See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/47409598

Modeling and Propagation of Near-Field Diffraction Patterns: A More Complete

Approach

Article  in  American Journal of Physics · September 2004

DOI: 10.1119/1.1767102 · Source: OAI

CITATIONS READS
37 144

2 authors, including:

Glen D Gillen
California Polytechnic State University, San Luis Obispo
42 PUBLICATIONS   329 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Glen D Gillen on 07 August 2015.

The user has requested enhancement of the downloaded file.

Modeling and propagation of near-field diffraction patterns: A more
complete approach
Glen D. Gillena) and Shekhar Guha
Air Force Research Laboratory, Materials and Manufacturing Directorate, WPAFB, Ohio 45433
�Received 5 December 2003; accepted 7 May 2004�
We discuss the origins and regions of validity of various near-field diffraction models. The complete
Rayleigh–Sommerfeld model is found to accurately represent intensity distributions for axial
distances up to and including the location of the aperture, a region where commonly used models
fail. We show that near-field diffraction theory can be applied to the refraction of light at an interface
between two different media yielding results that demonstrate the validity of Snell’s law in the
presence of diffraction. Calculations using near-field diffraction and Fourier optics are compared to
experimentally measured intensity distributions. © 2004 American Association of Physics Teachers.
�DOI: 10.1119/1.1767102�

I. INTRODUCTION tens10 to hundreds11 of microns. Modeling beam propagation

within complex systems on this scale presents new
Diffraction occurs when an aperture is illuminated by challenges.12,13
light. In spite of the historical importance of diffraction �the The aim of this paper is to discuss various diffraction
understanding of which first established the wave nature of theory models, probe the regions of validity of each, and
light�, a complete description of the distribution of light im­ apply appropriate models to some specific examples. The
mediately beyond the illuminated aperture is not available in discussion could be incorporated into an advanced under­
the usual textbooks of optics.1– 6 The treatment of near-field graduate or beginning graduate level course in optics. We
diffraction in undergraduate textbooks ranges from simply first derive and compare various models for near-field light
propagation. Then we calculate some representative two-
stating the equations to be used for near-field and far-field
dimensional intensity patterns and beam profiles for the near-
diffraction regions,1 to briefly discussing the physical basis
field diffraction of infrared laser light. In the first example
for the Fresnel and Fraunhofer diffraction models,2 to deriv­
we apply near-field diffraction theory to the traditional ex­
ing Fresnel and Fraunhofer diffraction using scalar diffrac­ ample of the refraction of light at an interface between two
tion theory.3,4 For a more detailed discussion of scalar dif­ media. The locations and intensity profiles are discussed as
fraction theory and derivations of models other than Fresnel functions of the incident angle of the light and the refractive
and Fraunhofer, students are usually referred to graduate- index of the medium. Finally, we investigate the propagation
level textbooks.5,6 of light beyond the aperture and through a lens, yielding
The simplest case of diffraction of light is that of plane some interesting results that are verified experimentally.
waves traveling beyond a hard, planar aperture. The com­
plexity of accurately modeling and predicting the electric
II. THEORETICAL MODELING
field and intensity distributions at every point beyond the
aperture can be computationally quite intensive. Frequently, Studies of the propagation of light beyond an aperture
various approximations are implemented to decrease the were extensively performed in the late 19th century by
level of complexity and computational times required to Kirchhoff,14 Sommerfeld,15 and Rayleigh,16 among many
model the diffracted beam propagation. others. Further progress in finding solutions that accurately
Over the past few decades, interest has slowly grown in describe the electric field distributions for all points beyond
reexamining near-field diffraction theory because of ad­ the aperture were hampered by the complexity of the inte­
vancements in laser technology, the continual shrinking of grals to be performed. Over the years, several different mod­
technological components, and advancements in computer els have been developed for the electric field and intensity
technology. One of the many recent advances in laser tech­ distributions for points beyond the aperture. The models
nology is the generation of terahertz radiation.7 The conver­ make different assumptions and approximations to simplify
sion of light wavelengths from the nanometer regime to the the mathematics and reduce the computational time. Table I
centimeter regime also results in a proportionate growth in summarizes some of the models and the approximations used
the scale of near-field diffraction patterns, from the micro­ for each. As a result of the speed of present computer tech­
scopic �sub-micron� scale to the macroscopic �millimeter and nology, numerical results for even the full wave equations
centimeter� scale. The presence of near-field optics on this can now be obtained using a desktop computer.
macroscopic scale has fueled efforts to accurately We begin by considering the situation depicted in Fig. 1.
characterize7,8 and model near-field propagation of terahertz Monochromatic plane waves of light travel in the positive z
electromagnetic radiation.9 direction and are incident upon an aperture in the z�0 plane.
In addition to laser technology expanding the scales of A volume of space after the z�0 plane is enclosed by two
near-field beam propagation, manufacturing technology con­ surfaces, S 0 and S 2 . S 0 is planar in shape and lies in the z
tinues to shrink the scale of optical components. Micro- �0 plane, and S 2 is an arbitrary surface in the z�0 region.
electrical-mechanical systems and optical micro-electrical­ The enclosed volume has a small hollow sphere with a sur­
mechanical components are now reaching the dimensions of face of S 1 . This sphere surrounds the point of interest, P 1 .

