Electrostatic accelerators

F. Hinterberger
Helmholtz-Institut fr Strahlen- und Kernphysik, University of Bonn, Germany
Abstract
The principle of electrostatic accelerators is presented. We consider Cockcroft
Walton, Van de Graaff and Tandem Van de Graaff accelerators. We resume
high voltage generators such as cascade generators, Van de Graaff band generators, Pelletron generators, Laddertron generators and Dynamitron generators.
The specific features of accelerating tubes, ion optics and methods of voltage
stabilization are described. We discuss the characteristic beam properties and
the variety of possible beams. We sketch possible applications and the progress
in the development of electrostatic accelerators.
1

CockcroftWalton accelerator

The CockcroftWalton accelerator [1] is the prototype of an electrostatic accelerator. The scheme of the
accelerator is shown in Fig. 1. The first accelerator of this type was built in 1932 by J.D. Cockcroft and
E.T. Walton. They achieved a high voltage of about 700 kV and they studied the first nuclear reactions
with an accelerator. In their experiment a proton beam of about 400 keV kinetic energy was used to
investigate the nuclear reactions 7 Li + p 4 He +4 He and 7 Li + p 7 Be + n. Cockcroft and Walton
shared the Nobel Prize in physics for 1951.
1.1

High voltage generator

The high voltage generator of the CockcroftWalton accelerator is the well known cascade generator or
voltage multiplier circuit invented by H. Greinacher [2] in 1921. It consists of a HV transformer, HV
capacitors and HV diodes. In Fig. 1 a cascade generator with four stages is shown. The capacitors are
stacked in two vertical columns capped by a large rounded terminal electrode. The capacitors in the
pushing column P are charged during the negative half-period of the sinusoidal AC voltage, those of the
smoothing column S during the positive half-period. In steady state the voltages at the break points of
the pushing column P amount to U0 + U0 sin t, U0 + 3U0 sin t, U0 + 5U0 sin t and U0 + 7U0 sin t,
those of the smoothing column S to 2U 0 , 4U0 , 6U0 and 8U0 . Here, U0 is the amplitude of the AC voltage
from the transformer. With n stages a high voltage of 2nU 0 can be achieved. Extracting a DC current
causes a mean voltage drop U and a ripple U which are proportional to the DC current I,
U
U
U

= 2nU0 U U ,


2 3 3 2
1
I
n + n + n ,
=
fC 3
4
12
I n(n + 1)
=
.
fC
2

(1)

Here, f is the frequency of the AC voltage and C the capacitance of the capacitors. In order to keep U
and U small one should choose the frequency f and the capacitance C as large as possible and limit the
number n of cascade units. Typical values are f = 0.510 kHz, C = 110 nF and n = 35. The ripple
U can be reduced by an additional RC filter.
The value of the electric field at the surface of the electrodes must kept small in order to avoid excessive corona discharges. Therefore, the intermediate electrodes and the terminal electrodes are formed
such that sharp edges and spikes at the outer surface are avoided. Aluminum is commonly used because
it can be spun to the desired spherical shape. Polished stainless steel has a higher breakdown limit but is

95

F. H INTERBERGER

I
8U0
7 U0 + U0 sin t

6U0

5 U0 + U0 sin t

4U0

3U0 + U0 sin t

2U0

U0 + U0 sin t
~ U = U0sin t
P

Fig. 1: Scheme of a CockcroftWalton accelerator [3]. P: pushing column, S: smoothing column, I: ion source,
T: target

~ at the surface of rounded electrodes with a local

more difficult to form. The value of the electric field | E|
curvature radius r is simply given by
~ =U .
|E|
(2)
r
For open air accelerators the electric field should not exceed 3 MV/m. The maximum possible voltages
of such machines are in the order of 1.5 MV.
1.2

Accelerating tube

The accelerating tubes are installed in an evacuated ceramic isolator, see Fig. 1. The full high voltage is
distributed among several tubes yielding a stepwise acceleration of the beam. Thus, the local value of
~ is reduced and the voltage stability is increased. For the same reason all inner
the accelerating field |E|
electrodes are rounded and polished. As a rule of thumb the electric field at the surface of polished stainless steel electrodes in high vacuum must not exceed 100 kV/cm. Higher fields cause violent discharges
due to the emission and acceleration of electrons at the electrode surface.
In order to avoid distortions of the ion trajectories by charge islands on the isolator walls the
accelerating tubes should be overlapping such that an ion cannot see the isolator walls. This is of great
importance in view of the low electric rigidity of low energy ions. Similarly the metal electrodes must
be free of oil, hydrocarbon deposits and all kind of insulating deposits.
The ion beam starts at the plasma boundary of the ion source. It is extracted and formed with
the aid of an extraction electrode. The acceleration by the rotational symmetric electric fields of the
accelerating tube acts like a focusing lens. A proper adjustment of this focusing yields a beam envelope
with a beam waist at the target. The electrostatic focusing is discussed in the lecture on electrostatic
lenses. The ion source, the electrostatic focusing system at the exit of the ion source, the power supplies
and a motor-driven AC voltage generator are housed inside of the HV terminal. The optimum adjustment
of the ion source and the extraction system is done by remote control. The high voltage is measured with
a voltage divider using Ohms law.
A serious problem of positive-ion electrostatic accelerators is caused by electrons which are released by secondary emission processes from the electrodes and walls and by collisions of the ions with
the atoms of the residual gas in the accelerating tube. These electrons are accelerated in the opposite
2
96

