вход по аккаунту


Design, characterization, and fabrication of microwave and millimeter-wave antennas

код для вставкиСкачать
By Timothy H. Hwang
A thesis submitted to the Faculty of the University of
Delaware in partial fulfillment of the requirements for the
degree of Masters of Science in Electrical Engineering
Fall 2006
Copyright 2006 Timothy H. Hwang
All Rights Reserved
UMI Number: 1440627
UMI Microform 1440627
Copyright 2007 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, MI 48106-1346
Timothy H. Hwang
Dr. Dennis W. Prather, Ph.D.
Professor in charge of thesis on behalf of the
Advisory Committee
Dr. Gonzalo R. Arce, Ph.D.
Electrical Engineering Department Chair
Eric Kaler, Ph.D.
Dean of the College of Engineering
Daniel Rich, Ph.D.
There are several people whom I would like to thank for
their help and support.
Without them, none of this would’ve
been possible.
I would first like to thank my advisor, Dr. Dennis
Without his generous heart, brilliant mind, and
savvy business sense, my research would be non-existent.
is because of him I was able broaden my horizons and gain a
fundamental understanding of electromagnetics.
I would also like to thank some of my group members,
especially, Dr. Shouyuan Shi, Christopher Schuetz, Dr.
Changjun Huang, Dr. Ahmed Sharkawy, Dr. Iulian Codreanu,
Rownak Shireen, Zhaolin Lu, Elton Marchena, and Binglin Miao.
Without their knowledge and patience, I would have never
been able to accomplish my research.
I have enjoyed my time at the University of Delaware
and would like to thank the following people for making the
lab an enjoyable time: Timothy Creazzo, Timothy Hodson,
Iftekhar Mirza, Jesse Samluk, Brandon Redding, and James
Lastly, I would like to thank my parents and family for
their support.
Without it, I would have never been able to
come this far and accomplish so much.
List of Tables ......................................
List of Figures....................................
Abstract ...............................................
Chapter 1: INTRODUCTION ...........................
Introduction.............................................. 1
Basic Electromagnetic Theory.............................. 7
Types of Antennas......................................... 9
Parameters of Antennas................................... 11
Microstrip Patch Antennas................................ 16
Advantages and Disadvantages ........................... 19
Feed Techniques ........................................ 20
Horn Antenna............................................. 23
Antenna Description......................................
Fabrication ............................................
Experimental Setup .....................................
Experimental Results ...................................
Introduction............................................. 49
Antenna Description......................................
Antenna Fabrication ....................................
Experimental Setup .....................................
Experimental Measurements ..............................
REFERENCES ...........................................
List of Tables
Table 1. Efficiency of antenna in E- and H-plane with
respect to frequency. ................................ 71
List of Figures
Figure 1. Atmospheric transmission of the atmosphere [1].. 3
Figure 2. Typical changes of antenna amplitude pattern shape
from reactive near field toward the far field [6]. ... 14
Figure 3. Basic Microstrip Patch Antenna Configuration [7].
..................................................... 16
Figure 4. Typical Microstrip Patch Antenna Return Loss... 18
Figure 5. Predicted E- and H-plane patterns of a rectangular
microstrip patch. .................................... 18
Figure 6. Aperture-Couple Microstrip Patch Antenna
Configuration ........................................ 22
Figure 7. Pyramidal horn and coordinate system [7]....... 25
Figure 8. Frequency response of stacked patch antenna.... 27
Figure 9. Stacked Patch Antenna Configuration............ 28
Figure 10. Relationship between W1 and Resonant Strength
[10]. ................................................ 29
Figure 11. Relationship between W2 and Resonant Strength
[10]. ................................................ 30
Figure 12. Relationship between H1 and Resonant Strength
[10]. ................................................ 31
Figure 13. Simulated E- and H-plane patterns of double
stacked microstrip patch antenna. .................... 33
Figure 14. Addition of a microstrip line-to-waveguide
transition. .......................................... 34
Figure 15. Expected frequency response from the addition of
a microstrip line-to-waveguide transition. ........... 34
Figure 16. Proposed K-band antenna design................ 35
Figure 17. Processes to fabricate feed substrate and
patches. ............................................. 41
Figure 18. Horn antenna fabrication...................... 42
Figure 19. Transmission line characteristics of SMA
connector. ........................................... 44
Figure 20. Antenna Measurement Setup..................... 45
Figure 21. Comparison of experimental and simulated results.
..................................................... 47
Figure 22. Comparison of experimental and simulated results
without impedance mismatches. ........................ 47
Figure 23. Comparison between experimental and simulated
radiation pattern. ................................... 48
Figure 24. Rectangular plot of response versus angle..... 48
Figure 25. Proposed CPW feed substrate [4]............... 51
Figure 26. Proposed W-band antenna design................ 55
Figure 27. Feed substrate fabrication process............ 60
Figure 28. Antenna measurement setup..................... 62
Figure 29. Transmission response of antenna.............. 67
Figure 30. S11 return loss response of antenna........... 67
Figure 31. Comparison of E- and H-plane radiation with
respect to the standard gain horn antenna. ........... 68
Figure 32. Evidence for substrate modes.................. 68
Figure 33. Radiation pattern of antenna.................. 70
Millimeter-waves are of particular interest
because they act as the only part of the terrestrial
blackbody emission spectrum that transmits well through
atmospheric obstructions such as fog and smoke.
millimeter-waves possess the ability to penetrate thin
dielectrics such as clothing, which can be used at security
checkpoints to determine if a person is concealing any type
of weapons, explosives, or any other harmful objects.
to the lack of sufficient detectors and sources, extensive
research of millimeter-waves and their applications to
imaging systems has not been carried out.
In this thesis, I present two new antennas for use in
an imaging array.
Both structures will use a stacked patch
configuration to expand bandwidth and a waveguide-to-horn
transition to increase the gain and create a narrow
The main difference between the antenna designs
is the feed structure.
This results from the difficulty to
scale down microwave designs to millimeter-wave frequencies.
At K-band, an aperture-coupled feed allows design
flexibility while at W-band, a coplanar waveguide feed
creates a planar configuration. Experimental results will
show the possibility of creating an antenna at W-band for
use in a millimeter-wave imaging system.
Millimeter-wave imaging systems that utilize focal
plane array systems generate a considerable amount of
interest because of their applications in spectroscopy, low
visibility conditions, and security measures.
wavelengths between one millimeter and one centimeter, until
recently, millimeter-waves remained a relatively unexplored
part of the electromagnetic spectrum.
The primary factor
inhibiting the widespread use of millimeter-waves in various
applications is the lack of good sources and detectors
within this frequency regime.
Millimeter-waves engender
widespread curiosity because they act as the only part of
the terrestrial blackbody emission spectrum that transmits
well through atmospheric obstructions such as fog and smoke.
All natural objects whose temperatures are above absolute
zero emit passive millimeter-wave radiation.
attenuation in poor weather conditions such as fog, clouds,
snow, and rain make millimeter-waves much more effective
than infrared radiation in such conditions.
millimeter-wave receivers possess approximately 10 times
better noise performance than infrared detectors, however,
millimeter-wave radiation levels are approximately 108 times
smaller when compared to infrared levels [1].
temperature contrast between air and the radiating object
make up the remaining 103 differences.
Because of its
inherent properties, millimeter-wave imaging systems can
currently compare to current infrared imaging systems in
Shown in figure 1, electromagnetic radiation
windows occur at 35 GHz, 94 GHz, 140 GHz, and 220 GHz.
is at these frequencies that sources and detectors will have
the capability of penetrating inclement weather and opaque
objects; potentially creating fertile research opportunities
for the exploration of a millimeter-wave imaging system.
