A compact, soft X-ray FEL at KVI Hans Beijers Kernfysisch Versneller Instituut University of Groningen, The Netherlands Introduction Science drivers ZFEL characteristics Roadmap to ZFEL Promising application of X-band high-acceleration gradient linacs, i.e. X-ray FELs. New proposal at University of Groningen, if accepted will be the first X-band X-ray FEL. Outline of talk is as follows:

Introduction Kernfysisch Versneller Instituut Zernike Institute for atomic, nuclear and (astro-) particle physics accelerator /radiation physics theoretical physics Kernfysisch Versneller Instituut F. Zernike (1888-1966) I am from KVI, situated in the city of Groningen in The Netherlands. KVI is a university-based institute that focuses on atomic, nuclear and particle physics. Main instrument is presently a K=600 superconducting cyclotron used for both fundamental and applied research. We are working on a proposal, together with the Zernike Institute for Advanced Materials to build in Groningen at the KVI site a soft X-ray FEL. Zernike institute focuses on two main themes, I.e. biomolecules and nanostructured materials. New proposal is called ZFEL, after Frits Zernike, our local Nobel laureate who got Nobel prize in 1953 for inventing phase-contrast microscopy. He probably would have liked the ZFEL proposal a lot, since it is the most ideal microscope to investigate the nanoworld. biomolecular and bio-inspired functionality nanostructured materials for electromagnetic functionality Zernike Institute for Advanced Materials

Hard and soft X-ray FELs SCSS Hard X-ray FEL’s LCLS: 15 GeV, NC, S-band SCSS: 8 GeV, NC, C-band SwissFEL: 5.8 GeV, NC, C-band Eur-XFEL: 20 GeV, SC, L-band FLASH: 1.2 GeV, SC, L-band FERMI@Elettra: 1.5 GeV, NC, S-band EU-XFEL Swiss-FEL Soft X-ray FEL’s FLASH FERMI@Elettra Two X-ray lasers in operation, first one is the soft X-ray laser FLASH, a superconducting 1.2 GeV soft X-ray FEL at DESY in Hamburg. The second is the hard X-ray, normal conducting LCLS at Stanford. Both X-ray lasers are heavily overbooked, and several other lasers are being built right now, among them SCSS, SwissFEL, Eur-XFEL and FERMI. The superconducting linacs use L-band, and the normal conducting ones either S-band or C-band. Let us now look into the properties that make these X-ray sources so special.

Explosion of T4 lysozyme (C, N, O, S) induced by radiation damage. XFEL properties Short wavelengths: < 1 Å – 100 nm (10 keV – 10 eV) High transversal coherence and, when seeded, also longitudinal coherence Very intense: 1010 – 1013 photons/pulse Short pulse lengths: < 1 – 100 fs Explosion of T4 lysozyme (C, N, O, S) induced by radiation damage. R. Neutze, et al, Nature 406, 752 (2000). Gaffney et al., Science 316 (2007) 1444 First important property is the short wavelength, ranging from say 100 nm down to less than 0.1 nm. They share this short-wavelength property with the X-rays produced in the insertion devices of the present 3rd generation synchrotron light sources. What sets the X-ray FEL’s apart are the other three properties, I.e. the high degreee of coherence, almost full spatial coherence, and when seeded also temporal coherence. Then, as befits a laser, the pulses are very intense, ranging from 1010 to 1013 photons/pulse. And finally, the pulses are also very short, between typically 100 fs and less than a fs. These properties make X-ray lasers the ideal tool to visualize the nanoworld in action. The short wavelength makes it possible to determine molecular structure, and the high coherence that we can determine the structure of single molecules, no need for crystallization anymore. And because the pulses are so short we can even see chemical reactions happen in real time, so we can actually make molecular movies. Of course the high intensity destroys the molecule, but this happens slow enough that the diffraction information, from which the structure is determined, is recorded before the molecule has been fallen apart. So you can see that X-ray FEL’s are indeed the ideal tools to investigate the nanoworld in action.

Science drivers Hierarchical biology Chemical reactions Protein folding, ‘biological’ water, structure and function, membranes, viruses, cells Chemical reactions Catalysis, interfaces, combustion, molecular movies Matter in extreme conditions Phase diagram borders, end of scales Atoms to materials Inorganic/bio clusters Emergence of solid state properties from atom clusters Fundamental interactions QED and symmetries, axions, dark matter Correlated materials Phase transitions, emergent phenomena, high Tc superconductivity, magnetism Mesoscale physics Fatigue, fracture, strain, radiation damage, nucleation, disordered materials Complex materials New battery and photovoltaic materials, phase transitions Studying chemical reactions in real time and the formation and dynamics of biological structures is only one area of the nanoworld that can be investigated. Physicists at KVI are proposing fundamental physics experiments investigating QED and exotic forms of matter. Also the emergence of solid state properties from those of single atoms can be studied in detail with clusters that are of increasing size. The atom-size resolution, both spatial and temporal, that X-ray FEL’s offer are ideal to investigate many aspects of material physics, among others correlated materials with high-temperature superconductivity as an example. Aso various mesoscale phenomena that are important for various industriel processes can be followed in detail. And lastly I can mention new and complex materials that can be studied with X-ray lasers. As you can see most prospective users are biologists and biochemists and material scientists.

