Benefits of the Super-X divertor configuration for scenario integration on MAST Upgrade

The integration of good core and edge/pedestal confinement with strong dissipation of heat and particles in the divertors is a significant challenge for the development of fusion energy. Alternative divertor configurations offer potential advantages by broadening the operational space where a device can operate with detached divertors and acceptable power exhaust. First results from MAST Upgrade are presented from high confinement mode experiments with outer divertors in the Super-X divertor configuration, showing that the outer divertors naturally detach when the Super-X is formed with no discernible impact on the plasma core and pedestal. These initial findings confirm predicted benefits of the Super-X configuration in terms of facilitating scenario integration.


Introduction
The development of high-performance plasma scenarios in magnetic confinement fusion devices necessitates stable and favourable confinement in the plasma core (including thermal and energetic particles) and edge/pedestal regions and acceptable power exhaust.While the required levels of performance in each aspect vary between devices (e.g.ITER [1,2] and DEMO [3][4][5]), a key aspect is the compromise between core and edge/pedestal confinement and power exhaust.It has been widely observed that dissipation of power and particles in the divertors can degrade confinement in the core and edge (e.g [6][7][8] and references therein).Conversely, the achievement of 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.good core and pedestal confinement can exacerbate the challenge of handling power and particle exhaust (e.g [9][10][11].).
The development and study of integrated plasma scenarios that minimise trade-offs between performance of the confined plasma and power exhaust is a very active area of research, with solutions broadly falling into three categories.Firstly, radiating impurities have been used in devices with conventional divertor configurations to dissipate a significant proportion of the power crossing the separatrix before it reaches the divertors [12][13][14][15], sometimes in combination with improved core confinement regimes [16,17].Secondly, alternative divertor configurations are employed to decrease the divertor heat flux via a combination of spreading the heat flux over a larger surface area and promoting dissipation of particles, momentum and energy in the divertor to promote access to the detached divertor regime [18,19].Finally, novel materials are being developed to withstand higher sustained heat fluxes (e.g [20].) to help ease the constraints power exhaust places on scenario integration.MAST Upgrade is an excellent facility for studying the comparative benefits of alternative divertor configurations in a low aspect ratio tokamak [21].The device has significant flexibility to independently vary the shaping of the plasma core and divertor configuration, and is optimised to study the Super-X divertor configuration [22].The Super-X divertor configuration is characterised by placing the outer strike points at large major radius and a region of low poloidal field along the outer divertor legs is formed to increase the length of field lines connecting the outer mid-plane and outer divertor strike points.Extending the outer divertor legs to larger major radius, i.e. increasing R T , where the total magnetic field is reduced, is predicted to reduce the parallel heat flux density at the divertor target, q ∥,T .This reduction in q ∥,T is due to its proportionality to the local magnetic field, B tot,T .Assuming the total magnetic field at the divertor target is dominated by the toroidal field, B tot,T ∼ B ϕ ,T ∝ 1/R T , therefore q ∥,T ∝ 1/R T .In attached divertor conditions, reducing q ||,T is predicted to increase the target electron density, n e,T , by a factor R 2  T and decrease the target electron temperature, T e,T , by the same factor [23,24].This in turn is expected to facilitate the onset of divertor detachment (for example, in analytic models [25], and numerical simulations [24] the onset of detachment is predicted to scale as 1/R T ).However, these benefits could be mitigated somewhat by changes in transport across flux surfaces (e.g [26]) or the presence of significant flows as R T is increased [27].When the outer divertors are detached, it is predicted that gradients in q ∥ along the outer divertor leg, that can be approximated by the ratio of the total magnetic field at the X-point and divertor strike point, act to passively stabilise the movement of the detachment front from the target towards the X-point [25,28] although the impact of flows could act to partially mitigate this effect [27].In addition to increasing R T , in the Super-X configuration the length of field lines connecting the outer midplane to the outer divertor target, L ∥ , is typically higher in the Super-X configuration (typically up to a factor of ∼2 higher [29], depending on the precise details of the Super-X divertor configuration and the comparator 'standard' divertor).The simplified two-point model predicts increasing L ∥ can lead to an additional reduction in T e,T via T e,T ∝ L −4/7 ∥ and increase in n e,T via n e,T ∝ L 6/7 ∥ , suggesting increasing L ∥ is predicted to have a more subtle effect on the plasma properties at the divertor target compared with increasing R T .
In MAST Upgrade, the major radius of the outer strike points can be varied over a wide range, from ∼0.8 m to ∼1.45 m within tightly baffled divertor chambers (a larger variation in the strike point major radius is possible by moving the outer strike points inward to a major radius of 0.33 m, however this divertor configuration would be unbaffled, complicating comparison with other, baffled, configurations).By comparison, the conventional aspect ratio TCV tokamak can vary the major radius of the outer strike point from 0.62 m to 1.06 m [19], and was observed to have a modest impact on detachment access when the divertor was unbaffled, and a more pronounced effect in baffled divertor configurations [30,31].These observations are consistent with the generation of supersonic flows in the divertor [27].
Due to the low major radius of the inner strike points in low aspect ratio devices, managing heat fluxes to the inner divertors is more challenging.Therefore, MAST Upgrade typically operates in the connected double null topology, with a separation between the primary and secondary separatrices at the outer mid-plane typically less than 20% of the radial heat flux decay length, to minimise the power entering the inner divertors [32].
During the design and initial operation of MAST Upgrade, extensive modelling studies were performed that predicted the Super-X configuration would offer significant benefits, including reduced divertor heat flux in the attached divertor regime [33,34] and improved access to the detached regime [33,35] with negligible impact on the radial plasma profiles at the midplane, upstream of the divertors [34].Initial experiments in low confinement experiments [36] at low heating power confirmed these benefits of the Super-X configuration compared with a conventional divertor.The results presented here build on previous studies by performing initial experiments to study the impact of the Super-X divertor configuration in high performance H-mode pulses with significant auxiliary heating power.
The structure of this paper is as follows: a brief overview of MAST Upgrade and scenario integration issues are presented in section 2, the diagnostics used to characterise this experiment is presented in section 3, the strategy to maximise the divertor heat flux is presented in section 4, results from initial scenario integration experiments are presented in section 5, followed by conclusions in section 6.

