Electromagnetic Waves in Various Media

Dr. Rohit Narula

September 3, 2016

Our starting point is the di↵erential form of Maxwell’s equations in the SI

unit convention, which constitute a set of coupled, first-order, partial di↵erential
equations for both E and B. In their most general form they can be expressed
as:

r·D = ⇢, (Gauss’ law of electricity)

r·B = 0, (Gauss’ law of magnetism)
@B (1)
r⇥E = , (Faraday’s law of induction)
@t
@D
r⇥H = J + , (Ampere’s law)
@t
where the charge density ⇢ is the free or external charge density and the J
is the free current density.

1 Electromagnetic Waves in Vacuum

Given that in vacuum there is no charge ⇢ and therefore no possibility of current
J too, the Maxwell’s equation of Eq. 1 reads,

r·E = 0,
r·B = 0,
@B (2)
r⇥E = ,
@t
@E
r⇥B = µ0 ✏ 0
,
@t
where we have used the fact that for vacuum D = ✏0 E and B = µ0 H.
Applying the curl operator r⇥ to Ampere’s and Faraday’s laws we get,

1
 @B
r ⇥ (r ⇥ E) = r(r · E) r2 E =r⇥( )
@t
@ @2E
= (r ⇥ B) = µ✏0 2 ,
@t @t (3)
@B
r ⇥ (r ⇥ B) = r(r · B) r2 B = r ⇥ (µ0 ✏0 )
@t
@ @2B
= (r ⇥ E) = µ0 ✏ 0 2 .
@t @t
Using r · E = 0 and r · B = 0, we get,

@2E 2 @2B
r2 E = µ0 ✏ 0 , r B = µ 0 ✏ 0 . (4)
@t2 @t2
Eq.4 are second-order equations for both E and B but E and B are decoupled.
Both these equations satisfy the three-dimensional wave equation,

1 @2f
r2 f = , (5)
v 2 @t2
where v is the velocity of the wave. Eq.5 suggests that the EM waves de-
scribed by Eq.4 travel at a velocity,
1
v=p = 3 ⇥ 108 m s 1
, (6)
µ0 ✏ 0
which is the velocity of light c.
Since an arbitrary wave-packet can be constructed by a suitable superposi-
tion of plane-waves, we can try solutions that are plane-waves travelling in the
(without loss of generality) z-direction,

E(z, t) = E0 exp{i(kz !t)}, B(z, t) = B0 exp{i(kz !t)}. (7)

On invoking the condition that r · E = 0 and r · B = 0 we find that,

(E0 )z = (B0 )z = 0, (8)

which shows that both the electric field and magnetic amplitudes are trans-
verse to the direction of wave propagation k̂.
Invoking Faraday’s Law from Eq.2 we can also see that the E and B are in
phase and mutually perpendicular, as well as jointly perpendicular to the wave
vector k.

(k ⇥ E0 )
B0 = . (9)
!

2
 It’s important to get a feel for the numbers related to the frequencies (wave-
lengths) of the various parts of the electromagnetic spectrum (See Fig.1). For
instance, visible radiation extends roughly from 400 THz to 750 THz (400 nm
to 800 nm). Since E = h⌫ for a photon, another popular unit for describing
the energy and the closely-related frequency ⌫ is the electron-Volt (eV ) which
is obtained by dividing the energy of the photon by the electronic charge e. For
example, visible red corresponds to roughly 2 eV.

1.1 Energy and Momentum in Electromagnetic Waves in

Vacuum
The energy stored per unit volume in an electromagnetic field u is given by the
familiar expression,

3
 1 1
u(z, t) = (✏0 | E |2 + | B |2 ) (10)
2 µ0
For a monochromatic plane-wave the electric and magnetic contributions are
equal,
1
| B |2 = | E |2 , (11)
c2
and so we get,

u = µ0 ✏0 | E |2 = ✏0 E0 2 cos2 (kz !t + ). (12)

The Poynting vector S that gives the energy flux density, or energy per unit
area, per unit time transported by the fields is,
1
S= (E ⇥ B) = c✏0 E0 2 cos2 (kz !t + ) = cuẑ. (13)
µ0
The momentum density stored in a field P is given by,
1 1 u
P= S = ✏0 E0 2 cos2 (kz !t + ) = ẑ. (14)
c2 c c

