HOM ANALYSIS OF MULTICELL SUPERCONDUCTING CAVITIES

By
John Popielarski

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

Department of Electrical and Computer Engineering

2003
Professor Leo Kempel
 ABSTRACT

HOM ANALYSIS OF MULTICELL SUPERCONDUCTING CAVITIES

By
John Popielarski

Design of superconducting multi-cell cavities and evaluation of their electromagnetic (EM)

properties relies heavily on the use of numerical calculations by computer code. Analysis
of the cavity’s higher-order modes (HOMs) and its associated power couplers are greatly
facilitated by computer simulation. The complete HOM simulation of the Rare Isotope
Accelerator (RIA) six-cell 805 MHz for beams traveling about half the speed of light is

discussed. RF measurements are essential to verify numerical predictions. The calculations

are used, in practice, to design the cavity and perform perturbation studies. The measure-
ments are performed to verify as designed performance. Both are necessary for successful
linear accelerator design, especially for superconducting systems. The results of numer-
ical calculations of the EM properties of HOMs are compared to RF measurements on a

copper model of a 5-cell superconducting cavity. The EM properties of the fundamental

pass-band mode are also compared. The main EM properties of interest are the frequency,
shunt impedance and the strength of the coupling to the input coupler. The measurements
are done with a bead-pull system and a network analyzer.
 ACKNOWLEDGEMENTS

I thank the SRF group at NSCL and the ECE department at MSU for the this research op-
portunity. A special thanks to Walter Hartung, Terry Grimm, Felix Marti, Holger Podlech,
Leo Kempel and Steve Cossmann for their constant help and support. This work is sup-
ported by the US Department of Energy under grant DE-FG02-00ER41144. The elliptical

cavity development at NSCL is being carried out in collaboration with Thomas Jefferson
National Laboratory and INFN-Milano.

ii
In memory of Heather McPherson

iii
 TABLE OF CONTENTS

LIST OF FIGURES v

LIST OF TABLES vi

1 INTRODUCTION 1
1.1 Particle acceleration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 The Rare Isotope Accelerator . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Design and evaluation of superconducting cavities . . . . . . . . . . . . . . 2
1.4 Computer Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 THEORY 5
2.1 Particle acceleration with an RF cavity . . . . . . . . . . . . . . . . . . . . 5
2.2 Pill-box modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Multi-cell elliptical axisymmetric cavities . . . . . . . . . . . . . . . . . . 6
2.4 Perturbation by a metallic object . . . . . . . . . . . . . . . . . . . . . . . 8

3 FIGURES OF MERIT 10
3.1 Some important quantities in superconducting cavities . . . . . . . . . . . 10
3.2 Higher Order Modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.3 Field Flatness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.4 Geometric Shunt Impedance . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.5 Cavity Coupling Strength . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.6 Peak surface fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

4 SIMULATIONS AND MEASUREMENTS OF A 5-CELL PROTOTYPE 14

4.1 Field profiles and Ra Q0 . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.2 Coupling measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

5 DISCUSSION 19
5.1 Analysis of error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

6 CONCLUSION AND FUTURE WORK 21

6.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
6.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

BIBLIOGRAPHY 22

iv
 LIST OF FIGURES
1.1 Set-up for bead-pull measurements. The input antenna is mounted onto
the right end-cap. The power coupler on the left beam tube may be used
as a pickup, although an additional antenna on the left end-cap was used
in some cases. The inset shows the metallic needle and the fishing line
(controlled by the motor on the right) without the end-cap shown in the
main picture. The Qext is measured with the same set-up, adjusting the
input antenna for unity coupling. . . . . . . . . . . . . . . . . . . . . . . . 4

2.1 Six-cell axisymmetric cavity with power coupler port on the left and pick-
up antenna port on the right. The disks on the outside of the endcells at-
tached to the beam tubes are in place to hold the titanium helium vessel
and cavity tuner. Stiffening rings are also drawn in place; they are used to
maintain the shape of the cavity. . . . . . . . . . . . . . . . . . . . . . . . 7

4.1 (a) Raw data collected for the accelerating mode along the beam axis from
a metallic needle perturbation. (b) Data collected for a dipole mode 2 cm
off axis. (c,d) Comparison of normalized Ez profiles, taken from the mea-
surements in (a,b). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.2 The measured (a) and simulated (b) Ra Q0 as a function of beam veloc-
ity for the fundamental passband. The figures illustrate the misalignment
of the measured data due to inaccurate displacement measurements, which
can lead to large disagreement with the simulation results for certain dis-
crete points. Graphs (c) and (d) compare two modes in the the T M110
pass-band. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

v
 LIST OF TABLES
4.1 Summary of measured and simulated frequencies and shunt impedances.
The listed Ra Q0 values are the maximum values in the range 0 40 β
 


