# 8.14.

Problems 361

improving liquid crystal displays, and other products, such as various optoelectronic
components, cosmetics, and ”hot” and ”cold” mirrors for architectural and automotive
windows. 9
8.14 Problems Waveguides
8.1 Prove the reﬂectance and transmittance formulas (8.4.6) in FTIR.
8.2 Computer Experiment—FTIR. Reproduce the results and graphs of Figures 8.4.3–8.4.5.
8.3 Computer Experiment—Surface Plasmon Resonance. Reproduce the results and graphs of
Figures 8.5.3–8.5.7.
8.4 Working with the electric and magnetic ﬁelds across an negative-index slab given by Eqs. (8.6.1)
and (8.6.2), derive the reﬂection and transmission responses of the slab given in (8.6.8).
8.5 Computer Experiment—Perfect Lens. Study the sensitivity of the perfect lens property to the
Waveguides are used to transfer electromagnetic power efﬁciently from one point in
deviations from the ideal values of  = −0 and μ = −μ0 , and to the presence of losses by
reproducing the results and graphs of Figures 8.6.3 and 8.6.4. You will need to implement
space to another. Some common guiding structures are shown in the ﬁgure below.
the computational algorithm listed on page 329. These include the typical coaxial cable, the two-wire and mictrostrip transmission lines,
hollow conducting waveguides, and optical ﬁbers.
8.6 Computer Experiment—Antireﬂection Coatings. Reproduce the results and graphs of Figures
In practice, the choice of structure is dictated by: (a) the desired operating frequency
8.7.1–8.7.3.
band, (b) the amount of power to be transferred, and (c) the amount of transmission
8.7 Computer Experiment—Omnidirectional Dielectric Mirrors. Reproduce the results and graphs
losses that can be tolerated.
of Figures 8.8.2–8.8.10.
8.8 Derive the generalized Snel’s laws given in Eq. (8.10.10). Moreover, derive the Brewster angle
expressions given in Eqs. (8.11.4) and (8.11.5).
8.9 Computer Experiment—Brewster angles. Study the variety of possible Brewster angles and
reproduce the results and graphs of Example 8.11.1.
8.10 Computer Experiment—Multilayer Birefringent Structures. Reproduce the results and graphs
of Figures 8.13.1–8.13.2.

Fig. 9.0.1 Typical waveguiding structures.

Coaxial cables are widely used to connect RF components. Their operation is practi-
cal for frequencies below 3 GHz. Above that the losses are too excessive. For example,
the attenuation might be 3 dB per 100 m at 100 MHz, but 10 dB/100 m at 1 GHz, and
50 dB/100 m at 10 GHz. Their power rating is typically of the order of one kilowatt at
100 MHz, but only 200 W at 2 GHz, being limited primarily because of the heating of
the coaxial conductors and of the dielectric between the conductors (dielectric voltage
breakdown is usually a secondary factor.) However, special short-length coaxial cables
do exist that operate in the 40 GHz range.
Another issue is the single-mode operation of the line. At higher frequencies, in order
to prevent higher modes from being launched, the diameters of the coaxial conductors
must be reduced, diminishing the amount of power that can be transmitted.
Two-wire lines are not used at microwave frequencies because they are not shielded
and can radiate. One typical use is for connecting indoor antennas to TV sets. Microstrip
lines are used widely in microwave integrated circuits.

9.1. Longitudinal-Transverse Decompositions 363 364 9. Waveguides

Rectangular waveguides are used routinely to transfer large amounts of microwave where , μ denote the permittivities of the medium in which the ﬁelds propagate, for
power at frequencies greater than 3 GHz. For example at 5 GHz, the transmitted power example, the medium between the coaxial conductors in a coaxial cable, or the medium
might be one megawatt and the attenuation only 4 dB/100 m. within the hollow rectangular waveguide. This medium is assumed to be lossless for
Optical ﬁbers operate at optical and infrared frequencies, allowing a very wide band- now.
width. Their losses are very low, typically, 0.2 dB/km. The transmitted power is of the We note that ˆz·ˆ z = 1, ˆ
z×ˆ z = 0, ˆ
z · ET = 0, ˆz · ∇ T Ez = 0 and that ˆ
z × ET and
order of milliwatts. z × ∇ T Ez are transverse while ∇ T × ET is longitudinal. Indeed, we have:
ˆ

z × ET = ˆ
ˆ z × (x
ˆ Ex + y ˆ Ex − x
ˆ Ey )= y ˆ Ey
9.1 Longitudinal-Transverse Decompositions ∇ T × ET = (x
ˆ ∂x + y
ˆ ∂y )×(xˆ Ex + y z(∂x Ey − ∂y Ex )
ˆ Ey )= ˆ

In a waveguiding system, we are looking for solutions of Maxwell’s equations that are Using these properties and equating longitudinal and transverse parts in the two
propagating along the guiding direction (the z direction) and are conﬁned in the near sides of Eq. (9.1.4), we obtain the equivalent set of Maxwell equations:
vicinity of the guiding structure. Thus, the electric and magnetic ﬁelds are assumed to
have the form:
∇ T Ez × ˆ
z − jβ ˆ
z × ET = −jωμHT
∇ T Hz × ˆ
z − jβ ˆ
z × HT = jωET

E(x, y, z, t)= E(x, y)ejωt−jβz ∇ T × ET + jωμ ˆ
z Hz = 0
(9.1.1) (9.1.5)
H(x, y, z, t)= H(x, y)e jωt−jβz ∇ T × HT − jω ˆ
z Ez = 0
∇ T · ET − jβEz = 0
where β is the propagation wavenumber along the guide direction. The corresponding ∇ T · HT − jβHz = 0
wavelength, called the guide wavelength, is denoted by λg = 2π/β.
Depending on whether both, one, or none of the longitudinal components are zero,
The precise relationship between ω and β depends on the type of waveguiding struc-
we may classify the solutions as transverse electric and magnetic (TEM), transverse elec-
ture and the particular propagating mode. Because the ﬁelds are conﬁned in the trans-
tric (TE), transverse magnetic (TM), or hybrid:
verse directions (the x, y directions,) they cannot be uniform (except in very simple
structures) and will have a non-trivial dependence on the transverse coordinates x and
y. Next, we derive the equations for the phasor amplitudes E(x, y) and H(x, y). Ez = 0, Hz = 0, TEM modes
Because of the preferential role played by the guiding direction z, it proves con- Ez = 0, Hz = 0, TE or H modes
venient to decompose Maxwell’s equations into components that are longitudinal, that Ez = 0, Hz = 0, TM or E modes
is, along the z-direction, and components that are transverse, along the x, y directions. Ez = 0, Hz = 0, hybrid or HE or EH modes
Thus, we decompose:
In the case of TEM modes, which are the dominant modes in two-conductor trans-
mission lines such as the coaxial cable, the ﬁelds are purely transverse and the solution
E(x, y)= x ˆ Ey (x, y) + ˆ
ˆ Ex (x, y)+y z Ez (x, y) ≡ ET (x, y)+ˆ
z Ez (x, y) (9.1.2) of Eq. (9.1.5) reduces to an equivalent two-dimensional electrostatic problem. We will
     
transverse longitudinal discuss this case later on.
In all other cases, at least one of the longitudinal ﬁelds Ez , Hz is non-zero. It is then
In a similar fashion we may decompose the gradient operator:
possible to express the transverse ﬁeld components ET , HT in terms of the longitudinal
ones, Ez , Hz .
∇=x
ˆ ∂x + yˆ ∂y + ˆ
z ∂z = ∇ T + ˆ
z ∂z = ∇ T − jβ ˆ
z (9.1.3)
   Forming the cross-product of the second of equations (9.1.5) with ˆ z and using the
transverse BAC-CAB vector identity, ˆ z × (ˆz × HT )= ˆ z · HT )−HT (ˆ
z(ˆ z·ˆ z)= −HT , and similarly,
where we made the replacement ∂z → −jβ because of the assumed z-dependence. In- z × (∇
ˆ ∇ T Hz × ˆz)= ∇ T Hz , we obtain:
troducing these decompositions into the source-free Maxwell’s equations we have:
∇ T Hz + jβHT = jω ˆ
z × ET

∇ × E = −jωμH ∇T − jβˆ
(∇ z)×(ET + ˆ
z Ez )= −jωμ(HT + ˆ
z Hz ) Thus, the ﬁrst two of (9.1.5) may be thought of as a linear system of two equations
∇ × H = jωE ∇T − jβˆ
(∇ z)×(HT + ˆ
z Hz )= jω(ET + ˆ
z Ez ) in the two unknowns ˆ z × ET and HT , that is,
⇒ (9.1.4)
∇·E=0 ∇T − jβˆ
(∇ z)·(ET + ˆ
z Ez )= 0 z × ET − ωμHT = jˆ
βˆ z × ∇ T Ez
(9.1.6)
∇·H=0 ∇T − jβˆ
(∇ z)·(HT + ˆ
z Hz )= 0 ∇T Hz
z × ET − βHT = −j∇
ω ˆ

Hz .14) c2 ω2 ω2 √ 1 − c2 The quantity k = ω/c = ω μ is the wavenumber a uniform plane wave would ω have in the propagation medium . (9. negative.7) as follows: Although k2c stands for the difference ω2 μ − β2 .1. Waveguides  The solution of this system is: where the medium impedance is η = μ/. Some related deﬁnitions are the cutoff frequency and the cutoff wavelength k2c ηTM deﬁned as follows: z × (ˆ Using the result ˆ z × ET )= −ET .8) ηTE =  .15) jβ 1 conducting waveguide k2c takes on certain quantized positive values.1. in a TEM line. Using (9. we can write also:  ω2 η ω2c k2c = ω2 μ − β2 = − β2 = k2 − β2 (cutoff wavenumber) (9. HT − z × ET = ˆ ∇ T Hz ηTM β λc = 2π/kc . λc = (cutoff frequency and wavelength) (9.1. in a z × ET = − ˆ z × ∇ T Ez + ηTE∇ T Hz ˆ k2c dielectric waveguide k2c is positive inside the guide and negative outside it.10) c c ω 1 j Often. it is easily shown that the left-hand sides of the Note that λ is related to the free-space wavelength λ0 = 2πc0 /ω = c0 /f by the remaining four of Eqs.9.1.10) is expressed in terms of the wavelengths λ = 2π/k = 2πc/ω. Longitudinal-Transverse Decompositions 365 366 9.7) ηTE ηTM = η2 . Eq.5) and expressed ET .1. we have: k2c ηTM  An alternative and useful way of writing these equations is to form the following   1 ω ω2 linear combinations.6): β= ω2 2 − ωc = 1 − c2 and ω = ω2c + β2 c2 (9. With these deﬁnitions. it turns out that the boundary conditions for each waveguide type force k2c to take on certain values. we solve for ET and HT : 2π jβ ωc = ckc .1. (9.8). so that η/c = μ and ηc = 1/. We note the properties: jβ jωμ z × ET = − ˆ z × ∇ T Ez − ˆ ∇ T Hz k2c k2c ηTE ω2 (9. or ω in terms of β and ωc .1.1. k2c HT = − z × ∇ T Ez + ∇ T Hz ˆ is zero. and characterize the propagating modes. which can be jβ positive. Taking jβ 1 HT = − ∇ T Hz + z × ∇ T Ez ˆ the positive square roots of Eq.1.1. ηTM = η 1− (9. which are equivalent to Eq. μ. in a hollow (9.1. and λg = 2π/β.1.16). = 2 2 (9. (9. (9.1.1.1.5) take the forms: refractive index of the dielectric material λ = λ0 /n.12) ∇ T · ET − jβEz = − 2 ∇2T Ez + k2c Ez β βc ω ω kc jβ ∇ T · HT − jβHz = − 2 ∇2T Hz + k2c Hz kc . we may rewrite Eq. It follows from k2 = k2c + β2 that (9. HT λc in terms of Ez .9) ET = − ∇ T Ez − ηTE ˆ z × ∇ T Hz kc k2c (transverse ﬁelds) (9.1.1.16) We can then express β in terms of ω and ωc .13) jω jβ ηTM β c HT = − z × ∇ T Ez − 2 ∇ T H z ˆ k2c kc  where we deﬁned the so-called cutoff wavenumber kc by: Because βc/ω = 1 − ω2c /ω2 . or zero.1. (9.1.17) j ET − ηTE HT × ˆ z= ∇ T Ez 1 1 1 λ β = + ⇒ λg =  (9.11) λ2 λ2c λ2g λ2 1− 2 So far we only used the ﬁrst two of Maxwell’s equations (9. It is convenient at this point to introduce the transverse impedances for the TE and jωμ 2 ∇ T × ET + jωμ ˆ z Hz = z ∇T Hz + k2c Hz ˆ TM modes by the deﬁnitions: k2c jω 2 ∇T × HT − jω ˆ z Ez = − 2 ˆ z ∇T Ez + k2c Ez kc ωμ ω β βc jβ ηTE = =η . ηTM = =η (TE and TM impedances) (9. For example.

∇2T Ez + k2c Ez = 0 The Helmholtz equations (9. Longitudinal-Transverse Decompositions 367 368 9.9.1. Waveguides where ∇2T is the two-dimensional Laplacian operator: ∇2T = ∇ T · ∇ T = ∂2x + ∂2y (9.19) 2 ∇T Hz + kc Hz = 0 2 . ∇ T × (ˆ z × ∇ T Hz )= ˆ z ∇T Hz . it is necessary that the longitudinal ﬁelds Ez (x.1.19) now read: (Helmholtz equations) (9. It follows that in order to satisfy all of the last four of Maxwell’s equations (9. y) satisfy the two-dimensional Helmholtz equations: Fig.1.5).1.18) 2 and we used the vectorial identities ∇ T × ∇ T Ez = 0.1. and ∇ T · (ˆ z × ∇ T Hz )= 0. y). Hz (x. 9.1.1 Cylindrical coordinates.

