
Development of an X-band Photoinjector at SLAC*

A. E. Vlieks, G. Caryotakis, R. Loewen, D. Martin, A. Menegat
SLAC, 2575 Sand Hill Rd, Menlo Park, CA 94025, USA
E. Landahl, C. DeStefano, B. Pelletier, and N.C. Luhmann, Jr.
3001 Engineering III, Dept. of Applied-Science
Davis, CA 95616, USA

Abstract
As part of a National Cancer Institute contract to develop Solenoid
a compact source of monoenergetic X-rays via Compton Laser Mirror
backscattering, we have completed the design and Chamber
construction of a 5.5 cell Photoinjector operating at
11.424 GHz. Successful completion of this project will Interaction Chamber
result in the capability of generating a monoenergetic X- Quadrupoles
ray beam, continuously tunable from 20 - 85 KeV. The
immediate goal is the development of a Photoinjector
producing 7 MeV, 0.5nC, sub-picosecond electron
bunches with normalized RMS emittances of >.1 pi-mm-
mR at repetition rates up to 60 Hz.. This beam will then RF Gun
be further accelerated to 60 MeV using a 1.05 m
accelerating structure. This Photoinjector is somewhat
different than the traditional 1.5 cell design both because Accelerator
of the number of cells and the symmetrically fed input
Figure 1. Photoinjector Layout
coupler cell. Its operating frequency is also unique. Since
the cathode is non-removable, cold -test tuning was This RF gun is unique in its design because of the
somewhat more difficult than in other designs. We will number of cells used and its high operating surface
present results of "bead-drop" measurements used in field gradients. The basic gun parameters are listed in
tuning this structure. table 1. The RF power for the RF gun and accelerator
Initial beam measurements are currently in progress and will be derived from two 60 MW X-band Klystrons
results will be presented as well as results of RF driven by a common frequency generator. This will
conditioning to high gradients at X-band. Details of the permit independent phase and amplitude control of
RF system, emittance-compensating solenoid, and both the accelerator and Gun RF.
cathode laser system as well as PARMELA simulations
will also be presented. Number of Cells 5.5
Peak Surface Gradient/Power 200 MV/m @ 16 MW
1. Introduction RF Filling Time 65 ns
We have recently completed the design and construction Cathode Material Copper
of the main components of an X-band photoinjector and RF Pulse length 200 ns
are currently in the initial stage of testing. In this paper we
will describe the design and initial testing of the RF gun. Table 1. RF Gun Parameters
The essential components of the Photoinjector are shown
in Figure 1. The 0.6 T Emittance compensation Solenoid 2. Simulations
consists of a pair of identical coils positioned axially with
respect to each other. The coils are oppositely wound so The initial gun design was obtained using
that the field is zero in the center of the solenoid. The RF SUPERFISH[1]. It was used to identify the 6 modes
gun is positioned in the Solenoid with the cathode at this and establish the pi-mode at the correct frequency.
field null. A Laser Mirror chamber permits a UV laser (11.424 GHz.) It was also used to determine the Q-
beam to strike the cathode along a nearly axial path. The value, the R/Q and the frequency separation of the pi-
generated electron beam is further accelerated by a 1.05 m mode from its nearest neighbor. HFSS[2] and
accelerating structure. At the exit of the accelerator three MAFIA[3] were used to design the symmetrically fed
quadrupoles are positioned to focus the electrons to a coupler cell and establish the desired external Q of
narrow spot size in the Interaction chamber. A the gun. PARMELA[4] was used to optimize the
Spectrometer, not shown, is used to divert the beam away beam focusing and emittance as well as establish the
from its axial path into a beam dump. approximate location of the accelerator, quadrupoles

*Work supported by Department of Energy contract DE-A C-76SF00515
and National Cancer Institute Contract Number N01-CO-97113


and interaction chamber. The required design strength of Tuning posts were incorporated into each cell (4 for
the solenoid and quadrupoles were also determined using the first 5 cells and 2 for the coupler cell).
PARMELA. Figure 2 shows the electron beam profile as The individual cell frequencies were measured but no
it is accelerated through the gun and a .75 m accelerator attempt was made to tune the individual cells since
structure and finally focused at the interaction point. This they were already within 2 MHz. of their design
particular simulation was performed assuming a well values. The cells were diffusion bonded and the
formed "beer barrel" shaped laser spot at the cathode. The coupler cell irises were machined to its design value.
resulting beam spot size at the interaction point was The peak field flatness was measured using a "bead
approximately 30 microns with a normalized emittance of drop" scheme. In this method a small metal cylinder
0.75 pi-mm-mrad. was lowered into the gun using a precision stepper
motor. The frequency shift was measured by a
HP8510 network analyzer. Operation of the stepper
motor and sampling of the Network Analyzer was
performed using a LabVIEW program. The effect of
the mass of the bead and the diameter of the
supporting string were investigated to find a
combination, which gave reasonable and repeatable
results. The final choice of string was .02mm nylon
suture thread. A comparison of "bead-pull" and "bead-
drop" measurements was made on the cold-test
prototype (which had an axial hole in the cathode).
This verified that the two measurement techniques
gave the same results. Figure 3 shows the final field
profile after tuning and Figure 4 shows the total
measured frequency spectrum

