Present Capabilities and Future Concepts for Intense THz from SLAC Accelerators

Transcription

1 Present Capabilities and Future Concepts for Intense THz from SLAC Accelerators Alan Fisher SLAC National Accelerator Laboratory Frontiers of Terahertz Science SLAC 2012 September 6 1

2 THz from an Accelerated Electron Beam Narrow bandwidth, many cycles: Undulator radiation Corrugated pipe or dielectric tube (both under study at FACET) Broad bandwidth, short pulse: SR, ER: Synchrotron and edge radiation from dipole magnets TR: Transition radiation as beam crosses thin foil (LCLS and FACET) DR: Diffraction radiation as beam passes through a hole All are intense and coherent for wavelengths > bunch length Applications: Broadband THz source for user experiments Temporal profile and spectrum linked to the bunch: Beam diagnostic Coherent transition radiation (CTR) from highly compressed electron bunches in LCLS and FACET produces half-cycle THz pulses with energies of 0.5 mj and fields up to 6 GV/m (0.6 V/Å). 2

3 Options for THz Extraction at LCLS Undulator CTR Focusing mirror 33 mm ~7 m CER Annular mirror or Be foil Profile Monitor YAGS:DMP1: Feb :38:13 Dump magnets Diagnostics chamber X-Ray 3 < d < 100 mm y (mm) Visible ER CTR -5 3 mm Plane mirror Dump x (mm) Henrik Loos/Sep 5,

4 Coherent Transition Radiation in Undulator Hall Undulator CTR Dump magnets Calculations using: Beam energy = GeV Bunch charge = 250 pc Foil diameter = 22 mm 2:1 imaging optics Dump Henrik Loos/Sep 5,

5 CTR in the Undulator Hall : Under-Compression Calculations for an undercompressed beam: Bunch length = 7 µm RMS THz pulse energy = 380 µj Formfactor (%) Spectrum TR Efficiency Beam Form Factor TR Intensity 20 Current (ka) e-beam Electric Field (MV/cm) Wavenumber (cm -1 ) THz Peak Transient Time (fs) Time (fs) Henrik Loos/Sep 5,

6 CTR in the Undulator Hall : Peak Compression Calculations for a fully compressed beam: Bunch length = 2.5 µm RMS THz pulse energy = 1.5 mj Formfactor (%) Spectrum TR Efficiency Beam Form Factor TR Intensity 20 Current (ka) e-beam Electric Field (MV/cm) Wavenumber (cm -1 ) THz Peak Transient Time (fs) Time (fs) Henrik Loos/Sep 5,

7 Coherent Edge Radiation in the Dump Area Calculations using: CER from 1 st bend F/4 focusing optics Undulator 33 mm Focusing mirror Dump magnets Profile Monitor YAGS:DMP1: Feb :38: Visible ER 5 ~7 m Simulated ER 600 nm CER Diagnostics chamber Annular mirror or Be foil X-Ray 3 < d < 100 mm y (mm) 0 y (mm) mm x (mm) mm x (mm) Dump Henrik Loos/Sep 5,

8 CER in the Dump: Under-Compression Compare to Undulator Hall CTR: THz energy = 130 µj (66% lower) Diffraction in long beampipe loses wavelengths > 50 µm (< 200 cm 1 ) Formfactor (%) Spectrum TR Efficiency Beam Form Factor TR Intensity 20 Current (ka) e-beam Electric Field (MV/cm) Wavenumber (cm -1 ) THz Peak Transient Time (fs) Time (fs) Henrik Loos/Sep 5,

9 CER in the Dump: Peak Compression Compare to Undulator Hall CTR: THz energy = 700 µj (53% lower) THz pulse is not a half cycle Low frequency loss creates large negative spikes even at full compression Formfactor (%) Spectrum TR Efficiency Beam Form Factor TR Intensity Current (ka) e-beam Electric Field (MV/cm) Wavenumber (cm -1 ) THz Peak Transient Time (fs) Time (fs) Henrik Loos/Sep 5,

