A facility for Femtosecond Soft X-Ray Imaging on the Nanoscale

Transcription

1 A facility for Femtosecond Soft X-Ray Imaging on the Nanoscale Jan Lüning

2 Outline Scientific motivation: Random magnetization processes Technique: Lensless imaging by Fourier Transform holography Feasibility: Single shot imaging at LCLS Equipment and costs Outlook: Future capabilities and experiments

3 Occurrence of random processes in magnetism Two examples: 1. Science: Imaging of critical magnetic fluctuations Direct observation of random magnetization dynamics occurring in equilibrium at Curie temperature 2. Technology: Origin of speed limit for magnetization switching Deterministic (repeatable) processes are required for technology but Random processes determine technological limit! Imaging of random dynamics requires recording of snapshots Jan Lüning Andreas Scherz H.C. Siegmann Bill Schlotter Yves Acreman Jo Stöhr Stefan Eisebitt Christian Günther Wolfgang Eberhardt

4 Phase transitions Phase transitions play an important role in a wide variety of materials and applications Theory groups phase transitions in classes, which following common, universal scaling laws Computer simulations allow detailed analysis of phase transitions in model systems Ferro- to paramagnetic phase transition is example of 2 nd order phase transition for which the order parameter (correlation length) diverges at critical temperature T c

5 Magnetic phase transitions Susceptibility T¼T C M(T)/M 0 Magnetization T<T C Â(T;H! 0) T>T C Image of critical fluctuations is computer simulation of Ising model (Web page of Schwabl, TU Munich). T / T c Critical fluctuations in 3D are expected to be small and fast In 2D fluctuations expected to be larger

6 2D critical magnetic fluctuations: Size and dynamics 1 ns 30 µs (T c ) o = Scaling laws ³»(Tc )» o z= 2 Relaxation time 10 nm 1 µm Â(T c ) Â o = ³»(Tc )» o =º Correlation length» (T T c ) T c Input for scaling laws: ξ 0 is range of spin correlation in ferromagentic phase Relaxation time o from FMR line width Susceptibility Â(T) for 1.8 ML Fe/W from Back et al., Nature (95)

7 Experiment Grow thin film with T c just above room temperature Let sample temperature drift slowly through T c Measure Magnetization Susceptibility Record image of magnetic domain structure M(T)=Mo Â(T;H! 0) T=T C This yields: Images of critical fluctuations arbitrarily close to T c ξ(t) Test of scaling laws

8 The technology challenge: Smaller and faster size speed factor ~ 100 to go factor 30 x 150 ~ 5000 to go Speed limit determined using ultrafast SLAC electron pulse as pump pulse. Probe pulse was subsequent imaging of static magnetic domain pattern.

9 The technology challenge: Smaller and faster To understand what is happening, we need to probe while the magnetic field is applied and right afterwards. 100 fs 10 ps Speed limit determined using ultrafast SLAC electron pulse as pump pulse. Probe pulse was subsequent imaging of static magnetic domain pattern.

10 The length of current x-ray probe pulses not enough intensity need to repeat over and over blind to random processes Repeatable process can be studied by pump-probe To understand why switching is not deterministic anymore one needs snapshot probe

11 Laser-generated current pulses Photoconductive Auston Switch Photoemission into Vacuum SLAC e - beam pulse not compatible with x-ray probe Generate laser-based current pulse for - magnetic field pulse - spin current pulse Estimated Peak Current: Experiment: ma Potential for >>10 A peak current Estimated Pulse Record Rise Time: image 5 ps of magnetization Pulse dynamics in the picosecond during range Estimated Pulse and Width: right 10 -after 200 ps the magnetic field pulse Estimated Peak Field: T

12 Fourier Transform X-ray spectro-holography

13 Soft x-rays provide magnetic contrast XMCD Fe L 2,3 a σ σ σ σ M M M M Cross-section data from Jeff Kortright (LBNL)

14 Digital image reconstruction Single Fourier transformation of scattering intensities yields the auto-correlation of sample, which contains image of sample due to the off-axis geometry in FT holography. (correlation theorem) Autocorrelation (Patterson map) Sample Mask 2 µm 10% - 90% intensity rise over about 50 nm RCP Intensity in image center, which contains self-correlation of apertures, is truncated.

