Ab-initio výpocty magnetismu clusteru a nanostruktur:
Transkript
Ab-initio výpocty magnetismu clusteru a nanostruktur:
Ab-initio výpočty magnetismu clusterů a nanostruktur: Proč a jak? Ondřej Šipr Fyzikálnı́ ústav Akademie věd České republiky, Praha http://www.fzu.cz/~ sipr 16. března 2010 / Odborný seminář Nové technologie LS 2010 Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Where does magnetism come from? ◮ Classically: Magnetic field is something produced by moving electric charges that affects other moving charges ◮ Special relativity: Magnetism is a fictitious force needed to guarantee Lorentz invariance when charges move ◮ ◮ One observer may perceive a magnetic force where a moving observer perceives only an electrostatic force Dealing with magnetism in the framework of Dirac equation (instead of Schrödinger equation) is ideologically simple Where does magnetism come from? ◮ Classically: Magnetic field is something produced by moving electric charges that affects other moving charges ◮ Special relativity: Magnetism is a fictitious force needed to guarantee Lorentz invariance when charges move ◮ ◮ One observer may perceive a magnetic force where a moving observer perceives only an electrostatic force Dealing with magnetism in the framework of Dirac equation (instead of Schrödinger equation) is ideologically simple Where does magnetism come from? ◮ Classically: Magnetic field is something produced by moving electric charges that affects other moving charges ◮ Special relativity: Magnetism is a fictitious force needed to guarantee Lorentz invariance when charges move ◮ ◮ One observer may perceive a magnetic force where a moving observer perceives only an electrostatic force Dealing with magnetism in the framework of Dirac equation (instead of Schrödinger equation) is ideologically simple Where does magnetism come from? ◮ Classically: Magnetic field is something produced by moving electric charges that affects other moving charges ◮ Special relativity: Magnetism is a fictitious force needed to guarantee Lorentz invariance when charges move ◮ ◮ One observer may perceive a magnetic force where a moving observer perceives only an electrostatic force Dealing with magnetism in the framework of Dirac equation (instead of Schrödinger equation) is ideologically simple Magnetický moment ◮ Magnetický moment µ je mı́rou snahy zmagnetizovaného objektu orientovat se v souhlau s vnějšı́m magnetickým polem ◮ Moment sı́ly τ působı́cı́ na magnetický moment v magnetickém poli: τ = µ×H ◮ Potenciálnı́ energie magnetického momentu v magnetickém poli: U = −µ · H ◮ Přı́klad: Pro uzavřenou smyčku µ = µ0 I S Magnetický moment ◮ Magnetický moment µ je mı́rou snahy zmagnetizovaného objektu orientovat se v souhlau s vnějšı́m magnetickým polem ◮ Moment sı́ly τ působı́cı́ na magnetický moment v magnetickém poli: τ = µ×H ◮ Potenciálnı́ energie magnetického momentu v magnetickém poli: U = −µ · H ◮ Přı́klad: Pro uzavřenou smyčku µ = µ0 I S Magnetický moment ◮ Magnetický moment µ je mı́rou snahy zmagnetizovaného objektu orientovat se v souhlau s vnějšı́m magnetickým polem ◮ Moment sı́ly τ působı́cı́ na magnetický moment v magnetickém poli: τ = µ×H ◮ Potenciálnı́ energie magnetického momentu v magnetickém poli: U = −µ · H ◮ Přı́klad: Pro uzavřenou smyčku µ = µ0 I S Magnetický moment ◮ Magnetický moment µ je mı́rou snahy zmagnetizovaného objektu orientovat se v souhlau s vnějšı́m magnetickým polem ◮ Moment sı́ly τ působı́cı́ na magnetický moment v magnetickém poli: τ = µ×H ◮ Potenciálnı́ energie magnetického momentu v magnetickém poli: U = −µ · H ◮ Přı́klad: Pro uzavřenou smyčku µ = µ0 I S Two ways of moving an electron (A quick and dirty introduction to magnetism) ◮ Orbiting ◮ Spinning Orbital magnetic moment Classical expression for magnetic moment: µorb = µ0 I S =⇒ µorb = − µB S where µB is Bohr magneton µB ≡ eµ0 ~ 2me and L is angular momentum devided by ~. For electron orbiting around an atom, the z-component of orbital magnetic moment is thus (z) µorb = − mℓ µB , where mℓ is the magnetic quantum number. Orbital magnetic moment Classical expression for magnetic moment: µorb = µ0 I S =⇒ µorb = − µB S where µB is Bohr magneton µB ≡ eµ0 ~ 2me and L is angular momentum devided by ~. For electron orbiting around an atom, the z-component of orbital magnetic moment is thus (z) µorb = − mℓ µB , where mℓ is the magnetic quantum number. Spin magnetic moment Electron spin: Picture of a rotating charged sphere fails. . . µorb = − µB L vers. µspin = − 2 µB S L is angular momentum connected with orbital motion S is angular momentum connected with “spinning” For electron around an atom, the z-component of spin-related angular momentum is S (z) = ± 1 ~ , 2 hence we get for a z-component of spin-related magnetic moment (z) µspin = ± µB . Spin magnetic moment Electron spin: Picture of a rotating charged sphere fails. . . µorb = − µB L vers. µspin = − 2 µB S L is angular momentum connected with orbital motion S is angular momentum connected with “spinning” For electron around an atom, the z-component of spin-related angular momentum is S (z) = ± 1 ~ , 2 hence we get for a z-component of spin-related magnetic moment (z) µspin = ± µB . Magnetická anizotropie ◮ Magnetické vlastnosti materiálů závisı́ na směru (orientaci) Energie systému závisı́ na směru spontánnı́ magnetizace ◮ Různé zdroje (druhy) magnetické anizotropie