First lasing of the ELBE MID-IR FEL

Abstract

First lasing of the mid-infrared free-electron laser at ELBE was achieved on May 7, 2004. The Radiation Source ELBE at the Forschungszentrum Rossendorf in Dresden is currently under transition from commissioning to regular user operation. Presently the electron linac produces an up to 18 MeV, 1 mA (cw) electron beam which is alotted to generate various kinds of secondary radiation. After the successful commissioning of the bremsstrahlung and channeling-X-ray facilities during 2003 stable lasing has now been observed in the IR range (15 to 22 μm). The oscillator FEL is equipped with two planar undulator units, both consisting of 34 hybrid permanent magnets with a period of 27.3 mm (Krms = 0.3...0.8). The distance between the two parts is variable and the gaps can be adjusted and tapered independently. At 19.6 μm an optical power of 3 W was outcoupled in a macro pulse of 0.6 ms duration using an electron beam energy of 16.1 MeV and an energy spread of less than 100 keV. The micropulse charge was 50 pC and its width slightly above 1 ps. With the installation of a second acceleration module for additional 20 MeV shorter wavelengths will become available in the near future.

Full text

Page 1

FIRST LASING OF THE ELBE MID-IR FEL
P.Michel, F.Gabriel, E.Grosse, P.Evtushenko, T.Dekorsy, M.Krenz, M.Helm,
U.Lehnert, W.Seidel, R.W¨ unsch, D.Wohlfarth, A.Wolf
Forschungszentrum Rossendorf, Germany
Abstract
First lasing of the mid-infrared free-electron laser at
ELBEwasachievedonMay7, 2004. TheRadiationSource
ELBE at the Forschungszentrum Rossendorf in Dresden is
currently under transition from commissioning to regular
user operation. Presently the electron linac produces an
up to 18 MeV, 1 mA (cw) electron beam which is alot-
ted to generate various kinds of secondary radiation. After
the successful commissioning of the bremsstrahlung and
channeling-X-ray facilities during 2003 stable lasing has
now been observed in the IR range (15 to 22 µm). The os-
cillator FEL is equipped with two planar undulator units,
both consisting of 34 hybrid permanent magnets with a pe-
riod of 27.3 mm (Krms = 0.3...0.8). The distance be-
tween the two parts is variable and the gaps can be adjusted
and tapered independently. At 19.6 µm an optical power of
3 W was outcoupled in a macro pulse of 0.6 ms duration
using an electron beam energy of 16.1 MeV and an en-
ergy spread of less than 100 keV. The micropulse charge
was 50 pC and its width slightly above 1 ps. With the in-
stallation of a second acceleration module for additional
20 MeV shorter wavelengths will become available in the
near future.
INTRODUCTION
At Forschungszentrum Rossendorf a superconducting
Electron Linac with high Brilliance and low Emittance
(ELBE) has been constructed which can deliver a 1 mA
cw beam at 40 MeV [1]. The electron beam is used to
generate infraredlight(FreeElectronLasers), X-rays(elec-
tron channelling), MeV-Bremsstrahlung, fast neutrons and
positrons. Table 1 gives an overview of the secondary
beams at the ELBE facility and the associated fields of sci-
ence.
OVERVIEW OF THE ELBE FEL
The ELBE accelerator is fed by a grid-pulsed thermionic
gun operating at 250 keV. The gun can deliver 450 ps long
pulses and bunch charges up to 77 pC at 13 MHz or 4 pC
at 260 MHz. A macro pulser chops the electron beam with
adjustable duty cycle and allows to generate a very flexible
time structure. By means of two RF buncher cavities oper-
ating at 260 MHz and 1.3 GHz the pulses are compressed
down to 10 ps upon injection into the first accelerator mod-
ule.
The Linac uses standing wave RF cavities (1.3 GHz) de-
signed for the TESLA test facility at DESY [2]. Two 9-cell
Table 1: Secondary radiation sources at ELBE and their
scientific application
0...20 MeV
γ-radiation -astrophysics
-radiation damage of cells
-study of phase transitions in
liquid metals
-semiconductor physics
-radiochemical and biologi-
cal experiments
-materials studies for fusion
reactors
0...30 keV positrons
-defect studies in solids
-nuclear physics
10...100 keV X-rays
5...150 µm
infrared FEL
0...30 MeV neutrons
superconducting niobium cavities are contained in a cry-
omodule cooled with superfluid helium at about 2 K. Each
cavity has its own RF coupler and is driven by a 10 kW
CPI klystron amplifier. The standard accelerating gradient
amounts to 10 MeV/m and the beam energy after acceler-
ation is 20 MeV. Downstream the first cryomodule a mag-
netic chicane is used for bunch length variation. After the
installation of the second LINAC module in near future the
full energy of ELBE of 40 MeV will be available.
Infrared radiation in the 5–25 µm range will be produced
at ELBE in the undulator U27 [3]. It consists of two 34-
pole sections with a length of 0.98 m each. The undula-
tor structure has a period of λu=27.3 mm and consists of
NdFeB permanent magnets and poles of decarburized iron
(hybride type; the units were test modules for the TTF-
Facility at DESY [4] and are modified for use as a “pas-
sive” undulator). The sections are mounted on carriages
(delivered by DANFYSIK) such that the distance between
the two sections is adjustable for phase-matching. The
gaps (minimum=13.8 mm, corresponding to Krms=0.70)
of both sections can be varied independently. For high-gain
lasing it is possible to introduce a taper of the field. Both
sections were scanned and adjusted using a calibrated Hall
probe setup at DESY. After installation at ELBE the field
distribution was checked using the pulsed wire method [5].
The inner dimension of 10x34 mm2of the stainless steel
vacuum chamber in the undulator required a well designed
resonator. The cavity requirements are summarized in the
following Table 2.
To optimize the extraction ratio over the whole wave-
length range we use 5 mirrors with different hole sizes in
8
P. Michel et al. / Proceedings of the 2004 FEL Conference, 8-13
MOAIS04 Available online at http://www.JACoW.org

