EVOPEDIA

← all systems · see it in The System

Tokamak

A tokamak is a machine which uses a powerful magnetic field generated by external magnets to confine plasma in the shape of an axially-symmetric torus: toroidal and poloidal coils and a central solenoid, with a current driven in the plasma itself, twist the field lines into helices so the hot fuel stays off the walls long enough to fuse. Walked as the device, its physics and limits, its control, and the people and institutions that build and run it.

Properties published scienceour proposal

Sourced fields, and our own readings.

born at1958 (T-1 at the Kurchatov Institute)
born precisionyear (one source says late 1957)
boundaryInside: the vacuum vessel and the plasma it holds; the magnet system (toroidal field coils, central solenoid, poloidal field coils, superconducting or copper); the plasma current and the helical field it makes; heating and current drive (ohmic, neutral beams, radio-frequency waves); fuelling (gas, pellets, tritium injection); the first wall, limiters and divertor; diagnostics and the real-time plasma control system; cooling, cryogenics and pumping. Outside: the grid that powers it, the fusion reaction itself (sys:nuclear-fusion, a sibling walk), the plasma's own physics (sys:plasma), the neutrons once they leave the shield, the tritium plant and supply chain, and the laboratories, states and collaborations that design, fund and run tokamaks.
nested levelA kind of magnetic confinement fusion device (with the stellarator, pinches and mirrors); built from magnets, a vessel, a plasma, heating, fuelling, wall and control systems; embedded in national and international fusion programmes (Kurchatov, Culham, PPPL, JET, ITER) and in the wider effort to make fusion power.
scaletechnological

Measures (50) published science

Discharge (pulse) durationseconds or minutes in modern tokamaks; 1056 s on EAST (world record pulse with improved confinement, 2023)
Vertical instability growth rate held (TCV)more than 1.4 kHz, held by a controller acting at 10 kHz
Tearing prediction horizon (DIII-D)25 ms ahead (reward computed on the prediction); one instability forecast 300 ms before the disruption
Control cycle (TCV magnetic controller)10 kHz (one command every 0.1 ms)
First tokamak in operation (T-1)1958 (end of 1958; one source says late 1957) year
Kruskal-Shafranov limitq > 1 (earlier pinches ran near q = 1/3; tokamaks at q much greater than 1) dimensionless safety factor
Disruption timescaleconfinement lost in milliseconds ms
Plasma current to keep fusion alphas in the fuelabout 3 MA
Major disruption rateon the order of a few percent of shots fraction of shots
Runaway electron current in a major disruption (ITER-class)up to 12 MA deposited on a small area (plasma current about 15 MA)
ITER design disruption budget2600 events over its lifetime
Greenwald density limitn_GW [10^20 m^-3] = I_p [MA] / (pi a^2); exceeded by up to 10 times in some cases (encyclopedia)
Ohmic heating ceilingabout 2-3 x 10^7 K
Operating temperature for a reactorgreater than 10 keV (over 10^8 K)
Ideal ignition floor (D-T)4.3 keV: below it bremsstrahlung exceeds alpha heating
Fermi's self-sustaining temperature (1944)about 50,000,000 K
