enhanced T i is mainly attributed to the stabilized ITG.
In the above situation, Q i is comparable to Q e .Different Q e /Q i can be achieved by varying heating methods (electron or ion heating) and the enhanced T i can always be observed by impurity seeding.These different situations are also discussed, including 1. Q i driven by TEM is comparable to ITG; 2. Q e is much higher than Q i ; 3. Q e is lower than Q i , possibly corresponding to strong RF heating combined with weak NBI, pure RF and strong NBI heating combined with weak RF, separately.R/L Ti is varied to illustrate these situations, which are 5.1, 4.6 and 7, respectively.The turbulence driven Q i of the first two cases matches well with the measured Q i for similar discharges but with less NBI heating power [16,44] as shown in figures 12 and 13, respectively.However, for the case with R/L Ti = 7, the corresponding Q i is much larger (almost ten times larger) than the experimental measured one not shown in figure 13 because once R/L Ti is beyond the threshold of ITG, Q i will be very sensitive to R/L Ti (strong stiffness).Pure ITG in the plasma can be achieved by NBI dominating heating [45].The effect of neon on turbulence in these three situations is investigated as shown from figures 11-13.It is found that turbulence is always stabilized when neon is included.In figure 11, when R/L Ti = 5.1, TEM and ITG coexist, but the growth rate of TEM is larger allover k θ ρ s .With increasing impurity content, Q i driven by these two different instabilities declines at almost the same rate.In figures 12 and 13, only TEM and ITG can be found when R/L Ti = 4.6 and 7, respectively.Their driven Q i always declines with increasing neon content.Therefore, the enhanced T i by neon seeding in EAST experiment regardless of heating methods can always be attributed to stabilization of turbulence.We discuss which kind of turbulence is stabilized, and how it depends on the Q i driven source, which is related to heating methods.

Effect of different R/Lnz on turbulence
Impurity effect on instability can be influenced by the impurity density gradient, especially with hollow or peaked profiles, corresponding to the negative and positive R/L nz , respectively.In the above modelling, R/L nz is set to be similar to positive R/L ne .In this section, the effect of negative R/L nz on the instability is investigated.It is found that with negative R/L nz , behaviors of TEM and ITG are different from the situations with positive R/L nz as shown in figure 14.For ITG, contrary to positive R/L nz , it is unstabilized with negative R/L nz .For TEM, it is always stabilized whether R/L nz is positive or negative, but stabilization effects can be more efficient with negative R/L nz .With unstabilized ITG and stabilized TEM when R/L nz is negative, their overall Q i increases, which does not match the experimentally observed enhanced T i , implying that the positive R/L nz is much closer to the real situation.Thus, in the following modeling, R/L nz is always set to be positive.

