PHYS 7440 Assignment 1 Due Jan. 22, 2003 Reading: 1. For those interested in the general history of solid state theory, see: Out of the Crystal Maze: Chapters from the History of Solid-State Physics, by Lillian Hoddeson, Ernest Braun, Jürgen Teichmann and Spencer Weart, Oxford University Press, Inc. (1992). 2. Read Chapter 6 in Marder. 3. Related material on the Sommerfeld free-electron model is in most of the reserve books e.g., Ashcroft and Mermin, Chap. 1, 2. Comments: Many of the most successful models of phenomena in solids are based upon 'educated guesses' for the form of the low energy many-body wavefunctions as particular expansions in terms of single particle wavefunctions. Upon selection of such an approximate many-body wavefunction, a variational calculation (more on this later) is applied to determine a 'self-consistent field' Hamiltonian which yields approximate eigen energies useful in determination of thermal and transport properties. Examples include the Hartree-Fock Equation and associated determinental wavefunctions, the BCS 'gapequation' and trial wavefunction for superconductivity and the Laughlin wavefunction for the Fractional Quantum Hall Effect. The first problem involves looking at simple product wavefunctions. 1. Marder 6.1 (This notation means Problem 1 in Marder's Chapter 6 problems). 2. Marder 6.2 Use whatever program you like (Mathematica is nice) to plot the energy versus electron number. 3. Marder 6.4 Plot the density of states functions. 4. Marder 6.5 In this problem, determine i) the wavefunctions, ii) eigen energies, iii)the conditions on wavevector, k, iv)the nature of k-space, v)the appropriate k-space density and vi) the average energy per particle at zero temperature. What replaces the Fermi sphere in this case? Assume that the box is deformed so that the z-dimension is very short, but y=x and retaining xyz=V. Sketch how the points in k-space are distributed and show how these states are populated in the ground state. Problem Set 1 Solutions Phys. 4340 Spring 1995 This problem set emphasizes some of the simple pictures which we will use for understanding the structures of real solids. The first two problems consider the simple picture of a solid being composed of spherical atoms of a give radius. The third problem deals with a specific pair potential and shows how you can take a simple form and adjust the parameters to reproduce some of the measured properties of the solid. Rosenberg 1.1. The density of b.c.c. iron is 7900 kg-m-3 , and its atomic weight is 56. Calculate the side of the cubic unit cell and the interatomic spacing. The name of the game here is to figure out what the b.c.c. structure is and then determine how large the unit cell must be to get the correct number of iron atoms per unit volume to match the measured density. I’ll treat this problem in three steps: 1. Determine the weight of an iron atom. 2. Determine the number of atoms in the b.c.c. cubic unit cell. 3. Determine the size of the unit cell necessary to get the correct density. Weight of an iron atom. The atomic weight is 56. The atomic weight essentially tells us the number of protons plus neutrons and these weight about the same amount, namely 1.67x10 -27 kg. Therefore, the atomic weight of an iron atom is: Weight 56 1.67 10 27 kg 9.4 10 26 kg Number of atoms per bcc unit cell. bcc unit cell from Rosenberg. Rosenberg shows the cubic bcc unit cell in Fig. 1.2 (a) on page 3 of the text. I’ve reproduced in here. As you can see, the standard cubic bcc unit cell has atoms located at the corners of the cube (there are eight corner locations) and one atom at the center of the cube where the body diagonals cross. Remember that the crystal is constructed by packing a huge number of these units together. Therefore, you can see that, although the central atom is entirely within a single cell, the corner atoms are partially inside several cells which meet at the corner. In fact, each corner atom is shared by eight cells. Thus, there is 1/8 of each corner atom inside each cell. Since there are eight corners, the total contribution is one atom for the corners and one atom at the center or two atoms per bcc cubic unit cell. Density of bcc iron. The cubic cell has a side length of L, so that the volume is L3. Therefore, since there are two atoms per cell, the mass density is: 29.4 1026 kg L3 Since we are told that 7900 kg - m-3 , then we can solve for L: 29.4 1026 kg L 13 0.288 nm OK, so the result is that the cell is a bit under 3 Angstroms on a side. If you look at the structure, you can see that for any given atom, the nearest neighbor atom is reached by traveling half-way along the body diagonal of the bcc cell. The atoms located along the edges are somewhat further away. Therefore, the nearest neighbor interatomic spacing is just half the length of the body diagonal. We can calculate the body diagonal length via Pythagorus: 1 1 Body diag. length = 2 2 1 2 Face diag. length Cubic side length 2 2 2Cubic side length Cubic side length 2 2 3 Cubic side length 2 0.249 nm Now here’s the real point as I see it: The atomic number of iron can be determined through mass spectrometry in the gas phase i.e., it is a property of the single atom which