1850 J. Opt. Soc. Am. A / Vol. 25, No. 8 / August 2008 Yusuf Z.

Umul

Modified diffraction theory of Kirchhoff

Yusuf Z. Umul
Faculty of Engineering and Architecture, Department of Electronics and Communication Engineering,
Cankaya University, Öğretmenler Cad., No. 14, Yüzüncü Yil, Balgat, 06530, Ankara, Türkey
([email protected])

Received March 26, 2008; revised May 13, 2008; accepted May 15, 2008;
posted May 20, 2008 (Doc. ID 94264); published July 2, 2008
The diffraction theory of Kirchhoff is reinterpreted and a new form of a surface diffraction integral is developed
by using the axioms of the modified theory of physical optics, which leads to the exact scattered fields by con-
ducting bodies. The new integral is arranged according to the interpretation of Young, and the diffracted waves
are expressed in terms of a line integral. The method is applied to the diffraction problem by a semi-infinite
edge contour. © 2008 Optical Society of America
OCIS codes: 000.3860, 050.1960, 260.0260, 260.1960.

1. INTRODUCTION ranged by using a modified surface normal and a new

form will be derived in terms of the unit vectors of inci-
The diffraction theory of Kirchhoff is obtained from the
dence and scattering. It is obvious that the modified form
integral solution of the Helmholtz equation by using the
of the Kirchhoff theory will lead to the rigorous field ex-
Green’s functions [1]. It defines the scattered fields by an
pressions from scattering by apertures and conducting
aperture in terms of a surface integral, which is named
bodies. The line integral reduction of the scattering inte-
the Kirchhoff–Huygen integral [2]. This method is widely
gral will also be studied. The importance of this reduction
used in optics in order to investigate the diffraction and
is the separation of the diffracted field from the geometri-
propagation of light [3,4]. The diffraction theory of Kirch-
cal optics (GO) wave. The given type of diffraction process
hoff gives the value of the field at an observation point in
was first supposed by Young [12] in a qualitative manner
terms of the incident wave on a suitable aperture. A simi-
and applied to the diffraction theory of Kirchhoff by Ru-
lar approach is the physical optics (PO), which is widely
binowicz [13]. However the asymptotic evaluation of the
used in electromagnetic scattering problems [5]. The
line integral does not yield the exact solution [14], al-
method of PO is generally obtained for the reflected and though the approach suggests a reasonable physical ex-
reflected scattered waves whereas the theory of Kirchhoff planation of the diffraction phenomenon. In this study we
diffraction takes into account the transmitted and trans- will also derive the classical potential function in a novel
mitted scattered fields by an aperture. In fact the two way by using vector algebra. The diffraction of plane
methods are common in the general construction of the waves by a semi-infinite edge contour will also be inves-
scattering integral and lead to similar results. The most tigated as an application of the new method. The results
important defect of these theories is the incorrect dif- will be plotted numerically.
fracted waves that are evaluated from the edge point con- In this study the variables of the cylindrical and spheri-
tributions [6]. The physical theory of diffraction (PTD) cal coordinates are represented by 共␳ , ␾ , z兲 and 共r , ␪ , ␾兲,
was developed by Ufimtsev in order to obtain the exact respectively. A time factor of exp共jwt兲 will be considered
diffracted fields by defining fringe currents that complete and suppressed throughout the paper.
the uniform surface currents of PO [7]. These fringe cur-
rents are evaluated from the exact solutions of some ca-
2. MODIFIED THEORY
nonical problems. But this process makes PTD dependent
on the rigorous solutions. Umul introduced a modified The behavior of light, like diffraction, reflection, or propa-
form of PO, named the modified theory of physical optics gation, is obtained from the solution of the Helmholtz
(MTPO), in order to directly derive the exact scattered equation, which can be given by
fields from the surface integral of PO [8]. He defined the ⵜ2u + k2u = 0, 共1兲
surface currents, induced by the incident waves, accord-
ing to three axioms, that consider the structure of the for u expresses the field. The integral solution of Eq. (1)
edge diffraction process. The MTPO was later expanded can be represented as
for the wedge diffraction problems by conducting surfaces
[9]. Recently the line integral reduction of MTPO surface
integrals has been derived for edge and wedge diffraction
u共P兲 = 冕冕
S
ជ dS,
关u共Q兲ⵜQG共P,Q兲 − G共P,Q兲ⵜQu共Q兲兴 · n

[10,11].
共2兲
It is the aim of this study to reconsider and extend the
diffraction theory of Kirchhoff by taking into account the in terms of Green’s functions [15]. P and Q are the obser-
axioms of the MTPO. The surface integral will be rear- vation and integration points, respectively. S is the sur-

1084-7529/08/081850-11/$15.00 © 2008 Optical Society of America

Yusuf Z. Umul Vol. 25, No. 8 / August 2008 / J. Opt. Soc. Am. A 1851

face of diffraction. G共P , Q兲 is the Green’s function, which

is a function of both the observation and integration
points. The point of Q changes on the surface of S. n ជ is the
unit normal vector of the surface, which is directed to-
ward P. It is mentioned in [8,16] that the defect of PO and
Kirchhoff ’s theory arises from the constant value of n ជ.
This defect is eliminated in MTPO by defining a variable
unit normal vector for the scattering surface. The variable
nជ is reasonable since the edge diffracted rays define
changeable normal vectors that divide the angle between
the incident and diffracted rays into two equal parts.
Since PO and Kirchhoff ’s theory considers a constant nor-
mal vector, the edge contribution is also defined according
ជ , yielding incorrect diffracted waves. The variable
to this n Fig. 2. (Color online) Surface of integration for the Kirchhoff
theory.
nature of the unit normal vector at the diffraction point is
given in Fig. 1. The cosine of the angle between the inci-
dent (or diffracted) ray and n ជ 1 is equal to for the choice of the integration surface. First of all the
surface must include all of the incident rays or the wave-
␾−␣
cos ␯ = cos , 共3兲 front according to the Huygens principle. The second cri-
2 terion is the unnecessity of considering a closed surface
that also contains regions where the incident rays do not
according to the geometry in Fig. 1. It is apparent that directly hit.
this term gives the exact diffraction coefficient of Keller Equation (4) can be rearranged as
[17] for the incident diffracted waves. We will rewrite Eq.

