Вы находитесь на странице: 1из 25

E.

Kapon/Optics I (2016/2017)/Chapter 5: Interference

Chapter 5

Interference

The superposition of several waveforms produces peculiar interference patterns that depend
on both the nature of the interfering waves as well as on the interference geometry. A close
examination of such interference patterns gives important clues on the characteristics of the
interfering waves: their wavelength, coherence, … . This chapter will present the
fundamental features of interference patterns resulting from the superposition of two or
several beams of light. The results are applied to the design of several types of
interferometers constructed specifically for analyzing the spectral components of a given
light beam.

5.1 Two-Beam Interference


We consider first the interference pattern of two linearly polarized, inhomogeneous plane
waves, represented by their complex electric field vectors (see Optics I, ch. 1.3.1):

{
E j (r, t) = Re E 0j (r )e
i(ω j t− k j ⋅r+ φ j )
} ; j = 1, 2.

Note that the amplitude of these waves depends on the position r. Although the treatment
developed will be quite general, it is useful to imagine such two beams produced in Young’s
double slit experiment (Fig. 5.1). In this geometry, a plane wave impinges on a screen in
which two thin slits are cut. The transmitted wave then consists of two cylindrical waves,
each centered at the corresponding slit. Each cylindrical wave is a special case of the
inhomogeneous plane wave referred to above. These two waves interfere on a planar screen
positioned far away (a distance of many wavelengths) from the two slits. The interference
pattern produced on the screen will be the major interest in this chapter and will be discussed
in detail for several important special cases.
Since the interference term that we will calculate arises from the scalar product
E1 ⋅ E 2 of the inhomogeneous plane waves, we will consider for simplicity two fields
polarized in the same direction. Moreover, we will also choose their wave vectors such that
k1 //k 2 , as perpendicularly oriented k-vectors will also imply normal components of the
electric fields, which do not contribute to the interference term. Generalization of the
developed formalism to the case of polarized beams is rather straightforward.

1
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

Figure 5.1: Young’s double-slit interference configuration.

5.1.1 Interference Pattern


The measurable quantity at the screen corresponds to the time average of the Poynting vector
of the sum of the fields at the point of observation:

ε 2
S = E1 + E 2
T µ T

where the averaging time T depends on the temporal response of the detector employed.
Expanding the term inside the time-averaging bracket yields:
2 2 2
E1 + E 2 = E1 + E2 + 2E1 ⋅ E2

We first evaluate the “self” terms and expand them using the explicit complex form of the
fields

2 2
2 1
E j = Re E j ( r) e (
i ω j t+φ j )
E je ( j j ) + E j*e ( j j )
i ω t+φ −i ω t+φ
=
4 ; j=1,2
1  2 2i(ω jt+φ j )
+ E j*2 e ( j j ) + 2 E j 
2
= E j e
−2i ω t+φ

4 

Supposing that each wave contains a single wavevector component, the spatial part of the
inhomogeneous plane wave is written as

E j (r ) = E j 0 e
− ik j ⋅r

(In the more general case, the field would correspond to a linear superposition of all
wavevector components of this form.) Assuming that the detector has a time response that is
much slower than the frequency of light, i.e., T >>1/ω , we obtain after time averaging

ε 2 1 ε 2 1 ε 2
Ej = E j (r) = E j0 ≡ I j ( r) ; j = 1, 2
µ T >>1/ω 2 µ 2 µ

This is just the (spatially dependent) intensity of each wave evaluated individually.
We now proceed to calculate the interference term. This term reads:

2
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

1
E 1e ( 1 1) + E 1*e ( 1 1) ⋅ E 2 e ( 2 2 ) + E 2*e ( 2 2 )
( )( )
i ω t+φ −i ω t+φ i ω t+φ −i ω t+φ
2E1 ⋅ E 2 =
2
1
E 1 ⋅ E 2 e ( 1 2 ) 1 2  + E 1* ⋅ E 2*e ( 1 2 ) 1 2  + E 1 ⋅ E 2*e ( 1 2 ) 1 2  + E 1* ⋅ E 2 e ( 1 2 ) 1 2 
i ω +ω t+φ +φ  −i ω +ω t+φ +φ  i ω −ω t+φ −φ  −i ω −ω t+φ −φ 
=
2
{ }
and its time averaging yields

1
2E1 ⋅ E 2 T >>1/ ω
= E 1 ⋅ E 2*e i( Δωt +Δφ ) + E * 1 ⋅ E 2e−i( Δωt +Δφ )
2 T

In this expression, the frequency and phase differences are defined by Δω ≡ ω1 − ω 2 and
Δφ ≡ φ1 − φ 2 , respectively. The intensity corresponding to the interference term thus becomes

