Several Issues and Questions for Discussion Related to HGHG cascade

Transcription

1 1 Several Issues and Questions for Discussion Related to HGHG cascade L.H. Yu BNL

2 2 Points for Discussion Concept of High spectral flux source High spectral flux source (Part of talk given by Timur, C.C. Kao, J. Rose, T. Shaftan, L.H. Yu at BNL Forum) Estimate of requirement of coherence of HGHG and relation to energy modulation based on phase noise Double bunches and regenerative bunches to improve synchronization, stability, and reduce noise

3 3 The need for an Xray Source with High Spectral Flux (~ ph/s/10 4 BW) SR based IXS is severely limited by spectral flux (10 13 ~ ph/s/10 4 BW) Today charge excitations are studied with 100 mev resolution, significantly greater than the energy scale of interest O(meV), for example in High T c. A factor of is needed! Note: the next generation of IXS (Inelastic Xray Scattering) instrument being proposed at the SPring8 and NSLSII will only have a factor of O(10) more flux

4 Inelastic Xray Scattering This technique probes dynamics in condensed matter systems: Higher energyresolution allows the study of lower energy excitations: Example: Sound modes in organic bilayers These are very slow vibrational modes in biological membranes which may control their function: Best existing experiments cannot resolve these modes (ΔE=1.5 mev). With High Spectral Flux Source we will have the resolution to cleanly resolve such features for the first time. ESRF Thanks to K. EvansLutterodt Chen et al. PRL (2001)

5 5 Concept of High spectral flux source Develop high spectral flux (SF) source SF=N ph /Δε ph Need a lot of photons per second in a narrow bandwidth Achieve photons/sec/0.01%bw at ε ph =10 kev Note: SF is not brightness: no specification for transverse phase space (photons/sec/0.01%bw/ mm 2 mrad 2 )

6 6 Concept Goals versus Storage Ring Modern synchrotron light sources SF photons/s/0.01%bw around ε ph =10 kev Currently lacking factor of ~1000 in spectral flux Undulator radiation: SF I N u E Increase current I < x2 Increase undulator length L und < N u x5 Increase energy E x2 Issues: Increase in all three is unlikely Even if you could do it factor 20 increase in heat load Useful photons are only ~10 4 from total radiated flux

7 7 Concept Goals versus current FEL R&D Modern Xray FEL sources LCLS FEL SF photons/s/0.01%bw at ε ph =8.5 kev European XFEL SF photons/s/0.01%bw around ε ph =10 kev Issues Lacking factor of 10 to reach in spectral flux Total power in fundamental harmonic may be more than stateofart high heat load optics can handle Natural bandwidth for XFEL ~10 4 is factor 1000 larger than the desirable bandwidth

8 8 Concept Goals versus StateofArt (Summary) High Spectral Flux Source

9 9 Conceptual Solution: HGHG FEL with high rep rate Gun Accelerator Xrays Increase accelerator repetition rate Decrease accelerator energy Use narrow band FEL SASE FEL: In saturation FEL energy per pulse: W ~ ρ E q distributed in bandwidth of (SASE) Δε ph /ε ph ~ ρ W of 10 kev photons per pulse SF = W / Δω f rr f rr ph/sec/0.01%bw per pulse Thus f rr ~ 10 6 is required to reach ph/sec/0.01%bw for SASE FEL Repetition rate may be lowered if High Gain Harmonic Generation FEL (HGHG FEL) is used λ ph HGHG FEL: In saturation Δε ph / ε ph = Δz for Fourier limited radiation spectrum of flat top pulse of 50 fs Lower repetition rate Low heat load on user optics Electron beams with repetition rate of MHz are within reach. FEL

10 10 Cascaded HGHG Generate 1 MHz & 1 nc =1 ma electron beam Compress to 2.5 ka of peak current Accelerate to 6 GeV Seed with 250 nm, 50 fs seed laser Using cascade 1 st cascade nm 2 nd cascade nm 3 rd cascade 10 2 nm 4 th cascade 2 nm 4 Å 5 th cascade 4 1 Saturate FEL radiation in the 5 th cascade achieving Fourierlimited (fully coherent) and tunable output New possibility: Echo (G. Stupakov) reduce to 23 stages?

