E. Allaria on behalf of the FERMI team Work partially supported by the Italian Ministry of University and Research under grants FIRB-RBAP045JF2 and FIRB-RBAP06AWK3 1
2 Outline Elettra and the FERMI FEL project FERMI parameters High Gain Harmonic Generation FEL experimental results at FERMI Coherence properties Spectral characterization Evidence of mb effects on FEL Conclusions
FERMI at the ELETTRA LABORATORY SINCROTRONE TRIESTE is a nonprofit shareholder company of national interest, established in 1987 to construct and manage synchrotron light sources as international facilities. FERMI@Elettra FEL: 100 4 nm HGHG, Sponsors: Italian Minister of University and Research (MIUR) Regione Auton. Friuli Venezia Giulia European Investment Bank (EIB) European Research Council (ERC) European Commission (EC) ELETTRA Synchrotron Light Source: up to 2.4 GeV, top-up mode.
4 FERMI main features FERMI@Elettra single-pass FEL user-facility. Two separate FEL amplifiers will cover the spectral range from 100 nm (12eV) to 4 nm (320 ev). The two FEL s will provide users with ~100fs photon pulses with unique characteristics. high peak power short temporal structure tunable wavelength variable polarization seeded harmonic cascade 0.3 GW s range sub-ps to 10 fs time scale APPLE II-type undulators horizontal/circular/vertical longitud. and transv. coherence COURTESY A. NELSON
FERMI Layout 5 Laser Heater X-band linac tunnel BC1 BC2 PI L1 L2 L3 L4 FEL1 undulator hall Transfer Line FEL2 PADReS Photon Beam Lines experimental hall FEL1 slits DIPROI FEL2 I/O mirrors & gas cells
6 FEL-1 and FEL-2 FERMI s two FELs will cover different spectral regions. FEL-1, based on a single stage high gain harmonic generations scheme initialized by a UV laser will cover the spectral range from ~100 nm down to 20nm. FEL-2, in order to be able to reach the wavelength range from 20 to ~4 nm starting from a seed laser in the UV, will be based on a double cascade of high gain harmonic generation. The nominal layout uses a magnetic electron delay line in order to improve the FEL performance by using the fresh bunch technique. Other FEL configurations are also possible in the future (e.g. EEHG).
7 HGHG mechanism Modulator Dispersive section Radiator
8 High Gain Harmonic Generation - HGHG HGHG scheme has been proposed as a way to partially solve the lack of seeding sources at short wavelengths. seed laser 5l modulator compressor radiator HGHG l planar Bunching at harmonic l APPLE II e-beam Compared to SASE devices, generally more compact and nearly full temporally coherence output; many spectral parameters more easily controlled (e.g., pulse length, chirp). L.H. Yu et al. PRL 91, 074801 (2003) After the initial HGHG demonstration experiment done at Brookhaven BNL, HGHG and Coherent Harmonic Generation (GHG) have been demonstrated and explored in other facilities (UVSOR-II(JP), Elettra SR-FEL(IT), Max-Lab FEL (SE), SPARC(IT), SDUV- FEL(CN), SLAC(USA)).
FEL gain and stability 9 With the FEL optimized we can measured an FEL growth along the radiators that can suggests exponential gain.the FEL gain has been measured both for circular and planar polarization and are in agreement with expectations (l g ~ 2,2.5 m). FEL growth is in good agreement with FEL simulations using the measured and inferred beam parameters. 32.5 nm, 450pC When the machine is stable the FEL stability is very good. FEL-1, 38 nm, 500pC, ~40min.
FEL mode and transverse coherence Quality of the FEL mode strongly depend on the undulator tuning. For a good resonance condition a nice Gaussian TEM00 mode is produced. As expected dimensions of the FEL mode depend on the FEL wavelength. 32.5 nm, 450pC Double slits Young s experiment have been done to characterize the transverse coherence. 32.5 nm, 450pC Quantitative analysis is ongoing. FEL at 32.5 nm, 6 radiators, 450pC, compression ~3. Slit separation = 0.8 mm, width = 20 mm
FEL bandwidth (ev) 11 Relative bandwidth of the FEL is smaller than the bandwidth of the seed laser. In the frequency (energy) domain the FEL spectrum is larger than the one of the seed laser. Since we expect the FEL pulse to be shorter than the seed laser the spectrum broadening does not necessary implies a degradation of the longitudinal coherence of the FEL pulse. s SEED rms = 4.7meV(0.098%) s FEL rms =14meV(0.038%) For the N th harmonic we expect in the ideal case the FEL pulse to be about a factor N shorter than the seed laser. For the 8 th harmonics this is in agreement with what has been measured in terms of bandwidth increase and suggests that longitudinal coherence is preserved.