1195 Am. J. Phys. 72 �9�, September 2004 http://aapt.org/ajp © 2004 American Association of Physics Teachers 1195
Table I. Various diffraction theory models and their approximations.

Model Approximation
�� S1
� E i �E i �
� U� � U � •n̂ ds 1 �4 � E i � P 1 � , �4�

Full wave equations None

Complete Rayleigh–Sommerfeld Boundary conditions where E i represents either the x, y, or z component of the
on surfaces electric field for each point on the surface, and E i ( P 1 ) is the
Approximate Rayleigh–Sommerfeld � �� ith component at position P 1 . If we use the Sommerfeld
Fresnel–Kirchhoff � �� radiation condition6 that the radiation across the surface S 2 is
Fresnel � composed of only outward waves and let the distance from
z31� ��x1�x0�2��y1�y0�2�2
4� P 1 to the surface of S 2 become very large, the integral over
Fraunhofer � S 2 also vanishes. Equation �3� now becomes
z1� �x21�y21�
�

E i� P 1 � �
1
4�
�� S0
� E i �E i �
� U� � U � •n̂ ds 0 . �5�

We use the divergence form of Gauss’ theorem, several Rayleigh–Sommerfeld boundary conditions on the aper­
vector identities, and a particular choice of a vector function ture plane state that in the aperture region of the aperture
to arrive at Green’s scalar theorem,6 plane the field distribution U and its normal derivative are

��
exactly as they would be in the absence of the screen; every­
� V�V�
� U� � U � •n̂ ds where else in the aperture plane either the field distribution
S U or its normal derivative is exactly zero. Both of these

���
boundary conditions on U in the aperture plane are satisfied
� � U� 2 V�V� 2 U � d v , �1� by the following Green’s function:
�

where U and V are arbitrary scalar fields that are functions of e �ik � r� 2 � r� 0 � e �ik � r� 1 �r� 0 �
U� � , �6�
position. The only restriction on the functions U and V are � r� 2 �r� 0 � � r� 1 � r� 0 �
that they be smooth and continuous at all points in space.
The surface integration is over all surfaces: S 0 , S 1 , and S 2 . where �r 0 denotes the vector from the origin to a point on the
The volume integration is over the volume enclosed by S 0 z�0 plane, r� 1 denotes the vector from the origin to point
and S 2 , with the exception of the volume enclosed by S 1 . If P 1 , and �r 2 denotes a mirror image of point P 1 on the nega­
U and V also are restricted to be solutions of the Helmholtz tive side of the z�0 plane. The scalar function U in Eq. �6�
wave equation, satisfies the Rayleigh–Sommerfeld boundary conditions and
represents a point source located at r� 1 and a mirror image
� � 2 �k 2 � U�0 and � � 2 �k 2 � V�0, �2� point source located at r� 2 oscillating exactly 180° out of
then the volume integral is equal to zero, and Eq. �1� be­ phase with each other. If we substitute Eq. �6� into Eq. �7�, it
comes can be shown that the field at P 1 is