E LECTROSTATIC ACCELERATORS

+
+
+
+
+

+ U+ +

+
+
++
+
+
+
+
+
+
+

+
+
+

Uc

Fig. 2: Scheme of the Van de Graaff generator [3]

direction and generate hard X-rays when hitting the electrodes of the HV terminal. In addition the unwanted electron current makes a substantial contribution to the total current. This electron load can be a
limitation on terminal voltage. Therefore, a high vacuum is needed in order to minimize electrons generated from the collisions of ions with the atoms of the residual gas in the accelerating tube. In addition,
electron suppression systems are installed in modern electrostatic accelerators. For instance permanent
magnets can be installed in the accelerating tube in order to deflect the electrons near the point of origin
without causing significant distortions of the ion trajectories.
2
2.1

Van de Graaff accelerator

Van de Graaff generator

The first belt-charged electrostatic generator was developed by R. J. Van de Graaff in 1931. The Van
de Graaff generator [4] is made up of a motor-driven belt (made of rubber, vulcanized fabric or another
flexible insulating material) stressed between two rollers (pulleys), an insulating column and a spherical
or rounded high-voltage terminal electrode which is installed on top of the insulating column as shown
in Fig. 2. The belt is electrically charged by a brush or comb of metallic wires which is connected to
a DC voltage source. The amount of electric charge sprayed onto the belt is controlled by the voltage
Uc . The charge, which can be negative or positive depending on the polarity of the source, is carried by
the belt to the terminal electrode. Here, the charge is transferred from the belt to the terminal electrode
by a second brush or comb of metallic wires. Electric charges accumulate on the external surface of the
terminal electrode. The resulting terminal voltage is a function of the diameter of the terminal electrode.
The terminal behaves like a spherical capacitor which has capacitance
C = 40 r = (1.11 1010 F/m) r ,

(3)

where r is the radius of the terminal electrode. An air insulated electrode with r = 1 m yields C =
111 pF. If a spherical terminal of radius r 1 is enclosed within a grounded concentric shell of radius r 2 ,
the capacitance C is given by
C = 40

r1 r2
r1 r2
= (1.11 1010 F/m)
.
r2 r 1
r2 r 1
3
97

(4)

F. H INTERBERGER

I
I belt
I tot
Ires
Ibeam
Icor
U0 U
Fig. 3: Typical current-versus-voltage functions for the corona current I cor , beam current Ibeam , resistive current
Ires and total current Itot in a Van de Graaff accelerator

The terminal voltage U is proportional to the charge Q accumulated on the terminal electrode,
1
Q.
C

U=

(5)

The achievable equilibrium voltage U 0 depends on the current Ibelt transported by the belt, the beam
current Ibeam of the accelerating tube, the current I res through the resistor chain including the current
due to the residual resistance of the insulating column and the current I cor due to the corona discharge
and secondary electrons in the tube. The corona current I cor is highly nonlinear rising rapidly above a
threshold value. In Fig. 3 the typical current-versus-voltage functions for those current components is
shown.
Assuming a constant charging current I belt the equilibrium voltage U0 is reached asymptotically.
The differential equation reads
1 dQ
1
dU
=
= (Ibelt Ibeam Icor Ires ) .
dt
C dt
C

(6)

This equation can be solved analytically at low terminal voltages where the highly nonlinear corona
current Icor is negligibly small. Introducing the effective resistance R of the resistor chain and the
insulating column, the differential equation may be written


U
dU
1
Ibelt Ibeam
.
(7)
=
dt
C
R
We define the time constant = RC and introduce the equilibrium voltage U 0 ,
U0 = R(Ibelt Ibeam ) .

(8)

Thus, the solution of the differential equation may be written

U (t) = U0 + [U (0) U0 ]et/ .

(9)

This equation holds as long as the currents I belt and Ibeam and the effective resistance R are constant and
the corona currents are negligible. The terminal will reach an equilibrium potential U 0 where the charge
4
98

E LECTROSTATIC ACCELERATORS

+
+
+
I
+
+ +
+ Uc
Uc +
+
+
+
+
+
+
+
+
++ +
+
+
+
+
+
+
+
+
Uc
U
I+ c
+

Fig. 4: Scheme of the Pelletron and Laddertron HV generator [3]

