ArticlePDF Available

The Gyrotron at 50: Historical Overview


Abstract and Figures

Gyrotrons form a specific group of devices in the class of fast-wave vacuum electronic sources of coherent electromagnetic wave radiation known as electron cyclotron masers (ECMs) or cyclotron resonance masers (CRMs). The operation of CRMs is based on the cyclotron maser instability which originates from the relativistic dependence of the electron cyclotron frequency on the electron energy. This relativistic effect can be pronounced even at low voltages when the electron kinetic energy is small in comparison with the rest energy. The free energy for generation of electromagnetic (EM) waves is the energy of electron gyration in an external magnetic field. As in any fast-wave device, the EM field in a gyrotron interaction space is not localized near a circuit wall (like in slow-wave devices), but can occupy large volumes. Due to possibilities of using various methods of mode selection (electrodynamical and electronic ones), gyrotrons can operate in very high order modes. Since the use of large, oversized cavities and waveguides reduces the role of ohmic wall losses and breakdown limitations, gyrotrons are capable of producing very high power radiation at millimeter and submillimeter wavelengths. The present review is restricted primarily by the description of the development and the present state-of-the-art of gyrotrons for controlled thermonuclear fusion plasma applications. The first gyrotron was invented, designed and tested in Gorky, USSR (now Nizhny Novgorod, Russia), in 1964.
Content may be subject to copyright.
The Gyrotron at 50: Historical Overview
Gregory S. Nusinovich &Manfred K. A. Thumm &
Michael I. Petelin
Received: 16 September 2013 / Accepted: 9 January 2014 /
Published online: 9 February 2014
#Springer Science+Business Media New York 2014
Abstract Gyrotrons form a specific group of devices in the class of fast-wave vacuum electronic
sources of coherent electromagnetic wave radiation known as electron cyclotron masers (ECMs)
or cyclotron resonance masers (CRMs). The operation of CRMs is based on the cyclotron maser
instability which originates from the relativistic dependence of the electron cyclotron frequency
on the electron energy. This relativistic effect can be pronounced even at low voltages when the
electron kinetic energy is small in comparison with the rest energy. The free energy for generation
of electromagnetic (EM) waves is the energy of electron gyration in an external magnetic field. As
in any fast-wave device, the EM field in a gyrotron interaction space is not localized near a circuit
wall (like in slow-wave devices), but can occupy large volumes. Due to possibilities of using
various methods of mode selection (electrodynamical and electronic ones), gyrotrons can operate
in very high order modes. Since the use of large, oversized cavities and waveguides reduces the
role of ohmic wall losses and breakdown limitations, gyrotrons are capable of producing very
high power radiation at millimeter and submillimeter wavelengths. The present review is
restricted primarily by the description of the development and the present state-of-the-art of
gyrotrons for controlled thermonuclear fusion plasma applications. The first gyrotron was
invented, designed and tested in Gorky, USSR (now Nizhny Novgorod, Russia), in 1964.
Keywords Gyrotron.Cyclotron resonancemaser.Electron cyclotronheating and currentdrive .
Millimeter waves
1 Introduction: The Gyrotron Fundamentals
Gyrotrons are members of a specific family of devices in the class of vacuum electronic sources of
coherent microwave radiation known as electron cyclotron masers (ECMs [1]) or cyclotron
J Infrared Milli Terahz Waves (2014) 35:325381
DOI 10.1007/s10762-014-0050-7
G. S. Nusinovich
Institute for Research in Electronics and Applied Physics, University of Maryland, College Park,
Maryland 20742-3511, USA
M. K. A. Thumm (*)
Karlsruhe Institute of Technology, IHM and IHE, Kaiserstr. 12, 76131 Karlsruhe, Germany
M. I. Petelin
Institute of Applied Physics, RAS, 603600 Nizhny Novgorod, Russia
resonance masers (CRMs [2]). The operation of electron cyclotron masers is based on the
cyclotron maser instability [36] which gives rise to electromagnetic perturbations in the motion
of electrons gyrating in an external (quasi-) uniform magnetic field. This instability originates
from the relativistic dependence of the electron cyclotron frequency on the electron energy.
According to the definition given in Ref. [7], the gyrotrons are cyclotron resonance masers in
which the interaction of helical electron beams with electromagnetic waves takes place in nearly
uniform waveguides near their cutoff frequencies. This definition is illustrated by Fig. 1
reproduced from Ref. [8], which shows the arrangement of the simplest gyrotron oscillator
configuration, a single-cavity device known as the gyromonotron.
In this device, a hollow electron beam are emitted from the cathode ring of the magnetron
injection gun (MIG) in the DC electric field caused by the voltage applied between the cathode
and the anode. Then the electrons move towards the interaction space (cavity) in the increasing
magnetic field, thus experiencing the adiabatic compression which increases their transverse
(orbital) momentum in accordance with the conservation law p
/B=const. In the cavity,
electrons gyrating with the cyclotron frequency Ω=eB/m
γinteract with an electromagnetic
(EM) field under the cyclotron resonance condition
In (1)ωand k
are the EM wave angular frequency and axial wavenumber, v
is the electron
axial velocity and sis the cyclotron resonance harmonic number (s=1, 2, ). After passing through
the cavity a spent electron beam moves in the decreasing magnetic field experiencing adiabatic
decompression and reaches the collector. The EM field excited in the cavity radiates through an
open end into an up-tapered output waveguide and propagates through the output window.
As follows from (1), the changes in the electron energy in the process of interaction with the
EM wave may cause axial bunching due to the changes in the electron axial velocity and orbital
bunching due to the dependence of the electron cyclotron frequency on the electron energy. As
also follows from (1), in the case of interaction with traveling waves the spread in electron axial
velocities can result in significant inhomogeneous Doppler broadening of the cyclotron reso-
nance band, which can lead to efficiency deterioration. To mitigate this effect, the gyrotrons
operate in waves with small k
s, i.e. near cutoff frequencies where the broadening is weak. This
operation is illustrated by the dispersion diagram shown in Fig. 2. Clearly, in such a case the
orbital electron bunching is dominant because the Doppler term in (1)issmall.
Fig. 1 Arrangement of a single-cavity gyrotron (also known as gyromonotron)
326 J Infrared Milli Terahz Waves (2014) 35:325381
One can distinguish three stages in the electron interaction with the EM field [7]: (a) energy
modulation, (b) orbital bunching, and (c) bunch deceleration. These stages are illustrated by
Fig. 3. During the first stage, a ring of electrons is displaced towards the region of the
accelerating field where v
E<0. During this stage, the energy of some electrons increases,
while that of others with different initial gyrophases decreases. On the average, the energy of
the ring increases so that the electrons absorb energy of the EM field. During the second stage,
the modulation of electron energies, which causes small changes in the electron cyclotron
frequency (relativistic effect of the dependence of the electron mass on its energy!), results in
the formation of an electron bunch. When the EM frequency slightly exceeds the electron
cyclotron frequency or its resonance harmonic (ω>sΩ
), a bunch is formed in the decelerating
phase of the EM field. (Note that this process of bunching may take place even in the absence
of the EM wave, e.g., in drift sections of multi-stage devices, i.e. it is inertial). Then, during the
third stage the bunch is decelerated so that the electrons transfer energy of their gyration to the
EM field.
After these introductory remarks explaining the physics of gyrotron operation, let us start
the main part of this review, which describes the history of the gyrotron development. Before
starting this description we would like to remind our readers that by now more than 2,000
papers have been published and presented on various aspects of gyrotron theory, codes,
experiments and applications by a large number (possibly, about one thousand) contributors.
Therefore, it is practically impossible to mention in this review everybody and everything.
Bearing this in mind, we would like to apologize for omitting some names and some results
which may look important in the eyes of some readers.
Fig. 2 Dispersion diagram: the parabola shows the dispersion curve of a smooth-wall waveguide with cutoff
frequency ω
, straight lines show the cyclotron wave beam line in the region of forward-wave (solid line,
positive k
) and backward-wave (dashed line, negative k
) interaction. The gyromonotron operation corresponds
to the intersection of the beam line with the waveguide dispersion curve near the bottom of the parabola
J Infrared Milli Terahz Waves (2014) 35:325381 327
Fig. 3 Ring of electrons gyrating
around one axis interacts with a
synchronous component of the EM
field. Dashed lines show the initial
distribution of electrons
328 J Infrared Milli Terahz Waves (2014) 35:325381
2 History of Cyclotron Resonance Masers
In principle, the prehistory of the cyclotron resonance masers can be dated back to the early
1920s[1,2,9]. An important turn happened in the late 1950s, when fundamental CRM
mechanisms were adhered with the relativistic dependence of the electron cyclotron frequency
on the electron energy. This profound approach [36] (photos of its pioneers R. Q. Twiss, J.
Schneider, A. V. Gaponov and V. V. Zheleznyakov are shown in Fig. 4)wasusedtoexplain
cyclotron instabilities of non-equilibrium magnetized plasmas as well as the performance of a
number of cyclotron resonance microwave amplifiers and oscillators driven by helical and
trochoidal electron beams.
Papers by Twiss [3] and V. V. Zheleznyakov [6] dealt with the linear theories developed for
the simplest theoretical models. In particular, Zheleznyakov [6] considered EM wave propa-
gation in a homogeneous flow of electrons moving in a static magnetic field and immersed into
a uniform dielectric medium; all electrons having the same longitudinal v
and the same
transverse v
velocities relative to the magnetic field. The wave propagating along the
magnetic field has the frequency and the wavenumber obeying the dispersion equation
In (2), nis the dielectric refractive index, ω
mis the beam plasma frequency (n
the electron beam density), β
/cis the electron orbital velocity normalized to the speed of
light cand Ω
is the initial electron cyclotron frequency. (Later, similar fourth-order dispersion
equations were derived by many authors for various, e.g. cylindrical and Cartesian, geome-
tries.) At low beam densities (ω
0), under the resonance condition (1), which for this model
can be rewritten as ω(1nβ
,Eq.(2) reduces to the cubic form
4ck c2k2
Let us recall that in this model the wave propagates along the magnetic field, so k
presence of electrons, this axial wavenumber can be represented as k
(1+ ν)wherek
±ωn/c. The small term νis due to the presence of electrons. When this term has an imaginary
part while the wave frequency is real, the wave grows along the magnetic field, which means a
convective instability. Zheleznyakov [6] also analyzed the absolute instability of the waves that
is the case when the axial wave number is real, but the wave frequency is complex; then the
wave field grows in time at any point.
Fig. 4 CRM founders (from left to right): Richard Quentin Twiss (1920-2005), Jürgen Schneider (1931-2012),
Andrey Victorovich Gaponov (1926) and Vladimir Vasilyevich Zheleznyakov (1931)
J Infrared Milli Terahz Waves (2014) 35:325381 329
The wave instability described by this equation (as well as by Eq. (2)) originates from two
above mentioned mechanisms of electron bunching: when the wave phase velocity v
smaller than the speed of light (n> 1 - slow wave) the axial bunching prevails; when the wave
propagates faster than the speed of light (n< 1 - fast wave) the orbital bunching caused by the
relativistic dependence of the electron mass on energy dominates. Finally, when this phase
velocity is equal to the speed of light, these two terms cancel each other and the last term in (2),
which is the same as the term in the RHS of (3), vanishes.
J. Schneider [4] used the simplest quantum mechanical approach to the mono-energetic
ensemble of electrons in a homogeneous external magnetic field. He showed that from the fact
that Landau levels for the energy of a relativistic electron in the external magnetic field are
non-equidistant (due to the relativistic dependence of the electron mass on energy) it follows
that coherent cyclotron radiation in an ensemble of such electrons is possible. Schneider
concluded his short paper [4] with an optimistic phrase: It does not appear unlikely that this
effect could be used for a new type of maser.
A. V. Gaponov [5,10] summarized both classical and quantum approaches to the stimulated
cyclotron radiation of relativistic electrons. He showed that the axial bunching is related to the
recoil effect of radiating electrons, while the orbital bunching is caused by the dependence of
the electron oscillation frequency on the electron energy. It was shown [10], first, that both
quantum and classical approaches yield the same results and, second, that, when electrons
gyrating in the external magnetic field interact with an EM wave propagating along this field
with the phase velocity equal to the speed of light, two mechanisms of electron bunching
cancel each other and, hence, the electron ensemble behaves as an ensemble of linear
oscillators. In a certain sense, this conclusion drawn for an ensemble of electrons anticipated
the autoresonance effect predicted for a single electron a few years later [11,12].
