You are on page 1of 3

492 J. Opt. Soc. Am. A / Vol. 27, No.

3 / March 2010 Wei Guo

Light-scattering theory of diffraction

Wei Guo
Department of Natural Sciences, Queens University of Charlotte, 1900 Selwyn Avenue,
Charlotte, North Carolina 28274, USA (weiguoguo@hotmail.com)

Received August 27, 2009; revised January 13, 2010; accepted January 14, 2010;
posted January 14, 2010 (Doc. ID 116311); published February 23, 2010
Since diffraction is a scattering process in principle, light propagation through one aperture in a screen is dis-
cussed in the light-scattering theory. Through specific calculation, the expression of the electric field observed
at an observation point is obtained and is used not only to explain why Kirchhoff ’s diffraction theory is a good
approximation when the screen is both opaque and sufficiently thin but also to demonstrate that the
mathematical and physical problems faced by Kirchhoff ’s theory are avoided in the light-scattering theory.
© 2010 Optical Society of America
OCIS codes: 260.1960, 290.4210.

When light propagation through one aperture in a screen In addition to the mathematical problem, Kirchhoff ’s
is discussed, it is usually treated as a diffraction process theory has other flaws from the viewpoint of physics.
by following Kirchhoff ’s theory. More specifically, the elec- First, the observed field is assumed to be formed by wave-
tric field ␺共rជ o兲 observed at an observation point rជ o is as- lets from the wavefront of the incident field inside the ap-
sumed to originate from the incident field ␺inc inside the erture (Huygens’s principle), but it is unclear what actu-
aperture and is determined through the following integral ally drives the wavelets away from the wavefront. Second,
relation [1]: the integral relations in Eqs. (1) and (2) do not take the

冕 冋 冉 冊 册
width of the screen into consideration [9] and, as a result,
1 eikR i ជ
R cannot satisfy the requirement that when the width ap-
␺共rជ o兲 = − ជ · ⵜ␺inc + ik 1 +
n ␺inc da,
4␲ R kR R proaches zero, that is, when the screen is absent, the ob-
served field must reduce to the incident field. Third, if the
共1兲 cross-section of the aperture shrinks to nonexistence, the
where nជ is a normal to the screen and points in the direc- observed field must become the transmitted field through
tion toward the observation point, R = 兩Rជ 兩 = 兩rជ o − rជ 兩 the sepa- the screen in the absence of the aperture; unlike what the
ration between rជ o and an arbitrary point rជ in the aper- preceding equations suggest, the transmitted field should
ture, and k the wavenumber of ␺inc. The surface not strictly be equal to zero in principle. All these math-
integration in Eq. (1) is only over the aperture, with da as ematical and physical problems can be traced to the fact
the area element. In its form in Eq. (1), Kirchhoff ’s theory that Kirchhoff ’s theory is built on Green’s theorem, which
is often criticized [1,2] to have a mathematical problem of is a mathematical identity but not a physical principle.
inconsistency. One way to correct this problem is to intro- In the present work, light propagation through one ap-
duce either a Neumann or a Dirichlet Green function [1]. erture in a screen is revisited from the viewpoint of light
For example, for the Dirichlet Green function, the rela- scattering. As pointed out in [1], scattering and diffraction
tion in Eq. (1) becomes are not logically separate. Based on the light-scattering
theory, the validity of Kirchhoff ’s theory is also examined.

␺共rជ o兲 =
2␲i
k
冕 冉eikR
R
1+
i
kR
冊 n
R

ជ ·R
␺incda. 共2兲
For the purpose of simplicity, the present discussion is
limited to the domain of scalar diffraction, which, as
pointed out in [1], suffices in optics. Selected papers on
On the other hand, if the observed field is viewed to be scalar light diffraction can be found in [10].
created after the incident field is scattered from the edge The screen is assumed to be constituted by identical at-
of the aperture, Eq. (1) can even be argued to be a rigor- oms and to occupy the space between z = 0 and z = d. Thus,
ous solution of a certain boundary-value problem [3]. The the width of the screen is d. The volume of the space is
latter approach, known as the boundary diffraction denoted as V. Inside the screen exists a cylindrical aper-
theory, is also used to discuss the half-plane problem ture whose symmetric axis coincides with the ẑ axis of the
[4,5]. A different interpretation of Eq. (1) is due to Kottler, coordinate system; the cross-section of the aperture is S.
who showed that this equation is also an exact solution of According to the light-scattering theory, for an incident
a certain saltus problem [6,7]. See [8] for other methods plane field approaching the screen from the z ⬍ 0 region

commonly used to remove the mathematical inconsis- ␺inc共rជ 兲 = eik·rជ , where kជ is the field’s wave vector 共k = 兩kជ 兩兲, the
tency in Kirchhoff ’s theory. observed field ␺m at rជ o (assumed to be in the z ⬎ 0 region)