delivered by the belt is equal to the charge leaving the the terminal electrode. In modern machines fast
closed-loop controls are used in order to achieve a fast adjustment and a high stability of the terminal
voltage U . Then, the time dependence of U is determined by the characteristics of the closed loop
control. The terminal voltage can be controlled by regulating the charging current I belt and by varying
the amount of corona load current with adjustable needles.
The maximum achievable voltage U is strongly limited by the corona discharge from the terminal.
Occasionally, a corona streamer will develop into a spark to ground, discharging the terminal completely.
The corona discharge is caused by the breakdown of insulation due to excessive electric fields at the surface of the terminal. The breakdown potential of an insulated electrode in gas depends on the radius
of curvature, smoothness of the terminal surface, electrode material, surface contaminations, shape and
material of the insulating supports, potential distribution along the insulator and last but not least composition and pressure of the surrounding gas. The breakdown electric field in air at atmospheric pressure
is about 3 MV/m although limitations on maintaining ideal surface conditions lead to lower electric field
limits.
In order to achieve higher voltages than about 1.5 MV the Van de Graaff accelerator is enclosed
in a high pressure vessel. The use of compressed gas to increase the breakdown potential is based on
Paschens law which predicts a linear increase of sparking potential with pressure. However, the sparking
potential curves always deviate from linearity above some value of pressure. In addition, breakdown
potentials depend strongly on gas composition. Today, dry and purified sulfur hexafluoride (SF 6 ) gas or
mixtures of nitrogen and carbon dioxide (80% N 2 and 20% CO2 ) are used with pressures up to 20 bar.
Another important point is a uniform and smooth voltage distribution along insulator, belt and accelerating tube. This can be achieved by closely spaced equipotential rings connected to resistor voltage
divider. The rings are also connected to corresponding accelerating-tube electrodes and to field control
bars near the belt.
2.2

Pelletron and Laddertron generator

The Pelletron [5] charging chains (see Fig. 4) were developed by Herb and collaborators from National
Electrostatics Corporation (NEC) in the mid 1960s as an improvement over the older Van de Graff
charging belts. These belts suffered from a number of operational difficulties including terminal voltage
5
99

F. H INTERBERGER

instability and susceptibility to spark damage. Also, they generated belt dust which necessitated frequent
cleaning inside the accelerator tank. The chain rapidly proved to be more durable than the old belts, while
producing a greater terminal stability than had been possible before. It eliminated the belt dust problem
as well. The chain does not limit ultimate terminal potential, and it is in use in electrostatic accelerators
up to and above 25 MV. A different type of charging chain called Laddertron, was developed by High
Voltage Engineering Company (HVEC). The name derives from the fact that this chain had originally H
shaped metallic carriers, which looked like a ladder.
Pelletron chains are made of metal pellets connected by insulating nylon links. Laddertron charging chains are made of metal plates. The metal pellets or plates are charged using the effect of influence
in an electrical field (see Fig. 4). For positive charging, the electric field between the negatively biased inductor electrode and the pulley pushes electrons off the pellets while they are in contact with the
grounded drive pulley. Since the pellets are still inside the electric field as they leave the pulley, they
retain a net positive charge. The chain then transports this charge to the high-voltage terminal, where the
reverse process occurs. When it reaches the terminal, the chain passes a negatively-biased suppressor
electrode which prevents sparking as the pellets make contact with the terminal pulley. The positive
charge flows smoothly via the terminal pulley to the terminal. Most Pelletrons and Laddertrons employ
down-charging as well as up-charging. Down-charging works identically to up-charging, except the
inductor/suppressor polarities are reversed, and it effectively doubles the charging capacity of the chain.
The typical charging currents are in the order of 100300 A. High voltages up to 30 MV have been
reached. The terminal voltages can be regulated and controlled by varying the charging voltage U c between inductor/suppressor electrodes and pulleys. Charging voltages up to 50 kV are used. By reversing
the polarities of the charging voltages the Pelletron and Laddertron generators can easily be used to accelerate electrons and negative ions. The charging chain for high voltage generation exhibit an excellent
voltage stability, a high reliabilty and a long lifetime (over 50 000 hours).
2.3

Dynamitron generator

CD
C

RF
Fig. 5: Scheme of the Dynamitron HV generator. D: Driver electrodes, C: Coupling electrodes, T: Terminal, V:
High-pressure vessel

6
100

E LECTROSTATIC ACCELERATORS

The Dynamitron generator [6] has been developed by Radiation Dynamics Inc. (RDI), Long
Island, since 1958. The Dynamitron generator (see Fig. 5) is based on an rf oscillator operating at
frequencies between 30 kHz and 300 kHz. Driver electrodes D in the form of half-cylinders transmit
the rf power via capacitive coupling to coupling rings C. A CockcroftWalton-like cascade generator
generates the DC voltage. In modern machines semiconductor diodes are used. RF chokes provide
the connection of the rectifier cascade between ground and high voltage terminal. Like in the classical
CockcroftWalton generator the achievable high voltage is proportional to the number n of rectifier
stages,
U = 2nfc U0 .
(10)
Here, U0 is the amplitude of the alternating voltage at the driver electrode D and f c the coupling factor.
A rectifier stage consists of two rectifier diodes. The voltage amplitudes U 0 reach values up to about
100 kV. The advantage of this power supply is a high reliability due to the absence of moving parts
inside the pressure vessel. Another advantage compared to the classical CockcroftWalton generator
are the very low capacitances of the system. Therefore possible spark-overs are rather harmless. The
regulation loop of the Dynamitron generator can be modelled like in a standard electronic power supply.
Therefore the stabilization of the terminal voltage is substantially faster than the classical Van de Graaff
belt generator or the Pelletron generator.
A variant of the Dynamitron power supply was developed at the Budker Institute of Nuclear
Physics in Novosibirsk. There, the capacitive coupling is replaced by inductive coupling.
2.4

