Lecture 3 - Electromagnetic Waves II 9

3. Electromagnetic Waves II

Last time, we discussed the following.

1. The propagation of an EM wave through a macroscopic media: We discussed how the wave interacts
with the media and how all of the details of inter-atomic and atom-EM wave interactions can be
described by a constitutive relation. Examples we discussed last time included the change of the speed
of the propagation, reflection and refraction, gain and loss, and tunneling through a thin slab.
2. Macroscopic Maxwell’s equations: We derived the macroscopic Maxwell’s equations. These equations
can describe all the EM interaction with macroscopic media with a linear dimension > 10 nm. The
constitutive relation can be measured from experiments. It can also be calculated by taking into account
all the microscopic interactions. For example, to the first order, the dielectric constant of a
homogeneous medium can be determined by the polarization vector P which is the total dipole moment
in the medium. We will show you how this can be done later in the class when we discuss microscopic
interaction between light and matters.
ε = ε0 + P / E (3.1)

3.1. Generation of EM waves

observer

r’
r

source

3.1.1. Antenna basics

In the discussions so far, we have only studied the behavior of a given EM wave (e.g. a plane wave) and its
interaction with macroscopic media but we have not discussed how the EM wave is generated. We will see
in this section how a time-variant current source can generate an EM wave. We will solve the wave equation
with the inclusion of the current source as follows.
K K K
∇ × ∇ × E − k 2 E = iωµ J (3.2)
Once the electric field is obtained, we can calculate the magnetic field by:

EECS 598-002 Nanophotonics and Nanoscale Fabrication Winter 2006, P.C.Ku

Lecture 3 - Electromagnetic Waves II 10

K
K ∇×E
H= (3.3)
iωµ
K
The solution to (3.2) can be written as follows for any observer at r outside the source distribution region.
K K K K K K
E ( r ) = iωµ ∫
source
G ( r , r ') ⋅ J ( r ')dr ' (3.4)

where G is a dyadic2 Green’s function which satisfies

K K
∇ × ∇ × G − k 2G = I δ ( r − r ') (3.5)
I is an identity matrix ( I = xx
ˆ ˆ + yy
ˆ ˆ + zz ˆˆ + θθ
ˆˆ = rr ˆ ˆ + ϕϕ
ˆ ˆ ). You can easily verify by inserting (3.4) into the LHS
K
of (3.2) and using (3.5), you get the RHS of (3.2). Note also that because r is outside the source, we can
interchange ∇ × ∇ × and the volume integral. Please note all the differential operators below including (3.5)
K
act on r unless otherwise specified.

To solve G , first we notice that ∇ × ∇ × G = −∇2G + ∇∇ ⋅ G . If we take the divergence of (3.5) and use the
K
vector identity ∇ ⋅ ( ∇ × A) = 0 , we get:

K K
− k 2 ∇ ⋅ G = ∇δ ( r − r ') (3.6)
Substituting (3.6) back into (3.5), we have:

⎛ ∇∇ ⎞ K K
(∇ 2
+ k 2 ) G = − ⎜ I + 2 ⎟ δ ( r − r ')
⎝ k ⎠
(3.7)

We can verify that G now can be written in terms of a scalar function:

⎛ ∇∇ ⎞
G =⎜I + 2 ⎟g (3.8)
⎝ k ⎠

where the scalar function g satisfy:

K K K K
(∇ 2
+ k 2 ) g ( r , r ') = −δ ( r − r ') (3.9)

K K K K
To solve g in (3.9), we notice that g ( r , r ') should depend only on r − r ' but not the absolute location of
K K K
r ' . Therefore we can arbitrarily set r ' at origin. After the choice of r ' , we can easily see the solution for g
must be spherical symmetric around the origin. (3.9) becomes:

d 2 g (r) dg ( r )
r2 2
+ 2r + k 2 r 2 g ( r ) = −δ ( r ) (3.10)
dr dr
The solution to (3.10) is:

eikr
g (r) = C (3.11)
r

KK
2
Dyadic G is a direct product of two vectors. For example, if G = AB , its index notation becomes
Gij = Ai B j . In matrix notation, the direct product of two vectors can be represented by a 3x3 matrix.