The MAST Upgrade scenario integration challenge
MAST Upgrade is a low aspect ratio spherical tokamak, with typical plasma major (R) and minor (a) radii of 0.8 m and 0.5 m respectively (R/a ∼ 1.6), capable of reaching a plasma current up to 2 MA (1.0 MA has been achieved to date), maximum toroidal field on axis of 0.8 T (0.72 T to date) and auxiliary heating from on-axis and off-axis neutral beam injectors, capable of injecting up to 2.5 MW each (to date, the maximum total injected power is ∼3.7 MW).Typical operating scenarios are in close proximity to a connected double null topology to minimise the heat flux to the inner divertors [32] and facilitate H-mode access [37].Initial Ohmic heated experiments to study access to detachment in conventional and Super-X divertor configurations on MAST Upgrade, could not unambiguously attach the outer divertors in the Super-X configuration, even at the lowest operating density of the device, as shown in figure 1, corresponding to approximately 10% of the Greenwald density limit, where non-disruptive locked modes and concomitant strike point splitting were common [38].However, this data suggests that the detachment threshold, in terms of outer mid-plane separatrix density, is substantially lower in a Super-X divertor configuration.In the conventional divertor configuration, the total D α emission from the outer divertor increases approximately linearly with the separatrix density, which is typically observed in detachment experiments (e.g.since initial studies described in [39]).In the Super-X configuration, the outer divertor D α emission initially increases with increasing separatrix density, then rolls over at separatrix densities exceeding ∼5 × 10 18 m −3 , approximately twice the detachment threshold.This roll-over behaviour in the outer divertor D α emission is not commonly observed, because and is thought to be due to low electron temperature and density conditions that prevail when the Super-X divertor configuration is detached in MAST-U, where plasmamolecule interactions produce a significant proportion of the Dα emission.At separatrix densities significantly exceeding the detachment threshold in the Super-X configuration, the electron temperature at the divertor target drops below 1 eV, reducing the source of molecular ions and their associated D α emission [36,40], causing the total emission from the divertor to fall with increasing separatrix density.
The significantly lower detachment threshold in the Super-X configuration, and therefore larger operational space with detached divertors, greatly facilitates the integration of detached divertors with strong auxiliary heating and good core and edge confinement.However, it potentially poses a challenge to test the limits of scenario integration if the outer divertors remain detached in the Super-X configuration at the highest levels of heating power currently available on MAST Upgrade.This motivated the development of a plasma scenario that intentionally maximises the exhaust challenge for the divertors, to enable studies of the impact of fuelling and seeding that are typically used to moderate divertor power and particle fluxes on the core and edge/pedestal regions.