10 CTR in the Dump Line Compare to Undulator Hall CTR: X-ray FEL off: THz energy is 10 40% lower due to larger beam size X-ray FEL on: Energy spread increases, dispersion makes beam 10 larger, THz energy is 90% lower Energy jitter of 0.1% creates ~1mm of jitter in the THz source point Finite R 53 creates a y-z pulse tilt of order of the bunch length X-Ray Undulator Dump magnets CTR Plane mirror Dump Henrik Loos/Sep 5,

11 LCLS and FACET 2 km 1 km Positron Bunch Compressor PEP-II SLC Experimental Halls LCLS Electron Bunch Compressor W Chicane Compressor FACET Experimental Area Undulator Hall THz Sources 11

12 LCLS Parameters Electrons Energy GeV Charge per bunch nc Repetition rate 120 Hz Bunch length ( z ) µm ( t ) fs Size at THz foil ( x,y ) µm X Rays Photon energy ev Pulse energy mj 12

13 LCLS THz Source in the Undulator Hall THz is generated by coherent transition radiation in a 10-µm-thick Be foil Between undulator and beam dump: Both electrons and x rays pass through foil To e-beam dump and user hutches THz table X rays and electrons 31 m End of undulator 13

14 Beryllium Foil Assembly Pneumatic actuator Foil Diamond window Bellows 14

15 Be Foil and Holder Foil in Foil out 15

16 Optics Enclosure on the THz Table 16

17 Optics Enclosure on the THz Table Ti:sapphire laser for THz pump and optical probe Enclosure for laser safety, also purged with dry air (no THz absorption) Foil X rays and electrons Laser 17

18 The FACET User Facility at SLAC Facility for Advanced Accelerator Experimental Tests Provides highly compressed e (and soon e + ) bunches at high energy and with high charge Uses the first 2 km of the SLAC linac Experiments include: Plasma wakefield acceleration Dielectric wakefield acceleration Ultrafast magnetic switching Beam diagnostics with Smith-Purcell radiation Terahertz radiation 18

19 FACET Parameters (2012) Design Typical Best Electron energy GeV Charge per bunch pc Repetition rate Hz Bunch length ( z ) < 25 < 20 µm ( t ) < 83 < 67 fs Size at experimental IP ( x ) µm ( y ) µm Standard Optics Double Waist Size at THz foil ( x ) µm ( y ) 6 36 µm 19

20 THz Table and the FACET IP W Chicane (Compressor) IP Table THz source: Coherent transition radiation from two 1-µm-thick Ti foils m before main focus at experiments on the IP (Interaction Point) Table Allows parasitic operation and use of THz for beam diagnostics But electron beam at THz foil is larger than at IP 20

21 Upstream IP Table THz table Plasma oven Upstream IP table Downstream IP table 21

22 IP Tables 22

23 THz Table with Dry-Air Enclosure 23

24 Layout of the THz Table Michelson Interferometer Top View 1-µm Ti foil 1-µm Ti foil From bunch compressor CCD Visible Insertable silicon plate To IP THz Side View 24

25 Optics for the Upstream Foil Foil To Michelson OAP Knife edge Joulemeter 25

26 Michelson Interferometer Fixed mirror Scanning mirror Pyroelectric detector THz beamsplitter 26

27 Finding the Peak Electric Field Model the electric field at the focus with a simple Gaussian: This neglects the radial polarization and the negative lobes before and after peak x y z E E0exp x y z Then the energy in the pulse is related to the electric field by: U E dx dy dz E 2 3/ x y z Dependence on bunch charge q, bunch compression z : Field E 0 ~ q/ z Energy U 0 ~ E 02 z ~ q 2 / z Coherence: Less pulse energy, due to incoherent emission, for wavelengths x,y,z Measurements needed: Pulse energy U 0 Pyroelectric joulemeter and electron energy loss Pulse duration t = z /c Michelson interferometer Size at focus x,y Pyroelectric camera or knife-edge scans 27

