Optics I 2017 Chapter2 PDF
Optics I 2017 Chapter2 PDF
Optics I 2017 Chapter2 PDF
Chapter 2
Propagation of light
In free space (vacuum), light propagates with constant phase velocity c. Most real situations,
however, involve propagation in matter. In that case, the perpetual interaction between the
electromagnetic wave and the atoms, molecules, or solid constituting the medium of
propagation, modifies the characteristics of the propagating wave. In particular, when the
propagating light is of frequency that matches resonant absorption due to electronic
transitions in the medium, strong attenuation of the electromagnetic field occurs.
In this chapter, however, we will mainly be concerned with light of non-resonant
frequency that is different from that of electronic transitions in the medium. In this case,
weak attenuation of the amplitude of the electromagnetic field results and, more importantly,
the phase velocity changes. In addition to the propagation in a homogeneous medium with
such non-resonant interaction, we will also formulate the phenomena of reflection and
refraction at boundaries between different homogeneous media. The case of geometrical
optics as a limiting case of light propagation will also be discussed.
1
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
p (t ) = p 0 cos(ωt )
This electric dipole radiates with a characteristic electromagnetic field (in vacuum) given by
p0 ω 2 cos ( kr − ω t ) E
E= 2
sin (θ ) ; B=
4πε 0 c r c
where θ is the polar angle. The radiation pattern of such a dipole is illustrated schematically
in Fig. 2.2.
In the radiation zone, far away from the dipole, the electric and magnetic fields are
perpendicular to each other, and the time-averaged Poynting vector becomes:
ε 0c 2 p02ω 4 sin 2 θ
I= S = E0 =
T
2 32π 2ε0c 3 r 2
Notice that the intensity varies as the inverse of the fourth power of the wavelength (
I ~ λ−4 ~ ω 4 ).
€
To cast a relationship between the amplitudes of the incident and the scattered wave,
we use a classical, damped harmonic oscillator model to describe the atom. The displacement
x (t ) of the electron (charge -e), viewed as a mass m attached to the nucleus of the atom by a
spring, then obeys the equation of motion
2
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
2 F −eE
x˙˙ + σx˙ + ω0 x = =
m m
where ω 0 is the non-damped frequency of the oscillator, σ is the damping constant and E is
the electric field providing the driving force. We treat the electromagnetic wave in the scalar
€ iωt iωt
approximation. For a monochromatic field of the form E(t ) = E0 e we expect x (t ) = x0 e ,
and by inserting into the equation of motion we get
−eE0
−ω 2 x0 + iωσx0 + ω02 x0 =
m
From these expressions for the response represented by the complex amplitude x0, we see that
the largest response in amplitude occurs at resonance ( ω = ω 0 ); this response has a
€
Lorentzian lineshape (i.e., dependence on frequency) and its peak decreases with increasing
damping σ. The frequency dependence of the amplitude and phase of the oscillator are
depicted in Fig. 2.3.
Figure 2.3: Frequency dependence of the amplitude (left) and phase (right) of a harmonic
damped oscillator. The solid line is for smaller damping, dashed for larger damping.
We now use this simple model to derive an expression for the polarization of N such
oscillators (per unit volume). Starting with the definition:
3
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
P = −Nex0 = εχ E
we obtain
Ne 2 E 0 (ω ) 1
P (ω ) =
m ω 0 − ω 2 + iσω
2
Writing the complex dielectric constant in terms of the complex refractive index n˜ (ω ) as
ε(ω )
= n˜ 2 (ω )
ε0 €
and using our result
€ε ω
( ) = 1+ Ne 2 1
ε0 ε0 m ω − ω 2 + iσω
2
0
we obtain
€
Ne2 ω02 − ω 2 − iσω
n˜ 2 (ω ) = 1+
ε 0 m (ω02 − ω 2 )2 + σ 2ω 2
€ Ne 2 ω 02 − ω 2
2
n r (ω ) = 1+
ε0 m (ω 02 − ω 2 ) 2 + σ 2ω 2
2
n 2 ω = − Ne σω
i ( )
ε0 m (ω 02 − ω 2 ) 2 + σ 2ω 2
€
4
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
The real part of the refractive index (depicted in Fig. 2.4) yields the correction to the phase
velocity in the medium filled with the given atoms/molecules, whereas the imaginary part
determines the attenuation constant.
Figure 2.4: Frequency dependence of the real part of the refractive index near the
resonant frequency ω 0 .
The simple model described in this section provides the two important elements
€
characterizing the scattering of an electromagnetic wave by a scattering center: the spatial
distribution of the scattered wave and the frequency response of the medium. Next, we use
these characteristics to describe the impact on the propagating wave in two important limiting
cases, the dilute scattering medium and the dense scattering medium (see Fig. 2.5). In a dilute
medium, the distance between scattering centers is much larger than the wavelength, whereas
it is smaller than the wavelength in a dense medium.
Figure 2.5: Schematic illustration of a plane wave impinging on a dilute (left) and a dense
(right) scattering medium. The scattering centers are represented by filled circles.