In [13] A. V. Gaponov reduced the dispersion equation for a traveling-wave CRM to a
cubic equation similar to the one derived by J. Pierce for a linear-beam traveling-wave tubes
[14]. In Ref. [13] all terms in the dispersion equation were derived for arbitrary configurations
that facilitated the use of the results in various experiments.
The nonlinear theory of CRMs was, first, developed by V. K. Yulpatov for the simplest
model of a CRM monotron. In his talk at the All-Union Conference on Microwave Electronics
in 1960 (Kharkov), a self-consistent interaction of a mono-energetic ensemble of gyrating
electrons with a monochromatic EM wave was described. It was shown that the interaction
efficiency can be at the level of several tens of percents. This theory agreed with results of the
first experiments with CRMs, in which all electrons move in crossed electric and magnetic
fields with the same drift velocity v
[15] along trochoidal trajectories (such CRMs
were named trochotrons). Later, this simple theory and theoretical progress made by many
other authors was summarized in 1967 in Ref. [2].
3 From Cyclotron Resonance Masers to Gyrotrons
The pioneering papers [36] explained a number of contemporary experiments [1620]and
encouraged further research in CRMs [2134] (A more complete list of references can be
found elsewhere [2]). During the first half of 1960s, as described in Refs. [7,35], the highest
powers (about 1 kW CW) and efficiencies (on the order of 10%) were achieved in experiments
with Ka-band CRMs utilizing trochoidal electron beams (trochotrons) [26]. CRMs with helical
electron beams were not so efficient (see, e.g., [35]) because the spread in electron axial
velocities inherent in such beams caused the inhomogeneous Doppler broadening of the
cyclotron resonance line defined by the condition (1).
330 J Infrared Milli Terahz Waves (2014) 35:325381
The solution of the axial velocity spread problem arrived as a trivial partial case from a
generalized based on the kinetic equation and the Einstein coefficients analysis of the
stability of homogeneous magnetized relativistic plasma [28]. Indeed, it is quite clear that,
&The electron distribution is monoenergetic and, hence, all electrons have the same
cyclotron frequency Ω,
&The wave propagates perpendicular to the static magnetic field (k
the cyclotron resonance condition (1) holds for all electrons independently on their pitch
factors α=v
. In this regard, it should be noticed that, while the initial spread in the pitch
ratios of electrons in helical beams can be quite large, the initial spread in electron energies is
very small (practically absent).
A corresponding practical method of eliminating (or, at least, mitigating) the inhomoge-
neous Doppler broadening is conceptually illustrated by Fig. 5: here electrons leaving an equi-
potential cathode on the left and guided by an external magnetic field enter an equi-potential
volume (on the right) where they interact with an RF field composed of waves propagating
perpendicular to the external magnetic field. (This concept is essentially the same as the
gyrotron arrangement shown in Fig. 1, which originated from it.)
The simplest theoretical model of this robust CRM represents a 2D mirror resonator
composed of two parallel metallic plates and driven by a homogeneous plasma flow, in which
electrons are proceeding along helical trajectories. Self-consistent structures and start currents
for the highest-Q quasi-optical modes of the CRM shown in Fig. 5were calculated in the late
1963 by introducing the mono-energetic plasma dielectric conductivity [28] into the structure
described, following to L. A. Weinstein [36], with the Wiener-Hopf factorization method. The
first gyrotron prototype designed with the use of this theory is shown in Fig. 6reproduced
from Ref. [8].
This gyrotron prototype was operating in the TE
-mode of a rectangular waveguide
excited near cutoff; the radiation was extracted via an open output cross-section and then
propagated as the TE
-wave through the uptapered rectangular output waveguide. An
electron beam was formed in a triode-type, three-electrode electron gun with scales of non-
uniformities of the static electric and magnetic fields that were much larger than an electron
gyro-radius. Therefore the electron motion could be treated as the gyration around an
adiabatically moving guiding center. The cathode was situated in the stray magnetic field of
the solenoid, so the electrons moving towards the interaction region experienced adiabatic
magnetic compression enhancing the energy associated with electron gyration. Results of the
first experiments with this device (6 W of the CW power at the fundamental cyclotron
Fig. 5 Configuration of CRM without Doppler broadening of the cyclotron resonance
J Infrared Milli Terahz Waves (2014) 35:325381 331
resonance in the X-band) were reported at the All-Union Electronics Conference in Moscow
(1964) in the presence of L. A. Weinstein who predicted a good future for this concept in spite
of the fact that at that time a CRM with a trochoidal electron beam produced 300 W CW.
Indeed, it was soon recognized that the CRMs with trochoidal electron beams (trochotrons)
are not very attractive for wavelength shortening. This conclusion followed from the fact that
the external magnetic field H
!0should be increased proportionally to the operating frequency,
while possibilities to simultaneously increase the external electric field E
!0are limited by
breakdown. This limitation leads to reduction of the axial velocities that, in turn, increases the
deteriorating role of the space charge effects on the beam quality.
In 1965, after increasing the beam voltage from 8 to 19 kV, a second harmonic 25 GHz
CRM utilizing a helical electron beam and operating in the TE
-mode of a circular cross-
section cavity delivered 190 W CW power, thus confirming the potentials of this concept [25].
It should be noticed that initially these devices did not have a specific name distinguishing
them from other varieties of CRMs. This step was done in 1966 in Saratov at the All-Union
Conference on Microwave Electronics where some experiments with fundamental and second
harmonic CRMs at the kilowatt CW power level were reported. There, V. T. Ovcharov advised
to introduce a specific designation for the new version of CRMs, and the name gyrotron
suggested by A. L. Goldenberg was unanimously adopted.
Practically, at the same time, it was recognized that the vacuum electron devices based on
the cyclotron maser instability have an important similarity with microwave tubes driven by
linear electron beams (O-type devices): in both types of devices the electron bunching leading
to the coherent radiation is inertial. Therefore for any device driven by a linear electron beam
Fig. 6 Layout of the first gyrotron (reproduced from Ref. [8])
332 J Infrared Milli Terahz Waves (2014) 35:325381
one can propose a CRM-counterpart. Such comparison of the key members of the two families
is shown in Table 1which was, first, presented at the same conference in Saratov in 1966.
Here, the first device in the second row represents the gyromonotron shown in Fig. 1.The
second device is a two-cavity gyroklystron amplifier, in which the first cavity serves for
modulating electron energies by the input signal. Then, this energy modulation causes the
orbital bunching shown in Fig. 3. This bunching has an inertial nature, i.e. it can proceed in the
absence of any wave field in the drift space. Finally, in the last cavity these bunches are
decelerated by the RF field, i.e. their kinetic energy is transformed into the energy of EM
oscillations. The three last devices shown schematically in Table 1are the forward-wave
amplifier (gyro-traveling-wave tube), the hybrid device comprising the input cavity with the
drift stage and the output traveling-wave section (twystron) and the backward-wave oscillator
(gyro-BWO), respectively. Strictly speaking, there is an important distinction between the
conventional linear-beam backward-wave oscillator (BWO) invented by R. Kompfner [37]
and the gyro-BWO. In the former device, the wave phase velocity v
and the wave
group velocity v
have opposite signs: the phase velocity is positive, i.e. the wave
phase propagates in the same direction as the streaming electrons, while the group velocity is
negative, i.e. the wave energy propagates backward forming together with the forward
streaming electrons an internal feedback loop resulting in the excitation of oscillations. On
the contrary, in the gyro-BWO, both the phase and the group velocities are negative.
In 1967, efficiencies of gyrotrons reached 50% at the fundamental cyclotron resonance and
approached 20% at the second cyclotron harmonic; also, the simplest two-cavity X-band
gyroklystron amplifier operating in the TE
-mode with about 70% efficiency was demon-
strated [35]. Combined methods of electrodynamical and electronic mode selection were
proposed for further power enhancement. The research was awarded with the State Prize of
the USSR.
Prior to moving forward with describing the progress in gyrotron development it makes
sense to, at least briefly, explain the issue of above mentioned mode selection, because this was
a key to success in escalation of gyrotron power. Gyrotrons, whose two main components are a
thin annular beam of electrons gyrating in the external magnetic field and a smooth-wall open
resonator, are very selective devices, which can operate in very high-order modes. First of all,
the selectivity is determined by the cyclotron resonance condition, because only the modes
whose frequencies obey this condition can strongly interact with EM fields. Typically,
electrons make 5-10 turns in the interaction space, so the corresponding cyclotron resonance
band, which consists of the regions of cyclotron absorption and reabsorption, can be on the
order of 10% of the cyclotron frequency. However, the region of the coherent reabsorption,
Tab le 1 Basic types of cyclotron resonance masers and their linear-beamcounterparts.
J Infrared Milli Terahz Waves (2014) 35:325381 333
which is the only region where coherent EM radiation is possible, is smaller than that of the
absorption, and its width is on the order of 1-2%. This means that only a small fraction of all
modes of an overmoded circuit can be excited by an electron beam. [Note that, as the voltage
increases, the relativistic cyclotron frequency decreases and, therefore, during the voltage rise,
this excitation windowmoves down in frequency because the change in the beam voltage
causes the change in the cyclotron frequency ΔΩ/Ω=e(ΔV
)]. This fact, as
well as a possibility to inject a thin annular electron beam in the strongest peak of the beam
coupling to a desired mode of a cylindrical waveguide or resonator, forms the basis for a so-
called electronic mode selection. There are also various possible methods of electrodynamical
mode selection. The first of them is based on the fact that the profile of slightly irregular open
waveguides, which serve as resonators, can be made in such a way that diffractive losses of
modes with one axial variation, which have the smallest axial wavenumber (and, hence, the
lowest group velocity), will be much smaller than those of modes with a larger axial index
q>1[38]. Thus, at moderate values of beam currents only the modes with q = 1 can be
excited. (In coaxial resonators, as will be discussed below, there are additional opportunities
for mode selection.)
In 1968, the development of trochotrons was stopped, and the Soviet Ministry of Electron-
ics started funding the research for higher power gyromonotrons and gyroklystrons; the
customer being represented by a commission led by M. B. Golant was inspecting researchers
every second year to accept the following results:
&A free-running gyrotron program (non-officially called O-Mega,forone Megawatt)
was aimed to reach one megawatt output power at any frequency and in any short (~ 10
μs) pulse. In 1968, an X-band gyrotron operating in the TE
-mode delivered 140 kW
pulsed power; in 1970, X-band pulsed gyrotrons (one of versions used a coaxial cavity)
delivered 400 kW pulses in the TE
and TE
modes; and, at last, in 1973, with using a
superconducting solenoid, the pulse power of a 42 GHz gyrotron operating in the TE
mode exceeded 1 MW [39].
&In 1971, Istokand the Radiophysical Research Institute (NIRFI) demonstrated a Ka-
band gyroklystron with 11 kW CW power in the TE
output mode. A pulsed Ka-band
gyroklystron operating in the TE
mode was developed. (After some upgrades, similar
0.5 MW pulse power gyroklystrons operated as high-power output amplifiers of the
phased array radar Ruza[40].)
The limitations on the length of this review paper do not allow us to describe the history of
the development of all types of gyro-devices for various applications. Therefore below we
limit our story primarily by the description of the progress in the development of gyrotrons for
controlled fusion plasma applications.
4 1970s: Beginning of the Development of Gyrotrons for Fusion Plasma Applications
We would like to start this Section from the notice that many results described here have been
already overviewed in Ref. [41].
From the late 1950s, the concept of controlled thermonuclear fusion reactors as possible
sources of energy was actively pursued in many countries around the globe. Various kinds of
reactor configurations, such as tokamaks, stellarators, open mirror machines etc were pro-
posed. One of the most attractive configurations has been the tokamak; this abbreviation stands
for the Russian words toroidal chamber (kamera) with magnetic coil (katushka). Initially, it
334 J Infrared Milli Terahz Waves (2014) 35:325381
was assumed that to heat up plasmas in tokamaks it is enough to use electric power and treat
the tokamak as a transformer in which a plasma propagating inside the torus plays the role of
the secondary winding which can be heated due to its finite conductivity (Ohmic heating).
Very soon, however, it was recognized that this Ohmic heating is efficient only at low plasma
temperatures. Thus, some additional, auxiliarymethods of heating should be used. The
methods of auxiliary plasma heating proposed in the beginning of the 1960s included the
heating by neutral particle beams and various methods of heating by RF power. The latter
methods were based on various resonances between the injected EM field and the plasma
components. Three dominant methods have been distinguished:
the method based on the cyclotron resonance between injected RF power and plasma ions
(so-called, ion cyclotron resonance heating or ICRH),
the method using the lower hybrid resonance (LHRH), and
the method based on the cyclotron resonance between injected EM waves and plasma
electrons (electron cyclotron resonance heating or ECRH).
