Optica Applicata, Vol. XLVI, No.

2, 2016
DOI: 10.5277/oa160215

Propagation in dielectric rectangular waveguides

KIM HO YEAP1*, KAI HONG TEH1, KEE CHOON YEONG1, KOON CHUN LAI1, MEY CHERN LOH2
1
Faculty of Engineering and Green Technology, Tunku Abdul Rahman University,
Kampar, Perak, Malaysia
2
Centre of Photonics and Advanced Materials Research,
Kampar, Perak, Malaysia
*
Corresponding author: yeapkh@utar.edu.my

We present a fundamental and accurate approach to compute the attenuation of electromagnetic

waves propagating in dielectric rectangular waveguides. The transverse wave numbers are first ob-
tained as roots of a set of transcendental equations developed by matching the fields with the surface
impedance of the wall. The propagation constant is found by substituting the values of transverse
wave numbers into the dispersion relation. We have examined the validity of our model by com-
paring the computed results with those obtained from Marcatili’s equations and the finite element
method. In our results, it is shown that the fundamental mode is identical with that found in a per-
fectly conducting waveguide. Our analysis also shows that a hollow waveguide is found to have
much lower attenuation than its dielectric counterparts. Since the cutoff frequency is usually af-
fected by the constitutive properties of the dielectric medium, for a waveguide designed for wave
with the same cutoff frequency, hollow waveguides turn out to be relatively larger in size.

Keywords: dielectric rectangular waveguide, transverse wave numbers, attenuation coefficient, surface
impedance, propagation constant, fundamental mode.

1. Introduction
Dielectric waveguides have been commonly used in integrated optics, as well as,
millimeter and submillimeter circuits, as transmission lines, filters, optical couplers,
reflectometers, power divider/combiner, resonators and phase shifters [1–7]. During
wave propagation, it is important to ensure that energy loss in the waveguides is min-
imized [8]. Hence, the availability of a mathematical model which is able to predict
accurately the propagation coefficient of waves kz is important in the design of wave-
guides. Wave propagation in a circular dielectric waveguide has been accurately com-
puted based on the fundamental approach formulated by Stratton [9–13]. In Stratton’s
approach, the circular symmetry of the waveguide allows the analytical equations for
the eigenmodes to be expressed in a single variable, i.e., the radial distance r. Hence,
the propagation coefficient could be accurately computed from a single analytical
equation. However, unlike its circular counterpart, the expression for the propagation
318 KIM HO YEAP et al.

coefficient in a rectangular waveguide involves a 2D Cartesian coordinate and is, there-

fore, inherently more complicated to be formulated. Thus far, a similar rigorous and
fundamental technique such as that suggested by Stratton is not available for the case
of a dielectric rectangular waveguide. The existing formulations found in literatures
– either closed forms [14–19] or numerical solutions [19–33], have imposed certain
simplifications and assumptions in the process of derivation. The most apparent sim-
plification found in these literatures is allowing the solution for Helmholtz equation
y x
to be separated into pure E pq and E pq modes, where p and q represent respectively
the number of maxima of the electric fields in the x and y directions of the waveguide.
In a practical dielectric waveguide, however, a superposition of these two modes is
necessary to satisfy the boundary conditions [34–36]. Among these available tech-
niques, Marcatili’s approach [19–21] turns out to be the most widely implemented.
This is partly due to its ability to produce simple and straight-forward analytical solu-
tions; and partly, because the approach gives reasonably good results. By neglecting
the presence of fields at the edges of the rectangular waveguide, Marcatili has devel-
oped a pair of transcendental equations to solve for the transverse wave numbers, i.e.,
kx and ky. The propagation coefficient kz is then solved by relating kx and ky with the
wave number in the waveguide material k.
In [34], we have formulated a fundamental and accurate technique to compute
the propagation coefficient kz in an imperfectly conducting rectangular waveguide.
Here, we develop further this approach to the case of a dielectric rectangular wave-
guide. Like the method proposed by Marcatili, kz in our method is computed by relating
k with kx and ky, which are first numerically solved. However, unlike Marcatili’s ap-
proach, the fields at the edges of the waveguide are taken into consideration during
y
formulation. Since our method accounts for the concurrent presence of both E pq and
x
E pq modes, the propagation coefficient for either mode can be conveniently computed
using the same set of characteristic equations. We shall demonstrate that our method
gives more realistic results since it models closely the actual propagation of waves in
a practical waveguide.