Diagnostics
MAST Upgrade is equipped with an array of high-resolution diagnostics to characterise the plasma core and divertor regions.The principal diagnostics used in this study are shown schematically in figure 2. In the plasma core, radial profiles of n e and T e are measured with 1 cm resolution at 130 spatial points from R = 0.305 m to R = 1.48 m, which extends beyond the separatrices, into the scrape-off layer on the high-field side and low-field side with a time resolution of up to 240 Hz [41].Radiation losses from the plasma core are estimated using a 32 channel resistive bolometer system [42] with a time resolution of ∼25 Hz (depending on the measured signal:noise).The main chamber neutral pressure is monitored with a Bayard-Alpert ion gauge at a time resolution of 1 kHz.The magnetic equilibrium is estimated using the EFIT++ equilibrium code [43,44].
MAST Upgrade is designed to have up-down symmetry from the equatorial mid-plane, including the closed divertor chambers.To date, all experiments on the device have been performed with the ion ∇B drift pointing downwards, which is expected to result in more power entering the lower divertor (based on results reported from MAST [32] and C-Mod [45]) therefore the lower divertor is more comprehensively diagnosed.Radial profiles of T e and divertor ion saturation current (I + sat ) are measured with arrays of Langmuir probes to infer the ion saturation current density (J + sat ), n e and the divertor heat flux (via the relationship describing the heat flux parallel to magnetic field lines q ∥ = (γT e + E pot ) J + sat where γ is the sheath heat transmission coefficient and E pot is the potential energy released per incident ion [46], neglecting contributions from electric currents in the SOL [47] due to the relatively high density and collisionality of these experiments) at a spatial resolution of ∼1 cm and temporal resolution of ∼1 kHz, with probe tips inclined at 10 • to the surface of the divertor tiles to enable measurements at low field line incidence angles.The heat flux to the upper and lower divertor tiles are also measured independently using infrared cameras that have tangential views across the divertor chamber with a typical spatial resolution of 1.5-3 mm and temporal resolution of 2.25 ms.The power radiated from the plasma in the divertor chambers is estimated using resistive bolometers, including arrays of 16 channels mounted to the vacuum vessel in both divertors, and an additional 16 channels in the lower divertor mounted slightly above the chamber [42].Profiles of n e and T e along the lower outer divertor leg are measured using a divertor Thomson scattering system [48] capable of measuring T e in the range of 40-0.5 eV with a time resolution of 30 Hz at 12 locations from R = 1.03 m to 1.44 m.Spectra of passive light emission in the near-UV and visible light regions of the electromagnetic spectrum, with variable spectral, spatial and temporal resolution [40].The neutral pressure in the divertors is monitored using similar Bayard-Alpert ion gauges as used in the main chamber.

Maximising the heat flux entering the divertors
As explained in section 1, in future high power tokamaks, it is envisaged that the unmitigated peak divertor particle and heat fluxes will exceed material limits in the attached divertor regime.These fluxes will be moderated via fuelling and impurity seeding to promote losses of momentum and energy, which will facilitate access the detached divertor regime.In MAST Upgrade H-mode pulses with the maximum auxiliary heating available, ∼3.5 MW, in the Super-X divertor configuration, in the absence of fuelling (except to fuel the plasma core) or impurity seeding, the outer divertors are typically detached.
This paper investigates an optimal plasma scenario to maximise the outer divertor target temperature was developed to maximise the probability of attaching the outer divertors, to enable studies of the impact of fuelling and/or seeding to detach them.This analysis is applied to the divertor conditions between ELMs in H-mode.To guide scenario development, a two-point model [49] was used to estimate the dependence of the outer divertor target temperature T t (eV) on plasma scenario parameters in the absence of dissipation of momentum and energy in the scrape-off layer and divertor: where n e,sep is the electron density at the outer mid-plane (m −3 ), L || is the length of field lines connecting the outer midplane and outer divertor strike point (m), κ 0e is the Spitzer conductivity divided by T 5/2 (∼2800 eV −7/2 Wm −1 ) [49], m i is the deuterium ion mass (kg), γ is the sheath heat coefficient (assumed to be ∼9), e is the electron charge (C) and q || is defined as: R out,mp is the radial position of the separatrix at the outer mid-plane (m), λ q is the exponential decay length of the heat flux at the outer mid-plane (m), B θ,mp and B tot,mp are the outer mid-plane poloidal and total magnetic field components respectively (T) and P SOL is defined as: where P ohm is the Ohmic heating power (W), P NBI,abs is the absorbed neutral beam heating power estimated by TRANSP [50] (W), P rad,core is the power radiated from the plasma core estimated by foil bolometers (W) [42] and Ẇ is the rate of change of plasma stored energy (W).Estimation of the twopoint model parameters from experiments is described in the following sections.

