Approaches for the generation of femtosecond x-ray pulses

1. Approaches for the generation of femtosecond x-ray pulses Zhirong Huang (SLAC)

2. The Promise of X-ray FELs Ultra-bright Ultra-fast

3. Single Molecule Imaging with Intense fs X-ray R. Neutze et al. Nature, 2000

4. Introduction • Femtosecond (fs) x-ray pulses are keys to exploring ultra-fast science at a future light source facility • In typical XFEL designs based on SASE the photon pulse is similar in duration to the electron bunch, limited to 100~200 fs due to short-bunch collective effects • Great interests to push SASE pulse length down to ~10 fs and even below 1 fs • A recent LCLS task force studied upgrade possibilities, including short-pulse approaches • I will discuss and analyze several approaches in the next 1800000000000000000 fs!

5. Outline of the Talk • Temporal characteristics of a SASE FEL • Optical manipulation of a frequency-chirped SASE • Compression • Slicing: single-stage and two-stage • Statistical analysis • Electron bunch manipulation • Spatial chirp • Enhancing undulator wakefield • Selective emittance spoiling (slotted spoiler) • Sub-femtosecond possibilities

6. Temporal Characteristics of a SASE FEL E(t)=j E1(t-tj), tj is the random arrival time of jthe- E1: wave packet of a single e- after Nu undulator period Nu Coherence time coh determined by gain bandwidth 

7. bunch length Tb coh • Sum of all e- E(t) • SASE has M temporal (spectral) modes with relative intensity fluctuation M-1/2 • Its longitudinal phase space is ~M larger than Fourier transform limit • Narrower bandwidth for better temporal coherence • shorter x-ray pulse (shortest is coherence time)

8. 1 % of X-Ray Pulse Length • LCLS near saturation (80 m) • bunch length 230 fs • coherence time 0.3 fs • number of modes ~ 700 statistical fluctuation sw/W ~ 4 % Shortest possible XFEL pulse length is only 300 as!

9. Optical manipulations of a frequency-chirped SASE

10. X-ray Pulse Compression • Energy-chirped e-beam produces a frequency-chirped radiation • Pair of gratings to compress the radiation pulse C. Pelligrini, NIMA, 2000 • No CSR in the compressor, demanding optics • Pulse length controlled by SASE bandwidth and chirp

11. SASE FEL Monochromator X-ray Pulse Slicing • Instead of compression, use a monochromator to select a slice of the chirped SASE ω monochromator short x-ray slice t compression • Single-stage approach

12. Chicane SASE FEL FEL Amplifier Monochromator Two-stage Pulse Slicing C. Schroeder et al., NIMA, 2002 • Slicing after the first undulator before saturation reduces power load on monochromator • Second stage seeded with sliced pulse (microbunching removed by bypass chicane), which is then amplified to saturation • Allows narrow bandwidth for unchirped bunches

13. Analysis of Frequency-chirped SASE • Statistical analysis (S. Krinsky & Z. Huang, PRST-AB, 2003) Frequency-chirp • coherence time is indep. of chirp u • frequency span and frequency spike width coh ~ u • A monochromator with rms bandwidth sm passes MF modes

14. Minimum Pulse Duration • The rms pulse duration st after the monochromator ω u t • Minimum pulse duration is limited to for either compression or slicing • Slightly increased by optical elements (~ fs)

15. One-stage Approach • SASE bandwidth reaches minimum (~r~10-3) at saturation  minimum rms pulse duration = 6 fs (15 fs fwhm) for 1% energy chirp • st minimum for broad sm choose sm ~ sw to increase MF (decrease energy fluctuation) and increase photon numbers

16. Two-stage Approach • Slicing before saturation at a larger SASE bandwidth leads to a longer pulse Ginger LCLS run • Synchronization between sliced pulse and the resoant part of chirped electrons in 2nd undulator ~ 10 fs

17. Electron Bunch Manipulations

18. Spatially Chirped Bunch P. Emma & Z. Huang, 2003 (Mo-P-52) 200-fs e- bunch 30-fs x-ray Undulator Channel • FEL power vs. y’ offset for LCLS • Gain is suppressed for most parts of • the bunch except the on-axis portion

19. E = 4.5 GeV, sz = 200 mm, V0 = 5 MV 1.0 m +2sy 0 -2sy y vs. z at start of undulator ? • No additional hardware for LCLS • RF deflector before BC2 less jitter • Beam size < 0.5 mm in linac  FWHM x-ray pulse ~ 30 fs

20. Courtesy S. Reiche

21. Using Enhanced Wakefield • Ideal case (step profile) with various materials for the vacuum chamber to control wakefield amplitude 4 fs (FWHM) • Change of vacuum chamber to high resistivity materials (graphite) is permanent, no long pulse operation S. Reiche et al., NIMA, 2003

22. Where else can we access fs time? Large x-z correlation inside a bunch compressor chicane LCLS BC2 Easy access to time coordinate along bunch 2.6 mm rms 0.1 mm rms

23. Slotted-spoiler Scheme 1 mm emittance 5 mm emittance 1 mm emittance P. Emma et al. submitted to PRL, 2003 (Mo-P-51)

24. ParmelaElegantGenesis Simulation, including foil-wake, scattering and CSR

25. 2 fsec fwhm fs and sub-fs x-ray pulses • A full slit of 250 mm  unspoiled electrons of 8 fs (fwhm) •  2~3 fs x-rays at saturation (gain narrowing of a Gaussian electron pulse) • stronger compression + narrower slit (50 mm)  1 fs e- •  sub-fs x-rays (close to a single coherence spike!)

26. Statistical Single-Spike Selection Unseeded single-bunch HGHG (8  4  2  1 Å ) I8/ I18 8 Å 1 Å sub-fs spike Saldin et al., Opt. Commun., 2002

27. Selection Process Set energy threshold to reject multi-spike events (a sc linac helps)

28. Conclusions • XFEL can open both ultra-small and ultra-fast worlds • Many good ideas to reduces SASE pulse lengths from 100 fs to ~ 10 fs level • Optical manipulations are limited by SASE bandwidth, available electron energy chirp, and optical elements • Electron bunch manipulations and SASE statistical properties may allow selection of a single coherent spike at sub-fs level • Time for experimental investigations