Peter Oppeneer. Department of Physics and Astronomy Uppsala University, S Uppsala, Sweden

Transcription

1 Light Matter Interactions (II Peter Oppeneer Department of Physics and Astronomy Uppsala University, S Uppsala, Sweden 1

2 Outline Lecture II Light-magnetism interaction Phenomenology of magnetic spectroscopies Electronic structure theory, linear-response theory Theory/understanding of magnetic spectroscopies Optical regime Ultraviolet and soft X-ray regime 2

3 In the beginning First observation of magneto-optics Magneto-optical Faraday effect (1845 Observation of interaction light-magnetism enormous impact on development of science! Faraday effect E q F (-M = -q F (M q F (M ~ lin. M Michael Faraday (

4 Magneto-optical Kerr effect Kerr (1876 Kerr (1878 Zeeman (1896 pol. analysis pol. analysis intensity measurement tan K b a r r r r Rotation of the polarization plane & ellipticity: Completely a magnetic effect! One of the best tools in magnetism research! 4

5 Magneto-optical Voigt effect Voigt effect (1899 ^ 45 o q V Very different from Faraday effect; Voigt effect is quadratic (even in M Kerr effect Imaging of magnetic domains using Voigt and Kerr effect in reflection Woldemar Voigt ( Voigt effect (Courtesy R. Schäfer 5

6 Magnetic circular and linear dichroism M E E MCD (I I /(I I Magnetic Circular Dichroism odd in M E ^ MLD (I I^/(I I^ Magnetic Linear Dichroism even in M 6

7 Development of light-magnetic material interaction Inv. Faraday effect Ultrafast magn.switching Spin Hall effect Vector magnetometry MO Kerr loops Electron MCD Non-linear magneto-optics XMLD XMCD, XRMS Domain imaging Magn. circ. dichroism Zeeman effect Kerr effect Voigt effect Faraday effect (1845 7

8 Recent Examples of Light Magnetism Interactions Observation of the spin Hall effect Kato, Myers, Gossard & Awschalom, Science 306, 1910 (2004 Spin Hall effect Kerr rotation image Dyakonov & Perel, JETP Lett. 13, 467 (1971 Reflection image Material: non-magnetic n-gaas [110] MO Kerr rotation detection ~ 10-5 deg. 8

9 Spin Hall effect in heavy metals Gives rise to spin-orbit torque Direct observation of SHE in pure heavy-metal difficult because of short spin lifetime and spin diffusion length Miron et al, Nature 476, 189 (2011 Liu et al, Science 336, 555 (2012 J c MOKE detection could be possible due to penetration depth 9

10 Experimental direct observation of spin Hall effect Pt j =10 7 A/cm 2 SH s (exp xz 1880 [Wcm] 1 SH s (th xz 1890 [Wcm] 1 Stamm, Murer, Berritta, Feng, Gabureac, Oppeneer & Gambardella, PRL 119, (2017 Excellent agreement with experiment Estimated l s =11.4±2 nm for pure Pt Accurate MOKE measurements of SH conductivity in heavy metals feasible with nrad sensitivity 10

11 Optically induced magnetization Due to nonlinear opto-magnetic effect, the inverse Faraday effect: Induces magnetization M Kimel et al, Nature 435, 655 (2005 Could potentially lead to a fast, optically driven magnetization reversal 11

12 All-optical writing of magnetic domains GdFeCo Stanciu et al, PRL 99, (2007 FePt Lambert et al, Science 345, 1337 (2014 Due to inverse Faraday effect? Background all-optical magn. recording Erasing & writing with fs-laser pulses Approx times faster recording? (symposium Th. Rasing, A. Kirilyuk 12

13 Ultrafast magnetism Measurement of ultrafast magnetic response with time-resolved magneto-optics x z y q K ( t, M, R( t s M ( t, n( t fs laser pulse t E e Ni p fs laser pulse (pump Beaurepaire, Merle, Danois, Bigot, PRL 76, 4250 (1996 Magnetization decay in <250 fs Hofherr et al, PRB 96, R (2017 Very fast decay ~40 fs 13

14 Theoretical description of light magnetism interaction 14 Use 2 nd level: Combination of Maxwell-Fresnel theory and ab initio quantum theory Fresnel equation for modes in material: Geometry & materials boundary conditions: i p i s pp ps sp ss r p r s E E r r r r E E t t i i i i ss i s t s t t i i t t i i ss i s r s n n n t E E n n n n r E E q q q q q q q cos cos cos 2 cos cos cos cos E p E s And: ab initio theory for calculation of (w q a b Polarization analysis or intensity measurement

