X-Rays and Magnetism Joachim Stöhr Stanford Synchrotron Radiation Laboratory Past Present Future Present: Size > 0.1 mm, Speed > 1 nsec Future: Size Ultrafast Nanoscale Dynamics.
Download ReportTranscript X-Rays and Magnetism Joachim Stöhr Stanford Synchrotron Radiation Laboratory Past Present Future Present: Size > 0.1 mm, Speed > 1 nsec Future: Size Ultrafast Nanoscale Dynamics.
X-Rays and Magnetism Joachim Stöhr Stanford Synchrotron Radiation Laboratory Past Present Future Present: Size > 0.1 mm, Speed > 1 nsec Future: Size < 0.1 mm, Speed < 1 nsec Ultrafast Nanoscale Dynamics Growth of X-Ray Brightness and Magnetic Storage Density Non Resonant X-Ray Scattering Relative Intensity: 1 Relative Intensity: (hn / mc2)2 hn ~ 10 keV, mc2 = 500 keV Fe metal – L edge Kortright and Kim, Phys. Rev. B 62, 12216 (2000) Soft X-Rays are best for magnetism! Core level binding energies give: Element specificity Chemical state specificity Magnetic Spectroscopy and Microscopy Part 1: Nanoscale Magnetism Real Space Imaging X-Rays have come a long way…… 1895 1993 PEEM-II at ALS Po la riz ed X-ra ys • • • • Full Field Imaging Electrostatic (30 kV) 20 - 50 nm Resolution Linear and circular polarization PEEM Contrast Mechanisms Topological Elemental Norm. Electron Yield 0.5 Tb 0.4 Fe 0.3 Co 0.2 0.1 600 800 1000 1200 1400 Photon Energy [eV] Magnetic Chemical bonding Photon Energy (eV) 4 Norm. Electron Yield Absorption Coefficient 10 8 6 4 2 0 690 700 710 720 730 740 3 2 Fe FeO x 1 0 703 705 707 709 711 Photon Energy [eV] Use soft x-rays – L edges of Fe, Co, Ni 713 Exchange Bias – Exchange Coupling FM 1 W FM 2 AFM coercivity increase (uniaxial anisotropy) exchange bias, loop shift (unidirectional anisotropy) Magnetization(a.u.) 1.0 0.5 0.0 -0.5 -1.0 -1500 -1000 -500 0 500 1000 1500 Applied Field (Oe) A ferromagnet behaves different when in contact with an antiferromagnet. No appropriate understanding yet – 45 years after discovery Conventional techniques cannot study the magnetic interface Co NiO Co on NiO(001) Analysis of dichroism contrast 3 dimensional spin structure s s [010] s 2mm Bare NiO(001) NiO after deposition 2nm Co on NiO(001) Ni spins near antiferromagnetic surface rotate in-plane couple parallel to Co Spectromicroscopy of Ferromagnets and Antiferromagnets AFM domain structure at surface of NiO substrate s [010] s 2mm NiO XMLD 0.10 8 TEY (a.u.) TEY (a.u.) 0.15 0.05 Co XMCD 4 0 0.00 868 870 872 874 Photon Energy(eV) 777 778 779 Photon Energy (eV) H. Ohldag, A. Scholl et al., Phys. Rev. Lett. 86(13), 2878 (2001). FM domain structure in thin Co film on NiO substrate Interface Spectroscopy 0.02 Ni A L2-edge M MxOy Upon Co deposition on NiO 2ML NiO Ni 2ML Co CoO Electron Yield (arb. units.) XAS Line shape is sensitive to transition Ni NiO Co/NiO Model B 0.01 0.00 868 0.10 870 872 874 Co CoO Co/NiO Model Co L3-edge 0.05 Linear combination of metal and oxide spectra possible 0.00 776 778 780 Photon Energy (eV) Experiment 3: Interface Microscopy Co Co NiO CoNiO NiO AFM NiO FM Ni(O) FM Co XMLD XMCD XMCD Chemically induced interfacial spins provide the magnetic link Only a small fraction of interfacial spins is pinned Method Experiment +0.5% Co/IrMn 15 Co 10 -0.5% Loops of interfacial Mn spins Bias polar XMCD Asymmetry (%) 5 0 -5 -10 -15 0.3 Mn 0.2 0.1 0.0 -0.1 -0.2 -0.3 -3k -2k -1k 0 1k 2k 3k Applied Field (Oe) azimuthal A small fraction (4%) of interfacial spins is pinned – they are the origin of exchange bias! Exchange Bias: A new x-ray look at an old problem 4 crucial experiments 1.) The bare antiferromagnetic surface. F. U. Hillebrecht, H. Ohldag et al., Phys Rev. Lett. 86(15), 3419 (2001). 2.) The antiferromagnet surface coupled to a ferromagnet. H. Ohldag, A. Scholl et al., Phys. Rev. Lett. 86(13), 2878 (2001). 3.) The interface between both. H. Ohldag, A. Scholl et al., Phys Rev Lett. 87(24), 7201 (2001). 