# Fusion power **Fusion power** is the proposed generation of electricity from the heat of [[Nuclear_fusion|nuclear fusion]] — the joining of light nuclei into heavier ones, the process that powers the [[Sun|Sun]]. The physics is not in doubt and the fuel is effectively unlimited; what has resisted seventy years of work is the engineering condition that the [[Plasma_(physics)|plasma]] must satisfy simultaneously in three quantities at once, and that condition is what the microsim on this page is about. In the microsim below the reader sets three numbers and watches a power balance tip. Temperature T in keV, particle density n, and energy confinement time tau_E together fix three curves: the fusion power computed from the deuterium–tritium reactivity, the unavoidable radiation loss `P_br ∝ n²·√T` from [[Thermal_radiation|bremsstrahlung]], and the transport loss `3·n·k_B·T/tau_E` that describes heat leaking out of the confined region.[^nrl][^spec-p55] Where the first exceeds the sum of the other two, the plasma heats itself and the point crosses the ignition curve, `n·tau_E ≥ 1.5×10²⁰ s/m³` near 25 keV.[^lawson-std] Because the requirement depends on temperature too, the figure of merit usually quoted is the triple product, `n·T·tau_E ≥ 3×10²¹ keV·s/m³`, and a table of real machines — JET, the [[Tokamak|ITER]] design point and [[Inertial_confinement_fusion|NIF]] — sits on the plane as points rather than as claims.[^lawson-std][^spec-p55] Two presets read the same criterion from opposite ends: a magnetic preset with a thin plasma held for seconds, and an inertial preset with a compressed pellet that lives for picoseconds. On the [[Physics]] flagship this article serves the *Fusion power* section of Part IV — Branches and fields, and it owns that section. Its companion at the other end of the [[Nuclear_binding_energy|binding-energy]] curve is [[Nuclear_fission]]. ## Terminology Several words that sound like synonyms are not. *Breakeven* means the fusion power equals the heating power delivered to the plasma, a ratio written Q = 1. *Ignition* is stronger: the charged fusion products alone sustain the temperature, so no external heating is needed. *Scientific gain* compares fusion energy with energy delivered to the target or plasma; *engineering gain* compares electricity out with electricity in at the wall plug, and it is always the smaller and always the one that matters commercially. *Thermonuclear* means the reacting nuclei are hot and randomly moving, as opposed to beam-driven schemes where they are not. ## Background Two nuclei must come within about a femtometre for the [[Nuclear_force|nuclear force]] to act, and to get there they must overcome their mutual [[Coulomb's_law|Coulomb]] repulsion. Classically this demands energies far above anything a plasma reaches; in practice [[Quantum_tunnelling|tunnelling]] through the barrier does most of the work, which is why fusion happens at all at stellar temperatures. The Sun's core runs at only 1.6×10⁷ °C, well below the temperature any terrestrial scheme needs, and compensates with density, gravitational confinement and an immense amount of time.[^murphy-fusion] ### Cross section The reaction rate per unit volume is `n1·n2·<sigma·v>`, where `<sigma·v>` is the reactivity — the cross-section averaged over the thermal velocity distribution. Its temperature dependence is brutal at low T and flattens at high: below about 20 keV the deuterium–tritium reactivity rises roughly as T², so a factor of two in temperature buys a factor of four in power, and this steepness is why fusion schemes fail suddenly rather than gradually.[^nrl] Deuterium–tritium is chosen over every other candidate because its reactivity peaks at the lowest temperature by a wide margin. The sim reads `<sigma·v>` from a fitted evaluation rather than integrating a cross-section at run time.[^nrl][^spec-p55] ### Lawson criterion The criterion is a power balance, and writing it out shows where each term comes from. The plasma gains `P_fus` from fusion, of which only the charged products — for D–T, the 3.5 MeV alpha out of the 17.6 MeV total — stay behind to heat it. It loses energy two ways: by radiating, principally as bremsstrahlung scaling as `n²·√T`, and by transport, at a rate `3·n·k_B·T/tau_E`, which is simply the stored thermal energy divided by the time it takes to leak away.