Design of the Source Development Lab bunch compressor

Abstract

The accelerator at the Source Development Lab at BNL consists of a
1.6 cell RF photocathode electron gun followed by a 230 MeV SLAC-type
linac that includes a magnetic chicane bunch compressor. The nominal
specifications call for a 10 ps FWHM bunch of 2nC charge to be
compressed in time by a factor of 25 at an energy of 85 MeV. The design
of the compressor magnets and the beam dynamics from the gun through the
magnetic chicane are described

Full text

Page 1

DESIGN OF THE SOURCE DEVELOPMENT LAB BUNCH
COMPRESSOR
W.S. Graves, I. Ben-Zvi, E.D. Johnson, S. Krinsky, J. Skaritka, M.H. Woodle,
L.-H. Yu, National Synchrotron Light Source, Brookhaven National Lab, Upton, NY 11973
T.O Raubenheimer Stanford Linear Accelerator Center, Stanford, CA
Abstract
The accelerator at the Source Development Lab at BNL
consists of a 1.6 cell RF photocathode electron gun fol-
lowed by a 230 MeV SLAC-type linac that includesa mag-
netic chicane bunch compressor. The nominal specifica-
tions call for a 10 ps FWHM bunch of 2nC charge to be
compressed in time by a factor of 25 at an energy of 85
MeV. The design of the compressor magnets and the beam
dynamics from the gun through the magnetic chicane are
described.
1 INTRODUCTION
Next generation light sources and linear colliders will op-
erate with electron beams that occupy very small 6D phase
space volumes. The Source DevelopmentLab (SDL) is de-
signed to take advantage of the latest methods for produc-
ing these high brightnessbeams, includingthe use of short-
pulse lasers to tailor electron beam pulse shapes produced
by RF photocathode guns, the latest generation BNL-type
photocathode, and magnetic pulse compression to squeeze
the beam longitudinally to a very short bunch length and
high peak current. At the NSLS the primary use for this
beam will be the production of short wavelength light
through the FEL interaction and by Thomson scattering of
an infrared laser. Our FEL uses the NISUS wiggler [1]
as part of a high-gain amplifier which can be operated in
several configurations which provide tunable radiation at
wavelengths from 1µm to below 100 nm. In combination
with the gun laser, Thomsonscattering fromthe linac beam
can provide tunable x-rays from 100 to 300 keV.
This paper briefly reviews the accelerator, then focuses
on the design of the magnetic bunch compressor and the
beam dynamics through the compressor.
2ACCELERATOR COMPONENTS
The electron source [2] is a radio-frequency photocath-
ode developed by a collaboration from BNL, SLAC, and
UCLA. It consists of a 1.6 cell RF structure driven at 2856
MHz. The maximum gradient is 140 MV/m, yielding an
exit energy of approximately 8 MeV.
To produce electrons in the RF gun, the photocathode
is illuminated with 266 nm light generated by frequency
tripling the fundamental from a Ti:Sapphire laser. Up to
0.4 mJ of UV light is available with a tunable pulse length
ranging from 60 fs to 20 ps which should be more than suf-
ficient to produce several nancoulombs of charge from a
copper cathode. A square transverse intensity profile with
a 65 degree wavefront tilt is used to match the beam onto
the RF photocathode. The square intensity profile is opti-
mal for emittance correction. The wide bandwidth of the
Ti:Sapp laser allows for longitudinal pulse shaping so that
nonlinear emittance correction may be investigated.
Thelinac currentlyconsistsof fourSLAC-typeconstant-
gradient linac tanks operatingat 2856 MHz, with provision
for installation of a fifth section. The first two linac tanks
are used to accelerate the beam to approximately 84 MeV,
where it enters the magnetic chicane. The final two linac
tanks following the chicane accelerate the beam to a maxi-
mum energy of 230 MeV.
3CHICANE DESIGN
Figure 1: Magnetic chicane.
The SDL bunch compressor is made up of three major
components: the dipole mangets, the vacuum vessel sys-
tem with the beam diagnostics, and the mechanical support
system. The 4 dipoles are fabricated from solid blocks of
1008 low carbon steel. Steel dowel pins at the magnet mid-
plane accurately locate the yoke blocks. Each magnet uses
eight air-cooled pancake coils of 16 turns. Current density
is held to less than 1.5A/mm2to avoid thermal distortion.
The coils also extend away from the iron to allow natural
convection cooling. The field strength is 4.5 kG, magnetic
length is 19 cm, and the gap is 3 cm. The magnetic design
was optimized with the three dimensional code TOSCA.
The aluminum vacuum chamber is machined in halves
and seam-welded at its periphery. Two CCD cameras are
used as electron beam and synchrotron radiation diagnos-
1197
0-7803-4376-X/98/$10.00  1998 IEEE

