High performance power handling in the absence of an H-mode edge in negative triangularity DIII-D plasmas

A grand challenge for magnetic fusion energy and for the extrapolation of operational scenarios in current fusion devices to future reactors is the integration of a high-performance core with edge plasma conditions compatible with sustainable plasma material interaction. Compatible edge plasma conditions include steady state heat fluxes and transient energy fluences below the limits imposed by materials (∼10 MW m⁻² and ∼1 MJ m⁻², respectively), local plasma temperatures below the threshold for physical sputtering (∼5 eV) [1], and a controllable divertor detachment state.

In the tokamak approach, future devices are planned to operate in high-confinement (H-mode) regimes [2] following the higher energy confinement with respect to low-confinement (L-mode) regimes [3]. The higher confinement manifests through the formation of an edge pedestal and steep gradient region, reducing the overall volume of the confined plasma necessary for a reactor, with the drawback of transient heat fluxes due to edge localized modes (ELMs) [4]. Integration of high-performing ELM-free scenarios [5, 6] (I-mode [7, 8], EDA H-mode [9], QH-mode [10], RMP ELM-suppressed regimes [11, 12], QCE regime [13], XPR regime [14]) with dissipative divertor conditions is being pursued in various tokamaks. The ability to robustly achieve H-mode grade confinement and reactor-relevant pressure levels with an L-mode-like edge [15] over a wide operational space [16] and the intrinsic stability to ELMs with a high power threshold for L-H mode transition [17, 18] has made scenarios with negative triangularity (NT) shaping attractive for the integration of core and edge plasma conditions [19–23].

Scaling of the scrape-off layer (SOL) heat flux width λ_q , detachment accessibility and its impact on core confinement need to be further addressed to assess the viability of NT as a reactor candidate. Experiments in the DIII-D tokamak in a NT shape with moderate NT ($\delta_{\mathrm{avg}}\sim -0.15$) indicated a quasi-continuum exhaust with λ_q 30%–50$\%$ wider than H-mode discharges in a similar shape [24]. A reduction in λ_q with NT shaping with respect to positive triangularity (PT) L-mode discharges was observed in TCV [25], consistent with the reduction in edge turbulence [26]. Gyrokinetic simulations (GBS, TOKAM-3X) observed a reduction in λ_q with NT due to a suppression of resistive ballooning modes driven by plasma shaping [27, 28]. In TCV, NT detachment experiments were attempted, but were unable to achieve detachment in Ohmic discharges with only intrinsic radiation due to the higher densities needed to detach compared to PT [29].

This letter reports on the first achievement of detached divertor conditions in NT discharges with intrinsic radiation, its integration with high confinement, and the first characterization of the parametric dependencies of detachment access in NT discharges with current, field and power. Detachment was achieved at high input power ($\unicode{x2A7D}$10 MW), high current ($\unicode{x2A7D}$1 MA) enabled by access to high normalized densities (up to 1.7 times the Greenwald density), with elevated normalized plasma pressures $\beta_\mathrm N \unicode{x2A7E} 2$ and H-mode confinement factors $H_{98-y2} \sim 1$ at the onset of detachment. A further confinement degradation of 20%–30$\%$ was experienced in the transition to deep detachment conditions.

Experimental setup. A NT armor was installed in DIII-D in 2023 with dedicated diagnostics for the initial characterization of the divertor/SOL in NT discharges [30]. Four new toroidal rows of graphite divertor tiles on the outboard side of the device, covering inner and outer strike points (ISP and OSP, respectively) enabled operation at high power and large negative δ ($\delta_{\mathrm{avg}}\sim -0.5$). Lower single null diverted discharges with negative upper and lower triangularity ($\delta_u/\delta_l \sim -0.35/-0.65$) with deuterium as the main fuel were obtained at plasma current $I_\mathrm p$ = 0.6–1.0 MA and toroidal magnetic field $B_\mathrm T$ = 2 T with injected power $P_{\mathrm{INJ}}$ from neutral beam ($P_{\mathrm{NBI}}$ = 0–10 MW) and electron cyclotron heating ($P_{\mathrm{ECH}}$ = 1.5 MW) with both ion B $\times\nabla$ B drift direction into (Fwd $B_\mathrm T$) and out of (Rev $B_\mathrm T$) the active (lower) divertor (see figure 4 for divertor geometry). Access to dissipative divertor conditions was investigated via upstream density ramps.

