Chapter 15

Accelerators for Protons and Other Heavy Charged Particles

Alejandro Mazal and Annalisa Patriarca

CONTENTS
15.1 Introduction........................................................................................................................................................... 295
15.2 Types of Accelerators for Proton Therapy................................................................................................................ 296
15.2.1 Circular Accelerators.................................................................................................................................... 296
15.2.1.1 The Cyclotron............................................................................................................................... 296
15.2.1.2 The Synchrocyclotron.................................................................................................................... 297
15.2.1.3 The Isochronous Cyclotron........................................................................................................... 298
15.2.1.4 The Synchrotron........................................................................................................................... 298
15.2.2 Linear Accelerators...................................................................................................................................... 299
15.2.2.1 Dielectric Wave Linear Accelerators (DWAs)..................................................................................300
15.3 Types of Accelerator for Therapy with Heavy Ions...................................................................................................300
15.3.1 Synchrotron Facilities for Carbon-Ion Therapy............................................................................................300
15.3.2 Cyclotron Facilities for Carbon-Ion Therapy................................................................................................ 301
15.3.3 Linac Facilities for Carbon-Ion Therapy....................................................................................................... 301
15.4 New On-Going Developments in Accelerators........................................................................................................ 301
15.4.1 Laser-Based Particle Accelerators................................................................................................................. 302
15.4.2 Fixed Field Alternating Gradient Particle Accelerators.................................................................................. 302
15.5 General Technical Design Considerations............................................................................................................... 302
15.5.1 Extraction................................................................................................................................................... 302
15.5.2 Vacuum and Cryogenic Systems................................................................................................................... 302
15.5.3 Ironless Magnetic Circuits........................................................................................................................... 302
15.6 Beam Transport...................................................................................................................................................... 302
15.7 Economic and Functional Aspects.......................................................................................................................... 303
15.7.1 Capital Costs............................................................................................................................................... 303
15.7.2 Running Costs............................................................................................................................................304
15.7.3 Treatment Capacity.....................................................................................................................................304
15.7.4 Cost of Treatment.......................................................................................................................................304
15.8 Perspectives............................................................................................................................................................304

15.1 Introduction The possibility of using such particles for radiotherapy

was suggested as early as 1946 (Wilson 1946). However, to
Modern radiotherapy is principally based on photon beams reach depths of the order of 20 cm–30 cm in tissue, proton
produced by electron linear accelerators. However, the depth energies of several hundred MeV are required. The produc-
dose distribution of photon beams is not optimal for treating tion of such beams is possible only with huge and expensive
deep-seated tumours because of the photon attenuation and accelerators such as those used for atomic and nuclear physics
dispersion, resulting in a continuous decrease of absorbed research from the 1950s. At the time when the reference
dose as a function of depth. Therefore, the use of multiple book of Raju (1980) was written, the accelerators used for
beam combinations is required for such beams. The elec- clinical applications were limited; for example, for protons,
trons produced by linear accelerators can also be used clini- there was the Harvard 160 MeV cyclotron, the 187 MeV
cally, but due to the importance of scattering, the depth dose synchrocyclotron in Uppsala and in the Soviet Union, the
distribution is not ideal (see Figure 24.3). Heavier charged degraded beams from the 680 MeV synchrocyclotron of
particles (protons, helium or carbon ions, etc.) have more Dubna, the 7.2 GeV synchrotron in Moscow and the 1 GeV
favourable properties with a rapid increase of dose at the end synchrocyclotron in Gatchina, and a few other research
of their range, commonly referred to as the Bragg peak (see facilities in the world. Since that time, many new proton
Sections 3.7 and 25.2). therapy facilities have been installed throughout the world,

295
296 Part C: Equipment

most of them based on dedicated turnkey equipment espe- cases, up to 330 MeV to pass through the patient’s body)
cially designed for medical use. and beam intensities of several tens or hundreds of nano-
The principal components of a proton therapy facility are amperes to deliver dose rates of the order of 2 Gy min−1 or
the following: more. Such accelerators are commonly used for atomic and
nuclear physics research, but specific performance character-
• The accelerator with its associated systems; istics are required for clinical use, such as stability, maximal
• The beam transport system allowing beam delivery uptime and dedicated control systems. The status of particle
into one or more rooms delivered through fixed therapy, including accelerators and projected new systems,
lines or isocentric gantries; together with economic and strategic considerations, has
• The shielding assembly; been reviewed by Bonnett (1993), Schwartz et al. (1995),
• The beam delivery and shaping devices; Amaldi et al. (2010), Peach et al. (2011).
Electrostatic accelerators (van der Graaf, tandems, etc.) are
• The patient verification and positioning system;
unable to produce sufficiently high energies for clinical appli-
• The machine control system and a full set of ancil- cations. With current technology, circular accelerators are
lary systems. the standard solution adopted nowadays and have the best
cost–performance ratio, while linear accelerators are under
In this chapter, we describe the general principles and the construction as a new option.
specifications of the different types of accelerators used for
therapy with protons or heavier charged particles as well as
the related systems. The beam-shaping devices used clinically 15.2.1 Circular Accelerators
are described in Chapter 25. Supplementary information
The general principle of circular accelerators is the following:
may be found, for instance, in the report on Clinical Proton
a magnetic field bends the trajectory of particles in spiral or
Dosimetry published by the International Commission on
quasi-circular trajectories, and a single cavity can accelerate
Radiation Units and Measurements (ICRU 1998) or in
the particles at each rotation until extraction. The two main
Owen et al. (2014). Pearson et al. (2016) list about 50 facili-
types of circular accelerators used for clinical applications are
ties based on the use of cyclotrons and 20 facilities based on
cyclotrons (with sub-types such as synchrocyclotrons and iso-
synchrotrons that are in clinical operation now or will be in
chronous cyclotrons) and synchrotrons.
the near future. CERN (2017) published a report on accel-
erators for medical applications, including detailed aspects
for protons and heavy ions. 15.2.1.1 The Cyclotron
Neutron therapy, which gives dose distributions similar to
Figure 15.1 presents a schematic diagram of a cyclotron, con-
photon beams but with high linear energy transfer (LET)
ceived and patented by E.O. Lawrence in 1932 (Lawrence
local dose deposition (see Section 4.5.4), was still considered
and Livingston 1932). The ion source (e.g. a plasma produced
in the 1980s as useful for some radioresistant tumours. The
by an electric arc from a tungsten filament in hydrogen gas)
beams were produced from protons or deuterons interact-
injects the protons into the centre of the machine. A high-
ing with a target. The energy of these primary particles was
frequency alternating high voltage is applied to two hollow
between 14 MeV and 70 MeV, mostly produced by cyclotrons.
semi-circular electrodes (called dees because of their shape).
This method of treatment has now practically disappeared.
When a proton is in the electric field between the two elec-
Low-energy (epithermal) neutrons produced in nuclear
trodes, the sector of negative polarity exerts a force of attrac-
reactors are used for radiation therapy in a few centres,
tion and therefore, produces acceleration. A magnetic circuit
combined with an injection of boron-10 concentrated in
and a set of coils are used to create a strong magnetic field
the tumour, in what is called boron neutron capture therapy
perpendicular to the trajectory of the particle. When the pro-
(BNCT). The interaction selectively produces lithium-7 and
ton re-enters the cavity of the electrodes, no electric force
alpha particles, with a high local deposition of energy at
acts on it. The magnetic field B bends the particle of charge
the cell level. The technique is applied for highly resistant
tumours such as gliomas and, for technical reasons of energy
and accessibility, cutaneous melanomas. New accelerators 4
are under development to improve the beam specifications
1
in terms of energy, efficiency and reliability using what are 2
called accelerator-based neutron sources (ABNS), for exam-
ple with proton beams incident on a solid lithium-7 target 9 3
(Koivunoro et al. 2015). 5

