Beam dynamics for an X-band LINAC driving a 1 keV FEL R. Bartolini Diamond Light Source Ltd and John Adams Institute, University of Oxford XB10 Workshop Daresbury, 02 December 2010

Outline Introduction A brief look at the NLS project and users’ requirements Can we do the same with an X-band Linac? First optimisation of the 2.25 GeV Linac and AP challenges layouts and performance a comparative look at L-band vs X-band collective effects (CSR and wakefields) control of the e– phase space distribution Conclusions XB10 Workshop Daresbury, 02 December 2010

Layout of NLS photoinjector 3rd harmonic cavity BC1 BC2 BC3 laser heater accelerating modules collimation diagnostics spreader FELs IR/THzundulators gas filters experimental stations 3 free-electron lasers covering the photon energy range 50 eV–1 keV, - GW peak power - 20 fs pulses - laser HHG seeded for longitudinal coherence high brightness electron gun, operating (initially) at 1 kHz 2.25 GeV cw superconducting linac Operates with 50-200 pC L-band gradient 15 MV/m (shallow minimum of cost optimisation) Total length from gun to final tube = 290 m

Can we do the same with an X-band Linac? NLS science case had pressing requirements for high rep rate Phase 1 – 1 kHz Phase 2 upgrade – 100 kHz and 1MHz  SC L-band technology 15 MV/m  150 m active length to reach 2.25 GeV Assuming lower gradient in X-band structures for higher repetition rate 35 MV/m can operate at 1 kHz (see C. Christou’ talk) 35 MV/m  65 m active length to reach 2.25 GeV XB10 Workshop Daresbury, 02 December 2010 4

Accelerator Physics challenges Soft X-ray are driven by high brightness electron beam 2.25 GeV n  1 m ~ 1 kA  /   10–4 This requires: a low emittance gun (norm. emittance cannot be improved in the linac) acceleration and compression through the linac keeping the low emittance Requirements for the operation of seeded FELs Assuming a 20 fs FWHM seed laser pulse we need an electron bunch with constant slice parameters over 20 fs plus the relative time jitter between the electron bunch and the laser seed pulse. constant slice parameters on a length of 100 fs – or longer no residual energy chirp (or very limited) low sensitivity to jitter The slice parameters to control are not only slice current, emittance, energy spread but also slice offset and angle and Twiss parameters

Design and Optimisation of LINACs driving FELs (WIP) Parameters used in the optimisation Accelerating section and 3HC amplitude and phase, Bunch compressors strengths (R56) Validation with full start-to-end simulation Gun to FEL (time dependent) including space charge in the gun, CSR, longitudinal space charge, wake-fields in RF cavities (elegant) Astra Elegant GENESIS Gun S01 LH X4H S02 S03 S04 BC1 X01 X02 X03 X04 X05 X06 BC2 DL undulators 4 + 12 tubes S band operating at 12 MV/m + 96 tubes X-Band T53 First compression at 450 MeV – total length from gun to last tube  190 m Gun S01 LH X4H S02 S03 BC1 X01 X02 X03 X04 X04 X05 X06 BC2 DL undulators 4 + 8 tubes S band operating at 12 MV/m + 106 tubes X-Band T53 First compression at 350 MeV – total length from gun to last tube  180 m

Linac design Additional S band cavities operating at 12 MV/m and transition to X-band cavities as soon as possible Two compression stages at 450 MeV and 1.3 GeV made of two C-chicane with 4 dipoles; Linearise 4HC before S-module at 135 MeV – reduced gradient requirements (blow 35 MV/m) 96 T53 X-band cavities (0.53 m) – 1.8 GeV in X band + 450 MeV in S band Doublet focussing structure every 16 tubes – 1.2 m length for diagnostics Spreader design based on a dog leg: further upgrade based on NLS spreader (no issues expected)

NLS S-band gun (WIP) A new S-band gun has been optimised to operate at higher rep rate (1 KHz) Modified BNL-ATF type design operating at 100 MV/m to reach 1 kHz (instead of 120 MV/m) Emittance below 0.2 um 8 ps FW bunch < 5 keV en. spread 7 A peak Courtesy J.H. Han 8

X-band cavities wakefields for lower gradient SLAC X-band cavity has a/ = 0.18 – 45MV/m Wakefield used in this simulations is scaled from the SLAC X band structure using the analytical formula (K. Bane 2003 - Z. Huang private communication) Assuming a/ = 0.13 – 35MV/m Changing a, keeping g and L fixed (readjusting b to keep the resonant frequency)

Ratio of wakes (a/ = 0.187 to a/ = 0.13) LCLS - NLC design SLAC high rep rate design Transverse wakefields with a low rep rate design are expected to be 3-4 time stronger This factor was included in the elegant simulations

50 pC operation For SASE with short pulses – 20 fs FWHM 8 ps FW at start 0.8 ps FW at 460 MeV compression ratio 10 60 fs FW at 1.3 GeV compression ratio 13

Slice analysis of electron beam at the FEL Current distribution Norm emittance delta gamma XB10 Workshop Daresbury, 02 December 2010

FEL time dependent simulations (50 pC) FEL SASE time dependent simulations confirm the good slice beam quality achieved in the optimisation 3 GW power at 40 m 5 m FWHM (17 fs) 1 keV pulse

200 pC operation For seeded FEL with short HHG seed – 20 fs FWHM 8 ps FW at start 1.5 ps FW at 460 MeV (C 5) 100 fs FW at 1.3 GeV (C 15)

Slice analysis of electron beam at the FEL (WIP) Current distribution Norm emittance 3D Xie gain length 60 fs flat region XB10 Workshop Daresbury, 02 December 2010

Comparative Optimisation L-band vs X-band Can use 2 compressors instead of 3 the bunch from the S-gun is shorter (higher gradient at gun S-band gun even if operated at higher rep. rate 100 MV/m used) Clash between the requirements start with X-band structure asap (use more X-band structure asap) which implies first compressor asap but do not compress too much too early (space charge issues) changing to X-band requires a short bunch in order not to reintroduce a curvature from the RF acceleration (which is bad for compression in second compressor) Reduction of energy chirp easier with X-band than with L-band Usual conditions for the control of the optics functions at the compressors

Conclusions and ongoing work Thank you for your attention We have started the analysis of an X-band linac to drive a Soft X-ray FEL with a repetition rate of 1 kHz for ultrashort pulses (20 fs) in SASE mode and for seeded FEL (20 fs seed on flat top of 100 fs) using scaled version of the SLAC X-band cavity to (a/ = 0.13) Beam dynamics optimisation at low charge (50 pC) for SASE appears to be straightforward. Higher charge seeded operation is more demanding but appears to be feasible. Cheaper option for NLS (see R. Walker’s talk tomorrow) Thank you for your attention