X-ray Free Electron lasers Zhirong Huang Lecture Outline XFEL basics XFEL projects and R&D areas Questions and Answers Bright light sources from relativistic electrons Synchrotron radiation Undulator radiation Electrons emit with random phase radiation intensity N (g is Lorentz factor, N is number of electrons ~109) Free Electron Laser (FEL) • Produced by resonant interaction of a relativistic electron beam with EM radiation in an undulator electron beam photon beam undulator e- beam dump l1 • Radiation intensity N2 • Tunable, Powerful, Coherent radiation sources Three FEL modes Light Source Brightness (Brilliance) 10 orders of magnitude! Undulator Radiation l1 lu forward direction radiation (and harmonics) undulator parameter K = 0.94 B[Tesla] lu[cm] LCLS undulator K = 3.5, lu = 3 cm, e-beam energy from 3 GeV to 15 GeV to cover l1 = 30 Å to 1.2 Å Can energy be exchanged between electrons and copropagating radiation pulse? Resonant Interaction of Field with Electrons Electrons slip behind EM wave by l1 per undulator period (lu) + + + + x lu K/g e- vxEx > 0 vxEx > 0 l1 vxEx > 0 vxEx > 0 x-ray z vxEx > 0 + + + Due to sustained interaction, some electrons lose energy, while others gain energy modulation at l1 E P. Emma t e- losing energy slow down, and e- gaining energy catch up density modulation at l1 (microbunching) E Microbunched beam radiates coherently at l1, enhancing the process exponential growth of radiation power t FEL Micro-Bunching Along Undulator electron beam photon beam undulator e- beam dump S. Reiche log (radiation power) distance What is SASE? Shot noise originates from discrete nature of electrons … .. . .. SASE Dz electron arrival time t is random spontaneous emission amplified by FEL interaction quasi-coherent x-rays SASE FEL Electron Beam Requirements radiation wavelength (e.g., 1 Å) transverse emittance: eN < 1 µm at 1 Å, 15 GeV FEL parameter peak current undulator period relative energy spread: <0.04% at Ipk = 3 kA, K 3, lu 3 cm, … beta function undulator ‘field’ = 0.93∙Blu FEL gain length: 18LG ≈ 100 m for eN 1.5 µm We must increase peak current, preserve emittance, and maintain small energy spread so that power grows exponentially with undulator distance, z, P(z) = P0 ∙exp(z/LG) FEL power reaches saturation at ~18LG SASE performance depends exponentially on e- beam quality ! (challenge) Slippage leads to coherence length and spiky structure Due to resonant condition, light overtakes e- beam by one radiation wavelength l1 per undulator period (interaction length = undulator length) + + + + - z x-rays e- ~1 µm Nl1 + + + Slippage length = l1 × N undulator periods: (at 1.5 Å, LCLS slippage length is: ls ≈ 1.5 fs << 100-fs pulse length) P. Emma Each part of optical pulse is amplified by those electrons within a slippage slippage length (an FEL slice) Coherence length is slippage over ~2LG (lc ≈ ls/10) ML ≈ Dz/lc independent radiation sources (modes) length Dz SASE temporal characteristics • E(t)=j E0(t-tj), tj is the random arrival time of jth eNu l E0: wave packet of a single e- l lc 2c • lc ~ 500 l1 = 200 as (LCLS 1.5 Å) • Sum of all packets E(t) bunch length Dz Statistical intensity fluctuation determined by number of longitudinal modes Due to noise start-up, SASE is chaotic light with ML coherent modes (i.e., spikes in intensity profile): z = 50 m temporal spikes appear bunch length Dz ML coherence length lc Longitudinal phase space is ML larger than Fourier Transform limit SASE energy fluctuation is… DW 1 W M ← 50 % of X-Ray Pulse Length → L ML is not constant – reduced by increased coherence during exponential growth, and increased with reduced coherence after saturation LCLS near saturation (~50 fs): ML ≈ 200 DW/W ≈ 7 % FEL startup from e- beam noise ~10 kW ~1 MW ~0.1 GW spiky temporal structure ~10 GW All vertical axes are log scale BW = 0.6% BW = 0.15% BW = 0.10% narrow bandwidth BW = 0.08% FEL Bandwidth set by FEL Parameter, (~10-3) LCLS spectrum spike width ~ l1/(2Dz) Bandwidth ~ 