Characterisation of the scrape-off layer in JET-ILW deuterium and helium low-confinement mode plasmas

,


Introduction
The ITER Pre-fusion Power Operation phase foresees experiments in hydrogen and helium (He) plasmas to demonstrate plasma operation in low-confinement mode (L-mode) [1].ITER -and future powerproducing reactors -require divertor detachment to reduce the heat flux onto the divertor target to powers <20 MW m −2 [2][3][4].Experiments for the study of the scrape-off layer (SOL) and detachment in plasmas with He as the main ion species have previously been performed in JET with an all-carbon wall, JET-C [5,6], the carbon-walled DIII-D [7], and the water-cooled tungsten divertor tokamak EAST [8].This paper presents experiments with He and deuterium (D) plasmas performed in JET ITER-like wall (JET-ILW), which has beryllium wall tiles and a tungsten divertor [9].
It is necessary to simulate detachment for the planning of operating current reactors and the design of future reactors.Current edge plasma simulations for hydrogenic plasmas do not quantitatively reproduce all experimental observations.In hydrogenic simulations, divertor current at the onset of detachment and the radiated power is under-predicted by a factor of 2 by the 2D edge fluid codes SOLPS-ITER and EDGE2D-EIRENE [10,11].Mass rescaling of molecular charge exchange rates changes the divertor neutral density by 100% [12].Helium plasmas offer a way to validate codes and SOL physics understanding through comparison with the more typical D plasmas.He has the highest first ionisation energy, longer mean-free-path of ionisation, and lacks molecules.

Experimental setup
Experiments with 4 He and D neutral beam injection (NBI) heated L-mode plasmas were conducted in JET-ILW with JET-typical machine parameters of ⃗ ×∇ towards the divertor, an on-axis toroidal field,  T , of 2.5 T, a plasma current,  p , of 2.4 MA, and an edge safety factor,  95 , of 3.2.The experiments were conducted in a vertical-horizontal divertor configuration -optimised for diagnostics and edge modelling [13] -where the high-field side (HFS) strike point is on the HFS vertical plate and the low-field side (LFS) strike point is on the horizontal plate (Fig. 1).Fuelling was performed through gas injection to the HFS https://doi.org/divertor either continuously, for fuelling ramps, or intermittently to hold the plasma at set densities, for fuelling steps.NBI heating was applied with the same species as the plasma main ion, at  NBI = 1 MW in D and 1 MW, 2 MW, and 5 MW in He.The highest total heating power,  T , was 3.5 MW and 7 MW for D and He, respectively.Higher   in L-mode confinement were permissible in the He plasmas due to the higher L-H transition power threshold in He [14].The purity across all He pulses was  He =  He ∕ e = (49.5 ± 0.2)%, as measured by the sub-divertor optical Penning gauge [15].The experimental setup, characterisation, and modelling efforts of the D plasmas have been previously described in Refs.[10,13,[16][17][18][19][20].
The He experiments utilised argon frosting of the divertor cryogenic panel for improve density control but the pumping rate in these plasmas is unknown.A study of Ar-frosting in the MkII-GB configuration of JET found a maximum pumping rate for He of 95 m 3 s −1 .The pumping rate was not constant through the pulse and dropped to as low as 20 m 3 s −1 .The pumping rate for D 2 was 125 m 3 s −1 [21].Thus the He pumping performance in this works plasmas is also likely poor.
Sweeps of the strike points over the divertor targets were conducted to produce continuous profiles of the ion saturation current density,  sat , electron temperature,   , and electron density,   , measured by the divertor Langmuir probes (LPs, Fig. 1).Integration over the   profile provided the total ion current to the target plates,  div .The LP measurements of   have an ion mass and charge dependence, and require an assumption of the charge state of impinging ions.The effective charge was assumed to be  ef f = 1 and 1.5 for D and He, respectively.For D plasmas,  div is directly proportional to the ion flux  D+ div but for He plasmas, total He ion flux is between 0.5−1 ×  D+ div depending on the ratio of He +2 to He+ impinging the targets.The peak plasma pressure at the target was estimated from the LPs   and   profiles by The total gas pressure in the sub-divertor was measured using a capacitance manometer (baratron) located below the cryogenic pump [22,23].Baratrons are gas-species independent, such that the measured pressure is of molecular D 2 and atomic He.The radiated power was measured by the JET bolometry system [24].The lines-of-sight of the vertical and horizontal cameras are shown in Fig. 1.Tomographic reconstructions from the two bolometers provided 2D profiles of the radiated power.
The LFS edge line-averaged electron density, ⟨  ⟩ edge , measured by the far-infrared interferometer [25] in Fig. 1, was used as a proxy for the SOL upstream electron density and as the primary independent parameter.

