Nuclear Instruments and Methods in Physics Research A 763 (2014) 510–516

Contents lists available at ScienceDirect

Nuclear Instruments and Methods in

Physics Research A
journal homepage: www.elsevier.com/locate/nima

Development and discharge characteristics of negative hydrogen ion

source for the SKKUCY-9 cyclotron
Yeong Heum Yeon a, Mitra Ghergherehchi b, XiangJie Mu b,
Khaled Mohamed Mohamed Gad a, Jong Seo Chai b,n
a
Sungkyunkwan University, Department of Energy Science, Suwon, Korea
b
Sungkyunkwan University, College of Information & Communication Engineering, Suwon, Korea

art ic l e i nf o a b s t r a c t

Article history: A negative hydrogen ion source was designed and manufactured for the SKKUCY-9 cyclotron at
Received 25 April 2014 Sungkyunkwan University. This ion source is a penning ionization gauge type source. We analyzed
Received in revised form the electromagnetic field distribution for simulation of electron temperature, electron trajectory, and
26 May 2014
average electric field in the anode using CST Particle Studio. We applied the ion bombardment position
Accepted 12 June 2014
Available online 1 July 2014
by electric field simulation data and confirmed the thermal distribution of the cathode. The anode
structure was developed using these simulation results. The penning discharge of plasma in the ion
Keywords: source occurred at 0.4 T. This ion source operates with a hydrogen flow of 1–5 cm3/min. The ignition
Negative hydrogen ion source voltage is approximately 1.7 kV and the arc current varies from 0.1 A to 2 A. The structure of the ion
Electromagnetic field simulation
source was constructed and the experimental results are in good agreement and can be useful
Discharge characteristics
information for the operation of a negative ion source.
& 2014 Elsevier B.V. All rights reserved.

1. Introduction The combined electric and magnetic fields make the electrons
oscillate, and these electrons are trapped in the plasma with a high
Sungkyunkwan University has developed the compact infor- probability of ionization effects. The electron emission takes place
mation-technology based SKKUCY-9 cyclotron for producing 18F. at the two cold cathodes, and the energy of the electron emission
It has a fully automated system and chemistry modules. The must be enough to excite hydrogen in the anode.
SKKUCY-9 cyclotron accelerates H
ions to obtain a 9 MeV proton The hydrogen is ionized with electrons, and this ionized
beam with a carbon stripper. In this cyclotron, an internal penning hydrogen produces an ion bombardment, which increases the
ionization gauge type ion source is used for producing the H
ions. temperature of the cathodes. As the temperature of the cathodes
The schematic of the penning ionization gauge type ion source is increases because of the ion bombardment, the increasing emis-
shown in Fig. 1. sion rate of electrons also results in hydrogen plasma discharge.
After L. R. Maxwell investigated penning discharge in 1931, The voltage drop of the plasma switches from a few thousand volts
penning ionization gauge-type ion sources were used as internal to a few hundred volts when the cold type cathode turns into a
sources in cyclotrons in the 1940s, because they have a simple thermo-cathode at a specific temperature. The temperature dis-
design and do not need an inflector for introducing beams [1–4]. It tribution of the cathodes is important for electron emission and
was used not only in cyclotrons but also in linear accelerators self-discharge of the plasma. The design of the anode is optimized
forproducing multiply charged ions [5,6]. by calculating this temperature distribution. The arc defining
In such an ion source, the H
ions are produced in an arc radius was confirmed by the temperature distribution of cathode.
discharge with a magnetic field. Applying a high voltage, two cold- The diameter of the anode was confirmed with the arc defining
type cathodes generate electrons for producing hydrogen plasma. radius.
The electric field between the two cathodes and the one hollow The combined electric and magnetic fields were optimized for
anode plays an important role in extracting H
ions from the slit designing the structure and positioning the cathodes and anode.
of the ion source. The distribution of the electric and magnetic fields was considered
for the H
ion production processes in hydrogen plasma, because
electron collision processes such as dissociative electron attach-
n
Corresponding author at: Sungkyunkwan University, College of Information &
ment and polar dissociation with neutral molecules in the ground
Communication Engineering, 2066, Seobu-ro, Jangan-gu, Suwon-si, Korea. state are the main processes. The behavior of the electron-volume
E-mail address: [email protected] (J.S. Chai). process is very important for producing H
ions.