5
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Lateral scattering:
Here, the distance from the scattering center to the point of observation varies considerably
for the different wavelets. Hence, the random phases of the wavelets lead to an incoherent,
−4
weak lateral scattering, with scattered intensity that varies as λ . This Rayleigh scattering
mechanism explains, for example, the blue color of sun light scattered by particles in the
−4
atmosphere. For larger scatterers, the distinct λ wavelength dependence disappears; the
resulting wavelength-independent scattering, so-called Mie scattering, is responsible, for
example, for the white color of milk.
Forward scattering:
In this situation the number of wavelengths within the distance to the observation point does
not vary appreciably from one scattering center to another, and the scattered wavelets sum up
in a coherent way. The constructive interference in the forward direction leads to the forward
propagation of the beam.
Lateral scattering:
Since there are many scattering centers within a distance of one wavelength, many scattered
wavelets result with all possible phases, and the interference is completely destructive. There
is thus no laterally scattered wave in this case.
Forward scattering:
The forward propagating wave is the result of the coherent superposition of the scattered
wavelets. The various interfering wavelets differ in their phase. In fact, the delay in the
secondary (scattered) wave with respect to the primary (incident) one gives rise to a smaller
phase velocity, which is reflected in the value of the refractive index being larger than one
(see Fig. 2.6).
Note that the frequencies of the incident wave and the resulting forward wave are the
same ( ν 0 ). The smaller phase velocity of the propagating wave can be written then
c
V p < c ⇒ Vp =
€
n
6 €
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.6: Contributions of the primary and scattered waves to the sum of the
electromagnetic field amplitude, for two different secondary waves.
7
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.7: Application of Huygens’ principle for constructing a phase front starting from
a given one. The reconstructed phase front is the envelope of the spherical wavelets whose
origins lie on the original one.
Note that Huygens’s wavelets have the same frequency and phase velocity as those of
the primary, incident wave. Moreover, the direction of propagation of the reconstructed
wavefront is the same as that of the incident one. This conjecture is formally justified by the
more fundamental Kirchoff’s diffraction theory (see Chapter 6 of Optics I). An extension of
this principle to the case of non-isotropic media will be presented in the discussion on
birefringence, in Chapter 4 of Optics I.
As an example, we will now apply Huygens’s principle to derive the laws of
reflection and refraction of a plane wave incident on a planar boundary between two different
homogeneous media characterized by refractive indices n i and n t ≠ n i .
Reflection
€ €
In this case the incident and reflected plane waves, both of angular frequency ω , propagate in
the same medium of index ni. Thus, their wavelengths are the same: λi = λr = (2πc /ω ) /n i .
Using the Huygens construction procedure, we see that the reflected wave has its planar
wavefronts inclined by the same angle as those of the incident€one with respect to the
interface (see Fig. 2.8). We thus obtain the law of reflection for a plane wave impinging on a
planar surface: €
θi = θr
8
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.8: Application of Huygens’s construction to the reflection of a plane wave off a
planar surface.
Refraction
We now consider the wave transmitted to the other side of the interface. In this case the
wavelengths λi = (2πc /ω ) /n i and λt = (2πc /ω ) /n t of the incident and transmitted plane
waves are different. Hence, the propagation directions of the incident and transmitted waves
are in general different, a process referred to as refraction. Using the Huygens construction as
sketched in Fig. 2.9 we get the relation
€ €
λi λt
=
sin θ i sin θ t
where θ i and θ t are the angles for the incident and transmitted plane waves. Inserting the
values of the wavelengths in the two media we obtain
c ni c nt
=
sin θi sinθ t
This result is equivalent to the well-known Snell’s law describing the refraction of a light ray
at a planar boundary (see Fig. 2.10):
ni sin θ i = nt sin θ t
9
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Note that the light rays employed in geometrical optics correspond in the wave formulation to
the propagation directions, i.e., the direction normal to the phase fronts of the propagating
waves.
B
l= ∫ n( s)ds
A
where s is the curvilinear abscise measured along the trajectory. The time t it takes for a light
ray to propagate along a distance ds is related to the optical path through the relation
€
ds 1 l
t=∫ = ∫ dsn ( s) =
c /n(s) c c
In Fig. 2.11(a), the “path of choice” S is indicated by the solid curve connecting the
points A and B. The optical path length corresponding to S is stationary in the sense that
several other “similar” paths (the dashed lines almost coincident with S, in Fig. 2.11(a)) will
have almost the same optical path as that of S. Often the length of the optical path exhibits a
minimum with respect to small variations of the physical path; hence the path of choice
corresponds often to the least time for propagating between the two points.
10
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.11: Fermat’s principle: (a) A path S whose optical path length is stationary with
respect to small variations; (b) application to the problem of refraction at the boundary
between two homogeneous media.
where v1 = c /n i and v 2 = c /n 2 are the phase velocities in the two media and x is the (parallel
to the interface) distance between A and the point where the ray crosses the interface. The
rate of change of this €
time with respect to variations in x should vanish according to Fermat’s
principle. This gives the condition
€ €
dt x − (a − x)
= + =0
dx v1 h 2 + x 2 v b 2 + (a − x)2
2
which yields
sin θ1 sin θ 2
=
v1 v2
Expressing the phase velocities in term of the refractive indices, we recover Snell’s law:
n1 sin θ1 = n2 sin θ 2 .
