Divertor

The divertor is the component that absorbs the harshest heat and particle loads inside a fusion reactor. It concentrates the heat leaking out of the plasma and the “ash” produced by burning, guides them to one place, and dumps them out of the reactor. On this page we build up an understanding of why the divertor is needed, why its design is so difficult, and where the research frontier lies today, starting from intuition.

Start with Intuition (High School)

You can grasp the essence of the divertor by thinking of it as the reactor’s “exhaust port.”

First, the plasma in a fusion reactor is at roughly 100 million degrees. We want to keep confining this heat forever, but no perfect thermos exists, and heat and particles always leak out little by little. If the leaked heat is not caught somewhere, the reactor wall would burn across its entire surface. So we deliberately gather what leaks into one place and build a dedicated area to process it. This is the divertor.

Picture a kitchen range hood. Instead of letting cooking smoke and heat spread throughout the room, it sucks them into a single point of the hood and sends them outside. The divertor works the same way: it guides the heat and particles leaking from the plasma to a limited surface at the bottom of the reactor (usually the lower part), cools them there all together, and pumps out the ash.

There are two things to collect. One is heat. The other is “ash.” In fusion, deuterium and tritium react to produce helium, but this helium is spent residue, so if left alone it accumulates in the fuel and dilutes it. The divertor also acts as a vacuum cleaner that pumps out this helium ash.

The biggest headache here is that “the heat concentrates too much into a narrow region.” It would be easy if the leaked heat spread thinly over a large area, but in reality it gathers into a very narrow band. As an analogy, it is like the heat when you focus sunlight to a single point with a magnifying glass. Heat comparable to the inner wall of a rocket engine rushes onto the divertor’s target plates. How to ease this concentration is the heart of divertor design.

Understand the Physics (Undergraduate)

The keys to understanding the divertor are the shape of the magnetic field lines and the physics of the thin layer through which heat flows.

The Separatrix and the X-point

In a tokamak plasma there are two kinds of surfaces formed by magnetic field lines (magnetic surfaces). One is the closed magnetic surface that circulates around inside the doughnut in a loop; the other is the open magnetic surface that runs into the reactor wall. The special magnetic surface that forms the boundary between the two is called the separatrix.

The separatrix forms because the current in external coils cancels the poloidal field (the field component that circulates within the cross section), creating $B_p = 0$ at a certain point. This point is called the X-point. Near the X-point the field lines appear to cross like the letter X. Inside the X-point the surfaces are closed; outside they are open, and the open field lines connect to the divertor’s receiving plates (target plates).

In other words, the X-point is the “branch point” that separates the confinement region from the exhaust region. Leaked particles ride the open field lines, pass through the X-point, and are reliably carried to the divertor.

The Scrape-off Layer (SOL)

The thin layer just outside the separatrix, occupied by open field lines, is called the scrape-off layer (SOL). Particles that enter it flow away along the field lines to the target plates.

Particles move fast in the direction along the field lines (the parallel direction) but diffuse only slowly across the field lines (the perpendicular direction). The competition between these two determines the thickness of the SOL. The faster the parallel transport, the sooner particles escape to the plates before spreading in the perpendicular direction, so the layer carrying the heat becomes thin. In fact, this thickness (the heat flux width $\lambda_q$) is only a few mm.

The Heat Flux Concentration Problem

If we call the heat leaking from the whole reactor $P_{SOL}$, it concentrates into a narrow band along the separatrix. The rough magnitude of the heat flux $q_\perp$ reaching the target plate perpendicularly can be estimated as follows.

$$q_\perp \sim \frac{P_{SOL}}{2\pi R \, \lambda_q} \sin\theta$$

Here $R$ is the major radius of the device, $\lambda_q$ is the heat flux width, and $\theta$ is the shallow angle at which the field lines strike the plate. Because the small $\lambda_q$ (a few mm) sits in the denominator, $q_\perp$ can easily shoot up to the 10 MW/m² class. This reaches the same order of magnitude as the heat flux radiated by the solar surface, and hitting the plate with it directly would exceed the material’s heat-removal limit.