http://dx.doi.org/10.1016/j.nima.2014.06.048
0168-9002/& 2014 Elsevier B.V. All rights reserved.
 Y.H. Yeon et al. / Nuclear Instruments and Methods in Physics Research A 763 (2014) 510–516 511

Fig. 1. Schematic of the PIG ion source.

For deciding the anode diameter, we calculated the arc-

defining radius, because the anode should cover all of the plasma.
We assumed that the anode diameter is twice the arc-defining Fig. 2. The structure of the cathodes and the anode in CST Particle Studio.
diameter. High-energy primary electrons accelerated at a cathode
sheath produce vibrationally excited molecules, and low-energy
electrons produce H
ions by dissociative attachment to the
vibrationally excited molecules [7]. The density difference of Table 1
electrons and ions produces the cathode sheath and ion accelera- The cathode materials, ignition voltage, and work function.
tion to the cathodes. This acceleration increases the cathode
Material Ignition voltage (V) Work function
temperature. The ion bombardment results in a specific diameter
assumed to be the arc-defining diameter. Nickel 3600 5.15
We analyzed the electromagnetic field distribution for simula- Aluminum 3500 4.28
tion of the electron temperature, electron trajectory, and average Copper 2300 4.65
electric field in the anode using CST Particle Studio [8]. The Carbon 2300 5.0
Tungsten 2100 4.55
transition region length was calculated from the simulation
Molybdenum 1800 4.6
results. The positioning cathodes and anode are related to this Tantalum 1700 4.25
transition region length. The anode diameter was decided by
ANSYS simulation [9]. We simulated the electromagnetic field
distribution and the ion bombardment position. The temperature
distribution of the cathode with ion bombardment using the
results of CST Particle Studio was simulated by ANSYS. of the cathodes and the anode in CST Particle Studio is shown in
Fig. 2. The cathode diameter of 10 mm and anode diameter of
12 mm were chosen at first. The thickness of the cathode was
2. Simulation procedure for the penning ion source 2 mm. The length of the anode was 12 mm. The distance between
the cathodes and the anode was 5 mm. These values were variable
In a penning ionization gauge type ion source, the two parameters in the optimization of the ion source structure.
cathodes play important roles in creating a self-maintained dis- The voltages applied to the cathodes create an electric field
charge plasma. In the cathode, there are cathode sheaths where distribution at the source, and this distribution decides the plasma
high-energy electrons accelerate. The high-energy electrons are length. An electric field distribution simulation was done for
confined by a magnetic field. To confirm the anode diameter, we several cathode materials. The ignition voltage of the cathodes in
divided the plasma into two regions: the central region and the hydrogen gas was applied for simulating the electron energy and
surrounding region. When we apply a voltage higher than average electric field. The ignition voltage of the cathodes is shown
the discharge ignition voltage, a self-maintained discharge will in Table 1 [10]. This information is related to the transition region
be generated in the anode central region with low-pressure gas. length. This transition region length decides the distance between
The generation of a self-maintained discharge is determined by the cathodes and the anode.
the appropriate selection of the cathode material. The discharge of The fluid equation is applied for calculating the length of the
plasma must be continuously maintained to obtain a continuous transition region in the reflex discharge
ion current from the source.  
The internal penning ionization gauge type ion source must eE
ne ¼ n0 exp
x
have transversal ion extraction in the center of the cyclotron with Te
the puller which extracts the H
ions to the starting position of
the acceleration. The electron emission from a cold-type cathode is where ne, no, E, Te, x are the electron density, the charge density,
perpendicular to the ion extraction, and it has an important role in electric field strength, electron temperature and the transition
producing ions in the anode. region length respectively. In the plasma edge, the ratio of the
CST Particle Studio was used for analyzing the electromagnetic electron density and charge density is 0.2 according to Chen's
field distribution in the cathodes and the anode. The basic structure experiment [11]. As a result of the equation below, we can obtain
512 Y.H. Yeon et al. / Nuclear Instruments and Methods in Physics Research A 763 (2014) 510–516

