Theory - Classical Harmonic Crystals - Dell.10 12
Theory - Classical Harmonic Crystals - Dell.10 12
Theory - Classical Harmonic Crystals - Dell.10 12
Introduction
Relax rigid lattice assumption of all preceding discussion. Investigate motion of
lattice ions/atoms, etc. about equilibrium lattice sites.
Two Assumptions in what follows:
!
1) Mean equilibrium position of each ion is a BL site, and R denotes
this position.
2) Excursions of each ion away from its equil. pos. are SMALL
compared with interionic spacing. (define more precisely in what follows)
Leads to Harmonic approximation
to make life easier
!
!
Now: Denote position of ion whose mean position is R by r! ( R) .
! ! !
! !
! !
At any given time r ( R) = R + u( R) , where u( R) is deviation from equil. pos.
Internal Energy:
All BL vectors
!
!
But all BL sites are equivalent, so "! R! = N! R! and thus
!
R!
( )
( )
! !
!
N
N
R - R is also BL vector and we
U = $ ! R ! R" = $! R
are summing over all BL vectors.
! !
!
2 R#
2 R#0
R"
Now relax static approx.
!
! !
Atom whose average pos. is R will generally be found at r ( R)
Thus
( )
! ! ! ! ! !
! ! ! !
1
1
U = #! r ( R) ! r ( R") = #! R ! R" + u( R) ! u( R")
2 R!R!"
2 R!R!"
! !
P.E. now depends on dynamical variables u( R) .
! ! 2
! P( R)#
"
$
H =%
+U
!
2M
R
!
R ! !
r ( R)
! !
u( R)
! ! ! !
! !
1) Assume u( R) are small in sense that u( R) ! u( R") is small for all atom pairs
! !
that have appreciable interaction, ! R ! R" .
! !
! ! ! ! 1 ! ! 2 ! 1 ! ! 3 !
f (! + a) = f (! ) + a !f (! ) + (a !) f (! ) + (a !) f (! ) +....
2
3! !
! ! !
!
! !
!
Apply this to each term of 1
with R ! R" = ! , and u( R) ! u( R") = a :
! ! !
! !
! ! ! !
(! ( R ! R") + u(
, order
First
R) ! u( R") #! ( R ! R") +
*
1 *
U = /)
! 2 ! !
! ! !
!
!
1
!
!
3
$
&
2 RR" *
u( R) ! u( R") #' ! ( R ! R") + O(u ) +....*
%
+
. 2nd order
2
Zeroth
order
N
=
2
(
R
)
R 0
Rewriting on next pg
N
2
( R) +
R 0
1
1
) + O (u 3 ) + ...
u
(
R
)
u
(
R
)
(
R
R
)
+
u
(
R
)
u
(
R
)
(
R
2 RR
4 RR
[(
) ]
Linear in u ( R) u ( R) :
Coeff. of u (R) is
Eval. at
this R
order
harm.
1
=
4
2 ( r )
u ( R ) u ( R) ( R R) uv ( R ) u ( R) , where ( r ) =
r r
RR
, = x , y , z
Frequently ignore Uequil. (constant, indep. of us and Ps), and then U = Uharm.
Starting point for essentially all theories of lattice dynamics. Further corrections u3 and u4
are known as anharmonic terms.
Harmonic P.E. usually written in more general form:
harm.
1
=
2
u ( R ) D ( R R)u ( R)
RR
, = x , y , z
Relaxes restriction
to only pairwise
interactions
1
2
)
u
(
R
)
D
(
R
R
)
u
(
R
RR
, = x , y , z
This is equivalent to prev. expression for Uharm. (assuming pairwise interactions) if we take
D ( R R) = RR
(
R
R
)
(
R
R)
Eq. 2
represents general interaction (through D) between displaced ion at R
with displaced ion at R . It need not be simple, direct pairwise interaction. Could
take place via another mechanism, e.g., distortion of electronic arrangement around
ions (contributes to P.E. of xtal), with arrangement depending on ionic configuration.
Also Coulombic interaction among ions is not simply pairwise, e.g., displacement of
ion i directly affects ion j, but also affects ion j through interacting with ions k, l, m, .,
and their direct interaction with j.
Generally very complex.
du ( R)dP( R)
du ( R)dP ( R) .
And d is the volume element in crystal Phase Space, d =
Write
Now
H =
2
P( R)
+ U harm.
2M
Ignoring Uequil.
2
u ( R) = u ( R) ; du ( R) = 2 du ( R)
3
P( R) = P ( R); dP( R) = 2 dP ( R)
1
2
R,
1 u ( x)
ln u ( x) =
x
u ( x) x
Be careful with
notation in the ! !
following: u and u( R)
1 1 1
(du represents du x du y du z = 2 2 2 du x du y du z )
de
P
( R) 2 1
3 N
=
du ( R)dP ( R) exp
u ( R) D u ( R)
2
M
2
Here N is # of
Bravais Lattice
points
Integral is indep. of T ()
u=
1
1
[ 3N ln + ln(const.)]