So the design uses two tricks. One is to make $\sin\theta$ small, that is, to strike the plate with the field lines as shallowly as possible (a few degrees) so that the heat spreads thinly over a large surface. The other is detachment, described next.

Deepen the Theory (Graduate)

The undergraduate-level estimate alone does not make a real reactor divertor viable. Operation that keeps the plasma from “touching” the plate, and the radiation control needed for it, become necessary.

Detachment (Non-contact Plasma)

Operation that slams the heat straight into the plate is called attached. In contrast, the state in which the plasma temperature is dropped below a few eV just in front of the plate, greatly reducing the pressure and particle flux, is called detachment (non-contact plasma).

In detachment, several processes chain together just before the plate. As the plasma temperature drops, collisions with neutral particles become effective, and charge exchange, ionization, and recombination proceed. Momentum is lost through collisions with neutrals (momentum loss), and energy is dissipated by radiation and recombination. As a result, most of the heat and pressure are removed before reaching the plate, and the heat flux to the target drops by an order of magnitude. The plasma behaving as if it “floats” off the plate is the origin of the name.

However, detachment tends to become unstable, and if the detached region creeps up to the X-point, it degrades confinement. Control that keeps the detachment front near the target is an important operational challenge.

Radiative Cooling by Impurity Seeding

To sustain detachment stably, impurity gas is deliberately introduced. When nitrogen, neon, argon, and the like are seeded into the SOL or divertor region, these atoms and ions scatter energy in all directions as light through line-spectrum radiation. This is called radiative cooling.

The power density lost to radiation can be expressed using the electron density $n_e$, the impurity density $n_z$, and the temperature-dependent radiative loss function $L_z(T_e)$ as follows.

$$P_{rad} = n_e \, n_z \, L_z(T_e)$$

$L_z(T_e)$ is determined for each element, and the key is to choose an element that radiates efficiently in the divertor’s temperature range. The aim is to let the heat concentrated on the plate escape as radiation into a wide solid angle rather than onto the plate surface. Controlled well, most of the leaked heat can be processed by radiation. On the other hand, if impurities penetrate all the way to the core of the main plasma, they cool and degrade confinement, so a configuration that confines them to the divertor region and an exhaust design are indispensable.

Tungsten Monoblock Targets

The material-side solution is also part of the theory. Tungsten (W) is chosen as the armor material for the target. Its melting point of about 3422 degrees is the highest of all metals, its thermal conductivity is high, and it resists sputtering (being knocked out) by hydrogen.

What ITER adopts is the monoblock structure. A hole is drilled through the center of a prismatic tungsten block, a copper-alloy (CuCrZr) cooling tube is passed through it, and the gap is joined with a soft copper layer. Pressurized water flows through the cooling tube, and a swirl tape (a twisted metal strip) is placed inside to strengthen turbulence and enhance heat removal. This structure is designed to withstand the steady 10 MW/m² class, and even higher heat fluxes transiently. The ITER divertor is divided into 54 cassettes, made replaceable by remote handling.

Because different materials are stacked, stress from the difference in thermal expansion acts on the joints. This thermal stress, together with the degradation from neutron irradiation described later, are the two major constraints of material design.

Research Frontier (PhD)

The divertor is considered one of the last remaining obstacles to the practical realization of fusion, and active research continues.

SOL Width Scaling

One of the biggest unsolved problems is how the heat flux width $\lambda_q$ changes with reactor size. An empirical rule gathered from many tokamak experiments (known as the Eich scaling) found that $\lambda_q$ is essentially determined by the poloidal field and depends little on the size of the device. If this holds for large reactors as well, then in ITER and prototype reactors (DEMO) $\lambda_q$ would remain extremely narrow at a few mm, making the heat concentration even more severe.

On the other hand, theories based on turbulent transport (such as the heuristic drift model), and higher-performance operating regimes, are discussed as possibly widening $\lambda_q$. Whether the extrapolation of SOL width scaling holds, and what physics determines it, is a central research theme that will shape prototype reactor design. Since turbulence in the SOL is at play here, fundamentals such as Debye shielding, which lets you understand the shielding length of a plasma, form the foundation.

