Professional Documents
Culture Documents
( )2
2L R−1 2D
ls = = . (3)
R−1 2R R−1
Equations (1)–(3) indicate that the geometry of the resonator is unambiguously determined fixing two of the
three parameters D, L, and R.
Figure 2. (Transmissivity) spectrum of a Cantor dielectric resonator at the stage of growth M = 2. D = 15 mm, R = 2.0,
𝜖d = 3.78 1 − j2 ⋅ 10−4 , 𝜖a = 1.
Figure 3. Longitudinal distribution of the normalized field amplitude |E|∕ ||E0 || at the center of the waveguide broad wall
of the Cantor dielectric resonator with the parameters as in Figure 2. Resonance frequency f1 = 2.26 GHz.
When inserted in metallic rectangular waveguide, a Cantor resonator exhibits a transmissivity spectrum
with forbidden bands and narrow transmission peaks in the single-mode frequency range of the waveguide
[Chiadini et al., 2009]. At the frequencies corresponding to the peaks a resonance phenomenon occurs,
yielding energy concentration and field localization in the central air gap [Chiadini et al., 2014].
The transmissivity T of the resonator can be computed using the well known transmission matrix method
[Collin, 1992]. The waveguide resonator can be considered as the cascade of seven alternate dielectric- and
air-filled waveguide sections. Numbering the sections from the leftmost one by an index m, the dielectric-filled
waveguide sections are characterized by odd index values while the air-filled ones by even index values.
Because the kind of discontinuity between sections does not allow higher order modes to be excited, each
section can be modeled by the transmission matrix
[ ( ) ( )]
cos km dm jZ0m sin km dm
P = ( ) ( ) , (4)
−1
m jZ0m sin km dm cos km dm
where km and Z0m are the propagation constant and the wave impedance of the fundamental TE10 mode,
respectively, and dm is the length of the waveguide section (d1 = d3 = d5 = d7 = D, d2 = d6 = ls , d4 = lc ). The
transmission matrix P of the waveguide resonator reads
P =P ⋅P ⋅…⋅P . (5)
1 2 7
Figure 4. Longitudinal distribution of the normalized field amplitude |E|∕ ||E0 || at the center of the waveguide broad wall
of the Cantor dielectric resonator with the parameters as in Figure 2. Resonance frequency f2 = 2.99 GHz.
Figure 5. Transmissivity at 2.45 GHz of the Cantor dielectric resonator with same parameters as in Figure 2 but with a
varying fractal parameter R.
| |2
| 2Z0 |
T = |( ) ( )| (6)
| Z P +P +Z Z P +P |
| 0 11 12 0 0 21 22 |
where Z0 is the wave impedance of the TE10 mode in the air-filled waveguide.
( )
As an example, consider a resonator in fused silica with relative permittivity 𝜖d = 3.78 1 − j2 ⋅ 10−4 ,
D = 15 mm and R = 2 inserted in a WR340 rectangular waveguide. The transmissivity spectrum in the stan-
dard single-mode frequency band of the waveguide (2.20–3.30 GHz) is plotted in Figure 2. Two resonance
peaks can be observed at frequencies f1 = 2.26 GHz and f2 = 2.99 GHz, respectively. At these frequencies con-
structive interference between multiple reflections from dielectric interfaces gives rise to a standing wave in
the structure with strong field localization in the central air gap. The longitudinal profile of the field amplitude
|E| at the center of the waveguide broad wall, normalized to the amplitude of the incident field |E0 | on the
first dielectric interface, is plotted in Figures 3 and 4 at the frequencies f1 and f2 , respectively. These profiles
were computed using the transmission line model which gives the standing wave solution in each waveguide
section
( ) Z ( )
Em (z) = Em (0) cos km z − j 0m Em (0) sin km z , (7)
Zm (0)
Figure 6. Longitudinal distribution of the normalized field amplitude |E|∕ ||E0 || at the center of the waveguide broad wall
of the Cantor dielectric resonator with the parameters as in Figure 2. Resonance frequency 2.45 GHz, fractal parameter
R = 1.5119. The gray vertical band features the central air gap of the resonator.