the electric field in the transition region. reflected electrons [12]. Two types of anodes, a tapered anode and
a cylindrical anode were simulated for calculation of the secondary
Te
E ¼ 1:6 electron number. We simulated the number of secondary electrons
ex
after deciding the cathode material and the location of the
We applied average electric field and average temperature of cathodes and the anode.
electrons between the cathodes and the anode to find the transi- The emission electron temperature and applied voltage were
tion region length. used for the thermal distribution of the cathodes. The thermal
The transition region length was calculated. The appropriate distribution of the cathodes is important for their life time. The
cathode material was chosen for the optimization of conditions simulation of the thermal distribution was done by ANSYS to
such as lifetime, the center gap of the cyclotron, and price. confirm the choice of the cathode material also. The thermal
A particle tracking solver is used for the static field distribution distribution of the cathodes is needed because the ion bombard-
simulation in the cathodes and the anode. The magnetic field is ment temperature is the main reason for switching a cold-type
simulated by coil construction in CST Particle Studio. The magnetic cathode to a thermo-cathode. We used steady-state thermal-
field direction is perpendicular to the extraction direction of the H
electric conduction analysis for the solution of the thermal and
ions. When the emitted electrons or the plasma ions are moving, electric fields. These coupled-fields simulate joule heating for
they feel a Lorentz force, which is perpendicular to both the particle resistive materials. We determined the ion bombardment area
velocity and the field. This force has no effect on the velocity with the results of the CST Particle Studio electron trajectory and
component parallel to the magnetic field, but in the perpendicular confirmed the anode diameter with the thermal distribution area.
plane, it forces the particles to rotate in a cyclotron orbit. We used a permanent magnet for the electrons oscillation and
The frequency of the particle's motion is known as the confinement in a test chamber. The magnetic field distribution was
cyclotron frequency. This value is given by simulated in Opera Tosca [13]. The potential was calculated using a
! finite element method. The position of the permanent magnet was
eB decided by the structure of the test chamber. The permanent
ωce;i  ;
me;i magnets were located in upper cover of the test chamber and the
bottom of it. The distance between the permanent magnet and the
where e, B, and me;i are the electron charge, the magnetic field, and
ion source center point was 45 mm. We applied a ‘vacodym 863 tp
the electron/ion mass respectively. The Larmor radius is defined as
[14]’ permanent magnet B–H curve. The positions of the perma-
υ? nent magnets are shown in Fig. 3.
r ce;i  ;
ωce;i
!
where υ ? is the velocity perpendicular to Β . The Larmor radius
decides the central region of the plasma. We calculate this Larmor 3. Experimental setup
radius using CST Particle Studio. The electron emission was
calculated using a thermionic emission model. The velocity, The penning ionization gauge type ion source consists of two
energy, and electric field are simulated for designing the ion tantalum cathodes, one copper-tungsten alloy anode, austenitic
source. stainless steel support pipes, oxygen-free highly conductive
The secondary electron emission was also calculated by CST copper pipes, and quartz. The copper pipes are connected with
Particle Studio to select the type of anode. There are three basic an arc power supply for applying high voltage to the tantalum
types of secondary electrons: backscattered, rediffused, and elastic cathodes. The quartz tube is installed in between stainless steel

Fig. 3. The simulation of permanent magnet.