Advanced Divertor Configurations

Assuming a narrow $\lambda_q$, research that changes the magnetic configuration itself to thin out the heat is also progressing.

The Super-X divertor is a configuration that stretches the outer divertor leg far out to a larger major radius. The magnetic field at the plate position is weaker and the magnetic flux spreads, so the receiving surface increases; the connection length also grows, making detachment easier to trigger. Demonstration experiments are being carried out on MAST-U in the UK.

The snowflake configuration creates a second-order null near the X-point, broadening the region where $B_p \approx 0$. The magnetic flux spreads greatly near the X-point, and heat can be distributed to multiple targets. The name comes from the field lines looking like a snow crystal.

These are promising, but they require extra coils and complex field control, and compatibility with the whole reactor is a challenge.

Liquid Metal Divertors and Transient Heat Loads

Solid tungsten becomes embrittled by neutron irradiation, and its thermal conductivity drops. As a way to fundamentally avoid this wear and degradation, the liquid metal divertor, which keeps flowing liquid lithium or liquid tin, is being studied. Because the surface is constantly renewed, wear is less of a problem, while flow control and impurity contamination are challenges.

Transient heat loads also remain a difficult problem. ELMs (edge-localized modes), in which energy from the periphery erupts periodically, and disruptions, in which the plasma suddenly collapses, deliver heat to the target that momentarily exceeds material limits. Suppressing and controlling ELMs, and avoiding and mitigating disruptions, are research fields directly tied to divertor lifetime. The divertor’s heat and particle handling is also closely intertwined with the design of the first wall and plasma-facing materials.

Check Your Understanding

Q1. Which combination correctly describes the main roles of the divertor?

A Heating the plasma, and generating the magnetic field B Concentrating and processing the heat that leaks from the plasma, and exhausting impurities such as helium ash C Breeding fuel inside the reactor, and shielding neutrons D Driving the plasma current, and maintaining the vacuum

Q2. Which is the correct explanation of the separatrix and the X-point?

A The separatrix is the surface where the field is maximal, and the X-point is where the plasma current is maximal B The separatrix is the boundary between closed and open magnetic surfaces, and the X-point is the branch point where the poloidal field becomes zero and separates the confinement region from the exhaust region C The separatrix is the divertor target plate itself, and the X-point is its center D The separatrix is the surface of highest temperature, and the X-point is where the temperature becomes zero

Q3. Why is it a problem when the heat flux width lambda_q is small?

A Because the heat spreads too thinly over a large surface and the receiving plate cannot be cooled B Because the leaked heat concentrates into a narrow band of a few mm, and the heat flux on the target plate reaches the 10 MW/m2 class, exceeding the material's heat-removal limit C Because the field lines hit the plate perpendicularly and are reflected D Because impurities more easily penetrate into the main plasma

Q4. What kind of state is detachment (non-contact plasma)?

A A state that slams the plasma directly into the plate to transfer heat efficiently B A state that cuts the magnetic field and completely expels the plasma out of the reactor C A state in which the plasma temperature is dropped below a few eV just in front of the plate, dissipating heat and pressure through radiation and collisions with neutral particles, and lowering the heat flux reaching the plate by an order of magnitude D A state that raises the plate temperature to the melting point of tungsten and melts it

Q5. Why is SOL width scaling important for prototype reactor design?

A Because lambda_q always widens the larger the device is, so heat concentration is automatically eased in large reactors B Because the empirical rule predicts lambda_q stays at a few mm regardless of device size, and if that is correct, heat concentration becomes very severe even in large reactors, so the validity of the extrapolation shapes the design C Because lambda_q is determined by the toroidal field alone, so coil design can be simplified D Because lambda_q is proportional to the operating temperature, so lowering the temperature makes the problem disappear

Related Topics

• Debye shielding
• First wall
• Plasma-facing materials