1084-7529/10/030492-3/$15.00 © 2010 Optical Society of America


Wei Guo Vol. 27, No. 3 / March 2010 / J. Opt. Soc. Am. A 493

must be a superposition of the incident field and fields


␺e = ␺d − ˜␣共1 − ˜␣ĜV兲−1Ĝ⍀␺d
scattered from all the atoms, not just from those forming
the edge of the aperture [3]: + ˜␣2共1 − ˜␣ĜV兲−1Ĝ⍀共1 − ˜␣ĜV兲−1Ĝ⍀␺d − ¯ , 共9兲

␺m共rជ o兲 = ␺inc共rជ o兲 + 4␲k2␣n 冕 ⍀⬘


G共rជ o,rជ 1兲␺e共rជ 1兲dx1dy1dz1 ,
where ␺d = 共1 − ˜␣ĜV兲−1␺inc describes multiple scattering of
the incident field by the atoms inside the slab and is thus
nothing other than the diffused field inside the slab. Due
共3兲 to the presence of the aperture, ␺e is certainly different
from ␺d and must be modified by those light-scattering
where ␣ and n are the polarizability and number density events between atoms formally in V and ⍀; see the other
of the atoms respectively, G共rជ , rជ 1兲 = eik兩rជ −rជ 1兩 / 共4␲兩rជ − rជ 1兩兲 the terms on the right-hand side of Eq. (9).
retarded Green’s function representing a spherical wave Substitute Eq. (9) into Eq. (6) to express ␺m in terms
radiated from rជ 1 to rជ , and ␺e共rជ 兲 the field experienced by an of ␺d:
atom at rជ . Note that, due to the aperture, the atoms can
occupy only a volume ⍀⬘, which is one part of V and ob- ␺m = ␺inc + ˜␣ĜV␺d − ˜␣Ĝ⍀␺d − ˜␣2ĜV共1 − ˜␣ĜV兲−1Ĝ⍀␺d
tained when ⍀, the volume of the aperture, is excluded
from V. Note also that the screen becomes a slab when its + ˜␣2Ĝ⍀共1 − ˜␣ĜV兲−1Ĝ⍀␺d + ¯ . 共10兲
aperture is filled with atoms identical to those forming
Each term on the right-hand side of Eq. (10) represents a
the screen; the density of these filled atoms must be n,
different event. For example, the second term means that
too. Consequently, the observed field ␺m from the screen
the diffused field ␺d is scattered from the slab directly to
must be equal to the field from the slab minus the field
the observation point and forms, with the incident field
from the atoms filled inside the aperture:
(the first term), the transmitted field ␺t through the slab.


See [12] for a multiple-scattering formulation of light
␺m共rជ o兲 = ␺inc共rជ 0兲 + 4␲k2␣n G共rជ o,rជ 1兲␺e共rជ 1兲dx1dy1dz1 propagation through a dielectric slab. On the other hand,
V in the third term, the diffused field should be understood


to be scattered by atoms formally inside the aperture to
the observation point. The remaining terms describe
− 4 ␲ k 2␣ n G共rជ o,rជ 1兲␺e共rជ 1兲dx1dy1dz1 . 共4兲

other events in which the diffused field is first multiply
scattered between atoms formally in the aperture and in
Similarly, the field ␺e, which is created as the incident the slab before finally being scattered toward the obser-
field is multiply scattered among the atoms in the volume vation point.
⍀⬘, must satisfy the following integral equation: Equation (10) is an exact description of light propaga-
tion through the screen. Nevertheless, for the purpose of


simplicity, so that the mathematical formulation will not
␺e共rជ 兲 = ␺inc共rជ 兲 + ˜␣ G共rជ ,rជ 1兲␺e共rជ 1兲dx1dy1dz1 swamp the physics intended to be illustrated, all higher-
V order terms on the right-hand side of Eq. (10), starting


from the fourth one, are ignored in the following discus-
− ˜␣ G共rជ ,rជ 1兲␺e共rជ 1兲dx1dy1dz1 , 共5兲 sion. This simplification is valid as long as ˜␣ is sufficiently
⍀ small. One way to achieve this requirement, for example,
is to require that the incident field be off-resonant with
where ˜␣ = 4␲k2␣n. the atoms. In the real space, the observed field ␺m subse-
quently simplifies to
Define two operators ĜV and Ĝ⍀ by their action on any
function f共rជ 兲 as ĜVf → 兰VG共rជ , rជ 1兲f共rជ 1兲dx1dy1dz1 and Ĝ⍀f
→ 兰⍀G共rជ , rជ 1兲f共rជ 1兲dx1dy1dz1 to transform Eqs. (4) and (5),
respectively, to
␺m共rជ o兲 ⯝ ␺t共rជ o兲 − ˜␣ 冕

