A tokamak is a device that confines an extremely hot plasma within a toroidal, or doughnut-shaped, chamber by combining externally generated magnetic fields with a magnetic field produced by an electric current flowing through the plasma. It is the most extensively developed configuration for magnetic-confinement fusion, a field concerned with sustaining the conditions under which light atomic nuclei combine and release energy. The name derives from a Russian abbreviation commonly rendered as “toroidal chamber with magnetic coils.”
Tokamaks are important both as experimental instruments for studying high-temperature matter and as candidates for future electricity-generating fusion reactors. Their development has established many of the operating regimes, heating methods, superconducting technologies, and plasma-control techniques used in fusion research. A successful power-producing tokamak must do more than generate fusion reactions: it must sustain a sufficiently productive plasma, remove concentrated exhaust heat, withstand neutron irradiation, reproduce its fuel supply, and convert the released energy into useful power.
Historical development
The tokamak concept emerged in the Soviet Union in the early 1950s, principally through the work of Igor Tamm and Andrei Sakharov. Their proposals combined a strong toroidal magnetic field with a current-carrying plasma, seeking to overcome the rapid particle losses that affected simpler confinement arrangements. Soviet experimental development proceeded under Lev Artsimovich at the Kurchatov Institute. Early machines established the basic configuration, but reliable measurements of plasma temperature and confinement were essential before its advantages could be accepted internationally.
At the 1968 International Atomic Energy Agency conference in Novosibirsk, Soviet researchers reported unexpectedly high electron temperatures and improved confinement in the T-3 tokamak. These results attracted considerable attention because other magnetic configurations had struggled to reach comparable performance. In 1969, a team from the Culham laboratory in the United Kingdom used laser-based Thomson scattering to make independent temperature measurements on T-3. The measurements confirmed electron temperatures of approximately a kiloelectronvolt, equivalent to more than ten million kelvin, and helped establish the tokamak as a leading approach.
During the 1970s and 1980s, laboratories built larger machines and developed auxiliary heating, shaped plasma cross-sections, and more sophisticated magnetic control. The Princeton Large Torus demonstrated substantial ion heating with neutral beams in 1978. In 1982, experiments on ASDEX in West Germany discovered the high-confinement mode, or H-mode, which became central to later reactor designs. Large facilities—including the Joint European Torus, the Tokamak Fusion Test Reactor, and JT-60—were designed to investigate plasmas closer to reactor conditions.
Experiments with deuterium and tritium subsequently demonstrated significant fusion production. JET first used a deuterium–tritium mixture in 1991, while TFTR began its deuterium–tritium campaign in 1993. JET produced about 16 megawatts of fusion power in 1997, with a plasma fusion gain of approximately 0.67. In its final deuterium–tritium campaign in 2023, JET produced 69 megajoules of fusion energy over approximately five seconds; the result was announced in 2024. These achievements demonstrated increasingly capable fuel handling and plasma operation, but did not constitute net electricity production.
The subsequent development path has emphasized superconducting magnets, longer pulses, integrated plasma control, and reactor-relevant materials. The international ITER project in southern France represents the largest coordinated effort to study a tokamak plasma in which heating by fusion products becomes a dominant part of the energy balance.
Magnetic geometry and plasma equilibrium
Charged particles moving through a magnetic field spiral around its field lines. This motion makes it possible to restrict their movement across the field without placing a material container directly against the hot plasma. A purely toroidal field, however, is insufficient for good confinement. Because the field is stronger on the inside of the torus than on the outside, particle drifts tend to separate charges and produce losses. A tokamak adds a poloidal field, directed around the short circumference of the torus, so that the combined field lines wind helically around the chamber.
The toroidal field is generated mainly by coils encircling the vacuum vessel. The poloidal field arises partly from the toroidal plasma current and partly from external coils that establish position and shape. Together these fields form nested magnetic surfaces. In an ideal axisymmetric equilibrium, particles and heat are predominantly confined within these surfaces, although collisions, turbulence, and electromagnetic fluctuations cause transport across them.
Two basic dimensions are the major radius , measured from the machine’s central axis to the plasma’s magnetic axis, and the minor radius , characterizing the plasma’s cross-sectional size. Their ratio, , is the aspect ratio. Plasma cross-sections are often elongated vertically and made approximately D-shaped. Elongation and triangularity can improve performance, but stronger shaping also increases demands on stability and position control.
