221B Lecture Notes

Notes on Spherical Bessel Functions

1 Denitions
We would like to solve the free Schrodinger equation

h
2
2m

1
r
d
2
dr
2
r
l(l + 1)
r
2

R(r) =
h
2
k
2
2m
R(r). (1)
R(r) is the radial wave function (x) = R(r)Y
m
l
(, ). By factoring out
h
2
/2m and dening = kr, we nd the equation

d
2
d
2

l(l + 1)

2
+ 1

R() = 0. (2)
The solutions to this equation are spherical Bessel functions. Due to some
reason, I dont see the integral representations I use below in books on math-
emtical formulae, but I believe they are right.
The behavior at the origin can be studied by power expansion. Assuming
R
n
, and collecting terms of the lowest power in , we get
n(n + 1) l(l + 1) = 0. (3)
There are two solutions,
n = l or l 1. (4)
The rst solution gives a positive power, and hence a regular solution at the
origin, while the second a negative power, and hence a singular solution at
the origin.
It is easy to check that the following integral representations solve the
above equation Eq. (2):
h
(1)
l
() =
(/2)
l
l!

i
+1
e
it
(1 t
2
)
l
dt, (5)
and
h
(2)
l
() =
(/2)
l
l!

i
1
e
it
(1 t
2
)
l
dt. (6)
1
By acting the derivatives in Eq. (2), one nds

d
2
d
2

l(l + 1)

2
+ 1

h
(1)
l
()
=
(/2)
l
l!

i
1
(1 t
2
)
l

l(l + 1)

2
+
2(l + 1)it

t
2

l(l + 1)

2
+ 1

dt
=
(/2)
l
l!
1
i

i
1
d
dt

e
it
(1 t
2
)
l+1

dt. (7)
Therefore only boundary values contribute, which vanish both at t = 1 and
t = i for = kr > 0. The same holds for h
(2)
l
().
One can also easily see that h
(1)
l
() = h
(2)
l
(

) by taking the complex

conjugate of the expression Eq. (5) and changing the variable from t to t.
The integral representation Eq. (5) can be expanded in powers of 1/.
For instance, for h
(1)
l
, we change the variable from t to x by t = 1 + ix, and
nd
h
(1)
l
() =
(/2)
l
l!

0
e
i(1+ix)
x
l
(2i)
l

1
x
2i

l
idx
= i
(/2)
l
l!
e
i
(2i)
l
l

k=0
l
C
k

0
e
x

x
2i

k
x
l
dx
= i
e
i

k=0
(i)
lk
(l + k)!
2
k
k!(l k)!
1

k
. (8)
Similarly, we nd
h
(2)
l
() = i
e
i

k=0
i
lk
(l + k)!
2
k
k!(l k)!
1

k
. (9)
Therefore both h
(1,2)
l
are singular at = 0 with power
l1
.
The combination j
l
() = (h
(1)
l
+ h
(2)
l
)/2 is regular at = 0. This can be
seen easily as follows. Because h
(2)
l
is an integral from t = 1 to i, while
h
(1)
l
from t = +1 to i, the dierencd between the two corresponds to an
integral from t = 1 to t = i and coming back to t = +1. Because the
integrand does not have a pole, this contour can be deformed to a straight
integral from t = 1 to +1. Therefore,
j
l
() =
1
2
(/2)
l
l!

1
1
e
it
(1 t
2
)
l
dt. (10)
2
In this expression, 0 can be taken without any problems in the integral
and hence j
l

l
, i.e., regular. The other linear combination n
l
= (h
(1)
l

h
(2)
l
)/2i is of course singular at = 0. Note that
h
(1)
l
() = j
l
() + i n
l
() (11)
is analogous to
e
i
= cos + i sin . (12)
It is useful to see some examples for low l.
j
0
=
sin