8
7
6
15.2 Types of Accelerators for Proton Therapy
Accelerators producing proton beams for medical use gener- FIGURE 15.1 The principle of the cyclotron: (1) magnetic circuit; (2)
magnetic field; (3) coils; (4) gas injection; (5) ion source; (6) high-fre-
ally have energies between 230 MeV and 250 MeV to pene-
quency high voltage; (7) accelerating electrodes (dees); (8) extraction;
trate tissues to a depth of the order of 32 g cm−2 (and in some (9) vacuum systems.
Chapter 15: Accelerators for Protons and Other Heavy Charged Particles 297

q moving with a velocity v into a circle of radius r. It emerges 15.2.1.2 The Synchrocyclotron

again into the space between the dees exactly when the elec-
When the particle energy is such that relativistic effects cause
tric field has changed its direction, being accelerated again
an increase in the mass of the accelerated protons, the fre-
and going on a circular trajectory with a larger radius, and
quency of rotation of the protons, ω, decreases (Equation
so on.
15.2), so the frequency of the electric field between the dees
The operating principle of the cyclotron is thus expressed
of the cyclotron can no longer be constant if synchronicity is
by equating the centrifugal force on a particle of mass m at a
to be retained. As the magnetic field in a conventional cyclo-
distance r to the Lorentz force exerted on it by the magnetic
tron must decrease with radius to provide vertical focusing
field B:
of the beam, this also affects the frequency of rotation of
q v B = m v 2 /r or qB r = mv (15.1) the particles. To maintain synchronicity between the elec-
tric field and the passage of the particles, a modulation of
The force, and therefore the trajectory are perpendicular to the frequency is necessary. Thus, the frequency is reduced
the magnetic field. The charge, the magnetic field and the when the mass of the proton increases with energy at large
sub-relativistic mass are constant, and therefore, an increase radius, requiring a longer time to move through the dee.
in velocity results in an increase in the radius of the trajectory, This principle is applied in the synchrocyclotron (also called
with an angular frequency of rotation ω that depends only on a frequency-modulated cyclotron), conceived independently
the magnetic field and on the mass and charge of the particle: by Veksler (1946) in Russia and McMillan in America, and
patented by McMillan in 1952 (McMillan 1952).* The fre-
w = v /r = q B /m (15.2) quency variation is obtained by using a variable capacitor in
the high-frequency circuit. In practice, this is achieved by a
Typical values for conventional cyclotrons accelerating pro-
rotary condenser (or rotating capacitor), which is a complex
tons are magnetic fields of 1 T to 2 T and frequencies of sev-
device, or by a solid-state equivalent amplifier. The disadvan-
eral MHz (tens to hundreds). The accelerating electric field
tage of the synchrocyclotron is the production of a pulsed
between the dees remains at this constant frequency, stay-
beam instead of a continuous one, but the usual repetition
ing in phase with the passage of the particles, providing a
frequencies of the macro pulses (e.g. 450 Hz) are still well
microstructure of the beam with pulses at the high radiofre-
adapted to clinical use of the beam. The proton therapy facil-
quency (RF) but a macroscopic continuous wave beam. There
ities at Harvard (Boston, USA), Uppsala (Sweden) and Orsay
are various other considerations in addition to these guiding
(France), originally based on research machines converted for
principles of operation. Among the principal ones are the
clinical use, are examples of synchrocyclotrons that were used
problems of stability of phase, focusing and extraction (see
for treating patients during the period 1973–2015. Several of
Section 15.5). The magnetic field tends to decrease near the
the most compact recent solutions for clinical accelerators are
edges of the magnets, curving the lines of force and impart-
also based on synchrocyclotrons (Derenchuk 2013).
ing radial components to the field. This results in focusing
Ion Beam Applications (IBA, Louvain, Belgium) has com-
forces that keep the particles oscillating around the central
mercialised a 50 tons, 2.5 metres diameter, 6.5 teslas peak
plane; the magnetic field must decrease with the radius to
magnetic field, superconducting synchrocyclotron producing
keep these oscillations bounded.
a proton beam of 230 MeV working with pulses at 1 kHz.
The electric field between the dees not only accelerates the
It is named Proteus One and is designed with a short trans-
protons but also focuses and defocuses the beam, with a net
port system towards a single treatment room, the full system
effect depending on the phase of the particle. To maintain
requiring a space of about 30 m × 15 m. Clinical applica-
synchronisation, the particle must arrive at the accelerating
tions with the first delivered system started in 2017 in Nice,
gap when the voltage is increasing, during a vertical defocus-
France.
ing stage. The beam diameter increases with radius at first with
Mevion Medical Systems (Littleton, MA) has reduced the
defocusing of the electric field, and after half the final radius,
size of the system even more with a 20 tons, 1.8 metres diam-
the magnetic forces focus the beam again. Since the magnetic
eter, 9 teslas peak magnetic field, superconducting synchro-
field must decrease with radius to keep the beam focused, there
cyclotron producing a proton beam of 250 MeV working
is also a change in the equilibrium between the centrifugal and
with pulses at 500 Hz. The most singular characteristic of
the magnetic forces. If the field decreases more slowly than 1/r,
this system is that the synchrocyclotron is directly mounted
the net effect is radial oscillations with a final focusing effect.
on an isocentric gantry that rotates 190º around the patient.
Vertical and radial oscillations are known as betatron oscilla-
The required space is of the order of 200 m 2 with a height of
tions, as they were first described for those machines.
about 10 m, depending on the final shielding design. Clinical
A detailed analysis of the mathematics of the basic effects
applications started in 2013 in St Louis, Missouri.
lies outside the scope of this chapter and can be studied in
specialist references (e.g. Rosenblatt 1968).
The conventional cyclotron does not directly fit the specifica-
tions needed for radiation therapy (e.g. for the required ener-
gies). Different approaches to acceleration have been developed * Veksler and McMillan independently realised that the frequency did
not have to be precisely matched to the speed of rotation of all the
to solve these issues, producing a fixed energy beam either with particles, because particles going too fast would not be accelerated as
pulsed beams (synchrocyclotron) or with quasi-continuous and much, and vice versa. This meant that the synchronicity was inher-
high-intensity currents (isochronous cyclotron). ently stable – so-called phase stability.
298 Part C: Equipment

