High harmonics from gas, a suitable source for seeding FEL from vacuum-ultraviolet to soft X-ray region (XUV) UMR 7639 G. LAMBERT Palaiseau - FRANCE 22/08/11

UMR 7639 Laserlab Integrated Infrastructures Initiative RII-CT TUIXS European project (Table top Ultra Intense XUV Sources) FP6 NEST-Adventure n European Research Council Paris ERC project M.E. Couprie, M. Labat, O. Chubar Synchrotron Soleil, Gif-sur-Yvette, France D. Garzella, B. Carré CEA, DSM/SPAM, Gif-sur-Yvette, France T. Hara, H. Kitamura, T. Shintake, Y. Tanaka, T. Tanikawa SPring-8/RIKEN Harima Institute, Hyogo, Japan B. Vodungbo, J. Gautier, A. Sardinha, F. Tissandier, Ph. Zeitoun, S. Sebban, V. Malka LOA ENSTA-Paristech, Palaiseau, France J. Luning LCPMR, Paris C.P. Hauri Paul Scherrer Institute, Villigen, Switzerland M. Fajardo Centro de Física dos Plasmas, Lisboa, Portugal Thanks to JSPS

XUV Free Electron Lasers (FEL) : SASE 1 FLASH (2004, down to 4.1 nm), SCSS (2007, 50 nm), SPARC (2009, 160 nm)… SASE: Self Amplified Spontaneous Emission -High number of photon : photons -Short pulse duration (sub ps) peak power (GW) -Relatively high repetition rate (tens Hz) -High wavelength tuning -Variable polarization -Good wavefront (Bachelard, PRL 106, 2011) -Good spatial coherence (Ischebeck, NIMA 507, 2003) very performing tool for user experiment but… λ, t Accelerator

SASE limitations 2 1 Wavelength (nm) Intensity (arb. unit.) Weak gain at short wavelength, single pass long undulator (tens m) -Relatively important shot to shot variations: intensity, temporal/spectral profile jitter for pump-probe experiments limited temporal coherence (Saldin, Opt. Commun. 202, 2002)

λ, t -Coherent Improvement of the temporal coherence -Intense Decreasing of the saturation length => shorter undulator How to reduce/supress SASE limitations: “seeding fully coherent light” in XUV ? 3 Accelerator SASE Seeded About the same gain XUV radiation : Harmonics produced from gas (HHG) λ, t External source Previous demonstrations in IR with CO 2 and Ti: Sa lasers, then in UV with crystals (Yu, Science 289, 2000)

Why proposing High Harmonics Generated in gas (HHG) ? 4 Rare gas Laser Lens ~10 14 W/cm 2 Odd harmonics Harmonic order Number of photon Cut-off ● Spatially and temporally coherent ● Relatively tunable ● Short pulse duration ( fs) ● Conversion efficiency (10 -5 to ) ~ 100 nJ - nJ -Limited energy per pulse Compared to SASE shot noise?

-Regular and quasi-perfect Gaussian shape -Spectral width / ~5 -Pulse duration / ~20 Unseeded: 1 ps HHG: 50 fs Seeded: 50 fs Strong improvement of the temporal coherence/SASE -High amplification Direct seeding with HHG at 160 nm at SCSS: fundamental spectrum SCSS Test Accelerator (Japan) at 150 MeV with 4.5 m long undulator sections Wavelength (nm) Intensity (Arb. Un.) SASE x 13.7 HHG x 9270 Seeded 2 μJ (40 MW) 5 (Lambert, Nature Physics 4, 2008)

Intense and coherent emission at short wavelength while E=150 MeV Unseeded Seeded: 3 nJ (50 kW) Intensity (arb. un.) Wavelength (nm) Unseeded x 125 Intensity (arb. un.) Seeded: 4 nJ (75 kW) -Odd and even harmonics from 2 nd (80 nm) to 7 th (23 nm) -Clear amplification -Regular and quasi Gaussian spectral shape -Spectral narrowing -Shorter pulse duration Improvement of the temporal coherence Wavelength (nm) Unseeded H5 H6 H7 Intensity (arb. un.) Wavelength (nm) H2H3 Non Linear Harmonic spectra (Tanikawa, EPL 94, 2011) Seeded: 55 pJ (1 kW)

Evaluation of the seed level requirement for observing coherent emission For E seed ≥88 pJ: -Stable emission -Regular and quasi Gaussian spectral profiles Seeding in XUV ! Wavelength (nm) Normalized intensity (arb. un.) Vertical position (mm) E FEL =12.2 nJE seed =88 pJ HHG E seed =0 pJ E seed =88 pJ E seed =14 pJ E FEL =1.4 nJ E seed =0 pJ E FEL =0.87 nJ HHG seedE seed =88 pJ Vert. pos. (mm) 88 pJ (160 nm) ―> 10 pJ (spatial overlapping) or ~ 200xP e, shot noise At 13 nm (P e, shot noise E seed < 0.7 nJ (200x10Wx10x35fs) 8 Very weak level of injection Below 13 nm is a challenge! (Lambert, EPL 88, 2009)