0 52. The dipole modes were evaluated at a radius of 2 cm. . . . . . . . . . 15
4.2 Summary of measured and simulated coupling strengths. The energy ratio
is of the total energy to the energy in the end cell. The simulated 5-cell
Qext values were calculated by multiplying the simulated energy ratios by
the simulated Qext . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

vi
 CHAPTER 1

INTRODUCTION

1.1 Particle acceleration

The early use of particle accelerators in the 1930’s became important to the area of nuclear
research. Since then, particle accelerators have become important to a widespread area
of research and applications. These areas include research in elementary particle physics,
nuclear physics, synchotron radiation and medical applications.
Acceleration of particles can be acheived by several methods. The DC voltage of a

battery can be used as a simple method to increase the kinetic energy of a particle, where
the the kinetic ebergy gained is equal to the electrical energy lost by the battery. Crofton
and Walton developed the first mordern accelerator using this method, by charging then
disicharging capacitors to achieve a potential difference of 120,000 volts[1]. Other meth-

ods, such as the cyclotron, betatron, synchotron and linear accelerators utialize time varying
electric and magnetic fields in order to increase the particle’s kinetic energy.
The linac is composed of a line of resonent cavities which are intended to give the
particle beam energy, accelerating it as it travels down the line. The resonant cavities are

powered by RF sources, usually with frequencies in the range of 300-3,000 MHz. The
accelerating gap of the RF cavities, a cavity cell, must have the accelerating field pointing
in the same direction as the particle beam velocity. This is done by allowing a half of an
RF cycle to power the beam. The beam is accelrated with synchronous energy, meaning
each time the beam enters a new cell it is timed such that it sees the nominal accelerating

field. With the changing velocity of the particle beam as its energy increases, either the cell
length or the RF frequency changes in order to maintain synchronous operation. This calls
for different cavity designs for each stage of the linac[2].
The main objective of high energy accelrators is to impart a large amount of energy to

the particle beam by achieving a high accelerating field in the resonant cavities wtihout a

1
substantial amount of losses. Reduction of the losses is directly related to the reduction of
the surface resistance in the cavity walls. The use of superconductivity greatly reduces the
surface resistance of the cavities.

1.2 The Rare Isotope Accelerator

The rare isotope accelerator (RIA) concept is to utilize state-of-the-art superconducting

cavities to supply instense beam of exotic isotopes for research in nuclear physics[2]. The
linac provides the heavy ions acceleration of up to 400 MeV per nucleon using several
different types of superconducting structures. The design of RIA is significantly helped on

the experienced gained from the Spallation Nuetron Source (SNS). Part of the RIA linac
will use the same accelerating cavities and cryostats produced for the SNS project.

1.3 Design and evaluation of superconducting cavities

While the use of computer simulation greatly facilitates the design of radio frequency (RF)

cavities, it is important to confirm simulations with measurements. Much of the highe-

order mode (HOM) analysis for the 6-cell cavity was done with MAFIA [6]. In this thesis,
computer simulation results for a 5-cell axisymmetric copper cavity are compared with RF
measurements. The cavity is a prototype of a 6-cell superconducting cavity designed to

accelerate beams at 47% the speed of light. The 6-cell cavity is being developed for the
Rare Isotope Accelerator (RIA) [3, 4]. Additional MAFIA simulations were done for the
5-cell case for comparison with the measurements.
The field profile is measured via a “bead pull” perturbation method in order to calculate
Ra Q0 . The field profile is also used to evaluate the field flatness of the cavity.

Some important HOM properties that are obtained through computational means can
also be obtained through bench-top measurements. Quantities that are independent of the
cavity material are most desirable because a copper prototype may be used to confirm

2
simulated results in the design process. There are other important quantities for cavity
design, including peak electric and magnetic fields, but they are not readily measured.

1.4 Computer Simulations

MAFIA [6] and ANALYST [7] were used for numerical calculation of cavity eigenmodes.

