529 - pr 01 - a more detailed treatment of the phonon dispersion in metals: in deriving the Bohm-Staver

relation [26.8], we regarded the ions as point-particles, interacting only through Coulomb forces,

2 2
2
0
/
/ (2 ) 1
[26.8]
3 / 2 /
F
e
F F
Zm Zm k m Zm m
c v Z
k M M M M mk e mk
e c
t t
= = = =
p p
(1.1)

A more-realistic model would take the ions as extended distributions of charge, and allow for the
impenetrability of the ion-cores by an effective ion/ion interaction in addition to the Coulomb interaction. Since
the core/core repulsion is short-ranged, it leads to no difficulties in the usual treatment of lattice-vibrations and
can be described by a dynamical matrix
c
D in the manner described in chapter 22. We can therefore treat
lattice vibrations in metals by the methods of chapter 22, provided that we take the full dynamical matrix as
modified by the Coulomb interaction,

| | | |
0 1
; screening effects ; core/core repulsion effects
c c coul coul c
= + = + = = D D D D D D D D (1.2)

Note problem-statement slightly-rewritten: Take the i
th
ion to be at position ( ) + R u R to have charge-
distribution ( ( )) = r R u R , so that the electrostatic force on this i
th
ion is given by
3
( ) ( ( )d r
}
E r r R u R , where ( ) E r is the electric field reduced by electronic screening
1
due to all other ions
(whose charge density is ( ( )) ( )
other other

'=
' ' = =

R R
r R u R r .

(a) expand this additional electrostatic interaction to linear order in the ionic displacements u .

Let this be ion-i. The i
th
interaction energy is
i
E i i
dU d = - F u , where the force
i
F is due to all other ions,

3 3
( ) ( ( ))
i i i
d dq d r d r = = = =
} } } }
F F E E E r r R u R (1.3)

In which: the electric field ( ) = E E r is due to all other ions, each of charge Ze + , but this density is
encapsulated by ( ) ( ( ))
other

'=
' ' =

R R
r r R u R mentioned above. Using Coulombs Law,

3 3
3 3 3
( ) ( ( )) d dq d r d r
'=
' ' '
' ' ' ' ' ' ' = = = =
' ' '

} } } }
R R
r r r r r r
E r E r R u R
r r r r r r
(1.4)

In (1.3) and (1.4), the expansion in ( ) u R and ( ) ' u R appears as,

( ( )) ( ) ( ) ( ) ( );
( ) ( ) ( ) ( );
other


2
2
= = V - +
' ' ' ' ' = V - +
r R u R r R r R u R u
r R r R u R u
O
O
(1.5)

Thus, the interaction on the i
th
ion appears as,

1
The screening theory of chapter 17 was based on the assumption that the external potential
ext ind
| | | = + was a weak perturbation
on the electron-gas. Since this is not the case for the potential of the ions,
ions ext
| | > , we have ( ) (1/ ) ( )
ion
| c | = q q

(note q-
space). One can find a linear relation between the deviations of | and
ion
| from their equilibrium-values. One must take the system
perturbed by the ions to be not a free electron gas, but rather a gas of e- in the presence of the full equilibrium periodic potential (with
eigen-energies given by a transcendental relation). The formula describing the screening is cumbersome compared to the simple
Thomas-Fermi or Lindhard forms, but worthy of pursuit. These more-cumbersome models of screening are said to contain band
structure physics (e.g., band structure effects is the formal lingo). We ignore these in this problem; see (1.17).

( )( )
max
3 3
3
3 3
3
0
( ( )) ( ( ))
( ) ( ) ( ) ( ) ( ) ( )
i
E i i i
i
U d d r d r d
r r
u u d r d r du


' ' '
'=
'=
| |
'
' ' ' ' = - = | -
|
'
\ .
| |
'
' ' ' ' ' ' ~ V V |
|
'
\ .

} } } }

} } }
R R
u
R R
r r
F u r R u R r R u R u
r r
r R r R R r R r R R
r r
(1.6)

Notice the switch to index-notation here in (1.6); it makes computing the derivatives defining the dynamical
matrix much easier.

Assuming the electronic screening is described by the static dielectric constant
2
( ) c q , show that the dynamical
matrix appearing in
2
[22.57] ( ) Me c c = D k must be taken to have matrix elements,

( )
2
2
4 ( )
( ) ( ) ( ) ( ) ( ) ; ( ) ;
( )
C
nq q
D D V V V V
q q
v
v v v v v v
t
c
=
= + + + =

K u
q
k k k K k K q (1.7)

Formula for dynamical matrix: The dynamical matrix due to this interaction (1.6) is computed as,

( )
( )
max
2 2
2
3 3
3
0
3
3
( ) | |
( ) ( )
( ) ( ) ( )
( ) ( ) ( )
i i
i i E E
i
i
U U
D D
u u u u
u
r r
d r d r du
u u
u
r r
D d r d
v v
v v
v
v v


' ' '
' '
= =
'=
'=
c c
' = = =
' ' c c c c
| | ( ' ' ' ' ' V
' c
|
( ' =
| ' c c ( ' V
\ .
'

'

} } }

} }
u 0 u 0
u
R R
R R
R R
R R
r R r R R
r r r R r R R
r r
D
max
3
0
; ( , , , , );
i
r du
v v
'
| |
' ' ' | =
|
\ .
}
u
r r R R u D D
(1.8)

Derivatives: Ignoring all else, and focusing just on the bracketed term, and abbreviating ( ) c ' ' ' V = r R and
the other one as ( ) c V = r R , we have the mixed 2
nd
derivative of
v
D appearing as,

( )
( )
( )( )
2 2
( ) ( ) ( )
( ) ( ) ( )
( ) ( )
u
c u c u
u u u u
u
c c u c c c c
u
v
v v
v
v v v

o o o
( ' ' ' ' ' V
c c
( ' ' ( = =

' ' c c c c (
V

c
' ' ' ' ' ( = = = = V V

c
r R r R R
r R r R R
r R r R
D
(1.9)

Restoring the terms our notation concealed, we put (1.9) into (1.8)

max
3 3
3
0
( ) ( ) ( );
i i
i
r r
D d r d r du D
v
v v

' '
'=
'
| |
'
' ' ' ' = V V | =
|
'
\ .

