Journal of Physics: Conference Series

PAPER • OPEN ACCESS

Multipacting studies of the coaxial coupler for BNCT DTL

To cite this article: M X Fan et al 2019 J. Phys.: Conf. Ser. 1350 012194

View the article online for updates and enhancements.

This content was downloaded from IP address 36.81.57.188 on 07/06/2020 at 04:57

10th International Particle Accelerator Conference IOP Publishing
Journal of Physics: Conference Series 1350 (2019) 012194 doi:10.1088/1742-6596/1350/1/012194

Multipacting studies of the coaxial coupler for BNCT DTL

M X Fan1,2*, H C Liu1,2, A H Li1,2, P H Qu1,2, Y Wang1,2, X L Wu1,2, B Li1,2, Q

Chen1,2, J Peng1,2, A X Wang1,2 and S N Fu1,2
1
Institute of High Energy Physics, Chinese Academy of Sciences (CAS), [100049]
Beijing, China
2
Dongguan Neutron Science Center, [523803] Dongguan, China

*[email protected]

Abstract. Multipacting is a phenomenon in which electrons grow sharply under certain

conditions in a RF structure. It may lead to the breakdown or even damage to the equipment.
Therefore, it is very important to calculate the Multipacting range in the RF equipment design.
Since the phenomenon is too complicated to use the formula to fully predict it, numerical
simulation is employed. There are many computer codes (such as Track3P, MultiPac, CST PS,
etc.) used to simulate the phenomenon, but most of them are not commercial. In this paper,
theories used in coaxial line for predicting multipacting are introduced; the CST PS is chosen
to simulate the multipacting of coaxial coupler for BNCT DTL; finally, methods of suppressing
multipacting are discussed.

1. Introduction
Boron Neutron Capture Therapy (BNCT) is considered one of the most promising tools for treating
certain types of cancer [1], it’s developed and researched by an increasing number of workgroups.
Now, a new BNCT facility contains a 75 keV H- Ion Source, a 3.5 MeV Radio Frequency
Quadrupole (RFQ) and a solid Li target is under construction in China Spallation Neutron Source
(CSNS) campus. Considering the neutron flux and difficulty of target manufacturing, a 3.5 to 10 MeV
Drift Tube Linac (DTL) is planned to construct as an upgrade solution after successful obtaining the
neutron beam. Preliminary physical design of the DTL shows that the required peak and average
power does not exceed 600 kW and 200 kW independently, therefore, the coaxial type coupler is
chosen. Preliminary design of the coaxial coupler draws on the successful experience of coupler used
in many machines such as ADS injector-1 RFQ, TRASCO RFQ, JPARC DTL and so on, the process
of it is quite standard and similar to the design of CSNS DTL coupler [2][3], so the detailed
parameters will not be discussed here.
Multipacting in coaxial coupler has been studied for many years and it can be a limit factor of the
coupler. Different analytical formulas are used to predict the multipacting barriers; however, they
often base on homogeneous electromagnetic fields and the estimated results are rough for complex
structures. Therefore, the numerical calculation is quite essential for reliable prediction of multipacting
barrier. In this paper, theories commonly used to calculate the multipacting of coaxial coupler are
introduced firstly, and the two-point 1st order multipacting point of BNCT DTL coupler is calculated
according to the theory. Section 2 introduces the detailed multipacting simulation process with CST
Particle studio (CST PS), after that, the results are compared with theory ones. Finally, the suppression
of multipacting is simulated and analyzed, which is helpful for the new coupler design.
Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
10th International Particle Accelerator Conference IOP Publishing
Journal of Physics: Conference Series 1350 (2019) 012194 doi:10.1088/1742-6596/1350/1/012194

2. Theory
There are many theories to compute the multipacting barrier in coaxial line and the most commonly
used one is the Scaling laws. This theory holds that the multipacting in standing wave (SW) coaxial
line is only caused by electric fields and the one-point and two-point multipacting both may occur.
Based on experience and analysis, it gives the following formulas [4]:
Pone-point ~ (fd)4Z Ptwo-point ~ (fd)4Z2 (1)
where f is the operating frequency, d and Z is the outer diameter of the coaxial line and the line
impedance respectively. The average impact energy obeys the following laws:
Ei ~ (fd)2 (2)
Although the proportional relationship between the multipacting point and the coaxial line parameters
is given, the scale factor also needs to be obtained by many computations or simulations; it could be
difficult for new designers.
Another concise theory is proposed by Z. Zheng in Facility for Rare Isotope Beams (FRIB). They
experienced strong multipacting in high power conditioning of the coaxial coupler for half wave SC
resonators (HWR); the first eight resonators used a DC bias insert into the transmission line to
suppress the multipacting [5]. However, the number of couplers is quite a lot, suppression of
multipacting directly in the coupler is an economical and effective solution. In this theory, the radial
electric field (Er) between the inner and outer conductor is assumed to be constant, the electric field in
coaxial coupler is:
1
V  b 
E  ln( )  (3)
r  a 
the input power of the coaxial line is:
1
A V 2  b 
P (4)
  a 
ln( )

