What is thermonuclear fusion?
Thermonuclear fusion is one of the most promising alternatives to satisfy the need in electrical energy of humanity for the centuries to come. It is also one of the most challenging technologies to master, and nobody expects an industrial realization before the mid-century. ITER stands for International Thermonuclear Reactor and is a key milestone towards better understanding and mastering of some (but not all) technological bricks, before designing the prototype of an electrical plant. Its initial acceptance as an international project between United States, Europe (through EURATOM), Japan and Russia (then USSR) goes back to 1988, and the project was officially launched in June 2005 with Cadarache in the south of France as its location. Since then, India, China and South Korea have also joined the project. ITER is not the only thermonuclear reactor launched in the last decade; others exist such as CFETR in China, JT60SA in Japan, or W7X in Germany, where another type of reactor – a stellerator instead of a tokamak like ITER – is also tested.
Thales Microwave and Imaging Sub-Systems has been involved in the development of thermonuclear fusion reactors since the 1970s. Many of our power tubes have been developed for this purpose, such as tetrodes (DIII-D, JET, WEST (formerly TORE SUPRA)), klystrons (JET, WEST) and gyrotrons (W7X, WEST, TCV, in partnership with KIT and EPFL). MIS’ involvement in the ITER project extends from microwave tubes (ICRH and ECRH) to other kinds of heating systems (NBI). Besides, MIS is involved in projects running in parallel to new reactors, the main one being IFMIF (not addressed in the document). More globally, supporting high power technologies for fusion is a major ambition for MIS, which is also a partner of CEA on the LMJ (Laser Mega Joule) project. This white paper sets the trends of heating and current drive power technologies in the scope of tokamaks and stellerators developments.
Thermonuclear fusion and tokamaks
Reactors based on thermonuclear fusion retrieve the mechanical energy released when two very light elements (hydrogen isotopes: deuterium and tritium) combine to generate one helium nuclei. The total mass of output constituents (helium + neutron) is lower than the addition of the mass of the light nuclei (deuterium + tritium), and this difference is available in form of kinetic energy which could be used to heat up water and therefore could be transformed into electrical power, as in every thermal reactor. The same physical phenomena is at stake in a star like the sun, the main difference being that deuterium-tritium (D-T) fusion is used instead of proton-proton (p-p), because the probability (or cross section) of fusion is much higher, leading to a much higher efficiency. The D-T fusion reaction can be summarized with the following equation (between brackets are the kinetic energies of the elements):
D (~20-30keV) + T (~20-30keV) _ 4He (3.52MeV) + n (14.06MeV)
A gas of deuterium and tritium is first ionized to form a plasma which is heated up to 100 - 150 million degrees (corresponding to a kinetic energy of 20 - 30keV for deuterium and tritium nuclei), vs. 20 million degrees which is the estimated temperature inside the sun; these 20keV allow D-T fusion and correspond approximately to a speed of 600km/s for ions, and 40000km/s for electrons. At this temperature range, deuterium and tritium nuclei merge to generate helium (with a kinetic energy of 3.5MeV) and neutron (with a kinetic energy of 14MeV). Part of neutron energy can be recovered and transformed into electricity, while the helium nucleus is being kept inside the plasma and cooled down (and therefore it helps to maintain the plasma temperature), before being extracted from the reactor through a complex device called divertor. As a rule of thumb, 0.4g of tritium produces 2GJ (two billion of Joules) of energy. Tritium is an unstable isotope (lifetime 12.3 years) and cannot be found in nature, but it can be synthesized inside the reactor walls by colliding neutrons with lithium elements. Therefore, one can say that 1g of deuterium plus 10g of lithium will produce 30MWh, the same amount of energy that is achievable with 500g of uranium or with 10 tons of coal.
Another difference with stellar thermonuclear process is the way the plasma is confined. There are three solutions to confine plasmas at such temperature: either by gravitation (like in a star, of course inapplicable here), by inertia (like in the laser mega-joule project), or by magnetic confinement, which is the technology used in reactors: the plasma is confined into a toroidal volume by a 2 to 8 teslas toroidal magnetic field. This magnetic confinement, together with stability criteria against radiation perturbations (so-called Greenwald criteria), imposes a maximum on the plasma density n, slightly above 1020 nuclei/m3, 300 000 times less than the atmosphere density.
Due to inhomogeneity in the magnetic field (which decreases while approaching the center of the torus), a plasma only confined by a toroidal field will be unstable. An additional (ten times weaker) poloidal magnetic field is required to maintain its homogeneity and local neutrality.