Once. HT are . 1 ∂ ∂Ez 1 ∂2 Ez These equations are to be solved subject to the appropriate boundary conditions for ρ + + k2c Ez = 0 ρ ∂ρ ∂ρ ρ2 ∂φ2 each waveguide type. the ﬁelds Ez . Hz are known. the transverse ﬁelds ET .

(9. z. (9.1. H → −E . that is. we 1 ˆ ˆ (∂ρ Hz )−ρ z × ∇ T Hz = φ ˆ (∂φ Hz ) summarize the form of the above solutions in the two types of coordinate systems.  → μ. Below. The transverse gradient and Laplace operator are in cylindrical coordinates: With the ﬁeld solutions at hand. (9.1. Cylindrical Coordinates The relationship between cartesian and cylindrical coordinates is shown in Fig. 18. 9. we have x = ρ cos φ and y = ρ sin φ. making the substitutions: jβ jβ 1 Ex = − ∂x Ez + ηTE ∂y Hz Hx = − ∂x Hz − ∂y Ez k2c k2c ηTM . (9. To get the full x. we obtain: The cross-sections of practical waveguiding systems have either cartesian or cylin- drical symmetry. such as the rectangular waveguide or the coaxial cable. we obtain for the longitudinal components: jβ 1 jβ 1 Eρ = − ∂ρ Ez − ηTE ∂φ Hz Hρ = − ∂ρ Hz + ∂φ Ez (∂x + ∂y )Ez + kc Ez = 0 2 2 2 k2c ρ k2c ηTM ρ (9. the equations for HT may be obtained from those of ET by a so-called duality transformation.19) is straightforward.1. Using z × ∇ T Hz = y the identity ˆ ˆ ∂x Hz − x ˆ ∂y Hz . the ρ ∂ρ ∂ρ ρ2 ∂φ2 above solutions must be multiplied by the factor ejωt−jβz .16) and (9.24) (∂2x + ∂2y )Hz + k2c Hz = 0 jβ 1 jβ 1 1 Eφ = − ∂φ Ez + ηTE ∂ρ Hz Hφ = − ∂φ Hz − ∂ ρ Ez k2c ρ k2c ρ ηTM Eq. 9.2 Power Transfer and Attenuation From the triangle in the ﬁgure.1. The cylindrical ˆ Eρ + φ Cartesian Coordinates coordinates version of (9. resulting in a complete solution of Maxwell’s equations for 1 ∂ ∂Hz 1 ∂ 2 Hz 2 ρ + + k c Hz = 0 the guiding structure. Duality is discussed in greater detail in Sec.1.16) are: The cartesian component version of Eqs. μ→ (duality transformation) (9.21) E → H.25) jβ jβ 1 Ey = − 2 ∂y Ez − ηTE ∂x Hz Hy = − 2 ∂y Hz + ∂x Ez kc kc ηTM These imply that η → η−1 and ηTE → η−1 TM .1. z×ρ Noting that ˆ ˆ and ˆ ˆ =φ ˆ = −ρ z×φ ˆ . ρ The decomposition of a transverse vector is ET = ρ ˆ Eφ .1.1. (9.2. (9.1.16) becomes for the transverse components: For either coordinate system.16).1. t dependence of the propagating ﬁelds.1.20) . one can determine the amount of power transmitted .23) computed from Eq.1. y.

as well as the transmission losses. The total power carried by the ﬁelds ∂ ˆ 1 ∂ .22) ∂ρ ρ ∂φ ρ ∂ρ ∂ρ ρ2 ∂φ2 vector over the cross-sectional area of the guide: . 1 ∂ ∂ 1 ∂2 along the guide direction is obtained by integrating the z-component of the Poynting ∇T = ρ ˆ +φ ∇ 2T = ρ + (9. along the guide.1.

where n ˆ is the outward PT = Pz dS .2. that is. the ohmic losses per unit conductor area are calculated by Eq.8) dz Ca 2 Cb 2 mechanisms. δ= = skin depth (9. we have: 1 Pz = Re(ET × H∗ z T )·ˆ (9. then α = αd + αc (9.2.2. Power Transfer and Attenuation 369 370 9.8. (2. the corresponding  attenuation coefﬁcient is calculated from Eq. 9. the ﬁelds are determined on the assumption α= = diel = αd + αc (attenuation constant) (9.2. Third.9) provide a systematic methodology by which to calculate the ω2 μI 1 ω2 μ ω tan δ transmitted power and attenuation losses in waveguides.2.9) applies also to the dielectric losses so that in general Ploss arises from two parts. First. Waveguides Second. In practice. or by its conductivity σd —the two being related by σd = ω tan δ. For example.2.2. the effective dielectric constant of the medium may have a negative imaginary part I that includes both conductive and polarization losses. Figure It is easily veriﬁed that only the transverse components of the ﬁelds contribute to 9.2. Finally.7) to this area. Equations (9. one due to the dielectric and one due to the conducting walls. if αd and αc are the attenuations due to the ohmic losses in the dielectric and in the conducting walls. Eq. (2. Then. Adding similar terms for all the other conductors gives the total power loss per unit z-length: e−jβz → e−jβc z = e−(α+jβ)z = e−αz e−jβz (9. For weakly lossy dielectrics (I ).2.1 shows such an inﬁnitesimal conductor area dA = dl dz. where Pz = Re(E × H∗ )·ˆ z (9.7) 2σ 2σ σδ ωμσ The transmission losses can be quantiﬁed by replacing the propagation wavenumber β by its complex-valued version βc = β − jα.2.5) 2β 2 β 2c 1 − ω2c /ω2 examples later on.4). where α is the attenuation constant. Applying Eq.6) dA dldz 2 For waveguides with conducting walls. We will apply it to several αd = = tan δ =  (dielectric losses) (9.1) S 2 normal to the conductor. after setting μ = 1/c2 and βc/ω = 1 − ω2c /ω2 .2.8. Pz can be written in the form: cross-sectional periphery of the conductor.2. (2.8.2. (2. the transmission losses are due primarily to where Rs is the surface resistance of the conductor given by Eq. we may make the approximation: where Ca and Cb indicate the peripheries of the conductors.6. .2. (ω)=  − jI .1)–(9. The conductor losses are more complicated to calculate.4) The ohmic losses in the dielectric can be characterized either by its loss tangent. the corresponding complex-valued wavenumber βc is obtained by the replacement:   β = ω2 μ − k2c → βc = ω2 μ(ω)−k2c Fig.2.22):   ω2 μI ω2 μI βc = ω2 μ( − jI )−k2c = β2 − jω2 μI = β 1−j β−j β2 2β Ploss  αc = (conductor losses) (9. the following Ploss P + Pcond approximate procedure is adequate.1 Conductor surface absorbs power from the propagating ﬁelds. say tan δ. (9.7). with I =  tan δ. Rs = =η = . where dl is along the the power ﬂow. More generally. the losses ωμ ω 1 2 are due to absorption and scattering by imperfections. the magnetic ﬁelds on the conductor surfaces are determined and the corre- 1 sponding induced surface currents are calculated by Js = nˆ × H.2) dPloss dPloss 1 2 = = Rs |Js |2 (9.10) 2PT 2PT that the conductors are perfect. In dielectric waveguides. ohmic losses in (a) the conductors and (b) the dielectric medium ﬁlling the space between    the conductors and in which the ﬁelds propagate.3)   dPloss 1 1 The quantity α is the sum of the attenuation constants arising from the various loss Ploss = = Rs |Js |2 dl + Rs |Js |2 dl (9. (9. The Integrating Eq.6) around the periphery of the conductor gives the power loss per z-dependence of all the ﬁeld components is replaced by: unit z-length due to that conductor.2.2.9.9) 2PT Resulting in the attenuation constant.

3.10) by applying it sepa- rately to the TE and TM modes.2) takes the simple form in all three cases: Integrating over the cross-sectional area of the guide gives the dielectric loss per unit waveguide length (i. We recall from Eq.2) 1 2 2ηT 2 Pdiel = ωI |E|2 dS 2 S Applying this to the TE case. the consistency of the system requires that S 2 2ηTE S 2ωμ S ηTE = ηTM . TE. 1 1 1 Pz = Re(ET × H∗ z= T )·ˆ |ET |2 = ηT |HT |2 (9. both Ez and Hz vanish. 1 HT = z × ET ˆ (9. 2 S 2 S From any point view. we ﬁnd. (9. the electric and magnetic ﬁelds Pdiel ω2 μI αd = = satisfy: 2PT 2β The TM case is a bit more involved. or equivalently in (9.3) η 1 1   These are the same as in the case of a uniform plane wave. and TM modes 371 372 9. are given by. and TM cases. Using Eq. For example. ηTM dPloss 1   1   in the TEM. = Re Jtot · E ∗ = ωI E · E ∗  dV 2 2 Because of Eq. the power ﬂow per unit cross-sectional area described by the Poynting vector Pz of Eq. or in (9.1) from Problem 9. that is.13. (9. η.5). ηT Jtot = J + jωD.11. TEM.1.13) implies ω = βc. arising from both a conduction and polarization current.e. (9. The electric ﬁeld ET is   . β2 + k2c = ω2 μ. and the ﬁelds are fully transverse. TEM modes In TEM modes.2. or ω = βc.9.3.17) vanish. y dependence.9. which by virtue of Eq.3.2.16). where ηT is the transverse impedance of the particular mode type. we ﬁnd. (9.1. except here the ﬁelds Pdiel = ωI |E|2 dS = ωI |Ez |2 + |ET |2 dS 2 S 2 S are not uniform and may have a non-trivial x. one obtains the condition k2c = 0. Waveguides Eq. (1.1. 1 after using the result. One can set 1 1 Pdiel = ωI |E|2 dS = ωI |ET |2 dS Ez = Hz = 0 in Maxwell equations (9.3.6) that the losses per unit vol. TE.5) for αd can also be derived directly from Eq. It also implies that ηTE .3.. ηTE .1).1.1. ηTM must both be equal to the medium impedance η.2. if 1 1 β PT = Re(ET × H∗ z dS = T )·ˆ |ET |2 dS = |ET |2 dS the right-hand sides of Eq. HT = z × ET ˆ (9.17).1) ume in a dielectric medium. (9. (9. (9. z-length). Thus.

3.4) PT = |ET |2 dS = ∇ 2 4 |∇ T Ez | dS = 2 |Ez |2 dS ∇ T · ET = 0 2ηTM S 2β S kc 2kc S .5). determined from the rest of Maxwell’s equations (9. which read: 1 β2 1 β2 = ωI ∇ |Ez |2 + |∇ E T z | 2 dS = ω I 1 + |Ez |2 dS 2 S k4c 2 kc2 S 2 ∇ T × ET = 0 1 ω β ωβ (9.1.

HT are related to each other in the same way as in the case of uniform plane waves propagat- ing in the z-direction. 1 1 1 zero Ez and Hz components. (For backward moving ﬁelds. 11. The time-varying propagating ﬁelds will be given by 2PT ωβ β 2k2c Eq. 9.1. two-wire. (9. (9.1.1 and discuss the cases of the coaxial. (9.19) is a hybrid solution with non. we look at the specialized forms of these equations Pz = Re(ET × H∗ z= T )·ˆ |ET |2 = η|HT |2 (9. Here. their cross-product points in the z-direction. TE. and TM modes the Poynting vector Pz of Eq.16) and (9. ET (x. TE. y). and strip lines.1. replace β by −β. and TM modes.) We explore this electrostatic point of view further in Sec. with ω = βc. and they satisfy: .5) 2 2η 2 in the cases of TEM.1). 1 β2 These are recognized as the ﬁeld equations of an equivalent two-dimensional elec- ω I 1 + P 2 k2c ω2 μI trostatic problem. they are perpendicular to each other.3). that is. Once this electrostatic solution is found. (9. the magnetic ﬁeld αd = diel = = is constructed from Eq.3.2) will be: The general solution described by Eqs.2. Because of the relationship between ET and HT . One common property of all three types of modes is that the transverse ﬁelds ET .3 TEM.3.