A technique based on the electro-optic conversion of
millimeter-wave energy into sidebands on an optical carrier
with subsequent filtering and photodetection of the
resultant sidebands will serve as the basis for a
millimeter-wave imaging system [2].
To collect incoming
signals, a millimeter-wave imaging array will consist of
adjacent antennas that will gather passive millimeter-waves.
Minimizing interference between adjacent array antennas in a
focal array system will necessitate narrow beamwidths.
Figure 1. Atmospheric transmission of the atmosphere [1].
In order to achieve wide bandwidth, high gain, and a
narrow beamwidth, two similar antennas are designed and
The first antenna is designed at microwave
frequencies and combines a waveguide-to-horn transition with
a double-stacked aperture-coupled microstrip patch antenna.
The second antenna is a scaled down version of the first
antenna except the second design employs a coplanar
waveguide fed square-slot instead of an aperture-coupled
The feeding technique at millimeter-wave frequencies
for microstrip antennas creates problems in the design
At millimeter-wave frequencies, classical feeding
techniques, such as coaxial probe or edge feeds create
several problems.
The patch itself can severely degrade
performance, soldering of probe feeds creates repeatability
problems, and edge-feeding provides very little room for the
feed network and associated devices.
The proposed aperture-
coupled feeding technique possesses many intrinsic
properties that make it an attractive feeding technique for
millimeter wave applications.
Conventional single layer microstrip antennas
inherently suffer narrow operation bandwidth.
To circumvent
this problem and increase the frequency bandwidth, we
implemented a stacked patch antenna technique where a
microstrip transmission line directly feeds the bottom patch
while the top patch couples parasitically to the bottom
Wide-band operation of this type of microstrip
antenna has been demonstrated at microwave frequencies (1-10
GHz) [3].
A single patch configuration facilitates a high
back-radiation level since the slot operates near resonance,
as this does not occur in the stacked patch configuration
the antenna design exploits the latter arrangement.
Coupling two patches together generates two new resonant
frequencies that differ from the individual resonant
frequencies of the single patches. The size of the top patch
differs slightly from the bottom patch, which allows the two
adjacent resonant frequencies to expand bandwidth.
Combining this structure with a waveguide-to-horn transition
will help to increase the bandwidth, increase the gain, and
create a narrow beamwidth.
A coplanar waveguide (CPW) configuration conveniently
provides a signal line and ground plane on the same side of
the substrate, which eliminates the need to align the
microstrip and slot in the aperture-coupled design.
using a CPW feed with a widened tuning stub, a square slot
antenna for broadband operation is designed.
results show that the impedance matching for the proposed
antenna strongly depends on the location of the tuning stub
in the square slot, and the impedance bandwidth is mainly
determined by the width and length of the tuning stub.
properly choosing the location and size of the tuning stub,
Reference [4] showed it was possible to obtain a wide
impedance bandwidth of 60%, which is 1.9 times that of a
conventional CPW-fed square slot antenna with a simple
tuning stub.
In this thesis, the objective is to design,
characterize, and fabricate two different antennas for a mmW
imaging system.
These antennas will be shown to have wide
bandwidth, high gain, and a narrow beamwidth.
Based on this
thesis, the feasibility of a K-band microwave antenna and a
W-band millimeter-wave antenna are demonstrated.
The main
difference between the antennas lies in the feeding
It will be shown that the new design
implementing a waveguide-to-horn transition will provide an
easier impedance match between free space and the top patch.
When the operating frequency of the antenna increases, the
size of the device decreases, creating a situation that
requires a new antenna measurement setup, a new antenna
fabrication process, and a feed structure that allows
compatible excitation with commercially available machines.
By converting to a CPW-feed structure from an aperturecoupled feed, the device results in easier alignment process.
It will be shown that both of these designs show promise for
millimeter-wave antenna design.
Basic Electromagnetic Theory
In order to fully appreciate how antennas operate and
why they are important devices, a fundamental understanding
of Maxwell’s equations is required.
Equations 2.1-2.4
represent Maxwell’s equations and provide the fundamental
basis for all electromagnetic theory [5]:
∇ × E = −M i −
= −M i − M d
∇ × H = Ji + Jc +
= Ji + Jc + Jd
∇ ⋅ D = qev
∇ ⋅ B = qmv ,
The definitions and units of the quantities are:
= Electric field intensity
= Magnetic field intensity
= Electric flux density
= Magnetic flux density
= Impressed electric density
= Conduction electric current density
= Displacement electric current density
= Impressed magnetic current density
= Displacement magnetic current density
= Electric charge density
= Magnetic charge density
For two systems to communicate between one another,
electromagnetic energy can facilitate the transport of
information between two systems.
The Poynting vector P
(equation 2.5) describes the energy flux of an
electromagnetic field and points in the direction of energy
Its magnitude is the power per unit area crossing a
surface, which is normal to it:
P = E×H .
In the analysis of radiation problems, the usual procedure
involves specifying the sources and then computing the
fields radiated by the sources.
By introducing an auxiliary
function, known as the vector potential, it is possible to
simplify the analytical process of finding E and H.
vector potential A for an electric current source J is
described by the following relationship:
∫∫∫ J
e − jkR
dv′ .
Also, equation 2.7 describes the relationship between the
vector potential F and a magnetic current source M:
e − jkR
dv′ .
The total fields of E and H are related to the vector
potentials by using the following equations:
E = E A + EF =
H = HA + HF =
∇× HA − ∇× F
∇× A−
∇ × EF .
The far-field radiation due to a current source J can be
found using the following relationships:
Similarly, a magnetic source M will radiate fields in the
far field as:
Types of Antennas
Antennas act as transitional devices that assist the
transport of electromagnetic energy.
Whether the antenna
acts as a transmitter or receiver, the antenna acts as an
intermediary device.
Different types of antennas provide
different radiation patterns for different applications;
some of these devices include: loop antennas, array antennas,
aperture antennas, microstrip antennas, and horn antennas.
The loop antenna provides a straightforward, economical, and
flexible design, but acts as a poor radiator and is rarely
used for transmission in radio communications.
antennas are necessary in the design of antennas with very
directive properties.
Aperture antennas are the most
prevalent at microwave frequencies and may adapt the shape
of a waveguide or horn.
The two antennas of most concern for this thesis are the
horn and microstrip antennas, but the antenna design will
incorporate loop, array, and aperture elements.
The microstrip antenna comes in a variety of designs,
which allows designers to use different feeding methods, a
planar or conformal configuration, and integration into
various applications.
Some of these applications include:
high-performance aircraft, spacecraft, satellites, missiles,
cars, and even cell phones.
The basic structure of a
microstrip antenna consists of a metallic patch on top of a
grounded substrate.
Although the existence of the horn antenna dates back
to the late 1800’s, the horn antenna is still used today as
a feed element for large radio astronomy, satellite tracking,
and communication dishes.
It is commonly seen as an element
of phased arrays and also serves as the universal standard
for calibration and gain measurements.
The structure of the
horn consists of nothing more than a hollow pipe of
different cross sections, which has been tapered to a larger
The type, dimensions, and amount of taper dictate
the overall performance of the antenna.
Parameters of Antennas
To possess the ability to fully characterize and
describe an antenna, various parameters need to be defined.
Radiation Pattern
An antenna radiation pattern provides a graphical
representation of the radiation properties of an antenna in
the far field.
