EAST Tokamak Pushes Plasma to 1.65 Times the Greenwald Density Limit

What happened in Hefei

China's Experimental Advanced Superconducting Tokamak (EAST), often called the "artificial sun," has run stable plasmas at densities well above the Greenwald limit. That limit is the rule of thumb that has set how much fuel a tokamak can safely hold for nearly four decades. Researchers reported line-averaged electron densities between 1.3 and 1.65 times the Greenwald value. EAST normally runs at 0.8 to 1.0 times that value.23 The paper, titled "Accessing the density-free regime with ECRH-assisted ohmic start-up on EAST," appeared in Science Advances in January 2026.11

Prof. Ping Zhu of Huazhong University of Science and Technology and Associate Prof. Ning Yan of the Hefei Institutes of Physical Science co-led the study.11 The author list includes Dominique Franck Escande, whose theory the experiment was built to test.11 The Institute of Plasma Physics at the Chinese Academy of Sciences, Huazhong University and Aix-Marseille University worked on it together.2

The popular headline says scientists thought this limit was "unbreakable." That overstates it, but the result itself holds up. The real news is not that a tokamak went past the Greenwald number. It is that the machine did so in a way that confirms a specific theory about why the limit exists, using wall materials that future power plants are expected to use.

Why density matters

In deuterium-tritium fusion at about 150 million kelvin, fusion power rises with the square of plasma density.11 Double the density and the output roughly quadruples.18 The catch is that tokamaks have a density ceiling. When operators push past it, the plasma usually becomes unstable, loses confinement and can dump energy into the vessel walls.3 So machines are normally run below it.6

Martin Greenwald identified the limit in 1988.1 It is often described as a law, but it is a fitted formula: it summarizes what experiments show rather than deriving the result from first principles.16 Chris Eaglen of IChemE's nuclear technology special interest group, who was not involved in the study, said engineers treated it as a firm ceiling for decades. Machine sizes, fueling plans and safety margins were all set around it.17

The theory behind the result

This is why the story is about proof, not just a new record. The Greenwald formula describes where plasmas fail but does not explain why. The fusion community has long believed the trigger sits in the boundary between the plasma and the wall, but the mechanism has never been fully understood.3

Plasma-wall self-organization (PWSO) theory, proposed by Escande and colleagues at France's CNRS and Aix-Marseille University, offers an explanation.11 When hot plasma strikes the wall, it knocks loose atoms, a process called sputtering. Those impurities drift into the plasma and radiate away energy. If that radiation outruns the heating, the plasma cools and collapses.

The theory predicts two possible outcomes. In the normal regime there is a hard density limit. In a "density-free" regime, the ceiling moves up to very high values.210 "Density-free" does not mean unlimited density. It means the practical cap is pushed far out of the way.106

Reports disagree on when the theory was first published. One dates the founding work to 201717, while another puts the key formulation in 20228. This may simply be the gap between early papers and the developed model, but it is a reminder that the theory is fairly new.

Also worth noting: the power-dependent density-limit scalings derived from PWSO had already matched results on several tokamaks. Stellarators were also known to reach higher density limits when started with more microwave heating power.2 So EAST was testing a prediction that already had some support, not an untried idea.

Why tungsten made the difference

The result depends heavily on what the inner walls are made of. Before EAST, the Joint Texas Experiment Tokamak (J-TEXT) at Huazhong University tested PWSO with carbon walls.2 There, more microwave heating or higher starting gas pressure raised the density limit, but the plasma stayed inside the regime that still has a limit.2

Carbon walls shed impurities mainly through chemical reactions. Tungsten walls shed them mainly when particles physically knock atoms loose. The theory singled out that second case as the route into the density-free regime.2 Tungsten also releases very few particles at low temperatures, which is why the theory treats metal walls as essential.10 EAST's all-tungsten divertor made it a natural test bed.

The method itself was simple. Operators raised the starting fuel-gas pressure and added electron cyclotron resonance heating, a form of microwave heating, as each discharge began. This shaped how the plasma and wall interacted from the first moment.11 Cooler conditions at the divertor targets meant less tungsten was knocked loose, so less energy was lost to radiation.38

The effect fed on itself: a cleaner plasma kept the edge cooler, and a cooler edge produced fewer impurities.18 Density limits also rose as wall conditions improved over repeated shots, which fits what the theory predicts.

This matters for materials science because tungsten is the wall material expected in commercial reactors. Analysts point to this as what separates EAST from earlier attempts.18 In this reading, the wall is part of the solution rather than just a material that has to survive the plasma.

Is it really a first?

The coverage disagrees here, and the details matter. Several outlets present the result as a historic first. Others point out that the Greenwald limit has been beaten before. DIII-D in San Diego did it in 2022.418 ASDEX Upgrade in Germany and DIII-D both reached up to about 1.5 times the limit using frozen fuel pellets, but only briefly.108 The Madison Symmetric Torus in Wisconsin went about ten times past the limit, though with a weak magnetic field and plasma too cool for fusion.19

On duration, a 2024 DIII-D run exceeded the limit by about 20 percent for roughly 2.2 seconds. EAST's discharges lasted about six to seven seconds at much higher densities.108 The fair summary: EAST is the first tokamak to confirm the predicted density-free regime, with metal walls and at fusion-relevant conditions. It is not the first machine to go above the Greenwald line.318

Some secondary write-ups should be treated with caution. One describes densities of 50 percent above the limit, lithium wall coatings, neutral beam heating and millisecond magnetic feedback.9 Another credits controlled impurity injection and advanced divertor shapes.5 None of this matches the method described in the paper and by trade outlets, which centers on gas pressure and microwave heating at startup.112 Readers should rely on the paper and the institute's own account.

Limits of the result

The most important caveat is easy to miss. The density-free regime has so far been reached only during the startup phase of the discharge.4 The paper's title points to ohmic start-up. Yan said the next step is to try the same approach in high-confinement (H-mode) operation, the regime a real reactor would use.11

EAST has separately held H-mode plasma for 1,066 seconds above 100 million degrees Celsius.19 France's WEST machine later broke that duration record.14 However, long confinement and high density have been hard to achieve together, and these density runs were a separate campaign.810 Showing both at once in the same discharge is the real test.

Eaglen's view is the most balanced. He said the result means reactors "may not need to be as large or as conservative in density assumptions." He also said it "improves confidence in future reactor designs rather than accelerating timelines."17 Some outlets predict commercial fusion in the 2030s or the first net-energy tokamak by 203059, but the experiment does not support those timelines. Materials that can withstand years of neutron bombardment remain unsolved.15 ITER is not scheduled to begin full fusion operations until 2039.18

AI and the bigger picture

This result came from theory guiding experiment, not from AI. But the two approaches are likely to meet. In 2024, an AI system predicted a major plasma instability up to 300 milliseconds in advance and adjusted the reactor to prevent it.11 AI controllers like that do best with a clear map of where stable operation is possible. PWSO now gives tested boundaries for where density limits come from, which future control systems could be built around. That is a reasonable expectation rather than something the coverage reports.

Overall, the EAST result does not make fusion easy. It does turn one of the field's oldest rules of thumb into a mechanism that can be explained, tested and engineered around. Commentators also note it narrows the gap between tokamaks and stellarators, which have long found high densities easier to reach.18 The next experiments will show whether the density-free regime survives in H-mode. If it does, tokamak designers may be able to drop much of the safety margin they have built around the Greenwald limit.