A central measure of magnetic winding is the safety factor . For a simple circular, large-aspect-ratio plasma,
where and are the toroidal and poloidal magnetic-field components. Roughly, describes the number of toroidal circuits a field line makes per poloidal circuit. Its radial profile matters because surfaces with rational values of can support resonant magnetic perturbations and instabilities.
Large-scale equilibrium is described using magnetohydrodynamics, in which plasma pressure is balanced by magnetic forces:
For axisymmetric systems, this balance leads to the Grad–Shafranov equation, a principal tool for calculating plasma shape and reconstructing experimental equilibria. Another important parameter is beta,
which compares plasma pressure with magnetic pressure. Higher beta can increase fusion output for a given magnetic field, but the usable value is constrained by instabilities, current distribution, and conducting structures surrounding the plasma.
Fusion reactions and measures of performance
The most extensively studied fuel for an initial generation of tokamak power plants is a mixture of the hydrogen isotopes deuterium and tritium. Their reaction is
Approximately 3.5 megaelectronvolts are carried by the helium nucleus, or alpha particle, and 14.1 megaelectronvolts by the neutron. Because alpha particles are electrically charged, a sufficiently well-confined plasma can retain much of their energy as heat. Neutrons are uncharged and escape magnetic confinement, depositing their energy in surrounding structures and blankets.
Fusion performance depends jointly on temperature, fuel density, and energy confinement. Deuterium–tritium plasmas relevant to reactors typically require ion temperatures of order 10–20 kiloelectronvolts, corresponding to roughly 100–200 million kelvin. Such temperatures do not imply that the entire machine has a comparable heat content: a tokamak plasma is extremely dilute, with representative densities around particles per cubic metre. Nevertheless, the total stored plasma energy and concentrated energy fluxes can be substantial.
The energy-confinement time is conventionally written
where is the stored plasma thermal energy and is the relevant power loss. The Lawson criterion expresses the requirement that density and confinement be sufficient at a given temperature for fusion heating to overcome losses. The related fusion triple product, , is widely used to compare performance. Its precise target depends on fuel composition, temperature, impurities, and whether the objective is scientific breakeven, ignition, or a practical power balance.
Plasma fusion gain is commonly defined as
A value of means that fusion power equals the external heating power delivered to the plasma under the stated accounting convention. It does not mean that the facility produces net electricity. Electrical losses in heating systems, magnet refrigeration, pumps, and other equipment remain outside this ratio. Ignition denotes a condition in which fusion-product heating can maintain the plasma without external heating.
ITER’s reference objective is 500 megawatts of fusion power from 50 megawatts of externally supplied plasma heating, corresponding to , for pulses lasting several hundred seconds. This is a design goal rather than an achieved operating result. ITER is not intended to generate electricity. Reactor studies must additionally calculate thermal conversion efficiency and recirculating electrical power, making plant-level energy gain a different and more demanding measure.
Plasma heating, current drive, and fuel supply
A tokamak discharge usually begins by introducing a small quantity of gas into an evacuated vessel and applying an electric field to ionize it. In conventional operation, the central solenoid acts as the primary winding of a transformer, while the plasma acts as a single-turn secondary. The induced plasma current provides both part of the confining magnetic field and resistive, or ohmic, heating.
Ohmic heating becomes progressively less effective as the plasma becomes hotter because electrical resistivity decreases strongly with electron temperature. It is therefore useful for starting a discharge but insufficient by itself for reactor-grade conditions. Auxiliary heating systems transfer additional energy through energetic particles or electromagnetic waves. Their deposition profiles also influence pressure, current, rotation, and stability, so they are tools for controlling the plasma rather than simply increasing its temperature.
In neutral-beam injection, ions are accelerated outside the tokamak and converted into energetic neutral atoms. Neutral particles can cross the confining magnetic field; after entering the plasma, collisions and ionization produce fast ions that transfer energy to the background particles. Beam injection can also supply toroidal momentum and drive current. High-energy systems use negative ions as an intermediate stage because they can be neutralized more efficiently than positive ions at energies approaching a megaelectronvolt.
Radio-frequency heating uses waves selected to interact with particular plasma motions. Ion-cyclotron systems couple energy through resonances associated with ion motion, while electron-cyclotron systems use millimetre waves that can deposit power in a localized region. Lower-hybrid waves are particularly important for efficient non-inductive current drive in suitable operating regimes. Wave propagation, absorption, and accessibility depend on density, magnetic field, frequency, and the particle velocity distribution.