The usual way to implement the poloidal field is to generate it in an inductive way through a huge toroidal current (around 17MA for ITER) inside the plasma. In this case, reactors work like mega transformers, the toroidal plasma being at the secondary. It is how tokamaks work (for Toroidalnaya Kamera s Magnitnaya Katushka).
Like for any transformer, the toroidal current (and therefore the poloidal field) is pulsed. So to get a steady-state production of electricity, an additional non-inductive toroidal current is mandatory. Stellerators, which are the main alternative to tokamaks, does not require toroidal current.
The final objective is to generate in the reactor significantly more power that must be injected to ignite the plasma, maintain it at the working temperature, and sustain the toroidal current. The ratio between power released through the fusion process (Pf) and injected power (Pi) is the Power Amplification Factor, noted Q. Q is the most popular factor of merit of a tokamak, and it is fixed by three parameters, which are the plasma density and temperature (n and T), and the plasma relaxation time t; this relationship is known as the Lawson criteria.
As n and T are more or less given, the only parameter which could be enhanced for future reactors is t, which is the ratio between the mechanical energy stored inside the plasma and the power lost through the divertor. To increase Q, one has to increase t, and a scaling law states that the only solutions to do so are:
- To increase the radius of the vacuum vessel, as the losses are proportional to the plasma envelop, and the stored energy is proportional to its volume.
- To increase the value of the toroidal field, to cope with higher toroidal currents (today around 5T in the core of the plasma, the main limitation coming from the Nb3Sn superconducting coils).
Q=1 is called the breakeven. The quest for Q=1 was launched in the 1980s at TFTR in Princeton, and a full generation of reactors has been designed to reach this goal (DIII-D in US, JET in UK, TORESUPRA in France, JT60U in Japan), with a radius from 1.6 to 3.4m, and a toroidal current above 2MA.
As a result, relaxation times between 0.1 and 1s have been recorded in the 1990s (the record being at JET with 1s).
ITER is designed with a radius of 6.2m, and a relaxation time of 6s is expected for Q ~ 10. CFETR in China, a reactor of the same generation, is designed with a radius close to 5m.
Q =1.25 has been reached in 1996 in Japan (JT60U at Naka, with the old deuterium-deuterium fusion process). Knowing that part of the power generated by the plasma is ultimately lost through various other physical processes (bremsstrahlung radiation of electrons, etc.), a minimum value of Q = 2 is required to guarantee a self-regeneration of the plasma.
To be able to produce electrical energy (by taking into account additional losses outside the reactor), one should get Q = 10 : it is one of the objectives of ITER, with an output power of the plasma (Pf) close to 500MW during 5 to 8 minutes – the current worldwide record being owned by TORE SUPRA with a 6’30’’ plasma, but the output power was 200 times less than ITER’s target. Another objective is to demonstrate Q = 5 in a steady-state mode. This is why the need of external power sources is estimated at around 100MW for phase 1, split between three technologies (NBI, ECRH and ICRH), as we will see.
Later on, an industrial reactor such as DEMO (foreseen successor of ITER) will require Q = 40 at least; DEMO should be the first reactor where the plasma output power (2000MW) will be ultimately transformed into electrical power.
External power systems are mandatory to ignite the plasma and to reach the required temperature to start fusion operation. For future industrial reactors, the fusion will generate enough power to be self-capable to sustain the plasma temperature, but for reactors under development, it is not possible. Four technologies are available and actually used to heat up the plasma.
A fifth way does exist, as the plasma toroidal current is directly generating plasma heating, due to Ohm’s law. But this ohmic process is very inefficient above a temperature of 30MK, as the plasma resistance falls down when temperature increases.
Four heating technologies may appear oversized for just heating plasma, but current reactors are not industrial plants and require flexibility to fully master physical processes still at the R&D stage. Besides, and maybe more importantly, external sources are also used:
- To drive a non-inductive current in the plasma (so-called “current drive”), to mitigate the “transformer effect” of the tokamak.
- To stabilize magneto-hydro-dynamic (MHD) plasma instabilities.
Heating systems for a thermonuclear reactor
The first heating technology is known as neutral beam injection (NBI). Injection of supra-thermal neutral particles (deuterium) will effectively heat up the plasma through a succession of collisions with ions and electrons. NBI is a highly reliable heating technology, which also favors on-axis current drive and plasma rotation, with a positive impact on plasma mobility. A deuterium gas is first ionized in a dedicated plasma chamber, and ions are then accelerated by an electrostatic field. After reaching the targeted kinetic energy, ions are neutralized and injected in the tokomak vessel with a velocity parallel to the plasma current. The neutralization is mandatory both to respect the plasma neutrality and to be able to inject particles through the magnetic confinement wall.
