STATUS OF THE UCLA HIGH-GAIN

INFRARED FREE ELECTRON LASER *

M. Hogan, C. Pellegrini, J. Rosenzweig, G. Travish
UCLA Department of Physics, Los Angeles, CA 90024, USA

A.Varfolomeev
Kurchatov Institute, Moscow, Russia

A compact, infrared (10-20 µm), high-gain FEL is being

commissioned at the Particle Beam Physics Laboratory II. OVERVIEW
(PBPL) at UCLA. A 60 cm long undulator with a period of
1.5 cm and an undulator parameter K≈1 has been built to be
used in conjunction with the PBPL beam. Experiments will a. The Beamline
focus on FEL physics pertinent to proposed short wavelength
The beam is produced in an S-band RF copper
devices. Of particular interest is exploration of startup from
photocathode gun driven by a frequency-quadrupled, pulse-
noise, self amplified spontaneous emission (SASE), beam
compressed Nd:YAG laser (UV) [2]. Solenoids control the
parameter effects on gain, and output power fluctuations.
highly divergent beam and provide for emittance compensation
Beam micro-bunching due to the FEL action will also be
[3] into the linac. A Plane Wave Transformer linac (PWT)
measured using coherent transition radiation. Here we present
accelerates the electrons from an injection energy ~4 MeV to a
an overview of the relevant diagnostics, FEL simulation
final energy of ~17 MeV [4].
results and proposed experiments.
Six of the quadrupoles are used to match the four phase-
space parameters needed for injection into the undulator. The
I. INTRODUCTION magnetic center of the beamline is passively aligned to
~100µm using machined brackets, optical tables, and linear
The Free Electron Laser has shown potential as a light bearings (rails). This tolerance was chosen based on the
source in the infrared, UV and, as recent proposals indicate, in performance simulations of our FEL. A second dipole after the
the XUV and X-ray regime. While oscillator experiments have undulator will bend the electron beam away from the optical
provided a number of verifications and enhancements to theory pulse to facilitate the IR optics/diagnostics.
and operational experience, few high-gain amplifier systems
have operated in the optical regime. This paper describes the b. Diagnostics
UCLA IR FEL – a system designed to study critical issues in An unsaturated high-gain FEL is highly (exponentially)
high-gain systems and to improve the operational FEL and sensitive to certain beam-parameter fluctuations. Thus, beam
accelerator experience with the requisite high-brightness diagnostics on the UCLA system are designed for single-bunch
beams. (shot-to-shot) operation. Beam position, size, charge and
The UCLA experiment was designed to study issues emittance are measured using the following:
important to future short wavelength devices at a minimum of • Stripline beam position monitors (BPMs) for non-
cost and space. The short-period undulator, combined with our destructive measurements.
moderate-energy beam produces radiation in the infrared (IR), • Phosphor screens and video cameras.
where a large number of diagnostics are available, without the • Integrating Current Transformer (ICT).
added complexity of producing a higher-energy beam necessary • Slits (1D pepper pots) to measure the effective
for operation at shorter wavelengths. Further, working in the transverse emittance of the space-charge dominated
IR does not suffer from the beam noise problems associated beam [5].
with past microwave FELs. The lack of suitable sources at • A SLAC-like pulse-length monitor to make non-
short wavelengths makes the feasibility of start up from noise destructive shot-to-shot pulse-length measurements
(SASE) important [1]. Additionally, the difficulty of [6].
producing high reflectance mirrors makes an oscillator The first dipole magnet, in conjunction with the
configuration impractical for short wavelengths, so successful quadrupoles, is used as a spectrometer to measure the energy
operation in the high-gain regime is a necessary precursor to and the energy spread. The second dipole will also allow for a
designing short-wavelength devices. For these reasons the FEL crude energy measurement after the beam exits the undulator.
studies will begin from SASE in the high gain regime. Other diagnostics include Faraday Cups for charge

*Work supported by US DOE Grants DE-FG03-90ER40796 and DE-FG03-92ER40693.

