Gyrotron RF Heating

A vacuum tube the size of a phone booth that uses relativistic quantum mechanics to convert a 55-kilovolt electron beam into a 1-megawatt beam of 170 GHz microwaves — and then aims it like a scalpel into the heart of a burning plasma.

A gyrotron is a vacuum-tube microwave source that produces 1 MW of continuous-wave power at 170 GHz — a millimetre-wave frequency too high for conventional solid-state or klystron sources to reach at useful power levels. ITER's electron cyclotron resonance heating (ECRH) system uses 24 such gyrotrons, delivering 24 MW of millimetre-wave power to the plasma through 160 m of oversized corrugated waveguide. ECRH deposits energy directly into plasma electrons at their cyclotron resonance frequency, heating the plasma to fusion temperatures of 100–150 million degrees. Beyond bulk heating, ECRH can drive localised off-axis currents to suppress neoclassical tearing modes (NTMs) — large-scale magnetic islands that would otherwise degrade plasma confinement — by steering the gyrotron beam to deposit power in a region just a few centimetres wide inside the plasma. The ability to aim the beam with millisecond precision makes ECRH a unique real-time feedback control tool for plasma stability.

The gyrotron operates on the electron cyclotron maser instability. An annular beam of electrons is emitted from a magnetron injection gun (MIG) cathode at approximately −55 kV and accelerated by a coaxial anode structure. As the beam enters the bore of a large superconducting solenoid (approximately 6.7 T for a 170 GHz fundamental-mode gyrotron), the electrons are adiabatically compressed: their transverse (orbital) velocity increases relative to their axial (drift) velocity, so that the pitch angle α = v⊥/v∥ approaches 1.3–1.5. The electrons are only weakly relativistic (γ ≈ 1.05–1.1), but the relativistic mass increase is sufficient to cause azimuthal phase bunching: faster electrons slow down and slower electrons speed up relative to the cyclotron wave, so the beam spontaneously organises into rotating clumps that emit coherently. In a high-order transverse-mode open resonant cavity (typically TE31,11 or TE32,9 for 170 GHz, 1 MW ITER gyrotrons), the bunched beam radiates at the cyclotron frequency f₂ = eB/(2πmₑγ), depositing its transverse kinetic energy into the microwave field. A quasi-optical mode converter (a Vlasov launcher plus a series of phase-correcting mirrors) transforms the high-order whispering-gallery cavity mode into a free Gaussian beam, which exits through a chemical vapour deposition (CVD) diamond window — the only material that combines low millimetre-wave absorption loss with the thermal conductivity (~1,500 W/m·K) needed to survive MW-level transmission for 1,000-second pulses. A single-stage depressed collector recovers 25–30% of the spent-beam energy, boosting wall-plug efficiency to approximately 50%.

In plain terms

Imagine a spinning top that wobbles in a circle at exactly the right speed to match the rhythm of a sound wave — and every time it wobbles, it hands energy to that wave. A gyrotron does the same thing with electrons spinning inside a powerful magnetic field: the electrons wobble at exactly 170 GHz, and their energy pours into a beam of microwave radiation powerful enough to heat plasma to 100 million degrees.

Also called: gyrotron, electron cyclotron resonance heating (ECRH), ECH source, electron cyclotron maser, ECRH gyrotron.

Why it matters

ECRH is the only external heating method in ITER that can deposit energy in a precisely defined, steerable spatial location inside the plasma cross-section — essential for suppressing magnetic islands that would otherwise cause a plasma disruption.
24 gyrotrons at 1 MW each provide ITER's baseline 24 MW ECRH system, operating at 170 GHz for pulse lengths of up to 3,600 seconds — a continuous-wave power level and duration that no prior millimetre-wave source approached before the ITER development programme.
Wall-plug efficiency of ~50% (with depressed collector) means the 24 MW of delivered microwave power requires only ~48 MW of electrical input, directly affecting the energy balance of any future fusion power plant that relies on ECRH for startup and current drive.
The CVD diamond output window is the single most technically demanding component in the gyrotron: at 1 MW and 170 GHz, no other material has simultaneously survived the dielectric loss heating, thermal stress, and neutron irradiation environment over 1,000-second pulses.
Gyrotron technology developed for fusion has secondary applications in materials processing, DNP-NMR spectroscopy, and W-band radar, making it a dual-use enabling technology for advanced manufacturing and scientific instrumentation.