11
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Ei ( r, t ) = E 0i cos ( ki ⋅ r − ω i t )
Er ( r, t ) = E 0r cos ( k r ⋅ r − ω r t + φr )
Et ( r, t ) = E 0t cos ( k t ⋅ r − ω t t + φt )
The boundary conditions for the electric and magnetic fields at the interface are found
using the integral version of Faraday’s law, with a contour C taken across the interface (see
Fig. 2.12):
d
∫ C E ⋅ dl = − dt ∫ S B ⋅ dA
Taking the limit where the width h of this contour vanishes we get:
h → 0 ⇒ E tang,1 − E tang,2 = 0
⇒ E tang,1 = E tang, 2
€
12
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
That is, the tangential component of the (total) electric field vector is continuous upon
crossing the interface. A similar argument, based on the integral version of Ampere’s law,
leads to the condition that the tangential component of the magnetic field H is also
continuous at the interface.
Since the projection of a vector onto a plane can be represented by the cross vector
product of that vector and the unit vector normal to the plane, we can rewrite the boundary
condition for the electric field in the form
uˆ n × (E i + E r ) = uˆ n × E t
For this relation to hold, the arguments of the electric field functions should have the same
spatial and time dependence at the interface. Thus, we should have
€
(ki ⋅ r − ω i t ) int erface = (kr ⋅ r − ω rt + φ r ) int erface = ( kt ⋅ r − ω t t + φt ) int erface
Since this relation must be valid for all values of r and t, we obtain the conditions
(1) ω i = ω r = ω t = ω
(at the interface)
(2) k i ⋅ r = k r ⋅ r + φ r = k t ⋅ r + φ t
The physical significance of the first condition is that the polarization vector, which gives rise
to the radiated reflected and transmitted fields in response to the incident field, oscillates at
€ frequency as the incident one. This condition yields
the same
k i = k r = n iω /c
(€
ki − kr ) ⋅ r = const (at the interface)
As the vector r lies in the plane of the interface, this equation defines the plane of the
interface, and we must have:
(k i − k r ) // uˆ n
Hence, this difference vector lies in the plane containing k i and k r , and we can therefore
write:
€
uˆ n × (k i − k r ) = 0
€ €
or
€ ki sinθ i = k r sinθ r
where the angles define the directions of the wave vectors (see Fig. 2.12). Finally, since the
ω
norms of these k vectors are equal, k i = k r = n i , we get:
c
€ 13
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
θi = θ r
This result is identical to that obtained using the Huygens principle in 2.2.1. Note that the
vectors {k i ,k r , uˆ n } are co-planar. The plane containing these vectors is referred to as the
plane of incidence.
The second condition also gives
€ (ki − kt ) ⋅ r = const (at the interface)
Following a similar reasoning as for the incident and reflected waves, we find
(k i − k t ) // uˆ n
which yields
€ uˆ n × (k i − k t ) = 0
and hence
€ ki sinθ i = k t sin θt
Dividing both sides by k0 = ω /c , we get the condition for the transmission angle:
ni sinθ i = nt sinθ t
This result is €
Snell’s law, also derived previously using the Huygens principle in 2.2.1. Note
that the vectors {k i ,k t , uˆ n } are co-planar as well. Thus, the vectors {k i ,k r ,k t , uˆ n } all lie in the
plane of incidence.
Fresnel’s equations relate the amplitude of the electric field of the incident wave to
those of the reflected and transmitted waves. These relations are obtained by applying the
€ conditions to the electric and magnetic field €vectors at the interface of the two
continuity
homogeneous dielectric media. We distinguish two polarization cases.
The three-dimensional geometry and the directions of the electric field vectors are illustrated
in Fig. 2.13 for the case where the electric field vectors are perpendicular to the plane of
incidence. The projections of the electric and magnetic field in the plane of incidence are
shown for this case in Fig. 2.14.
14
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.13: Geometry of the reflected and transmitted waves for electric field vectors
perpendicular to the plane of incidence.
Figure 2.14: Projections of the electric and magnetic fields in the plane of incidence for the
case of a “perpendicular electric field”.
