Plasma boosts its own particles, quiets itself, and bubbles trouble

A fusion plasma quietly supercharged its own particles — by 2.5 times

Shoot particles into a fusion plasma at 40,000 volts of energy and some come back out at 102,000 — no explosion required.

Here is what happened. On the EXL-50U spherical torus — a compact, roughly ball-shaped fusion device in China — a team injected beams of neutral hydrogen particles carrying 40 kilo-electron-volts of energy. When they measured what was bouncing around inside the plasma afterward, some particles had climbed to 102 keV. That is 2.5 times the energy they went in with. And critically: the plasma stayed calm the whole time. No major eruption, no confinement collapse, just a quiet internal process handing energy to the particles. Think of it like a surfer paddling into a stretch of small, overlapping ripples — each one too weak to do much alone — and somehow being launched by a combined push far bigger than any single wave. The mechanism the team points to is small-scale magnetic reconnection: tiny pockets of the magnetic field snapping and rearranging on centimetre scales, releasing energy in bursts that happen to land on passing fast particles. The team backed this up with particle-in-cell simulations — computer models that track how individual charged particles respond to electromagnetic fields — and the numbers matched reasonably well. Why does this matter? In a burning fusion plasma, energetic particles — including the helium produced by the fusion reaction itself — need to stay confined long enough to heat the surrounding plasma. If small-scale reconnection is routinely boosting particle energies, that is potentially free internal heating we hadn't planned for. It also means our models of how fast-ion populations evolve may be missing something. The catch is significant though. This is four plasma shots on one small experimental device. The spherical torus geometry is more compact than a standard tokamak, and results don't automatically scale to ITER or any large machine. The team is honest: this needs reproduction elsewhere before anyone gets too excited.

A plasma's own pressure can silence one of its most dangerous instabilities

A plasma under pressure shifts its own magnetic geometry — and that shift turns out to calm down one of fusion's nastiest internal waves.

When a fusion plasma gets hot and dense, its pressure physically pushes the magnetic field outward. The centre of the plasma shifts slightly away from the geometric centre of the tokamak — this displacement is called the Shafranov shift, named after the Soviet physicist who described it. It is a known effect, but what it actually does to stability has been hard to pin down because you need very expensive simulations to model it properly. A team using the ORB5 simulation code — running on grids of up to 512 by 1024 by 256 points with 40 million computational particles per species — has now done exactly that. Their result: when you properly include the pressure from energetic particles in the magnetic equilibrium calculation, the growth rate of Toroidal Alfvén Eigenmodes drops by roughly 90%. Alfvén Eigenmodes are a type of wave that can form inside a tokamak plasma, driven by fast particles rattling around inside the magnetic cage. Think of them like resonant vibrations in a guitar string — they only grow when driving conditions are just right, but when they do, they can kick fast particles out of the plasma and cause heating loss. A 90% reduction in growth rate is the difference between a string vibrating wildly and barely moving. The team also found something counterintuitive: this strong linear stabilization does not reduce how large the waves get at saturation — they still reach similar peak amplitudes. But the heat and particle transport those waves carry is reduced. The wave vibrates; it just carries less cargo. I simplified here. These are linear and early nonlinear simulations; a full burning plasma is more complex. But this is the kind of physics that has to be modelled correctly before ITER turns on.

Helium bubbles in liquid metal blankets are more disruptive than expected

The liquid metal meant to breed fusion fuel has a fizzy problem — helium barely dissolves in it and then bubbles out in complicated ways.

You probably know that fusion reactors don't run on plain hydrogen — they run on tritium, a rare heavy variant that barely exists in nature. The plan is to breed tritium inside the reactor walls using a liquid blanket made of lead-lithium alloy: neutrons from the fusion reaction hit lithium atoms and produce fresh tritium. It is the best idea we have. But the blanket is not just a passive target — it is a hot, flowing liquid that has to survive inside one of the most hostile environments humans have ever built. One of the things that happens inside that liquid is helium production. When neutrons hit lithium, you get tritium and helium as a byproduct. Helium barely dissolves in liquid metals — think of trying to dissolve air into cold cooking oil, and you will get the idea. It comes out of solution almost immediately and forms bubbles. Those bubbles create local pressure differences at the interface between the helium and the liquid metal, and the physics of that interface turns out to be non-trivial. A team used classical molecular dynamics simulations — modelling the motion of individual atoms using established force descriptions — to map out how interfacial tension and bubble radius change as a function of temperature and the exact ratio of lead to lithium in the alloy. They found non-ideal behaviour: the mixture does not act like a simple blend, especially near a composition of about 80% lithium and 20% lead. Why does this matter? If bubble behaviour is hard to predict, it complicates the engineering of tritium extraction systems and could affect how evenly the blanket heats and flows. The catch: these are simulations of idealised systems; actual blanket conditions involve radiation damage, flow turbulence, and impurities that no simulation at this scale can yet fully capture.

Look at what these three papers are each pointing at, and a pattern emerges. The first says that small, fast internal events inside a plasma — events we often treat as noise — can move energy in ways that are larger than expected. The second says that the plasma's own pressure, properly accounted for, can suppress one of the instabilities we fear most. The third says that even the engineering systems we think we understand — the liquid blankets sitting outside the plasma — have unresolved physics in them at the atomic level. None of this is panic-inducing. But it does tell you that fusion science in 2026 is still very much a field of surprises: things we assumed were stable turn out to be calmer than expected, and things we assumed were solved turn out to have hidden complexity. The honest takeaway is that the models are getting better, but reality keeps adding detail. That is what progress actually looks like.

EXL-50U is an active machine — if the ion re-acceleration result holds up, the team will likely publish follow-up shots with better diagnostics within months, and that will be the real test. On the blanket side, the EU is running the ITER Test Blanket Module programme, with results from early tritium-breeding experiments expected to filter out from 2026 onward; helium bubble behaviour will be a live question there. The open thing I'd most want answered: does the Shafranov shift stabilisation survive at the plasma pressures ITER will actually run at, or does it weaken at higher energetic particle fractions?

• Wikipedia: Tritium breeding in fusion reactors
• Wikipedia: Alfvén wave
• Wikipedia: Magnetic reconnection