1 Electromagnetic field by a localized source (lecture

4)
Consider the electromagnetic fields caused by time-dependent charge and current densities
localized in a constrained region of space. Here we will mainly study the fields by an electric
dipole. Later, the analysis is extended to the full multipole expansion.
Assume harmonic time dependence e−iωt —arbitrary time dependences can be dealt with using
Fourier analysis of their components. The charge density ρ and current density j are
ρ(x, t) = ρ(x)e−iωt
j(x, t) = j(x)e−iωt
and the physical quantities correspond to the real parts of the comelx quantities. The elect-
romagnetic potentials and fields are also time-harmonic and the sources are assumed to be
located in an otherwise empty space.
Let us start from the vector potential A in Lorentz gauge,
j(x0 , t0 ) 0 |x − x0 |
Z Z
µ0 3 0
A(x, t) = dx dt0 δ(t + − t) (1)
4π |x − x0 | c
and, by writing A(x, t) = A(x)e−iωt , we obtain
0
eik|x−x |
Z
µ0 ω
A(x) = d3 x0 j(x0 ) , k= (2)
4π |x − x0 | c
The magnetic field is, according to definitions, H = µ10 ∇ × A and, outside the source region,
the electric field equals E = iζk0 ∇ × H, where ζ0 = µ0 /0 is the impedance of free space.
p

When the current density j(x0 ) is given, the electromagnetic field can be calculated from the
integral above, at least in principle. Let us study the case where the source region (size d)
is much smaller than the wavelength: d  λ = 2πc/ω. We can distinguish three regimes of
interest:
(i) Near zone (static regime): d  r  λ
(ii) Intermediate zone (induction regime): d  r ∼ λ
(iii) Far zone (radiation regime): d  λ  r
In the near zone (i) kr  1 and the exponential part of the integrand for the vector
potential can be set to unity, and the inverse distance can be presented using series of spherical
harmonics Ylm :
Z
µ0 X 4π Ylm (θ, ϕ)
lim A(x) = l+1
d3 x0 j(x0 )(r0 )l Ylm
∗
(θ0 , ϕ0 ) (3)
kr→0 4π l,m 2l + 1 r

1
We can see that the near fields vary harmonically in time but are static in their character: no
wave solution follows for the spatial dependence. Above, we have made use of the relation
l
1 X 1 r<
= 4π Y∗ (θ0 , ϕ0 )Ylm (θ, ϕ)
l+1 lm
(4)
|x − x0 | l,m
2l + 1 r>

In the far zone (iii), kr  1 and the exponential part of the vector potential varies strongly
and dictates the character of the vector potential. We can approximate
x x
|x − x0 | ≈ r − n̂ · x0 , n̂ = = (5)
|x| r
When the leading term is desired in kr, the inverse distance can be replaced by r. The vector
potential is of the form
µ0 eikr
Z
0
lim A(x) = d3 x0 j(x0 )e−ikn̂·x (6)
kr→∞ 4π r
Therefor, the vector potential behaves like an outgoing spherical wave (eikr /r) with angular
dependence. It can be shown that the electromagnetic field is also of the form of a spherical
wave and thus is a radiation field. (Note that this part of the analysis is vlid for localized
source regions of arbitrary size.)
Now that kd  1 the integral can further be developed into series:
µ0 eikr X (−ik)n
Z
lim A(x) = d3 x0 j(x0 )(n̂ · x0 )n (7)
kr→∞ 4π r n n!

where the magnitude for the nth term is (1/n!) d3 x0 j(x0 )(k n̂ · x0 )n and thus becomes rapidly
R

smaller with increasing n. In this case, the main contribution to radiation comes from the
first non-vanishing term in the sum.
In the intermediate zone (ii), all powers of kr need to be accounted for, and no simple
limits can be taken. The vector potential is then written with the help of the expansion for
the exact Green’s function in the form
X (1) Z
A(x) = µ0 ik hl (kr)Ylm (θ, ϕ) d3 x0 j(x0 )jl (kr0 )Ylm
∗
(θ0 , ϕ0 ) (8)
l,m

where we have made use of the expansion

0 ∞ l
eik|x−x | X (1)
X
∗
= ik jl (kr< )hl (kr> ) Ylm (θ0 , ϕ0 )Ylm (θ, ϕ) (9)
4π|x − x0 | l=0 m=−l

2
 (1)
where r< = min(r, r0 ), r> = max(r, r0 ), and jl and hl are the spherical Bessel and Hankel
functions.
Again when kd  1, the jl -functions can be approximated and the result is of the same
form as the near zone result, when the following replacement is carried out:
1 eikr
→ l+1 [1 + a1 (ikr) + a2 (ikr)2 + . . . + al (ikr)l ] (10)
rl+1 r
The coefficients ai derive from the explicit expansions of the Hankel functions. This end result
allows us to see the transition from the near-zone kr  1 static field to the far-zone kr  1
radiation field.