For ITER, 16MW should be injected per equatorial port. 2 equatorial ports will be equipped in the first phase of the project, and a third port is scheduled in phase 2; an additional pulsed NBI (called DNB for Diagnostic Neutral Beam) is also required for diagnosis purposes. 16MW means 16A / port, with 1MeV deuterium atoms.
Previous reactors were using positive ions, but were limited to 100keV ions. As the required kinetic energy for deuterium atoms in large reactors such as ITER is around 1MeV to get a good penetration of the plasma, and because high energy positive ions cannot be neutralized, the only way to inject atoms at such energy is to use negative species.
This concept has been tested in Japan on JT60 with arc sources, but the negative ion source designed to be implemented in ITER, called SPIDER (Source for the Production of Ions of Deuterium Extracted from a Radio Frequency), is new and derived from a device (named ELISE) previously studied at MPI Garching (Germany). So the ITER NBI technology will be first demonstrated by the consortium RFX at Padova, Italy, in the frame of the Neutral Beam Test Facility (NBTF), before being installed on the ITER tokamak.
Two test benches will be designed, manufactured and tested. The first one, SPIDER, demonstrates the production of a high flux (40A at 100keV) of negative deuterium atoms. Designed by RFX and manufactured under Thales supervision, SPIDER is composed of eight cylindrical plasma chambers, each one measuring 20cm x 30cm diameter. The deuterium gas is ionized by a RF field (1MHz) inside the chambers, and a set of four grids extracts the negative ions (and free electrons) and accelerates them up to 100keV; the free electrons are then selectively deflected with a magnetic field. The eight cylindrical vessels are made of copper coated with molybdenum (to protect them against erosion), and an atomic cesium layer pull down the work function of the metal, as the vessels' walls also act as cathodes to supply the plasma with new electrons.
SPIDER test bench has been transferred to Padova by end of 2017 and will be tested in 2018 and 2019; it will be followed by the second part of NBTF, called MITICA (Megavolt ITER Injector & Concept Advancement), itself composed of an electrostatic five-stage accelerator (to reach 1MeV) and by others BLC (Beam Line Components): a neutralizer, a residual ion dump to split ions from neutral atoms, and a removable calorimeter for diagnosis purpose.
Heating systems for a thermonuclear reactor: ICRH
Other heating technologies are based on microwaves. Their principles are similar to microwave ovens, an RF wave propagates inside a media (gas, liquid, solid) which is characterized by a resonance frequency (for food, this resonance is at 2.45GHz and corresponds to vibrations to water molecules). By exciting this resonance, the RF wave transfers its electromagnetic energy to (water) molecules, which accelerate (or vibrate). The resonance is associated with a collisional process, which transforms the kinetic energy of the accelerated molecules into heat.
The same is true inside a thermonuclear plasma, but with other kinds of resonances, gyromagnetic by nature: each charged particle surrounded by a magnetic field B is circling around field lines with a propulsion equal to qB/m, m being the mass of the particle and its q charge. For a magnetic field of a few teslas, the gyromagnetic resonance is in the range of a few tens of MHz for light ions such as hydrogen isotopes, and between 100 and 200GHz for electrons, depending on the magnetic field.
There is another difference with traditional microwave heating: as plasmas are composed of free ions and electrons, they are conductive, and electromagnetic waves are reflected by conductive media. They do not propagate, except in an evanescent way (and with some exceptions in plasma, as we will see). Fortunately, as for every conductor, there is a cutoff frequency above which electrons inertia forbid them to move in phase with the RF wave: the conductivity falls down and the media becomes transparent to electromagnetic waves, as no exchange of energy is possible anymore. This cutoff frequency is known as the electron plasma frequency ƒp of the media, and is proportional to the square root of the plasma density n (typically, ƒp = 9.√n).
For a metal, where n is very high, this cutoff happens in the ultraviolet range. For a thermonuclear plasma, n is close to 1020/m3 and ƒp is in the range of 90GHz. Ions gyromagnetic resonances (qB/m) are far below the plasma cutoff because of the mass of ions (a few GeV), and electron gyromagnetic resonance is slightly above (the mass of electrons is only 511keV). These background elements will help understand how RF plasma heating technologies work. But it should be remembered that despite the major concern that heating up the plasma is, two other topics are just as key and impact the design of the RF systems: driving a continuous plasma toroidal current (current drive), and stabilizing plasma instabilities, which rely on other physical process.
