Nanostructures Research Group

CENTER FOR SOLID STATE ELECTRONICS RESEARCH

Time-Dependent Perturbation
Theory
David K. Ferry and Dragica
Vasileska
Arizona State University
Tempe, AZ 85287-5706
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
0 0
0
0
0
0
|
|
) , ( ) , (
> =
c
> c
+ = +
c
c
n H
t
n
i
t r H t r
t
i
n n

0
) 0 (
0 0
| | ) ( > = > n E n r H
n
Objective: To calculate the transition probability from a state |n> to a state |m>.These occur
under the influence of a perturbation ,which slightly displaces the system from equilibrium and
causes the transition. The perturbation is usually time dependent.
To calculate the change in the eigen state (r ,t) due to the perturbation, V(r ,t), we solve the
TDSE.
) , ( ) ( '
) , ( ' ) , (
0
t r V r H H
t r H t r
t
i
+ =
+ = +
c
c

where H
0
(r) is the operator for the total energy of the unperturbed system and V(r,t) is the
perturbing potential.

We first consider the unperturbed system
The stationary part of the wave function is assumed to
satisfy the eigen value equation
The solution is then of the form,
0
0
| ) , (
) 0 (
> = +

n e t r
t iE
n
n

1
2
3
4
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
The wave functions form a complete and orthonormal set and so the total wave function can be
expressed as a linear combination of the individual eigen functions for each state.
0
| ) ( ) , (
) , ( ) ( ) , (
) 0 (
> = +
+ = +

n e t a t r
t r t a t r
t iE
n
n
n
o
n
n
n

5
6
Inserting this into the TDSE 1( for
the perturbed system), and using
equation 3, we get on simplification,
0
/
0
/
| ) , ( ) ( |
) 0 ( ) 0 (
> = >
c
c


n t r V e t a n e
t
a
i
t iE
n
n
t iE
n
n
n n

7
We now use the property that two non identical states are orthogonal to each other, by
pre multiplying the above equation with and integrating over
all space to get:
| ) , ( 0
*
0
) 0 (
m e t r
t iE
m
m
< =

0
0
/
0
0
/ ) (
| ) , ( | ) ( |
) 0 ( ) 0 ( ) 0 (
> < = > <
c
c


n t r V m e t a n m e
t
a
i
t iE
n
n
t E E i
n
n
n n m

8
n m,
o mn
V
0
0
/
| ) , ( | ) (
) 0 (
> < =
c
c

n t r V m e t a
t
a
i
t iE
n
n
m
n

9
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Let us assume that we have a harmonic potential. Then
t i
e r V t r V
0
) ( ) , (
0
e
=
mn
t i t i
V e n r V m e n t r V m
0 0
0 0
0
0 0
| ) ( | | ) , ( |
e e
= > < = > <
and,
Substituting this into equation 9, we get,
10
mn
t E E i
n
n
m
V e t a
t
a
i
n m

/ ] [
0
) 0 ( ) 0 (
) (
e

=
c
c
11
What we have obtained is a complete set of equations for the expansion coefficients a
m
(t) and
their time dependence. Matrix methods can be used to solve this problem but it is generally a
computationally taxing task.
The approach that we adopt is as follows. We assume that initially the system was in a state k
before the perturbation is turned on :

From the above expression, we have
0 , 0 0
| ) 0 ( | ) 0 ( | ) 0 , ( ) = ) = ) =

k a n a k r
k k n
n
n
o
= ) 0 (
n
a
k n
k n
=
=
0
1
To understand the meaning of these coefficients, consider the normalization of
) , ( t r
1 ) (
) ( ) ( ) (
| ) ( ) ( ) , ( | ) , (
1 ) , ( | ) , (
2
2
*
,
0
0
/ ) ( *
) 0 ( ) 0 (
= =>
= =
) ( = ) (
= ) (

m
m
m
m
m
m m
n m
t E E i
n m
t a
t a t a t a
n m e t a t a t r t r
t r t r have We
n m



It can also be shown that,
) ( ) , ( | ) , (
) 0 (
t a t r t r
k k
= ) + (
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
2
2
2
2
) 2 / (
) 2 / sin(
) (
) ( | |
1
| ) ( | ) (
(

=
= =
t
t
t g
tg V t a t W
mk
mk
mk
mk mk m mk
e
e
e
e

The next task is to calculate the amplitude a

m
(t), from the coupled set of differential equations. If
V(r,t) is small perturbation, we can make an approximation that even at t>0 we have, In
this case we only have to solve the following equation

k n n
t a
,
) ( o =
mk
t E E i
k
m
V e t a
t
a
i
k m

/ ] [
0
) 0 ( ) 0 (
) (
e
=
c
c
12
1
Integrating this equation from 0 to t,
we get
mk
t E E i
k m
m m
V e
E E i
a t a i
k m
] 1 [
] [
)] 0 ( ) ( [
/ ] [
0
) 0 ( ) 0 (
0
) 0 ( ) 0 (