The transformer mechanism has a finite magnetic-flux swing, which limits the duration of inductively driven operation. Long-pulse or steady-state tokamaks therefore require non-inductive current from beams, radio-frequency waves, and the bootstrap current. The bootstrap current arises from pressure gradients and the differing motions of trapped and passing particles in toroidal geometry. Advanced operating scenarios seek to exploit it without producing unstable current profiles.
Fuel can be supplied by gas injection or by injecting frozen pellets, commonly made from hydrogen isotopes. Pellets travel into the plasma and ablate, depositing particles farther inside than edge gas injection usually permits. Fueling must be coordinated with pumping, density control, and removal of helium produced by fusion. A reactor fuel cycle must recover unburned deuterium and tritium from exhaust streams, separate impurities, and return usable fuel to storage and injection systems.
Transport, confinement regimes, and instabilities
Actual tokamak confinement is not determined simply by the spiraling motion of individual particles. Collisions in toroidal geometry produce neoclassical transport, but observed heat losses are often much greater because of small-scale turbulence. Instabilities driven by temperature and density gradients generate fluctuating electric and magnetic fields, which move energy and particles across magnetic surfaces. Understanding these processes requires kinetic descriptions as well as fluid models.
A major discovery was the transition from low-confinement mode, or L-mode, to high-confinement mode, or H-mode. First observed on ASDEX in 1982, H-mode develops an edge transport barrier after certain operating conditions are reached. Strongly sheared plasma flows help suppress turbulence near the boundary, creating a steep pressure pedestal. Since the edge temperature helps determine how much pressure can be sustained farther inward, this relatively narrow region has a large effect on overall performance.
The pedestal can also drive edge-localized modes, usually abbreviated ELMs. These instabilities expel particles and energy in short bursts. Although some particle exhaust is useful, large ELMs can deliver damaging transient loads to plasma-facing components. Experiments investigate mitigation through resonant magnetic perturbations, controlled pellet injection, and regimes with small or absent large ELMs. Suppressing one instability can alter density, confinement, or impurity accumulation, so successful operation requires a balanced scenario.
Larger-scale instabilities include kink modes, tearing modes, and resistive-wall modes. Tearing modes change magnetic topology through reconnection and form magnetic islands that can enhance heat transport. Neoclassical tearing modes are sustained partly by disturbances to the bootstrap current. Precisely aimed electron-cyclotron current drive can replace missing current near an island and help suppress it. Sawtooth oscillations periodically redistribute heat and particles in the central plasma and can interact with other instabilities.
A disruption is a rapid loss of a controlled discharge, often involving both a thermal quench and a current quench. Thermal energy can be deposited abruptly on surrounding surfaces, while changing currents generate electromagnetic forces. Under some conditions, strong induced electric fields accelerate runaway electrons to relativistic energies. Their concentrated impact can cause severe local damage.
Disruption research combines avoidance, prediction, and mitigation. Diagnostic signals are used to identify developing instabilities, while control systems adjust heating, fueling, magnetic fields, or current profiles. Injecting large quantities of material, including shattered cryogenic pellets, can spread energy loss through radiation and modify current decay. Reactor-scale effectiveness must account for three-dimensional radiation patterns, material assimilation, electromagnetic loads, and runaway-electron behavior.
Plasma exhaust and interactions with materials
The confined plasma is separated from material surfaces by a boundary region through which particles and heat escape. Early tokamaks often used limiters: solid components that defined the outer edge of the plasma. Many high-performance machines instead employ a divertor, in which specially shaped magnetic fields guide escaping particles toward dedicated target plates away from the main confined region.
A divertor configuration contains a magnetic X-point and a separatrix, the surface separating closed magnetic field lines from lines that intersect material targets. Outside the separatrix lies the scrape-off layer. Heat travels rapidly along its field lines, concentrating exhaust power into narrow regions at the divertor strike points. This geometry protects much of the main chamber but creates demanding local conditions for target materials and cooling systems.