3. (∂2x + ∂2y )Hz + k2c Hz = 0 jβ jβ Hx = − ∂x Hz .3. ET = ηTE HT ×ˆ z.3.4. Ey = −ηTE Hx And.3. Thus.6) propagating in the z-direction. The Poynting vector k2c takes the form: ET = ηTE HT × ˆ z 1 1 1 β2 Pz = Re(ET × H∗ z= T )·ˆ |ET |2 = ∇T Ez |2 |∇ (9.3.3. .1.17) that ET is completely determined from HT .9.4 Rectangular Waveguides Next.6) is: dielectric material . except the wave impedance is replaced by ηTE .2) then takes the form: 9.7) walls.3. the cylindrical coordinate version: Fig. we may assume that the lengths 2 2ηTE 2 2 kc a. Waveguides TE modes ∇2T Ez + k2c Ez = 0 TE modes are characterized by the conditions Ez = 0 and Hz = 0. 9. and TM modes 373 374 9. The guide is typically ﬁlled with air.2. the relationship of ET and HT is identical to that of uniform plane waves jβ HT = − ∇ T Hz (TE modes) (9. 9. we discuss in detail the case of a rectangular hollow waveguide with conducting 1 1 1 1 β2 Pz = Re(ET × H∗ z= T )·ˆ ∇ T H z |2 |ET |2 = ηTE |HT |2 = ηTE 4 |∇ (9.11) The relationship of ET and HT is identical to that of uniform plane waves propagating 2 2ηTM 2ηTM k4c in the z-direction. TEM. that is. μ may be assumed. TE.4. Without loss of generality.16). (9. but the wave impedance is now ηTM .8) k2c kc Ex = ηTE Hy .1 Rectangular waveguide. jβ The ﬁeld HT is determined from the second of (9.10) for TE modes are obtained from the equations: 1 HT = z × ET ˆ ηTM ∇2T Hz + k2c Hz = 0 Again.1. all ﬁeld components ET = − ∇ T Ez k2c (TM modes) (9. Hy = − 2 ∂y Hz (9. but any other The cartesian coordinate version of Eq.1. (9. b of the inner sides satisfy b ≤ a. as shown in Fig. The Poynting vector of Eq. (9. It follows from the second of Eqs.

we begin by looking jβ ∂Hz jβ 1 ∂Hz (9. Eφ = −ηTE Hρ ∂2x Hz (x)+k2c Hz (x)= 0 where we used HT × ˆ z = (ρ ˆ Hφ )×ˆ ˆ Hρ + φ ˆ Hρ + ρ z = −φ ˆ Hφ . It also follows that: the following equations.3. Hφ = − 2 for solutions of Eq. Because there is no y-dependence. the Helmholtz equation k2c ∂ρ kc ρ ∂φ reduces to: Eρ = ηTE Hφ . However. It follows from the ﬁrst of Eqs. HT = η− 1 z × ET . TM modes only the former will satisfy the boundary conditions.9) Hρ = − .3.17) that HT is Hz (x)= H0 cos kc x (9. and hence Hy = 0 and Ex = 0.1) completely determined from ET .1. (9.3.) Therefore. In this case.3. the solution is: TM modes have Hz = 0 and Ez = 0.8) that ∂y Hz = 0. that is. (9. it follows from the ﬁrst of (9.16).6): jβ jβ jβ Hx (x)= − ∂x Hz = − 2 (−kc )H0 sin kc x = H0 sin kc x ≡ H1 sin kc x k2c kc kc . which are dual to the TE equations (9.8) that depend only on x. Therefore.4. (9. The most general solution is a linear combination of cos kc x and sin kc x. 1 ∂ ∂Hz 1 ∂2 Hz 2 ρ + + kc Hz = 0 The simplest and dominant propagation mode is the so-called TE10 mode and de- ρ ∂ρ ∂ρ ρ2 ∂φ2 pends only on the x-coordinate (of the longest side.1. The ﬁeld ET is determined TM ˆ where H0 is a (complex-valued) constant. so that all ﬁeld components for TM modes are obtained from from Eq.

5. it is parallel to the left and be constants. y) = H1 sin kx x cos ky y Ex (x.2) that will satisfy the TE boundary conditions are fc = ωc /2π.9. Ey (x) vanishes because sin kc x does. z. x = a. and wavelength λc = 2π/kc = c/fc are: cos kx x and cos ky y. y) = E2 sin kx x cos ky y a a 2a Fig.5 Higher TE and TM modes E0 = −ηTE H1 = −ηTE H0 = −jη H0 kc ωc where we used ηTE = ηω/βc.8) that: The dominant mode is the one with the lowest cutoff frequency or the longest cutoff wavelength.4.5.2 Electric ﬁeld inside a rectangular waveguide. (9.8) becomes: Ey (x)= E0 sin kc x jωt−jβz Ey (x. y)= F(x)G(y) Hz (x)= H0 cos kc x Hz (x. jβkx jβky H1 = H0 .3.4.4. z. y. G (y)+k2y G(y)= 0 (9.2 depicts the electric ﬁeld Ey (x)= E0 sin kc x = E0 sin(πx/a) of this mode where we deﬁned the constants: as a function of x.4) ωc = . H2 = H0 k2c k2c ωky ωkx E1 = ηTE H2 = jη H0 . it is normal to the top and bottom sides. H1 = H0 kc (9. fc = . the longitudinal magnetic ﬁeld will be: cnπ cn 2a Hz (x.4. λc = (TEn0 modes) (9. the mode TE10 having n = 1.2) jβ ω 9.5.5) a 2a n It then follows from the rest of the equations (9. t)= E0 sin kc x e F (x) G (y) F (x)G(y)+F(x)G (y)+k2c F(x)G(y)= 0 ⇒ + + k2c = 0 (9. we write: right sides. y)= H0 cos kx x cos ky y (TEnm modes) (9.4) k2c = k2x + k2y (9.5.4. y) = H2 cos kx x sin ky y Ey (x. Thus. the boundary condition requires: = −k2x .4. that is.and G-terms must y-direction. the non-zero ﬁeld components are: To construct higher modes.1) to satisfy: nπ kc a = nπ ⇒ kc = (9.2) This requires that kc a be an integral multiple of π: where the constants k2x and k2y are constrained from Eq. The most general solutions of (9. we look for solutions of the Helmholtz equation that are factorable in their x and y dependence: Hz (x. t)= H1 sin kc x ejωt−jβz (9. On the right F (x) G (y) side. On the left side. x = 0. y) = E1 cos kx x sin ky y π cπ c (9.5. t)= H0 cos kc x ejωt−jβz Hx (x)= H1 sin kc x ⇒ Hx (x.4. Because the electric ﬁeld is in the Because these must be valid for all x.3) a These are the so-called TEn0 modes.6) Hy (x.5) kc = . fc = .5. It has: Hx (x. = −k2y or F(x) G(y) Ey (a)= E0 sin kc a = 0 ⇒ sin kc a = 0 F (x)+k2x F(x)= 0 . y (inside the guide).4. E2 = −ηTE H1 = −jη H0 ωc kc ωc kc . y. the boundary conditions require that there be F(x) G(y) no tangential electric ﬁeld at any of the wall sides. the F. y. Rectangular Waveguides 375 376 9. ωc = . Waveguides Then. But. Thus. 9.5.3.3) Then. z. In summary. Eq. the corresponding electric ﬁeld will be: jβ Ey (x)= −ηTE Hx (x)= −ηTE H0 sin kc x ≡ E0 sin kc x kc where we deﬁned the constants: jβ Fig. independent of x and y. 9. (9. The corresponding cutoff frequency ωc = ckc .1) Assuming perfectly conducting walls. λc = 2a (TE10 mode) (9.

bandwidth for the TE10 mode. y)= E0 sin kx x sin ky y (TMnm modes) (9. and erratic operation. Similarly.1.8) If the frequency ω is greater than the cutoff frequencies of several modes. β relationship (9.5.5. 1 jωky 1 1 jωkx 1 On the other hand. λnm =  2  2 (9. and the wave will attenuate in the z- a b direction. then f01 ≤ f20 and the two smallest frequencies H1 = − E2 = E0 . A mode with cutoff frequency ωc will propagate only if its frequency is ω ≥ ωc .5. Thus. the operating bandwidth becomes: 0. where fc = f10 = c/2a. E2 = − E0 f10 ≤ f01 .3.1.6) a b This follows from the ω. except both n and m bandwidth is the interval [fc . ky b = mπ ⇒ kx = . y = b.10):  correspond to the TEnm modes. not just hollow conducting wave- guides. we must now choose the solutions sin kx and sin ky y. If b ≤ a/2. and f01 = c/2b. that is. the wavenumber β is real-valued and the wave will propagate.9) frequency. We will see later that the total amount of transmitted power in this mode is propor- Waveguide modes can be excited by inserting small probes at the beginning of the tional to the cross-sectional area of the guide. if in addition to having the waveguide. kx a = nπ .5. but with x and a replaced by y and If we arrange the cutoff frequencies in increasing order. The family of TM modes can also be constructed in a similar fashion from Eq. However. In terms of the wavelength must be non-zero (otherwise Ez would vanish identically. the cutoff frequencies λ = c/f . and the single-mode bandwidth is deter- that they are given by the same expressions as (9. Thus. a ≤ λ ≤ 2a. Conversely. with a penetration depth δ = 1/α: The cutoff frequencies fnm = ωc /2π = ckc /2π and wavelengths λnm = c/fnm are: e−jβz = e−αz  2  2 n m 1 fnm = c + . If ω < ωc .) The two cases b ≤ a/2 and b ≥ a/2 are depicted in Fig. longitudinal electric ﬁelds is: These remarks apply to all waveguiding systems. then 1/a ≤ 1/2b and therefore.8).5.6 Operating Bandwidth vanish on the top wall.) Thus. This interval deﬁnes the is parallel to all walls. Waveguides The boundary conditions are that Ey vanish on the right wall. ky = (9. the frequency must be restricted to the interval Assuming Ez (x. ωc1 = 0.3. Higher TE and TM modes 377 378 9. the wave will attenuate exponentially along the guide direction. Ey are the same as It is evident from this ﬁgure that in order to achieve the widest possible usable before. β = −jα. it will. 2fc ]. in which case f01 = f10 and the smallest frequencies are f10 and f20 . 9.5). in coaxial cables the lowest mode is the TEM mode having no cutoff Ez (x. Thus. then 2a 2b none of the modes can propagate. the guide dimensions must satisfy b ≤ a/2 so that the These conditions imply that kx . For example. The boundary conditions on Ex . This may The conditions require that kx a and ky b be integral multiples of π: cause undue amounts of dispersion. (9. y)= E0y sin kx a cos ky y = 0 .5. (9. TE and TM modes with non-zero cutoff frequencies do exist and place an upper limit on the usable bandwidth of the TEM mode. . say. Because we assumed that b ≤ a. the lowest mode has no cutoff.† one has no control over which modes will actually be carrying the transmitted signal. The TE0m modes are similar to the TEn0 modes.7) ω < ωc . ky will be given by Eq. (9. y)= F(x)G(y).2). then. x = a. and that Ex 9. then all 2a 2b n m + of these modes can propagate. we obtain the same equations (9. nπ mπ or λ < λc . H2 = E1 = − H0 ηTM ωc kc η ηTM ω c kc η are f10 and f01 (except when b = a.” of the desired mode. so that only the lowest mode will propagate. the cutoff wavenumbers of these modes These ω2 − ω2c kc = k2x + k2y take on the quantized values: ω2 = ω2c + β2 c2 ⇒ β2 = c2  2  2 If ω ≥ ωc . the operating bandwidth of the guide. b)= E0x cos kx x sin ky b = 0 lowest mode can propagate. it follows that always jβkx jβky E1 = − E0 . and wavelengths are the same as in Eq. f20 = c/a = 2f10 .10) and one ﬁnds optical ﬁbers.6. in The rest of the ﬁeld components can be worked out from Eq. the smallest cutoff frequencies are f10 = c/2a.6).5 ≤ a/λ ≤ 1.5. determined in terms of E0 : In rectangular waveguides. All waveguiding systems are operated in a frequency range that ensures that only the Ey (a. Because Ez ωc1 < ω < ωc2 . we must require that Ez vanish on all walls. except now the constants are mined by the next cutoff frequency. ab. But if nπ mπ kc = + (TEnm modes) (9. ωc1 < ωc2 < ωc3 < · · · .5. (9. β becomes imaginary. f20 ≤ f01 . to ensure single-mode operation.5. Ex (x.10).9. or. if ω is less than all cutoff frequencies. and in addition. if a/2 ≤ b ≤ a. b. The probes are chosen to generate an electric ﬁeld that resembles the ﬁeld † Murphy’s law for waveguides states that “if a mode can propagate. distortion. where we used ηTM = ηβc/ω. If several modes can propagate simultaneously. so that the two lowest k2c k2c cutoff frequencies are f10 and f20 .

95 S 2.05 C 0. that is.01 73.17 14.49 11.6. We may also calculate the distribution of electromagnetic energy along the guide.84 1.00 Ka 27 kW 0.7. a b a b 1 1 late the energy density of the ﬁelds and determine the velocity by which electromagnetic We = we (x.019 WR-159 1.1 Operating bandwidth in rectangular waveguides. Most practical guides 0 0 2ηTE follow these side proportions.074 39.7): 1 1 1 Pz = |ET |2 = |Ey (x)|2 = |E0 |2 sin2 kc x 2ηTE 2ηTE 2ηTE The transmitted power is obtained by integrating Pz over the cross-sectional area Fig.7.56 8.) We have chosen one example from each microwave band. If there is a “canonical” guide.20 12.007 magnetic ﬁelds are: WR-284 2.14 21. as name a b fc fmin fmax band P α measured by the time-averaged energy density.311 9.87 49.55 1. The energy densities of the electric and WR-510 5. πx a sin2 kc x dx = sin2 dx = (9.05 17.5fc = 0.6.59 0. We recall a b that the non-zero ﬁeld components are: 1 1 Wm = μ |H1 |2 sin2 kc x + |H0 |2 cos2 kc x dxdy = μ |H1 |2 + |H0 |2 ab 0 0 4 8 Hz (x)= H0 cos kc x . (9. of the guide: a b widest bandwidth.10 2. Using the integral (9. Energy Density.1) .5 kW 2.64 7.110 we = Re E · E∗ = |Ey |2 2 2 4 WR-62 0.90 18.5 kW 1. if we integrate them over the cross-sectional area of the guide.08 2.1 lists some standard air-ﬁlled rectangular waveguides with their naming designations. and Group Velocity unit z-length.622 0.90 0.50 X 250 kW 0.7 MW 0. 1 PT = |E0 |2 sin2 kc x dxdy sion b must be chosen to be as large as possible.75 (9. Power Transfer.80 V 7. Ey (x)= E0 sin kc x (9.9 MW 0. we also require to have the maximum power transmitted.6.80 112. cutoff frequencies in GHz.74 Inserting the expressions for the ﬁelds.3) c 0 0 a 2 f = 1.70 K 50 kW 0.3. Hx (x)= H1 sin kc x .3) and an identical one for the cosine case.1)  a and using ηTE = ηω/βc = η/ 1 − ω2c /ω2 .7. inner side dimensions a.28 0. b in inches.1 Characteristics of some standard air-ﬁlled rectangular waveguides. 2fc ].7.7.00 W 3.45 2.80 75.00 Ku 140 kW 0.043 1 1 1 WR-90 0. we calculate the time-averaged power transmitted in the TE10 mode.176 WR-42 0. E0 = −ηTE H1 = −jη H0 (9.40 6. we ﬁnd: Next. we ﬁnd: 1 1 Table 9. it will have b = a/2 and be operated at a frequency a a that lies in the middle of the operating band [fc .60 3.20 L 9 MW 0.6.4) 4ηTE 4η ω2 band. and attenua.9.795 3. We also calcu.7 Power Transfer. 1 1 ω2c 2 tions in dB/m (the power ratings and attenuations are representative over each operating PT = |E0 | ab = |E0 |2 ab 1− (transmitted power) (9.42 0.2) kc ωc The Poynting vector is obtained from the general result of Eq.583 2 2 4 WR-15 0. minimum  and maximum recommended operating frequencies in GHz. wm = μ |H1 |2 sin2 kc x + |H0 |2 cos2 kc x 4 4 Because these quantities represent the energy per unit volume. we = |E0 |2 sin2 kc x . we obtain: Table 9.16 1.370 1 1 1 wm = Re μH · H∗ = μ |Hx |2 + |Hz |2 WR-28 0.10 0.148 0.52 WR-10 0. Noting the deﬁnite integral.7. we will obtain the energy distributions per 9. 9.08 26.05 59.34 2. Waveguides where jβ ω H1 = H0 .60 26.40 40. that is. the dimen.71 4. and Group Velocity 379 380 9. power ratings. y) dxdy = |E0 |2 sin2 kc x dxdy = |E0 |2 ab 0 0 0 0 4 8 energy ﬂows down the guide and show that it is equal to the group velocity. b = a/2. Energy Density.