FEL tuning 12 52 nm The fine FEL tuning around 52nm has been achieved by changing the seed laser wavelength of ~1 nm (0.4%). After tuning of the seed laser wavelength, the undulator resonance is changed accordingly to maximize the FEL power. Tuning the FEL in a large spectral range (30-60nm) done using the Optical Parametric Amplifier. The OPA is implemented in the control system and according to the users requirements seed laser wavelength and FEL harmonic number can be chosen according. Typical time needed for wavelength tuning was of the order of 10 minutes, much shorter for fine tuning. 58.7 nm 49.2 nm 39.8 nm 29.9 nm
FEL wavelength stability 13 In addition to the very narrow spectrum FERMI is characterized by excellent spectral stability. Both short and long term measurements show that the spectral peak jitter of less than 1 part in 10 4. Reported data refer to an electron beam of 350pC at 1.24GeV compressed about a factor 3. The 6 radiators are tuned to 32.5nm. FEL photon energy fluctuations fluctuations FEL bandwidth fluctuations fluctuations ~ 38.19eV = 1.1meV (RMS) = 3e-5 (RMS) = 22.5meV (RMS) = 5.9e-4 (RMS) = 3% (RMS)
HGHG with no e-beam chirp 14 Energy modulation of the phase space by the seed. Energy modulation converted into spatial modulation. Electron beam current strongly modulated at the seed wavelength, sharp spike indicate strong harmonic components. Spectral analysis of the bunching show strong harmonic components.
HGHG with linear e-beam chirp 15 Energy modulation and linear chirp Density modulation and compression Wavelength shift associated to the compression
HGHG with quadratic e-beam chirp 16 Energy modulation on the quadratic chirp Due to the nonlinear chirp different part of the beam suffer from different compression and wavelength shift. Spectrum broadening. Density modulation and compression vary along the bunch
Summary of chirp effects 17 Electron beam phase space has significant effects on the bunching spectrum and as a consequence on the output FEL spectrum. In case that microbunching instabilities in the LINAC perturb the electron beam phase space introducing nonlinearities and regions with different chirp this should be evident on the spectrum of the FEL.
e-beam predictions with start to end simulations at FERMI For the design beam (800pC 800A) e-beam extensive start to end simulations have been done to predict the electron beam properties at the FEL. In particular effort has been spent to study the ub. Calculated e-beam phase space Those structures at ~10 fs (3mm) although are weak are affecting the FEL and are evident on the spectrum of the FEL. Numerical simulation of FEL radiation Studies on the FEL spectra may also be useful to estimate the properties of the electron beam phase space
Periodic modulation of the e-beam energy (1/2) current (A) gamma Output power (W) The effect periodic structures on the e-beam phase space has been investigated with FEL numerical simulations looking at the FEL spectrum. The nominal (*) electron bunch of FERMI obtained with start to end numerical simulations (impactelegant) has been numerically manipulated in order to artificially create the energy modulation with different values of modulation strength and period. Analysis of the strength of the modulation Analysis of the cases with l mod =10 mm and modulation strength 100, 250, 500, 1000keV 1200 1000 800 600 400 200 2.0x10 9 1.5 1.0 0.5 100keV 250keV 500keV 1000keV 0 140 120 100-200fs -100 0 100 200 time 0.0 2236 2235 2234-400fs -200 0 200 400 time x10 6 80 2233 60 40 2232 (*) nominal is the e-beam used in CDR not the one currently used at FERMI 20 0 39.6 39.8 40.0 nm 40.2 40.4 2231 2230-200fs -100 0 100 200 time
Periodic modulation of the e-beam energy (2/2) power (W) gamma current (A) Analysis of the period of the modulation Analysis of the cases with DE=500keV and different modulation period: 5, 10, 50, 100 mm 2236 2235 2234 2233 2232 2231 5 mm 10 mm 50 mm 100 mm 1200 1000 800 600 400 200 2230-200fs -100 0 100 200 time 0-200fs -100 0 100 200 time 2.0x10 9 1.5 1.0 0.5 x10 6 60 50 40 30 20 10 0.0 0-400fs -200 0 200 400 time 39.6 39.8 40.0 nm 40.2 40.4