[^lawson-std][^nrl] Set gain equal to loss, and the density cancels partly against itself, leaving a condition on the *product* n·tau_E rather than on either separately: for D–T ignition, `n·tau_E ≥ 1.5×10²⁰ s/m³` near 25 keV.[^lawson-std] That single product is the reason two utterly different machine families exist. A magnetic device works at n ≈ 10²⁰ m⁻³, so it needs tau_E of order 1.5 s — a plasma held steady for seconds.[^derived-fp] An inertial device abandons confinement entirely and lets the pellet fly apart, giving tau_E of order 10⁻¹¹ s, which forces n ≈ 10³¹ m⁻³, more than a thousand times solid [[Density|density]].[^derived-fp] The same inequality, satisfied eleven orders of magnitude apart. The sim's two presets are exactly these two solutions of one equation, and it is worth moving the sliders from one to the other to see that neither is a different physics. ### Triple product: density, temperature, time Because `<sigma·v>` itself depends on temperature, n·tau_E alone does not say how close a machine is to ignition unless T is quoted with it. Multiplying through gives the fusion triple product, `n·T·tau_E`, which for D–T ignition must exceed about 3×10²¹ keV·s/m³ and which is nearly flat in T across the operating window of 10–25 keV.[^lawson-std] This is the number by which machines are compared, and its convenience is that a tokamak and a laser target can be plotted on the same axis despite sharing almost nothing else. The arithmetic makes the engineering visible. At n = 10²⁰ m⁻³ and T = 15 keV, the triple product requires tau_E ≥ 2 s.[^derived-fp] Confinement time is not a material property that can be specified; it is an emergent result of turbulence, geometry and current, and it is the quantity every magnetic design is really trying to buy. Note also the unit conversion the sim exposes: 1 keV of [[Temperature|temperature]] is 1.16×10⁷ K, so a 15 keV plasma is at 1.7×10⁸ K, ten times the Sun's core.[^derived-fp][^murphy-fusion] ### Energy capture In a D–T plant four fifths of the energy leaves as a 14.1 MeV [[Neutron|neutron]], which no [[Magnetic_field|magnetic field]] can hold. It is absorbed in a surrounding blanket, where it both delivers heat to a conventional [[Steam_turbine|steam]] cycle and breeds replacement tritium from [[Lithium|lithium]]. That double duty is not optional: tritium does not occur naturally in useful quantities, so a D–T plant that does not breed its own fuel has no fuel.[^lawson-std] ## Plasma behavior Above a few electronvolts a gas is fully ionised and becomes a plasma, in which charged particles respond collectively to fields they themselves generate. That collective response is the source of every difficulty. A confined plasma supports a catalogue of instabilities — kinks, tearing modes, edge-localised modes — which redistribute energy far faster than collisions would, and the resulting anomalous transport is what sets tau_E in practice rather than any calculable diffusion coefficient. [[Magnetohydrodynamics|Magnetohydrodynamics]] describes the large-scale motions well enough to design a machine; the turbulence that governs confinement is a computational problem, not an analytic one. ## Methods Confinement schemes fall into two families defined by which side of the Lawson product they attack, plus a long tail of alternatives that have not scaled. ### Magnetic confinement A charged particle spirals about a magnetic field line, so a closed field configuration can hold a thin plasma almost indefinitely in principle. The [[Tokamak|tokamak]] — a torus with a strong toroidal field and a driven plasma current — is the most developed geometry and holds every performance record; the [[Stellarator|stellarator]] achieves the same twist with external coils alone, trading engineering complexity for the absence of a disruptive plasma current. Both need field strengths that today mean [[Superconductivity|superconducting]] magnets and [[Cryogenics|cryogenic]] plant.