11 11 HGHG Xray FEL at 10 kev FEL undulator parameters Period 18 mm 0.25nm 0.124nm K = Gap 6 mm Electron beam parameters Energy 6 GeV Normalized emittances 1.5 μm rad Energy spread 0.01% Peak current 2.5 ka Bunch charge 1 nc Betatron wavelength <β> = 200 m Distance along FEL, m FEL radiation parameters SASE HGHG FEL parameter ρ Gain length 10.5 m Saturation length 200 m 80 m (last cascade) Saturation power 1.8 GW 1.4 GW Pulse length 400 fs 50 fs Number of photons (10 kev) Radiation bandwidth(rms) (Fourierlimited) Repetition rate to reach MHz 0.17 MHz Beam current to reach ma 0.17 ma Heat load: 14W HGHG compared with SR 10kW (NSLSII) HGHG FEL peak power, GW

12 12 Electron beam requirements intrinsic energy spread modulation energy energy period chirp time modulation amplitude bunch length energy spread Charge/Energy: 1nC at 6 GeV Final bunch length: 400 fs (FWHM) Final energy spread: <10 4 (RMS) Energy modulation: energy chirp < within 50 fs Density bunching: to be studied Emittance: 1.5 μm normalized Seed laser Electron bunch time jitter: <50 fs (pp) DESY and MIT achieved 40 fs RMS and improving 50fs nm 400fs

13 Estimate on energy modulation requirement and bandwidth Based on phase noise 1.If we have a modulator at 250nm, the energy modulation produced is Δγ/γ=10 2, then to generate sufficient bunching, the dispersion should generate a phase shift of Δθ~1. Take Δθ~1, then the dispersion should be dθ/(dδγ/γ)= Δθ/( Δγ/γ)=1/10 2 = Assuming due to space charge or CSR effect, there is an energy chrip in the electron bunch. The energy chirp within 50fs is assumed to be δγ/γ=5*10 5. Then, it causes an additional undesired phase shift of dθ/(dδγ/γ)* δγ/γ=10 2 *5*10 5 =5*10 3.

14 Estimate on energy modulation requirement and bandwidth Based on phase noise 3. This phase shift moves the zero point in the field by δl=λ*δθ/2π=250nm*5*10 3 /(2π)=0.2nm 4. Since the laser pulse length is 50fs=16μm, the phase error of 0.2nm gives the wavelength shift of δλ/λ=0.2nm/16μm=1.24* Assuming this wavelength shift is random, it generates a bandwidth increase as 1.24*10 5. Thus to achieve bandwidth of 1.24*10 5 we require the energy chirp error < 5*10 5, within 50fs. 6. However, if we want output to be Fourier transform limited for a flat top pulse of 50 fs at 0.1nm, the bandwidth should be 7.3*10 6 instead of 1.24*10 5. Thus we need to reduce the energy modulation by a factor of 12.4/7.3~1.7, thus the required chirp should be < 5*10 5 /1.7=3*10 5

15 Estimate on energy modulation requirement and bandwidth Based on phase noise 7. Can we achieve this small energy modulation? 8. For echo method, does this apply? 8. How about density modulation?

16 16 Regenerative pulses Regenerative pulse Pulse 266nm stacking delay G Linac Linac FEL Ch 266nm Seeding pulse C 266nm Mod B Rad Linac 53nm L 800nm 266nm Regenerative seeded scheme Regenerative ignition scheme Greatly reduces requirements on synchronization and energy stability Greatly reduces requirements on drive and seed laser powers self ignition Needs detailed analysis With new synchronization technology go to within 50 fs, this method will provide further improvement

17 17 Doublebunch to reduce noise? t Long wavelength, use lower current Short wavelength, use higher current Most noise comes from the first stage of cascade at long wavelength. Lower compression may have smaller energy modulation. Can this achieve required modulation tolerance? Can double bunches be generated by one RF gun? Or one injector? Can double bunches be generated by one RF gun within 2 RF cycles?