15
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Bi B B
− cosθ i + r cosθ r = − t cosθ t
µi µr µt
∂B ∂E x ∂B
∇×E=− ⇔ =− y
∂t ∂z ∂t
and using
E x = E 0 cos(kz − ωt )
By = B0 cos(kz − ωt )
c
E0 = B
n 0
ni nt
(E 0i − E 0 r )cosθ i = E 0t cosθ t
µi µt
E + E = E
0i 0r 0t
ni n
cos θi − t cos θ t
E 0r
= µi
µt
r⊥ ≡
E 0i ⊥ n i n
cos θi + t cosθ t
µi µt
ni
2 cosθ i
E µi
t ⊥ ≡ 0t = 1 + r⊥ =
E 0i ⊥ ni n
cosθ i + t cosθ t
µi µt
The projections of the electric and magnetic fields in the plane of incidence for this case are
shown in Fig. 2.15. Here, the condition for the continuity of the tangential component of E
gives:
16
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
ni n n
E 0i + r E 0 r = t E 0t
µi µr µt
The amplitude reflection and transmission coefficients are thus given by:
nt n
cosθ i − i cos θt
E 0r
r// ≡ = µt µi
E 0i // n i n
cosθ t + t cosθ i
µi µt
and
n
2 i cos θi
E 0t µi
t// ≡ =
E 0i // n i n
cos θt + t cos θi
µi µt
Figure 2.15: Projections of the electric and magnetic fields in the plane of incidence for the
case of a “parallel electric field”.
The expressions given above for the reflection and transmission coefficients constitute the
basic Fresnel’s equations.
A special important case concerns non-magnetic media, for which we can set
µi = µt = µ0 . In this case, the amplitude reflection and transmission coefficients become:
17
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
n i cos θ i − n t cos θ t
r⊥ = n cos θ + n cos θ
i i t t
Polarization s (TM)
t = 2n i cosθ i
⊥
n i cosθ t + n t cos θ i
n t cosθ i − n i cosθ t
r// = n cosθ + n cosθ
t i i t
Polarization p (TE)
€ t = 2n i cos θ i
// n t cos θ i + n i cosθ t
Here, we have introduced the common notation for the perpendicular (s or TM polarization)
and parallel (p or TE) directions of the electric field with respect to the plane of incidence.
€
Note that the incidence and transmission angles are not independent. In fact, using
Snell’s law,
ni sinθ t
=
nt sin θi
the refractive indices can be eliminated from the expressions for the amplitude reflection and
transmission coefficients, and we finally obtain:
Normal Incidence:
In this case the incident ray is perpendicular the plane defining the interface between the two
media. We then obtain:
θi = 0 ⇒ θ t = 0
18
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
r = −r = nt − ni
// ⊥
nt + ni
2n i
t // = t⊥ =
nt + n i
This leads to the relationship between the reflection and transmission coefficients:
Note that this relationship is valid only in the absence of losses (e.g., due to scattering,
absorption) at the interface.
For non-normal incidence, Fresnel’s equations reveal a rich dependence on the angle
of incidence. We classify this dependence according to the relative values of refractive
indices involved.
External Reflection
For such reflection the incident ray arrives in the low-index medium, that is ni<nt. The
geometry and the incident and transmitted rays involved are shown in Fig. 2.16.
The dependence of the reflection and transmission coefficients on the angle of incidence is
shown in Fig. 2.17. In this case, for a particular incidence angle θ i = θ B , θi + θt = π /2 and
hence r// = 0 . Thus, only the s-polarization component is reflected; hence, the reflected field
corresponding to an incident field with both s- and p-polarization components (non-
polarized) becomes polarized. This occurs at a particular angle of incidence called the
Brewster angle. The value of Brewster’s angle can be calculated with the aid of Snell’s law:
π
ni sinθ i = nt sin − θ i = n t cos θi
2
which yields
19
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
tan θ B = n t n
i
Figure 2.17: Angular dependence of the amplitude reflection and transmission coefficients
in the case of air/glass interface (n=1.5) for external reflection.
Internal Reflection
In this case the incident ray arrives in the medium with the higher refractive index, that is
ni>nt (see Fig. 2.18). The angular dependence of the reflection and transmission coefficients
is shown in Fig. 2.19.
20
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.19: Angular dependence of the amplitude reflection and transmission coefficients
in the case of air/glass interface for internal reflection.
From Snell’s law, we now have θt > θi , and for a critical angle of incidence θi = θ c
we obtain θt = π /2 , i.e., the transmitted wave becomes a surface wave. For larger angles of
incidence, there is no real solution for the transmitted angle; this situation is referred to as
total internal reflection. Snell’s law gives directly the value of the critical angle for obtaining
such total reflection:
ni sinθ i = nt
or
θc = sin−1 (n t /n i )
As examples, for a glass/air interface one finds θc = 41.8 , and for a water/air interface one
has θc = 49.8 .
It is important to note that the reflection coefficients can be negative or positive, and
for angles beyond the total internal reflection angle become complex numbers. In general,
they can be written:
r⊥,// = r⊥,// e iφ
The phase-shift angle φ is plotted in Fig. 2.20 as a function of the angle of incidence for
several cases.
€
€
21
€
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.20: Phase shift angle of reflection coefficient as a function of angle of incidence,
for several geometries.