These three methods required sources of EM power of very different frequency ranges: tens
up to hundreds of MHz for ICRH, several GHz for LHRH and tens up to hundreds of GHz for
ECRH. The sources of the first two frequency ranges were available (tetrodes for ICRH and
klystrons for LHRH), but there were obvious problems with using these methods for bulk
plasma heating (see, e.g. Ref. [42]). The ECRH method offered an easy access of the
millimeter-wave power to the plasma center, but there were no sources with sufficient power
level at millimeter wavelengths.
The situation was drastically changed when plasma physicists realized that such sources
exist. This event took place after the first experiment was carried out at Gorky Radiophysical
Research Institute (NIRFI) in 1970 by B. G. Eremin and A. G. Litvak [43]. In that experiment,
the self-focusing of electromagnetic waves generated by one of the first gyrotrons was
observed in plasma. Right after the first results of this breakthrough became known at the
Kurchatov Institute, V. V. Alikaev and his chief V. M. Glagolev visited Gorky and discussed
experimental results with Gaponov, Litvak, Petelin and Eremin. It was realized that the
millimeter-wave energy produced by such gyrotrons in sub-millisecond pulses can be suffi-
cient for substantial increase of electron temperature in such tokamaks as the TM-3 tokamak
existing by that time at the Kurchatov Institute in Moscow. (Note that later the interaction of V.
M. Glagolev with Gorkys physicists and, especially, with V. Yu. Trakhtenhertz led to the
development of a new concept of a cyclotron resonance maser with background plasma [44],
the concept which was later used for describing coherent radiation of the electrons trapped in
Van Allen belts in the magnetospheres of the Earth and Jupiter [45].)
Therefore, as soon as the gyrotron power and pulse duration reached the level sufficient for
substantial increase of electron temperature in small-scale tokamaks existing at that time, the
plasma physicists promptly started to use these gyrotrons for ECRH. This step was made in the
USSR where the gyrotrons reached the required level at the end of the 1960s. There were two
experiments carried out in parallel in Moscow and Leningrad (now St. Petersburg) on plasma
heating with gyrotrons in the early 1970s: one was done at the Kurchatov Institute by the
group led by V. V. Alikaev [46] and another one at the Ioffe Institute by the group led by V. E.
Golant [47]. Alikaevs group clearly identified the mechanism of millimeter-wave absorption
in plasma and demonstrated almost doubling of the electron temperature due to plasma heating
by gyrotron radiation: from approximately 250 eV to about 450-490 eV. Efficient heating was
observed not only at the fundamental, but also at the second harmonic of the electron cyclotron
frequency in the plasma. The gyrotron used in this experiment radiated about 40 kW power at
J Infrared Milli Terahz Waves (2014) 35:325381 335
the wavelength close to 1 cm in pulses of about 0.5 ms duration, i.e. the millimeter-wave
energy per pulse was about 20 Joules.
Since that time, the gyrotron development was actively supported by the State Committee
for Atomic Energy and, for a number of years, the group led by V. V. Alikaev was the world
leader in ECRH. (Note that in the first half of the 1970s the foreign plasma physicists could
not get any information about the millimeter-wave power sources used in these experiments.)
This gyrotron development was a result of active collaboration between academia and
industrial partners (company Salut, key persons S. D. Bogdanov and V. I. Kurbatov); the
head of this collaborative team was V. A. Flyagin.
By the mid-1970s the gyrotron power and the radiation frequency were practically
doubled: for example, in 1975 it was reported [48] that gyrotrons operating at 4.5 mm
wavelength produce 80 kW power in 0.6 ms pulses. In addition to the development of these
relatively long-pulse gyrotrons for immediate use in plasma experiments, the research group at
NIRFI (starting from 1978, the Institute of Applied Physics (IAP)) was actively studying
possibilities for further increase of the gyrotron power. It was clear that this raise of power
requires studying of such issues like efficiency optimization [49], possibilities of single-mode
operation in high-order modes where the competition between modes with close frequencies
may be an obstacle [50] and a search for the optimal start-up scenario, which would allow one
to initially excite the desired mode and then drive the gyrotron to the point of stable operation
in this mode with the highest efficiency [51]. To be able to describe the operation of high-
power gyrotrons more accurately, in addition to the previously developed nonlinear theory
based on the cold-cavity approximation of the axial distribution of the resonator field, the self-
consistent theory was developed [52] in which the effect of a high-current electron beam on the
resonator field structure was taken into account. For accurate description of gyrotron operation
at high current densities it was also important to include the role of space charge effects. The
theory describing these effects was developed in a series of papers [53].
It was also vitally important to develop electron-optical systems allowing for accurate
design of magnetron-type electron guns forming beams of electrons gyrating in external
guiding magnetic fields. It was desirable to form such beams with a large ratio of the
orbital-to-axial velocities and small velocity spread at high current densities [5456]. All these
theoretical efforts were taking place in parallel with active experimental studies of various
gyrotrons. Results of many of these experiments were published in the mid-1970, although
many of these results were obtained much earlier. Some of the important experimental
accomplishments were published in [39,5759]. In brief, Ref. [57]reportedthegeneration
of 1.5 kW power in the continuous-wave (CW) regime at 0.92 mm wavelength (i.e. the
radiation frequency was 326 GHz). This result was obtained in operation at the second
cyclotron harmonic. Also, in Ref. [59] the operation at the second cyclotron harmonic with
a very high electronic efficiency was reported: 40% in the CW and 43% in the pulsed regimes.
Note that Ref. [59] contained a figure showing the gyrotron arrangement (this figure is
reproduced here in Fig. 7) which greatly helped gyrotron designers in other countries to
facilitate their designs. In Ref. [39] the first attempts to operate in whispering gallery modes,
which are the modes having azimuthal indices much larger than the radial ones, were
described. In Ref. [39], also the possibility of improving the resonator mode selectivity by
using a coaxial insert was experimentally demonstrated.
The portfolioof gyrotron studies in Gorky by that time was extremely diversified. It
included not only gyrotron oscillators, but also gyroklystron amplifiers. Because of obvious
reasons not much was published about the development of gyroklystrons for quite a long time;
however, one can find in Ref. [35] that it was possible to realize the operation of a gyroklystron
with 70% electronic efficiency. That experiment was carried out in about 1970; the device
336 J Infrared Milli Terahz Waves (2014) 35:325381
operated in the X-band, and the CW power level was about one kW. Another step deserving
mentioning was the development of gyrotron oscillators with mechanically tunable frequency
[60]. This continuous frequency tuning was realized by several means: one of them was using
the resonators with axial slots. A gyrotron with such resonator was used for some studies in the
gas spectroscopy [61].
In the next series of experiments performed in the second half of the 1970s, the emphasis
was made on achieving the MW power level by operating in very high-order whispering
gallery modes. This led to demonstration of stable operation in the TE
-mode in a 100 GHz
gyrotron producing 1.1 MW power with 34% efficiency in less than 0.1 ms pulses [62]. At the
same time, long-pulse gyrotrons developed for the next tokamak T-10, achieved at 3.6 mm
wavelength, required by the tokamak magnetic field, the power level of 300 kW in pulses of
10-20 ms duration. The photo of a short-pulse, 1.1 MW, 100 GHz gyrotron is shown in Fig. 8
reproduced from Ref. [62]. By that time it was realized that it is very convenient to operate in
the whispering gallery modes with two radial variations, because the electron beam positioned
on the inner peak of such modes propagates far enough from the wall for avoiding the beam
interception and, at the same time, close enough to the wall to avoid significant potential
depression in short pulses.
It should be noted that long-pulse gyrotron operation in high-order modes required a certain
modification of the gyrotron geometry. In the case of a simple gyrotron arrangement sche-
matically shown in Fig. 1, it was necessary to rapidly uptaper the output waveguide in order to
provide an acceptable level of power density deposition at the collector in long-pulse or CW
regimes. Such a rapid variation of a waveguide wall radius would inevitably cause significant
Fig. 7 Arrangement of a second harmonic gyrotron (reproduced from Ref. [59])
Fig. 8 Photo of a 100 GHz, 1.1 MW gyrotron operating in the TE
-mode (reproduced from Ref. [62])
J Infrared Milli Terahz Waves (2014) 35:325381 337
transformation of an operating high-order mode into spurious modes with different radial
indices. That would greatly complicate the conversion of such radiation into a Gaussian wave
beam with linear polarization required for transmission of the millimeter-wave power to the
plasma chamber. Instead, it was proposed to transform the operating mode into such a wave
beam directly in the gyrotron output coupler. Such built-in quasi-optical mode converters were
proposed, designed and tested in the 1970s for various modes: for symmetric modes and
modes with one azimuthal variation in Ref. [63] and for whispering gallery modes in Ref. [64].
Schematics of these quasi-optical (QO) converters are shown in Fig. 9.
This success of the gyrotron development in the USSR had initiated a strong interest in
gyrotrons in many countries. In 1976 the gyrotron development was started in the USA, in
1978 it was started in France and Australia (in the latter case, it was planned to develop a
gyrotron for millimeter-wave spectroscopy [65]) and about the same time it was also started in
Japan and Peoples Republic of China. In USA the gyrotron development began practically at
the same time at Varian and at the NRL. The first gyrotron for plasma ECRH was developed at
NRL [66] and used for ECRH heating at the tokamak ISX-B in Oak Ridge (here ISXstands
for Impurity Study Experiments). First results were reported in Ref. [67]. The 35 GHz gyrotron
[66] produced up to 150 kW peak power in 10-20 ms pulses with 31 % efficiency. This
millimeter-wave energy per pulse was sufficient for increasing the electron plasma temperature
from 850 eV to 1250 eV [67]. It should be noted that the NRL gyrotron team led by V. L.
Granatstein was actively working not only on experiments, in which an active role was played
by M. Read [66], but also on the theory of gyrotrons and, more broadly, of cyclotron resonance
masers. Among the people actively working on the theory, at least several names must be
mentioned: P. Sprangle [68] , Y. Y. L a u [ 69], K. R. Chu [70], and W. Manheimer [71]. An
Fig. 9 Left QO converter of symmetric modes [63]; right QO converter of whispering gallery modes [64]. The
waves leaving the waveguide cut 1 are transformed by a parabolic mirror 2 into a wave beam (shown as a set of
parallel rays) with linear polarization of the electric field
338 J Infrared Milli Terahz Waves (2014) 35:325381
overview of their contributions can be found in the review [72]; some of their results are
presented below.
The gyrotron development for plasma applications at Varian was a little slower because it
was necessary to develop in parallel a gyrotron oscillator and a gyrotron amplifier both
operating at the frequency of 28 GHz. Results of this development were reported in Ref.
[73]: the gyrotron oscillator reached 175 kW and 107 kW in the pulsed and CW modes,
respectively, at an efficiency of 22.5%. In the gyroklystron amplifier, 25 kW peak power level
was realized in the high-gain regime and 65 kW peak power was achieved at the lower gain
values near 30 dB with about 10% efficiency. A photo of the CW gyrotron oscillator is shown
in Fig. 10.
Note that the power was extracted by a triple-miter-bend coupler through the side wall; the
wave coupling system used for extraction of the operating TE
-mode in this manner was
accompanied by significant undesirable mode conversion and reflection which caused some
negative consequences (for more details see Ref. [73]). Later this concept was greatly modified
in line with the Soviet concepts described above.
In the late 1970s, R. Temkin started to form at MIT his gyrotron team aimed at the studies
of gyrotron capabilities, novel gyrotron concepts and carrying out short-pulse experiments
(few microseconds) with MW-class gyrotrons. Formation of this team greatly benefited from
having strong solenoids available in the Francis Bitter National Magnet Lab and from the
support of the MIT Plasma Fusion Center led by M. Porkolab. To the best of our knowledge,
the first paper on gyrotron related topics was published by this group in 1979 [74]. In that
Fig. 10 The first 107 kW CW, 28 GHz Varian gyrotron with triple-miter-bend output coupler for the operating
-mode (courtesy of CPI)
J Infrared Milli Terahz Waves (2014) 35:325381 339
paper, some issues in the design of high-power, high-frequency gyrotrons for plasma heating
were discussed and some references to unpublished MIT reports on gyrotron related topics
were given.
