Optoelectronic Systems Lab., Dept. of Mechatronic Tech.

, NTNU

Chap 4 Fresnel and Fraunhofer Diffraction

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Content
4.1 Background 4.2 The Fresnel approximation 4.3 The Fraunhofer approximation 4.4 Examples of Fraunhofer diffraction patterns

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

z 3 "" T / 4P[( x  \ ) 2  ( y  L ) 2 ]2 max

z "" k (\ 2  L 2 ) / 2

(\ ,L )

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.1 Background
These approximations, which are commonly made in many fields that deal with wave propagation, will be referred to as Fresnel and Fraunhofer approximations. In accordance with our view of the wave propagation phenomenon as a system, we shall attempt to find approximations that are valid for a wide class of input field distributions.

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.1.1 The intensity of a wave field Poyntings thm. X X X S ! EvH

X 1 X S ! ( 0 E 2 )V 2 X 1 1 ! 0 E2 2

X S I w E2
When calculation a diffraction pattern, we will general regard the intensity of the pattern as the quantity we are seeking.
5

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.1.2 The Huygens-Fresnel principle in rectangular coordinates Before we introducing a series of approximations to the HuygensFresnel principle, it will be helping to first state the principle in more explicit from for the case of rectangular coordinates. As shown in Fig. 4.1, the diffracting aperture is assumed to lie in the plane, and is illuminated in the positive z direction. According to Eq. (3-41), the Huygens-Fresnel principle can be stated as 1 U ( P0 ) ! j

U ( P1 )

jkr 01

r01

cos Uds

(1)

X and the ector r01 here U is the angle bet een th e out ard nor al n pointing ro P0 to P . 1
6

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

L
\
y
y

P0

P1
Fig. 4.1 Diffraction geometry
Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

The term cosU is given exactly by

cos ! z r01

and therefore the Huygens-Fresnel principle can be rewritten

z U ( x,y ) ! j

U (\ ,L ) r 2 01
tance r01

jkr01

d\dL

(2)

where the

en exactly by

r01 ! z 2  ( x- 2 )  ( y- 2 )
8

(3)
Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

There have been only two approximations in reaching this expression. 1.One is the approximation inherent in the scalar theory

. The secon is the ass mption that the observation istance is many wavelengths rom the apert re r01 "" .

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.2 Fresnel Diffraction

Recall, the mathematical formulation of the Huygens-Fresnel , the first Rayleigh- Sommerfeld sol.

1 e jkr01 U ( po ) ! U ( p1 ) r01 cos(r01 , a n ).ds jP z

The Fresnel diffraction means the Fresnel approximation to diffraction between two parallel planes. We can obtain the approximated result.
( j ? x \ ) 2  ( y L ) 2 A e jkz g 2z U ( x, y ) ! d\ dL g U (\ ,L )e jPz k

(1)

10

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

y
\
"" j
e

K x  \ 2  y  L 2 2z

x z

(Why?) (wave propagation)

wave propagation z Aperture Plane Observation Plane Corresponding to

The quadratic-phase exponential with positive phase k ? x \ ) 2  ( y L ) 2 A j ( i.e, ,for z> 2z

11

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Note:

The distance from the observation point to an aperture point

r01 ! z  ( x  \ )  ( y  L )
2 2

2 1 2

x \ 2 y L 2 ) ( ) ! z 1  ( z z

=b Using the binominal expansion, we obtain the approximation to r ! z  ! z

x \ z

y L z

?x  \ z

 y L

A
Dr. Gao-Wei Chang

12

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

as the term

x  \ 2 y L 2 ( z )  ( z )
is sufficiently small. The first Rayleigh Sommerfeld sol for diffraction between two parallel planes is then approximated by
jk [ z
1 ( x\ ) 2  ( yL) 2 ] 2z

e 1 U ( x, y) ! U (\,L) jP 7

r01

z d\ dL

13

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

( integrand is supposed to be well approximated by the first term only in the binomial expansion, i.e, r01 ! z In addition, the aperture points and the observation points are confined to the ( \ , L ) plane and the (x,y) plane ,respectively. )

z cos( r , a n ) ! r ) , the r01 in denominator of the

Thus, we see
( j ? x \ ) 2  ( y L ) 2 A e jkz 2z d\ dL U ( x, y ) ! U (\ ,L )e jP z k