ε
I1 (r ) I2 (r )e [
i Δωt +Δϕ ( r )] −i[ Δωt +Δϕ ( r )]
2E1 ⋅ E 2 T
= + I1 (r ) I2 (r )e
µ T

where the total phase difference is now defined by


€ Δϕ (r ) = (−k1 ⋅ r1 + φ1 ) − (−k2 ⋅ r2 + φ2 )
    
ϕ 1 ( r) ϕ 2 ( r)

Note that this total phase difference is produced due to the initial phase difference between
the two waves, Δφ = φ1 − φ1 (which can be time-dependent), and the phase difference
accumulated during the propagation to the interference point, which is given by
(−k1 ⋅ r ) − (−k 2 ⋅ r ) .
We can thus write the intensity distribution at the observation point as

I (r ) = I1 (r ) + I2 (r ) + 2 I1 (r ) I2 (r ) cos[Δωt + Δϕ (r,t )]
T

This result constitutes the interference pattern, as measured by a detector whose time
response is much slower than the frequency of light. It is seen to be dependent on the total

phase difference Δϕ (r ) between the two interfering beams, which depends on the difference
in propagation distance between the initial point (e.g., the slit) and the point of observation. It
can also be written in an alternative way, using the difference in propagation time between
each slit and the observation point. Approximating each beam as a plane wave, we can write
for the phase accumulated by such beam between the slit and the interference point as

ωj
k j ⋅ r j = k j rj = r = ω jτ j
c j

where τ j denotes the propagation time between each slit and the interference point. We thus
have

Δϕ ( r) = Δφ + (ω1τ1 − ω 2 τ 2 )

and the interference pattern now reads:

3
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

I (r ) = I1 (r ) + I2 (r ) + 2 I1 (r ) I2 (r ) cos[Δωt + (ω1τ1 − ω 2τ 2 ) + Δφ ]
T

It should be pointed out that for a fixed observation point the times τ 1 and τ 2 are fixed as
well; thus, the time dependence enters through the phase term Δωt , and, if the initial phases
€ on time, through Δφ ( t) .
depend
Note that the interference term depends on time, and thus the measured interference
pattern is in general non-stationary. We consider now several important special cases of two-
beam interference.

Plane Waves of Different Wavelength

For two frequencies sufficiently different such that Δω >>1/T , the interference term
averages out to zero, and we obtain:

I(r ) = I1 (r ) + I 2 (r )

That is, the interference pattern corresponds to the incoherent superposition of the two
intensity patterns. Note that the same situation holds if the difference in initial phases Δφ ( t)
varies in time rapidly enough.

Plane Waves of Equal Wavelength

In this case the interference term does not depend upon time, as the frequencies are identical,
and the initial phase difference Δφ ( t) is constant in time. We thus obtain:

I(r ) = I1 (r ) + I 2 (r ) + 2 I1 (r )I 2 (r ) cosΔϕ (r )

The interference pattern then corresponds to a constant background on which a sinusoidal


€ €
function of the accumulated phase is superimposed. This pattern is shown schematically in
€ Fig. 5.2 as a function of the phase difference.

Figure 5.2: Interference pattern of two waves of identical wavelengths but different
intensities.

For the special case of equal, uniform intensities of the two interfering waves, i.e.,
I1 (r ) = I2 (r ) = I0 , we have:

4
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

I(r ) = 2I 0 + 2I0 cosΔϕ (r ) = 2I0 [1+ cosΔ ϕ (r )]

or:

I(r ) = 4I 0 cos 2 [Δϕ (r ) 2]

The variation of the intensity across the intensity pattern can be expressed for this special
case in terms of the difference in propagated distance:

I(r ) = 2I 0 [1+ cos(kΔr)]

or, equivalently, in terms of difference in time propagation:

I(r ) = 2I 0 [1+ cos(ω Δτ )]

Plane Waves of Nearly Equal Wavelength

If the two frequencies are sufficiently close such that Δω <<1/T , the interference term does
not average out to zero. (We also assume here perfect plane waves such that the difference in
initial phases does not vary with time.) Expanding the time dependent interference term, we
get:

cos[Δωt + Δϕ (r )] = cosΔωt cosΔϕ (r ) − sinΔωt sinΔϕ (r ) ≈ cosΔϕ (r ) − sinΔωt sinΔϕ (r )

and the time average yields (to first order in ΔωT )


€ 1
cos[Δωt + Δϕ (r )] ≈ cosΔϕ (r ) − ΔωT sinΔϕ (r )
T 2

We thus obtain:
€ 1
I (r ) = I1 (r ) + I2 (r ) + 2 I1 (r ) I2 (r ) {cosΔϕ (r ) − ΔωT sinΔϕ (r )}
2

In the limit of sufficiently fast detector response, we recover the interference pattern obtained
for ω1 = ω 2.