18 18 EXTRA

19 19 Crude estimate Longitudinal Space Charge Collective effects Collective effect impedances 100 Coherent Synchrotron Radiation (SteadyState) Linac wake Impedance, Ohm/m Zlsc( λ, 150) Zcsr( λ) Zlw( λ) final 1 compressor initial Energy modulation amplitude σa transverse bunch size, γi beam energy, R bending radius in chicane s dipole, a iris size in cavity, λ modulation wavelength. Lowenergy accelerator: to be calculated Undulator chamber wake: to be calculated λ Modulation wavelength [μm]

20 20 1D beam dynamics estimate Beam phase space (Δγ/γ in %, Δt in ps) Compression scenarios: Without linearizer High peak current is only required in the last stage 4 1 Å Idea: use natural nonlinearity of RF wave and produce a sharp spike at beam head Sharp spike comes with drastic increase in energy spread With linearizer: Flat beam with required peak current Linearizer amplitude controls final bunch shape Final beam phase space strongly depends on realistic intrinsic energy spread Choice of accelerator parameters Optimization is required: minimize max linac energy, maximize jitter tolerances I p, 0.1 x ka Final phase space before FEL 0.3 Energy GeV 0.25 and ES % 0.2 RMS Bunchlength ps Final phase space before FEL Energy GeV and ES % RMS Bunchlength ps Final phase space before FEL No linearizer Full compensation Energy GeV and ES % RMS Bunchlength ps Optimum BROOKHAVEN SCIENCE 0.8 ASSOCIATES 1

21 21 Energy modulation amplitude Induced energy modulation Relative energy modulation at 6 GeV I A Alfven current L action length δ initial density modulation Crude estimate; need careful consideration of Landau damping through uncorrelated energy spread Example of energy modulation driven by: LSC at 150 MeV throughout 1 st compressor CSR through the 2 nd compressor LW through the last linac Each time driven by 1% initial density modulation The beam parameters are demanding Careful analysis is required Energy modulation amplitude Δγ1 ( λ) γ f Δγ2 ( λ) γ f Δγ3 ( λ) γ f Our requirement λ C 1 Modulation wavelength, um

22 22 Gun Parameters at injector exit Our specs CAD BNL 1 CAD BNL 2 Rossendorf 3 Frequency (MHz) Rep. rate (MHz) 0.2(HGHG) (0.14) (SASE) Charge (nc) Emittance (μm) h:8.1, v: Energy spread corr. (kev) n/a Energy spread uncorr. (kev) 5??? Bunch length (ps)

23 23 Drive and seed lasers Gun drive laser ERL gun example JLab gun, 100 ma is generated by 10 W 527 nm light focused on 4 mm ø spot JLab NEA GaAs cathode: QE is ~2.5% (10% max) Estimate using 1% QE (Cs 2 Te, 266 nm) for High Spectral Flux Source: Gun drive laser energy for 1 nc is 400 nj Gun drive laser power for 1 ma (with 10x losses) is 4 W Seeding laser 50 fs and 0.5 GW 25 μj per pulse at 266 nm 200 khz rep rate 50 W (with 10x losses) Drive and seeding laser repetition rate 200 khz 1 MHz MIT lasers: 2 ps, (266) nm at 110 khz Synchronization of ~50 fs pp is required between electron beam and seed laser 10 fs RMS synchronization has been discussed for MIT FEL proposal High Spectral Flux Source High Spectral Flux Source laser system requires intensive R&D High Spectral Flux Source FEL might require alternative R&D solutions like ignitionseeding

