Appendix A Properties of Yttrium–Iron–Garnet (YIG)
Structure and Physical Properties
Chemical formula Y3Fe5O12 Crystal structure Cubic Number of formula units per unit cell 8 Lattice constant (25 ◦C) 12.376 A˚ Mass density (25 ◦C) 5172 kg/m3 Octahedral Sublattice 27 −3 Site density Na 8.441 × 10 m Angular momentum Ja 5/2 Land´e g factor ga 2 Magnetization at 0 K Ma (0 K) 391.5 kA/m Tetrahedral Sublattice 27 −3 Site density Nd 12.66 × 10 m Angular momentum Jd 5/2 Land´e g factor gd 2 Magnetization at 0 K Md (0 K) 587.2 kA/m Molecular Field Constants [1]
Octahedral λaa 735.84 Tetrahedral λdd 344.59 Nearest-neighbor λad 1100.3 Phenomenological Exchange Constant [2] 2 −16 2 Defined by hex = λex∇ m (3.104) λex 3 × 10 m Macroscopic Magnetic Properties
Magnetization at 0 K Mtot (0 K) 196 kA/m Magnetization at 298 K Mtot (298 K) 140 kA/m Curie temperature Tc 559 K
333 334 References
Anisotropy Constants 3 First-order cubic (4.2 K) [3] Kc1 (4.2 K) −2480 J/m 3 First-order cubic (295 K) [3] Kc1 (295 K) −610 J/m 3 Second-order cubic (4.2 K) [4] Kc2 (4.2 K) −118.0J/m 3 Second-order cubic (273 K) [4] Kc2 (273 K) −26.0J/m Optical Properties [5–7] Refractive index (1.2 μm) n 2.2 Faraday rotation (1.2 μm) φF 240 deg/cm Absorption coefficient (1.2 μm) α 0.069 cm−1 μ 2 − × −4 Cotton–Mouton tensor elements (1–3 m) g44MS 1.14 10 2 × −5 ΔgMS 5.73 10
Cotton–Mouton rotation (1.14 μm) ΦCM −160 deg/cm
References
[1] E. E. Anderson, “Molecular field model and magnetization of YIG,” Phys. Rev., vol. 134, p. A1581, 1964. [2] E. H. Turner, “Interaction of phonons and spin waves in yttrium iron garnet,” Phys. Rev. Lett., vol. 5, p. 100, 1960. [3] P. Hansen, “Anisotropy and magnetostriction of gallium-substituted yt- trium iron garnet,” J. Appl. Phys., vol. 45, p. 3638, 1974. [4] P. Escudier, “L’anisotropie de l’aimantation; un param´etre important de l’´etude de l’anisotropie magn´etocrystalline,” Ann. Phys. (Paris), vol. 9, p. 125, 1975. [5] M. J. Weber, Ed., CRC Handbook of Laser Science and Technology, Vol. IV Optical Materials Part 2: Properties. Boca Raton, FL: CRC Press, 1986, p. 291. [6] R. V. Pisarev, I. G. Sinii, N. N. Kolpakova, and Y. M. Yakovlev, “Magnetic birefringence of light in iron garnets,” Sov. Phys. JETP, vol. 33, p. 1175, 1971. [7] G. A. Smolensky, R. V. Pisarev, I. G. Sinii, and N. N. Kolpakova, “Cotton- Mouton birefringence of ferrimagnetic garnets,” J. Physique Colloque, vol. C1 32, pp. c1–1048, 1971. Appendix B Currents in Quantum Mechanics
B.1 Density of States
The valence electrons in a metal can be modeled as free electrons contained in a box the size of the sample. Because of the boundary conditions at the faces of the box, the allowed values of wave number k are quantized. The actual conditions at the surface of a metal may be very complicated, but much of solid-state physics depends primarily on the fact that the modes are quantized, rather than on exactly what the boundary conditions of a particular sample happen to be. Consequently, it is common to make the simple assumption of periodic boundary conditions. In this assumption, the wave function of an electron at x = L should equal to that at x = 0, where L is the length of the sample along the x-direction. For this to be true, we have exp(ikL) = exp(0) = 1, which gives the discrete values of wave number k =2πn/L, where n is an integer. In k space, the total number of states N in a sphere of radius |k| is given by the volume of the sphere divided by the volume occupied by one state. Thus
4 1 4πk3L3 N =2 πk3 =2 , (B.1) 3 (2π/L)3 3(2π)3 where we have assumed the sample is a cube with length L on each side so that 3 the minimum volume in k space is ΔkxΔkyΔkz =(2π/L) . The factor of 2 accounts for the degeneracy in the energy of spin-up and spin-down electrons.1 In the free electron approximation, the Hamiltonian H is merely the kinetic energy operator (cf. Table 1.1)
p2 2 H = = − ∇2 , (B.2) 2m 2m 1 When we consider ferromagnetic metals, this degeneracy is lifted by the inter- action with the internal exchange field causing a splitting of the energy band diagram.