e
e
0
k m for =
) 1 ( ) (
] [
1
0
) 0 ( ) 0 (
=
=
t w i
mk
mk
m
k m mk
mk
e
iw
V
t a i
E E w Defining

e
13
mk
t iw
mk
mk
m
V e
w
t w
i t a
mk
2 /
) 2 / sin( 2
) (

= 14
2 2
| ) ( | ) ( . | ) ( | t a t W by this denote We m to k state from made is transition a that y probabilit the is t a
m mk m
=
15
t e e 2 ) ( =
}

mk mk
d g
) ( 2
) ( 2 ) (
lim
mk
mk mk
t
E
g
o t
e to e
=
=

The following can also be shown
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
To summarize, in the limit t->, the probability W
mk
(t) is
The above limit can only be used in the case when one
scattering event is finished before the other starts i.e.,
when the scattering is infrequent .
S
mk
is called transition rate i.e., the probability for the transition in a time of interval t
mk k m mk
mk mk mk
S t E E V t
E t V t W
= =
=
) ( | |
2
) ( 2 | |
1
) (
0
) 0 ( ) 0 (
2
2
2
e o
t
o t

) ( | |
2
0
) 0 ( ) 0 ( 2
e o
t

=
k m mk mk
E E V S
15
The formula in equation 15 was named by Fermi (as
with some measure of importance) The Golden
Rule.
Summary of approximations made in deriving Fermis Golden Rule
Weak perturbation which gives no depletion of the initial state a
k
(t)=1 for all times
The limit t->, can only be used when we are in a weak scattering regime. It led to
Energy conservation in the scattering process which is not necessarily satisfied under
heavy scattering.
Summary of Notation
k state of out rate scattering total k k S k
k state of out scattering of y probabilit total k k W k P
m state to k state f rom rate transition S
m state to k state f rom scattering of y probabilit W
k
k
mk
mk
= I
=

'
'
) , ( ) (
) , ( ) (
'
'
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Example that illustrates the application of Fermis Golden rule
Consider a potential barrier of the form :
We can view this as a scattering potential. Use Fermis Golden rule to calculate the
reflection probability and compare it with the exact result of
otherwise z U
d z V z U
s
s
0 ) (
0 ) (
0
=
< < =
) (
2
;
) 1 ( 4 ) ( sin
) ( sin
) (
0
2
0 0
2
2
V E
m
k
V
E
V
E
kd
kd
E R
B
=
+
=

Solution
We calculate the matrix element for the scattering event as follows.
dz e V
L
V
z k k i
d
z
k k
) ' (
0
0 ' ,
1

}
=
Evaluating this straight forward integral,
2 / ) ' (
] 2 / ) ' sin[(
2 / ) ' ( 0
' ,
d k k
d k k
d e
L
V
V
d k k i
z
k k

=

The total probability of scattering out of state k is
2
2
2
0
' '
2
'
'
) ' (
] 2 / ) ' sin[(
4 ) ( ' ) (
2
) (
|
|
.
|

\
|

= =
}

k k
d k k
L
V
E E dk
T L
E E V
T
k P
z
k k
z
k k
k
kk
o o
t

'
'
;
2
) ' (
2
'
2
'
dk
m
k
E d
m
k
E Since
k
k

=
=
From energy conservation we have |k|=|k|, but we consider only k=-k for
reflection. Also using the dispersion relation, we convert the integration into
Energy domain and the result is
) ( sin ) (
2
3 3
*
2
0
kd
k L
m TV
k P
z

=
Now T in the expression is the time it takes for electrons to cross
the length L
z
. Therefore
k
m
Lz
v
Lz
T

*
= =
) ( sin ) (
2
2
2 2
*
0
kd
k
m V
k P
|
|
.
|

\
|
=

If we consider very weak scattering then V0<<E

and it is very simple to show that R
B
(E) reduces
to the expression from Fermis Golden Rule
Fermis Golden Rule result
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Application 2: Elastic scattering of Electrons
Consider the coulomb interaction between a charge center Ze and the electron as a perturbation.
The perturbing potential is given by .
qn
V
L
r
e Ze
r V
T
d
Ld r
c
tc
= =

;
4
) (
/ 2
It should be noted that Coulomb
interaction is first order time
independent, which means that only
the w=0 term will be non zero

Before we proceed with the solution, we need to introduce
the concept of SCATTERING CROSS SECTION
The probability of scattering out of state k is given by
Evaluating this, by using the parabolic dispersion relation and reducing the integral to energy
domain, we get
}}}

O
= ) , ( '
) 2 (
) , ( ) (
' 3
3
'
'
k k W k d k k W k P
k
t
) (
2
) (
'
2
'
' k k
k
kk
E E V
T
k P =

o
t

2
3 2
| |
4
*
) ( Vkk
Tk m
k P
t
O
=
Again, T in the expression is the time
it takes for electrons to cross the
length L
z
.

