Physics 505 Fall 2007

Homework Assignment #6 Due Thursday, October 25

Textbook problems: Ch. 4: 4.1, 4.7, 4.8, 4.9
4.1 Calculate the multipole moments q
lm
of the charge distributions shown as parts a)
and b). Try to obtain results for the nonvanishing moments valid for all l, but in each
case nd the rst two sets of nonvanishing moments at the very least.
a) z
a
a
q
q
q
q

a
a
x
y
The multipole moments are given by
q
lm
=
_
(x)r
l
Y

lm
(, ) d
3
x = qa
l
[Y

lm
(

2
, 0)+Y

lm
(

2
,

2
)Y

lm
(

2
, )Y

lm
(

2
,
3
2
)]
This is given in terms of associated Legendre polynomials by
q
lm
= qa
l

2l + 1
4
(l m)!
(l + m)!
P
m
l
(0)[1 + (i)
m
(1)
m
(i)
m
]
The moments vanish unless m is odd. Writing m = 2k + 1 gives
q
l,2k+1
= 2qa
l
[1 i(1)
k
]

2l + 1
4
(l (2k + 1))!
(l + (2k + 1))!
P
2k+1
l
(0)
= 2qa
l
[1 i(1)
k
]Y
l,2k+1
(

2
, 0)
Note that by parity this vanishes unless l is odd. Hence only the odd l and m
moments are present. The lowest non-trivial ones are
q
1,1
= q

1,1
= 2qa(1 i)
_
3
8
and
q
3,3
= q

3,3
= 2qa
3
(1 + i)
1
4
_
35
4
q
3,1
= q

3,1
= 2qa
3
(1 i)
1
4
_
21
4
b)

a
x
y
z
q
q
a
q 2
In this case, we have
q
lm
= qa
l
[Y

lm
(0, 0) + Y

lm
(, 0)]
for l > 0 and q
00
= 0. By azimuthal symmetry, only the m = 0 moments are
non-vanishing. Hence
q
l0
= qa
l
_
2l + 1
4
[P
l
(1) + P
l
(1)] = qa
l
[1 + (1)
l
]
_
2l + 1
4
l > 0
We end up with even multipoles
q
l0
= qa
l
_
2l + 1

l = 2, 4, 6, . . .
Explicitly
q
20
= qa
2
_
5

q
40
= qa
4
_
9

c) For the charge distribution of the second set b) write down the multipole expansion
for the potential. Keeping only the lowest-order term in the expansion, plot the
potential in the x-y plane as a function of distance from the origin for distances
greater than a.
The expansion of the potential is
(x) =
1
4
0

l,m
4
2l + 1
q
lm
Y
lm
(, )
r
l+1
=
1

l=2,4,...
qa
l
2l + 1
_
2l + 1

Y
l0
(, )
r
l+1
=
q
2
0

l=2,4,...
a
l
r
l+1
P
l
(cos ) =
q
4
0
a
2
r
3
(3 cos
2
1) +
In the x-y plane we have cos = 0, so the lowest order term is
=
q
4
0
a
_
a
r
_
3
+
We all know what 1/r
3
looks like when plotted, but here it is
1.5 2 2.5 3 3.5 4
0.2
0.4
0.6
0.8
1
d) Calculate directly from Coulombs law the exact potential for b) in the x-y plane.
Plot it as a function of distance and compare with the result found in part c).
For three charges, the potential is simply the sum of three terms, one for each
charge. In the x-y plane, if r is the distance from the origin we have
=
q
4
0
_
1

r
2
+ a
2

1
r
+
1

r
2
+ a
2
_
=
q
2
0
r
_
1
1
_
1 + (a/r)
2
_
=
q
4
0
a
2
_
1
(r/a)

