Fusion Density Limit Broken at EAST as China Readies BEST Tokamak

A ceiling in fusion physics gives way

China's Experimental Advanced Superconducting Tokamak (EAST), the device in Hefei usually called the country's "artificial sun," has run stable plasma at densities well above a limit that has held back tokamak research for decades. The team reported entering a "density-free regime" that theory had predicted, in which the plasma stays stable even as its density climbs past the usual empirical ceiling.1 The results were published in Science Advances on January 1, 2026. 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 work.10

The result is drawing new attention this month because China is also building EAST's successor. On October 1, Chinese state media reported that the project campus for the Burning Plasma Experimental Superconducting Tokamak (BEST) had been handed over and put into use, and that the main machine had entered a critical assembly phase.1213 Read together, the two stories describe a program that is solving physics problems and building hardware at the same time. That pairing deserves more attention than any single record.

What the Greenwald limit is, and how EAST got past it

In deuterium-tritium fusion, fuel works best at about 150 million kelvin. At that temperature, fusion power rises with the square of plasma density.1 Packing in more fuel is one of the strongest levers available. Tokamaks, however, have long run into an upper density bound. Above it, the plasma tends to become unstable and confinement collapses.4 That bound is the Greenwald limit, named after American physicist Martin Greenwald, who described it in 1988.9

The EAST team did not use bigger magnets or more brute force. Its approach came from a theory called plasma-wall self-organization (PWSO), first proposed by D.F. Escande and colleagues at France's CNRS and Aix-Marseille University. PWSO says a density-free regime can open up when the plasma and the reactor's metal walls reach a delicate balance, with physical sputtering as the dominant process.10 In practice, the researchers set the starting fuel gas pressure carefully and used electron cyclotron resonance heating (ECRH) while each discharge was starting up. This cut impurity buildup and energy losses from the very first moments.1 Controlling conditions at the divertor target plate reduced sputtering of tungsten from the walls, and that let the plasma move past the limit into the new regime.2

The reported results are concrete. EAST reached line-averaged electron densities between 1.3 and 1.65 times the Greenwald limit, and the team said the data agreed closely with PWSO predictions.2 Nuclear Engineering International notes that EAST normally operates at 0.8 to 1.0 on the same scale.3 Popular Mechanics led with the top figure, describing densities 65 percent beyond the limit.9

Theory first: the scientific-computing angle

This result rests on theory and modeling, not on a new machine. The Institute of Plasma Physics team built a theoretical model of how the plasma edge and the wall organize themselves. The model identified radiation at the plasma boundary as the trigger for the density limit, and the experiment was designed to test it.2 The theory also predicts two operating basins: a normal one with a hard density ceiling, and a density-free one reached through particular start-up conditions.3 Power-dependent scaling laws derived from the theory had already been supported by data from several tokamaks.3

The experiment came in stages. Before EAST, researchers tested parts of PWSO on the J-TEXT tokamak at Huazhong University. J-TEXT has carbon walls, where chemical sputtering matters. Those runs showed that more ECRH power or higher prefill pressure raised the density limit, but the plasma stayed within the limited basin.3 EAST's tungsten walls, where physical sputtering dominates, gave the conditions the theory said were needed to reach the density-free basin.3 One account of the method describes two series of runs. In one, prefill pressure changed while ECRH power stayed at about 600 kW. In the other, heating power changed while pressure stayed fixed. In both, the density limit flattened out because of the available heating power, not because of a hard physical ceiling.8 If that holds, more heating could push density higher still.

Measurement was just as important. World Nuclear News credits EAST's precise diagnostics for density, temperature, radiation and impurities, along with its efficient ECRH system, as key technical support for the work.2 The broader point is that fusion progress increasingly depends on predictive models that tell experimenters where to look. ScienceDaily's related coverage recalls a 2024 result in which an AI system forecast a major plasma instability up to 300 milliseconds ahead and adjusted the reactor to avoid it.1 The EAST work is not an AI story. It belongs to the same trend, though: computation and theory are guiding operating strategy instead of trailing behind it.

Where the coverage agrees and where it drifts

On the main facts, coverage is consistent. The Chinese Academy of Sciences, World Nuclear News, Nuclear Engineering International and Popular Mechanics all describe the same method: ECRH-assisted ohmic start-up, managing plasma-wall interactions, and testing PWSO.10239 Outside physicists quoted by a UK student newspaper were positive. One called the results very promising and worth exploring on other devices. Another argued they show tokamak power can rise without spending billions on stronger magnets.7

Some reporting goes well past the evidence. One widely shared write-up says the team used neutral beam injection and lithium wall coatings, frames the problem around edge-localized modes, and suggests the result could speed commercial fusion by a decade or allow fusion electricity by 2027.5 None of these claims appears in the primary accounts. Those accounts describe gas prefill and ECRH during start-up, with tungsten sputtering as the key factor.210 Readers should treat the bolder claims as editorializing.

Hackaday gives the most useful corrective. It argues that the main news is confirmation of PWSO theory, and it notes that density limits interact with the transition out of high-confinement H-mode.6 This is the real caveat. The team's own next step is to apply the method during high-confinement operation and try to reach the density-free regime under high-performance conditions.10 Until that works, the result is a validated mechanism, not a demonstration at power-plant-relevant conditions. The researchers do say the approach is relevant to stellarator start-up and to next-generation burning-plasma devices.24

From physics to steel: BEST

That makes the hardware timeline important. BEST's main machine is now in what state media call "four-layer nested assembly." Working outward, it consists of a vacuum vessel, a thermal shield, toroidal field magnets and internal precision components. More than half of the assembly tasks are reportedly done.12 Construction is due to finish by the end of 2027, with deuterium-tritium power-generation demonstrations planned around 2030.13 Argentine coverage citing Chinese Academy of Sciences figures puts BEST's target at 20 to 200 megawatts of fusion power, with net energy gain as the main goal. It also notes that neutron damage to materials, tritium handling and continuous heat conversion remain unsolved.15

The supporting hardware is advancing as well. The Institute of Plasma Physics reported that a 582-tonne superconducting magnet and a high-temperature superconducting central solenoid coil both passed testing. The coil held a steady 60 kiloamperes and stored 6.03 megajoules. Six years of work on the two components produced 47 patents and 14 technical standards. This builds on EAST's January 2025 record of 1,066 seconds of high-confinement plasma.1218

The reading

The density result is real and, by fusion standards, unusually clean. A theory made a prediction, a carbon-walled machine tested part of it, and a tungsten-walled machine confirmed the rest.3 It does not show that fusion power is imminent. It shows that one long-standing operating limit depends partly on how a discharge is started, which means engineers can design around it. Density feeds directly into fusion output, so that matters.1

The more important story may be strategic. One commentator argues that the key question in fusion may be less about who first produces net energy and more about who can manufacture superconductors, magnets and vacuum systems at scale. On that measure, Hefei's growing cluster of suppliers is a real advantage.18 The same analysis warns that a 2030 "electricity" demonstration could be limited and experimental, not continuous power to the grid.18 That seems the right way to read China's schedule. Its physics is credible, its timelines are ambitious, and BEST, now about halfway built, will be the real test of both.