2. Formulation
2.1. Fields in Cartesian coordinates
For electromagnetic waves propagating in the z direction of a rectangular waveguide, as
shown in Fig. 1, Helmholtz equations are expressed in Cartesian coordinates as [35–37]

2 2
∂ ψz ∂ ψz
---------------- - + ( k 2 – k z2 ) ψ z = 0
- + ---------------- (1)
2 2
∂x ∂y

where ψz is the z component of a two dimensional vector phasor ψ that depends on

the cross-sectional coordinates. To derive the field components in the waveguide, ψz can
be substituted with the longitudinal electric Ez and magnetic Hz fields.
Propagation in dielectric rectangular waveguides 319

y
z

x
0 a

Fig. 1. Dielectric rectangular waveguide.

The transverse field components are obtained by substituting Ez and Hz into

Maxwell’s source free curl equations. Rearranging the transverse fields by expressing
them in terms of the longitudinal fields, we obtain the following:

–j dH z dE z 
-  k z --------------
H x = --------------------- - – ω ε -------------
- (2)
2 2  dx dy 
kx + ky

–j dH z dE z 
-  k --------------
H y = --------------------- - – ω ε -------------
- (3)
2 2  z dy dx 
kx + ky

–j dE z dH z 
-  k z -------------
E x = --------------------- - – ω μ -------------- (4)
2 2  dx dy 
kx + ky

–j dE z dH z 
-  k -------------
E y = --------------------- - – ω μ -------------- (5)
2 2  z dy dx 
kx + ky

where ε and μ are the permittivity and permeability of the inner core material, respec-
tively. For a non-magnetic material, μ is identical with the permeability of free space μ 0.
Generally, the permittivity ε of a lossy material is complex and is given as [34, 38]
σ
ε = ε 0 – j -------- (6)
ω
where ε 0 and σ are respectively the permittivity and conductivity of the material, and
ω is the angular frequency. However, since the conductivity of the dielectric material
is almost negligible, the imaginary part in (6) can be neglected and ε in the inner core
is usually taken as a real value. The permittivity at the wall, on the other hand, could
be either complex (for conductor) or real (for dielectric), depending on the cladding
material at the wall.

2.2. Fields in a dielectric rectangular waveguide

In general, the dielectric constant of a dielectric waveguide is higher than its surround-
ing medium, which, in most cases, is the air. This allows the fields to be confined most-
320 KIM HO YEAP et al.

ly within the waveguide and decays in evanescence beyond the boundary of the
waveguide. Since the fields are concentrated at the core of the waveguide, the resultant
tangential electric field Et and the normal derivative of the tangential magnetic field
∂Ht /∂an are at their minimal (but not necessarily zero) at the boundary of the wave-
guide. Using the method of separation of variables to solve (1), the longitudinal fields
can be expressed as:

E z = E 0 sin ( k x x + φ x ) sin ( k y y + φ y ) (7)

H z = H 0 cos ( k x x + φ x ) cos ( k y y + φ y ) (8)

where E0 and H0 are the constant amplitudes of the fields. The phase parameters φ x
and φ y , are referred to as the field’s penetration factors in the x and y directions, re-
spectively [34]. The penetration factors account for the remaining fields at the bound-
ary, which decays exponentially beyond the boundary. Since both Et and ∂Ht /∂an are
either at their maximum or zero at the centre of the waveguide, i.e.,

kx a ky b
sin ------------ + φ x = sin ------------ + φ y = ± 1 or 0 (9)
 2   2 

where a and b are the width and height of the waveguide, respectively, then, the pene-
tration factors can be found as:

pπ – k ax
φ x = -------------------------- (10)
2

qπ – k y b
φ y = --------------------------- (11)
2
y x
In order to account for the coexistence of E pq and E pq modes, both longitudinal
fields must be present. Hence, substituting the longitudinal fields (7) and (8) into (2)
to (5), the transverse fields are obtained as:

j ( k z k x H 0 + ω ε d k y E 0 ) sin ( k x x + φ x ) sin ( k y y + φ y )
H x = -----------------------------------------------------------------------------------------------------------------------------------
2 2
- (12)
kx + ky

j ( k z k y H 0 + ω ε d k y E 0 ) cos ( k x x + φ x ) sin ( k y y + φ y )
H y = ------------------------------------------------------------------------------------------------------------------------------------
2 2
- (13)
kx + ky

j ( k z k x E 0 + ω μ d k y H 0 ) cos ( k x x + φ x ) sin ( k y y + φ y )
E x = -------------------------------------------------------------------------------------------------------------------------------------
2 2
- (14)
kx + ky
Propagation in dielectric rectangular waveguides 321

j ( k z k y E 0 + ω μ d k x H 0 ) sin ( k x x + φ x ) cos ( k y y + φ y )
E y = ------------------------------------------------------------------------------------------------------------------------------------
2 2
- (15)
kx + ky

where ε d and μ d are the permittivity and permeability of the dielectric material, re-
spectively.