EECS 598-002 Nanophotonics and Nanoscale Fabrication Winter 2006, P.C.Ku

Lecture 3 - Electromagnetic Waves II 11

To determine the constant C, we integrate (3.9) over a volume including the origin and let the volume go to
zero:
K
∫ ∇ gdr = −1
2

dg
⇒ ∫ ∇g ⋅ ndS
ˆ = 4π r 2 = −1 (3.12)
S
dr r =δ → 0

1
⇒C =
4π
Combining (3.4), (3.8), (3.11) and (3.12), we have:
K K
K K ⎛ ∇∇ ⎞ eik |r − r '| K K
E ( r ) = iωµ ⎜ I + 2 ⎟ ⋅ ∫ K K J ( r ')dr ' (3.13)
⎝ k ⎠ source 4π | r − r ' |

The magnetic field is (from (3.3) and (3.13)):

K
K ∇× E
H=
iωµ
K K (3.14)
eik |r − r '| K K
= ∇× ∫
source
4π |
K K J ( r ')dr '
r − r ' |

Before we proceed, we have to remember that all the quantities in (3.13) and (3.14) are in the frequency
K
domain. We have dropped their e − iωt dependence. If J ( r ')e − iωt is a static source, ω = 0 and k = 0 .

3.1.2. General properties of near field and far field

Depending on the distance between the observer and the source, we can study two extreme cases, the
K K K K K K K
near field ( k | r − r ' | 1 ) and the far field ( k | r − r ' | 1 ). In the far field, we have | r − r ' |≈ r − rˆ ⋅ r ' . The
electric field becomes:
K K ⎛ ∇∇ ⎞ eikr K KK K
E ( r ) = iωµ ⎜ I + 2 ⎟ ⋅
⎝
∫
k ⎠ 4π r source
J ( r ')eik ⋅r ' dr ' (3.15)

The integral in (3.15) results in a function that depends only on θ and ϕ. We can define a vector current
moment as:
K K KK
K
f (θ , ϕ ) = ∫
source
J ( r ')e − ik ⋅r ' dr ' (3.16)

In the far field region, we only keep terms on the order of 1/ kr and neglect all the higher order terms.
Using:
∂ ˆ1 ∂ 1 ∂
∇ = rˆ +θ + ϕˆ (3.17)
∂r r ∂θ r sin θ ∂ϕ

(3.15) becomes:

eikr K
( )
K K
E ( r ) = iωµ I − rr
ˆˆ ⋅
4π r
f = iωµ
eikr ˆ
4π r
θ fθ + ϕˆ fϕ ( ) (3.18)

EECS 598-002 Nanophotonics and Nanoscale Fabrication Winter 2006, P.C.Ku

Lecture 3 - Electromagnetic Waves II 12

This is an outgoing wave with a spherical wave front. The electric field is perpendicular to the propagation
direction. At large distance, the wave can be approximately by a plane wave.

K K K K
In the near field, we have eik |r − r '| = 1 + ik | r − r ' | + " ≈ 1 and the observer is at a distance many times smaller
than the wavelength from the source. (3.13) becomes:
K K K
⎛ ∇∇ ⎞ J ( r ') K
E ( r ) = iωµ ⎜ I + 2 ⎟ ⋅ ∫ K K dr ' (3.19)
⎝ k ⎠ source 4π | r − r ' |

This is a quasi-static field because of the absence of the oscillating exponential term. Note the field is not
truly static since there is an implicit time harmonic factor e − iωt . Similarly, the magnetic field is:
K K K
J ( r ') K
H (r ) = ∇ × ∫ K K dr ' (3.20)
source
4π | r − r ' |
K K
Because in near field region, we have k | r − r ' | 1 , the contribution from the 2nd term in the parenthesis of
(3.19) dominates and the magnetic field can usually be neglected (because magnetic field gets
differentiation once while the electric field gets differentiation twice). We will see an example in the following
when we discuss the dipole radiator. But we notice that if magnetic field can be neglected, the solution of
fields satisfies the electrostatic equation or the Poisson equation:
K
∇⋅ E = 0 (3.21)
Or in terms of the potential φ :
∇2φ = 0 (3.22)

3.1.3. Dipole radiation

The most fundamental antenna is a Hertzian dipole which consists of a current-carrying wire with an
infinitesimal length l:
K K
ˆ δ ( r ')
J ( r ') = zIl (3.23)
Substituting (3.23) into (3.13), we get:
K K
K K ⎛ ∇∇ ⎞ eik |r − r '| K K
E ( r ) = iωµ ⎜ I + 2 ⎟ ⋅ ∫ K K zIl ˆ δ ( r ')dr '
⎝ k ⎠ source 4π | r − r ' |
⎛ ∇∇ ⎞ eikr
= iωµ ⎜ I + 2 ⎟ ⋅ ˆ
zIl (3.24)
⎝ k ⎠ 4π r
⎛ ∇ ∂ ⎞ eikr
= iωµ Il ⎜ zˆ + 2 ⎟
⎝ k ∂z ⎠ 4π r

Now we make the coordinate transformation to the spherical coordinate by

∂ ˆ1 ∂ 1 ∂
∇ = rˆ +θ + ϕˆ
∂r r ∂θ r sin θ ∂ϕ
∂ eikr ⎛ 1⎞ eikr
= ⎜ ik − ⎟ cos θ (3.25)
∂z 4π r ⎝ r⎠ 4π r
zˆ = rˆ cos θ − θ sin θ
ˆ

