Plasma Physics

Tokamak Confinement

Toroidal field + plasma current → twisted helical field lines confine 100-million-degree fuel

A tokamak (Russian acronym, 1950s) is a toroidal magnetic confinement device for fusion plasma. Confinement is achieved by combining: (1) toroidal field B_T from external coils (3-12 T at the magnetic axis in modern designs), (2) poloidal field B_P induced by plasma current I_p (driven by transformer action with the plasma as secondary), giving twisted helical field lines that close on flux surfaces — preventing the radial drift charged particles would otherwise experience in a pure toroidal field. Q factor = fusion power / heating power; breakeven Q=1, ignition Q=∞. JET (1997): Q=0.67 with 16 MW fusion. JT-60SA (2023, Japan): plasma operations begin. ITER (under construction in France, first plasma 2034 expected): Q=10 target, 500 MW fusion from 50 MW input. SPARC (CFS, MIT spinout) and Tokamak Energy (UK) pursue compact high-field tokamaks using REBCO superconductors. Successor concept: stellarator (Wendelstein 7-X 2015+) — twisted coils, no plasma current.

  • InventorsSoviet 1950s
  • FieldToroidal + poloidal
  • QP_fusion / P_heat
  • BreakevenQ=1
  • JET 1997Q=0.67
  • ITER goalQ=10

Interactive visualization

Press play, or step through manually. The visualization is yours to drive — try it before reading on.

Open visualization fullscreen ↗

Watch the 60-second explainer

A condensed visual walkthrough — narrated, captioned, under a minute.

Why tokamaks matter

  • Fusion energy goal. Demonstrated D-T fusion produces 17.6 MeV per reaction with no long-lived high-level radioactive waste, no greenhouse gases, and fuel from seawater plus lithium — the long-term clean baseload story. ITER's Q=10 target would be the first sustained net-energy plasma, the engineering proof that fusion can run a power plant.
  • Plasma physics frontier. A 100-million-degree plasma is the most extreme matter humans manipulate routinely on Earth. Studying it has driven turbulence theory, magnetohydrodynamic stability theory, atomic physics in extreme conditions, and computational plasma simulation — with spillover into astrophysics (stellar interiors, accretion disks).
  • ITER as megaproject. 35 nations sharing 25 billion dollars in capital, 1 million components manufactured worldwide, 23,000 tonnes total mass — among the largest engineering collaborations in history. Even before producing its first plasma, ITER has driven advances in cryogenics, superconducting cable, vacuum technology, and remote handling.
  • SPARC and the high-field path. CFS spun out of MIT in 2018 with a thesis: REBCO high-temperature superconductors allow compact 12 T tokamaks, and confinement scales as B^4. Lower volume but same fusion gain at one-tenth the cost. SPARC first plasma is expected mid-2020s, demonstration plant ARC to follow. Other private players (Tokamak Energy, Helion, TAE) chase parallel concepts.
  • Stellarators as backup. Wendelstein 7-X (Greifswald, Germany) achieves confinement with no plasma current — relying entirely on shaped 3D coils. No current means no disruptions and steady-state operation indefinitely, but the shaped coils are ferociously expensive to manufacture. W7-X has demonstrated confinement-time scaling competitive with tokamaks since 2015.
  • Diagnostics technology. Tokamaks need to measure plasma density, temperature, current, and radiation profiles in real time at millimeter resolution. Thomson scattering, charge-exchange recombination spectroscopy, electron cyclotron emission, motional Stark effect — every diagnostic that started in fusion now appears in semiconductor processing, space science, and industrial plasmas.
  • Materials science. Plasma-facing components survive 14 MeV neutron flux, 100 MW/m^2 transient heat loads, and tritium retention. Tungsten armor, beryllium first walls, reduced-activation steel structures, and ITER-grade Nb3Sn superconductors all originated as fusion-driven research and now feed back into nuclear, aerospace, and high-magnetic-field applications.

Common misconceptions

  • Fusion is solved. Net energy gain has been demonstrated only briefly: NIF in 2022 (target gain 1.5, but laser efficiency drops the wall-plug Q far below 1) and JET pulses (Q=0.67). No experiment has produced sustained net electricity from fusion. ITER aims to demonstrate Q=10 for hundreds of seconds; commercial power plants are decades and many engineering breakthroughs away.
  • Tokamak is the only design. Stellarators (Wendelstein 7-X, the Helically Symmetric eXperiment) confine without plasma current. Z-pinches like Sandia's MagLIF compress fuel to fusion conditions in nanoseconds. Inertial confinement (NIF, Lockheed) uses lasers or particle beams. Field-reversed configurations (TAE Energy) and magneto-inertial schemes (Helion) explore different physics regimes entirely.
  • Magnetic field stops nuclei. Magnetic fields confine charged plasma — the deuterium-tritium fuel and the alpha particles produced by fusion. The fusion neutrons (14 MeV, neutral) escape unimpeded and deposit their energy in the surrounding blanket, which is also where lithium absorbs them to breed tritium. Confining neutrons is impossible; the design instead intercepts them productively.
  • Plasma is a hot gas. A fusion plasma is fully ionized, with electron and ion populations weakly coupled, transport governed by collective MHD and turbulence rather than neutral-gas hydrodynamics. Pressure can balance magnetic-field tension, kink and ballooning instabilities can disrupt confinement on milliseconds, and sheared flows can heal turbulent transport. Plasma physics is its own discipline.
  • Bigger always means better. The historical scaling rule is volume (a^3) for confinement time, so big helps. But high-field compact designs scale as B^4 — a 12 T REBCO tokamak the size of JET would outperform ITER on paper. Plasma-facing component damage scales unfavorably with size; cost scales as a^2 to a^3. The optimal is hotly contested.
  • Tokamaks are clean. D-T fusion produces neutrons that activate the surrounding structural materials, and tritium is radioactive (12.3 year half-life) and mobile through hot metals. A fusion reactor's first wall reaches dozens of dpa damage over its lifetime and its blanket components become low-level waste. Cleaner than fission by orders of magnitude, but not absolutely clean.