T-3 electron temperature and confinement (1968)1000 eV; confinement at least 50 times the Bohm limit
PLT ion temperature (1978)about 6 x 10^7 K (8,000 eV), eight times T-3's record
Sakharov MTR design power (1951)900 MW D-D reactor: major radius 12 m, minor radius 2 m, 5 T, density 3 x 10^20 m^-3, ion temperature 100 keV
ECRH absorbed fraction, T-3 1973at least 30 kW absorbed of 40 kW gyrotron power at 35 GHz (at least 0.75)
ECRH heating efficiency, T-1070-80%
Radiated fraction in q > 1 regimes (early tokamaks)did not exceed 30% of the power delivered
T-15 heating and fieldneutral beams 9 MW, ECR and ICR 6 MW each; 3.6 T on axis from Nb3Sn superconducting coils (design 3.5 T) MW, T
TCV size and actuatorsmajor radius 0.88 m, vessel 1.50 m high and 0.512 m wide; 19 actuated coils
TCV real-time magnetic sensors34 flux loops, 38 field probes, 19 coil-current measurements sensors
EAST heating and current drive power32 MW: NBI 8 MW, lower hybrid 10 MW, ECRH 2 MW at 140 GHz, ICRH 12 MW
DIII-D field, current and heatingup to 2.2 T, 2.0 MA, 23 MW T, MA, MW
JET peak D-T fusion power (DTE1, 1997)16.1 MW (ELM-free hot-ion H-mode); 22 MJ fusion energy in a steady pulse
JET gain Q (1997)0.67: 16 MW of fusion power for 24 MW of heating ratio
Scientific gain Qsci reached (TFTR, JET)0.27 (TFTR), 0.64 (JET) ratio
JET fusion energy per pulse (DTE2, 2021)59 MJ (world record at the time); 69 MJ later (record standing as of 2025)
JET neutrons produced in DTE28.5 x 10^20 neutrons
JET tritium used (T and DTE2 campaigns)1 kg (240 g into the plasma by gas injection, 763 g to the neutral beams); about 100 g in DTE1
JET wall surface temperature limitsberyllium limiters below 900 C; tungsten-coated CFC divertor tiles 1200 C
JET operating life1983 to December 2023; decommissioning expected to last until 2040 years
Tritium half-life12.32 years
Tritium consumption per GW-year of fusion power55.8 kg per 1000 MW of fusion power per year
Tritium injected relative to tritium burned20 to 200 times ratio
Tritium processing time needed for self-sufficiency1-4 h, with burn fraction times fuelling efficiency above 2%
Outside tritium needed at TBR 0.9 (STEP-scale)8-11 kg per full power year for a 1500-2000 MW device kg/yr
STEP steady heat loadsfirst wall below 1.3 MW/m2; divertor strike points 5-10 MW/m2
Tritium start-up inventory (3 GW plant)about 11 kg at 2% and 4 h processing; under 5 kg at 5% and 1 h
ITER divertor heat load (D-T phase)as high as 20 MW m^-2 MW/m2
SPARC design (field, size, gain)B0 = 12.2 T, R0 = 1.85 m, a = 0.57 m; Q > 2 goal, Q about 11 and 140 MW projected at 8.7 MA T, m
SPARC volume relative to ITERroughly 2%
D-T reaction productsalpha particle 3.5 MeV, neutron 14.1 MeV
ITER goalsQ = 10 momentarily, Q = 5 at steady state; about 500 MW of fusion power from about 50 MW of heating ratio
ITER first plasma (expected)2033-2034; deuterium-tritium operation planned from 2039 year
ITER costprojected 18-22 billion euros for construction and operations (higher estimates disputed by ITER); initial budget close to 6 billion euros EUR
Tokamak programme scale (early 1970s US)a tokamak cost about $500,000; the US annual fusion budget was about $25 million USD