on the left and right walls: dω ω Js = ±x ˆ × H = ±x ˆ × (x ˆ Hx + ˆ zHz )= ∓y ˆ Hz = ∓y ˆ H0 cos kc x vgr = . density. We = Wm . Taking differentials of  |H0 |2 .8) by integrating |Js |2 vgr = = =c 1− (9. whereas vph ≥ c. (9. the energy transport velocity will be the ratio ven = PT /W . consistent with the theory of = Rs |H0 |2 + |H1 |2 + Rs b|H0 |2 = |H0 |2 + |H1 |2 + |H0 |2 relativity.7.3) were derived assuming the boundary conditions for deed. we ﬁnd: the waveguide are given by Js = n ˆ × H.5) and noting that 1/η = 1/ μ = c. ven = vgr . a b where we used Eq. Waveguides Although these expressions look different.3 for uniform plane waves. which gives: 2 |Js | = |H0 |2 cos2 kc x + |H1 |2 sin2 kc x . or Ploss = 2 Rs |Js |2 dx + 2 Rs |Js |2 dy 2 0 2 0 a b vgr vph = c2 (9.7. the magnitudes of the For uniform plane waves and TEM transmission lines. we have: The equality of the electric and magnetic energies is a general property of wavegui- ding systems.7. The total energy density per unit length will be: 1 W = We + Wm = 2We = |E0 |2 ab (9. which follows from Eq. β2 ω2 μ μ |H1 |2 + |H0 |2 = μ |H0 |2 2 + |H0 |2 = μ|H0 |2 2 = 2 |E0 |2 = |E0 |2 The surface currents and tangential magnetic ﬁelds are shown in Fig.2).6.2). and transport velocity given in Eq. In. this gives Js = ∓y ˆ(±H0 )= y ˆ H0 .8) dβ ω ω2 around the four walls. For a rectangular waveguide. We note that vgr = βc2 /ω = c2 /vph .8. respectively. Using √ Eqs. we have ω = βc. we ﬁnd:  Fig. (left and right walls) both sides. (top and bottom walls)  dω βc2 ω2c The power loss per unit z-length is obtained from Eq.9.2.4. they are actually equal. Information trans- a Rs a 2b mission down the guide is by the group velocity and. For any dispersion relationship between ω and β. we have ω2 = ω2c + β2 c2 .7. Thus. 2. The induced surface currents on the inner walls of ships between the constants in (9. 2 2 a Using |H0 |2 +|H1 |2 = |E0 |2 /η2 from Sec. the group and phase velocities are deﬁned by Similarly.7.1. where the unit vector n ˆ is ±x ˆ and ±y ˆ on the left/right and bottom/top walls. that is. on the bottom and top walls.6) W ω Js = ±y ˆ × H = ±y ˆ × (x ˆ Hx + ˆ zHz )= ±(−ˆ z Hx + x ˆ Hz )= ±(−ˆ z H1 sin kc x + x ˆ H0 cos kc x) This is equal to the group velocity of the propagating mode. (1. The ﬁeld expressions (9.7. and |H0 |2 = (|E0 |2 /η2 )ωc2 /ω2 . 9. 9.4.7) dβ β At x = 0 and x = a.1 Currents on waveguide walls. In par- kc ωc η ticular.5) 4 According to the general relationship between ﬂux. We also encountered it in Sec. we obtain: 9.8. vph = (group and phase velocities) (9.4) and (9. we ﬁnd 2ωdω = 2c2 βdβ.8. it is less than c.7.1.8 Power Attenuation .7.10). PT ω2 ven = =c 1 − c2 (energy transport velocity) (9. the energy transport velocity is equal to the group 1 1 velocity. Power Attenuation 381 382 9. 9.7. Thus. using the property β2 /k2c + 1 = (β2 + k2c )/k2c = k2 /k2c = ω2 /ω2c and the relation. (9. (9. perfectly conducting wall surfaces.1).9) = Rs |H0 |2 cos2 kc x + |H1 |2 sin2 kc x dx + Rs |H0 |2 dy 0 0 The energy or group velocity satisﬁes vgr ≤ c. so that vgr = surface currents are on the four walls: vph = c. (9.

9. (9. The losses due to the ﬁlling dielectric can be determined from Eq.2.2.7.2. The attenuation constant is computed from Eqs. Rs a|E0 |2 2b ω2c In this section.4): . (9.9) and (9.5). we calculate the attenuation coefﬁcient due to the ohmic losses of the Ploss = 1+ 2η2 a ω2 conducting walls following the procedure outlined in Sec.

9. The maximum operating . Waveguides The cutoff frequency of the TE10 mode is fc = c/2a = 3. Power Attenuation 383 384 9.8.71 GHz.

2 |E0 |2 ab 1 − c2 4η ω Attenuation Coefficient Power Transmitted 0.8. bandwidth is the interval [fc .64.71. and the recommended interval is Rs a|E0 |2 2b ω2c [4. 7.1 1.1) is plotted in dB/m versus frequency in Fig.42] GHz. 9. (9.5 which gives: bandwidth . 2fc ]= [3. the calculated attenuation 2PT 1 ω2 constant αc from Eq.8. 7. 1+ Ploss 2η2 a ω2 αc = =  Assuming copper walls with conductivity σ = 5.05] GHz.2.8×107 S/m.

9 Reﬂection Model of Waveguide Propagation 0. as shown in Fig. then. but then the attenuation increases and the transmitted power Fig. wavevectors will be: At 10 GHz. 0.8.5 MV/m. Thus. 9. 9. a = 4. Because ω = 1. is 3 MV/m. Thus. the maximum allowed electric ﬁeld in the guide is E0 = 1. thus the smaller the guide dimensions. which gives a safety factor of two over the dielectric breakdown design it to be operated at 10 GHz. Its inner dimensions are a = 1. equivalently. For a 0. and  the maximum transmitted power assuming the maximum electric ﬁeld is one-half of the Because of the factor 1 − ω2c /ω2 in the denominator of αc and the numerator of PT .5 MV/m.75×30 GHz cm a= = = 4.6.2 Attenuation constant and transmitted power in a WR-159 waveguide.8.6. αc increases with decreasing b. dielectric strength of air. (9. Because Rs depends on f like f 1/2 . The power transmitted PT is calculated from Eq.59 and b = a/2 = H= k × E1 + k × E 1 0. 0 0 0 1 2 3 4 5 6 7 8 9 10 11 12 0 1 2 3 4 5 6 7 8 9 10 11 12 The main tradeoffs in a waveguiding system are that as the operating frequency f f (GHz) f (GHz) increases. the larger the attenuation.6. operating band.4) assuming a maximum breakdown Example 9.2. Then. Its value in dB/m is obtained by αdB = 8.125 k=x ˆ kx + ˆ z kz = x ˆ k cos θ + ˆ z k sin θ √ cm.06 ηb ω2c 1− ω2 0.7. Consider the C-band WR-159 air-ﬁlled waveguide whose 1ˆ 1 ˆ characteristics were listed in Table 9.   ˆ E1 e−jk ·r = y ˆ E1 e−jk·r + y E=y ˆ E1 ejkx x e−jkz z = E1 + E 1 ˆ E1 e−jkx x e−jkz z + y Example 9. while the power is almost zero.   Solution: If f is in the middle of the operating band. the guide dimensions are halved.04 0. a = 2.1 for the ponent of the form: C-band and X-band waveguides. 9.25 cm.02 bandwidth given ratio a/b. then the incident and reﬂected (from the right wall) Eq. Eq. the attenuation constant becomes very large near the cutoff frequency. we require b = a/2 = 2.75c 0.8.8.28 MW = 280 kW.8.5ωc .5 This is in units of nepers/m.0193 cm.104 dB/m.8.1.0386 and b = 2.1) 0. it will increase by a factor of 2. Then. so will the power. (9. η η . Because the area ab is reduced by a factor of four. The electric and magnetic ﬁelds will be the sum of an incident and a reﬂected com- The results of these two cases are consistent with the values quoted in Table 9. or. re. because f is doubled.1. Assume copper walls with conductivity σ = 5.7.2: WR-159 Waveguide. The power in megawatt scales is plotted in Fig.1: Design a rectangular air-ﬁlled waveguide to be operated at 5 GHz.12 MW. Solving for a. the attenuation will increase to the value k = −x ˆ kx + ˆ z kz = −x ˆ k cos θ + ˆ z k sin θ √ αc = 0. the dimensions of the guide must decrease in order to maintain the operat- ing band fc ≤ f ≤ 2fc . voltage of E0 = 1. Calculate the guide dimensions. A physical explanation of this behavior is given in the next section.08 2b ω2c 1+ 1 Rs a ω2 α (dB/m) PT (MW) αc =  (attenuation of TE10 mode) (9.9.686αc .25 and b = 1. WR-159 and WR-90. fc ≤ f ≤ 2fc .795 inches.75c/a. If θ is the angle of incidence.5 cm f 5 An intuitive model for the TE10 mode can be derived by considering a TE-polarized uniform plane wave propagating in the z-direction by obliquely bouncing back and forth For maximum power transfer.4) gives PT = 1. Then.1. then f = 1. where fc = c/2a. decreases as it is proportional to the guide’s area.8×107 S/m.5fc = 0. we ﬁnd 9. The operating frequency must lie in the middle of the of air of 3 MV/m. the attenuation constant in dB/m. PT = 1. (9. we have ωc /ω = 2/3. The dielectric strength of air between the left and right walls of the waveguide. This trend is noted in Table 9.12/4 = 0.037 · 2 2 = 0.037 dB/m. the factor √ Rs /b will increase by a factor of 2 2.1) gives αc = 0.

creating a standing wave. This model clariﬁes also the meaning of the group velocity. we see that Ey = E1 e−jkx x + E1 ejkx x e−jkz z the effective speed in the z-direction of the common-phase points will be vph = c/ sin θ 1 so that vph vgr = c2 .3) that:  ω2c vz = c sin θ = c 1− = vgr (9.  ωc ω2c cos θ = .9. For highly conducting walls.9. These ﬁeld expres.1. where the electric ﬁeld was taken to be polarized in the y direction.4. (9.1 shows a rectangular cavity with z-length equal to l formed by replacing These are identical to Eq. Indeed. which is recognized to be ηTE . the wave is bouncing left and right at normal incidence. Hz = cos θ E1 e−jkx x − E1 ejkx x e−jkz z η The boundary condition on the left wall. However.2. Thus. The phase velocity can also be understood geometrically. or.1 Reﬂection model of TE10 mode. A forward-moving wave shown in Fig. ω  ωc .10. Then.3) ω ω2 The ratio of the transverse components. causing the wave to zoom through guide almost at the speed of light. Reﬂection Model of Waveguide Propagation 385 386 9.9.3. 1 Hx = − sin θE0 sin kx x e−jkz z having a very high Q of the order of 10.9. and does not propagate towards the z-direction. On the other hand.3) implies also that at ω = ωc . The Ey = E0 sin kx x e−jkz z boundary conditions on the cavity walls force the ﬁelds to exist only at certain quantized resonant frequencies.4).000. for very large frequencies. the resonances are extremely sharp.1 Rectangular cavity resonator (and induced wall currents for the TEn0p mode.1) Higher TE and TM modes can also be given similar geometric interpretations in terms 1 of plane waves propagating by bouncing off the waveguide walls [890]. cavities can be used not only to efﬁciently store electro- j magnetic energy at microwave frequencies. E1 = −E1 = jE0 /2.9. through Eq.9. (9. The plane wave is bounc- ing left and right with the speed of light c.) 1 − c2 ω . kc a = nπ. From the indicated triangle at points 1. Hx = − sin θ E1 e−jkx x + E1 ejkx x e−jkz z η (9. Indeed. sin θ = 1− (9. requires that E1 + E1 = 0. This is equal to the group velocity. we have sin θ = 0. We may write 9. (9. implies. Waveguides The boundary condition on the right wall requires sin kx a = 0. It follows from the wavevector triangle in the ﬁgure that the angle will bounce back and forth from these walls. x = 0. η (9. −Ey /Hx . 9.9.9. that αc will be inﬁnite. the component of this velocity in the z-direction will be vz = c sin θ.9. Indeed.4.10 Resonant Cavities therefore.9.4) Hx sin θ ω2 Fig.9. is the transverse impedance.9).3) provided we identify β with kz and kc with kx . or θ = 0. the above expressions simplify into: Cavity resonators are metallic enclosures that can trap electromagnetic ﬁelds. resulting in a standing-wave pattern along of incidence θ will be given by cos θ = kx /k = kc /k. we have: Ey η η ηTE = − = =  (9. but also to act as precise oscillators and to Hz = cos θE0 cos kx x e−jkz z η perform precise frequency measurements. we observe in the sions become component-wise: rightmost illustration of the above ﬁgure that the planes of constant phase are moving obliquely with the speed of light c. 9. that is. the transmitted power is zero and this also Fig. as the sending and receiving ends of a waveguide by metallic walls. that is. Fig. 9.2. the angle θ will tend to 90o . which gives rise to the same condition as (9.10. it follows from Eq. 9. the z-direction.5) ω2 Eq. (9.2) Because of their high Q .