Simulations where done on simple 5 and 6-cell (no power couplers or pick-up antennas
attached) axisymmetric geometries in 2D. Necessary 3D simulations, such as the design of
the power coupler, were done with a single-cell geometry. The simulation results facilitated
design the shape and location of the fundamental power coupler (FPC) and confirm that an

additional port is not required for a HOM damper.

1.5 Measurements

The measurements were done with a vector network analyzer (VNA) and a bead pull mech-
anism, as shown in Figure 1.1. Various beads were investigated, including both metallic

and dielectric elliptical shaped beads, as well as spherical beads and circular disks. The
metallic needle is most sensitve to the axial electric field[9]. Since the axial component
of the electric field is the component sought for measurement, the most accurate measure-
ments were acheived with the thin metallic needle, as shown in the inset of Figure 1.1. The

thin metallic needle is actually a hyperdermic needle cut to about 7 mm in length. Simple
wire antennas of different shapes and sizes were used to couple to different modes. The an-
tennas were mounted on the beam tube end-caps, on which holes were drilled at the center
(beam axis) and at various distances off axis, so the bead pull could be placed precisly at
the center or at a known distance and angle off axis.

The Qext is measured in transmission, using the same setup as in Figure 1.1. The FPC
antenna has sections of varying lengths, giving the antenna different penetration distances.
The input or driven antenna on the right end-cap is made close to match, but more impor-

3
Figure 1.1: Set-up for bead-pull measurements. The input antenna is mounted onto the
right end-cap. The power coupler on the left beam tube may be used as a pickup, although
an additional antenna on the left end-cap was used in some cases. The inset shows the
metallic needle and the fishing line (controlled by the motor on the right) without the end-
cap shown in the main picture. The Qext is measured with the same set-up, adjusting the
input antenna for unity coupling.

tantly the input antenna should not disrupt the cavity fields.

4
 CHAPTER 2

THEORY

2.1 Particle acceleration with an RF cavity

A time varying electromagnetic field is used in a linear accelerator to accelerate a particle

beam as the beam travels through the cavity. For simplicity, the particle bunch can be
represented by an ideal particle of charge q measured in coulombs. The acceleration is

achieved by a time rated change in momentum d p dt. The equation of motion for the
particle is given by

 q  E v  B
dp
(2.1)
dt

with the momentum p  

γm0 v. The Lorentz factor γ, and the rest mass m0 of the particle

is used to specify p.

Suppose this change in momentum is achieved along a path where a single component
of the electric field, let’s say the longitudinal component, is very high compared to other
components of the EM fields. The change in momentum pz is now related to the z com-

ponent of the electric field Ez in a cylindrical coordinate system and its charge q. The

particle feels a force along the z direction.

Fz
 qEz (2.2)

The energy gained by the particle can be determined from the line integral of this force
along the path the particle traverses the cavity. It is convenient to define a potential differ-
ence the particle sees due to the electric field, the cavity voltage Vc , by evaluating the line

integral of Ez along the path the particle travels through the cavity. This cavity voltage is
an important figure of merit of a superconducting cavity, and is described in Section 3.4.

5
 2.2 Pill-box modes

The cavity voltage defined in Equation (3.2) relies on the longitudinal, or axial, electric field
in the cavity. In order to visualize the EM fields in a cavity, it is relatively straight forward to
solve the boundary value problem based on Maxwell’s equations for an ideal pillbox cavity.

The various solutions to Maxwell’s equations are EM field distributions that generally have
different resonant frequencies. These solutions are called pill-box modes. Significantly,

two types of modes are derived, transverse electric T E and transverse magnetic T M . 
The different mathematical solutions are denoted by the integer subscripts mnp, referring

to the angular, radial and axial variations in the field distributions, e.g. a T M 110 mode
refers to a dipole (one angular variation per π 2 radians) transverse magnetic (no axial
component of the magnetic field) with one radial variation and no axial field variation.
The accelerating mode for cavities are usually the first resonant magnetic monopole,

the T M010 mode. This mode is called the fundamental mode because it is the first mode
in the frequency spectrum. This provides a good means of accelerating the particle as the
axial electric field is very high along center axis of the cavity, or the beam axis, and there
are no other EM field components along the beam axis.

2.3 Multi-cell elliptical axisymmetric cavities

The accelerating cavity requires the passage of the particle through the cavity. This is
done with cylindrical tubes placed along the radial center of the cavity as shown in Figure
2.1. The diameter of these beam tubes are chosen as to not allow the accelerating mode
to propagate, i.e. the cut-off frequency of the cylindrical beam tube must be above the

frequency of the RF source. The corners of the cavity are of concern when attempting to
achieve a high accelerating field for the particle. High surface electric fields near the exit
and entrance of the cavity can be dangerous because of their relationship to field emission.
High surface magnetic fields have the ability to quench the superconductor.