Results and discussion
The D plasmas showed the expected rollover behaviour of  div,LFS against ⟨  ⟩ edge with no difference observed between fuelling ramps and steps (Fig. 2a).The peak of  div rollover against ⟨  ⟩ edge is considered here to be the detachment onset density, as used in previous studies [5,8,19]. div,LFS and  div,HFS were fit for each pulse with Gaussian process regression [26] to find the average LFS rollover edge density in  NBI = 1 MW He  e,det,He = (2.94 ± 0.09) × 10 19 m −3 and in D  e,det,D = (2.6 ± 0.1) × 10 19 m −3 .The HFS rollover occurred at ∼10% lower edge density than the LFS in both species (Fig. 2i).For  NBI = 1 MW in He,  e,det,HFS,He = (2.60 ± 0.04) × 10 19 m −3 and for D,  e,det,HFS,D = (2.3 ± 0.1) × 10 19 m −3 .Detachment of the HFS at lower upstream densities than the LFS is typical behaviour in tokamaks when the ⃗  × ∇ drift is towards the divertor, as ⃗  × ⃗  drifts drive a flow of particles from the LFS divertor into the PFR and from the PFR into the HFS divertor [27].In JET-C, rollover of  div,LFS occurred at ∼30% higher ⟨  ⟩ edge in He than D [5].
Ohmic heating power,   , was 30% higher in He than D  NBI = 1 MW plasmas (Fig. 2d) due to the higher Spitzer resistivity of the higher    of the He 2+ core plasma.The radiation from the confined region as measured by bolometry,  rad,bulk , and total radiated power,  rad,tot , was the same within measurement uncertainty between  NBI = 1 MW He and D for ⟨  ⟩ edge <  e,det,He (Fig. 2e and f).The power across the separatrix was estimated by  SOL =   +  NBI −  rad,bulk (Fig. 2j).For ⟨  ⟩ edge <  e,det,He ,  SOL was 10% higher in He than D, due to the higher   in He plasmas, which contributes to the observed higher  div rollover density in He.In D plasmas,  rad,bulk and  rad,tot are proportional to   with increasing ⟨  ⟩ edge and show no relationship to  div rollover.The continued increase of radiated power with increasing ⟨  ⟩ edge is a defining feature of detachment [28].In He however, for ⟨  ⟩ edge >  e,det,He ,  rad,bulk and  rad,tot increase at a greater rate than   with increasing ⟨  ⟩ edge .The increase in  rad,bulk reduces  SOL by ∼15%, to the same as in the D plasmas.
The sub-divertor neutral pressure was observed to be independent of heating power in the He plasma (Fig. 2b).The higher pressure observed in D than He is likely due to the higher recycling at the targets, but the sub-divertor pressure was found to be unsuitable as a proxy for the target neutral pressure as it is governed by the throughput of the pumping plena and the pumping speed of the cryogenic pump, which is different for the two species [29].
In He plasmas with  NBI = 1 MW of NBI heating,  div,LFS was up-to four times lower than in  NBI = 1 MW D (Fig. 2a).Currently, the cause of the difference is unclear, and is likely caused by a combination of factors.One factor is the effective electron cooling (or 'ionisation cost'),  ion , associated with the ionisation of the recycling neutrals by the SOL plasma, which is higher for He than D for   ≳ 10 eV.The ion current to the divertor in high-recycling conditions is approximately described by, where  rad,imp is the impurity radiated power,  rec is the recombination rate and plasma [30,31].As  rad,bulk was the same between the two species for ⟨  ⟩ edge <  e,det,He , for Eq. ( 1)  rad,imp is assumed to also be the same in both species.As  SOL was observed to be 10% higher in He than D. In high-recycling conditions (  ∼ 30 eV,   = 2 × 10 19 m −3 )  ion,He ≈ 35 eV for He 0 ionisation and  ion,D ≈ 24 eV for D 0 ionisation [32], which produces a 25% reduction from  div with D to  div with He.An additional energy loss to the ionisation of the singly charged ion, e + He +1 ⟶ He +2 + 2e, will further reduce ion current.However, the distribution of charge states of He in the divertor is unknown and AMJUEL currently does not include  ion for the reaction.The charge exchange rate is ∼ 10× lower for He than D and may also contribute to the lower ion current.Simulations of the He plasmas with SOLPS-ITER is expected to allow analysis of quantities that cannot be experimentally measured and determine the contributions to the lower recycling in He compared with D plasmas.The ⟨  ⟩ edge of  div rollover scales with NBI power; increasing heating in He from  NBI = 1 MW to 2 MW (Fig. 2 blue up triangles) increases  e,det,He by 10% and from 1 MW to 5 MW (Fig. 2 green down triangles) by 25%.In He  NBI = 5 MW plasmas,  total,LFS,pk and  e,LFS,pk do not reduce to match measurements in He  NBI = 1 MW plasmas at ⟨  ⟩ edge >  e,det,He,5 MW .The effect of higher core temperature lowering resistivity The peak  sat,LFS in He was found to be insensitive to ⟨  ⟩ edge , increasing by ∼30% from low-density to mid-density compared to ∼90% in D (Fig. 3b and d).The He  sat,LFS profile was broader than in D. It is difficult to determine the peak  sat,HFS due to a gap in the LPs that obscures wherever the strike point has been observed, however the current to the HFS target was generally higher than the LFS for both species (Fig. 3a, c, and e).