Radiation pattern measurements are usually
normalized with respect to their maximum value, which
corresponds to a value of zero decibels on a polar plot.
The radiation pattern will most likely exhibit many
different lobes (unless it is an isotropic radiator), which
classify into three different types: major or main, side,
and back lobes.
An isotropic radiator radiates equally in
all directions, is lossless, and provides a reference for
the directive properties of antennas.
Most antennas possess many radiation lobes, which are
defined portions of the radiation pattern bounded by regions
of relatively weak radiation intensity.
The major lobe
contains radiation in the direction of maximum radiation.
Any lobe that radiates in any unwanted direction represents
Back lobes radiate 180° degree with respect to
a side lobe.
the main beam of the antenna.
Side and back lobes need to
be minimized since they usually radiate in undesired
The level of minor lobes is usually expressed
as a ratio of the power density in the lobe in question to
that of the major lobe.
Attainment of a side lobe level
smaller than –30 dB usually requires very careful design and
The designed antennas radiate with linear polarization
and to characterize the performance of these antennas, the
antenna is described in terms of its principal E- and Hplane patterns.
The E-plane possesses the electric field
vector while the H-plane contains the magnetic-field vector.
Both the E-plane and H-plane are in the plane containing the
direction of maximum radiation.
Field Regions
To detect the amount of radiation from an antenna at
different distances, three regions are defined: reactive
near-field, radiating near-field, and far-field.
The region
closest to the antenna, the reactive near-field, encloses a
space of radius R < 0.62 D 3 / λ , where λ is the wavelength and
D represents the largest dimension of the antenna.
fields dominate in this region. The radiating near-field
region lies between the reactive near-field region and the
far-field region.
The radiation fields begin to dominate
and the angular field distribution of the antenna relies on
the distance from the antenna.
0.62 D 3 / λ ≤ R ≤ 2 D 2 / λ .
This occurs when
In the far-field region, the angular
field distribution becomes independent of the distance from
the antenna.
The field components are essentially
transverse and the angular distribution is independent of
the radial distance where the measurements are made.
far field exists at any distance where R > 2 D 2 / λ .
amplitude of the radiation for an antenna varies as the
observation distance increases from the source.
change in both magnitude and phase.
The fields
A typical progression
of the shape of an antenna, with largest dimension D, is
shown in Figure 2 [6].
Figure 2 shows that in the reactive
near-field region, the radiation retains a more spread out
and nearly uniform pattern, with slight variations.
As the
observation point increases to the radiating near-field
region, the pattern begins to smooth and form lobes, whereas
in the far-field region, the pattern develops into a welldefined pattern with no variations.
Figure 2. Typical changes of antenna amplitude pattern shape
from reactive near field toward the far field [6].
The beamwidth of an antenna radiation pattern gives a
visual representation of the directivity an antenna and is
understood to be the angular separation between two
identical points on opposite side of the pattern maximum.
conventional beamwidth measurement, the half-power beamwidth
(HPBW) represents the angles at which the main beam reaches
half its maximum value.
The beamwidth of an antenna denotes
a very important figure of merit and often a trade-off
exists between the beamwidth and the side lobe level; that
is, as the beamwidth decreases, the side lobe increases and
vice versa.
The beamwidth describes the resolution
capabilities of an antenna and its ability to distinguish
between two adjacent radiating sources.
A narrow beamwidth
will be very important for an imaging array.
Directivity versus Gain
The directivity of an antenna compares the radiation
intensity in the maximum direction of the antenna to the
radiation intensity averaged over all directions.
average radiation intensity is commensurate to the total
power radiated by the antenna divided by 4π .
4π U
The gain of an antenna relates the directivity to the
efficiency of the antenna and is a ratio between the
radiation intensity in the direction of maximum radiation
and the radiation intensity of an isotropic radiator.
difference between gain and directivity is that gain
incorporates antenna losses while the directivity does not.
The bandwidth of an antenna is comprised of a range of
frequencies, such that the antenna performs to a specified
For a wide bandwidth antenna, useful bandwidth
corresponds to a return loss measurement of below –10 dB.
Having a wide bandwidth will allow the capture of radiation
over a wide frequency range [7].
measured experimentally by:
Percent bandwidth can be
High Frequency - Low Frequency
Center Frequency
Microstrip Patch Antennas
The microstrip patch antenna is nothing more than a
radiating patch on one side of a dielectric substrate and a
ground plane on the opposite side as shown in figure 3 [7].
Figure 3. Basic Microstrip Patch Antenna Configuration [7].
Usually the patch conforms to some geometric shape such as a
square, rectangle, circle, triangle, ellipse, or some other
common shape.
This simplifies analysis and allows for a
better prediction of antenna performance.
A rectangular
patch of length L usually retains dimensions of
0.3333λ0 < L < 0.5λ0 , where λ0 is the free space wavelength.
patch thickness is much smaller than the free space
wavelength such that t << λ0 (where t is the patch thickness).
The dielectric substrate thickness h usually falls in the
range 0.003λ0 < h < 0.05λ0 .
Typical dielectric constants of the
substrate ( ε r ) lie in the range 2.2 ≤ ε r ≤ 12 [8].
fields between the patch edge and ground plane provide the
fundamental radiating mechanism for microstrip patch
A thick dielectric substrate having a low
dielectric constant creates an antenna with better
efficiency, larger bandwidth, and better radiation, however,
this configuration leads to larger antenna dimensions.
design a compact microstrip patch antenna, a substrate with
a higher dielectric constant should be employed, but using a
higher dielectric constant results in a less efficient
antenna and narrower bandwidth.
Thus, a compromise must be
reached between antenna dimensions and antenna performance.
A typical frequency response of a microstrip patch antenna
is shown in figure 4.
The radiation pattern characteristics
are shown in figure 5.
Figure 5 shows the moderately
directive properties of a single patch antenna.
Figure 4. Typical Microstrip Patch Antenna Return Loss.
Figure 5. Predicted E- and H-plane patterns of a rectangular
microstrip patch.
Advantages and Disadvantages
Microstrip patch antennas provide many enviable
advantages for wireless applications.
They have found their
niche in many communications applications because of their
low profile structure.
Microstrip patch antennas can be
found in handheld wireless devices such as cell phones and
Some of the advantages of microstrip antennas
include: 1) light weight and low volume, 2) low profile
planar configuration which can be easily made conformal to
the host surface, 3) low fabrication cost, 4) supports
linear and circular polarizations, 5) the adeptness to be
easily integrated with microwave circuits, 6) the efficacy
of dual and triple frequency operations, and 7) mechanically
robust when mounted on rigid surfaces.
Some of their major
disadvantages include: 1) narrow bandwidth, 2) low
efficiency, 3) low gain, 4) low power handling capacity, and
5) surface wave excitation.
Microstrip patch antennas possess a very high antenna
quality factor (Q), which represents the losses associated
with the antenna.
A large Q means the antenna will have
narrow bandwidth and low radiation efficiency.
radiation, ohmic, dielectric, and surface wave losses exist
within the antenna.
Increasing the thickness of the
dielectric substrate will lower the Q, but as the thickness
increases, an increasing fraction of the total power
delivered by the source goes into a surface wave.
waves represent unwanted power loss since they cause the
degradation of antenna characteristics by scattering fields
at the dielectric bends.
Feed Techniques
The flexibility of microstrip patch antennas allows
designers to choose between various feed techniques.
device can use a microstrip line to directly feed RF power
to the radiating patch, a coaxial probe, or a non-contacting
scheme that couples electromagnetic energy between the
microstrip line and the radiating patch.