ZFEL characteristics Wavelength range: 0.8 – 50 nm  Emax= 2.1 GeV Repitition rate: 10 Hz – 1 kHz Total length: < 200 m  high accel-grad. structures Our material scientist colleagues request a minimum wavelength of 0.8 nm, for which you need a maximum electron energy of 2.1 GeV with a repition rate between 10 Hz and 1 kHz. In order that the entire facility fits on the KVI site its total length should not exceed 200 meters, which is only possible when the length of the accelerator is minimized, I.e. using high-gradient acceleration structures that is an X-band linac.

XFEL: lasing Self-Amplified Spontaneous Emission (SASE) EM wave – electron interaction (m-bunching) ZFEL Present X-ray linacs use the SASE process, which stands for self-amplified spontaneous emission. Here a relativistic electron bunch moves through an undulator. Quasi-monochromatic synchrotron radiation is emitted in the forward direction because photons emitted on subsequent turning points of the electron trajectory have to interfere constructively, and this leads to the following relation between the photon wavelength and the electron energy. The photon wavelength decreases inversely proportional to the square of the electron energy. This can also interpreted as a double relativistic effect. The electrons see in their rest frame a Lorentz-contracted undulator period, given by lambda_u/gamma, and in the lab-frame these photons are blue-shifted with a factor 1/2gamma. The undulator K-factor describes the decrease of the average longitudinal velocity of the electrons because of the wiggle motion in the undulator field. The emitted photons are spontaneously emitted and therefore incoherent. This is the basic process of undulator radiation in third-generation synchrotrons. Exponential gain occurs because of the back reaction of the electromagnetic field on the electron motion. When the electromagnetic field is strong enough it leads to a so-called microbunching of the electron beam, that is the electron bunch acquires a density modulation with a scale length on the order of the radiation wavelength. This microbunching is also called the FEL instability and becomes increasingly pronounced as the electrons progress along the undulator and leads to coherent amplification so that the emitted power scales as N squared instead of N. This is the basic reason why the radiated power of FEL’s is some 10 orders of magnitude larger than the power of third-generation light sources.

SASE SASE has no longitudinal coherence ! Here you see the temporal and frequency spectrum of a SASE FEL, which is clearly not very coherent. Actually each individual spike is a coherent pulse. The red line is the average, and the broken line represents the electron bunch. Longitudinal coherent radiation can be produced by not using the SASE process, but by injecting resonant photons from a seeding laser into the undulator. Then the FEL is not used as an oscillator, but as an amplifier. The problem is of course that it becomes increasingly more difficult to built a seed laser for smaller wavelengths. Various schemes have been developed to overcome this problem, for example HGHG which stands for high gain harmonic generation. The FERMI facility at Trieste will be the first facility making use of this principle. Another way to make longitudinally coherent pulses is by using very short electron bunches, shorter than the so-called cooperation length, this means electron bunches shorter than 1 fs. SASE has no longitudinal coherence ! Longitudinal coherence can be realized by Seeding at l1 or n x l1 using e.g. HGHG Making very short electron bunches, i.e. < 1 fs

XFEL: e- beam requirements FEL power gain length (1 D): High peak current: Ie~ 1 – 3 kA (bunch compression) Excellent e- - g transverse overlap: small transversal emittance Small wavelength/energy bandwidth: The parameter rho is called the Pierce-parameter and should be as large as possible in order to minimize the gain length. FEL saturation typically occurs after 15 gain lengths when the microbunching has developed completely. I_A is the so-called Alfven current and equals 17 kA. I is the peak current and sigma_x the transversal bunch length and k_u the undulator wavenumber. The Bessel factor JJ is of order unity and describes the geometry of the undulator. In order to maximize rho one has to maximize the peak current and minimize the transversal bunch size. Typical currents are between 1 and 3 kA. The requirement on normalized beam emittance comes from the desire to maximize the overlap between electrons and photons in the undulator. It follows that the normalized emittance should be smaller than gamma times the photon wavelength divided by 4pi. The smaller the photon wavelenght the smaller the normalized emittance must be. For ZFEL this translates to a value of the normalized emittance of only 0.26 mm mrad. A higher emittance immediately translates into a less efficient FEL process and longer gain length. Finally, there is also a demand on the maximum fractional energy width of the lelectrons. A change in electron energy immediately leads to a change in photon wavelength, and when the photon wavelength falls outside the FEL gain profile amplification stops. The fractional energy width should be smaller than the Pierce parameter rho, which typically is in the range of 5x10-4. Regular-spaced e- bunches with rep.rate up to 1 kHz High acceleration gradient (X-band linac) Small emittance and energy spread must be preserved during acceleration !