Therefore,
k
m
A V
L
T
z

* O
= =
) , (
1
| |
2
* 1
) , (
2
2
2
| u o
t
| u
A
Vkk
m
A
P =
|
.
|

\
| O
=

To get the total scattering cross section,

we need to integrate over entire solid
angle
) , ( ) sin(
0
2
0
| u o | u u o
t t
} }
= d d
s
Physical explanation of scattering cross section
The scatterer can be viewed as a solid obstacle with
cross sectional area
s
which scatters a fraction of the
incident electrons equal to
s
/A.
Usually we have electrons moving through a medium
of certain density (Ni) of scattering centers and we need
to find the probability that an electron is scattered in time
t .
Since each scatterer scatters
s
/A fraction of electrons,
the total number of scatterers encountered in that time t
is, NiAVt.
Therefore, Ps(t)=s/A* NiAVt=t/, where is the mean
time between scattering events

Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
Therefore to evaluate T, we need to find the total scattering
cross section. where,
) , ( ) (cos
1
1
2
0
| u o | u o
t
} }

= d d
s
2
2
2
| ' |
2
* 1
) , ( Vkk
m
A
O
|
.
|

\
|
=
t
| u o
The matrix element is calculated from }}}

O
=
r k k i
kk
e r rV d V
) ' ( 3
'
) (
1
The volume of the box to which we normalize all wave functions is completely arbitrary and
unphysical and will never appear in the final result for some measurable quantity. We now define
=k-k, as being the momentum transfer in the scattering process. Since we have freedom in
choosing the coordinate system, we can chose to be aligned with the z-axis, .r= r cos
|
.
|

\
|
= =
+
O
=
(

O
=
O
=
O
=

}
} } } } } }
2
sin 4 ) cos 1 ( 2
1
1
2
4
) (cos
4 4
) (cos
1
2 2 2 2
2
2
2
0
/
2
'
1
1
cos
0
/
2
2
0
2
cos
/ 2
2
0
1
1 0
2
'
u
u |
|
c |
t
tc
u |
tc tc
| u
| |
u |
t
u |
t
k k
Ld
Ze
r i
e e
dr e r
Ze
V
d e dr
r
e
r d
Ze
e
r
e Ze
d d dr r V
r i r i
Ld r
kk
r i
Ld r
r i
Ld r
kk
2
2
0
2
2 2 2
2
0
2
2
2
2 2
2
2
2
2
2
2
2
*
4
;
1
2
sin 4
1 2
1
2
sin 4
1
2
*
| ' |
2
* 1
) , (
e Zm
a
Ld k
a
Ld
Ld
k
Ze m
Vkk
m
A


tc
u u
c t t
| u o =
|
|
.
|

\
|
+
|
.
|

\
|
|
|
.
|

\
|
=
|
|
.
|

\
|
+
|
.
|

\
|
|
|
.
|

\
|
|
.
|

\
|
= O
|
.
|

\
|
=
If

then scattering is isotropic as

2
2
2
|
.
|

\
|
((
k
L
d
t
1 4
2 2 2
(( =
d
L k
u | u o of ent independ const a L
d
= =
2
0
4
/ 4 ) , (
and if 1 4
2 2 2
)) =
d
L k
Low energy electrons
|
.
|

\
|
=
2
sin 4
1
) , (
4 4
2
0
u
| u o
k a
then
Scattering is anisotropic and peaked around
=0; High energy electrons are hardly deflected
Nanostructures Research Group
CENTER FOR SOLID STATE ELECTRONICS RESEARCH
To get the total cross section, we need to integrate over the entire solid angle to get
2 2
0
4
2 2
0
4
2
1
1
2
2
0
2
2 2
2
2
1
1
2
0
2
0
2
) 2 ( 1
1
16
1
1
16
1 2
2
2
2
) 1 (
2
1 cos ;
) cos 1 (
2
1
1
) (cos
2
2
kLd a
Ld
a
Ld
dy
y a
Ld
dx dy x y x d d
a
Ld
s
s
+
=
+
=
|
|
.
|

\
|
|
|
.
|

\
|
=
= + = =
|
|
.
|

\
|
+
|
|
.
|

\
|
=
}
} }
+

t o

u
u

u | o

t
v N
v N
i s
i s
o
t o
t = =
1 1
2 2
0
4
) 2 ( 1
1 16
*
1
d
d
i
kL
a
L
m
k
N
+
=
t
t

Brooks Herring result
2
0
4
16
*
1
a
L
m
k
N Electrons Energy Low
d
i
t
t

=
2
2
0
2
2 2 2
0
4
4
* 4
1 16
*
1
k a
L
m
k
N
Ld k
a
L
m
k
N Electrons Energy High
d
i
d
i
t t
t

= ~
When L
d
-> (depletion region of a semiconductor),

s
and are large.

t / 1
It is important to note that once L
d
gets comparable to the inter impurity spacing, (~ N
i
-1/3
), the
coulomb potential of nearby impurities overlap. It is then necessary to modify the Brooks
Herring theory. More appropriate in those cases is the Conwell - Weisskopf theory, which is
not discussed here.