The NT geometry and the DIII-D coil and power supply current limits resulted in an open divertor configuration with short parallel connection length ($L_{\parallel}$) and short divertor poloidal leg length. Baseline NT shapes with $B_\mathrm T = 2$ T and $I_\mathrm p = 1$ MA in DIII-D have parallel connection lengths from outboard midplane to target of 8 m (outer) and 30 m (inner) about a factor of 3 and 2 shorter, respectively, than in a typical DIII-D PT discharge with strike point on the DIII-D lower divertor shelf at the same $I_\mathrm p$ and $B_\mathrm T$. The shorter $L_{\parallel}$ is a result of the short divertor leg (5 cm vs 10–15 cm) and of the lower $B_\mathrm T$ at the X-point (1.6 T vs 2.3 T) due to the larger X-point major radius.

Sustainment and controllability of dissipative divertor conditions at high performance. Power detachment conditions at the OSP were obtained with H-mode level confinement $H_{98-y2}\sim 1$ and reactor-relevant pressure $\beta_\mathrm N \sim 2$, with a reduction in target electron static pressure up to a factor of 2 but no reduction in particle flux. Sustainment of detached conditions was investigated by adding $\beta_\mathrm N$ feedback via neutral beam heating to pre-programmed density ramps in discharges with $I_\mathrm p$ = 0.6–0.8 MA and ion B $\times\nabla$ B drift out of the divertor (figure 1). As the line-averaged density was raised from 5–$8 \times 10^{19}$ m⁻³ through the detachment onset density of $6.5 \times 10^{19}\,$m⁻³ at 4.5 s, $P_{\mathrm{NBI}}$ was increased from 4.5–6 MW to maintain the β_N target of 2.1 due to the degradation in core confinement ($H_{98-y2}\sim 1.3 \rightarrow1$ during the detachment phase). Edge effective charge $Z_{\mathrm{eff}}$ remains low (∼1.5) in all conditions. In the detachment onset phase, $T_{\mathrm{e-OSP}}$ is reduced to the lowest measurable by Langmuir probes (∼ 3–4 eV) without a clear reduction in ion saturation current ($J_{\mathrm{sat}}$) at the outer strike point, indicating the absence of deep detachment conditions (as typically observed also in PT discharges in this drift direction [31]), but with a reduction in peak heat flux of 4 × and plasma pressure of 2 × as derived by the Langmuir probes. The two transient reductions in $T_{\mathrm{e-OSP}}$ at 2.3 and 4.3 s are due to radial strike point sweeps placing the Langmuir probe in the private flux region. Similar attempts in Fwd $B_\mathrm T$ resulted in deeper detachment conditions with a 2 × reduction of target $J_{\mathrm{sat}}$, further degraded performance ($H_{98-y2} \sim 0.7$) and the inability to match the pre-detachment $\beta_\mathrm N$ target to within 15%.

