Band Theory of Ferromagnetism Angle Resolved Photoelectron Spectroscopy (ARPES) Phys. 342/555:Ch12B (2008) The Band Theory of Ferromagnetism Model including both band structure and correlation of same spin electrons. Stoner and Wolfarth Ansatz: E↑ (k) = E(k) − E↓ (k) = E(k) − Define n↑ − n↓ R= N I(n↑ + n↓)/2N In↑ Stoner I is a parameter describing energy reduction due to correlation N In↓ N M= μB N V IR N ~ IR E↓ (k) = E (k) + N R= μB N n↑ − n↓ V ~ E↑ (k) = E (k) − ~ E (k) = E(k) − N I(n↑ + n↓ ) N Phys. 342/555:Ch12B (2008) The Band Theory of Ferromagnetism Calculate R n↑ − n↓ R= N 1 1 1 R = ∑ f↑ − f↓ = ∑ N k N k exp( E˜ (k) − IR /2 − E F /kT) + 1 1 1 − ∑ N k exp( E˜ (k) + IR /2 − E F /kT) + 1 There are certain conditions when R is non-zero even when B=0, i.e. Ferromagnetism Expand for small R Δx 2 Δx f (x − ) − f (x + ) = − f '(x)Δx − f '''(x) 2 3! 2 Phys. 342/555:Ch12B (2008) The Band Theory of Ferromagnetism 1 1 1 ∂f (k) ∂ 3 f (k) 3 R = ∑ f↑ − f↓ = ∑ IR − ( IR) ∑ ˜ N k N k ∂E (k) 24N k ∂E˜ (k) 3 Look at f’ and f’’’. First derivative is negative the third is positive. Ferromagnetism occurs when R>0 For R>0 T=0 I ∂f (k) −1− ∑ >0 N k ∂E˜ (k) ∂f (k) I V − ∑ = ˜ N k ∂E (k) (2π) 3 N I ∂f (k) V − ∑ = ˜ N k ∂E (k) (2π) 3 N ⎛ ∂f ⎞ ∫ dk⎜⎝− ∂E˜ ⎟⎠ Spin 1V ˜ ∫ dkδ( E − E F ) = 2 N D(E F ) Phys. 342/555:Ch12B (2008) The Band Theory of Ferromagnetism Define V ˜ D(E F ) = D(E F ) 2N Stoner Criterion The Curie Temperature is the temperature where Ferromagnetism goes to 0 ID˜ (E F ) > 1 I ∂f (k) −1− ∑ =0 N k ∂E˜ (k) Phys. 342/555:Ch12B (2008) The Band Theory of Ferromagnetism Calculated I Calculated D(E) Phys. 342/555:Ch12B (2008) The Band Theory of Ferromagnetism Consider applying a magnetic Field 1 1 1 R = ∑ f↑ − f↓ = ∑ N k N k exp( E˜ (k) − IR /2 − μB B − E F /kT) 1 1 − ∑ N k exp( E˜ (k) + IR /2 + μB B − E F /kT) R = D˜ (E F )(IR + 2μB B) N 2 N ˜ M = μB R = D(E F )(IM + 2μB B) V V First order when T=0 N D˜ (E F ) M = 2μ B V 1− ID˜ (E F ) 2 B χ= μ 0M B χ= χ Pauli 1− ID˜ (E F ) Phys. 342/555:Ch12B (2008) The Temperature Behavior of a Ferromagnetic in the Band Model A simple model to account for band theory Consider only d-electrons at Fermi energy In Ni there are ~0.54 d holes per atom. So magnetic moment is μBeff~0.54μB Phys. 342/555:Ch12B (2008) Band Model of Ferromagnetism Model, density of spin up and down bands is a delta function, separated in energy by exchange spliting I. μB ⎡ IR IR ⎤ ˜ D(E) = δ (E − E F − μB B − ) + δ (E − E F + μB B + )⎥ ⎢ 2 2 ⎦ μB ⎣ eff Fermi Energy half way between the two δ functions ⎡ ⎤ μB eff ⎢ ⎥ 1 1 R= − ⎢ ⎥ μB ⎢ exp[(−μ B − IR ) /kT] + 1 exp[(μ B + IR ) /kT] + 1⎥ B B ⎣ ⎦ 2 2 IμB eff TC = μB 4k Again we want R>0 for Ferromagnetism Look at B=0 with the abbreviations μB ˜ R= R μ B Phys.eff342/555:Ch12B (2008) Band Model of Ferromagnetism ˜T ⎡ ⎤ R 1 1 C = Tanh − R˜ = ⎢ ⎥ ˜ ˜ exp[2 R T exp[−2 R T /T] + 1 /T] + 1 T ⎣ ⎦ C C Look R˜ = 1 forT = 0 R˜ = 0 forT = TC Tc is the Curie Temperature M(T)=0 When T<<TC When T~TC −2TC /T ˜ R = 1− 2e 1/ 2 ⎛ ⎞ T R˜ = √ 3⎜1− ⎟ ⎝ TC ⎠ Phys. 342/555:Ch12B (2008) Band Model of Ferromagnetism Looks pretty good! Phys. 342/555:Ch12B (2008) Band Model of Ferromagnetism Power law wrong T<<TC Wrong Problem is low energy spin excitations Phys. 342/555:Ch12B (2008) Band Model of Ferromagnetism What does this predict above the Curie Temperature? ˜T ⎡ ⎤ R 1 1 = Tanh C R˜ = ⎢ − ⎥ T ⎣exp[−2R˜ T /T] + 1 exp[2 R˜ T /T] + 1⎦ C C Only have magnetization when B≠0: expand around small R and B μB 1 μB TC ˜ ˜ ˜ B0 R= R= B0 + R 2k T − TC 2kT T C Curie-Weiss Law χ= T − TC C Curie-Weiss Constant Phys. 342/555:Ch12B (2008) Currie-Weiss Law For T>>TC Currie-Weiss Works C χ= T − TC For T~TC Currie-Weiss does not works: (T-TC)-4/3 C related to saturation magnetization: Does not work, gives a C much too small! We could use Stoner I and definition of Tc to determine Tc using the measured value of the exchange splitting: Doesn’t work Tc is much too large! The failure of this picture is not because we have the band structure wrong, but because we do not have the low energy magnetic excitations done properly! Phys. 342/555:Ch12B (2008) Calculated Spin Density of State Half Metallic!!! Phys. 342/555:Ch12B (2008) Half Metallic QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) Point Defects QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Phys. 342/555:Ch12B (2008) ARPES (angle resolved photoemission spectroscopy) Momentum conservation EB ~0 k|| ( f ) = k|| (i ) + k|| ( p ) k|| (i ) = k|| ( f ) = 2 mE k ( f ) sin (ϑ ) Energy conservation Ek = hν − Φ 0 − EB Hence, the E B and the k|| (i ) can be obtained. F. Reinert et. al., New J. Phys. 7, 97(2005). Phys. 342/555:Ch12B (2008) New ARPES in SERF This is being built currently Scienta R4000 Phys. 342/555:Ch12B (2008) Phys. 342/555:Ch12B (2008) Photoemission Matrix Element Fermi golden rule Hamiltonian dω 2π = |< Ψf | H'| Ψi >|2 {δ (E f − E i − ω )} dt e e2 2 ( A • P + P • A) − eΦ + | A | H = H0 + 2mc 2 2mc Pick guage so scaler potential is 0: |A|2 is always small and if we use commutator [ P, A] = −i ∇ ⋅ A | A |2 dσ ∝|< Ψf | 2 A • P − i ∇ • A | Ψi >|2 {δ (E f − E i − ω )} dΩ Second term usually small A(r,t) = A0 exp(−iωt + iq • r ) Phys. 342/555:Ch12B (2008) Photoemission Matrix Element: II dσ ∝|< Ψf | P | Ψi > • A0 |2 {δ (E f − E i − ω )} dΩ Use commutation relations of H0 with P and r, yielding dσ ∝|< Ψf | r | Ψi > • A0 |2 {δ (E f − E i − ω )} dΩ dσ ∝|< Ψf | ∇V | Ψi > • A0 |2 {δ (E f − E i − ω )} dΩ Single particle representation Ψ0N = φ i ΨiN −1 : initial − state ψ f = uk Φi, j (N −1) : final − state Hole in ith orbital Phys. 342/555:Ch12B (2008) Photoemission Matrix Element: III dσ ∝|< uk | A • P | φ i >< Φ i, j | Ψi >|2 {δ (E f − E i − ω )} dΩ For an independent particle representation, there is no rearrangement of charge so < Φi, j | Ψi >= δi,0 dσ ∝|< uk | A • P | φ i >|2 {δ (E f − E i − ω )} dΩ Symmetry Rules: Very important Uk always even in plane of emission to have signal in detector Phys. 342/555:Ch12B (2008) Photoexcition in a periodic system Momentum picked up by lattice: G Vertical