28 Energy in a THz Pulse Joulemeters aren t black for THz Pyroelectric manufacturers provide rough correction factors of 2 to 4 LCLS: 200 to 400 µj, for 350 pc and 20 fs RMS (full compression) FACET: 460 µj, for 3 nc and 130 fs RMS (but with intermediate compression) Compare THz energy to electron energy loss when LCLS foil is inserted Beam dump serves as e spectrometer Transition radiation as beam enters, exits foil Entry THz goes through window Foil in 30% reflection loss on diamond Exit THz is not collected Measured loss: 2.9 mj for 350 pc and 20 fs So about 1 mj should be collected Calibration of THz joulemeter? Energy Loss (MeV) 8.7 MeV loss for 350 pc = 2.9 mj Foil out Repetitions 28

29 Bunch Profile: Finding the Form Factor Electric field E of a bunch of N electrons at positions t j : N i tj E( t) E ( t t ) E( ) E ( ) e j 1 1 j 1 where E 1 is the field of one electron. Energy in the pulse is related to the longitudinal form factor f( ): 2 2 d 2 i t 2 d j ( ) 1( ) 1( ) ( ) 2 j 2 U dte t E e N E f Interferometer gives the pulse energy in an autocorrelation with a delay : d 2 2 i U( ) U0 Re E1( ) f( ) e 2 The power spectrum U( ) is then (neglecting the DC component): 2 2 U( ) E ( ) f( ) E 1 is essentially constant at THz frequencies, and so U( ) gives us f( ) 2. 1 j 29

30 Michelson Interferometer Scans LCLS FACET 30

31 Michelson Spectra LCLS FACET Spectral width depends on bunch compression LCLS bunches are usually shorter and so have higher THz frequencies Reflections from detector layers modulate the spectra 31

32 Restoring Low Frequencies Finite size of beampipe and window remove low-frequency components Must compensate for filtering to get longitudinal bunch profile from the spectrum Model the low-frequency loss as a high-pass filter attenuating the field by: F ( ) 1 exp 2 2 hp 0 Filter causes the negative dips about the central interference peak in the time domain Fit region with filtered Gaussian beam to find 0 Then discard Gaussian assumption, but keep 0 Spectral compensation: Divide spectrum by filter function F hp Fit a parabola near f = 0, where 1/ F hp diverges Bunch reconstruction isn t very sensitive to this region 32

33 Kramers-Kronig Retrieval of the Phase and Profile If we express f( ) in terms of its magnitude ( ) and phase ( ), then: ln f( ) ln ( ) i ( ) Since ln and are real and f(t) is causal, the Kramers-Kronig relations give: 2 ln ( ) ( ) ( ) P d The magnitude ( ) and phase ( ) then let us compute the bunch profile f(t). 33

34 Profiles at Three Bunch Compressions High Compression Medium Compression Low Compression 34

35 Transverse Size at the THz Focus LCLS Pyroelectric camera LCLS pyroelectric camera: Pixel size = 100 µm Spot size 500 µm FWHM 200 µm RMS FACET knife-edge scans: x = 1.4 mm RMS, y = 1.1 mm RMS Why is the size at FACET much larger? FACET Knife-edge scans Horizontal Profile Vertical Profile 35

36 Standard Electron Optics near THz Table Large x beam size at THz table lowers pulse energy Beam size minimum is at experimental IP 36

37 Transverse e-beam Size Simulated beam with standard optics: x = 1.2 mm, y = 6 µm Measured with optical transition radiation 37

38 Double Waist to Reduce Transverse e-beam Size Simulation Comparing standard optics to a circular 85-µm beam Double-Waist optics Focus remains at IP, but x size at foil is reduced to x = 317 µm, y = 36 µm But an even smaller x would help, since the spectrum peaks at 1 THz (300 µm) Circular 85 x 85 µm 2 Elliptical 1200 x 6 µm 2 /c (mm 1 ) 38