Incoherent Quasi-Monochromatic Waves

If both waves have the same carrier frequency ω1 = ω 2 = ω , the only source for time
dependence of the interference term will be the time variation in the initial phase difference
Δφ ( t) . This is caused by the random nature of the wave trains in a quasi-monochromatic
wave (random lengths of wave trains with random phases). This results in interference of
waves originating from the same wave train if the time difference for arrival at the
observation point is smaller than the coherence time of the wave impinging on the slits (i.e.,
the average temporal length of a wave train). Otherwise, waves originating from different

5
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

wave trains will interfere. In this case, if the rate of change of mutual phase is large,
compared to the frequency response of this detector, the interference pattern will be washed
out.
The interference pattern consists of maxima and minima in the measured intensity,
referred to as bright and dark interference fringes. For the simple case of two interfering
plane waves of equal wavelengths (see Fig. 5.2), we have

I (r ) = I1 (r ) + I2 (r ) + 2 I1 (r ) I2 (r ) cosΔϕ (r )

Therefore, the maximum and minimum intensity values at the interference pattern are given
by:

2
Imax = I1 + I2 ± 2 I1I2 =
min
( I1 ± I2 )
The average of these intensities is:
€ I max + Imin
I≡ = I1 + I2
2

In addition, we have

Imax − Imin
2 I1I2 =
2

Therefore, we can rewrite the interference pattern as


€ Imax + Imin Imax − Imin
I= + cosΔϕ (r )
2 2

The contrast of the modulation of the interference pattern thus depends on the
intensity ratio of the interfering beams. This contrast can be quantified in terms of the
visibility parameter,€defined as

Imax − Imin 2 I1I2 2 ρ


v≡ = ≡
Imax + Imin I1 + I2 1+ ρ

where the intensity ratio is given by ρ = I1 /I 2 or ρ = I2 / I1 . Note that 0 ≤ v ≤ 1, with v = 0


for ρ = 0 and v = 1 for ρ = 1. The interference pattern can then be rewritten as

I = I [1+ v cosΔϕ (r )] ; with 0 ≤ v ≤ 1

When the interference term depends on time, visibility values lower than the one set
by the intensity ratio will be obtained; the degradation in visibility will depend on the
detector response time.
Notice that, although the visibility may be related to the coherence of the interfering
waves, it is also a function of their relative intensity ρ . To obtain a parameter expressing

6 €
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

solely the coherence of the beams, one needs to factor out the intensity ratio. This is done by
introducing the so-called degree of coherence, introduced in Chapter 1 of Optics II.

5.1.2 Young’s Double-Slit Experiment


Consider Young’s double-slit experiment with two parallel slits separated by a distance a,
placed at a distance s from the screen where the interference pattern is observed (see Fig.
5.3). The position on the screen (normal to the direction of the slits) is specified by the
coordinate y. A plane wave of frequency ω is incident on the slits with the wave fronts
parallel to the plane in which the apertures are drilled, so that φ1 = φ 2 . We assume slits of
infinitesimal widths such that cylindrical waves emerge and interfere at the screen. (The case
of slits of finite width is referred to in Chapter 6.) Moreover, the interference pattern is
measured across a limited width on the screen such that we can assume that the illuminations
due to the two slits are uniform and equal, i.e., I1 = I 2 ≡ I 0 .

Figure 5.3: Arrangement of Young’s double-slit experiment.

The phase difference between the waves, accumulated due to the propagation from
each slit to the screen, is given by

Δϕ ( y ) = k ( r1 − r2 ) = kΔr

€ where the wavenumber is k = 2π / λ . The distance from each slit to the observation point y is,
respectively,
2 1/2
r1,2 = s 2 + ( y ± a 2 ) 

and hence the difference in distance is

1/2 1/2
  a 
2   a 
2
2 2
Δr = r1 − r2 = s +  y +   − s −  y −   .
  2     2  

For a screen situated sufficiently far away from the slits such that s >> a, y , we can
expand this expression in a Taylor series and obtain, to first order:

7
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

 2 1/2  ( y − a 2 )2  
1/2
 ( y + a 2) 
Δr = s 1+  − 1+  
 s2   s2  

 ( y + a 2 )2 ( y − a 2 )  ya
2

≅ s 1+ −1− = ≅ θa
 2s 2 2s 2  s

In this expression, we used the direction angle θ (see Fig. 5.3) defined by

y
sin θ ≅
s

The phase difference is thus given approximately by


€ 2π ya
Δϕ ( y) = kΔr ≅
λ s

and the interference pattern can be written:


2
I( y ) = 4I 0 cos (π ya λ s)

This pattern exhibits bright and dark fringes with visibility v = 1. The bright fringes
correspond to the maxima of the cosine function:

πya mλs  y λ
Maxima: = mπ ⇒ y m = θ m ≅ m = m  ; m = 0, ± 1, ...
λs a  s a

whereas the dark fringes correspond to the minima:


€ πya π λs mλs  y λ mλ 
Minima: = + mπ ⇒ ym = + θ m ≅ m = +  ; m = 0, ± 1, ...
λs 2 2a a  s 2a a 

The angular position of the interference fringes depends on the wavelength of light
used and on the slit separation. Thus, knowledge of the slit separation and the distance from

the screen can be used to measure the wavelength of light. Note that the position of the
fringes in the interference pattern (with respect to y=0) depends on the difference in initial
phases.