where we used the following deﬁnite integrals (valid because kc = nπ/a. top. Speciﬁcally. Resonant Cavities 387 388 9.6). where p is a non-zero integer. the standing. 0 0 2 0 0 2 wave patterns along the transverse directions require a = nλx /2 and b = mλy /2. The directions of these currents are shown in Fig.10. l = pλg /2 = a l sin2 kc x dx = cos2 kc x dx = . f101 = = √ (9.1. z) = −jE0 sin kc x Ae−jβz + Bejβz . which also changes the sign of ηTE = ηω/βc.10. all three cartesian components of the wave vector  The surface currents induced on the walls are related to the tangential magnetic ﬁelds are quantized. For simplicity. z) on the front/back walls. 9. or put differently. (9. z = 0 and z = l. z) = H0 cos kc x Ae−jβz + Bejβz . It follows that the Q -factor will be: coefﬁcients A. if present.2. For the same reason. sin2 βz dz = cos2 βz dz = (9. may be calculated by integrating the electric energy density over the cavity volume: a b l 1 1 W = 2We = 2 |Ey (x. Thus.2) specializes into: W ωμ (k2c + β2 )(abl) Hz (x.1.2. For a cubic cavity. x = 0 and x = a. Eqs. H1 = H0 (9.7) to replace Hx (x. H1 = H0 (9. therefore. and Q = 12724. or doubling the answer for the left. Q=ω = 2 Ploss 2Rs kc l(2b + a)+β2 a(2b + l) β For the TEn0p mode we have β = pπ/l and kc = nπ/a.10. (9. Starting with a linear combination of forward The power loss can be computed by integrating the loss per unit conductor area. (9.10.10.7) δ n2 2 1 p2 2 1 on the left/right walls. Waveguides Because the tangential components of the electric ﬁeld must vanish at the end-walls. Eq. Using the integrals (9. that is. a = W Q=ω (9. z) = H0 cos kc x sin βz . z) have z-dependence of the form sin βz.   1 bl al ab β Ploss = Rs |J s |2 dA = Rs H02 + (H02 + H12 ) + H12 Hx (x. or β = pπ/l. we ﬁnd f101 = 7. integrated over all six cavity sides.07 GHz. such as dielectric 3δ a 2π a 2 losses. ω101 = .10.10.6) 1 β2 ω = Rs H02 l(2b + a)+ 2 a(2b + l) Ey (x.3. the where we substituted H12 = H02 β2 /k2c .86×10−5 resonance centered at frequency ω. β = pπ/l) : these walls must coincide with zero crossings of the standing wave. cavities can be excited by inserting small probes It is easily veriﬁed that the electric and magnetic energies are equal. we may write Q = ω/Δω. we ﬁnd: ω Ey (x.10. or The ohmic losses are calculated from Eq.10. B must be chosen as A = −B = j/2. = |E0 |2 (abl)= μ|H0 |2 2 (abl)= μ |H0 |2 (abl) 8 8 ωc 8 k2c say 2a. (9. E0 = ηH0 n2 p2 ωc 1 + Q= a2 l2 As expected. the Q and the lowest resonant frequency are: Ploss √ where W is the total time-averaged energy stored within the cavity volume and Ploss is a cπ 2 ω c Q= . As in waveguides. requires the quantization conditions: β = pπ/l + + + a2 a b l2 l b and kc = nπ/a. we obtain the ﬁeld components: Eq. and backward waves in the TEn0 mode. Wave propagation in the z-direction is by total internal . and front sides. z) = jH1 sin kc x Ae−jβz − Be jβz .4) b = l. (9. cm.5) pπ/β. Therefore. W that generate ﬁelds resembling a particular mode.10. an a a l l integral multiple of half-wavelengths must ﬁt along the z-direction. over the six wall sides.2) 2 walls 2 4 2 kc   (9. we ﬁnd: Hz (x.8) the total power loss due to the wall ohmic losses (plus other losses.2. 9. and     (9. kx = nπ/a and ky = mπ/b. as shown in Fig.10. and therefore.11 Dielectric Slab Waveguides 4 vol 2 0 0 0   1 1 ω2 1 k2c + β2 A dielectric slab waveguide is a planar dielectric sheet or thin ﬁlm of some thickness. so is the frequency of the wave ω = c kx2 + k2y + β2 : by J s = n ˆ × Htan . δ = 7. z) = −jE0 sin kc x sin βz .5).) The ratio Δω = Ploss /W is usually identiﬁed as the 3-dB width of the For an air-ﬁlled cubic cavity with a = 3 cm. z)|2 dx dy dz = |E0 |2 sin2 kc x cos2 βz dx dy dz 9.1) ⎨H0 sin βz ⎪ (left & right) a b l 2 |J s | = H0 cos2 kc x sin2 βz + H12 sin2 kc x cos2 βz 2 (top & bottom) ⎪ ⎪ ⎩ 2 Such modes are designated as TEnmp or TMnmp .9.3) kc Rs in terms of the skin depth δ. Then.10. we ﬁnd for the currents on the six sides:  2  2  2 ⎧ 2 nπ mπ pπ ⎪ 2 ωnmp = c + + (resonant frequencies) (9.6). the vanishing of Ey (x.11.6) also describe backward-moving waves if one replaces β by −β. we consider the case H1 sin2 kc x (front & back) TEn0p . By requiring that Ey (x. Using Eq. z) = −H1 sin kc x cos βz . E0 = ηH0 4 kc ωc where ωc = ckc . The Q of the resonator can be calculated from its deﬁnition: The lowest resonant frequency corresponds to n = p = 1.

if x ≤ −a Fig. the continuity of Ey follows from the continuity of Hx . ⎪ ⎪ − 2 ∂x Hz (x)= − H1 cos kc x . x ≥ a.) For the even modes. and exist effectively within a skin depth distance 1/αc from the slab. if −a ≤ x ≤ a and outside the guide.8) kc αc αc Inside the slab. the quantity kc2 must be imaginary. sin kc2 x and cos kc2 x. the solutions of Eqs. For simplicity.8) using the same.3) −α2c = k20 n22 − β2 α2c = β2 − k20 n22 Similarly. if x ≤ −a ∂x Hz (x)+k2c1 Hz (x)= 2 0 for |x| ≤ a where we deﬁned the constants: (9. E3 = − ηTE H3 (9. decaying exponentially with distance from the slab. k2c takes different values inside and outside the guide: ⎪ ⎪ c c ⎪ ⎪ jβ jβ ⎪ ⎪ ⎩ − ∂x Hz (x)= H3 eαc x . and only eαc x Hz at x = a and x = −a implies that: to the left. then only both media. (9. the ﬁelds attenuate exponentially with the transverse distance . We note that ω.11.2) read: ∂2x Hz (x)+k2c Hz (x)= 0 for |x| ≤ a (9.1) become in this new notation: k2c = k20 n21 − β2 k2c = k20 n21 − β2 ⇒ (9. ⎩ H3 eαc x . Eqs. x ≤ −a.) therefore also Hx must be continuous. If αc is positive. 9.1 shows a typical electric ﬁeld pattern as a function of x. The corresponding Helmholtz equations in the Ey (x)= E2 e−αc x . where ηTE = ωμ0 /β is the same inside k2c2 = ω2 2 μ0 − β2 = ω2 0 μ0 n22 − β2 = k20 n22 − β2 (outside) and outside the slab. (9. are continuous across the dielectric interfaces at x = −a propagate at large x distances from the slab (they would correspond to the rays refracted and x = a. if x≥a (9.11. k2c2 = −α2c outside and k2c1 = k2c inside: the slab must satisfy 1 > 2 . The continuity of the solution e−αc x is physically acceptable to the right of the slab.11. E2 = ηTE H2 . k2c = ω2 μ − β2 . the ﬁelds would electric ﬁelds. if x≥a (9. if −a ≤ x ≤ a models for the conﬁning mechanism of waves propagating in optical ﬁbers.11. There. for if it is real. Setting kc1 = kc and kc2 = −jαc . the normal components of the magnetic ﬁeld Bx = μ0 Hx and from the inside into the outside. and odd TE modes (referring to the evenness or oddness of the resulting electric ﬁeld. the electric ﬁeld has the form: where k0 = ω/c0 is the free-space wavenumber. however. Because Ey = −ηTE Hx and ηTE is the same in If we set kc2 = −jαc .5) ⎪ ⎪ exist as evanescent waves outside it. The cutoff wavenumber kc appearing in the Helmholtz equation for Hz (x) ⎨ jβ jβ depends on the dielectric constant of the propagation medium. Ey . they also Hz (x)= H2 e−αc x . (9.11. β are the same inside ⎧ ⎪ ⎪ E cos kc x . the solutions outside will be e±αc x .11. that is.11.2) ∂2x Hz (x)+k2c2 Hz (x)= 0 for |x| ≥ a jβ jβ jβ E1 = ηTE H1 . Hz . In order for the waves to remain conﬁned in the near The boundary conditions state that the tangential components of the magnetic and vicinity of the slab. This follows from matching the tangential ﬁelds at all times t ⎨ 1 and all points z along the slab walls. n1 > n2 .4) ∂2x Hz (x)−α2c Hz (x)= 0 for |x| ≥ a The two solutions sin kc x and cos kc x inside the guide give rise to the so-called even Fig. if x≥a (even TE modes) (9. and outside. ⎨ 1 The propagating ﬁelds are conﬁned primarily inside the slab.1 Dielectric slab waveguide. 9. Hx (x)= − 2 ∂x Hz (x)= − H2 e−αc x . Thus. Waveguides x.6) ⎪ ⎪ −α α fore.7) ⎪ ⎪ regions inside and outside the guide are: ⎩ E3 eαc x .3. the solutions are sin kc1 x and cos kc1 x. or equivalently.4) have the form: reﬂection from the left and right walls of the slab. that is. if x ≤ −a k2c1 = ω2 1 μ0 − β2 = ω2 0 μ0 n21 − β2 = k20 n21 − β2 (inside) −α2c αc (9.11. Thus. we assume that the media to the left and right of the slab are the The corresponding x-components Hx are obtained by applying Eq. Dielectric Slab Waveguides 389 390 9. e±jkc2 x . the dielectric constants inside and outside appropriate value for k2c . and similarly for the refractive indices.9.11.11. if −a ≤ x ≤ a ⎪ ⎪ kc kc ⎪ ⎪ ordinate. Such waveguides provide simple ⎧ ⎪ ⎪ H sin kc x .11. Similarly.1) The electric ﬁelds are Ey (x)= −ηTE Hx (x). To guarantee total internal reﬂection.11. Eqs. ⎧ ⎪ ⎪ jβ jβ We only consider TE modes and look for solutions that depend only on the x co.11. (9.11.

Eqs. v for each value of R. if x≥a (even TE modes) (9. (9.11. if −a ≤ x ≤ a 1 H1 cos kc a = 1 H2 e−αc a and 1 H1 cos kc a = − 1 H3 e−αc a (9. we have for the solutions of Eq.18) provide three E2 = E3 = E1 eαc a cos kc a = E1 eαc a sin kc a (9.11) or (9.9) odd functions of x for the two families of modes.11. if x≥a (odd TE modes) (9.20) ⎪ ⎪ Similarly.21) in the equivalent forms: Hz (x)= H2 e−αc x .11. β.11. αc .11.11. we may rewrite Eqs. if x≥a (9.11) ⎪ ⎪ E sin kc x .11. if x ≤ −a ωa 2πf a 2πa The boundary conditions imply in this case: R = k0 aNA = NA = NA = NA (9.11. These solutions are obtained at the kc intersections of the curves v = u tan u and v = −u cot u with the circle of radius R. E2 = −E3 = E1 eαc a sin kc a = −E1 eαc a cos kc a (9.11. the single-mode condition reads 2πaNA /λ < 2. there are two solutions.11. (odd modes) (9.11.9. two even and one odd. if −a ≤ x ≤ a kc ⎨ 1 Ey (x)= E1 sin kc a e−αc (x−a) . there may be several possible solution pairs u. For π ≤ R < 3π/2. if x≥a (odd TE modes) (9. that is. we ﬁnd for the electric ﬁeld constants: ⎩ −E1 sin kc a eαc (x+a) . the free-space kc wavelength. if −a ≤ x ≤ a ⎨ 1 Ey (x)= E2 e−αc x .† For π/2 ≤ R < π.11.11. (9.19) ⎪ ⎪ ⎩ E1 cos kc a eαc (x+a) .10) ⎨ 1 kc αc kc αc Ey (x)= E1 cos kc a e−αc (x−a) . one even αc = −kc cot kc a (9.11. (9.15) where R is the normalized frequency variable: ⎪ ⎪ ⎩ E3 eαc x . the continuity of Hx implies (after canceling a factor of −jβ): ⎧ ⎪ ⎪ E cos kc x .18).11. there are three solutions.4): Next. (9.11. .11. (9. if x ≤ −a kc Given the operating frequency ω.11. Expressing E2 and E3 in terms of E1 . To solve them.13).7) and (9.11.11.11.3) and (9.2 shows the solutions for various values of the radius R αc corresponding to various values of ω.21) αc c20 For the odd TE modes. v2 + u2 = R2 . we discuss the numerical solutions of these equations. if x ≤ −a Eqs.12) requires that: kc ω2 2 cos kc a = sin kc a ⇒ αc = kc tan kc a (9.14) ⎪ ⎪ ⎩ H3 eαc x . for the electric ﬁeld constants: Because the functions tan u and cot u have many branches.9) and (9.11.11. 0 ≤ R < π/2. and. Fig.22) ⎧ v2 + u2 = R2 v2 + u2 = R2 ⎪ ⎪ E sin kc x .11.16) where NA = n21 − n22 is the numerical aperture of the slab and λ = c0 /f .11. The consistency of the last equation requires: It is evident from the ﬁgure that for small enough R. and ⎪ ⎪ H cos kc x .11.405. Dielectric Slab Waveguides 391 392 9. Deﬁning the dimen- ⎧ sionless quantities u = kc a and v = αc a.11. we summarize the forms of the electric ﬁelds in the two cases: Similarly.12) αc equations in the three unknowns kc . Waveguides We note that the electric ﬁelds Ey (x) given by Eqs. and † for an optical ﬁber.11.11.23) c0 c0 λ αc  H2 = H3 = H1 eαc a cos kc a = −H1 eαc a sin kc a (9. if −a ≤ x ≤ a ⎨ 1 (9. we add the two equations (9.15) are even or H1 sin kc a = H2 e−αc a and − H1 sin kc a = H3 e−αc a (9.18) and one odd.13) or (9. if x ≤ −a The resulting electric ﬁeld is: v = u tan u v = −u cot u (even modes) . 9.13) α2c + k2c = k20 (n21 − n22 )= (n1 − n22 ) (9. where a is the core radius. there is only one solution and it is even.11.17) that is.11.3) to eliminate β: The consistency of the last equations in (9.10) imply: αc ⎧ H2 = −H3 = H1 eαc a sin kc a = H1 eαc a cos kc a (9. (9.