39 Peak Electric Field LCLS Charge: q = 0.35 nc Energy: 100 U µj Size: x,y = 200 µm Length: 6 z 7 µm Electric and magnetic fields: 2.7 E 5.8 GV/m = 0.6 V/Å 9 B 19 T Quasi-half-cycle pulse Broad spectrum peaking at 10 THz FACET Charge: q = 3 nc Energy: 400 U µj Size: x = 1.4 mm, y = 1.1 mm Length: 20 z 40 µm Electric and magnetic fields: 0.36 E 0.68 GV/m = 0.07 V/Å 1.2 B 2.3 T Quasi-half-cycle pulse Broad spectrum peaking at 1 THz Higher fields should be possible with a smaller x at the foil 39

40 Detector Response THz detectors are poorly calibrated and are not spectrally flat Significant etalon effects (reflections from detector layers) Infratec has strong modulation Dip in response near 0.8 THz cuts out part of spectrum Not suitable for interferometer Compare to model (Henrik Loos) Compression: Low Medium High Detector Model Gentec has thinner layers 60-GHz modulation When 60 GHz is filtered, 240 GHz becomes visible Better, but not ideal Unfiltered Filtered 40

41 Water Vapor Before and after adding the dry-air enclosure at FACET Comparison to transmission through 1 m of humid air Without dry-air enclosure Dips near 0.75 and 1.2 THz 3 bunch compressor settings With dry-air enclosure Much improved, but there may still be some absorption Plan to improve the dry-air purge Compression: Low Medium High Humid air 41

42 Initial Concept: THz Pump and X-Ray Probe Bragg crystals delay the x-ray probe with a U-turn, so that they arrive at the sample after the THz pump. 42

43 Thin versus Thick Crystal Initial concept used a silicon crystal etched to 8 µm in center Advantage: In-band x rays diffract, but out-of-band x rays pass through Would permit parasitic operation Disadvantages: Electrons also pass through crystal Crystal must be thin to reduce hard radiation in user hutches downstream Heat deposited by x rays generates acoustic wave in the thin membrane Vibration disrupts Bragg diffraction Weak damping restricts operation to 1 Hz Newer concept uses thicker (100 µm) crystal with electron chicane Advantages: Electrons travel around crystal: No hard radiation Vibrations damped by clamping crystal to support Disadvantage: Not suitable for parasitic use 43

44 Concept: THz Pump and X-Ray Probe 44

45 Advantages and Disadvantages Advantages: Single bunch makes both THz and x-rays: No jitter No long transport line: Lower cost and no jitter from vibrating parts 3 Si crystals at 45 ±2.5 span photon energies from 7.59 to 9.56 kev Disadvantages: Undulator Hall would become an awkward user hutch Limited time for experimental setup: One shift every two weeks Access during run is possible, but slow: Insert e-beam stopper after linac Narrow diffraction bandwidth reduces x-ray energy Better match for narrow FEL bandwidth from hard x-ray self seeding Soft x rays are more difficult Multilayer mirrors, not crystals, needed to turn beam through 180 Usually reflectivity is low (away from grazing) and bandwidth is narrow To avoid absorption, x rays must not hit foil: Stronger chicane dipoles Must place sample in vacuum rather than dry air 45

46 Three Silicon Crystals Span a 2-keV Range Silicon Lattice Planes Energy (kev) at Angle (deg) h k l

47 Concept: Multilayer Mirror for Soft X Rays Reflectivity for S polarization of a calculated multilayer mirror 150 layers of 0.82-nm Cr on 1.13-nm Ti Suitable only over a narrow energy range Reflectivity versus Energy at 45 Incidence Reflectivity versus Angle for 450-eV Photons 47