So far we have assumed ideal plane waves, i.e., waves of infinite coherence length.
For quasi-monochromatic light of finite coherence length lc , we obtain (with a sufficiently
slow detector) interference between two waves originating from the same wave train only if
the difference in propagations paths Δr satisfies the relation

Δr < lc

This is equivalent to the condition:

€ 8
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

a
y <l or y < lc s / a
s c

Therefore, fringes with clear visibility will be observed only for this range of y. From this
expression, we find that the “last” bright fringe, with m = M , will be located at the position

Mλ s lc s
yM ≈ ≈ ⇒ M ≈ lc / λ = c /( λΔ ν )
a a €
where we have used the linewidth Δν = c /lc of the light source used to illuminate the slits.
The number of bright fringes observed in the case of a finite coherence length is thus:

2M + 1 ≈ 1+ 2lc / λ = 1+ 2c /(λΔν )

For m > M , waves originating from different wave trains interfere, and since their
relative phases fluctuate randomly, their interference pattern will smear out completely. Thus,
counting the number of bright fringes in Young’s experiment provides a measurement of the
coherence length (or the linewidth) of a quasi-monochromatic light source. This is one
example of how interference experiments can be used to characterize the coherence
properties of a light beam.

5.1.3 Fringes of Equal Thickness


Thin film structures (a film of soap, a multilayer dielectric mirror, …) yield interference
patterns due to the superposition of beams reflected at the different boundaries of the layers.
In the case of a single thin layer with a refractive index close to that of its surrounding, the
interference pattern is formed mainly by two beams. We consider here the case where the
thickness of such film varies with position, and show that the locus of a given interference
fringe can in fact serve as a measure of the film thickness.

Thin Film Wedge

Consider two glass plates in contact at x=0 with their surfaces inclined at a small relative
angle α <<1 (Fig. 5.4). An air gap of variable thickness δ ( x ) ≅ xα is thus formed between
the two glass surfaces. A plane wave impinging at near normal incidence to the two plates
will yield two near-normal reflected waves that will interfere. The resulting interference
pattern is observed in reflection from above (see Fig. 5.4). The phase difference between
these two reflected waves will hence vary along x€as:

Δϕ ( x ) = 2kδ ( x ) + π

9
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

Figure 5.4: Cross section of a wedge-shaped thin film of air showing paths of interfering
light rays.

Note that the additional π phase shift arises from the facts that the refractive index of air is
smaller than that of glass, and thus the Fresnel reflection coefficient for normal incidence,
which reads
n − nt
r= i
ni + n t

yields a negative reflection coefficient at that interface.


A set of dark and bright fringes is obtained due to the interference of the two beams.
The condition for obtaining interference maxima is:

2k δ( x ) = (2m + 1)π ; m = 0,±1,±2,...


or:
 1 λ
δ( x ) = m +  ; m = 0,±1,±2,...
 2 2

The bright fringes are thus located at

 1 λ
Bright fringes: xm = m +  ; m = 0,±1,±2,...
 2  2α

and indicate positions of equal thickness of the air gap. Similarly, the positions of the dark
fringes correspond to

2δ ( x ) = mλ ; m = 0,±1,±2,...

and they are located at:

10
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

λ
Dark fringes: xm = m ; m = 0,±1,±2,...

Note the appearance of a dark fringe for zero thickness of the air film: this arises from the
additional π phase shift due to the reflection at the air/glass interface. Hence, the
transmission at that point is complete. It should also be pointed out that the wavelength in the
above expressions is the one in the medium of the thin film (i.e., the vacuum wavelength
divided by the refractive index of the medium). In the example of Fig. 5.4, this medium is air.

Newton’s Rings

Consider now another thin film interference configuration, in which a spherical glass surface
of radius R is in contact with a planar surface of glass (Fig. 5.5). A thin film of air of varying
thickness then forms between the two surfaces. Its thickness is given by

(
δ ( x ) = R − R 2 − x 2 = R 1− 1− x 2 R 2 )
where x=0 is chosen at the point of contact.

Figure 5.5: Cross-section of spherical surface in contact with a flat surface.