Where it lives

ITER ECRH system — 24 gyrotrons × 1 MW at 170 GHz, 24 transmission lines, 5 launchers; first gyrotrons delivered by Europe and Japan, 2018–
Wendelstein 7-X (Germany) — 10 MW ECRH from 140 GHz gyrotrons used as primary heating source for the world’s most advanced stellarator
DIII-D (General Atomics, USA) — 110 GHz ECRH system used for current profile control and NTM suppression experiments
JT-60SA (Japan, ITER satellite tokamak) — 110 GHz and 138 GHz gyrotrons providing ECRH for plasma startup and heating
DEMO (concept) — ECRH via high-efficiency gyrotrons is baselined for plasma startup and current drive, driving continued development toward >2 MW per unit

Key numbers

Output frequency (ITER): 170 GHz (Fundamental cyclotron resonance of electrons in ITER's 5.3 T plasma field; free-space wavelength 1.76 mm)
Output power per unit (ITER spec): 1 MW CW (Continuous-wave; pulse duration up to 3,600 s; 24 units × 1 MW = 24 MW total ECRH power)
Wall-plug efficiency (with depressed collector): ~50% (Single-stage depressed collector recovers 25–30% of spent-beam kinetic energy; without it, efficiency is ~35%)
Cathode voltage: ~−55 kV (Emitter potential; anode at ~+35 kV relative to cathode; beam energy ~80 keV total)
Superconducting solenoid field (cavity): ~6.7 T (At the interaction cavity; determines fundamental cyclotron frequency; NbTi or Nb3Sn coils at 4.2 K)
Relativistic factor (γ) of beam electrons: γ ≈ 1.05–1.10 (Weakly relativistic; the fractional relativistic mass shift is the physical origin of azimuthal phase bunching)
CVD diamond window diameter / thickness: 80 mm / ~1.5 mm (PACVD (plasma-assisted CVD) diamond; thermal conductivity ~1,500 W/m·K; only material viable at 1 MW, 170 GHz, 1,000 s)
ITER ECRH transmission line length: ~160 m (Corrugated HE11-mode waveguide from RF building to ITER tokamak building; each TL handles up to 1.2 MW input)

Surprising facts

An ITER gyrotron operating at 1 MW CW for 1,000 seconds delivers 1 GJ of microwave energy through a single CVD diamond disk 80 mm across and 1.5 mm thick — the same energy stored in roughly 240 kg of TNT, transmitted through a crystal grown in a laboratory.
The 170 GHz electromagnetic wave generated inside the gyrotron cavity has a free-space wavelength of 1.76 mm; yet the resonant cavity sustaining it operates in a TE31,11 transverse mode, meaning 31 azimuthal and 11 radial field oscillations fit across its cross-section — an extreme over-moded design made necessary because single-mode cavities at 1 MW would melt.
At 6.7 T and fundamental cyclotron resonance, 170 GHz corresponds to electrons that are only weakly relativistic (γ ≈ 1.06), but the 0.6% relativistic mass shift is the entire basis of the phase-bunching mechanism: without Einstein’s E = γmc², the gyrotron would not work.
ITER’s 24-gyrotron ECRH system will be able to locate and heat a region of plasma only a few centimetres wide inside a torus 6 m in diameter — at a distance of 160 m from the gyrotrons, through a multi-mirror transmission line — with a beam-steering accuracy sufficient to suppress a magnetic island rotating at kilohertz frequencies.