[^lawson-std] ### Inertial confinement The alternative is to give up on holding the plasma and instead compress a millimetre-scale fuel capsule so violently that it burns before it disassembles. Lasers or X-rays ablate the capsule surface, the reaction drives the remainder inward, and densities of order 10³¹ m⁻³ are reached for tens of picoseconds. On 5 December 2022 the National Ignition Facility reported the first laboratory target gain above unity, 3.15 MJ of fusion yield from 2.05 MJ of laser energy delivered to the target — a scientific-gain result, not an engineering one, since driving those lasers took far more than 2.05 MJ from the grid.[^nif][^derived-fp] ## Common tools Every scheme shares a toolkit. Plasmas are heated ohmically by their own current until resistivity falls too low, then by neutral-beam injection and by radio-frequency waves tuned to cyclotron resonances. Diagnostics must be non-contact almost by definition: interferometry for density, Thomson scattering for temperature, spectroscopy for impurities, [[Neutron_detection|neutron counting]] for the fusion rate. Control is increasingly a data problem, and [[Machine_learning|machine learning]] has moved from analysis into real-time disruption prediction. Underlying all of it is magnet technology, which is why high-temperature [[Superconducting_magnet|superconducting magnets]] have changed the design space more in the last decade than any plasma-physics result. ## Fuels Four reactions are seriously discussed, and they trade reactivity against neutron production. The three Murphy tabulates give the shape of the choice: | reaction | energy released | temperature required | energy density | |---|---|---|---| | 4 ¹H → ⁴He | 26.7 MeV | ~10⁹ K | 153 million kcal/g | | ²H + ²H → ⁴He | 23.8 MeV | ~10⁸ °C | 137 million kcal/g | | ²H + ³H → ⁴He + n | 17.6 MeV | ~4.5×10⁷ °C | 81 million kcal/g | Deuterium–tritium releases the least energy per gram of the three and is nevertheless the universal first choice, because it requires by far the lowest temperature.[^murphy-fusion] Even so, D–T at 81 million kcal/g carries about five times the energy density of ²³⁵U fission at 16.8 million, and D–D about eight times.[^murphy-fusion][^manual10][^derived-fp] [[Deuterium|Deuterium]] is abundant in seawater; [[Tritium|tritium]] must be bred. D–³[[Helium-3|He]] and proton–[[Boron|boron]]-11 are [[Aneutronic_fusion|aneutronic]] and would avoid neutron damage altogether, at the cost of temperatures and triple products far beyond anything demonstrated. ## Material selection The blanket and first wall face a 14 MeV neutron flux that displaces atoms, swells and embrittles [[Steel|steel]], and transmutes it into radioactive species. Reduced-activation ferritic steels are designed so that the activated products decay to handleable levels in decades rather than millennia. The divertor, where the plasma exhaust lands, sees heat fluxes comparable to a rocket nozzle and is built from tungsten; [[Beryllium|beryllium]] has served as a low-Z first-wall material because high-atomic-number impurities radiate the plasma's energy away catastrophically.[^lawson-std] ## Accident scenarios and the environment The characteristic safety argument for fusion is structural rather than procedural: the plasma holds at most a few grams of fuel at any instant, and any disturbance — an impurity influx, a loss of field, a leak — cools it and stops the reaction. Nothing about the reaction is self-amplifying in the way a neutron chain is, so no configuration runs away.