�� S0
� V�V�
� U� � U � •n̂ ds 0
� � P1��
E
kz 1
i2 �
�� E z�0
e ik �
� 2 1�� 1
ik � �
dx 0 dy 0 , �7�

� �� S2
� V�V�
� U� � U � •n̂ ds 2 where

� �� S1
� V�V�
� U� � U � •n̂ ds 1 �0. �3�
� � �� x 1 �x 0 � 2 � � y 1 �y 0 � 2 �z 21 . �8�

Equation �7� is the full Rayleigh–Sommerfeld solution. The

If we let V represent a component of the electric field, and only assumptions are the boundary conditions on the sur­
the radius of S 1 go to zero, it can be shown �see Ref. 6, p. faces S 0 and S 2 . The Rayleigh–Sommerfeld diffraction in­
42� that the integral over S 1 becomes tegral is more commonly expressed in its simplified form as

E� � P 1 � �
kz 1
i2 �
��� E z�0
e ik �
�2
dx 0 dy 0 , �9�

where it has been assumed that � ��. 6 Equation �9� is also

known as the Fresnel–Kirchhoff diffraction integral.
If Eq. �8� is expanded and we assume that

�
z 31 � �� x 1 �x 0 � 2 � � y 1 �y 0 � 2 � 2 , �10�
4�

Fig. 1. Representation of volumes and surfaces used in establishing a dif­ and � 2 �z 2 in the denominator of Eq. �9�, we arrive at the
fraction theory. Fresnel �paraxial� near-field diffraction integral,

1196 Am. J. Phys., Vol. 72, No. 9, September 2004 Glen D. Gillen and Shekhar Guha 1196
 ke ikz 1
� 1� P 1 � �
E
i2 � z 1

� ��� E z�0 e � ik/2z 1 � ((x 1 �x 0 )

2 �(y �y ) 2 )
1 0 dx 0 dy 0 .

�11�
Equation �11� can be expressed in cylindrical coordinates for
a circular aperture of radius a as
2 � e ikz 1
� � P1��
E
ikz 1
2
�e � ikr 1 /2z 1 �� 0
a 2
E z�0 e � ikr 0 /2z 1 J 0 � �
kr 0 r 1
z1
r 0 dr 0 ,

�12�
where r 0 and r 1 are the radial coordinates in the planes z
�0 and z�z, respectively, and the point P 1 is (x 1 ,y 1 ,z).