M ≤ 2R/π < M + 1. . the u-range (9. m = 0.29) v2 + u2 = R2 If one had an approximate solution u. 9.11.34) have been implemented in the tan(u − Rm )= (9. Just such an approximate solution. there will be M+ 1 solutions indexed by m = 0.11. . we determine M as that integer satisfying Eq.m.11. (9. . M.11. Fig. Waveguides sin u cos(mπ/2)− cos u sin(mπ/2) tan(u − mπ/2)= cos u cos(mπ/2)+ sin u sin(mπ/2) Therefore. 1. . correct to order O(F): If m is even. . . m = 0. if R where u1 (m). Without going into the detailed justiﬁcation of this method.11. 2 .24). . . .9. . the approximation is as follows: u = Rm + w1 (m)u1 (m)+w2 (m)u2 (m) . Dielectric Slab Waveguides 393 394 9. .31)     1 π/4 + Rm 2R Vm = √ − Rm M = ﬂoor (maximum mode number) (9.27) G(u) 2 where G(u) is the derivative F (u).11. In general. u2 (m) are approximate solutions near and far from the cutoff Rm .11. M (9.29)–(9.26) deﬁnes a branch of tan u. there will be M + 1 solutions.2 Even and odd TE modes at different frequencies. The M + 1 branches of tan u  and cot u being intersected by the R-circle are those contained in the u-ranges: F(u)= u tan(u − Rm )−v = u tan(u − Rm )− R2 − u2 (9.11. . . M (9. 1. .11. we must ﬁnd the unique solution of the following system in the u-range Rm ≤ u < Rm+1 : v = u tan(u − Rm ) (mth mode) (9.11. a branch of cot u. and falls in the interval: w1 (m).11. and if m is odd. .32) Newton’s iteration is: Rm ≤ u < Rm+1 . where F(u) is deﬁned by to even modes if m is even and to odd modes if m is odd. the largest integer less than 2R/π: w1 (m)= exp −(R − Rm )2 /Vm2 . which converges very fast provided it is close to the true solution. one could reﬁne it by using Newton’s method. with usage: ⎩ − cot u .11. . which will correspond tion F(u)= 0.v.25 cos(π/4) π This solution serves as the starting point to Newton’s iteration for solving the equa- Then.err] = dslab(R. alternating between even and odd.Nit).11.25) ln 1.11.11.30) so on. 1. Nit do: F(u) (9. % TE-mode cutoff wavenumbers in a dielectric slab This follows from the trigonometric identity: . w2 (m) are weighting factors: Mπ (M + 1)π  ≤R< (9. if m is odd [u. that is.26) where for i = 1. whether even or odd. accurate to within one percent of the true solution.11. or. m = 0. . to ﬁnd the mth mode. w2 (m)= 1 − w1 (m) (9.11. (9. (9. M (9. 1. if m is even The solution steps deﬁned in Eqs.34) u v u ⎧ ⎨ tan u .33) mπ u=u− Rm = .22) into a single case by v u R2 noting the identity: G(u)= + + (9. v for the mth mode. We can combine the even and odd cases of Eq.24) 1 + 2R(R − Rm ) − 1 π R−m 2 2 u1 (m)= . u2 (m)= R 2 R+1 Given a value of R. . .11. was given by Lotspeich [930].28) MATLAB function dslab.

given the frequency f .5553 0.11. v are the (M + 1)-dimensional vectors of solutions.44 Electric Fields kc = k1 cos θ = k0 n1 cos θ 7 6 The value of β for each mode will generate a corresponding value for θ.ac.4641. we have: The frequency radius is R = 5. kx of k1 with an angle of incidence θ. the quantity αc decreases and the effective skin depth 1/αc increases. The at- 1 0 1 0 tenuation wavenumber αc outside the slab can also be expressed in terms of the total 5 1 2 internal reﬂection angles:   Ey (x) / E1 4 2 3 0 αc = β2 − k20 n22 = k0 n21 sin2 θ − n22 v 3 2 3 Since the critical angle is sin θc = n2 /n1 . the solution is ωm aNA /c0 = 2πfm aNA /c0 = mπ/2.9808 . The quantity fc is the vector of The lowest non-zero cutoff frequency is f1 = 8.9792458 of Appendix A.5207 8.5675 17. If we denote by k1 = k0 n1 the wavenumber within the slab. a. we calculate k0 = 6.381×10−8 .1: Dielectric Slab Waveguide. n2 = 1.7837 9.885×10−15 using only three Newton itera- [be. then the wavenumbers [be. Problem 9. αc would be slightly different if a more precise value of c0 interval fm ≤ f < fm+1 .33).6410] GHz.2: For the Example 9.1629 nepers/cm. n1 = 2. % dielectric slab guide tions.3971 10. err is the value of F(u) at the end of the iterations. and no (the Lotspeich approximation) iterations would result in the errors 2.11. Waveguides where Nit is the desired number of Newton iterations (9.11.35) where the upper limit is 4f1 .5664 rad/cm. The critical and total internal reﬂection angles of the four modes are found to be: Fig.9519 7. implying that there will be a the M + 1 cutoff frequencies deﬁned by the branch edges Rm = mπ/2. all θs are greater than θc .6359 4.0793 1.2832 and k1 = 12. if f = 5 GHz. The meaning of fm is that there are m + 1 propagating modes for each f in the the calculated values for kc .and x-components kz . m = 0. kc are the z. 9. that is. αc in cm .err] = dguide(f.2838 2.4414. 36. Nit = 2–3.n2. Dielectric Slab Waveguides 395 396 9. kc .1960o .) Thus.8359 7.ac.11.1585 3. we may also express αc as: 1 −1  αc = k0 n1 sin2 θ − sin θ2c (9.9. . The cutoff frequencies fm are in GHz. 9.13 studies the sensitivity of the Example 9.9037 25.Nit) β. 34. is found to be 4.11. and therefore. are as follows: β = k1 sin θ = k0 n1 sin θ (9. Using two.4071 5. and therefore. β = 1.8210 7. In terms of the ray picture of the propagating wave.0609o } k1 1 2. err. and β. causing the ﬁelds The related MATLAB function dguide.a.6497 10. 4. is used. n2 . For example. −1 where f is in GHz. Rm = single solution if f is in the interval 0 ≤ f < f1 .11. such as 29.6603 GHz. (The vectorial relationships are the same as those in Fig. fm = . depicted in Fig. outside the slab to be less conﬁned. n1 . f = 30.err] = dguide(f. 1.11. which gives 2R/π = 3. The refractive indices of the slab and the surrounding dielectric are n1 = 2 and n2 = 1.7603 11. of the slab.5649 rad/cm.5 cm at frequency f = 30 GHz.3.3205 As required.11. the half-length a.5100o . the angles of total internal reﬂection are quantized according to the values of the propagation wavenumber β for Solution: The solution is obtained by the MATLAB call: the various modes.9.8275o . . one. αc .11.2718 9. 51.3 TE modes and corresponding E-ﬁeld patterns.Nit). and αc = 1. 9.9105 3. M = 3. .36) TE Modes for R = 5. Nit = 3. a = 0. and 0. M (9.n1.2777 12.9808.kc. 4aNA We note that the function dguide assumes internally that c0 = 30 GHz cm. and the refractive indices n1 .1.n1. kc = 1. We note that as the mode number m increases.kc.0000 θ = asin = {77. and u. . c0 . mc0 The frequency range over which there are only four solutions is [25.6603 2 3. Determine the propagating TE modes of a dielectric solutions to small changes of the parameters f . 65.5. It has usage: The approximation error.3920 rad/cm. n2 as functions of x. a in cm.   slab of half-length a = 0.1. The electric ﬁeld patterns are also shown in the ﬁgure eters β.37) 0 0 1 2 3 4 5 6 7 −3 −2 −1 0 1 2 3 u x/a Example 9.058.029×10−4 .fc. The number of iterations is typically very small.11.n2.   n2 θc = asin = 30o n1 m u v β kc αc fm   β 0 1.   3 5. or. The resulting solutions. kc .m uses dslab to calculate the solution param.3248 5.a.fc.

see Eq. Assuming only x-dependence for the Hz component.3. resulting in the two con- where we deﬁned the asymmetry parameter δ: ditions: n2s − n2c 1 1 αc δ= (9.12.12 Asymmetric Dielectric Slab which combine to deﬁne the allowed range of β for the guided modes: The three-layer asymmetric dielectric slab waveguide shown in Fig.12.2) cos(kf a + φ) = sin(kf a + φ) tan(kf a + φ) = n2f − n2s kf αc kf ⇒ (9.12. x ≤ −a αs α2c = β2 − k20 n2c α2c − α2s = k20 (n2f − n2s )δ = k20 (n2s − n2c ) Since we assumed that μ = μ0 in all three regions. nc ≤ ns ≤ ≤ nf (9. αc are assumed to cos(kf a − φ) = sin(kf a − φ) tan(kf a − φ) = be real. On the other ⎧ hand.1 Three-layer asymmetric dielectric slab waveguide.12. we brieﬂy discuss the properties of the TE and TM propagation modes. such as air. x≥a (9. The remaining two components. 2 2 x ≤ −a The solutions. If we orient the coordinate axes as shown in the above ﬁgure. that is. where φ is a parameter to be determined. decaying exponentially in the substrate and cover. z dependence of the ﬁelds is assumed to be jβ ωμ the usual ejωt−jβt . Waveguides 9. β ≥ k0 ns . 9. To achieve propagation where the lower limit β = k0 ns deﬁnes the cutoff frequencies. can be written in the following form. |x| ≤ a satisfy the relations (we assume μ = μ0 in all three media): ⎪ ⎪ ⎪ ⎪ kf ⎪ ⎨ β k2f = k20 n2f − β2 k2f + α2s = k20 (n2f − n2s ) Hx (x)= −j H1 sin(kf a + φ)e−αc (x−a) . Hx and Ey . the continuity of Ey across the boundaries x = ±a implies the same for the Hx components. This gives in the three regions: hence. which automatically satisﬁes the continuity conditions at the two boundaries x = ±a: ⎧ ⎪ ⎨ H1 sin(kf x + φ) . x ≤ −a In this section. it must satisfy the Helmholtz equations in the three regions: (∂2x + k2f )Hz (x)= 0. Because kf . (9.12.8). 2 2 x≥a (∂x − αs )Hz (x)= 0. Above the ﬁlm is a dielectric cover or cladding nc . the corresponding transverse wavenumbers will be jαs and −jαc .12.12. we assume that the refractive indices satisfy nf > ns ≥ nc . αs . x≥a (9. it follows that β must satisfy the inequalities.12. by total internal reﬂection within the ﬁlm.3) k0 A thin dielectric ﬁlm nf of thickness 2a is deposited on a dielectric substrate ns .6) 1 1 αs Note that δ ≥ 0 since we assumed nf > ns ≥ nc . These quantities ⎪ β ⎪ ⎪ −j H1 cos(kf x + φ) . ⎪ |x| ≤ a Hz (x)= H1 sin(kf a + φ)e−αc (x−a) . kf αs kf .4) ⎪ ⎪ ⎩ Fig.1 is a typical β component in integrated optics applications [912–933]. The case of the symmetric dielectric slab of the previous section is obtained when nc = ns . Asymmetric Dielectric Slab 397 398 9.12. 9. β ≤ k0 nf . The t. the transverse wavenumber kf within the ﬁlm will be real-valued. operating frequency ω or f in Hz. and β ≥ k0 nc . Ey = −ηTE Hx ηTE = the decay constants αs and αc within the substrate and cladding must be positive so k2f β that the ﬁelds attenuate exponentially with x within both the substrate and cladding.12.1) ⎩−j H1 sin(kf a − φ)eαs (x+a) . −H1 sin(kf a − φ)eαs (x+a) . √ Let k0 = ω μ0 c0 = ω/c0 = 2πf /c0 = 2π/λ0 be the free-space wavenumber at the are obtained by applying Eq. then Hx = − ∂x Hz .13). TE modes We consider the TE modes ﬁrst. (9.5) ⎪ ⎪ α ⎪ ⎪ c ⎪ ⎪ ⎪ ⎪ β α2s = β2 − k20 n2s ⇒ k2f + α2c = k20 (n2f − n2s )(1 + δ)= k20 (n2f − n2c ) (9.9. |x| ≤ a (∂x − αc )Hz (x)= 0.