At the same time, in France, at Thomson-CSF, the development of 35 GHz gyrotrons was
started by G. Mourier with co-workers. During the 1970s their activity was primarily limited
to theoretical studies. At the end of the 1970s, gyrotron development was also started in Japan
and in the Peoples Republic of China. However, since results of this development appeared in
the open literature only in the beginning of the 1980s, this activity is described in the next
5 1980s: Gyrotron Expansion over the Globe, Escalation of the Gyrotron Power
and Frequency Increase
5.1 Soviet Gyrotrons and Experiments on ECRH in the 1980s
In the late 1970s-beginning of the 1980s, a new, larger-scale tokamak T-10 was put into
operation at the Kurchatov Institute [75]. The gyrotron complex for plasma ECRH on this
tokamak initially consisted of four gyrotrons operating at about 3.6 mm wavelength and
producing about 200 kW power each in 150 ms pulses. A prototype of such a long-pulse
gyrotron equipped with a built-in quasi-optical mode converter and having the output radiation
in form of a Gaussian wave beam with linear polarization is shown in Fig. 11 reproduced from
Ref. [76].
Let us note that Ref. [76] was an invited paper published in the first special issue on
gyrotrons of the International Journal of Electronics. Seven such special issues edited by V. L.
Granatstein with colleagues were published from 1981 to 1992. They represent an important
collection of gyrotron papers. Also, many important papers on gyrotron development have
been published in a series of Special Issues on High-Power Microwave Generation of the IEEE
Transactions on Plasma Science. To better organize these special issues, the IEEE-PS Editor S.
Gitomer was inviting Guest Editors for each of them. The first issue appeared in 1985. Then,
starting from 1988 such issues were published regularly every second year.
The layout of the T-10 gyrotron complex is shown in Fig. 12 reproduced from Ref. [77].
Figure 13, taken from the same report, shows this complex assembled with the tokamak T-10.
In the first plasma experiments, about 700 kW of total millimeter-wave power was injected
into T-10. The efficiency of the ECRH was about 90%. As a result, due to ECRH the electron
temperature reached a 2.6 keV level [75]. Later, the total number of gyrotrons was increased
up to 11 [78]. To gyrotrons operating at 3.6 mm wavelength, several 3.0 mm gyrotrons were
added. This allowed plasma physicists to increase the level of the total millimeter-wave power
injected into the plasma to 4 MW and to manipulate with the power deposition profile. As a
Fig. 11 Prototype of a long-pulse gyrotron with a built-in quasi-optical mode converter for the T-10 tokamak
(reproduced from Ref. [76])
340 J Infrared Milli Terahz Waves (2014) 35:325381
result, the temperature of the electron component reached 8 keV and the pulse duration was
long enough (0.1 s) for a certain increase of the ion temperature up to about 1 keV due to
Coulomb collisions with the electrons. The gyrotron development in the USSR sponsored by
the State Committee on Atomic Energy (manager N. S. Cheverev) was focused on providing
Fig. 12 Layout of the gyrotron complex for the tokamak T-10 (reproduced from Ref. [77]
J Infrared Milli Terahz Waves (2014) 35:325381 341
the Kurchatov Institute by gyrotrons for T-10. The export of Soviet gyrotrons to foreign
laboratories was not allowed.
5.2 Gyrotrons at Varian in the First Half of 1980s
This period of time (1980s) was the time of blooming of the gyrotron development at Varian.
Major steps in this development were described in the review paper [79], and the most
important facts are reproduced below. After CW operation at more than 200 kW level at
28 GHz was demonstrated in 1980, the development of 60 GHz gyrotrons was started. Those
gyrotrons were scaled versions of the previously developed 28 GHz gyrotrons. They also
operated in the symmetric TE
-mode. To avoid the competition between this mode and the
-mode having a close frequency, a step-profile resonator was used (this concept was
earlier described in Ref. [76]). In such a resonator schematically shown in Fig. 14, the radius of
the first part corresponds to the cutoff frequency of the first mode (TE
-mode in the Varian
gyrotron), while the radius of the second part corresponds to the cutoff frequency of the second,
higher-order mode (TE
-mode in the Varian gyrotron). As a result, this pair of partial modes
forms a normal mode occupying all volume of the resonator, while other competing modes are
localized in one of the two parts only and, therefore, have much higher starting currents.
There were also several versions of such gyrotrons scaled for various users to 56 GHz and
70 GHz. Pulse operation at 200 kW 60 GHz in 100 ms pulses was demonstrated in early 1982,
and CW operation at the same power level was achieved in 1984.
The next step was the development of 100 kW CW gyrotrons at 140 GHz. (By that time,
the short-pulse operation of a 140 GHz gyrotron at the 100 kW power level was demonstrated
at MIT [80].) This phase was started in 1984. To handle the ohmic losses, which are increasing
Fig. 13 Photo of the gyrotron complex attached to the tokamak T-10 (reproduced from Ref. [77])
342 J Infrared Milli Terahz Waves (2014) 35:325381
with the operating frequency, it was decided to increase the resonator radius and operate at the
next symmetric TE
-mode. Later, also step-profile cavities operating in the TE
superposition of modes were used. A power level of 100 kW CW at this frequency was
successfully demonstrated in 1984. In addition, the power levels of 200 kW and 150 kW were
achieved in the 1 ms and 100 ms pulses, respectively.
The next goal of the Varian gyrotron group was formulated in 1986 as the generation of
1 MW CW power at 140 GHz. Before starting to describe the Varian activity in the second half
of the 1980s, however, it makes sense to describe the activity of other gyrotron developers in
the first half of the 1980s.
5.3 Gyrotron Development at Hughes Aircraft
During the first half of the 1980s the development of 60 GHz gyrotrons for plasma application
was sponsored by the US Department of Energy in parallel at Varian and at Hughes Aircraft.
The Hughes group led by J. Tancredi had the goal to develop 250 kW CW, 60 GHz gyrotrons.
Very soon gyrotron prototype was developed which at low duty produced 285 kW at 44%
efficiency [81]. This group showed itself as a strong competitor of Varian, but its activity lasted
for several years only and then was terminated for a number of reasons.
Fig. 14 Step-profile resonator (complex cavity): (a) the axial structure of a normal mode, (b) partial modes of
each stage which form the normal mode (the mode notation H corresponds to TE)
J Infrared Milli Terahz Waves (2014) 35:325381 343
5.4 Gyrotrons at MIT in the First Half of 1980s
During this period of time, MIT researchers carried out detailed analysis of the linear theory of
gyrotrons (see relevant references in [42]) and started short-pulse experiments in the 140 GHz
frequency range. Although MIT gyrotrons were operating at a low duty cycle (repetition
frequency of several Hz, few microsecond pulse duration) the gyrotrons were designed to be
scalable to CW operation. In the first gyrotron designed for operating in the TE
-mode the
power level reached 75 kW in early 1982 [42]. After one year, the power of this gyrotron,
which was also operating in the TE
-mode with even higher power and efficiency, was
increased up to 175 kW [82]. Later, following the Soviet approach, the MIT gyrotrons were
designed to operate in whispering gallery modes with a large azimuthal index, but with only
two radial variations. The first experiments with a gyrotron operating in the TE
-mode were
described in Ref. [83]. The design methodology for MW-class gyrotrons was formulated in
Ref. [84].
5.5 Gyrotrons at NRL in the First Half of 1980s
During this period of time, there were several directions in which gyrotron oscillators were
studied at NRL.
One of them was the analysis of the efficiency enhancement by using tapering either of the
resonator profile or the external magnetic field (or both) [85,86]. For example, in Ref. [85]the
transverse or orbital efficiency, which characterizes the efficiency of conversion of the energy
associated with electron gyration in the external magnetic field into the microwave energy, was
studied. It was shown that in a gyrotron with a simple sinusoidal axial structure of the resonator
field, the transverse efficiency, which in the case of a constant external magnetic field does not
exceed 36%, in the case of an optimal magnetic field tapering can reach 75%. (Soviet studies
of these methods of optimization were briefly overviewed in Ref. [87].) These theoretical
predictions were verified by experiments [88]. One more step important for more accurate
analysis of the processes of wave-beam interaction and gyrotron design was the development
of the self-consistent gyrotron theory at NRL [89], which was done along the lines of Ref. [52].
Another important issue was a concept of gyrotrons with step-profile resonators. The NRL
Ka-band gyrotron [90] was designed for operating in the TE
-mode, which was a great
step in terms of the difference between radial indices in each part of the cavity (so far, in all
other step-profile resonators the radial indices differed by one only). Two major steps were
done in this experiment: the use of a novel resonator concept with a large step and the profiling
of the external magnetic field done in line with theoretical recommendations [86] and
experimental results [88]. As a result, the mode selection allowed achieving the 340 kW
power level, while the magnetic field profiling yielded the efficiency of 54%. The maximum
efficiency of 63% was realized at a lower power level (150 kW) [90].
Another direction of the NRL activity was a quasi-optical gyrotron proposed in Ref. [91,
92]. This version of a gyrotron (or the electron cyclotron maser) is schematically shown in
Fig. 15 reproduced from Ref. [91].
Here a sheet beam of electrons guided by an external magnetic field oriented in the z-
direction interacts with an electromagnetic wave propagating in the y-direction. (Note that a
similar cylindrical configuration of an amplifier, in which an electron beam and an electro-
magnetic wave propagate in perpendicular directions, was studied earlier in Refs. [93,94].)
Such configuration has two obvious advantages: superior mode selection (like in lasers) and
possibility to tune the oscillation frequency by varying the distance between mirrors. At the
same time, the interaction efficiency of such a configuration should be lower than in a classical
344 J Infrared Milli Terahz Waves (2014) 35:325381
cylindrical configuration because in the case of strong wave reflection from both mirrors two
waves propagating in opposite directions form a standing wave pattern and, therefore,
electrons with different guiding centers interact with an EM field of non-equal intensities.
(Later it was proposed in [95] and proven experimentally in [96] to mitigate this efficiency
reduction by a slight tilting of the electron beam with respect to the resonator.) The develop-
ment of such gyrotrons took place not only at NRL, but also at EPFL Lausanne. Results of this
work were overviewed in Ref. [97,98].
5.6 Gyrotrons Besides Plasma Heating
During this period of time the gyrotron development was focused on plasma heating applica-
tions, but not limited by them. There were also some efforts to develop gyrotrons operating at
much higher frequencies. Since the gyrotron operating frequency, as follows from (1)forthe
case of negligibly small Doppler term, requires corresponding magnetic fields, there are some
obstacles and some ways to avoid them. The obstacle is the maximum value of magnetic fields
realizable in cryomagnets whose cost should be not prohibitively high. This maximum value is
somewhere in the range of 15-20 T, that allows gyrotrons to operate at the fundamental
cyclotron resonance at frequencies up to almost 0.5 THz.
One obvious possibility to move towards higher frequencies is to operate at cyclotron
harmonics. Above, we already mentioned Refs. [57,59] describing first experiments with
second harmonic gyrotrons. One experiment with the third harmonic gyrotron was described
in Ref. [99]. That gyrotron operated at the frequency close to 55 GHz and delivered about
150 kW peak power with 10% efficiency. A step-profile resonator was used which formed the
coupled normal mode. This was just a demonstration of gyrotron ability to operate
at cyclotron harmonics higher than the second one. Harmonic operation at near THz frequen-
cies was demonstrated much later and will be described below (see Section 7).
Another possible way to avoid the limitation imposed by available magnetic fields is the
Doppler frequency up-conversion. As follows from Eq. (1), when the Doppler term becomes
comparable with the operating frequency, the device can operate at frequencies much higher
Fig. 15 Schematic representation of a quasi-optical cyclotron maser (reproduced from Ref. [91])
J Infrared Milli Terahz Waves (2014) 35:325381 345
than the electron cyclotron frequency or its harmonics. Equation (1) for this case can be
rewritten as
In (4), n=c/v
is the wave refractive index, which in the case of fast waves (v
smaller than one. As follows from (4), to realize substantial Doppler frequency up-conversion
in the case of fast wave operation, one needs relativistic electron axial velocities, i.e. it requires
high beam voltages. At the same time, the operation with phase velocities close to the speed of
light allows one to maintain the cyclotron resonance under conditions when the changes in the
relativistic electron cyclotron frequency due to electron energy modulation and the changes in
the Doppler term compensate each other to the extent required for efficient operation. Such a
device operating under conditions close to the autoresonance discussed above in Section Iwas
proposed in Ref. [100] and named the cyclotron autoresonance maser or CARM. This concept
was actively studied in the late 1970s in the USSR (see the review paper [101]) and later
attracted attention of other researchers, see, e.g., Ref. [102]. So far, we described the devices
operating with fast waves. In principle, the Doppler frequency up-conversion can also be
realized in the case of operation in slow waves even at low voltages [100,101,103]. However,
as the wavelength shortens, such devices face the same difficulties with small-period slow-
wave structures and device miniaturization as conventional linear-beam devices based on the
coherent Cherenkov or Smith-Purcell radiation. Moreover, above we considered the region of
the normal Doppler effect (β
). However, the slow-wave CARMs can also operate in the
region of the anomalous Doppler effect (β
), as described in [100,101,104106]. In the
latter case the device can operate even being driven by a beam of electrons injected with
=0 (linear beam) [100], because electrons radiating EM waves under the anomalous
Doppler condition lose the axial momentum, but gain the orbital one [104].