48 Concept: THz Transport up to Klystron Gallery Transport THz from foil to an optical table in the Klystron Gallery 19-m path, including an 8-m vertical section through a penetration Characterize pulse before and after transport: energy, focus, spectrum Severe diffraction at these long wavelengths Large-diameter mirrors and tubing: 200 mm (8 inches) Frequent refocusing with off-axis parabolic (OAP) or toroidal mirrors Under vacuum to avoid water vapor or convection (UHV unnecessary) Hope for approval and funding in 2013 Future concept: Add 20 m to bring THz to Sector-20 laser hutch Location of new FACET laser, synchronized with electrons To pre-ionize the plasma for studies of plasma-wakefield acceleration Space for THz user experiments, including THz pump and laser probe 48

49 Concept: Transport Line at Top of Penetration Pumping flange Flange with mirror mount Alignment viewport View HeNe alignment beam directly or with a camera Bellows Flexibility Breaks up reflections from pipe wall 200-mm (8-inch) tubing 49

50 Concept: THz Transport to the NEH From Undulator Hall to Near Experimental Hall 100 m THz can t be transported along a straight path THz arrives 50 to 100 ns after x rays Probe/pump instead of pump/probe Remedy: Two electron bunches First bunch is optimized for THz High bunch charge and peak compression Second bunch, 100 ns later, makes x rays in the FEL Bunch charge and compression set to user s preference Integer number of RF buckets plus an optical trombone delay determines pump/probe spacing LCLS tested two bunches with 8.4-ns spacing in 2010 Now planning more tests with variable spacing 50

51 2 Bunches = 4 Pulses Each bunch emits both x rays and THz Arrival order: X ray 1, THz 1, X ray 2, THz 2 Early x-ray pulse interferes with pump-probe measurements Also, final THz pulse after pump-probe might be a problem Remedy: Suppress lasing by the first bunch High charge and full compression High laser-heater energy for first bunch Kicker to make first bunch oscillate in the undulator But the first bunch emits spontaneous in the undulator Could interfere with some experiments 51

52 Concept: Bypass Beamline for the THz Bunch Add a fast kicker before the undulator, sending bunch into a bypass 1 st bunch goes through bypass, parallel to undulator Put THz foil in the bypass Bunch makes THz but no x rays 2 nd bunch goes through the FEL Bunch makes x rays but no THz No trailing THz pulse after pump/probe Soft x rays are not absorbed by foil 52

53 Concept: THz for LCLS-2 Two bunches in each 120-Hz pulse Two ways to share them: Bunch 1 to Undulator 1, Bunch 2 to Undulator 2 Each user gets 120-Hz x rays, without THz Bunch 1 to bypass, Bunch 2 alternates between undulators Each user gets THz and x rays at 60 Hz 53

54 Comparing THz at LCLS and FACET LCLS FACET Location 30 m past undulator 10 m before FACET IP Source CTR from 10-µm Be foil CTR from 1-µm Ti foil Beam(s) Both electrons and x rays Electrons only Bunch length 20 fs RMS 65 fs RMS THz spectrum 3 30 THz THz THz electric field 6 GV/m 0.7 GV/m Pulse energy 400 µj µj Access to tunnel 8 hours every 2 weeks 8 hours every week Transport line Requires 100-m transport line and 2 e bunches. No plans now. 20-m line to Gallery possible in May extend to hutch later. Future concepts THz pump and x-ray probe Easy access to THz in a hutch 54

55 Summary Present capabilities: Intense, broadband, quasi-half-cycle, THz pulses produced by coherent transition radiation at LCLS and FACET The two sources are complementary: Shared instrumentation development Same potential user community Future concepts include: THz pump and x-ray probe, on a single table and with a single bunch Transport line to a hutch at FACET These ideas come with limitations in cost, access, and beam time. In planning the future role of THz here, SLAC will be guided by your ideas. Tell us where you see exciting opportunities in THz science, and what unique capabilities SLAC could offer to pursue them. 55