[^murphy-fusion] The real hazards are different in kind. A superconducting magnet quench dumps stored [[Magnetic_energy|magnetic energy]] as heat in the coil. Tritium is a mobile beta emitter that permeates metals, so inventory control rather than criticality control dominates licensing. Activated structure is a genuine waste stream, though shorter-lived than fission's, and no long-lived actinides are produced. ## Potential military usage Uncontrolled fusion has been militarily practical since the 1950s, in thermonuclear weapons where a fission primary supplies the compression and temperature. Civil research overlaps that expertise in inertial confinement, which is why large laser facilities are funded through weapons-stewardship programmes as well as energy programmes. Magnetic confinement has no comparable weapons pathway. ## Economics Fusion's cost structure resembles [[Nuclear_power|fission's]] more than gas: very high capital cost, negligible fuel cost, and an economics decided almost entirely by capacity factor and cost of capital. No fusion plant has generated a kilowatt-hour for a grid, so every published cost is a projection from a design rather than a measurement. The honest statement of the position is the Portal Book's: controlled fusion has never been net energy positive at the plant level, and neither ITER nor NIF is a power plant — ITER carries no generating equipment at all.[^murphy-fusion] ## Regulation Because fusion produces no fissile material and cannot undergo a criticality excursion, several jurisdictions have moved to regulate fusion devices under radiation-source rules rather than reactor-licensing rules, which materially changes the cost and schedule of a first-of-a-kind plant. The [[Nuclear_fuel_cycle|safeguards]] questions that dominate fission regulation apply instead to the tritium inventory and to the neutron source's potential for transmuting fertile material.[^iaea] ## Geopolitics Fusion research has been unusually collaborative for a strategic technology: ITER is a seven-party construction project and the plasma-physics literature is open. Competitive pressure has shifted to supply chains — superconducting tape, tritium handling, tungsten fabrication — and to private programmes pursuing compact high-field designs on shorter timescales.[^iaea] ## Advantages The fuel is effectively unlimited and geographically unconcentrated; the reaction cannot run away; no long-lived actinide waste is produced; and the [[Energy_density|energy density]] exceeds fission's by five to eight times per gram.[^murphy-fusion][^derived-fp] Output is not weather-dependent, so a fusion plant is dispatchable baseload without [[Energy_storage|storage]]. ### Helium production Every reaction on the table produces [[Helium|helium]], which is inert, non-radioactive and commercially valuable. A D–T plant makes it at roughly one gram per 100 GJ of fusion energy — a waste-disposal question only in the sense that the ash must be pumped out continuously, since it dilutes the fuel and radiates energy away.[^derived-fp] ## Disadvantages The engineering problem is genuinely hard and has been consistently underestimated for seventy years. A D–T plant must breed its own tritium with a breeding ratio above one while surviving the neutron flux that does the breeding, and no facility has demonstrated the combination. Capital cost per unit output is projected to be high, and the neutron-activated structure is a waste stream even if a short-lived one. Against fission, fusion has no operating record at all; against [[Wind_power|wind]] and [[Solar_cell|solar]], it competes with technologies whose costs have fallen while fusion's schedule has not. ## History The first controlled-fusion programmes began in the early 1950s in Britain, the United States and the Soviet Union, largely in secret and largely in pinch devices whose confinement collapsed to instabilities. The field's decisive result came in 1968, when Soviet measurements on the T-3 tokamak reported electron temperatures an order of magnitude above anything achieved elsewhere; independent confirmation by a visiting British team converted most of the world's programmes to the tokamak within a few years.[^lawson-std] The Joint European Torus, operating from 1983, produced 16 MW of fusion power in a deuterium–tritium pulse in 1997, the peak-power record for magnetic confinement and a scientific Q of roughly two thirds.[^jet] On the inertial side, laser compression experiments began in the 1970s and culminated in the National Ignition Facility's target gain above unity in December 2022.[^nif] ITER, under construction since 2010, is designed for 500 MW of fusion power from 50 MW of injected heating, a scientific gain of ten.