Lastly, it was possible to avoid limitations on the gyrotron frequency, which were imposed
by available superconducting magnets, by using pulsed solenoids. Successful realization of
this concept was demonstrated in Ref. [107]. To minimize the volume occupied by the
magnetic field and, hence, to reduce the energy losses, the pulsed solenoid was wound directly
on a thin stainless body of a gyrotron. The photo of this gyrotron is shown in Fig. 16.
The pulse duration of the magnetic field was a few ms, which allowed this field to easily
penetrate into the tube body. The high-voltage power supply provided single pulses of 80 μs
pulse duration with up to 70 kV and up to 20 A voltage and current, respectively. In those
experiments, the magnetic field was varied from 14 T up to 27 T and, correspondingly, a
sequence of various whispering gallery modes was excited. At the magnetic field of 15 T, the
gyrotron operated at 0.8 mm wavelength and produced 120 kW peak power with 15%
efficiency. At the wavelength of 0.6 mm (0.5 THz frequency) the gyrotron delivered
100 kW with more than 8% efficiency. Lastly, at the highest magnetic field, the wavelength
achieved was 0.54 mm and the power was at the 60 kW level. Much later, some improvements
Fig. 16 Gyrotron with a pulse solenoid (reproduced from Ref. [107]). This 0.5 THz gyrotron produced 100 kW
346 J Infrared Milli Terahz Waves (2014) 35:325381
in the construction of pulsed solenoids allowed researchers to reach a 1 THz frequency barrier
at a several kW power level [108](seeSection7).
One more variety of CRMs should be briefly mentioned here, although it was never used for
plasma heating and current drive. This is a so-called large-orbit gyrotronor LOG, which is a
device in which all electrons rotate about the waveguide axis. This concept was proposed in
Ref. [109] and later explored by many group of researchers [155]. A remarkable feature of this
concept is its superior mode selectivity becausein such a gyrotron anelectron beam excites only
oscillations of modes whose azimuthal index is equal to the cyclotron harmonic number.
5.7 Expansion of the Gyrotron Geographyin the Second Half of the 1980s
5.7.1 A. Europe
As was mentioned above, the theoretical studies of gyrotron physics in Europe were started at
Thomson-CSF in the mid-1970s. In the late 1970s in Germany first theoretical investigations
on the large-signal theory were performed by the group of H. Döring at RWTH Aachen [110].
For a short time, there were also some basic gyrotron studies at EMI-Varian [111], Kings
College London [112] and Cambridge Electronics Consultants [113,114]inGreatBritain.
However, a real systematic development of gyrotrons for plasma applications in Europe was
started in 1983 and initially included [115]:
The industrial development of a 100 GHz, 100 ms, 200 kW gyrotron at Thomson-CSF,
Vèlizy, France
The development of a 120 GHz, 100 ms, 200 kW quasi-optical gyrotron at CRPP
Lausanne, Switzerland and
The study of mode competition in highly overmoded and complex cavities with the goal
to develop a 150 GHz gyrotron at KfK Karlsruhe, Germany.
Having this fact in mind, let us describe this development in some details.
5.7.2 France: Thomson CSF
The results of the development of gyrotron oscillators at Thomson CSF in Vélizy, France, for
fusion plasma applications at 8 GHz (TE
-mode), 35 GHz (TE
-mode), 100 GHz (TE
mode) and 110 GHz (TE
-mode and TE
-mode) are summarized in Refs. [116,117].
Chronologically, the first one was a 35 GHz gyrotron which was followed by a series of
100 GHz tubes with different cavity designs based on the known concepts of step-profile and
tapered cavities [11 8]. Note that the gyrotron development at Thomson-CSF also included
theoretical studies; in particular, French scientists were the first who developed the self-
consistent gyrotron theory outside the Soviet Union; that theory was presented in Santa Fe
in 1981 [119]. (Soon after that, a similar theory was developed in the US at NRL [89].) Among
experimental results, the most impressive was, possibly, the 1 MW, 1s, 45% efficiency
operation of the 8 GHz gyrotron [116] for lower hybrid heating on the FTU tokamak in
Frascati, Italy.
5.7.3 Switzerland: EPFL and ABB
In parallel to the activities in the field of quasi-optical gyrotrons at 91, 100 and 200 (second
harmonic) GHz [98], the team of M.Q. Tran at EPFL Lausanne in collaboration with ABB
J Infrared Milli Terahz Waves (2014) 35:325381 347
Baden, Switzerland, also developed an 8 GHz TE
-mode gyrotron with 350 kW output power
in 0.5s pulses at 35% efficiency for the low-hybrid heating experiments at the FTU (Frascati
Torus Upgrade) Tokamak. Also, a 39 GHz, TE
-mode tube with 250 kW output power in
100 ms pulses at 42% efficiency was developed for the TCA tokamak at EPFL [116,120].
5.7.4 Germany: KfK and Philips-Valvo
In the middle of the 1980s, gyrotron development for fusion plasma heating in Germany was
started by a team guided by G. Hochschild at the Karlsruhe Center for Nuclear Research
(KfK in German notations) in collaboration with Philips-Valvo company in Hamburg. The
first step of the KfK gyrotron program was intended to develop a TE
-mode tube capable of
delivering 200 kW at 150 GHz in 100 ms pulses [121]. After initial ms-short-pulse experi-
ments in 1986 (120 kW at 20% efficiency) [122] the design frequency was changed to
140 GHz for the second harmonic ECRH (at 2.5 T) and electron heat wave experiments (with
modulated output power) in high-density plasmas of the stellarator W7-AS at IPP Garching.
The prototype tube was manufactured by Philips-Valvo. In 0.5 ms pulses, 300 kW were
obtained at 33.4% efficiency. In long-pulse operation (up to 0.5s pulses), 120 kWoutput power
at 26% efficiency was achieved [123].
Later, headed by M. Thumm and following the Russian approach, the KfK gyrotrons were
designed to operate in asymmetric volume modes. The first experiments with a 500 kW
140 GHz gyrotron operating in the TE
-mode were described in Ref. [124]. The KfK
activity was not limited by gyrotron experiments. Also, the theoretical group led by E. Borie
was doing active studies [121,123], which were, to some extent, motivated by the course of
lectures given by G. Nusinovich in 1984 at KfK [125]. These theoretical activities at KfK were
supplemented by the work on the gyrotron theory at TU Hamburg-Harburg in the group of K.
Schünemann [126].
Independently of these 140 GHz activities at KfK, the Philips-Valvo company (K. Behm),
as a competitor of Varian, developed a 70 GHz, CW TE
-mode gyrotron. That gyrotron
achieved 140 kW at 30% efficiency in CW operation [127]. However, since finally the W7-AS
stellarator was equipped with five Varian 70 GHz gyrotrons, Philips-Valvo stopped its gyrotron
development program.
5.7.5 Asia
In Asia, the most active gyrotron development took place in Japan where two industrial
companies were involved in the gyrotron business: Toshiba and Mitsubishi. Toshiba was
developing gyrotrons in collaboration with Fukui and Kyoto Universities. This collaboration
team, first, developed a 22 GHz gyrotron delivering more than 20 kW power with more than
40 % efficiency and then started developing gyrotrons at higher frequencies (70 GHz) and
higher power levels (200 kW) [128]. Mitsubishi Electric Corporation started in the mid-1980s
the development of a 120 GHz gyrotron operating in the TE
-mode. [129].
By the end of the 1980s, the collaboration team of the Electron Tube Division of Toshiba
Corporation and the Japan Atomic Energy Research Institute (JAERI) developed a 120 GHz
gyrotron operating in the whispering gallery modes and delivering more than 0.5 MW in 1 ms
pulses [130]. According to Ref. [130], JAERI started the 120 GHz gyrotron development with
both companies (Toshiba and Mitsubishi) in 1986. Both companies built gyrotrons operating
in the TE
-mode and delivering 150 kW in 10 ms pulses. Then, the second stage of the
gyrotron development took place in the late 1980s with the goal to develop a 0.5 MW
gyrotron operating in a whispering gallery mode. The mode of choice was the TE
-mode in
348 J Infrared Milli Terahz Waves (2014) 35:325381
a tapered cavity. It turned out that the gyrotron was able to deliver in the TE
-mode more
than 0.5 MW with 30.2% efficiency. It was also possible to operate in the TE
-mode where
the power exceeded 0.6 MW and the efficiency was more than 36%. That gyrotron was
developed for plasma heating experiments at the tokamak JT-60 Upgrade. It should be noted
that the gyrotron operating in the whispering gallery mode was equipped with a built-
in quasi-optical mode converter like the one schematically shown in the right part of
Fig. 9.
To conclude this brief geographical overview, let us note that during the 1980s the above
mentioned group of the Sydney University in Australia started the development of frequency
tunable gyrotrons for spectroscopic applications [131]. In 1988, this group established a
fruitful cooperation with the Fukui University [132]. Although the gyrotrons developed in
the process of this collaboration were not intended for plasma heating, some of them were used
for active plasma diagnostics based on collective Thomson scattering.
Another place on the globe where gyrotron development was started in the mid-1980swas
the Institute for Space Research (INPE) in Sao Jose dos Campos, Brazil. That group was
focused on the gyrotron theory [133,134]. To the best of our knowledge, some gyrotron
experiments at frequencies around 30 GHz were started there, but they were not very
successful and, therefore, this gyrotron activity was terminated in the late 1990s.
5.8 Gyrotron Related Activity in US and USSR at the end of the 1980s
By that time, Varian in USA became a gyrotron provider of practically all plasma installations
around the globe outside the USSR. The Table 2below, which is reproduced from Ref. [79],
fully describes that situation. It shows the gyrotron operating frequencies and the power
achieved in different pulse durations. More than 100 of such gyrotrons were delivered to
various plasma labs in the US, Europe and Japan.
By that time, T-V George of DoE, who for a long time was responsible for gyrotron
development in the US, formed the national gyrotron collaboration team, which included the
MIT group responsible for conceptual designs and short-pulse experiments, the theory group at
the University of Maryland, including T. M. Antonsen, Jr. and B. Levush, the University
Tab le 2 Summary of Varian gyrotrons for ECRH at the end of the 1980s (reproduced from Ref. [79]).
Frequency (GHz) Power (kW) Pulse length (ms)
28 200 40, 75, CW
35 200 CW
53 200 100, CW
56 200 CW
60 200 100, CW
70 200 100, CW
106 100, 400 100
140 200 1
150 100
100 CW
1000* 1*
400* CW*
*Under development
J Infrared Milli Terahz Waves (2014) 35:325381 349
of Wisconsin group (R. Vernon) responsible for design of mode converters and
launchers, the Varian team led by H. Jory and K. Felch and the users from General
Atomics, including R. Callis, R. Prater and J. Lohr. The coordinated efforts of this
collaboration team resulted in many successful steps, some of which are described
In the USSR, the experiments with a 100 GHz coaxial-cavity gyrotron resulted in produc-
ing 2.1 MW power in short (less than 0.1 ms) pulses [135]. Important progress was made in the
gyrotron theory: at least, such results deserve mentioning as the self-consistent theory of non-
stationary processes in gyrotrons [136], which was later generalized for the case of multi-mode
interaction [137], the theory of automodulation instability [138] and improved codes for
designing magnetron-type electron guns, which made it possible to include into consideration
the space charge effects, self-magnetic fields, beam instabilities and formation of quasi-laminar
beams (see Section 10.8 in Ref. [56] and references therein). The industrial companies Salut
and Toriywere involved in the development of gyrotrons at various frequencies [139].
Table 3reproduced from Ref. [139] shows various Soviet tokamaks, stellarators and open
mirror machines where industrial gyrotrons of different frequencies manufactured by the
Salutcompany were used.