Scrape-off layer heat flux decay length
The exponential decay length of the heat flux in the scrapeoff layer (SOL), λ q , is a key parameter to determine the heat flux density (see equation ( 2)), of which a substantial fraction would be dissipated in detached divertor regime.In experiments, λ q is typically estimated using spatially resolved measurements of the temperature of the divertor tiles in the attached divertor regime, where heat dissipation in the SOL and divertor is negligible, to estimate the unmitigated divertor heat flux (e.g [45,51,52] on individual devices and in cross-machine studies [53]).Radial profiles of the divertor heat flux are then fit to a convolution of Gaussian and exponential profiles [54] to estimate the peak heat flux, the radial heat flux decay length λ q and the Gaussian profile width, S, that largely determines the heat flux profile in the private flux region.A similar approach is applied here, where infrared thermography measurements of the lower outer divertor temperature profiles, and the derived heat fluxes, are routinely made.For the purposes of this study, estimates of λ q were compared with previously published studies of its dependencies on plasma scenario parameters [ 51,53].These previous studies found a strong inverse dependence of λ q on the poloidal field at the outer mid-plane (λ q,Eich ∝ B −1. 19  pol,mp , λ q,Thornton ∝ B −0.68 pol,mp ), and relative insensitivity to other parameters including machine size, plasma density, P SOL and toroidal magnetic field.In a given device, where the plasma minor radius does not vary significantly, B pol,mp has a linear relationship with I p .Measurements of λ q in H-mode plasmas with attached outer divertors in conventional divertor configurations over a range of 0.14 ⩽ B pol,mp ⩽ 0.27, equivalent to 450 kA ⩽ I p ⩽ 1000 kA are shown in figure 3.Only measurements taken from 60% to 90% of the ELM cycle (i.e.60%-90% of the time between adjacent ELMs) are used in this analysis and frames acquired during an ELM crash were excluded from the analysis.Over this period of the ELM cycle, it is expected that the pedestal parameters have stabilised after the ELMs [55].Comparison of the total lower outer divertor power load with P SOL suggests over 30% of P SOL is deposited to the attached lower outer conventional divertor, which is consistent with previous results from MAST [32].
The dependence of λ q on I p is in reasonable agreement with the previously published scalings, shown in figure 3, however the absolute values are 30% higher than predictions from the Thornton scaling obtained on MAST [51] and 80% higher than the Eich scaling.The cause of this disagreement is unclear.The results in the following sub-sections will also show strong interdependencies between I p and other parameters that have been observed to more weakly influence λ q , such as P SOL and the separatrix density.However, for the purposes of this study, the dependence of λ q on I p is most of interest.

Separatrix density
In H-mode, the radial profiles of electron temperature and density exhibit steep gradients in the edge pedestal region, thus making estimates of the plasma conditions at the equilibrium separatrix at the outer mid-plane highly uncertain when errors in the equilibrium reconstruction are considered.The approach adopted in this study builds on previously reported findings [56,57], to use a two-point model that includes the variation in magnetic field strength along field lines [58] to estimate the outer mid-plane separatrix temperature, assuming all the heat is electron conducted from upstream to the target and no volumetric momentum or power losses occur: where L ||x and L ||t are the distance along field lines from the outer mid-plane to the primary X-point, and outer divertor target respectively (m), s is the distance along magnetic field lines and q || is defined in equation ( 2).This equation is evaluated with magnetic flux and field strength estimates from EFIT++ equilibrium reconstruction on field lines 1 cm outboard of the primary separatrix.
Radial profiles of electron temperature, T e , and density, n e , are measured at 1 cm spatial resolution with a Thomson scattering diagnostic [55].A Bayesian profile fitting code that includes the spatial instrument function of the finite sampling volumes in the plasma is used to fit the density and temperature profile measurements of the edge n e and T e profiles to a modified Tanh function to enable the pedestal separatrix density, n e,sep , for a given separatrix electron temperature, T e,sep , to be estimated.Typical values of T e,sep for 450 kA shots is 25 eV, rising to 40 eV at higher plasma current.
This approach was applied to typical H-mode scenarios in connected double null topology and conventional divertor configurations spanning a range of plasma current, 450 kA ⩽ I p ⩽ 1,000 kA, toroidal field on axis 0.42 T < B ϕ < 0.64 T, injected neutral beam power 1.1 MW < P NBI < 3.5 MW, fraction of the Greenwald density limit, f GW , 0.4 ⩽ f GW ⩽ 0.6 to form a database to ascertain how the mid-plane separatrix density, n e,sep , pedestal top density, n e,ped and line-average core density <n e > vary with I p .These parameters were averaged over the H-mode period of each shot, only including measurements of electron temperature and density profiles taken from 60% to 95% of the ELM cycle.This database is summarised in figure 4.
Despite the large variation in the plasma scenarios in the database, there is a clear linear positive trend of n e,sep with I p , as observed on other devices, such as ASDEX Upgrade [14].There is a significant scatter in <n e > in this database as the core line-average density steadily increases in H-mode phase of these shots.However, it is observed in all of the shots analysed that n e,sep and n e,ped are not strongly affected as <n e > increases in individual shots.In a given plasma scenario at constant I p , the increase in <n e > is due to the rising electron density near the magnetic axis, n e,core , as shown in figure 5, and a concomitant increase in the radial density gradient in the core.
The data shown in figure 4 does however show that both <n e > and n e,sep increase with I p , however the 600 kA scenarios generally have higher <n e > and 1000 kA have lower <n e > than a simple linear trend with I p because they were  developed relatively recently, so fewer variants been developed (reducing scatter in the data) and have optimised extent as other scenarios.In all of these shots, no attempts were made to reach the H-mode density limit by strongly fuelling the plasma core, although a small level (∼10 21 particles/s) of gas fuelling from a highfield side gas valve was used to moderate core MHD activity.In principle, higher fuelling could elevate n e,sep , for example the H-mode density limit on ASDEX Upgrade has n e,sep /n GW = 0.4-0.5 [57], compared with this database where n e,sep /n GW = 0.15-0.2.
From this database, a simple linear regression of n e,sep with I p yields n e,sep (10 19 m −3 ) = 1.86 × I p (MA).This