G共rជ o,rជ 1兲␺d共rជ 1兲dx1dy1dz1 , 共11兲

where, as shown in [12],


␺m = ␺inc + ˜␣ĜV␺e − ˜␣Ĝ⍀␺e , 共6兲 ជ
␺d共rជ 兲 = eik·rជ ⬜D0兵关kz共␪兲 + k cos ␪兴eikz共␪兲z

+ 关kz共␪兲 − k cos ␪兴e2ikz共␪兲d−ikz共␪兲z其,


␺e = ␺inc + ˜␣ĜV␺e − ˜␣Ĝ⍀␺e . 共7兲

See [11] for an illustration of the same operator approach ␺t共rជ 兲 = 2D0kz共␪兲ei关kz共␪兲−k cos ␪兴deik·rជ .
applied to the formulation of light propagation inside a
In the expressions of the diffused and transmitted fields,
random medium. Shift ˜␣ĜV␺e in Eq. (7) to the other side which are all obtained analytically, rជ ⬜ is the projection of
and multiply the resultant expression by 共1 − ˜␣ĜV兲−1 to the position vector rជ on the xy plane, ␪ the angle between
get kជ and ẑ, kz共␪兲 = 冑k2 cos2 ␪ + ˜␣ the ẑ component of the dif-
fused wave’s wave vector, and D0 = 2k cos ␪兵关kz共␪兲
␺e = 共1 − ˜␣ĜV兲−1␺inc − ˜␣共1 − ˜␣ĜV兲−1Ĝ⍀␺e , 共8兲 + k cos ␪兴2 − 关kz共␪兲 − k cos ␪兴2e2ikz共␪兲d其−1. Note that kz共␪兲 is a
function of an effective index of refraction of the screen
which, through iteration, leads to a serial solution of ␺e: 冑1 + ˜␣共k cos ␪兲−2, which is determined by the atoms
494 J. Opt. Soc. Am. A / Vol. 27, No. 3 / March 2010 Wei Guo

through ˜␣. Similarly, because of kz共␪兲, the amplitudes of Kirchhoff ’s theory, that is, that in the aperture the values
␺d and ␺t all rely on the index of refraction and show a of the field are equal to the values of the incident field, is
certain resonance effect coming from the denominator of in principle invalid. But, since on the z1 = d plane the dif-

D0. See [12] for an illustration of such resonance. fused and incident fields are all proportional to eik·rជ 1⬜, Eq.
Equation (11) reveals that, in the light-scattering (13) is in fact roughly proportional to Eq. (2), making the
theory, the observed field is formed approximately by the latter equation a valid approximation of the field at the
transmitted field in the absence of the aperture and the observation point. Note that, in most cases, it is the rela-
field scattered from those atoms formally in the aperture tive magnitude of the observed field that is important.
after the atoms are excited by the diffused field. Com- Hence, the validity of Kirchhoff ’s theory must also re-
pared with Kirchhoff ’s theory, the light-scattering theory quire the screen involved to be a thin screen.
has several advantages. For example, since the light- From the discussion in the preceding two paragraphs,
scattering theory explains light propagation through the it is clear that, under the condition of an opaque, thin
screen from Maxwell’s equations, it does not have the screen, Kirchhoff ’s theory becomes approximately paral-
problems faced by Kirchhoff ’s theory. In order to under- lel to the light-scattering theory. Therefore, it should not
stand this point, it is necessary to note, for example, that be a surprise to find that Kirchhoff ’s theory is able to give
in the absence of the aperture, that is, when ⍀ → 0, the results in excellent agreement with experiment. The
observed field ␺m directly reduces to the transmitted field mathematical similarity between the relations in Eqs. (2)
␺t [see Eq. (11)] and that in the absence of the screen, that and (13), however, should not transcend the fact that they
is, when d = 0, the observed field reduces to the incident are different explanations of light diffraction. While in
field [see Eq. (3)]. Eq. (2) the observed field is explained to be formed by
In order for ␺m in Eq. (11) to be compared with ␺共rជ 0兲 in wavelets from the wavefront of the incident field inside
Eq. (2), which does not have the mathematical problem of the aperture, the observed field in Eq. (13) should be un-
inconsistency, it is convenient to transfer the volume in- derstood to be formed by the fields scattered from the at-
tegral in the former equation into a series of surface inte- oms formally in the aperture.
grals over the aperture surface S at z1 = d through re- In conclusion, the problem of light propagation through
peated application of integration by parts over z1: a screen with one aperture is analyzed in the light-


