Boron Powder, Steel Limits, and the Plasma That Tears Itself Apart

Sprinkling boron powder into a plasma cuts energy spikes by 76%

What if the answer to one of fusion's nastiest wall-damage problems was just: drop a little powder in?

A team working on the DIII-D tokamak at General Atomics in San Diego tried exactly that — they fed tiny amounts of boron powder into a running plasma and watched what happened to the plasma's stability. The problem they were tackling is called an ELM — an Edge Localized Mode. Think of it like a pressure cooker that periodically vents a burst of steam. In a fusion plasma, ELMs are sudden eruptions of hot particles that slam into the reactor wall in milliseconds. In ITER — the big international machine now being assembled in France — each ELM could deliver roughly as much concentrated heat as a lightning bolt to a wall surface the size of a dinner plate. Repeat that thousands of times and the wall erodes. What the team found is striking. At a boron injection rate of 4.5 milligrams per second — about the mass of a small raindrop per second — ELM frequency dropped by 76% compared to the reference plasma. At a higher rate of 9.7 milligrams per second, ELMs disappeared entirely for stretches of around 300 milliseconds. That is short in human terms, but for a plasma it is a meaningful gap of quiet. The mechanism seems to be that boron impurities enhance a low-frequency turbulence at the plasma edge, which increases particle leakage between ELMs and keeps the pressure from building to the point where an ELM fires. The team also found, for the first time, that low-atomic-weight impurities can decouple two different instability boundaries simultaneously — opening a possible route to a high-performance operating mode called super-H mode. The catch: this was five discharges in one experimental session on a mid-size machine. Nobody yet knows whether boron accumulation on the wall causes its own damage over long operation, or whether this approach scales to ITER-size plasmas. ELM-free is a promising step. ELM-free and wall-safe is the harder goal.

Future fusion power plants hit a structural wall beyond 20 tesla

There is a point where making fusion magnets stronger stops being a physics problem and becomes a steel problem.

A research team used the D0FUS tokamak design code to map out exactly where the structural limits fall for future fusion power plants running very high magnetic fields — and found a hard ceiling they hadn't fully quantified before. Here is the engineering problem. The coils that confine a fusion plasma are crushed from all sides by enormous electromagnetic forces. For today's machines, standard structural steel handles this. But as you push toward stronger magnets — which means tighter plasma confinement, which means more power out — the forces scale faster than the material can cope. Think of designing a skyscraper: beyond a certain height, the walls at the base have to be so thick to carry the load that there is no usable space left inside the building. The same thing happens in a tokamak. The structural layers surrounding the magnet coil — collectively called the radial build — become so bulky they physically cannot fit within a viable reactor footprint. The team found that for a DEMO-class power plant producing 2 gigawatts of fusion power, the standard steel and wedge configuration cannot be made to work beyond 20 tesla of peak magnetic field. For context, a hospital MRI runs at 1.5 to 3 tesla. The high-field machines being designed today are pushing toward that 20 T limit. Three levers help push the boundary further. Switching to a high-strength steel alloy called CHSN01, using a bucking mechanical architecture instead of a wedged one, and reducing the magnetic flux demand on the central solenoid each independently improve things. Combined, the team found, they enable compact machines with a major radius under 4 metres. The catch: this is analytical modelling, not a built machine. CHSN01 steel still needs validation under the radiation conditions inside a real fusion reactor, which no material has fully faced yet.

Step-by-step simulation of a plasma catastrophically tearing itself apart

The plasma temperature crashes to below 10 percent of its value in a few milliseconds — here is exactly how that happens.

A team ran detailed three-dimensional simulations of a plasma disruption — fusion's equivalent of an engine seizing — modelled on the Chinese CFETR reactor design, using a code called NIMROD. They traced, step by step, how the plasma unravels. The instability at the centre of the story is called a resistive wall mode, or RWM. Here is the analogy: picture a spinning top. A perfectly spinning top stays upright. But if it starts to wobble at a frequency that the surface beneath it responds to with a slight lag, the wobble can amplify until the top falls. In a tokamak, the plasma is the top and the metal wall surrounding it responds to magnetic disturbances with a small delay. When a particular magnetic distortion — in this simulation, a spiral pattern called the n=1 mode — grows faster than the wall can respond, the wobble takes over. Once the RWM gained hold in the simulation, magnetic field lines that had been neatly nested like layers of an onion began to tangle and cross. Within a few milliseconds, the core electron temperature crashed to below 10 percent of its initial value. Just before the final collapse, a brief current spike appeared — the plasma trying to conserve its magnetic flux as its internal structure shrank, like a figure skater pulling their arms in and briefly spinning faster before falling. The team found that wall conductivity matters enormously: a more resistive wall accelerates the disruption. Plasma resistivity, by contrast, barely changes the outcome at fusion-relevant conditions. The catch: this is a single-fluid simulation. Real disruptions involve kinetic effects, energetic particles, and impurity dynamics not included here. The sequence of events is credible. The precise timings and wall-load magnitudes need more complete physics to be trusted for engineering design.

Take these three papers together and a single picture forms: fusion is not one problem, it is a stack of problems that each have to be solved without breaking the others. The boron powder result says you can calm the edge of the plasma — but only if you accept some impurity in the fuel and some uncertainty about the wall over time. The structural limits paper says you can push magnets harder — but only if you redesign the steel cage, and only up to a point that is closer than some roadmaps assume. The disruption simulation says that when the plasma cage fails, it fails catastrophically and fast — and the wall conductivity you chose for one reason now determines how bad the crash is. None of these findings is fatal. All three are real steps toward understanding what a working power plant actually demands. But they are also reminders that the engineering margins in fusion are tight, and that progress on one constraint often tightens another. That is not pessimism — it is just the honest shape of where the field is.

The boron injection result from DIII-D will need replication across more discharges and ideally on a larger machine — watch for follow-up campaigns at DIII-D or results from JET's final dataset being analysed now. On the magnet side, the Commonwealth Fusion team is expected to publish further operating results from their SPARC high-field coil program in the coming months, which will test some of these structural assumptions with real hardware. The open question I'd most want answered: does the ELM-free period enabled by boron injection degrade the plasma's energy confinement in ways the stored-energy numbers don't yet show?

• Wikipedia: Edge-localized mode (ELMs) — background on why they matter
• Wikipedia: Tokamak disruptions — what they are and why engineers worry about them
• ITER Organization: Why controlling ELMs is a priority for ITER