An X-Ray Free Electron Laser Oscillator Kwang-Je Kim ANL & the U of Chicago Claudio Fest Oct. 1-2, 2010 Avalon, CA

Claudio’s Workshops KJK Catalina Oct Claudio’s workshops helped me and others, I am sure, to kick-start my accelerator physics career. Thank you, Claudio!

KJK Catalina Oct Era of Hard X-Ray (   1 Å) FEL Commenced in 2009 with the Operation of LCLS (SASE) 3 LINAC Coherent Light Source LCLS Project start 1999 RIKEN/SPring-8 XFEL 2011 European XFEL Facility 2014 LCLS August, LCLS, April 2009 I=500 A I=3000A April 10, 2009 User experiment September, 2009 BNP paper in 1984 Claudio’s proposal to use an RF photocathode gun,  x ~1 , and the SLAC linac

4 X-ray FELs beyond LCLS  More facilities –Higher average brightness (Euro XFEL, SCRF) –Compact FELs ( SPring-8) –Coherent soft x-rays with harmonic generation  Options with lighter ( 1-50 pC) bunches with lower emittance ( 0.1 mm-mr) –Atto-second time resolution –meV spectral resolution with an XFELO KJK Catalina Oct 2010

5 A hard x-ray FEL oscillator (XFELO) will provide full coherence with ultra-pure spectrum  An X-ray pulse is stored in a diamond cavity  multi-pass gain & spectral cleaning  Provide transform limited BW –  ~10 -7   meV  Zig-zag path cavity for wavelength tuning Originally proposed in 1984 by Collela and Luccio and resurrected in 2008 (KJK, S. Reiche, Y. Shvyd’ko, PRL 100, (2008) 5

KJK Catalina Oct High reflectivity and narrow Bandwidth with near backscattering from diamond crystals Photon energy (keV) Courtesy of Yuri Shvyd’ko

KJK Catalina Oct Representative Parameters  Electron beam: –Energy  7 GeV –Bunch charge ~ pC  low intensity –Bunch length (rms)  1 (0.1 ps)  Peak current 20 (100) A –Normalized rms emittance  0.2 (0.3) mm-mr, energy spread (rms) ~ 2  –Constant bunch rep ~1 MHz  Undulator: –L u = 60 (30) m,  u  2.0 cm, K=1.0 – 1.5  Optical cavity: –2- or 4- diamond crystals and focusing mirrors –Total round trip reflectivity > 85 (50) %  XFELO output: – 5 keV      25 keV –Bandwidth:  ~ 1 (5)  10 -7, pulse length (rms) = 500 (80) fs –# photons/pulse ~ 1  Blue color in the above indicates short-pulse mode for relaxed tolerances

KJK Cat alin a Oct Tunable X-ray Cavity  Two crystal scheme –a very limited tuning since  must be kept small  A tunable four crystal scheme –Any interesting spectral region can be covered by one chosen crystal material –Simplify the crystal choice  Diamond as highest reflectivity & best mechanical and thermal properties 8 R. M.J.Cotterill, APL, 403,133 (1968) KJK & Y. Shvyd’ko, PRSTAB (2009)

KJK Catalina Oct XFELO will drastically improve hard x-ray techniques developed at 3 rd generation light sources, as well as new applications in areas complementary to SASE  High resolution spectroscopy –Inelastic x-ray scattering  Mössbauer spectroscopy –10 3 /pulse, 10 9 /sec Moessbauer  s (14.4 keV, 5 neV BW)  X-ray photoemission spectroscopy –Bulk-sensitive Fermi surface study with HX-TR-AR PES  X-ray imaging with near atomic resolution (~1 nm) –Smaller focal spot with the absence of chromatic aberration  Applications we have not thought of yet! 9

10 XFELO Modeling  Analytical (KJK, R. Lindberg) –Gain calculation, super-mode theory for evolution in optical cavity  GENESIS (S. Reiche) –(x,y) asymmetric, single wavefront  Slow:1 month computing from noise to saturation!  Reduced 1-D FEL code (R. Lindberg) –Transverse dependence integrated out assuming Gaussian mode –Fast and reasonable agreement with GINGER and GENSIS  GINGER (W. Fawley) –(x,y) symmetric  much faster than GENESIS –Implemented a correct crystal response KJK KEK Dec 21, 200 9

Crystal Phase Shift and Cavity Length Detuning  Amplitude reflectivity for near normal incidence x-rays  XFELO works near y~0. The angular spread effect is small   -dependent phase shift  can be corrected by cavity length adjustment KJK KEK Dec 21,

Filtering by crystals expedite and stabilize the development of the ultra-narrow spectrum. Spectrum saturation takes much longer than intensity saturation KJK KEK Dec 21,

KJK Catalina Oct Ginger Simulation of XFELO Spectrum After 500 Passes (Two Diamond Crystal Cavity, 50  m and 200  m, R. Lindberg)

KJK Catalina Oct Technology Development for XFELO  Electron injector  Diamond crystal –Reflectivity –Heat load dynamics  Grazing incidence mirrors  Stability of optical elements 14