冕冕 冋 册
(2) as
1 ⳵ G共P,Q兲 ⳵ u共Q兲

冕冕
1 u共P兲 = u共Q兲 − G共P,Q兲 dS. 共5兲
ជ 1dS, 4␲ ⳵ n1 ⳵ n1
u共P兲 = 关u共Q兲ⵜQG共P,Q兲 − G共P,Q兲ⵜQu共Q兲兴 · n
4␲ S
S

共4兲 The partial derivatives can be expressed as

by considering a variable unit normal vector for the scat- ⳵ G共P,Q兲 ⳵ G共P,Q兲
terer’s surface. In fact we think of all the points on the = cos共n ជ 兲,
ជ 1,R 共6兲
integration surface as separate points of diffraction as Qe, ⳵ n1 ⳵R
in Fig. 1. Since the points are continuous along the reflec-
tion or transmission surface, the stationary phase evalu-
ation of the diffraction integral gives the GO fields. The ⳵ u共Q兲 ⳵ u共Q兲
= cos共n ជ 兲,
ជ 1,R 共7兲
i
points that are in the edges of the surface are discontinu- ⳵ n1 ⳵ Ri
ous and lead to the edge diffracted waves. Figure 2 shows
the incident and scattered rays with the integration sur- according to the geometry in Fig. 3. New unit vectors can
face of S. The surface does not wholly cover the observa- be defined as
tion point 共P兲 since only the direct rays create the field at
P. The lower semisphere of the surface, which is not
taken into account in Fig. 2, does not have any contribu- ជ
R
tion on the wave at P. For this reason it is enough to con- sជ r = , 共8兲
ជ储
储R
sider the upper semisphere on which the incident field di-
rectly hits according to Huygens’s principle. The area of
the surface that will be considered is determined accord-
ing to the structure of the incident wave in free space. If a
smaller surface is chosen for integration, which have not
contain all the incident rays, fictitious edge diffraction ef-
fects, which have no physical reality, occur at the obser-
vation point. For this reason two criteria can be defined

Fig. 1. (Color online) Edge diffraction process. Fig. 3. (Color online) Geometry of the scattering.
1852 J. Opt. Soc. Am. A / Vol. 25, No. 8 / August 2008 Yusuf Z. Umul

ជ
R
tinuities on the surface. These discontinuous geometric
i
sជ i = , 共9兲 places may be edges, corners or tips. It is shown by the
ជ 储
储R works of Umul that the MTPO yields the correct dif-
i
fracted waves for such discontinuities [8,9,18]. As men-
in the directions of scattering and incidence, respectively. tioned in Section 1, line integrals that give the exact dif-
The partial derivatives can be rewritten as fracted waves for scattering problems by conducting
⳵ G共P,Q兲 ⳵ G共P,Q兲 ⳵ G共P,Q兲 ␤−␣ bodies were recently derived [10,11]. In this paper we will
= cos ␯ = sin , 共10兲 study the line integral reduction of the modified diffrac-
⳵ n1 ⳵R ⳵R 2 tion integrals of Kirchhoff and compare the results with
the potential function of the boundary diffraction wave
⳵ u共Q兲 ⳵ u共Q兲 ⳵ u共Q兲 ␤−␣ (BDW) theory.
= cos共␯ + ␤ − ␣兲 = − sin ,
⳵ n1 ⳵ Ri ⳵ Ri 2
共11兲
3. VECTORIAL DERIVATION OF THE
when Fig. 3 is taken into account. The modified integral of POTENTIAL FUNCTION OF THE BOUNDARY
Kirchhoff can be written as DIFFRACTION WAVE THEORY
u共P兲 =
4␲
1
冕冕 冋
S
u共Q兲
⳵ G共P,Q兲
⳵R
The theory of BDW has its foundations on the works of
Young. He proposed a qualitative interpretation of the dif-
fraction phenomenon [12]. Afterward the mathematical

册
studies of Rubinowicz [13] put forward a new quantitative
⳵ u共Q兲 ␤−␣
+ G共P,Q兲 sin dS, 共12兲 theory on edge diffraction. The advanced form of the
⳵ Ri 2 theory of BDW was developed by Miyamoto and Wolf
[19,20] and Rubinowicz [21]. They developed the same po-
when Eqs. (10) and (11) are used in Eq. (5). Equation (12)
tential function by separately following different ways.
can be expressed in terms of unit vectors as