6
 High accelerating fields are desired along the beam axis of the cavity and high surface
fields need to be limited. This can be achieved by perturbing the geometry of the cavity so
the beam ports have rounded (elliptical) edges at the iris and along the equator.
Figure 2.1 shows a multi-cell cavity, with a coaxial power coupler. Aside from the

coupling ports, the cavity is axisymmetric about the beam axis. The modes in the elliptical
cells are classified by analogy to the modes in an ideal pillbox (T Mmnp and T Emnp where
the subscripts mnp refer to the azimuthal, radial and longitudinal variations). In general,
elliptical cavity modes are not necessarily pure transverse magnetic (TM) or transverse
electric (TE), although they are more like one or the other. One mode of a single cell cavity

splits into 6 pass-band modes in a six cell cavity. Additionally, multi-pole modes have
two different polarizations which split in frequency due to asymmetries. The longitudinal
electric field of the first six modes, or the T M010 passband, for the six-cell cavity have 6

variations. Note that the fundamental mode π 6 is no longer the accelerating mode; the

accelerating mode is the π-mode.

Figure 2.1: Six-cell axisymmetric cavity with power coupler port on the left and pick-up
antenna port on the right. The disks on the outside of the endcells attached to the beam
tubes are in place to hold the titanium helium vessel and cavity tuner. Stiffening rings are
also drawn in place; they are used to maintain the shape of the cavity.

7
 2.4 Perturbation by a metallic object

MAFIA can calculate the relevant mode properties directly from their theoretical defini-
tions. For measurement, manipulation of the theoretical definitions is needed in order to
relate the mode properties to the S-Parameters that can be measured with a network an-


alyzer. An important figure of merit, the geometric shunt impedance R a Q0 , which is
described in Section 3.5, can be measured with the help of perturbation theory.
Using Slater’s perturbation theory, it is possible to calculate the normalized field strength,
 
E0 and/or H0 , by introducing a small metallic perturbation and measuring the change in

resonant frequency. A metallic sphere may be used to determine E z along the beam axis
for all TM monopoles, because for these modes the other field components are very small.
For other modes, a metallic needle can be used in combination with a calibration con-
stant. This constant may be derived theoretically [8, 9] or determined experimentally. The

metallic needle changes the resonant frequency significantly due only to the presence of
an electric field, and the frequency change is greatest when the needle is oriented in the
direction of the electric field [9]. With a small change in resonant frequency, ∆ω ωo ,
Slater’s perturbation can be approximated for a sphere by:

∆ω
 
4πω0 a3 εE0
2 1
2

µH0
2
(2.3)


2 2 
where E0 and H0 are normalized by dividing by twice the cavity’s stored energy and a is
the radius of the sphere. The technique used herein is a simple method to quickly measure
the Ra Q0 of a particular mode and compare with MAFIA. Therefore, it is only desired to
measure the z-component of the electric field on or near the beam axis. A needle was used


with calibration through a T M010 π measurement with a sphere.

8
 Sphere.

For TM monopole modes, Ez is the only field component along the axis, which simplifies
Equation (2.3). The Ra Q0 for T M monopoles can be calculated numerically from

 
∆ω exp iω0 z cβ dz  2

 
Ra
Q0  4πε0 ω2
0a
3
(2.4)

allowing us to obtain Ra Q0 for a given cβ, or beam velocity, from the bead-pull measure-
ment of ∆ω as a function of z.

Needle.

A needle was also used, (see Figure 1.1) to evaluate Ez . Since multi-pole modes have no

axial electric field along the beam axis and vary radially, they are measured by moving off
axis in the direction corresponding to the strongest coupling: Ra Q0 can be determined for

any ρ, φ coordinate. This is done by coupling to the mode with an off-axis antenna and

doing the bead-pull 2 cm off axis and 180 in rotation from the antenna, where a strong

enough signal for a measurement can be found. The two polarizations of the HOMs are
measured by rotating the antenna and bead until that the other polarization is coupled (at a
different frequency). With a needle of volume πa2 l,

 
∆ω exp iω0 z cβ dz  2

 
Ra
Q0  k   4πε0 ω2 2
0a l
(2.5)

The proportionality constant k in Equation (2.5) may be determined experimentally from

either a known cavity, such as a pillbox, or from a monopole mode measurement. The

Ra Q0 results in this paper were obtained from needle measurements, using a proportion-


ality constant obtained from the T M010 π mode calibrated with the sphere standard.