can be determined independently of any solid state property. Similarly, the density of the iron is a measurable property. Therefore, we can predict what the atomic spacing must be like depending upon which crystal structure the solid takes on. In the case of iron, it actually does turn out to prefer the bcc structure. Furthermore, x-ray diffraction data confirms that the bcc unit cell in iron has a cubic cell width of 0.287 nm. Thus, the atomic properties allow simple predictions about the solid state. Rosenberg 1.2. The density of fcc.gold is 19,300 kg-m-3 , and its atomic weight is 197. Calculate the separation between the close-packed planes. On the assumption that the atoms may be thought of as spheres which just touch one another, estimate the atomic radius. This problem is similar to the last one, but now requires that you spend time thinking about the fcc structure. As an aside, the bcc, fcc, and the hexagonal close-packed structures are the crystal structures taken on by the vast majority of elemental solids. Further, when you generalize the structures so that different atoms can reside around the lattice points, you find that these structures are chosen by the majority of known materials. As Rosenberg explains on pages 3-6, the fcc and the hcp structures can both be constructed by producing layers of close-packed spheres. You have all probably seen this type of packing in fruit displays or in piles of cannonballs. Let’s proceed as we did in 1.1, and use the standard cubic fcc unit cell, determine the number of atoms in the cell, calculate the cubic side to get the correct mass density, and then figure out the orientation of the close-packed planes (along the cubic body diagonal) to get the spacing. Again, Rosenberg has provided the pictures, in this case, Fig. 1.5 on page 5. For the fcc structure, there are eight atoms in the cube corners as for the bcc case, so we have one atom total from the corner positions. Next, we have six atoms at the face centers. In this case, each face atom is shared between only two unit cells, so each contributes half an atom to the cubic fcc cell. Thus, there are four atoms in the fcc cubic unit cell, one from the corners and three from the faces. The gold atomic mass is 3.3x10-25 kg, so the cube has a size of: 43.3 10 25 kg L -3 19,300 kg - m 13 0.409 nm From here on, it’s ‘just’ solid geometry. Now, Fig. 1.5 (c) shows that the close-packed planes are perpendicular to the body diagonal of the cubic fcc cell. Each close-packed plane has the atoms arranged at the corners of equilateral triangles. Here’s a picture of several of the equilaterals within the cubic fcc cell: Further, the next plane up the stack is an identical set of equilateral triangles, but shifted over so that the atoms lie above the centers of the triangles in the plane below. Therefore, the basic unit is a threesided pyramid constructed of equilateral triangles. I’ve pulled out this basic pyramid in the picture, so that you can see it. In any case, if you look at the way the equilaterals are packed into the cube, you can convince yourself that the side of one of these triangles is just the same length as half of the face diagonal. Therefore, any edge of the pyramid is of length 2L 2 L 2 . What we want is the height of the top atom of the pyrimid above the floor. Again, you can get the length by construction of a right triangle. In this case, things are straight-forward because all the triangles we care about have interior angles of 300. We find that the height is found by calculating the base length of a right triangle with a side of length L 2 6 and a hypotenuse of L 3 2 2 . Then, Spacing between planes = L 3 0.236 nm If the atoms are spheres, then close packing means that they just touch at the middle of the side of each equilateral triangle. Thus, Atomic radius = L 2 2 0.145 nm Rosenberg 1.7. 9 The potential energy of a pair of atoms in a crystal is of the form A r B r when their separation is r. The equilibrium separation is 0.28 nm and the dissociation energy is 8x10-19J. Calculate A and B. Find the effective modulus of elasiticity for the pair of atoms and calculate the force which would be necessary in order to reduce their spacing by 5%. Here’re the main points which you should get from this problem: First, once you have a few pieces of experimental information such as the atomic spacing and the dissociation energy (the energy necessary to separate the atoms from each other) then it is possible to take a hypothetical pair potential and adjust its parameters so that it reproduces the experiment. Second, once you have fitted the pair potential, it is possible to make predictions of other properties. To the extent that your predictions work out well, you can assume that the potential is a good description of the bonding strength of the atoms. We will see that the Lennard-Jones 6-12 potential turns out to do an excellent job of describing the noble gas solids. The 1-9 potential which Rosenberg has you think about does a pretty good job of describing solids like NaCl where bonding is approximately due to Coulombic attraction between ions. Fitting A and B to data. First, let’s look at the potential and see what it is that we are