Open contentions (6) open contention

Where sources or readings disagree. Both stay on the map.

Partners and parts published science

Built from / builds (14)

Sits inside / holds (4)

Trades with (valence) (22)

expressed by (35)

floor filled by (24)

Kinds (6)

wired (1)

Notes published science

Layers

The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.

the boardroom (shared floors)

  1. L5: The confined plasma: confinement and transport. Where power, fuel, heating, alphas and impurities meet in one plasma: energy confinement time, L-mode and H-mode (edge transport barrier above a power threshold), internal barriers, bootstrap current, turbulent transport; Bohm scaling beaten by an order of magnitude in the 1960s
  2. L6: Limits and control: the operating space (q above 1, Greenwald density, pressure and tearing limits, ELMs, disruptions) held by the plasma control and protection system, which reads magnetics, profiles, density, wall heat and the programme's targets and commands coils, heating and fuel in real time
  3. L7: Output and purpose: fusion power and gain (Q), records, burning plasma, ITER scenarios and power-plant designs; the laboratories, states and collaborations that set the goals and read the results

tokamak (reader's view: power and fuel in, magnets, the plasma and its equilibrium, heating, confinement, limits and control, fusion out and the programme)

  1. L1: What it draws on and sits in: electric power through the coil power supplies, deuterium and tritium fuel, the evacuated and baked vessel, cryogenics and cooling water
  2. L2: The magnets and their field: toroidal field coils (D-shaped), the central solenoid and poloidal field coils, copper or superconducting; field lines twisted into helices with safety factor q > 1
  3. L3: The plasma and its equilibrium: a current induced in the plasma (the transformer's secondary), a vertical field for force balance (Grad-Shafranov), the shape (elongation, triangularity, D-shape, X-point)
  4. L4: Heating, current drive and fuelling: ohmic heating, neutral beams, electron and ion cyclotron and lower hybrid waves, gas and pellets, and alpha self-heating in D-T
  5. L5: Confinement and transport: energy confinement time, L-mode and H-mode (edge transport barrier), internal barriers, bootstrap current, turbulence; the triple product
  6. L6: Limits and control: q, density (Greenwald), pressure and tearing limits, ELMs, sawteeth, disruptions and runaway electrons; the divertor; diagnostics feeding a real-time plasma control and protection system
  7. L7: Output and purpose: fusion power (alphas in, neutrons out), heat to the cooling system, gain and records; the mission set by laboratories and collaborations (JET, ITER, SPARC, STEP) toward power plants

way in: Alphas: self-heating from D-T fusion (inner)

  1. L1: D-T reactions in the hot core release 3.5 MeV alpha particles
  2. L2: The charged alphas are held by the same field as the fuel; they stay in when the plasma current is about 3 MA
  3. L3: Alphas collide with fuel nuclei and heat them, reducing the external heating needed
  4. L4: Alpha effects: alpha-driven Alfven eigenmodes; at ignition self-heating carries the burn alone

way in: Density and fluctuations: interferometry, reflectometry, soft x-rays

  1. L1: Microwave and far-infrared beams crossing the plasma; soft x-ray emission
  2. L2: Interferometers, reflectometers, soft x-ray detectors
  3. L3: Line-averaged density and spectra of density perturbations (modes at 156 kHz in a JET D-T pulse)
  4. L4: Modes identified and located (toroidal Alfvén eigenmodes with n = 3 or 4 driven by alphas)
  5. L5: Density held by real-time feedback in long pulses (EAST 1056 s: density well controlled)

way in: Fuel in: deuterium and tritium

  1. L1: Deuterium and tritium gas; frozen D-T pellets (seeded with a little xenon in STEP's design)
  2. L2: Tritium gas injection modules, gas valves, pellet injectors (centrifuge-propelled); neutral beams also carry fuel
  3. L3: Ionised in the plasma; beam atoms re-ionise, are trapped and add to the fuel
  4. L4: Density held by injection and cryopumping; most injected tritium leaves unburnt (injection 20-200 times consumption)

way in: Fusion products: neutrons, gamma rays, lost alphas

  1. L1: 14 MeV neutrons and gamma rays leaving the plasma, untouched by the magnetic field; escaping fast alphas
  2. L2: In-vessel-calibrated neutron yield monitors, neutron spectrometers (MPRu, diamond), a 2D neutron/gamma camera, a fast-ion loss detector and Faraday cups
  3. L3: Absolute fusion power from neutron counting, activation and gamma spectrometry
  4. L4: Tomographic reconstruction of the alpha-particle source; alpha losses mapped

way in: Heating in: neutral beams and radio-frequency waves

  1. L1: Microwaves and radio waves from gyrotrons and klystrons outside the torus; fast neutral atoms from beam injectors
  2. L2: Coupling: waves of the right frequency and polarisation resonate with electrons or ions; neutral beams cross the field and ionise
  3. L3: Energy given to electrons or ions; current driven without the transformer (lower hybrid, ECCD)
  4. L4: Collisions spread the energy to the bulk plasma (ECRH heating efficiency 70-80% on T-10)

way in: Impurities from the wall

  1. L1: Wall, limiter and divertor materials (beryllium, tungsten, carbon, lithium; ceramic and silica in the earliest machines)
  2. L2: Atoms sputtered by hot ions or outgassed from the container enter the plasma
  3. L3: Impurities radiate and cool the fuel; heavier atoms are worse
  4. L4: Core impurity accumulation degrades fusion power (JET pulse #99972)

way in: Kinetic profiles: Thomson scattering, charge-exchange spectroscopy, electron cyclotron emission