[^iter][^derived-fp] ## Future development The near-term programme divides between the intergovernmental route — ITER, then a demonstration plant — and privately funded compact machines betting that high-temperature superconducting magnets let a much smaller device reach the same triple product, since fusion power in a tokamak scales steeply with field strength. Both converge on the same unproven step: a tritium breeding blanket that works, under neutron irradiation, for years. Until that is shown at scale, every date for [[Electric_power|fusion electricity]] is a projection. ## See also - [[Lawson_criterion]] — the ignition condition the sim plots - [[Tokamak]] - [[Inertial_confinement_fusion]] - [[Magnetic_confinement_fusion]] - [[Stellarator]] - [[Nuclear_fusion]] - [[Aneutronic_fusion]] - [[Nuclear_binding_energy]] ## References [^murphy-fusion]: Murphy, Thomas (2021). *Energy and Human Ambitions on a Finite Planet*, Chapter 15, pp. 285–294: Eq. 15.3, the three fusion reactions 4 ¹H → ⁴He + 26.7 MeV, ²H + ²H → ⁴He + 23.8 MeV and ²H + ³H → ⁴He + n + 17.6 MeV, with the temperatures they require (~10⁹ K, ~10⁸ °C and ~4.5×10⁷ °C) against the Sun's core at 1.6×10⁷ °C, p. 285; the energy densities 153, 137 and 81 million kcal/g for p–p, D–D and D–T against 16.8 million kcal/g for ²³⁵U fission, p. 285 with pp. 273–274; and the statement that controlled fusion has never been net-positive and that neither ITER nor NIF is a power plant, pp. 285–287. Claims drawn from elsewhere in the pp. 285–294 range are marked (page to pin). https://open.umn.edu/opentextbooks/textbooks/energy-and-human-ambitions-on-a-finite-planet [^nrl]: *NRL Plasma Formulary*, U.S. Naval Research Laboratory, sections on fusion reactivities and on bremsstrahlung radiation (pages to pin): the fitted `<sigma·v>` evaluations for D–T and the other candidate reactions from which the sim reads its reactivity, and the bremsstrahlung power scaling `P_br ∝ n²·√T` together with its numerical constant. No Portal Book in this cluster carries either. [^lawson-std]: The Lawson power balance, the triple product, the confinement geometries and the blanket and material requirements are standard results of the fusion literature, taken here from Freidberg, Jeffrey P. (2007). *Plasma Physics and Fusion Energy*, Cambridge University Press (pages to pin), and Wesson, John. *Tokamaks*, Oxford University Press (pages to pin), which also supply the T-3 tokamak result of 1968 and its confirmation. No Portal Book in this cluster treats plasma confinement. [^iter]: ITER Organization, published design parameters for the ITER tokamak (page to pin): 500 MW of fusion power from 50 MW of injected heating power, a scientific gain Q = 10, and the absence of any electricity-generating equipment in the design. [^nif]: Lawrence Livermore National Laboratory and the U.S. Department of Energy, announcement of the National Ignition Facility result of 5 December 2022 (page to pin): a fusion yield of 3.15 MJ from 2.05 MJ of laser energy delivered to the target, the first laboratory demonstration of target gain above unity, reported explicitly as a target-level rather than a facility-level gain. [^jet]: United Kingdom Atomic Energy Authority / EUROfusion, records of the Joint European Torus deuterium–tritium campaigns (page to pin): the 1997 pulse producing a peak fusion power of 16 MW, the magnetic-confinement peak-power record, at a scientific gain of roughly 0.65. [^iaea]: International Atomic Energy Agency, material on fusion energy status, safeguards and regulation (page to pin): the treatment of fusion devices under radiation-source rather than reactor-licensing frameworks in several jurisdictions, the international composition of the ITER project, and the tritium-inventory focus of fusion safeguards. No Portal Book covers fusion regulation or geopolitics. [^manual10]: Wikitube MicroSim Guide, sub-manual 10 *Earth, Energy and Environment*, §4.2 "Binding energy per nucleon: why fusion and fission both pay": the binding-energy curve, the statement that fusion pays on the low-A