scattering theory. Since it is built on Maxwell’s equations,
␺m共rជ o兲 ⯝ ␺t共rជ 0兲 − ˜␣ dx1dy1共Ad − A共1兲d2/2 + A共2兲d3/3! the light-scattering theory does not have the inherent
S problems of Kirchhoff’s theory. The light-scattering theory
is also used to find under what conditions Kirchhoff’s
− A共3兲d4/4! + ¯ 兲, 共12兲 theory is valid.
where A = G共rជ o , rជ 1兲␺d共rជ 1兲 and A共n兲 denotes the nth deriva-
tive of A with respect to z1; both A and A共n兲 are evaluated REFERENCES
at z1 = d. Unlike Kirchhoff ’s theory, where the selection of 1. J. D. Jackson, Classical Electrodynamics (Wiley, 1975).
the surface over which the surface integral is carried out 2. M. Nieto-Vesperinas, Scattering and Diffraction in Physical
Optics (Wiley, 1991).
has some degree of freedom—sometimes curved surfaces 3. E. W. Marchand and E. Wolf, “Consistent formulation of
are used for convenience [13], the light-scattering theory Kirchhoff ’s diffraction theory,” J. Opt. Soc. Am. 56,
requires that the aperture surface S be a plane. 1712–1722 (1966).
For an opaque screen, the transmitted field ␺t is negli- 4. S. Ganci, “A general scalar solution for the half-plane
gibly small in magnitude and can be entirely ignored in problem,” J. Mod. Opt. 42, 1707–1711 (1995).
5. Y. Z. Umul, “Young–Kirchhoff–Rubinowicz theory of
the expression of ␺m. For this reason, one (unjustified) as- diffraction in the light of Sommerfeld’s solution,” J. Opt.
sumption in Kirchhoff ’s theory that there is no field ex- Soc. Am. A 25, 2734–2742 (2008).
cept in the aperture is practically acceptable. Note that 6. F. Kottler, “Diffraction at a black screen, Part I: Kirchhoff ’s
this assumption leads directly to the absence of the trans- theory,” in Progress in Optics, Vol. IV, E. Wolf, ed. (North-
Holland, 1965), pp. 281–314.
mitted field in Eq. (2). Kirchhoff ’s theory thus never 7. F. Kottler, “Diffraction at a black screen, Part II:
works unless the screen is at least opaque. Electromagnetic theory,” in Progress in Optics, Vol. VI, E.
If in addition the opaque screen is so thin that the se- Wolf, ed. (North-Holland, 1967), pp. 331–377.
ries in the integrand in Eq. (12) is practically equal to its 8. S. Ganci, “Equivalence between two consistent
first term alone, the observed field further simplifies to formulations of Kirchhoff ’s diffraction theory,” J. Opt. Soc.
Am. A 5, 1626–1628 (1988).


9. A. Roberts, “Electromagnetic theory of diffraction by a

␺m共rជ o兲 ⯝ − 2˜␣dD0kz共␪兲eikz共␪兲d G共rជ o,rជ 1兲eik·rជ 1⬜dx1dy1 , circular aperture in a thick, perfectly conducting screen,” J.
Opt. Soc. Am. A 4, 1970–1983 (1987).
S
10. K. E. Oughstun, ed., Selected Papers on Scalar Wave
共13兲 Diffraction, Vol. MS51 of SPIE Milestone Series (SPIE
Optical Engineering Press, 1992).
where the expression of ␺d is used. A comparison between 11. W. Guo and S. Prasad, “Multiple scattering of light from a
Eqs. (2) and (13) immediately demonstrates that in the random slab: a density-fluctuation approach,” Opt.
Commun. 212, 1–6 (2002).
surface integral, it is the diffused field, not the incident 12. W. Guo, “Multiple scattering of a plane scalar wave from a
field, that should be included in the integrand. In other dielectric slab,” Am. J. Phys. 70, 1039–1043 (2002).
words, the second (unjustified) assumption made in 13. M. V. Klein and T. E. Furtak, Optics, 2nd ed. (Wiley, 1986).

You might also like