  1. L1: Laser light scattered by the plasma's electrons, charge-exchange spectral lines, electron cyclotron emission
  2. L2: Thomson scattering system, charge-exchange recombination spectroscopy, ECE radiometer (core Te from ECE on JET)
  3. L3: Real-time profiles of electron density and temperature, ion temperature and rotation
  4. L4: Profiles reconstructed and mapped on magnetic flux coordinates with EFIT
  5. L5: A dynamic model predicts normalised pressure and tearing likelihood 25 ms ahead

way in: Magnetics: flux loops, field probes and coil currents

  1. L1: The magnetic field and flux round the plasma, made by the coils and the plasma current
  2. L2: Wire loops that measure magnetic flux, probes that measure the local field, coil-current measurements (34, 38 and 19 on TCV)
  3. L3: Sampled every control cycle (10 kHz) and handed to the controller
  4. L4: Equilibrium reconstruction: an inverse problem finding the current distribution that satisfies Grad-Shafranov force balance and best matches the measurements (LIUQE, EFIT)
  5. L5: State estimates: plasma current, position, shape, X-point; or read directly by a learned policy without a shape observer

way in: Power in: electricity to the coils

  1. L1: Electric power from the supply
  2. L2: Thyristor-based power supplies apply voltages to the coils
  3. L3: Coil currents make the toroidal and poloidal fields; a rising solenoid current induces the plasma current
  4. L4: The plasma current makes the poloidal field that twists the lines and heats the plasma ohmically

way in: The programme: experiment plans and control targets

  1. L1: Experimental goals written by physicists: shapes, currents, scenarios (ITER-like shapes, snowflakes, droplets)
  2. L2: Targets specified at a high level and combined into a reward function
  3. L3: The policy and its targets compiled into an executable for real-time control at 10 kHz
  4. L4: Loaded into the TCV control framework; a traditional controller forms the plasma, then hands over at a set time

way in: Wall heat: pyrometers and infrared cameras

  1. L1: Heat on the limiters, first wall and divertor tiles
  2. L2: An array of pyrometers and infrared cameras
  3. L3: Real-time vessel temperature monitoring
  4. L4: Alarm generation against the surface limits (beryllium 900 C, tungsten-coated tiles 1200 C)

way out: Coil voltages: shaping and holding the plasma

  1. L1: The plasma's position, current and shape follow (elongation, triangularity, X-points, two 'droplets')
  2. L2: Power supplies change coil currents; the poloidal field changes
  3. L6: The controller sends voltage commands to the coils (a single network to all 19 coils on TCV; PID loops conventionally)

way out: Fuel and gas: density control

  1. L1: Gas puffs and pellets enter the plasma; density held
  2. L6: Fuel and gas injection, one of the three main plasma control actuators

way out: Heat and ash out: divertor, wall, cooling water

  1. L1: The wall conducts heat to water cooling; pumps remove gas and helium ash
  2. L3: The divertor takes them to target plates, away from the main wall; detachment spreads the load
  3. L5: Heat and particles leave the confined region along open field lines

way out: Heating power: beams and waves turned up or down

  1. L1: Plasma pressure rises or falls
  2. L4: Beams and waves deposit energy and drive current
  3. L6: Total beam power and triangularity chosen by the controller; the plasma control system sets the individual beams and coils

way out: Light out: the visible glow and other radiation

  1. L1: Visible light dominated by the hydrogen-alpha line (656 nm) reaches cameras and eyes
  2. L5: The plasma radiates; impurities radiate more

way out: Neutrons out: shield, blanket, tritium breeding

  1. L1: In reactor designs they meet lithium in a breeding blanket and make tritium and heat; they activate the hall
  2. L2: Neutrons pass the vessel; shields of water, plastics and boron concrete absorb them
  3. L5: Each D-T reaction sends a 14.1 MeV neutron out of the plasma

way out: Results out: records, scenarios, trained staff

  1. L1: Published papers and design bases read by the next machines (SPARC's projections rest on the ITER Physics Basis)
  2. L7: Results compiled: fusion energy records, isotope physics, scenarios for ITER, staff experience

way out: Transients: ELMs, disruptions, runaway beams onto the wall

  1. L1: Thermal and electrical loads land on the vessel in milliseconds; runaway beams can burn through it
  2. L6: Control or limits fail: ELM crash, thermal and current quench, runaway avalanche