Page 2

tics. The electron beamprobeis injectedthroughthe center
of the vacuum chamber. It has a YAG crystal[3] and mir-
ror to observe beam shape. It also has 3 slots 0.3, 0.6, and
1.0 mm wide, respectively, so that slice emittance may be
studied. The slots may also be used to select a fixed en-
ergy bandwidth for transmission to the wiggler. The local
emittance and local energy spread are key determinants of
single-pass FEL performance. The probe may be moved
to any tranverse position in the chamber by a stepper mo-
tor. The maximum displacement of the beam is 14 cm at
84 MeV. The compressor may be operated at any value
from zero field to full strength. The synchrotron radia-
tion monitorisused toobserveemission asthe beampasses
through the final bend magnets. A variable position mirror
is needed to accommodate potential magnet repositioning.
The CCD mountis also moveableto accomodatevariations
in the direction of radiation emission.
A six strut system was selected to support the magnet
girder. The struts simplify survey and alignment and take
the place of a separate stand. The struts are bolted to a steel
ground plate which is bolted and grouted to the floor.
Drift distances are 38 cm between outer magnets and
25 cm between the central magnets. For a given com-
pressor length it is desirable to maximize these outer drifts
(from magnets 1 to 2, and 3 to 4) in order to minimize
the emission of coherent synchrotron radiation (CSR). The
separated magnet configuration also reduces coupling be-
tween adjacent magnets. The outer magnets may be placed
in either of two positions; the normal operating position is
shown in Figure 1, the other position has them placed ad-
jacent to the inner magnets. The latter position generates a
greater portion of the path-length difference in the bends,
and will be used to investigate the effects of coherent syn-
chrotron emission on beam emittance and energy distribu-
tion.
4BUNCH COMPRESSION DYNAMICS
There are several methods of producing short electron
bunches, but they are not all applicable to the high charge
(≥ 1 nC) regime. The electron bunch length immediately
adjacent to the cathode surface is approximately equal to
the input laser pulse length. It is possible to produce short
bunches in the RF gun through two mechanisms.
Ti:Sapp lasercan produce100fs pulses directly,or a longer
laser pulse may be timed at an advanced RF phase so that
velocity bunching [4] occurs in the first 1/2 cell of the gun.
Bothofthesemethodsareapplicabletosmallbeamcharges
(? 1 nC) only. At higher charges the low energy and high
charge density in the gun cause both bunch-lengthening
and emittance growth due to space charge forces. This pa-
per describes bunch compression via magnetic rotation at
higher energies, where space charge forces are small.
The bunch compression, ∆L, due to energy spread is
The
∆L = R56
∆E
E
+ T566
?∆E
E
?2
(1)
whereR56=?ds η/ρ, η isthe dispersion,ρ is thebending
radius inside the compressor magnets, ∆E/E is the corre-
latedspreadinbeamenergyproducedbyrunningthebunch
off-crest in the linac, and T566is the appropriatesecond or-
dermatrixelement. For the chicaneconfigurationwith four
equal-length magnets, the dominant R56term can be writ-
ten [5]
R56= 2θ2(Ldft+2
3Lmag)
(2)
where θ is the bend angle in each magnet, Ldftis the drift
distance between magnets, and Lmagis the length of each
magnet. The drift distance from magnet two to three does
not affect the compression, and may be made larger or
smaller to accomodate diagnostics or beamline length con-
straints. Equation 2 shows that the drifts contribute more
to ∆L than do the bends. For a fixed chicane length, CSR
will be reduced by maximizing the drifts.
Nonlinear components of the longitudinal bunch distri-
bution are produced by curvature of the RF waveform and
longitudinal wakefields generated in the linac. These non-
linearites may be partially corrected [6] by the T566ele-
ment of the magnetic field. For our design T566 = −7.4
cm and ∆E/E = 1% yielding a second order path length
difference of just 7.4µm, too little to correct aberrations
in the distribution. Investigations are underway to study
longitudinal pulse shaping to balance the curvature due to
wakefields and RF.
Conservation of the longitudinal emittance requires that
the uncorrelated energy spread of the bunch grow in in-
verse proportion to the bunch length reduction. The max-
imum energy spread that the FEL will tolerate sets a limit
on the amount of compression that may be achieved. The
input beam to the compressor is expected to have an uncor-
related RMS energy spread of .04% and a FWHM bunch
length of 10 ps. The FEL and scattering experiments will
tolerate approximately1% energyspread at the compressor
(0.4% at the wiggler), hence we may compress the beam
by a factor of 25 for a compressed FWHM of a few hun-
dred femtoseconds. Figures 2 and 3 show the results of a
PARMELA simulation of 2 nC of charge represented by
1000 macro particles. The peak current after compression
is 10 kA in a bunch length of 200 fs FWHM. The actual
performance will depend on the details of the bunch distri-
bution and wakefields generated.
Short,highcurrentbunchesgeneratesignificantcoherent
synchrotronradiation(CSR) at wavelengthsnear the bunch
length and longer. The CSR power generated by N parti-
cles bending through an angle θ is [7]
P[watts] = 2.42 × 10−20
N2
ρ2/3[m] σ4/3[m]
θ
2π
(3)
where ρ is the the bending radius and σ is the RMS bunch
length. The SDL chicane parameters are ρ = 1.25 m, θ =
0.25, and σ = 50µm. This yields an output power of 65
kW. The CSR is emitted predominantlyin the final bend of
the chicane. The total emitted CSR energy is 0.4 mJ near
60µm wavelength. We plan to detect the CSR spectrum for
1198