14

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Furthermore, Eq(1) can be rewritten as

U ( x, y )

U (\ ,L )h( x  \ , y  L )d\
g

dL

(2a)

where the convolution kernel is

Obviously, we may regard the phenomenon of wave propagation as the behavior of a linear system.

e jkz jk 2 h ( x, y ) ! exp[ ( x  y 2 )] jP z 2z

(2b)

15

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Another form of Eq.(1) is found if the term e is factored outside the integral signs, it yields

k (x y ) z

( x\ yL ) j j (\ 2 L 2 ) e jkz j 2 z ( x2 y2 ) g Pz 2z ]e U ( x, y) ! e d\ dL g [U (\ ,L )e jPz

24

(3)

which we recognize (aside from the multiplicative factors) to be the Fourier transform of the complex field just to the right of the aperture and a quadratic phase exponential.

16

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

We refer to both forms of the result Eqs. (1) and (3), as the Fresnel diffraction integral . When this approximation is valid, the observer is said to be in the region of Fresnel diffraction or equivalently in the near field of the aperture. Note: In Eq(1),the quadratic phase exponential in the integrand
j k ( x \ ) 2  ( y L ) 2 2z

17

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

do not always have positive phase for z> .Its sign depends on the direction of wave propagation. (e.g, diverging of converging spherical waves) In the next subsection ,we deal with the problem of positive or negative phase for the quadratic phase exponent.

18

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.2.1 Positive vs. Negative Phases Since we treat wave propagation as the behavior of a linear system as described in chap.3 of Goodman), it is important to descries the direction of wave propagation. As a example of description of wave propagation direction, if we move in space in such a way as to intercept portions of a wavefield (of wavefronts ) that were emitted earlier in time.

19

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

f t ! f zt
t

f ( z, t )
z

tc

f ( z  zc , t )
z t

tc f (t  tc ) ! f ( z, t  tc )

zc f ( z  2 zc , t )
z

2tc

2 zc

In the above two illustrations, we assume the wave speed v=zc/tc where zc and tc are both fixed real numbers.
20

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

In the case of spherical waves,

Diverging spherical wave

T r

Converging spherical wave

r
k

21

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Consider the wave func.

j k r

,where

r ! ar r

T and r > and k ! a k k ! ak P If

ak

 a r ,then
( Positive phase)

1 j k r 1  j k r ! e e r r

implies a diverging spherical wave. Or if

a k ! a r
22

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

1 j k r 1  j k r e ! e r r
i lies a c er i

( Negative phase) s erical a e.

Note For spherical wave ,we say they are diverging or converging ones instead or saying that they are emitted earlier in time or later in time.

23

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Specifically, for a time interval tc > , we see the following relations,

Earlier in time

 j 2Tv ( t tc )

 j 2Tvt

 j Tv t  tc

Positive phase

The term e standing for the time dependence of a traveling wave implies that we have chosen our phasors to rotate in the clockwise direction.

 j 2Tvt

24

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Therefore, we have the following seasonings Earlier in time Positive phase (e.g., diverging spherical waves) Later in time Negative phase (e.g., converging spherical waves) Note Earlier in time means the general statement that if we move in space in such a way as to intercept wavefronts (or portions of a wave-field ) that were emitted earlier in time.

25

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

To describe the direction of wave propagation for plane waves, we cannot use the term diverging or converging .Instead .we employ the general statement ,for the following situations.
ay

Propagation direction

Uc " 0

az

Spatial distribution of wavefronts

26

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

The phasor of a plane wave,

 j 2TEy

, (where E > )

multiplied by the time dependence gives

e  j 2TEy e  j 2Tvt ! e

 j 2Tv ( t tc

, where

1 ! Ey c v

We may say that ,if we move in the positive y direction , the argument of the exponential increases in a positive sense, and thus we are moving to a portion of the wave that was emitted earlier in time.