XFEL linac courtesy R. Bakker

ZFEL parameters Stage 1 Stage 2 Beam energy (GeV) 1.0 2.1 Bunch charge (pC) 10 - 100 Norm. emittance (mm mrad) Peak current (kA) 1.5 Energy spread (MeV) 0.9 Rep. Frequency (Hz) 10 - 1000 Stage 1 Stage 2 Undulator period (mm) 15 Undulaor par. K 1.2 Pierce param. r 9.7e-4 4.6e-4 Gain length (m) 0.711 1.5 Saturation length (m) 13.3 27 Photon wavelength (nm) 3.4 0.8 We propose to build ZFEL in 2 stages, in the first stage the electrons are accelerated to 1 GeV, which leads to a minimum wavelength of 3.4 nm using an in-vacuum undulator with a period of 15 mm and undulator parameter K=1.2. Using values of the bunch charge, normalized emittance, peak current and energy spread as given in the table gives a gain length of only 71 cm and a total undulator length of 13.3 m. In stage 2 the energy will be increased to 2.1 GeV giving us photons of 0.8 nm and a total undulator length of 27 m. in order to realize this soft X-ray FEL with an accelerator as short as possible we intend to use high-gradient X-band acceleration structures. A possible layout of ZFEL has been proposed by our SLAC colleagues and has been discussed in great detail in the previous session.

Generic ZFEL layout SLAC scheme 3 S-band injector sections X-band linac section: 8 x 0.5 m TW = 400 MeV energy gain per section 2 X-band linearizer cavities Stage 1: 2 X-band linac sections Stage 2: 5 X-band linac sections Courtesy: Chr. Adolphsen Here you see the first-stage layout. Electrons are produced in a S-band RF photogun which accelerates the electrons to 7 MeV. Then the electrons enter the 1st S-band linac where they are accelerated to 200 MeV, then follows the 1st beam compressor. Then the electrons enter the main linac which consists of 2 X-band sections each yielding an energy gain of 400 MeV. Between the two X-band sections is another bunch compressor. After exiting the main linac the electrons enter the undulator where the FEL process takes place. In the second stage another 3 X-band linac sections will be added to bring the electron energy to more than 2.1 GeV. Each linac section is powered by 2 X-band klystrons.

ZFEL RF Photogun Collaboration with TU Eindhoven (J. Luiten) 266 nm, 10 ps, 1 kHz rep. rate > 0.2 mJ/pulse E = 6.9 MeV Q = 0.1 – 1 nC en  1 mm mrad Courtesy: Jom Luiten The RF photogun will be delivered by the Technical University of Eindhoven. This photogun has already been built and produces bunches with charges up to 1 nC with a normalized emittance smaller than 1 mm mrad. The RF photogun has 2.5 cells and the microwaves are coaxially injected into the gun. This prevents emiitance increase.

X-band accel. structures High-gradient accelerating structures for ZFEL SLAC T53 or H60 structures high acceleration gradient and low breakdown rate design optimization (e.g. a/l) We will only accelerate single bunches, so the long-range wakefields are of no concern and we can simply use the SLAC T53 undamped structures which can produce acceleration gradients of 100 MV/m with very small breakdown rates. A very important asppect is of course that the high beam quality obtained from the photogun must be preserved during the acceleration process and this has to be insured by detailes sensitivity studies of the beam dynamics to various errors. In addition to the single-particle beam dynamics also various collective effects will have to be studied in great detail.

Design studies Single-particle dynamics: B-field errors, magnet misalignments and chromatic aberrations Maximum-allowed emittance growth determines tolerances Collective effects: short-range wakefields, CSR emission and beam instabilities S2E simulations Basic problem: no designs as yet for 1 kHz, high-power X-band klystrons and modulators

Road to ZFEL Submitting ZFEL proposal in 2011 RF photogun and diagnostics, X-band test stand Build up expertise and setting up collaborations Design studies and developing subsystems, e.g. seed lasers, diagnostics, timing and synchronization systems, alignment and feedback etc. Construction First light 5 years after start project

Thank you for your attention Thanks The CLIC group for having me here Many colleagues from the X-band and XFEL communities for very useful discussions and advice (Walter Wuensch, Sami Tantawi, Chris Adolphsen, Hans Braun, …….. Thank you for your attention