Effect of different impurities on turbulence
As neon and argon have different effects on raising T i , their impacts on instabilities will be compared in this section.In addition to the divertor radiative experiments, an impurity is also injected to improve the wall condition and boron is   applied.To better compare impacts of different impurities on instabilities, boron with lower atomic number ( 5) is also included.The atomic numbers of neon and argon are 10 and 18, respectively.As turbulence is always stabilized in different situations when neon is included as presented above, modeling is mainly focused on pure ITG and pure TEM.In the modeling, neon and boron are fully ionized while argon is partially ionized and its average charged number is 16 calculated by STRAHL.
The effect of impurity on ITG is firstly investigated as shown in figure 15.It can be seen that with increasing δZ eff , Q i in the situation with boron declines most quickly, implying that boron is most efficient to enhance T i .This can be explained by the mechanism of stabilization of ITG by   impurity, which is through the dilution of main ions to change its threshold.Here, the impurity dilution effect is evaluated for the lost main ions replaced by impurity, which is δZ eff * n e /(Z-1).It can be seen that with fixed δZ eff , boron with the lowest Z of the three impurities is most effective to dilute main ions.The growth rates and frequency spectra are also checked with δZ eff = 0.6 in figure 15(b).As expected, the situation with boron has the lowest growth rate.Additionally, it is interestingly noted that the frequency of the high part k θ ρ s of boron is lower than that of the other two impurities in figure 15(c).
As for TEM, contrary to ITG, the impurity with highest Z is most efficient at enhancing T i in figure 16(a), corresponding to the lowest growth rates in figure 16(b).The influence of impurity seeding on TEM is complicated.On the one hand, the increased collisionality can stabilize TEM through detrapping trapped electrons [20,28,29].On the other hand, the mass and charge of impurity can influence TEM by impacting the dispersion relations through the Larmor radius, V th and quasineutrality condition [20][21][22].In order to evaluate how these three factors (collisionality, mass and charge) contribute to the stabilization of TEM, relevant modeling is performed as shown in figure 17.For the cases with impurity included, collisionality is set to be same with fixed δZ eff = 0.4.
In order to study the impact of collisionality, the charge of neon is set high enough (Z = 1000) to minimize the dilution effect on main ions (the dilution effect can be evaluated with δZ eff * n e /(Z-1)).In this situation, neon is set to be not included in the dispersion relation and only contributes to collisionality through δZ eff .As shown in figure 16, when only collisionality is included, the TEM can be stabilized.To study the charge impact, neon will be included in the dispersion relation by recovering its own Z, which is 10, while its mass is set to be the same as that of the main ions (deuterium) to eliminate   the influence of mass.It can be seen that with the same collisionality, when charge is also considered, TEM can be further stabilized.Following the above modeling, when recovering neon's mass and further replacing it with argon's mass, it can be seen that TEM can be stabilized in succession.Hence, it can be concluded that additional increased collisionality, mass and charge contribute to the stabilization of TEM.But as those three factors couple to each other, it is difficult to compare their own effects on TEM.So it is hard to say which factor plays the most important role in stabilizing TEM.
Therefore, it can be inferred that in pure RF heating scenarios where Q i could be dominated by TEM, even with same δZ eff (same increased collisionality), the reason why argon is more efficient than neon to raise central T i is that argon can be more significant to stabilize TEM due to its heavier mass and higher charge.Furthermore, when concentrations of neon and argon are the same, argon can bring more δZ eff (collisionality) because of its higher charge, which can be further beneficial to stabilize TEM.

Effect of toroidal rotation on turbulence
Toroidal rotation can be beneficial to stabilize turbulence by increasing the E × B shear flow, which can also be helpful to raise T i .Increased rotation is observed after impurity seeding in scenarios with NBI heating, especially with argon seeding.To investigate the impact of such increased toroidal rotation on turbulence, modeling by including different factors (impurity or toroidal rotation) is performed.In the modeling, the radial electric field induced by NBI [46] and the increased toroidal rotation being 10 kV m −1 and 15 km s −1 , separately.Gradients of toroidal rotation are not included.δZ eff is set to be 0.8.
In figure 18, R/L Ti is set to be 7 and only ITG is found.It can be seen that when rotation is included, in the range of k θ ρ s > 0.38, the growth rate of ITG declines and  this stabilization effect can be comparable to dilution effect brought by impurity, implying the increased toroidal rotation also play roles in enhancing T i .This may explain the contradiction in section 3.5 that in the modeling it is found impurity with low Z can be more efficient to stabilize ITG, while in experiment with NBI dominating heating plasma impurity with high Z is more evident to raise central T i .As for the TEM dominating regime with R/L Ti = 4.6 in figure 19, in the range of k θ ρ s > 0.4, TEM can be efficient to be stabilized by the toroidal rotation.However, different from ITG, stabilization effects on TEM by impurity are more efficient than such increased toroidal rotation.This means that for the plasma whose Q i is dominated by TEM, the enhanced T i is mainly attributed to the impurity intrinsic stabilization effect on TEM, which accords with the experimental observation that in pure RF heating plasma even though toroidal rotation declines after impurity seeding, the central T i is still enhanced.Additionally, in the low range of k θ ρ s , it is found that both TEM and ITG are marginally influenced by such increased toroidal rotation.