} } }
u
R R
r R r R R R
r r
(1.10)

Inverse-space: Fourier transforming using the transform
( )
( ) ( )e
' -
' =

i k R R
R
D k D R R , the (1.10) then is,

2
More precisely: to capture the physics of the presence of a normal mode of frequency e , we should use ( , ) ( ) c c e c = = q q .
However, we often have
D
e e << , so ( , ) ( , 0) ( ) c c e c c = = q q q (c.f., [17.60]. Note, also, that this is an expression of the
adiabatic approximation).

max
( )
( ) 3 3
3
;
0
3 3
3
;
( ) ( )
( ) ( )
( ) ( )
i i i
i
i
D D D e
r r
e d r d r du
r r
D e e d r d r
v v v
v
v
v


' -
' ' ' '
' '
-
'=
' - -
'=
' = =
| |
'
' ' ' = V V |
|
'
\ .
|
( '
' ' ' = V V
(

'

\

} } }

} }
i k R R
R
u
i k R R
R R R
i k R i k R
R R R
k R R
r R r R
r r
r R r R
r r
max
0
i
du
'
|
|
|
.
}
u
(1.11)

b) show that if the electronic screening is not taken into account ( 1 c = ), then (1.7) predicts, in the long-
wavelength limit, a longitudinal normal mode at the ionic plasma frequency, [26.2]. we have,

2 2
2 2 2
plasma-frequency
4 ( ) 4
[26.2] ; ;
from chapter 1
i e
p p p
n Ze n Ze K Zm
M M M m
t t
e e

(
O = = = = =
(

(1.12)

0
th
-order expansion: In the long-wavelength limit,
2
( ) ( ) ( ) ( ) V V V
v v v
+ =V - +
k
K k K K k k O . However,
V
v
V
k
is incredibly cumbersome, as we can see here,

( )
( )
| |
2 2
2 2
2
( ) ( ) ( ) ( )
4 a mess
( )
nk k k k nk k k k
V
k k
v v
v
c c
t
c
V V
V = =
k k
k
k k
(1.13)

Thus, lets truncate the series at 0
th
order
3
. We also say that for long wavelengths, there is no core/core
repulsion effects, meaning,

1
0
( ) ( ) ( ) 0; lim ( ) ( ) 0;
C C
V V D D
v v v v

+ = = =
k
K k K k k 0 O (1.14)

With (1.14), we are left with the dynamical matrix (1.7) appearing as,

2 2
2
2 2
0 0
4 4 ( ) ( ) ( ) ( )
( ) lim ( ) lim ;
( ) (0)
k k k k n n
D V M
k k k
v v
v v v
t t
e c c
c c

= = =
k k
k k 0 0
0 k (1.15)

Note that
v v
c c o = , that (0) 1 c = , and the definition ( ) Ze = 0 ; this can make (1.15) into,

2 2 2
2 2 2
2
2 2 2
4 ( ) ( ) ( ) ( ) ( )
( )
4 4 4
(0) (0) (0)
p
k k n n n Ze k k k n Z e
M
k k k M M
v
v v
v
t
e o t e t t
c c c
= = = = = O
0 0 0 0 0
0
(1.16)

(c) show that when screening is taken into account via the Thomas-Fermi dielectric function [26.4] all phonon-
frequencies vanish linearly with k at long wavelengths, though the dispersion relation is not of the simple
Bohm-Staver form (1.1). writing the Thomas-Fermi model for the dielectric-function, and estimating the
derivative
4

0
( )
F
n g

c c ~ that appears, we get,

2 2 0
0 0 0 2 2
[26.4] ( ) 1 ( ) ; [17.51] 4 ; ( ) ;
F
F F
k mk
k e n T T n g
k

c t c
t
= + = c << c ~ = k (1.17)

Using (1.17) the dispersion relation (1.16), in which
2 1/ 2 1 3
(1 ) ( ) x x x

+ = + O , appear as,

3
Further justification for 0
th
order expansion: for electron-jellium oscillation of wavelength in a box of side-length L, a plasma
oscillation is on the order of ~ 2
plasma
L . That is huge compared to the incredibly-short Fermi wavelength.
4
Some numbers:
2
0
2 2 2 2
0 0 0
0
2 2 2 2 2/3 1
16 4 4 1 1
0 0 0
/ 3
4 4 ( ) ( ) 0.815 2.95 ;
s s F
F F
s F
k r r mk
me
F k k a a a
r a k
k e n e k k A
t t
t t
t t

= c = = = = = =

( )
0 0
2 2 2
3 3/ 2
2 2
0
0
( ) ( )
( ) 4 2 ( )
1 ( / ) 4
p
k k
p p k k
F
F
k n Z e k n Z e
k ck k
k M M mk
k k e mk
t
e t t e
t
O
= = O + O = = =
+
0 0
O (1.18)

Idea: compute bohm-staver (1.1) velocity, vs. Thomas-Fermi screened velocity (1.18) for various elements. In
order to avoid issues with units, use the estimate in footnote-2