The four most
conventional feeding techniques include the microstrip line,
coaxial probe, proximity coupling, and aperture coupling.
Directly connecting a stripline to the edge of the patch
makes fabrication easier, however, problems emerge.
feed technique leads to surface waves and increases the
probability of spurious feed radiation, which decreases the
bandwidth of the antenna.
The coaxial probe scheme provides
auspicious properties since this feeding technique can be
placed at any desired location inside the patch in order to
match with its input impedance.
The main problem with this
feeding technique is that a hole needs to be drilled in the
substrate to excite the patch.
At millimeter-wave
frequencies, this will be a very arduous task.
coupling uses an electromagnetic coupling scheme that will
eliminate spurious feed radiation and provide high bandwidth,
however, this technique creates a complicated fabrication
scenario since the two layers require alignment.
The first antenna design uses an aperture-coupled feed
as the feed technique of choice.
The aperture-coupled feed
provides significant advantages in the design process.
Since the ground plane isolates the feed network from the
radiating element, this configuration helps to prevent
spurious radiation.
The aperture-coupled feeding
configuration allows the fabrication of devices with a
higher dielectric constant in the feed substrate, which
effectively reduces the size of the device.
Thus, the
aperture-coupled feed allows the designer more freedom.
this type of feed technique, the ground plane separates the
radiating patch and the microstrip feed line.
in figure 6a.
This is shown
A slot or an aperture in the ground plane
couples the energy between the patch and the feed line.
Usually centered under the patch, the coupling aperture
leads to a decreased cross polarization due to symmetry of
the configuration.
The only requirement for the aperture is
that it should be precisely positioned below the microstrip
patch and above the feed line, which is shown in figure 6b).
The shape, size, and location of the aperture determine the
amount of coupling from the feed line to the patch.
slot couples energy from the stripline to the patch and can
be either resonant or nonresonant.
If resonant, in addition
Figure 6. Aperture-Couple Microstrip Patch Antenna
to the patch resonance, the slot provides another resonance,
which results in an antenna with 10-15% impedance bandwidth,
however, the resonant slot possesses strong backlobe
radiation which substantially reduces the gain of the
For this reason, a nonresonant slot is usually
preferred in aperture-coupled microstrip patch antennas.
The position of the aperture above the feed line creates a
stronger coupling because of the higher concentration of
fields at the slot.
This distance corresponds to a quarter-
wavelength from the end of the microstrip line.
designing an aperture-coupled microstrip antenna, the feed
substrate should posses a high dielectric, while the patch
substrate utilizes a thick low dielectric constant material.
This will optimize radiation from the patch.
The major
disadvantage of this feed technique is that it is difficult
to fabricate due to multiple layers, which also increases
antenna thickness.
This feeding scheme also provides narrow
Horn Antenna
The horn antenna serves as the universal standard for
all antenna calibration and gain measurements.
different types of horn antennas exist, however, the
pyramidal horn configuration serves as the primary horn
antenna of interest.
A pyramidal horn antenna combines the
design of E- and H-plane sectoral horns.
The tangential
components of the E- and H-fields over the aperture of the
horn are approximated in [6] by:
⎛π ⎞
E ′y ( x′, y′) = E0 cos ⎜ x′ ⎟ e − j[ k ( x′ / ρ2 + y′ / ρ1 ) / 2]
⎝ a1 ⎠
H x′ ( x′, y′) = −
⎛π ⎞
cos ⎜ x′ ⎟ e − j[ k ( x′ / ρ2 + y′ / ρ1 ) / 2]
⎝ a1 ⎠
and the equivalent current densities by:
J y ( x′, y′) = −
⎛π ⎞
cos ⎜ x′ ⎟ e − j[ k ( x′ / ρ2 + y′ / ρ1 ) / 2]
⎝ a1 ⎠
⎛π ⎞
M x ( x′, y′) = E0 cos ⎜ x′ ⎟ e − j[ k ( x′ / ρ2 + y′ / ρ1 ) / 2] .
⎝ a1 ⎠
The pyramidal horn radiates fields that correspond to a
combination of the E- and H-plane horn antennas.
Both the
E-plane and H-plane patterns of the pyramidal horn antenna
are identical to their respective sectoral counterparts.
To construct a pyramidal horn, the dimensions pe and ph
should be equal, such that [7]:
⎡⎛ ρ ⎞ 2 1 ⎤
pe = (b1 − b) ⎢⎜ e ⎟ − ⎥
⎢⎣⎝ b1 ⎠ 4 ⎥⎦
1/ 2
1/ 2
⎡⎛ ρ ⎞ 2 1 ⎤
ph = (a1 − a ) ⎢⎜ h ⎟ − ⎥
⎢⎣⎝ a1 ⎠ 4 ⎥⎦
where pe and ph are the depths of the horn in the E- and Hplanes respectively, b1 and a1 denote the length and width
of the horn, and
the aperture.
b and a represent the length and width of
The horn radiates most of its energy along
the z-axis (θ=0).
The directivity of a pyramidal horn can
be calculated from the following equation [7]:
⎛ 32 b1
π λ ρe / λ
Dp = ⎝
⎞⎛ 32 a1
⎠⎝ π λ ρ h / λ
ρe / λ ρ h / λ
Antenna Description
Pozar and Croq designed [10] a microwave wide-band
aperture-coupled stacked microstrip antenna using stacked
elements that will increase the impedance bandwidth.
Published results have shown that aperture-coupled multilayer microstrip antennas exhibit the ability to obtain
bandwidths as large as 50%, which is greater than those of
dipoles and slots and comparable to the horn antenna [9].
By combining this structure with a waveguide-to-horn antenna
Figure 7. Pyramidal horn and coordinate system [7].
transition, it is possible to create a microwave antenna
with wide bandwidth, high gain, and a narrow beamwidth.
Pozar’s design forms the basis for the microwave
antenna design in this thesis.
The coupling between stacked
patches generates two new resonances.
These new resonant
frequencies do not have a simple or direct relation to the
former ones, which makes it difficult to make general
statements about their characteristics.
Figure 8a shows the
return loss performance of a microstrip patch antenna with
the use of a bottom patch only.
If the bottom patch is
replaced with the top patch from the design specifications,
figure 8b shows the respective S11 performance.
When the
two patches are combined, figure 8c shows the generation of
a wide bandwidth.
Figure 8. Frequency response of stacked patch antenna.
Five parameters which are most critical in the design
of a K-band antenna have been characterized and analyzed;
these parameters include: 1) the dimensions of the lower and
upper square patches, W1 and W2 ,respectively, 2) the
thicknesses of the two substrates supporting the patches,
H1 and H 2 ,respectively, and 3) the length of the coupling
slot A1 .
For the discussion of parameter analysis, the
parameters have the following dimensions: W1 = 3.5 mm, W2 =
3.8 mm, H1 = 0.50 mm, H 2 = 1.0 mm, and Al = 3.2 mm.
Figure 9. Stacked Patch Antenna Configuration.
In this analysis, it is assumed that W1 , W2 , H1 , H 2 ,
and Al remain constant unless otherwise noted.
When the
ratio of W1 / Al increases, the coupling to the top patch
decreases. This occurs because the top patch becomes
isolated from the excitation field of the slot, causing the
coupling to the fringing field to decline.
When the size of
the bottom patch W1 decreases, the resonant frequency of
that patch begins to converge to the behavior of the top
The antenna begins to operate as if the bottom patch
doesn’t exist when W1 approaches a value of zero.