24 Cavity and WallPlug Power Gradient 17 MV/m Cavity power = 22.4W, Beam power = 17 kw, Loaded Q ~ 1.7x10 7, near limit of practical values P g 2 V acc Δ ω2q = 1 + R 2 Q ext ω Q 0 ext 2 P g 17.5kW (Δω = 40 Hz) For 6 GeV, = 353 x 17.5kW = 6177,5 kw 2.5 GeV chirp for compression= 147x 0.5kW = 73.5 kw Total 8.5 GeV, 6.25 MW RF Refrigerator power = 22.4 W RF + 1.5W HOM + 1 W static = 25W/Cavity 25 W x 500 cavities = 12.5 kw at 2º K AC power (RF) = 2 x 6.25 MW (50% efficiency) = 12.5 MW AC power (cryo) = 600 (2º 300º) x12.5kw = 7.5MW Total AC power = 20 MW

25 HOM power dissipated in cavity Due to low beam current (1mA) and relatively low repetition rate (1 MHz) the power dissipated at 2 K is moderate up to 100 GHz. Below cutoff HOM mode dampers limit HOM field amplitudes to insignificant values as far as power dissipation in the cavity* Above cutoff analysis shows even small amount of damping is sufficient Work to continue Frequency Band (GHz) Beam Power lost (W) Assumed Q Power dissipated in cavity (W) *Following Merminga et al, LINAC 2000 Frequency of 1300 MHz is chosen for smaller size, lower stored energy

26 26 LINAC Cavities ½ 9cell linac cavity Frequency (MHz) E_zero_T (MV/m) Q 1.3* * *10 10 Operating T (K) Power/cav (W) Stored Energy (J) Total linac energy 8.5 GeV Energy gain per cavity 17MV 1 meter active length m ends + 0.5m focusing/compression = 2m/cav 500 cavities 1 kilometer linac 11.2 kw RF 2K 1 ma beam x 17 MV = 17 kw/cavity + ~0.5 kw field control = ~17.5 kw/cav RF 6 GeV electron beam + 2.5GeV coasting beam 6.25 MW RF 20 MV/m 425 cavities, 850 m length, cryolosses K 15 MV/m 567 cavites, 1133 m length, cryolosses 10

27 27 gun Preliminary High Spectral Flux Source layout Injector chicane I chicane II FEL Linac X Linac Linac linearizer dispersive section seed Xrays Ebeam Seed modulator radiator HGHG cascade RF frequency 1.3 GHz Installed linac energy is 8.5 GeV Total compression ratio is 7 x 5.5 = ~38 Optimization is needed: similar optimization process was implemented at LCLS Energy ΔE Phase R 56 Injector Linearizer 3 rd st chicane nd linac nd chicane rd linac MeV MeV deg. cm

28 Light Source Parameters NSLSII: Electron beam: 0.5 A, 3 GeV, 1 nm rad, 0.1% e.s. Undulator: 14 mm period, 2 m long, K 2.2 LCLS FEL Electron beam: 120 na, GeV, 42 pm rad, 0.008% e.s. Undulator: 30 mm period, 120 m long, K=3.7 European XFEL Electron beam: 58 μa, <50 GeV, 26 pm rad, 0.008% e.s. (@30 GeV) Undulator: 45 mm period, 311 m long, K=4 K.J. Kim, S. Reiche and Y. Shvyd ko (ERL07): FELO = FEL Oscillator Electron beam: 25 ma, 7 GeV, 6 pm rad, 0.02% e.s. (Cornell ERL) Undulator: 1.9 cm period, 2330 m long, K=1.4