trying to do. A sketch of the function looks like this: V 1/ r 9 ro r 1/ r -Vo Sketch of the 1-9 pair potential. This particular pair potential is designed to describe the potential energy between ions. The attractive potential is of the Coulombic form. The strongly repulsive short range piece is supposed to mimic the repulsion between the electron clouds due to the Pauli exclusion principle. Between the two terms, we see that the function develops a potential well. The potential well causes the two atoms which interact to reach a ground state in which they are separated by the distance r0. Furthermore, it takes and energy V0 to pull the atoms apart. We fit the function to the experimental data by adjusting A and B so that we get the correct potential well depth and the correct position for the well minimum. First, we generate an equation relating A and B to r0 by setting the derivative of the potential to zero (true at the minimum point): dV dr 9 r r 0 A B 2 0 10 r0 r0 Therefore, we have a relation between A and B, or we can think of it as an equation for the equilibrium position. A r0 9 B 1/ 8 or 9A Br08 Since we have two unknowns, we need a second relation to finally solve the problem. Of course, the idea is that: V r0 V0 This equation yields a second relation: A B V0 r09 r0 or B A V0 r0 r08 We can combine these results to find that: V0 r09 8 9 B V0 r0 8 A We can immediately plug in the experimental values for the equilibrium distance and the dissociation energy to find: A 1.06 10 105 J - m 9 B 2.52 10 28 J - m A better way to look at it and to write it. OK, so we found the values for A and B. We can press on from here to calculate further properties, but these numbers don’t really help in understanding the physics. There is a much better way to write the pair potential, which I believe helps to clean things up. What you do is take the formal relations we found above between A and B and the equilibrium position and dissociation energy. Plug those back into the pair potential and do some minimal algebra and you get a new form for the potential: V r V0 r0 9 r0 9 r 8 r It’s the same potential, still in terms of two parameters, but now we get some physical insight into what the parameters are! Of course, the equilibrium position, r0, now appears as a natural length scale for the potential, while the dissociation energy, V0, appears as a natural energy scale. Much nicer! Elastic forces. Now we have a form form the potential, so we can begin predicting other properties. As we get into the case of solids, we’ll find that elastic properties, sound wave frequencies and velocities in particular, are among the most important of these properties which are easiest to get out. Here, we only consider a pair of atoms, so Rosenberg asks for the ‘elastic’ properties of the pair. The essential concept here is that for small changes in separation (compared with the size of the equilibrium spacing) the potential looks approximately parabolic i.e., the potential looks like a harmonic oscillator potential. Defining the deviation from equilibrium spacing as: r r r0 what I’m saying is that it should be possible to write the potential as: V r r0 V0 1 2 kr 2 where k is the modulus of elasticity. The ‘better’ way to write the potential is also a better way to write it if you want to derive this type of small oscillation approximation. To determine the approximate potential, let’s do a Taylor expansion to second order around the equilibrium position: dV V r r0 V r0 dr 1 d 2V r 2 dr2 r r0 r2 Or 3 r r0 Evaluating the derivatives gives us: dV dr r r 0 V0 r09 r 9 10 9 02 0 8 r0 r0 and d 2V dr2 r r 0 V0 r09 r 90 11 18 03 8 r0 r0 9 V0 r02 The first result is the standard observation that at the equilibrium spacing, there is no net force on the system (so the first derivative of the potential is zero). The second result says that, aside from the numerical factor of 9, the ‘natural’ parameters can only be combined in one way to yield a constant with the units of the elastic modulus. Anyway, the potential is then: 1 V V r r0 V0 9 02 r 2 2 r0 so that the elastic modulus is: k 9 V0 r02 91.8 J/m 2 To get the force necessary to decrease their spacing by 5%, let’s continue to use our small deviation approximation for the potential. Then, the force is determined by the negative first differential of the potential: F V k r To reduce the spacing by 5% means that r 0.05 0.28 nm 0.014 nm i.e., that the required force is 1.29x10-9 N. As an aside, Rosenberg’s elastisity modulus is defined just like our k, except she has pulled in the factor of the equilibrium spacing so that she has a new modulus, , such that: F % length change or kr0 2.57 10 8 N Plotting the potential. This part is simple once you have written the potential in the way I have above. As I say, the natural scales of length and energy are the equilibrium spacing and dissociation energy. Defining a dimensionless distance and potential, respectively as, x r r0 and v V V0 , then the potential function to plot is: v x 1 1 1 9 9 8 x x Here’s the Mathematica code necessary to plot it: In[1]:= v[x_]:=(1/8)((1/x)^9 -9/x) In[4]:= Plot[v[x],{x,0.1,6}, PlotRange -> {-2,6}, Frame -> True, FrameLabel -> {"r/r0","V/V0"}] Out[4]= -Graphics-