1.2 Calculating Time-Averages

Given the fact that the frequency of visible light ( 600 THz) is much higher
than the time period of most macroscopic measurements, we are usually only
interested in the time-average of the quantities that derive from the fluctuating
cosine term cos2 (kz !t ).
Using the mathematical fact that,
Z T
1 1
cos2 (kz !t )dT = , (15)
T 0 2
we get,
1
✏0 E0 2 ,
hui =
2
1
hSi = c✏0 E0 2 ẑ, (16)
2
1
hPi = ✏0 E0 2 ẑ.
2c
The intensity or the average power per unit area transported by an electro-
magnetic wave is given by,
1
✏0 E0 2 .
I = hSi = (17)
2
The momentum transfer is p = hPiAc t, so the radiation pressure or the
average force per unit area exerted on a perfect absorber is given by,

4
 1 p 1 I
P = = ✏0 E0 2 = , (18)
A t 2 c
which of course doubles if the material in question is perfectly reflecting.

2 Electromagnetic Waves in Linear, Homoge-

nous Media
In electromagnetic wave theory linear media is taken to mean that the con-
stitutive relations that describe material response/properties (needs to calcu-
lated/determined separately) are given by,

D = ✏0 E + P = ✏0 ✏r E =, ✏E
1 1 1 (19)
H= B+M = B = H.
µ0 µ0 µr µ
Here both ✏ and µ are scalars, while in a more general non-isotropic material
they are described by a tensor.
The term homogenous refers to the fact that the material properties do not
vary from point to point in the material. Of course, real materials are made
of a collection of atoms and there is certainly a varying of electron density, for
example. For homogeneity to apply, the length scale of such a variation (⇠
nm) must be much smaller than the wavelength of the electromagnetic field
in question (⇠ µm).
For the linear, isotropic and homogenous case, the Maxwell’s equations re-
duce to:

r·E = 0,

r·B = 0,
@B (20)
r⇥E = ,
@t
@E
r⇥B = µ✏ .
@t
The only di↵erence between Eqs. 2 for the vacuum and Eqs. 20 for the linear,
isotropic and homogenous media is the prefactor µ✏.
Thus, the velocity of wave propagation v through a linear, homogenous ma-
terial moves at speed,
1 c
v=p = , (21)
µ✏ n
where n is the familiar refractive index of a material.

5
3 Reflection and Transmission at Normal Inci-
dence
Let us first consider the various boundary conditions at the interface that will
help us relate both the electric and magnetic fields right at the interface.

3.1 Normal component (B? )

Using Coulomb’s Law r · B = 0 we integrate over the entire ”pillbox”, given

by the red boundary in Fig. 3.1. Applying Gauss’ theorem we have,
Z I
r · BdV = B · dS = 0. (22)
V S
We can calculate the surface integral in Eq. 22 over the flat ends (S1 and
S2 ) and the curved part (S3 ) via,
Z Z Z
B · dS + B · dS + B · dS = 0. (23)
S1 S2 S3
As we shrink the ”pillbox” dimension h we observe that the contribution in
Eq. 23 due to S3 must go to zero, giving us,

B1? A + B2? A = 0, (24)

leading to,

B1? = B2? . (25)

6
3.2 Tangential component (Ek )

Using Ampere’s Law,

@B
r⇥E = , (26)
@t
and integrating over the loop shown in Fig.3.2 and applying Stokes’ theorem
we get,
Z I Z
@
r ⇥ E · dS = E · dl = B · dS. (27)
S C @t S
Again, shrinking the loop in Fig.3.2 we get,

E1k = E2k (28)

7
3.3 Normal component (D? )

Using Gauss’ Law r · D = ⇢f we integrate over the entire ”pillbox”, given

by the red boundary in Fig. 3.3. Applying Gauss’ theorem we have,
Z I Z
r · DdV = D · dS = ⇢f dV. (29)
V S V

We can calculate the surface integral in Eq. 29 over the flat ends (S1 and
S2 ) and the curved part (S3 ) via,
Z Z Z
D · dS + D · dS + D · dS = ⇢S A. (30)
S1 S2 S3

As we shrink the ”pillbox” dimension h we observe that the contribution in

Eq. 30 due to S3 must go to zero, giving us,

D1? A + D2? A = ⇢S A, (31)

leading to,

D1? + D2? = ⇢S , (32)

where ⇢S is the surface free charge density at the interface, which is zero for
a dielectric.