15.2.1.3 The Isochronous Cyclotron A superconducting version was developed by Varian (after

acquisition of the Accel company). In 2016, such machines
Another way to solve the relativistic effect on the particle
were operational in Munich, Switzerland, San Diego and
mass is to increase the magnetic field strength B towards the
Baltimore, with several other centres under construction
outside of the circular dees (Equation 15.1 and Equation
(including the two National Health Service [NHS] centres
15.2). This solution, however, has limitations in a cyclotron,
in the UK). The Varian isochronous cyclotron is a supercon-
because it is associated with beam defocusing, resulting in a
ducting cryogenic system of 90 tons, diameter of 3.1 metres,
loss of particles due to collisions with the walls of the vacuum
magnetic field close to 4 teslas, producing a 250 MeV beam
chamber and the magnetic pole.
of up to 800 nA quasi-continuous current (Derenchuk 2013).
The required focus can be re-established by using alter-
nating gradients of the magnetic field along the trajectory.
The accelerator has several separated sectors in a spiral con- 15.2.1.4 The Synchrotron
figuration (alternation of peaks and valleys – Figure 15.2).
The diameter of a cyclotron increases with the required
At the boundary of these sectors, the magnetic field plays
energy, and in consequence, so do the weight and the cost
a role equivalent to an optical lens, focusing and defocus-
of the magnetic circuit and other accessories. If we consider
ing the beam. It is then possible to increase the magnetic
only the final orbit in the trajectory of the particle within a
field with increasing radius to compensate for the increase in
cyclotron, and we try to keep the particles under accelera-
mass of the particles by bringing the pole tips closer together
tion in that ring, Equation 15.1 shows that the velocity v of
without losing focusing and thereby, maintain a constant
the particle can be increased (taking also into account the
frequency for the accelerating potential (and therefore, a
change in mass) at constant radius r, with a variable magnetic
continuous beam). Such machines are known as isochronous
field B in the annular region occupied by the acceleration
cyclotrons. The acceleration is provided by resonating radio-
chamber.* This is the principle of operation of a synchrotron
frequency cavities connected to the accelerating electrodes in
(Figure 15.3), developed by McMillan (1945) and Veksler
the valleys. Ion Beam Applications (IBA, Belgium) installed
(1946). The particles are accelerated up to several MeV by a
an isochronous cyclotron in Boston in 2001, and there were
linear accelerator (e.g. consisting of a radio-frequency quad-
more than 80 machines in operation in 2019. The IBA iso-
rupole [RFQ]) before being injected into a ring that includes
chronous cyclotron is a normal-temperature device of 220
bending dipoles, focusing quadrupoles (and sometimes sex-
tons, a diameter of 4.3 metres, a magnetic field of 2.2 teslas
tupoles), injection and extraction modules, one or more RF
and a power of 320 kW, producing a beam of 230 MeV with
cavities, and the vacuum system. At each revolution, a high-
a quasi-continuous beam current of 300 nA. The conception
frequency cavity, similar to those used for electron linacs (see
and production have been shared with Sumitomo, Japan,
Section 11.2.1), produces an acceleration synchronised with
who also offers the same accelerator.
the angular frequency of the particles. The deflecting dipoles
increase the magnetic field to maintain the trajectory of the
protons at a constant diameter until they reach the desired
energy and are extracted. This type of accelerator thus
makes it possible to produce a beam with variable energy,
with the disadvantage that one obtains a pulsed beam of low

5 4

7 3
1 2
1
2
8
3 4 5 6 6

FIGURE 15.3 Principle of the synchrotron: (1) ion source; (2) RFQ
(radio-frequency quadrupole); (3) bending dipoles and focusing quad-
rupoles; (4) injection; (5) bending dipoles with variable magnetic field;
FIGURE 15.2 The isochronous spiral separate sectors cyclotron: (6) linear section, focusing; (7) radiofrequency cavity for acceleration;
(1) magnet yoke; (2) coils (may be superconducting); (3) valley; (4) elec- (8) extraction (as installed e.g. in Loma Linda, CA).
trodes, stems and resonators for acceleration in a valley; (5) radiofre-
quency generator and ion source; (6) magnet peak; (7) extraction (as * This variation in the magnetic field in the synchrotron is an alternative
installed e.g. in Boston, MA). to the variation in frequency in the synchrocyclotron.
Chapter 15: Accelerators for Protons and Other Heavy Charged Particles 299