Electron Gun Technologies for XFELO  A photo-cathode based injector satisfying XFELO requirements should be feasible –The LCLS S-band NC RF PC demonstrated ultra-low  x  x =0.14  m, Q=20 pC, f b =120 Hz –The PITZ L-band NC RF PC may be suitable a pulsed XFELO:  x =  m, Q=100 pC, f b =1 MHz, t macro =800  s, f macro =10 Hz –Cornell DC PC for ERL could also work if 750 kV DC voltage can be achieved  x =0.2  m, Q=20 pC, f b =1.3 GHz, CW –LBNL 200 MHz, NC RF Gun (CW) (design) could be configured for XFELO application  A thermionic-cathode based injector appears also feasible by modifying the RIKEN/Spring-8 pulsed DC –RIKEN injector by Togawa and Shintake: Hz,  x =0.6  m with 3 mm diameter – Reduce diameter to 1 mm to obtain  x =0.2  m with 1/9 th bunch charge (sufficient for an XFELO) –Replace DC voltage by 100 MHz RF similar to LBNL –Reduce the bunch rep rate to a few MHz by pulsed HV on a gate electrode (T. Shintake) KJK KEK Dec 21,

Operation of a Gate Electrode ( M. Borland & X. Dong) KJK Catalina Oct

A thermionic cathode may provide a higher stability? KJK Catalina Oct CeB6 cathode after 20,000 hrs of operation Laser spot for LCLS gun ( D. Dowells)

KJK KEK Dec 21, A possible injector concept for XFELO 18 1.Gate electrode HV pulser, 1 MHz, 10 kV, ~3ns. 2.RF cavity with therm­ionic cathode, 100 MHz, 1 MV. 3.Solenoid. 4. Quadrupole triplet. 5.Chicane. 6. Horizontal jaw slit as an energy filter to select out a 0.5-ns segment in each rf period. 7. Quadrupole triplet. 8. Short solenoid. 9. Monochromator of the beam energy, f=600 MHz. 10. Sub-harmonic pre-buncher (velocity bunching), f=300 MHz. 11. Booster linac section, 66 MeV, f=400 MHz. 12. Bunch compressor I. 13. SC linac section, 542 MeV, f=1300 MHz. 14. Bunch compressor II. 15. Main SC linac, f=1300 MHz

KJK KEK Dec 21, KJK, HBEB Maui X-Ray Optics R&D for XFELO  Diamond crystal –High-reflectivity –Head-load dynamics –Damage issue  Grazing incidence focusing mirror –Reflectivity and phase front quality  Positional and angular stability  Advances in these technologies are eagerly sought after by broader synchrotron radiation community 19

KJK Catalina Oct Topograpy, R and  E data (Sumitomo sample, S. Stoupin & Y. Shvyd’ko) Optical Properties of HPHT Synthetic Diamond Crystals Crystal supplier: Element 6, Sumitomo, TISNUM (Moscow)

KJK Catalina Oct Reflectivity and spectral width measurement at APS sector-30 in good agreement with theory 21 S. Stoupin, Y. Shyv’dko, A. Cunsolo, A. Said, S. Huang C(995) E H = keV (Nature Physics 6(2010)196)

Radiation damage issues  Power density on crystals is about 4 kW/mm 2 ( 2  photons/s/mm 12.4 keV) ~ 30 times higher than that of the APS undulators  Estimate shows that all atoms will be ionized in 250s in the absence of recombination ( Robin Santra)  Various recombination processes may prevent an irreversible damage  However, photo-electrons are lost at the surfaces  charge build-up may lead to structural change at the surface  Graphitization?  Possible remedies –Isotopically pure 12 C crystals, cryogenic temperature –Attaching a thin conducting layer, e.g. graphene  We plan for a thorough theoretical and experimental study KJK Catalina Oct Graphitization of a diamond crystal at an APS HHLM ( high heat load monochromator at the APS. The crystal surface became darkened without apparent performance degradation after 1 year of exposure.

KJK KEK Dec 21, Heat Load Dynamics  As an intracavity x-ray pulse hit crystals, r-dependent temperature rise  T  crystal expansion   E/E =   T (  L/L=    Is this <10 -7 ?  Yes, if cooled to a cryogenic temperature:T< 100K –Inter-pulse  E/E <10 -7 due to high thermal-diffusivity –Intra-pulse  E/E <10 -7 due to  <10 -7 and if the expansion time < pulse duration (~ps)  Theories suggest the expansion is slow. We will pursue experimental study using laser heating 23 S. Stoupin and Y. Shvyd’ko, PRL 

KJK Catalina Oct Grazing Incidence, Curved Mirror  JTEC –Developing a technique combining elastic emission machining (EEM, slow) and electrolytic in-process dressing (ELID, fast) to fabricate a smooth surface to <nm height error and 0.25 mrad figure error –Issue: large ration in the sagital and meridional depths –Such mirrors are sought after by “every body” in SR business  Other ways of focusing –Curved crystal surface, CRL,.. 24 H. Mimura, et. al. RSI 79, , 2008

KJK Catalina Oct Null-detection FB stabilization at APS Sector 30 (S. Stoupin, F. Lenkszus, R. Laird, Y. Shvy’dko, S. Whitcomb,..)  The stability of IC3 signal indicates the angular stabilization of the 3 rd crystal pair within 50 nrad is achieved (~1 Hz BW)  (S. Stoupin, F. Lenkszus, R. Laird, Y. Shvy’dko, S. Whitcomb,., RSI, 81, , 2010) 25 IC0 IC1 IC2 IC3 IC0 IC1 IC2 IC3 Feedback correction signal HRM PZT

Where can an XFELO be implemented?  Euro-XFEL –The length of the macro-pulse ( 1 ms) is sufficient for a pulsed XFELO ( being studied by J. Rossbach) –Current plan is to use 600 m SCRF linac for 14 GeV with =23.7 MeV/m. By using the full length of tunnel (720 m) at =10 MeV/m, CW XFELO can be 7 GeV  Plan for JAEA-KEK ERL includes an XFELO KJK KEK Dec 21, (1) In a straight section of the loop Beam dump (2) In an additional single-ended branch Integration with an ERL light source From Hajima’s talk at X’ian (Sept. 2009)