1
_
1 + (r/a)
2
_
The exact potential looks like
1.5 2 2.5 3 3.5 4
0.2
0.4
0.6
0.8
1
Divide out the asymptotic form in parts c) and d) to see the behavior at large distances
more clearly.
If we divide out by 1/r
3
, the approximate and exact potentials are
1.5 2 2.5 3 3.5 4
0.2
0.4
0.6
0.8
1
where the straight line is the approximation of c) and the sloped line is the exact
result. The approximation improves as r a.
4.7 A localized distribution of charge has a charge density
(r ) =
1
64
r
2
e
r
sin
2

a) Make a multipole expansion of the potential due to this charge density and deter-
mine all the nonvanishing multipole moments. Write down the potential at large
distances as a nite expansion in Legendre polynomials.
This charge distribution is azimuthally symmetric. As a result, only m = 0
moments will be nonvanishing. Furthermore, noting that
sin
2
= 1 cos
2
=
2
3
[P
0
(cos ) P
2
(cos )]
we may write down the moments
q
l0
=
_
(r, )r
l
Y

l0
(, ) r
2
dr dd(cos )
= 2
_
2l + 1
4
_
(r, )r
l
P
l
(cos ) r
2
dr d(cos )
=
2
64
2
3
_
2l + 1
4
_

0
r
l+4
e
r
dr
_
1
1
P
l
(cos )[P
0
(cos ) P
2
(cos )] d(cos )
=
1
48
_
2l + 1
4
(l + 5)[2
l,0

2
5

l,2
]
As a result, we read o the only nonvanishing multipole moments
q
00
=
_
1
4
, q
20
= 6
_
5
4
The multipole expansion then yields the large distance potential
=
1
4
0

l,m
4
2l + 1
q
lm
Y
lm
(, )
r
l+1
=
1
4
0

l
_
4
2l + 1
q
l0
P
l
(cos )
r
l+1
=
1
4
0
_
1
r

6
r
3
P
2
(cos )
_
(1)
b) Determine the potential explicitly at any point in space, and show that near the
origin, correct to r
2
inclusive,
(r )
1
4
0
_
1
4

r
2
120
P
2
(cos )
_
We may use a Greens function to obtain the potential at any point in space. In
general (since there are no boundaries, except at innity)
G(x, x

) =
1
|x x

|
=

lm
4
2l + 1
r
l
<
r
l+1
>
Y

lm
(

)Y
lm
(, )
However, for azimuthal symmetry, it is sucient to focus on the m = 0 terms in
the expansion
G(x, x

) =

l
r
l
<
r
l+1
>
P
l
(cos )P
l
(cos

) + (m = 0)
Then
(x) =
1
4
0
_
(x

)G(x, x

) d
3
x

=
1
4
0
2
64
2
3
_

0
r
4
e
r
r
l
<
r
l+1
>
dr

_
1
1
[P
0
(cos

) P
2
(cos

)]P
l
(cos

)P
l
(cos ) d(cos

)
=
1
4
0
1
48
_
1
r
l+1
_
r
0
r
l+4
e
r

dr

+ r
l
_

r
r
3l
e
r

dr

_
[2
l,0

2
5

l,2
P
2
(cos )]
Instead of writing this out in terms of incomplete Gamma functions, it is better
just to integrate for l = 0 and l = 2. The result is
=
1
4
0
1
24
_
1
r
_
24 e
r
(24 + 18r + 6r
2
+ r
3
)
_

1
r
3
P
2
(cos )
_
144 e
r
(144 + 144r + 72r
2
+ 24r
3
+ 6r
4
+ r
5
)
_
_
Note that as r the e
r
factors are exponentially small. As a result, we
simply reproduce (1) in this limit. On the other hand, as r 0, a Taylor
expansion yields
=
1
4
0
__
1
4
+
_

_
r
2
120
+
_
P
2
(cos )
_
(2)
Obtaining the correct l = 2 term involves the cancellation of the rst ve terms
in the Taylor expansion. Note that the leading terms in the nal expression have
the correct powers of r
l
P
l
(cos ) in order to satisfy Laplaces equation.
c) If there exists at the origin a nucleus with a quadrupole moment Q = 10
28
m
2
,
determine the magnitude of the interaction energy, assuming that the unit of
charge in (r ) above is the electronic charge and the unit of length is the hydrogen
Bohr radius a
0
= 4
0
h
2
/me
2
= 0.529 10
10
m. Express your answer as a
frequency by dividing by Plancks constant h.
The charge density in this problem is that for the m = 1 states of the 2p level
in hydrogen, while the quadrupole interaction is of the same order as found in
molecules.
We rst note that if we put the correct units of electronic charge e and Bohr
radius a
0
into the charge distribution , the potential near the origin (2) becomes
=
e
4
0
a
0
_
1
4

1
120
_
r
a
0
_
2
P
2
(cos ) +
_
where the overall minus sign is due to the negative charge of the electron. (We
take e > 0). The interaction energy is then
W =
_