Page 2

X-rays
Positron
Production
Neutron TOF
Experiments
Brems-
strahlung
Free Electron Laser (IR)
Neutrons
chicane
Accelerator Hall
RF-generators
Accelerator
Electronics
Experiment
Control
IR Laboratories
X-raysPosi-
trons
10 m
Figure 1: Layout of the radiation source ELBE.
Table 2: Design parameters of the U27 resonator
resonator type stable, near concentric, sym-
metric
cavity length 11.53 m, stabilized < 0.5 µm
tilt stability
< 6 µrad
Rayleigh range1 m
mirror diameter7.5 cm
radii of curvature5.94 m
g2
0.88
diameterof the
outcoupling holes
1.5, 2.0, 3.0 and 4.5 mm
the upstream mirror chamber. The Au-coated Cu-mirrors
are mounted on a revolvable holder (wheel), which is fixed
to a high-precision rotational stage. Angular adjustment
of the mirror wheel is performed using piezoelectric inch-
worm UHV motors, which provide both coarse and fine
adjustments. A similar construction with 3 mirrors of dif-
ferent curvature is used in the downstream chamber. Here,
the angular as well as longitudinal adjustment are designed
with micrometers and a flexible bar for fine tuning driven
by DC-motors outside the vacuum. The translation stage of
thismirrorwillbe usedtoadjustthe cavitylength to<5µm
accuracy.
To ensure the stability of the resonator at wavelengths
down to 3 µm we require the mirror angular adjustment to
have a resolution and stability in the order of 6 µrad. For
the initial alignment of the mirror angles an accuracy in the
order of 20 µrad is required. To achieve this accuracy we
built an alignment system consisting of two collinear He-
Ne lasers using insertable adjustment apertures inside the
cavity.
PC
DC-Mot.
setting? L ? cavity detuning
stabilization? (L + ? L) < ? 0.5 ? m
laser
splitter
70%
30%
undulator
outcoupling mirror
mirror
Figure 2: Schematic view of the resonator length control
system. Its time constant is one Hz and hence fast in com-
parison to the thermal time constant of the resonator.
A Hewlett-Packard interferometer system is used to
monitor and stabilize the resonator length (see Fig. 2).
The interferometer beam is split into two beams (70% and
30%). The low-intensity beam monitors the position of the
upstream mirror using one of five retro reflectors installed
adjacent to each outcoupling mirror. The high intensity
beam passes through the same resonator chamber as the
main laser and the electron beam. However, constraints
on the width of the vacuum chamber do not leave enough
space in the cavity for a separate parallel interferometer
beam. Therefore, the latter will pass diagonally from one
side of the upstream cavity mirror to a retro reflector on the
other side of the downstream mirror. The control electron-
ics for the two interferometer arms include a servo system
to control and stabilize the relative distance between the
two cavity mirrors using the motorized micrometer drive
on the translation stage of the downstream chamber. There
is no active tilt stabilization.
Estimating the maximum intracavity laser power, up to
15 W (cw regime) can be absorbed in the mirrors despite
P. Michel et al. / Proceedings of the 2004 FEL Conference, 8-13
9
FEL Prize Talk and New Lasing