, j
1
=
sin

2

cos

, j
2
=
3
2

3
sin
3

2
cos ,
n
0
=
cos

, n
1
=
cos

2

sin

, n
2
=
3
2

3
cos
3

2
sin ,
h
(1)
0
= i
e
i

, h
(1)
1
= i

2

i

e
i
h
(1)
2
= i

3
2

3

3i

e
i
.
h
(2)
0
= i
e
i

, h
(2)
1
= i

2
+
i

e
i
h
(2)
2
= i

3
2

3
+
3i

e
i
.
(13)
2 Asymptotic Behavior
Eqs. (8,9) give the asymptotic behaviors of h
(1)
l
for :
h
(1)
l
i
e
i

(i)
l
= i
e
i(l/2)

. (14)
By taking linear combinations, we also nd
j
l

sin( l/2)

, (15)
n
l

cos( l/2)

. (16)
3 Plane Wave Expansion
The non-trivial looking formula we used in the class
e
ikz
=

l=0
(2l + 1)i
l
j
l
(kr)P
l
(cos ) (17)
can be obtained quite easily from the integral representation Eq. (10). The
point is that one can keep integrating it in parts. By integrating e
it
factor
3
and dierentiating (1 t
2
)
l
factor, the boundary terms at t = 1 always
vanish up to l-th time because of the (1 t
2
)
l
factor. Therefore,
j
l
=
1
2
(/2)
l
l!

1
1
1
(i)
l
e
it

d
dt

l
(1 t
2
)
l
dt. (18)
Note that the denition of the Legendre polynomials is
P
l
(t) =
1
2
l
1
l!
d
l
dt
l
(t
2
1)
l
. (19)
Using this denition, the spherical Bessel function can be written as
j
l
=
1
2
1
i
l

1
1
e
it
P
l
(t)dt. (20)
Then we use the fact that the Legendre polynomials form a complete set of
orthogonal polynomials in the interval t [1, 1]. Noting the normalization

1
1
P
n
(t)P
m
(t)dt =
2
2n + 1

n,m
, (21)
the orthonormal basis is P
n
(t)

(2n + 1)/2, and hence

n=0
2n + 1
2
P
n
(t)P
n
(t

) = (t t

). (22)
By multipyling Eq. (20) by P
l
(t

)(2l + 1)/2 and summing over n,

n=1
2l + 1
2
P
l
(t

)j
n
() =
1
2
1
i
n

1
1
e
it

n=0
P
l
(t

)P
l
(t)dt =
1
2
1
i
n
e
it

. (23)
By setting = kr and t

= cos , we prove Eq. (17).

If the wave vector is pointing at other directions than the positive z-
axis, the formula Eq. (17) needs to be generalized. Noting Y
0
l
(, ) =

(2l + 1)/4 P
l
(cos ), we nd
e
i

kx
= 4

l=0
i
l
j
l
(kr)
l

m=l
Y
m
l
(

k
,

k
)Y
m
l
(
x
,
x
) (24)
4
4 Delta-Function Normalization
An important consequence of the identity Eq. (24) is the innerproduct of two
spherical Bessel functions. We start with

dxe
i

kx
e
i

x
= (2)
3
(

). (25)
Using Eq. (24) in the l.h.s of this equation, we nd

dxe
i

kx
e
i

x
=

l,m

,m

(4)
2

d
x
drr
2
Y
m
l
(

k
)Y
m
l
(
x
)Y
m

l
(
x
)Y
m

l
(

)j
l
(kr)j
l
(k

r)
=

l,m
(4)
2

drr
2
j
l
(kr)j
l
(k

r)Y
m
l
(

k
)Y
m
l
(

). (26)
On the other hand, the r.h.s. of Eq. (25) is
(2)
3
(

) = (2)
3
1
k
2
(k k

)(

)
= (2)
3
1
k
2
sin
(k k

)(

)(

). (27)
Comparing Eq. (26) and (27) and noting

l,m
Y
m
l
(

k
)Y
m
l
(

) = (

), (28)
we nd

0
drr
2
j
l
(kr)j
l
(k

r) =

2k
2
(k k

). (29)
5