III. CALCULATED RESULTS

Figure 2 illustrates the on-axis calculated intensities for
the three models discussed thus far. The integration of the
diffraction integral for each model derived in Sec. II was
done using Mathcad. Figure 2�a� is the calculated on-axis
intensity using the complete Rayleigh–Sommerfeld model,
Eq. �7�, and Fig. 2�b� is calculated using the approximate
Rayleigh–Sommerfeld or Fresnel–Kirchhoff model, Eq. �9�;
Fig. 2�c� is calculated using Fresnel’s paraxial approxima­
tion, Eq. �11�. Each graph of Fig. 2 is calculated as a function
of position along the z-axis with x 1 �y 1 �0, for incident
plane waves with a wavelength of 10 �m and a round aper­
ture with a radius of 100 �m in the aperture plane. The
amplitude of the electric field of the incident light is assumed
to be unity.
According to Fig. 2�a�, if we were to place a detector at Fig. 2. Calculated on-axis intensity distributions for �a� the complete
Rayleigh–Sommerfeld, �b� approximate Rayleigh–Sommerfeld, and �c� the
position (0,0,z), where z is on the order of 1000a, or 1000 Fresnel diffraction model. In �b� and �c� the arrow represents the lower limit
aperture radii, and observe the central intensity as we reduce of the validity of the model.
z, we would first observe an increasing intensity as a func­
tion of smaller distances to the aperture, which is what we
would expect. But, the increasing axial intensity does not
simply continue to increase all the way to the location of the the complete Rayleigh–Sommerfeld model. The vertical ar­
rows in Figs. 3�a� and 3�b� are at the same longitudinal dis­
aperture, as would be expected for a 1/z 2 intensity depen­
tance as those in Figs. 2�b� and 2�c�. As would be expected,
dence for a point source. Instead, a primary maximum is
the deviation of each approximate model is gradual; starting
reached at a position of 10a and an intensity oscillation is slightly beyond the lower validity threshold and deviating
observed for smaller values of z, with an overall amplitude farther and farther from the complete model as the aperture is
decrease. It also would be expected that as z asymptotically approached. The difference between the complete Rayleigh–
approaches the position of the aperture, the intensity should Sommerfeld model, Eq. �7�, and the approximate model, Eq.
approach that of the incident plane waves, or unity for this �9�, is primarily an amplitude inaccuracy, whereas the
example. Closer inspection of Fig. 2�a� reveals that the ob­ paraxial approximation, Eq. �12�, has both amplitude and
served number of on-axis maxima is equal to a/� �which phase inaccuracies.
also is observed for other calculations performed using a In the same spirit as Fig. 2, Fig. 4 is an image plot as a
variety of a/� values�. For optically dense materials, n�1, function of the distance from the aperture, z, and the radial
the longitudinal location of the diffraction pattern also is pro­ distance from the z-axis, using the complete Rayleigh–
portional to n beyond the aperture, and will be discussed in Sommerfeld model and the same aperture and laser field as
Sec. IV. in Fig. 2. The integration of Eq. �7� is repeated for every
The vertical arrows in Figs. 2�b� and 2�c� represent the point in a grid in the y-z plane. If a screen were to be placed
lower region of validity of each model, as defined in Table I. at a particular distance from the aperture, then the observed
For Fig. 2�b�, the vertical arrow is at z��, while the arrow image would be radially symmetric with a radial intensity
in Fig. 2�c� is at z� �� 4/3/4�. Figure 3 is a closer view of the profile equal to a vertical stripe through Fig. 4. To help illus­
longitudinal region where the approximate models begin to trate the intensity profiles and radial symmetries, Fig. 5
deviate from the axial intensity distributions calculated using shows a collection of radial intensity distributions for various

1197 Am. J. Phys., Vol. 72, No. 9, September 2004 Glen D. Gillen and Shekhar Guha 1197
Fig. 3. Comparison of the complete Rayleigh–Sommerfeld model to the �a�
approximate Rayleigh–Sommerfeld and the �b� Fresnel models.

distances from the aperture as calculated directly from Eq.

�7�, for points along the (0,y,z) line where z is held constant.
The four beam profiles of Fig. 5 are a demonstration of the
variety of radial intensity distributions that can be observed
as a function of distance from the aperture. Figure 5�a� is
calculated for a distance of 10 aperture radii, 1 mm, or the
location of the primary maxima. A single spot would be ob­
served on an image screen placed at this location with the
intensity at the center four times greater than that of the
incident light on the aperture plane. Figure 5�b� is calculated
for a distance of 4.9 aperture radii, 490 �m, or the location
of the first minima. If an image screen were to be placed Fig. 5. Calculated radial intensity beam profiles for axial distances of �a� 10,
there, a single ‘‘doughnut’’ shape would be observed. Figure �b� 4.9, �c� 1.86, and �d� 0.05 multiples of the aperture radii. The vertical
5�c� shows the calculated radial intensity distribution for a axes are normalized by an incident intensity on the aperture of unity.
distance of 1.86 aperture radii, or the location of the third
maxima. At this location, a bright central spot would be ob­
served circumscribed by two blurry circles of light. As the
and size, eventually fading together �as the contrast between
image screen is moved closer to the aperture, the central
the maxima and minima decreases� to form a near uniform
region oscillates between a maximum and a minimum and
spot with a radius equal to that of the aperture and an inten­
the number of rings surrounding it increases in both number
sity of unity, as seen in Fig. 5�d�. The latter represents a
calculation for a distance of only 5 �m from a 200 �m
diameter aperture.