 Y.H. Yeon et al. / Nuclear Instruments and Methods in Physics Research A 763 (2014) 510–516 513

support pipes and copper pipes for insulation, because the stain- the ion source slit. We applied a voltage of 5 kV to the copper
less steel support pipes are in contact with the anode and the electrode. The position of the copper electrode was 1 cm from the
copper pipes sustain the cathodes. A Teflon tube was chosen at ion source slit.
first for the insulator because of its high insulation, but it has a low The copper electrode was also inserted in the SKKUCY-9
melting point of 600 K. It is burned when hydrogen plasma is cyclotron for measuring the H
current. We applied a voltage of
formed in the ion source. When it burns, the high voltage between 5 kV to the copper electrode to extract the H
ions, just like a
the stainless steel support pipes and the copper pipes creates a puller in the test chamber. However the puller extraction voltage
short circuit. The short circuit damages to the arc power supply, was 40 kV, the same as a dee voltage. We applied only 5 kV to the
because the high current flows to the arc power supply. A resistor electrode because of the discharge between the electrode and the
protects the arc power supply against the high current. The ion source. The puller structure is complicated than the electrode
resistor consumes power coming from the short circuit and and its position is far from the ion source. We used a simple
damages the high power supply. copper electrode to check the H
current because we wanted to
The ion source was installed to the test vacuum chamber to measure the H
current without RF electric field effects.
check the self-maintained discharge plasma. The vacuum chamber
is the same as the vacuum chamber used for a cyclotron system,
and it was made of aluminum due to its non-magnetic properties. 4. Results and discussion
The permanent magnets were installed in the upper and lower
side of the vacuum chamber. A vacuum window, and vacuum 4.1. Electric field and thermal distribution simulations
sensor gauge were connected to the vacuum chamber, and a feed
through was connected to an ampere meter for checking the H
The trajectory of electrons was simulated by CST Particle Studio
ion current. A power supply was included for extraction. We with electric and magnetic field distributions. These simulations
checked the gas flow rate using a gas flow meter before the ion were done to decide the length of the transition region. The length
source gas inlet. between the anode and the cathode was decided using this length,
A self-maintained discharge was confirmed in the test chamber. which was determined by the emission electron temperature.
We measured the H
current using a copper electrode in front of The electric field strength between the cathode and the anode
is shown in Fig. 4 for a tantalum cathode. We started by assuming
that the length of the anode was 12 mm. The distance between the
cathode and the anode is optimized by calculation of the transition
region length.
The equipotential lines of the cathode and anode are shown in
Fig. 5. The electrons accelerate to the anode because of the voltage
difference between the cathode and the anode. The electrons
accelerate perpendicular to the direction of the equipotential lines.
The electron trajectory is shown in Fig. 6. We began with a
cathode diameter of 10 mm, but the space of this diameter allows
accelerating electrons to have enough energy for creation of self-
discharge plasma. Thus, we decreased the cathode diameter to
8.5 mm in order to reduce the size of the ion source.
The appropriate temperature of electrons emitted from tanta-
lum cathodes is 140 eV, and the electric field is 395 V/cm in the
cathode sheath (from the simulation). The length of the transition
region is 5.8 mm, and the total length of the transition region
should be approximately 12 mm (two cathode sheaths). The total
Fig. 4. The electric field strength between the tantalum cathode and the anode. plasma height should be 36 mm from the calculation result,

Fig. 5. The equipotential lines of the cathode and anode structure.

