B. D. Patterson, R. Abela and J. F. van der Veen, Paul Scherrer Institut, CH-5232 Villigen
The Paul Scherrer Institut is planning the construction of an X-ray Free Electron Laser. The PSI-XFEL will produce 60 fs pulses of coherent X-rays in the wavelength range 0.1 – 10 nm, with a peak brightness approximately 1010 times that of a third-generation synchrotron. The brightness, coherence and short pulses provide opportunities for performing novel science in the fields of chemistry, biochemistry, solid state physics and materials science. The PSI-XFEL could be operational in the year 2016.
What is an XFEL, and how is the PSI-XFEL special?
The active medium in an X-ray free electron laser consists of a 100 femtosecond pulse of 109 relativistic electrons moving in the sinusoidal field of an undulator: a periodic linear array of alternately-poled permanent magnets. As in a synchrotron light source, the transverse acceleration from the Lorentz force causes the electrons to emit X-radiation, but in an XFEL, the undulator is sufficiently long that the growing radiation field influences the trajectory of the electrons. At the "resonance condition", where the radiation overtakes an electron by exactly one wavelength per undulator period, certain of the electrons gain energy and others lose energy, thus splitting the pulse into 105 "microbunches". As it moves along the undulator, the microbunched electron pulse then radiates as if it were a single charge of 109 e, producing an intense, coherent pulse of "superradiant" X-rays.
Besides the PSI-XFEL there are presently three other projected XFELs worldwide: in Stanford, USA (2009), Hyogo, Japan (2011) and Hamburg, Germany (2014). The maximum electron energy and hence the overall XFEL length (800 m) are significantly lower at PSI than for the other projects. This is made possible by PSI innovations in the high-brightness electron source technology, including nanometer-scale field-emitting tips and initial acceleration in a pulsed field of 1 MV across a 4 mm gap, followed by a novel two-frequency RF-cavity. The individual X-ray pulses will be very similar to those of the larger projects (see Table 1). While the Swiss, US and Japanese XFELs will emit 60-120 pulses per second, the Hamburg machine, due to the use of superconducting accelerator technology, will produce 10 trains of 3000 pulses per second, with a minimum pulse spacing of 200 ns.
With sufficient resources, the PSI-XFEL could have enhanced capabilities which are not presently foreseen at the other projects. These include: rapid tuning of the XFEL wavelength for spectroscopic investigations, the option of circular polarization for magnetic studies, additional beams of pulsed, broadband spontaneous radiation for time-resolved Laue crystallography and wavelength-dispersive spectroscopy, extension of the maximum photon energy to the ultra-narrow Mössbauer resonance of 57Fe at 14.4 keV, and modification of the electron pulses by "seeding" and / or "slicing" to yield substantially narrower spectral widths and / or pulse lengths than those given in the Table. Rapid switching of the electrons will allow the simultaneous operation of three PSI-XFEL branches, covering the photon energies 12.4 – 4.0, 4.4 – 0.4 and 1.2 – 0.13 keV.
|Table 1: PSI-XFEL specifications|
|Maximum electron energy||6 GeV|
|Photon wavelength||0.1 – 10 nm|
|Photon pulse length (FWHM)||60 fs|
|Spectral width (FWHM)||0.1 – 0.9 %|
|Beam size at undulator exit (FWHM)||25 – 35 µm|
|Peak brilliance||1033 – 1031 ph/s/mm2/mrad2/0.1 % bw|
|Flux||2 x 1011 – 5 x 1012 ph/pulse|
|Pulse repetition rate||100 Hz|
The bright pulses and high degree of coherence of the XFEL radiation will allow time-resolved lensless imaging to be performed on a variety of systems, including ferroelectric and magnetic domains in thin films, biomolecular conformations in solution and molecular diffusion on surfaces. Time-dependent changes can be triggered by a switched electric or magnetic field (see Fig. 1) or by a fs optical laser pulse which is synchronized to the X-ray pulse. On the other hand, equilibrium fluctuations on the fs – ns time scale can be studied as a function of wavevector using X-ray Photon Correlation Spectroscopy . All of these techniques require the rapid collection of high-resolution 2D-images and the stepwise variation of a pulse delay. Furthermore, the excited sample must relax to equilibrium before the next pump-probe cycle is initiated. Such processes are compatible with the 100 Hz repetition rate of the PSI-XFEL.