An important operating objective is divertor detachment. In a detached divertor, cooling, atomic processes, and momentum losses reduce the temperature and pressure of the plasma reaching the target. Radiation from deliberately introduced impurities, together with hydrogenic excitation, ionization, and recombination processes, can distribute power through a larger volume. The detachment front must remain controlled: excessive cooling or impurity penetration can impair the core plasma and may trigger operational difficulties.
Materials must tolerate sputtering, thermal cycling, irradiation, and chemical interactions with fuel. Carbon-based components historically offered useful thermal properties and avoided the strong core radiation associated with high-atomic-number contaminants. Their disadvantages include erosion and tritium retention through co-deposition. Tungsten has a high melting point and relatively low erosion under many relevant conditions, making it important for reactor-facing designs. However, even small concentrations of tungsten in the core can cause substantial radiative losses.
Surface engineering involves more than choosing a bulk material. Divertor components commonly combine plasma-facing armour with a heat sink and actively cooled structures. Interfaces must survive repeated thermal stresses, and cooling channels must maintain acceptable temperatures without failure. Steady-state heat loads can reach the megawatt-per-square-metre range, while short transients impose additional limits.
Fusion neutrons create a separate materials problem. Their passage produces atomic displacements, changes in composition, and helium and hydrogen within structural materials. These effects can cause hardening, embrittlement, swelling, and degradation of thermal properties. Reactor components must consequently be designed for inspection, replacement, and remote handling. The lifetime of a divertor or blanket is determined by interacting thermal, mechanical, irradiation, and fuel-retention limits rather than by a single melting temperature.
Magnets, diagnostics, and real-time control
The magnet system must create strong fields while withstanding large mechanical forces. Conventional copper coils are suitable for many experimental machines, but resistive heating becomes costly during long pulses. Superconductivity allows high currents to circulate with very small electrical resistance, although the coils require cryogenic refrigeration, structural support, and protection against a transition to the normal conducting state.
Niobium–titanium and niobium–tin are established superconducting materials used in fusion magnet systems. Niobium–tin supports higher magnetic fields but is mechanically and industrially demanding. High-temperature superconducting tapes, especially rare-earth barium copper oxide conductors, offer access to stronger fields and potentially more compact configurations. Their engineering challenges include joints, mechanical strain, conductor uniformity, quench detection, and protection under high stored magnetic energy.
Stronger fields can improve reactor prospects because fusion power density depends strongly on the pressure that can be confined. Under simplified assumptions of fixed beta and temperature, fusion power density scales approximately as . This is not a complete reactor scaling law: neutron shielding, magnet stresses, exhaust geometry, and maintenance access may prevent the whole plant from shrinking in proportion to the plasma. High-field design is therefore an integrated engineering problem.
Tokamak diagnostics measure quantities that are often inaccessible to direct probes. Thomson scattering provides electron temperature and density information. Interferometry measures line-integrated electron density, while magnetic coils help reconstruct position, shape, and current. Electron-cyclotron emission provides temperature-related information under suitable conditions, and spectroscopy reveals impurity species, flows, and ion temperatures. Neutron detectors measure fusion production, while infrared cameras monitor surface heating.
These measurements feed real-time systems based on control theory and fast computation. Elongated plasmas require active vertical stabilization because they can move rapidly away from equilibrium. Controllers also regulate plasma current, shape, density, stored energy, impurity radiation, and divertor conditions. Some functions operate on submillisecond timescales; others manage the evolution of an entire discharge.
Reactor environments complicate both sensing and actuation. Neutron and gamma radiation can degrade components, access is restricted, and optical surfaces can become coated or damaged. A practical control architecture needs redundant measurements and models that continue to function when individual diagnostics fail. Techniques using machine learning have been investigated for disruption prediction and plasma control, but their deployment requires attention to unfamiliar operating conditions, reliability, and integration with machine-protection systems.
Experimental facilities and alternative tokamak configurations
Tokamak experiments serve different purposes, and their results cannot be ranked by a single record. Some machines emphasize high temperature or fusion power, others long duration, exhaust research, high magnetic field, or flexible control. A long plasma pulse does not necessarily involve deuterium–tritium fuel, large fusion gain, or reactor-level heat exhaust. Comparisons require the fuel, density, heating power, confinement regime, and duration to be specified together.
JET, located at Culham in the United Kingdom, was a major facility for deuterium–tritium research. It combined a large plasma volume with substantial auxiliary heating and developed experience with tritium processing and remote maintenance. Its later experiments used a beryllium main-chamber wall and tungsten divertor to investigate an ITER-relevant materials combination. TFTR at Princeton made complementary contributions to deuterium–tritium physics, including studies of energetic alpha particles.