Magnetická anizotropie ◮ Magnetické vlastnosti materiálů závisı́ na směru (orientaci) Energie systému závisı́ na směru spontánnı́ magnetizace ◮ Různé zdroje (druhy) magnetické anizotropie Tvarová magnetická anizotropie Energie interakce mezi dipóly: Edip−dip 1 X 1 = − 2πµ0 rij3 i 6=j " (rij · µi )(rij · µj ) µi · µj − 3 rij2 Protáhlé nebo planárnı́ objekty se snažı́ orientovat magnetizaci rovnoběžně s rovinou. # Magnetokrystalická anizotropie Orientace magnetizace ve směru preferovaném geometrickou strukturou krystalu ◮ U normálnı́ch (3D) látek obvykle malá ◮ U povrchů, vrstev, clusterů může být vysoká Magnetokrystalická anizotropie Orientace magnetizace ve směru preferovaném geometrickou strukturou krystalu ◮ U normálnı́ch (3D) látek obvykle malá ◮ U povrchů, vrstev, clusterů může být vysoká Magnetokrystalická anizotropie Orientace magnetizace ve směru preferovaném geometrickou strukturou krystalu ◮ U normálnı́ch (3D) látek obvykle malá ◮ U povrchů, vrstev, clusterů může být vysoká Konkurenčnı́ boj anizotropiı́ Tvarová anizotropie chce magnetizaci orientovat “in-plane”, magnetokrystalická anizotropie působı́ (zde) “out-of-plane”, výsledek závisı́ na tom, kdo je silnějšı́. Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Magnetism of Fe atom Magnetic properties of atoms are governed by Hund rules Electron configuration: 3d 6 4s 2 ◮ Spin magnetic moment: µspin = 4 µB ◮ ◮ First P Hund rule: Total atomic spin quantum number S = ms is maximum (as long as it is compatible with Pauli exclusion principle) Orbital magnetic moment: µorb = 2 µB ◮ Second P Hund rule: Total atomic orbital quantum number L = mℓ is maximum (as long as it is compatible with Pauli exclusion principle and first Hund rule) Magnetism of Fe atom Magnetic properties of atoms are governed by Hund rules Electron configuration: 3d 6 4s 2 ◮ Spin magnetic moment: µspin = 4 µB ◮ ◮ First P Hund rule: Total atomic spin quantum number S = ms is maximum (as long as it is compatible with Pauli exclusion principle) Orbital magnetic moment: µorb = 2 µB ◮ Second P Hund rule: Total atomic orbital quantum number L = mℓ is maximum (as long as it is compatible with Pauli exclusion principle and first Hund rule) Magnetism of Fe atom Magnetic properties of atoms are governed by Hund rules Electron configuration: 3d 6 4s 2 ◮ Spin magnetic moment: µspin = 4 µB ◮ ◮ First P Hund rule: Total atomic spin quantum number S = ms is maximum (as long as it is compatible with Pauli exclusion principle) Orbital magnetic moment: µorb = 2 µB ◮ Second P Hund rule: Total atomic orbital quantum number L = mℓ is maximum (as long as it is compatible with Pauli exclusion principle and first Hund rule) Magnetism of bulk Fe crystal ◮ Generally: Magnetism is suppressed in the bulk (with respect to atomic case) ◮ Intuitively: Electron are not free to orbit around atoms ◮ Spin magnetic moment is µspin ≈ 2.2 µB per atom ◮ Orbital magnetic moment is quenched (outright zero in non-relativistic case) Magnetism of bulk Fe crystal ◮ Generally: Magnetism is suppressed in the bulk (with respect to atomic case) ◮ Intuitively: Electron are not free to orbit around atoms ◮ Spin magnetic moment is µspin ≈ 2.2 µB per atom ◮ Orbital magnetic moment is quenched (outright zero in non-relativistic case) Magnetism of bulk Fe crystal ◮ Generally: Magnetism is suppressed in the bulk (with respect to atomic case) ◮ Intuitively: Electron are not free to orbit around atoms ◮ Spin magnetic moment is µspin ≈ 2.2 µB per atom ◮ Orbital magnetic moment is quenched (outright zero in non-relativistic case) Magnetism of bulk Fe crystal ◮ Generally: Magnetism is suppressed in the bulk (with respect to atomic case) ◮ Intuitively: Electron are not free to orbit around atoms ◮ Spin magnetic moment is µspin ≈ 2.2 µB per atom ◮ Orbital magnetic moment is quenched (outright zero in non-relativistic case) Surfaces are magnetism-friendly ◮ Atoms at surfaces exhibit some atomic-like characteristics ◮ Spin magnetic moment is larger than in bulk; for Fe it is µspin ≈ 2.5–3.0 µB per atom ◮ The orbital magnetic moment is increased by an even larger percentage, µorb ≈ 0.07–0.12 µB per atom Surfaces are magnetism-friendly ◮ Atoms at surfaces exhibit some atomic-like characteristics ◮ Spin magnetic moment is larger than in bulk; for Fe it is µspin ≈ 2.5–3.0 µB per atom ◮ The orbital magnetic moment is increased by an even larger percentage, µorb ≈ 0.07–0.12 µB per atom Surfaces are magnetism-friendly ◮ Atoms at surfaces exhibit some atomic-like characteristics ◮ Spin magnetic moment is larger than in bulk; for Fe it is µspin ≈ 2.5–3.0 µB per atom ◮ The orbital magnetic moment is increased by an even larger percentage, µorb ≈ 0.07–0.12 µB per atom Magnetism of iron: summary atom surface bulk µspin = 4 µB µspin = 2.5–3.0 µB µspin = 2.2 µB µorb = 2 µB µorb = 0.07–0.12 µB µorb = 0.05 µB A co anizotropie malých systémů ? ◮ Magnetická anizotropie vrstev je řádově většı́ než anizotropie krystalů ◮ Magnetická anizotropie clusterů je (až) řádově většı́ než anizotropie vrstev. ◮ ⇒ Kdož máš zájem na vysoké anizotropii, vrhni se na nanostruktury A co anizotropie malých systémů ? ◮ Magnetická anizotropie vrstev je řádově většı́ než anizotropie krystalů ◮ Magnetická anizotropie clusterů je (až) řádově většı́ než anizotropie vrstev. ◮ ⇒ Kdož máš zájem na vysoké anizotropii, vrhni se na nanostruktury A co anizotropie malých