15 Feasibility of single shot imaging at LCLS Minimize sample heating by 1. Phase contrast to lower absorption cross section 2. Reflection geometry to reduce film penetration 3. Multiple reference holography to better utilize photons Experiment Detector - Central beam stop OK for Fourier Transform holography - Higher dynamic range desirable Equipment and costs Outlook

16 1. Phase contrast soft x-ray holography Refractive index is complex: n = 1 - δ + iβ Co C. Mertens et al PRB ev FTH yields autocorrelation (correlation theorem): a a = Ŧ -1 ( Ŧ(a) 2 ) real part of AC Attenuation imaginary part of AC Phase shift On resonance λ x = 15 nm

17 1. Phase contrast soft x-ray holography Refractive index is complex: n = 1 - δ + iβ Co C. Mertens et al PRB Imaging with phase contrast before absorption resonance FTH yields autocorrelation reduces absorbed (correlation energy theorem): by factor of a >10 a = Ŧ -1 ( Ŧ(a) 2 ) real part of AC Attenuation imaginary part of AC Phase shift Before resonance λ x = 600 nm

18 2. Dichroic reflectivity Angle of total external reflection is dichroic Fe L 2,3 Phase Contrast 2 2 difference in angle of total external reflection yields respective reflectivity of domains with magnetization parallel and antiparallel to incident light of 50% and 1%.

19 3. Multiple reference Fourier transform holography SEM Linear signal improvement with number of references 5 µm 10 4 photons on detector 1 image 5 images 10 8 photons

20 Experiment: The IDEAL reflection sample design Side view Top view Cu single crystal substrate (green) 2 monolayer thin Fe film on substrate area titled upwards by 3 (red) 100 nm Au nanostructures, ideally cubes with 3 tilted surface (gold) 3 upwards tilt not required, but it would eliminate: substrate heating specula reflection from substrate, no central beam stop might be needed background scattering from surface roughness Fe film can be grown continuously on substrate

21 STEP 1: Unfocused, pink LCLS pulse LCLS parameters photons per pulse at 800 ev 5.7 µrad divergence (FWHM) at 800 ev about 150 m behind undulator 855 µm FWHM spot size at sample 10 7 photons per µm 2 within FWHM area of beam 10 7 photons on 5 µm sample area for 4 incidence

22 Sample heating analysis 50% and 1% reflectivity for 4 incidence angle and 705 ev (phase contrast) (similar to 100% contrast of F sample) 10 5 photons in scattering pattern needed for 5 µm diameter sample 5 10 reference beams incident photons required for overall zero magnetization Even if all energy remained within the thin film, i.e., assuming no ~1700 escaping photons Auger are absorbed electrons by or the fluorescence atoms photons, within the we less deposit: reflective domains of the Fe film (µ x = 1 / 600 nm -1 ) less than 1 mev per Fe atom in 2 Fe layers ~2900 about photons 2 mev are per absorbed Cu atom by in the top 4 10 mono 8 atoms layers of each Cu mono layer in regions below less reflective domains (µ x = 1 / 350 nm -1 ). This corresponds to a temperature raise of 1/50 ~10 of these K in Fe numbers film for high reflective domains. ~30 K in Cu substrate» Could use full unfocused beam!

23 Equipment and costs Small angle scattering setup (2Θ = 20 ) Layout of SSRL chamber ~150k + CCD Beam aperture chamber Sample chamber with Temperature control Earth field compensation Sample characterization - Susceptibility - Magnetization Sample manipulation Area detector Central beam stop OK for Fourier Transform holography Higher dynamic range desirable

24 STEP 2: Focused, pink LCLS pulse Spatial resolution is flux limited 116 µm FWHM source diameter 10:1 demagnification compatible with sample illumination and required momentum space resolution 10 µm FWHM spot size at sample with photons per pulse Focusing increases incident flux by factor of 10 5» 100 ev per Fe atom, 200 ev per Cu atom

25 STEP 3: Monochromatic LCLS pulse Intrinsic band width of LCLS sufficiently small for scattering from XMCD contrast Monochromaticity on the order of 5,000 would allow imaging with contrast originating from: XMLD (antiferromagnets) Element specific mapping of functional groups Study dynamics by pulse stretching Grating monochromator stretches pulse length C-value of PGM offers control over pulse stretching Reduces peak brightness by preserving integrated photon flux! Compare to stretching of electron beam pulse

26 STEP 4: Split LCLS pulse D 1 Ω, q max FEL pulse BS α M M D S Θ 2Θ D 2 Ω, q max X-ray probe X-ray probe Record consecutive images to follow ultrafast time evolution uneven pulse split X-ray pump X-ray probe

27 Future experiments Snapshot imaging on the nano-scale with femtosecond time resolution enables investigation of non-reproducible component of dynamics. Of interest to us are: Equilibrium dynamics Nanoscale ordering corresponds to fs ps time scale Spin and charge dynamics in correlated electron systems Short-range structure in liquids Non-equilibrium dynamics Spin relaxation in ferromagnets close to the Curie temperature

28 LCLS commissioning Unique contribution to LCLS characterization Pulse by pulse imaging of the X-ray source Mutual coherence function can be measured with non redundant pinhole array from each x-ray pulse. From mutual coherence function one can calculate an image of the source. Feasibility demonstrated at SSRL