27

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

In a similar fashion , we may deal with the situation for E

or U c

Uc

Propagation direction

28

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Note Show that the Huygens-Fresnel principle can be expressed by

e jkr X X U(p ) ! U(p ) cos( r , an )ds jP r

<pf> Recall the wave field at observation point P

xG xu U(p ) ! (G xn  U xn )ds 4T
29

( )

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

For the first Rayleigh Sommerfeld solution ,the Green func.

jk~ 1 r

e e G !  ~ r1 r1

jkr 1

(2)

Note we put the subscript -, i.e, G- to signify this kind of Green func.

Substituting Eq(2) into Eq.(1) gives

30

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

xG 1 U ( p0 ) ! (U xn )ds 4T

or

( )

x k 1 ( p0 ) ! ( xn )ds T
where the Green func. proposed by Kirchhoff

(4)

e jkr01 Gk ! r01
31

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

The term in the integrand of Eq.(4)

xGK ! (GK ) an xn x e j r01 ! ar01 ( ) an xr01 r01 1 j r01 j r01 2 ! o (r01 an ) j e e (1) r 01 r01 1 e j r01 ! o (r01 an )( j  ) r01 r01

32

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

as K

2T 1 "" or r1 P

r "" P
(5)

xGK 2T e j r 1 ! j cos(r 1, n ) xn P r1
Finally, substituting Eq.(5) into Eq.(4) yields
jkr01 1 X X ( p0 ) ! ( p1 ) cos(r01, an )ds r01 jP

33

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU 4.2.2 Accuracy of Fresnel Approximation

Recall Fresnel diffraction integral

g
k

j ?x \ 2  y L 2 A  e jkz 2z U x, y ! d\dL g U \ ,L e jPz 

Parabolic wavelet

(4.14)

observation point (fixed)

Aperture point (varying with )

We compare it with the exact formula

U x, y ! 1 jP
g

z
jkr01

g

U \ ,L

e XX r01d n d\dL a cos r01

r01

Spherical wavelet

(or

1 x  \ r01 ! z 1  2 z

where

x  \ y L r ! z   z z

y L  z

 1 x  \  y L 8 z z
2

.

34

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

since the binomial expansion

1  b

1 2

1 1 2 ! 1 b  b . 2 8
x  \ y L !  z z
2 2

where

  b 1 2  1  1 b ) ) The max.approx.error (i.e.,( 1 2 max

1 8
2 2 2 1 x  \ y L !  8 z z 2

and the corresponding error of the exponential

jkz b 8

is maximized at the phase

(or approximately 1 radian)

35

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

A sufficient condition for accuracy would be

x  \ y L z  P z z
a

<<1

T z P For example

?x  \ 

 y  L

A
( y) x,

1cm

( y) x,

( , ) 36

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

2 3.14 1v10 z 3 4 v 0.5 v10  6

2 2

!6

or z 0.4 m

X az

This sufficient condition implies that the distance z must be relatively much larger than T x  \  y  L ma P 37 Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Since the binomial expansion

1  b

1 2

1 1 1 2 ! 1  b  b  . ! 1  b  HOT 2 2 8
2 2

(high order term)

x  \ y L where b !  z z

we can see that the sufficient condition leads to a sufficient small value of b

However, this condition is not necessary. In the following, we will give the next comment that accuracy can be expected for much smaller values of z (i.e., the observation point (x , y) can be located at a relatively much shorter distance to an arbitrary aperture point on the ( , ) plane)
38

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

We basically malcr use of the argument that for the convolution integral of Eq.(4-14), if the major contribution to the integral comes from points ( , ) for which x and y, then the values of the HOTs of the expansion become sufficiently small.(That is, as ( , ) is close to (x , y) x  \ y L b!  z z

gives a relatively small value

1 Consequently,   b

1 can be well approximated by 1  b ) . 2

39

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

In addition it is found that the convolution integral of Eq.(4-14),

e U x, y ! g U  ,L e jP z 
g
j

e jkz or U x, y ! gU \ ,L e jPz 

x \ jT Pz

e jkz jT  2  Y 2 X U  ,L \ e d\dL jP z 7

where

X !