Near the point of contact, where this thickness is sufficiently small ( δ << R ) that we can
approximate the thickness (using a Taylor expansion) by

 x 2  x2
δ( x ) ≈ R1− 1+ 2  =
 2R  2R

The locus of constant thickness of the air film corresponds now to a ring centered at the point
of contact. The interference pattern consists of a set of rings, called Newton’s rings, that are
concentric with the point of contact at x=0. Using the same reasoning as for the case of the
wedge, we obtain the position of bright fringes:
1/ 2
  1 
Bright fringes: x m = Rλ m +  ; m = 0,±1,±2,...
  2 

and dark fringes:



11
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

1/ 2
Dark fringes: x m = (R λm) ; m = 0,±1,±2,...
In the more general case where a surface of an arbitrary profile is put in contact with a
“perfectly” planar surface, a similar analysis shows that the features of the surface can be
inferred from the corresponding interference pattern. In particular, the pattern of dark and
bright fringes follow the topography of the surface with height accuracy of the order of a
wavelength of the illuminating light.

5.2 Multiple-Beam Interference


In many cases, the interference pattern is formed due to the superposition of many partial
beams. The main difference between the case of multi-beam interference and that of two
beams is that now the condition for constructive interference is more stringent, and therefore
occurs for a smaller range of phase difference. Hence, the interference patterns usually
display sharper features. This is a characteristic that is particularly useful for applications in
interferometers, as we shall see below.

5.2.1 Interference Pattern


Similar to the two-beam interference case, we imagine now a system of multiple beams
formed when a wave impinges upon a screen containing a series of N slits (see Fig. 5.6).
Each slit produces a beam that subsequently arrives at a screen situated at a large distance
from the slits. We wish to evaluate the interference pattern produced by the superposition of
all these partial waves arriving at a given point on the screen.
It is useful to consider each of the beams arriving at the screen as a phasor, i.e., a
scalar wave characterized by amplitude and phase:


E j = U j e j ; j = 1,2,....N ;

where the phase is given by ϕ j ( r ) = −ik j ⋅ r + φ j and the amplitude of each phasor is related
to the intensity by U j =€ I j .

Figure 5.6: Model for calculating the interference pattern of a multiple-slit system.

12
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

As in the case of two-beam interference, the phase term contains the initial phase, φ j , as well
as the phase accumulated during propagation, −ik j ⋅ r . The different amplitudes account for
possible differences in the intensities of the partial waves.
The desired interference pattern is obtained by summing up all the phasors. This can
be conveniently done using a phasor diagram (see Chapter 3). The resultant phasor
corresponds to the vector addition of the phasors and the resultant intensity is then given by
the norm of this phasor (see Fig. 5.7). The interference pattern is thus given by:

2
N
2 iϕ j
I = E tot = ∑U e j
j =1

Note that the total intensity I of the interference pattern depends on the position on the screen
due to the spatial dependence of the phases ϕ i (r ) . In general, this interference pattern thus
depends on the intensities of the partial waves as well as on the relative phases. We now
consider two special cases of great importance.


Figure 5.7: Phasor diagram for calculating the interference pattern of N beams of varying
phase difference and amplitude.

5.2.2 Multiple-Slit Interference


We consider first the simple case in which the intensities of the partial waves are identical
and the phase difference between consecutive phasors is the same. This is, to a good
approximation, the case of the interference pattern produced by a plane wave impinging on a
series of slits of uniform spacing, or that formed due to multiple reflections from a multilayer
coating of thin films of equal thicknesses, deposited on a planar surface.
In this case we have Uj = U0 , ϕ j − ϕ j −1 = Δϕ = const. By inserting these parameters
into the expression for the interference pattern we obtain:
2 2 2
N N
iϕ j iΔϕ ( j−1) 1− e iΔϕN
I = I0 ∑ e = I0 ∑ e = I0
j=1 j=1 1− e−iΔϕ
2 2 2
N N
iϕ j iΔϕ ( j−1) 1− eiΔϕ N
I = I0 ∑e = I0 ∑e = I0
1− eiΔϕ
j=1 j=1

13
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

where we have used the summation formula for a geometric series. The intensity of the
interference pattern can be rewritten, as a function of the phase difference Δϕ , as

sin2 (Δ ϕN /2)
I = I0
sin2 (Δϕ /2)

The N-slit interference patterns show principal maxima for

Principal maxima: Δϕ /2 = mπ ; m = 0,±1,±2,...

for which the intensity is given by

(Δ ϕN /2)2 2
I → I0 2 = N I0
Δϕ →2mπ
(Δϕ /2)
At these values of the phase difference Δϕ , all phasors are in phase, and a coherent
superposition results; the intensity is much larger than the incoherent sum NI0 of the
intensities.
The intensity of the interference pattern drops to zero at a series of points located at
phase differences given by


Minima: Δϕ min = m ; m = ±1, ±2,... ± (N − 2)
N

In between the principal maxima, there are also N-2 subsidiary maxima, obtained each time
the argument of the sine function in the numerator reaches a maximum value:

π
Subsidiary maxima: Δϕ max = (2m + 1) ; m = 0,±1,±2,...
€ N

These subsidiary maxima have intensity values of

I0
Imax = ; m = 0,±1,±2,...
2  π 
sin (2m + 1) 
 2N 

We now apply the above expressions to the case of the multiple beam interference
pattern of a multiple slit system. Such interference pattern is illustrated for the case N=6 in