In particular, in addition to industrial gyrotrons for T-10, whose power at the frequency of
83 GHz reached the 0.5 MW level in less than 1 s pulses, the industrial development of
gyrotrons for ITER at the frequency of 140 GHz began (later the frequency was increased to
170 GHz). In the beginning of the 1990s such gyrotrons were able to deliver 0.5 MW power
in the pulses up to 0.7s duration [139]. The millimeter wave energy per pulse in these
gyrotrons was primarily limited by the capacity of output windows. In Soviet gyrotrons, these
windows were made either from single-crystalline alumina (sapphire) or from boron nitride
(BN); those were single-disk windows with periphery cooling [140]. In Varian gyrotrons by
that time edge-cooled single-disk beryllium oxide, aluminum oxide and sapphire windows and
face-cooled (with Fluor-Carbon liquid) double-disc sapphire windows were used, whose
capacity was quite similar to that of Soviet windows.
Tab le 3 List of Soviet plasma installations where industrial gyrotrons manufactured by Salutwere used
(reproduced from Ref. [139]).
Name, Type Institute Operating
Output Power
Unit, kW
No. of Gyrotrons in
T-10 Tokamak IAE, Moscow 75, 82, 166* 0.4-0.5 11
T-15 Tokamak 82*, 100* 0.5 24
T-7 Tokamak 62.5 0.2 2
Ogra, Magnetic Snare 37.5 0.1 1
T-14 Tokamak IAE, Troitsk 62.5 0.2 1
Liven-2 Stellarator IGP AS, Moscow 37.5, 75 0.2, 0.3-0.4 2
Tuman Tokamak Ioffe, Leningrad 54.5, 62.5 0.35, 0.15 1
Ambal Magnetic
INP SS AS, Novosibirsk 54.5*, 75* 0.5 6
Uragan Stellarator Phys.-Techn. Inst.,
37.5*, 54.5* 0.2, 0.5 2
*in planning
350 J Infrared Milli Terahz Waves (2014) 35:325381
So, it was desirable to find some window materials which would mitigate the restrictions on
millimeter-wave output power. Also desirable was to improve the efficiency of built-in quasi-
optical mode converters for transforming the high-order cavity modes with complicated E-field
structures into linearly polarized Gaussian wave beams.
6 1990s: Gyrotron Maturity, Megajoules per Pulse, Improved Output Couplers,
Depressed Collectors, Diamond Windows, 2 MW Coaxial-Cavity Gyrotrons
In the 1990sthemodern history of gyrotrons, i.e. the development of MW-class long-pulse
gyrotrons, began with the following principal achievements described elsewhere [8,114,141]:
two groups produced more than 1 Megajoule per pulse energy: CPI [142]ina110GHz
gyrotron operating in the TE
-mode at the power level of 0.4-0.5 MW in more than 2
second pulses and GYCOM [143] in a 140 GHz gyrotron operating in the TE
-mode at
the 0.5 MW level in 3 second pulses.
Demonstration of efficient and stable gyrotron operation in very high-order modes (e.g.
at 110 GHz, TE
at 140 GHz, and TE
or TE
at 170 GHz) [144]. This
solved the problem of thermal cavity loading due to ohmic wall losses. Realization of
stable efficient operation in such high-order modes was possible due to the progress in
understanding of problems associated with mode competition and start-up scenarios. Also
a new concept of the echelette type resonator, which can be treated as an extension of a
step-profile resonator was proposed and studied both theoretically and experimentally in
Russia [145,146].
Development of improved internal quasi-optical mode converters transforming high-order
gyrotron cavity modes into Gaussian RF beams [147]. Such converters with rippled-wall
launchers and adapted mirrors with non-quadratic surface contour functions generated high-
purity Gaussian-like output wave beams with low level of internal stray radiation [148].
Introduction of single-stage depressed collectors for energy recovery from spent electron
beams [149,150]. This led to increasing the wall-plug tube efficiencies up to about 50%
and mitigating collector and power supply problems.
Introduction of gyrotron vacuum barrier windows made of synthetic diamond [151,152]
which is still the only option for long-pulse and CW MW-class tubes.
Development of short-pulse prototypes of very-high order mode 1.5 - 2 MW coaxial-
cavity gyrotrons [153,154].
Since 1993 the progress on gyrotrons and various other types of gyro-devices is summa-
rized in M. Thumms annual laboratory report State-of-the-Art of High Power Gyro-Devices
and Free Electron Masers[155], including almost 1200 references on experimental results.
Therefore, some technical details of experiments will be omitted below.
6.1 Efficient Gyrotron Operation in Very High-Order Cavity Modes
For achieving stable efficient operation in very high order modes it was necessary to develop a
more accurate theory describing the interaction between modes during the voltage rise and
sensitivity of gyrotron operation to various misalignments and fluctuations. This work was
practically always in progress. There were several reviews describing the studies of beam-
wave interaction in details [156158]. In the 1990s, this theoretical work was focused on
improving the accuracy of calculations by using more sophisticated codes. One of the most
J Infrared Milli Terahz Waves (2014) 35:325381 351
important steps was done at the University of Maryland (later in collaboration with NRL),
where the group led by T. Antonsen developed the non-stationary, self-consistent code MAGY
[159,160]. This code, which takes into account the effects caused by the space charge fields,
resonator wall profile and many other important factors, was then widely used for designing
high-power gyrotrons in the US. Similar codes were also developed in Europe [161]andin
Russia [162]. Also, detailed analysis of various start-up scenarios has been performed [163,
164] and issues like the possibility to manipulate with the operating frequency by varying the
relation between the modulation-anode and beam voltages [165], technical noise in gyrotrons
[166] and sensitivity of gyrotron operation to the displacement of the electron beam axis with
respect to the cavity have been analyzed [167,168]. In addition to the progress in understand-
ing the wave-beam interaction, also significant progress was made in developing codes for
gyrotron electron-optical systems. This progress was described in the textbook of S. E.
Tsimring [56]. Another area of active research was optimization of gyrotron cavities, wave-
guides and mode converters. As known, standard cylindrical gyrotron cavities employ a linear
input taper and a nonlinear output up-taper. Mode conversion has been strongly reduced by
using smooth transitions between the cylindrical and taper sections. The resulting mode purity
has been calculated to be 99.9%.
The state-of-the-art of MW-class 110-170 GHz gyrotrons with quasi-optical output couplers
at the end of the 1990s is summarized in Table 4. In this Table, parameters in the longest
pulses are shown; for more details see Ref. [155].
6.2 Improved Quasi-Optical Mode Converters
Two general methods were utilized to provide low diffraction losses, resulting in a pencil-like
wave beam:
(1) Employing tailored aperture distributions at the radiating launcher such that the sidelobes
are reduced [147,148,183], which means pre-shaping of the wave beam before it is
launched to the mirror system. The shaped beam has lower fields at the cut edges (low
diffraction losses) and a nearly Gaussian angular spectrum. To achieve a sidelobe-free
fundamental Gaussian beam as the output mode, the launcher must have space-periodic
helical feed waveguide deformations (dimples) such that the incident rotating TE cavity
mode is converted to a suitable mode mixture generating the Gaussian beam distribution.
The interference of this mode mixture creates an RF-field bunching in the axial and
azimuthal directions. The overall efficiency can be up to 96-98%. In order to avoid
parasitic mm-wave generation in cavities produced by the dimples, the average diameter
of the launcher must be increased along its axis (e.g., linear taper) [177].
(2) In cases where a simple smooth-wall launcher was used in connection with a large quasi-
parabolic reflector, specific phase-correcting mirrors with non-quadratic surface contour
function allowed one to generate any desired amplitude and phase distribution of the
wave beam. Advanced iterative computer algorithms have been developed to provide the
optimized shapes of the mirrors [184187]. The overall conversion efficiency was in the
range of 90-92%. The computer codes have also been used for the phase retrieval from
measurements of only the amplitude, thus allowing detailed mode diagnostics and
reconstruction of the overall fields [184187]. The knowledge of phase and amplitude
at any point of the longitudinal axis of the beam makes a calculation of the field pattern at
all other points possible. Most of the MW gyrotrons have been equipped with one or
even several relief windows (BN, Al
or Sapphire) to reduce the level of parasitic stray
radiation inside the gyrotron.
352 J Infrared Milli Terahz Waves (2014) 35:325381
6.3 Single-Stage Depressed Collector (SDC)
The electronic interaction efficiency η
of the annular, helical electron beam in a gyrotron
cavity is typically in the range of 30-40%. This means the remaining power of the beam after
interaction (spent beam) is approximately 60-70% of the initial beam power. Fortunately, since
the interaction with the waves propagating near cutoff in cavities leaves the electron axial
momentum practically unchanged, the energy distribution in a spent beam has a definite
minimum value, so the energy recovery by using a depressed collector can be realized. A
schematic of such a gyrotron with an isolated collector is shown in Fig. 17.
Up to now, only single-stage depressed collectors have been employed in high-power
gyrotrons. They were introduced by K. Sakamoto et al. at JAERI [149] and B. Piosczyk
et al. at FZK [150]. In this case the total efficiency η
can be increased, in accordance with a
simple formula
ηTotal ¼ηintVb=VbVdep
Here, V
is the beam voltage and V
is the retarding voltage whose maximum value
corresponds to the minimum energy of the spent electron beam. For example, for η
= 80 kV, this results in η
Tot al
= 56%. There are numerous merits of using
the depressed collectors. First, the required voltage of the main power supply is significantly
reduced (e.g., from 80 kV to 50 kV). Next, the spent electron beam energy and, thus, the
collector heat load is significantly lowered, which allows the reduction of the collector size
and/or an increase of the gyrotron reliability as well as a reduction of the capacities of the
cooling and power supply systems. Lastly, decreasing the energy of electrons deposited to the
collector lowers the intensity of generated X-rays. Detailed design studies show that the
Tab le 4 Status of gyrotrons for ECRH and stability control in magnetic fusion devices at the end of the 1990s.
Institution Frequency Cavity
Power Efficiency Pulse
[GHz] [MW] [%] [s]
0.6 31 10
0.106 21 CW
GYCOM-M, IAP [171173]110TE
0.35 33 10
GYCOM-N, IAP [171174]110TE
0.5 33 1.0
, TOSHIBA Naka, Otawara [149,175]110 TE
0.35 48 (SDC) 5.0
110 TE
1.0 38 (SDC) 3.0
,CEA,CRPP,KIT[176178]118 TE
0.35 23 111
,Karlsruhe[124,150,179] 140.5 TE
0.46 51 (SDC) 0.2
GYCOM-M, IAP [171173] 140 TE
0.26 (0.1) 36 10 (80)
170 TE
0.65 45 (SDC) 1.0
GYCOM-N, IAP [171173] 140 TE
0.55 33 2.0
,TOSHIBA[180182] 170 TE
0.52 32 (SDC) 6.2
SDC: Single-Stage Depressed Collector
Communications & Power Industries, formerly VARIAN
Formerly KfK, then FZK
Formerly JAERI
Formerly Thomson TE
J Infrared Milli Terahz Waves (2014) 35:325381 353
efficiency of gyrotrons with two-stage depressed collectors can be in the range of 60% (Ref.
6.4 Synthetic Diamond Output Window
In order to define the appropriate concepts for the development of 1-2 MW, CW mm-wave
windows one has to compare the thermophysical, mechanical and dielectrical parameters of
possible window materials related to the load-failure resistance and the power-transmission
capacity at different temperatures. The features of beryllia, boron nitride, silicon nitride,
sapphire, Au-doped silicon, silicon carbide and CVD diamond are given in [155,189]. The
comparison of BeO, BN, Si
, sapphire and SiC clearly shows that there is no chance to use
these dielectrics as an edge-cooled, single-disk 1 MW, CW window at room temperatures.
Experiments at CPI in the USA and at NIFS and JAERI in Japan confirmed that even a double-
disk FC75-face-cooled sapphire window has a CW-power limit around 0.3-0.4 MW. Never-
theless, these materials are widely used at lower frequencies and in pulsed gyrotrons. Silicon
disks are very brittle. On the way to 1 MW, CW gyrotrons, IAP in Russia and FZK in
Germany developed 140 GHz prototype tubes with two output windows for two mm-wave
beams. The TE
-mode gyrotron at IAP used an internal 3 dB-diffraction-grating beam
splitter (2x0.37 MW, 3s) [190], whereas the coaxial-cavity tube at FZK operating in the
mode employed an internal mode converter to the degenerate, counter-rotating
mode in combination with a two-cut quasi-optical launcher (2x0.55 MW, 5 ms) [191].