Page 3

their high reflectivity of more than 99% . To stabilize the
mirror wheel temperature we installed a heater in the center
of the wheel. Independently of whether the laser is working
or not all components are at the same equilibrium temper-
ature slightly above the expected saturation temperature.
The mirror wheel is made of Cu to reduce mechanical ten-
sion between the mirrors and the surrounding material but
thermally isolated from the high-precision rotational stage.
The heat is dissipated directly to the outside of the vacuum
chamber (Peltier element or air cooling).
Behind the out coupling hole the divergent IR beam
passes through a CVD diamond vacuum window (thick-
ness 320 µm; useable aperture 8 mm) mounted at Brewster
angle. The adjacent optical transport system [6] guides the
beam to the diagnostic station using 4 (3 toroidal and 1
flat) gold plated copper mirrors. The optics of the system
was aligned by monitoring the spatial intensity profiles of a
He-Ne guide laser which in turn is aligned to the resonator
axis. The same laser will be used for indicating the position
of the IR beam in each user laboratory with an accuracy
better than 200 µm. Therefore, all optical components of
the transport system have to be transparent for IR radiation
and for 632 nm as well. Spot size and position of the waist
at the diagnostic table are independent of the wavelength.
Linear polarization is conserved. The transport system and
the diagnostic station [7] both are purged with dry nitrogen
to avoid absorption in air. From the main beam, approxi-
mately 10...40 percent of the total power are extracted by
different beam splitters for wavelength measurement and
power monitoring.
FEL COMMISSIONING
Electron beam parameters measurements and di-
agnostics
One of the first steps in the FEL commissioning was the
electron beam characterization done at a beam energy of
16 MeV. The transverse emittance was measured in the in-
jector with the multislit method, while the emittance of the
accelerated beam was measured with the quadrupole scan.
At the maximum design bunch charge of 77 pC the emit-
tance is measured to be 8 mm mrad. The gain reduction
factor due to the finite emittance was 0.97 at first lasing,
i.e., is almost negligible.
Since the FEL gain is linearly proportional to the beam
peak current it is highly desirable to minimise the electron
bunch length in the vicinity of the undulator. Previously,
the bunch length was measured to 1.5 ps (rms) immediately
at the accelerator exit using a Martin-Puplett interferometer
(MPI). At the undulator with the beam tuned for lasing this
value is however different. One can groupe the beam line
elements, which influence the bunch length, between the
accelerator exit and the undulator in three groups. These
are the magnetic chicane, the “S”–shaped part of the beam
line and drift spaces. The chicane can be used to adjust the
R56of the FEL beam line. For the FEL commissioning
Figure 3: Electron beam parameters dependence on the
second cavity phase (dots show the cavity gradient needed
for constant energy, squares the energy spread and the line
the bunchlength signal).
the MPI was installed right downstream of the undulator.
Additionally a single Golay cell detectors was installed up-
stream of the “S” shaped part of the beam line to measure
the total power of the coherent transition radiation, which
is in the first approximation inversely proportional to the
bunch length. Note that the two Golay cells installed at
the interferometer can be used in the same way for the
bunch length minimisation without scanning the interfer-
ometer. One of the key elements to tune the longitudinal
phase space is the second accelerator cavity. Adjusting the
cavity phase one changes the electron beam energy spread
and the bunch length in the undulator vicinity because of
the non-zero R56of the beam line. For that reason the en-
ergy spread and the Golay cell signals were measured as a
function of the second cavity phase. Fig. 3 shows results of
the measurements. The bunch length at its minimum can
be measured with the help of the interferometer. Here the
FWHM of the interferogram is about of 2.5 ps.
The most important observation to note is the follow-
ing: adjusting the cavity phase to have minimum energy
spread drastically increases the bunch length in the undula-
tor . In fact at the energy spread minimum the bunch length
is so long that it cannot be measured with the MPI. It turnes
out, however, that for lasing at the used (rather long) wave-
length it is more critical to minimize the energy spread of
the beam than to minimize the bunchlength.
The electron beam profile and position in the undulator
are measured with the help of OTR view screens. They
are made of beryllium and have a prism shape. The prism
has a 1 mm diameter hole with is precisely aligned to the
magnetic axis of the undulator. The view screens are used
for the optical cavity alignment as well, which is important
to ensure the overlap between the electron beam and the
optical mode. It is noteworthy that the view screens are ex-
tremely long-term reproducible and reliable, which means
that every time an inserted view screen takes the same po-
10
P. Michel et al. / Proceedings of the 2004 FEL Conference, 8-13
MOAIS04

