Nuclear Instruments and Methods in Physics Research A 483 (2002) 24–28

Characterization of an 800 nm SASE FEL at saturation

A. Tremainea,*, P. Frigolaa, A. Murokha, C. Pellegrinia, S. Reichea,
J. Rosenzweiga, M. Babzienb, I. Ben-Zvib, E. Johnsonb, R. Maloneb,
G. Rakowskyb, J. Skaritkab, X.J. Wangb, K.A. Van Bibberc, L. Bertolinic,
J.M. Hillc, G.P. Le Sagec, M. Libkindc, A. Toorc, R. Carrd, M. Cornacchiad,
L. Klaisnerd, H.-D. Nuhnd, R. Rulandd
a
UCLA, Los Angeles, CA 90095, USA
b
BNL, Upton, NY 11973, USA
c
LLNL, Livermore, CA 94551, USA
d
SLAC, Stanford, CA 94720, USA

Abstract

Visible to Infrared SASE Amplifier is a free electron laser (FEL) designed to saturate at a radiation wavelength of
800 nm within a 4 m long, strong focusing undulator. Large gain is achieved by driving the FEL with 72 MeV, high
brightness beam of BNL’s accelerator test facility. We present measurements that demonstrate saturation in addition to
the frequency spectrum of the FEL radiation. Energy, gain length and spectral characteristics are compared and shown
to agree with simulation and theoretical predictions. r 2002 Published by Elsevier Science B.V.

PACS: 41.60.C

Keywords: SASE; FEL; Saturation; Strong focusing; Undulator

1. Introduction was made several years ago [1], and recently

experiments have confirmed high gain, short
Several FEL schemes are used to generate wavelength SASE FEL theory [2–6]. X-ray SASE
coherent, high power radiation: oscillators, seeded FEL devices have been proposed [7,8], and recent
and single pass devices. Because hard X-ray experimental success shows the feasibility of
mirrors and seed sources are currently unavailable, building such facilities.
the SASE FEL is the most promising configura- In this paper, we describe a visible SASE FEL,
tion for obtaining coherent, X-ray radiation. The Visible to Infrared SASE Amplifier (VISA) [5,6]
proposal for a SASE FEL to generate hard X-rays that is R&D for the Linac Coherent Light Source
X-ray facility and lases at a fundamental wave-
length of 840 nm. The electron beam source and
*Corresponding author. Brookhaven National Laboratory,
Building 725, Brookhaven Avenue, Upton, NY 11973, USA.
matching beamline optics upstream of the undu-
Tel.: +1-516-344-2505; fax: +1-516-344-3029. lator are reviewed. Since the electron beam quality
E-mail address: [email protected] (A. Tremaine). is important to the performance of VISA, a

0168-9002/02/$ - see front matter r 2002 Published by Elsevier Science B.V.

PII: S 0 1 6 8 - 9 0 0 2 ( 0 2 ) 0 0 2 8 0 - 2
 A. Tremaine et al. / Nuclear Instruments and Methods in Physics Research A 483 (2002) 24–28 25

detailed analysis of the transport line and beam Eight diagnostic pop-in ports (spaced by 50 cm
parameters is given. The strong focusing undulator with the last port 25 cm from the end of the
and the high beam quality allow VISA to saturate undulator) are situated along the length of the
in only 3.6 m with a power gain length of 18.7 cm. undulator [14]. Both SASE and electron beam
These experimental results in addition to the parameters vs. distance measurements use these
frequency spectrum are described and compared pop-ins. A faraday cup downstream of the
to simulation and theory. undulator and an electronic BPM upstream of
the undulator are both used for charge measure-
ments.
2. Experimental setup

