Today in Physics 217: multipole expansion

‰ Multipole expansions
‰ Electric multipoles and their moments
• Monopole and dipole, in detail
• Quadrupole, octupole, …
‰ Example use of multipole expansion as approximate
solution to potential from a charge distribution (Griffiths
problem 3.26)

-q q
q q -q q -q
q q -q
-q -q q -q q

21 October 2002 Physics 217, Fall 2002 1

 Solving the Laplace and Poisson equations by
sleight of hand
The guaranteed uniqueness of solutions has spawned several creative
ways to solve the Laplace and Poisson equations for the electric potential.
We will treat three of them in this class:
‰ Method of images (9 October).
Very powerful technique for solving electrostatics problems involving
charges and conductors.
‰ Separation of variables (11-18 October)
Perhaps the most useful technique for solving partial differential
equations. You’ll be using it frequently in quantum mechanics too.
‰ Multipole expansion (today)
Fermi used to say, “When in doubt, expand in a power series.” This
provides another fruitful way to approach problems not immediately
accessible by other means.

21 October 2002 Physics 217, Fall 2002 2

 Multipole expansions

Suppose we have a known charge distribution for which we

want to know the potential or field outside the region where
the charges are. If the distribution were symmetrical enough
we could find the answer by several means:
‰ direct calculation using Gauss’ Law;
‰ direct calculation Coulomb’s Law;
‰ solution of the Laplace equation, using the charge
distribution for boundary conditions.
But even when ρ is symmetrical this can be a lot of work.
Moreover, it may give more precise information on the
potential or field than is actually needed.
Consider instead a direct calculation combined with a series
expansion…

21 October 2002 Physics 217, Fall 2002 3

 Multipole expansions (continued)
P
If the reference point for potential is
(and can be) at infinity, then
ρ dτ ′
V (r ) =
r ∫ ,
V r
where r 2 = r + r ′2 − 2rr ′ cosθ
2
r
2
 2
 r′  r′ 
= r  1 +   − 2 cosθ 
  r  r 
  dτ ′
≡ r2 (1 + ε ) .
ρ ( r ′)
r′ θ
If point P is far away from the charge
distribution, then ε 1. V

21 October 2002 Physics 217, Fall 2002 4

 Multipole expansions (continued)
1 1 1
So consider = . But first recall this infinite series:
r r 1+ε
∞
s!
(1 + x) s
= ∑ xn
n=0 (
n! s − n)!
s s ( s − 1 ) 2 s ( s − 1 )( s − 2 ) 3
= 1+ x+ x + x +…
1! 2! 3!
where x < 1 and s is any real number. (This is one form of the
binomial theorem.) Then
1 1 1 3 2 5 3 
=  1 − ε + ε − ε + …
r r 2 8 16 

21 October 2002 Physics 217, Fall 2002 5

 Multipole expansions (continued)

1 1
2 2
1 r′  r′  3  r′   r′ 
= 1 −  − 2 cosθ  +    − 2 cosθ 
r r  2 rr  8 r   r 
3 3  2
5  r′   r′   r ′  2 r′
−    − 2 cosθ  + … =   + 4 cos θ − 4 cosθ
16  r   r   r r

1
2 2 3 4
r′ 1  r′  3  r′  3  r ′  3  r ′ 
=  1 + cosθ −   +   cos 2 θ −   cosθ +  
r  r 2 r  2 r  2 r  8 r 

5  r′ 
3  r ′ 3 4r ′ 2  r′ 
2
 r′ 
3
−     + cos θ − 4   cosθ  
16  r   r  r r r

′ 2
r′  
r
  3 2 
−2   cosθ − 8 cos θ + 8 cos θ  + …
r r  

21 October 2002 Physics 217, Fall 2002 6

 Multipole expansions (continued)

Collect terms with the same powers of r ′ r , and ignore

3
higher powers than ( r ′ r ) , for now:
1 1
2
r′ r ′
  3 2 1
= 1 + cosθ +    cos θ − 
r r  r  r  2 2

r ′
  5
3
3 3  
+    cos θ − cos θ  + …
 r  2 2  

1
2
r′  r′ 
=  P0 ( cos θ ) + P1 ( cosθ ) +   P2 ( cosθ )
r  r r

′
r 
3 
+   P3 ( cosθ ) + …
r 
21 October 2002 Physics 217, Fall 2002 7
 Multipole expansions (continued)
∞ n
1 1  r′ 
Thus, =
r r n=0  r 
∑
  Pn ( cos θ )

1 ∞
n
 r′ 
V (r ) =
r n=0
∑∫
ρ ( r )   Pn ( cosθ ) dτ ′
′
r
V
1 1
= ρ ( r ′ ) dτ ′ + 2
∫ ∫ ρ ( r ′) r ′ cosθ dτ ′ Monopole, Dipole
r r
V V
1  3 1
+ 3 ρ ( r ′ ) r  cos θ −  dτ ′
∫ ′2 2 Quadrupole
r V 2 2
1 35 3  ′
+ 4 ∫
ρ ( ) 
r ′ r ′ cos 3
θ − cos θ  dτ + … Octupole
r V 2 2 
21 October 2002 Physics 217, Fall 2002 8
 Electric multipoles

This is a useful approximation scheme, the more useful the

further away point P is from the charges within V, because
one can neglect the higher-order terms in the series after the
desired accuracy is achieved.
‰ The monopole term:
1 q
Vmonopole ( r ) = ρ ( r ′ ) dτ ′ =
∫
r r
V

If a charge distribution has a net total charge, it will tend

to look like a monopole (point charge) from large
distances.