Acceleration tube

The acceleration tube is one of the most critical components of a Van de Graaff accelerator. The tubes
are constructed of insulating material, commonly ceramic or glass cylinders with vacuum tight seals
to metal-plate electrodes. The vacuum seals are extremely critical in view of the high external gas
pressures of up to 20 bar. The electrodes are connected to equipotential rings in the voltage generator
column to maintain a uniform potential distribution along the tube. Often, the electrodes have large
diameter holes (25 cm) in order to achieve a high pumping speed with vacuum pumps at the grounded
end. The electrodes produce an accelerating and focusing electric field for the charged particles. They
are arranged such that the insulating walls of the tube are protected from the particle beam in order to
suppress flashover and to shield the beam from isolated surface charges on the insulator walls.

Fig. 6: Acceleration tube segment from NEC. The overall length is 8" and the I.D. of the ceramic is 4". The
maximum voltage in SF6 gas is 330 kV. (Courtesy of G. A. Norton, NEC.)

7
101

F. H INTERBERGER

In modern designs the spacing between electrodes is decreased and the number of gaps is increased
resulting in the so-called uniform field tube. This allows to achieve higher voltage gradients along the
tube and higher terminal voltages. For illustration, a segment of the accelerating tubes from National
Electrostatics Corporation (NEC), Middleton, Wisconsin, USA is shown in Fig. 6. Historically, one
of the most serious problems of positive-ion electrostatic accelerators was the limitation on terminal
potential due to electron loading. The loading was caused by electrons released by secondary emission
processes from the electrodes and walls of the accelerating tube and by collisions of the beam particles
with atoms of the residual gas in the vacuum. The electrons traverse the tube in opposite direction and
increase the current so as to limit the terminal voltage at constant charging rate. In addition the electrons
are focused by the electric fields onto the ion source in the terminal, producing hard X-rays with an
energy distribution extending up to the terminal voltage. Such X-rays require expensive shielding of
the room in which the accelerator is housed. As a consequence modern accelerator tubes are equipped
with electron suppression systems. The suppression is achieved by (i) minimizing secondary-electron
emission by applying selected materials, (ii) preventing electrons from entering the accelerator tube, (iii)
minimizing ion-atom collisions by improving the high vacuum and (iv) by deflecting the electrons in
statu nascendi with weak magnetic fields installed within the tube. Another method was to use inclined
field tubes where the electric fields between the accelerating sections are inclined yielding a deflection
of back-streaming electrons.
3

Tandem accelerator

The concept of the Tandem accelerator was invented in order to achieve higher beam energies than with
single ended Van de Graaff machines. It has been specifically proposed by Bennett [7], Kallmann [8]
and Alvarez [9]. However, the first practical application was made by the High Voltage Engineering
Company in a machine constructed for the Chalk River Laboratory and reported by Van de Graaff [10].
A Tandem accelerator utilises the terminal high voltage twice in sequence in order to obtain output
energies of two or more times that available in a single acceleration, see Fig. 7. Negative ions produced
by an appropriate ion source I are accelerated from ground to the positively charged terminal. Inside
the terminal is a stripper, which uses a gas canal (usually nitrogen or argon) or a very thin carbon foil
(areal density about 5 g/cm2 ) to remove electrons from the incoming negative ions. The now positivelycharged ions experience a second boost of acceleration (hence the name Tandem accelerator) as they
exit the terminal and travel down the acceleration tube to ground at the high-energy end of the machine.
The resulting kinetic energies T of the beam depend on the charge q of the positive ions,
T = eU + qU = (e + q)U.

(11)

Here, e is the absolute value of the singly charged negative ions. The positive charge q of heavy ions can
be multiples of e. Thus the maximum possible kinetic energy depends on the ions, e.g.,
p, d :
3

He

2+

T = 2 eU ,

2+

T = 3 eU ,

32 16+

T = 17 eU .

, He
S

(12)

In principle, the final kinetic energy is a little bit higher since the negative ions are pre-accelerated to
about 50 keV by connecting the negative ion source to a negative potential with respect to ground.
Tandem accelerators are enclosed like single ended Van de Graaff machines in high-pressure vessels. Tandem accelerators, especially the ones used in nuclear physics research, can be very large. The
most powerful ones can reach high voltages between 20 and 30 MV. Vertical generators are usually installed in specially built towers. The primary benefits of the Tandem accelerator are much higher beam
energies for a given terminal voltage. The charge-exchange process results in a significant reduction of
beam intensity. However, for a wide range of applications a few microamperes of accelerated ions are
8
102

E LECTROSTATIC ACCELERATORS

HV

+ + + + + + + + + + + + + +

I
positive
ion
beam

negative
ion
beam

++
+
+
+
+
+
+
+
+ AS
+
+
+
+
D +++
+
+
+
+
+
+
+
+

Fig. 7: Scheme of Tandem accelerator with ion source and external beam preparation system [3]. I: negative ion
source, V: high-pressure vessel, B: belt generator, HV: high-voltage terminal, S: stripper foil or gas, A; analysing
magnet, AS: analysing slit, D: deflecting magnet, T: target.

