Professional Documents
Culture Documents
Lattice Vibrations
Lattice Vibrations
We introduce new concepts, namely phonons, dispersion relations and density of states. This lecture marks the beginning of a new part of the course. We will now be concerned with physical properties related to the atoms, and in particular to their vibrational modes. The models we will study are purely classical in the first part; only later will we introduce the quantum mechanical concept of phonons. We will mainly consider one-dimensional, linear models. This is a good approximation in certain symmetry directions, i.e. the [100], [110] and [111] directions in cubic structures. In these directions whole planes of atoms move in phase. The vibrations are called longitudinal if the displacement of the atoms is parallel to the wavevector (i.e. to the direction) of the propagating elastic wave. A transversal vibration has the wavevector orthogonal to the atomic displacements. 1. A single oscillator. We first consider the simplified case of only one oscillator. When a force is applied, the distance between atoms (x) becomes x+u, where u is the displacement from equilibrium. This case is treated in Hoffman, p. 43-44. The interatomic potential (see the lecture on crystal structure) can be written as a Taylor series for distances close to the equilibrium distance (lattice constant), a.
Since the first derivative is equal to zero at x=a, it is the second derivative of the potential that gives the lowest contribution to the force between the atoms, which in one dimension is: F = - d/dx = - d2(a)/dx2 (x-a) = - d2(a)/du2 u = - u, where is the force constant. The latter is of the order Ya, where Y is the bulk modulus (H p. 5354), or numerically from a few to a few hundred N/m. Vibrational frequencies in solids are of the order of 1012 to 1014 Hz (eq. 4.2). Strong force constants and light masses give rise to high frequencies. The vibration amplitude is usually a small fraction of the lattice spacing (eq. 4.5). 2. One-atomic linear chain (H p. 44-47). Here we consider a one-dimensional lattice in the x-direction. One atom (mass M) is sitting at each lattice point, x=(n+p)a (a is the lattice constant), and there is no basis. The force between any two atoms is assumed to be proportional to the difference between their displacements (see above, this is called the harmonic approximation), compare Hookes law. The force on atom n becomes: Fn = p (un+p-un), leading to the equation of motion,
M d2un/dt2 = p (un+p-un).
Here we have summed the interactions over all atoms at pn in the chain. We keep only the nearest-neighbor interactions in the following. Then the equation of motion becomes:
The solution becomes: 2 = (2/M) (1-cos(ka)) = (4/M) sin2(ka/2). This relation between the frequency and magnitude of the wavevector is called the dispersion relation (H p. 45). The dispersion relation is periodic in k with period 2/a. All information about the vibrations is present in the first Brillouin zone of the reciprocal lattice (k-space), i.e. for kvalues between -/a and /a.
The wavelength is 2/k; hence k-values outside of these limits represent wavelengths less than 2a. Vibrations with so short wavelengths are seen to include unphysical oscillations (nodes between the atoms), H fig.4.2, p. 46. The continuum model holds only in the long wavelength limit as k->0. Here the phase velocity vf=/k and the group velocity vg=d/dk of the wave are equal. For large k the phase and group velocities are no longer equal. At k=/a the group velocity is zero; we have a standing wave. 3. Two-atomic linear chain. Here we consider a linear lattice with a basis of two atoms having masses M1 and M2. They are positioned at x=(n+p)b and x=(n+p+1/2)b, respectively. Note that the lattice constant is now denoted b. The displacements are denoted un and vn, respectively. Considering only nearest-
neighbour interactions, the equations of motion are given and the dispersion relation derived by H, p. 47-48. The dispersion relation becomes
The dispersion relation now exhibits two branches, one with optical modes and one with acoustic modes. The nature of the vibrations at k=0 are shown in the figure below, for the case of an ionic compound.