Open questions

Parasitic mode competition threatens single-mode oscillation: In an overmoded gyrotron cavity — needed because single-mode cavities would be destroyed by ohmic heating at 1 MW — hundreds of competing transverse modes lie within the gain bandwidth of the electron beam. The operating mode (e.g. TE31,11) wins only because its interaction impedance with the beam slightly exceeds that of its neighbours. Small deviations in beam voltage, beam current, or magnetic field can cause the gyrotron to jump to an adjacent parasitic mode, abruptly changing its output frequency and directional pattern. This mode competition is not fully deterministic: even in carefully manufactured tubes, start-up into the correct mode requires a precisely choreographed sequence of voltage and magnetic field ramps that can require individual tuning for each physical tube.
CVD diamond transmits MW power with less loss than a pane of glass at visible light: At 170 GHz, the dielectric loss tangent of CVD diamond is approximately 3–10 × 10⁻⁵ — lower than fused silica by a factor of ~10 and lower than sapphire by a factor of ~100. This means a 1.5 mm thick CVD diamond disk transmits 1 MW of millimetre-wave power with less than 1% ohmic absorption. By comparison, a 1.5 mm thick pane of ordinary window glass at its best visible-light transmission wavelength absorbs roughly 1% of the incident power. The material that is nearly transparent to a megawatt of microwave energy is, gram for gram, one of the most thermally conductive and mechanically hard substances on Earth.

A short history

The theoretical basis of the electron cyclotron maser was independently predicted by Rudolf Schneider (Germany, 1959) and Aleksandr Gaponov (USSR, 1959), who recognised that relativistic azimuthal bunching of gyrating electrons would produce coherent cyclotron emission. The first experimental demonstration was by James Hirshfield and James Wachtel at Yale in 1964, who observed maser oscillation from a 9.4 GHz electron beam in a magnetic field (Phys. Rev. Lett. 12, 533, 1964). Practical gyrotron development was driven primarily by Soviet groups at the Institute of Applied Physics in Gorky (Nizhny Novgorod), led by Gaponov-Grekhov and Grigory Nusinovich, who progressively increased output power and frequency through the 1970s and 1980s. The term 'gyrotron' was coined in the USSR; Western development accelerated through fusion programmes at MIT (Francis Bitter Magnet Lab), General Atomics, and European laboratories from the late 1970s. The 1 MW / 170 GHz milestone required by ITER was first demonstrated by European and Japanese tube manufacturers (Thales, TOSHIBA, CPI) in the 2000s after a decade-long development programme. The ITER ECH&CD system was formally specified in the ITER Technical Basis (2002), with final gyrotron deliveries ongoing into the 2020s. Manfred Thumm (KIT, Germany) has documented the state of the art in his biennial review articles in the Journal of Infrared, Millimeter, and Terahertz Waves.

Sources

Hirshfield J.L. & Wachtel J.M. — 'Electron cyclotron maser', Phys. Rev. Lett. 12, 533–536 (1964). DOI: 10.1103/PhysRevLett.12.533 (https://link.aps.org/doi/10.1103/PhysRevLett.12.533)
Nusinovich G.S. — 'Introduction to the Physics of Gyrotrons' (Johns Hopkins University Press, 2004). ISBN: 978-0-8018-7921-5 (https://jhupbooks.press.jhu.edu/title/introduction-physics-gyrotrons)
Thumm M. — 'State-of-the-art of high-power gyro-devices and free electron masers', J. Infrared Millimeter Terahertz Waves 41, 1–98 (2020). DOI: 10.1007/s10762-019-00641-w (https://doi.org/10.1007/s10762-019-00641-w)
ITER Organization — 'External heating systems: Electron cyclotron resonance heating (ECRH)', iter.org (https://www.iter.org/machine/supporting-systems/external-heating-systems)
Henderson M.A. et al. — 'An overview of the ITER EC H&CD system and functional requirements', Fusion Sci. Technol. 53(1), 139–163 (2008); also IAEA ITR/P1-10 (https://www-pub.iaea.org/mtcd/meetings/pdfplus/2010/cn180/cn180_papers/itr_p1-10.pdf)
Dammertz G. et al. — 'Design of a RF window for 170 GHz, 1 MW gyrotron for ECRH', Infrared Phys. Technol. 55, 14–18 (2012). DOI: 10.1016/j.infrared.2011.07.002 [CVD diamond window] (https://www.sciencedirect.com/science/article/abs/pii/S1350449511001228)
Electron Cyclotron Maser review — Chu K.R., 'The electron cyclotron maser', Rev. Mod. Phys. 76, 489–540 (2004). DOI: 10.1103/RevModPhys.76.489 (https://link.aps.org/doi/10.1103/RevModPhys.76.489)