where b is the inner radius of the outer conductor, a is the outer radius of the inner conductor, V is
voltage between the outer conductor and inner conductor, for travelling wave A is 1, for standing wave
A is 0.25, η is the wave impedance in vacuum.
The electron trajectory is considered to be the difference between the outer and inner conductor radius
in a half-integer RF period. So:
 (2n  1)  (2n  1) eEr (b  a) 2 m
st    b  a  Er  (5)
    m
2
(2n  1) e
Er can also be represented by V/(b-a), therefore, the two-point multipacting barrier can be given:
1
A 4 (b  a)4 m2  b 
Pn   ln( )  (6)
(2n  1) 2  e2  a 
n is the order of two-point multipacting, m and e is the quality and charge of the electron.
According to this theory, the multipacting power of the main transmission part in BNCT coupler is 90
kW for the standing wave. Also, we can get the 1st order multipacting power as a function of
impedance by reducing the inner radius and increasing the outer radius (Figure 1); it could be helpful
for the design and modification of the coupler.

Figure 1. P1 vs. coupler impedance.

2
10th International Particle Accelerator Conference IOP Publishing
Journal of Physics: Conference Series 1350 (2019) 012194 doi:10.1088/1742-6596/1350/1/012194

3. Simulation and results

There are two solvers in CST PS studio, one is Tracking Solver (TRK), another is Particle In Cell
Solver (PIC). The difference of the two solvers is that only the PIC solver considers the interaction
between the electric field and the particles, it determines that more powerful computing resources are
needed in PIC Solver than TRK Solver, but the TRK Solver is proved to be useful in Ref [6], so the
solver used in this paper is TRK Solver.
In our simulation, the material is chosen to be SEE-Copper and the secondary emission model is
Furman-Pivi. This model was first proposed by Furman and Pivi in 2002 [7]. The whole process can
be shown in Figure 2.

Figure 2. Different components of Secondary emission.

To ensure the simulation accuracy, the EM field is calculated in EM Solver before the TRK Solver
setup (as shown in Figure 3). First, the mode of interest is import into the TRK Solver as the base for
calculation, the amplitude scaling factor and phase of the field can be changed in this process. Since
the Eigen-mode fields are normalized to 1 J in simulation, the scaling factor s can be evaluated by the
ratio of the actual average electric field in the axial direction of the cavity and the simulation electric
field at the same location. Next, we define the particle source on the target surface, in this step, the
initial emission energy and spread angle can be set, the number of fixed points is defined according to
the size of the emission face and the computation resources. Thirdly, the mesh properties, the
frequency range, the background properties and the boundaries conditions are defined, it must be
noted that if the mesh lines per wavelength are too small, the particles will increase indefinitely and
may lead to the software crash. Finally, in the TRK Solver menu, we select the fields which will be
used in the simulation and run the project, we generally run for 30-200ns.

Figure 3. Electric field distribution.

We can identify the multipacting phenomenon from the simulation results, and the most intuitive
method is to observe the particle vs. time curve and fit the curve with eαt, thus we can get the growth
rate α and time interval T between two occurrences of multipacting, the secondary emission yield
(SEY) can be easily got by eαT [8]. Another method is to calculate the average SEY in CST post-
processing, and this method will be used in our simulation. Also, the trajectories plotted in results will
be helpful for determining the type and location of the multipacting, as shown in Figure 4.

3
10th International Particle Accelerator Conference IOP Publishing
Journal of Physics: Conference Series 1350 (2019) 012194 doi:10.1088/1742-6596/1350/1/012194

Figure 4. Electron Trajectories plotted after simulation.

According to the methods and settings described above, six different places of the BNCT coupler are
simulated, and the initiation SEY line of multipacting is set to 1.2; due to the limit of article length,
three of the six results can be seen in Figure 5.

(a)

(b)

(c)
Figure 5. The particle source of different places (left) and their corresponding SEY
curve with s (right).

