nnicaPlasma Instabilities
Left alone, a hot plasma tends to lose its shape and start misbehaving all by itself. This tendency to misbehave is what we call plasma instability. Instabilities disrupt confinement and are one of the biggest factors limiting the performance of a fusion reactor. This page starts from the intuition of what an instability is, then works through the representative modes that cause trouble in a tokamak one by one, and finally builds up to the frontier of research on controlling them.
Start with Intuition (High School)
Section titled “Start with Intuition (High School)”Let us first think about what “unstable” means through a familiar example. If you lay a pencil down flat on a desk, it returns to its place or stays put even if you nudge it. That is stable. But if you try to balance the pencil on its sharpened tip, even the slightest tilt grows larger and larger until it topples over. That is unstable.
The key point is that in an unstable state, “a small deviation grows on its own.” Nobody needs to push hard. As long as there is a tiny fluctuation, it becomes the trigger and grows large by itself. Plasma instabilities work the same way: a confined plasma stores a great deal of energy, and that energy tries to move to a lower state, amplifying a small fluctuation all at once.
There is another example that is perfect for grasping the intuition. If you fill a cup with water and gently float oil on top, you get a stable arrangement with the light oil on top and the heavy water on the bottom. But if you try the opposite, placing the heavy water on top of the light oil, the water sinks down into the oil while the oil pushes up through the water to rise, and the boundary becomes a chaotic mess. A state with something heavy on top of something light is unstable. This phenomenon is called the Rayleigh-Taylor instability.
In a plasma, the role of the “heavy water” can be played by high-pressure plasma, and the role of the “light oil” by the magnetic field. When you hold plasma down with a magnetic field, you get an arrangement just like water sitting on top of oil, and the plasma tries to break through the field and burst outward. Many of the instabilities that occur inside a fusion reactor can be thought of as relatives of this Rayleigh-Taylor instability, which makes the overall picture easier to grasp.
In a device called a tokamak, plasma is confined inside a doughnut-shaped vessel by applying a helical magnetic field. If the way this helix is wound, the current flowing in the plasma, and the balance of the plasma pressure are even slightly off, the whole plasma can bend and warp, bulge locally, or tear apart. From the next section on, let us give each of these ways of misbehaving a name and examine them one by one.
Understand the Physics (Undergraduate)
Section titled “Understand the Physics (Undergraduate)”When treating instabilities as physics, the first thing to know is the framework of magnetohydrodynamics (MHD), which treats the plasma as a single fluid and describes its motion together with the magnetic field. The macroscopic instabilities handled by MHD fall broadly into current-driven and pressure-driven types. In the former, the current flowing in the plasma is the energy source; in the latter, the plasma pressure is the energy source.
The representative current-driven instability is the kink mode. A thin column carrying a current can find itself in a situation where, because of the magnetic field it generates, it is “better off bending into a helix” than staying straight. Just as a wire twists to form a kinked loop, the entire plasma column deforms into a helical shape. A close relative is the sausage mode, in which the plasma column pinches in here and there, forming thick and thin spots like a sausage. Where the column narrows, the magnetic field grows stronger and squeezes it further, accelerating the pinch.
The single most important parameter that appears here is the safety factor . The quantity tells how many times a magnetic field line goes around the doughnut the long way (the toroidal direction) for every one time it goes around the short way (the poloidal direction). Written as a formula, in the circular-cross-section approximation,
Here is the radius at which the field line runs, is the major radius of the doughnut, is the toroidal field, and is the poloidal field. The larger is, the more gently the field lines are twisted, and the more stable the plasma becomes. The name “safety” factor comes precisely from the fact that it measures the margin of stability.
For the kink mode, it is known that the external kink mode of a cylindrical plasma is stabilized when the safety factor at the plasma surface satisfies , and this is called the Kruskal-Shafranov limit. In real tokamaks, to also avoid higher-order modes, a value of roughly is used as a rule of thumb for stable operation. Inside the plasma, an internal kink mode appears that resonates on the surface where , and this causes the sawtooth oscillation. In a sawtooth oscillation, the central temperature rises gradually and then drops abruptly, repeating this cycle, and it was named because the waveform resembles the teeth of a saw. The period is roughly 10 to 100 ms.
The representative pressure-driven instability is the ballooning mode. On the outer side of the doughnut (the outboard side of the torus), the way the magnetic field curves is unfavorable for the plasma, in what is called the bad-curvature region. Here the high-pressure plasma tries to bulge out locally, just as a balloon swells where it is weak. As the name suggests, the image is of a balloon. How far a plasma can raise its pressure is measured by a quantity called the normalized beta , and an upper limit known empirically as the Troyon limit (a of roughly 2.8 for the case of a distant wall, and up to about 3 to 3.5 in real tokamaks) is known. Beta is the ratio of the plasma pressure to the magnetic field pressure, and it is an important indicator of confinement efficiency.
There is yet another instability that arises because resistivity comes into play. In an ideal plasma, the field lines and the plasma move together, and the field lines never reconnect. But a real plasma has a slight electrical resistance, and because of it the field lines can break and reconnect in a process called magnetic reconnection. When this reconnection proceeds at a resonant surface, a tearing mode develops there. To “tear” means to rip apart, and it describes how the field lines are torn and reconnected.