335 336 Appendix B Currents in Quantum Mechanics and when we assume an electron wavefunction of the form ψ ∼ eik·r,the Schr¨odinger equation Hψ = Eψ yields
2k2 = E. (B.3) 2m Combining (B.1) and (B.3), the number of states with energy less than E can be written 4πL3 2m 3/2 N =2 E3/2. (B.4) 3(2π)3 2 The density of states is defined as the number of states per unit energy, or dN g(E)= . (B.5) dE Thus, the density of states is given by 4πL3 2m 3/2 3 g(E)=2 E1/2 (B.6) 3(2π)3 2 2 L3 2m 3/2 √ = E. (B.7) 2π2 2
The Fermi energy is defined as the highest occupied energy state by an electron, at absolute zero temperature. Interactions at higher temperatures 2 2 typically occur near the Fermi energy, EF = kF/(2m), since electrons at the Fermi energy have nearby empty states. Consequently, the current is often calculated at the Fermi energy while assuming the parabolic band approxi- mation g(E) ∝ E1/2. Multiplying and dividing (B.7) by E and comparing the result with (B.4), we see that the relationship between the Fermi energy and the density of states can be expressed 3 N g(EF)= . (B.8) 2 EF At finite temperature, some of the electrons will be excited to states above the Fermi level. In this case, the density of states is related to the number of particles by ∞ N = g(E)f(E)dE , (B.9)
0 where − − 1 f(E)= 1+e(E EF)/(kBT ) (B.10) is the Fermi–Dirac distribution describing the probability of finding an elec- −23 tron with energy E at temperature T and kB =1.38 × 10 J/Kisthe Boltzmann constant. B.2 Electric and Spin Current Densities 337
Note that the above expressions assume degenerate spin-up and spin-down subbands. In a ferromagnetic metal, the molecular field interacting with the electron spin splits the bands as illustrated in Figure 10.1. If the magnetization is oriented upwards, then the Hamiltonian with the effects of the molecular field is (cf. Eq. (2.30)) −2 d 2 H = + μ μ σH , (B.11) 2m dx 0 B m where Hm is the molecular field and σ = ±1. Defining Em = μ0μBσHm,the density of states for the spin subbands can be expressed 1 g (E)= g (E − σE ) , (B.12) σ 2 m and the total density of states is given by g(E)= gσ(E) (B.13) σ = g↑(E)+g↓(E) (B.14) 1 = [g (E − E )+g (E + E )] . (B.15) 2 m m (B.16)
B.2 Electric and Spin Current Densities
Consider ψ to be a solution of the 1D Schr¨odinger equation
∂ψ 2 i = − ∇2ψ. (B.17) ∂t 2m Multiplying (B.17) by ψ∗ and subtracting the complex conjugate of the re- sulting equation, we obtain ∂ −i (ψ∗ψ)+ ∇·[ψ∗∇ψ − ψ∇ψ∗]=0. (B.18) ∂t 2m In obtaining (B.18), we have used the vector identity
∇·(f∇g)=∇f ·∇g + f∇2g. (B.19)
If we define the probability density of finding an electron as ψ∗ψ,wecan express the charge density as ρ = qψ∗ψ. We can also define a current density j that satisfies the conservation equation ∂ρ + ∇·j =0. (B.20) ∂t 338 Appendix B Currents in Quantum Mechanics
Multiplying (B.18) by the electronic charge q and comparing with (B.20), we make the identification −iq j = [ψ∗∇ψ − ψ∇ψ∗] 2m (B.21) q = Im(ψ∗∇ψ). m √ Choosing ψ =1/ keikx for the 1D case ensures the normalization dψ Im ψ∗ =1. (B.22) dx Under the two-band approximation, we treat the spin-up and spin-down carriers as being independent. Under this two-channel approximation, we can borrow standard techniques from quantum mechanics to solve for the prob- ability of carrier transmission across an interface, or even through a barrier with multiple interfaces.