measurement and Cherenkov radiators, in conjunction with a
streak camera, to measure the absolute pulse length. III. SIMULATION PREDICTIONS
c. Microbunching monitor The lack of experimental work on SASE optical FELs
Coherent transition radiation (CTR) can be used to necessitates relying on numerical simulations to predict the
measure the extent of bunching in the FEL. We plan on performance of our experiment. Earlier work in the IR on the
installing a foil at the exit of the undulator to study the Paladin FEL at LLNL has helped test high gain codes, but has
bunching. Calculations indicate that the expected 5% bunching not provided information on startup from noise [10].
factor should produce CTR in the FEL band comparable to the Fluctuations in FEL performance, especially from startup, and
FEL output itself [7]. sensitivities to system parameters are a critical issue in future
short-wavelength high-gain systems where output stability and
d. The Undulator saturation are significant to users. Simulations of the UCLA
A planar undulator 60 cm long with a 1.5 cm period, 5 system have been performed to investigate such sensitivities
mm fixed gap spacing and a greater than 7 kG peak field in hopes of performing experimental comparisons. Most of
awaits installation into the beamline. The undulator was the following work was performed with TDA3D [11] and
designed to provide IR radiation from modest beam energies (< includes 3D effects, diffraction, emittance and energy spread.
20 MeV) while maintaining a strong coupling (K~1). An rms
field uniformity of better than 0.18%, measured using both a a. Current
Hall probe and the pulsed wire technique [8], should assure The beam current is our easiest parameter to control and
good FEL performance. Additionally, the second integral of measure. Spontaneous emission can be differentiated from
the undulator field satisfies the requirement that the rms amplified (stimulated) radiation by observing the dependence
electron beam deflection in the wiggle plane (~105 µm) be on current: spontaneous emission is broadband and scales
less than the rms beam waist (~200 µm). It should be notes linearly with the current, while amplified radiation power, P,
that the construction of the FEL is not well suited to studying scales as P~Iexp(αI^4/3) where α is a constant and I is the
the effects of varying undulator parameters such as field beam current.
strength and error. Simulations of gain vs. current show that current
variations ~ 10% vary the FEL output power +40%/-25%.
Table 1: Electron Beam and FEL Parameters expected for the Current variations of this order are within our ability to
UCLA IRFEL. measure, and power (energy) fluctuations of a few percent are
within our detector / electronics bandwidth. The challenge will
Electron Beam Parameters [Expected]
lie in deconvolving a variation of beam current from
Energy 17 MeV
parameters such as beam size, pulse length, energy spread and
Energy Spread (uncorr.) 0.1%
emittance.
Current (peak) 200 Amps
Pulse Length (rms) ~5 psec b. Beam Size
Norm. Emittance (rms) 5 mm mrad Beam-size changes, such as those caused by space charge,
Undulator Parameters [Measured] affect the beam density as well as the matching into the
Total length 60 cm undulator. The FEL is sensitive to the overall (six
Undulator period 1.5 cm dimensional) beam density, however small changes in the
Peak field on axis 7.3 kG transverse beam size should cause predictable changes in the
Pole face gap (fixed) 5 mm FEL performance. Further, simulations predict FEL
Undulator parameter (K) ~1 performance is insensitive to achievable beam matching.
FEL parameter (ρ) [9] ~1 x 10-2 Regardless, matching is a technical issue that needs to be
FEL Parameters [Simulations @10.6 µm] resolved with experience in beam handling. Phosphor screens
Radiation wavelength 10-20 µm and BPMs should provide sufficient operator feedback on beam
size.
Power gain length 7.2 cm
SASE peak power 3 mW @ 7.2cm
c. Pulse Length
~1 W @ 60cm
Only a few “finite pulse” simulations have been
performed on our system. Slippage is a factor in the