冕冕 冋 册
Dubra and Ferrari proposed an alternative method for
1 ⳵ G共P,Q兲 ⳵ u共Q兲 the line integral reduction of the surface integrals [22].
u共P兲 = u共Q兲 sជ r − G共P,Q兲 sជ i · n
ជ 1dS, They took into account the Rayleigh–Sommerfeld inte-
4␲ ⳵R ⳵ Ri
S grals and directly evaluated one of the parts of the surface
integral without using any assumption for a plane wave
共13兲
incidence. The obtained line integral represents a contour
when the relations of integral along the diffracting edge.
In this section, we will derive the potential function of
cos共n ជ 兲 = sជ · nជ ,
ជ 1,R 共14兲 the classical BDW theory by using vector algebra. To our
r 1
knowledge, such an approach does not exist in literature.
ជ 兲 = sជ · nជ ,
ជ 1,R
cos共n 共15兲 We will take into account the functions that are the solu-
i i 1
tion of the homogeneous Helmholtz equation, which was
are taken into account. Equation (13) is a more general given by Eq. (1). A vector of
form than Eq. (12). The studies on the MTPO have shown
that the forms that are given in Eqs. (12) and (13) lead to 1
exact GO and edge diffracted fields [8,9,18]. The scatter- ជ=
V 共u ⵜ G − G ⵜ u兲 共18兲
ing integrals can be applied for reflection process by soft 4␲
and hard surfaces. The total field is equal to zero on a soft
surface (Dirichlet condition). The scattered field can be can be defined according to the integrand of the
evaluated by the integral of ជ gives
Kirchhoff–Huygen integral [2,21]. The curl of V

us共P兲 =
4␲
1
冕冕
S
G共P,Q兲
⳵ u共Q兲
⳵ Ri
共sជ i · n
ជ 1兲dS, 共16兲
ជ=
ⵜ⫻V
1
2␲
ⵜ u ⫻ ⵜG. 共19兲

in this case. The first derivative of the total field is equal

to zero on a hard surface (Neumann condition). This con- The equation of
dition gives the equation of

冕冕
ជ = 共ⵜG · ⵜ兲 ⵜ u − 共ⵜu · ⵜ兲 ⵜ G + 4␲k2V
2␲ ⵜ ⫻ ⵜ ⫻ V ជ,
1 ⳵ G共P,Q兲
uh共P兲 = u共Q兲 共sជ r · n
ជ 1兲dS. 共17兲
4␲ ⳵R 共20兲
S

In fact the terms of soft and hard were first defined for can be obtained by taking the curl of ⵜ ⫻ Vជ and using the
acoustics [5]. equations of ⵜ u = −k u and ⵜ G = −k G. At this point we
2 2 2 2

When a whole and continuous surface is considered will consider the geometry given in Fig. 4. P and Q repre-
both the PO and MTPO lead to the exact solutions for the sent the observation and diffraction points. S is a surface
reflected or transmitted fields. But PO and the actual dif- that divides the space into two parts. Two unit vectors can
fraction theory of Kirchhoff fail when there exist discon- be defined as
Yusuf Z. Umul Vol. 25, No. 8 / August 2008 / J. Opt. Soc. Am. A 1853

ជ =ⵜ⫻W
V ជ. 共31兲

Equation (30) can be rewritten as

ជ 兲 − 2k2关1 + cos共R
ⵜ ⫻ 兵共ⵜ ⫻ V ជ ,R
ជ 兲兴W
ជ 其 = 0, 共32兲
i

by taking into account Eq. (31). Equation (32) leads to

ជ 兲 − 2k2关1 + cos共R
共ⵜ ⫻ V ជ ,R
ជ 兲兴W
ជ = ⵜ␸ . 共33兲
i

The vector potential of the boundary diffraction wave

Fig. 4. (Color online) Geometry for the vector potential. theory is found to be
ⵜu ⫻ ⵜG ⵜ␸
ជ ជ ជ =
W − , 共34兲
R i R ជ ,R
ជ 兲兴 ជ ,R
ជ 兲兴
eជ 1 = , eជ 2 = . 共21兲 4␲k2关1 + cos共R i 2k2关1 + cos共R i
ជ 兩
兩R ជ兩
兩R
i
where ␸ is a scalar function. Equation (34) can be rewrit-
The gradients of u and G can be expressed as ten as

⳵u uG共eជ 1 ⫻ eជ 2兲 ⵜ␸
ⵜu = eជ 1 , 共22兲 ជ =−
W − , 共35兲
⳵ Ri ជ ,R
4␲关1 + cos共R ជ 兲兴 ជ ,R
2k 关1 + cos共R
2 ជ 兲兴
i i

⳵G by taking into account Eqs. (22), (23), (26), and (27).

ⵜG = eជ 2 . 共23兲 Equation (35) reduces to the potential function of the
⳵R Rubinowicz theory for ␸ equal to zero as
The direction of eជ 2 is chosen as in [19]. Equation (20) can ជ ,R
ជ 兲eជ
uG sin共R
be rewritten as ជ =−
W
i d
, 共36兲
ជ ,R
4␲关1 + cos共R ជ 兲兴
⳵ G ⳵ 2u ⳵ R i ⳵ u ⳵ 2G ⳵ R i
ជ=
2␲ ⵜ ⫻ ⵜ ⫻ V eជ 1 − ជ,
eជ 2 + 4␲k2V
⳵ R ⳵ Ri2 ⳵ R ⳵ Ri ⳵ R2 ⳵ Ri where eជ d is a unit vector that is perpendicular to the
plane that contains the incident and scattered rays. The
共24兲 ជ ,R
ជ 兲eជ as eជ ⫻ eជ .
minus sign is removed by defining sin共R i d 2 1
which reduces to

ជ = − k2 cos ␣ u
2␲ ⵜ ⫻ ⵜ ⫻ V 冉 ⳵G
⳵R
eជ 1 − G
⳵u
⳵ Ri
冊 ជ,
eជ 2 + 4␲k2V
4. LINE INTEGRAL REDUCTION OF THE
MODIFIED KIRCHHOFF INTEGRAL
The field, at the observation point, can be given by
共25兲
for ⳵Ri / ⳵R and ⳵R / ⳵Ri are equal to cos ␣. We will consider
the approximations of
u共P兲 = 冕冕
S
ជ · nជ dS,
V 1 共37兲