2.3. Constitutive relations

At the boundary of the dielectric waveguide, the ratio of the tangential electric field Et
to tangential magnetic field Ht is related to the surface impedance Zs as [34–36]
Et
----------------------
- = Zs (16)
an × Ht

where an is a normal unit vector; Zs can be expressed in terms of the electrical properties
of the two mediums [39]
1
Z s = ---------------------------------------- (17)
jω ( ε rd – ε r0 )b

where ω is the angular frequency, whereas ε rd and ε r0 are the relative permittivities of
the waveguide and the surrounding material, respectively. For simplicity, we are con-
sidering a single layer dielectric waveguide surrounded by air.
At the height surface of the waveguide where y = b, Ez /Hx = –Ex /Hz = Zs. Substi-
tuting (7), (8), (12), (14) and (17) into (16), the following relationships are obtained:

–Ex j E0 1
- = --------------------- ---------
------------- - k z k x – ω μ d k y tan ( k y b + φ y ) = ---------------------------------------- (18a)
2  H  ω ε
Hz j ( rd – ε r0 )b
2
kx + ky 0

Hx j H0
- = --------------------- ----------
--------- - k k + ω ε d k y cot ( k y b + φ y ) = jω ( ε r – ε r )b (18b)
Ez 2 2  E0 z x  d 0
kx + ky

Similarly, at the width surface of the waveguide where x = a, Ey /Hz = –Ez /Hy = Zs.
Substituting (7), (8), (13), (15) and (17) into (16), the following relationships are
obtained:

Ey –j E0 1
---------- = --------------------- ---------
- k z k y + ω μ d k x tan ( k x a + φ x ) = ---------------------------------------- (19a)
Hz   jω ( ε rd – ε r0 )b
kx + ky H0
2 2

–Hy –j H0
-------------- = --------------------- ----------
- k z k y – ω ε d k x cot ( k x a + φ x ) = jω ( ε r – ε r )b (19b)
Ez 2  
kx + ky E0
2 d 0
322 KIM HO YEAP et al.

In order to obtain non-trivial solutions for (18) and (19), the determinants of the
equations must vanish. This leads us to the following set of transcendental equations
jω μ 0 k y tan ( k y b + φ y ) 1
-------------------------------------------------------- + ---------------------------------------- ×
2
kx + ky
2 jω ( ε rd – ε r0 )b

jω ε 0 k y cot ( k y b + φ y )  k z k x 2
× ------------------------------------------------------- – jω ( ε rd – ε r0 )b =  --------------------
- (20a)
2 2
kx + ky  k x2 + k y2 

jω μ 0 k x tan ( k x a + φ x ) 1
-------------------------------------------------------- + ---------------------------------------- ×
2
kx + ky
2 jω ( ε rd – ε r0 )b

jω ε 0 k x cot ( k x a + φ x )  k z k y 2
× ------------------------------------------------------- – jω ( ε rd – ε r0 )b =  --------------------
- (20b)
2 2
kx + ky  k x2 + k y2 

In (20), the transverse wave numbers kx and ky are the complex variables to be
solved for. A root-searching algorithm can be used to find the roots of kx and ky.
The solutions of kx and ky are then substituted into the dispersion relation which relates
the transverse wave numbers with the propagation coefficient kz:

2 2 2
kz = k – kx – ky (21)

Here, the propagation coefficient kz is a complex variable which is denoted as kz =

= β z – jα z, where β z is the phase coefficient and α z – the attenuation coefficient of the
waves. Hence, by extracting the real and imaginary values from kz, both the phase and
attenuation coefficients could be obtained.