Superconducting facilities have emphasized sustained operation. EAST in China and KSTAR in South Korea investigate long-pulse confinement and plasma control with superconducting magnet systems. WEST in France focuses on tungsten plasma-facing components and long-pulse exhaust issues. JT-60SA, a joint Japanese–European project that achieved first plasma in 2023, was designed to study advanced operating scenarios and support the development of ITER and subsequent reactors.
ITER brings together China, the European Union, India, Japan, South Korea, Russia, and the United States. Its design includes a major plasma radius of approximately 6.2 metres, a reference plasma current of 15 megaamperes, and a toroidal magnetic field of 5.3 teslas at the plasma centre. It is intended to integrate burning-plasma physics with large superconducting magnets, tritium systems, remote handling, and reactor-relevant component technologies. Its objectives differ from those of a commercial generating station.
A spherical tokamak has a much smaller aspect ratio than a conventional tokamak, giving the plasma a shape sometimes compared with a cored apple. Experiments such as MAST in the United Kingdom and NSTX in the United States explored its ability to sustain high plasma pressure relative to magnetic pressure. The narrow central column, however, leaves limited room for magnets, neutron shielding, and a breeding blanket in a reactor configuration.
The stellarator is a distinct toroidal confinement concept rather than a tokamak subtype. It generates most of its magnetic twist through three-dimensional external coils and does not require the same large transformer-driven plasma current. Stellarators therefore address steady-state operation differently, but face demanding coil geometry and optimization problems. Comparative research helps distinguish limits inherent to magnetic confinement from those associated specifically with tokamak currents and axisymmetry.
Requirements for an electricity-producing reactor
A fusion power plant must convert the energy carried mainly by neutrons into useful heat and then electricity. A surrounding blanket would absorb neutron energy, transfer heat to a coolant, and shield sensitive structures. Depending on coolant and temperature, the balance of plant could use a steam turbine or another heat engine cycle. Thermal efficiency must be evaluated alongside the electrical consumption of refrigeration, pumps, heating, current drive, and fuel processing.
Tritium supply is a fundamental requirement. Tritium has a half-life of approximately 12.3 years and is not naturally available in quantities suitable for a large fusion industry. Reactor designs therefore incorporate a tritium-breeding blanket containing lithium. One important reaction is
The blanket must produce enough tritium to replace that consumed in fusion while allowing for processing losses, decay, inventory requirements, and startup needs. A calculated breeding ratio above unity does not alone demonstrate a complete self-sufficient fuel cycle; extraction rates, processing time, penetrations through the blanket, and material performance also matter.
Operation may be pulsed or approximately steady-state. Pulsed designs can rely more heavily on transformer-driven current but subject structures to repeated thermal and electromagnetic cycles. They may also require thermal storage or other arrangements to smooth electricity output between pulses. Steady-state designs reduce this interruption but require sufficient bootstrap current and efficient external current drive, potentially consuming a substantial share of generated electricity.
Fusion does not use the neutron-driven fission chain reaction characteristic of conventional nuclear reactors, and loss of plasma confinement terminates fusion conditions rapidly. Nevertheless, a plant contains significant hazards: radioactive tritium, neutron-activated materials, stored magnetic energy, hot coolants, and possible chemical reaction pathways. Safety analysis must address confinement barriers, accident heat removal, fuel inventories, coolant behavior, and maintenance exposure. Waste characteristics depend on material composition, neutron exposure, and the practical availability of recycling or disposal routes.
Economic performance depends strongly on availability and component lifetime. Blankets and divertors must be replaced in a radioactive environment, making remote-maintenance speed and reliability central plant parameters. A compact core is not automatically inexpensive if its components experience higher loads or require frequent replacement. Likewise, a high plasma does not guarantee favorable plant economics if current drive, heat removal, or maintenance imposes large costs.
Integrated reactor development must therefore demonstrate compatible achievements: sufficient fusion gain, dependable fuel breeding and recovery, survivable materials, controllable exhaust, efficient power conversion, and maintainable operation. Demonstration-plant studies, often grouped under the name DEMO, examine these coupled requirements and the engineering margins needed to connect a fusion facility to an electrical grid.