ⵜu ⬇ − jkueជ 1 , 共26兲
as the scattering integral of the modified diffraction
theory of Kirchhoff. Equation (31) is also valid for this
ⵜG ⬇ − jkGeជ 2 , 共27兲 ជ does not change. Equation (37)
case since the value of V
for k
1. As a result Eq. (25) gives can be rewritten as

ជ = jkuGk2 cos ␣共eជ − eជ 兲 + 4␲k2V

2␲ ⵜ ⫻ ⵜ ⫻ V 1 2
ជ,

ជ can be writ-
共28兲 u共P兲 = 冕冕
S
ជ 兲 · nជ dS,
共ⵜ ⫻ W 1 共38兲
under the conditions of Eqs. (26) and (27). V
ten as ជ is given by Eq. (36). We
in terms of the vector potential. W
jk propose a modified form of the Stokes theorem as
ជ ⬇−
V uG共eជ 2 − eជ 1兲. 共29兲

Equation (12) reads

4␲
冕冕
S
ជ 兲 · nជ dS =
共ⵜ ⫻ W 1 冖C
ជ · ជt
W
dl
ជ 1 · nជ
n
, 共39兲

ជ = 2k2V
ⵜ⫻ⵜ⫻V ជ 关1 + cos共R
ជ ,R
ជ 兲兴, 共30兲
i ជ is the actual unit normal vector of the surface. C is a
for n
ជ ,R
ជ 兲 denotes the cosine of the angle between closed loop that surrounds the surface of S. ជt is the unit
where cos共R i tangent vector of C. To prove this statement we will take
ជ ជ
R and Ri, which is also equal to cos ␣. It can be shown into account the geometry given in Fig. 5. S1 is the sur-
ជ is equal to zero [19]. This property shows that V
that ⵜ · V ជ face, which accepts n ជ 1 as the normal vector. nជ is the unit
can be derived from the curl of another vector as normal vector of the real scattering surface. ␸ is equal to
1854 J. Opt. Soc. Am. A / Vol. 25, No. 8 / August 2008 Yusuf Z. Umul

Fig. 5. (Color online) Surfaces of scattering.

Fig. 6. (Color online) Reflection geometry by a plane.
␤−␣
␸= . 共40兲 sជ r = cos ␤eជ x + sin ␤eជ y , 共46兲
2
according to Fig. 6. The incident field is supposed to be a
The vector direction of the potential function is evident by
plane wave, which has the representation of
the vector product of eជ 2 ⫻ eជ 1, which is out of the paper’s
plane for Fig. 5. For this reason, the direction of ជt is also ui = u0e−jk共x cos ␣+y sin ␣兲 . 共47兲
out of the figure’s plane and is equal to ជt1. eជ 1, eជ 2, and n
ជ1
The modified unit normal vector has the expression of
are in the same plane. ␸ also represents the angle be-
tween nជ and nជ 1. Equation (38) can be written as
ជ 1 = − sin
n 冉 冊 冉 冊
␤−␣
eជ x + cos
␤−␣
eជ y . 共48兲

冕冕 冕冕 dS1 2 2
ជ 兲 · nជ dS =
共ⵜ ⫻ W ជ 兲 · nជ
共ⵜ ⫻ W , 共41兲
1 1
ជ 1 · nជ
n The vector potential is found to be
S S1
u0 e−jkR sin共␤ − ␣兲
according to a well-known identity of surface integration ជ =
W e−jxy cos ␣ eជ z 共49兲
4␲ R 1 − cos共␤ − ␣兲
[23]. The surface integral, at the right-hand side of Eq.
(41), reads for a scattering point on the surface of the half-plane. The
unit tangent vector of the edge contour 共tជ兲 is equal to eជ z.
冕冕
S1
ជ 兲 · nជ
共ⵜ ⫻ W 1
dS1
ជ 1 · nជ
n
= 冖
C1
ជ · ជt
W 1
dl1
ជ 1 · nជ
n
, 共42兲 The actual unit normal vector of the surface 共n
scalar product of nជ 1 · nជ gives
ជ 兲 is eជ y. The

␤−␣
according to the Stoke’s theorem. C1 is a closed surface ជ 1 · nជ = cos
n , 共50兲
2
that surrounds the surface of S1. Since ជt1 is equal to ជt, C
can also be used instead of C1. It is important to note that when Eq. (48) is taken into account. As a result the line
we consider the contour of C on the scattering point of Q. integral of diffraction fields can be written as

冕
Its direction is perpendicular to the paper’s surface. This ⬁
assumption will be examined as an example. As a result, u0 e−jkRe sin共␤e − ␣兲 1
ud = dz⬘ , 共51兲
the edge diffracted fields are found to be 4␲ −⬁
Re 1 − cos共␤e − ␣兲 ␤e − ␣
cos

冖 dl 2
ud共P兲 = ជ · ជt
W , 共43兲
ជ 1 · nជ
n according to Eq. (44). A minus sign is used in the denomi-
C nator of the vector potential since the direction of the re-
which also leads to the equation of flected wave is reversed. Re is equal to 冑␳2 + 共z − z⬘兲2. ␤e is
the value of ␤ on the contour of the edge and has the value