Collaboration of the gyrotron teams led by K. Sakamoto at JAERI, Japan and M. Thumm at
FZK, Germany [151,152] on the development of chemical vapor deposition (CVD) diamond
windows resulted in solving this most serious problem in realization of MW-class gyrotrons.
Output window
RF beam
Electron beam
Mode converter
Electron gun
Ceramic Super conducting
magnet (SCM)
Mai n power
Fig. 17 Conceptual view of a gyrotron with power supply system for depressed collector [149]
354 J Infrared Milli Terahz Waves (2014) 35:325381
Nowadays synthetic diamond windows are installed on most of the high-power long-pulse
gyrotrons, with the exception of a cryogenic (LN
) sapphire window, which is mounted on a
118 GHz gyrotron and is capable of transmitting approximately 0.4 MW CW [176178]. At
present, CVD diamond is the only material for simple, edge-cooled (water) single-disk 1-2
MW, CW gyrotron windows [155,189,192194]. Current CVD capabilities allow samples of
up to 140 mm diameter and 2.5 mm thickness (572 carat!). The price is very high (about one
tenth of the tube costs) and there are only two suppliers: Element Six in the Netherlands and
Diamond Materials in Germany. CVD diamond is attractive due to its good mechanical
properties (bending strength: σ
=400 MPa), very small thermal expansion (α=10
modest dielectric constant (ε
=5.67 ± 0.01), low mm-wave attenuation (tan δ2.10
very high thermal conductivity (k=1900 W/mK at room temperature). Nuclear irradiation of
such windows with <10
neutrons /m
and 0.8 Gy/s γ/X-rays is acceptable and brazing
technologies are available. In the early stage of the development, some problems with
corrosion of the brazing material (aluminum) were reported, which were avoided by
employing other materials, e.g., Cu/Ag-based brazing of the disk to copper cuffs at about
750°-850°C, or by copper coating on the aluminum, etc. The transmission frequencies are
pn¼1;2;3……ðÞ ð6Þ
Here, fis the frequency, cthe speed of light, and dis the disk thickness. For example, in the
case of 140 GHz frequency a diamond disk with d=1.8 mm can be used. This thickness
corresponds to four half wavelengths in the dielectric.
6.5 2 MW Coaxial-Cavity Gyrotrons
In order to reduce the cost of ECH&CD systems for future fusion reactors and to allow
compact poloidal launchers for plasma stabilization by keeping the number of the required
gyrotrons and magnets as low as possible, higher mm-wave power per tube (1.5 MW or more)
is desirable. Capabilities of gyrotrons with conventional cylindrical cavities to reach such
power levels are limited because of high Ohmic wall losses and/or mode competition
problems. In coaxial resonators, the existence of the inner rod yields more flexibility in
providing mode selection and better control of an electron beam. These facts stimulated the
first studies of coaxial resonators [195] and the first experiments [135] mentioned above. It
should be noted that, while initially [195] it was proposed to use smooth-wall down-tapered
inner rods for providing mode selectivity, later G. Denisov proposed to use rods with axial
slots for improving the mode selectivity. This concept was elaborated in Ref. [196]. The
history of the development of gyrotrons with coaxial resonators was overviewed in Ref. [197].
Below, in this subsection, we focus on the progress made in the 1990-s [191,198,199]. In
addition, the inner conductor enables a specific voltage depression scheme for energy recovery
and ultra-fast frequency step tuning just by applying an appropriate voltage to this coaxial
insert [200]. Regarding technical limitations on the gyrotron power level imposed by the
output windows let us note that CVD-diamond windows with a transmission capability of 2
MW, CW are feasible [192194].
Successful activities on the development of 1.5 MW gyrotrons with coaxial cavities were
conducted at IAP Nizhny Novgorod (TE
at 140 GHz [201]) and FZK Karlsruhe (TE
at 140 GHz and TE
at 165 GHz) [154,199,202]. A schematic layout of the FZK 165 GHz
gyrotron is shown in Fig. 18.
Detailed description of this tube and results of experiments can be found elsewhere [154].
In brief, the coaxial longitudinally corrugated and slightly down-tapered insert shown in
J Infrared Milli Terahz Waves (2014) 35:325381 355
Fig. 18 is supported from the bottom of the gun and is in total approximately 1 m long. The
insert can be aligned with respect to the electron beam under operating conditions with high
accuracy (better than ±0.05 mm) The cooling of the insert is provided by a water flow of about
10 l/min which is sufficient even for CW operation. The amplitude of mechanical vibrations of
up to 0.03 mm has been measured. This stability of the insert fulfills completely the
requirements for a stable long pulse operation. The collector was insulated with respect to
Fig. 18 Schematic layout of the 165 GHz coaxial cavity gyrotron at FZK [154]
356 J Infrared Milli Terahz Waves (2014) 35:325381
the tube body and in addition, the body was insulated from the ground allowing operation in
depressed mode either by positive biasing of the body and the insert or by negative biasing of
the collector with respect to the ground. RF-measurements were performed with a pulse length
of typically 1 ms with a repetition rate of 1 Hz. In single-pulse operation, the pulse length was
extended up to 17 ms. Figure 19 (left part) shows the RF-output power P
and efficiency η
as functions of the beam current I
. For each value of the beam current the cathode voltage V
and the magnetic field were optimized for maximum P
=84A a maximum RF output
power as high as 2.2 MW was achieved with an efficiency η
=28%. A maximum output
efficiency of 30% (without depressed collector) was measured at the design output power
= 1.5 MW. The calculated values, which are in good agreement with the experimental
results, were obtained with the self-consistent, multi-mode FZK simulation code [161]using
the experimental parameters as input. The internal RF losses due to absorption and diffraction
were assumed to be 10%. Operation with a single-stage depressed collector has been per-
formed at the output power of 1.5 MW.
By applying a negative retarding voltage of about V
= 34 kV at the collector, it was
possible to increase the overall efficiency from 30% up to 49%, as shown in Fig. 19 on the
right. The experimental results of the Russian 140 GHz coaxial cavity gyrotron are comparable
[200,201]. The mechanical stability of the coaxial insert is a crucial issue for stable long pulse
and CW operation of coaxial cavity gyrotrons. Thus measurements of the mechanical vibra-
tions of the insert have been performed under operating conditions [202].
7 21st Century: Gigajoules per Pulse, 1 MW, CW Operation, Multi-Frequency
and Step-Frequency-Tunable Gyrotrons, THz Gyrotrons
In the 21st century the development of 1 MW, CW gyrotrons for nuclear fusion plasma heating
has been completed by the following achievements:
Improved beam tunnels. The beam tunnel is mounted in the electron beam compression
zone between the MIG and the cavity in order to suppress undesired parasitic oscillations
which would degrade the quality of the electron beam and produce a thermal overload in
this component [203206].
Improved electron beam sweeping over the collector wall. A combination of a vertical
magnetic field sweeping system (VFSS) and a transversal magnetic field sweeping system
; (calc.)
rf output power / MW
/ A
rf output eff iciency / %
rf - output effici ency / %
rf-output power
rf - output power / MW
collector voltage / kV
rf-output efficiency
Fig. 19 RF-output power and efficiency of the 165 GHz coaxial cavity FZK gyrotron: left maximum power
and efficiency as functions of the electron beam current; right power and efficiency as functions of the retarding
collector voltage (All other parameters were kept constant [202].)
J Infrared Milli Terahz Waves (2014) 35:325381 357
(TFSS) allows reducing the peak wall loading of the collector by a factor of nearly 2 as
compared to VFSS alone [207].
The availability of long-pulse gyrotrons with fast frequency tunability (several
GHz per second, tuning in 1.5-2.5% steps for approximately ten different fre-
quencies) would permit the use of a simple, fixed non-steerable mirror antenna for
local ECCD experiments and plasma stabilization [208,209]. Frequency change in
gyrotrons can be achieved by operation in different cavity modes and correspond-
ing variation of the cavity magnetic field. Challenges in this development are the
proper MIG, the broadband quasi-optical output coupler and the output vacuum
window which should be transparent at many frequencies. Only cavity-mode
series with nearly the same radius of the maximum of the electric field are
Already in the 1990s, D-band (114 170 GHz) short-pulse (1 ms) frequency tunable high-
power gyrotrons with broadband SiN Brewster angle window were tested at FZK [208]. In the
case of a conventional cylindrical cavity tube, in which such frequency tuning was realized, the
frequency spacing was 3.7 GHz [210,211] whereas the different modes of the coaxial-cavity
tube had a distance of 2.2 GHz [212].
For suppression of neoclassical tearing modes (NTMs) in tokamaks, gyrotrons operating
in power modulated regimes have been developed [213,214] and a new concept of
periodical steering of the wave beams from a gyrotron operating in two, periodically
switched frequencies (fast directional switch: FADIS) was proposed [215,216]. Also, an
alternative concept of a gyrotron with an azimuthally corrugated resonator wall was
proposed [217], where the switching from one standing wave with an azimuthal structure
cosm8to another wave (sinm8) with a slightly different frequency (the frequency
separation between these waves is proportional to the corrugation depth) can be realized
by using a short-pulse, low-power driver.
The self-consistent, multi-frequency code MAGY was used for studying modification of
the axial structure of modes with the beam voltage [218] and the start-up scenarios in
gyrotrons where two triplets of quasi-equidistant modes (one co- and another counter-
rotating with gyrating electrons) were analyzed [219]. Also the stability of operation in
very high-order modes was studied by considering quintets of quasi-equidistant modes
rotating in the same azimuthal direction [220].
In the course of the last decade, also a lot of attention was paid to the so-called after-cavity
interaction (ACI), which is the effect of additional interaction between the EM radiation
outgoing through the uptapered waveguide after the resonator and electrons in a spent
beam, which are moving towards a collector in a decreasing magnetic field. As a rule, this
ACI reduces the overall efficiency and, thus, plays a negative role. Therefore it is desirable
to avoid or, at least, mitigate this effect.
In the case of low-power gyrotrons for spectroscopic applications the 1 THz frequency
threshold was overcome by fundamental harmonic gyrotrons with pulsed magnets [221]
and second harmonic gyrotrons operating in cryomagnets [222].
7.1 Beam Tunnel (Compression Zone)
This component is especially prone to parasitic gyrotron-type oscillations since the perpen-
dicular energy of the electrons increases on the way to the cavity. Azimuthally symmetric high-
358 J Infrared Milli Terahz Waves (2014) 35:325381
order TE
gyro-backward waves are the most dangerous source of parasitic oscillations since
it is difficult to attenuate them, especially in high-electron-current MW gyrotrons [204,205].
Several improved beam tunnel damping structures have been developed:
&Arbitrarily non-axisymmetric conical metallic structures (IAP) [201]
&Conical silicon carbide (pure SiC) structures with weakly semi-conducting SiC in order to
avoid static charges (JAEA) [203]. SiC is a good mm-wave absorber.
&Conical alternating stack of BeO/SiC(60/40) ceramic damping rings and copper rings. The
copper rings are indented with a slot depth of about λ/4 in order to cut off azimuthal RF
currents and to break the azimuthal symmetry. The number, width and periodicity of the
indentations can vary irregularly (KIT). Such a structure can also be considered as a
grating with many different grating constants. This complicates the formation of a resonant
field structure and effectively damps TE
modes, in particular, symmetric TE
modes. A
schematic cross section of the modified beam tunnel is shown in Fig. 20 [204].
7.2 Improved Electron Beam Sweeping Over the Collector Wall
Vertical magnetic field sweeping systems (VFSSs) became a standard technique, which keeps
the time averaged specific heat dissipation by a spent electron beam at the collector within
technically acceptable values (< 500 W/cm
). As such collectors operate close to the techno-
logical limit, they are a high-risk component with a small safety margin.
To address this problem, KIT Karlsruhe in collaboration with IPP Greifswald investigated
and optimized a combined 7 Hz VFSS and a 50 Hz transverse field sweeping system (TFSS)
for the 140 GHz W7-X gyrotron with a cylindrical collector. The TFSS consists of 3 pairs of
TF-coils (Fig. 21, left), which are powered by a 3-phase AC-supply, thus generating a
transverse field, which rotates with 50 Hz frequency [207]. At the position of the coils the
collector is made of stainless steel in order to reduce the influence of eddy currents. This
section of the collector is not hit by the electron beam. By proper tuning of both magnetic
sweeping systems an almost perfectly flat power deposition profile along the collector area was
obtained. The peak loading was reduced by a factor of two as compared to VFSS alone as
shown in Fig. 21 on the right [223].