Excitiation Phys. 342/555:Ch12B (2008) Projection of Bulk onto Surface 100 surface of a fcc material: Top is the Surface Brillouin zone and bottom shows the bulk Brillouin zone: This shows the projection of the bulk onto the surface. The solid line represents a rod connection all points in the bulk zone with the one point in the surface zone with the same k\\ The perpendicular component of the momentum of the photoexcited state inside the solid is undertermined experimentally. The value of k(perpendicular) could be anywhere on a rod in k-space with fixed k]] E i = E kin + eφ − ω Phys. 342/555:Ch12B (2008) 3-D assuming Free Electron Bands Fcc Cu: The vertical axis is normal to the surface and the horizontal axis is in two different directions in the bulk. The top shows circles of a the free electron bands at different energy. The dashed lines show the k versus E location for transiton seen with a detector at a fixed angle. The effects of bands crossing zone boundaries is shown at the bottom. Phys. 342/555:Ch12B (2008) Excitation process in Cu Normal Emission Cu(100) The position, number, and intensity of the peaks depends upon photon energy. This is never a density of states. What causes the width?? Phys. 342/555:Ch12B (2008) Data for Normal Emission Cu(100) Intensity vs. Photon energy for energy 0.13 below EF Data vs. Photon Energy Dispersion assuming free electron final bands Phys. 342/555:Ch12B (2008) Al data: procedure Dashed Free Electron Solid DFT Phys. 342/555:Ch12B (2008) Normal Emission: k\\=0 Surface State Direct Transitions Plasmons Free Electron bands Real Data342/555:Ch12B for normal Emission Phys. (2008) Normal Emission: This is the kind of measurement that should be made by anyone wanting to know the bulk band structure. Bottom of band Phys. 342/555:Ch12B (2008) X-W band mapping: fix k|| M in SBZ Phys. 342/555:Ch12B (2008) Fix k|| sweep photon energy Phys. 342/555:Ch12B (2008) Γ−X direction in the Bulk Gap Wrong Band Width Wrong Phys. 342/555:Ch12B (2008) Γ−X direction in the Bulk Gap Wrong Band Width Wrong Many-body effect Phys. 342/555:Ch12B (2008) X-W in the bulk There is still a problem with the Gap when compared to DFT calculations! Phys. 342/555:Ch12B (2008) X-W in the bulk Phys. 342/555:Ch12B (2008) How did we do? Phys. 342/555:Ch12B (2008) Final States Γ−X Data Folded Back Band Structure Calculated Band Structure with dark lines where there is amplitude in the right direction. Phys. 342/555:Ch12B (2008) Final States X-W Data Folded Back Band Structure Calculated Band Structure Phys. 342/555:Ch12B (2008) Line Width: Imaginary part of interaction Phys. 342/555:Ch12B (2008) Absolute Measurement: 1988 Phys. 342/555:Ch12B (2008) Band Structure of Na Phys. 342/555:Ch12B (2008) Band Structure of Na Band Width Wrong!! Phys. 342/555:Ch12B (2008) Band Structure of Na Real part of Self Energy Re{Σ(k)} ≡ E(k) − E 0 (k) Phys. 342/555:Ch12B (2008) Band Width in Simple Metals Rs is an inverse measure of electron density Theory Looks Good, BUT!!! Phys. 342/555:Ch12B (2008)