22
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
For practical applications, one is often interested in the fraction of optical power
reflected or transmitted at an interface between two dielectric media. To calculate these
power reflection and transmission coefficients R and T, we use the expressions for the
intensity,
1c 1
I= S T
= εE 0 2 = cnε0 E 0 2
2n 2
Denoting the cross section of the beam cut by the interface by A (see Fig. 2.21), we get for
the reflectance:
€
2
I Acosθ r I r E 0r
R≡ r = = = r2
I i Acosθ i Ii E 0i
2
I A cosθ t I t cosθ t nt cos θt E 0t n t cos θt 2
T≡ t = = = t
Ii A cosθ i Ii cosθ i ni cos θi E 0i ni cos θi
or:
I r I t cosθ t
1= + = R+ T
Ii Ii cosθ i
For the special case of normal incidence, the expressions for the reflectance and the
transmittance reduce to:
2
n − ni
R = R// = R⊥ = t
n t + ni
2
n t 2n i 4n i n t
T = T// = T⊥ = =
ni nt + ni (nt + ni )2
The transmittance and reflectance are plotted as a function of the angle of incidence in Fig.
2.22. Note that, for parallel polarization, the reflectance vanishes and the transmittance is
€ angle.
unity at the Brewster
23
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
(a)
(b)
Figure 2.22: Reflectance and transmittance for perpendicular (a) and parallel (b)
polarization directions, in the case of air/glass interface.
24
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.23: Propagation close to total internal reflection conditions at a planar interface.
E t = E0t e i( k t ⋅ r−ωt )
with
€ n
ktx =kt i sinθ i
n
t
k t ⋅ r = ktx x + k ty y = k t sin θ t ⋅ x + kt cos θ t ⋅ y
and
€
ni2
cosθ t = ± 1 − sin θ t = ± 1 − 2 sin 2 θ i
2
nt
n i2 2
cosθ t = ±i 2 sin θ i −1
€ nt
and hence
25
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
n i2 2
kty = ±ik t
sin θ i −1 ≡ ±iκ y
n t2
where we have introduced the real parameter
ni2 2
€ κ y = kt sin θ i −1
nt2
€ n
ikt i sin θ i x
i( ktx x ±iκ y y−ωt ) κ y y
E t = E0t e = E0t e e n t e−iωt
The physical solution corresponds only to the exponentially decaying field (along y): the
electric field is evanescent along the y direction (see Fig. 2.24). As a result, the field does not
carry any energy € flux (Poynting vector) perpendicular to the interface in the case of total
internal reflection. In fact, the field decays to 1/e of its value at the interface within a
(normal) distance
−1
1 n i2 sin 2 θ i
Δy = = kt −1
κ y n t2
Figure 2.24: Evanescent wave obtained under the conditions of total internal reflection.
26
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.26: Beam splitter based on frustrated total internal reflection (FTIR).
27
€
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
and k0 = 2π / λ0 = ω /c .
The function S(r ) is called the eikonal, and plays in optics the role of a potential in
classical mechanics. The eikonal is proportional to the phase, and hence surfaces of constant
value of S correspond to the wave fronts, whereas its gradient gives the directions of the light
€
rays (see Fig. 2.27):
S (r ) wave fronts
∇S ( r ) ray direction
Figure 2.27: The eikonal S and its relation with wavefronts and rays.
Notice that, for the special case of a plane wave, we have k0 S (r ) = k ⋅ r , and then
[∇ 2
+ k 2 (r)]E (r) = 0
Note that this is an equation for the vector quantity E (r) , and hence represents three similar
€ E j (r); j = x, y,z
{ }
equations for the three components . For clarity, we consider each
component separately in the derivation that follows.
To calculate ∇ 2 E , we consider €the component ∇ 2 E j and write the Laplacian
explicitly: €
€ €
28
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
∇ 2E j ≡ [ ∇ ⋅ ∇ ]E j (r)
∂ ∂ ∂ ∂ ∂ ∂
= ê x + ê y + ê z ⋅ ê x + ê y + ê z E j (r)
∂ x ∂y ∂z ∂ x ∂y ∂ z
The first operation with the gradient operator on the field component yields (using
E (r) = E0 (r)exp(−ik0 S(r)) ):
∂ ∂ ∂
ê x + ê y + ê z E0j (r)e−ik0S(r)
€ ∂x ∂y ∂z
∂E0j −ik S(r) ∂ S −ik0S(r)
= ê x e 0 − ik0E0j e
∂x ∂x
The second operation with the gradient operator on this expression yields:
= −ik0 e−ik0S(r) ∇S(r)⋅ ∇E0j (r) − ik0E0j (r)∇S(r) + e−ik0S(r) ∇ ⋅ ∇E0j (r) − ik0E0j ∇S(r)
The last term in the preceding expression can be calculated by writing the gradient operator