3. Results and discussion

To validate our formulation, we compute the propagation coefficient kz of waves trav-
eling in a WR10 silicon waveguide with size 2.4×1.3 mm2. Since Marcatili’s formu-
lation has popularly been used in the design of dielectric rectangular waveguides [1–3],
we compare our results with those obtained from Marcatili’s approach and the finite
element method (FEM). The results from the FEM are simulated from Ansoft’s high
frequency structure simulator (HFSS). Unlike a hollow conducting rectangular wave-
y
guide in which TE10 is known to be the fundamental mode, Marcatili has suggested E 11
x
and E 11 to be the fundamental modes in a dielectric rectangular waveguide [19, 20].
In his analysis, however, none has been discussed on the condition when pq = 10.
WELLS has simulated the fields’ distribution in a dielectric rectangular waveguide [2].
It is shown in the results that some of the field patterns resemble closely that of the
TE10 mode, i.e., the cross-section exhibits half-wave field variation in the x-direction;
 Propagation in dielectric rectangular waveguides 323

2000
Attenuation [Np/m]

1500

1000

500

0
2.5 7.5 12.5 17.5 22.5 27.5 32.5
Frequency [GHz]
y y
Fig. 2. Attenuation of E 10 (dashed line) and E 11 (dashed-dotted-dotted line) in a dielectric rectangular
waveguide.

but almost uniform field distribution in the y-direction. WELLS’ result indicates that
mode 10 could have existed in a dielectric rectangular waveguide. However, it is not
certain if there is a switch in the fundamental mode from 10 to 11. Here, to further
analyze both modes, we have computed and compared the attenuation coefficient for
both pq = 10 and 11 modes using Marcatili’s transcendental equations. Since it is re-
vealed in WELLS’ paper [2] that when a shield is coated at the wall of the waveguide,
y y
E 11 changes to TE10, we have applied the equations which describe E pq in our calcu-
y y
lation. As can be seen in Fig. 2, E 10 has a lower cutoff frequency fc than that of E 11 .
This is to say that, notwithstanding the material used for the wall, pq = 10 remains un-
changed as the first mode to propagate in a rectangular waveguide. Indeed, such phe-
nomenon is to be expected. Since it has been found that the first mode in a circular
dielectric waveguide remains similar to that of its hollow conducting counterpart [40],
naturally, this phenomenon should not have changed for the case of a rectangular wave-
guide as well.
y
Figures 3 and 4 depict the attenuation of E 10 in the dielectric rectangular wave-
guide. As can be clearly seen in Fig. 3, the attenuation predicted by both Marcatili’s
transcendental method and our method agrees very well with HFSS simulation result
at frequencies at the vicinity of cutoff fc. As shown in both Figs. 3 and 4, the attenuation
and the cutoff frequency fc predicted by Marcatili’s closed form equation are somewhat
lower than the simulation results. Since the closed form equation is a simplification of
its transcendental form, the significant discrepancies found using this approximate
method should be of no surprise at all and can be attributed to the assumptions made
to simplify the formulation. After close inspection on the attenuation above fc, it could
be observed from Fig. 4 that the attenuation computed using our method agrees very
well with the simulation results and is, in fact, almost indistinguishable with each other;
Marcatili’s transcendental method, on the other hand, has overestimated the attenua-
tion exhibited in the dielectric waveguide. Hence, it is sufficient to say that although
Marcatili’s transcendental equation shows high accuracy below cutoff fc, it fails to give
accurate loss prediction for waves propagating above fc. One reason why our result is
 324 KIM HO YEAP et al.

1180
Attenuation [Np/m]

780

380

–20
2.5 7.5 12.5 17.5
Frequency [GHz]
y
Fig. 3. Attenuation of E 10 below cutoff, computed using Marcatili’s closed form equations (dashed
-dotted-dotted line), Marcatili’s transcendental equations (dashed line), our method (solid line), and HFSS
simulation (dashed-dotted line).

19.5
Attenuation [Np/m]

14.5

9.5

4.5

–0.5
18.2 18.3 18.4 18.5 18.6
Frequency [GHz]
y
Fig. 4. Attenuation of E 10 above cutoff, computed using Marcatili’s closed form equations (dashed
-dotted-dotted line), Marcatili’s transcendental equations (dashed line), our method (solid line), and HFSS
simulation (dashed-dotted line).