1 2 3 4 5

FIGURE 15.4 The principle of the proton linear accelerator: (1) source; (2) RFQ (radio-frequency quadrupole) for first bunching and acceleration;
(3) associated cavities for acceleration at medium energies; (4) beam bending and focusing; (5) last series of tanks with accelerating cavities for high
energy. (Modified from Benincasa, G., et al., High frequency proton linac in: The RITA Network and the Design of Compact Proton Accelerators,
Vol. 2, Part 1, Amaldi, U., et al., INFN Divisione Ricerca, Frascati, 1996.)

pulse-repetition frequency. Typical acceleration cycles are independent accelerator for proton therapy. A typical linear
between 1 second and 3 seconds. accelerator for proton therapy has as its main components
Examples of synchrotrons developed for proton therapy are an ion source, different stages of accelerating RF cavities,
those manufactured by Optivus for Loma Linda in California RF high-voltage sources and amplifiers, a vacuum conduit,
(Slater et al. 1992) and by Mitsubishi and Hitachi for cen- focusing magnets and a control system (Figure 15.4).
tres in Japan and the US. The proton synchrotron installed A common source for proton accelerators is the electron
in Loma Linda in 1990 is a zero-gradient (uniform field cyclotron resonance (ECR) ion source. It uses microwaves
inside the magnet gap), weak focusing type (the magnets have and magnetic fields in a small volume with a selected low-
a combined function of bending and focusing) designed by pressure gas (e.g. H) to accelerate free electrons and to ionise
Fermilab. It produces a beam of 70 MeV to 250 MeV. The the gas to produce plasma. Selected ions (e.g. protons) will
source is a dual plasmatron that feeds protons at 37 keV into a be extracted and accelerated at high intensities (hundreds of
1.6 m RFQ, which accelerates them to 2 MeV and injects them mA) in direct current mode and are then injected into the
into the synchrotron ring of 20 m circumference. The beam first stage of acceleration. Other types of source exist and are
is extracted by resonance with a frequency of 0.5 Hz, with in use or under study; for example the Duo Plasmatron, in
a flat top* of 400 ms and an initial beam current of 3 × 1010 which a cathode filament emits the electrons that produce
protons per pulse giving a maximal dose rate of 1 Gy min–1. the plasma.
This performance was subsequently upgraded (Alonso 1994; As for synchrotrons, a radio-frequency quadrupole (RFQ)
Coutrakon et al. 1994). is the usual component for the first low-energy stage of
Synchrotrons are widely used in Japan. Examples of strong focusing, bunching and accelerating high-current and low-
focusing solutions (where separate magnets are used, such emittance charged particle beams (e.g. from tens of keV to
as dipoles for bending and quadrupoles for focusing) exist a few MeV), achieving a high efficiency. It consists of a cyl-
in Tsukuba and a multi-ion system at Wakasa Bay. Hitachi inder with four internal tips or roads as longitudinal elec-
installed its first synchrotron-based proton therapy centre trodes, top–bottom in opposite phase to the left–right of an
in the US at the MD Anderson Cancer Center in Houston, RF electric field (see Figure 15.5). This produces an alternat-
Texas, in 2006. The accelerator has six bending magnets and ing focus–defocus effect with a final slight focusing one. The
a set of focusing and defocusing quadrupoles in a ring of tips have a non-flat corrugated or modulated surface on their
7 m × 7.8 m, producing a beam with a flat extraction between inner side, providing a longitudinal accelerating component
0.5 s to 5 s and a time between spills† from 2 s to 6.7 s to of the electric field. The RFQ was invented by Kapchinskij
deliver a beam with variable energies between 70 MeV and and Teplyakov in 1970 (Lombardi 2006).
250 MeV (Mohan et al. 2017).
The Protom Radiance 330 synchrotron installed in
Michigan, produces pencil beam energies from 70 MeV
to 250 MeV for treatments and up to 330 MeV for proton
radiography, delivered with a maximal range of 37.9 g cm−2
adjustable to within 0.1 g cm−2 (and 0.05 g cm−2 for low ener-
gies) (Nazaryan et al. 2014).

15.2.2 Linear Accelerators
Linear accelerators are used currently as injectors for syn-
chrotrons and are also under construction as a complete

* The flat top describes the point at which the particles have reached
their required energy and are extracted with a constant intensity dur-
ing a wide pulse.
† A spill is the term used to describe the extraction of part of the beam FIGURE 15.5 A radio-frequency quadrupole. (From Lombardi, A. M.,
from the ring. The Radio Frequency Quadrupole, CERN, Geneva, 2006.)
300 Part C: Equipment