Absorbed neutral beam power
To maximise the heat flux entering the divertors, it is important to maximise the auxiliary heating power.In MAST Upgrade, the main sources of heating are two co-current neutral beams, an on-axis beam situated at Z = 0 m with a tangency radius of 0.71 m and off-axis beam at Z = 0.65 m and a tangency radius of 0.8 m, both with a maximum injected power of 2.5 MW.As discussed in [59], the heating power from the on-and off-axis beams have significant differences, in particular, higher sensitivity of the heating from the off-axis beam to I p and elongation of the plasma core.The scenarios in this study are limited to those with elongation in the range 2.0-2.2, which has moderate impact on the fraction of injected NBI power absorbed by the plasma.However, there is a significant variation in the absorbed NBI power when I p is varied from 450 kA to 1000 kA.The influence of I p on absorbed NBI power from the on-and off-axis beams is estimated using interpretive TRANSP simulations of typical ELMy H-mode pulses studied in the preceding sections, shown in figure 6.To help isolate the effect of I p on NBI heating, anomalous fast ion diffusion is not applied in these simulations (for example, anomalous fast ion diffusivities of 0.5-3 m 2 s −1 were needed to match neutron camera measurements of neutron emission to match interpretive simulations on MAST [60,61] and MAST Upgrade [59]).This represents a best-case estimate of the absorbed NBI power.Additional losses of injected NBI power due to chargeexchange interactions with main chamber neutrals are also not considered here.
These TRANSP simulations show that the absorption of both beams is sensitive to the core density when f GW < 0.3, due to increased beam shine-through at lower density.At higher core density, the absorption of the on-axis beam is relatively insensitive to I p , compared with the scatter in the data.However, when f GW > 0.3, the absorption of the off-axis beam is highly sensitive to I p due to strong first-orbit losses at lower plasma current.This suggests that scenarios with I p ⩾ 750 kA are preferred to maximise the auxiliary heating power.

Estimation of optimal plasma parameters for scenario integration
The analyses presented in the preceding subsections provide estimates of the key plasma parameters and absorbed auxiliary heating power and their variation with plasma current, to inform the development of an optimal scenario that maximises the temperature at the outer divertor strike points, T t .To leading order, equation ( 1) indicates that T t is most sensitive to q 2 || /n 2 e,sep ; their variation with I p is shown in figure 7. The optimal I p to maximise the outer divertor target temperature is largely determined by the variation of the absorbed off-axis neutral beam heating power at typical core line average densities, i.e. 0.4 < n e /n GW < 0.6.There is a strong I p dependence of the absorbed off-axis neutral beam power at I p ⩽ 750 kA and a much weaker dependence at higher I p .Consequently, the mid-plane parallel heat flux, q ||,mid , increases strongly with I p at lower I p (<750 kA) and a weaker increase at higher I p , while the separatrix density increases linearly with I p .This suggests that the I p that maximises the outer divertor temperature is 750 kA.As discussed in the previous section, MHD instabilities can reduce fast ion confinement, as observed on MAST [60,[62][63][64], which vary between plasma scenarios if they cannot be avoided.Furthermore, additional auxiliary heating and current drive sources, such as a new electron Bernstein wave system that is planned for installation in 2025, will have a different density dependence on the absorbed heating power that may modify the optimum I p for these studies in future.