found to be in close agreement with the simulation result is that our method has not
only considered the interaction of fields at the boundary of the width and height sur-
faces, but also those at the four edges of the rectangular waveguide. By including the
analysis of fields at the edges, allowing the penetration of fields at the wall of the wave-
guide, as well as accounting for the superposition of modes, our method actually gives
a more realistic behaviour of the propagation of fields in the dielectric rectangular
waveguide.
Despite being popularly implemented in the millimeter and submillimeter circuits,
we found that data and analysis which compare the performance of both dielectric and
metallic waveguides are surprisingly rare in the literature. Here, we investigate the at-
tenuation in three different kinds of rectangular waveguides, i.e., a silicon waveguide,
a hollow copper waveguide and a silicon waveguide coated with a copper wall. The size
Propagation in dielectric rectangular waveguides 325

of both silicon waveguides remains as 2.4×1.3 mm2. Since the cutoff frequency fc of
a waveguide is dependent on the constitutive properties of the dielectric medium, as
follows [41, 42]:

1 pπ 2 qπ 2
f c = -------------------------------- ------------ + ------------ (22)
2π μ d ε d  a   b 

we have adjusted the size of the hollow conducting waveguide so as to give the same
cutoff frequency fc as the other two waveguides. The size of the hollow waveguide is
given as 8.28×4.49 mm2. The attenuation in the silicon waveguide is computed using
(20) and (21). The attenuations in both the hollow copper waveguide and the silicon
waveguide with the copper wall, on the other hand, are computed based on the equa-
tions in [34]. For convenience, we outline the transcendental equations for computing
the transverse wave numbers kx and ky in [34] as follows:

jω μ 0 k y tan ( k y b + φ y ) μ 0
-------------------------------------------------------- + ---------------------------
- ×
2
kx + ky
2 σc
ε c – j ----------
ω

σc
jω ε 0 k y cot ( k y b + φ y ) ε c – j ----------  k z kx 2
× ------------------------------------------------------- –
ω
---------------------------- =  --------------------
- (23a)
2
kx + ky
2 μ0  k x2 + k y2 

jω μ 0 k x tan ( k x a + φ x ) μ 0
-------------------------------------------------------- + ---------------------------
- ×
2
kx + ky
2 σc
ε c – j ----------
ω

σc
jω ε 0 k x cot ( k x a + φ x ) ε c – j ----------  k z ky 2
ω
× ------------------------------------------------------- – ---------------------------- =  --------------------
- (23b)
2
kx + ky
2 μ0  k x2 + k y2 

where εc and σc are respectively the permittivity and conductivity of the copper wall.
Like the case of the dielectric waveguide in this paper, the transverse wave numbers
are first numerically solved. The solutions are then substituted into (21) to obtain the
attenuation constant of the metallic waveguides.
Figure 5 depicts the attenuation of the dominant mode in the waveguides at fre-
quency f below cutoff fc, while Figs. 6 and 7 illustrate the attenuation beyond cutoff.
As can be observed in Fig. 5, at f below fc, the loss in the two silicon waveguides is
comparable to each other. At f above fc, however, Figs. 6 and 7 show that the silicon
326 KIM HO YEAP et al.

1180
Attenuation [Np/m]

780

380

–20
2.5 7.5 12.5 17.5
Frequency [GHz]

Fig. 5. Attenuation of the dominant mode below cutoff, in a silicon rectangular waveguide (solid line),
silicon rectangular waveguide with copper wall (dashed-dotted line), and hollow copper rectangular
waveguide (dashed line).

4
Attenuation [Np/m]

0
18.35 18.45 18.55 18.65
Frequency [GHz]

Fig. 6. Attenuation of the dominant mode immediately after cutoff, in a silicon rectangular waveguide
(solid line), silicon rectangular waveguide with copper wall (dashed-dotted line), and hollow copper
rectangular waveguide (dashed line).

1.6
Attenuation [Np/m]

1.2

0.8

0.4

0.0
20 21 22 23
Frequency [GHz]

Fig. 7. Attenuation of the dominant mode above cutoff, in a silicon rectangular waveguide (solid line),
silicon rectangular waveguide with copper wall (dashed-dotted line), and hollow copper rectangular
waveguide (dashed line).
Propagation in dielectric rectangular waveguides 327