Zoom In Zoom Out Reset image size

Detachment did not show the bifurcating behavior at the OSP typical of H-mode discharges, suggesting improved controllability of detached divertor conditions. In DIII-D PT H-mode discharges with ion B $\times\nabla$ B drift into the divertor, detachment shows a bifurcation (the detachment cliff [32]) which has been associated with the interplay between radial E × B drift and collisionality [33], dependent on radial profile scale lengths. The detachment cliff typically leads to the radiation front jumping from the divertor plate to just below the X-point and to hysteresis in the degree of detachment, both of which complicate control of divertor conditions. Controllability of divertor detachment in NT was tested with pre-programmed density perturbation to a discharge at the onset of detachment. A gradual evolution of the divertor radiation front as measured by C III radiation was observed. The movement of the C III radiation front away from the target plate correlated with a decrease in the target $J_{\mathrm{sat}}$ and an increase in the emission from high-n Balmer transitions (upper state n = 9–11) line-integrated through the radiation front, the latter two indicate the occurrence of recombination processes. Differences in geometry, radial transport and the impact of cross field drifts are discussed in the next paragraph to explain observed characteristics of access to detachment in NT discharges.

SOL characterization and access to detachment. Radial profiles of density ($n_\mathrm e$) and temperature ($T_\mathrm e$) typical of a NT plasma edge indicate reduced radial scale lengths and transport compared to PT L-mode discharges. $T_\mathrm e$ and $n_\mathrm e$ profiles were measured by Thomson scattering with the upstream separatrix location identified from power balance constraints [34]. NT discharges displayed a small $T_\mathrm e$ pedestal without an $n_\mathrm e$ pedestal. Electron temperature scale lengths at the separatrix $\lambda_{\mathrm{Te-sep}} = (1/T_e \mathrm dT_e/\mathrm dr)^{-1} \sim$ 4–10 mm approached values typical of H-mode discharges while electron density scale lengths $\lambda_{\mathrm{ne-sep}} = (1/n_e \mathrm dn_e/\mathrm dr)^{-1} \sim$ 8–20 mm remained intermediate between those typical of L and H-modes.

The NT SOL heat flux width λ_q at the OSP, measured by infrared thermography and Langmuir probes, was between 1.5 and 3.5 mm for the plasma currents explored in the NT campaign, shorter than PT L-mode discharges and L-mode multi-machine scalings and approached values consistent with H-mode drift-based scalings [35, 36], as previously observed in discharges with reduced NT in DIII-D [24]. A shorter λ_q compared to PT L-mode is consistent with a reduction in edge turbulence and the short parallel connection length. Edge scale lengths intermediate between L- and H-mode are possibly responsible for the observed absence of a detachment cliff. An inverse dependence of λ_q on $I_\mathrm p$ is observed and a detailed discussion of λ_q and its dependencies is planned in a separate manuscript.

Higher densities were necessary to access detachment in NT discharges compared to PT, while parametric dependencies of the detachment density on $I_\mathrm p$ and on power flowing into the SOL $P_{\mathrm{SOL}}$ remained consistent with PT scalings. Detached conditions at the OSP were accessed using density ramps via main chamber D₂ fueling in Ohmic and auxiliary heated discharges with $P_{\mathrm{INJ}} =$0–10 MW and $I_\mathrm p =$0.6–1.0 MA. Detachment onset conditions were identified by the decrease of $T_{\mathrm{e-OSP}}$ (measured by Langmuir probes) to $\unicode{x2A7D} 5$ eV, from the movement of the C III visible radiation front (associated with $T_\mathrm e \sim$ 8–12 eV) away from the divertor plate and from the rollover in $J_{\mathrm{sat}}$ (in discharges with ion B $\times\nabla$ B drift towards the divertor). $n_{\mathrm{e-sep}}$ and Greenwald fraction $f_{\mathrm{Gw}}$ at the onset of detachment are plotted in figure 2 in blue and red for discharges in Rev $B_\mathrm T$ and Fwd $B_\mathrm T$, respectively. Stars, hearts and triangles represent discharges with $I_\mathrm p$ = 0.6, 0.8 and 1.0 MA (at fixed $B_\mathrm T$), respectively. Higher $n_{\mathrm{e-sep}}$ was required to detach with increasing $I_\mathrm p$, consistently with the shorter $L_{//}$ and narrowing of λ_q . A linear regression carried out separately over the datasets with different drift directions, gives $n_{\mathrm{e-sep-det}} \sim I\,{_\mathrm p^{0.8}}\times P_{\mathrm{SOL}}^ {0.4}$ in Fwd $B_\mathrm T$ and $n_{\mathrm{e-sep-det}} \sim I\,{_\mathrm p^{0.58}}\times P_{\mathrm{SOL}}^{0.3}$ in Rev $B_\mathrm T$. The stronger $I_\mathrm p$ exponents in Fwd $B_\mathrm T$ is likely related to a measured stronger dependence of λ_q on $I_\mathrm p$. The $I_\mathrm p$ = 1 MA data point in Fwd $B_\mathrm T$ (empty triangle) represents the highest obtained density but did not achieve detachment, thus the $I_\mathrm p$ exponent represents a lower bound. Parametric dependencies are generally consistent with two point model and detachment scalings [37, 38] which indicate $n_{\mathrm{e-sep-det}} \sim L_{\parallel}^{-2/7} \lambda_q^{-5/7} P_{\mathrm{SOL}}^{0.5}\sim I_p P_{\mathrm{SOL}}^{0.5}$. However, the densities needed to achieve detachment were higher than in PT, with f_Gw at detachment onset larger than 1, compared to $f_{\mathrm{Gw}} \sim$ 0.6–0.8 for PT H-mode detachment in DIII-D in open divertor in Fwd $B_\mathrm T$. The 3 × difference in $L_{\parallel}$ together with the narrowing of λ_q generally explain the observed difference between PT and NT discharges.