adequate. In contrast to single ended machines, Tandem accelerators have both ends at ground with the
high-voltage terminal in the middle. With the ion source(s) external to the pressure vessel, maintenance
requiring entry into the tank and letting the acceleration tubes up to atmosphere is minimized.
4

Ion optics

An important point is the optimum design of the ion optics of an electrostatic accelerator. This can be
done with computer programs. If high currents are accelerated the space charge forces must be taken
into account. Various requirements have to be considered. Very often the ion beam is mass analysed by
a 90 double focusing analysing magnet. With such a system all elements from hydrogen to uranium can
be separated. To this end a sharp beam waist at the entrance and exit slits of the analysing magnet system
is required.
A technical requirement is a double waist at the terminal stripper of a Tandem Van de Graaff or
near the high energy end of a single ended Van de Graff. This can be achieved by preparing a double
waist at a well defined distance to the entrance of the accelerator tube. The electric field transition at the
accelerator entrance acts like a focusing lens. In nonrelativistic approximation the focal length of this
lens is given by
~
f = 4V /|E|
(13)
where V is the total acceleration potential through which the ions have been accelerated before reaching
~ is the electric field of the acceleration tube. The electric field | E|
~ is proportional to the
the lens and E
terminal voltage. Therefore the focal length f and the optimum position of the double waist in front of
the acceleration tube depends on the final energy of the ions. A possibility to circumvent this problem
~ is kept constant. Such a patented
is to vary the potential V by pre-acceleration such that the ratio V /| E|
injection system for focusing to a waist at the terminal stripper has been developed by High Voltage
Engineering Europe (HVEE).
Another important issue is a charge selector in the terminal of a Tandem. Heavy ions passing the
terminal stripper exhibit a finite charge distribution. Therefore a charge selection is needed in order to
9
103

F. H INTERBERGER

achieve a well defined beam energy.

At the high energy end of single ended Van de Graaff accelerators and Tandem accelerators a 90
analysing magnet is commonly used in order to prepare ion beams with a well defined energy and to
achieve a high energy resolution. To this end a sharp beam waist at the analysing slit of the 90 analyser
is needed. Sometimes the analysing system consists of a double focusing magnet with point-to-point
imaging between a narrow entrance slit and the analysing slit.

Fig. 8: Beam envelopes in a 15 MV Tandem accelerator from NEC (courtesy of G. A. Norton, NEC)

For illustration the ion optics of a 15 MeV Tandem from National Electrostatics Corporation
(NEC) is shown in Fig. 8. The beam from the ion source is first focused with an einzel lens and a
pre-acceleration tube to provide a double waist at the entrance slit of the inflection magnet. Another
double waist follows at the analysing slit of the double focusing inflection magnet. A system of electrostatic quadrupoles produces a sharp double waist in front of the acceleration tube which is transformed
by the focusing effect at the entrance of the acceleration tube into a weak double waist near the terminal
stripper. The defocusing effect at the exit of the low energy acceleration tube and the focusing and defocusing effects at the entrance and exit of the high energy acceleration tube are so weak that they cannot
be seen. In the terminal an offset quadrupole is used as charge selector followed by another electrostatic
10
104

E LECTROSTATIC ACCELERATORS

quadrupole. The beam envelopes are diverging to the high energy end of the Tandem. The double focusing analysing magnet produces a sharp double waist at the analysing slit. Then the beam is guided with
a switching magnet to the target. The final focusing is done with a magnetic quadrupole triplet.
5

Voltage measurement and control

The problem of voltage measurement is fundamental for many applications of electrostatic accelerators.
The most simple method consists in measuring the current in a string of calibrated resistors running from
terminal to ground. These resistors must be carefully shielded in order to be free from corona losses.
The resistor string can also be utilized to provide a uniform potential distribution between successive
equipotential rings. The total resistance is high. A typical value is 10 G per 1 MV yielding a current
of 100 A. A micro-ammeter measures the current at the grounded end of the resistor column. The
individual resistors must be calibrated to high precision. If other more precise voltage calibrations are
used, the current through the resistor column can still be used as a check of the relative voltage stability.
Another classical method is to use a generating voltmeter for observing the terminal voltage. The
operation depends on the electric charge induced by influence on an metal plate or vane in the electric
field near the grounded pressure vessel. In the usual form an insulated vane rotates at constant speed
behind a grounded shield covering the vane during half its travel. The alternating voltage induced on the
vane is amplified and rectified and its magnitude is a measure of the terminal voltage. The voltage scale
is usually calibrated against other standards. The most serious limitation is the distortion of electric fields
due to corona discharge from terminal. Mostly, the generating voltmeter is used as a relative instrument
in tuning the accelerator. With the advent of multiple shields around the terminal its usefulness is reduced
since it measures only the voltage of the outer shield.
The most precise method uses electric or magnetic deflection of the ion beams in order to measure
the beam energy and to calibrate the terminal voltage. An illustration is the 90 analysing magnet used
to measure and stabilize the beam energy of Tandem accelerators as shown in Fig. 7. Such systems yield
energy resolutions up to 10 000 corresponding to a relative energy spread (FWHM) of about 1 10 4 .
The mean energy can be stabilized to a precision of about 2 10 5 . In the analysing magnet system the
beam passes narrow entrance and exit slits. The magnetic field is highly stabilized to a level of 1 10 5
using NMR probes. A variation of the terminal voltage will cause a deflection of the beam to one side
or the other of the exit slit. The slit edges are insulated and the currents to the edges are measured
and compared. The detected left-right asymmetry (I l Ir )/(Il + Ir ) is used as error signal in order to
stabilize the terminal voltage with the aid of a fast control loop.
An important means of cross-calibrating the energy scales is provided by nuclear resonances.
Certain reactions have extremely sharp resonances. By measuring excitation functions in the region of
those resonances an absolute calibration of the analysing magnets is possible.
An interesting variant of stabilizing the voltage with analyser magnets is offered by High Voltage
Engineering Europe (HVEE), Amersfoort, The Netherlands. In order to achieve a very high voltage
stability two separate accelerator tubes are installed inside of the Dynamitron-like acceleration column,
see Fig. 9. One accelerator tube is used to accelerate and analyse a reference beam. For negative terminal
voltages a H beam is used, for positive voltages a proton beam. The beam energy is measured using
a standard 90 analysing magnet. A stabilization of the terminal voltage up to 1 10 5 can be achieved
with such a system.
6
6.1

