Transcript Slide 1

Advanced School on Synchrotron and Free Electron Laser Sources and their
Multidisciplinary Applications
Free Electron Lasers (II)
Enrico Allaria
[email protected]
Sincrotrone Trieste
Outline
-Basic concepts of light-electron interaction in a Free-Electron Laser
-Why a free electron laser
-How it work
- Different schemes for FEL
-FEL aplifier
-FEL oscillator
-Self Amplified Spontaneous Emission FEL (SASE)
-Coherent Harmonic Generation FEL (CHG)
- Application to the FERMI project at Elettra
- Recent experimental results on the Elettra storage ring FEL
The FERMI project at Elettra
(on behalf of the FERMI FEL group)
The FERMI project at Elettra
FERMI will be a photon source based on a multi-stage harmonic generation process.
It will be one of the first FEL user facilities in the world operating at wavelengths in the
VUV to soft X-ray range, based on the CHG scheme.
The project is making full use of the existing LINAC, previously
used for the electron filling of the Elettra storage ring.
The FEL design is based on a “start-to-end” approach.
This means that one has to keep track of the electron-beam dynamics and
preserve its quality from the gun, through the LINAC, up to the end of the
undulators chain.
This design and realization of a facility like FERMI relies on many different expertises:
Accelerator and laser physics, electron and light diagnostics, high level engineering, ...
Overview of the ELETTRA laboratory
Outline
•Presentation of the two FERMI FEL’s
•Numerical simulations for FEL 1 and FEL2
•Problem of FEL sensitivity to fluctuation of input
parameters
•Discussion and open issues
Harmonic generation: the principle (1/2)
l/n
l
e- beam from the Linac
Modulator
Dispersive section
From Perseo,
a numerical code by L. Giannessi
9104
Bunching
Bunchingcoefficie
coefficient
nt
Energy
Energy
(MeV)
9052
9000
895
2
4
890
0
150
2
100
3150
zP(nm
hase)
4200
5 250
6 300
0.6
0.4
0.4
0.2
0.2
00 1
1
22
33
44
Harmonic
Harmonic number
number
55
66
77
Harmonic generation: the principle (2/2)
l/n
l
e- beam from the Linac
Radiator
200
100
0
(um )
zz (mm)
Distance along the undulator
 GW output signal
100
Spectrum profile
200
Power Spectrum (a.u.)
Power (W)
109
s t~ 200 fs
Pow. Spec. (arb. units)
Evolution of the harmonic signal
units)
Power
Power(arb.
(arb. units)
Pulse profile
Dl/l
~ transform limited
89.2
89.3
89.4
89.5
89.6
89.7
89.8
wavelength (nm )
Wavelength
(nm)
89.9
90
FEL undulators
FERMI layout
Bunch
compressor
Bunch
compressor
Transport
line
FEL-1
FEL-2
Injector
Linear accelerator
FEL
Parameter
Medium bunch
Long Bunch
Beam energy
1.2 GeV
1.2 GeV
Peak current
0.8 KA
0.5 KA
Uncorrelated energy spread
200 KeV
150 KeV
Normalized Emittance
1.5 mm·mrad
1.5 mm·mrad
Bunch length
~ 0.6 ps
~ 1.4 ps
Two different ondulator lines for two different spectral regions:
•FEL-1 covers the spectral region between 100 and 40 nm
•FEL-2 the region between 40 and 10 nm
Both FEL’s are based on the Harmonic Generation scheme and make use of APPLE II type
ondulator for producing light with variable polarization
Total length of FEL-1 ~ 23 m
FEL-1 (100 - 40 nm)
l = l/n
Input seed laser
(240  l  360 nm)
Modulator
Parameter
Value
Type
Planar
Structure
One segment
Period
10 cm
K
Length
Radiator
Dispersive section
Parameter
Value
Radiator
Value
Type
Apple
Structure
~ 6 Segments
>5
Period
6.5 cm
3.04 m
K
2.4 - 4
Segment length
2.34 m
Break length
1.06 m
Total length
19.34 m
R56
Length
~ 32 mm
~1m
Electrons interact with an external laser field in the first
ondulator, the energy modulation produced by this interaction
is transformed into spatial modulation (bunching) to the laser
wavelength and to its harmonics.
Bunched electrons emit coherently into the radiator tuned to the desired harmonic and FEL process
is initiated.