Figure 7. Longitudinal distribution of the normalized field amplitude |E|∕ ||E0 || at the center of the waveguide broad wall
of the Cantor dielectric resonator with the parameters as in Figure 2. Resonance frequency 2.45 GHz, fractal parameter
R = 1.6058. The gray vertical band features the central air gap of the resonator.
where zm ∈ [0, dm ] is the longitudinal abscissa and Zm (0) is the input impedance of the mth line section. The
input impedance Zm (0) can be iteratively calculated starting from that of the rightmost line section, using
the recursive formula
( )
Zm+1 (0) + jZ0m tan km dm
Zm (0) = Z0m ( ) . (8)
Z0m + jZm+1 (0) tan km dm
with m = 7, 6, … , 1 and Z8 (0) = Z0 . Finally, the field amplitude profile can be computed by means of
equation (7) starting from the leftmost section and enforcing the boundary conditions
E1 (0) = (1 + Γ) E0 (9)
and
( )
Em (0) = Em−1 dm−1 , m = 2, … , 7 (10)
Figure 8. Longitudinal distribution of the normalized field amplitude |E|∕ ||E0 || at the center of the waveguide broad wall
of the Cantor dielectric resonator with the parameters as in Figure 2. Resonance frequency 2.45 GHz, fractal parameter
R = 1.7701. The gray vertical band features the central air gap of the resonator.
Figure 9. Longitudinal distribution of the normalized field amplitude |E|∕ ||E0 || at the center of the waveguide broad wall
in the central gap of the Cantor dielectric resonator with the parameters as in Figure 2. Resonance frequency 2.45 GHz,
fractal parameter R = 2.1499. The gray vertical band features the central air gap of the resonator.
The localization factor 𝜂 is defined as the ratio between the maximum field amplitude |EM | in the central air
gap and the amplitude of the incident field |E0 |
|EM |
𝜂= . (12)
|E0 |
In order to be useful for microwave heating the resonator must be tuned at a frequency used for industrial
microwave applications. In the following the frequency f0 = 2.45 GHz, which is widely used for microwave
heating, is considered. Consequently, a suitable resonator geometry has to be determined in order the res-
onator to exhibit resonance at the desired frequency. So, in this example, the problem of designing a fused
silica resonator exhibiting resonance at 2.45 GHz has to be faced. To this end, the transmissivity at 2.45 GHz
was computed, keeping fixed the thickness D = 15 mm of the dielectric layers and varying the fractal param-
eter R. Figure 5 shows that there are four values (R = 1.5119, R = 1.6058, R = 1.7701, and R = 2.1499) in the
range R ∈ [1, 3], which give structures resonating at 2.45 GHz. The normalized field amplitude profile in these
structures is reported in Figures 6– 9, where the gray vertical bands feature the central air gap. It can be seen
that the field distribution exhibits one or more lobes depending on the length of the central air gap, while the
localization factor is 𝜂 ≃ 5 for the one lobe configuration and 𝜂 ≃ 6 for the other configurations.
However, the localization factor depends on the relative permittivity of the dielectric layer constituting the
resonator. To investigate this property we considered other two resonators, realized with dielectric layers of
( ) ( )
plexiglas (𝜖d = 2.62 1 − j9 ⋅ 10−3 ) and alumina (𝜖d = 9.9 1 − j10−4 ), respectively. As for the fused silica
resonator, the tuning procedure was performed computing the transmissivity at 2.45 GHz with D = 15 mm
and R ranging from 1.5 to 3.0. For the plexiglas resonator four transmissivity peaks were found at R = 1.5054,
R = 1.5936, R = 1.7416, and R = 2.0489. Figure 10 shows the longitudinal profile of the normalized field
amplitude for R = 2.0489, corresponding to a structure with a single field lobe in the central gap, exhibiting
a localization factor 𝜂 ≃ 2.5. The field profiles for the other values of R show multiple lobes in the central gap
of the resonator with localization factors 3.0 ≲ 𝜂 ≲ 3.5. For the alumina resonator the transmissivity peaks
Figure 12. Pictures of the prototype of the WR340 waveguide applicator with a two-stage Cantor resonator inserted
inside; (a) plexiglas resonator, (b) alumina resonator, and (c) applicator in the configuration used for measuring the
longitudinal field amplitude profile.
lie at R = 1.5282, R = 1.6319, R = 1.8245, and R = 2.3525. Figure 11 shows the field profile of the one lobe
resonator (R = 2.3525) for which the localization factor is 𝜂 ≃ 9.6. Also, in this case the field profiles for the
other values of R show multiple lobes and localization factors 6.5 ≲ 𝜂 ≲ 9.5. Results show that given M = 2
and D = 15 mm, the higher the real part of the relative permittivity the higher the localization factor. Further
simulations performed with M = 2 and different values of D confirmed this behavior.