explicitly:
29
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
∇ 2E j (r) = −ik0 e−ik0S(r) ∇S(r)⋅ ∇E0j (r) − ik0E0j (r)∇S(r) + e−ik0S(r) ∇ ⋅ ∇E0j (r) − ik0E0j (r)∇S(r)
+e−ik0S(r) ∇ 2E0j (r) − ik0 ∇E0j (r)⋅ ∇S − ik0E0j (r)∇ 2 S(r)
= e−ik0S(r) ∇ 2E0j (r) − ik0E0j (r)∇ 2 S(r) − k02E0j (r) ∇S(r) − 2ik0 ∇E0j (r)⋅ ∇S(r)
2
2
Substituting into the Helmholtz equation, and using k 2 (r ) = k 0 n 2 (r ) , we obtain
[∇ 2
+ k 2 (r)]E0j (r)
2€
[
= e−ik0 S(r ) ∇ 2E0j (r) − ik0E0j (r)∇ 2 S(r) − k 0 E ]
0j (r) ∇S(r) − 2ik 0∇E0j (r) ⋅ ∇S(r) + k 0 n (r )E0j (r ) = 0
2 2 2
or:
€
∇ 2E0j (r) + k 0 n 2 (r )E0j (r) − k 0 E0j (r) ∇S(r) + i −2k 0∇E0j (r) ⋅ ∇S(r) − k 0E0j (r)∇ 2 S(r) = 0
2
2 2
{ }
The real part of this equation gives:
€
[ ]
∇ 2E0j (r) + k 0 n 2 (r ) − ∇S(r) E0j (r) = 0
2 2
or
€ λ 2
2
∇S E0j = n 2E0j + 0 ∇ 2E0j
2π
To obtain the case of ray optics, we take the limit of vanishing wavelength:
€ λ0 2 2
2
∇S E0j = n E0j + ∇ E0j
2
→ n 2E0j
2π λ0 →0
30 €
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
This finally yields the wave equation in the geometrical optics limit, known as the eikonal
equation:
2
∇S (r ) ≅ n 2 (r )
or
∇S (r ) ≅ n (r )
€
Thus, the norm of the gradient of the eikonal at each point in space is approximately equal to
the refractive index at that local. For a given distribution of the refractive index n(r ) , solving
€
the eikonal equation yields the wave fronts S(r ) . Then, calculating back the gradient of the
eikonal ∇S (r ) finally gives the trajectory of the optical rays.
Note that the variation in the eikonal along a certain trajectory is related to the optical
path as defined in 2.2.2. To see this, we express the variation in S across a differential
segment ds along the ray trajectory (see Fig. 2.2.8) as
dS = ∇S (r ) ds ≅ n (r ) ds
Thus, the variation in the eikonal along the ray trajectory is equal to the total optical path.
Recalling Fermat’s principle, it follows that ∇SAB /c , and hence ∇SAB , is stationary about the
actual trajectory.€
€ €
Figure 2.28: Integration path along a trajectory of a light ray for calculating the total
variation in the eikonal S.
31
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
The case of paraxial propagation can be formulated starting from the scalar spherical
wave U (r ) = U 0 e − ikr / r , which is a solution of the Helmholtz equation
(∇ 2 + k 2 )U (r ) = 0
with the wavenumber k = 2π / λ . The coordinate r represents the distance from the point
source of the spherical wave to the point where its amplitude is U(r) (see Fig. 2.29(b)). In
the paraxial approximation, the rays stay close to the principal axis of propagation, which is
defined here as the z-direction. We can therefore write:
€ 1
1 x2 + y2 2
r = (x + y + z
2 2 2
) 2
= z1+
z2
x2 + y 2 x2 + y 2
≈ z1 + = z +
2z2 2z
€
and hence the spherical wave solution can be approximated by
x2 +y2
e− ikz − ik
2z
U( r) ≈ U 0 e
x2 + y2
z+
2z
1 1 x2 + y 2 1 x2 + y 2 1
≈ 1 − = − ≈
x 2 + y 2 z 2z2 z 2 z3 z
z+
2z
x2 +y2
U − ik
U(r) ≈ 0 e− ikze 2z
≡ A(r)e−ikz
z
x 2 +y 2
U −ik
A(r) ≡ 0 e 2z
and is referred to as a paraboloidal wave, since its constant phase surfaces consist of
paraboloids of revolution (about the z axis). Note that the main z-dependence of the solution
−ikz
U is contained in the phase factor e whereas the envelope function A(r) exhibits relatively
slow variations of the amplitude along z. The phase fronts of a paraboloidal wave are
sketched in Fig. 2.29(c). They essentially consist of the segments of a spherical wave cut by a
cone that subtends a small angle about the optical axis, where the paraxial approximation
holds. €
32
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
∂2
∇ 2 = ∇ 2⊥ +
∂z 2
Using
∂2 ∂2 ∂2 A − ikz ∂A −ikz
2 U (r) ≈ 2 A(r)e −ikz
= 2 e − k 2 A(r)e −ikz − 2ik e
∂z ∂z ∂z ∂z
and the fact that the z-variation of A(r) is relatively slow over a distance of a wavelength,
i.e.,
∂2 A ∂A
€ 2
<< k << k 2 A
∂z ∂z
∂A −ikz
0 = ∇ 2U + k 2U ≅ e−ikz∇ 2⊥ A − 2ik e
∂z
and finally obtain the paraxial version of the scalar wave equation:
€ ∂A(r)
∇ 2⊥ A(r) − 2ik ≈0
∂z
Paraboloidal waves are solutions of this equation, as can be verified by direct substitution.