Fig. 5.8. We assume a system of N infinitely thin slits, with spacing a, and no initial phase
difference. The phase difference between the waves arriving from adjacent slits is given by:


Δϕ = ka sinθ = asin θ
λ

where θ is the angle between the forward direction and the vector pointing to the observation
point on the screen. Applying the expressions obtained above, we find that the principal
maxima are obtained at angles

14
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

λ
sinθ m = m ; m = 0,±1,±2,...
a

The first minimum is obtained at

λ
θ ≈ sinθ =
Na

This should be compared to the first minimum obtained in a double slit interference pattern,
which is

λ
θ ≈ sinθ =
a

Notice that the first minimum is obtained now when the phase difference between adjacent
waves (i.e., arriving from adjacent slits) is only 2π / N . Thus, a much narrower principal
maximum can be obtained for N>>1.

Figure 5.8: Interference pattern for a system of N=6 slits separated by a distance a (solid
line), compared with the corresponding interference pattern for a double-slit (dashed line).

5.2.3 Fabry-Perot Optical Cavity


Another class of important multiple beam interference patterns corresponds to exponentially
decreasing amplitudes of phasors. This is typical of optical systems in which the amplitude of
a wave is constantly decreased due to consecutive reflections, as is the case of multiple
reflections from a thin film or in a Fabry-Perot cavity. We consider the latter case as a model
for such systems.
An ideal Fabry-Perot cavity consists of two parallel mirrors of (amplitude) reflectance
r1,r2 (and transmittance t1,2 = 1− r1,2 ), spaced by distance L (see Fig. 5.9). This system
represents, e.g., a slab of glass, polished on both parallel facets to produce reflectors due to
the difference in refractive index between the glass and the air ambiance. We wish to
calculate the transmission coefficient for a plane wave of wavelength λ incident on one of
the mirrors. The transmitted wave consists of a series of waves that undergo a different
number of reflections as they circulate in the optical cavity formed by the two mirrors. The
amplitude of the transmitted wave is thus given by

15
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

Figure 5.9: A model of a parallel-mirror Fabry-Perot optical cavity.

Eout = Ein (1− r1 ) (1− r2 ) eikL

+Ein (1− r1 ) r2 r1 (1− r2 ) e3ikL

+Ein (1− r1 ) r2 r1r2 r1 (1− r2 ) e 5ikL

+...

= Ein (1− r1 ) (1− r2 ) eikL 1+ ( r1r2 ei2kL ) + ( r1r2 ei2kL ) +...
2

 

where the wave number is k = 2π λ . We now define r1r2 ≡ r and Δϕ ≡ 2kL (phase
accumulated in a roundtrip inside the cavity), and rewrite the transmitted field as

E out = E in (1− r1 )(1− r2 )e ikL [1+ reiΔ ϕ + r 2e 2iΔϕ + ...]

Summation of this infinite series corresponds to the evaluation of the phasor diagram shown
in Fig. 5.10.

Figure 5.10: Phasor diagram for a Fabry-Perot optical cavity.

16

E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

By summing the series, we get for the transmitted intensity

2
2 2 iΔϕ 2 2 2 1
I out = t t I 1+ re
1 2 0 +... = t t I
1 2 0
1− reiΔϕ
1 t12 t22 I 0
= t12 t22 I 0 =
(1− reiΔϕ ) (1− re−iΔϕ ) 1+ r 2 − 2r + 4r sin 2 Δϕ 2
2
where I0 = E in is the intensity of the incident wave.
The transmitted intensity is thus given by:

t12 t 22 I0
Iout = 2
(1− r) + 4r sin 2 Δϕ 2
The maximum transmitted intensity, obtained for Δϕ = 0 , is

t12 t 22 I0
Imax ≡ 2
(1− r)
Defining the finesse of the cavity as
€ 1
πr 2
F ≡
1− r

the transmitted intensity can be written as the Airy function


€ Imax
Iout (Δ ϕ ) =
 2F 
2

1+ sin2 Δϕ 2
 π 
The form of this function is shown in Fig. 5.11 for different finesse values.
The finesse of the Fabry-Perot cavity can be rewritten in terms of the intensity
2
reflection coefficients R1,2 = ( r1,2 ) , which yields

1 1
π ( r r ) 2 π ( R1R2 ) 4
F = 12 = 1
1− r1r2 1− ( R1R2 ) 2

For a symmetric Fabry-Perot cavity with R1 = R2 = R , the finesse is


€ 1
πR 2
F =
1− R

For example, a mirror reflectivity of R = 0.998 yields a finesse of F ≅ π 2 ⋅10−3 ~ 103 .