IV. VERIFICATION OF SNELL’S LAW USING

NEAR-FIELD DIFFRACTION
We next apply the complete Rayleigh–Sommerfeld dif­
fraction model to the refraction of light at the interface of
two different media and compare the results to those pre­
dicted using Snell’s law of refraction and geometrical optics.
For plane waves incident upon the aperture at an angle other
than normal, a position-dependent phase term must be in­
Fig. 4. Calculated intensity distributions as a function of both radial and cluded at the aperture plane. The electric field of the incident
axial distance from the aperture using the complete Rayleigh–Sommerfeld light can be written as
model. White represents the maximum intensity, black the minimum inten­
sity, and grey in between. E� � x 0 ,y 0 ,0� �e i2 � n 0 x 0 sin( � )/� x̂ , �13�

1198 Am. J. Phys., Vol. 72, No. 9, September 2004 Glen D. Gillen and Shekhar Guha 1198
 profile offset from x�0° on the image plane by z tan(�), or
577 �m at z�1 mm. The resulting intensity profile is illus­
trated in Fig. 6�b� and is centered on x�578 � m. The dif­
ference between the transverse locations of the two is within
a step size �30 �m� of the Rayleigh–Sommerfeld calculation.
The geometrical shadow of the 200 �m diameter pinhole for
light incident at an angle of 30° would be an ellipse with a
minor diameter of 200 �m in the y dimension, and a major
diameter of 231 �m in the x dimension. The uniform inten­
sity would have a value of 0.866 for an incident electric field
amplitude of unity. This value is slightly higher than the ratio
Fig. 6. Radial intensity profiles for plane waves with a wavelength of 10 of the peak heights of Figs. 6�a� and 6�b� of 0.825. But,
�m, incident upon a 200 �m diameter aperture, and an aperture-image plane neither of these intensity profiles of Fig. 6 is close to uni­
distance of 1 mm for �a� n�n 0 �1, � �0°, �b� n�n 0 �1, � �30°, and �c�
form, as predicted using geometrical optics. Closer inspec­
n�4, n 0 �1, � �30°.
tion of Fig. 6�b� reveals that the intensity distribution is not
just simply wider and lower than that of Fig. 6�a�, as would
be expected using geometrical optics, but actually has some
where n 0 is the refractive index of in the medium before the
asymmetries in the outer-lying rings; another result of the
aperture, x 0 is the radial distance from the center of the ap­ wave nature of light using diffraction theory.
erture in the x direction, � is the angle of incidence, and x̂ is Finally, to truly compare the results of Rayleigh–
the polarization direction of the laser light. The amplitude of Sommerfeld calculations to Snell’s law, we include an opti­
the electric field is again assumed to be unity. If we substitute cally dense medium in the region after the aperture. The
Eq. �13� into Eq. �7� and use n as the refractive index of the refractive index chosen for this calculation is that of germa­
medium after the aperture, we obtain nium, a common optical material for long-wave infrared ap­

