Experimental Demonstration of Emittance Compensation with Velocity Bunching
M. Ferrario,
1
D. Alesini,
1
A. Bacci,
3
M. Bellaveglia,
1
R. Boni,
1
M. Boscolo,
1
M. Castellano,
1
E. Chiadroni,
1
A. Cianchi,
2
L. Cultrera,
1
G. Di Pirro,
1
L. Ficcadenti,
1
D. Filippetto,
1
V. Fusco,
1
A. Gallo,
1
G. Gatti,
1
L. Giannessi,
4
M. Labat,
4
B. Marchetti,
2
C. Marrelli,
1
M. Migliorati,
1
A. Mostacci,
1
E. Pace,
1
L. Palumbo,
1
M. Quattromini,
4
C. Ronsivalle,
4
A.R. Rossi,
3
J. Rosenzweig,
5
L. Serafini,
3
M. Serluca,
6
B. Spataro,
1
C. Vaccarezza,
1
and C. Vicario
1
1
INFN-LNF, Via Enrico Fermi, 40–00044 Frascati, Rome, Italy
2
INFN-Roma ‘‘Tor Vergata,’’ Via della Ricerca Scientifica, 1-00133 Rome, Italy
3
INFN-Milano, Via Celoria 16, 20133 Milan, Italy
4
ENEA, Via Enrico Fermi, 00044 Frascati, Rome, Italy
5
Department of Physics and Astronomy, University of California, Los Angeles,405 Hilgard Avenue, Los Angeles, California 90095, USA
6
INFN-Roma I, Piazza Aldo Moro 5, 00185 Roma, Italy
(Received 18 September 2009; published 5 February 2010)In this Letter we report the first experiments aimed at the simultaneous demonstration of the emittancecompensation process and velocity bunching in a high brightness electron source, the SPARC photo-injector in INFN-LNF. While a maximum compression ratio up to a factor 14 has been observed, in aparticular case of interest a compression factor of 3, yielding a slice current of 120 Awith less than
2
m
slice emittance, has been measured. This technique may be crucial in achieving high brightness beams inphotoinjectors aiming at optimized performance of short wavelength single-pass free electron lasers orother advanced applications in laser-plasma accelerators.
Ultrashort electron bunch production is a subject of investigation that has attracted increasing attention in re-cent years, spurred by a large number of applications,spanning short wavelength free electron lasers (FEL),THz radiation production, linear colliders, and plasmawake field accelerators. Space charge effects at low energyprevent the generation of short electron bunches (
<
1 ps
)withasignificantamountofcharge(
>
10 pC
)directlyfromthe electron source, leading to emittance degradation andbunch elongation within a few centimeters downstream thecathode. As such, bunch compression is always necessaryto shorten the electron pulse to the required length thusachieving a high peak current. The most popular andeffective device used thus far is the magnetic compressorin which a bunch with a time-energy correlation (or chirp)is driven along an energy-dependent path length by adispersive, nonisochronous beam transport section, con-sisting, in its simplest form, of four dipoles placed in achicane configuration. The process of magnetic compres-sion may often unacceptably degrade the beam quality,however, due to significant emittance growth caused bycoherent synchrotron radiation effects in bends [1].On the other hand, a new method termed velocity bunch-ing, able to compress the bunch using rectilinear trajecto-ries at relatively low energy [2], which must thus beintegrated into the emittance compensation process [3],has been proposed in [4]. The longitudinal phase spacerotation in the velocity bunching process is based on acorrelated time-velocity chirp in the electron bunch, insuch a way that electrons on the tail of the bunch are fasterthan electrons in the bunch head. This rotation occursinside the longitudinal potential of a traveling rf wave(longitudinal focusing) which accelerates the beam insidea long multicell rf structure and simultaneously applies anoff crest energy chirp to the injected beam. This is possibleif the injected beam is slightly slower than the phasevelocity of the rf wave so that when injected at the zerocrossingfieldphaseit slipsbacktophaseswherethe fieldisaccelerating, but is simultaneously chirped and com-pressed.The keypointis thatcompressionand accelerationtake place at the same time within the same acceleratorsection, the initial one following the gun.In order to prevent irreversible emittance growth duringbunch compression the key issue is to preserve the lami-narity of the beam with an envelope propagated as close aspossible to a Brillouin-like flow, represented by an invari-ant envelope [5] as generalized to the context of beamcompression and thus increasing
I
during acceleration.For these kind of beams, mismatches between the spacecharge correlated forces and the external focusing gradientproduce slice envelope oscillations that cause normalizedemittance oscillations. It has been shown that to keep suchoscillations under control during thevelocity bunching, thebeam has to be injected into the rf structure with a laminarenvelope waist (
0
¼
0
) and the envelope has to bematched to the accelerating and focusing gradients insuch a way to stay close to an equilibrium mode [5,6].