Beam properties and applications

Beam properties

A great advantage of electrostatic accelerators is the fact that the beam energy can easily be varied over
a wide range of energies.

11
105

F. H INTERBERGER

Fig. 9: Cross-section of two acceleration tubes mounted in a coaxial high-voltage power supply (courtesy of H.
van Oosterhout, HVEE)

The maximum terminal voltages of the CockcroftWalton accelerators are typically between 200
and 1000 kV, thus the kinetic energies of the ions are rather low. Depending on the ions and the high
voltage generator typical beam currents vary between 1 A and several 10 mA. Due to the progress in the
field of ion sources not only light ions like hydrogen and helium but all kinds of heavy ions up to uranium
can be accelerated. Using pulsed ion sources also pulsed beams with a high time-of-flight resolution can
be prepared.
The Van de Graaff and the Tandem Van de Graaff accelerator offer a great variety of ion beams
from hydrogen up to uranium. Depending on the ion source and the voltage generator the beam currents
vary between 10 nA and several 10 mA. Typical kinetic energies of hydrogen beams are in the range
1 MeV up to 40 MeV. With Tandem accelerators heavy ions can be accelerated up to several hundred
MeV. A great advantage of Van de Graaff and Tandem Van de Graaff accelerators is the excellent beam
quality. Energy resolutions up to 10 000 corresponding to a relative energy spread of 1 10 4 can be
achieved. Also excellent transverse beam emittances can be achieved corresponding to the theoretical
limit given by the extraction system near the ion source. Thus, experiments with a high angle and energy
resolution can be performed. The easy energy variation of the beams allows systematic energy scans in
fine energy steps.
6.2

Application of electrostatic accelerators

The CockcroftWalton accelerator is still used as pre-accelerator for large accelerator facilities. In
medicine and industry it is used as a cheap neutron generator. Here, deuterons with a kinetic energy
between 400 and 800 keV are bombarded onto a deuterium or tritium target. The nuclear reactions
d + d n +3 He and d + t n +4 He yield neutrons with kinetic energies around 2 MeV and 14 MeV,
respectively. CockcroftWalton accelerators are also used in solid state physics, ion beam modification
of materials, atomic physics, ion beam analysis of materials like Rutherford backscattering (RBS), particle induced X-ray emission (PIXE), particle induced gamma emission (PIGE), nuclear reaction analysis
(NRA) and elastic recoil detection (ERD), astrophysics and ecological research.
Van de Graaff and Tandem Van de Graaff accelerators are still in use for basic research in nuclear
physics. Besides that they are used in many applications. The capability of producing a large variety of
12
106

E LECTROSTATIC ACCELERATORS

ion beams with an energy of 100 keV up to several 100 MeV make the Van de Graaff and Tandem Van
de Graaff accelerators suitable for:
Materials modification
Ion implantation and ion beam mixing
Materials analysis

Rutherford backscattering spectroscopy (RBS),

particle induced X-ray emission (PIXE),
particle induced gamma ray emission (PIGE),
nuclear reaction analysis (NRA),
elastic recoil detection (ERD),
resonance scattering analysis (RSA),
accelerator mass spectrometry (AMS) and
ion micro beam applications (-beam).

Particle production
Medical, security
Radiation production
X-ray imaging
The ion micro beam application makes full use of the high beam quality of Van de Graaff and Tandem
Van de Graaff beams. A famous application of AMS is the carbon dating of ancient probes by measuring
the 14 C concentration, see Fig. 10.
Small compact electron accelerators are of great industrial importance. Dynamitron generators are
commonly used as high voltage generators in order to accelerate very high beam currents. These systems
service a wide range of industrial applications:

Production of X-rays,
Sterilization of medical products,
Wire and cable cross-linking,
Tyre and rubber pre-cure treatment,
Shrink wrap sheet products,
Thin films polymer cross-linking,
Heat shrinkable tubing and plastics,
Polymer tube cross-linking,
Bulk polymer modification,
Sheet foam materials,
Silicon wafer processing,
Specialty automotive wire,
Food irradiation,
Purification of gases,
Treatment of waste water and toxic wastes,
Advanced composites modification and
Scissioning of long chain polymers.