2 Example, LCLS relative spectral spike width: Dz = 50 fs bunch length: width = 5×10-6 Dz = 5 fs bunch length: width = 5×10-5 Dz = 0.5 fs bunch length: width = 5×10-4 Spectral properties are similar to temporal domain, except that everything is inverted… SASE 1D Summary Power gain length: 3.5 m Exponential growth: P(z) = P0 exp(z/LG) Startup noise power: P0 ≈ 2g mc3/l1 1.5 kW (spontaneous radiation in two gain lengths) Saturation power: Psat ≈ × e-beam power 20 GW Saturation length: Lsat ≈ lu/ ≈ 18LG 60 m FWHM bandwidth at saturation: ≈ 2 0.1% Coherence length at saturation: lc ≈ l1/() 0.2 fs Transverse coherence Z=25 m Z=37.5 m Z=62.5 m Z=75 m S. Reiche Z=50 m Z=87.5 mm Single mode dominates close to 100% transverse coherence Harmonic Radiation also Generated: ln l1/n FEL gain creates e- energy and density modulation at l1 Near saturation, strong bunching at fundamental wavelength also produces rich harmonics For example, ~1% of fundamental power in 3rd harmonic l E l t linear regime, before saturation E t non-linear regime, near saturation l1 LCLS may produce up to 25 keV in 3rd harmonic photons at ~100 MW l3 Peak Brightness Enhancement From Storage Ring Light Sources To SASE Undulator in SR # of photons ΩxΩy ΩZ B Ne (2πex) (2πey) D Z 10 -3 10 ps c 1023 SASE NeNlc (l1 2)2 D Z 10 -3 100 fs c compressed 1033 Nlc: number of electrons within a coherence length lc Enhancement Factor 6 7 Nlc~10 to 10 102 102 1010 to 1011 XFEL accelerator system emittance corrector Linac rf photocathode gun Linac Linac SASE Undulator Pulse compressors • Photocathode rf gun exn ~ 1 m m, Ip ~ 100A • Bunch compression Ip ~ 2-5 kA, Dt ~ 1-100 fs • Acceleration 3–20 GeV, l ~ lu/(2g2) adiabatic damping • ex ~ exn/g ~ l/4, g/g < ~ 10-3 Undulator 100-m long, segmented, a few mm tolerance Projects undertaken at US, Germany, Japan, Korea, Swiss, Italy… Linac Coherent Light Source (LCLS) at SLAC X-FEL based on last 1-km of existing 3-km linac 1.5-15 Å (14-4.3 GeV) Proposed by C. Pellegrini in 1992 Injector (35º) at 2-km point Existing 1/3 Linac (1 km) (with modifications) New e- Transfer Line (340 m) X-ray Transport Line (200 m) Undulator (130 m) Near Experiment Hall Far Experiment Hall LCLS: world’s first hard x-ray FEL 1.5 Å SASE wavelength range: 30 – 1.2 Å Photon energy range: 0.4 - 10 keV Pulse length FWHM 5 – 100 fs (5- 500 fs for SXR only) Pulse energy up to 4 mJ ~95% accelerator availability Smaller charge, shorter x-rays L0 TCAV0 heater 3 wires 3 OTR L1X 3 wires 2 OTR 3 OTR z1 L2-linac BC1 stopper z2 4 wire scanners old screen L1S DL1 L3-linac TCAV3 4 wire scanners + 4 collimators m wall gun BSY DL2 vert. dump undulator BL signal FEL signal 20 pC, photon energy @ 840 eV Low charge mode developed by J. Frisch et al. Simulations* suggest a few fs electron and x-ray pulse duration. A 3-pC bunch is capable of generating attosecond FEL, but diagnostics is very challenging * Y. Ding et. al, PRL 2009 More to come Spring-8 SACLA 2011 SASE Wavelength range: 3 – 0.6 Å Photon energy range: 4 - 20 keV Pulse length (10 fs FWHM) Pulse energy up to 1 mJ more to come: PAL-XFEL (2015) SwissFEL (2016) LCLS-II (2018) … European XFEL ~ 2015 25 Also for soft x-ray FELs FLASH @ DESY Operates down to 4 nm Next-Generation Light Source (NGLS), LBL What comes next for XFELs? SASE temporal coherence can be drastically improved by seeding (self or external seeding) SASE seeded Precise control x-ray properties similar to optical lasers Compact coherent sources Harmonic generation for seeding High Gain Harmonic Generation (HGHG) L.-H. Yu, PRA, 1991 beam chicane radiator modulator BNL 2003 FERMI FEL, 2011 Echo Enabled HG G. Stupakov, PRL, 2009 seed laser 2 seed laser 1 beam modulator 1 modulator 2 radiator Self-Seeding1,2 First undulator generates SASE X-ray monochromator filters SASE and generates seed Chicane delays electrons and washes out SASE microbunching Second undulator amplifies seed to saturation chicane 1st undulator 2nd undulator grazing mirrors slit SASE FEL Seeded FEL grating Long x-ray path delay (~10 ps) requires large chicane that take space and may degrade beam quality Reduce chicane size by using two bunches3 or single-crystal wake monochromator4. 