The effects of inserting a lossy dielectric load in the central air gap of the resonator have been described in
a recent paper [Chiadini et al., 2014]. Both cylindrical vertical loads and longitudinal planar loads placed at
the center of the waveguide broad wall in the middle of the central air gap have been considered. The load
perturbs the resonator causing a detuning which makes the field localization do not occur anymore at the
resonance frequencies of the unperturbed resonator (the structure without the load). As a consequence, the
geometry of the resonator must be modified to retune the resonance frequency of the loaded structure at
f0 = 2.45 GHz. As for the unperturbed resonator, the retuning procedure consists in computing the trans-
missivity of the perturbed resonator by varying the parameter R. It has been shown that for both loads, the
retuning can be successfully performed, obtaining, at the frequency f0 , field localization just inside the load
with amplitude noticeably greater than the incident field. Numerical simulations showed that the localiza-
tion factor is lower than that one of unperturbed structure. Nevertheless, the field localization inside the
load should allow more efficient heating processes of low loss materials to be performed if their dielectric
properties do not dramatically change during the treatment [Chiadini et al., 2014].
Figure 14. Longitudinal distribution at the center of the waveguide broad wall of the normalized field amplitude
|E|∕ ||E0 || of a two-stage Cantor plexiglas resonator with D = 15 mm and R = 2.4063. Resonance frequency 2.45 GHz.
Solid line: calculated distribution. Stars: measured values.
Figure 15. Longitudinal distribution at the center of the waveguide broad wall of the normalized field amplitude
|E|∕ ||E0 || of a two-stage Cantor plexiglas resonator with D = 20 mm and R = 2.4109. Resonance frequency 2.45 GHz.
Solid line: calculated distribution. Stars: measured values.
of the waveguide. The fractal parameter R determining the length of the air gaps between the dielectric blocks
was evaluated by means of the tuning procedure described in the previous section.
The prototype was used to evaluate the field localization in the central gap of the unperturbed resonators.
To this end, the upper largest waveguide wall was replaced by a plate having a low-perturbing longitudinal
narrow slot (370 mm × 5 mm) in its middle, equipped with a centimeter ruler and a movable carriage mount-
ing a small probe antenna and the diode detector. In this configuration the applicator looks as a common
waveguide slotted line (Figure 12c).
Blocks of three different thicknesses (D = 15 mm, D = 20 mm, and D = 30 mm) were used for the plexiglas
resonator. For D = 15 mm the tuning procedure gave R = 2.0489 corresponding to a resonator with central and
lateral air gaps of length 𝓁c = 112.2 mm and 𝓁s = 28.7 mm, respectively, and overall length L = 229.6 mm. For
D = 20 mm the parameters of the resonator were R = 2.4109 , 𝓁c = 96.8 mm, 𝓁s = 28.3 mm and L = 233.4 mm.
Finally, for D = 30 mm calculation gave R = 3.3412 , 𝓁c = 73.2 mm, 𝓁s = 25.6 mm and L = 244.4 mm.
For the alumina resonator blocks with thicknesses D = 15 mm were used. In this case the tuning procedure
gave R = 2.3525 corresponding to a resonator with central and lateral air gaps of length 𝓁c = 77.2 mm and
𝓁s = 22.2 mm, respectively, and overall length L = 181.5 mm.
Figure 13 shows the schematic of the experimental setup used for measurements. It includes a microwave
generator (Hameg HM8135 RF Synthesizer) generating a microwave 2.45 GHz signal with the amplitude mod-
ulated by an internal 1 kHz square wave, a N(f )-WR340 coaxial-waveguide adapter (Elenova Telecomunicazioni
model WNT002), a WR340 directional coupler (Connecticut Microwave Corporation model 366004) for moni-
toring the incident power, a three-port WR340 isolator (CoberMuegge LLC WG340IS3), the Cantor applicator,
Figure 16. Longitudinal distribution at the center of the waveguide broad wall of the normalized field amplitude
|E|∕ ||E0 || of a two-stage Cantor plexiglas resonator with D = 30 mm and R = 3.3412. Resonance frequency 2.45 GHz.
Solid line: calculated distribution. Stars: measured values.
Figure 17. Longitudinal distribution at the center of the waveguide broad wall of the normalized field amplitude
|E|∕ ||E0 || of a two-stage Cantor alumina resonator with D = 15 mm and R = 2.3525. Resonance frequency 2.45 GHz. Solid
line: calculated distribution. Stars: measured values.
a WR340-WR284 waveguide transition, and a WR284 matched load (HP model S910A). Power measurements
were carried out with a microwave power meter (Anritsu ML2438A) and RF power sensor (Anritsu 2472D).