29 High SF Xray source: undulator radiation Undulator Radiation (UR) spectrum at ~10 kev Calculation parameters: E beam = 3 GeV, I beam = 0.5 A, ε x /ε y = 1/0.01 nm, σ γ /γ = 0.1%, β x /β y = 2 m. UR spectra per 0.1% BW 30 m away vs aperture size 10x10 mm, 3x3 mm, 1x1 mm. Relative bandwidth of a user experiment is 10 4 Natural bandwidth of central cone from an undulator is 1/N periods 10 2 Reducing aperture = loosing large factor in flux Photons outside 10 4 bandwidth are useless and create harmful heat on user equipment 7x10 15 Ph/s/0.1%bw undulator ~10 kev radiation 19 mm, 3 m, K= m Δε ph /ε ph = Photon Energy, kev 15 aperture 20keV

30 30 Can Undulator Radiation be used as High Spectral Flux Xray source? undulator ~10 kev radiation aperture 19 mm, 3 m, K= m use10 4 as our unit bandwidth Undulator Radiation (UR) spectrum at ~10 kev (NSLSII) 7x10 15 Δε ph /ε ph = Natural bandwidth of central cone from an undulator is 1/N periods 10 2 Reducing aperture = loosing large factor in flux Photons outside 10 4 bandwidth are useless and create harmful heat on user equipment Ph/s/0.1%bw Aperture size: 10x10 mm, 3x3 mm, 1x1 mm Photon Energy, kev 15 20keV

31 31 Betatron wavelength=190m Gainlength =10.5m, current=2500amp, emittance=1.5mu Wiggler period=1.8cm, energy spread 1e4, at 6GeV Sturation at 80m with 1.4GW. If Betatron wavelength=300m, gainlength is 9m But saturation alos at 80m but power drops to 0.3GW. This plot is for stage from 0.25nm to 0.124nm HGHG set zwaist=1.5 set zmax1=6 set nstepz1=9 set nstepz2=20 set nstepz3=160 set npart=96000 set zmax3=200 set aw01= set aw02= set xlamdb2=190 set prad01=1e9 set dpsdgm=0.6 set rxbeam=

32 32 Charge/Energy: 1nC at 6 GeV FEL output feasibility of undulator design machine footprint, cost, Electron beam requirements modulation energy energy period chirp time modulation amplitude bunch length energy spread Final bunch length: 120 fs (RMS) determines bunch peak current for FEL should be long enough for 5 stages in FBT Final energy spread: <10 4 (RMS) High HGHG efficiency?product of initial energy spread, bunch length and compression ratio Energy chirp: 0 intrinsic energy spread increases FEL bandwidth Energy modulation: wavelengthdependent impacts on FEL bandwidth energy chirp < within 50 fs Density bunching: TBD by ebeam dynamics impacts on energy modulation Emittance: 1.5 μm normalized determines FEL gain feasibility of gun design

33 33 FEL bandwidth Free Electron Laser narrow bandwidth Some experimental data Small gain regime: Linewidth narrowing in VEPP3 optical klystron 625 nm (V. Litvinenko et al., IEEE QE V27, 1993) High gain regime: DUVFEL at BNL 266 nm (L.H. Yu et al., PRL91 V7, 2003) High gain regime: at DUVFEL bandwidth is proven to be Fourierlimited σ ω x σ t = constant that depends on the radiation line shape Longer bunch narrower bandwidth Intensity (a.u.) Spectra of HGHG and SASE at 266 nm HGHG 0.23 nm FWHM SASE x10 5 Wavelength (nm)

34 34 HGHG FEL High Gain Harmonic Free Electron Laser narrow bandwidth seed dispersive section output We developed DUVFEL at SDL and tested HGHG properties in UV modulator radiator Measured FEL bandwidth Δε ph /ε ph = 266 nm (L.H. Yu et al., PRL91 V7, 2003) DUVFEL bandwidth is proven to be Fouriertransformlimited spectrum of flat top: Δε ph /ε ph = λ/δz Intensity (a.u.) Comparison of HGHG radiation with nonsaturated SASE HGHG 0.23 nm FWHM SASE x10 5 Wavelength (nm)