8
3.4 Tangential component (Hk )

Using Faraday’s Law,

@D
r⇥H=J + , (33)
@t
and integrating over the loop shown in Fig.3.4 and applying Stokes’ theorem
we get,
Z I Z Z
@
r ⇥ H · dS = H · dl = J · dS + D · dS. (34)
S C S @t S
Again, shrinking the loop in Fig.3.4 we get,

H1k l H2k l = Js? l (35)

and finally,

H1k H2k = Js? (36)

where Js? is the surface free current at the interface which is of course zero
for a dielectric.

9
 We now consider an interface between two linear, homogenous media labelled
as 1 and 2 as shown in Fig. 3.4 and write out the terms for incident (I) E and
B fields,

EI (z, t) = E0I exp{i(k1 z !t)}x̂,

1 (37)
BI (z, t) = E0I exp{i(k1 z !t)}ŷ,
vI
the reflected (R) fields,

ER (z, t) = E0R exp{i( k1 z !t)}x̂,

1 (38)
BR (z, t) = E0R exp{i( k1 z !t)}ŷ,
v1
and finally, the transmitted (T ) fields,

ET (z, t) = E0R exp{i(k2 z !t)}x̂,

1 (39)
BT (z, t) = E0R exp{i(k2 z !t)}ŷ.
v2
These are combined with the boundary conditions derived above in Eqs. 30,
28, 32 and 36 that after some algebra lead to relationships between the reflected
(E0R ) and transmitted (E0T ) amplitudes, with the known incident amplitude
(E0I ).
We finally get,
µ1 v1
= ,
µ2 v2
1
E0R =( )E0I , (40)
1+
2
E0T =( )E0I .
1+

10
 Further, we can calculate the reflection coefficient R and the transmission
coefficient T as,
✓ ◆2
IR E0R 2 n1 n2
R= =( ) = ,
II E0I n1 + n2
(41)
IT E0T 2 4n1 n2
T = =( ) = .
II E0I (n1 + n2 )2
Clearly, R + T = 1, as expected from energy conservation.

4 Reflection and Transmission at Oblique Inci-

dence

Now let us consider that the incoming wave (EI ) impinges on the surface at
at an oblique angle as shown Fig. 4. Following a similar procedure as for the
case of normal incidence above, but after much more algebra we get,

11
 p
cos(✓T ) 1 [(n1 /n2 ) sin ✓I ]2
↵= =
cos(✓I ) cos(✓I )
µ1 v1
= ,
µ2 v2
(42)
↵
E0R = ( )E0I (Fresnel’s equation),
↵+
2
E0T = ( )E0I (Fresnel’s equation).
1+

There is an angle (✓B ), called the Brewster’s angle at which the reflected
wave is completely absent (See Fig. 4) and is given by:
2
1
sin2 (✓B ) = 2 2
(43)
(n1 /n2 )
Further, we can calculate the reflection coefficient R and the transmission
coefficient T as,
✓ ◆2
IR E0R 2 ↵
R= =( ) = ,
II E0I ↵+
✓ ◆2 (44)
IT E0T 2 2
T = =( ) =↵ ,
II E0I ↵+

12
 that are plotted in Fig. 4.

5 Absorption and Dispersion in Conducting Me-

dia
Writing the Maxwell’s equations again we have,

r·D = ⇢,
r·B = 0,
@B (45)
r⇥E = ,
@t
@D
r⇥H = .J +
@t
Since we are now dealing with a conductor we invoke the well-known Ohm’s
Law (J = E) that describes a material property and is often a good phe-
nomenological description of material behaviour particularly at low values of
electric field E. Thus, we can rewrite Maxwell’s equations above in Eqs. 46 as,
⇢
r·E = .
✏
r·B = 0,
@B (46)
r⇥E = ,
@t
@E
r⇥B = µ( E + ✏ ).
@t
Using the equation of continuity,
@⇢
r·J + = 0, (47)
@t