4.) Where are the pinned interfacial spins? To be published. Exchange Bias Model from X-Rays ideal AFM poly AFM The Future - PEEM-III M D • • Aberration corrected PEEM Estimated 2 nm resolution O L L A L SM L D S L D X E t 50-70 ps • Picosecond dynamics 330 ns LL L I Part 2: Ultrafast Magnetization Dynamics Switching with Exchange Fields Oersted fields Exchange fields Oersted fields are long range and weak Exchange fields are short range and strong Magnetic Switching by Spin Injection • Nanoscale: <1000 Å M t t+ t t+ • Both “in-plane” (black) and “outof-plane” (red) M Proposed Solution • Use exchange field of injected spin current • Strong field, short pulse • Picosecond switching t0 • Optimal at small sizes t0 + t The Ultrafast Worlds Creating spin current •Spin moment relaxes into direction of bulk ferromagnet •Spin polarization reaches maximum at approximately 10 nm •Theoretically, spin polarization can be ~100% in some metals Spin injection FM NM current Polarization is spin coherence length or spin flip length ~ 1 nm for ferromagnets (or 10 fs) ~ 1 µm for noble metals ( or 10 ps) ~ 100 µm for semiconductors (or 1 ns) X-ray experiments can observe: effect, size, sign and dynamics Negative Damping by Spin Injection Minority spin injection Nano-structure for Spin Injection 2 nm 1 – 1000 nm 30 nm 40 nm NiFe Cu Si3N4 Co Focused Ion Beam Holes 1 µm Si3N4 • Current flows through Co to become spin polarized PEEM microscopy • Spin polarized current enters the NiFe layer changes the domain structure Oersted Switching by Current through Contact Holes Current is not polarized, switching due to Oerstedlike field 100 contact holes diameter 40 nm j 1012 A/m2 initial Fe 0.8 nm Cu 10 nm Si3N4 30 nm Cu 160 nm after 180 mA H 10 mm 10-100 mm Measuring Precession by Pump/Probe Technique • Pump: induce spin polarization by current pulse • Probe: Image with delayed photon pulse Current pulse 50 ps 330 ns • Vary the delay between current and photon pulse • Vary the strength of the current pulse Photon pulse X-Ray Free Electron Lasers Peak Brightness [Phot./(s · mrad2 · mm2 · 0.1%bandw.)] Part 3: The Future X-Ray FELs initial future ERLs 3rd Gen. SR SPPS 2nd Gen. SR Laser Slicing FWHM X-Ray Pulse Duration [ps] • SASE gives 106 intensity gain over spontaneous emission • FELs can produce ultrafast pulses (of order 100 fs) LINAC COHERENT LIGHT SOURCE 0 Km 2 Km 3 Km Concepts of the LCLS: Based on single pass free electron laser (FEL) Uses high energy linac (~15 GeV) to provide compressed electron beam to long undulator(s) (~120 m) Based on SASE physics to produce 800-8,000eV (up to 24KeV in 3rd harmonic) radiation Analogous in concept to XFEL of TESLA project at DESY Proposed Schedule and Budget • FY2003-2004 – Prepare preliminary designs • FY2005 – Procure undulator – Construct injector • FY2006-2007 – Convert linac,install undulator, begin FEL commissioning • FY2008 – Complete civil construction, characterize photon beam Estimated Total Project Cost : M$ 221 + M$ 47 = M$ 268 Example: Nanoscale Magnetism Reciprocal Space Imaging = Speckle Sample is non-periodic – no Bragg peaks Nanoscale Magnetism PEEM versus SPECKLE PEEM – X-ray Absorption Speckle – Coherent X-ray Scattering • Photon-in / electron-out • Photon-in / photon-out • Spatial resolution set by electron optics • Spatial resolution set by x-ray wavelength: (775 eV) 16 Å) • No strong external magnetic fields • Magnetic and electric fields • Equilibrium dynamics: > 1msec • Equilibrium dynamics: > 1msec (now) fsec (in future) • Pump-probe: ultrafast and single shot • Pump-probe: no single shot – space charge limit Technique of choice for dynamics, future X-FELs Incoherent vs. Coherent X-Ray Scattering Small Angle Scattering -40 40 -20 0 20 scattering vector q (mm-1) Coherence length larger than domains, but smaller than illuminated area information about domain statistics 20 0 -20 -40 40 log (intensity) -40 -20 0 20 40 scattering vector q (mm-1) Coherence length larger than illuminated area -40 40 -20 0 20 scattering vector q (mm-1) Speckle true information about domain structure 20 0 -20 -40 40 log (intensity) -40 -20 0 20 scattering vector q (mm-1) 40 Imaging by Coherent X-Ray Diffraction Phase problem can be solved by “oversampling” speckle image Transmission