Experimental evidence of microbunching instabilities at FERMI? When FEL is optimized to maximize the power on the on axis mode the typical FEL spectrum has a single spike. When working with a single compressor scheme this is well reproducible and robust. BC1 BC1&BC2 BC1&BC2 After several shifts on FEL with BC1&BC2 it has been possible to have tens of mj from the FEL using the beam compressed with both BC1 and BC2. While the produced power was of the same order than what produced with the single compressor scheme, the measured spectrum showed a structure. Due to the limited time dedicated to the optimization, results can not be considered conclusive. FEL results seems to suggest that the microbunching is stronger for the beam compressed with BC1 and BC2 than with only BC1. More studies are needed. MAC meeting January 2012 21
22 E-beam structures in beam dump beam tail Depending on LINAC settings (charge, compression, optics), We often see detailed microstructure in images of the electron beam energy spectrum measured in the Main Beam Dump (after the FEL). These structures may suggest the presence of some mb, but. Since the seeding and FEL process are locally modifying the electron beam phase space it is possible to recognize the portion of the beam that is contributing to the FEL. In the reported case, representative of RUN10, the FEL was typically optimized seeding close to the spectrum peak. Beam tail Beam head beam head Local hole produced by the seeding
FEL spectrum changes FEL intensity (a.u) FEL wavelength (nm) 23 Beam head Beam tail In the usual working point the FEL has good stability, high intensity and a clean spectrum: l=32.45nm Dl=0.04nm (FWHM) Dl/l=0.12% (FWHM) Moving the seeding toward the head of the bunch we see first a degradation of the spectrum and a small wavelength shift. Resonance condition Further toward the head, the spectrum splits in two with a new emission that has a very different wavelength (red shifted) and noisy. Seed laser delay (ps)
24 Analysis of spiky spectrum In the case of the spikes in the spectrum, there continues to be one peak that is at close to the expected wavelength (harmonic of the seed laser) but the spikes at longer wavelength are not compatible with the measured spectrum of the seed laser. Single shot spectrum Dl=0.03nm (FWHM) Dl/l=0.1% (FWHM) Average over 100 shots spectra Dl=0.34nm (FWHM) Dl/l=1 % (FWHM) Dl=0.15nm (FWHM) Dl/l=0.5% (FWHM) The small wavelength shift of the main peak could be associated to a change of chirp in the electron beam when approaching the head of the bunch. The spikes at longer wavelength could be explained with a mb structure that affect the produced bunching.
FEL intensity (a.u) Undulator tuning effects on FEL spectra 25 Resonanc e condition Seed laser delay (ps) FEL spectra has been acquired as a function of the seed delay as shown before but for different tuning of the undulators. FEL spectra has been Setting the resonance at shorter wavelength does not induce clear changes in the FEL spectra. Nominal undulator seetings Opening undulator gaps Setting the resonance to longer wavelength allow to favorite the occurrence of the spike structure. Closing undulator gaps Closing undulator gaps Spike structure is favorite and present in the whole bunch when the resonance condition is set to the wavelength of the spikes.
mb or not? 26 FERMI FEL spectra can show structures and a significant wavelength shifts with respect to the exact seed-laser harmonic. Possible interpretation: Spectrum modifications are associated to e-beam phase-space nonlinearities (chirp, modulations). Spiky and noisy aspect of FEL spectra could be explained with electron beam distortion due to mb developed in the LINAC. NOT yet clear: The fact that the spikes can be generated along the entire bunch (also where the peak current is low). The fact that the wavelength shift associated with the spiky structure is discrete and asymmetric (only to the red) We are starting to prepare a program for dedicated studies on this.
27 Conclusions FERMI started producing high quality FEL pulses in the EUV Good stability Polarization control Good spectral properties Good coherence Easily tunable Experiments to exploit the FERMI radiations started on the three beamlines. FEL studies are ongoing at FERMI to characterize the HGHG and to define the limits for FERMI.
28 Success at FERMI has been the result of a concerted and unified effort by the entire FERMI team and the support staff at Sincrotrone Trieste. The physics commissioning team thanks all the people involved in the project (including consultants, guests and advisory committee members) that contributed to the design, construction and commissioning of FERMI over the the past 6 years. Special thanks to: P. Craievich, G. De Ninno, S. Di Mitri, W. Fawley, E. Ferrari, L. Froehlich, B. Mahieu, G. Penco, C. Spezzani, M. Trovo
Thank you 29