Note that the value of the finesse depends critically on the reflectivity. It can be

shown that the introduction of absorption or scattering of the waves during propagation in the
cavity is equivalent to lowering the reflectivity of the mirrors, which, in turn, can drastically

17
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

reduce the finesse. Conversely, the introduction of optical gain (as in laser cavities) increases
the finesse.
The significance of the finesse can be appreciated by examining the variation of the
transmitted intensity with Δϕ . The intensity drops to its half maximum value when

 2F 
2
2
I = Imax 2 ⇒ sin Δϕ 2 = 1
 π 

which yields the half width at half maximum of the transmission peak:

2
 Δϕ1/ 2  2  π 
  ≈ 
 2   2F 
π
⇒ Δϕ1/ 2 ≈
F

Figure 5.11: The Airy function for different finesse values.

The full width at half maximum (FWHM) of the transmission peak is thus given by


δ = 2Δ ϕ1/ 2 =
F

On the other hand, the phase difference between the main transmission peaks corresponds to

Δϕ ≡ Δ = 2π

which is called the free spectral range of the Fabry-Perot interferometer (see Fig. 5.12).
Thus, the Finesse is given by

Δ
F =
δ

The higher the finesse is, the better resolved are the transmission peaks.

18
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

Figure 5.12: Free spectral range and FWHM of a Fabry-Perot interferogram.

5.3 Interferometers
Since the details of the interference patterns discussed above depend on the wavelength of the
interfering waves, they can be used to analyze the spectral content of the incident beam, the
size of objects that fix the optical path lengths, or even the motion of the system in which the
interference takes place. The special design issues involved in constructing interferometers
are briefly discussed in this section.

5.3.1 Diffraction Grating


A diffraction grating consists of a surface that is corrugated periodically with a pitch a. A
plane wave incident on such a grating will be scattered by the surface relief pattern. Each
period of the corrugation acts as a line source, and the reflected beam pattern corresponds to
the interference pattern of a set of multiple slits spaced by a distance a. As the resulting
interference pattern depends on the wavelength of the incident beam, the spectral content of
the incoming wave can be resolved provided that the interference peaks for two given
wavelengths are sufficiently separated in angle.
To calculate the resolving power of such a diffraction grating, consider again the
condition for the principal maxima (ch. 5.2.2),

Δϕ max = 2πm

Assuming an incoming wave at normal incidence, the phase difference between adjacent
partial waves at an angle θ (with respect to the incidence direction) reads

Δϕ (θ ) = kasin θ

Thus, the principal maxima are obtained at angles satisfying the grating equation:
€ asin θ m = λm

The variation in angle of a given principal maximum with changing wavelength is related to
the angular dispersion of the grating, which follows from the grating equation:

19
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

dθ m m
D≡ =
dλ acos θ

It can be thus seen that a higher order m and a smaller pitch a lead to higher dispersion and
hence better wavelength resolution.
The resolving power € depends also on the width of each principal maximum. This
width is given by the condition for hitting the first minimum next to a principal maximum,
i.e., the point at which an additional variation of


δ (Δϕ ) =
N

is obtained. To calculate the angle at which this minimum is set, we evaluate the differential:
€ 2π
δ (Δϕ (θ )) = kacosθδθ =
N

which yields the angular width of each principal maximum:


€ λ
δθ =
Nacos θ

Equipped with the expressions for the angular positions of the principal maxima and
their angular widths, particularly as a function of the wavelength, we can now evaluate the
conditions for resolving two spectral contents of wavelength difference Δλ . For a
wavelength difference of Δλ between two adjacent spectral lines, the angular difference
between the corresponding principal maxima is

dθ m Δλm
Δθ = Δλ = ΔλD =
dλ acosθ

The two wavelengths will be just resolved in the interference pattern if this angular
separation is equal to the angular width of the interference maximum

Δ θ = δθ
or
λ
= mN
Δλ

Thus, for high resolving power it is desirable to work with gratings with a large number N of
lines and at a high diffraction order m.

20

E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

5.3.2 Fabry-Perot Interferometer


In a Fabry-Perot cavity of sufficiently high finesse, the change in transmission characteristics
as a result of changing the wavelength of the incident beam can be used for spectral analysis.
The difference in phase shift between the partial waves transmitted through the cavity is
given by

Δϕ = 2kL = 4 πL λ

A variation in this phase difference can be obtained by changing the cavity length or by
changing the wavelength while fixing L (or by both). In practice, the input wave contains
several spectral components€λ1 , λ2 , λ 3 ... that need to be resolved. The length L is scanned such
that each wavelength provides the conditions for transmission maximum at a different length.
In this way, a spectrum of the input wave is obtained. The spectral components are well
resolved if the length for transmission maximum for one component corresponds to the
length that gives minimum transmission for the other wavelengths.
The requirements for resolving two wavelengths separated by δλ are obtained by
considering the change in Δϕ that corresponds to Δϕ1/ 2 . (This requirement is rather arbitrary;
the ability to resolve the two wavelengths depends on the details of the spectra involved, e.g.,
the intensity of each spectral line). We have then