RF heating of ions within the plasma is called ICRH (for "Ion Cyclotran Radiofrequency Heating"). ICRH involves both ions and electrons (both species are heat up) and is based on complex plasma physics.
Electromagnetic waves with frequencies matching the gyromagnetic resonance of ions (between 30 and 70MHz) show in theory a high coupling with the plasma. But as these waves are far below the plasma cutoff frequencies (90 GHz), only evanescent waves should enter the plasma. Fortunately, in a tokamak, due to the strong toroidal magnetic field, the electrical conductivity is highly anisotropic and under specific polarization conditions, electromagnetic waves can propagate radially. Such radial waves are called “Fast Alfvén magneto-static Waves (FW)” and they can access to the plasma core, with two cutoff frequency linked to the plasma density: they are evanescent where the plasma density is too low (i.e. close to the wall of the vessel) and eventually where it is too high. By putting the ICRH antennas very close to the plasma, it is possible to excite these fast radial waves which will reach the plasma core where they will propagate. But ironically, they cannot couple directly with ions, even at the cyclotron frequency. It is because at the gyromagnetic resonance, FW show a circular right polarization, opposite to ions which are turning left! No transfer of energy is possible.
The problem can be bypassed in different ways: either by exciting harmonics of the ions cyclotron frequency, or by taking advantage of the fact that there are more than one ion species in the plasma. And here two cases are possible: if one of the species is seldom (for example a minority specie composed of traces of 3He), one can demonstrate the existence of a hybrid resonance (involving both species) which is very close to the cyclotronique frequency of the minority specie; both resonances are coupled, and by exciting the hybrid resonance, which is possible, this reactive energy will transfer itself to the minority specie cyclotronique resonance, giving birth to supra-thermal particles which give up their energy to the plasma by multiple collisions. This very efficient mode of is called the Minority Species mode (MH). The other case takes advantage of having two majority species. In this case, under certain conditions, FW are partially converted to shorter wavelength modes known as "Bernstein waves", strongly absorbed by electrons with a positive impact both on plasma heating and on current drive (approximately 20kA/MW for ITER). It is the Conversion Mode (MC).
All in all, knowing that ions mass range between 2 and 5GeV, that the magnetic field is not homogeneous and that the current drive excitation process is wide band, the ICRH excitation must also be wide band, between typically 30 and 70MHz.
Total required RF power is in the range of 10-20MW. This very high power is supplied by tetrodes coupled with microwave cavities, which can be tuned over the full operating bandwidth. The power sources work in CW mode and must sustain very high VSWR (Voltage Standing Wave Ratio), the plasma impedance being the subject of very fast evolution (approximately 1Hz) between a resistive load and a short circuit, due in particular to ELM instabilities (see below). ICRH technology is also useful to stabilize some plasma instabilities such as internal torsion modes ("saw teeth instabilities").
For ITER, 20MW CW should be injected in the plasma per equatorial port (leading to 9MW/m² at antenna level). 1 equatorial port will be equipped in the first phase of the project, and a second port will be introduced in the second phase. Such power corresponds to 9.2MW/m² at the boundary of the plasma. At source level, and taking into account the various losses in the RF transport system, it represents 24MW of generated RF power, spread between nine 2.5MW units. The cavities coupled to the sources must provide power at various different frequencies between 35 and 65MHz.
RF power sources working in this frequency range are tetrodes: the principle of these vacuum electron devices is the same than triodes (the current extracted from the cathode is modulated by the grid and the electrons are accelerated by the high voltage between anode and grid, generating the high RF power), but to screen the cathode field from the anode voltage (RF modulated), a second grid G2 is added. Tetrodes can produce up to typically 1MW CW in normal VSWR conditions. The RF wave propagates between G2 and anode, and RF losses heat up the G2 grid, by nature badly cooled. For this reason, the grid is manufactured using a very specific pyrolytic graphite technology. The 1MW output power is nevertheless a limit for traditional tetrodes, and THALES proposes a specific design, called diacrode, where the cathode is twice higher (l/2 instead of l/4, where l is the RF wavelength): up to 1.5MW output power has been demonstrated with TH628 diacrodes, with VSWR 2.0. The first demonstrator has been supplied in 2016 to ITER India organization, and the next step will be to demonstrate the coupling of two diacrodes to reach the required 2.5 MW in the same VSWR conditions.