In  NBI = 1 MW plasmas at low-density, He and D both have a peak  ,LFS of ∼60 eV and at mid-density, ⟨  ⟩ edge ≈ 2.2 × 10 19 m −3 , the peak deceases to ∼30 eV, Fig. 4b and d, respectively.The same broader profile for He observed in  sat is also apparent in   .At highdensity, ⟨  ⟩ edge ≈ 3.6 × 10 19 m −3 , the peak  ,LFS measured by the LPs for He was <10 eV and D was <5 eV, but LPs are known to overestimate   in high-recycling and detached conditions [33,34].For two of the  NBI = 1 MW D pulses included in this work (JPN 94759 and 94771),   was derived from deuterium Balmer line emission, which showed the LFS strike point at ⟨  ⟩ edge =  e,det,D to be  ,LFS ≈ 2.5 eV and at highdensity,  ,LFS ≈ 0.8 eV [17].More accurate temperature measurements are not available for the He plasmas but from the D estimate, it is likely that at high-density  ,LFS ≪ 10 eV but not as low as for D.
The LFS target peak electron density,   , at mid-density is a factor of two lower for  NBI = 1 MW He plasmas than D plasmas.The lower   and similar   combine to the low  total,LFS,pk observed for the He plasmas.
The 2D radiated power distribution was dependent on ⟨  ⟩ edge and had similar profiles for the two species at low-density but diverged as density increased.For  NBI = 1 MW at low-density, the peak emission was close to the HFS vertical target and, although  rad,tot was the same in both species, D had higher power density, with a peak power of 0.30 MW m −3 , whereas He was more distributed, peaking at 0.18 MW m −3 (Fig. 5a and b).At mid-density, the peak radiated power in He was observed above the X-point, whereas in D the peak was still on the HFS (Fig. 5c and d).
For ⟨  ⟩ edge ≈ 3.6 × 10 19 m −3 >  e,det,He , the total radiated power was higher in He than D. The He peak power density was almost doubled going from ⟨  ⟩ edge = 2.2 × 10 19 m −3 to 3.6 × 10 19 m −3 but in D, across a similar density increase, the peak power was unchanged.This coincides with the majority of radiated power in He concentrating to the confined region above the X-point and towards the core along the inside of the separatrix (Fig. 5e), which was also observed with He plasmas in JET-C [5].For a similar edge density in D plasmas, the LFS divertor was the dominate radiator (Fig. 5f).
In  NBI = 5 MW He plasmas the power distribution with ⟨  ⟩ edge scaling behaves similar to the  NBI = 1 MW He plasmas (Fig. 6).At low-density the 2D profile was similar to He 1 MW at low-density.At mid-density the peak emission was in the region above the X-point, but with strong emission from the LFS divertor.For ⟨  ⟩ edge >  e,det,He,5 MW in 5 MW He, radiated power was concentrated above the X-point.The peak power density in the high-density case, Fig. 6c, is 1.5 MW m −3 .
In D plasmas the rollover of  div,LFS is accompanied by a similar rollover of the estimated peak LFS target plasma pressure,  total,LFS,pk (Fig. 2g). total,LFS,pk is reduced in  NBI = 1 MW He for ⟨  ⟩ edge >  e,det,He but does not increase and rollover with increasing edge density.The peak electron temperature at the LFS target,  e,LFS,pk , decreases with increasing ⟨  ⟩ edge for both species, starting at the lowest upstream density of the experiments, ⟨  ⟩ edge ≈ 1×10 19 m −3 (Fig. 2h).The reduction in particle and power flux to the targets without a reduction in the plasma pressure further upstream in the SOL in D plasmas is driven by losses to impurity radiation, volumetric recombination, molecular assisted dissociation (MAD), and ion-neutral momentum loss [28,35,36].These processes drop  div,LFS to 18% of peak and reduce  e,LFS,pk to <0.8 eV.In He plasmas, the long mean-free-path for He 0 ionisation combined with lower divertor densities pushes the ionisation front to above the X-point as ⟨  ⟩ edge increases and  e,LFS,pk decreases, developing a region of intense radiation above the X-point seen in Fig. 5.With the ionisation front inside the confined region, power is lost to the greater  rad,bulk , reducing  SOL and subsequently  div,LFS , but not to the extent seen by D detachment, and without reducing  e,LFS,pk to the <5 eV seen in D plasmas.