As W1
decreases, the upper resonance becomes strongly excited at a
higher frequency.
Illustrated in figure 10, a 10% increase
in W1 will reduce the lower resonant frequency by 4%, while
the upper resonant frequency only decreases by 2.6%;
Conversely, when the size of W1 decreases by 10%, the lower
resonance increases by 1.1%, while the upper resonant
frequency increases by 4.3% [10].
This clearly shows that
changing the dimensions of one patch will affect both
Figure 10. Relationship between W1 and Resonant Strength
Analysis of W2 on the device performance will show
that as W2 decreases, coupling to the lower resonance
Figure 11 shows that when W2 becomes very small,
the device behaves as if the resonance emerges from the
bottom patch by itself.
Coupling between the fringing
fields of the bottom and top patches becomes negligible.
This seems logical since increasing W1 and reducing W2 have
the same effect of decoupling the top patch from the
fringing field of the bottom patch.
A 20% variation of W2
will affect the resonant frequencies by 6 to 7%.
Figure 11. Relationship between W2 and Resonant Strength
Figure 12 shows that as the thickness of the feed
substrate H1 increases slightly, both patches become
removed from the coupling slot.
Intuitively, increasing H1
would cause both resonances to decrease, however, this did
not occur.
As H1 increases from its original value of 0.5
mm to 1.2 mm, coupling to the lower resonance strongly
reduces and the upper resonance increases to a maximum
before decreasing.
Figure 12. Relationship between H1 and Resonant Strength
When H 2 approaches zero, both resonances show
analogous behavior.
This implies that the patches appear as
a single patch to the slot when placed very close together,
even though the coupling between them generates another
resonance at a high frequency.
Conversely, large values of
H 2 will isolate the top patch from the structure, causing
the behavior of the overall structure to resemble that of a
single patch antenna.
It was also seen that the coupling
between two patches affected the frequency of the upper
resonance, which increased as H 2 decreased.
The influence of the length of the slot on the strength
of the resonances possesses a similar effect as the patches.
When Al increases from its original value to 3.6 mm, the
lower resonance immensely intensifies while the upper
resonance decreases to some extent.
Conversely, when Al
decreases from its original value to 2.8 mm, the strength of
the lower resonance will diminish while the upper resonance
slightly increases [10].
The three resonators of this structure interact
strongly and are intimately coupled.
For the two resonant
patches that determine the bandwidth of the antenna, any
variation of the geometric parameters will result in a
modification of the relative excitation of the resonances.
The effects of these changes are interdependent, which
allows design flexibility.
The radiation pattern of this
structure is shown in figure 13.
From the plot, it is
apparent that more energy radiates in the E-plane, while the
H-plane is slightly more directive.
Figure 13. Simulated E- and H-plane patterns of double
stacked microstrip patch antenna.
When these elements are constructed together, the
addition of a waveguide transition on top shown in figure 14,
will give the radiation pattern a different shape, but the
S11 return loss parameters become slightly more broadband;
this is shown in figure 15.
By varying the different
dimensions of the antenna, the directivity and gain of the
antenna will change accordingly.
By adding a horn to the
waveguide transition, this will ensure that when the
antennas are placed in an array, the beamwidths of each horn
antenna won’t interfere with the adjacent antenna.
Figure 14. Addition of a microstrip line-to-waveguide
Figure 15. Expected frequency response from the addition of
a microstrip line-to-waveguide transition.
Figure 16 shows a microwave antenna designed to operate
in the frequency range of 16-20 GHz.
The feed substrate has
a thickness of 0.4 mm, the bottom patch substrate employs a
thickness H1 of 0.508 mm, and the top patch substrate
possesses a thickness H 2 of 0.787 mm.
The top and bottom
patches have square dimensions of W1 = 4.4mm and W2 =4.65 mm
Aw =0.4 mm.
The slot has a length Al =3.5 mm and width
The waveguide transition element possesses
dimensions of Wg = Lg =14 mm and height of H g =5 mm.
The horn
antenna will have pe = ph =45 mm and a length of h =50 mm.
Figure 16. Proposed K-band antenna design.
When deciding which materials to use, the antenna was
designed using commercially available materials.
[8] suggests using a material with a dielectric constant of
2.2 ≤ ε r ≤ 12 for the patch substrate.
A thick dielectric
substrate having a low dielectric constant is desirable,
however, a material with a higher dielectric constant allows
the design of a compact microstrip antenna.
(ε r = 4.4 )
Originally, a FR4 epoxy
was chosen as the material
of choice for the feed substrate, however, after extensive
research, results showed the lossy properties of epoxy
between 16-20 GHz.
Alternatives to FR4 epoxy are the TMM
substrates provided by Rogers Corporation.
TMM materials
provide a rigid substrate that allows the use of
conventional printed circuit board techniques.
possesses a very similar dielectric constant
(ε r = 4.4 ) ,
(ε r = 4.5)
to FR4
however, its loss tangent of 0.002 at 10 GHz
makes TMM4 a much better material at higher frequencies.
FR4 epoxy on the other hand has a loss tangent of 0.016 at 1
Clearly, TMM4 will perform better at higher
For the patch substrates, both patches are fabricated
on RT/duroid 5880, also provided by Rogers Corporation.
RT/duroid 5880 is a glass microfiber reinforced PTFE
composite designed for exacting stripline and microstrip
circuit applications.
The dielectric constant of a material
usually changes with increasing frequency, however,
RT/duroid 5880 remains constant over a wide frequency range.
Both TMM4 and RT/duroid 5880 come with 35 microns of copper
electrodeposited on both sides.
The loss tangent of
RT/duroid 5880 decreases from 0.001 at 1 GHz to 0.0009 at 50
GHz [11].
Fabrication of the proposed K-band antenna consists of
a series of sequential fabrication steps.
Each antenna
element will be constructed separately and combined to
construct the device described above.
The proposed antenna
consists of five elements: the microstrip feed, the ground
plane with rectangular aperture, the top and bottom patches,
and the waveguide-to-horn transition.
All elements are
fabricated in separate steps, however, since the microstrip
feed and rectangular aperture lie on opposite sides of the
feed substrate, alignment marks ensure the proper placement
of both elements.
Photolithography represents the utmost critical step in
the antenna fabrication process.
By applying a light-
sensitive photoresist to the wafer and controlling the
amount of light exposed the wafer, photolithography allows
the fabrication of three-dimensional devices.
An important
performance measurement in photolithography is the
resolution of each image, which is the adeptness to discern
between two closely spaced features on the wafer surface.
Employing the correct amount of light and the right resist
will allow almost any design to be patterned on the wafer at
the scales required for this project [12].
For the feed substrate and patches, positive
lithography is employed, which will create a pattern on the
wafer that replicates the pattern on the mask.
This occurs
because positive photoresist, when exposed to UV light, will
undergo a photochemical reaction and become soluble and
soften in the developer.
In this manner, the developer
solution will remove regions of resist exposed to light,
whereas regions of resist not exposed to light underneath
the opaque mask pattern remain on the wafer.
For these
devices, AZ 4330 photoresist and AZ 400K photoresist
developer are used.
The first step in the photolithography process involves
cleaning the wafer.
Shown in figure 17a), this process
consists of spinning the substrate at 3000 rpm for thirty
During the thirty seconds, the wafer undergoes an
acetone-methenol-isoproponal (AMI) cleaning approach, where
each chemical is sprayed on the wafer for ten seconds each.