Zoom In Zoom Out Reset image size

A larger effect of cross field drifts on in/out divertor profile asymmetries leads to a wider difference in detachment onset density in Fwd $B_\mathrm T$ versus Rev $B_\mathrm T$ ($f_{\mathrm{Gw}}\sim$ 1.3 vs. 0.9–1.0) compared to PT discharges. In discharges with matched upstream density and input power in attached divertor conditions, $n_\mathrm e$ at the OSP was up to three times higher in Rev B_T compared to Fwd $B_\mathrm T$ (figure 3-top). The dominant component in the drift-driven plasma flux is proportional to $E\times B / |B|^2$. Edge fluid simulations with the multi-fluid edge code UEDGE [39] comparing PT, NT and NT shapes with longer divertor legs indicated that the larger integrated E × B flux in NT discharges, leading to a larger effect of cross-field drifts in NT shapes compared to PT, was mostly due to the 30$\%$ lower $B_\mathrm T$ at the X-point. Consistently lower $n_{\mathrm{e-sep}}$ (up to 30$\%$) was needed to achieve OSP detachment in Rev $B_\mathrm T$ with respect to Fwd $B_\mathrm T$, while this difference can be up to only $10\%\unicode{x2013}20\%$ in equivalent PT discharges [32]. The higher density needed in Fwd $B_\mathrm T$ is consistent with the E × B drift rarefying the OSP due to the combination of radial and poloidal drifts. Rev $B_\mathrm T$ configuration also resulted in more balanced divertor conditions with ISP and OSP detaching at nearly the same density. UEDGE simulations, with the inclusion of cross-field drifts and charge-state resolved carbon impurities, performed for discharges with $I_\mathrm p = 0.6$ MA and $P_{\mathrm{INJ}} = 4$ MW, are able to generally reproduce the large difference in detachment onset density in the different drift direction (figure 3-bottom), while still somewhat underestimating the density at detachment onset. The simulations were performed adjusting radial transport coefficients (particle D and thermal χ diffusivities) to match upstream $n_\mathrm e$, $n_\mathrm C$, $T_\mathrm e$ profiles in attached conditions. D and χ were assumed to be unchanged through the density ramp. The simulations in Fwd $B_\mathrm T$ exhibit a detachment cliff which is not observed experimentally, possibly due to the degradation in confinement approaching detachment. The agreement of fluid simulations with experimental data indicates that geometry and radial transport (input to the simulations) can explain differences from PT discharges.