Asymmetric Dielectric Slab 399 400 9. we may For a given operating frequency f . Waveguides Since the argument of the tangent is unique up to an integral multiple of π. or. . All allowed propagating invert the two tangents as follows without loss of generality: modes must satisfy Rm ≤ R. the value of R is ﬁxed.9.12.

√ kf a + φ = arctan αc + mπ 1 1  2R − arctan δ kf mπ + arctan δ ≤ R ⇒ m≤ 2 2 π .

αs This ﬁxes the maximum mode index M to be: kf a − φ = arctan kf √ .

14) π . 2R − arctan δ which result in the characteristic equation of the slab and the solution for φ: M = ﬂoor (maximum TE mode index) (9.12.

.

M. 1 1 αs 1 αc kf a = mπ + arctan + arctan (9.7) 2 2 kf 2 kf Thus. . .12. δ = 0.11.12. 1. . .14) reduces to Eq. there are (M + 1) modes labeled by m = 0. In the symmetric case. The corresponding . (9.25) of the previous section. and (9.

.

12. 9. Eqs. (9. Eqs. w2 − v2 = R2 δ The characteristic equation (9. v. 1. . .8) 2 2 kf 2 kf 1 1  2πfm a  mπ + arctan δ where the integer m = 0. where f = c0 /λ0 . Then. kf (αc + αs ) kf (αc − αs ) For each of the M+1 propagating modes one can calculate the corresponding angle of tan(2kf a)= . M. . kf = k0 nf cos θ.12. (9.10) u = kf a .7) and the Rm = n2f − n2s ⇒ = 2 2 2πa  2 fm (9.7) is preferred for the numerical solution.11) to ﬁnd the radius Rm of the mth But if we follow the ray starting at A along the zig-zag path AC → CS → SB.12.12. 2.12.12.2 Ray propagation model.12. 1.7) and (9. . to   e−jkf x e−jβz    β  k2f  u2 β = k20 n2f − k2f ⇒ nβ = = nf − 2 = n2f − 2 2 2 (9. the propagation ray model [925]. z) is proportional.1) provide four equations in the four unknowns {β. that is.8) may also be written in the following forms: which can be written more simply as fm = f Rm /R.12.12. m = 0. αs . z + l) should be k0 k0 k0 a e−jkf x e−jβ(z+l) (9. Eq.12) Similarly.2.12. denoted by 2ψc and 2ψs . as shown in Fig. v = αs a . kf . and we obtain: will have traveled a total vertical roundtrip distance of 4a and will have suffered two 1 1  total internal reﬂection phase shifts at points C and S. the ﬁeld of the upgoing ray at the point B at (x. we introduce the dimensionless variables:  2πf a  2 a 2 R = k0 a n2f − n2s = nf − n2s = 2π nf − n2s c0 λ0 (9.13) 2 2 . 1. . the ray mode. .12.7) can be given a nice interpretation in terms of the Once these are solved for the three unknowns u. m = 0. up constant β. we set v = 0 in the characteristic √ equation (9.16) To determine the number of propagating modes and the range of the mode index m. . or kf . w = αc a Then.12. 9. We Rm = mπ + arctan δ . The form of Eq. u = Rm and w = Rm δ.12. (9.15) three equations (9.12. corresponds to the mth mode. To this end. 1 1 αc 1 αs cutoff frequencies are obtained by setting: φ= mπ + arctan − arctan (9. 2. w. .9) k2f − αc αs k2f + αc αs total internal reﬂection of the equivalent ray model by identifying kf with the transverse propagation wavenumber.12. The ﬁeld of the upgoing ray at a point A at (x. tan(2φ)= (9. . . (9. or equivalently. c0 2 nf − ns Using the trig identities tan(θ1 +θ2 )= (tan θ1 +tan θ2 )/(1−tan θ1 tan θ2 ) and tan(θ)= c0 tan(θ + mπ). αc .12.11) Fig.12. (9.7) and (9. αc }. the effective index nβ = β/k0 can be obtained from: to a constant amplitude. .12.1) can be written in the normalized forms:     1 1 v 1 w u= mπ + arctan + arctan 2 2 u 2 u u2 + v2 = R2 (9. αs .12.

Thus. 0. Universal mode curves for TE modes as given for example by Eq.8.9 e−jkf (x+4a) e2jψs e2jψc e−jβ(z+l) 0.16) and therefore the extra accumulated phase 4kf a − 2ψs − 2ψc must be equal to a multiple of 2π. the ﬁeld at point B would be 1 0.8 m=0 This must match (9. (7.9. the phase terms are exactly those appearing in Eq. that is.3).6 2 4kf a − 2ψs − 2ψc = 2mπ ⇒ kf a = mπ + ψs + ψc 2 2 2 0.5 b 3 As seen from Eq.7): 0.12.12. (7. (9.4 4 . Asymmetric Dielectric Slab 401 402 9.12.8.3). Waveguides recall that the reﬂection coefﬁcients have the form ρ = e2jψ for total internal reﬂection.7 1 1 1 1 0.

.

1. R:† ⎪ ⎪ β ⎪ ⎪ −j E1 cos(kf x + φ) . ψs = arctan kf kf kf kf 0.12. 9. the boundary conditions at x = ±a require: TM modes f c αc cos(kf a + φ) = sin(kf a + φ) tan(kf a + φ) = pc The TM modes are obtained by solving Eqs. which ranges over the standardized interval 0 ≤ b ≤ 1. and for ⎪ ⎪ β ⎩−j E1 sin(kf a − φ)eαs (x+a) . Hy = Ex . we must solve in each region: ⇒ (9.12. 10. tan ψs = ⇒ ψc = arctan .3.20) ⎪ ⎪ ⎩ −E1 sin(kf a − φ)eαs (x+a) . Then.12. Similar curves can be developed for TM modes.3 with one branch for each value of m = 0.19) Ex (x)= −j β E1 sin(kf a + φ)e−αc (x−a) .12.12. x ≤ −a the three asymmetry parameter values δ = 0.12. Ex = − ∂x Ez . (9. continuity of the tangential ﬁeld Hy because Hy = Ex /ηTM = Ex ω/β = Dx ω/β. |x| ≤ a ⎛ ⎞ ⎛ ⎞ ⎪ ⎪ k  ⎪ ⎪ f b b + δ⎠ ⎨ 2R 1 − b = mπ + arctan ⎝ ⎠ + arctan ⎝ (9.23) ⎪ ⎨ E1 sin(kf x + φ) .12. 1. .12. (9. 9. Eq.22) f s αs cos(kf a − φ) = sin(kf a − φ) tan(kf a − φ) = ps jβ 1 β kf αs kf (∂2x + k2f )Ez = 0 . x≥a (9. the Ex component is: ⎧ Then. w=R b+δ (9. . Assuming x-dependence only.21) 1−b 1−b ⎪ ⎪ αc ⎪ ⎪ ⎪ ⎪ It is depicted in Fig.11) by δ = 10 means of a universal mode curve [927] deﬁned in terms of the following scaled variable: 0 0 1 2 3 4 5 6 7 8 9 10 11 R v2 β 2 − k20 n2s b= = 2 2 2 (9.12.3 tan ψc = . ps = = 2 (9.4): f n2f f n2f ⎧ pc = = 2.1 for a concrete placement ﬁeld Dx = Ex across the interfaces at x = ±a. .10) in each region and applying the kf αc kf boundary conditions. 0.3 Universal mode curves. See Example 9. . αc αs αc αs 0. αs A vertical line drawn at each value of R determines the values of b for the propagating The boundary conditions require the continuity of the normal component of dis- modes. x≥a (9.12.12.12. ⎪ |x| ≤ a c nc s ns Ez (x)= E1 sin(kf a + φ)e−αc (x−a) . Thus.2 δ=0 A similar interpretation can be given for the TM modes. .1 δ=1 It is common in the literature to represent the characteristic equation (9. (9. which is equivalent to the example that includes both TE and TM modes. so that    u = R 1 − b. 2.17) R2 k0 (nf − ns ) Fig.12.11) takes the universal form in terms of the variables b. x ≤ −a †R is usually denoted by the variable V.18) where φ is a parameter to be determined. v = R b. ηTM = k2f ηTM ω where we deﬁned the ratios: The solution for Ez (x) is given by a similar expression as Eq.

m = 0. .12. Asymmetric Dielectric Slab 403 404 9. 1. . . Waveguides Inverting the tangents we obtain: which can be written more simply as fm = f Rm /R. . M. . where f = c0 /λ0 .9.

The corresponding angles of total internal reﬂection in the equivalent ray model are αc obtained by solving kf = k0 nf cos θ. kf a + φ = arctan pc + mπ kf Because pc > 1. we observe that the maximum mode index M and the cutoff fre- .

quencies fm will satisfy the following inequalities for the TE and TM cases: αs kf a − φ = arctan ps kf MTM ≤ MTE .12.30) These give the characteristic equation of the slab and φ: .TE ≤ fm.TM (9. fm.

.

(9. A simple and effective iterative method for solving such char- . Numerical Solutions 1 1 αs 1 αc kf a = mπ + arctan ps + arctan pc (9. The TE case is also included 2 2 kf 2 kf by setting ps = pc = 1.12.27).24) Next we look at the numerical solutions of Eqs.12.

.

. acteristic equations was given in Ref.25) denote the right-hand side of Eq. The modiﬁed iteration has the form: un+1 = rF(un )+(1 − r)un The number of propagating modes and the range√ of the mode index m.26) The problem then becomes that of ﬁnding the ﬁxed-point solutions u = F(u). until two successive un values become closer √ 1 1  2R − arctan pc δ to each other than some speciﬁed error tolerance. . w. By replacing v. (9.31) The allowed propagating modes must satisfy Rm ≤ R. and proceeds iteratively. 1. w0 = R δ (9. 2. 2. .12.12. tan(2φ)= (9.  2 2 u0 = R . v0 = 0 . . . are again determined by setting v = 0. 1. (9. the iteration starts with the initial values: 1 1  Rm = mπ + arctan pc δ . v. . or. and w = Rm δ: Explicitly. we can write: 1 1 v 1 w F(u)= mπ + arctan ps + arctan pc 2 2 u 2 u kf (ps αs + pc αs ) kf (pc αc − ps αs ) tan(2kf a)= . but fails in 2 2 2 (9. n = 0. The k2f − ps pc αs αc k2f + ps pc αs αc method suggested in Ref.27): 2 2 kf 2 kf     and. .9). let F(u) 1 1 αc 1 αs φ= mπ + arctan pc − arctan ps (9. [963] is to use the iteration: In terms of the normalized variables u. 1.12. . R.12. u = Rm . such as tol = 10−10 : mπ + arctan pc δ ≤ R ⇒ m≤ 2 2 π      1 1 vn 1 wn which ﬁxes the maximum mode index M to be: un+1 = r mπ + arctan ps + arctan pc + (1 − r)un 2 2 un 2 un √ . 2. [963]. .12.12. 1 1 v 1 w u = mπ + arctan ps + arctan pc 2 2 u 2 u initialized at u0 = R. m = 0. we have:     un+1 = F(un ) . as in Eq. w in terms of u. say tol. for n = 0.27) others. . This simple iteration does converges in many cases. We have found that a simple modiﬁcation that involves the introduction of a u +v =R “relaxation” parameter r such that 0 ≤ r ≤ 1. enables the convergence of even the most w2 − v2 = R2 δ difﬁcult cases.

M. 2R − arctan pc δ if |un+1 − un | < tol.Nit] = dguide3(a. .mode.12.ns. 1.tol).kf. The corresponding cutoff  2 wn+1 = R2 δ − vn+ 1 frequencies are obtained by setting: 1 1  The MATLAB function dguide3.r.as. .ac.nf.32) π vn+1 = R2 − u2n+1 The (M + 1) modes are again labeled by m = 0.28)  (9.12. .nc.fm.m implements the method and has usage: 2πfm a  mπ + arctan pc δ Rm = 2 nf − ns 2 ⇒ = 2 2 2πa  2 fm (9.29) [be. then exit. . c0 2 nf − ns c0 . else continue M = ﬂoor (maximum TM mode index) (9.12.