Initial development and study of H-mode scenario with a Super-X divertor configuration
To perform initial studies of the relative benefits of the Super-X divertor configuration on H-mode core and pedestal confinement and power exhaust, a plasma scenario was developed at I p = 750 kA with 3 MW of neutral beam injection (of which, interpretive TRANSP simulations predict ∼1.5-2.2MW is absorbed) and ∼0.3 MW of Ohmic heating with 0.5 MW being radiated from the plasma core (measured using resistive bolometers), suggesting that P SOL ∼ 1.3-2.0MW.The rate of change of stored energy term in equation ( 3) is estimated to be negligible.The outer divertor strike points were swept from a conventional to a Super-X divertor configuration with constant core shape from 0.25 s to 0.38 s (shown in figure 8), 50 ms after the transition from L-to H-mode.During the H-mode phase of the shot, the plasma stored energy was 100 kJ, corresponding to an energy confinement time normalised to the ITER H-mode scaling, H 98(y,2) ∼ 1.0 [65].As the outer divertor strike points are swept to larger major radius from the conventional to the Super-X configuration, the outer strike point major radius increases from 0.76 m to 1.36 m, the field line connection length from the outer mid-plane to outer divertor targets increases from 12 m to 22 m, poloidal flux expansion (defined as (B pol /B ϕ ) mp /(B pol /B ϕ ) t ) increases from 3 to 20, resulting in a reduction of the target field line incidence angle (defined as the angle between a unit vector parallel to the surface of the divertor tiles and the normalised magnetic field vector) from 8 • to 1 • .These plasmas did not feature any extrinsic impurity seeding or divertor fuelling.
The magnetic geometry of the Super-X divertor configuration conforms to its initial description in [22], with increased strike point major radius and connection length, within the operational constraint to avoid excessive plasma power and particle fluxes to the vacuum vessel above the divertor tiles, in the gap between the divertor tiles and the divertor baffles.This was achieved by introducing secondary X-points in the divertor chamber to prevent magnetic connection of the outer mid-plane and the region above the divertor tiles in the Super-X divertor configuration via magnetic flux surfaces.This technique limits the mid-plane flux that can reach the divertor tiles to 15 mm (as shown in figure 8), or two mid-plane heat flux exponential decay lengths (see figure 3) at this plasma current.

Impact on mid-plane profiles
In the phase of the shot where the outer strike point was swept outward, the global confinement parameters such as stored energy are well maintained, however there is a modest increase in the core line-average density from 2.9 × 10 19 m −3 -3.2 × 10 19 m −3 , and an increase in the plasma stored energy from 90 kJ to 100 kJ, resulting in a commensurate increase in the normalised energy confinement time H 98(y,2) from 0.8 to 1.0.The change In plasma stored energy, and normalised confinement time, is correlated with the acceleration of an n = 1 mode that persists throughout the NBI heated phase of the shot, and is thought to be unrelated to the changes in the divertor configuration.The plasma gas fuelling rate was constant throughout this phase, suggesting that the increasing density trend is due to improved particle confinement in H-mode.The time variation in the mid-plane radial profiles of n e , T e and T i are shown in figure 9.The continual rise of the core lineaverage density is also evident in the radial profiles, most notably in the plasma core, but the and ion temperature profiles do not change significantly over this period.
These observations suggest that the changes in divertor configuration do not significantly affect the mid-plane profiles.Simple analytic two-point models that neglect volumetric momentum and power losses assume heat is transported along field lines via electron conduction [23,66] predict a slight increase in the electron temperature at the mid-plane separatrix according to: where f R is the ratio of the major radii of the outer strike point and the separatrix at the outer mid-plane and other terms have been defined earlier.For constant q || and taking L || and f R from the conventional and Super-X equilibria as presented earlier in this section, the model predicts a ∼10%, or ∼4-5 eV, increase in the T e,sep .This modest increase in T e,sep is too small to discern from variation in the temperature profiles as the pedestal evolves and noise inherent in the profile data.Moreover, this model assumes that the outer divertor strike points are attached in both divertor configurations, which will be discussed in the following subsection.