Ponderomotive rf focusing force are actually too weak in atravelling wave structure [7] to provide sufficient beamfocusing. A long solenoid around the accelerating structurePRL
104,
054801 (2010)PHYSICAL REVIEW LETTERS
week ending5 FEBRUARY 2010
0031-9007
=
10
=
104(5)
=
054801(4) 054801-1
Ó
2010 The American Physical Society
is a convenient replacement to provide the necessaryfocusing.In this configuration the matching condition for thetransverse rms envelope
is given by
¼
1
k
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
I
0
4
0
I
A
1
þ
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1
þ
4
"
n
0
kI
A
I
0
2
s v uut
;
(1)where
k
¼
eB
sol
mc
,
B
sol
is the solenoid field,
I
A
¼
17 kA
theAlfve´n current,
"
n
the normalized emittance,
0
and
I
0
arethe values for the current and the energy, respectively, atinjection into the compressor. The previous relation repre-sents a new exact equilibrium solution (including the emit-tance contribution which in practical cases of velocitybunching is small but not negligible; previously derivedapproximate solutions are discussed in Ref. [8]) of thebeam envelope equation:
00
þ
0
0
þ
k
2
¼
I
2
I
A
3
þ
"
2
n
2
3
;
(2)where
0
%
2
E
acc
,
E
acc
½
MV
=
m
being the acceleratingfield. The main approximation leading to solution (1) con-sists on the assumption that the beam current grows line-arly in the compressor as
I
¼
I
0
=
0
. Nevertheless thisresult is confirmed by observations performed in severalsimulations of the rf compressor [6], indicating that bestperformances in terms of final beam brightness areachieved near the condition of beam flow at constantenvelope. The matching conditions (1) guarantee beamlaminarity preservation during acceleration and compres-sion, but the final value of the emittance is strongly af-fected by the phase of the emittance oscillation in thevelocity bunching, that cannot be easily predicted by thetheory. The fine-tuning of the emittance compensationprocess can be performed by varying the injection spotsize via a parametric scan of the gun solenoid field [9], aswill be shown later.In this Letter we report the first experimental observa-tion of effective emittance compensation in a velocitybunching device. The experiment described here was per-formed at the SPARC photoinjector, during the final com-missioning stage of the SPARC project [10]. Experimentsaiming at the demonstration of the velocity bunching prin-ciple have been previously performed in other laboratories[11–16] in photoinjector configurations not specifically
designed for emittance compensation, thus limiting theachieved beam brightness. This simultaneous demonstra-tion of emittance compensation with significant beamcompression obtained using the velocity bunching tech-nique, has indeed been one of the central goals at the beamphysics program at SPARC. For this reason, the first two
S
-band traveling wave accelerating structures downstreamthe 1.6 cells
S
-band rf gun, are embedded in long sole-noids, consisting of 13 coils each with a maximum field of 1.8 kG, in order to keep the space charge induced emit-tance oscillations under control as the bunch iscompressed.The SPARC beam diagnostics allow rms beam envelopemeasurements on four screens: three screens are located atthe entrance of each rf structure while the fourth is locatedat the linac exit. Downstream the linac exit the rms nor-malized emittance
"
n
is measured using the quadrupolescan technique, while an rf deflecting cavity allows hori-zontal slice emittance measurements [17] as well as bunchlength measurements with a resolution of 60 fs [18].During the experiments the drive laser was operatedwith a 7.3 ps FWHM long pulse, having a rms spot radiusof
350
m
. The bunch charge was 280 pC giving a maxi-mum slice current of about 30 A without compression. Atthe gun exit the beam energy was 4.4 MeV correspondingto a peak field on the cathode of about
100 MV
=
m
, limitedby rf breakdown. This field limit is the central reason our rf photoinjector does not achieve emittance performancessimilar the one obtained with the new rf gun design atthe Linac Coherent Light Source [19]. When the beam wasaccelerated on crest, using an accelerating field amplitudeof
20 MV
=
m
in the first two sections and
10 MV
=
m
in thefinal section, the final energy was 148 MeV with an energyspread of 0.1% and an rms energy stability better than0.1%. The rms bunch length measured at the linac exitwas 3 ps, with a measured minimum rms projected emit-tance of
1
:
5
m
in both planes with the gun solenoid set to2.46 kG (long solenoids off).For linac-driven FEL applications, the bunch slice pa-rameters are most relevant to the amplification process.The SPARC FEL operating at 530 nm has a typical slip-page length of the order of
250
m
; we have thus takenthis value as the bunch slice length in our analysis. In Fig.1the beam current profile and the slice emittance is plottedvs the longitudinal bunch coordinate.The first measurements made at SPARC in the velocitybunching mode were devoted to study the longitudinalfocusing effects as a function of the injection phase, with-
FIG. 1 (color online). Horizontal slice emittance and corre-sponding current profile (continuous lines), for the noncom-pressed beam. Gun solenoid set to 2.46 kG, long solenoids off.