Page 4

sition with a 10 µm accuracy.
ELBE is equipped with a strip-line beam position mon-
itor (BPM) system. The resolution of the system is about
10 µm. There are two phenomena, which make this sys-
tem very useful during the FEL operation. First, there is
an energy drift observed for the first 2–3 hours every time
the linac is switched on. A BPM located in a dispersive
region is used to monitor the electron beam energy and to
compensate the drift. The second phenomenon is the de-
pendence of the R56of the “S” shaped beam line on the
electron beam path through it. Both phenomena are to be
investigated more detailed in the future.
Observation of the spontaneous radiation
The general idea for the first FEL turn–on was to observe
the spontaneous undulator radiation and to maximize it by
systematic adjustment of the optical cavity and the elec-
tron beam parameters. First, the spontaneous radiation was
observed downstream of the undulator so that the optical
cavity was not incorporated in the measurements. For that
purpose a mirror was inserted in to the beam line behind
the last dipole deflecting the beam to the dump. The spon-
taneous radiation was outcoupled off the beam line through
a KRS–5 window and focused by a parabolic mirror on a
liquid nitrogen cooled MCT detector. An accurate align-
ment of the setup was essential for the spontaneous radia-
tion measurements. For the first observation of the sponta-
neous radiation we had to use an extremely strong averag-
ing of the data, however, that was an important step for the
commissioning, since once the spontaneous radiation was
observed we could optimize the machine using it as a tune
signal.
Setting the optical cavity length
The adjustment of the FEL cavity to the correct length
is an important prerequisite for the achievement of lasing.
The cavity length of the FEL has been determined by em-
ploying an external frequency stabilized fs mode-locked
Ti:sapphire laser (Femtolasers, Austria) [8]. The fs laser
is operated at 78.0 MHz, i.e. the 6thharmonic of the FEL.
A 390 MHz reference signal is derived from the RF elec-
tronics of the gun, which is used for stabilizing the repe-
tition rate of the fs laser with a phase-lock loop at its 5th
harmonic. This synchronization scheme reduces the timing
jitter of the fs laser to 500 fs. The pulse train of the fs laser
operating at 800 nm with 15 fs pulse duration is directed
through the outcoupling hole into the FEL cavity. The light
re-emitted through the outcoupling hole is detected via a
beam splitter and a fast photodiode. When perfect syn-
chronism of the fs laser and the FEL cavity is achieved,
the detected optical pulse is enhanced due to constructive
superposition of pulses circulating in the cavity. This re-
sults in an increase of the detected pulse intensity by a fac-
tor of five. The correct cavity length is determined by this
method with an accuracy of some µm, i.e. a relative accu-
racy of 10−7. Since the expected FEL operation covers a
Figure 4: FEL spectra measured at different optical cavity
detuning.
cavity detuning range of several 10 µm, this accuracy was
sufficient to start lasing at the preset cavity length.
FIRST RESULTS
First lasing of the mid-infrared free-electron laser at
ELBE was achieved on May 7, 2004. At 19.6 µm an op-
tical power of 3 W was outcoupled using an macro-pulsed
electron beam with anenergy of 16.1 MeV and an energy
spread of less than 100 keV. The bunch charge charge was
50 pC.
The optical spectrum of the FEL
The FEL spectra were measured with a Czerny-Turner
type spectrometer (SpectraPro-300i from ARC) which con-
tains a turret with three different gratings (75 l/mm, blazed
at 8 µm; 60 l/mm, blazed at 15 µm; 30 l/mm, blazed at
30 µm). For these measurements we used the side exit slit
equipped with a single MCT detector. Fig. 4 shows how
the spectral width decreases with the detuning of the cavity
length.
Detuning curves
Up to now we operate the FEL in a pulsed mode only us-
ing an MCT detector to measure the FEL power as a func-
tion of time. In that mode the small signal gain is measured
by fitting an exponential function to the rising slope of the
MCT signal in its very beginning. The amplitude of the
MCT detector close to the macropulse end is associated
with the saturated power. The saturation power as well as
the small signal gain are measured as a function of the op-
tical cavity detuning. Typical results of such measurements
are shown in Fig. 5.
Such measurements are made in the beginning of every
FEL run, since the active cavity length stabilization is not
yet commissioned and the cavity length may change be-
tween the runs. The measured detuning curve appears in
P. Michel et al. / Proceedings of the 2004 FEL Conference, 8-13
11
FEL Prize Talk and New Lasing