waveguide surrounded with the copper wall exhibits considerably higher loss. Since
wave propagation is generally confined within the waveguides, radiation loss is prac-
tically negligible in both types of waveguides. Hence, the two main factors which con-
tribute to the loss in a waveguide are the dielectric and conduction losses [41]. The loss
in the metallic waveguide is found to be higher mainly because, besides having dielectric
loss at the silicon core, it also experiences conduction loss at the copper wall. With
the absence of the outer conducting wall, the dielectric silicon waveguide, on the other
hand, only experiences dielectric loss. This study confirms the notion that dielectric
waveguides are generally believed to have lower loss, compared to their metallic coun-
terparts [1–3]. It is worthwhile noting, however, that there is one exceptional case in
which the loss in a metallic waveguide could be significantly lower than dielectric
waveguides. As shown in Figs. 5 to 7, the loss in the hollow conducting copper wave-
guide is considerably lower than that in the dielectric silicon waveguide. This is be-
cause air has generally much lower dielectric loss than any other kind of dielectric
materials. The low loss found in hollow waveguides is also the reason why hollow con-
ducting waveguides are widely used in radio receiver systems built particularly to de-
tect the extremely weak extraterrestrial signals at millimeter and submillimeter
wavelengths [43–46]. However, it could also be seen here that while hollow wave-
guides exhibit much lower attenuation, they come at the expense of size. For wave-
guides which allow signals with the same cutoff frequencies to propagate, the size of
the hollow waveguide is usually larger. After close inspection on (22), we can find that
the size of the hollow waveguide is about ε r1/2d
times larger than its dielectric counter-
part. It is also worthwhile noting that the fabrication cost for hollow conducting wave-
guides is usually higher than dielectric waveguides as well. This is partly due to the
highly conducting material which is more expensive than dielectric; and partly also,
because the process involved in the fabrication of metallic waveguides is usually more
laborious. Unlike dielectric waveguides which generally require only the technique of
lithography, etching and dielectric deposition, fabricating metallic waveguides may re-
quire the additional step of electroforming the conducting layer onto the dielectric core.
Electroforming is an electrodeposition process which involves immersing the wave-
guide (which is solely dielectric at this stage) into a conducting electrolyte so as to
allow metallic ions to build up at the outer layer, forming a metallic coating at the wave-
guide. This additional step will certainly contribute to the cost in fabricating metallic
waveguides.

4. Conclusion
A fundamental and accurate technique to compute the propagation constant of waves
in a dielectric rectangular waveguide is proposed. The formulation is based on match-
ing the fields to the constitutive properties of the material at the boundary. At the wave-
guide wall the surface current density divided by the tangential electric field is matched
with the surface impedance of the wall. Doing so, we obtain two sets of equations which
describe the surface impedance at the width surface and another two sets at the height
328 KIM HO YEAP et al.

surface. The equations admit non-trivial solutions only when their determinants are
zero. The expansion of the determinants lead to transcendental equations, whose roots
are the allowed values for the transverse wave numbers in the x and y directions, i.e.,
kx and ky , respectively, for different modes. The wave propagation constant kz could
be found by relating kx, ky , and kz using the dispersion relation.
The attenuation curves obtained are in good agreement with those obtained from
the finite element method (FEM). An important implication of this work is that the
fundamental mode is observed to be pq = 10. It is also observed that hollow conducting
waveguides exhibit much lower attenuation than dielectric waveguides. This can be
explained by the low dielectric loss in free space, compared to other dielectric mate-
rials. Although more superior in preserving the energy of the waves, for a signal with
the same fc to propagate, the hollow waveguide is generally much larger in size com-
pared to its dielectric counterparts.
Acknowledgements – Part of this work has been supported by the Fundamental Research Grant Scheme
FRGS funded by the Ministry of Education, Malaysia (project: FRGS/2/2013/SG02/UTAR/02/1).