Conclusion and discussion
In EAST H-mode plasma, enhanced central T i is commonly observed by impurity seeding through more peaked profiles.Levels of enhanced T i are related to seeding impurity types and argon can be more efficient at raising T i than neon.In shots with NBI, toroidal rotation can be enhanced, especially when argon is seeded.In addition, turbulence is observed to be suppressed.
The underlying mechanism of such enhanced T i is investigated by extensive modeling with QuaLiKiz.In addition, ONETWO is applied to evaluate the power deposition (Q i and Q e ).Modeling is first focused on neon seeding plasma heated by RF and NBI.It is found that at ρ = 0.4 where T i begins to become more peaked, the enhanced T i is mainly attributed to the stabilization of ITG.In addition, even though the largest growth rate of TEM is larger than ITG, Q i is dominated by ITG.Furthermore, as the enhanced T i is always observed in EAST H-mode plasma regardless heating methods, R/L Ti is adjusted to change turbulence property to realize different Q i /Q e , which corresponds to the different heating scenarios.Three other different situations are discussed, including 1. Q i driven by TEM is comparable to ITG; 2. Q e is much higher than Q i (only TEM exists); 3. Q e is lower than Q i (only ITG exist), possibly corresponding to strong RF heating combined with weak NBI, pure RF and strong NBI heating combined with weak RF, respectively.It is found that the enhanced T i can always be explained by the turbulence stabilization.Types of stabilized turbulence (ITG, TEM or both of them) depending on the dominating driven source of Q i .
Impacts of impurity species and toroidal rotation on turbulence are also investigated.It is found that with fixed δZ eff low Z impurity (neon) can be more efficient to stabilize ITG because it can more efficiently dilute main ions, whilst high Z (argon) is more beneficial to stabilize TEM due to its higher charge and heavier mass.Furthermore, with the increased δZ eff and the increased toroidal rotation in experiments, the stabilization effect of TEM by impurity is much stronger than toroidal rotation.However, such increased toroidal rotation plays an important role in stabilizing ITG.
Based on the above modeling, observations that argon is more efficient at raising T i can be well explained.On the one hand, in pure RF heating where Q i is mainly driven by TEM even with decreased toroidal rotation, argon can be more efficient to stabilize TEM and such stabilizing effect can be much stronger than that brought by toroidal rotation.On the other hand, in the hybrid heating scenarios, with increasing NBI heating, Q i can be dominated by ITG.Under this circumstance, except argon's stronger ability to stabilize TEM, more increased toroidal rotation by seeding argon can also benefit to turbulence stabilization, especially for ITG, which can also contribute to the increased T i .That is why when Q i is dominated by ITG, even though neon has a stronger ability to stabilize ITG, seeding argon can still more evidently Another nonnegligible difference between argon and neon is their profiles form in the core plasma, which can impact the impurity's effect on turbulence.However, so far, for EAST, density profiles of argon and neon are unavailable.In the modeling, R/L nz is assumed to be the same as R/L ne , which is positive.It turns out that this assumption is much more coincident with the real situation when compared to the negative R/L nz .In the future, with the improvement of the EAST impurity diagnostic, the measured impurity profile will be applied to the modeling.Moreover, the mechanism of increased toroidal rotation in the scenarios with NBI heating requires further investigation.In addition, in some discharges, the safety factor q is also influenced by impurity seeding [47].
In the modeling, the overestimation of stabilization of TEM by QuaLiKiz is solved by benchmarking collmult with GKW.Thanks to the fast calculation of QuaLiKiz by making some rational approximation, extensive modeling by QuaLiKiz becomes highly efficient.Furthermore, its ability to output the three most unstable at a same k θ ρ s , as well as the corresponding Q i , makes evaluation of Q i separately driven by different turbulences become possible, which is very meaningful for the case when ITG and TEM coexist at same k θ ρ s .