NameWavelength (nm)stateSeeding processCountry SCSS Test Accelerator 60 / 160demonstratedDirect HHGJapan SPARC160 / 266demonstratedDirect HHG / HGHGItaly sFlash38-13Due Direct HHGGermany Fermi100-10Due HGHG, direct HHG?Italy SwissFEL5Due 2018EEHG, direct HHGSwitzerland SPARX15Due ?HGHG, direct HHG?Italy Overview of the HHG seeded FELs 9 (Giannessi, Proceedings FEL10 JACOW (2011) Labat WE0B3) (Togashi, Optics Express 19, 2011)

Keep the simplicity of the classical HHG setup Already obtained -fs pulse duration -Full coherence -High repetition rate: kHz HH currently First MHz HH in xenon (J. Boullet et al. optics letters, 34, 1489 (2009)) To be improved -Intensity at short wavelengths -Tuneability (only odd HH): => need to considerably chirp the driving laser and/or change the gap of the undulator -Wavefront: diffraction limited beam (aberration-free) => need to drastically clip the IR or HH beam or use adaptive optics for IR or HH -Variability of the polarization -Stability of the shot to shot intensity Harmonics properties relevant for seeding What has to be improved on HHG for seeding FEL in future? 10 Two colour mixing

Lens Ti: Sa ω Harmonics Noble gas IR Filter ~10 14 W/cm 2 Technical principle: BBO crystal 2ω2ω ω +2ω X 25 Neon -double harmonic content even types: 2x(2n+1) from 2  2x(2n) from the mixing -increase of the number of photons -redshift: High order harmonics generated with a two-colour field (Lambert, NJP 11, 2009) 11

Ar, ω +2ω He, ω +2ω Ne, ω +2ω Xe, ω +2ω Ar, ω He, ω Ne, ω Xe, ω 100 µm thick BBO crystal, and with the optimization parameters corresponded to ω: E ω <6 mJ, L C =7-9 mm and P G =30-35 mbar Al cut Wavelength (nm) Intensity (arb. un.) (x 100) (x 25) (x 0.5) -Flat spectra (same intensity level for odd and even harmonics) -Increase limited at high wavelengths due to an already relatively high efficiency for Xe and Ar => ω+2ω technique compensates the weak efficiency at short wavelengths Harmonic spectra obtained with either ω or ω+2ω technique (Lambert, NJP 11, 2009) 12

I 2ω ~1 I ω ~0.8 H17 H15 H19 H21 H23 I ω ~1.2 I 2ω ~2.9 I ω ~3.9 I 2ω ~5.4 I ω ~4.4 H22 H18 H14 H27 H25 Φ=20 mm /6 rms  /5 rms /17 rms Φ=40 mm Φ=20 mm 100  m thick BBO crystal -iris clipping technique: change the focusing geometry/energy clean the major part of the distortions in the outer part of the beam: to /6 rms -ω (L C =8 mm and P G =30 mbar) to ω+2ω (L C =4 mm and P G =16 mbar) -very high increase on 2x(2n+1) type of even harmonics (50 nJ) due to strong blue/IR and distortions limited:  /5 rms -iris clipping: limited decrease of intensity But distortions about /17 rms: First aberration-free high harmonic beam Wavelength (nm) Energy per pulse (nJ) I ω and I 2ω in W.cm Φ=40 mm Optimization of both flux and wavefront (Ar gas) (Lambert, EPL 89, 2010) 13

Already obtained -fs pulse duration -Full coherence -High repetition rate: kHz to MHz soon -Intensity at short wavelengths -Tuneability: both odd and even harmonics Use parametric amplifier (  m) -Wavefront: aberration-free beam -Simple system To be improved -Variability of the polarization -Stability of the shot to shot intensity? Then next? 15

Circularly polarized HH -Using circular IR beam -Two step setup: linear IR beam + polarizer (Vodungbo, Optics Express 19, 2011) 16

User applications in XUV: SASE-FEL vs HH In XUV performances are close: weak flux harmonics weak temporal coherence Single shot coherent diffraction imaging of nano-object (Ravasio et al. PRL 103 (2009) 32 nm, 1μJ, 20 fs Reconstructed image with 119 nm resolution 17

Pump-probe experiments Natural synchronization between Laser and HH Coherent diffraction imaging of magnetic domains (Vodungbo, EPL 94, 2011) : -Magnetic domain orientation: 45°+- 10° -Width distribution of magnetic domains: 65 nm +- 5 nm 18 q~ λD/a D 1200 s Co M 2,3 edge at 60 eV Aligned magnetic domains MFM a

Ultra-fast demagnetization -much shorter time scale -better temporal resolution for HH? And/or due to XFEL at 800 eV (L edge) Integrated density gives magnetization : I ~ M 2 (Boeglin, Nature 465, 2010) 19