References
[1] HAMILTON D.P., GREEN R.J., LEESON M.S., Simple estimation formulas for rectangular dielectric
waveguide single-mode range and propagation constant, Microwave and Optical Technology
Letters 49(3), 2007, pp. 503–505.
[2] WELLS C.G., Analysis of shielded rectangular dielectric rod waveguide using mode matching, PhD
Thesis, University of Southern Queensland, Australia, 2005.
[3] DUDOROV S., Rectangular dielectric waveguide and its optimal transition to a metal waveguide, PhD
Thesis, Helsinki University of Technology, Finland, 2002.
[4] KUZNETSOV M., Expressions for the coupling coefficient of a rectangular-waveguide directional
coupler, Optics Letters 8(9), 1983, pp. 499–501.
[5] LEUNG K.W., SO K.K., Rectangular waveguide excitation of dielectric resonator antenna, IEEE
Transactions on Antennas and Propagation 51(9), 2003, pp. 2477–2481.
[6] SALEHI M., BORNEMANN J., MEHRSHAHI E., Wideband substrate-integrated waveguide six-port power
divider/combiner, Microwave and Optical Technology Letters 55(), 2013, pp. 2984–2986.
[7] LIU J., SAFAVI-NAEINI S., CHOW Y.L., ZHAO H., New method for ultra wide band and high gain
rectangular dielectric rod antenna design, Progress in Electromagnetics Research C 36, 2013,
pp. 131–143.
[8] ENG GEE LIM, ZHAO WANG, JING CHEN WANG, LEACH M., RONG ZHOU, CHI-UN LEI, KA LOK MAN,
Wearable textile substrate patch antennas, Engineering Letters 22(2), 2014, pp. 94–101.
[9] STRATTON J.A., Electromagnetic Theory, Chapter 9, McGraw-Hill, New York, 1941.
[10] COLLIN R.E., Field Theory and Guided Waves, Chapter 5, IEEE Press, New York, 1991.
[11] BUCK J.A., Fundamentals of Optical Fibers, Wiley, New York, 2004.
[12] YEAP K.H., THAM C.Y., YEONG K.C., WOO H.J., Wave propagation in lossy and superconducting
circular waveguides, Radioengineering 19(2), 2010, pp. 320–325.
[13] YASSIN G., THAM C.Y., WITHINGTON S., Propagation in lossy and superconducting cylindrical wave-
guides, [In] Proceedings of the 14th International Symposium on Space Terahertz Technology,
Tucson, Arizona, 2003, pp. 516–519.
[14] KNOX R.M., TOULIOS P.P., Integrated circuit for millimeter through optical frequency range, [In]
Proceedings of the MRI Symposium on Submillimeter Waves, Polytechnic Press, 1970, pp. 497–516.
Propagation in dielectric rectangular waveguides 329

[15] KUMAR A., THYAGARAJAN K., GHATAK A.K., Analysis of rectangular-core dielectric waveguides:
an accurate perturbation approach, Optics Letters 8(1), 1983, pp. 63–65.
[16] CAI Y., MIZUMOTO T., NAITO Y., Improved perturbation feedback method for the analysis of rectan-
gular dielectric waveguides, Journal of Lightwave Technology 9(10), 1991, pp. 1231–1237.
[17] YOUNG P.R., COLLIER J., Solution of lossy dielectric waveguide using dual effective-index method,
Electronics Letters 33(21), 1997, pp. 1788–1789.
[18] XIA J., MCKNIGHT S.W., VITTORIA C., Propagation losses in dielectric image guides, IEEE Trans-
actions on Microwave Theory and Techniques 36(1), 1988, pp. 155–158.
[19] MARCATILI E.A.J., Dielectric rectangular waveguide and directional coupler for integrated optics,
The Bell System Technical Journal 48(7), 1969, pp. 2071–2102.
[20] CHARLES J., BAUDRAND H., BAJON D., A full-wave analysis of an arbitrarily shaped dielectric wave-
guide using Green’s scalar identity, IEEE Transactions on Microwave Theory and Techniques 39(6),
1991, pp. 1029–1034.
[21] OGUSU K., Numerical analysis of the rectangular dielectric waveguide and its modifications, IEEE
Transactions on Microwave Theory and Techniques 25(11), 1977, pp. 874–885.
[22] HORN L., LEE C.A., Choice of boundary conditions for rectangular dielectric waveguides using
approximate eigenfunctions separable in x and y, Optics Letters 15(7), 1990, pp. 349–350.
[23] CHENG Y.H., LIN W.G., Investigation of rectangular dielectric waveguides: an iteratively equivalent
index method, IEE Proceedings J – Optoelectronics 137(5), 1990, pp. 323–329.
[24] KIM C.M., JUNG B.G., LEE C.W., Analysis of dielectric rectangular waveguide by modified effective
-index method, Electronics Letters 22(6), 1986, pp. 296–298.
[25] KRAMMER H., Field configurations and propagation constants of modes in hollow rectangular
dielectric waveguides, IEEE Journal of Quantum Electronics 12(8), 1976, pp. 505–507.
[26] GOELL J.E., A circular-harmonic computer analysis of rectangular dielectric waveguides, The Bell
System Technical Journal 48(7), 1969, pp. 2133–2160.
[27] RAHMAN B.M.A., DAVIES J.B., Finite-element analysis of optical and microwave waveguide prob-
lems, IEEE Transactions on Microwave Theory and Techniques 32(1), 1984, pp. 20–28.
[28] VALOR L., ZAPATA J., Efficient finite element analysis of waveguides with lossy inhomogeneous
anisotropic materials characterized by arbitrary permittivity and permeability tensors, IEEE Trans-
actions on Microwave Theory and Techniques 43(10), 1995, pp. 2452–2459.
[29] SCHWEIG E., BRIDGES W.B., Computer analysis of dielectric waveguides: a finite difference method,
IEEE Transactions on Microwave Theory and Techniques 32(5), 1984, pp. 531–541.
[30] SOLBACH K., WOLFF I., The electromagnetic fields and the phase constants of dielectric image lines,
IEEE Transactions on Microwave Theory and Techniques 26(4), 1978, pp. 266–274.
[31] EYGES L., GIANINO P., WINTERSTEINER P., Modes of dielectric waveguides of arbitrary cross sectional
shape, Journal of the Optical Society of America 69(9), 1979, pp. 1226–1235.
[32] MABAYA N., LAGASSE P.E., VANDENBULCKE P., Finite element analysis waveguides of optical, IEEE
Transactions on Microwave Theory and Techniques 29(6), 1981, pp. 600–605.
[33] ATHANASOULIAS G., UZUNOGLU N.K., An accurate and efficient entire-domain basis Galerkin’s method
for the integral equation analysis of integrated rectangular dielectric waveguides, IEEE Transac-
tions on Microwave Theory and Techniques 43(12), 1995, pp. 2794–2804.
[34] YEAP K.H., THAM C.Y., YASSIN G., YEONG K.C., Attenuation in rectangular waveguides with finite
conductivity walls, Radioengineering 20(2), 2011, pp. 472–478.
[35] YEAP K.H., THAM C.Y., YEONG K.C., YEAP K.H., A simple method for calculating attenuation in
waveguides, Frequenz – Journal of RF-Engineering and Telecommunications 63(11–12), 2009,
pp. 236–240.
[36] YEAP K.H., THAM C.Y., YEONG K.C., LIM E.H., Full wave analysis of normal and superconducting
microstrip transmission lines, Frequenz – Journal of RF-Engineering and Telecommunications
64(3–4), 2010, pp. 59–66.
330 KIM HO YEAP et al.