A series of RF cavities (cells) is needed after the first stages the size of the accelerator small. As the planned accelerat-
to produce proton beam energies between 60 MeV and 230 ing field to be achieved with HGIs is up to 100 MV m –1 for
MeV. A simple first model (Widerøe accelerator, Section a nanosecond pulse, this would result in a proton DWA of
11.2.2) is a series of drift tubes (DTL) of increasing length, about 2 m in length (Alonso 2011). After the cooperation
where the ions drift at a constant velocity into each tube and of LLNL with TomoTherapy in 2007, the company CPAC
are accelerated by oscillating fields in the gaps between tubes. (Compact Particle Accelerator Corporation) was established
To avoid problems of size and RF losses at high frequencies, to take charge of the commercialisation of the DWA. In
the gaps are surrounded by a resonant tube (Alvarez accel- 2012, a DWA prototype with a 20 MV m–1 gradient was
erator) or presented as a series of coupled cavity arrays. Beam successfully tested with no gantry. Three-dimensional beam
focusing can be achieved with quadrupoles. With rather scanning with a fixed beam is forecast with an upgrade of the
strong electric field gradients (e.g. 20 MV m−1 to 30 MV m−1), maximum energy to 215 MeV. For the time being, the feasi-
it is necessary to avoid resonant growth of surface electron bility and the clinical application of a DWA accelerator must
currents (multipactoring), which lead to RF energy losses, still be proved, and its development seems to have stopped.
affect the electric fields and may lead to a cavity breakdown.
Bogomolov (1972, 1994) presented concepts in linear ion
accelerator technology based on a travelling-wave electro-
magnetic field.
15.3 Types of Accelerator for
In 1998, the first design of the 62 MeV to 200 MeV LIBO Therapy with Heavy Ions
(LInac BOoster) was achieved. The prototype, accelerating Heavy ions are of great interest for clinical applications based
from 62 MeV to 74 MeV using a 1.3 m long, 3 GHz side cou- on their physical dose distribution using the Bragg peak and
pled linac (SCL) with a gradient of 15.7 MV m−1, consisting also on their relative biological effectiveness (RBE). The
of four accelerating tanks, 23 half-cell-plates each, was tested most usual beam is of carbon ions (see Chapter 25). Initially,
successfully first at CERN (Amaldi et al. 2004) and then their use was based on accelerators built for research in phys-
at the Institute Nazionale di Fisica Nucleare – Laboratori ics (e.g. in Berkeley), but they have since evolved into com-
Nazionali del Sud (INFN-LNS), using the LNS cyclotron mercial turnkey clinical solutions (e.g. Chiba and four other
as injector (De Martinis et al. 2012). The energy modula- facilities in Japan and more than 10 in the world).
tion, obtained by adjusting the number of active klystrons
and adjusting the power sent to the last active one (Amaldi
15.3.1 Synchrotron Facilities for Carbon-Ion Therapy
et al. 2009), is an evident advantage in terms of radiation
protection issues because of the absence of passive degraders. The standard accelerator for particle therapy with heavy ions
A system called LIGHT (Linac Image-guided Hadron is the synchrotron because of the ease with which the heavy
Technology) is under development by Advanced Oncotherapy ions’ energy can be increased. The pioneer work at Berkeley
(Meyrin, Switzerland), based on a four-stage accelerator: started in 1977 at the Lawrence Berkeley Laboratory’s (LBL)
the source, an RFQ increasing the proton beam energy to 5 Bevalac accelerator, producing ions from 4He to 28Si. Beams
MeV, a low-speed accelerator based on side coupled drift tube of 20Ne with energies of 450 MeV u−1* and 585 MeV u−1 have
linacs (SCDTLs) increasing the energy to between 20 MeV been the most commonly used (Chu et al. 1985).
and 40 MeV, and a high-speed accelerator based on a coupled The Heavy Ion Medical Accelerator (HIMAC) operating
cavity linac (CCL), following the LIBO-tested design. It is at the National Institute of Radiological Sciences in Chiba,
intended to produce a 230 MeV proton beam in a total length Japan, since 1994 is based on a linac injector and a dual ring
of 25 m, with fast energy switching (200 times per second) synchrotron (indeed, two accelerators superposed) producing
and low neutron production. The first system is planned to beams from protons up to Xe and energies from 100 MeV u−1
be installed in London and to be operational in 2019–2020. to 800 MeV u−1. To produce carbon ions, the ECR source pro-
duces a C2+ beam, which is injected into an RFQ and Alvarez
linac and accelerated to 6 MeV u−1. The carbon beam is fully
15.2.2.1 Dielectric Wave Linear Accelerators (DWAs)
stripped with a thin carbon foil and injected into the synchro-
The acceleration technique of DWA was patented in 1996 tron (a 12-dipole system), where the particles execute multi-
at Lawrence Livermore National Laboratory (LLNL) ple traverses of the ring (or multiple turns). The beam is then
(Sampayan et al. 1998). An electromagnetic wave is generated extracted from the ring (referred to as a spill). The spill can
by an external laser source, inducing very short (nanosecond) have a variable duration. In the HIMAC synchrotron, multiple
current pulses, which travel through a tube surrounded by energies can be extracted in one cycle. This is achieved with
dielectric material (see Figure 15.6). This electromagnetic multiple energy steps, which can be extracted in a slow extrac-
wave excites an electric field, accelerating the protons pro- tion spill. The facility has been upgraded with three new treat-
duced by a dedicated ion source. High electric field gradi- ment rooms connected to the existing HIMAC accelerator.
ents are obtained in the so-called Blumline transmission line.
High-gradient insulators (HGIs) alternating insulators and
conductive material form the inner wall of the accelerator * u is one unified atomic mass unit, being 1/12th of the mass of a car-
bon nucleus. 1 u ≈ 1.66 × 10 –27 kg ≈ 931.5 MeV/c2. It represents
tube. HGIs prevent the generation of electron avalanches, approximately the rest mass of one nucleon (see Section 1.2.2.3). The
multipactoring and breakdowns (Caporaso et al. 2009), and approximate kinetic energy for a given ion of atomic mass A is then
the accelerating electric field can be increased while keeping obtained by multiplying the nominal energy (MeV u –1) by A.
Chapter 15: Accelerators for Protons and Other Heavy Charged Particles 301

FIGURE 15.6 The principle of the dielectric wall accelerator. (From Mazal, A., Habrand, J. L., Delacroix, S., Datchary, J., Dendale, R., Desjardins,
L., et al., Bull. Cancer, 97 (7), 831–846, 2010. Modified from Caporaso et al. patent, USA, 1996)