13
107

F. H INTERBERGER

Fig. 10: 14 C accelerator mass spectrometry using a 3 MeV Tandem Pelletron from NEC (courtesy of G. A. Norton,
NEC)

Progress in electrostatic accelerator development

In this section we will sketch the progress in the development of electrostatic accelerators. In the 1960s ad
1970s big Tandem Van de Graaff accelerators were developed and installed worldwide for basic research
in nuclear and heavy ion physics. The culmination were Tandem accelerators achieving terminal voltages
of more than 20 MV like the facilities in Oak Ridge and Yale, USA, in Daresbury, U. K. and the Vivitron
facility in Strasbourg, France. But there was simultaneously a growing demand for small electrostatic
accelerators for applications in medicine, biology and industry. This trend continues and goes to compact
and customized electrostatic accelerators. In the following, we sketch a few examples.
Beginning in 1971, the Budker Institute of Nuclear Physics in Novosibirsk started the development
and manufacturing of high-power electron accelerators of the ELV-type for use in industry (see Fig. 11).
The acronym ELV means ELectron Rectifier (Russian: Viprjamitel). The covered energy range is 0.2
2.5 MeV with beam currents 25800 mA and beam powers 20500 kW. The high voltage generator is a
cascade generator with parallel inductive coupling. The electron beam passes through a thin titanium foil
into air. The beam intensity is homogeneously distributed over a large circular area of 1.7 m diameter
using two scanning magnets.
Another manufacturer of industrial electron accelerators is Radiation Dynamics Inc. (RDI), Long
Island (USA). RDI was founded in 1958, now RDI-IBA is part of the IBA Technology Group, Louvainla-Neuve, Belgium. During the first ten years RDI primarily supplied accelerators to universities and
governmental research facilities, the majority of which are still operating today. Since the late 1960s
RDI has supplied over 250 industrial electron beam systems. The Dynamitron generator insures smooth
and reliable operation at high voltages. The generator avoids the use of large capacitors, which are
particularly vulnerable to spark damage at voltages above 1 MV. Since the Dynamitron has very low
stored energy, the chance of spark damage is minimal. Even when sparking does occur, the unit may
14
108

E LECTROSTATIC ACCELERATORS

Fig. 11: View of the ELV-4 accelerator from BINP, Novosibirsk (courtesy of V. V. Parkhomchuk, BINP)

be quickly restarted and returned to normal operation. The standard Dynamitron models are designed
for high voltages between 0.55 and 5.0 MV. The available beam currents are between 10 and 160 mA,
depending on the voltage. The maximum beam power is about 200 kW.
Concerning positive ion accelerators compact customized electrostatic accelerators are produced
by High Voltage Engineering Europe (HVEE), Amersfoort, The Netherlands and by National Electrostatics Corporation (NEC), Middleton, Wisconsin, USA.
HVEE produces air insulated accelerator systems up to 500 kV, single ended accelerator systems
Singletron up to 6 MV and Tandem accelerator systems Tandetron up to 6 MV terminal voltage. A
modular design allows customized systems from standard components. The acceleration voltage of the
air insulated systems is supplied by a free standing generator consisting of a RF driver and a voltage
multiplier and rectifier stack. A high voltage stability is achieved using a fast stabilization circuit. The
acceleration tube consists of glass insulators and aluminum electrodes. A low voltage gradient makes
conditioning of the acceleration tube superfluous. Secondary electrons are suppressed by electron catcher
and an internal magnetic suppression system.
The high voltage generator of the Singletron and Tandetron accelerator systems is a SF 6 insulated
Dynamitron power supply. It is characterized by a very high voltage stability and a low voltage ripple
resulting in an excellent beam energy resolution. The most compact versions are the coaxial Singletron
and Tandetron accelerators (see Fig. 12) where the HV power supply is built around the acceleration tube.
In the other versions the power supply is a self-supporting unit in a separate section of the pressure vessel.
The new generation of single ended Van de Graaff accelerators is equipped with a unique ion source
exchange system that allows for ion source exchange without opening the pressure vessel. Complete
Singletron and Tandetron systems for ion implantation and ion beam analysis and Tandetron systems for
15
109

F. H INTERBERGER

Fig. 12: Photo of 5 MV Tandetron in service position from HVEE (courtesy of H. van Oosterhout, HVEE)

accelerator mass spectrometry are available.