N
d
3
x =
e
4
0
a
0
_

N
_
1
4

1
120
_
r
a
0
_
2
P
2
(cos ) +
_
d
3
x
where
N
is the charge density of the nucleus. Since
_

N
d
3
x = Ze gives the
total charge of the nucleus, we write
W =
e
2
4
0
a
0
_
Z
4

1
240a
2
0
1
e
_

N
r
2
(3 cos
2
1)d
3
x +
_
where we have used P
2
(x) =
1
2
(3x
2
1). Using z = r cos , this may be rewritten
as
W =
e
2
4
0
a
0
_
Z
4

1
240a
2
0
1
e
_

N
(3z
2
r
2
)d
3
x +
_
=
e
2
4
0
a
0
_
Z
4

Q
240a
2
0
+
_
where we have used the (classical) denition of the nuclear quadrupole moment
Q =
1
e
_

N
(3z
2
r
2
)d
3
x
The rst term is the electrostatic interaction energy. The quadrupole interaction
energy (expressed as a frequency) is
W/h =
e
2
4
0
hc
Qc
480a
3
0
=
Qc
480a
3
0
1 MHz
where 1/137.036 is the ne structure constant, and where we have put in
the numerical value of Q. This nuclear quadrupole interaction with the electric
eld of the electron cloud typically gives rise to radio frequency resonances (in
the low megahertz range) that may be detected using the process of nuclear
quadrupole resonance (NQR). Since NQR is sensitive to the electronic structure
(ie chemical bonds), it has seen some application towards explosives detection.
In particular, nitrogen is a common element in many explosives, and since
14
N
has a non-zero quadrupole moment, NQR can be used to detect what sorts of
nitrogen compounds may be present in a sample.
4.8 A very long, right circular, cylindrical shell of dielectric constant /
0
and inner and
outer radii a and b, respectively, is placed in a previously uniform electric eld E
0
with its axis perpendicular to the eld. The medium inside and outside the cylinder
has a dielectric constant of unity.
a) Determine the potential and electric eld in the three regions, neglecting end
eects.
Since the cylinder is very long, we treat this as a two-dimensional problem. In
this case, the potential admits a general expansion
=

m
[
m

m
+
m

m
] cos(m
m
)
(where the m = 0 term should actually be
0
+
0
log ). Furthermore, by
orienting the electric eld along the +x direction, we may use the
symmetry of this problem to eliminate the phases
m
. As a result, we are able to
write the potential as an expansion in each of the three regions
=
_

1
= A
0
+

m
A
m

m
cos(m) E
0
cos , > b

2
= B
0
+ C
0
log +

m
(B
m

m
+ C
m

m
) cos(m), a < < b

3
= D
0
+

m
D
m

m
cos(m), < a
For each value of m, there are four unknowns, A
m
, B
m
, C
m
and D
m
. On the other
hand, there are also four matching conditions (D

and E

both at a and at b).

Note, however, that when m = 1 these matching conditions yield homogeneous
equations which only admit the trivial solution
A
m
= B
m
= C
m
= D
m
= 0 m = 1
(Although the m = 0 case has to be treated separately, it is easy to see that
C
0
= 0. The remaining constants must satisfy A
0
= B
0
= D
0
, and may be taken
to vanish, since an overall constant added to the potential is unphysical.) Thus
we may focus on m = 1 and write
=
_
_
_

1
= (A
1
E
0
) cos , > b

2
= (B + C
1
) cos , a < < b

3
= D cos , < a
(3)
We may obtain the electric eld by taking a gradient
E

=
_
_
_
E
1

= (A
2
+ E
0
) cos , > b
E
2

= (B + C
2
) cos , a < < b
E
3

= Dcos , < a
E

=
1

=
_
_
_
E
1

= (A
2
E
0
) sin , > b
E
2

= (B + C
2
) sin , a < < b
E
3

= Dsin , < a
(4)
The matching at = a is

0
E
3

= E
2

=a
, E
3

= E
2

=a
or
(
0
/)D B + Ca
2
= 0, D B Ca
2
= 0
This may be solved for C and D in terms of B
C =
1
0
/
1 +
0
/
Ba
2
, D =
2
1 +
0
/
B (5)
Similarly, the matching at = b is
E
2