A new downsized accelerator has been built in Gunma, INFN-LNS) with an injector capable of delivering protons
Japan, and started operation in 2010, with carbon-ion ener- and C+6 carbon ions. The linac booster, CABOTO (Carbon
gies from 140 MeV u−1 to 400 MeV u−1 and a maximum BOoster for Therapy in Oncology), a 3 GHz SCL with 400 Hz
dose rate of 1.6 Gy min−1 per litre (isoeffective to a photon repetition rate and 1.5 µs pulse length, will be capable of
dose rate of 5 Gy min−1). Based on this design, other instal- accelerating carbon ions from 300 MeV u−1 to 430 MeV u−1
lations have been constructed in Tosu (SAGA accelerator) in in 22 m. The resulting accelerating gradient is 25 MV m−1.
2013 and in the Kanagawa Prefecture (the i-Rock Ion Beam Subsequently, another CABOTO design has been proposed
Radiation Oncology Center in Kanagawa, which is able to (Degiovanni et al. 2010). The injector is a 120 MeV u−1 K480
produce a higher energy, up to 430 MeV u−1). superconducting cyclotron (190 tons and 4 metres diameter
The facility at Heidelberg, Germany (with the first carbon- with a superconducting electron beam ion source producing
ion rotational gantry) has been the base for the development 108 C+6 in 3 µs at 300 Hz). The carbon-ion booster has 18 mod-
of centres in Shanghai and in Marburg, all of them based on ules fed by klystrons, each with 12 MW peak power. The inno-
synchrotrons with the support of Siemens. A facility in Pavia, vation of this linac is in the operating frequency, 5.7 GHz, and
Italy, and MedAustron in Austria have been conceived on also in the higher accelerating gradients, 40 MV m−1, allowing
synchrotrons designed with the support of CERN. the total length to be reduced to 24 m to accelerate ions from
120 MeV u−1 to 400 MeV u−1.
Multi-ion accelerators are also being studied for clinical
15.3.2 Cyclotron Facilities for Carbon-Ion Therapy applications, based on an optimal choice of specifications for
IBA is developing together with the JINR (Joint Institute of particular parameters. An intriguing idea is that of using a
Nuclear Research, Dubna, Russia) a 400 MeV u−1 supercon- beam of antiprotons; as a result of antiproton–proton annihi-
ducting isochronous cyclotron as the first of its type. Carbon lation, the height of the Bragg peak is approximately double
and alpha particles will be accelerated to 400 MeV u−1, while that of a proton beam of the same energy and entrance dose
protons will be accelerated to 265 MeV. The magnet yoke level (Hall 2006; Holzscheiter et al. 2016).
is planned to have a 6.6 m diameter with a total weight of
700 tons. The magnetic field will be 4.5 T in the peaks and
2.45 T in the valleys. The superconducting coils will be 15.4 New On-Going Developments
enclosed in a cryostat, and the rest of the cyclotron will be
‘warm’ (Pearson et al. 2016). A potential site for installation
in Accelerators
of the first unit is in Caen, France. Much research and development is being devoted to new
alternatives to accelerate protons and heavy particles for clini-
cal applications. As well as reducing the cost and size of the
15.3.3 Linac Facilities for Carbon-Ion Therapy
accelerator, the goals are to improve the performance in terms
Different proposals are being built to use linacs for carbon of energy, energy spread, beam intensity, intensity modula-
therapy. The TERA Foundation proposed the design of a tion, stability, uptime, servicing and operating conditions.
Cyclinac-based carbon-ion therapy facility (Amaldi 2007). Two of these projects are described here: laser-based particle
The first project planned the use of a 300 MeV u−1 supercon- accelerators (Section 15.4.1) and the Fixed Field Alternating
ducting cyclotron (350 tons, 5 metres diameter developed at Gradient (FFAG) particle accelerators (Section 15.4.2).
302 Part C: Equipment

15.4.1 Laser-Based Particle Accelerators 15.5.2 Vacuum and Cryogenic Systems

High-intensity lasers (several hundred TW), which can To avoid disturbing the acceleration of the beam by absorption
deliver ultra-short pulses (e.g. approximately 50 femtosec- or scattering, a high vacuum (e.g. 10 –6 mbar [10 –4 Pa]) is main-
onds at a wavelength of 0.82 µm, with an energy of 1 J and tained in the accelerating chamber. This vacuum must also be
repetition rate of 10 Hz), will ionise a solid or gas target, maintained along the path of the beam. For negative ions, the
creating a plasma state. Subsequent interaction of the pulse vacuum requirements are more severe to minimise interactions.
with the created plasma leads to the generation of longitudi- The most recent developments of compact accelerators are
nal electric fields that ultimately accelerate the charged parti- based on the use of superconducting coils, which require the
cles. The generated electric fields can be greater than several use of cryogenic systems based, for example, on cryostats.
hundreds of GV m−1, arising from either electron expulsion Vacuum and cryogenic systems are critical parts, which
from the target (in the case of solid targets) or excitation determine the capacity of a facility to return to clinical opera-
of the plasma waves (in the case of gas targets) induced by tion after a breakdown or maintenance. This is of paramount
the light pressure of the high-intensity laser. These fields can importance for medical applications.
be used to accelerate electrons and protons up to the ener-
gies required for clinical applications in a few millimetres
15.5.3 Ironless Magnetic Circuits
(Malka et al. 2002). Fourkal et al. (2002) described simula-
tions showing that under optimal interaction conditions, a Magnetic fields are limited in strength by the saturation of
petawatt laser field could accelerate protons up to energies of the iron in the magnetic circuit and by the current neces-
300 MeV. Protons at energies between 10 MeV and 50 MeV sary in the coils. The maximum magnetic fields produced by
have been produced experimentally but with very low repeti- superconducting coils are limited by saturation of the iron
tion frequencies. Increasing the laser intensity, frequency and core. Ironless magnetic circuits have been proposed to solve
focusing, as well as optimising the target design, have been these problems, reducing costs and giving the possibility to
proposed as solutions for developing compact devices for the change the energy at the Cyclotron (Minervini et al. 2018).
next generation of particle accelerators. Other aspects, such
as the beam energy spectra (Fourkal et al. 2003), intensity
and stability, need further improvements.
15.6 Beam Transport
15.4.2 Fixed Field Alternating Gradient It is common practice for accelerator facilities devoted to
Particle Accelerators physics research to have several beam lines. This is because
of the cost and the size of accelerators and the multiple users
A new idea for accelerating protons and heavier ions (such involved in separate research programmes.
as carbon) has emerged, the so-called non-scaling fixed field Similarly, for medical use, the high cost of the accelerator is
alternating gradient (NS-FFAG) accelerator (Keil et al. a justification to distribute the beam amongst several rooms,
2007; Barlow et al. 2010). This combines the fixed magnetic usually between two and four, with the option to dedicate
field, high current capability and rapid repetition rate of the some of them to specific clinical applications. Fast switching of
cyclotron with the variable energy extraction of the synchro- the beam between these different rooms is required to increase
tron. A major research programme is underway to investigate the patient throughput, taking advantage of the fact that the
whether this technology is feasible.* time for patient set-up is usually longer than the beam-on time.
The beam is transported after extraction into the treatment
rooms along conduits kept under vacuum. Electric and mag-
15.5 General Technical Design Considerations netic systems are used to modify the trajectory (e.g. dipoles)
and to focus the beam in a similar way to optical lenses (e.g.
15.5.1 Extraction quadrupoles). In the research facilities, most of the rooms have
The extraction of the beam can be achieved by various meth- fixed horizontal beams. A particular case of optimisation of
ods. Electrostatic deflectors are often used in conventional the beam transport system for therapy is the design of com-
cyclotrons. In a synchrocyclotron, electromagnetic channels pact and isocentric gantries that direct the beam around the
cancel the main magnetic field and let the particles come patient (Figure 15.7). While this is a common approach with
out in a straight line. An electromagnet kicker can be used, electron beams, the required dipoles to bend the protons are
particularly for synchrotrons. When negative ions are being much bigger and are expensive (even more so for heavy ions),
accelerated, a different approach can be used. It is possible to which acts as an economic constraint to their development.
strip off the electrons from the ions with a thin target once The basic expression to calculate the rotational radius of accel-
they have reached the desired energy. The resulting positive erated charged particles is through the magnetic rigidity Br:
ions will emerge from the accelerating chamber with a curved
trajectory opposite to that of the orbit of the negative ions. B r = p /q = m v /q (15.3)

