LLNL modelling maps NSTX-U divertor heat Flux limits for graphite and lithium

NSTX-U’s vessel interior at Princeton Plasma Physics Laboratory, the spherical tokamak at the centre of LLNL’s divertor heat flux modelling

(Image courtesy of Michael Livingston/PPPL Office of Communications)

New modelling from Lawrence Livermore National Laboratory quantifies NSTX-U divertor heat flux limits for graphite and lithium-coated plasma-facing components. Using the UEDGE code coupled to a two-dimensional wall transport solver, the study finds that graphite tiles reach their sublimation threshold once peak heat flux passes roughly 7 MW per square metre, while a 5mm lithium coating can cap surface temperature below approximately 700°C but risks pushing core lithium concentration past a 3% fuel-dilution limit.

LLNL’s UEDGE modelling shows graphite divertor surface temperature crossing the approximately 1200°C sublimation threshold during a typical NSTX-U-length discharge at higher input powers

(Image courtesy of Islam et al., Nuclear Fusion 66 106017 (2026), CC BY 4.0)

Graphite divertor tiles hit their heat flux ceiling early

Graphite is planned as NSTX-U’s primary initial plasma-facing component material, without active cooling, a choice that shapes how much heating power the machine can safely apply. Across simulated input heating powers from 4 MW to 10 MW, surface temperature rises continuously through a discharge with no mechanism to arrest it. Carbon radiation from sputtered impurities dissipates part of that load, reaching about 2.5 MW at the highest input power modeled, but that is not enough on its own. Once peak heat flux exceeds approximately 7 MW per square metre, surface temperature crosses the roughly 1200°C point at which graphite starts to sublimate and shed carbon into the plasma. Avoiding that outcome at higher heating powers requires operating at the highest core densities scanned in the study, narrowing the safe operating window for a machine meant to explore higher-power regimes.

The snowflake divertor spreads heat but creates a hot secondary strike point

An alternative magnetic geometry, the snowflake-minus divertor, splits the heat load across four strike points instead of concentrating it on one. It extends the field-line connection length and expands the divertor volume available for cooling, a design candidate for future high-power operation. Most of the power still lands on the primary strike points, but a churning-mode instability near the magnetic X-points can redirect substantial divertor heat flux onto the secondary strike point, SP#2, where field lines strike at close to a 90-degree angle. A separate geometric projection, built to test the worst case for a protruding tile edge rather than to represent expected conditions, put the heat flux at SP#2 as high as 500 MW per square metre at that incidence angle under the strongest churning-mode transport assumed. That figure is an unbuffered upper bound rather than an expected operating value, since it excludes compensating effects such as enhanced local neutral recycling that a full three-dimensional model would capture.

Lithium vapor shielding controls divertor heat flux at the cost of fuel dilution

For the second part of the study, the graphite tiles are coated with a 5mm layer of lithium, with lithium emission sourced dynamically from a coupled surface model rather than fixed as an input. Lithium evaporation stays low below roughly 500 to 600°C, then rises sharply as surface temperature climbs further. That evaporated lithium radiates in the divertor plasma and caps surface temperature below approximately 700°C even as core heating power increases, a behaviour the study terms vapor shielding. The trade-off is upstream contamination. For a 5-second discharge, any input power above 4 MW pushes upstream lithium concentration past the 3% limit treated as the ceiling for avoiding fuel dilution, and above 6 MW that ceiling is breached within a few seconds. Increasing core plasma density widens the safe window somewhat, because higher density increases the frictional drag that keeps lithium ions confined near the target rather than migrating upstream. The lithium layer itself proves durable within the simulated timescale: the deepest depletion recorded, in the highest-power case, was about 0.15mm out of an initial 5mm thickness after three seconds of exposure.

The findings point to active cooling and geometric baffling for future PFCs

Both material options come across as partial solutions rather than settled answers. Graphite cannot handle sustained high-power operation without a narrow, high-density operating envelope, and lithium successfully manages the heat problem while creating a fuel-dilution problem of its own. Closing that gap, the study argues, needs three things: thicker lithium layers to prevent localized strike-point depletion, active cooling approaches such as fast-flowing liquid lithium systems, and geometric baffling or additional deuterium gas puffing near the divertor to confine lithium and limit its migration into the core. These are steps the authors treat as necessary for steady-state operation, not incremental refinements to the current approach.

Source: Islam, M.S., Umansky, M.V. and Soukhanovskii, V.A., “Modeling of divertor heat flux limits and lithium vapor shielding in NSTX-U using UEDGE code,” Nuclear Fusion 66 (2026) 106017.