15 15 Dependence of the dielectric tensor on fields 1 (,,, ( w E B k ( ( ( ( ( (,,, ( 0 j i j i j i E E O B B O B E O E O B O k O E B k w The dielectric tensor depends on external fields Use a Taylor expansion for effects to lowest order: All the (linear phenomena can be described, using the Fresnel formalism

16 Magnetic effects in Fresnel equations Typical tensor: Onsager relations Magnetic parity ^ yx 0 xy ^ xy ( H, w ( H, w xy ( H, w ( H, w odd even Magn. effects probe always ~M or ~M 2 (to lowest order Examples: MCD wd 2c Re n xy Odd in M; MLD wd 2cn Im ^ 2 xy ^ even 0 M 2 16

17 Magneto-optical Kerr and Faraday effects polar Kerr effect, normal incidence Faraday effect, normal incidence Use n 2 i xx xy Assume xy xx or s xy s xx q K s ( w xy xy i K q 1/ 2 F i F 1/ 2 s xx(1 4is xx / w c (1 4is xx / w 2d s ( w 17

18 Classification of magnetic spectroscopies Linear (odd in M spectroscopies: Polarization analysis Intensity Classification criteria: 1. Magnetic parity Transmission Faraday L MCD C 2. Transmission or reflection Reflection P-MOKE L-MOKE L T-MOKE RMS 2 quantities 1 quant. C L 3. Polarization or intensity 4. Linearly or circ. polarized light Suitable for (element-selective study of ferro-, ferrimagnets 18

19 Even-in-M magnetic spectroscopies Quadratic (even in M spectroscopies: Polarization analysis Intensity Transmission Voigt L MLD L Reflection birefringence R-Voigt L R-MLD L 2 quantities 1 quant. Suitable for (element-selective study of antiferromagnets (and ferromagnets as well 19

20 Linear-response theory Im[ xx ( w] e 2 w 2 m V n n' un. occ. Re{ x n' n x nn' } ( w w nn' Lifetime broadening, 1/t ~ 0.4 ev Re[ xy ( w] e 2 w 2 m V n n' un. occ. Im{ x n' n y nn' } ( w w nn' (for 1/t -> 0 Lifetime broadening happens and needs to be taken into account 20

21 Lifetime broadening linear magneto-optics calc. Optical frequencies: lifetime G1/t 0.03 Ry Oppeneer, Handbook of Magnetic Materials, Vol. 13 (

22 22 Origin of magneto-optical effects s s ˆ ˆ ˆ ( 1 ( ( 2 ˆ 0, 2 r B r V r V m H xc N e Effective Kohn-Sham Hamiltonian: Vary the two magnetic interactions (exchange & spin-orbit to deduce how magnetic spectra depend on these. Exchange field Spin-orbit coupling 2 }/ ( 1 ( { ( 0 s r m r n r n Spin-density (2x2: ( ( ( ( ( ( r n r n r m r n r n r n B

23 Effect of SOI and exchange interaction 2 e Full form of SOI: H SO 2 4m c (small relativistic effect Exchange splitting, 1 2 ev (3d atom 2 1 r dv dr L s ( L S Ni ex SOC Spin-orbit coupling breaks crystal symmetry Spin-orbit splitting ~20 mev 1 ev 23

24 Leading order quantity: spin-orbit coupling Leading quantity determining the valence band MO effect is spin-orbit coupling L.S Kerr and Faraday effect scale linear in the SOC, not in the exc.-splitting! Scaling of SOI Scaling of exc.int. Ni 24

25 What about the X-ray regime? X-ray magnetic circular dichroism E ( z, t ( e x ie y e iw/ c( n z iwt Co XMCD Understand origin of and perform ab initio calculations for XMCD & XMLD at L-edges 25

26 Note on importance of XMCD sum rules Thole et al, PRL 68, 1943 (1992 Carra et al, PRL 70, 694 (1993 Atomic spin moment Atomic orbital moment The XMCD sum rules are not exact but are intensively used, because they allow an element-selective determination of the spin & orbital moment on a 3d element in a material. (Lecture E. Goering 26

27 We had: Definition of refractive index (X-ray regime n 2 In x-ray regime: i xx xy n 1 ( i( 2 n xx (nonmagnetic: Small quantities! Co These quantities can be obtained from XAS, XMCD & Faraday effect measurements 27