9
 CHAPTER 3

FIGURES OF MERIT

3.1 Some important quantities in superconducting cavities

The beam stability can be predicted from the HOM frequency ( f ), geometric shunt impedance
(Ra Q0 ) and coupling strength (Qext ), which are explained in this chapter.The quality fac-
tor (Q0 ) is also important for the system performance. Because the RF measurements are
done on a room temperature copper model of the cavity, its Q 0 is much lower than an actual
superconducting niobium cavity. Therefore, a room temperature measurement of Q 0 is not

very useful. All of the other quantities were calculated numerically and measured.

3.2 Higher Order Modes

The higher order modes of the cavity is the term used to describe cavity modes that may be

excited by the accelerated beam and can become harmful to the beam stability. These are
all the modes that are not used to accelerate the beam. In order to determine if the HOM is
harmful, an HOM analysis must be done. This includes determining the frequencies of the
HOMs and evaluating what type of damage the mode can do to the beam. If the HOM is
determined to be harmful, the mode must be damped in order to design a functional cavity.

The RIA 6-cell β

 0
47 cavity has been designed without a port used specifically for
an HOM damper. This is because a HOM analysis was done and very few modes were
deemed to be problematic, and those that were will be damped by both the pick-up antenna
and the fundamental power coupler. The HOM analysis was done via computer simulation

to show this[3].

10
 3.3 Field Flatness

The accelerating mode should have equal energy in each of the cells of the cavity, giving
the beam an equal kick from each cell. This gives the cavity a greater accelerating voltage
while minimizing peak surface fields. The field flatness of the cavity is a measure of how

evenly the the energy is distributed in the cavity for the accelerating mode. The field flatness
can be described by
max ∆ Ez  peak



Σ Ez
peakn
 N
(3.1)

The field flatness, determined from Equation (3.1), is often checked with a bead pull mech-
anism as in Figure 1.1. Since the square of the electric field is proportional to the change in
the resonant frequency, a bead pull measurement using a thin metallic needle is suitable to


record the square of the longitudinal electric field in each cell. The ∆ E z
peak
 is the greatest
 
change in the maximum Ez between any two cells, and the Σ Ez  peakn
 N is the average
 
maximum Ez across all the cells.

3.4 Geometric Shunt Impedance

The shunt impedance and quality factor of the superconducting cavity are two important

figures of merit. Both of these depend on the surface resistance of the cavity walls. Dividing
the shunt impedance by the quality factor gives a useful material-independent quantity
(Ra Q0 ). The Ra Q0 of a cavity mode is an integral of the longitudinal electric field
(Ez ). For TM monopole modes, Ez exists along the beam axis, where the other electric and

magnetic field components are very small. For all other modes, E z goes to zero along the
beam axis. Since beam instabilities [5] can occur due to any mode with non-zero E z on
or near the axis, it is important to calculate Ra Q0 for modes other than the accelerating
mode.

The quality factor of the cavity is the ratio of the stored energy in the cavity times the
angular frequency to the power dissipated in the cavity walls. The power dissipated, Pc ,

11
is dependent on the surface resistance, Rs , of the cavity walls. The shunt impedance is
determined from the average voltage seen by the beam, Vc , and the dissipated power, Pc .
The beam voltage is calculated from the longitudinal electric field along the beam axis
a 0 , for monopole modes, or near the beam axis a   0 for multi-pole modes:


  a  z exp  
d iω0 z
Vc
  0
Ez ρ
cβ
dz (3.2)

 
where ωo is the resonant angular frequency of the mode, cβ is the velocity of the beam, d
is the total length of the cavity, and ρ is the radial cylindrical coordinate. Thus, Vc is the
integral of the longitudinal force on a particle traveling at the speed cβ, accounting for the
time and space dependence in the field. The particle travels on the crest of the wave by
making the cell length 21 βλ, thus the particle passes through the cell on 1 2 of an RF cycle.