[37] BLACKLEDGE J., BABAJANOV B., Three-dimensional simulation of the field patterns generated by an in-
tegrated antenna, IAENG International Journal of Applied Mathematics 43(3), 2013, pp. 138–153.
[38] WEI B.L., XIONG C., YUE H.W., WEI X.M., XU W.L., ZHOU Q., DUAN J.H., Ultra wideband wireless
propagation channel characterizations for biomedical implants, IAENG International Journal of
Computer Science 42(), 2015, pp. 41–45.
[39] KOLUNDZIJA B.M., OGNJANOVIC J.S., Electromagnetic Modeling of Composite Metallic and Dielectric
Structures, Artech House, 2002.
[40] BALANIS C.A., Advanced Engineering Electromagnetics, Wiley, New York, 1989.
[41] CHENG D.K., Field and Waves Electromagnetics, Addison Wesley, 1989.
[42] YEAP K.H., YEONG K.C., CHONG K.H., WOO H.J., RIZMAN Z.I., Plots of field distribution in a rec-
tangular waveguide, International Journal of Electronics, Computer and Communications Technol-
ogies 1(2), 2011, pp. 5–13.
[43] YEAP K.H., THAM C.Y., NISAR H., LOH S.H., Analysis of probes in a rectangular waveguide, Frequenz
– Journal of RF-Engineering and Telecommunications 67(5–6), 2013, pp. 145–154.
[44] YEAP K.H., LAW Y.H., RIZMAN Z.I., CHEONG Y.K., ONG C.E., CHONG K.H., Performance analysis of
paraboloidal reflector antennas in radio telescopes, International Journal of Electronics, Computer
and Communications Technologies 4(1), 2013, pp. 21–25.
[45] WITHINGTON S., Terahertz astronomical telescopes and instrumentation, Philosophical Transactions
of the Royal Society A: Mathematical, Physical and Engineering Sciences 362(1815), 2004, pp. 395
– 402.
[46] PHILLIPS T.G., KEENE J., Submillimeter astronomy, Proceedings of the IEEE 80(11), 1992, pp. 1662
–1678.

Received August 19, 2015

in revised form November 6, 2015