冖
ជ ,R
ជ 兲 eជ · ជt of ␾. Equation (51) can be rewritten as
1 u共Q兲G共P,Q兲sin共R

冕
i d
ud共P兲 = dl, 共44兲 1 u0 ⬁
e−jkRe
4␲ ជ ,R
ជ 兲 ជ 1 · nជ
n
1 + cos共R i ud = dz⬘ , 共52兲
C
4␲ ␾−␣ −⬁
Re
sin
when Eq. (36) is taken into account. As an example of the 2
usage of the line integral reduction, we will take into ac-
count the reflection geometry given in Fig. 6. The re- the stationary phase evaluation of which gives
flected diffracted field will be evaluated. A half-plane ex- e−j共␲/4兲 1 e−jk␳
ists in the 共x , z兲 plane. The edge contour lies on the z axis ud = u0 , 共53兲
at x = 0. If the edge was on two axes, then sជ r would be a 2冑2␲ ␾ − ␣ 冑k ␳
sin
three-dimensional vector. The expressions of sជ j and sជ r can 2
be given by
which is exactly the diffracted field expression of the geo-
sជ j = cos ␣eជ x + sin ␣eជ y , 共45兲 metrical theory of diffraction (GTD) [17]. It is apparent
Yusuf Z. Umul Vol. 25, No. 8 / August 2008 / J. Opt. Soc. Am. A 1855

that the reflection boundary occurs at ␾ = ␣ as the dif- which reflects from Qs. Since the edge diffracted fields are
fracted field approaches to infinity at this region. The rea- investigated in Section 4, we will focus on the second line
son for this inconsistency is the nature of the GTD field. integral in this section. The GO wave can be evaluated by
The GTD fields represent the first terms in the high fre- the line integral of

冖
quency asymptotic expansions of the exact solutions. For
dl
this reason they are not valid for near fields where the uGO = ជ · ជt
W . 共57兲
multiplication of k␳  1 or the overall argument is smaller C2
ជ 1 · nជ
n
than one.
The singularity of the potential function occurs from the
term of 1 − sជ j · sជ r. By considering this point, the integrand
5. GEOMETRICAL OPTICS WAVE of Eq. (57) can be written as
The line integral reduction of the Kirchhoff integral also
ជ · ជt
W f共P,Q兲
gives the GO fields since the surface integral contains
both the diffracted and GO waves. The GO fields are = , 共58兲
ជ 1 · nជ
n g共P,Q兲
transmitted or reflected waves that do not affected by the
presence of the edge discontinuity. A more correct repre- for g共P , Q兲 is equal to zero for Q = Qs. The GO wave can be
sentation of the modified Stoke’s theorem must be rewritten as

冕冕 冖 冕
dl 2␲
f共P,Q兲
ជ 兲 · nជ dS =
共ⵜ ⫻ W ជ · ជt
W + uGO , 共54兲 uGO = lim ␧d␸ , 共59兲
1
C
ជ 1 · nជ
n ␧→0 0
g共P,Q兲
S

because the line integration over the edge contour of C which also yields to the equation of

冕
gives only the edge diffracted fields. uGO represents the 2␲
f共P,Q兲
GO wave. In this section we will deal with the evaluation uGO = lim ␧d␸ . 共60兲
of the GO fields by using Eq. (54). Investigation of GO 0 ␧→0 g共P,Q兲

waves in the context of the BDW theory was put forward

The geometry in Fig. 8 is taken into account. It is impor-
in detail by Miyamoto and Wolf [19].
tant to note that the reflection angle is equal to the angle
A surface of S, which is bounded by the contour of C1, is
of incidence at Qs. Equation (60) leads to the expression of
taken into account. The geometry is given in Fig. 7. Qs is

冕
2␲
the point of reflection. The vector potential can be defined ␧
as uGO = f共P,Qs兲 lim d␸ , 共61兲
0 ␧→0 g共P,Q兲
uG共sជ j ⫻ sជ r兲
ជ =
W , 共55兲 which can be evaluated by using the rule of L’Hospital
4␲关1 − sជ j · sជ r兴 since g共P , Qs兲 is equal to zero. As a result one obtains

冕
for the reflected GO fields. sជ j is in the image direction of sជ i f共P,Qs兲 2␲
according to the plane of the reflection surface. The vector uGO = d␸ , 共62兲
potential is analytic (has no singularity) everywhere on S g⬘共P,Qs兲 0
except the point of reflection 共Qs兲, on which the scalar
product of sជ j · sជ r is equal to one. If we isolate Qs from S by where the derivative of g共P , Q兲 is with respect to ␧. The
a contour of C2, then the modified Stoke’s theorem reads GO field reads

冕冕 冖 冖
dl dl f共P,Qs兲
ជ 兲 · nជ dS =
共ⵜ ⫻ W ជ · ជt
W − ជ · ជt
W . uGO = 2␲ . 共63兲
1
ជ 1 · nជ
n ជ 1 · nជ
n g⬘共P,Qs兲
C1 C2
S
Now we will consider the problem of scattering that is
共56兲 given in Fig. 6 and evaluate the reflected GO wave. The
The line integral, along C1, gives the edge diffracted stationary point of QS can be isolated with a circle, which
waves. The integration along C2 leads to the GO wave,

Fig. 8. (Color online) Geometry for the evaluation of the line in-
Fig. 7. (Color online) Scattering surface and its boundary. tegral of GO waves.
1856 J. Opt. Soc. Am. A / Vol. 25, No. 8 / August 2008 Yusuf Z. Umul

␤−␣ ␧
sin ⬇− , 共68兲
2 2R

for small values of ␧ [19]. Equation (67) yields

u0 e−jk共x cos ␣+y sin ␣兲

冕
2␲
uGO = − lim2Rd␸ , 共69兲
4␲ Rs 0 ␧→0

which gives the result of

Fig. 9. (Color online) Geometry for the GO field evaluation.
uGO = − u0e−jk共x cos ␣+y sin ␣兲 . 共70兲
has a radius of ␧, as in Figs. 7 and 8. Afterward Eq. (63)
can be applied for the evaluation of the GO field. The po- It is apparent that Eq. (70) expresses the exact reflected
tential function can be given by GO field.