x \

and

Y!

y L Pz

Pz

40

jkz

T Pz

?x  

2   y L 2 A
2

d dL

2

y L  Pz

d\dL

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

can be governed by the convolution integral of the function with a second function (i.e., U( , )) that is smooth and e jT X Y slowly varying for the rang 2 < X < 2 and 2 < Y < 2. Obviously, outside this range, the convolution integral does not yield a significant addition.
2 2

( Note For one dimensional case

g

jTX 2

dX ! 1 is governed by

e jTX dX

we can see that

g

is well approximated by

e jT X

Y
Y

e jT X

dXdY !

dXdY
Dr. Gao-Wei Chang

41

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Finally, it appears that the majority of the contribution to the convolution integral for the range - < X < and - < Y < or the aperture area comes from that for a square in the ( , ) plane with width and Pz 4 centered on the point = x, = y \ y L (i.e., the range 2 < <2 and 2< <2 or Pz
Pz

\

< 2 Pz and y  L < 2 P z )

As a result within the square area, the expansion

1 1 ! 1 b  b2 . 2 8 as well approximated, since

1  b 12

 \ L !  z z

4 Pz

X az

is small enough.
42

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

From another point of view, since the Fresnel diffraction integral

T ?x \ 2  y L 2 A  j e jkz z U x, y ! d\dL U \ ,L e j z7

?x \ 2  y L 2 A  e d\dL U \ ,L e jPz Corresponding square area

jkz j

yields a good approximation to the exact formula

1 U P0 ! jP
where r

e j r01 X X cosr01 , a n ds U P 7 1 r01

z x \ z y L z

43

T Pz

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

we may say that for the Fresnel approximation (for the aperture area or the corresponding square area) to give accurate results, it is not necessary that the HOTs of the expansion be small, only that they do not change the value of the Fresnel diffraction integral significantly.

Note
From Goodmans treatment (P. 9 7 ), we see that

X

jTX 2

dX
jTX 2

can well approximate

g

dX

or

jTX 2

dX

Where the width of the diffracting aperture is larger than the length of the region 2 < X < 2
44

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

For the scaled quadratic-phase exponential of Eqs.(4-14) and Eq.(4-1 ), the corresponding conclusion is that the majority of the contribution to the convolution integral comes from a square in the ( , ) plane, with width 4 Pz and centered on the point ( = x , = y) In effect, 1. When this square lie entirely within the open portion of the aperture, the field observed at distance z is, to a good approximation, what it would be if the aperture were not present. (This is corresponding to the light region)

45

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

2. When the square lies entirely behind the obstruction of the aperture, then the observation point lies in a region that is, to a good approximation, dark due to the shadow of the aperture. 3. When the square bridges the open and obstructed parts of the aperture, then the observed field is in the transition (or gray) region between light and dark.

For the case of a one-dimensional rectangular slit, boundaries among the regions mentioned above can be shown to be parabolas, as illustrated in the following figure.

46

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4 Pz

2 Pz

2 Pz

47

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

The light region Wx W+x

2 Pz, x 2 Pz , x

Thus, the upper (or lower) boundary between the transition (or gray) region and the light region can be expressed by

x  w

! 4Pz (or x  w ! 4Pz )

48

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.2.3 The Fresnel approximation and the Angular Spectrum In this subsection, we will see that the Fourier transform of the Fresnel diffraction impression response identical to the transfer func. of the wave propagation phenomenon in the angular spectrum method of analysis, under the condition of small angles. From Eqs.(4-15)and (4-1 ), We have

U ( x y) !

g

U (^ L )h(E  ^
jk

y  L )d\dL

Where the convolution kernel (or impulse response) is

h( x y ) !

e
j

xk (x y ) x
Dr. Gao-Wei Chang

49

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

The FT of the Fresnel diffraction impulse response becomes

jkz

F [h( x, y )] !