The shunt impedance (linac definition) is given by

Ra

Vc2
Pc

(3.3)

3.5 Cavity Coupling Strength

The coupling strength of the power coupler is determined from the external quality factor

(Qext ). The Qext of the accelerating mode must be small enough to accommodate the beam
loading of the accelerating mode; the Qext of the HOMs determine the rate of damping of
beam induced fields in the HOMs and hence the beam stability. The Q ext calculations for
multi-cell cavities can be difficult because simulations require large mesh sizes due the to

larger structure and the asymmetric geometry. However, a 3D code can be used to calculate
the single cell Qext for each mode. Multi-cell modes can be more easily calculated with
2D meshes; the 2D calculation give an energy ratio (the ratio of the total stored energy to
the energy in the cell adjacent to the coupler). Multiplying the Qext of the single cell cavity
mode by the energy ratio for each multi-cell mode in the multi-cell pass-band provides an

12
adequate approximation of the multi-cell Qext values. If all of the cells have the same
energy, the 6-cell Qext is 6 times the single cell Qext . This method requires a knowledge
of which single cell mode corresponds to each multi-cell passband. This can be determined
by viewing graphical simulation results, but it is not always straightforward.

The external quality factor, Qext , is a ratio of the stored energy times ωo to the power
emitted out of the power coupler. If other losses are assumed to be small, the overall
Q of the cavity, the loaded Q, (QL ), can be expressed in terms of Q0 and Qext only; it
is measured with the VNA from the bandwidth of the resonance. Typically, operational
values of Qext are very large for superconducting cavities, i.e. the coupling is weak. The

Qext can be determined from QL via a measurement of the transmission coefficient (S21 ).
With unity coupling (matched load),

Qext
  
2 QL S21 2 (3.4)

If the input antenna does not have unity coupling, a correction factor may be determined
from the reflection coefficient (S11 ). It is preferred to be near unity coupling in order to

minimize systematic errors in the measurement.

3.6 Peak surface fields

The average accelerating field, Eacc , is the field the beam sees during its passage through
the cavity. Using Equation (3.2) to determine Vc along the trajectory of the beam, Eacc is

the ratio of the cavity volatage to the length of the cavity. In order to minimize undesireable
effects from high surface fields, the ratios of peak surface fields to the accelerating field
should be minimized. The peak electric field is determined from computer simulation;
these peak fields are near the iris of the ellipical cavities and are not easily measured. For

the peak magnetic field, the peak field is near the equator of the elliptical cavity.

13
 CHAPTER 4

SIMULATIONS AND MEASUREMENTS OF A 5-CELL

PROTOTYPE

4.1 Field profiles and Ra  Q0

MAFIA was used to calculate the Ez profiles on or near the beam axis for all relevant
HOMs [3]. Figure 3 compares two MAFIA profiles with the measured profiles. The mode

classification is determined from the field distribution, a step which is also required to
predict the multi-cell Qext , as explained in Section 1.1.
Figures 3a and 3b show the raw data collected: the S21 phase shift (∆φ) as a function of
bead position. The bead position is determined by measuring the distance and time of the

bead travel. The S21 phase shift is proportional to the change in resonant frequency due to
the perturbation. The frequency change is calculated from the phase shift by interpolating
a measurement of phase as a function of frequency for the unperturbed cavity mode. Using
a conducting needle (radius = 0.25mm, length = 6.34 mm), the maximum ∆φ was about 7 

for the T M010 π mode (Figure 3a), which corresponds to 3.00 kHz change in frequency.


For the T M010 π 5 mode (Figure 3b), the conducting needle was placed off-axis 2 cm, and


the maximum phase change was only about 1 2 , corresponding to a ∆ f max of 574 Hz.
Figures 3c and 3d show a comparison of MAFIA to the measurements. Prior measurements
were done with the conducting sphere, and the calibration constant k was determined to

be 1.15 from the monopole measurement. Data manipulation is needed to account for a
change in the direction of the Ez field (e.g. the field changes direction four times in Figure
3c) and then Ra Q0 can be numerically calculated from Equation (2.5) with the calibration
constant. The results are shown in Table 1.

The cavity is designed to accelerate the beam from β

 0
40 to β  0
52, where βc is
the beam velocity. Hence, the dependence of Ra Q0 on β is of interest. Figure 4.2a shows
the dependence of the measured Ra Q0 on β, as obtained via Equation (2.4). MAFIA

14
 0 0

∆φ(degrees)

∆φ(degrees)
−2
−0.5

−4

−1
−6
TM010,π TM110,π/
a. z b. 5
z
−8 −1.5
0 10 20 30 40 cm 0 10 20 30 40 cm

Ez Ez
1 1

0 0

−1 −1 MAFIA
c. z d. MEASURED
0 10 20 30 40cm 0 10 20 30 40 cm

Figure 4.1: (a) Raw data collected for the accelerating mode along the beam axis from a
metallic needle perturbation. (b) Data collected for a dipole mode 2 cm off axis. (c,d)
Comparison of normalized Ez profiles, taken from the measurements in (a,b).