Fig. 20 Corrugated beam tunnel for suppression of parasitic oscillations [204]
J Infrared Milli Terahz Waves (2014) 35:325381 359
GYCOM in Russia introduced a more sophisticated, shaped (conical) profile of the inner
collector surface in order to increase the target area of the electron beam in the lower part of the
collector and thus to reduce the power deposition density [224].
7.3 Current State-of-the-Art of MW-Class CW Gyrotrons
The present state-of-the-art of industrial megawatt-class long-pulse fusion gyrotrons (devel-
oped at 140 GHz for the stellarator W7-X and at 170 GHz for the tokamak ITER) with TEM
output, CVD-diamond window, and single-stage depressed collector (SDC) is summarized in
Table 5. Photographs of the tubes are shown in Fig. 22 [206,225]. As follows from Table 5,
nowadays there are four companies in the world, which at frequencies above 100 GHz can
deliver MW-power level gyrotrons providing pulses exceeding 1000 seconds, i.e. gigajoule
energy in a single pulse.
In 2005, the first CPI 140 GHz long-pulse gyrotron was tested at IPP Greifswald and
delivered about 0.9 MW power in 30 minute pulses [226]. A few months later, the European
140 gyrotrons (KIT-CRPP-CEA-TED collaboration) outperformed this result: 0.92 MW at 30
min. pulse duration, 97.5% Gaussian mode purity and 44% efficiency [223,227229]. The
Japanese 170 GHz ITER gyrotron holds the energy world record of 2.88 GJ (0.8 MW, 60 min.)
Fig. 21 TED-gyrotron collector with three pairs of TF-coils (VF-coil removed) around the stainless steel part
(left) and corresponding profile of the average collector temperature increase ΔT along the vertical coordinate z
for combined collector sweeping TFSS (dots). (The profile for VFSS only is also shown by squares as a
reference) [223]
Tab le 5 State-of-the art of industrial megawatt-class long-pulse gyrotrons (f140 GHz)
Company/Institution Frequency Cavity mode Power Efficiency Pulse length
(GHz) (MW) (%) (sec)
CPI [226] 140 TE
0.9 35 (SDC) 1800
TED/KIT/CRPP [228] 140 TE
0.92 44 (SDC) 1800
GYCOM/IAP [232] 170 TE
1.0 53 (SDC) 1000
1.0 55 (SDC) 800
0.8 57 (SDC) 3600
360 J Infrared Milli Terahz Waves (2014) 35:325381
and the efficiency record of 57% at 1 MW, 800s for tubes with an output power of more than
0.5 MW [203,213,230,231]. This very high efficiency has been obtained by operation in the
so-called hard-excitation region, which can be achieved in gyrotrons utilizing either diode-type
or triode type electron guns. Especially easy it can be realized in the case of triode-type guns
with a modulating anode where the electron orbital-to-axial velocity ratio can be controlled
during the beam voltage rise by varying the relation between the mod-anode and beam voltages.
The Russian 170 GHz ITER gyrotron achieved 1 MW in 1000s pulses at 53% efficiency [224,
232]. The reliability of these megawatt-class gyrotrons is approximately 90% [233]. The
Japanese and Russian ITER gyrotrons operate in liquid-Helium-free superconducting magnets.
A 75 GHz, 0.8 MW, 0.1s TE
GYCOM gyrotron holds the world record efficiency of
70% (with SDC) and Gaussian output mode purity of 98% [232]. In Japan there is the
development of a 28 GHz, 1 MW tube with TE
mode cavity for the GAMMA10 tandem
mirror at the University of Tsukuba and a 77 GHz, 1 MW, 10s gyrotron with TE
cavity for the Large Helical Device (LHD) at Toki, Japan [234]. The aims of this recent
development is to get new record values of ion confining potential and electron temperature in
GAMMA10 and to upgrade the EC H&CD system of LHD with an option to allow Collective
Thomson Scattering (CTS) diagnostics.
Table 6summarizes running EC H&CD systems with installed gyrotron power > 2 MW
and frequency > 70 GHz.
7.4 Coaxial-Cavity and Advanced Cylindrical-Cavity Gyrotrons
A 2 MW, CW, 170 GHz coaxial-cavity gyrotron operating in the TE
mode for ECH&CD
in ITER was under development in cooperation between European Research Institutions
(EGYC) [235], which is a collaboration between CRPP, Switzerland; KIT, Germany; HEL-
LAS, Greece; CNR, Italy and ENEA, Italy.
The design of critical components like electron gun, beam tunnel, cavity and quasi-optical
mm-wave output system has been studied in a short-pulse coaxial gyrotron (pre-prototype) at
KIT. At the magnetic field of B= 6.87 T a maximum mm-wave output power P
Fig. 22 Megawatt-class long-pulse cylindrical cavity gyrotrons for ITER (170 GHz) and W7-X (140 GHz) [206,225]
J Infrared Milli Terahz Waves (2014) 35:325381 361
Tab le 6 Operated EC H&CD systems at tokamaks and stellarators with installed power > 2 MW and frequency > 70 GHz.
Plasma Device Institution ECRH Power Frequency Pulse Duration Company Coupled Energy
(Place) [M W] [GHz] [s] [M W x s]
D III-D Tokamak GA San Diego 6.0 (6 × 1.0) 110 5.0 CPI 3.4 × 5.0
TCV Tokamak CRPP Lausanne 3.0 (6 × 0.5) 82.7 2.0 GYCOM 2.7 × 2.0
1.5 (3 × 0.5) 118 2.0 CEA / CRPP / KIT / TED 1.3 × 2.0
JT 60-U Tokamak JAEA Naka 4.0 (4 × 1.0) 110 5.0 JAEA / TOSHIBA (SDC) 2.9 × 5.0
ASDEX-U Tokamak IPP Garching 3.0 (3 × 1.0) 140 (105) 10.0 GYCOM (SDS) 2.4 × 10.0
2.0 (4 × 0.5) 140 2.0 GYCOM 1.6 × 2.0
T-10 Tokamak NRI Kurchatov Moscow 1.0 (2 × 0.5) 129 0.5 GYCOM 0.8 × 0.4
2.0 (4 × 0.5) 140 0.5 1.6 × 0.4
FTU Tokamak ENEA Frascati 2.0 (4 × 0.5) 140 0.5 GYCOM 1.6 × 0.5
3.0 (3 × 1.0) 77 10.0 NIFS/TSUKUBA/JAEA/TOSHIBA(SDC) 2.3 × 5.0
0.5 (1 × 0.5) 82.7 2.0 GYCOM 0.4 × 2.0
LHD Stellarator NIFS Toki 1.6 (2 × 0.8) 84 3.0 GYCOM (SDC) 1.0 × 3.0
0.4 (1 × 0.5) 168 1.0 NIFS/TOSHIBA (SDC) 0.3 × 1.0
W7-AS Stellarator IPP Garching 0.5 (1 × 0.5) 70 2.0 CPI (SDC) 0.45 × 2.0
0.8 (1 × 0.8) 140 1.0 GYCOM (SDC) 0.65 × 1.0
1.5 (3 × 0.5) 140 3.0 GYCOM 1.2 × 3.0
362 J Infrared Milli Terahz Waves (2014) 35:325381
=93 kV and I
= 80 A with an efficiency of about 30% at 1 ms pulse
duration in non-depressed collector operation [236]. If a CVD-diamond window would be
installed instead of the fused silica window, the output power would be 2.3 MW with an
efficiency of 31% due to lower mm-wave absorption in diamond. The Gaussian output mode
purity is almost 96%. Reduced length of the quasi-optical launcher has been obtained by an
internal helical line of mirrors with mode-converting and phase-correction non-quadratic
surface contour functions optimized by iteratively solving the scalar diffraction integral
equation [237,238]. These techniques have been extended to launchers that work at
many frequencies. High conversion efficiencies of over 96% for several modes have
been reported, which opens a way for high efficiency multi-frequency gyrotrons [237]
(see 7.5).
The measured mm-wave output power and efficiency versus the cathode voltage are shown
in Fig. 23 [236]. The figure also contains the results of numerical simulations performed with
the KIT multi-mode, self consistent code SELFT taking into account 5% rms spread in
electron transverse velocities. The agreement between experiment and simulations is very
good, if approximately 10% of the generated output power in the nominal mode is assumed to
be lost inside the tube due to stray radiation, Ohmic losses and absorption in the output
RF tests of the industrial prototype gyrotron (TED) for ITER were done at the EU test
facility at Lausanne in December 2011 [239]. The industrial prototype did show excellent
voltage stand-off. It was possible to excite the nominal operating TE
mode. The output RF
beam profile was in good agreement with the expected profile. The output power reached the
level of almost 2.1 MW in short-pulse (1 ms) operation with SDC operation corresponding to
46% efficiency. The results have been achieved without any optimization after only 4 days of
operation. However, since at the next day an internal water-cooled alumina pipe stray radiation
load broke, the tube was lost and the EU ITER gyrotron development strategy changed to the
back-up solution of a 1 MW conventional cylindrical cavity gyrotron operating in the TE
Fig. 23 Measured and calculated gyrotron output power and efficiency vs. beam voltage at I
= 80 A and cavity
magnetic field of 6.87 T [236].
J Infrared Milli Terahz Waves (2014) 35:325381 363
Nevertheless, the development of coaxial-cavity gyrotrons will continue at KIT, because
very-high frequency gyrotrons (240 GHz and above) for ECH and efficient non-inductive
ECCD in future nuclear fusion reactors (like DEMO), probably, can generate unit-power of
1 MW and more, only if they employ coaxial cavities.
The gyrotron development teams in Japan [231,234,240], Russia [232] and USA [241]are
testing the power limits of conventional cylindrical-cavity CW gyrotrons. Table 7summarizes
the present state-of-the-art of advanced short-pulse 1.2 - 2 MW gyrotrons with TEM
output operating in the frequency range of 110-170 GHz.
7.5 Frequency Step-Tunable Gyrotrons for Suppression of Neoclassical Tearing Modes
One of the major roles of local EC H&CD in ITER and other tokamaks is the suppression of
Neoclassical Tearing Modes (NTMs) [209,242,243]. In order to compensate for the lack of
current in the center (O-point) of a magnetic island, external current drive is required inside the
island. The use of frequency step-tunable gyrotrons could enhance greatly the flexibility and
the performance of the ITER EC H&CD system. Two design options have been analyzed in
In case of a system based on a limited number of frequency steps in the range of 10-35 GHz
with steerable launchers already a two-frequency gyrotron, which is technically the simplest
variant of a multi-frequency gyrotron, could greatly enhance flexibility and performance, e.g.
NTM stabilization through the upper launcher at the higher frequency and CD at mid-radius
through the mid-plane launcher at the lower frequency.
A system based on frequency step-tunable sources with frequency steps of 2-3 GHz could
be used together with simple, fixed launcher structures without loss of performance. Gyrotrons
with ultra-broadband CVD-diamond windows and fast-tunable superconducting gyrotron
magnets (sweeping speed 0.2 T/5s corresponding to 1GHz/s) would allow stepwise frequency
tuning in the seconds time-scale in the full D-band (110-170 GHz) [208211].
At present, GYCOM/IAP is developing in collaboration with IPP Garching and KIT an
industrial, frequency-tuneable 1 MW gyrotron with almost 50% efficiency (SDC) for the new
ASDEX Upgrade ECH system. A four-frequency tube (105, 117, 127 and 140 GHz (with the
cavity mode at 140 GHz)) delivered in 10s pulses 0.8 MW at 105 GHz and 0.9 MW at
140 GHz (two-frequency gyrotron) [232]. The single-disk CVD-diamond window has max-
imum transmission for these two frequencies. After the installation of a circularly brazed CVD-
diamond Brewster window with specific internal and external mirrors mounted closely to the
window disk, the GYCOM/IAP group operated this gyrotron also at the two intermediate
frequencies. However, the CVD-diamond disk broke due to successive RF arcing in front of
the disk. Therefore GYCOM is now developing a tuneable travelling-wave window consisting
of two CVD-diamond disks and two mirrors (see Fig. 24).
In contrast to the GYCOM approach, KIT is developing a step-frequency tunable D-band
MW gyrotron using an elliptically brazed CVD-diamond window. The cavity mode at
140 GHz is also TE
. The dimensions of the diamond disk are 139 x 95 x 1.7 mm
(Fig. 25)[244].
The elliptically shaped disk has bee