H
j

( F

fx fy
,

)!

z  g e

k (x  y ) z

 j T ( f x x  f y y)

dxdy

The integral term

g e
e
-j

T 2 2 (x  y ) z

 j 2T ( f x x  f y y )

dxdy

can be rewritten a
T ( f x ( P z )  f y ( Pz ) ) g P g

T ( p q ) dpdq Pz

where

p ! x  Pz

and

q ! y  Pz
50

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU ( eca se t e e e ts

T j [ x  x (P z Pz

( f
x

x Pz

jT ]! ( x  Pz Pz

fx

where

p ! x  Pz

f
y

P
2

2 jT [ y  2 y (P z ) Pz

 ( f y Pz)

jT ]! Pz

( y  Pz f y )
q

where q ! y  Pz as a result,
HF (

jkz

T ( f x ( Pz )  f Pz

y ( Pz )

1 jPz

T ( p q ) dpdq=1 Pz
51

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU so  jTPz ( 2  2 ) jkz x y

HF(

)!

e e

On the other hand, the transfer function of the wave propagation phenomenon in the angular spectrum method of analysis is expressed by

Ha

jk ! e y

-( ( x ) -( -

under the condition of small angles (as noted below the term)
jkz 

(P x)  (P y)
1 (P )2 1 (P f )2 fx  y ) 2 2
f 2 f 2 ) x y

can be approximated by

e j z  jTPz ( e e

j z (1

T (because k ! ) P
52

) ,

, otherwise

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

(Note: because

!z o1

x \ y L 1 ( ) ( ) z z
2

1 2

For Fresnel approximation, the sufficient condition ma be

""

T 4P [

 y L 2] x \

max

The obliquity factor co That i U ! co

an r o

then approache is small angle

XX an r

53

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Therefore, we have shown that the FT of the Fresnel diffraction impulse response

H f x f y !Ha f x f y
Which is the transfer function of the wave propagation phenomenon in the angular spectrum method of analysis under the condition of small angles.

54

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.2.4 Fresnel Diffraction between Confocal Spherical surfaces.

L

o1

r r
o1

o1

55

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

\  1 L ) ) 2( ) z z L  2\x   z  2Ly  ! z (1   2z 2z
2 2

1 r o1 $ z(1  2 (

as x y \ L

are all very close to zero, (i.e, the paraxial condition)

\x  Ly r o1 $ z  z  z
Recall the Rayleigh Sommerfeld sol, (for the paraxial condition
U ( x, y ) ! jP U (\ ,L )
j 2 P

e ro
56

jk r o

cos(

r o , a n ) d\ dL
Dr. Gao-Wei Chang

U (\ ,L ) e jPz

k  j (\ x L y ) d\ dL z

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

as a result, for the paraxial region,

U ( x, y ) !

jkz j

U (\ ,L ) e jPz

T (\x Ly ) Pz

d\dL

(including the paraxial representation of spherical phase)

This Fresnel diffraction eq. expresses the field U \ ,L observed on the right hand spherical cap as the FT of the filed U(x,y) on the left-hand spherical cap.

Comparison of the result with Eq(4-17),the Fresnel diffraction integral (including Fourier-transform-like operation)
Dr. Gao-Wei Chang

57

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

( x, y ) !

e e jPz

jkz

k 2 2 (x  y ) 2z

g [

(\ ,L )

k 2 2 (\ L ) ] 2z

j

2T ( x\  y L ) d\dL Pz

quadratic phase parabolic phase

Note: Recall

58

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

The two quadratic phase factors in Eq(4-17)are in fact simply paraxial representations of spherical phase surfaces, (since the Rayleigh Sommerfeld sol. can be applied only to the planar screens), and it is therefore reasonable that moving to the spheres has eliminated them. For the diffraction between two spherical caps, it is not really valid to use the Rayleigh-Sommerfeld result as the basis for the calculation (only for the diffraction between two parallel planes). However, the Kirchhoff analysis remains valid, and its predictions are the same as those of the Rayleigh-Sommerfeld approach provided paraxial conditions hold.