Impact on outer divertor conditions
Sweeping the outer divertor legs out to larger major radius is expected to have a substantial impact on the plasma conditions in the divertors, in particular by reducing divertor target temperature and improving access to the detached divertor regime [22,24,25,33,34].The effect of varying the major radius of the strike point is predicted by simplified models [23], in good agreement with predictive multi-fluid simulations in the attached divertor regime [24], that T e,t ∝ where R OSP and R OMP are the major radii of the outer strike point and the separatrix at the outer mid-plane respectively.These predictions are in qualitative agreement with previous Ohmic heated L-mode experiments on MAST-U [36,40], observing that the Super-X configurations facilitates access to the detached divertor regime.
In the experiments presented here, the outer divertor strike points are swept to larger major radius using equilibrium shape controllers in feedforward control that were designed not to influence where the separatrix intersects the mid-plane at the high and low field-sides and the X-point position.The gas fuelling from the main chamber was held constant at a low level during this phase and no fuelling or seeding from the divertors was applied.As the outer strike points are swept outward from 0.25 to 0.38 s, the neutral pressure in the main chamber rises from 0.8 mPa to 1.1 mPa, and the lower divertor neutral pressure rises from 0.12 Pa to 0.19 Pa.The divertor neutral compression (defined as the ratio of the measured gas pressure in the lower divertor chamber and the main chamber) is approximately 150 throughout the period where the outer strike points  are swept outward.No fuelling from the divertor was used in this experiment.
In the Super-X divertor configuration, there is a substantial reduction in the outer divertor ion saturation current density and divertor parallel heat flux density profiles, as shown in figure 10.The reduction in the parallel heat flux to the divertor target greater than the factor ∼2 expected due to increased total flux expansion, due to additional power loss mechanisms, leading to a reduction in the electron temperature.This can be illustrated by the relationship between the divertor parallel heat flux, q ∥,t and the ion saturation current density in attached divertor conditions, j + sat , q ∥,t = γT e j + sat [49].However, strong reductions in the divertor ion saturation current density is suggestive of substantial dissipation of momentum and particles, via volumetric recombination and plasma-neutral interactions that occur at low temperatures (i.e.T e < 5 eV) [36,40,67].
In the phase of the shot with a Super-X divertor configuration, the divertor Thomson scattering diagnostic can provide useful estimates of the electron density and temperature profiles.The profile measurements from 0.380 ms in the shot are shown in figure 11, at the same time as the divertor Langmuir probe profile data was recorded.The data indicate T e < 1.5 eV close to the separatrix, and that T e near the target is below 0.5 eV, as there is negligible detected scattered light from spatial locations with a major radius of 1.26 m in the polychromator channels closest to the laser wavelength, to resolve temperatures down to 0.5 eV.These observations are corroborated by measurements from the DMS spectrometer, indicating that the electron temperature is lower than 1.8 eV, based on the D 2 Fulcher band emission in this region.
In the conventional and Super-X phases of this scan, electron pressure profiles at the mid-plane and outer divertors were measured with Thomson scattering and divertor Langmuir probes respectively, with characteristic radial profiles shown in figure 12.The mid-plane pressure profiles are mostly unaffected by the changes in divertor configuration and there is reasonable pressure balance in the conventional divertor configuration, indicating the outer divertors are attached.Conversely, there is a significant target pressure reduction in the Super-X configuration.There is considerable scatter in the Langmuir probe inferred electron temperatures in the Super-X configuration, due to issues concerning the interpretation of the MAST-U Langmuir probe data in detached conditions [46] where the electron temperature is low.This prevents observation of a substantial reduction in the Langmuir probe inferred divertor electron temperature between the Conventional and Super-X divertor configurations.Through the plasma sheath entrance conditions, the target pressure, p t , is correlated with the target temperature, T e,t , and the ion target flux j + sat via p t ∝ j + sat √ T e,t .The strongly reduced ion target flux in the Super-X configuration is therefore consistent with the observed pressure reduction between the mid plane and the target in the Super-X configuration, even though the reduction in p t between the conventional and Super-X configurations is likely over-estimated, providing evidence that the Super-X is detached.

Conclusions and discussion
The results presented here provide first evidence that substantial reductions in the outer divertor power and particle fluxes are possible in H-mode with the Super-X divertor configuration with negligible impact on global energy confinement and pedestal profiles in the confined plasma.These experiments were performed in a plasma scenario designed to maximise the outer divertor temperature with the aim of preventing the divertor naturally detaching in the Super-X configuration to study the impact of additional fuelling and seeding to detach the divertors on the core and pedestal.This provides an important initial demonstration of the benefits of the Super-X configuration in integrating good core and pedestal confinement with plasma exhaust in MAST Upgrade.
While these initial results are very positive, there are several important open questions that need to be addressed to ascertain how they extrapolate to future devices.More detailed spectroscopic measurements will enable deep studies of the atomic and molecular processes governing detachment in Hmode between ELMs for comparison with previously reported results in L-mode [36,40,67] and to test model predictions.The experiment reported here has concentrated in characterising the plasma state between ELMs.Building on this work, experiments are currently underway to extend these studies to plasma scenarios with regular, higher frequency ELMs to characterise the plasma state during and between ELMs, to estimate the degree to which the ELM energy can be buffered by a detached Super-X divertor configuration.
Further studies at higher heating power and particle pumping are needed to attach the outer divertors in the Super-X configuration (guided by an assessment of the implications of additional heating power and particle pumping [33]).With attached outer divertors, studies of power exhaust and scenario integration can progress to deepen our understanding of the impact of fuelling and impurity seeding on exhaust and scenario integration.In turn, this will enable investigation of the degree to which impurities are retained in the divertors and whether the sensitivity of the detachment front to changes in density, radiated power and heating power is reduced in divertor configurations with high total flux expansion, predicted by theory and simulations [25,58].