u0 e−jkR sin共␤ − ␣兲
ជ =
W e−jk共xs−␧兲cos ␣ eជ z , 共64兲
4␲ R 1 − cos共␤ − ␣兲 6. APPLICATION: DIFFRACTION OF PLANE
WAVES BY A SEMI-INFINITE EDGE
according to Fig. 9. xs is the value of x at Qs and is the CONTOUR
stationary phase point. Rs is equal to 共x − xs兲cos ␣
+ y sin ␣. Equation (64) can be expanded as In this section we will investigate the diffracted fields by
a semi-infinite edge. The geometry of the problem is given
ជ
W u0 e−jkR 1 in Fig. 10. The plane is illuminated by an incident plane
= e−jk共xs−␧兲cos ␣ eជ z , 共65兲 wave of
ជ 1 · nជ
n 4␲ R ␤−␣
sin
2 ui = u0ejk共x cos ␣+y sin ␣兲 . 共71兲

ជ 1 · nជ . The line integral for the

by considering the term of n We will only deal with the reflected diffracted waves in or-
GO waves can be constructed as der to lessen the number of equations since the construc-

冕
2␲
tion of the incident diffracted waves are similar to the re-
u0 e−jkR ␧ flected diffracted fields. The unit vectors of the directions
uGO = lim e−jk共xs−␧兲cos ␣ d␸ , 共66兲
4␲ ␧→0 0
R ␤−␣ of the rays can be given by
sin
2
sជ i = − eជ x cos ␣ − eជ y sin ␣ , 共72兲
which can be further written as
2␲ sជ j = − eជ x cos ␣ + eជ y sin ␣ , 共73兲

uGO =
u0
4␲
e−jkxs cos ␣
e−jkRs
Rs
冕 lim
␧→0
␧
␤−␣
d␸ , 共67兲
sជ r = − eជ x cos ␤ + eជ y sin ␤ sin ␩ − eជ z sin ␤ cos ␩ , 共74兲
0 sin
2
according to Fig. 10. sជ j represents the direction of the in-
by taking the operation of limit in the integral. The sine cident wave’s image. The vector and scalar products of sជ j
term can be approximated as and sជ r are found to be

Fig. 10. (Color online) Geometry of the semi-infinite edge contour.

Yusuf Z. Umul Vol. 25, No. 8 / August 2008 / J. Opt. Soc. Am. A 1857

sជ j ⫻ sជ r = − eជ x sin ␣ sin ␤ cos ␩ − eជ y cos ␣ sin ␤ cos ␩ ជ 1 ⫻ sជ i = nជ 1 ⫻ sជ r .

n 共78兲

+ eជ z共sin ␣ cos ␤ − cos ␣ sin ␤ sin ␩兲, 共75兲 Equation (77) shows the cosine of the angle between n ជ1
and the actual unit normal vector of the surface. The unit
tangent vectors of C are equal to eជ z. The diffracted field
sជ j · sជ r = cos ␣ cos ␤ + sin ␣ sin ␤ sin ␩ , 共76兲 can be found to be

冕冏 冏
⬁
Wz
ជ 1 · nជ can be evaluated
respectively. The scalar product of n ud = dz⬘ . 共79兲
ជ 1 · nជ
n
as 0 x=0

ជ can be determined to be
W
sin ␣ + sin ␤ sin ␩
ជ 1 · nជ =
n , 共77兲 u0 e−jkR sជ j ⫻ sជ r
冑2共1 − sជ i · sជ r兲 ជ =−
W ejkx⬘ cos ␣ , 共80兲
4␲ R 1 − sជ j · sជ r

from the formula of according to Eq. (36). The function of

共cos ␣ sin ␤ sin ␩ − sin ␣ cos ␤兲冑1 − cos ␣ cos ␤ + sin ␣ sin ␤ sin ␩
f1 = , 共81兲
共1 − cos ␣ cos ␤ − sin ␣ sin ␤ sin ␩兲共sin ␣ + sin ␤ sin ␩兲

ej共␲/4兲 1 e−jk␳
will be defined for the line integral. The diffracted field ud = u0 , 共83兲
can be written as 2冑2␲ ␾ + ␣ 冑k ␳
cos
2

冕
⬁ for the edge diffracted wave. Equation (83) is the exact
u0 e−jkR1
ud = f1 dz⬘ , 共82兲 diffracted field of the GTD [17], which is the dominant
2冑2␲ 0
R1 term in the high frequency asymptotic expansion of Som-
merfeld’s rigorous solution [24]. The reflection boundary
occurs at ␾ = ␲ − ␣ according to Eq. (83). The corner dif-
where R1 is equal to 冑␳2 + 共z − z⬘兲2. When the line integral
fracted rays can be evaluated by using the edge point
technique, which can be defined as
of diffraction is examined, it can be seen that the station-