Page 5

Figure 5: The saturation power (line) and the FEL net gain
(dots) vs. optical cavity detuning.
correspondence with the shape predicted by theory. It is
also clearly observed that the detuning which optimizes the
FEL gain differs from the detuning that optimizes the satu-
rated power, which is also expected according to the theory.
Optical cavity loss measurements
The optical cavity losses can be measured similar to the
small signal gain. At the end of the macropulse the electron
beam is turned off instantaneously. Then, the characteristic
decay time of the MCT signal carries the information on
the cavity losses, which includes diffraction losses at the
undulator vacuum chamber as well as the losses on the mir-
rors and the outcoupled beam. Thus, the total cavity losses
were measured as a function of the FEL wavelength. The
losses were also calculated using numerical code GLAD.
The measurements are in reasonable agreement with the
calculations. However, one has to note that the calculations
are also limited in accuracy, probably of the same order of
magnitude as the measurement accuracy. In the code there
is no element like a tube which could simulate the vacuum
chamber. For that reason a set of apertures was used to that
purpose, which also causes some systematic error. More
detailed numerical calculations of the cavity losses are in
progress. Results of both the measurements and the calcu-
lations are shown in Fig. 6.
Electron beam energy spectrum
After the electron beam passes the undulator it is de-
flectedtothebeamdumpbyadipole. Thereisaquadrupole
doublet between the dipole and the dump, which is nor-
mally adjusted to have maximum transmission of the elec-
tron beam to the dump. However, it can also be also ad-
justed to image the electron beam energy spectrum on a
view screen. Fig. 7 shows both the electron beam energy
spectra when the FEL is off and when the FEL turns on.
An increase of the energy spread is observed as predicted
by Madey’s second theorem. The change of the electron
Figure 6: Optical cavity losses.
05 10152025
0.0
0.2
0.4
0.6
0.8
1.0
Intensity,a.u.
X,mm
FEL off
FEL on
Figure 7: Electron beam spectrum change when the FEL
turns on.
beam mean energy is measured as well, which allows one
to estimate the amount of energy transferred from the elec-
tron beam to the optical beam.
Autocorrelation IR pulse length measurements
To characterize the ultrashort pulses generated by the
FEL we built a non-collinear background-free autocorre-
lator. As SHG medium we use a CdTe crystal [9]. CdTe is
transparent for a wide wavelength range in the FIR, thus, a
good candidate for SHG of the FEL radiation.
In Fig. 8 we present the autocorrelation function mea-
sured at the maximum-power point of the detuning curve.
From the measured autocorrelation we calculate a pulse
duration of 2.1 ps, assuming a Gaussian temporal pulse
shape. The FWHM of the spectrum is approx. 220 nm.
The calculated time-bandwidth product is 0.46 which indi-
cates Fourier-transform limited operation.
12
P. Michel et al. / Proceedings of the 2004 FEL Conference, 8-13
MOAIS04