A segmented radiator allows to insert electron optics and diagnostics between modules .
FEL-1 time independent simulations
Input parameters
Medium bunch
Beam energy
1.2 GeV
Peak current
0.8 KA
Uncorrelated energy
spread
200 KeV
Normalized
Emittance
1.5 mm·mrad
Seed power
100 MW
Within such approach we use ideal electron bunches whose
parameters are those predicted for the FERMI Linac.
Simulations show the possibility to reach saturation and
output power of several GW within 6 radiator segments for the
considered wavelength range (100-40nm).
Output power (GW)
4
3
2
FEL at 40 nm
FEL at 60 nm
1
0
0
5
z (m)
10
15
20
Energy spread (
8
6
4
2
0
5
z (m) 10
15
Bunching
20
0.6
0.4
0.2
0.0
0
5
z (m) 10
15
20
Medium electron beam at the end of the LINAC
Phase space
Parameter
Value
Beam energy
1.2 GeV
Peak current
0.8 KA
Uncorrelated energy spread
150 KeV
Normalized Emittance
1.5 mm·mrad
Bunch length (flat part)
~ 0.6 ps
mean energy (
2233.0
The analysis of the electron bunch shows the
presence of a cubic chirp in the electron-energy
profile with some noisy modulation period of the
order of 10 mm. Similar microbunching modulation
has been found also in the current profile.
2232.0
2231.0
Note the energy “chirp”
2230.0
-400 -200 0 200 400
time (fs)
-6
3.0x10
Emittance (mm mrad)
2000
800
0.7
1500
0.6
current (A)
energy spread (
0.8
0.5
700
1000
0.4
0.3
-300
300
500
0
0.2
-400 -200
0
200
time (fs)
400
-400 -200
0
200
time (fs)
400
2.5
2.0
1.5
1.0
0.5
0.0
-400 -200
0
time (fs)
200
400
FEL1 simulation using medium bunch (40nm)
2233.0
0.10
2233.0
2232.5
0.06
2231.5
0.04
2231.0
0.02
2230.5
0.00
2232.0
1.0
2231.5
2231.0
0.5
electron energy (
2232.0
2232.5
1.5
FEL output power (GW)
0.08
electron energy (
Input seeding power (GW)
100fs seed
2230.5
2230.0
-400
-200
0
200
time (fs)
400
0.0
2230.0
-400
-200
Pulse width (rms) = 66 fs
Photon number
= 6.5e+13
Bandwidth (%)
= 0.03
Ratio to Fourier limit = 1.9
Output spectrum
•The particle file has been simulated with the
optimized setup for FEL1 using a seed pulse of
100fs rms centred on the flat part of the bunch.
6
•This produces an output pulse of the order of 1.5 120x10
GW and less than 70 fs long.
100
•The cubic chirp and the microbunched structure in
electron energy profile are responsible for the
80
increase of the bandwidth with respect to the
60
Fourier limit.
0
200
time (fs)
400
40
20
0
39.8
39.9
40.0
40.1
Output wavelenght (nm)
40.2
FEL-2 (40 - 10 nm)
seed laser
l/n
l
Total length FEL-2 ~ 37.5 m
“fresh bunch” break
l/n
First Stage
l/(nm)
Second Stage
Radiator
Parameter
Value
Type
Apple
Structure
Segmented
Period
5 cm
Segment length
2.4 m
0.04
K
1.1 - 2.8
0.02
Break length
1.06 m
0.00
Total length
19.7 m
Fresh-bunch
first stage
Fresh-bunch
second
0.10
stage
0.08
0.06
-400
-200
0
200
400
•The electrons that are used for the first stage are not useful in the second stage because they
have a too large energy spread.
•A fresh bunch break is used in order to delay the electrons with respect of the photons
•In the second modulator the produced 40 nm radiation is superposed to a fresh part of the
electron bunch and a new harmonic generation is performed
FEL-2 time independent simulations
Beam parameters
Medium bunch
Beam energy
1.2 GeV
Peak current
0.5 KA
Uncorrelated energy spread
150 KeV
Normalized Emittance
1.5 mm·mrad
Bunch length
1.4 ps
250 rad1 power at 40 nm (MW)
1.2
200
1.0
Power (GW)
0.8
150
0.6
100
0.4
50
0.2
0
0.0
0
3.2
2
4z (m)6
8
rad1 energy spread (
10nm
20nm
0
10
2.8
z(m)
5
Energy spread (
10
15
20
4.0
3.0
2.4
0.4
•Simulations show the possibility of reaching
saturation within the 20 meter of radiator for the 20
nm case with more than 1 GW of output power.
