For decades, every tokamak hit the same wall. Increase plasma density beyond a certain point — roughly proportional to plasma current divided by cross-sectional area — and disruptions follow. The plasma becomes unstable, dumps its energy into the containment walls, and the experiment ends. This is the Greenwald density limit, named after Martin Greenwald who formalized the empirical scaling in 1988.
The limit appeared across every major tokamak: JET, DIII-D, ASDEX Upgrade, EAST. Different sizes, different magnetic configurations, different fueling methods — same ceiling. The universality of the limit made it look like physics: something fundamental about how magnetically confined plasma behaves at high density.
Ping Zhu and Ning Yan's team at China's EAST tokamak demonstrated that the limit is not physics. It is procedure.
Their approach, grounded in Plasma-Wall Self Organization (PWSO) theory, targets the first moments of plasma operation. By controlling initial fuel gas pressure and applying electron cyclotron resonance heating during startup, they optimized the plasma-wall interaction from the beginning of the discharge. This minimized impurity accumulation — the real killer at high density — and reduced energy losses through the boundary layer.
The plasma went well past the Greenwald limit. It stayed stable.
The insight is that the density limit was always a secondary effect. The primary mechanism was impurity buildup from uncontrolled wall interactions during standard startup. Impurities radiate energy out of the plasma core, cooling the edge faster than it can be reheated, which drives instabilities. At low densities, this radiation is manageable. At high densities, the uncontrolled impurity influx overwhelms the energy balance. The limit tracks density only because density amplifies the radiative losses from impurities that were already there.
Remove the impurities by managing the wall interaction from the start, and density is no longer the binding constraint.
What makes this result structurally interesting is how long the limit persisted as a “law.” Greenwald's scaling held for nearly forty years across every machine. It appeared in reactor design studies, informed ITER's operating parameters, and constrained theoretical projections of fusion power output. The limit shaped what engineers thought was possible. And it was never about the plasma — it was about how the plasma was started.
The procedure created the evidence for its own inevitability. Every tokamak started the same way, hit the same impurity-driven ceiling, and confirmed the same scaling. The limit was real, reproducible, and universal — for the startup procedures in use. Change the procedure, and the universality disappears.