2 Electromagnetic field of an electric dipole

If only the first term in kd is kept in the expansion of the vector potential, one obtains
µ0 eikr
Z
A(x) = d3 x0 j(x0 ) (11)
4π r
which holds everywhere outside the source region (this follows from the intermediate-zone
results above). With the help of partial integration,
Z Z Z
d x j = − d x x (∇ · j) = −iω d3 x0 x0 ρ(x0 )
3 0 3 0 0
(12)

where the substitution term disappears (the source region is constrained) and, according to
the continuity equation, iωρ(x0 ) = ∇ · j(x0 ). The vector potential is thus

iµ0 ω eikr
A(x) = − p , (13)
4π r
d3 x0 x0 ρ(x0 ).
R
where p is the electric dipole moment p =
The electromagnetic fields are
ck 2 eikr 1
H = (n̂ × p) (1 − )
4π r ikr
1  2 eikr 1 ik eikr 
E = k (n̂ × p) × n̂ + (3n̂(n̂ · p) − p)( 2 − )
4π0 r r r r
We note that the magnetic field is always transverse but that the electric field has both
longitudinal and transverse components.

3
 In the far zone,
ck 2 eikr
H = (n̂ × p)
4π r
E = ζ0 H × n̂

which shows the typical form of a spherical wave.

In the near zone,
iω 1
H = (n̂ × p) 2
4π r
1 1
E = (3n̂(n̂ · p) − p) 3
4π0 r
The electric field is, except for the harmonic time dependence, that of a static lectric dipole.
The field ζ0 H is smaller, by a factor of kr, than the field E so, in the near zone, the field is
electric in its nature. In the static limit k → 0, the magnetic field disappears and the near
zone extends to infinity.
The power radiated by the vibrating dipole moment p as per solid angle is
dP 1
= Re(r2 n̂ · E × H∗ )
dΩ 2
c2 ζ0 4
= k |(n̂ × p) × n̂|2 ,
32π 2
where n̂ × p) × n̂ gives the polarization state. If all components of p are in the same phase,
dP c2 ζ0 4 2 2
= k |p| sin θ (14)
dΩ 32π 2
which is the typical radiation pattern of an electric dipole (θ is here measured from the
direction of p). Independently of the phases of the components for p:n, the total radiated
power is
c2 ζ0 k 4 2
P = |p| (15)
12π

3 Scattering by small spherical particles in the electric

dipole approximation
Light scattering by particles clearly smaller than the wavelength can be studied in the ap-
proximation, where the incident field induces an electric dipole moment to the particle. The

4
dipole fluctuates in a certain phase with the incident field and thus scatters radiation in di-
rections differing from the propagation direction of the incident field. In this case, the dipole
moments can be computed using electrostatic methods.
Assume that a monochromatic plane wave is incident on a small scatterer located in free
space. Let the propagation direction and polarization vector of the incident field be n̂0 and
ˆ0 :
Ei = ˆ0 E0 eikn̂0 ·x
Hi = n̂0 × Ei /ζ0
where k = ω/c and the time dependence has been assumed harmonic (e−iωt ). These fields
induce a dipole momentn p in the small particle and the particle radiates energy in (almost)
all directions. In the far zone, the scattered fields are of the form
1 2 eikr
Es = k ((n̂ × p) × n̂)
4π0 r
Hs = n̂ × Es /ζ0
where n̂ is the dirction of the observer and r the distance from the scatterer. The power
scattered in direction n̂ with polarization ˆ per unit solid angle divided by the incident flux
density is the so-called differential cross section

dσ r2 2ζ10 |ˆ∗ · Es |2
(n̂, ˆ, n̂0 , ˆ0 ) = 1 ∗ (16)
dΩ 2ζ0 0
|ˆ · Ei |2

where the complex conjugation of the polarization vectors is important for proper treatment
of circular polarization. Furthermore,
dσ k4
(n̂, ˆ, n̂0 , ˆ0 ) = |ˆ∗ · p|2 , (17)
dΩ (4π0 E0 )2
where the n̂0 , ˆ0 -dependence is implicit in p. We can see that the differential and total cross
sections of the dipole scatterer are both proportional to k 4 :een and λ−4 :een (Rayleigh’s law).
Assume that the scatterer is a small sphere (radius a) with the relative permittivity r = /0 .
According to electrostatics, the dipole moment of the sphere is
 − 1
r
p = 4π0 a3 E i (18)
r + 2
so that
dσ r − 1 2 ∗
= k 4 a6 | | |ˆ · ˆ0 |2 (19)
dΩ r + 2

5
The polarization dependence is purely that of electric dipole scattering. The scattered radia-
tion is polarized in the plane defined by the dipole moment ˆ0 and the vector n̂.
For unpolarized incident radiation, the differential cross sections in different polarization
states of the scattered field are
dσk k 4 a6 r − 1 2
= | | cos2 θ
dΩ 2 r + 2
dσ⊥ k 4 a6 r − 1 2
= | |
dΩ 2 r + 2
where θ is now the scattering angle.
The degree of polarization is
dσ⊥ dσk
dΩ
− dΩ sin2 θ S21 (θ)
P (θ) = dσk
= 2
=− (20)
dσ⊥
+ 1 + cos θ S11 (θ)
dΩ dΩ

and the differential cross section summed over the polarization states of the scattered field is
dσ r − 1 2 1
= k 4 a6 | | (1 + cos2 θ) ∝ S11 (θ) (21)
dΩ r + 2 2
where S11 (θ) and S21 (θ) are elements of the scattering matrix. The total scattering cross
section is
8π 4 6 r − 1 2
Z
dσ
σ= dΩ = k a| | (22)
(4π) dΩ 3 r + 2
The scattered radiation is 100% positively polarized at the scattering angle θ = 90◦ . It was
the polarization characteristics of the blue sky that got Rayleigh interested in scattering by
small particles.