With a tunable XFEL wavelength, X-ray absorption spectroscopy (XAS) can also be used to follow a time-dependent process. For example, it is proposed that a chemical reaction be "gently" initiated on a catalytic surface by using a sub-ps pulse of coherent terahertz radiation . After a variable time delay, the chemical or electronic environment of the reactive species is determined by near-edge (XANES) or far-edge (EXAFS) absorption spectroscopy. It is also foreseen to use time-resolved pump-probe XAS to follow laser-initiated photo-chemical reactions in solution .
The time-dependent behavior of solid materials following excitation by a sub-ps laser pulse is a wide field of inquiry, including topics such as non-equilibrium melting, the coherent creation and control of phonons, local structural changes upon the photo-excitation of chromophores, etc. Since the advent of "sliced" synchrotron beams, sub-ps X-ray pulses [6-8] have become available, albeit with very low intensities. At the sliced "FEMTO" beamline of the Swiss Light Source, it has been possible to coherently create and control phonons in a bismuth crystal . Because the PSI-XFEL will deliver vastly more photons per pulse, such pump-probe experiments in solids will become orders of magnitude easier to perform.
The XFEL will also provide several new tools to biochemists. The time-dependent behavior of photosensitive biomolecules in crystalline form can be studied by time-resolved, pump-probe Laue crystallography. Using isolated, 100 ps pulses at a synchrotron, structural changes accompanying the photodetachment of CO from the heme group in myoglobin have been recorded in the ns – µs regime  (see Fig. 2). With spontaneous radiation from an incoherent undulator downstream from the XFEL, the same number of broad-bandwidth photons now available in a 100 ps synchrotron pulse will be available in a 100 fs pulse, thus extending time-resolved Laue crystallography into the sub-ps regime.
The "holy grail" of XFEL-based structural biology is the structural determination, with atomic resolution, of large, individual biomolecules, without the requirement of crystallization. Indeed, the highly-important membrane proteins can generally not be crystallized, and hence their structures remain largely unknown. It is proposed that the hard X-ray XFEL beam be focused down to 100 nm, and that individual protein molecules be synchronously injected, one-by-one, into the XFEL pulses. Of course, the intense X-ray pulse will quickly destroy the molecule (see Fig. 3), but from sliced pulses shorter than several fs, sufficient scattered photons may possibly be obtained to allow a structural determination. Even with the XFEL in operation, many hurdles must be overcome before this goal can be achieved: pulse shaping and focusing, particle preparation and injection, data collection and, importantly, combining the data from many individual, randomly-oriented molecules.
The qualitative advances in performance by an X-ray free electron laser over all existing X-ray sources make accurate predictions of the relevant photon-matter interactions difficult. Nonetheless, novel science will certainly be made possible by the enormous increases in peak brightness, coherence and time resolution. The Paul Scherrer Institut welcomes input from interested parties inside and outside of Switzerland regarding the conceptual design of the PSI-XFEL facility.
 S. Eisebitt, et al, Nature 432, 885 (2004).
 C. Gutt, O. Leupold and G. Grübel, Thin Solid Films 515, 5532 (2007).
 H. Ogasawara, et al, Proc. 27th FEL Conf. (2005).
 W. Gawelda, et al, Phys. Rev. Lett. 98, 057401 (2007).
 R.W. Schoenlein, et al, Science 287, 2237 (2000).
 S. Kahn, et al, Phys. Rev. Lett, 97, 074801 (2006).
 P. Beaud, et al, Phys. Rev. Lett. 99, 174801 (2007).
 F. Schotte, et al, Science 300, 1944 (2003).
 R. Neutze, et al, Nature 406, 752 (2000).
[Released: May 2008]