NEC produces three distinct classes of electrostatic accelerators. The U-series are Pelletron accelerators with terminal voltages from about 4 MV up to about 25 MV. They are mainly used for basic
nuclear structure research. The S-series are Pelletron accelerators with terminal voltages from <1 MV
up to about 5 MV for applications involving MeV ion implantation, surface analysis, micro-beam applications, electron beam and X-ray production up to 5 MeV, neutron generation and advanced research. In
addition open-air accelerators up to 250 kV are available. The NEC high gradient acceleration tube used
in the Pelletron accelerators is an all metal/ceramic assembly, see Fig. 6. Titanium electrodes are bonded
to alumina ceramic using aluminum diffusion bonding. There are no organic compounds in the vacuum.
Limiting apertures and an external magnetic field suppress secondary electrons in the tube. In order to
provide higher beam currents the Pelletron accelerators can be equipped with up to four charging chain
systems. For maximum stability, the control system uses a liner driver in addition to the corona probe
to increase terminal voltage stability and decrease ripple. Signals from the capacitive pick off plates are
inverted and amplified providing a potential to the liner driver. Stabilities of better than 10 4 are typical.
For accelerator mass spectrometry, especially radiocarbon dating, a 3 MV Tandem accelerator system
(see Fig. 10) and a new single stage 250 kV accelerator system is available.
The progress in building compact electrostatic accelerators pioneered the development of commercial ion implanters. Semiconductor manufacturer use ion implantation for doping processes in silicon
integrated circuits. Doping and modifying silicon and other semiconductor wafers involves generating
an ion beam and steering it into the substrate so that the ions come to rest beneath the surface. The most
commonly implanted ions are arsenic, phosphorus, boron, boron difluoride, indium, antimony, germanium, silicon, nitrogen, hydrogen, and helium. The scheme of an ion implanter is shown in Fig. 13.
The ion energies of ion implanters must be variable over a wide range. High current implanters can
produce currents up to about 30 mA with ion energies from 1 keV to about 200 keV. Medium-current
implanters are designed for maximum dose uniformity and repeatability. Their beam currents are in the
range of 1 A to 5 mA, at energies from 2 keV to about 900 keV. High-energy implanters provide ion
beams in the MeV range with beam currents for singly-charged ions up to about 1 mA and for multiplycharged ions up to about 50 A. High-energy implanters with a large energy range can also be used for
many medium-current applications in the energy range 10900 keV. Major manufacturers of commercial
ion implanters are Axcelis Technologies, Beverly, MA, Varian Semiconductor Equipment Associates,
Gloucester, MA, Applied Materials Implant Division, Horsham, U.K., Sumitomo Easton Nova (joint
venture with Axcelis), Tokyo, Japan, Nissin Electric, Kyoto, Japan and Ibis Technology, Danvers, MA.

16
110

E LECTROSTATIC ACCELERATORS

HV

IMPL

ACC

HV
W
I

Fig. 13: Scheme of an ion implanter. I: ion source, A: analysing magnet, HV: high voltage terminal, ACC:
electrostatic accelerator, H,V: horizontal, vertical electrostatic raster scanning, IMPL: ion implantation chamber,
W: silicon wafer.

Conclusion

The basic features of electrostatic accelerators are described in this paper. The principles of Cockcroft
Walton, Van de Graaff and Tandem Van de Graaff accelerators are presented. This presentation includes
high voltage generators such as the CockcroftWalton cascade generator, the Van de Graaff band generator, the Pelletron and Laddertron generator and the Dynamitron generator. The beam properties and the
application of electrostatic accelerators in research and industry are sketched. The progress in the development of compact and customized electrostatic accelerators is emphasized. Electrostatic accelerators
are very much alive.

Acknowledgements
The author wishes to thank G.A. Norton, H. van Oosterhout and V.V. Parkhomchuk, for helpful information and discussions.

Copyright notice
Figures 1, 2, 4 and 7 reprinted from from F. Hinterberger, Physik der Teilchenbeschleuniger und Ionenoptik, Springer-Verlag Berlin Heidelberg New York 1997, Chapter 2 with kind permission of Springer
Science and Business Media (copyright 2005).
Figures 6, 8 and 10 by courtesy of Dr. G.A. Norton, National Electrostatics Corp. (NEC), Middleton, Wi 53562-0310, USA ([email protected]).
Figures 9 and 12 by courtesy of H. van Oosterhout, High Voltage Engineering Europe (HVEE),
Amersfort, The Netherlands ([email protected]).
Figure 11 by courtesy of Dr. V.V. Parkhomchouk, Budker Institute of Nuclear Physics (BINP),
Novosibirsk, Russia ([email protected]).

References
[1] J.D. Cockcroft and E.T. Walton, Proc. Roy. Soc. (London) A136 (1932) 229, A144 (1934) 704.
[2] H. Greinacher, Z. Phys. 4 (1921) 195.
[3] Figure reprinted from F. Hinterberger, Physik der Teilchenbeschleuniger und Ionenoptik, SpringerVerlag, Berlin, 1997, Chapter 2, with kind permission of Springer Science and Business Media
(copyright 2005).
[4] R.J. Van de Graaff, Phys. Rev. 38 (1931) 1919A.
17
111

F. H INTERBERGER

[5]
[6]
[7]
[8]
[9]
[10]

R.G. Herb, Nucl. Instrum. and Methods 122 (1974) 267.

M.R. Cleland and M.R. Morgenstern, Nucleonics 18 (1960) 52.
W.H. Bennet and P.F. Darby, Phys. Rev. 49 (1936) 97, 422, 881.
H. Kallmann, German patent 696998, 9 February 1938.
L.W. Alvarez, Rev. Sci. Instrum. 22 (1951) 705.
R.J. Van de Graaff, Nucl. Instrum. and Methods 8 (1960) 195.

18
112