Thank you for your attention

5 How to perform a HHG seeding experiments ? Various undulator configurations e-e- Electron source Chicane Spectrometer device HHG Laser Gas Refocusing system 1)Generate HHG with as much as possible photons 2)Refocus HHG at the electron beam size inside the undulator for strong interaction 3)Align precisely HHG beam on the undulator axis 4)Tune HHG wavelength with FEL one 5)Synchronize ebeam and HHG below ps time scale

Stabilization of the FEL wavelength Intensity (arb. un.) Wavelength (nm) Seeded Data file number Intensity (a. u.) Wavelength (nm) ~60 s Intensity (a. u.) ~300 s Wavelength (nm) Data file number SASE 7

Stabilization of the FEL wavelength nm ± 0.10 nm rms Data file number Intensity (a. u.) Wavelength (nm) ~60 s Intensity (a. u.) ~300 s Wavelength (nm) Data file number +-57% nm ± 0.63 nm rms +-37% 7 -λ stabilized by a factor of 6 -Still important intensity variation (mainly due to synchronization trouble)

Lens Ti: Sa ω Harmonics Noble gas IR Filter ~10 14 W/cm 2 Theoretical and technical principles: BBO crystal 2ω2ω Semi-classical model in three steps: ω +2ω X 25 Neon -double harmonic content even types: 2x(2n+1) from 2  2x(2n) from the mixing -increase of the number of photons -redshift: High order harmonics generated with a two-colour field HHG IpIp x W E =ex.Esin  t Tunnel ionization t~T/4 Radiating recombination t~T Initial state t~0 Oscillation in the laser field t~3T/4 (Lambert, NJP 11, 2009) 11

General set-up Spherical multilayer mirror Transmission grating (2000 lines/mm) CCD Camera Spectrometer 800 nm 1 kHz < 7 mJ 35 fs BBO crystal, (100  m or 250  m thickness) Gas cell (4 to 7 mm long) Aluminium filters (200 nm thickness) Lens, f=1.5 m Calibrated XUV photodiode Iris Grid CCD Camera Wavefront sensor  y/L Hole arrayCCD camera L yy  Aberrated spot centroid position Aberrated wavefront Reference spot centroid position Reference flat wave front

2ω2ω BBO Waist ratio between ω and 2ω: iris Cleaning the distortions with the 2 nd harmonic generation Harmonics from ω+2ω Gas cell IR beam wavefront ω Filter Wavefront sensor A. U. 1.4 time smaller for V and H sizes From ωFrom ω+2ω HH source sizes A. U A. U. From ωFrom ω+2ω HH footprint sizes

Summary of wavefront optimization /6 rms /5 rms /17 rms ω +2ω, Φ=20 mm ω +2ω,Φ=40 mmω,Φ=20 mm (4) RMS (0.100λ), PV (0.483λ) Horizontal dimension (mm) Vertical dimension (mm) ω, deform. mirror (52 em act.), Ф=15 mm λ/10 rms  λ=32 nm (ω) to 44 nm (ω+2ω)

(b) (a) BBO crystal (100 μm thickness) Aluminium filters (200 nm thickness) SiO 2 flat mirror 45° incidence B 4 C/Mo/Si flat mirror λ/2 waveplate 800 nm 1 kHz < 7 mJ 35 fs Lens, f=1.5 m Iris Transmission grating (2000 lines/mm) Spherical multilayer mirror CCD Camera Optical system Gas cell ω H V H V H V ω 2ω2ω HHG from ω+2ω (c) ω, Hor. Pol. 2n+1 2n2n Φ 2n+1 2ω, Vert. Pol. Φ 2n

Time (fs) Intensity (10 14 W. cm -2 ) ω 2ω -variation of polarization between odd and even harmonics: Higher for longer wavelengths -variation of polarization from one order to another Aluminium filter

-80 nm-17 nm: intense XUV harmonic signal Ar (ω+2ω) intensity increase for 1 over 2 even harmonics Ne (ω+2ω) intensity about Ar (ω) intensity (typical reference) -13 nm : Ne (ω +2ω) about 0.3 nJ ( Lambert, EPL 88, 2009 ) ~ 200 x P SASE, noise at 13 nm (P SASE, noise ~ 10 W) => E seed ~ 0.7 nJ (200 x 10 W x 35 fs) -Intensity could have been more optimized with more energy and longer focusing Wavelength (nm) Ar, ω +2ω Ne, ω +2ω Ar, ω Estimation of the Ne spectra content: below 17 nm at 13 nm Energy per pulse (nJ) Al cut Perspectives of the two-colour HHG for seeding of FEL -Only relative control on the SHG/HHG ◘ I2w/Iw ok ◘  t =18 fs /100 μm not ok ◘ Δt2w/Δtw not ok Intensity (a. u.) Time (fs) ω 2ω2ω Δt 2ω tt 14