=
0
E
1

=b
, E
2

= E
1

=b
or
(
0
/)Ab
2
+ B Cb
2
= (
0
/)E
0
, Ab
2
B Cb
2
= E
0
Eliminating C using (5) gives rise to the simultaneous equations
_
b
2
1
1
0
/
1+
0
/
_
a
b
_
2
(
0
/)b
2
1
1
0
/
1+
0
/
_
a
b
_
2
__
A
B
_
= E
0
_
1

0
/
_
This yields a solution
A = E
0

1
(1
0
/)
_
1
_
a
b
_
2
_
b
2
B = E
0

1
(2
0
/)
where
= (1 +
0
/)
_
1
_
1
0
/
1 +
0
/
a
b
_
2
_
is b
2
times the determinant of the above matrix. Substituting B into (5) then
gives the remaining coecients
C = E
0

1
(1
0
/)2
0
/
1 +
0
/
a
2
D = E
0

1
4
0
/
1 +
0
/
These expressions may be simplied to read
A = E
0
b
2
(
2

2
0
)(b
2
a
2
)
( +
0
)
2
b
2
(
0
)
2
a
2
B = 2E
0

0
( +
0
)b
2
( +
0
)
2
b
2
(
0
)
2
a
2
C = 2E
0
a
2

0
(
0
)b
2
( +
0
)
2
b
2
(
0
)
2
a
2
D = 4E
0

0
b
2
( +
0
)
2
b
2
(
0
)
2
a
2
(6)
The potential and electric eld are obtained by substituting these coecients into
(3) and (4). For the potential, we have
b < :
1
= E
0
_
(
2

2
0
)(b
2
a
2
)
( +
0
)
2
b
2
(
0
)
2
a
2
b
2

_
cos
a < < b :
2
= 2E
0

0
b
2
[( +
0
) + (
0
)a
2
/]
( +
0
)
2
b
2
(
0
)
2
a
2
cos
< a :
3
= 4E
0

0
b
2

( +
0
)
2
b
2
(
0
)
2
a
2
cos
(7)
b) Sketch the lines of force for a typical case of b 2a.
For /
0
= 1.5, the electric eld lines look like
Note that we have actually plotted the electric displacement eld

D, as Gauss
law in vacuum



D = 0 ensures that the lines of electric displacement are
continuous and unbroken. The electric eld lines themselves are discontinuous at
the interface between dielectrics.
c) Discuss the limiting forms of your solution appropriate for a solid dielectric cylin-
der in a uniform eld, and a cylindrical cavity in a uniform dielectric.
A solid dielectric cylinder of radius b may be obtained by taking the limit a
0. In this case the expressions (6) and (7) simplify considerably. We give the
potential
=
_

1
= E
0
x + E
0
1
0
/
1+
0
/
b
2
x

2
, > b

2
= E
0
2
0
/
1+
0
/
x, < b
(8)
where x = cos . The potential
3
is irrelevant in this case. Here we see that the
potential
2
inside the cylinder is uniform (but corresponds to a reduced electric
eld provided >
0
). The potential outside is that of the original uniform electric
eld combined with a two-dimensional dipole.
For the opposite limit, we obtain a cylindrical cavity of radius a by taking the
limit b . In this case, we end up with
=
_

2
= E
0
2
0
/
1+
0
/
x E
0
2
0
/(1
0
/)
(1+
0
/)
2
a
2
x

2
, > a

3
= E
0
4
0
/
(1+
0
/)
2
x, < a
At rst glance, this appears to be considerably dierent from (8). However, note
that the physical electric eld we measure as is

E
0
= E
0
(2
0
/)/(1+
0
/).
In terms of

E
0
, we have
=
_

2
=

E
0
x

E
0
1
0
/
1+
0
/
a
2
x

2
, > a

3
=

E
0
2
1+
0
/
x, < a
which may be rewritten as
=
_

2
=

E
0
x +

E
0
1/
0
1+/
0
a
2
x

2
, > a

3
=

E
0
2/
0
1+/
0
x, < a
This agrees with (8) after the replacement
0
(and a b), as it must.
4.9 A point charge q is located in free space a distance d from the center of a dielectric
sphere of radius a (a < d) and dielectric constant /
0
.
a) Find the potential at all points in space as an expansion in spherical harmonics.
By symmetry, we may place the point charge on the z-axis at z = d. In this
case, the problem is azimuthally symmetric, and we may expand the potential
in Legendre polynomials instead of spherical harmonics. For the potential inside
the dielectric sphere, we take

in
=
q
4

l
_
r
a
_
l
P
l
(cos ) (9)
where the q/4 prefactor is taken for convenience (but can be absorbed into a
redenition of
l
if so desired). Note that we do not need any source term, since
there are no charges inside the sphere. On the other hand, the solution outside
the sphere is given by

out
=
1
4
0
q
|x d z|
+
0
where
0
is a homogeneous solution to Laplaces equation,
2