Page 3

024
Time ps
68
0.1
0.15
0.2
0.25
0.3
0.35
Current kA
024
Time ps
68
81.5
82
82.5
83
83.5
84
Energy MeV
Figure 2: PARMELA simulation of energy-chirped beam
entering chicane.
use as a bunch length diagnostic. The total beam power is
Pbeam= IV = 8.4 × 1010W. Thus the energy loss due to
CSR is 4×10−4/0.17 = 0.2%, a modest increase over the
1% induced energy spread.
Emittance dilutionis of courseanticipatedto accompany
the compression process, and calculation of its magnitude
represents an active area of research[8]. The SDL accel-
erator with its compressor should provide an experimental
platform for measuring these effects. Emittance growth is
minimized by bringing the beam to a tight focus (increas-
ing its divergenceangle) in the final chicane bend. There is
a quadrupole triplet installed just upstream of the chicane
for this purpose.
5SUMMARY
The designof the SDL electronbunchcompressorhasbeen
described including its structure, magnetic properties and
the electron beam dynamics. It is currently under con-
struction. This compressor is expected to compress a 10 ps
FWHM, 2 nCinto 200 fs FWHM bunchlength. Thisyields
a peak current of 10 kA.
6ACKNOWLEDGEMENTS
We thank Bruce Carlsten and Lloyd Young of LANL for
muchhelpwith thePARMELA code. Thisworkperformed
under the auspices of the U.S. Department of Energy Con-
tract DE-AC02-76CH00016.
0 0.250.50.75
Time ps
11.251.51.75
0
2
4
6
8
10
Current kA
00.25 0.5 0.7511.25 1.5 1.75
Time ps
81.5
82
82.5
83
83.5
84
Energy MeV
Figure 3: PARMELA simulation of compressed bunch ex-
iting chicane.
7REFERENCES
[1] D.C. Quimby et al. Development of a 10-meter wedged–pole
undulator. Nucl. Inst. and Meth., A285:281–289, 1989.
[2] D.T. Palmer et al. Microwave measurements of the
BNL/SLAC/UCLA 1.6 cell photocathode RF gun. In 1995
Particle Accelerator Conf., pages 982–984, 1995.
[3] W.S. Graves, E.D. Johnson, P.G. O’Shea. A high resolution
profile monitor. In IEEE Particle Accelerator Conference,
1997.
[4] X.J. Wang et al. Experimental observation of high-brightness
microbunching in a photocathode rf electron gun. Physical
Review E, 54:3121–3124, 1996.
[5] T.O. Raubenheimer, P. Emma, S. Kheifets. Chicane and wig-
gler based bunch compressors for future linear colliders. In
1993 Particle Accelerator Conf., pages 635–637, 1993.
[6] B.E. Carlsten. Nonlinear subpicosecond electron-bunch com-
pressor. Technical report, Los Alamos Nat. Lab, 1995. LA-
UR-95-3933.
[7] J.B. Murphy et al. Longitudinal wakefield for an electron
moving on a circular orbit. Technical report, National Syn-
chrotron Light Source, 1996. BNL 63090.
[8] B.E. Carlsten, T.O. Raubenheimer. Emittance growth of
bunched beams in bends. Phys. Rev. E, 51:1453–1470, 1995.
1199