Summary
Experiments in JET-ILW L-mode He and D plasmas showed that He plasmas have lower recycling, lower sub-divertor pressures, low target plasma pressures, 10% higher  SOL , and broader divertor target profiles in  sat and  ,LFS .A ∼ 10% higher ⟨  ⟩ edge for the onset of detachment was found for He compared with D plasmas, partially due to higher Ohmic heating in He.
In He plasmas with  NBI = 1 MW,  div,LFS in He is up to 70% less than in D. Partially caused by the higher ionisation cost of neutral He compared with D and an unquantified loss to the ionisation of He + .Additional processes involved in the lower recycling in He plasmas are unknown but simulations of the experiments presented here in future work will elucidate the physics behind this difference.
A concentration of radiated power within the confined region above the X-point, occurring at edge densities less than the detachment threshold, was seen in 2D bolometry reconstructions.In conditions approaching ⟨  ⟩ edge =  e,det,He , He atoms recycled at the target can penetrate further through the plasma and into the confined region due to their longer ionisation mean-free path compared to D. Moving the ionisation front to above the X-point increases the power radiated within the confined region, reducing  SOL and thus  div,LFS at the divertor target.

Fig. 1 .
Fig. 1.Schematic of the JET vessel structure (grey) and magnetic equilibrium (black) for the He pulse JPN 101265 with relevant components shown, the cryogenic pump (red), target Langmuir probes (magenta), far-infrared interferometer (green), vertical and horizontal bolometry lines of sight (blue), and baratron pressure gauge (yellow).(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2 .
Fig. 2. (a) LFS divertor ion current from integration over the LPs, (b) baratron subdivertor neutral pressure, (c) NBI heating power, (d) Ohmic heating power, (e) total radiated power measured by bolometry, (f) radiated power from within the region confined by the separatrix, (g) peak estimated pressure on the LFS target, (h) peak electron temperature on the LFS target, (i) HFS divertor ion current, and (j) power crossing the separatrix into the SOL as a function of LFS edge line-averaged electron density.The legend in (b) applies to all sub-plots.The He pulses with  NBI = 1 MW of heating and strike point sweeps (JPN 101211/3) are marked with red squares, the 1 MW He density ramp (JPN 101265) with pale red circles, 2 MW He sweeps (JPN 101208-10) with blue up triangles, 5 MW He sweeps (JPN 101214, 101267/8) with green down triangles, 5 MW He ramp (JPN 101266) with pale green circles, 1 MW D sweeps (JPN 81473/4/6/7, 81490/2, 94767, 94770-72, 95884/7) with grey circles, and 1 MW D ramps (JPN 94759/61/62/64, 95232/35, 100559) with black diamonds.The detachment onset density is indicated with vertical dashed lines in black and red for 1 MW D and 1 MW He, respectively.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3 .
Fig. 3. Divertor target profiles for  NBI = 1 MW of the He (red) and D (black) ion saturation current density from the Langmuir probes on the high-field side (a, c, e) and low-field side (b, d, f).The top row (a, b) shows low-density sweeps (⟨  ⟩ edge ≈ 1.3 × 10 19 m −3 , He JPN 101213, D JPN 81473/4), the middle row (c, d) shows mid-density sweeps (2.2 × 10 19 m −3 , He JPN 101211/3, D JPN 81473/4), and the bottom row (e, f) shows high-density sweeps (3.6 × 10 19 m −3 , He JPN 101211, D JPN 94770).The He signal is scaled by ×4 in (e) and (f) for readability.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4 .
Fig. 4. Low-field side divertor target profiles for  NBI = 1 MW of the He (red) and D (black) electron density (a, c) and electron temperature (b, d) from the Langmuir probes.On the top row (a, b) the low-density sweeps are shown, ⟨  ⟩ edge ≈ 1.3 × 10 19 m −3 , He JPN 101213, D JPN 81473/4.The bottom row (c, d) shows mid-density sweeps, 2.2 × 10 19 m −3 , He JPN 101211/3, D JPN 81473/4.High-density is excluded due to unreliability of LPs in those conditions.Note that the scale for electron density is different for the two cases.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 5 .
Fig. 5.Total radiated power density tomographic reconstructions from bolometry of  NBI = 1 MW He (a, c, e) and D (b, d, f) plasmas for the low, mid and high-density cases by row.