Analysis of Molecular Vibrations

Marc Alomar April 25, 2013

Abstract The goal of this exercise is to show how nite group theory can simplify the analysis of molecular vibrations. This problem was rst considered by E. Wigner1 , and most of this analysis is based on his work. In particular, I analyze the vibrations of a triangular molecule, rst with the classical Lagrangian formalism, and then with the formalism of group theory. Both approaches lead to the same results with regard to the normal modes.

Lagrangian Formalism

In this section we will study the planar vibrations of a triangular molecule. We can describe this system by a set of three masses m, sitting on the vertices of an equilateral triangle and connected by springs of constant k . Following a classical dynamics approach, we will use the Lagrangian formalism to nd the frequencies and normal modes of this system.

2
y
1

x1

Nachr. Ges. Wiss. Gttingen 133 (1930). An English translation can be found in the book Applied Group Theory, by Arthur P. Cracknell.

The equilibrium positions in terms of the center of mass reference frame are 8 1 p 1 = 1 e e > 2 x <r 2 3 y 1 r 2 = p e 3 y > : 1 1 r 3 = 2 e x 2p e 3 y 1

It is convenient to describe the departure from equilibrium in terms of the displacement vector ~ ri = (xi , yi ). These coordinates set the position of the masses with respect to its equilibrium position, as described in the gure. Therefore, we can write the position vector of each mass as ~ r=r i + ~ ri . A general expression for the potential energy is 1 V (xi , yi ) = k (|~ r1 ~ r2 | 1)2 + (|~ r2 ~ r3 | 1)2 + (|~ r1 ~ r3 | 1)2 2 If we limit our analysis to small amplitude vibrations, we can write V in terms of a Taylor expansion about the equilibrium point. The zero and rst order will be zero, and we are left with the second order term, V = 1 ~ rT H ~ r 2

where ~ r T = (x1 y1 x2 y2 x3 y3 ) is the displacement vector and H is the Hessian matrix evaluated at the equilibrium point ~ r = 0. The Hessian matrix is given by 0 1 p p 5 3 1 3 1 0 p B p3 3 3 3 0 0 C B C p p B C 1B 1 3 2 0 1 3 C p p H= B C 3 3 0 3 3 p3 C 4B p B C @ A 1 0 1 3 3 3 p p 0 0 3 3 3 3 We can nd the normal modes H~ u = !2 ~ u. The solutions are 8 ~ u1 > >!1 = 3 > > > !2 = 1.5 ~ u2 > > > < ! = 1. 5 ~ u3 3 > !4 = 0 ~ u4 > > > > > !5 = 0 ~ u5 > > : !6 = 0 ~ u6 and frequencies if we solve the eigenvalue problem

p p = 3, 1, 0, 2, 3, 1 p p = 0. 5 ( 3 , 1 , 3, 1, 0, 2) p p = 0. 5 ( 1 , 3, 1, 3, 2, 0) p = (0, 1, 3, 0, 0, 1) = (1, 0, 1, 0, 1, 0) p = (0, 2, 3, 1, 0, 0)

The zero frequency modes represent the translational and rotational modes. Thus, we have found that this system has one non-degenerate and two degenerate normal modes.

Group theory analysis

We now proceed to the analysis of the same problem using the tools of group theory. For a given molecule, we rst identify its symmetry group G, i.e. the group of transformations that leave the equilibrium positions invariant. Then, we study how these transformations modify the displacement vector ~ r. It turns out that when we transform a normal mode under G, its frequency doesnt change, because the relative positions of the molecule are the same. In fact, the normal modes form an irreducible representation of the group. We only have to identify the completely reducible representation D that transforms the displacement vectors, and decompose it. Each irreducible representation corresponds to a normal mode with a characteristic frequency. For each normal mode, the dimension of the irreducible representation is equal to the number of linearly independent normal modes. Lets return to our example. The rst step is to identify the point group that describes the symmetry of the triangle. There are six symmetry operations that leave 0 the equilibrium positions invariant: rotations through 120 and 240 ( C3 and C3 ), three reections i in a plane that contains the atom i and bisects the opposite edge, and the identity operation E . In fact, each of these three dierent kinds of operations form a conjugacy class. An important result of group theory states that the number of conjugacy classes is equal to the number of irreducible representations. This group has three irreducible representations: two one-dimensional, A1 and A2 , and one twodimensional, E1 . The character table is shown below. A1 A2 E1 D E 1 1 2 6 2 C3 1 1 -1 0 3 1 -1 0 0 z Rz x, y Rx , R y x2 + y 2 , z 2 x2 y 2 , xy xz, yz