0.6




0.5




0.4

0.3
Figure 2. Beam profile along Injector axis.

For a Gaussian-shaped laser spot, the electron beam 0.2
diameter and emittance increases to 40 microns and 1.1
pi-mm-mrad. For both cases, the laser spot at the cathode 0.1
had transverse dimensions of 1 mm and a temporal
(longitudinal) dimension of 800 fs. 0
0 100 200 300 400 500 600 700
Position (arb. units) aev-3/29/02
3. Cell Tuning Figure 3. Final post-braze Field profile, measured
The tuning of the final RF gun proceeded through several
using "bead-drop" technique
steps. Initially a prototype was built using the dimensions
from the simulations. The 2b dimension was deliberately 4. Gun Preparation
undersized to permit frequency tuning by removing Since the operating peak surface gradient of the RF
material. This was done on a cell-to-cell basis using Gun is rather high, 200MV/m, careful attention to
pretuned end caps. The prototype was then assembled and surface cleanliness is necessary. After cold test
the coupler cell (cell 6) was retuned to obtain the design measurements were completed, the gun was vacuum
external Q value and frequency. The changes required in fired at 750 C. for 8 hours to remove hydrogen and
the final 2b dimension were small enough that an etching surface contamination. An RF window assembly was
procedure was used. A bead-pull measurement was then then connected to the input waveguide, a vac-ion
used to verify that the peak fields in each cell were near pump was installed at one port of a Tee at the Gun
identical. The cells were then diffusion bonded and output beamline and a Gate valve was connected to
remeasured. The only change due to bonding was to the other port. See Figure 1. The total assembly was
lower the pi mode frequency by 2 MHz. The resulting
cavity dimensions were used for the actual RF gun cells.


1 that the initial and final spikes are actually double
peaked. This is due to the shock excitation of the
1 1 . 2 8 6 5
neighboring resonance of the pi mode. This was
0 . 8
verified by performing a Fourier transformation to
the frequency spectrum of Figure 4 with a realistic
0 . 6 input pulse shape. The resulting time domain pulse
1 1 . 4 1 5 0
shape is virtually identical to Figure 6. (in the
1 1 . 3 8 4 2
0 . 4 absence of the forward power excursion at the 200 ns
1 1 . 3 4 6 8 1 1 . 4 2 4 4 region).

0 . 2

1 1 . 3 0 9 5 0 . 1 2



0 0.1
1 1 . 2 8 1 1 . 3 2 1 1 . 3 6 1 1 . 4 1 1 . 4 4

F r e q u e n c y ( G H z )
a e v - 3 / 2 8 / 0 2
0 . 0 8
Figure 4. Frequency profile of RF Gun
0 . 0 6
then baked out in a vacuum oven at 130 degrees for 24
hours to remove water vapor. While under vacuum, the 0 . 0 4
valve was closed so that the assembly could be installed
in the Solenoid magnet while remaining under vacuum. 0 . 0 2
In this way, the gun was high power conditioned without
breaking vacuum. 0
- 4 0 0 - 2 0 0 0 2 0 0 4 0 0 6 0 0 8 0 0
t i m e ( n s )
5. High Power Conditioning
Figure 5. Gun incident RF Power
After briefly running at 5 MW with a wide, 400ns, RF
pulse to study the pulse shapes, the gun was run up to 24
0.006
MW with a 75-80 ns pulse. At this power level it was run
for 46 minutes without breakdown events. The peak
surface gradient at this power level was 168 MV/m. The 0.005
power was then lowered and the pulse width widened to
100 ns. The power was then increased to 20 MW or 175 0.004
MV/m. Again lowering the power, the pulse width was
widened to 150 ns. The power was gradually raised to 16- 0.003
16.5 MW or 180 MV/m. Finally the pulse width was
widened to its design value of 200 ns and run up to 15 0.002
MW. This corresponds to 185MV/m. Little faulting was
observed Final processing will continue at a later date. 0.001
Typical pulse shapes of the incident and reflected RF
signals are shown in Figures 5 and 6. As can be seen 0
- 4 0 0 -200 0 200 400 600 800
from Figure 5, the pulse is flat for the first 200 ns and Time (ns) aev-6/602
then suffers from significant amplitude variations. This is

caused by the reflected signal returning to the Klystron, Figure 6. Reflected RF Power from Gun

and disturbing the subsequent Klystron output power. The 6. References
round trip distance to the Klystron is almost exactly 200 [1] James Billen and Lloyd M. Young, POISSON,
ns. For our purposes this will not present a problem since SUPERFISH reference manual, LA-UR-96-1834.
we require only a 200 ns pulse. For added flexibility [2] Agilent, High-Frequency Structure Simulator v5.6
however, additional waveguide length will be added in [3] R.Klatt, F.Krawczyk, W.R.Novender, C.Palm,
the near future. T.Weiland, "MAFIA- A 3-D Electromagnetic CAD system
In Figure 6, the typical general form of a reflected RF for Magnet, RF Structures and Transient Wake-Field
signal from a standing wave cavity can be seen. The Calculations", Reports at the 1986 Linear Accelerator
initial and final spikes are due to the filling and emptying Conf, Stanford, USA 6/2-6/1986
of the cavity while the intermediate region shows the [4] Lloyd Young, PARMELA reference manual, LA-
UR96-1835, January 8, 2000,. The version of PARMELA
approach to equilibrium of an over-coupled cavity. used in this work is a modified version due to E. Colby, at
The spike in the middle is due to the perturbation caused SLAC
by the forward power pulse shape. It is interesting to note