冕
ary phase evaluation of the integral gives the edge dif- ⬁
1 f共a兲
fracted fields by the edge contours of C. The edge point f共x兲e−jkg共x兲dx ⬇ e−jkg共a兲 , 共84兲
jk g⬘共a兲
contribution, at z⬘ = 0, leads to the corner diffracted ray at a

the corner of the edge. The corner diffracted fields com- for a typical diffraction integral. The first derivative of the
pensate the discontinuity of the edge diffracted waves at phase function is equal to −cos ␪ at the corner point of the
the transition regions. quarter-plane. R1 is r since z⬘ = 0. The corner diffracted
Equation (82) will be taken into account. The phase wave is found to be
function of the integral is equal to R1. The stationary
phase point can be found by equating the first derivative u0 f1兩z⬘=0 e−jkr
uc = − , 共85兲
of the phase function to zero. The operation gives zs = z. jk2冑2␲ cos ␪ r
This value of z⬘ requires that ␩s = ␲ / 2 and ␤s = ␲ − ␾. The
second derivative of the phase function is equal to 1 / ␳ at when Eq. (84) is used in Eq. (82). The term of f1兩z⬘=0 can be
the stationary phase point. As a result one obtains defined as

f1 冏 z⬘=0 =
共cos ␣ sin ␤c sin ␩c − sin ␣ cos ␤c兲冑1 − cos ␣ cos ␤c + sin ␣ sin ␤c sin ␩c
共1 − cos ␣ cos ␤c − sin ␣ sin ␤c sin ␩c兲共sin ␣ + sin ␤c sin ␩c兲
, 共86兲
1858 J. Opt. Soc. Am. A / Vol. 25, No. 8 / August 2008 Yusuf Z. Umul

of cos ␪ at the denominator of the corner diffracted

wave [25–27].

7. UNIFORM DIFFRACTED FIELDS AND

NUMERICAL ANALYSIS
Since the field expressions have singularities in certain
points, their uniform expressions must be obtained in or-
der to analyze them numerically. In this section, we will
use the uniform theory of diffraction (UTD) [28,29] in or-
der to represent the uniform versions of the fields. The
transition function of the UTD is composed of Fresnel
functions. The uniform version of the first edge diffracted
wave is well-known in the literature and can be expressed
as

ud = − u0ejk共x cos ␣−y sin ␣兲sgn共␰d兲F关兩␰d兩兴, 共91兲

where ␰d is equal to
Fig. 11. (Color online) Geometry of the corner diffracted ray.
␾+␣
␰d = − 冑2k␳ cos . 共92兲
which yields to the equation of 2

f1 冏 z⬘=0 =
sin ␪ sin共␾ + ␣兲冑1 + sin ␪ cos共␾ − ␣兲
关1 + sin ␪ cos共␾ + ␣兲兴共sin ␣ + sin ␪ sin ␾兲
,
sgn共x兲 is the signum function, which is equal to 1 for x
⬎ 0 and −1, otherwise. F关x兴 is the Fresnel function and
can be defined by the integral of
共87兲
ej共␲/4兲
冕
⬁
2
when the relations of F关x兴 = e−jt dt. 共93兲
冑␲ x
cos ␤c = − sin ␪ cos ␾ , 共88兲
The plot of the reflected diffracted field is shown in Fig.
sin ␤c sin ␩c = sin ␪ sin ␾ , 共89兲 12 versus the observation angle. The distance between
the observation point and the origin is equal to ␳ = 6␭
sin ␤c cos ␩c = − cos ␪ , 共90兲 where ␭ is the wavelength. The angle of incidence is 60°.
It can be seen that the reflection boundary occurs at ␾
are taken into account according to Fig. 11. The transition = 120°, which is exactly equal to ␲ − ␣. The amplitude of
region of the edge diffracted wave occurs at ␪ = ␲ / 2. Since the diffracted wave is 0.5, which is half of the reflected
the edge contour of C does not continue for z⬘ ⬍ 0, at z⬘ wave’s amplitude.
= 0 the edge diffracted wave suddenly vanishes by show- The uniform expression of the diffracted wave at the
ing a discontinuity, which is determined by the function corner of C1 can be written as

uc = − u02冑2ejk␳ cos共␾+␣兲
冏
f1 z⬘=0冑sin ␪ cos
␾+␣
2
cos
␪ + 共␲/2兲
2
sgn共␰d兲F关兩␰d兩兴sgn共␰c兲F关兩␰c兩兴 共94兲
cos ␪

where ␰c is that the edge diffracted wave suddenly goes to zero and
thus has a discontinuity at ␪ = ␲ / 2. The corner diffracted
␪ + 共␲/2兲
␰c = − 冑2kr cos . 共95兲 ray, which is a spherical wave, compensates this disconti-
2 nuity of the edge diffracted field. The total field is continu-
ous everywhere.
Figure 13 depicts the variation of the total, edge dif-
fracted, and corner diffracted waves with respect to the
observation angle of ␪. ␾ is taken as 90°. r is equal to 6␭
and ␳ is r sin ␪. The total diffracted field can be given by
the equation of
8. CONCLUSION
ut = ud + uc . 共96兲
In this paper, we investigated a modified version of the
The edge diffracted field varies with respect to the obser- diffraction theory of Kirchhoff, which will lead to the ex-
vation angle since ␳ is a function of ␪. It can be observed act scattered fields for conducting bodies, by using the
Yusuf Z. Umul Vol. 25, No. 8 / August 2008 / J. Opt. Soc. Am. A 1859

phenomena of edge diffraction. The achievement of the

MTPO comes from the modified normal vector, which is
defined as in Fig. 1. When the incident ray is perpendicu-
lar to the tangent of the edge contour the consideration of
the diffraction integral, as in Eq. (51), leads to the correct
result. But since the unit normal vector varies in three di-
mensions in the process of the corner diffraction, the us-
age of the normal vector, which varies in a plane, yields
incorrect field expressions. For this reason Eq. (82) and
the geometry of Fig. 10 are more suitable for the evalua-
tion of the edge diffracted waves.