59

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.3 The Fraunhofer approximation

From Eq(4-17), We see
U ( x, y ) !
z 2T ( x\  yL ) d\dL Pz

e e Pz
2

2z

( x 2  y 2) g

g [U (\ ,L ) e

2z

(\ 2 L 2)

(4-17)

If the exponent

k [ ( 2z
We have
T z "" P or

)]
max

\ L

a a
60

z ""

\ L

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

The observed filed strength U(x,y) can be found directly from a FT k j0 of the aperture function itself (because j 2 z (\ L ) ) e 1
2 2

That is, Eq.(4-17)with the Fraunhofer approximation becomes

e e U ( x, y ) !

jkz

k 2 2 (x  y ) g 2z

jPz

g

U (\ , )

 j 2T ( f x\  f y )

d\d

(4-25)

(Aside from the multiplicative phase factors, this expression is simply the FT of the aperture distribution) where

fx

x and Pz

y Pz

(4-2 )

61

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Note Recall the different forms of Fresnel diffraction integral

U ( x, y ) ! U ( x, y ) !

jkz j

U (\ ,L ) e jP z

T [ ( x \ )  ( y L ) ] d\dL ..................(4 - 14) Pz

g U (
jkz

, )h( x  , y  )d d .............................(4 - 15)

where the Fresnel diffraction impulse response

h x y

!e

jP z

k z

y

(4-1 )

and that of Eq(4-17)

62

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Comparison of Eqs(4-15)and (4-1 ) with Eqs.(4-25)and (4-2 ) tell us that there is no transfer function for the Fraunhofer (or far-field) diffraction since Eqs(4-25) and (4-2 ) do not include impulse response. Nonetheless, since Fraunhofer diffraction is only a special case of Fresnel diffraction, the transfer function Eq(4-21) remains valid throughout both the Fresnel and the Fraunhofer regimes. That is, it is always possible to calculate diffracted field in the Fraunhofer region by retaining the full accuracy of the Fresnel approximation.
Treating the wave propagation phenomenon as a linear system

63

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.4 Examples of Fraunhofer diffraction patterns

4.4.1 Rectangular Aperture If the aperture is illuminated by a unit-amplitude, normally incident, monochromatic plane wave, then the field distribution across the aperture is equal to the transmittance function .Thus using Eq.(4-25), the Fraunhofer diffraction pattern is seen to be

U ( x, y ) !

k 2 2 (x  y ) jkz 2 z e e j

jPz

F{U (\ ,L )}
f X ! x / Pz fY ! y / P z
64

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.4.2 Circular Aperture Suggests that the Fourier transform of Eq.(4-25) be rewritten as a Fourier-Bessel transform. Thus if K is the radius coordinate in the observation plane, we have

U(

e jkz k 2 ! exp( j ) F {U ( q )} jP z 2z p ! r / Pz

65

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.4.3 Thin Sinusoidal Amplitude Grating In practice, diffracting objects can be far more complex. In accord with our earlier definition (3- ),the amplitude transmittance of a screen is defined as the ratio of the complex field amplitude immediately behind the screen to the complex amplitude incident on the screen . Until now ,our examples have involved only transmittance functions of the form

t A (\ ,L )

1 0

in the ape tu e ut the ape tu e

66

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

Spatial patterns of phase shift can be introduced by means of transparent plates of varying thickness, thus extending the realizable values of tA to all points within or on the unit circle in the complex plane. As an example of this more general type of diffracting screen, consider a thin sinusoidal amplitude grating defined by the amplitude transmittance function

1 m \ L t A \ L !  cos2Tf \ rect rect 2 2 2w 2w

(4-33)

where for simplicity we have assumed that the grating structure is bounded by a square aperture of width 2w. The parameter m represents the peak-to-peak change of amplitude transmittance across the screen and f0 is the spatial frequency of the grating.
67

Dr. Gao-Wei Chang

Optoelectronic Systems Lab., Dept. of Mechatronic Tech., NTNU

4.4.4 Thin sinusoidal phase grating

(\ or x)
Binary phase grating m j[ sin ( 2 f 0 )] U ( ,y ) ! e 2 rect (

2w

)rect (

2w

68

Dr. Gao-Wei Chang