performance of this system; however, over the short undulator
being initially used the output power is not degraded severely. response time of the detector. By removing the Winston Cone
Further work is needed to quantify (through simulations and and aperturing the field of view of the detector to limit the
experiment) this effect. collected background, the SNR can be increased by several
The pulse length is also a factor in much the same way additional orders of magnitude. Other detectors such as
that beam size is. The variation of the pulse length due to Mercury-Cadmium-Telluride photodiodes may offer the
laser fluctuations and space charge are still an experimental advantage of faster response times and/or higher quantum
uncertainty. efficiencies, while only needing to be cooled to liquid Nitrogen
temperature. Both the spontaneous emission and the amplified
d. Energy Spread signal should be well within our sensitivity, and studies of
Wakefields (primarily from the linac) are expected to SASE FEL radiation production should be feasible.
produce a correlated energy spread ~ 1%. This spread can be
ameliorated by running the linac “off crest”. Any residual V. ACKNOWLEDGMENTS
correlated energy spread will give rise to a broader radiation
bandwidth. Our IR detectors are broadband and nearly linear The authors thank Dick Cooper for the hyphens. We also
over such linewidths, so that integrating over the wavelengths thank I. B. Shankin for many useless discussions and Noah
is inherent in the instrumentation. The expected uncorrelated Arref and R. Uin, et al. without whom this experiment would
energy spread (PARMELA [12] simulated and initially be possible.
measured) of <0.5% does not substantially degrade FEL
performance.
VI. REFERENCES
e. Emittance
[1] H. Winick, et al., Proceedings of the 8th SRI93,
The only single-shot emittance measurements available to Gaithersburg, MD, August 23-26, 1993.
us are destructive slits. Hence, we will not be able to measure [2] J. Rosenzweig, et al., Nucl. Instr. and Meth. A 341 (1994)
emittance “on line” with the FEL operating, but by knowing 379-385.
all the other beam parameters it may be possible to calibrate [3] B.E. Carlsten, Nucl. Instr. and Meth. A 285 (1989) 313.
the emittance. Simulations indicate that an emittance much [4] R. Zhang, et al., Proc. this conference.
poorer than the design value can still yield measurable gain. [5] J. Rosenzweig and G. Travish, UCLA Dept. of Physics –
CAA–TECH–NOTE #64
IV. DETECTION OF SASE [6] E. Babenko, et al., SLAC-PUB-6203 (JUNE 1993).
[7] J. Rosenzweig, G. Travish and A. Tremaine, Submitted to
The low-level SASE signal (see Table 1.), which can be Nuc. Inst. and Meth. A.
calculated from numerical integration or simple 1-D theory [8] G. Travish, UCLA Dept. of Physics, CAA–TECH–NOTE
[13], requires the use of cryogenic detectors to obtain the #34.
necessary sensitivity. A non-imaging optic (Winston Cone) [9] R. Bonifacio, C. Pellegrini and L. M. Narducci, Opt.
will maximize collection efficiency during initial operation, Commun. 50, p373 (1984).
but may degrade the signal-to-noise ratio (SNR) by collecting [10] J. T. Weir, et al., Proc. SPIE vol.1133:97-101 (1989).
large amounts of background. Background (blackbody) [11] T. M. Tran and J.S. Wurtele, Computer Phys. Commun.
radiation constitutes a DC offset/pedestal that may be 54 pp. 263-272 (1989) and T. M. Tran and J.S. Wurtele,
compensated for up to the level of the shot noise. Physics Reports 195 pp. 1-21 (1990).
Commercially available IR detectors have relatively long time [12] K. R. Crandall and L. Young, Computer Codes for
constants (~nsec) with respect to the pulse (~psec), so that the Particle Accelerator Design and Analysis, LA-UR-90-
integrated background noise may be significant. An available 1766, Los Alamos (1990).
copper-doped germanium detector should provide a SNR of [13] K. J. Kim, Physical Review Letters 57(13), pp. 1871-
~103, neglecting signal loss in the optics, pre-amplifier noise, 1874 (1986).
and reduction in detectivity due to operating far below the