4
10th International Particle Accelerator Conference IOP Publishing
Journal of Physics: Conference Series 1350 (2019) 012194 doi:10.1088/1742-6596/1350/1/012194

It should be noted that the simulated value of the on-axial electric field corresponds to the target value
when s is 0.314. As can be seen from Figure 5, the multipacting phenomenon is more likely to occur in
the upper part of the coupler. From the simulation in Figure 5a, the SEY is unstable when s is larger
than 0.25. If we fit the SEY curve in this range, the SEY is less than 1.2 when s is larger than 0.26,
thus multipacting may not occur on target power; but the multipacting range is quite wide and it may
extend the high power conditioning time. In the taper and down part of the coupler, the multipacting
range is narrow and no multipacting will happen on the operation power. We also simulate the outer
surface of the inner conductor; the SEY curve is quite similar to that of outer conductor. Due to the
electric field of inner conductor is larger than that of the outer conductor; the multipacting range of it
is narrower than that of the outer conductor.
The maximum SEY of Figure 5a is 1.46 while s is 0.16, we can calculate the forward input power is
about 79 kW, which is quite close to the calculation results, the difference between the theoretical and
simulation results may be due to the theoretical results only considering the first order multipacting.
Although the multipacting may occur in some power range, we can judge that they are soft barriers,
and the suppression of multipacting in couplers for room temperature cavities are not very difficult
(such as adding bias on the coupler when high power conditioning ), the maximum SEY can be
accepted.
From the design perspective, another simulation is carried out to research the suppression of
multipacting. Generally speaking, the multipacting barrier can be suppressed by changing the
impedance of the coupler; this method will totally change the design of the coupler and is very
complicated. So we try to enlarge the outer diameter, reduce the inner diameter or add choke structure
to the coupler while maintaining the impedance to 50 ohm. After every change, the electric field is re-
simulated and the coupling coefficient is re-optimized. The position of particle source is defined as
that of Figure 5a and the comparison is shown in Figure 6.

Figure 6. The comparison of different methods to suppress the multipacting barrier.

From Figure 6, in low power level (about s<0.17), we can see that adding a choke structure on the
outer conductor is the most obvious method to suppress multipacting, and the least method is to reduce
the radius of the coaxial line. Both the method of adding choke structure and increasing the radius,
their essence are to reduce the electric field of outer conductor on multipacting position. Theoretically
speaking, we can increase the radius infinitely and add a choke large enough to suppress multipacting
in all conditioning power level for the BNCT DTL coupler. However, considering the limit of size, the
structure stability, the price and many other aspects, the impact of these two methods is limited.
Meanwhile, we can see that all these methods have improved the SEY value to some extend when s is
larger than 0.21. In high power level, if these methods are all invalid, the most effective way for
multipacting suppression is to use the design of multiple sections with different impedance according
to Ref [5]. Finally, it can be conclusion that the method of adding choke and increasing radius can

5
10th International Particle Accelerator Conference IOP Publishing
Journal of Physics: Conference Series 1350 (2019) 012194 doi:10.1088/1742-6596/1350/1/012194

suppress multipacting obviously by improves the multipacting power in low power level while
increasing the SEY in high power level. It could be helpful to the design of a new coupler.

4. Conclusion
The theoretical and simulation analysis of the BNCT DTL coupler is described and methods of
suppressing multipacting are discussed in this paper. The theoretical calculation results show that the
two-point 1 order multipacting power is 90 kW which is quite close to the simulation results. The
comparison of different suppression methods shows that two of them are quite useful in low power
level. However, this is just a step in the coupler design, the mechanical and thermal analysis will be
done in the near future.

References
[1] A.J. Kreiner et al. 2007 Nucl Instru Meth B 261 751-754.
[2] M. X. Fan et al. 2017 High Power Laser and Particle Beams 29 29065105-.
[3] M. X. Fan et al. 2018 Proc.Int.29th Linear Accelerator Conf. (Beijing/China) p 767-770.
[4] E. Somersalo et al. 1996 Proc. Int. Particle Accelerator Conf. vol 3 p 1500-1502.
[5] Z. Zhang et al. 2014 Nuclear Instruments and Methods in Physics Research Section A 735, 596-
601.
[6] G. V. Romanov 2008 Proc. Int. 24th Linear Accelerator Conf. (Victoria/Canada) p 166-168.
[7] M. A. Furman and M. T. F. Pivi 2002 Physical review special topics-accelerators and beams 5
p 124404.
[8] B. Xu et al.2013 Chinese Physics C 37 p 127003.