systémů ? ◮ Magnetická anizotropie vrstev je řádově většı́ než anizotropie krystalů ◮ Magnetická anizotropie clusterů je (až) řádově většı́ než anizotropie vrstev. ◮ ⇒ Kdož máš zájem na vysoké anizotropii, vrhni se na nanostruktury K čemu je anizotropie dobrá ? Obracet směr spinu je energeticky řádově snažšı́ než přidávat či odebı́rat elektrony −→ spintronika ! Why all the fuss with µorb ? ◮ µorb is small but important ! ◮ It is a manifestation of spin-orbit coupling, which is the mechanism behind the magnetocrystalline anisotropy ◮ Under certain assumptions, magnetocrystalline anisotropy energy (MAE) can be estimated as k ∆EMAE = const × µorb − µ⊥ orb k where µorb and µ⊥ orb are orbital magnetic moments for two perpendicular directions of the magnetization M Why all the fuss with µorb ? ◮ µorb is small but important ! ◮ It is a manifestation of spin-orbit coupling, which is the mechanism behind the magnetocrystalline anisotropy ◮ Under certain assumptions, magnetocrystalline anisotropy energy (MAE) can be estimated as k ∆EMAE = const × µorb − µ⊥ orb k where µorb and µ⊥ orb are orbital magnetic moments for two perpendicular directions of the magnetization M Why all the fuss with µorb ? ◮ µorb is small but important ! ◮ It is a manifestation of spin-orbit coupling, which is the mechanism behind the magnetocrystalline anisotropy ◮ Under certain assumptions, magnetocrystalline anisotropy energy (MAE) can be estimated as k ∆EMAE = const × µorb − µ⊥ orb k where µorb and µ⊥ orb are orbital magnetic moments for two perpendicular directions of the magnetization M Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Clusters: Who they are? ◮ Free clusters — giant molecules, surrounded by vacuum ◮ Supported clusters — adsorbed on a surface Clusters: Who they are? ◮ Free clusters — giant molecules, surrounded by vacuum ◮ Supported clusters — adsorbed on a surface Comparing free and supported clusters ◮ How do the magnetic properties change if clusters are deposited on a substrate ? ◮ Take analogous systems (identical sizes, identical geometries) and have a look ◮ Focus rather on the trends than on particular values Comparing free and supported clusters ◮ How do the magnetic properties change if clusters are deposited on a substrate ? ◮ Take analogous systems (identical sizes, identical geometries) and have a look ◮ Focus rather on the trends than on particular values Comparing free and supported clusters ◮ How do the magnetic properties change if clusters are deposited on a substrate ? ◮ Take analogous systems (identical sizes, identical geometries) and have a look ◮ Focus rather on the trends than on particular values Calculational procedure for supported clusters Impurity Green function method ◮ Calculate electronic structure of the “host” system (clean surface) ◮ Supported clusters are treated as a perturbation to the clean surface and the ◮ ◮ Green’s function of the new system (cluster plus substrate) is obtained by solving the Dyson equation. Focus on planar FeN on Ni(001) and CoN clusters on Au(111) Calculational procedure for supported clusters Impurity Green function method ◮ Calculate electronic structure of the “host” system (clean surface) ◮ Supported clusters are treated as a perturbation to the clean surface and the ◮ ◮ Green’s function of the new system (cluster plus substrate) is obtained by solving the Dyson equation. Focus on planar FeN on Ni(001) and CoN clusters on Au(111) Calculational procedure for supported clusters Impurity Green function method ◮ Calculate electronic structure of the “host” system (clean surface) ◮ Supported clusters are treated as a perturbation to the clean surface and the ◮ ◮ Green’s function of the new system (cluster plus substrate) is obtained by solving the Dyson equation. Focus on planar FeN on Ni(001) and CoN clusters on Au(111) Calculational procedure for supported clusters Impurity Green function method ◮ Calculate electronic structure of the “host” system (clean surface) ◮ Supported clusters are treated as a perturbation to the clean surface and the ◮ ◮ Green’s function of the new system (cluster plus substrate) is obtained by solving the Dyson equation. Focus on planar FeN on Ni(001) and CoN clusters on Au(111) Shapes of clusters (free or supported) FeN / Ni(001) CoN / Au(111) Only nearest-neighbor substrate atoms are shown. Average magnetic moments 2.3 Co 3.1 supported ◮ Supported clusters: nearly monotonous decay of µspin and µorb with N ◮ Free clusters: quasi-oscillations, quite large amplitides 2.2 in Co spheres 3.0 2.9 2.8 2.1 2.0 1.9 spin spin in Fe spheres spin 1.8 Fe 2.7 spin 1.7 supported 1.6 2.6 2 4 6 8 10 2 Number of atoms in cluster 4 6 8 Number of atoms in cluster 0.9 0.5 0.8 supported supported 0.7 0.2 in Co spheres in Fe spheres 0.3 orb 0.4 Fe orb orb Co 0.6 orb 0.5 0.4 0.3 0.1 0.2 0.0 2 4 6 8 Number of atoms in cluster 10 2 4 6 Number of atoms in cluster 8 Average magnetic moments 2.3 Co free supported 3.1 ◮ Supported clusters: nearly monotonous decay of µspin and µorb with N ◮ Free clusters: quasi-oscillations, quite large amplitides 2.2 in Co spheres 3.0 2.9 2.8 2.1 2.0 1.9 spin spin in Fe spheres spin 1.8 Fe 2.7 spin free supported 1.7 1.6 2.6 2 4 6 8 10 2 Number of atoms in cluster 4 6 8 Number of atoms in cluster 0.9 0.5 free supported 0.7 0.2 in Co spheres in Fe spheres orb 0.4 0.3 free supported 0.8 Fe orb orb Co 