€
33
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.29: Schematic illustration of the wavefronts of (a) plane wave; (b) spherical wave;
(c) paraboloidal wave.
z
→ z + iz0 ≡ q ( z)
x2 + y 2
U0 −ik
A(r )
→ e 2 ( z+iz0 )
z + iz0
We now rewrite
1 z − iz 1 1
= 2 02 = 2 −i 2
q ( z) z + z0 ( z + z0 ) z ( z + z02 ) z0
2
34
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
z 2 + z02 z0 2
R( z) ≡ = z1 +
z z
2 z 2 + z02 λ λz0 z 2 z 2
2
W ( z) ≡ = 1 + 2 ≡ W0 1 +
z0 π π z0 z0
W ≡ λz0
0 π
we obtain:
1 1 λ
= −i
q ( z) R( z) πW 2 ( z)
Using these newly defined parameters, the scalar wave solution, termed a Gaussian beam,
reads:
A0
ρ2
U0 W0 −ρ 2 W 2 (z )
−ikz−ik
2R(z)
+iζ (z )
U(r) = e ⋅e
z0 W (z)
ζ (z) = tan−1 (z z0 )
where ρ 2 ≡ x 2 + y 2 .
The phase fronts and the transverse extension of a Gaussian beam propagating along
€
the z-axis are represented schematically in Fig. 2.30.
€
Figure 2.30: Schematic representation of a Gaussian beam and its phase fronts.
35
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
By direct inspection of the mathematical expressions defining the Gaussian beam, one
can derive its important features. Such beam exhibits Gaussian intensity distributions within
the x-y plane at each point z along its propagation axis (Fig. 2.31). It is further characterized
by several parameters:
defines the contours in the x-y plane that entrap 86% of the beam intensity (Fig.
2.32). The minimal beam width, located at z=0, is called the beam waist:
€
W (z = 0) = W0
(ii) The radius of curvature of the phase fronts is given by R(z) (Fig. 2.33). It is
minimal at z = ±z0 :
R( z = z0 ) = 2z0
€
It tends to infinity at z = 0 (planar phase fronts) and approaches the value for a
spherical wave centered at z=0 at large distances from the waist:
€
R(z) → z
€ z→∞
(iii) The Rayleigh range z0 represents the zone where the beam remains collimated
(i.e., of width near 2W0):
2
2z0 = 2π W0 / λ
36
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
Figure 2.32: Variation of the width of a Gaussian beam along its propagation axis.
Figure 2.33: Variation of the radius of curvature of the phase fronts of a Gaussian beam
along its propagation axis.
Figure 2.34: Width of a Gaussian beam along its propagation axis, indicating the Rayleigh
range z0.
It is interesting to note that the divergence (half) angle of a Gaussian beam is given by
(see Fig. 2.32)
W ( z) 1 1 W
θ0 ≅ = W0 2 + 2
z
→ 0 = const
z z z0 → ∞
z0
37
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
λ
θ0 =
W0 π
Thus, the product of the minimum cross section of the beam (at z = 0):
A0 = πW 02
€
and the solid divergence angle:
Ω 0 = πθ 02
is constant:
A0 Ω0 = (πW 02 ) ⋅ (πθ 20 ) = λ2
Since a Gaussian beam can be taken as a good approximation of the “real” beam generated
for example by a laser, it can be interesting to see how its properties can be manipulated by
inserting a thin lens into its path. First of all, we need to find out how a plane wave
propagates through a lens composed of a dielectric material having refractive index n and
maximum thickness d0 (see Fig. 2.35). As it turns out, the effects of the presence of such a
lens can be modeled through the introduction of a spatially dependent phase difference,
caused by the continuously varying lens thickness. In order to define this phase difference,
we must then write an expression for the length δd, as defined in Fig. 2.35:
2 2 2
1
x + y
δd( x, y ) 2
= R− R − x + y ( 2
= R 1− 1−
2
)
R2
2 2 2 2
x +y x +y
≈ R 1 −1+ =
2R 2 2R
€
38
€
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
R
From this expression, and from the fact that f = (where f is the focal length of the
n −1
lens), we can estimate the distance d traveled in the dielectric by the wavelet impinging on
the lens in the (x,y) point,
x2 + y2 x2 + y2
⇒ d ( x, y ) = d0 − δ€d = d0 − = d0 − .
2R 2 f ( n −1)
The different phase picked up by each wavelet in passing through the lens results in a
transmission coefficient, t, which is proportional to
€
k0 2 2
−i( n−1)k0 d ( x,y) −i( n−1)k0 d0
+i
2f
(x +y )
t ( x, y) ∝ e =e e .
Figure 2.35: Propagation of a plane wave through a thin lens: definition of relevant
parameters. R, n, and d0 are, respectively, the radius of curvature, the refractive index, and
the maximum thickness of the lens.