PRL
104,
054801 (2010)PHYSICAL REVIEW LETTERS
week ending5 FEBRUARY 2010
054801-2
out any additional external focusing (long solenoids off).Figure2shows the measured rms bunch length
t
versusthe injection phase
’
of the first traveling wave structure.The corresponding measured rms compression factor
C
¼
t
ð
0
Þ
=
t
ð
’
Þ
, the bunch length after on crest accelerationdivided by the bunch length after compression, is alsoshown. Significant bunch compression occurs only at aphase shift of 85
forward of crest. At this injection phasethe beam energyreduces to 100MeVand the energyspreadgrows up to 1%.The strong compression regime occurs when the phaseshifts from
À
85
to
À
95
, as expected, with almost con-stant final energy and energy spread observed. The shortestmeasured rms bunch length is 210 fs (
63
m
), limited bythe longitudinal beam emittance. The final two measure-ments also illustrate the effect of overcompression whenthe phase setting exceeds
À
95
. In Fig.2the results of PARMELA [20] simulations are also shown. The agree-ment is quite satisfactory even in the over-compressionregime.A systematic study of the emittance compensation pro-cess has been carried out using the moderate compressionratio of 3, as that one foreseen for the SPARX FEL project[21]. The SPARX linac has been in fact designed to pro-duce a 1 kA beam at 1.5 GeVin two compression stages: alow energy velocity bunching with a compression factor 3followed by a magnetic compressor at higher energy. Wehave considered this quite conservative choice for thecompression factor the most suitable for a proof of princi-ple experiment with the emphasis on the emittance com-pensation process, whose effectiveness in this regime hasnot been demonstrated so far. The final rms normalizedemittance has been optimized as a function of the gunsolenoid field for a given value of the long solenoids of 450 G, as shown in Fig.3. The observed asymmetry in
x
and
y
planes is likely due to an offset of the solenoidmagnetic axis with respect to the linac rf axis [22]. InTableIthe measured beam parameters and related experi-mental conditions are summarized. The second columncontains the measured data of the uncompressed beam(on crest acceleration) for comparison. To estimate thechromatic effects induced by the solenoids we have mea-sured the emittance with solenoids on and off also for theuncompressed beam. We expect a significant reduction of chromatic effect with a careful beam based alignment of the long solenoids.The lowest achieved emittances in thevelocity bunchingmode are
"
nx
¼
2
:
12
m
and
"
ny
¼
1
:
45
m
(to be com-pared with
"
nx
¼
6
:
2
m
and
"
ny
¼
4
:
0
m
with longsolenoids off, marked with
Ã
in the table). Comparing thegeometric mean of the emittances with and without com-pression(long solenoidsturnedon inbothcases) the resultsare even more remarkable. With a slice peak current of 120 A this bunch exhibits the highest beam brightness(
$
10
14
A
=
m
2
) so far obtained by the SPARC injector. InFig.4(left plot) the measured envelopes are shown incomparison with simulations for three different conditions:
FIG. 2 (color online). Measured rms bunch length (black) andcorresponding compression factor (red) of a 280 pC beam versusthe phase of the first traveling wave structure. PARMELAsimulations are also shown with dashed red line.FIG. 3 (color online). Measured rms normalized emittance at
C
¼
3
as a function of the gun solenoid field, long solenoid set to450 G. For comparison black and green dots represent bestresults without compression, long solenoid off.TABLE I. Measured beam parameters.NocompressionCompressionratio 3Charge (pC) 280 280Gun solenoid field (kG) 2.460 2.460Long solenoid field (G) 660 4500* 0*Injection phase (deg) 0
À
88
Energy (MeV) 148 100Energy spread (%) 0.1 1.0Rms bunch length (ps) 3.0 0.97
"
nx
(
m
) 1.85 2.121.4* 6.2*
"
ny
(
m
) 1.65 1.451.5* 4.0*Geometric mean
ffiffiffiffiffiffiffiffiffiffiffiffiffiffi
"
nx
"
ny
p
(
m
) 1.75 1.751.45* 5.0*
PRL
104,
054801 (2010)PHYSICAL REVIEW LETTERS
week ending5 FEBRUARY 2010
054801-3
Search History:
Result 00 of 00
00 results for result for
p.
Recommended
More From This User
Notes
Load more