0.6 orb 0.5 0.4 0.3 0.1 0.2 0.0 2 4 6 8 Number of atoms in cluster 10 2 4 6 Number of atoms in cluster 8 3.0 2.8 2.6 2.4 2.2 B] 2.2 Co supported 2.0 2 3 4 5 6 7 1.8 spin 2 3 4 5 6 7 8 9 in Co spheres [ Fe supported 3.2 spin in Fe spheres [ B] Effect of coordination on µspin 1.6 2.0 0 2 4 Number of neighboring Fe atoms 0 2 4 Number of neighboring Co atoms ◮ µspin decreases if coordination number increases ◮ Estimates of µspin possible for large clusters ◮ For free clusters, big scatter of data 6 3.0 2.8 2.6 2.4 2.2 B] 2.2 Co supported 2.0 2 3 4 5 6 7 1.8 spin 2 3 4 5 6 7 8 9 in Co spheres [ Fe supported 3.2 spin in Fe spheres [ B] Effect of coordination on µspin 1.6 2.0 0 2 4 Number of neighboring Fe atoms 0 2 4 Number of neighboring Co atoms ◮ µspin decreases if coordination number increases ◮ Estimates of µspin possible for large clusters ◮ For free clusters, big scatter of data 6 3.0 2.8 2.6 2.4 2.2 B] 2.2 Co supported 2.0 2 3 4 5 6 7 1.8 spin 2 3 4 5 6 7 8 9 in Co spheres [ Fe supported 3.2 spin in Fe spheres [ B] Effect of coordination on µspin 1.6 2.0 0 2 4 Number of neighboring Fe atoms 0 2 4 Number of neighboring Co atoms ◮ µspin decreases if coordination number increases ◮ Estimates of µspin possible for large clusters ◮ For free clusters, big scatter of data 6 3.0 2.8 2.6 2.4 2.2 B] 2.2 Co supported 2.0 2 3 4 5 6 7 1.8 spin 2 3 4 5 6 7 8 9 in Co spheres [ Fe supported 3.2 spin in Fe spheres [ B] Effect of coordination on µspin 1.6 2.0 0 2 4 0 3.0 2.8 2.6 2.4 2 4 6 B] Number of neighboring Co atoms 2.2 Co free 2.0 1.8 spin in Co spheres [ Fe free 3.2 spin in Fe spheres [ B] Number of neighboring Fe atoms 2.2 1.6 2.0 0 2 4 Number of neighboring Fe atoms 0 2 4 Number of neighboring Co atoms ◮ µspin decreases if coordination number increases ◮ Estimates of µspin possible for large clusters ◮ For free clusters, big scatter of data 6 Comparison between free and supported clusters: summary ◮ Substrate acts as an adult supervisor for the free clusters ◮ ◮ It suppresses the tendency of magnetic moments to oscillate with cluster size It makes µspin to depend linearly on the coordination number ◮ For free clusters, this trend appears as well but only for spherical and/or larger clusters Comparison between free and supported clusters: summary ◮ Substrate acts as an adult supervisor for the free clusters ◮ ◮ It suppresses the tendency of magnetic moments to oscillate with cluster size It makes µspin to depend linearly on the coordination number ◮ For free clusters, this trend appears as well but only for spherical and/or larger clusters Comparison between free and supported clusters: summary ◮ Substrate acts as an adult supervisor for the free clusters ◮ ◮ It suppresses the tendency of magnetic moments to oscillate with cluster size It makes µspin to depend linearly on the coordination number ◮ For free clusters, this trend appears as well but only for spherical and/or larger clusters Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Clusters and magnetism: Problems . . . ◮ Magnetic properties of large assemblies of clusters are macroscopic: no principal problems with measuring them ◮ To understand nanomagnetism, one needs to study individual small systems ◮ Magnetization of individual clusters (of just few atoms) cannot be measured by macroscopic methods ◮ Recent progress in studying magnetism of clusters is, to a large extent, associated with spectroscopy Clusters and magnetism: Problems . . . ◮ Magnetic properties of large assemblies of clusters are macroscopic: no principal problems with measuring them ◮ To understand nanomagnetism, one needs to study individual small systems ◮ Magnetization of individual clusters (of just few atoms) cannot be measured by macroscopic methods ◮ Recent progress in studying magnetism of clusters is, to a large extent, associated with spectroscopy Clusters and magnetism: Problems . . . ◮ Magnetic properties of large assemblies of clusters are macroscopic: no principal problems with measuring them ◮ To understand nanomagnetism, one needs to study individual small systems ◮ Magnetization of individual clusters (of just few atoms) cannot be measured by macroscopic methods ◮ Recent progress in studying magnetism of clusters is, to a large extent, associated with spectroscopy Clusters and magnetism: Problems . . . ◮ Magnetic properties of large assemblies of clusters are macroscopic: no principal problems with measuring them ◮ To understand nanomagnetism, one needs to study individual small systems ◮ Magnetization of individual clusters (of just few atoms) cannot be measured by macroscopic methods ◮ Recent progress in studying magnetism of clusters is, to a large extent, associated with spectroscopy X-ray absorption spectroscopy howto ◮ X-rays go in, x-rays go out, absorption coefficient is measured as a function the energy of the incoming x-rays x-rays in x-rays out sample ◮ Most of the absorption goes on account of the photoelectric effect on core electrons X-ray absorption spectroscopy howto ◮ X-rays go in, x-rays go out, absorption coefficient is measured as a function the energy of the incoming x-rays x-rays in x-rays out sample ◮ Most of the absorption goes on account of the photoelectric effect on core electrons Absorption via core electron excitation Absorption via core electron excitation ◮ By tuning the energy of incoming x-rays, electrons