The field localization in the applicator was analyzed by detecting the longitudinal field distribution by using
a detector probe (HP 447B) and a standing wave ratio (SWR) meter (HP 415E). In order to measure the nor-
malizing incident field |E0 |, a first measure is carried out without inserting the dielectric resonator into the
applicator. In this case, being the applicator terminated on a matched load, no standing wave occurs in the
waveguide and the probe detects a signal whose level, constant along the slotted waveguide, is proportional
to the square of the incident field. Successively, the dielectric resonator was inserted into the waveguide, the
probe was placed at the beginning of the central air gap and was moved toward the matched load for mea-
suring the field profile |E(z)| along the waveguide. During the field profile acquisition the incident power was
kept constant at the value detected when the normalizing incident field was measured.
Figures 14–16 show the longitudinal normalized field profile |E|∕|E0 | at the center of the waveguide largest
wall in the central gap of plexiglas resonators with dielectric blocks of thickness 15 mm, 20 mm, and 30 mm,
respectively. For comparison, the normalized profiles calculated by using the transmission line model in each
layer are also plotted. Figure 17 shows the measured field profile in the alumina resonator together with the
computed one. As can be seen, the agreement between calculated profiles and experimental data is excellent
for plexiglas resonators and good for the alumina resonator. This agreement shows the robustness of the
design procedure with respect to errors affecting the permittivity and the thickness of the dielectric blocks,
and their position in the waveguide.
4. Conclusions
In order to show the usefulness of dielectric resonators in microwave heating, we considered Cantor res-
onators since they offer performance better than the perturbed periodic ones. A microwave applicator was
realized by inserting a Cantor dielectric resonator in a metallic rectangular waveguide. The architecture of this
resonator was determined according to the triadic Cantor fractal construction. In particular, the Cantor frac-
tal construction scheme at the second stage of growth was considered, giving a very simple resonator which
consists of four dielectric layers of equal thickness separated by three air gaps with different lengths. At res-
onance frequencies the field localization phenomenon occurs, making the field reach an amplitude at the
resonator center significantly greater than the amplitude of the incident field. Analysis of field localization for
different low loss dielectric materials constituting the resonator showed that the higher the real part of the
material relative permittivity the higher the localization factor. A prototype of the Cantor applicator has been
realized by inserting a two-stage Cantor resonator in a WR340 aluminum waveguide. Plexiglas and alumina
resonators have been tested. Measurements of the field amplitude longitudinal profiles at the center of the
waveguide broad wall show a good agreement with profiles obtained by numerical simulations. These results
indicate that the proposed applicator can be easily realized, its design being very robust with respect to errors
affecting both the permittivity and the thickness of the dielectric blocks of the resonator and their position in
the waveguide.
Acknowledgments References
The data used in producing this
manuscript are available from author Chiadini, F., V. Fiumara, I. M. Pinto, and A. Scaglione (2003), Self-scaling properties of the reflection coefficient of Cantor prefractal
upon request. Readers may contact multilayers, Microwave Opt. Technol. Lett., 37, 339–343.
the first author Francesco Chiadini at Chiadini, F., V. Fiumara, I. Gallina, I. M. Pinto, and A. Scaglione (2008a), Filtering properties of defect-bearing periodic and triadic Cantor
fchiadini@unisa.it. multilayers, Opt. Commun., 281, 633–639.
Chiadini, F., V. Fiumara, I. Gallina, and A. Scaglione (2008b), Field localization inside a lossy dielectric slab by means of Cantor dielectric
multilayers, J. Appl. Phys., 103, 063104.
Chiadini, F., V. Fiumara, I. Gallina, S. T. Johnson, and A. Scaglione (2009), Cantor dielectric filters in rectangular waveguides, Electromagnetics,
29, 575–585.
Chiadini, F., A. Diovisalvi, V. Fiumara, and A. Scaglione (2014), A Cantor multilayer traveling wave applicator for microwave heating:
Numerical analysis and design, J. Appl. Phys., 116, 214503.
Collin, R. E. (1992), Foundations for Microwave Engineering, 2nd ed., McGraw-Hill, New York.
Jaggard, D. L., A. D. Jaggard, and P. V. Frangos (2000), Fractal electrodynamics: Surfaces and superlattices, in Frontiers in Electromagnetics,
edited by D. H. Werner and R. Mittra, pp. 1–47, IEEE Press, New York.
Lakhtakia, A., K. E. Weaver, and A. Verma (2002), Transmission through Cantor filters revisited, Optik, 113, 510–512.
Metaxas, A. C., and R. J. Meredith (1993), Industrial Microwave Heating, Peter Peregrinus, London.