side of the ⁵⁶Fe peak, the per-gram comparison in which D–D fusion yields about eight times what fission yields (137 against 16.8 million kcal/g), and the pitfall that controlled fusion has never been net-positive and that ITER and NIF are not power plants. [^spec-p55]: Matter & Energy Cluster contract, `_registry/plans/PHYSICS_SECTIONS.md` row P55: the sim concept for this page — the Lawson diagram, with T in keV, density n and confinement time tau_E as the reader's three controls; fusion power from `matter.fusion.reactivity('DT', T)` set against bremsstrahlung `P_br ∝ n²·√T` and transport loss `3·n·k_B·T/tau_E`; the ignition curve `n·tau_E ≥ 1.5×10²⁰ s/m³` near 25 keV and the triple product `n·T·tau_E ≥ 3×10²¹ keV·s/m³`; the table of machines (JET, the ITER design point, NIF) as points on the plane; and the magnetic and inertial presets. [^derived-fp]: Computed for this article from the criteria and constants cited above: at n = 10²⁰ m⁻³ the ignition product `n·tau_E ≥ 1.5×10²⁰ s/m³` requires tau_E ≥ 1.5 s, while at tau_E = 10⁻¹¹ s it requires n ≥ 1.5×10³¹ m⁻³, the eleven-order-of-magnitude split between the magnetic and inertial solutions of the same inequality; the triple product `3×10²¹ keV·s/m³` at n = 10²⁰ m⁻³ and T = 15 keV gives tau_E ≥ 2 s; the temperature conversion 1 keV = 1.160×10⁷ K, so 15 keV = 1.74×10⁸ K, about ten times the Sun's core temperature of 1.6×10⁷ °C; the per-gram ratios 81 ÷ 16.8 = 4.8 for D–T against ²³⁵U fission and 137 ÷ 16.8 = 8.2 for D–D; the NIF target gain 3.15 ÷ 2.05 = 1.54; the ITER scientific gain 500 ÷ 50 = 10; and the helium production rate, 17.6 MeV = 2.82×10⁻¹² J per reaction and 4 amu = 6.64×10⁻²⁷ kg of helium per reaction, giving 2.4×10⁻⁶ kg of helium per 10¹¹ J, i.e. of order one gram per 100 GJ of fusion energy. ## Bibliography - Murphy, Thomas (2021). *Energy and Human Ambitions on a Finite Planet*, Chapter 15. CC BY-NC. Portal Book 097; the fusion reaction energies, temperatures and energy densities used here. - *NRL Plasma Formulary*, U.S. Naval Research Laboratory. The reactivity fits and the bremsstrahlung scaling behind the sim. - Freidberg, Jeffrey P. (2007). *Plasma Physics and Fusion Energy*, Cambridge University Press. - Wesson, John. *Tokamaks*, Oxford University Press. ## Further reading - International Atomic Energy Agency, fusion energy programme documentation — status reports, the Fusion Device Information System, and the safeguards and regulatory discussion referenced above. - ITER Organization, published machine description and design parameters. - Lawrence Livermore National Laboratory, National Ignition Facility result documentation for the December 2022 ignition shot. ## External links - [*Energy and Human Ambitions on a Finite Planet*](https://open.umn.edu/opentextbooks/textbooks/energy-and-human-ambitions-on-a-finite-planet) (Murphy, 2021), Chapter 15 — the fusion energetics behind this page - The Wikipedia pair's *External links* section lists the laboratory, agency and project sites (IAEA, ITER, LLNL, EUROfusion) for current machine parameters and results; those are the canonical places to check any number on this page against its latest value. <!-- MATTERSIM:BEGIN g33 — Matter & Energy Cluster microsim (framework build, specs/sims/Fusion_power.json); do not hand-edit inside --> **Microsim — three.js (Wikitube framework):** *Fusion power* <div class="wt-sim" data-src="https://wikitube-3d-microsims.netlify.app/matter/Fusion_power.html" data-title="Fusion power"></div> *Built from `MICROSIM_GUIDE/specs/sims/Fusion_power.json`; part of the [[PORTAL_Matter|Matter portal]] spine (section sims and See-also variants).* <!-- MATTERSIM:END --> ## Wikipedia : Wikitube **Strict pair:** [Wikipedia](https://en.wikipedia.org/wiki/Fusion_power) : [Wikitube](https://en.wikitube.io/wiki/Fusion_power) · pinned revision [1371534300](https://en.wikipedia.org/w/index.php?oldid=1371534300) · 2026-09-11 ## Previous hub tags Hubs: `Life_Physics`. Portals: [[PORTAL_Physics]], [[PORTAL_Energy]]. --- *Matter & Energy Cluster child articles, wave 1 · 2026-09-11 · drafted · Physics row P55 · sim pending (matter/Fusion_power).*