The same expression of t holds for a double structure (Fig. 2.36), since in this case
d
x2 + y2 x2 + y2
d ( x, y ) ≅ d1 + d2 − −
2R1 2R2
1 1 1 1 1
+ = ( n −1) + ≡
f1 f 2 R1 R2 f
€ x2 + y2
⇒ d ( x, y ) = d −
2 f ( n −1)
€
and
€
39
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
ik0
x 2 +y 2 1 1 1
t ( x, y ) ∝ e 2f ; = ( n −1) −
f R1 R2
Before we proceed further, we would like to remind the reader that a Gaussian beam can be
written as
A0
ρ2
U 0 W 0 − ρ 2 W 2 (z) −ikz−ik 2R(z) +iζ (z)
U(r) = e ⋅e .
z0 W (z)
In passing through a lens (Fig. 2.37), the phase of such a beam changes in the following way:
€
ρ2 ρ2 ρ2
−ikz − ik + iζ (z) → −ikz − ik + iζ (z) + ik
2R(z) 2R(z) 2f
kρ 2 1 1
= −ikz − i − + iζ (z).
2 R(z) f
Therefore, the presence of the lens introduces a variation in the radius of curvature R of the
beam, which becomes
−1
1 1
R' ( z) = − .
R(z) f
40
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
1 1
If R>f, − < 0 , so that R' ( z) < 0 . This implies that our Gaussian beam will converge
R(z) f
after going through the lens, acquiring its minimum width 2W 0 ' at a distance z’ from the lens.
We can calculate the new waist size and distance from the lens in the following manner.
Given that for the incoming
€ beam W and R can be written as
€
1€
2 2
z λz0
W ( z) = W 0 1+ z ; W 02 =
0 π
z 2
R( z) = z 1+ 0
z
R z2 R
= 1+ 02 ⇒ z02 = −1 ⋅ z 2
z z z
2
R 2 2
−1 z + z
π2 4 z R2
⇒ W = =
λ2 R 2 R
−1 z −1
z z
R R2 λ 2 R2
⇒ −1 = =
z π2 4 π 2W 4
W
λ2
z 1 R
⇒ = ⇒ z= 2
R R2λ 2 λR
1+ 2 4 1+
π W πW 2
41
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
2 2 R2 R2
z 0 = zR − z = 2
− 2
λR λ R 2
1+ 1+
πW 2 π W 2
2 2
2 λR
2 2 λR
2
R +R −R R
πW 2 πW 2 π2 4
= 2
= 2
= W0
λ R 2 λ R 2 λ2
1+ 1+
π W 2 π W 2
4
λ4
4
R4λ 4 πW 2
R 4 4
π W π 4W 4 λ R W4
W04 = 2
= 2
= 2
λ R 2 π W 2 2 π W 2 2
1+ 1+ 1+
π W 2 λ R λ R
W
⇒ W0 = 1/2
.
π W 2 2
1+
λ R
According to what we saw previously, on the right-hand side of the lens we have that
−1
R → R' = 1 − 1
R f ,
W → W ' = W
The formalism developed above €can be used to describe the effects of a thin lens on a
Gaussian apodized plane wave. Like a homogeneous plane wave, this particular kind of wave
has a well-defined propagation direction (along its wave vector), but it is characterized by a
Gaussian intensity profile across its wave front (see Fig. 2.38). For this particular case, we
have that
R' = − f πW 2 πW 0 2 z
⇒ = − = − 0
W = W 0 λR' λf f
42
€
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
and also that z = 0; 2z0 >> f . As a consequence, for the beam waist after the lens ( W 0 ' ) we
can write
W0 f f
€ W0 ' = 1
≈ W0 = W0 €
z 2 2 z0 π W02
1+ 0 λ
f
λf W0 '
= = θ0 f ≅ θ0
π W0 f
f f 2
z' = ≈ f 1+ ≈ f .
f 2 z0
1+
z0
The lens diameter, Dlens, should at least be equal to 2W0 to achieve maximum focusing. If we
assume Dlens = 2W 0 , the waist of the focused beam is
€
λf 2λ
W0' = 2 = ⋅ F# ,
πD π
€
f
where F# = is the F-number of the lens.
d
€
Figure 2.38: Focusing of an apodized plane wave (note the Gaussian intensity profile in the
x-y plane).
43
E. Kapon/Optics I (2016/2017)/Chapter 2: Propagation of Light
2.5 Summary
ε (ω ) Ne2 1
-Refractive index dispersion = n˜ 2 (ω ) = 1 +
ε0 ε 0 m ω 0 − ω 2 + iσω
2
E(r ) = E0 (r)e
−ik0 S( r )
-Eikonal
2 λ0 2 2
-Eikonal equation ∇S E0 = n E + ∇ E
2
→ n 2E
2π λ0 →0
x2 +y2
U − ik
-Parabolidal wave U(r) ≈ 0 e− ikze 2z
≡ A(r)e−ikz
€ z
∂A
-Paraxial approximation ∇ 2⊥ A − 2ik ≈0
∂z
A0
ρ2
U0 W0 −ρ 2 W 2 (z )
−ikz−ik
2R(z)
+iζ (z )
-Gaussian beam U(r) = z W (z) e ⋅e
0
ζ (z) = tan−1 (z z0 )
44