•In the 10 nm case saturation is not reached within
the six modules however ~ 500MW are obtained.
2.0
0
2
4
6
8
10
rad1 bunching at 40 nm
0.4
0.3
0.2
0.1
0.0
0.3
0.2
0.1
0.0
0
2
4
6
8
10
10nm
20nm
0
Bunching
5
z (m) 10
15
20
10nm
20nm
0
5
z(m) 10
15
20
Long electron beam at the end of the LINAC
Parameter
Value
Beam energy
1.2 GeV
Peak current
0.5 KA
Uncorrelated energy spread
100 KeV
Normalized Emittance
1.5 mm·mrad
Bunch length
1.4 ps
Phase space
•The longitudinal phase space presents a quadratic
chirp and residual fast time fluctuations of the
mean energy.
•The useful part of the bunch is about 1.4 ps long
and presents an energy spread which is of the
order of 100keV. The current is of the order of 0.5
kA.
Note the energy “chirp”
0.8
energy spread (
mean energy (
2288
2287
2286
2285
2284
1200
1000
0.6
current (A)
2289
0.4
800
600
400
0.2
200
2283
0
0.0
-800
-400
0
time (fs)
400
800
-800
-400
0
time (fs)
400
800
-800
-400
0
time (fs)
400
800
FEL2 simulation using the long bunch (10nm)
2289
2288
1.0
2288
0.8
2287
2287
60
2286
40
2285
20
2284
250fs seed
2286
0.6
2285
0.4
2284
0.2
2283
2283
0
0.0
-800
-600
-400
-200
0
200
400
600
-800
800
-600
-400
-200
time (fs)
30x10
Output spectrum
•The length of the electron bunch allows the
use of a seed pulse of 250 fs rms placed on
the tail of the bunch.
•After the cascade the 10nm coherent
emission is produced from the head of the
bunch.
•Energy chirp and microbunching lead to a
broadening of the bandwidth.
0
time (fs)
200
400
600
800
6
25
20
15
10
5
0
9.97
Pulse width (rms)
Photon number
Bandwidth (%)
= 170 fs
= 2.2e+13
= 0.04
9.98
9.99
10.00
10.01
Output wavelenght (nm)
10.02
10.03
electron energy (
1.2
FEL output power (GW)
80
2289
electron energy (
Input seeding power (MW)
100
Jitter on initial conditions
Studies of output power sensitivity to input jitter
(e.g., energy, emittance, energy spread, peak current, seed power, beam offset and tilt)
Series of time independent simulation runs have been performed varying imput parameter of the
electron bunch as predicted from Gun and Linac studies.
Parameter
Shot-to-shot
variation (rms)
FEL 1 at 100 nm
Emittance
10 %
time independent simulations
Peak current
8%
Mean energy
0.1 %
Energy spread
10 %
Seed power
5%
e-beam axis offset
100 mm
2.8
e-beam tilt
10 mrad
2.6
33.0x10
GW9
Power (W)
Simulations show a critical sensitivity to the
Input mean energy is varied ONLY
All other parameters assumed constant
=> Global output power standard deviation: 9.6%
2.4
2.2
2.0
2 GW
electron mean energy responsible of strong
1.8
fluctuations of the output power
1.6
DE/E = 0.1 %
1.4
2342
2344
2346
2348 2350
Energy
2352
2354
2356
Simultaneous multi-parameter variation
Simultaneous variation of the following parameter has been considered :
energy, current, uncorrelated energy spread, transverse emittance, initial transverse
position and tilt
FEL 1 at 100 nm
FEL 2 at 40 nm
Energy variation projection
Output power global standard
deviation: 16.5%
Energy variation projection
Output power global standard
deviation: 33%
9
1.4x10
9
1.2
2.5
Power (W)
Power (W)
3.0x10
2.0
1.5
1.0
0.8
0.6
0.4
1.0
0.2
2342
2344
2346
2348 2350
Energy
2352
2354
2356
2346
2348
2350
Energy
2352
Simulation of 100 jittered electron bunches
•100 electron bunches have been propagated starting from the Gun through the Linac
considering possible noise sources (timing, phase and amplitude jitters)
•The study has been performed for the “Medium bunch” Linac configuration used for FEL1
•The central part has been considered as the useful one for the FEL process
Analysis of 100 jittered electron bunches
Start to end simulation confirm predictions for jitters
Quantity
Mean Value
Std. Dev.