Figure 1 .
Figure 1.Upper-outer divertor total ion flux in Ohmic heated experiments with density ramps to estimate the detachment threshold, in terms of the outer mid-plane separatrix density, in conventional (red) and Super-X (blue) divertor configurations.Lower-outer divertor total Dα emission.Detachment onset is observed when the divertor ion flux starts to decrease with increasing density mid-plane separatrix density.The approximate range of outer mid-plane separatrix density where splitting of the outer divertor legs, indicating the onset of non-disruptive locked modes, is highlighted in grey.

Figure 2 .
Figure 2. Principal diagnostics used in this study, including Thomson scattering (TS) to measure ne and Te in main chamber and divertor (note-every 5th measurement location for the core Thomson system is plotted for clarity), resistive foil bolometers (Bolo) to estimate the power radiated from the plasma in the main chamber and divertor, tangentially viewing IR thermography (IR) to estimate the divertor heat flux, Langmuir probes (LP) embedded in the divertor tiles to estimate ne, Te and J + sat at the probe locations, and main chamber and divertor fast ionisation gauges (FIG) to estimate the neutral pressure in these regions.

Figure 3 .
Figure 3. Measurements of λq in MAST-U compared with previously published scalings of its dependence on plasma current.

Figure 4 .
Figure 4. Dependence of the outboard mid-plane separatrix density (circles), pedestal top density (triangles) and core line-average density (crosses) and Greenwald density limit (dashed line) on plasma current in typical H-mode scenarios with attached conventional outer divertor configurations.The dotted line is a linear fit ne,sep (10 19 m −3 ) = 1.86 × Ip (MA).

Figure 5 .
Figure 5. (top) The time variation in core line-average density (<ne>), density at the major radius of the magnetic axis (ne,core), outer pedestal top density (n e,ped ) and outer midplane separatrix density (ne,sep) in a typical ELMy H-mode shot, (bottom) mid-plane Dα emission, intermittent due to ELMs.

Figure 6 .
Figure 6.Estimates of the fraction of the injected neutral beam power absorbed by the plasma from the on-axis beam (left) and off-axis beam (right) as a function of the core line-average density as a fraction of the Greenwald density limit.

Figure 7 .
Figure 7.The variation in upstream parallel heat flux and reciprocal of the mid-plane separatrix density (above) and q 2 ∥ /n 2 e,sep , where a maximum in this data is indicative of hottest outer divertor temperatures.

Figure 8 .
Figure 8. (Left) Equilibrium reconstructions in conventional (red) and Super-X (blue) divertor configurations in shot 47154 at 0.25 s and 0.38 s respectively and (right) radial profiles of parallel connection length, poloidal flux expansion and field line incidence angle.

Figure 9 .
Figure 9. Mid-plane radial profiles of electron density (top left), electron and ion temperature (top right) measured using Thomson scattering and charge-exchange recombination spectroscopy (CXRS) respectively.The average of these radial overlaid in The stored energy, by reconstruction (middle).The core plasma Dα emission and the timings of the Thomson scattering and CXRS measurements (bottom) are denoted by solid and dashed lines respectively.

Figure 10 .
Figure 10.Divertor parallel heat flux (left) and particle flux parallel to field lines (right) measured Langmuir probes in the conventional (red) and Super-X (blue) phases of shot 47154 as a function of normalised poloidal flux, ψ N .

Figure 11 .
Figure 11.Divertor Thomson scattering measurements of ne (left) and Te (right) profiles along the lower outer divertor leg from shot 47 154 at 0.380 s.The profile measurements closest to the divertor tile are highlighted.

Figure 12 .
Figure 12.Mid-plane (green squares) and outer divertor electron pressure profiles multiplied by 2 in conventional (red) and Super-X (blue) divertor configurations estimated by Thomson scattering (Thomson) and Langmuir probes (LP) respectively.