The microwave power is transported from the source to the tokamak vessel through coaxial waveguides, and is injected into the plasma by using inductive antennas located very close to the plasma. 4MW are lost between the sources and the antenna, partly in the waveguides, partly in the windows which isolate the tokamak vacuum vessel from the outside. The antennas radiate 10MW each and are composed of inductive loops (strap antenna array) protected by Faraday shields in order to select the fast wave mode.
Heating systems for a thermonuclear reactor: ECRH & ECCD
RF heating of electrons within a thermonuclear plasma is called ECRH (for Electron Cyclotron Radiofrequency Heating). This technology heats up free electrons, which transfer their energy to the plasma by thermal coupling between electrons and ions. As the electron cyclotronique (EC) waves are slightly above the plasma cutoff frequency, they propagate almost like in free space, and when they reach the region of the plasma where the gyromagnetic frequency fits the RF frequency (with some spread due to Doppler effect), they are absorbed on a thickness of a few centimeters: ECRH heating is highly localized. This technology is complex from a technological perspective, but the interaction process is simpler and better understood than ICRH, and simulations are also more accurate. For this reason, ECRH is now widely used in almost all tokamaks.
Besides being very efficient in plasma heating, EC waves can couple with plasmas in other ways which generate on-axis and off-axis current drive (ECCD). On-axis current is key to reach a steady-state electricity production, and off-axis helps to stabilize locally plasma instabilities, such as:
- Neoclassical Tearing Modes (NTM) or “magnetic islands”, linked to plasma current gradients (the ECRH frequency is a tradeoff between plasma heating and NTM stabilization efficiency),
- Edge Localized instability Modes (ELM), due to pressure gradients at the edge of the plasma which are very significant in the widely used (because very efficient) confinement mode of thermonuclear plasmas, called H-mode.
Current drive can accommodate a wide range for RF frequencies (typically from 160 to 230GHz for a toroidal magnetic field around 5T). As of today, a single frequency is used, like in ITER where 170GHz corresponds to gyromagnetic resonance in the core of the plasma (B = 5,3T), and is compatible with current drive. But more and more systems will require multi-frequency ECRH, to deal with inhomogeneity in the magnetic field, or to get more flexibility in current drive injection.
In ITER, RF sources for ECRH are supplying 20MW CW to one equatorial port, but the power could be rerouted towards four upper ports, mostly for NTM stabilization (one additional equatorial port is scheduled for phase 2). The output power is provided by 24 “1MW class” oscillators called gyrotrons.
Supplying 1MW CW at 170GHz is out of reach by far for any traditional RF source, and gyrotrons are the only potential candidates. These microwave tubes are based on a complex interaction process between an annular electron beam and RF modes into a microwave cavity (the so-called ECRM (Electron Cyclotron Resonant Maser) effect, which uses the relativistic behavior of electrons submitted to high accelerating electrical field (80-100kV) and a high magnetic field).
The RF wave is generated on a very high mode (TE32,9 for TH1509 designed for ITER) to limit RF losses on the cavity walls of the tube, and is then transformed into a high purity Gaussian mode (HE11) which propagates in corrugated guides for ITER; at these frequencies, traditional waveguide propagation cannot be considered, as ohmic losses on the walls of the guide would melt them. Another issue is the design and manufacturing of low losses diamond windows, which isolate the vacuum zones (the inner parts of the tubes, and the tokamak vessel) from the zone exposed to air.
In the ECRH case, the RF power is injected in the tokamak vessel by using complex steerable (±12° toroidal for the equatorial port, ±5° poloidal for the upper ports) antennas called launchers, capable to focus the RF power towards specific locations of the plasma through mechanical mirrors. For example, the current drive efficiency is highly dependent on the beam angle, which itself is correlated to the RF frequency. The huge springs which steer the mirrors are made of stainless steel pipes filled with helium under high pressure (200 bars typical), and the motion of the mirrors is controlled through the helium pressure. As steering is not sufficiently fast to inject RF power in real time to stabilize plasma instabilities, another device allows switching on and off the RF beam, one solution being to use a diplexer and to detune the frequency of the gyrotrons by 10-20MHz.
Two additional gyrotrons working at 120GHz are also required. These gyrotrons, working in short pulses, will be used to startup operation and do not need steerable launchers.
