Physics 216 Spring 2012

The Optical Theorem

1. The probability currents
In the quantum theory of scattering, the optical theorem is a consequence of the conser-
vation of probability. As usual, we dene
( x, t) |( x, t)|
2
,

j( x, t) =
i
2m
_

( x, t)

( x, t) ( x, t)

( x, t)
_
. (1)
The conservation of probability is expressed by

j +

t
= 0 .
For a stationary state solution, ( x, t) is independent of time, in which case, /t = 0.
Hence, it follows that



j = 0. Integrating this equation over the volume of a sphere
(centered at the origin) of radius R r
0
, where r
0
is the range of the potential, it follows
that _
V

j d
3
x =
_
S

j r da = r
2
_
S

j r d = 0 , (2)
where S is the surface of a sphere of radius R centered at the origin.
In scattering theory, the asymptotic form for the wave function is a stationary state that
represents the incoming plane wave an an outgoing (scattered) spherical wave,

k
( x) = e
i

k x
+ f(, )
e
ikr
r
, as r . (3)
In spherical coordinates, we have,

= r

r
+O
_
1
r
_
.
Inserting eq. (3) into the denition of the probability current [eq. (1)], it follows that

j =

m
_

k + k r
|f(, )|
2
r
2
_
+

j
int
, (4)
where

j
int
=
k
2mr
(

k + r)
_
f(, )e
i(

k xkr)
+ f

(, )e
i(

kxkr)
_
(5)
is the contribution to the probability current due to the interference between the incoming
plane wave and the outgoing (scattered) spherical wave.
1
2. Evaluation of lim
r
e
i

kx
To compute the cross section, we need the asymptotic forms for the probability currents
given in eqs. (4) and (5). In particular, we need to make sense of
1
lim
r
e
i

k x
, where r | x| . (6)
Strictly speaking, the above limit does not exist. However, this limit does exist in the sense
of distributions. This is not surprising, since plane waves are an idealization of the initial
state of the scattering process. In reality, the initial state is more realistically represented by
a wave packet with some spread of initial momenta

k. Employing the plane wave simplies

the mathematical analysis, although with a price of dealing with certain quantities such as
eq. (6) that must be carefully treated.
To evaluate the limit in eq. (6), we begin with the well-known expansion of the plane
wave in terms of spherical waves,
e
i

kx
= 4

=0

m=
i

(kr)Y

m
(

k)Y
m
( r) , (7)
where Y
m
( r) Y
m
(
r
,
r
) and Y
m
(

k) Y
m
(
k
,
k
). That is, the unit vectors r and

k are
specied by polar and azimuthal angles (
r
,
r
) and (
k
,
k
), respectively. Asymptotically,
we have as r ,
j

(kr) =
sin(kr
1
2
)
kr
+O
_
1
r
2
_
=
1
kr
_
sin kr cos(
1
2
) cos kr sin(
1
2
)

+O
_
1
r
2
_
.
Inserting this result into eq. (7) yields
lim
r
e
i

kx
= 4

=0

m=
i

kr
_
sin kr cos(
1
2
) cos kr sin(
1
2
)

Y

m
(

k)Y
m
( r)
= 4
_
sin kr
kr

even

m=
Y

m
(

k)Y
m
( r)
i cos kr
kr

odd

m=
Y

m
(

k)Y
m
( r)
_
, (8)
after using
sin(
1
2
) =
_
i
1
, for odd ,
0 for even ,
cos(
1
2
) =
_
0 , for odd ,
i

for even .
To evaluate the sums in eq. (8), we consider the completeness relation,

=0

m=
Y

m
(

k)Y
m
( r) = (

k r) , (9)
1
The limit r really means that we take the dimensionless quantity kr .
2
where the delta function above means
(

k r) = (
k

r
) = (cos
k
cos
r
)(
k

r
) ,
so that _
d
k
(

k r) = 1 .
Noting that
Y
m
( r) = Y
m
(
r
,
r
+ ) = (1)

Y
m
(
r
,
r
) = (1)

Y
m
( r) ,
it follows that

=0

m=
(1)

Y

m
(

k)Y
m
( r) = (

k + r) . (10)
Adding and subtracting eqs. (9) and (10) then yields

even

m=
Y

m
(

k)Y
m
( r) =
1
2
_
(

k r) + (

k + r)
_
,

odd

m=
Y

m
(

k)Y
m
( r) =
1
2
_
(

k r) (

k + r)
_
.
Using the above results to evaluate eq. (8), we end up with
lim
r
e
i

k x
=
2i
kr
_
e
ikr
(

k + r) e
ikr
(

k r)
_
. (11)
That is, lim
r
e
i

kx
can be understood as a distribution (often called a generalized function)
that can be expressed in terms of delta functions. Distributions acquire meaning when
multiplied by a smooth function and integrated over an appropriate region. In principle, one
can derive the entire asymptotic series for e
i