B.3 Reflection and Transmission at a Boundary
Consider an electron that is incident on a planar interface between two media as illustrated in Figure B.1. The electron will be reflected with probability
Medium 1 Medium 2
t r x
Fig. B.1. Geometry for calculating the reflection and transmission coefficients of a particle at a boundary between two materials. amplitude r, and transmitted with probability amplitude t. In medium 1, the wave function is
ψ1 = ψinc + ψref (B.23) 1 − = √ eik1x + re ik1x , (B.24) k1 (B.25) while in medium 2, the wave function is B.4 Tunneling Through a Barrier 339 t ik2x ψ2 = ψtr = √ e . (B.26) k2 The reflection and transmission coefficients can be found by requiring both the wave function and its derivative to be continuous at the boundary x =0. − + Requiring ψ1(0 )=ψ2(0 ) gives 1 1 √ (1 + r)=√ t. (B.27) k1 k2 Similarly, requiring dψ/dx to be continuous at x = 0 gives k1 (1 − r)=t k2. (B.28)
Solving (B.27) and (B.28) simultaneously gives the desired reflection and transmission amplitudes: k − k r = 1 2 , (B.29) k + k √1 2 2 k k t = 1 2 . (B.30) k1 + k2 It is straightforward to show that the probabilities of reflection and transmis- sion sum to 1: r2 + t2 =1. (B.31) These equations can be readily applied to the transmission and reflection of spin-up and spin-down electrons at an interface between a normal metal and a ferromagnetic metal. To do this, we make the identifications k1 = kN and k2 = kσ.Thus,wehave
kN − kσ rσ = , (B.32) kN + kσ √ 2 kN kσ tσ = , (B.33) kN + kσ and 2 2 rσ + tσ =1. (B.34)
B.4 Tunneling Through a Barrier2
Consider the quantum mechanical tunneling of particles with energy E through a potential barrier ) V −a/2 Fig. B.2. Tunneling through a potential barrier. A particle incident from the left has a finite probability of tunneling through the barrier, even though the particle energy E is less than the barrier height V . where V>E(Figure B.2). Assuming a free particle solution to the Schr¨odinger equation Hψ = Eψ, the wavefunction will have the form (for 1D problem) ψ = Ae+ikx + Be−ikx, (B.35) where the two terms represent plane waves of amplitude A and B traveling along the +xˆ and −xˆ directions, respectively. The wavenumber k is a measure of the oscillatory nature of ψ and can be obtained by substituting (B.35) into the Schr¨odinger equation 2 − ∇2ψ + U(x)=Eψ (B.36) 2m to yield 1 − k(x)= 2m(E U(x)). (B.37) We write the wavefunction for the particle in the three regions as3 ⎧ ⎨ Aeikx + Be−ikx x<−a/2, −κx κx − ψ(x)=⎩ Ce + De a/2 Since U(x) is piecewise constant, we can define 1 √ k = 2mE (B.39) outside the barrier, where U(x) = 0, and within the barrier U(x)=V and we define the real quantity κ: 1 − κ = 2m(V E). (B.40) 2 We rely heavily on the mathematics in [1]. 3 The normalization is not important in this case, since we are interested in the ratio of the transmitted to incident amplitudes, and the initial and final media both have the same k. References 341 We interpret 1/κ as an exponential decay length characteristic of tunneling through a barrier. The wavefunction in each region is determined by requiring that both ψ and dψ/dx be continuous across the interfaces x = −a/2and x = a/2. The boundary conditions at x = −a/2 yield the equations Ae−ika/2 + Beika/2 = Ceκa/2 + De−κa/2, (B.41) ikA e−ika/2 − ikB eika/2 = −κC eκa/2 + κD e−κa/2, (B.42) while the boundary conditions at x = a/2 give Ce−κa/2 + Deκa/2 = Feika/2, (B.43) −κC e−κa/2 + κD eκa/2 = ikF eika/2. (B.44) Normalizing the equations by A, we have four equations in the four unknown normalized coefficients. The ratio of the transmitted and incident wave amplitudes can thus be uniquely determined as [1] F e−ika = , (B.45) A cosh κa + i(ε/2) sinh κa where κ k ε = − . (B.46) k κ We are usually interested in quantities such as the transmissivity of the barrier defined as 2 F T = . (B.47) A For a high and wide barrier (κa 1), we obtain κk 2 T ≈ 16 e−2κa . (B.48) k2 + κ2 Note, however, that when we apply a bias between the two ferromagnetic electrodes, the potential U(x) will contain a linear term. Consequently, the solutions for ψ(x) within the barrier will no longer have the exponential form e±κx and will instead be replaced by Airy functions. References [1] E. Merzbacher, Quantum Mechanics, 3rd ed. New York: John Wiley & Sons, 1998. Appendix C Characteristics of Spin Wave Modes This appendix provides a summary of the characteristics of spin wave modes, as described in Chapters 5 and 6. We identify • H0 as the total internal field, • MS as the saturation magnetization, • HDC as the applied DC bias field, • g as the Lande’ g factor (2 for spin), • q and m as the charge (negative) and effective mass, respectively, of an electron, and • d as the film thickness • w as the width of the microstrip transducer • s as the separation between the transducer and the surface of a magnetic film and define ω0 = −γμ0H0, ωM = −γμ0MS, gq γ = (which is negative). 2m The coordinate system is generally defined so that H0 and HDC are along the zˆ direction. For simplicity, magnetocrystalline anisotropy is neglected in the following equations. C.1 Constitutive Tensors C.1.1 Polder Susceptibility Tensor