X-ray Microscope Reconstruction from Speckle Intensities 5 mm (different areas) S. Eisebitt, M. Lörgen, J. Lüning, J. Stöhr, W. Eberhardt, E. Fullerton (unpublished) Spin Block Fluctuations around Critical Temperature Magnetization Tc Temperature t = (T-Tc) / Tc T < Tc T Tc T > Tc Collaborators Stanford Synchrotron Radiation Laboratory Advanced Light Source Hendrik Ohldag Andreas Scholl Christian Stamm Frithjof Nolting (now SLS) Hans Christoph Siegmann Simone Anders (now IBM) Yves Acremann Stanford University: BESSY Scott Andrews Stefan Eisebitt Bruce Clemens Marcus Lörgen Wolfgang Eberhardt IBM Almaden Erik Fullerton Charles Rettner Jan-Ulrich Thiele Summary X-FELs will deliver: unprecedented brightness and femtosecond pulses Understanding of laser physics and technology well founded FELs promise to be extraordinary scientific tools Applications in many areas: chemistry, biology, plasma physics, atomic physics, condensed matter physics The End Spin Injection mm electron flow mm elec tron flow m m m Spin injection a nd da mping torque m m m Fast Magnetization Dynamics is governed by Landau-Lifschitz-Gilbert equation: 1 dM - M H dt Angular momentum change 1 M Precession torque 1 Tesla field: 90o rotation in 10 ps M dM dt Gilbert damping torque Typically << 1, 100 ps We want to understand on atomic level controls switching time, ~1 optimal Stoner Excitations: Changing the Magnetization by Electron Scattering X-FEL Radiation – Electric and Magnetic Fields Part 3: Ultrafast Magnetization Dynamics Switching with Oersted Fields ca. 200 BC 1995 Experimental Principle •High field pulses up to 5 T •High beam precision allows multiple shots on the same spot C. H. Back, R. Allenspach, W. Weber, S. S. P. Parkin, D. Weller, E. L. Garwin, H. C. Siegmann, Science 285, 864-867 (1999) Precession Torques Maximum torque Minimum Torque Simulation Results Small damping, many precessions Medium damping, fewer precessions Large damping, magnetization creeps into field Results show high sensitivity to damping ( optimal ~1) Experiment •State-of-art media (courtesy of Komag) Pattern read by PEEM e X-Rays • • • • Photoelectron emission microscopy image Same image drawn to simulation scale Comparison with simulation shows must be larger than 1 Larger than expected from all previous results But media have distribution of nanostructured regions Need to control nanostructure to understand why is so large Resonant Magnetic Soft X-ray Scattering e’ Fe e M I n exp( i q rn ) fn charge 2 magnetic -XMCD fn e' e Fn(0) i (e' e) Mn Fn(1) where Fn(i) are complex = f1 + i f2 Note: at resonance f1 = 0 Kortright and Kim, Phys. Rev. B 62, 12216 (2000) Resonant Magnetic Scattering Cross-Section e’ e M Charge magnetic Circular dichroism XMCD magnetic Linear Dichroism XMLD J. P. Hannon, G. T. Trammell, M. Blume, D. Gibbs, Phys. Rev. Lett 61, 1245 (1988) Motivation • Present methods of writing are unfavorable: – present recording time ~1 ns – unfavorable torque and dependent on thermal activation • Understand switching of soft and hard materials on sub-nanosecond scale – faster switching – avoid configuration with small torque on magnetization Magnetization and Spin Dynamics Magnetism ruled by four fundamental interactions: Exchange interaction => produces magnetic order on atomic scale, magnetic stiffness, TC , TN, spin-spin scattering, coherence time of spin excitations Spin-orbit interaction => produces magneto-crystalline anisotropy, spin-phonon (thermal) excitations, friction (Gilbert damping) Zeeman interaction => produces macroscopic spin alignment, torque (Landau-Lifshitz), magnetic switching Dipolar interaction => produces shape anisotropy, magnetic domain structure and motion Energy/atom time scale length scale Exchange eV fs Spin-orbit meV- meV ps - ns nano (nm) atomic Zeeman < meV ps - ns > nano Dipolar < meV ps - ms > nano Time scale of various processes, (leading to spin lattice relaxation) Stoner excitation: 10-15 sec (Femtosec) Emission of a spin wave : 10-12 sec (Picosec) Absoption of spin waves by the lattice: 10-10 sec