� � P1��
E
nz 1
i�
�� e i2 � n 0 x 0 sin( � )/�
plications, with a linear refractive index of n�4. For each
point on the x-axis, and a distance z�1 � m away, Eq. �14� is
integrated using n 0 �1, n�4, and � �30°. The resulting in­
�
� 2 �
e i2 � n � /�
1�
�
i2 � n � �
dx 0 dy 0 x̂. �14�
tensity profile is illustrated in Fig. 6�c�. If we use Snell’s law
of refraction, the image would be centered on x�126 � m,
which is the precise location of the minimum in the middle
Equation �15� is integrated over all points in the aperture of Fig. 6�c�.
plane for each point on the image plane. The integral is then The greatest difference between Figs. 6�b� and 6�c�, other
repeated for points along the line (x,0,z p ), where z p is con­ than their locations on the x-axis, is the difference in their
stant, and is the distance from the aperture to the longitudinal intensity profiles. As the refractive index of the medium after
position of the image plane of interest. The beam intensity the aperture changes, so does the relative distance to a par­
profiles as calculated are a function of x using this method ticular interference pattern. For example, the intensity profile
and are displayed in Fig. 6. The experimental conditions for calculated for Fig. 6�b� is very similar to the beam profile
Fig. 6 are those of a collimated CO2 laser, with a wavelength calculated for n�1, � �0° and a longitudinal distance closer
of 10 �m, incident upon an aperture with a radius of 100 to the aperture by a factor of 4 �the n value of germanium�.
�m. The beam waist is assumed to be much larger than the The net result of beam propagation through a nonabsorbing,
aperture �so a constant field amplitude across the aperture optically dense medium after the aperture is a diffraction
can be assumed�. The distance from the aperture to the image pattern similar to that of Fig. 4, except the longitudinal axis
plane is chosen to coincide with the primary maxima of the is increased by a factor of n.
interference pattern for beam propagation through air or
vacuum after the aperture.
V. PROPAGATION OF NEAR-FIELD PATTERNS
We first calculate the beam profile using Eq. �14� for light
waves normally incident upon the aperture and for a refrac­ THROUGH A LENS
tive index of the medium after the aperture equal to that Modeling light propagation through a hard aperture is just
before the aperture plane, n�n 0 �1. The resulting beam pro­ one step in accurately modeling beam propagation through a
file is shown in Fig. 6�a� and is centered at x�0. This beam complex optical train of components. The addition of a lens
intensity profile is the same as that of Fig. 4�a� which was to the system is an example of applying near-field diffraction
calculated using Eq. �7� for the same experimental condi­ theory to more complex systems. The lens is placed after the
tions. If we use geometrical optics and Snell’s law, the ex­ aperture so that the entire optical ‘‘train’’ for this example
pected intensity profile would be a perfect copy of the uni­ consists of a circular hard aperture and a lens some distance
form ‘‘flat-top’’ intensity profile incident upon the aperture beyond the aperture.
and centered around x�0, that is, a beam profile having a The general approach for modeling this example is de­
constant intensity with amplitude of unity would be expected picted in Fig. 7. The incident light on the aperture, located in
from �100 � m to �100 � m with sharp edges. Figure 6�a� plane 1 (z�0), is assumed to be a plane wave traveling in
shows that if we incorporate the wave nature of light and a the �z direction. The electric field for every point in plane 2
mathematically rigorous model of diffraction theory, we ob­ is calculated using a suitable near-field diffraction theory.
serve a much different intensity profile. The electric field is then calculated for plane 3 by incorpo­
If we perform a similar calculation, except allow for an rating the changes to the electric field induced by the lens.
angle of incidence of 30°, we might expect a similar beam Finally, by using near-field diffraction theory, the electric

1199 Am. J. Phys., Vol. 72, No. 9, September 2004 Glen D. Gillen and Shekhar Guha 1199
Fig. 7. Experimental setup for the propagation of near-field diffraction pat­
terns through a lens. The 5 in. focal length lens is placed 6.25 in. away from
a 2 mm diameter pinhole. The incident light is from a 10.6 �m wavelength,
CO2 laser.

field at the point of interest, P 4 , is calculated by propagating

the electric field in plane 3 to the point of interest.
The near-field diffraction model chosen to calculate the
electric field distribution on plane 2, just before the lens, is
that of Fresnel diffraction, or Eq. �12�. For this example, the
paraxial approximation was chosen over the complete
Rayleigh–Sommerfeld model because the axial distance be­
tween the aperture and the lens is well within the region of
validity of this approximation. If we use Eq. �12�, we can
express the electric field everywhere on plane 2 as

2 � e ikz 2
� ikr 2 /2z
E� 2 � r 2 ,z 2 � � e 2 2

ikz 2

� ��
0
a 2
E � r 1 � e � ikr 1 /2z 2 J 0 � �
kr 1 r 2
z2
r 1 dr 1 .

�15�

As the mathematical model for the lens, the Fourier optical

thin lens approximation was chosen.6 Thus, planes 2 and 3
are coplanar, and the electric field at plane 3 is the electric
field at plane 2 multiplied by a quadratic phase transforma­
tion term,
2
E 3 � r 3 ,z � �E 2 e �i � k/2 f � r 3 e �ikn 1 � 0 , �16�

where f is the focal length of the lens, k is the wave number

of the light, n 1 is the refractive index of the lens material,
and � 0 is the lens thickness at the center.
The final step in this example is to calculate the electric
field for points of interest beyond the lens. Once again, we Fig. 8. Beam intensity profiles measured over distances of �a� 313 mm, �b�
373 mm, and �c� 406 mm from the lens. The image scans for �a� and �b�
are interested only in the electric field at points far from the cover an area of 2 mm by 2 mm, while the image for �c� is an area of 2.625
axial position of the lens and choose to use the paraxial ap­ mm by 2.625 mm. �a� The experimental and calculated on-axis relative
proximation using plane 3 as the aperture plane and Eq. �16� intensities versus axial distance from the lens.
as the input electric field. The electric field at point P 4 is

