A 1,100 plasma campaign on a spherical tokamak in Oxfordshire hit the machine’s highest pressure without a single disruption, where the real result was 4 separate ways to tame the plasma edge

A vacuum vessel in Oxfordshire shaped less like a doughnut than a cored apple.

Inside it the machine made and destroyed more than 1,100 separate plasmas across one campaign.

It reached the highest pressure it has ever held, and held it.

It is also what makes the engineering brutal, because the central column has almost no room for shielding and sits in the worst neutron environment in the machine.

Nothing collapsed. No shot ended in the sudden loss of control that engineers dread.

The number worth keeping is not the pressure.

It is four.

Why the edge of a plasma is the part that wrecks the machine

A confined plasma can sit in a high performance state where a steep pressure barrier forms just inside the wall, and that barrier is what makes good confinement possible.

Barriers like that do not hold quietly. Pressure builds against the limit until the edge lets go in a burst, dumping energy outward in a fraction of a second.

Those bursts are survivable in a small research device. Scaled to a power plant they are not, because each one strips material from the surfaces facing the plasma.

So the field has a specific problem rather than a vague one. Keep the confinement that the barrier provides and get rid of the bursts that come with it.

Every route to that is a compromise, and the compromise is what has to be chosen before a reactor is designed rather than after.

Confinement needs the barrier. The barrier throws the bursts.

What the campaign actually produced

The machine got to four separate operating regimes in which the edge stayed quiet, all within the same vessel and the same run of experiments.

Two of them are alternative high confinement states where the edge sits naturally below the threshold for bursting rather than being forced there.

A third leaks energy out steadily at the edge instead of in bursts, turning a violent intermittent load into a continuous mild one.

The fourth uses small applied magnetic fields to deliberately roughen the edge, which suppresses the bursts at a cost in confinement.

Together with the pressure record and the absence of disruptions, that is the substance of the result.

Four routes, one vessel, one campaign.

Why a run with no disruptions counts separately

Bursts at the edge and a disruption are different failures and they are often blurred together.

A burst throws a slice of energy at the wall and the plasma carries on. A disruption is the plasma losing control of itself entirely and dumping everything it holds in milliseconds.

That second event drives enormous currents through the vessel structure, and the magnetic forces that follow can shift multi ton components.

Running a full campaign at record pressure without one is therefore not a footnote. Pressure is exactly the quantity that pushes a plasma toward the stability limits where disruptions live.

Going higher than the machine had ever gone and staying controlled is the part that would have been lost in a single sentence about pressure.

Bursts damage surfaces. Disruptions move hardware.

Why doing all four in one machine is the point

Any one of these regimes has been shown somewhere before. Comparing results between different machines is where the difficulty starts.

Two tokamaks differ in shape, in wall material, in heating method and in size, and every one of those changes how an edge behaves.

Demonstrating all four in a single device with a single wall and a single diagnostic set removes most of that ambiguity in one go.

What a reactor designer gets out of it is a menu rather than a bet. If one route fails at reactor scale, three others are documented under directly comparable conditions.

Getting from a physics result to a licensed plant is a separate mountain, which is why the industry keeps producing licensing papers ahead of hardware.

The four regimes, the pressure record and the absence of disruptions are reported by a nuclear service.

One machine gives comparable answers. Four machines give arguments.

What this is and what it is emphatically not

This device is not chasing energy gain and was never built to. It is a physics machine, and the useful output is measurement rather than power.

The spherical shape is the reason it is worth running. Squashing the plasma toward a sphere improves confinement for a given magnetic field, which is the whole argument for compact fusion designs.

It is also what makes the engineering brutal, because the central column has almost no room for shielding and sits in the worst neutron environment in the machine.

Nothing in this campaign addresses that. A quiet edge does not help a column that has to survive years of bombardment in a space too narrow to protect.

Control is the other open front, where machine learning is being pointed at instabilities through work like a plasma guardian.

The campaign framing and the record plasma are described by a government notice.

The plasma edge behaved four different ways on request, and the column in the middle is still the problem.