where B is the magnetic field intensity, r is the bending

* http://gow.epsrc.ac.uk/ NGBOViewGrant.aspx?GrantRef=EP/ radius, p the momentum, q the charge, m the mass and v the
E032869/1. velocity of the particle.
Chapter 15: Accelerators for Protons and Other Heavy Charged Particles 303

e f g h i j

FIGURE 15.7 Typical isocentric rotational gantry for proton beam delivery. (a) and (c) Quadrupole magnets; (b) 45° dipole magnet; (d) 135°
dipole magnet (e) and (i) nozzle (see Figure 25.1); (f) isocentre; (g) gantry support; (h) rotating gantry structure; (j) patient positioner. (Modified
from the original drawing, courtesy of IBA, Belgium.)

Compared with electron clinical beams, the magnetic equipment, including the accelerator, transport and delivery
rigidity of protons is three times higher, and for carbon ions, systems (such as the isocentric gantries). These costs can rep-
it is three times higher than for protons. An electron gantry resent from 1/3 to 2/3 of the total cost of installing and
has a radius of the order of 1.5 m and a weight slightly lower operating a particle facility. Other costs include the building
than 10 tons; for protons, the radius is between 4.5 m and (bunkers, services and clinical areas) and other equipment
6 m with weights of the order of 100 tons, and for carbon, (cooling, heating, ventilation and air conditioning), safety
the first gantry built in Heidelberg has a diameter of 13 m, is and radiation protection, imaging and dosimetry devices,
25 m long and weighs 670 tons. With heavy-ion accelerators, anaesthesia, etc.
fixed beam-lines (horizontal, vertical or oblique) are still Because of the high investment and functional costs,
sometimes chosen because of the high cost of gantries. Much most national agencies for the evaluation of health technol-
smaller ion gantries have recently been developed in Japan ogy have produced reports on the justification for installing
(at the National Institute of Radiological Sciences [NIRS], particle therapy centres (ANDEM 1995 and more recently
superconducting magnets are attached to a cylindrical rotary France, the Netherlands, Belgium and Canada, among oth-
body, 11 m in diameter and 13 m long).* ers). While most of these reports identify proven anatomical
The basic concepts relating to beam delivery and gantries sites that are optimally treated by particles (eyes, base of the
for proton therapy can be found in the following references: skull and paediatrics), most insist on the need to collect data
Coutrakon et al. (1991), Pedroni (1994), Renner et al. (1994) from clinical studies for the other anatomical sites. While
and Pedroni et al. (1995). the US and Japan have established centres at institutional
and regional levels, other countries install a national cen-
tre with a network of referring centres (e.g. Sweden, with a
national centre in Uppsala). The costs are known for turnkey
15.7 Economic and Functional Aspects proton centres; they are more difficult to obtain for carbon-
15.7.1 Capital Costs ion facilities, but they are at least two or three times higher.
The investment cost required to convert an existing
A high proportion of the investment and functional costs nuclear physics research accelerator to medical use, as per-
of a proton or a heavy-ion clinical facility is related to the formed in the period 1960–1990, was very dependent on the
detailed local situation. As an example, in 1991, at the Orsay
* www.nirs.qst.go.jp/ENG/news/press/2016/160108.html.
304 Part C: Equipment