EECS 598-002 Nanophotonics and Nanoscale Fabrication Winter 2006, P.C.Ku

Lecture 3 - Electromagnetic Waves II 13

(3.24) becomes:

K K eikr ⎛ ⎡ i ⎛ i ⎞2 ⎤ ⎡ i ⎛ i ⎞ ⎤⎞
2

E ( r ) = −iωµ Il ⎜ rˆ2 cos θ ⎢ + ⎜ ⎟ ⎥ + θˆ sin θ ⎢1 + + ⎜ ⎟ ⎥ ⎟ (3.26)

4π r ⎜⎝ ⎣⎢ kr ⎝ kr ⎠ ⎦⎥
⎟
⎣⎢ kr ⎝ kr ⎠ ⎦⎥ ⎠

3.1.4. Near and far fields for dipole radiators

In the far field, (3.26) reduces to:

K K eikr ˆ
E ( r ) = −iωµ Il θ sin θ (3.27)
4π r
The wavefront is a spherical outgoing wave and radiation pattern consists of two side lobes with no electric
field along the z axis. The polarization of the electric field is perpendicular to the direction of the propagation.
The magnetic field can be calculated by (3.3) to be:
K K eikr
H ( r ) = −ikIl ϕˆ sin θ (3.28)
4π r

In the near field, (3.26) reduces to:

K K 1 ⎛ ⎡⎛ i ⎞2 ⎤ ⎡⎛ i ⎞ 2 ⎤ ⎞
E ( r ) = −iωµ Il ⎜ r 2 cos θ ⎢ ⎜ ⎟ ⎥ + θ sin θ ⎢⎜ ⎟ ⎥ ⎟
ˆ ˆ
4π r ⎜⎝ ⎣⎢ ⎝ kr ⎠ ⎦⎥
⎟
⎣⎢⎝ kr ⎠ ⎦⎥ ⎠ (3.29)
3
iIl ⎛ 1 ⎞
= ⎜ ⎟ ⎡ rˆ2 cos θ + θ sin θ ⎤⎦
ˆ
4πωε ⎝ r ⎠ ⎣
The field fades away quickly with r −3 in contrast to far field dependence of r −1 . Note the field is quasi-static
and has the r̂ component.

3.1.5. Radiation from a Moving Charge

The dipole radiator is one special type of radiation sources. Since the dipole source usually consists of lots
of oscillating (or moving) charges. It is interesting to study the radiation from a single moving charge. It has
applications for example in particle detectors (using Cherenkov radiation) and synchrotron radiation. The
K
charge density of a moving charge with a trajectory r0 '(t ) is:
K K K
ρ ( r ', t ) = qδ ( r '− r0 '(t )) (3.30)
The current density is therefore:
K K K
dr '(t ) K K
J ( r ', t ) = q 0 δ ( r '− r0 '(t )) (3.31)
dt
K K
To solve the electric field with (3.13), we need to convert J ( r ', t ) to the frequency domain. To do that, we
use the Fourier transform as follows.
K K K
K dr '(t ) K K KK
J ( k , ω ) = ∫ dr ' ∫ dtq 0 δ ( r '− r0 '(t ))ei (ωt − k ⋅r ')
dt
K (3.32)
dr0 '(t ) i (ωt − kK⋅rK0 '( t ))
= ∫ dtq e
dt

EECS 598-002 Nanophotonics and Nanoscale Fabrication Winter 2006, P.C.Ku

Lecture 3 - Electromagnetic Waves II 14

We consider two cases. In the first case, the particle is moving at a constant velocity along the z-axis (i.e. no
acceleration.) (3.32) becomes:
K K K K
J ( k , ω ) = 2π qv0δ (ω − k ⋅ v0 )
K
(3.33)

From (3.33), the electric field will have terms with wavevectors given by
k = ω /( v0 cos θ ) (3.34)
where θ is the angle from the z-axis to the propagation direction. Remember that the wavevector itself
needs to satisfy the Maxwell’s equations, i.e. k = ωn / c . We have:

c c
v0 = > (3.35)
n cos θ n
That is in order to generate radiation from a charge moving at a constant velocity, the velocity has to be
greater than the speed of light in the media being considered. The radiation generated in such a way is
called the Cherenkov radiation. That’s why usually a moving charge without any acceleration does not
radiate.

The second case we will consider is a charge moving with acceleration. Because of the acceleration, the
integral in (3.32) will have terms that can generate EM waves. It is generally hard to evaluate such an
integral. But in summary an accelerated charge radiates.

EECS 598-002 Nanophotonics and Nanoscale Fabrication Winter 2006, P.C.Ku