δ(Δϕ ) = δ (2kL) = − Lδλ
λ2

and by putting

π
δ(Δϕ ) = Δϕ1/ 2 =
F

we obtain

π λ2 λL
δλ1/ 2 = =λ
F 4 πL 4F

Thus, the resolving power is

€ λ 4L
≈F
δλ λ

Recalling that the condition for maximum transmission reads

Δϕ = 2kL = 2π m ; m = 0,±1,±2,...

we have

λ
L= m ; m = 0,±1,±2,...
2
and hence the resolving power can be rewritten:

21
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

λ
≈ 2mF
δλ

To maximize the resolving power, it is then important to increase the finesse and the ratio of
cavity length to wavelength.

5.3.3 Mach-Zehnder and Michelson Interferometers


These two interferometers are based on the superposition of two beams between which an
optical path difference is introduced.
In a Mach-Zehnder interferometer, the incoming beam is split into two beams, and a
path difference is introduced by means of deflecting mirrors (see Fig. 5.13a). The two beams
are recombined using a second beam splitter. The superposition of the two beams results in a
transmitted beam (constructive interference in the forward direction) or a reflected beam
(destructive interference in the forward direction) depending on whether the path difference
is an even or odd multiple of half wavelengths. The interferogram can thus be used to
measure small differences in lengths. Dynamically changing the difference in optical path
length (e.g., by changing the refractive index in one arm of the interferometer) introduces
modulation in the transmission, and is often used to achieve data transmission in optical fiber
systems.
The Michelson interferometer also employs beam splitting, but into two perpendicular
beams, to introduce an optical path difference (see Fig. 5.13b). The two beams are
recombined using the same beam splitter, and the path difference is controlled using a
movable mirror. As in the case of a Mach-Zehnder interferometer, the transmission depends
on the optical path difference and thus variation in the position of the movable mirror on the
order of a fraction of a wavelength can be monitored. This kind of interferometer
demonstrated the invariance of the speed of light in intertial frames of reference in the
famous Michelson and Morley experiment. It is also useful for measuring coherence
properties and the spectrum of light beams (see Optics II).

Figure 5.13: Schematic illustration of a (a) Mach-Zehnder and (b) Michelson


interferometer.

22
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

5.3.4 Sagnac Interferometer


The Sagnac interferometer consists of a ring optical cavity in which two counter-propagating
waves circulate and interfere at an exit of the ring (see Fig. 5.14). The interference pattern
depends on the optical path length of each circulating wave. When the ring rotates about its
center, the effective optical path of one wave increases while that of the other one decreases.
This results in a shift in the interference fringes that is proportional to the angular velocity of
rotation. Thus, such interferometer can serve to detect rotations.
As a simple model for the Sagnac interferometer, consider a ring consisting of four
mirrors forming a square whose half diagonal is given by R (see Fig. 5.14). When this ring
rotates at an angular velocity ω , each mirror moves at a speed v = ωR at 45 degrees with
respect to the light beam. Consequently, the wave that circulates in the sense of rotation will
take more time to propagate along a side of the square as compared with a wave propagating
in the opposite sense. These times are given by

2R
t+ =
c −v/ 2
2R
t− =
c+v/ 2

Thus, the difference in time for circulating the ring is given by



8R 2ω
Δt = 4 ( t + − t− ) ≅
c2
or

4 Aω
€ Δt ≅
c2

Figure 5.14: Schematic illustration of the Sagnac interferometer.

23
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

where A is the area enclosed by the ring and v << c was assumed.
Considering that the oscillation period of the circulating (monochromatic) wave is
T = λ /c , we find that the fractional displacement in the fringes is

Δt 4 Aω
ΔN = =
T cλ

That is, the fringe displacement is proportional to the angular velocity of rotation.
The above treatment is classical (non-relativistic), which is obviously inadequate.
€ yields the same result.
However, more accurate analysis

24
E. Kapon/Optics I (2016/2017)/Chapter 5: Interference

5.4 Summary

-Two-beam interference

I (r ) = I1 (r ) + I2 (r ) + 2 I1 (r ) I2 (r ) cos[Δωt + Δϕ (r,t )]
T

Imax − Imin 2 I1I2 2 ρ


-Fringe visibility v≡ = ≡
€ Imax + Imin I1 + I2 1+ ρ

-Double slit interference I( y ) = 4I 0 cos 2 (π ya λ s)



sin2 (Δ ϕN /2)
-Multiple slit interference I = I0
sin2 (Δϕ /2)

1
πR 2
-Finesse F =
1− R

Imax
-Airy function € Iout (Δ ϕ ) =
 2F 
2

1+ sin2 Δϕ 2
 π 

25

Вам также может понравиться