Table 4.1: Summary of measured and simulated frequencies and shunt impedances. The
listed Ra Q0 values are the maximum values in the range 0 40 β 0 52. The dipole
 

modes were evaluated at a radius of 2 cm.
Mode Frequency (MHz) Ra Qo (Ω)
Simulated Measured Simulated Measured
T M010 π 5 794.0 794.4 .0652 .0431
T M010  2π 5 797.0 797.7 .268 .515
T M010  3π 5 800.6 801.5 3.94 3.28
T M010  4π 5 803.5 804.7 37.8 25.6
T M010 π 804.7 805.8 162.2 156.4
T M110 π 1133 1132 6.744 3.984
T M110  4π 5 1139 1140 13.095 26.47
T M110  3π 5 1147 1149 7.130 5.202
T M110  2π 5 1156 1158 3.534 5.577
T M110 π 5 1161 1164 .8562 1.542

15
 d c.

Figure 4.2: The measured (a) and simulated (b) Ra Q0 as a function of beam velocity for
the fundamental passband. The figures illustrate the misalignment of the measured data
due to inaccurate displacement measurements, which can lead to large disagreement with
the simulation results for certain discrete points. Graphs (c) and (d) compare two modes in
the the T M110 pass-band.

16
was also used to calculate the dependence on β via Equations (3.2) and (3.3) and the re-
sults are shown in Figure 4.2b. The maximum Ra Q0 was used when evaluating beam
stability issues [3]. Because the locations of the zeros rely heavily on exact bead displace-
ment measurements, measuring the maximum Ra Q0 over the range of interest gives better

agreement without a complicated set-up.

4.2 Coupling measurements

The measurement of Qext is done on the accelerating mode in order to design the input
coupler. The Qext can be calculated from Equation (3.4) and measurements of Q L and the

S-Parameters. The VNA is connected to the power coupler and an antenna in the beam
tube. The antenna is constructed such that it is close to a matched load.
Calculation of Qext for a single cell cavity was done in [3] using ANALYST. MAFIA
could also be used to compute Qext using the procedure in [10]. The multi-cell Qext was

computed from 2D MAFIA simulations. The predicted Qext for some HOMs are presented
in Table 4.2.

17
Table 4.2: Summary of measured and simulated coupling strengths. The energy ratio is
of the total energy to the energy in the end cell. The simulated 5-cell Q ext values were
calculated by multiplying the simulated energy ratios by the simulated Q ext .
Frequency (MHz) Ratio 
Qext 5 cell
Mode Simulated Measured Simulated Simulated Measured

T M010 single 805.0 799.7 –

2 170  106

2 820  106

T M010 π 5 794.0 794.4 23.75

5 154  107

3 116  107

T M010 2π 5 797.0 797.7 6.817

1 479  107

1 040  107

T M010 3π 5 800.6 801.5 3.744

8 124  106

6 266  106

T M010 4π 5 803.5 804.7 2.839

6 161  106

4 480  106

T M010 π 804.7 805.8 5.319

1 115  107

6 380  106

T M020 single 1725 1725 –

3 410  105

4 420  105

T M020 π 5 1702 1703 10.93

3 727  106

2 773  106

T M020 2π 5 1712 1713 3.828

1 305  106

3 816  105

T M020 3π 5 1726 1728 3.055

1 042  106

2 805  105

T M020 4π 5 1741 1743 4.280

1 459  106

3 579  105

T M020 π 1752 1754 13.17

4 491  106

1 250  106

T M110 1cell 1154.70 1149.5 –

9 580  105

1 090  106

18
 CHAPTER 5

DISCUSSION

5.1 Analysis of error

The measured results in Tables 1 and 2 provide good agreement with the simulated results
for the fundamental pass-band. However, there are significant discrepancies in R a Q0 at
certain velocities. The Qext measurements are in good agreement for the multi-cell pass-
band modes and single cell modes, while HOM dipoles do have a discrepancy.
The Ra Q0 measurements were all done with a calibrated needle, although measure-