21 October 2002 Physics 217, Fall 2002 9

 Electric multipoles

‰ The dipole term:

1 rˆ rˆ
V ( r ) = 2 ρ ( r ′ ) r ′ cosθ dτ ′ = 2 ⋅ ρ ( r ′ ) r ′dτ ′ = 2 ⋅ p ,
∫ ∫
r V r V r
where p = ∫ ρ ( r ′ ) r ′dτ ′ is called the dipole moment.
V
As usual, for surface, line and point charges, we have
n
p = σ ( r ′ ) r ′da′, p = λ ( r ′ ) r ′d ′, p =
∫ ∫ ∑ qi ri′ .
S C i =1
The simplest dipole has two point charges, ± q ,
separated by a displacement vector d that points from –q
to +q.
21 October 2002 Physics 217, Fall 2002 10
 The dipole potential and field
qd cosθ p cosθ P
Vdipole = 2
=
r r2
Edipole = −—Vdipole
θ r
 ∂Vdipole 1 ∂Vdipole ˆ 
= −  rˆ + θ 
 ∂r r ∂θ  q
d
2 p cosθ p sin θ ˆ 1 -q
= ˆ
r+ θ∝ 3
3 3
r r r
Dipole moment is defined the same way in cgs and MKS.
Expressions for potential and field still need a factor of 1 4πε 0
to convert from cgs to MKS.

21 October 2002 Physics 217, Fall 2002 11

 Quadrupole, octupole,…

A simple way to envision what the higher-order multipoles

“look like” is to construct them from the lower-order ones:
take two of the lower-order ones, invert one, and place the
two in close proximity.
-q q
q q -q q -q
q -q
-q -q q -q q
Dipole Quadrupole Octupole

The monopole moment (charge) is a scalar. The dipole

moment is a vector. Higher order multipole moments are
represented by higher-order tensors: the quadrupole moment
is a second-rank tensor, etc.
21 October 2002 Physics 217, Fall 2002 12
 Example of the use of multipole expansions

Griffiths problem 3.26: A sphere of radius R, centered at the

origin, carries charge density
R
ρ ( r ,θ ) = k 2 ( R − 2r ) sin θ ,
r
where k is a constant and r and θ are the usual spherical
coordinates. Find the approximate potential for points on the
z axis, far from the sphere.
Scheme: start by calculating the monopole term. If it’s not
zero, then it’s a good approximation to the potential, since it’s
larger by r/r’ than the dipole term. If it is zero, move on to
the dipole term. And so on…

21 October 2002 Physics 217, Fall 2002 13

 Example (continued)

Monopole moment (charge):

2π π R
1  2
q = ∫ ρ ( r ′ ) dτ ′ = kR ∫ ∫ ∫  r 2 ( R − 2r ) sin θ  r sin θ drdθ dφ
V 0 00
2π π R
= kR ∫ dφ ∫ sin 2 θ dθ ∫ ( R − 2r ) dr
0 0 0
2π π
R
= kR ∫ dφ ∫ sin θ dθ
2  Rr − r 
2
=0
 0
0 0

No net charge, so move on to the dipole term.

21 October 2002 Physics 217, Fall 2002 14

 Example (continued)

Dipole moment, or lack thereof:

p = r cosθρ ( r ′ ) dτ ′
∫
V
2π π R
1  2
= kR ∫ ∫ ∫ r cosθ  2 ( R − 2r ) sin θ  r sin θ drdθ dφ
0 00 r 
2π π R
(
= kR ∫ dφ ∫ sin 2 θ cosθ dθ ∫ Rr − 2 r 2 dr )
0 0 0
2π π R
 sin θ  3
= kR ∫ dφ   ∫
 3  0 0
Rr − 2 r (
2
dr = 0 )
0

21 October 2002 Physics 217, Fall 2002 15

 Example (continued)

Quadrupole “moment” (simple only because this is

spherical):
23 1
Q = r  cos θ −  ρ ( r ′ ) dτ ′
∫
2
 2 2
V
2π π R
=
kR
2 ∫ ∫∫r
2
( 1
)  2
3 cos θ − 1  2 ( R − 2r ) sin θ  r sin θ drdθ dφ
2
r 
0 00
2π π R
=
kR
2 ∫ ( )
dφ ∫ 3 cos2 θ − 1 sin 2 θ dθ ∫ Rr 2 − 2 r 3 dr ( )
0 0 0

kR  π   R 4
R 4 
kπ R 2 5
= ( 2π )  −   −  = ≠0
2  8  3 2  48
21 October 2002 Physics 217, Fall 2002 16
 Example (continued)

Thus, for a point way up the z axis,

1 kπ 2 R 5  1 
V (r ) ≅ 3
Q= × for the MKS answer 
z 48 z3  4πε 0 

21 October 2002 Physics 217, Fall 2002 17