from one core level only participate ◮ ◮ Chemical selectivity Dipole selection rule ◮ Angular-momentum selectivity (whether they are s, p, d states. . . ) Absorption via core electron excitation ◮ By tuning the energy of incoming x-rays, electrons from one core level only participate ◮ ◮ Chemical selectivity Dipole selection rule ◮ Angular-momentum selectivity (whether they are s, p, d states. . . ) Absorption via core electron excitation ◮ By tuning the energy of incoming x-rays, electrons from one core level only participate ◮ ◮ Chemical selectivity Dipole selection rule ◮ Angular-momentum selectivity (whether they are s, p, d states. . . ) Absorption via core electron excitation ◮ By tuning the energy of incoming x-rays, electrons from one core level only participate ◮ ◮ Chemical selectivity Dipole selection rule ◮ Angular-momentum selectivity (whether they are s, p, d states. . . ) Absorption edges (Absorpčnı́ hrany) hrana excitovaný elektron K L1 L2 L3 1s 2s 2p1/2 2p3/2 ◮ Zvyšuje-li se energii fotonů, absorpčnı́ koeficient klesá ◮ Vzroste-li energie tak, že může dojı́t k excitovánı́ dalšı́ho vnitřnı́ho elektronu, absorpčnı́ koeficient se skokem zvýšı́ Absorption edges (Absorpčnı́ hrany) hrana excitovaný elektron K L1 L2 L3 1s 2s 2p1/2 2p3/2 ◮ Zvyšuje-li se energii fotonů, absorpčnı́ koeficient klesá ◮ Vzroste-li energie tak, že může dojı́t k excitovánı́ dalšı́ho vnitřnı́ho elektronu, absorpčnı́ koeficient se skokem zvýšı́ Absorption edges (Absorpčnı́ hrany) hrana excitovaný elektron K L1 L2 L3 1s 2s 2p1/2 2p3/2 ◮ Zvyšuje-li se energii fotonů, absorpčnı́ koeficient klesá ◮ Vzroste-li energie tak, že může dojı́t k excitovánı́ dalšı́ho vnitřnı́ho elektronu, absorpčnı́ koeficient se skokem zvýšı́ Jemná struktura absorpčnı́ hrany: EXAFS a ti druzı́ ◮ Nı́zké energie fotoelektronu (do ∼50 eV) ◮ ◮ ◮ XANES (X-ray Absorption Near Edge Structure) NEXAFS (Near Edge X-ray Absorption Fine Structure) Vysoké energie fotoelektronu (∼100–1000 eV) ◮ EXAFS (Extended X-ray Absorption Fine Structure) ◮ Určovánı́ struktury, zejména nekrystalických látek Jemná struktura absorpčnı́ hrany: EXAFS a ti druzı́ ◮ Nı́zké energie fotoelektronu (do ∼50 eV) ◮ ◮ ◮ XANES (X-ray Absorption Near Edge Structure) NEXAFS (Near Edge X-ray Absorption Fine Structure) Vysoké energie fotoelektronu (∼100–1000 eV) ◮ EXAFS (Extended X-ray Absorption Fine Structure) ◮ Určovánı́ struktury, zejména nekrystalických látek Jemná struktura absorpčnı́ hrany: EXAFS a ti druzı́ ◮ Nı́zké energie fotoelektronu (do ∼50 eV) ◮ ◮ ◮ XANES (X-ray Absorption Near Edge Structure) NEXAFS (Near Edge X-ray Absorption Fine Structure) Vysoké energie fotoelektronu (∼100–1000 eV) ◮ EXAFS (Extended X-ray Absorption Fine Structure) ◮ Určovánı́ struktury, zejména nekrystalických látek Jemná struktura absorpčnı́ hrany: EXAFS a ti druzı́ ◮ Nı́zké energie fotoelektronu (do ∼50 eV) ◮ ◮ ◮ XANES (X-ray Absorption Near Edge Structure) NEXAFS (Near Edge X-ray Absorption Fine Structure) Vysoké energie fotoelektronu (∼100–1000 eV) ◮ EXAFS (Extended X-ray Absorption Fine Structure) ◮ Určovánı́ struktury, zejména nekrystalických látek Jemná struktura absorpčnı́ hrany: EXAFS a ti druzı́ ◮ Nı́zké energie fotoelektronu (do ∼50 eV) ◮ ◮ ◮ XANES (X-ray Absorption Near Edge Structure) NEXAFS (Near Edge X-ray Absorption Fine Structure) Vysoké energie fotoelektronu (∼100–1000 eV) ◮ EXAFS (Extended X-ray Absorption Fine Structure) ◮ Určovánı́ struktury, zejména nekrystalických látek XMCD howto XMCD (X-ray Magnetic Circular Dichroism): sample is magnetized, measure the difference between absorption of left and right circularly polarized x-rays photon helicity x-rays parallel magnetization antiparallel µXMCD = µ(+) − µ(−) XMCD howto XMCD (X-ray Magnetic Circular Dichroism): sample is magnetized, measure the difference between absorption of left and right circularly polarized x-rays photon helicity x-rays parallel magnetization antiparallel µXMCD = µ(+) − µ(−) L2,3 edge of magnetic systems L3 EF L2,3-edge XANES 6 L3 isotropic absorption 4 2p3/2 2p1/2 XMCD sum rules: By adding, subtracting and dividing the peak areas, chemically-specific µspin , µorb and µorb /µspin can be obtained L2 2 0 2 L2,3-edge XMCD L2 (transition metals) 0 difference (+) (-) - -2 -4 -10 0 10 20 30 energy [eV] Z Z (d) (d) (∆µL3 − 2∆µL2 ) dE ∼ (∆µL3 + ∆µL2 ) dE ∼ µspin + 7Tz (d) 3nh (d) µorb (d) 2nh L2,3 edge of magnetic systems L3 EF L2,3-edge XANES 6 L3 isotropic absorption 4 2p3/2 2p1/2 XMCD sum rules: By adding, subtracting and dividing the peak areas, chemically-specific µspin , µorb and µorb /µspin can be obtained L2 2 0 2 L2,3-edge XMCD L2 (transition metals) 0 difference (+) (-) - -2 -4 -10 0 10 20 30 energy [eV] Z Z (d) (d) (∆µL3 − 2∆µL2 ) dE ∼ (∆µL3 + ∆µL2 ) dE ∼ µspin + 7Tz (d) 3nh (d) µorb (d) 2nh L2,3 edge of magnetic systems L3 EF L2,3-edge XANES 6 L3 isotropic absorption 4 2p3/2 2p1/2 XMCD sum rules: By adding, subtracting and dividing the peak areas, chemically-specific µspin , µorb and µorb /µspin can be obtained L2 2 0 2 L2,3-edge XMCD L2 (transition metals) 0 difference (+) (-) - -2 -4 -10 0 10 20 30 energy [eV] Z Z (d) (d) (∆µL3 − 2∆µL2 ) dE ∼ (∆µL3 + ∆µL2 ) dE ∼ µspin + 7Tz (d) 3nh (d) µorb (d) 2nh What can XMCD do for us? ◮ Through the sum rules, XMCD can inform about µspin and µorb separately ◮ For compounds, magnetic state of each element can be explored separately ◮ Employing sum