E� � P 4 � �
2 � e ikz 4
ikz 4
� 0
�
E� 3 � r 3 � e � ikr 3 /2z 4 J 0
2
� �
kr 3 r 4
z4
r 3 dr 3 . E 4 � r 4 ,z 4 � �
id 1
� � ��
d 21d 42
exp �i
kr 241
2d 42
1�

d 1

d 42
�17�

The substitution of Eq. �15� into Eq. �16�, and Eq. �16� into
Eq. �17�, leads to a complicated function. The quadratic ex­
� � � � ��
0
1
exp �i
kr 211
2d 21
1�
d1
d 21
ponential and Bessel function terms of the infinite integral of
the resulting equation can be reduced to a quadratic expo­
nential term and a modified Bessel function, eliminating the
�J 0 � � d 1 r 11r 41
r dr ,
d 21d 42 11 11
�18�

infinite integral.17 Further simplifications lead to the follow­ where r 11 , r 41 , d 21 , d 42 , and d 1 are the dimensionless quan­
ing expression for the electric field for any point P 4 (r 4 ,z 4 ): tities:

1200 Am. J. Phys., Vol. 72, No. 9, September 2004 Glen D. Gillen and Shekhar Guha 1200
 r1 r4 cussed the approximations and limitations of each. The com­
r 11� , r 41� , �19a� plete Rayleigh–Sommerfeld model is applied to the tradi­
a a
tional geometrical optics example of refraction of light at an
z 2� � z 4 �z 2 � � interface between two different optical media. The results of
d 21� , d 42� , �19b� the calculation not only agree with predictions using geo­
2�a2 2�a2
metrical optics, but also demonstrate the wave nature of light
1 1 1 and the diffraction behavior of physical optics. We also dem­
d 1� � � . �19c� onstrated that electromagnetic wave propagation through
d 21 d 42 f
more complex optical systems can be modeled using near-
The distance a is the aperture radius. field diffraction techniques. The propagation of near-field
For this example, the experimental conditions chosen are diffraction patterns through a simple optical system is dem­
those of CO2 laser light �10.6 �m wavelength� incident upon onstrated as they manifest themselves farther down the opti­
a 2 mm diameter circular aperture and passing through a 5 cal train, and are magnified from the sub-millimeter to the
in. focal length lens placed 6.25 in. beyond the aperture. This decimeter scale. The observed on-axis intensity variations
particular arrangement is chosen using hindsight from a pre­ are consistent with calculations using Fresnel diffraction in­
vious experimental setup that produced large variations in tegrals and Fourier optics.
the transverse beam profiles as a function of the longitudinal
distance from the lens. Figures 8�a�– 8�c� are examples of a�
Electronic address: [email protected]; also at Anteon Corporation,
beam profiles measured for a variety of axial distances from Dayton, Ohio 45433.
the lens. The beam profiles of Figs. 8�a� and 8�b� span a 1
Frank L. Pedrotti and Leno S. Pedrotti, Introduction to Optics �Prentice–
space of 2 mm by 2 mm, and the beam image of Fig. 8�c� Hall, Englewood Cliffs, NJ, 1987�.
2
Francis A. Jenkins and Harvey E. White, Fundamentals of Optics
spans 2.625 mm by 2.625 mm. As we look at the beam
�McGraw–Hill, New York, NY 1976�, 4th ed.
profile just beyond the lens and slowly move farther away, 3
Grant R. Fowles, Introduction to Modern Optics �Dover, New York, 1975�,
the observed image is initially a large Gaussian-style spot 2nd ed.
that grows in central intensity and narrows in width. After 4
Eugene Hecht, Optics �Addison–Wesley, San Francisco, CA, 2002�, 4th
reaching a maximum intensity and minimum beam waist, the ed.
beam profile does not exhibit typical Gaussian characteris­ 5
Max Born and Emil Wolf, Principles of Optics �Cambridge U.P., New
