Equation of Motion and Dynamical Matrix

Here, we recall the equations of motion (EoM) and the derivation of the dynam-
ical matrix for the nuclear motion in a crystal. Let us assume a perfect crystal
with periodic boundary conditions at 0 Kelvin. We dene R as the position
vectors to the -th atom within an arbitrarily chosen 'zero' unit cell, and the
position vector to the n-th unit cell is denoted by Rn . Hence the equilibrium
position of atom in unit cell n can be written as Rn = Rn + R . If we denote
the Cartesian components of the displacement vector of atom in unit cell n
with sni , where i takes the three values x, y, z , we can write for the kinetic
energy:
N X r X 3  2
X M dsn i (t)
T = . (1)
n=1 =1 i=1
2 dt
Here, M is the mass of atom , and N is the number of unit cells, r the number
of atoms per unit cell, and i stands for the three Cartesian coordinates x, y, z .
Hence the total number of degrees of freedom is 3rN . The potential energy W
 as a function of the nuclei coordinates x  can be expanded into a Taylor series
in the atomic coordinates.
X  W (x 
W (sni ) = W (Rn ) + sni
ni
sni x=Rn
2 W (x
 
1XX
+ sni sn0 0 i0 + O(s3 ) (2)
2 ni 0 0 0 sni sn0 0 i0 x=Rn
ni

The linear terms in s vanish, on account of our assumption that we expand

around the equilibrium conguration (vanishing forces), and we can introduce
the force constants
2 W (x
 
n0 0 i0
ni = (3)
sni sn0 0 i0 x=Rn
Note that we neglect higher terms than the harmonic ones. In order to obtain the
equations of motion we dene the Lagrange function for our problem = T W ,
and recall the EulerLagrange equations
d
. = 0. (4)
dt sni sni
Inserting the expressions for the kinetic energy T and the potential expansion
W into the Euler-Lagrange equation and taking account of the symmetry of the

1
 0 0 0
force constants nni i = ni
n0 0 i0 leads to:

d2 sni X 0 00
M = nni i sn0 0 i0 . (5)
dt2 n0 0 i0

With the ansatz

1
sni (t) = uni eit (6)
M
we obtain 0 0 0
X n i
2
uni = ni un0 0 i0 (7)
n0 0 i0
M M0
which is an eigenvalue equation for the 3rN normal frequencies . On account of
0 0 0
the translational symmetry, the force constants nni i  which can be identied
as the ith component of the force on atom in unit cell n, when atom 0
in unit cell n0 is displaced a unit distance in i0 direction  only depend on the
0 0 0 0 0
dierence n n0 , i.e. nni i = ii (n n0 ). In order to take account of the
translational symmetry we make the ansatz:

uni = ci eiqRn (8)

and we nd
0 0
X
i
Di (q)c0 i0 = 2 (q)ci (9)
0 i0

where we have introduced the dynamical matrix as

" #
0 0
X 0 i0 (n)
i
Di (q) = i eiqRn (10)
n
M M
0

Thus, we have arrived at an eigenvalue equation (9) which is only 3rdimensional,

but now the eigenfrequencies are functions of q which is a wave vector inside
the rst Brillouin zone.