rules on experimental data may require substantial theoretical input What can XMCD do for us? ◮ Through the sum rules, XMCD can inform about µspin and µorb separately ◮ For compounds, magnetic state of each element can be explored separately ◮ Employing sum rules on experimental data may require substantial theoretical input What can XMCD do for us? ◮ Through the sum rules, XMCD can inform about µspin and µorb separately ◮ For compounds, magnetic state of each element can be explored separately ◮ Employing sum rules on experimental data may require substantial theoretical input Problems with sum rules Approximations involved in its derivation: ◮ Transition to a well-defined single band ◮ Core hole neglected ◮ ... ◮ Could be at least partly corrected for by renormalization, scaling and/or varying the L2,3 edge integration ranges ◮ Justified when studying series of related systems: One can assume that the corrections will be similar for each member of the series Problems with sum rules Approximations involved in its derivation: ◮ Transition to a well-defined single band ◮ Core hole neglected ◮ ... ◮ Could be at least partly corrected for by renormalization, scaling and/or varying the L2,3 edge integration ranges ◮ Justified when studying series of related systems: One can assume that the corrections will be similar for each member of the series Problems with sum rules Approximations involved in its derivation: ◮ Transition to a well-defined single band ◮ Core hole neglected ◮ ... ◮ Could be at least partly corrected for by renormalization, scaling and/or varying the L2,3 edge integration ranges ◮ Justified when studying series of related systems: One can assume that the corrections will be similar for each member of the series Problems with sum rules Approximations involved in its derivation: ◮ Transition to a well-defined single band ◮ Core hole neglected ◮ ... ◮ Could be at least partly corrected for by renormalization, scaling and/or varying the L2,3 edge integration ranges ◮ Justified when studying series of related systems: One can assume that the corrections will be similar for each member of the series Problems with sum rules Approximations involved in its derivation: ◮ Transition to a well-defined single band ◮ Core hole neglected ◮ ... ◮ Could be at least partly corrected for by renormalization, scaling and/or varying the L2,3 edge integration ranges ◮ Justified when studying series of related systems: One can assume that the corrections will be similar for each member of the series Problems with sum rules Approximations involved in its derivation: ◮ Transition to a well-defined single band ◮ Core hole neglected ◮ ... ◮ Could be at least partly corrected for by renormalization, scaling and/or varying the L2,3 edge integration ranges ◮ Justified when studying series of related systems: One can assume that the corrections will be similar for each member of the series Tz term: A feature, not a bug ! 3 IA (∆µL3 − 2∆µL2 ) dE = µspin + 7Tz nh integrated isotropic absorption spectrum d component of the local spin magnetic moment number of holes in the d band d component of magnetic dipole operator IA µspin nh Tz Tz = Z D T̂z E = 1 2 [σ − 3r̂(r̂ · σ)]z Tz is a measure of the asphericity of spin magnetization µspin can be obtained only in combination with 7Tz ! Tz term: A feature, not a bug ! 3 IA (∆µL3 − 2∆µL2 ) dE = µspin + 7Tz nh integrated isotropic absorption spectrum d component of the local spin magnetic moment number of holes in the d band d component of magnetic dipole operator IA µspin nh Tz Tz = Z D T̂z E = 1 2 [σ − 3r̂(r̂ · σ)]z Tz is a measure of the asphericity of spin magnetization µspin can be obtained only in combination with 7Tz ! Tz term: A feature, not a bug ! 3 IA (∆µL3 − 2∆µL2 ) dE = µspin + 7Tz nh integrated isotropic absorption spectrum d component of the local spin magnetic moment number of holes in the d band d component of magnetic dipole operator IA µspin nh Tz Tz = Z D T̂z E = 1 2 [σ − 3r̂(r̂ · σ)]z Tz is a measure of the asphericity of spin magnetization µspin can be obtained only in combination with 7Tz ! Osnova Magnetismus: Kdo je kdo, co je co Motivace: Proč nanosystémy ? Magnetismus volných a adsorbovaných clusterů — srovnánı́ Rentgenová absorpčnı́ spektroskopie snadno a rychle Součtová pravidla pro XMCD spektroskopii: důvěřuj, ale prověřuj Does Tz term really matter ? ◮ For bulk systems it is usually negligible ◮ For surfaces, monolayers or wires, absolute value of 7Tz is about 20 % of µspin ◮ For investigating trends of µspin within a series of systems, what matters is how Tz varies from one system to another ◮ If variations in Tz are small, neglect of Tz causes just an overall shift of the deduced µspin ◮ Could Tz vary in such a way that the overall trends of µspin +7Tz and µspin would be quite different ? Does Tz term really matter ? ◮ For bulk systems it is usually negligible ◮ For surfaces, monolayers or wires, absolute value of 7Tz is about 20 % of µspin ◮ For investigating trends of µspin within a series of systems, what matters is how Tz varies from one system to another ◮ If variations in Tz are small, neglect of Tz causes just an overall shift of the deduced µspin ◮ Could Tz vary in such a way that the overall trends of µspin +7Tz and µspin would be quite different ? Does Tz term really matter ? ◮ For bulk systems it is usually negligible ◮ For surfaces, monolayers or wires, absolute value of 7Tz is about 20 % of µspin ◮ For investigating trends of µspin within a series of systems, what matters is how Tz varies from one system to another ◮ If variations in Tz are small, neglect of Tz causes just an overall shift of the deduced µspin ◮ Could Tz vary in such a way that the overall trends of µspin +7Tz and µspin would be quite different ? Does Tz term really matter ? ◮ For bulk systems it is usually negligible ◮ For surfaces, monolayers or wires, absolute value of 7Tz is about 20 % of µspin ◮ For investigating trends of µspin within a series of systems, what matters is how Tz varies from one system to another ◮ If variations in Tz are small, neglect of Tz causes just an overall shift of the deduced µspin ◮ Could Tz vary in such a way that the overall trends of µspin +7Tz and µspin would be quite different ? Does Tz term really matter ? ◮ For bulk systems it is usually negligible ◮ For surfaces, monolayers or wires, absolute value of 7Tz is about 20 % of µspin ◮ For investigating trends of µspin within a series of systems, what matters is how Tz varies from one system to another ◮ If variations in Tz are small, neglect of Tz causes just an overall shift of the deduced µspin ◮ Could Tz vary in such a way that the overall trends of µspin +7Tz and µspin would be quite different ? To find out wheter Tz may turn nasty or not ◮ Take a series of supported clusters of different sizes ◮ For each cluster size, evaluate average of d components of µspin N 1 X (j) µspin N j=1 and of [µspin + 7Tz ]/nh (j) (j) N 1 X µspin + 7Tz (j) N n j=1 ◮ h Compare how µspin and [µspin + 7Tz ]/nh depend on the cluster size To find out wheter Tz may turn nasty or not ◮ Take a series of supported clusters of different sizes ◮ For each cluster size, evaluate average of d components of µspin N 1 X (j) µspin N j=1 and of [µspin + 7Tz ]/nh (j) (j) N 1 X µspin + 7Tz (j) N n j=1 ◮ h Compare how µspin and [µspin + 7Tz ]/nh depend on the cluster size To find out wheter Tz may turn nasty or not ◮ Take a series of supported clusters of different sizes ◮ For each cluster size, evaluate average of d components of µspin N 1 X (j) µspin N j=1 and of [µspin + 7Tz ]/nh (j) (j) N 1 X µspin + 7Tz (j) N n j=1 ◮ h Compare how µspin and [µspin + 7Tz ]/nh depend on the cluster size To find out wheter Tz may turn nasty or not ◮ Take a series of supported clusters of different sizes ◮ For each cluster size, evaluate average of d components of µspin N 1 X (j) µspin N j=1 and of [µspin + 7Tz ]/nh (j) (j) N 1 X µspin + 7Tz (j) N n j=1 ◮ h Compare how µspin and [µspin + 7Tz ]/nh depend on the cluster size Systems Clusters: FeN , CoN Substrates: Ni(001), Au(111) Clusters on Au(111) N=1–7 Clusters on Ni(001) N=1–9 Results: µspin and [µspin + 7Tz ]/nh 0.9 0.8 0.7 ( spin + 7 Tz) / nh [ B] no Tz with Tz 1 3 5 7 9 [ B] 3.2 spin 3.0 2.8 Fe on Ni 1 3 5 7 9 no. of Fe atoms N FeN /Ni(001) CoN /Ni(001) FeN /Au(111) CoN /Au(111) Results: µspin and [µspin + 7Tz ]/nh B] 0.7 ( 0.7 + 7 Tz) / nh [ 0.8 0.8 spin 0.9 ( spin + 7 Tz) / nh [ B] no Tz with Tz 1 3 5 7 9 1 3 5 7 9 2.1 B] 2.0 [ [ B] 3.2 spin spin 3.0 1.9 2.8 Fe on Ni 1.8 Co on Ni 1 3 5 7 9 no. of Fe atoms N 1 3 5 7 9 no. of Co atoms N FeN /Ni(001) CoN /Ni(001) FeN /Au(111) CoN /Au(111) Results: µspin and [µspin + 7Tz ]/nh 1 3 5 7 B] 9 1 3 5 7 5 7 B] [ B] 1.9 spin spin 3.0 2.8 1.8 3 3.4 2.0 [ B] [ 0.7 1 3.2 spin 0.8 9 2.1 Fe on Ni 0.9 ( spin + 7 Tz) / nh [ B] 0.7 ( 0.7 + 7 Tz) / nh [ 0.8 0.8 spin 0.9 ( spin + 7 Tz) / nh [ B] no Tz with Tz 3.2 3.0 Co on Ni Fe on Au 2.8 1 3 5 7 9 no. of Fe atoms N 1 3 5 7 9 no. of Co atoms N 1 3 5 7 no. of Fe atoms N FeN /Ni(001) CoN /Ni(001) FeN /Au(111) CoN /Au(111) Results: µspin and [µspin + 7Tz ]/nh B] + 7 Tz) / nh [ spin 0.8 0.7 0.8 0.6 0.4 ( B] 0.9 ( spin + 7 Tz) / nh [ B] 0.7 ( 0.7 + 7 Tz) / nh [ 0.8 0.8 spin 0.9 ( spin + 7 Tz) / nh [ B] no Tz with Tz 0.2 7 9 1 3 5 7 9 1 2.1 1.9 spin spin 3.0 spin 7 1 2.8 3.2 3.0 Co on Ni 3 5 7 2.1 B] [ B] [ B] [ 2.0 1.8 5 3.4 3.2 Fe on Ni 3 B] 5 2.0 [ 3 spin 1 Fe on Au 1.9 1.8 Co on Au 2.8 1 3 5 7 9 no. of Fe atoms N 1 3 5 7 9 no. of Co atoms N 1 3 5 7 no. of Fe atoms N FeN /Ni(001) CoN /Ni(001) FeN /Au(111) 1 3 5 7 no. of Co atoms N CoN /Au(111) Results: µspin and [µspin + 7Tz ]/nh B] + 7 Tz) / nh [ spin 0.8 0.7 0.8 0.6 0.4 ( B] 0.9 ( spin + 7 Tz) / nh [ B] 0.7 ( 0.7 + 7 Tz) / nh [ 0.8 0.8 spin 0.9 ( spin + 7 Tz) / nh [ B] no Tz with Tz 0.2 7 9 1 3 5 7 9 1 2.1 1.9 spin spin 3.0 spin 7 1 2.8 3.2 3.0 Co on Ni 3 5 7 2.1 B] [ B] [ B] [ 2.0 1.8 5 3.4 3.2 Fe on Ni 3 B] 5 2.0 [ 3 spin 1 Fe on Au 1.9 1.8 Co on Au 2.8 1 3 5 7 9 no. of Fe atoms N 1 3 5 7 9 no. of Co atoms N 1 3 5 7 no. of Fe atoms N FeN /Ni(001) CoN /Ni(001) FeN /Au(111) 1 3 5 7 no. of Co atoms N CoN /Au(111) CoN /Au(111): Tz changes the picture completely ! 0.8 ◮ For CoN on Au(111), trends of µspin and of [µspin + 7Tz ]/nh are exactly opposite. ◮ Ignoring variations in Tz would lead to a false estimate of how µspin per atom depends on the cluster size 0.6 0.4 ( spin + 7 Tz) / nh [ B] no Tz with Tz 0.2 1 3 5 7 2.0 spin [ B] 2.1 1.9 1.8 Co on Au 1 3 5 7 no. of Co atoms N CoN /Au(111): Tz changes the picture completely ! 0.8 ◮ For CoN on Au(111), trends of µspin and of [µspin + 7Tz ]/nh are exactly opposite. ◮ Ignoring variations in Tz would lead to a false estimate of how µspin per atom depends on the cluster size 0.6 0.4 ( spin + 7 Tz) / nh [ B] no Tz with Tz 0.2 1 3 5 7 2.0 spin [ B] 2.1 1.9 1.8 Co on Au 1 3 5 7 no. of Co atoms N CoN /Au(111): Tz changes the picture completely ! 