From experiment to power plant

The path from an ITER-class research tokamak to a commercial power plant requires three demonstrations beyond Q=10. First, tritium self-sufficiency: lithium-bearing blankets must absorb fusion neutrons and breed more tritium than the plasma consumes — a tritium breeding ratio above 1.05 with current-day technology. Second, materials with multi-year lifetimes under 14 MeV neutron flux: reduced-activation ferritic steels (EUROFER, F82H) are the leading candidates but need component-level testing in a fusion neutron source like IFMIF-DONES. Third, steady-state operation: pulsed transformer-driven plasmas waste capital, so non-inductive current drive (lower hybrid waves, electron cyclotron, neutral beams, or stellarator topology) must keep plasma current flowing indefinitely. ITER addresses none of these directly; demonstration plants like EU-DEMO (target 2050s) aim to combine all three.

Frequently asked questions

Why does a pure toroidal field fail (drift)?

A pure toroidal field B_T varies as 1/R (stronger on the inner side of the torus, weaker on the outer side) because of the geometry of current loops. This gradient produces a charge-dependent drift: ions drift upward, electrons downward (or vice versa, depending on sign conventions). The resulting vertical charge separation creates an electric field, and the resulting E cross B drift pushes the entire plasma radially outward into the wall in milliseconds. Adding a poloidal field component twists the field lines so that particles average over the full poloidal circuit and the drifts cancel.

What is the safety factor q (rotational transform)?

The safety factor q is the number of toroidal turns a magnetic field line makes per single poloidal turn around the axis. Equivalently, q = (r/R)(B_T/B_P) for a circular cross-section. The name comes from MHD stability: configurations with q above 1 throughout the plasma resist the dangerous m=1 internal kink instability. Typical tokamaks operate with q at the magnetic axis around 1 and at the edge around 3 to 4. Field lines on rational q surfaces (q = m/n) close on themselves after m toroidal turns, and these resonant surfaces are where MHD instabilities like tearing modes first appear.

What is the H-mode (high-confinement) regime?

Discovered on ASDEX in 1982, H-mode is a state where heating power above a threshold creates a steep edge transport barrier — energy and particle confinement times double abruptly. The mechanism: sheared E cross B flow at the edge tears apart turbulent eddies, suppressing transport across the last few centimeters. ITER's design relies on H-mode to reach Q=10. Downsides: Edge-Localized Modes (ELMs), periodic bursts of energy that erode plasma-facing components if not mitigated by resonant magnetic perturbations or pellet pacing.

Why is plasma current necessary?

Plasma current I_p creates the poloidal magnetic field that combines with the toroidal field to make helical field lines. Without I_p, the toroidal field alone cannot confine the plasma (drifts push it into the wall). I_p also heats the plasma ohmically (P = I^2*R) at low temperature — though resistivity falls as T to the -3/2 power, so ohmic heating becomes ineffective above 1 keV and additional auxiliary heating (neutral beams, RF waves) is needed to reach fusion temperatures.

What is the Greenwald density limit?

The Greenwald limit is an empirical maximum line-averaged plasma density n_GW (in units of 10^20 m^-3) equal to I_p (in MA) divided by pi*a^2 (a = minor radius in m). Above this density, the plasma becomes unstable: confinement degrades, edge cooling triggers radiative collapse, and a disruption — sudden loss of confinement releasing all stored energy in milliseconds — often follows. The limit is a problem because fusion power scales as n^2; designs need either higher current (limited by stability) or operation at fractions of n_GW.

How does ITER differ from JET (size, materials)?

JET (1983 to 2023, Culham UK) had a major radius of 3 m, plasma volume 80 m^3, copper toroidal coils, and a maximum 4 T field. ITER (under construction at Cadarache, France) has major radius 6.2 m, plasma volume 840 m^3 (10x larger), niobium-tin (Nb3Sn) low-temperature superconducting coils producing 5.3 T on the axis, and tungsten plasma-facing components instead of JET's beryllium. The size difference is critical: confinement time scales roughly as a^2 to a^3, so ITER's larger volume is what allows it to reach Q=10 where JET could only reach 0.67.