k x
as r by employing the full asymptotic
series for j

(kr) in eq. (7). Eq. (11) can also be viewed as a particular generalization of the
Riemann-Lebesgue lemma.
3. Derivation of the Optical Theorem
We can now compute the large r behavior of eq. (5). Using eq. (11),
lim
r
e
i(

k xkr)
=
2i
kr
_
e
2ikr
(

k + r) (

k r)
_
.
Hence, it follows that
lim
r
(

k + r)e
i(

kxkr)
=
4i
kr

k (

k r) ,
after employing the identities, (

k + r)(

k + r) = 0 and (

k + r)(

k r) = 2

k(

k r).
3
Consequently,

j
int
=
k
2mr
4i
kr

k (

k r) [f

(, ) f(, )]
=
4
mr
2

k (

k r)Imf(, ) . (12)
The eect of the (

k r) in eq. (12) is to set = 0 (the corresponding value of is irrelevant).

Hence, eqs. (4) and (5) yield

j =
k
m
_

k
_
1
4
kr
2
(

k r) Imf( = 0)
_
+
r
r
2
|f(, )|
2
_
+O
_
1
r
3
_
. (13)
Using eq. (2), we must evaluate the following integrals,
_

k r d =
_
cos d = 0 ,
_

k r (

k r) d =

k r

r=

k
= 1 ,
_
|f(, )|
2
d =
T
, (14)
where
T
is the total cross-section for scattering.
2
Therefore, eqs. (2) and (13) yield

4
k
Imf( = 0) +
T
= 0 ,
which is the celebrated optical theorem,
Imf( = 0) =
k
4

T
. (15)
The derivations presented in Sections 13 above were inspired by Askold M. Perelomov
and Yakov B. Zeldovich, Quantum Mechanics: Selected Topics (World Scientic Publishing
Company, Singapore, 1998), Chapter 2.4.
4. Another derivation of the Optical Theorem
It is instructive to present a second proof of eq. (15) using the abstract formulation
of scattering theory. In abstract scattering theory, the integral equation for scattering is
represented by the Lippmann-Schwinger equation,
|
(+)
= |

k +
1
E H
0
+ i
V |
(+)
, (16)
2
Recall that the denition of the cross-section is d =

j
sc
r r
2
d/j
inc
, where

j
sc
= k r|f(, )|
2
/(mr
2
)
and j
inc
= k/m. It follows that d/d = |f(, )|
2
.
4
where the (+) superscript (which corresponds to the sign of the i) indicates that the scat-
tered wave corresponds to outgoing spherical waves, and H
0
is the free-particle Hamiltonian
(in absence of the scattering potential V ). The states |

k are eigenstates of H
0
with corre-
sponding eigenvalues E =
2
k
2
/(2m). The transition operator T is then dened by
T|

k = V |
(+)
. (17)
The scattering amplitude f(, ) is related to the matrix elements of the transition op-
erator,
f(, ) =
4
2
m

|T|

k , (18)
where (, ) are the polar and azimuthal angles of the vector

k in a coordinate system in
which

k = k z. Setting

k

=

k is equivalent to setting = 0; thus the forward scattering
amplitude is
f( = 0) =
4
2
m

k|T|

k .
Using eqs. (16) and (17), it follows that

k|T|

k =

k|V |
(+)
=
(+)
|V |
(+)

(+)
|V
1
E H
0
i
V |
(+)
. (19)
We now employ the formal operator identity,
1
E H
0
i
= P
1
E H
0
+ i(E H
0
) , (20)
where P is the principal value. Under the assumption that V is hermitian,
Im
(+)
|V |
(+)
= Im
(+)
|V
1
E H
0
V |
(+)
= 0 ,
since the diagonal elements of hermitian operators are real. Hence, eqs. (17)(20) yield
Im

k|T|

k =
(+)
|V (E H
0
) V |
(+)
=

k|T

(E H
0
) T|

k .
To evaluate the above matrix element, we insert a complete set of eigenstates of H
0
.
Then, when (E H
0
) acts on |

, it yields the eigenvalue

_
E
2
k
2
/(2m)
_
. That is,
the operator delta-function is replaced by a c-number delta-function. Hence,
Im

k|T|

k =
_
d
3
k

k|T

|T|

k
_
E

2
k
2
2m
_
. (21)
Since k |

k| and k

| are both positive, the following identity can be used,

_
E

2
k
2
2m
_
=
_

2
2m
(k
2
k
2
)
_
=
m

2
k
[(k k

) (k + k

)] =
m

2
k
(k k

) ,
after noting that (k +k

) = 0 since its argument can never be zero. Thus, the integral given
in eq. (21) can be evaluated by writing d
3
k

= k
2
dk

d using the delta-function to perform

the integration over k

. The end result is given by

Im

k|T|

k =
mk

2
_
d|

|T|

k|
2
Finally, using eqs. (14) and (18) we recover the optical theorem given by eq. (15).
5