0.8 ◮ For CoN on Au(111), trends of µspin and of [µspin + 7Tz ]/nh are exactly opposite. ◮ Ignoring variations in Tz would lead to a false estimate of how µspin per atom depends on the cluster size 0.6 0.4 ( spin + 7 Tz) / nh [ B] no Tz with Tz 0.2 1 3 5 7 2.0 spin [ B] 2.1 1.9 1.8 Co on Au 1 3 5 7 no. of Co atoms N Sum rules themseleves can be checked L2,3-edge XANES 6 L3 ◮ Calculate the spectra theoretically ◮ Aply XMCD sum rules ◮ Compare µspin and µorb derived from XMCD spectra with moments calculated directly ◮ Upper integration boundary Ecut chosen so that there are exactly 10 electron states up to Ecut in the d band isotropic absorption 4 L2 2 L2,3-edge XMCD 0 2 0 difference (+) (-) - -2 -4 -10 0 10 20 energy [eV] 30 Sum rules themseleves can be checked L2,3-edge XANES 6 L3 ◮ Calculate the spectra theoretically ◮ Aply XMCD sum rules ◮ Compare µspin and µorb derived from XMCD spectra with moments calculated directly ◮ Upper integration boundary Ecut chosen so that there are exactly 10 electron states up to Ecut in the d band isotropic absorption 4 L2 2 L2,3-edge XMCD 0 2 0 difference (+) (-) - -2 -4 -10 0 10 20 energy [eV] 30 Sum rules themseleves can be checked L2,3-edge XANES 6 L3 ◮ Calculate the spectra theoretically ◮ Aply XMCD sum rules ◮ Compare µspin and µorb derived from XMCD spectra with moments calculated directly ◮ Upper integration boundary Ecut chosen so that there are exactly 10 electron states up to Ecut in the d band isotropic absorption 4 L2 2 L2,3-edge XMCD 0 2 0 difference (+) (-) - -2 -4 -10 0 10 20 energy [eV] 30 Sum rules themseleves can be checked L2,3-edge XANES 6 L3 ◮ Calculate the spectra theoretically ◮ Aply XMCD sum rules ◮ Compare µspin and µorb derived from XMCD spectra with moments calculated directly ◮ Upper integration boundary Ecut chosen so that there are exactly 10 electron states up to Ecut in the d band isotropic absorption 4 L2 2 L2,3-edge XMCD 0 2 0 -2 Ecut -4 -10 0 10 difference (+) (-) 20 energy [eV] 30 Validity of sum rules 0.045 ( spin orb / nh orb /( spin + 7Tz) FeN / Ni(001) directly 0.035 0.05 spin 0.045 /( spin orb/nh + 7Tz) per Fe atom per Fe atom 0.7 per Fe atom 0.04 0.8 0.03 ( orb + 7Tz)/nh 0.055 + 7Tz) / nh 0.04 0.6 0.025 2 4 6 number of Fe atoms 8 2 4 6 number of Fe atoms 8 2 4 6 8 number of Fe atoms ◮ Trends of the “effective moments” (µspin + 7Tz )/nh and µorb /nh are reproduced well enough ◮ Applying sum rules to supported clusters in principle makes sense ◮ However, beware of the Tz term (see above) Validity of sum rules 0.045 ( spin orb / nh orb /( spin + 7Tz) FeN / Ni(001) 0.035 0.05 spin 0.045 /( spin orb/nh + 7Tz) per Fe atom per Fe atom 0.7 directly from XAS per Fe atom 0.04 0.8 0.03 ( orb + 7Tz)/nh 0.055 + 7Tz) / nh 0.04 0.6 0.025 2 4 6 number of Fe atoms 8 2 4 6 number of Fe atoms 8 2 4 6 8 number of Fe atoms ◮ Trends of the “effective moments” (µspin + 7Tz )/nh and µorb /nh are reproduced well enough ◮ Applying sum rules to supported clusters in principle makes sense ◮ However, beware of the Tz term (see above) Validity of sum rules 0.045 ( spin orb / nh orb /( spin + 7Tz) FeN / Ni(001) 0.035 0.05 spin 0.045 /( spin orb/nh + 7Tz) per Fe atom per Fe atom 0.7 directly from XAS per Fe atom 0.04 0.8 0.03 ( orb + 7Tz)/nh 0.055 + 7Tz) / nh 0.04 0.6 0.025 2 4 6 number of Fe atoms 8 2 4 6 number of Fe atoms 8 2 4 6 8 number of Fe atoms ◮ Trends of the “effective moments” (µspin + 7Tz )/nh and µorb /nh are reproduced well enough ◮ Applying sum rules to supported clusters in principle makes sense ◮ However, beware of the Tz term (see above) Validity of sum rules 0.045 ( spin orb / nh orb /( spin + 7Tz) FeN / Ni(001) 0.035 0.05 spin 0.045 /( spin orb/nh + 7Tz) per Fe atom per Fe atom 0.7 directly from XAS per Fe atom 0.04 0.8 0.03 ( orb + 7Tz)/nh 0.055 + 7Tz) / nh 0.04 0.6 0.025 2 4 6 number of Fe atoms 8 2 4 6 number of Fe atoms 8 2 4 6 8 number of Fe atoms ◮ Trends of the “effective moments” (µspin + 7Tz )/nh and µorb /nh are reproduced well enough ◮ Applying sum rules to supported clusters in principle makes sense ◮ However, beware of the Tz term (see above) Validity of sum rules 0.045 ( spin orb / nh orb /( spin + 7Tz) FeN / Ni(001) 0.035 0.05 spin 0.045 /( spin orb/nh + 7Tz) per Fe atom per Fe atom 0.7 directly from XAS per Fe atom 0.04 0.8 0.03 ( orb + 7Tz)/nh 0.055 + 7Tz) / nh 0.04 0.6 0.025 2 4 6 number of Fe atoms 8 2 4 6 number of Fe atoms 8 2 4 6 8 number of Fe atoms ◮ Trends of the “effective moments” (µspin + 7Tz )/nh and µorb /nh are reproduced well enough ◮ Applying sum rules to supported clusters in principle makes sense ◮ However, beware of the Tz term (see above) XMCD spectroscopy of clusters: Conclusions ◮ Sum rules can yield useful information about trends in (µspin + 7Tz )/nh and µorb /nh ◮ However, the Tz term is a bad guy — it can completely reverse the trends of (µspin + 7Tz )/nh and µspin XMCD spectroscopy of clusters: Conclusions ◮ Sum rules can yield useful information about trends in (µspin + 7Tz )/nh and µorb /nh ◮ However, the Tz term is a bad guy — it can completely reverse the trends of (µspin + 7Tz )/nh and µspin Praktický závěr (nejen pro magnetismus nanostruktur) Pokud vı́te, co děláte, můžete si dovolit leccos