1. J. Feldhaus et al., NIMA, 1997. 2. E. Saldin et al., NIMA, 2001. 3. Y. Ding, Z. Huang, R. Ruth, PRSTAB, 2010. 4. G. Geloni, G. Kocharyan, E. Saldin, DESY 10-133, 2010. Hard x-ray self-seeding @ LCLS 1 GW 25 GW 15 51 16 17 31 Geloni, Kocharyan, Saldin (DESY) FEL spectrum after diamond crystal Power dist. after diamond crystal Wide-band power Monochromatic seed power 10-5 5 MW 6 mm 20 fs 30 Self-seeding of 1-mm e- pulse at 1.5 Å yields 10-4 BW with low charge mode Bragg diagnostic with camera Chicane magnet X-rays Diamond mono chamber 31 J. Amann, P. Emma (LBL) 8.3 keV SASE spectrum (diamond OUT) 20 eV Seeded SASE 0.45 eV (510-5) chicane OFF insert diamon d & turn on chicane SASE diamond IN A well seeded pulse (not typical) seeded Fourier Transform limit is 5 fs 0.45 eV chicane ON Factor of 40-50 BW reduction Submitted to Nature Photon., 2012 Soft X-Ray Self-Seeding (SXRSS) Compact grating monochromator and chicane that fit in one undulator section (4m) QU08 QU07 (existing quad) Grating B1 (toroidal VLS) B2 B3 -0.9° -0.9° ( plane mirror) 18 mm +0.9° (existing quad) M3 B4 +0.9° 3.85 mm beam direction M1 = 500 – 1000 eV Bandwidth ~2×10-4 SXRSS U1-7 Slit M2 Dtchicane ~ 700 fs DELTA HXRSS U9-15 U17-32 (add 5 more in future?) D. Cocco, Y. Feng, J. Hastings et al., in collaboration with NGLS (LBL) Taper to enhance FEL efficiency FEL saturates due to significant E-loss Tapered undulator keeps FEL resonance and increase power x-rays e-beam Taper works much better for a seeded FEL than SASE 400 GW Taper seeded Taper SASE Notaper SASE LLNL microwave FEL T. Orzechowski et al. PRL (1986) W. Fawley, Z. Huang et al. NIMA (2002) LCLS-II simulation 8.3 keV -- 1.5 Å (13.64 GeV) 4 kA, 0.3 um emittance LCLS low charge parameters Optimized tapering starts at 16 m with 13 % K decreasing to 200 m 1.3 TW over 10 fs ~1013 photons 1.0 x 10-4 FWHMBW After self-seeding crystal W. Fawley, J. Frisch, Z. Huang, Y. Jiao, H.-D. Nuhn, C. Pellegrini, S. Reiche, J. Wu, FEL2011 C. Schroeder, FLS2012 Laser Plamsa Accelerator (LPA) Transverse gradient undulator (TGU)* FEL resonant condition By canting the undulator poles, generate a linear field gradient Sort e-beam energy by dispersion h so that y g+ g- x Resonance can be satisfied for all energies if * T. Smith et al., J. Appl. Phys. 1979 Compact XFELs driven by LPA 1GeV, 10kA, 10 MeV energy spread; 0.1um emittance; 5 fs (50 pC) 5-m SC undulator lu = 1 cm, au = 1.41 (G. Fuchert, NIMA 2012) Radiation wavelength l1 = 3.9 nm For TGU, dispersion h = 0.01 m, x = 100um, y = 15um cFLASH (compact FLASH)? Z. Huang, Y. Ding, C. Schroeder, submitted to FEL2012 Compact XFEL with TGU (3.9 nm) Single-shot spectrum 1 GW TGU TGU no TGU no TGUx100 TGU is insensitive to energy jitters (energy jitters transverse position jitters), no change in l1. Good for laser plasma accelerators (currently at a few % energy jitters) Good for a seeded FEL when wavelength is fixed Summary Driven by development of accelerator science and technology, fourth-generation x-ray source based on FEL mechanism has become a reality LCLS is opening up a new world of ultrasmall and ultrafast. The high demands from the x-ray community will drive continuous growth of such sources and innovative R&Ds. Thank you for your attention! Quiz 1 An experimenter places a monochromator with 1 eV bandwidth centered at 10 keV photon energy after the LCLS beam. If the SASE pulse length is estimated to be 10 fs fwhm. What is the expected rms intensity fluctuation for the filtered radiation? Quiz 2 How many 10 keV photons per pulse for 2 mJ hard x-ray FEL? Assuming SASE has 100% transverse coherence, the fwhm pulse duration is 50 fs. What is the number of photons per mode (the degeneracy parameter)? In this context, discuss what is the benefit of seeding?