The optical mode at k=0 has a frequency falling in the infrared range. For ionic crystals, such a transverse optical mode can be excited by infrared radiation of this frequency TO. The acoustic branch approaches zero as k->0, in this limit it has a velocity of sound. At k=/a, the sublattices vibrate independent of one another, they do not couple to one another. This model is a good approximation for ionic crystals, A+B-, for waves propagating in directions where an atomic plane contains only A+ or only B- ions. Link to applet showing the dispersion relation: http://solidstate.physics.sunysb.edu/teach/intlearn/phonon/phonon_dispersion.html
4. Finite chains. Phonons. We consider now a linear chain with N atoms and of length L=Na. The finite length gives rise to a finite number of vibrational modes and hence a finite number of allowed k-values. We treat the finiteness of real crystals by the method of periodic boundary conditions (H p. 48-50). In our one-dimensional case we put u(x)=u(x+L). This is like we had a finite chain winding so that its beginning were connected to its end! For a plane wave solution this leads to the condition exp(ikNa)=1, which is satisfied only when k=m2/Na, with m=0,1,2,3.....(N/2-1),(N/2). There are N allowed k-values (modes) within the first Brillouin zone. The example below has N=10.
The distance between two modes is k=2/Na=2/L. The number of modes per unit |k|-range is the inverse of this, but multiplied with two to take into account both positive and negative k. The density of states, the number of modes per unit frequency range is now obtained from g() d = (L/) (dk/d) d, and hence g() = L/vg. The energy of the lattice vibrations is quantized. The quanta (quasiparticles) are called phonons. The energy of a vibrational mode is described with a harmonic oscillator model, H p.50-51.
1 E (k ) = (l + ) h ( k ) 2
The mode can be occupied with any number, l, of phonons. The average occupancy, <l>, depends on the frequency and is given by the Planck distribution function, H p. 56. The phonons are bosons and so their number is not conserved. The phonon energy is given by
E ph = h (k )
The phonons are due to relative motions (vibrations) in the crystal, hence they do not impart momentum to the crystal as a whole. The phonon however acts if it has momentum p=(h/2)k , leading to conservation rules for the k-vector in scattering processes. These conservation rules must include the reciprocal lattice vector G, since scattering might lead to k-values outside the first Brillouin zone and a G is needed to get back to the first zone (compare diffraction where k=k+G) .
5. Three dimensional crystals (H p.51-53). Each atom has now three degrees of freedom. Hence for each k-vector there are now three modes; two transverse one and one longitudinal. Below we picture the longitudinal and transverse vibrations for the case of planes of atoms moving in phase.
When there is no basis, there exist three (acoustic) branches, one longitudinal and two transverse ones. In the case of a basis of p atoms, there exists in total 3p branches of which three are acoustic and 3(p-1) are optic ones. Each branch contains N modes as above, where N now is the number of primitive cells. Note that k is now a vector and its allowed values are within the first Brillouin zone of the threedimensional reciprocal lattice. In general, the dispersion relation, (k), has to be calculated numerically. They are usually depicted only in certain symmetry directions in the Brillouin zone. Some general relations for the density of states can be obtained, though. To obtain the density of states we first apply periodic boundary conditions in x-, y- and z-directions, leading to g(k)=V/83 for each branch of modes. We use the notation k=|k|. A general expression for the density of states is expressed as an integral over a surface of constant frequency in k-space.
g ( ) = (V / 8 3 ) dS / v g ,
where the integration is over a constant frequency surface in reciprocal space. The next figure depicts the dispersion relations for lead and copper, in certain symmetry directions. We observe three branches, two transverse and one longitudinal. The force constants for the transverse ones are generally lower than for the longitudinal one, hence T(k)<L(k). In some directions the curves are similar to the result of the one-atomic linear chain model. The local maxima in some of the curves are due to interactions between atoms further apart than the nearest neighbours. In some directions the two transverse branches are degenerate.
Our next example is the dispersion relation in certain symmetry directions for the ionic compound NaCl. Here we have a basis of two atoms, and hence there are three acoustic and three optical branches. The transverse branches are seen to be degenerate. Note the similarity to the theoretical result for the two-atomic linear chain, especially in the [111] direction. This is because in this direction the atomic planes are composed of either only Na+ ions or solely Cl- ions. The atomic planes move in phase and so the problem is essentially onedimensional in this direction.
Diamond also has a basis of two atoms in each primitive cell, although all atoms are equal. It also has optic and transverse branches and considerably more complicated dispersion relations.