13
 and Ohm’s Law we get for a linear, homogenous material,
@⇢
= (r · E) = ⇢, (48)
@t ✏
from which it follows,
( /✏)t
⇢(t) = exp ⇢(t = 0). (49)
Eq. 49 states that the charge density ⇢(t) dissipates exponentially with a
time-constant ⌧ = ✏ . For a good conductor, its conductivity may be consid-
ered extremely large, and taking the limit ! 1, ⌧ ! 0 and therefore the
charge density ⇢(t) quickly diminishes to zero. In the steady-state where all the
free charges have reached the surface of the conductor we can again rewrite the
Maxwell’s equtions as,

r·E = 0,
r·B = 0,
@B (50)
r⇥E = ,
@t
@E
r⇥B = µ( E + ✏
).
@t
After applying the curl operator to Ampere’s and Faraday’s Law above we
get a modified wave equation for E and B:

@2E @E
r2 E = µ✏
+µ ,
@t2 @t (51)
@2B @E
r2 B = µ✏ 2 + µ .
@t @t
These still admit plane wave solution of the form:
e
E(z, t) = E0 expi(kz !t)
,
e
(52)
B(z, t) = B0 expi(kz !t)
,
where,

e
k = k + i,
r r !1/2
✏µ ⇣ ⌘2
k=! 1+ +1
2 ✏! (53)
r r !1/2
✏µ ⇣ ⌘2
=! 1+ 1 .
2 ✏!

The imaginary part of e

k  leads to attenuation of the wave as,

14
 z i(kz !t)
E(z, t) = E0 e e ,
z i(kz !t)
(54)
B(z, t) = B0 e e ,
where E0 = E0 ei E and B0 = B0 ei B are complex.
The skin depth d = 1/ denotes the distance required to attenuate to 1/e of
its original value, and describes how far the wave penetrates into the conductor.
The real part of ek k determines wavelength, the propagation speed, and the
index of refraction as:
2⇡
, =
k
!
v= , (55)
k
ck
n= .
!
While both the fields E and B remain transverse to the direction of wave
propagation k̂, they do not remain in phase with each other.
Expressing e k as a complex number ek = Kei we can show that,
s r
p ⇣ ⌘2
K =| ek |= k 2 + 2 = ! ✏µ 1 + . (56)
✏!
= arctan(/k).

The complex amplitudes E0 and B0 turn out to be related via,

Kei
B0 ei B
E0 e i E ,
= (57)
!
and comparing the imaginary part of the above equation we find that,

B E= . (58)
On the other hand, the real part of Eq. 57 implies that
s r
B0 K ⇣ ⌘2
= = ✏µ 1 + . (59)
E0 ! ✏!

15
6 Dispersion and Group Velocity
6.1 Dispersion

Dispersion describes material behaviour in which the frequency of wave prop-

agation, say ! is a function of wavelength , or alternatively the wave vector
k, such that we have a dependence !(k). A familiar manifestation of dispersion
is the separation of white light into its constituents by a glass prism. This is
because the refractive index n depends on the frequency of the incident light.
More generally, all the material properties of a material such as permittivity
✏, permeability µ and the conductivity depend of frequency thus exhibiting
dispersion.

6.2 Group Velocity

Let’s imagine a wave packet a(x, t = 0) consisting of various wavelengths

but centred around 0 . We can Fourier decompose this wave packet into it’s
spatial frequency constituents eikx as,
Z 1
a(x, t = 0) = dkA(k)eikx , (60)
1

that is simply a sum of plane waves eikx each with amplitude A(k).
Now, let’s propagate these waves through time t. This works by simply
propagating each wave A(k)eikx by a phase factor of e i!t . Here ! is also
centred around a frequency !0 = !(k0 ). This gives us,
Z 1
a(x, t) = dkA(k)eikx !t . (61)
1

Expanding !(k) in a Taylor series we get:

16
 @!(k)
!(k) ⇡ !0 + |k=k0 (k k0 ) (62)
@k
This leads to rewriting Eq. 63 as,
Z 1
@!(k)
a(x, t) = eik0 x !0 t dkA(k)ei(k k0 )(x @k t) . (63)
1

The term eik0 x !0 t describes a perfect monochromatic wave with wavevector

k0 that is moving with phase velocity !0 /k0 within the envelope of the wave
R1 @!(k)
packet. The other term 1 dkA(k)ei(k k0 )(x @k t) describes the movement
of the wavepacket as a whole. As it depends on the position and time only
through the combination (x @!(k) @k t), we can deduce that the group velocity vg
of the wave packet is,

@!(k)
vg = . (64)
@k

17