Gamma
2231.89
0.09%
Current (A)
718
6.6%
Incoherent energy
spread
0.32987
19.5%
Normalized emittance
1.35
12.4%
FEL1 results with 100 jittered electron bunches
Results of FEL simulations with the start to end jittered files are in agreement with time
independent predictions and confirm the crucial dependence of output power on fluctuations of
electron mean energy. On the contrary, the central wavelength shows a very weak dependence
on input parameter fluctuations.
Quantity
Mean Value
Std. Dev.
Average pulse width
(fs)
73.2
Average photon
number
7.1e+13
23%
Average central
wavelength (nm)
40.002
0.013%
Average bandwidth
0.033%
Average Fourier factor
2.2
13%
The energy spread problem
FEL1 and FEL2 have been optimized for electron bunches with an incoherent energy spread of 100200keV. Larger energy spread can compromise the FEL performance.
In the case of FEL1 at 40 nm output power larger than 1GW is still possible for  lower than
450keV.
The sensitivity is dramatic in the case of FEL2, that for =300 show only 100MW at 10nm.
Performance is better using the medium bunch also for FEL2.
9
1.2x10
FEL(40nm)
'500MW'
'1GW'
'3GW'
2.0
1.5
1.0
0.5
1.0
0.8
0.6
0.4
0.2
250
300
350
400
Energy spread (keV)
450
500
200
3
10
250
300
350
400
Energy spread (keV)
450
500
300
350
400
Energy spread (keV)
450
500
9
6
Output power @ 10nm (W)
10
medium bunch
long bunch
'10MW'
'100MW'
'300MW'
9
0.0
200
Output power @ 40nm (W)
Output power @ 40nm (W)
2.5
Output power @ 10nm (W)
3.0x10
2
9
9
8
7
6
5
FEL1(40nm)
'500MW'
'1GW'
'3GW'
200
250
4
2
10
8
6
4
2
10
7
6
4
2
300
350
Energy spread (keV)
400
450
500
medium bunch
long bunch
'10MW'
'100MW'
'300MW'
200
250
CHG on the Elettra Storage
Ring FEL
(on behalf of the SR-FEL group - Sincrotrone Trieste )
CHG scheme on a storage ring
ELETTRA FEL Layout
Back mirror
Optical cavity – 32.4 m
Front mirror
Nanospetroscopy
beamline
Optical klystron – 4.6 m
Experimental
hutch
First CHG evidence on 29 April 2007
•
Seed @ 780nm  laser @ 260nm (3rd harmonic)
Ebeam = 0.75 GeV
Ibeam = 0.57 mA (single bunch)
Seed:
λ = 783 nm
pulse length ≈ 100 fs
pulse energy ≈ 2 mJ
rep. rate = 1KHz
CHG characterization
Considering the difference in the number of photon per pulse and taking into account the difference
between the pulse length of synchrotron radiation (~35 ps) and coherent signal (~120 fs), the ratio
between peak powers can be estimated to be of the order of 10^4
This corresponds to what one can expect from a qualitative calculation using the parameter of our
setup
Seed 780nm  CHG 390nm (2nd) & 260nm (3rd)
Beam instability at 0.75 GeV and timing jitter/drifts
prevent the optimization of the experimental parameters
Seed at 390nm  Ebeam = 1.1 Gev
– better stability
– from spectra we can estimate CHG gain with respect to
spontaneous emission
Spectral stability
a): Spectrum of the coherent emission at 132 nm (linear polarization). The integration
time is 1 ms; the spectrum is obtained after subtraction of the background due to
spontaneous emission. b): Spectrum of spontaneous and coherent emission for the
case in which the radiator is tuned at 203 nm, i.e., slightly mismatched with respect
to the second harmonic of the seed laser (198.5 nm).
Quadratic dependence of CHG on bunch current
Quadratic dependence of the coherent harmonic
at 132 nm vs. (normalized) bunch current. Dots
represent experimental data; the curve is a t
obtained using a quadratic function.