0
= 0. Expanding
in Legendre polynomials allows us to write

out
=
q
4
0

l
_
r
l
<
r
l+1
>
+
l
_
a
r
_
l+1
_
P
l
(cos ) (10)
Note that r
<
= min(r, d) and r
>
= max(r, d). Since we must match the parallel
electric eld and perpendicular electric displacement at r = a, we may take r
<
= r
and r
>
= d when using
out
in the matching equations. For the parallel electric
eld, we have
E
in

=
1
r

in

r=a
=
q
4

l
a
P

l
(cos ) sin
E
out

=
1
r

out

r=a
=
q
4
0

l
_
a
l1
d
l+1
+

l
a
_
P

l
(cos ) sin
Matching these gives

l
=

0
_
a
l
d
l+1
+
l
_
(11)
On the other hand, for the perpendicular electric displacement, we have
D
in
r
=

in
r

r=a
=
q
4

l
l
l
a
P
l
(cos )
D
out
r
=
0

out
r

r=a
=
q
4

l
_
la
l1
d
l+1

(l + 1)
l
a
_
P
l
(cos )
Matching gives

l
=
a
l
d
l+1

(l + 1)
l

l
(12)
Solving (11) and (12) yields

l
=
2l + 1
l +

0

(l + 1)
a
l
d
l+1

l
=
(

1)l
l +

0

(l + 1)
a
l
d
l+1
As a result, the interior and exterior potential, given by (9) and (10), has the
form

in
=
q
4d

l
2l + 1
l +

0

(l + 1)
_
r
d
_
l
P
l
(cos )

out
=
q
4
0

l
_
r
l
<
r
l+1
>
+
(

1)l
l +

0

(l + 1)
a
2l+1
(rd)
l+1
_
P
l
(cos )
(13)
Note that, without the dielectric sphere (so that =
0
), this reduces to

in
=
q
4
0

l
r
l
d
l+1
P
l
(cos )

out
=
q
4
0

l
r
l
<
r
l+1
>
P
l
(cos )
which is simply the free space result
=
q
4
0
|x d z|
b) Calculate the rectangular components of the electric eld near the center of the
sphere.
Near the center, we may expand
in
in (13).

in
=
q
4d
_
1

+
3
1 + 2

r
d
cos +
5
2 + 3

_
r
d
_
2
3 cos
2
1
2
+
_
=
q
4
0
d
_
1 +
3
2 +

0
z
d
+
5
3 + 2

0
3z
2
r
2
2d
2
+
_
The electric eld is then

E =

in
=
q
4
0
d
2
_

3
2 +

0
z +
5
3 + 2

0
x x + y y 2z z
d
+
_
Very close to the center, the eld is nearly uniform, and pointed in the z direc-
tion (assuming q > 0). The presence of the dielectric modies the point charge
result

E = (q/4
0
d
2
) z by the factor 3/(2 + /
0
).
c) Verify that, in the limit /
0
, your result is the same as that for the
conducting sphere.
For /
0
, the potential (13) reduces to

in
=
q
4
0
d

out
=
q
4
0
_

l=0
r
l
<
r
l+1
>

l=1
a
2l+1
(rd)
l+1
_
P
l
(cos )
=
q(a/d)
4
0
r
+
q
4
0

l=0
_
r
l
<

a
2l+1
r
l+1
<
_
1
r
l+1
>
P
l
(cos )
=
q
4
0
_
a/d
r
+
1
|x d z|

a/d
|x (a
2
/d) z|
_
which is indeed the correct result for a conducting sphere. Note that the l = 0
term in the sum had to be treated with care when taking the limit /
0

. This results in the q(a/d)/4
0
r term corresponding to an uncharged (and
ungrounded) conducting sphere having non-zero potential when the charge q is
brought near it.