After cleaning the wafer, the substrate is accelerated to
5000 rpm for 10 seconds to dry.
The wafer is then
decelerated to apply a coat of liquid photoresist.
figure 17b), AZ 4330 is spun at 3000 rpm for 15 seconds,
which will apply a resist layer of approximately 5 microns.
Figure 17c) shows the resist undergoing a soft bake after
resist has been applied to the wafer.
The soft bake process
improves adhesion, promotes resist uniformity on the wafer,
and yields better linewidth control during etching.
17d) shows aligning the mask to the correct location of the
resist-coated wafer.
Depending on the reading on the
intensity meter, trial experiments showed that the radiation
intensity should be adjusted such that the wafer is exposed
for 225 mJ/cm2.
After exposure, the pattern becomes visible
after a development using AZ 400K for 90 seconds.
Development embodies the most critical step for creating the
pattern in the photolithography process.
In figure 17e),
the developer dissolves away the soluble areas of the
photoresist, leaving visible patterns of islands and windows
on the surface.
A chemical or physical etch process involves
selectively removing unwanted material from the wafer
Accurately reproducing the mask features on the
wafer signifies a successful etch.
In figure 17f), a wet
etch consisting of ferric chloride removes the copper
On the feed substrate, the microstrip side is
processed first.
Before etching can occur on this side, the
rectangular aperture side must be coated with a protective
resist layer.
This can be accomplished by using a swab to
apply photoresist onto the backside, which will prevent the
ground plane from being etched away.
Both the feed
substrate and patches are submerged in a ferric chloride
bath for three minutes to remove unwanted copper.
the ferric chloride bath to 55 °C accelerates the etching
When conclusive patterns become visible on the
substrate, the substrate undergoes an AMI cleaning to remove
the photoresist, as shown in figure 17g).
In a successful
fabrication run, the backside of the feed substrate is left
The backside of the feed substrate will then
undergo steps a)-g) of figure 17, to pattern the desired
aperture in the ground plane.
To fabricate the patches,
these devices will undergo fabrication steps 17a)-17g).
Figure 17. Processes to fabricate feed substrate and patches.
The waveguide-to-horn transition is fabricated
from .007” thick copper.
Shown in figure 18a), the
transparency printout provides a folding outline for the
The waveguide and horn are separately created and
soldered together.
When completed, the horn antenna, the
patches, and the feed substrate are glued together.
shown in figure 18.
Figure 18. Horn antenna fabrication.
This is
Experimental Setup
To accurately test the performance of the antennas, an
Agilent 8720ES Vector Network Analyzer (VNA) helps to
determine the performance of the antennas.
To perform
return loss and transmission measurements, the VNA is
calibrated using an Agilent 85056A 2.4 mm calibration kit,
which consists of using an open, a short, and a 50-ohm load.
A SMA edge-launch connector will excite the microstrip lines,
but the center electrode of the SMA connector needs to be
thinned down to 1/16
of an inch.
Failure to do so will
result in impedance mismatches and poor coupling, thus
invalidating the experimental results.
The SMA connectors
have a maximum operating frequency of 20 GHz.
Figure 19
shows the difference between a straight through between two
cables (which will be taken as our zero reference line), a 3
cm transmission line without thinning the center electrode,
and a 3 cm transmission line with a thinned electrode.
can be observed from figure 19, it is apparent that thinning
the center electrode makes an overwhelming difference in the
transmission line data.
A thinner electrode allows more
energy to be directly coupled to the microstrip line.
energy lost from center electrode to the microstrip line has
been minimized.
Figure 19. Transmission line characteristics of SMA
Figure 20 shows the radiation pattern measurement setup,
where two antennas are connected to an Agilent 8720ES Vector
Network Analyzer.
One of the antennas is placed on a
rotation stage, where Labview in conjunction with a Newport
motion controller, will control the rotation stage,
performing measurements at single degree increments.
antennas are separated by a distance of 6 feet, which for
the frequency of interest, lies in the far field.
Figure 20. Antenna Measurement Setup.
Experimental Results
A K-band microwave antenna for an imaging array has
been designed, characterized, and fabricated.
microstrip and slot designs are fabricated on a 0.4 mm thick
TMM4 double-sided copper clad substrate (•r = 4.5).
double-stacked patch antennas are patterned on RT/Duroid
5880 substrates (•r = 2.2) that possess thicknesses of
0.508mm and 0.787 mm.
The horn antenna consists of .007”
thick copper foil folded to form the horn shape.
combined structures form a high gain directional antenna.
The proposed antenna has been analyzed using Ansoft’s HFSS.
Simulations predict that the horn supplies a gain of 18 dB,
a very narrow beam width of ± 22°, and 30% bandwidth
centered at fo=17.5 GHz.
the simulated results.
Experimental results agree with
Shown in figure 21, the experimental
results show a strong resonance at 19.5 GHz.
The S11 return
loss measurements show that the antenna operates at a center
frequency of 17.5 GHz.
The center frequency is the midpoint
between the points at which the S11 measurements fall below
–10 dB.
Both simulated and experimental results exhibit the
same overall behavior.
The weaker response from the
fabricated antenna is attributed to the impedance mismatch
between the connector-to-microstrip line transition.
the mismatch is taken into consideration, figure 22 shows
that the performances of both antennas are almost identical.
The loss in energy from the impedance mismatch is determined
from figure 19.
As seen in figure 23, the antenna exhibits
a HPBW of approximately ± 30°.
Figure 24 shows that the
antenna possesses a front-to-back ratio of 18 dB.
Since the
beamwidth is very dependent on the dimensions of the horn
antenna, it is possible that folding the antenna out of
copper will lead to inexact dimensions.
What is promising
is that both simulation and experimental show the same
overall pattern.
H igh G ain P atc h F ed H orn A ntenna
S im ulat ed
E x perim ental
Response (dB)
15. 5
16. 5
17. 5
F requenc y (G H z )
18. 5
19. 5
Figure 21. Comparison of experimental and simulated results.
Figure 22. Comparison of experimental and simulated results
without impedance mismatches.
S im u la tio n
E x p e rim e n t a l
Figure 23. Comparison between experimental and simulated
radiation pattern.
N o rm a l iz e d R e s p o n s e vs A n g l e
-1 0
Response (dB)
-1 5
-2 0
-2 5
-3 0
-3 5
-4 0
A n g le (d e g re e s )
Figure 24. Rectangular plot of response versus angle.
A coplanar waveguide (CPW) structure bestows beneficial
properties to the antenna design, allowing the assemblage of
a signal line and ground plane on the same side of the
substrate, which dismisses the need for vias commonly used
with microstrip transmission lines.
Recently, the designs
of CPW-fed wide slot antennas received a great deal of
consideration because the CPW fed wide slot antennas have
the favorable properties of wide bandwidth and easy
integration with monolithic microwave integrated circuits.
The adoptions of a wide rectangular slot [13] or bow tie
slot [14] have led to marginally successful attempts to
increase the bandwidth of CPW-fed slot antennas.
Alternative broadband configurations such as a patch element
loaded in a circular slot [15] or a hybrid slot [16] have
additionally been employed to acquire a dual-resonance
response, however, these published CPW-fed broadband designs
perform with impedance bandwidths of less than 50%.
alternative approach for improving the impedance bandwidth
of CPW-fed structures is a CPW-fed square slot design.
This design implements a widened tuning stub to adjust the
coupling between the CPW feed line and the radiating slot by
varying the parameters of the location and size of the
tuning stub.