Test experiments
• Nanospectroscopy (Elettra):
– PEEM with gated detector
• CESYRA (CIMAINA/UniMi)
– TOF (mass spectrometry)
Nanospectroscopy
30
Seeded + Single bunch
(delay = 1050 ns)
Single Bunch
(delay = 200 ns)
intensity (a.u.)
25
20
gap1 = 23.56
gap2 = 31.4
phase1 = 0
phase2 = 32.8
energy: 9.1 eV
15
10
VB UPS on
Ag/W(110)
Imaging on SiO2
patterned sample
5
0
25
Difference (only Seeded)
20
15
3rd harmonic
λ = 133 nm
10
5
0
-2
0
2
4
6
8
kinetic energy (eV)
seeded
intensity (a.u.)
50
Seeded + Single Bunch
(delay = 1050 ns)
Single Bunch
(delay = 200 ns)
40
30
10
gap1 = 23.56
gap2 = 34.95
phase1 = 0
phase2 = 33.67
energy: 12.16 eV
0
50
20
4th harmonic
λ = 99.5 nm
40
30
Difference (only seeded)
20
10
0
2
4
6
8
kinetic energy (eV)
10
12
spontaneous emission
Comparison between CHG and NHG
•
Sketch of the experimental setup used for the investigation of harmonic generation in a
FEL. A powerful Ti:Saphire laser interacts with the electron bunch (e) within the
modulator and induces a modulation of the electrons' energy. After the conversion of
the energy modulation into spatial bunching, which occur into the magnetic chicane
(R56), the bunch enters the radiator and start emitting coherently at the resonant
wavelength and, eventually, at its harmonics. The produced coherent harmonic
radiation passes through a diaphragm (D) and is transported into a diagnostic area,
where temporal (PMT) and spectral (CCD) analyses are performed. The position of the
diaphragm de¯nes the angle of emission with respect to the undulator's axis
considered for the measurement.
CHG and NHG
Comparison between the third harmonic radiation produced in CHG (a) and NHG (b) configurations. In both
cases the modulator and the seed laser are in horizontal polarization. The radiator is tuned to the third
harmonic in horizontal polarization (a) and to the fundamental in horizontal polarization (b).
CHG and NHG
• Comparison of the harmonic radiation produced in CHG
configuration at the second (a,b) and third (c,d)
harmonics of the seed wavelength. The radiator is set in
horizontal (a,c) or in circular (b,d) polarization, while the
modulator and the seed are in horizontal polarization.
CHG and NHG
•
Coherent harmonic signals produced at the second harmonic of the seed wavelength in CHG
(a,b) and NHG (c,d) configurations. Figures (a,c) refer to a condition where the seed laser, the
modulator and the radiator are in planar polarization, while Figs.(b,d) refer to a condition where
both the seed and all undulators are set in circular polarization. Data reported in Figs.(a) and (c)
refer to the same experimental conditions, and can be used for a relative comparison. The
same holds for Figs.(b) and(d).
CHG and NHG
•
Measured angular distribution of the second harmonic in the case of CHG (squares)
and NHG (dots) with helical undulators. Measurements are well fitted by theoretical
curves,which have been obtained by integrating the expected Gaussian profile
(dashed line, CHG case) and the profile predicted in [9] (continuous line, NHG case),
over an angle of 0.09 mrad.
Perspectives for SR-CHG
• Seed with Ti:Sa 3rd harmonic (260nm)
• CHG down to 87nm (14.3 eV)
• Pump and probe beamline for time resolved
experiments
• Compatibility with normal operation mode at 2.0 GeV
(non symmetric SR filling)
People
SR-FEL Group
G.De Ninno, F.Curbis, E.Allaria, M.Trovò, L.Romanzin, M.Coreno,
E.Karantzoulis, C.Spezzani
Machine
Linac and Elettra operators
Laser
M.B.Danailov, A.Demidovich, R.K.Ivanov
Synchronization
P.Sigalotti
A.Carniel, F.Rossi, M.Ferianis,
Experiments
A.Locatelli, O.Mentes, M.A.Nino, R.Sergo, M.Pittana, G.Cautero
P.Piseri, G.A.Bongiorno, M.Amati, O.Nicoletti, L.Ravagnan, P.Milani