Table 1: Character table. The normal modes transform according to the irreducible representations of the group. Non-degenerate modes correspond to one-dimensional representations, whereas degenerate modes correspond to representations of dimension greater than 1. Our job is to nd a representation that describes the transformations of the displacement vectors. This representation will be completely reducible, and our task is to decompose it. Once we know the number of times that each irreducible representation appears, we will know the number of normal modes and its degeneracy. It turns out that its not necessary to nd the exact expression of the six 6x6 matrices; we only need their characters to decompose D in terms of the irreducible 3

terms. It is also well known that symmetry elements that belong to the same class have the same character, and then our problem is reduced to obtain the character of three operations: rotations, reections, and the identity. The character of the identity matrix is 6, and the rotation matrices have character 0, because none of the atoms remains at the same position (i.e., the diagonal only contains zeros). The reection matrices also have character 0. We can verify this result with the matrix 2 , 0 1 0 0 0 0 1 0 B 0 0 0 0 0 1 C B C B 0 0 C 1 0 0 0 B C 2 = B C 0 0 0 1 0 0 B C @ 1 0 0 0 0 0 A 0 1 0 0 0 0 The second atom remains at the same position, but the reection along the plane yz changes x0 ! x and y 0 ! y . The number of times mD a that each irreducible representation Da appears in D is obtained using the orthogonality relation, mD a = 1 X N g 2G
Da ( g ) D (g )

where N is the number of elements of the group and D (g ) is the character of the element g in the representation D. In this case we nd mA1 = 1, mA2 = 1 and mE1 = 2.Thus, the decomposition of D is D = A1 + A2 + 2 E1 However, if we only want the vibrational modes we have to substract the translational and rotational modes. Their irreducible representation can be identied using table 1. Translations along the x and y axis are described by E1 . This representation has dimension 2, so we only have to substract one E1 . Rotations correspond to A2 , and we are left with Dvib = A1 + E1 We conclude that the system has three normal vibrational modes: one non-degenerate, u1 , and two degenerate, u2 and u3 . However, we also know their symmetry properties. This information will allow us to obtain the eigenvectors. The eigenvector of the non-degenerate mode is very easy to obtain. In one-dimensional representations, the characters are equal to the representations. In this case, A1 is the trivial representation (D(g ) = 1, 8 g ). Thus, this normal mode must be symmetric 4

under any symmetry transformation, i.e. DA1 (g ) = , 8 g . The only vector that satises this relation is p p 1 ~ u1 = 1, 3, 0, 2, 1, 3 2 This mode is usually known as the breathing mode. The degenerate eigenvectors are more dicult to obtain. However, the matrix representation of E1 tells us how these modes transform. Most of the matrices are nondiagonal, but the matrix for 2 is particularly easy. We then know that, under the reection 2 , the normal modes u2 and u3 transform as u2 u2 DE 1 ( 2 ) = u3 u3 The mode u2 is antisymmetric under 2 , whereas u3 is symmetric. Therefore, the motion of particle 2 must be along the x axis in u2 , and along the y axis in u3 . The motion of particles 1 and 3 must follow precise symmetry relations: (x3 , y3 ) = (x1 , y1 ) in u2 , and (x3 , y3 ) = ( x1 , y1 ) in u3 . We can make a reasonable guess using these relations and the fact that both vectors are linearly independent, ( p p ~ u2 = 1 1 , 3 , 2 , 0 , 1 , 3 2 p p 1 ~ u3 = 2 3, 1, 0, 2, 3, 1 It turns out that these vectors also transform correctly under rotations, and we conclude that they are the normal modes of E1 .

u1

u2

u3