28 Basic electronic structure picture 1 Spin-splitting of 3d states due to exchange interaction 2 Helicity-dependent optical selection rules left: right m : m 1, 1, m s m 0 s 0 Leads to different absorption of left/right circ. pol. radiation (trans. probabilities 28

29 Ab initio calculated XMCD spectra - Effect of xc Very different size of SOI and xc! 2p 3/2 exc. split 3x0.3 ev XMCD Im[ ] To lowest order the XMCD does not depend on ex : SO split 15 ev 2p 1/2 ex XAS XMCD Many calculations ignore, but there is a small effect! Kunes et al, PRB 64, (

30 Comparison with experimental XMCD spectra XMCD Fe 0.5 Ni 0.5 d=50 nm n 1 ( i( Exchange-split core states give somewhat better results when compared to experimental spectra! Kunes et al, PRB 64, (

31 Quadratic in M effects: X-ray Voigt effect or XLMD Voigt effect k^m 45 o q V q V i V wd 2ic n n^ wd 2cn Im 2 ^ xy 2q V LMD ^ Voigt effect E Origin not understood... Magnetovolume effect? Why much smaller than MCD? Spin-orbit interaction? Cf. Faraday, Kerr: linear in SO n n^ E^ MLD (I I^/(I I^ 31

32 Measurements & ab initio calculations XMLD Im n^ n Mertins et al, PRL 87, (2001 XMLD Re n^ n XMLD / X-ray Voigt effect would be very small without exchange split core states! 2p 3/2 exc. split 3x0.2 ev SO split 15 ev ex =0 2p 1/2 32

33 Further results of ab initio calculations XMLD small effect ~ 5% good ab initio theory A XMLD wd c Im n n^ wd 2cn Im ^ Why do we have these spectral structures? 33

34 Simple model for X-ray magnetic spectroscopies Core-states are (k dispersionless X-ray spectroscopies: XAS 2 ]/ 3 Im[ ^ Sum of all transitions m Expand xy considering 2p -> 3d transitions Selection rules on m : xx xx i i xy xy 0 m m m XMCD Im[ ] Difference of transitions with m=+1 and m=-1 XMLD Im[ ^ ] Im[ ( / 2] Difference of transitions with m=0 and aver. m=+1 & -1 34

35 Understanding the shape of XMLD and XMCD Develop model theory and perform ab initio calculations Model theory (2p core: Neglect SO in valence states (~ mev Consider only 2p ->3d transitions Expand functions with respect to ex Im[ (w] a ( jd w,s ms 2 m,s, /2 m s D ms m- and spin-dependent 3d partial DOS SO (15 ev >> ex (1-3 ev >> ex ( ev > so-v (0.09 ev XMLD Im[ ^ ] Im[ ( / 2] Difference of transitions with m=0 and aver. m=+1 & -1 35

36 Dms : Model theory, results m and s-dependent partial DOS of unoccupied 3d states XAS 4 2 m m ( D ( D m m D D m m j j 3/ 2 1/ 2 XAS branching ratio: 2/3 no magnetism (M invariant XMCD 2 2 m m ( D ( D m m D D m m j j 3/ 2 1/ 2 L 2, L 3 equal & opposite odd in M, no f.o. effect of m-orbital degeneracy: XMLD d D D j 1/ 2, 3/ 2 de d de XMCD Small signal, proportional! related to energy deriv. XMCD even in M ( is odd, D is odd Absence of the crystal field 36

37 Experimental check of XMCD-XMLD relation m-orbital degeneracy: No crystal field, amorphous FeCo alloy Spin-pol. unocc. 3d DOS D D Leading quantity ex is very small, yet crucial! relation between XMCD-XMLD verified Kunes et al, JMMM 272, 2146 (

38 Explanation of the XMCD shape E E E E F ( D D ( E f ( E ( D D ( E Leading quantity for XMCD: exchange splitting of 3d DOS, determines XMCD shape XMCD signal at one edge 38

39 Explanation of the XMLD shape E E E E F ( D D ( E d de ( D D ( E Leading quantity for XMLD: exchange splitting of 2p level XMLD is quadratic in M d f ( E ( D D ( E de XMLD signal at one edge 39

40 How good are the assumptions in the model? It is excellent for XAS, XMCD and XMLD! Useful for studying the origin of XMCD and XMLD Kunes & Oppeneer, PRB 67, (