Deepen the Theory (Graduate)
Section titled “Deepen the Theory (Graduate)”What appears as a result of the tearing mode is the magnetic island. Magnetic surfaces that were originally nested concentrically reconnect locally around the resonant surface, and viewed in cross-section they form closed field-line structures shaped like islands. Inside a magnetic island, the field lines short-circuit, so the temperature and density quickly become uniform across the island. Since confinement means maintaining a large gradient between the center and the edge, the disappearance of the gradient inside the island means a corresponding drop in confinement performance.
Particularly troublesome is the neoclassical tearing mode (NTM). A tokamak plasma has a bootstrap current that flows spontaneously due to the pressure gradient. Because the pressure gradient vanishes inside a magnetic island, this current is also lost there. Since the missing current acts in a direction that widens the magnetic island further, once a small island (a seed island) forms and exceeds a critical size, the island begins to grow on its own. The NTM is a major factor limiting high-beta operation, with islands sitting on the and surfaces being especially problematic. A sawtooth that has become long-period (a monster sawtooth) can also create the seed island for an NTM.
The kink family also has an important mode in which resistive effects come into play. The resistive wall mode (RWM) is an external kink mode that ought to be stabilized by a surrounding perfectly conducting wall, but because of the wall’s finite resistance it seeps through over time and slowly grows. It is known that the RWM can be stabilized by plasma rotation or by active feedback control.
Among instabilities localized at the edge, the ELM (edge localized mode) is important. In the high-confinement H-mode state, a steep pressure wall called the pedestal forms at the plasma edge. This steep gradient, together with the large bootstrap current flowing there, drives a peeling-ballooning instability that combines both pressure drive and current drive, periodically expelling energy from the pedestal. This is the ELM. In particular, the Type I ELM occurs at a frequency of roughly 10 to 100 Hz and releases several to over ten percent of the pedestal energy all at once. The released energy concentrates as an instantaneous heat load on the divertor, which in large devices becomes a serious problem that can lead to damage of the wall material.
The most violent phenomenon is the disruption. This is a major event in which a plasma discharge suddenly collapses and terminates, and rather than being a single instability, it is a catastrophe in which several instabilities occur in a chain. When you exceed the Greenwald limit from raising the density too high, or the beta limit from raising beta too high, large magnetic islands grow and the magnetic surfaces break, leading to a disruption. The process proceeds in two stages. First, in the thermal quench (roughly 1 to 10 ms), the stored thermal energy is lost all at once, and then in the following current quench (roughly 10 to 100 ms), the plasma current drops rapidly. The sudden change in current generates large electromagnetic forces on the vessel, and halo currents and runaway electrons inflict serious damage on the equipment. How to avoid disruptions, and how to mitigate the damage when they cannot be avoided, is one of the central challenges of reactor design.
On the other hand, what gradually eats away at confinement is the microinstability. This is not a large deformation that would destroy the whole device, but a fine fluctuation occurring on the scale of ions and electrons within the plasma, which gives rise to turbulence. Turbulence carries heat and particles out across the gradient and is the true cause of anomalous transport, which far exceeds the predictions of classical theory. Representative examples include the ion temperature gradient mode (ITG), driven by the ion temperature gradient; the trapped electron mode (TEM), involving trapped electrons; and the electron-scale electron temperature gradient mode (ETG). Unlike MHD instabilities, these are described within the framework of kinetic theory, which treats the plasma down to the velocity distribution of the particles, and large-scale simulations based on gyrokinetics have become the mainstay of theoretical research. Their relationship to transport is treated in more detail on the transport page.
Research Frontier (PhD)
Section titled “Research Frontier (PhD)”In current research, the center of gravity is shifting from merely “suppressing instabilities after they occur” toward “predicting and avoiding them before they occur.”
For NTM control, electron cyclotron current drive (ECCD) has been established as a standard means. By targeting the O-point of the magnetic island (the center of the island) and driving current locally to make up for the missing bootstrap current, it stops the island from growing. The focus of research is the real-time control needed to keep the beam accurately aimed at the target location.
For ELM control, resonant magnetic perturbation (RMP) is regarded as promising. This is an approach in which a small helical magnetic field is added from external coils to slightly perturb the edge magnetic surfaces, changing a large Type I ELM into small, gentle emissions, or suppressing it entirely. ELM pacing by pellet injection (a technique that artificially triggers small ELMs to break up the energy release into small portions) is being studied in parallel. In large devices such as ITER, the heat load of uncontrolled Type I ELMs is unacceptable, so complete suppression or sufficient mitigation of ELMs has become a requirement that must be achieved.
Disruption prediction is an area where machine learning has been especially actively applied in recent years. Research is progressing on detecting the precursors of a disruption in real time from many diagnostic signals and moving to mitigation actions while there is still time. As mitigation means, techniques are being developed to inject impurities all at once by massive gas injection or shattered pellet injection, radiating away the stored energy gently while suppressing the generation of runaway electrons. The mechanism of runaway electron generation itself and its suppression are also active research topics.
In the field of microinstabilities and turbulence, validation efforts continue that compare gyrokinetic simulations against experimental measurements, aiming to predict turbulent transport from first principles. The self-regulating mechanism by which zonal flows generated by the turbulence itself weaken the turbulence, and the formation mechanism of transport barriers where confinement suddenly improves, remain central questions under study. In the literature, keywords such as ITG, TEM, ETG, zonal flow, gyrokinetics, and ExB shear appear frequently.