This facilitates the impedance matching of the
In [17], Giauffret
showed that using a square
slot loop will result in front-to-back ratios on the order
of 16 dB.
In this configuration, a metallic plane acts as a
conducting plane for the CPW line and as a reflector plane
for the radiating element.
Antenna Description
For the antenna described above, the antenna consists
of 4 elements: the CPW-fed square slot antenna, two stacked
patch antennas, and a horn antenna.
Figure 25 [17] shows
the geometry of the proposed CPW-fed square slot antenna
with a widened tuning stub.
A z-cut lithium niobate wafer
of thickness 150 μm (h) with relative permittivities of
ε z = 28 and ε x = ε y = 43 serves as the feed substrate.
niobate is used because of its highly electro-optic
A ground plane will cover the entire backside
of the wafer to diminish backside radiation.
The feed
substrate possesses a square radiating slot of dimensions
L × L , a 50 Ω CPW feed line having a center electrode of
width w f , and a gap of distance g between the center
Figure 25. Proposed CPW feed substrate [4].
electrode and ground planes.
Conjoined to the end of the
CPW feed line, the widened tuning stub of length l and
width w controls the amount of coupling.
The last
dimension of concern is the spacing S between the tuning
stub and edge of the ground plane.
[4] optimized the CPW-
fed square slot design to acquire 60% bandwidth between
1.56-2.88 GHz.
By using HFSS Ansoft and scaling down the
dimensions of the antenna to operate at W-band, the
dimensions are chosen to be g =15 μm, w f =45 μm, l =220 μm,
w =220 μm, S=40 μm, and L =300 μm.
The three parameters of
the tuning stub (i.e., l , w , and S) affect the broadband
operation of the proposed antenna, and the effects of the
design have been studied using HFSS Ansoft.
The first design process involves the impedance
matching of a 50 Ω CPW feed line to a widened tuning stub.
By varying the spacing (S), the effect on the impedance
matching shows that the resonant input resistance increases
as the spacing decreases.
Thus, the magnitude of coupling
between the widened tuning stub and ground plane behaves as
a function of distance.
The antenna exhibits strong
coupling between the tuning stub and ground plane when S=40
Chen showed that when the width of the tuning stub and
spacing are fixed, two resonant modes are excited. The
widened tuning stub contributes to the lower resonant mode,
whereas the fundamental resonant mode of the square slot
antenna bestows to the higher resonant mode.
Decreasing the
length l of the tuning stub imperceptibly perturbs the
higher resonant frequency, while the lower resonant
frequency expeditiously increases, shifting closer to the
higher resonant frequency.
Wide bandwidth operation will
arise when the combined resonant modes have a return loss
larger than 10 dB.
Properly selecting the tuning-stub
length generates two resonant modes at close frequencies
that result in a wide bandwidth operation.
The impedance bandwidth can be further improved by
adjusting the tuning-stub width ( w ).
Choosing w properly
will result in an impedance bandwidth larger than 50%.
Increasing the tuning-stub width will significantly increase
the higher frequency response.
It was shown in [4] that by
using a wide range of w from 0.59L to 0.9L, the proposed
antenna is relatively insensitive to the fabrication
tolerance for achieving a broadband operation.
The designed W-band antenna possesses dimensions of l
and w such that l = w .
square loop.
Patches are placed on top of the
Backside radiation will increase strongly with
increasing loop size, which shows that the antenna is poorly
excited for large dimensions of the loop.
This phenomenon
occurs because a slot-aperture excitation utilizes a
magnetic coupling effect as the dominant coupling mechanism.
To create an optimal coupling scenario between a square slot
and rectangular patch, the maximum of magnetic field for the
TM10 mode should be located in the center of the rectangular
patch antenna.
Also, the magnetic field in the slot loop is
considerably stronger in the loop edges parallel to the
radiating sides of the patch than in the two other edges of
the loop.
The patch excitation results largely from these
exciting edges.
When the dimensions of the loop increase,
the exciting edges become removed from the center of the
antenna, which deteriorates the coupling mechanism.
energy administered by the slot loop will directly couple to
both the patch antenna and to the backside of the substrate.
For a millimeter-wave antenna designed to operate in
the frequency range of 80-110 GHz, figure 26 shows that the
z-cut lithium niobate substrate has a thickness of 150 μm,
the bottom patch substrate has a thickness H1 of 0.127 mm,
and the top patch substrate possesses a thickness H 2 of
0.127 mm.
The top and bottom patches are fabricated on
RT/Duroid 5880 and have square dimensions of W1 = 0.75 mm
and W2 =0.83 mm, respectively.
The waveguide transition
element possesses dimensions of Wg = Lg =2.3 mm and height of
H g =1.2 mm.
The horn antenna will have pe = ph =8.5 mm and a
length of h =9 mm.
Figure 26. Proposed W-band antenna design.
Antenna Fabrication
Because the antenna will be used in conjunction with a
modulator to create a 2x2 millimeter-wave imaging array, a 3
inch lithium niobate wafer is used.
All processes will be
carried out for each element in the 2x2 array at the same
This helps to decrease the chance of human error and
increases the yield of the device fabrication.
In lithium-
niobate modulators, velocity matching between optical and
electrical signals is one of the largest challenges to
address due to the disparity in dielectric constants of
lithium niobate at optical and millimeter-wave frequencies.
While the typical index of refraction of the optical
waveguide near-infrared wavelengths is 2.15, the effective
index of the lithium niobate at millimeter-wave frequencies
is ~5-6, depending on the cut and field-orientation in the
anisotropic lithium niobate.
Thus, CPW’s patterned directly
on the surface of a planar lithium niobate substrate have an
effective index higher than that of the optical mode [18].
In order to correct this mismatch, a number of
techniques have been used.
Perhaps, the most common
technique is to introduce an intermediate low-index SiO2
buffer layer between the CPW electrodes and the lithium
Figure 27a) shows a 1.6 μm thick SiO2 layer is
deposited onto the wafer surface by a plasma enhanced
chemical vapor deposition (PECVD).
By using an Edwards FL
400 evaporator, seed layers of Ti and Au are evaporated onto
the lithium niobate substrate.
Shown in figure 27b), three
seed layers having a composition of Ti:Au:Ti and a thickness
of 30/100/30 nm respectively, accumulate on the lithium
niobate surface.
The first seed layer of titanium promotes
adhesion between the SiO2 buffer layer and gold, the second
seed layer of gold is used as a source for electroplating,
and the top seed layer will serve as a protective layer
during the electroplating process.
This top layer will
prevent shorting of the center electrode and ground planes
that can be realized during electroplating if a pinhole
exists in the photoresist.
SU-8 2025 negative photoresist
was chosen for this application because of its robust
physical properties after development, ability to form thick,
high-aspect ratio structures, and its adeptness to minimize
swelling. After cleaning the substrate using AMI cleaning
techniques, in figure 27c), a layer of SU-8 resist is
deposited on the substrate.
Possessing the inherent
properties of a negative photoresist, SU-8, when exposed to
light, will become insoluble and harden by cross-linking.
Once hardened, the cross-linked resist cannot be washed away
in solvents.
During exposure, the regions underneath the
opaque chrome of the mask remain unexposed to light
dissolves when exposed to the developer chemical since those
portions of the resist did not crosslink.
To create a
resist layer of approximately 25 microns, the resist is
accelerated to 4000 rpm at 100 r/s for 30 s.
To increase
the percentage of resolvable features during exposure, an
Edge Bead Remover with a swab removes the corner beads of
the wafer.