tics. Instead, the beam profile characteristics match those of York, 1999�, 7th ed.
beam profiles of near-field diffraction patterns, calculated in
6
Joseph W. Goodman, Introduction to Fourier Optics �McGraw–Hill, New
York, 1996�, 2nd ed.
Figs. 4 and 5. Figure 8�a� shows a beam intensity scan for a 7
B. I. Greene, J. F. Federici, D. R. Dykaar, R. R. Jones, and P. H. Bucks­
distance of 313 mm from the lens where a single radial mini­ baum, ‘‘Intererometric characterization of 160 fs far-infrared light pulses,’’
mum is observed in the center, similar to the first minimum Appl. Phys. Lett. 59, 893– 895 �1991�.
8
observed as an illuminated aperture in the near-field regime R. R. Jones, D. You, and P. H. Bucksbaum, ‘‘Ionization of Rydberg atoms
is approached �see Fig. 5�b��. Figures 8�b� and 8�c� are beam by subpicosecond half-cycle electromagnetic pulses,’’ Phys. Rev. Lett. 70,
intensity scans for axial locations of the next maxima and 1236 –1239 �1993�.
9
E. Budiarto, N.-W. Pu, S. Jeong, and J. Bokor, ‘‘Near-field propagation of
minima observed for increasing z. For each beam intensity terahertz pulses from a large-aperture antenna,’’ Opt. Lett. 23, 213–215
scan measured, the relative central intensity is displayed in �1997�.
Fig. 8�d� as a function of the distance from the lens. The 10
D. M. Bloom, ‘‘Grating light valve: revolutionizing display technology,’’
solid line displayed in Fig. 8�d� is a result of numerically Proc. SPIE 3013, 165–171 �1997�.
integrating Eq. �18� for points along the axis of propagation 11
L. Y. Lin, E. L. Goldstein, and R. W. Tkach, ‘‘Free-space micromachined
for the same experimental conditions. Calculated on-axis in­ optical switches with submillisecond switching time for large-scale optical
crossconnects,’’ IEEE Photonics Technol. Lett. 10, 525–527 �1998�.
tensities predict a single bright spot followed by oscillations 12
T. P. Kurzweg, ‘‘Optical propagation methods for system-level modeling
with an overall decreasing amplitude. The agreement be­ of optical MEM systems,’’ Ph.D. thesis, Electrical Engineering, University
tween the observed relative on-axis intensities and those pre­ of Pittsburgh, PA, 2002.
dicted using near-field diffraction is consistent out to a dis­ 13
S. P. Levitan, J. A. Martinez, T. P. Kurzweg, A. J. Davare, M. Kahrs, M.
tance of approximately 470 mm. Beyond this distance, the Bails, and D. M. Chiarulli, ‘‘System simulation of mixed-signal multi-
maxima and minima of the experimental beam scans begin to domain microsystems with piecewise linear models,’’ IEEE Trans.
blur together as the contrast between them decreases. Comput.-Aided Des. 22, 139–154 �2003�.
14
G. R. Kirchhoff, ‘‘Zur theorie der lichtstrahlen,’’ Ann. Phys. �Leipzig� 18,
663– 695 �1883�.
VI. CONCLUSIONS 15
A. Sommerfeld, ‘‘Zur mathematischen theorie der beugungsercheinun­
gen,’’ Nachr. Kgl. Akad. Wiss. Göttingen 4, 338 –342 �1894�
We have used Gauss’ theorem, several vector identities, 16
Lord Rayleigh, ‘‘On the passage of waves through apertures in plane
and Green’s scalar theorem to derive the complete and ap­ screens, and allied problems,’’ Philos. Mag. 43, 259–272 �1897�.
proximate Rayleigh–Sommerfeld diffraction integrals and 17
I. S. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and Products
the Fresnel �or paraxial� diffraction integral, and have dis­ �Academic, New York, 1980�, 4th ed., Eq. �6.633.2.�.

1201 Am. J. Phys., Vol. 72, No. 9, September 2004 Glen D. Gillen and Shekhar Guha 1201

View publication stats