As noted above, infrared radiation can be used to excite transversal optical (TO) phonons close to k=0 in ionic solids. The ions in the TO branch vibrate in opposite directions and the resulting dipole moment can couple to electromagnetic radiation. Only transverse vibrations can be excited since light is a transverse wave motion. The wavelengths of infrared radiation are of the order of 10 m, which is several orders of magnitude larger than the lattice constant, a. Hence the wavevector of the photon is very much less than /a in magnitude and is close to k=0 in the Brillouin zone.
We have an incoming photon (k) that is absorbed and gives rise to a phonon of wavevector q. The k-conservation rule becomes k=q. The absorption is centred around the TO frequency and is described by the Lorentz model: M d2u/dt2 + du/dt + Ku = -eE. Here M is the reduced mass, is the damping, K is the force constant and E is the local electric field. The infrared absorption by optical phonons has practical consequences. By Kirchhoffs law the absorption at each wavelength is equal to the thermal emittance of heat radiation at that wavelength. In certain applications a high emittance is desired, often in combination with a low absorption of solar radiation. a) The equilibrium temperature of a spacecraft needs to be quite close to room temperature due to the needs of the electronics. Surface coatings with high emittance can help regulate the temperature of the spacecraft. b) On earth, surfaces can be cooled to 10-20 degrees below ambient temperature in dry climates, by utilizing a material with tailored infrared absorption (emission). This is sometimes a nuisance (the frost on your car window even if the temperature has been a few degrees above zero) but might also be used in certain places for simple cool storage, in building design and for promoting dew condensation. b) Inelastic phonon scattering. To analyse inelastic scattering we need to consider the conservation of energy and k-vector. Energy conservation gives = o q. And k-conservation:
k = k o q + G.
The index 0 refers to the incoming photon and q to the created or absorbed phonon. In this technique the scattering of visible laser light is commonly used. Scattering can occur from acoustic phonons (Brillouin scattering) or optical phonons (Raman scattering), but for photon scattering q is very small.
c) Inelastic neutron scattering.
The energy of thermal neutrons is of the same order as that of the phonons. This means that the energy losses during scattering are easily measurable. Measurements of the energy change (from wavelength change) and the scattering direction, which gives k, are used to experimentally determine the dispersion relation. Unlike the photon based techniques we are not restricted to k0 here. A schematic picture of a triple-axis neutron spectrometer is shown below (adapted from Myers). A monochromatic beam is obtained by Bragg diffraction from a crystal, and the beam impinges on a sample. The scattered beams are analysed in another crystal using Bragg diffraction to determine the wavelengths. The whole apparatus can be rotated around M, also the analyzer can be rotated around S and the detector around A. This is a method to get a very detailed determination of the dispersion relation .
STUDY QUESTIONS
1. Estimate the atomic vibration frequency of an atom with mass 60 u, if the force constant is 10 N/m. 2. For the same atom, estimate the amplitude of the vibrations at room temperature and compare with a typical lattice constant. 3. A one-dimensional material consists of atoms with mass 35 u. The force constant between the atoms is 50 N/m. Calculate the velocity of sound and the highest vibration frequency. 4. In which materials can an electromagnetic wave excite the lattice vibrations? Which branch of modes can be excited? 5. Absorption of electromagnetic radiation in NaCl has a peak at a wavelength of 55 m. Calculate the frequency of the excited phonon and estimate its wave-vector, k. Calculate also the force constant for NaCl. 6. Give expressions for the frequencies of the acoustic and optical modes at the boundary of the first Brillouin zone, within a two-atomic chain model with same force constants. 7. How many lattice vibration modes are there in each branch in the Brillouin zone, if a solid contains (a) 100 and (b) 1000000 primitive cells? 8. Consider a solid with a basis consisting of four atoms. How many branches of acoustic and of optical vibration modes are there? 9. Draw schematically the direction of propagation and the vibrations of longitudinal and transverse modes in a three-dimensional structure. 10. Use periodic boundary conditions to determine the allowed k-values in one dimension. 11. Sketch schematically the dispersion relations of lattice vibrations for (a) a one-atomic linear chain and (b) a two-atomic linear chain. Indicate in the figures how one can determine the velocity of sound by a geometrical construct.