[ln 3 N term indep. of ] =
V
V
u=
[3N ] = 3nk
V
cV =
u
= 3nk B
T
n = # ions per
unit volume
Dulong-Petit Law
(n-3)a
(n-2)a
u(n-2)a
(n-1)a
na
(n+1)a
(n+2)a
(n+3)a
static
dynamic
u(na)
1
4
'
!!
RR"
,! =x, y,z
!
!
! !
!
!
#u ( R) ! u ( R")%! " ( R ! R")#u! ( R) ! u! ( R")%
$
&
$
&
e.g., R = (n+1)a,
R= na
harm.
1
#2!
2
= K " [u(na) ! u([n +1]a)] , where K =
2
#x 2
n
!
R!
Equations of motion
2,
"U harm.
" )K
$
&
!!
#
#
Mu(na) = !
=!
* (%u(n a) ! u ([ n +1]a )' "u(na)
"u(na) + 2 n#
.
nn
x=a
(x) is P.E. of
two ions a
distance x
apart
= ! $$(a)
n,n+1
A = 0 unless
n=n (and then A=1)
B = 0 unless
n = n+1 (and then B = 1)
"U harm.
!
= !K #$u(na) ! u ([n +1]a )%& + K #$u([n !1]a) ! u ( na )%&, and
"u(na)
!!
Mu(na)
= !K #$2u(na) ! u ([n !1]a ) ! u ([n +1]a )%&
Same Eq. of motion as for linear
chain of masses connected by
perfectly mass spring of spring
const. K
Boundary Conditions
For finite # of ions N must specify how ions at ends are to be treated.
Assume very large N (end effects relatively unimportant) and use
Born-von Karmen (cyclic) boundary conditions.
a
Linear chain
Join ions at either end of chain by same spring that connects internal ions.
Assume osc. solns. of form
a
Na
Boundary
u(na, t) ! ei(kna"!t )
a
2a
so kNa = 2! m, and k =
Allowed values
of k
2! m
a N
ei ( kna ) ei ( kna ) e
2
i na
a
Example: ( = 4a)
= ei ( kna ) 1
= 4a; k =
k =
2
=
2a
2
2 5
2 4a
; so k = +
= , and =
=
a
2a a 2a
k
5
Now
Substitute assumed solutions for u(na,t) into eq. of motion.
Mu(na, t ) = K [2u (na, t ) u ([n 1]a, t ) u ([n + 1]a, t )]
2K
(1 cos ka )
M
1 cos ka
or, using
1
= sin ka
2
(k ) =
1
sin ka
2
2K
(1 cos ka ) = 2 K sin 1 ka
M
M
2
k
2! / a
cos(kna t )
u (na, t )
sin(kna t )
Took positive root because is usually considered to be positive and because is an even
funct. of k; above solutions for k and -(k) are the same as those for -k and (k) .
K
1
K
1
sin ka t ) and cos(kna 2
sin ka t ) are the same
M
2
M
2
since cos(x) = cos(-x)
cos(+ kna + 2
Negative root
doesnt add
anything
.
Now there are N distinct values of k (between -/a and +/a, each with a unique frequency; so
There are 2N independent solutions (but only N normal modes) But sine solution is cosine
solution shifted in time by /2 -- can take linear combination of the two normal modes. Arbitrary
Motion of the chain is obtained by specifying N initial positions and N initial velocities so this
is a complete solution. These solutions are waves propagating along the chain with phase
velocity c = /k and group velocity v = /k.
(k)
K
1
(k ) = 2
sin ka
M
2
4K
M
(k ) = 2
K 1
K
ka = a
k
M 2
M
K
k << 2/a
-/a
/a
a-d
na
(n-1)a
(n+1)a
(n+2)a
(n+3)a
Denotes unit
cell; cells
separated by a
Now denote by u1(na) the displacement of ions that oscillate about equilibrium site
at na
, and by u2(na) the displacement of ions that oscillate about equilibrium site
at (na + d) .
Recall general form for P.E. (generalized a bit and specific to 1D no need for sum).
U
harm.
1
=
4
[
]
[
u ( R) u ( R ) ( R R ) u ( R) u ( R)]
RR
s =1, 2
Added sum over s (basis will yield 2 sums , one for u1s and one for u2s.)
Again a factor of two will arise from assumption of nn interactions and double sum over
R and R (sums are equivalent). As we assumed only nn interactions,
2u
2u
within
between
K>G
R R = d ,
or
( a d );
K =
; G=
2
2
a cell
cells
x d
x a d
( n 1) a + d ( n 1) a = d
na [( n 1) a + d ] = a d
a-d
(n-1)a
1) Single out ion
na
(n+1)a
(n+3)a
(n+2)a
in same cell.