In this section, we give a description of the 3. Experimental measurements

electron source, beam transport, and undulator.
The electron beam is derived from a 1.6 cell S- Electron beam parameters were measured im-
band photo-injector system after which the beam mediately after the 2nd linac, before the 2-dipole
is accelerated by two SLAC type linacs to a final dispersion section. Here, the beam had a charge of
energy of 72 MeV. Detailed measurements imme- 250 pC and a 60 A peak current. Quad-scans of the
diately after the 2nd linac confirm that a high projected emittance in the linac section gave an
brightness beam with an emittance of 0.8 mm mrad emittance of 0.8 mm mrad [9]. To longitudinally
and current of 60 A [9] can be generated by the compress the beam in the double bend line, and
ATF photoinjector. increase the peak current, an energy chirp (mea-
The electron beam then passes through a double sured after the linac section) of 0.15–0.2% rms is
bend consisting of two 201 deflecting dipole needed.
magnets and quadrupoles. When the electron The diagnostic pop-ins are used for SASE
bunch energy is chirped in the linac, the double energy vs. distance measurements along the
bend longitudinally compresses the electron undulator and Fig. 1 shows this result. The
bunch, thus increasing its peak current. Conse- deviation of the last two points from the linear
quently, the transverse emittance is increased fit for the previous five shows that the system has
through this section as we will further discuss in
the next section. After the dispersion section, a
quadrupole triplet is used for final matching of the
electron beam into the undulator.
The VISA 4 m long, strong focusing undulator
[10] has a 6 mm gap and a 1.8 cm period. The
undulator is made up of four 1 m sections and the
strong focusing quadrupole field requires align-
ment to within 30 mm. A laser interferometric
system specifically designed for VISA [11] aligns
the undulator to this tolerance. If this alignment
tolerance is not met, then VISA high gain is
reduced [12]. Dipole magnets placed on either side
of the propagation axis with opposite polarity
introduce strong focusing [13] in the undulator.
Simulations confirm [11] that the enhanced elec-
tron beam density reduces the saturation length by
40%. Eight steering magnets spaced by 50 cm are Fig. 1. SASE energy vs. distance. Deviation of last two points
situated along the undulator and used to keep the from the line fit of the previous five is evidence that the system is
electron beam and SASE radiation co-linear. in saturation.
26 A. Tremaine et al. / Nuclear Instruments and Methods in Physics Research A 483 (2002) 24–28

reached saturation after 3.6 m of undulator. The number of spikes is [15] Ns BLb =2pLc : While we
slope of the line produces a power gain length of see spikes when running with a non-compressed
Lg ¼ 18:7 cm. Defining gain as beam, none are observed with a compressed beam,
E as in the case of Fig. 2. In this last case,
G¼ ð1Þ 2pLc B50 mm, and an absence of spikes indicates
E0
a very short bunch length. An analysis on this
where E0 is the spontaneous energy inside the electron bunch compression is presented in the
coherency cone and bandwidth within the first next section.
field gain length and E is the energy measured at Additional measurements of VISA include those
the exit of the undulator, we obtain a value G ¼ on the non-linear harmonics and longitudinal
2  107 : electron beam microbunching. Here, we just state
A spectrum at saturation is shown in Fig. 2 with the general results as future publications will
a central wavelength at 840 nm. The wavelength describe both in detail. The gain lengths for the
for an FEL is given by the equation, 2nd and 3rd non-linear harmonics have been
 
lu K2 measured and confirmed to decrease with mode
l¼ 2 1þ ð2Þ number, n; compared to that of the fundamental.
2g 2
In addition, the spectra for the non-linear harmo-
where lu is the undulator period, g is the electron nics have been measured, and in a single shot the
beam energy, and K ¼ 1:26 is the undulator fundamental, 2nd and 3rd harmonics were cap-
parameter. For our parameters, Eq. 2 predicts a tured for the first time in a SASE FEL. Also, the
fundamental SASE radiation wavelength of fundamental and 2nd harmonic longitudinal mi-
840 nm, thus confirming the measurement in crobunching of the electron beam at the exit of the
Fig. 2. undulator has been measured. Using a micro-
When the bunch length, Lb ; is much longer than bunching coherent transition radiation (CTR)
a cooperation length, Lc ; we also expect to see experimental setup, shot-to-shot measurements of
spikes inside the spectral line. The expected microbunching for the two lowest modes vs. SASE
radiation were captured. These measurements
show the microbunching for both modes saturat-
ing at very high SASE radiation, another con-
firmation that the VISA FEL is in saturation.

4. Electron beam measurements and analysis

As the compression hypothesis was situated, a

need emerged to measure the current directly at
the location of the undulator. A single bolometer
approach was used to construct a CTR indicator,
which would be sensitive to the changes in the
beam longitudinal distribution [16]. Initial mea-
surements of CTR energy as a function of the linac
conditions indicated a strong compression, as the
CTR signal peaked up sharply within a narrow
sub 21 window of the linac RF phase. The
measurements with and without a filter of know
Fig. 2. VISA single shot spectrum at saturation. The central cut-off frequency demonstrated signal ratio of 0.68
wavelength is 840 nm, and the one spike indicates a bunch at the SASE operating point. Such a ratio indicates
length of the order of the cooperation length. a short bunch length (less than a 100 mm rms),
 A. Tremaine et al. / Nuclear Instruments and Methods in Physics Research A 483 (2002) 24–28 27