ments with a conducting sphere should provide accurate results for monopole modes.
The measured Qext values were generally lower than the simulated values. The energy
distribution could be affected by the input coupler. This could, in effect, lower the expected
energy ratio as simulated in the ideal case with MAFIA. The measured profiles of the HOMs

evaluated without the penetrating input coupler did have good agreement, corresponding to
an energy ratio agreement in the vicinity of  10%. The energy ratios in the presence of the
penetrating input coupler were not evaluated.
Several modes were simulated for the 6-cell cavity [3]. All the monopole, dipole,

quadrupole, and sextupole modes below their cut-off frequencies (both TM and TE) were
simulated using an axisymmetric geometry. The Ra Q0 , energy ratios, single-cell Qext
and frequencies were all considered to identify possible problematic modes.
Simple RF measurements were done on a copper model of a superconducting cavity at
room temperature to confirm some key RF properties. Though the results presented in this

thesis provide reasonable agreement with the simulation results, more accurate measure-
ments are possible, although this would require additional equipment and more complicated
set-ups.
The measurements of the field profile could be improved with an automated system,

such a PC connected to the VNA and an electronic displacement sensor, simultaneously

19
taking measurements. Additionally, the path of the bead should be parallel to the beam
axis. The bead can deviate from this path if the guides on the ends of the cavity are not
properly aligned.
The computer simulations provided accurate results and were a great facilitator in the

development of the 6-cell cavity, and the RF measurements provided a good confirmation
of the simulations.

20
 CHAPTER 6

CONCLUSION AND FUTURE WORK

6.1 Conclusion

This work described in this thesis was required to develop an elliptical cavity that is to be
part of the RIA linac where the beam travels around half the speed of light. The computer
simulation of the of the cavity indicated the cavity design was acceptable and should be
used in the RIA linac. RF measurements confirmed the results of the computer simulation.
Currently, two power couplers have been designed, constructed and tested for niobium

prototype cavities. Soon to be assembled is test cryo-module that will hold the two cavities
as if they were a part of the linac, and tested under realistic high-power conditions.

6.2 Future work

My future plans are to utilize the experience gained while doing this research and design,
build and test a new concept superconducting cavity. The fundamentals learned established
a good background in the field of cavity design that will be an invaluable tool in accelerator
physics.

21
 BIBLIOGRAPHY
[1] Edwards DA and Syphers M J 1993 An Introdutionto the Physics of High Energy
Accelerators (New York: JohnWiley and Sons) p 19

[2] Shepard K W, Delayen J R, Lyneis C M, Nole J, Ostroumov P, Staples J W, Brawly J,

Hovater C, Kedzie M, Kelly M P, Mammasser J, Piller C, Portillo M 1999 SC Driver
Linac for a Rare Isotope Facility in Proccedings of the 9th Workshop on RF supercon-
ductivity, Report LA-13782-C, LANL, Los Alamos, New Mexico p 345-351

[3] Grimm T, Hartung W, Marti F, Podlech H, York R C, Popielarski J, Wiess C, Kem-

pel L, Ciovati G, Kneisel P 2002 Input Coupling and Higher-Order Mode Analysis of
Superconducting Cavities for the Rare Isotope Accelerator in Proceedings of the Eight
European Particle Accelerator Conference (Geneva: EPS-1GA Publishing) p. 2241

[4] Compton C C, Grimm T L, Hartung W, Podlech H, York R C, Ciovati G, Kneisel P,

Barni D, Pagani C, Pierini P 2001 Niobium Development for the High-Energy Linac of
the Rare Isotope Accelerator in Proceedings of the 2001 Particle Accelerator Conference
(Piscataway, New Jersey: IEEE Publishing) p. 1044

[5] Padamsee H, Knobloch J and Hays T 1998 RF Superconductivity for Accelerators

(New York: John Wiley & Sons, Inc.) p 154

[6] Mafia 4, Computer Simulation Technology, Darmstadt, Germany (http://www.cst.de)

[7] Analyst, Simulation Technology & Applied Research, Inc., Wisconsin, USA
(http://www.staarinc.com)

[8] Maier L C and Slater J C Field Strength Measurements in Resonant Cavities 1952,
Journal of Applied Physics, Vol 23, p 68-77

[9] Ginzton E L 1957 Microwave Measurements (New York: McGraw-Hill) p 446

[10] Balleyguier P 1998 External Q Studies for APT SC-Cavity Couplers in Proceedings
of the XIX International Linac Conference (Argonne, Illinois: ANL-98128) p. 133

22