2
in cell (n + 1)a.
2
So
U harm. =
K
2
G
[u2 (na) u1 ([n + 1]a))]2
4
G
2
2
[
]
u
(
na
)
u
(
na
)
+
1
[u2 (na) u1 ([n + 1)a)]
2
2
within a cell
between cells
Mu1 (na ) =
=
u1 (na )
u1 (na ) 2
G
2
[u2 (na) u1 ([n + 1)a)]2
n [u1 (na) u2 (na)] + 2
n
K
[u1 (na) u2 (na)]2
4
u ( na )
= nn
u ( na )
u ([ n + 1]a )
= n , n+1
u ( na )
Solutions of form:
at site (n+1)a
u1 (na, t ) = 1e i ( kna t )
u 2 (na, t ) = 2 e i ( kna t )
2 m
a N
[M
[M
2 =
[M
] [
[K + Ge ] + [M
(K + G ) 2 + K + Geika 1 = 0
K +G 1
M
M
Rewriting
] [
(K + G ) 1 + K + Ge ika 2 = 0
K 2 + G 2 + 2 KG cos ka
] [
(K + G ) 1 + K + Ge ika 2 = 0
ika
(K + G ) 2 = 0
and
2
K + Geika
K + Geika
=
=
, but K 2 + G 2 + 2 KG cos ka = K + Geika
2
1
M ( K + G )
K 2 + G 2 + 2 KG cos ka
2
K + Geika
thus
=
4
1
K + Geika
2
(
ka )
cos ka 1
K +G 1
M
M
K 2 + G 2 + 2 KG 2 KG
(K + G)
2
Positive Root
+2 2
(ka )2
2
K +G K +G
KG
(ka )2
1
2
M
M
(K + G )
K + G K + G
KG
(ka) 2 + ....
1
2
M
M 2(K + G )
K + G KG
( K + G)
(ka) 2 + ....; so + = 2
O(ka) 2
M
M
M (K + G )
Negative Root
2
K +G K +G
KG
KG
+
(ka) 2 + ....; so
ka
M
M
2M (K + G )
2M ( K + G )
Use binomial
expansion on
radical
Binomial expansion
again; also note:
const. plus neg.
quadr. term. (ZB
value < Z. center
Linear in k, and
2 = + 1
2 = - 1
1
K +G 1 2
K +G K G
K + G 2 2 KG 2 =
M
M
M
M
Positive Root
K +G K G
K
2K
+
= 2 ; so + =
M
M
M
M
2 = - 1
K +G K G
G
2G
= 2 ; so =
M
M
M
M
2 = + 1
2
+
Negative Root
(k)
2( K + G )
M
KG
Slope =
ka
2M ( K + G )
1st BZ
2
K + Geika
=
1
K + Geika
K G
= 1 ( K > G )
So 2 =
1
K G
2K
M
+ optical modes
d
2G
M
a-d
Within cell ions are out-of-phase
- acoustic modes
-2/a
-/a
k = 2/
/a
2/a
a-d
1 corresponds to
; 2 corresponds to
K + G 2 KG = 1 +
1+ 2 +
cos ka
M
M
M K M
K
K
K K G
2 1 + cos ka
M M K
2K
M
1
+
O
(
)
2G ka
G
G
1
+
cos
ka
(
1
cos
ka
)
=
sin , and
K
M
M
2
2G
ka
sin
M
2
Amplitudes
G ika
e
2
K
=
1
G ika
1
1+ e
K
1+
For +: 2 = - 1
For -: 2 = + 1
Monatomic BL first
U harm. =
! ! " ! ! ! !
1
u(
# R)D( R ! R")u( R")
2 R!R! "
D(R R) = D(R R)
D(R R) = D(R R)
See A&M
R D( R) = 0
U harm.
= D ( R R)u ( R)
Mu ( R ) =
u ( R ) R ,
i ( k R t )
u ( R, t ) = e
u ( R + ni ai ) = u ( R ), N = N1 N 2 N 3
Allowed values of k
n5
n1
n2
k =
b1 +
b2 +
b3
N1
N2
N3
Next Page
ik R
From matrix algebra - real symm. matrix has 3 real e-vectors that satisfy
e-val. Eq. --- diagonalizes matrix D(k)); 1, 2, 3
D(k ) s (k ) = s (k ) s (k )
M s = D(k ) s = s (k ) s
2
3 normal modes with wave vector k will have pol. vectors s(k)
and frequencies
s ( k )
s (k ) =
M
As in 1D case
; when k 0, s ( k ) c s ( k ) k , >> a
OPTICAL MODES
Cells vibrate
as a unit
ZB
Schematic for p = 2
ZB
!
k
Important points:
1)
2)
3)
4)