Figure 1 .
Figure 1.From top to bottom, time traces of the plasma current and gas puff, heating power, and central Te and volume averaged ne, T i at ρ = 0 and 0.45, plasma's stored energy and poloidal beta, Z eff and line emission of NeVIII, power radiation from the bulk plasma and line emissions of tungsten, Dα and line emission of CVI.

Figure 2 .
Figure 2. (a) T i profiles from CXRS (charge exchange recombination spectroscopy) [35] and (b) the corresponding R/L Ti before and after neon seeding for discharge #94437.

Figure 4 .
Figure 4. From top to bottom, time traces of gas puff, central Te and Z eff and toroidal rotation by TXCS for argon seeding scenario (#102544) and neon seeding scenario (#103777), separately.Evolution of frequency-integrated spectral power of density fluctuations with k = 12 cm −1 in plasma core ρ = 0-0.4for #102544 is added at the bottom of the figure.

Figure 5 .
Figure 5. (a) T i profiles by TXCS and (b) the corresponding R/L Ti before and after argon seeding; (c) T i profiles by TXCS and (d) the corresponding R/L Ti before and after neon seeding.

Figure 7 .
Figure 7. T i profiles before and after different impurities seeding in discharges with (a), (b) RF and (c) NBI dominating heating power, respectively.

Figure 8 .
Figure 8. Normalized (a) growth rate and (b) frequency spectra versus k θ ρs by GKW and QuaLiKiz with the input shown in table 1, respectively.

Figure 9 .
Figure 9. (a) The possible regions satisfy experimental Q i and Qe with R/L Te = 9.5, (b) normalized growth rate and (c) frequency spectra versus k θ ρs after adjusting R/L Te , R/Lne and R/L Ti .

Figure 10 .
Figure 10.(a) Q i of ITG and TEM versus δZ eff , (b) normalized growth rate and (c) frequency spectra versus k θ ρs for the phases with and without neon seeding.The yellow region represents the possible experimental δZ eff (currently, only the line averaged Z eff is obtained on EAST, so the error-bars of δZ eff will be large).

Figure 11 .
Figure 11.(a) Q i of ITG and TEM versus δZ eff , (b) normalized growth rate and (c) frequency spectra versus k θ ρs for the phases with and without neon seeding with R/L Ti = 5.1.

Figure 12 .
Figure 12.(a) Q i of TEM versus δZ eff , (b) normalized growth rate and (c) frequency spectra versus k θ ρs for the phases with and without neon seeding with R/L Ti = 4.6.

Figure 13 .
Figure 13.(a) Q i of ITG versus δZ eff , (b) normalized growth rate and (c) frequency spectra versus k θ ρs for the phases with and without neon seeding with R/L Ti = 7.

Figure 14 .
Figure 14.Normalized growth rate and frequency spectra versus k θ ρs with different R/Lnz.

Figure 15 .
Figure 15.(a) Q i of ITG versus δZ eff , (b) normalized growth rate and (c) frequency spectra versus k θ ρs with different impurities.

Figure 16 .
Figure 16.(a) Q i of TEM versus δZ eff , (b) normalized growth rate and (c) frequency spectra versus k θ ρs with different impurities.

Figure 17 .
Figure 17.Normalized growth rate and frequency spectra versus k θ ρs for different cases to compare effects of collisionality, charge and mass.

Figure 18 .
Figure 18.Normalized growth rate and frequency spectra of ITG versus k θ ρs in conditions with or without impurities or toroidal rotation with R/L Ti = 7.

Figure 19 .
Figure 19.Normalized growth rate and frequency spectra of TEM versus k θ ρs in conditions with or without impurities or toroidal rotation with R/L Ti = 4.6.

Table 2 .
The mechanism of the enhanced T i by impurity seeding.