Heating systems for a thermonuclear reactor: LHCD
This last "heating" technology is only required for tokamaks (it is not useful in stellerators). By design, a tokamak works in pulsed mode and a solution is yet to be found to have them work in a steady-state, i.e. to inject a non-inductive toroidal current. This "current drive" consists of a RF wave propagating into the plasma, which transfers its momentum to the electrons: under specific conditions (called Landau resonance), the group velocity of the RF wave reaches zero and its energy is converted into electron kinetic energy in a direction parallel to the toroidal magnetic field. Supra-thermal electrons with velocity close to electromagnetic wave velocity "surf" on the wave, causing a net electrical current to appear as there are no more electrons accelerated than slowed down. As already explained, ECRH and to a less extent ICRH can be tuned in a current drive mode, but they lack efficiency. The Landau effect is much more efficient when applied to lower hybrid modes – “hybrid” meaning in this case that the excited resonance involves coupled ions and electrons.
Lower hybrid modes are excited between 3 and 5GHz (slightly above the ions plasma frequency), depending on plasma density and ion selectivity. The corresponding current drive technology is known as LHCD (for Lower Hybrid Current Drive). This frequency is far below the electron plasma frequency (approx. 90GHz), and for this reason lower hybrid modes cannot reach the core of the plasma: they generate off-axis current drive. On the other side, the H-mode confinement reduces plasma losses and increases the relaxation time τ but they are characterized by a sharp pressure gradient at the edge of the plasma, which favors plasma instabilities. This sharp gradient does not help the coupling of lower hybrid mode, but coupled LHCD helps to control the profile of current drive, and to shape the transport barriers of the H-mode confinement, with a positive impact on plasma stability. LHCD also helps reducing saw teeth instabilities.
Lower hybrid waves could also, in principle, be used to heat up the ions of plasma, but in a rather inefficient way as these waves cannot reach the core of the plasma. As a consequence, they are usually injected to optimize the current drive to the expense of plasma heating. Non inductive current drive up to 2 to 3 MA has been demonstrated with LHCD, approximately ten times more than what would have been possible with ECRH.
LHCD is not considered for ITER phase one (it is scheduled for the second phase of the project), but it is already implemented in some famous tokamak such as TORE SUPRA (aka WEST) in France, JET in UK (hotter at 3.6 GHz), EAST in China (at 2.45 GHz and 4.6 GHz), JT60U in Japan (at 2GHz) and KSTAR in Korea (at 5GHz). The requirement for ITER phase 2, if confirmed, will be 20MW CW per port, provided by 0.5MW klystrons (coupled by pairs) working around 5GHz. As far as THALES is concerned, high power CW klystrons state-of-the-art is 0.7MW CW at 3.6GHz (TH2103C); these klystrons equip the WEST experiment at Cadarache, and a lower power version of this tube equipped the JET experiment.
The coupling of lower hybrid waves to the plasma is achieved by antennas composed of alternate passive-active multi-junction (PAM) waveguides called grills, stacked in the poloidal direction. This phased array antenna controls the wave-front of the lower hybrid waves, to synthesize a travelling slow wave propagating in the toroidal direction. Like for ICRH antennas, LHCD antennas are positioned close to the plasma.
Conclusion
Starting from the thermonuclear fusion international road-map, which will take shape in the following decade through the first plasma in ITER (targeted for 2025), but also in JT60SA, CFETR and others, this white paper highlights the crucial importance of heating and current drive system technologies. These systems are diversified, both by nature and by their performances (frequencies, power, etc.). For each of them, Thales' ambition is to be a primary partner for national and international Agencies, first with the design and production of power sources (diacrodes, gyrotrons, ions sources or klystrons), and also for other parts of the transport line and diagnosis of heating and current drive systems. Thales relies on its industrial mastering of technologies based on vacuum and high voltages, electron beams, microwave, welding and brazing, ceramics and other dielectrics. These technologies, and the associated design competencies, are the output of decades of investments in developing and producing microwave tubes for defense, space, industry and large scientific infrastructures such as accelerators, which guarantee their industrial availability in the decades to come.
Thales has been investing in technologies and products for fusion reactors for 40 years, starting in the 1970s with L-band klystrons developed for CEA and MPI Garching, followed by gyrotrons in cooperation with well-known institutes such as KIT in Karlsruhe or EPFL in Lausanne, tetrodes and diacrodes, and more recently plasma sources for NBI. Maybe another 40 years will be necessary before reaching the industrial maturity and producing electricity. Long term partnerships are then mandatory, and Thales is to be considered as one of these long term reliable partners by the worldwide fusion community.