Sources (25) published science

Tokamak (Wikipedia)
ITER (Wikipedia; Wikimedia Enterprise via tools/wme) (behind its links)
Degrave et al., Magnetic control of tokamak plasmas through deep reinforcement learning, Nature 602, 414 (2022), doi:10.1038/s41586-021-04301-9 (Europe PMC full text, PMC8850200, CC BY 4.0) (behind its links)
Spherical tokamak (Wikipedia; Wikimedia Enterprise via tools/wme) (behind its links)
Smirnov V.P., Tokamak foundation in USSR/Russia 1950-1990, Nuclear Fusion 50, 014003 (2010), doi:10.1088/0029-5515/50/1/014003 (IOP open PDF) (behind its links)
Creely A.J. et al., Overview of the SPARC tokamak, J. Plasma Phys. 86, 865860502 (2020), doi:10.1017/S0022377820001257 (Cambridge open access, CC BY-NC-ND 4.0) (behind its links)
Degrave J. et al., Magnetic control of tokamak plasmas through deep reinforcement learning, Nature 602, 414-419 (2022), doi:10.1038/s41586-021-04301-9 (Europe PMC full text PMC8850200, CC BY 4.0) (behind its links)
Seo J. et al., Avoiding fusion plasma tearing instability with deep reinforcement learning, Nature 626, 746-751 (2024), doi:10.1038/s41586-024-07024-9 (Europe PMC full text PMC10881383, CC BY 4.0) (behind its links)
Giacomin M. et al., First-principles density limit scaling in tokamaks based on edge turbulent transport and implications for ITER, arXiv:2204.02911 (2022; Phys. Rev. Lett. 128, 185003) (behind its links)
Wurzel & Hsu, Progress toward fusion energy breakeven and gain as measured against the Lawson criterion, Physics of Plasmas 29, 062103 (2022), doi:10.1063/5.0083990 (arXiv:2105.10954v4, full text read) (behind its links)
Song Y. et al., Realization of thousand-second improved confinement plasma with Super I-mode in Tokamak EAST, Science Advances 9, eabq5273 (2023), doi:10.1126/sciadv.abq5273 (PMC9821864, CC BY-NC 4.0) (behind its links)
Maggi C.F. et al., Overview of T and D-T results in JET with ITER-like wall, Nuclear Fusion 64, 112012 (2024), doi:10.1088/1741-4326/ad3e16 (IOP open access, CC BY 4.0) (behind its links)
Joint European Torus (Wikipedia; Wikimedia Enterprise via tools/wme) (behind its links)
Abdou M. et al., Physics and technology considerations for the deuterium-tritium fuel cycle and conditions for tritium fuel self sufficiency, Nuclear Fusion 61, 013001 (2021), doi:10.1088/1741-4326/abbf35 (IOP open access) (behind its links)
Lord M. et al., Fusing together an outline design for sustained fuelling and tritium self-sufficiency, Phil. Trans. R. Soc. A 382, 20230410 (2024), doi:10.1098/rsta.2023.0410 (PMC11423675, CC BY 4.0) (behind its links)
Cane J. et al., Managing the heat: in-vessel components, Phil. Trans. R. Soc. A 382, 20230408 (2024), doi:10.1098/rsta.2023.0408 (PMC11423682, CC BY 4.0) (behind its links)
Meyer H. (STEP Plasma Team), Plasma burn: mind the gap, Phil. Trans. R. Soc. A 382, 20230406 (2024), doi:10.1098/rsta.2023.0406 (PMC11423683, CC BY 4.0) (behind its links)
Edge-localized mode (Wikipedia; Wikimedia Enterprise via tools/wme) (behind its links)
Wang L. et al., Integration of full divertor detachment with improved core confinement for tokamak fusion plasmas, Nature Communications 12, 1365 (2021), doi:10.1038/s41467-021-21645-y (PMC7921092, CC BY 4.0) (behind its links)
Divertor (Wikipedia; Wikimedia Enterprise via tools/wme) (behind its links)
Beidler C.D. et al., Demonstration of reduced neoclassical energy transport in Wendelstein 7-X, Nature 596, 221-226 (2021), doi:10.1038/s41586-021-03687-w (PMC8357633, CC BY 4.0) (behind its links)
Stellarator (Wikipedia; Wikimedia Enterprise via tools/wme) (behind its links)
Hassanein A. and Sizyuk V., Potential design problems for ITER fusion device, Scientific Reports 11, 2069 (2021), doi:10.1038/s41598-021-81510-2 (PMC7822851, CC BY 4.0) (behind its links)
Gong P, Kwok TWJ, Wang D et al. 2025. A multi-scale microstructure to address the strength-ductility trade off in high strength steel for fusion reactors. Nature Communications. doi:10.1038/s41467-025-58042-8 (PMC11926085) (behind its links)
High-confinement mode (Wikipedia; Wikimedia Enterprise via tools/wme) (behind its links)