REFERENCES
1. B. B. Baker and E. T. Copson, The Mathematical Theory of
Huygens’ Principle (Oxford U. Press, 1950).
2. M. Born and E. Wolf, Principles of Optics (Cambridge U.
Fig. 12. (Color online) Edge diffracted field at C. Press, 2003).
3. C. R. Schultheisz, “Numerical solution of the
Huygens–Fresnel–Kirchhoff diffraction of spherical waves
by a circular aperture,” J. Opt. Soc. Am. A 11, 774–778
(1994).
4. S. Ganci, “Equivalence between two consistent
formulations of Kirchhoff ’s diffraction theory,” J. Opt. Soc.
Am. A 5, 1626–1628 (1988).
5. P. Ya. Ufimtsev, Fundamentals of the Physical Theory of
Diffraction (Wiley-IEEE, 2007).
6. W. L. Stutzman and G. A. Thiele, Antenna Theory and
Design (Wiley, 1998).
7. P. Ya. Ufimtsev, “Elementary edge waves and the physical
theory of diffraction,” Electromagnetics 11, 125–160 (1991).
8. Y. Z. Umul, “Modified theory of physical optics,” Opt.
Express 12, 4959–4972 (2004).
9. Y. Z. Umul, “Modified theory of physical optics approach to
wedge diffraction problems,” Opt. Express 13, 216–224
(2005).
10. Y. Z. Umul, “MTPO based potential function of the
boundary diffraction wave theory,” Opt. Laser Technol. 40,
769–774 (2008).
11. Y. Z. Umul, “The theory of the boundary diffraction wave
Fig. 13. (Color online) Compensation of the edge diffracted for wedge diffraction,” J. Mod. Opt. 55, 1417–1426 (2008).
wave at the corner. 12. A. Rubinowicz, “Thomas Young and the theory of
diffraction,” Nature 180, 160–162 (1957).
13. A. Rubinowicz, “Die beugungswelle in der Kirchoffschen
axioms of the MTPO. The new Kirchhoff integral is also theorie der beugungsercheinungen,” Ann. Phys. 4, 257–278
defined for soft and hard surfaces for scalar waves. A po- (1917).
14. S. Ganci, “A general solution for the half plane problem,” J.
tential function, the curl of which gives the integrand of
Mod. Opt. 42, 1707–1711 (1995).
Kirchhoff ’s integral, is obtained with vector operations. 15. S. Silver, Microwave Antenna Theory and Design (McGraw-
The surface integral of the modified theory is reduced to a Hill, 1949).
line integral by using a modified form of Stokes’s theorem. 16. Y. Z. Umul, “Modified theory of physical optics solution of
Also the GO waves are evaluated with the aid of the same the impedance half plane problem,” IEEE Trans. Antennas
Propag. 54, 2048–2053 (2006).
line integral. The method is applied to the problem of dif- 17. J. B. Keller, “Geometrical theory of diffraction,” J. Opt. Soc.
fraction of plane waves by a semi-infinite half-plane. The Am. 52, 116–130 (1962).
edge and corner diffracted fields are evaluated by the 18. Y. Z. Umul, “Edge-dislocation waves in the edge diffraction
asymptotic approximation of the line integral of the process by an impedance half-plane,” J. Opt. Soc. Am. A 24,
507–511 (2007).
boundary diffraction wave theory. The results are exam- 19. K. Miyamoto and E. Wolf, “Generalization of the
ined numerically and it is observed that the plots are in Maggi–Rubinowicz theory of the boundary diffraction
harmony with the theory. wave—Part I,” J. Opt. Soc. Am. 52, 615–625 (1962).
It is important to emphasize the difference between 20. K. Miyamoto and E. Wolf, “Generalization of the
Eqs. (51) and (82). The distinction between two represen- Maggi–Rubinowicz theory of the boundary diffraction
wave—Part II,” J. Opt. Soc. Am. 52, 626–637 (1962).
tations also stresses the incorrect edge point contribution 21. A. Rubinowicz, “The Miyamoto–Wolf diffraction wave,”
of the PO method. As is mentioned in Section 2, the nor- Prog. Opt. 4, 201–240 (1965).
mal vector and the angle of reflection are considered in 22. A. Dubra and J. A. Ferrari, “Diffracted field by an arbitrary
their actual values; the method of PO leads to incorrect aperture,” Am. J. Phys. 67, 87–92 (1999).
23. F. B. Hildebrand, Advanced Calculus for Applications
edge diffracted waves. It is shown in Fig. 1 that the unit (Prentice-Hall, 1962).
normal vector varies by separating the angle between the 24. A. Sommerfeld, “Matematische Theorie der Diffraction,”
incident and diffracted rays into two equal parts in the Math. Ann. 47, 317–374 (1896).
1860 J. Opt. Soc. Am. A / Vol. 25, No. 8 / August 2008 Yusuf Z. Umul

25. S. Ganci, “Half-plane diffraction in a case of oblique 28. R. G. Kouyoumjian and P. H. Pathak, “A uniform
incidence,” J. Mod. Opt. 43, 2543–2551 (1996). geometrical theory of diffraction for an edge in a perfectly
26. Y. Rahmat-Samii, “Keller’s cone encountered at a hotel,” conducting surface,” Proc. IEEE 62, 1448–1461 (1974).
IEEE Antennas Propag. Mag. 49, 88–89 (2007). 29. Y. Z. Umul, “Uniform theory for the diffraction of
27. T. B. A. Senior and P. L. E. Ushlengi, “Experimental evanescent plane waves,” J. Opt. Soc. Am. A 24, 2426–2430
detection of the edge-diffraction cone,” J. Electron Microsc. (2007).
60, 1448–1449 (1972).