41 Magnetocrystalline anisotropy in XMLD Effect of crystal field: Cubic symmetry Combination of different m-orbital spin-pol. DOS XMLD d de (t t (e e 2g 2g g g M(001 : 1 1 M(111 : 2 1 t 2g e g xy xz yz :,, r r r x y 3z :, 2 r r 1 different combination of m-partial DOS probed, depending on M axis large magnetocrystalline anisotropy appears in XMLD spectra 2 Effect of val.-soi Kunes & Oppeneer, PRB 67, (

42 How about the 3p (M semi-core edges? 3d 3p ->3d transitions T-MOKE in XUV Permalloy XMCD La-O-Vorakiat et al, PRL 103, (2009 Ultrafast element-selective demagnetization of Fe and Ni in permalloy 42

43 Transversal MOKE at M edges Measured as intensity change of lin. polarized light in reflection (cf. Fresnel theory Demagnization mechanism: Compute ab initio xy for several cases: 1 frozen magnon excitations, 2 reduced exchange splitting (spin-flips, 3 increased electron temperature T e - construct the change in A(t wrt A(t=0 > least square fit with experiment R( M R( M A( t Asymmetry R( M R( M Turgut et al, PRB 94, R (

44 Comparison of experiment and ab initio theory T-MOKE T-MOKE Exp. Theory For Co Surprisingly small contribution from spin-flips (exch. split reduction Larger effect (2/3 is due to fast magnon excitation => reduction of M z 44

45 Quadratic in M effect in-near-normal reflection q SH Schäfer-Hubert effect (or Voigt effect in reflection 45 o Fe k^m Near-normal incidence detection at Fe 3p edges 45

46 Comparison to ab initio calculations Importance of treating exchange splitting and spinorbit interaction on equal footing Importance of hybridization of j z states Ab initio theory 2p exchange splitting 2p 3/2-3/2-1/2 1/2 3/2 s Valencia et al, PRL 104, (2010 2p 1/2 +1/2-1/2 s 46

47 j z -hybridization 3p semi-core level of Fe Strong mixing of j, j z states, SO splitting & exchange splitting equally large Strong j z -mixing Magneto-X-ray effects as large as at 2p s! -1/2 +1/2 3/2 1/2-1/2-3/2 No expansion in small quantity possible! 47

48 Summarizing light magnetic matter interaction Magnetic spectroscopy is a highly sensitive tool that can detect minute magnetizations (spin Hall effect Exchange and spin-orbit splitting work together in different ways in valence and X-ray regime to bring about light - magnetic matter interaction Ab initio quantum theory (effective single particle theory works well but it is needed to know about its limitations Current frontlines: 1 Ultrasensitive measurements to observe very small spin-orbit related effects (e.g. Inverse spin galvanic effect 2 Ultrafast limit of modifications & control of magnetization, experiments and suitable theory 3 Nonlinear magneto-optic effects 48

49 Literature S. W. Lovesey and S. P. Collins, X-Ray Scattering and Absorption by Magnetic Materials (Clarendon Press, Oxford, A.K. Zvezdin and V.A. Kotov, Modern Magnetooptics and Magnetooptical Materials (London, Taylor & Francis, P.M. Oppeneer, Magneto-optical Kerr Spectra, in Handbook of Magnetic Materials, Vol. 13, edited by K.H.J. Buschow (Elsevier, Amsterdam, 2001, pp W. Kleemann, Magneto-optical materials, in Handbook of Magnetism and Advanced Magnetic Materials, Vol. 4, edited by H. Kronmüller and S.S.P. Parkin (Wiley, New York, J. Stöhr and H.-C. Siegmann, Magnetism: From Fundamentals to Nanoscale Dynamics (Springer, Berlin,

50 50

51 Appendix I: Alternative way to describe X-ray spectra Other way of describing effects scattering formalism: f ( e' e F F 0 i(( e' e M F1 ( e' M ( e M 2 Charge scattering (XAS 1st order magnetic scattering (XMCD 2nd order magnetic scattering (XMLD 51

52 Deficiency of DFT-LDA for localized 4f states LDA Nearly localized f state => LDA+U better 52

53 Practicals problem: 1 Material with magnetization in the scattering plane E 2 Lin. pol. light E-vector at 45 to the magnetization 3 Consider R(+M-R(-M M Use the reflection coefficients to show that R(+M-R(-M is a measure of the magnetization and derive an expression for the magn. asymmetry: R( M R( M A R( M R( M 53