514 Y.H. Yeon et al. / Nuclear Instruments and Methods in Physics Research A 763 (2014) 510–516

Fig. 6. The electron trajectory.

because the internal penning ionization gauge type ion source is a

reflex-type ion source.
The total height of the ion source is 45 mm in order to cover the
plasma height and the anode. The height of the ion source is
related to the center gap of the SKKUCY-9 cyclotron. The center
gap is 70 mm, which is enough for positioning the ion source with
a puller.
The distances between the cathodes and the anode were
selected based on these simulation results. The distance between
the anode and the cathode was 6.5 mm to cover the transition
region length and the cathode sheaths. We assumed that the
radius of the hydrogen plasma's surrounding region was approxi-
mately 4 to 6 mm.
Tantalum cathodes are suitable for these conditions, and they
operate at a relatively low applied voltage for a long time because
Fig. 7. The arc defining radius.
of the low rate of cathode sputtering [15]. They are 2-mm thick
circular cylinder-type cathodes, and their diameter is 8.5 mm.
They are screwed to a cathode support block connected to an arc
power supply.
The anode must also cover the plasma. The diameter of the
anode was selected based on the simulation results of the particle
trajectory and thermal distribution of the cathodes. The current
density of electrons and ions is shown in Fig. 7. The arc defining
radius was confirmed by the particle trajectory simulation. The ion
bombardment position was decided by the electric magnetic field
simulation results. The energy consumption was applied to ANSYS
and we assumed an arc defining radius of 3 mm and determined
the anode radius to be 3.5 mm. The thermal distribution of the
cathode is shown in Fig. 8.
We considered two types of anode diameters: a tapered one
and a cylindrical one. Both the anodes were simulated with the
same internal diameter of 7 mm. The structure of the anode is Fig. 8. The thermal distribution simulation with ANSYS.
shown in Fig. 9. The number of secondary electrons was simulated
by CST Particle Studio to confirm which had more secondary simulation results, we choose the tapered anode with a diameter
electrons. The number of secondary electrons is related to the of 7 mm.
discharge of the plasma, because efficient ionization happens with The surface magnetic field of ‘vacodym 863 tp’ permanent
a high probability of electron-neutral collision occurring. magnet is 2 T. The magnetic field distribution is shown in Fig. 11.
The number of secondary electrons is approximately 3 times The magnetic field is 0.4 T in the cover of the test chamber, and it
higher with the tapered anode, as shown in Fig. 10. Using these has a symmetric distribution.
 Y.H. Yeon et al. / Nuclear Instruments and Methods in Physics Research A 763 (2014) 510–516 515

Fig. 9. The optimization of the anode structure.

Fig. 12. The self-maintained discharge in the ion source.

Fig. 10. The number of secondary electrons in two types of anodes.

Fig. 13. The H
current with increasing magnet power supply current.

We applied the checked ion source to the SKKUCY-9 cyclotron

Fig. 11. Magnetic field distribution of permanent magnet.
and measured the H
current. The position of the copper elec-
trode was 1 cm far from ion source slit. The copper electrode
was rectangular in shape and had a size of 5 cm by 2 cm.
4.2. The discharge characteristics in the ion source The thickness was 1 mm. It covered the H
ions motion. The
H
current increased as the magnetic field increased. The H

The ion source was tested with regard to plasma conditions such as current was maximized at 138 A of magnet power supply, which is
self-maintained discharge and ignition voltage. The self-maintained the operating current of the SKKUCY-9 cyclotron.
discharge is shown in Fig. 12. We can see the plasma through the The H
current was measured with increasing magnetic power
vacuum window. The H
current is 1 mA in the test chamber. The supply current for various electrode positions. We found the position
vacuum condition of the test chamber was 5  10
5 mbar. The gas of the electrode with the largest current value. The H
current is
flow rate is 5 cm3/min. The extraction voltage is 5 kV. shown in Fig. 13. The magnitude of the magnetic field was simulated
516 Y.H. Yeon et al. / Nuclear Instruments and Methods in Physics Research A 763 (2014) 510–516

Fig. 14. The magnetic field (at center) versus power supply current.