Fig. 4. Energy vs. distance. GENESIS simulation compared to

measured data different that in Fig. 1. Gray area indicates a
one-sigma shot noise fluctuations due to different seedings in
GENESIS.

which differs greatly from the beam parameters

measured at the linac. The test proved strong
bunch compression in the dispersive section of the
beam line and introduced a quantitative bench-
mark for the simulation studies.
To support the experimental results with simu-
lations, PARMELA was used to simulate ATF
photo-injection line and reproduce a particle
distribution similar to the one measured after
linac. In the dispersive section, CSR has to be
taken into account and the code ELEGANT was
found to be the most suitable. The control knob
for a non-linear compression in the simulations
was an energy offset of the electron beam with
respect to the nominal tune (resembling a natural
uncertainty in the experiment). The compression
was varied until the output ELEGANT file
matched CTR bunch length measurements, and
the result was an asymmetric beam with the peak
current 250 A, core horizontal emittance growth
up to 4 mm mrad and vertical emittance preserva-
tion. The phase space compression generated by
ELEGANT is shown in Fig. 3.
The ELEGANT output distribution was
roughly matched to the focusing lattice of the
undulator and imported into GENESIS for FEL
Fig. 3. Compression through the dispersion section. (a) Energy simulations. Fig. 4 shows the comparison of the
chirped electron beam exiting the linac section, before the GENESIS output for several shot noise seeds with
dispersion section. (b) Compressed beam after dispersion
the measured data, different from those measured
section, before the undulator. (c) Longitudinal beam size from
the linac to the undulator. in Fig. 1. The data points lie in the shot noise
28 A. Tremaine et al. / Nuclear Instruments and Methods in Physics Research A 483 (2002) 24–28

fluctuations of the simulations and the resulting Stanford Synchrotron Radiation Laboratory, Stanford,
gain length of 19.2 cm agrees closely with the CA, 1992.
[2] M.J. Hogan, et al., Phys. Rev. Lett. 81 (1998) 4867.
measured 18.7 cm gain length.
[3] J. Andruszkow, et al., Phys. Rev. Lett. 85 (2000) 3825.
[4] S. Milton, et al., Science 292 (2001) 2037.
[5] A. Murokh, et al., Proceedings of the 2001 Particle
5. Conclusion Accelerator Conference, Chicago, IL.
[6] A. Tremaine, et al., Proceedings of the 2001 Particle
Accelerator Conference, Chicago, IL.
VISA, a visible SASE FEL, saturated in 3.6 m [7] M. Cornacchia, et al., Linac Coherent Light Source
with a gain length of only 18.7 cm. Spectral (LCLS) Design Study Report, O ! Report SLAC-R-521,
measurements from the SASE radiation indicated Stanford Linear Accelerator Center, Stanford, CA, revised
that the bunch length at the undulator was shorter 1998.
[8] R. Brinkmann, G. Materlik, J. Rossbach, A. Wagner
than that measured in the linac section. CTR
(Eds.), Conceptual Design of a 500 GeV e+e Linear
measurements at the undulator confirmed this Collider with Integrated X-ray Laser Facility, DESY
hypothesis. Simulations generated the beam para- Report DESY97-048, Deutsches Elektronen-Synchrotron,
meters after the dispersion section and using these Hamburg, 1997.
parameters, FEL simulations showed good agree- [9] V. Yakimenko, Proceedings of the 2001 Particle Accel-
ment with the experimental measurements. erator Conference, Chicago, IL.
[10] R. Carr, et al., Phys. Rev. ST Accel. Beams, submitted for
publication.
[11] R. Ruland, et al., Proceedings of the 1999 Particle
Acknowledgements Accelerator Conference, New York, 1999.
[12] P. Emma, et al., Proceedings of the 20th International FEL
Conference (FEL98), Williamsburg, VA, 1198, SLAC-PB-
We would like to thank everybody from BNL, 7913.
LLNL, SLAC and UCLA who made this experi- [13] A.A. Varvolomeev, A.H. Hairetdinov, Nucl. Instr. and
ment a success. Meth. A 341 (1994).
[14] A. Murokh, et al., in: V.N. Litvinenko, Y.K. Wn (Eds.),
Proceedings for the 2000 International Free Electron Laser
Conference, Durham, NC, 2000 Elsevier Science B.V.,
References Amsterdam, 2001 p. II–35.
[15] Bonafacio, et al., Phys. Rev. Lett. 73 (1994) 70.
[1] C. Pellegrini, A 4 to 0.1 nm FEL based on the SLAC [16] A. Murokh, et al., Nucl. Instr. and Meth. A 410 (1998)
linac, Workshop of Fourth Generation Light Sources, 452.