Theory

Multislit Table of Contents | Project Listing | Multislit-Inhaltsverzeichnis | Liste der Projekte

Multislit Downloads | Entire Site Downloads | About this Site | Imprint

Page Contents

Fresnel Propagation

We are starting in a more general context with a one-dimensional, reduced field amplitude u~⁢(x0,z0) with lateral coordinate x0 at fixed longitudinal coordinate z0. We want to propagate the field to the coordinates x,z, namely to the reduced field u~⁢(x,z). We do this by Fresnel approximation of Huygens' integral:

u~⁢(x,z)=i(z-z0)⁢λ⁢∫u~0⁢(x0,z0)⁢e-i⁢π(z-z0)⁢λ⁢(x-x0)2⁢dx0

which is the convolution of u~⁢(x0,z0) with the Fresnel propagator

hL⁢(x)=iL⁢λ⁢e-i⁢k2⁢L⁢x2

for the propagation length L.

The Problem

The initial reduced field amplitude shall be unity, i.e., we are constraining ourselves to an incident planar wave propagating parallel to the longitudinal axis. The field will be cut off by the aperture. The aperture shall be an infinite slit, a rectangle function of width 2⁢a in x. We introduce unit-free coordinates:

y=xa,N=a2(z-z0)⁢λ  [y]=[N]=1

N is the Fresnel number of the slit, and y is the reduced transversal coordinate x taking on unity at the edges of the slit. Then the Fresnel integral becomes:

u~⁢(y)=i⁢N⁢∫-11ei⁢π⁢N⁢(y-y0)2⁢dy0

This integral is to be solved.

The Complex Fresnel Integral

The complex Fresnel integral is defined to be:

F~⁢(x)=C⁢(x)+i⁢S⁢(x)=∫0xei⁢π⁢t2/2⁢dt

with:

C⁢(x)=∫0xcos⁡(π⁢t2/2)⁢dt
S⁢(x)=∫0xsin⁡(π⁢t2/2)⁢dt

For the Fresnel integral holds:

F~⁢(-x)=-F~⁢(x)

There exist analytical approximations for the Fresnel integral:

F~⁢(x)≈{x⁢ei⁢π⁢x2/2|x|≪11+i2-iπ⁢x⁢ei⁢π⁢x2/2x≫1

where the second one holds according to Siegman also for values of x only slightly greater that unity. But for the transition region between the two regimes, no analytical approximation exists.

But Abramowitz and Stegun give a purely empirical rational approximation to the Fresnel integral, using:

F~⁢(x)=1+i2-(g⁢(x)+i⁢f⁢(x))⁢ei⁢π⁢x2/2

with:

f⁢(x)≈1+0.962⁢x2+1.792⁢x+3.104⁢x2
g⁢(x)≈12+4.142⁢x+3.492⁢x2+6.670⁢x3

The error of this approximation is ≤2⋅10-3 for 0≤x<∞, but Siegman does not tell whether absolute or relative.

Minor objections against the validity of this approximation arise when looking at the examples, but the overall performance of the formula seems to be very precise and very good.

Application to the Single Slit

We have obtained earlier for the integral describing the diffraction by a single slit in the paraxial approximation:

u~⁢(y)=i⁢N⁢∫-11e-i⁢π⁢N⁢(y-y0)2⁢dy0

Setting:

t=2⁢N⁢(y-y0)

equivalent to:

y0=y-12⁢N⁢t

the transformation is:

d⁢y0=-12⁢N⁢d⁢t

and for the integral thus holds:

u~⁢(y)=-i2⁢∫2⁢N⁢(y+1)2⁢N⁢(y-1)e-i⁢π⁢t2/2⁢dt=-i2⁢[(∫02⁢N⁢(y-1)ei⁢π⁢t2/2⁢dt)∗-(∫02⁢N⁢(y+1)ei⁢π⁢t2/2⁢dt)∗]=i2⁢[F~∗⁢(2⁢N⁢(1-y))+F~∗⁢(2⁢N⁢(1+y))]

The single slit integral is now expressed in terms of the Fresnel integral, for which the rational approximation by Abramowitz and Stegun is applicable.