Fusion's ash cloud, a boron pebble, and liquid metal walls

Fusion's own helium ash is accidentally suppressing dangerous turbulence

The helium ash that fusion reactions produce — waste, essentially — turns out to be calming the turbulence that was threatening to bleed heat out of the plasma.

When a fusion reactor runs, deuterium and tritium fuse into helium nuclei (called alpha particles) moving at enormous speed. The conventional worry is that these fast particles could destabilise the plasma. What a team using the CGYRO simulation code found, for a design called ARC V3A (being developed by Commonwealth Fusion Systems), is almost the opposite: those fast alphas are actually reducing turbulence in the inner half of the plasma. Think of the plasma as a pot of water on a gas stove. If you turn the heat up too fast, you get rolling turbulence that flings heat in all directions. Now imagine that the bubbles rising from the boil somehow organise themselves into little circular currents — whirlpools — that redirect the chaos. That is roughly what is happening here. The fast alpha particles trigger large-scale wave modes, and those modes amplify what physicists call zonal flows — ring-shaped streams inside the plasma that act like traffic roundabouts, redirecting turbulent eddies before they can carry heat outward. The result is a significant reduction in the heat and particles leaking out of the inner core at radius r/a ≤ 0.5 (meaning the inner 50% of the plasma cross-section). Why does this matter? In a fusion power plant, energy gain is measured by Q — the ratio of power out to power in. The ARC scenario they modelled reaches Q of about 22, meaning it produces 22 times the energy needed to run it. Less turbulence means more of that heat stays where it belongs. The suppression also gets stronger as alpha density increases — a self-reinforcing benefit as the reactor runs hotter. The catch: these are simulations, not measurements from a real device. The plasma profiles they used come from a medium-fidelity integrated model called MAESTRO, which is itself an approximation. The stabilising effect is also confined to the inner core — the outer half of the plasma is a different story. Nobody has measured this in a burning plasma yet.

A boron pebble rod survived 80 MW/m² in a real reactor — then shed half itself as dust

Researchers at DIII-D stuck a small rod of boron pebbles into the hottest part of a working tokamak and hit it with heat loads equivalent to 80,000 household hairdryers per square metre.

Boron is fusion's grease — spread it on the inside walls of a tokamak and it soaks up stray oxygen and carbon, keeping impurities out of the plasma. Today's reactors do this by periodically venting the machine and coating every surface. That works, but it means downtime. The idea behind a 'renewable boron pebble aggregate' is simpler: mount a small rod of compressed boron pebbles right in the divertor (the exhaust zone at the bottom of the reactor) and let the plasma erode it continuously, like a slow-burning candle replenishing its own coating. A team at General Atomics tested this concept for the first time in the DIII-D tokamak in San Diego. They mounted pebble rods on a sample holder and swept the plasma's exhaust point back and forth across them — roughly like dragging a welding torch along a surface. Heat fluxes reached 80 megawatts per square metre. For reference, the surface of the Sun is around 60 megawatts per square metre. The rods survived. The plasma was not contaminated badly enough to disrupt. So far, good. But here is the complication: about half the released boron did not coat the surrounding surfaces as intended. It flew off as millimetre-sized dust particles, scattered into the vacuum vessel, and briefly spiked the boron concentration in the plasma core during each sweep. That 50% dust loss is the main engineering puzzle now. A self-renewing boron source that also sprays debris into the machine is not quite the clean solution it sounds like on paper. The team found no lasting performance penalty in these short tests — the core plasma returned to normal quickly. Whether that holds in sustained, high-energy operation is a genuinely open question.

A free tool for simulating liquid metal reactor walls just got a critical physics fix

Some fusion reactor designs want to replace solid metal walls with flowing liquid metal — and the simulations to design those walls just became meaningfully more accurate.

One of the most extreme engineering problems in fusion is the wall: the surface closest to the plasma absorbs enormous heat and particle bombardment, and solid tiles crack, erode, and degrade. One proposed solution is to replace sections of the wall with a thin film of flowing liquid metal — lithium or tin — that continuously circulates, carries heat away, and self-repairs. The metal flows, the damage flows with it, and fresh material keeps arriving. To design such a system, you need simulations that accurately model how liquid metal behaves when it is immersed in the reactor's powerful magnetic fields. The tricky part is that the metal is electrically conductive, so it generates its own magnetic field as it moves. Think of it like this: if you stir a pot of water with a magnet, the water does not talk back. But if you stir a pot of mercury with a magnet, the mercury's motion creates its own electrical currents, which create their own magnetic field, which pushes back on the original magnet. That feedback loop is what previous open-source solvers were ignoring — they used something called the inductionless approximation, which assumes the metal's self-generated field is negligibly small. A team extended the open-source code FreeMHD2 (built on the engineering platform OpenFOAM) to model that feedback self-consistently. They verified it against textbook analytical solutions for duct flow, and validated it against physical measurements from the LMX-U liquid metal experiment. The honest caveat: the validation was done at relatively modest magnetic field strengths — far below what a fusion power plant would impose. The team also did not report precise error numbers in the abstract text available. This is an important tool upgrade, not a proof that liquid metal walls will work.

Look at these three stories together and you see fusion research closing in on the same hard problem from three angles: how do you keep a 100-million-degree plasma stable and contained for long enough to be useful, when the walls around it are being destroyed and the plasma itself is perpetually trying to leak? The alpha particle result says the plasma may be more self-stabilising than we thought — its own byproducts help it. The boron pebble experiment says renewable wall conditioning is physically plausible but messier than hoped. The liquid metal solver says the computational tools to design extreme-heat walls are maturing — slowly, carefully, with real validation. None of this is a straight line to a power plant. But I'd push back against the idea that fusion is stuck. What you are watching is the unglamorous, essential work of making the numbers trustworthy and the materials real. That is what progress looks like before anything gets built.

The DIII-D boron pebble team will almost certainly run longer exposure sequences to see whether that 50% dust loss stabilises or worsens — watch for follow-up publications from General Atomics in the next few months. On the simulation side, I am curious whether the FreeMHD2 team will next benchmark against higher-Rm conditions closer to actual fusion parameters; that would significantly strengthen the case for liquid metal divertors. The open question I most want answered: does the alpha particle turbulence suppression survive when the outer plasma region is also modelled — or does the benefit dissolve at the edges?

• Wikipedia: Divertor — what the exhaust zone of a tokamak actually does
• Wikipedia: Plasma-facing material — the wall problem explained
• Wikipedia: Alpha particle — what fusion ash actually is