Fig. 16. The H
current and vacuum condition with increasing gas flow rate.

parameters to the fluid equation. The length of the transition

region was 5.8 mm, and the total length of the transition region
was approximately 12 mm (two cathode sheaths). The distance
between the cathode and anode was 6.5 mm. The total anode
length was 15 mm for covering the plasma. The length between
the upper pocket cover and lower pocket cover was 45 mm. The
cathode diameter was 8.5 mm.
The thermal distribution due to ion bombardment to the
cathode was simulated to confirm the anode diameter. We
assumed an arc defining radius of 3 mm. The anode diameter
was 7 mm to cover the plasma arc cylinder. The type of anode was
tapered, because the number of secondary electrons was 3 times
higher than a cylindrical anode.
The ion source was well-designed and manufactured for the
SKKUCY-9 cyclotron. Self-maintained discharge occurred in a 0.4 T
Fig. 15. The H
current with increasing arc current and gas flow rate. magnetic field. The gas flow rate varied from 2 to 5 cm3/min. The
arc current varied from 0.3 to 2 A. The results of the H
current at
different arc currents, gas flow rates and magnetic fields are
for power supply currents between 80 A to 140 A. With increasing presented in this paper.
current, the magnetic field increases as shown in Fig. 14
The H
current for different gas flow rates is shown in Fig. 15. The
H
current increased when the arc current increased. The maximum Acknowledgements
H
current was 1.2 mA in the copper electrode 1 cm from the ion
source slit at a 5 cm3/min gas flow rate and 1.2 A of arc current. This
This work was supported by the IT R&D program of IITP/MSIP.
measurement was done at 138 A of magnet power supply which is
[10043897, Development of 500 cGy level radiation therapy
the operating current of the SKKUCY-9 cyclotron. The H
current and
system based on automatic detection and tracing technology with
vacuum condition with increasing gas flow rate was shown in Fig 16.
dual-head gantry for 30% reducing treatment time for cancel
We measured the H
current and the vacuum pressure of the
tumors].
SKKUCY-9 cyclotron as function of the gas flow rate. The measure-
ment was done under operating condition of the SKKUCY-9 cyclo-
tron. The H
current was increased with increasing gas flow. References
Ion plasma in the range of 5  10
6–6  10
5 mbar pressure
results the optimum plasma. Above 6  10
5 mbar the H
current [1] L.R. Maxwell, Rev. Sci. Instrum. 2 (1931) 129.
decreases with increasing gas flow. [2] A.T. Finkelstein, Rev. Sci. Instrum. 11 (1940) 94.
[3] R.S. Livingston, J. Appl. Phys. 15 (1944) 2.
[4] D.H. An, et al., Rev. Sci. Instrum. 79 (2008) 02A520.
[5] C.R. Anderson, K.W. Ehlers, Rev. Sci. Instrum. 27 (1956) 809.
5. Conclusions [6] B. Gavin, Nucl. Instrum. Methods 64 (1968) 73.
[7] J.M. Wadehra, Appl. Phys. Lett. 35 (1979).
The penning ionization gauge type ion source can be used for [8] CST-Particle studio, DE, 〈http://www.cst.com〉.
[9] ANSYSs Academic Research, Release 14.0.
many applications, we designed and developed this ion source for
[10] K.N. Leung, K.W. Ehlers, J. Appl. Phys. 52 (1981) 3905.
compact medical cyclotron. [11] Francis F. Chen, Phys. Rev. Lett. 234 (1962).
The electromagnetic field distribution was simulated to con- [12] F. Hamme et al., TUAPMP4, ICAP06, Chamonix, 2006.
firm the structure of the ion source. The average electron tem- [13] Vector Fields Limited, OPERA-3D User Guide & Reference Manual: 2012
Version, Oxford.
perature was 140 eV and the electric field was 395 V/cm in the [14] 〈http://www.vacuumschmelze.com〉.
cathode sheath. With these simulation results, we applied the [15] J.D. Gow, J.S. Foster, Rev. Sci. Instrum. 24 (1953) 606.