Protontherapy Centre (France), the cost was 1.7 M€ for a • The permanent staffing cost is also an important
fixed line. In 2010, a new centre was built at the same loca- component. It was more significant for research
tion, including a clinical building, a cyclotron and a gantry machines converted into clinical facilities, where
room, and the two existing fixed-beam rooms were adapted, maintenance and R&D staff were shared with a
for a total cost of 40 M€, of which 25 M€ was related to research centre. Single-room facilities are evolv-
the equipment. The present tendency (Beck 2015) is to move ing towards ‘no site operating staff ’ and a service
towards single-room facilities with a cost in the order of contract similar to those of conventional electron
20 M€ or less. The cost of photon linacs is much lower for the linacs. A review in Europe (Weber et al. 2017)
simple versions (e.g. a single low-energy accelerator around found that there is a ratio of 10:1 between the
1.3 M€), but the most complex ones (e.g. integrated with number of patients per year and the medical and
a magnetic resonance imaging [MRI] system) can cost as paramedical staff (doctors, physicists, therapists
much as 10 M€. The patents to integrate MRI with protons and nurses), excluding the technical and admin-
have already been filed, which will further increase the cost istrative staff.
of state-of-the-art facilities. Figure 15.8 shows the evolution
of the order of magnitude of the investment cost for a single- Estimates of the cost of operation (excluding staff salaries)
room proton therapy facility compared with a conventional for these facilities vary between US$250 and US$400 per
linac facility. It can be seen that the cost of proton therapy hour of operation. Between US$1000 and US$2000 can be
has been decreasing since the 1990s and is now approaching charged per hour when the beam is requested for experimen-
the cost of sophisticated high-energy x-ray equipment. tal use if all the costs are included (operating costs, salaries,
capital amortisation, experimental tools, etc.).

15.7.2 Running Costs
There are two major components of the running costs: 15.7.3 Treatment Capacity
The capacity of existing proton treatment centres ranges
• The service contract represents about 7–8% of the between 20 and 350 patients per year per room, as some
equipment cost and is essential to warranty the share their activity with physics research programmes.
uptime required for clinical operation. Present fig- New dedicated multi-room facilities are planning to treat
ures of uptime for modern turnkey facilities after a 1000–1500 patients per year, with one accelerator serv­
few years of operation are claimed by the companies icing several treatment rooms. The number of fractions
and users to be of the order of 97–98%, which is per patient is strongly dependent on the clinical proto-
similar to photon linacs. However, in general, more col in use: one fraction for stereotactic irradiations, four
effort is required for servicing to maintain these to five fractions for ophthalmic treatments, 10 fractions
values. for a boost after a photon treatment and more than 25
fractions for a treatment given entirely with protons. The
European survey (Weber et al. 2017) gave a mean value
of 221 patients/year (range 40–557) for 15 operating
centres.
It is important to take into account the ramp-up period
when starting a facility. Values of patients per year for the
first years of operation for recent facilities are presented in
Figure 15.9. After about 3 years, the patient throughput is
more or less stabilised or slowly increasing. For standard pho-
ton linacs, the ramp-up period is seldom larger than a few
months.

15.7.4 Cost of Treatment
As a result, the cost of individual treatments may be estimated.
As an example, it was found to be 1025 € (US$1320) per
fraction for protons versus 425 € (US$550) for x-rays (Goitein
and Jermann 2003). In the US and in most European coun-
FIGURE 15.8 Qualitative evolution of the investment costs of proton tries, it is covered by the national medicare system.
therapy facilities, from the converted research-based facilities, passing
through the regional multi-room hospital-based centres, to the insti-
tutional single-room facilities. It is compared with the evolution of the
linear-accelerator-based clinical facilities using photon and electron 15.8 Perspectives
beams with ancillary equipment, including the ongoing integration of
MRI-guided therapy. Data from the author (AM) based on personal In the early days of particle therapy, centres had to develop
communications from real projects. their own solutions for using the existing systems originally
Chapter 15: Accelerators for Protons and Other Heavy Charged Particles 305

TABLE 15.1
Types of Accelerator in Use and in Construction Worldwide in
2017 (Data Extracted from www.ptcog.ch)
Cyclotron Synchrocyclotron Synchrotron Total

In operation Proton 38 5 21 64
C-ion 11 11
In construction Proton 18 12 9 39
C-ion 2 2
Total 56 17 43 116
Percentage 48% 15% 37% 100%

accelerators are now related directly to the clinical require-

ments, including:

• Fast 3D scanning to minimise the impact of organ

movement;
• Combination of treatment beam and imaging
devices such as MRI (affecting the trajectories of
particles);
• Particle radiography and tomography requiring
higher energies;
• Research on the biological effects of very high dose
rates (e.g. close to 100 Gy s –1) requiring higher
beam intensities;
• Knowledge-based and fast planning based on rota-
tional techniques;
FIGURE 15.9 (a) Patients per year and per treatment room in 21 facili-
ties with more than 10 years of experience in proton therapy (some of • The combination of different types of beams.
them are treating very few patients as shown on the ‘Min’ curve). The
mean value of the more efficient facilities is 280 patients per year and per The evolution of accelerators for particle therapy must take
room. (b) Ramp-up of total number of patients per facility (independent these aspirations into account. However, the highest priority
of the number of rooms) for 25 new facilities. For these facilities, also, the must be to reduce their cost. This can be achieved through
most efficient had a mean value of 280 patients per year and per room. In mass production, improved cryogenics and ironless systems
general, 2 to 4 years were necessary to reach a plateau of reasonably stable
as well as the optimisation of throughput by fast switch-
activity. (Analysis done with data extracted from www.ptcog.ch)
ing between rooms, fast energy change and high dose rates.
Another goal is to reduce the uncertainties related to the
designed for physics research. Nowadays, many companies beam delivery systems and the full treatment process (see
over the world offer turnkey facilities for proton and carbon- Chapter 39). The role of research activities performed at oper-
ion therapy. Discussions about using other ions and innova- ating centres in synergy with industry is of the highest impor-
tive solutions are on-going, but they will take many years tance for such developments. This is becoming possible, as
before being operational, or they could ultimately fail. the number of facilities in construction is increasing expo-
In 2017, of a total of 116 accelerators in operation and nentially, at least for proton facilities, including single- and
under construction, around half were isochronous cyclo- multiple-room centres. The gap between the lowest-cost pro-
trons for proton therapy, 15% corresponded to the new ton facilities and the highest cost of modern linacs for pho-
tendency to use compact synchrocyclotrons for proton ton treatments has reduced. This, together with the results of
therapy, and 37% were synchrotrons, including all the on-going clinical protocols for different locations, will help to
carbon facilities (Table 15.1). Linear accelerators should assign the optimal role of particle-beam therapy among the
be installed in the near future. The specifications for various modalities available in radiation oncological therapy.