Zoom In Zoom Out Reset image size

Confinement degradation with deep detachment. Confinement degradation was observed for detached conditions beyond the onset of detachment with both ion B $\times\nabla$ B drift directed into and out of the divertor, possibly due to the open nature of the NT divertor. The steep reduction in core confinement ($H_{98-y2}$ reduced by up to 30$\%$) observed at the onset of detachment (figure 4-top, left—black) correlated with the sudden increase in $\lambda_{\mathrm{Te-sep}}$ (red). Examining radial T_e profiles (figure 4-top, right) for attached and detached times, the increase in $\lambda_{\mathrm{Te-sep}}$ corresponds to the loss of the $T_\mathrm e$ pedestal with detachment and edge $T_\mathrm e$ profiles which become comparable to those typical of L-mode discharges. The loss of the $T_\mathrm e$ pedestal at detachment was concomitant to the movement of the high field side (HFS) divertor radiation front above the X-point, leading to the formation of a stable high radiated power density region in proximity of the last closed flux surface (figure 4-bottom, as measured by bolometry). The overall confinement degradation at detachment was limited to ${\sim}30\%$ as core profiles (which drive a large part of the NT confinement improvement) as well as the edge density remain unaffected by detachment. The confinement degradation in Fwd $B_\mathrm T$ started before the onset of detachment with a continuous degradation of the edge $T_\mathrm e$ pedestal likely due to the larger in/out asymmetry in divertor conditions and the presence of HFS radiation above the X-point before detachment of the outer leg. These results highlight the need for divertor optimization towards improved core-edge integration and the possible benefits of operation in Rev $B_\mathrm T$ in the absence of H-mode.

Zoom In Zoom Out Reset image size

Conclusions. Experiments in the 2023 NT campaign in DIII-D achieved divertor detachment and an ELM-free edge [18] in plasmas with NT shaping ($\delta = -0.5$), demonstrating the potential for application of NT regimes to a core-edge integrated scenario. SOL heat flux widths approach values typical of inter-ELM H-mode discharges but more data across devices are needed to understand its scaling and extrapolation. The changes in radial transport and geometry are responsible for the very high detachment density while the parametric dependence of detachment density on $P_{\mathrm{SOL}}$, $I_\mathrm p$ remained similar to PT discharges. Confinement degradation and the loss of temperature pedestal limit the performance obtained in NT plasmas with dissipative divertor conditions, possibly limited by an open divertor with short poloidal leg length in these DIII-D plasmas. The absence of H-mode physics (e.g. difference in L-H power threshold, confinement degradation when approaching $P_{\mathrm{SOL}} \sim P_{\mathrm{LH}}$) and the similar confinement before detachment, together with the symmetrization of inner and outer divertor leg conditions (ISP and OSP detaching at nearly the same upstream density) make ion B $\times\nabla$ B drift direction out of the active divertor the favorable configuration for core-edge integration in NT plasmas. The ability to reproduce experimental observations of access to detachment in NT plasmas with edge fluid codes increases the confidence in future studies for the optimization of NT divertors in DIII-D and future devices.

This material was supported by the U.S. Department of Energy, Office of Science, Office of Fusion Energy Sciences, using the DIII-D National Fusion Facility, a DOE Office of Science user facility, under Awards DE-AC52-07NA27344, DE-FC02-04ER54698, DE-AC05-00OR22725, DE-SC0022270, DE-SC0016154, DE-FG02-97ER54415. This report is prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. This work has been carried out within the framework of the EUROfusion Consortium, via the Euratom Research and Training Programme (Grant Agreement No. 101052200—EUROfusion) and funded by the Swiss State Secretariat for Education, Research and Innovation (SERI). Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union, the European Commission, or SERI. Neither the European Union nor the European Commission nor SERI can be held responsible for them.