7880 2.45.12.0756 1.154191 1. % initialize iteration variables u.7599 3.865244 2. Waveguides where the inputs and outputs have the following meanings: Example 9.tol).51 if strcmpi(mode.v. % asymmetry parameter 3 2.6364 1.0935 3. % TE modes [be.0527 0. nf .0742 0.nc.1846 0./u)/2) + (1-r)*u. (9.atan(pc*sqrt(d)))/pi).5) tol = error tolerance (default tol = 10−10 ) nf=3. % number of iterations for TE and Nit = 66 for TM.kf.0000 2.0735 0.nf.5.2860 0. % TM modes Internally.as.2669 0.0247 78. % default value r=0.’TE’) % mode can also be entered in lower case ps = 1.5 did not work in this case and caused the TE iteration to diverge and the smaller value r = 0.20 4 – – – – 1. % u. an oxide a = half-width of slab in units of the free-space wavelength λ0 substrate with ns = 1.302025 2./u)/2 + atan(pc*w.28) and calculates β.15 d = (ns^2-nc^2)/(nf^2-ns^2). r = relaxation parameter (default r = 0. We note that all TIR angles are greater than the critical angles computed by: u = unew.fm.4745 0.1137 3. αs .12.46 end 1 3.8894 3. % half-thickness in units of la0 kf = transverse wavenumbers inside slab in units of k0 r=0.ns.1).434746 0.w v = zeros(M+1.2679 67. The default value of the w = R*sqrt(d)*ones(M+1.92 k0 = 2*pi.9616 1.47 (TE modes) R = k0*a * sqrt(nf^2-ns^2).451972 3.ns. the function determines M from Eq. αc .v) circle radius. % vector of mode indices 3 1.0760 – u = R*ones(M+1. % degrees unew = r*(m*pi/2 + atan(ps*v. . kf . αc = decay wavenumbers in substrate and cladding in units of k0 tol=1e-10. and cladding (nf > ns >= nc ) following MATLAB code generates both the TE and TM modes.55.’te’. nc=1.7545 41.12. Asymmetric Dielectric Slab 405 406 9.3 was chosen. The number of iterations were Nit = 57 Nit = 1.5894 49.1733 1.r.8327 32. ns=1.ac. if norm(unew-u) <= tol.3461 64. pc = 1. we consider the same benchmark example dis- cussed in [963] consisting of a silicon ﬁlm of thickness of 1 μm with nf = 3. fm as (M + 1)-dimensional column vectors representing the M + 1 modes.1).1028 77.5.6727 3. The TIR angles were computed by the following command: % while loop repeats till convergence while 1 thm = acos(kf/nf)*180/pi.1).8011 2. ﬁlm.9978 24. % (u.5112 55. relaxation parameter r = 0.2407 2. fm = cutoff frequencies in units of f = c0 /λ0 Nit = number of iterations it takes to converge to within tol [be.4794 2.1: For comparison purposes.416507 0. The ns .69 m = (0:M)’.w are (M+1)x1 vectors The β/k0 column is the effective phase index of the modes. o To clarify the computations.668932 2.kf.32 (TM modes) M = floor((2*R . with operating wavelength λ0 = 1.3413 2.r. % units of microns β = propagation wavenumbers in units of k0 = 2π/λ0 a = a/la0.nf.12. the essential part of the code is listed below: m β/k0 kf /k0 αs /k0 αc /k0 fm /f θm 0 3.872310 2.232789 1.3.5745 0.14) or (9.’tm’.9.Nit] = dguide3(a. break.ac.9915 0. 0 3. % oxide substrate | silicon film | air cover la0 = 1. a = 0. note k0*a = 2*pi*(a/la0) 2 2.5 fails to converge for the TE modes αs .2642 2.5. nc = refractive indices of substrate. end Nit = Nit + 1.55 μm.6926 0.nc.Nit] = dguide3(a. o else m β/k0 kf /k0 αs /k0 αc /k0 fm /f θm ps = nf^2/ns^2.5169 2. pc = nf^2/nc^2. and air cover.13 4 1.45.as.fm.v. % highest mode index 2 2. % la0 = 2*pi/k0 = 1 in the assumed units 1 3.tol). with the numerical outputs mode = ’TE’ or ’TM’ listed in the tables below.

.

that is. i. % kf in units of k0. % angles of total internal reflection .kf.TE < f < f4. break.60o w = sqrt(R^2*d + v. and therefore. near the operating frequency f = c0 /λ0 . and its TIR angle. but f4.*kf). kf/k0 = u/(k0*a) as = v/(k0*a). it so happened that in this example f falls in the be = sqrt(nf^2 . % beta in units of k0. v = sqrt(R^2 . % cutoff frequencies in units of f = c0/la0 of the characteristic equations. nf nf if Nit>1000. With reference to the inequality (9. ac = w/(k0*a). i. thm = acos(kf/nf).^2). the ﬁfth TM mode f4.e.u. % cutoff radius for m-th mode The convergence can be veriﬁed for all modes at once by computing the vector error norm fm = Rm/R. θc = arcsin = 16..12.30). Rm = m*pi/2 + atan(pc*sqrt(d))/2.TE is..47o .^2).TM is not excited.TM . very close to the critical angle. The ﬁfth TE mode is very weakly bound to the substrate side because its decay parameter αs is very small. end % safeguard against possible divergence end There are ﬁve TE modes and four TM ones. beta/k0 range f4.e. its cutoff frequency is very kf = u/(k0*a). ns nc θs = arcsin = 24.

atan(ps*as. we have Err = 2. 9. Determine the cutoff wavelength angles are in this example: λc of this guide. where λg = 2π/β and λ is the free-space wavelength. Fig. 9. (9. see e. TM (dashed lines/open circles).9. What are the dimensions of the guide in cm? 1 9. nf = 3. What are the dimensions a.3 An air-ﬁlled rectangular waveguide is used to transfer power to a radar antenna. (b) it can transmit the maximum possible power.2 maximum power and have the widest possible bandwidth? What is the cutoff frequency of the guide in GHz and the operating bandwidth? 0.55 μm. Determine this operating frequency in GHz and calculate the maximum TE.4 It is desired to design an air-ﬁlled rectangular waveguide operating at 5 GHz. b if the transmitted power is required to be maximum? 2 0. The op- 0. but we used the default value r = 0. b of the guide (in cm) if it is also required to carry 0.5 b b.12. u = kf a = (kf /k0 )k0 a = (kf /k0 )2π(a/λ0 ). Calculate the attenuation verges in 8 and 10 iterations respectively for the TE and TM modes.6 Determine the four lowest modes that can propagate in a WR-159 and a WR-90 waveguide.256.53 · 10−10 for TM. and Err = 2.9 erating frequency is 3 GHz and must lie exactly halfway between the cutoff frequencies of 0. Only one (M = 0) TE constant in dB/m due to wall ohmic losses.5λc ≤ . We note that the quantity kf*2*pi*a represents the power in Watts that can be transmitted without causing dielectric breakdown of air. by a safety m=0 margin of 3. 9. 9. 9.5 Show the following relationship between guide wavelength and group velocity in an arbitrary 0 air-ﬁlled waveguide: vg λg = cλ .5 cm×3 cm waveguide is operated at a frequency that lies in the middle of its This error is of the order of the assumed tolerance.4 What is that maximum power in MW? 3 0. more difﬁcult.4 displays the TE and TM solutions on the universal mode curves.3. nc = 1. 0 1 2 3 4 5 6 7 8 9 10 R Moreover. ns = 3. Assume copper walls.1 An air-ﬁlled 1. whose group velocity is 0.1 9.   TE10 mode with the widest possible bandwidth. Finally.5.g. The critical TIR 9. b for such a waveguide. Show that the operating wavelength band of the lowest mode is 0. The variable u in our units. m = (0:M)’.12./kf)/2 . Calculate the maximum power that The same MATLAB code applies here. Calculate the cutoff frequencies (in GHz) and cutoff wavelengths (in cm) of these modes. which con. and one TM mode are supported with parameters given in the table below.4 Universal mode curves. b such that b ≤ a/2. show that the λ and λg are related to the cutoff wavelength λc by: 1 1 1 Fig.8c.7 An air-ﬁlled WR-90 waveguide is operated at 9 GHz. = + λ2 λ2g λ2c 9. and (c) the operating frequency is 12 GHz and it lies in the middle of the operating band. TE (solid lines/ﬁlled circles). What are the dimensions a. The maximum electric ﬁeld within the guide may not exceed. Asymmetric Dielectric Slab 407 408 9. 9.8 these two modes.7 a.12. the breakdown ﬁeld of air 3 MV/m. Example 9. if the transmitted power 0.94 · 10−10 for TE10 mode band.atan(pc*ac. 0. The guide must meet the following speciﬁcations: The two lowest modes are TE10 and TE20 .6 1 is required to be 1 MW. Determine the smallest dimensions a. Waveguides M = length(be)-1.12. can be transmitted without causing dielectric breakdown of air.m*pi/2 .5 μm.3 4 9.19).2 It is desired to design an air-ﬁlled rectangular waveguide such that (a) it operates only in the Eq.13 Problems Err = norm(kf*2*pi*a . 0. a = 0. indeed. indeed.8 A rectangular waveguide has sides a.2: A second./kf)/2). example from [963] has the parameters λ0 = 1. dielectric strength of air is 3 MV/m.12.

.

265996 0. θs = arcsin = 80. What are the dimensions of the guide? 9. y) component satisﬁes: −11 The computational errors in the characteristic equations were Err = 1. Assume perfectly conducting walls. ∇T Ez |2 dS = k2c |∇ |Ez |2 dS . o mode β/k0 kf /k0 αs /k0 αc /k0 fm /f θm 9. show that the allowed range of the guide wavelength is λg ≥ λc / 3.52 · 10−11 for TM. TM 3.263384 0. Show that the Ez (x. θc = arcsin = 17.10 Computer Experiment: WR-159 Waveguide.6427 81.77 section.4902 0.11 A TM mode is propagated along a hollow metallic waveguide of arbitrary but uniform cross TE 3.63 · 10 for TE.8.63o .2553 3.4725 0. Moreover.13.2.1) and Err = 1.   S S .64o nf nf 9.9 The TE10 mode operating bandwidth of an air-ﬁlled waveguide is required to be 4–7 GHz.7142 81. Reproduce the two graphs of Fig. (kc = cutoff wavenumber) (9.2194 3.1064 0.46 a.1091 0. 9. √ ns nc λ ≤ λc .

1 and PT = |H1 |2 3 + |H1 |2 2k c 2α3c 9. 9. representing a where H1 is the amplitude deﬁned in Eq. from now more exact value).13 Show that if the speed of light c0 is slightly changed to c = c0 + Δc0 (e. write a program that reproduces all the results and graphs of Examples 9.11. the 1 + αc a c0 . B. Using the above result.11.13. αc change into: on set. Show that the electric and magnetic energy densities are given as follows.    b. Problems 409 410 9. Hint: Use the identity: ∇ T · (A∇ the slab: 9. show that the energy velocity is equal to the group velocity. and the second. where again. Waveguides b.12 Computer Experiment: Dielectric Slab Waveguide. Without loss of generality. a. Show that PT is given as the sum of the following two terms. then the solutions of Eq.29) for kc . (9. for scalar A. (9. H1 = 1. and the second. Using the MATLAB functions dslab and   βωμ0 akc + sin(akc )cos(akc ) βωμ0 sin2 (akc ) dguide.11.9.g. the power ﬂowing outside ∇T B)= ∇ T A · ∇ T B + A ∇ 2T B.2.5). kc Δc0 kc + Δkc = kc − the ﬁrst terms represent the energy contained within the slab. where the ﬁrst one represents the power ﬂowing within the slab.11.

which affect the solution through the value of R = aω0 c0−1 n21 − n22 . (9. Using the above expressions and Eq.1. c0 . consider the sensitivity of the solutions of Eq. calculate the corrected values when c0 = 30 and c = 29. μ0 (β2 + k2c )akc + μ0 (β2 − k2c )sin(akc )cos(akc ) μ0 (β2 + α2 )sin2 (akc ) Wm = + 4k3c 4α3c More generally. c. u + Δu = u + 1+v R . show the equality A small change in one or all of the parameters will induce a small change R → R + ΔR. Show that the solutions are changed to We = Wm    u ΔR Thus. (9. Compare with the values obtained if c0 is replaced by c inside the function dguide. n2 .9792458 GHz cm. n1 .11. the total energy density is W = We + Wm = 2We .  energy outside the slab: k2c a Δc0 αc + Δαc = αc − αc +   1 + αc a c0 μ0 (β2 + k2c ) akc + sin(akc )cos(akc ) μ0 (β2 − α2c )sin2 (akc ) We = 3 + 4kc 4α3c For Example 9. a.11.11.29) to any of the  param- eters ω0 .13).

1. where of Example 9. show that the changes due to a change a → a + Δa of the slab thickness are given by.12.b. show ﬁnally that k2c αc + Δαc = αc + Δa 1 + αc a ωβ(1 + aαc ) ωβ(1 + aαc ) ven = vgr = = 2 9.3) and (9.1.12.2) W dβ where PT is the time-averaged power transmitted in the z-direction through the cross- sectional area deﬁned by 0 ≤ y ≤ 1 and −∞ < x < ∞.12. and W is the energy contained in the volume deﬁned by the above area and unit-z-length. demonstrate that the energy transport (β2 + k2c )aαc + β2 ω 0 μ0 n21 aαc + β2 velocity is equal to the group velocity. show that From these results. dβ (β2 + kc2 )aαc + β2 ωβ = kc αc dω 1 + aαc kc + Δkc = kc − Δa 1 + αc a f. W = |Ey (x)|2 + |Hx (x)|2 + μ|Hz (x)|2 dx 2ηTE −∞ 4 −∞ Because of the substantial amount of algebra involved.11. ∞ ∞ 1 1   PT = |Ey (x)|2 dx .  d.14 For the dielectric slab waveguide shown in Fig.13.13). Reproduce all the results and Fig. show that u2 ΔR v + Δv = v + v + ωβμ0 (1 + aαc ) 1+v R PT = 2αc k2c In particular. ven = . Speciﬁcally. From parts (a.15 Computer Experiment.11.11. (9.13) with respect to ω.3)–(9. 9..4 deﬁned by Eqs. and show that ven = vgr . show that  2  μ0 (β + k2c )aαc + β2 W = ΔR Δa Δω0 Δc0 2n1 Δn1 − 2n2 Δn2 2αc k2c = + − + R a ω0 c0 n21 − n22 e. vgr = (9. make a separate graph of Fig. Combining the results of parts (e. Moreover. 9. i.11.e. break the calculation as follows: .4 that zooms into the PT dω neighborhood of the ﬁfth TE mode to make sure that it is indeed below cutoff.11. for simultaneous changes in all of the parameters. (9.f). 9. By differentiating Eqs.c). Asymmetric Slab Waveguide. consider the case of even TE modes 9.