Collective disturbances that can grow exponentially and destroy plasma confinement — the central challenge of fusion physics.
A plasma instability is any perturbation in a confined plasma that, once triggered, feeds on the free energy stored in pressure gradients, current distributions, or velocity shears to grow rather than decay. Because fusion plasmas must be held at temperatures exceeding 100 million kelvin, even a small instability that breaches confinement can terminate a discharge in microseconds. Understanding, predicting, and suppressing instabilities has therefore defined the research agenda of magnetic-confinement fusion for more than seven decades.[1]
In a neutral gas, random molecular collisions restore equilibrium after a perturbation. In a plasma, however, long-range electromagnetic interactions couple enormous numbers of particles collectively. A local displacement of plasma can alter currents and fields in a way that reinforces the original displacement — a positive-feedback loop. The growth rate of an instability is characterised by an e-folding time, often on the order of the Alfvén transit time (microseconds) for fast MHD modes, or the ion-transit time (milliseconds) for slower drift-wave turbulence.[2]
Instabilities are broadly grouped by the framework used to describe them:
Ideal MHD instabilities — driven by pressure or current gradients without breaking magnetic field lines. Examples include the kink mode, the interchange (or flute) mode, and the ballooning mode.
Resistive instabilities — require finite plasma resistivity to allow magnetic reconnection. The tearing mode and the resistive wall mode fall into this class.
Kinetic (microinstabilities) — arise from velocity-space features in the particle distribution function, such as trapped-electron modes and ion-temperature-gradient (ITG) modes that drive turbulent transport.
Energetic-particle instabilities — excited by fast ions from neutral-beam injection or fusion-born alpha particles, including toroidal Alfvén eigenmodes (TAEs).[3]
Uncontrolled instabilities manifest as disruptions in tokamaks — sudden losses of thermal and magnetic energy that can deposit megajoules onto plasma-facing components in milliseconds. In ITER, disruption mitigation is a dedicated engineering system. Stellarators, by contrast, avoid current-driven instabilities at the cost of geometric complexity. Inertial-confinement designs must contend with Rayleigh–Taylor instabilities during implosion.[1]
Modern experiments employ a suite of active and passive techniques: plasma shaping to access second-stability regimes, resonant magnetic perturbation (RMP) coils to suppress edge-localised modes (ELMs), electron-cyclotron current drive to stabilise neoclassical tearing modes, and real-time feedback control. Machine-learning controllers trained on disruption databases are now being tested to extend the boundary of stable operation.[2]