The wafer undergoes a prebake of two and a half
minutes at 135°C to promote adhesion between the top
titanium seed layer and SU-8.
For exposure, a total
intensity of 94 mJ/cm2 will create resolvable feature sizes.
Important to the exposure step is a good contact between the
resist and photomask because failure to do so will result in
irresolvable patterns.
In figure 27e), the substrate is
submerged in Microchem’s SU-8 developer for approximately 25
seconds or until the pattern is visible.
Regions of the
substrate where titanium is exposed will be removed using a
titanium etch consisting of a solution H2O: HF:H2O2 with
composition ratios of 20:1:1 for approximately 15 seconds.
A successful titanium etch will completely reveal the gold
In figure 27g), the substrate commences
electroplating by being submerged in a gold bathing solution,
at a current density of 4 mA/cm2, a temperature setting of
50 °C, and 900 rpm on a hot plate. A gold thickness of
approximately 20-25 microns will result after 6-8 hours,
depending upon the area of gold exposed to the gold bath.
After the proper thickness of gold is grown, figure 27h)
shows the SU-8 being removed using Remover PG at 80 °C for 1
If residual resist exists, a sonic bath is employed for
1 min.
In figure 27i), a titanium etch will remove the
remaining top seed layer of titanium, a gold etchtant
removes the unwanted seed layers of gold, and finally, a
titanium etch removes the superfluous seed layers of
Figure 27. Feed substrate fabrication process.
Originally, the devices were fabricated on a 150 μm
thick lithium niobate substrate.
The inconsequential yield
of these devices when fabricated on 150 μm thick substrate
due to the inherent fragile nature creates the desideratum
need of using a 500 μm thick lithium niobate substrate.
is possible to polish the 500 μm thick wafer to
approximately 150 μm by using a Ultrapol End/Edge Polisher.
Failure to polish the wafer to 150 μm will result in
undesirable substrate modes.
An analysis of S21
transmission lines data on a 500 µm thick lithium niobate
substrate showed that more energy was being lost in the
frequency range of 100-110 GHz than was expected.
the substrate down to approximately 250 µm reduced the
amount of energy lost, however, a second polishing reduced
the energy loss to a minimum.
The second polishing reduced
the substrate thickness to approximately 150 µm.
shown in figure 32.
This is
The discrepancy in the smoothness of
the graph stems from the use of different calibrations.
a 500-micron thick wafer is used instead of a 150 μm thick
substrate, a backside polishing process should follow the
etching process described above.
Experimental Setup
To characterize antennas at 95 GHz, a new measurement
setup is created.
Because of the fragile nature of the
fabricated devices, it is very important to create a setup
that won’t destroy the device.
Shown in figure 28, the
device is probed using a WR10-band picoprobe.
In the
frequency range of 75-110 GHz, the picoprobes provide a
maximum insertion loss of 1.75 dB, a maximum return loss of
15 dB, and the ability to reproduce repeatable measurements.
The probes serve as a tool to measure the S11 return loss
and S21 transmission performance of the antenna.
picoprobes connect to WR-10 waveguides, which in turn
connect to a 75-110 GHz Agilent 85104A mmW test set.
mmW test sets are connected to an Agilent 8510C microwave
receiver, which will serve as the source and detector for
all antenna measurements.
Figure 28. Antenna measurement setup.
To create an accurate testing measurement setup, a
standard calibration involving a standard short, open, and
50 ohm load accounts for the losses associated with the
waveguide and probing of devices.
point for all measurements.
This provides a reference
To account for all dielectric
and conductor losses, a transmission line pattern, which
replicates the CPW pattern of the antenna, possessing the
same length, width, and height of the device is fabricated.
By measuring the transmission line losses, it is possible to
determine the amount of energy being radiated by the source.
Once these calibrations and tests occur, it is possible to
quantify the amount of energy being radiated by the antenna.
The primary measurements used to characterize the
antenna at 95 GHz are the use of a S21 transmission
measurement and a radiation pattern measurement.
To ascertain the S21 parameters of the antenna, a known
standard is established.
Using the Friis transmission
equation, which relates the power Pr (delivered to the
receiver load) to the input power of the transmitting
antenna Pt , as a sanity check:
Pr ⎛ λ ⎞
⎟ G0t G0 r ,
Pt ⎝ 4π R ⎠
where λ represents the free-space wavelength, R
corresponds to the observation distance from target, and
G0t , G0 r are the transmitting gain and receiving gain of
their respective antennas.
Employing two 20 dB standard
gain horn antennas to calibrate the VNA will create a
reference point for all other antenna measurements.
the following parameters: R= 30 cm, •=3mm, G0t=100, G0r=100;
the Friis transmission equation predicts the response
expected from the VNA.
This measurement will serve as a
reference point that all other measurements will be compared
To measure the radiation pattern at W-band, current
technologies only measure the main lobe.
This is because
the fragile antenna needs to remain stationary.
Also, since
the picoprobes are also very brittle, it makes sense to keep
the fabricated device stationary.
To measure the radiation
pattern, one of the millimeter-wave test sets is placed on a
goniometer that has a ± 45° range of motion.
In conjunction
with a Newport ESP300 universal motion controller, Labview
possesses the capability to control the goniometer and take
measurements in single degree increments.
This will allow
the characterization of the main lobe of the antenna.
Experimental Measurements
The antenna exhibits a peak gain of 12 dB at 99 GHz,
which can be calculated by noting the difference in
radiation between measured antenna results and those of the
standard gain horn at the same position.
The antenna
displays an average gain of 10 dB in the frequency range of
90-110 GHz, which is shown in figure 29.
This antenna was
fabricated on a 150 μm thick lithium niobate substrate.
Figure 230 shows the S11 return loss performance of the
From the graph, it is apparent that the antenna
achieves broadband operation between 85-110 GHz, which
provides 25 GHz of useful bandwidth.
Also, the return loss
pattern does not exhibit much change with the addition of
the horn.
The importance of the horn can be seen in Figure
29, where the horn provides the structure with an absolute
gain of 8 dB and creates a bandwidth of approximately 25 GHz.
Without the horn, the CPW-fed double-stacked patch antenna
demonstrates 2 dB of gain.
The plot of the “CPW”
measurements results from probing the feed substrate alone
without the use of either the patches or horn.
From these
two figures, it is evident that the antenna obtains a wide
bandwidth and achieves high gain.
A comparison between the radiation patterns of the
standard gain horn antenna (SGHA) and the fabricated antenna
raises room for concern.
If two SGHAs are used to calibrate
the maximum amount of energy transported between two
antennas, then the amount of energy received by a SGHA
should be the consistent regardless of the transmitting
The integrals of the radiation patterns should be
equal, however, the ratio between both the E- and H-plane
and the SGHA are not.
The ratio between the H-plane and the
SGHA is 0.139, which means that 86% of the energy is being
When the same measurement was performed in the E-
plane, the antenna performed better, however, showing a
minor improvement.
Shown in figure 31, the ratio between
the radiation pattern of the antenna in the E-plane to the
SGHA is 0.155.
The E-plane transmits only 15.5% of the
energy that a SGHA would.
The possible sources of energy loss include: waveguideto-probe transition, probe-to-CPW transition, dielectric
losses, conductor losses, and substrate modes.
calibrating the setup properly, the average loss from the
waveguide-to-probe and probe-to-CPW transitions is 0.2 dB.
To test the conductor and dielectric losses of the antenna,
transmission lines were fabricated on 500 µm thick
These transmission lines match the dimensions
of the CPW-feed.
figure 32.
S21 transmission measurements are shown in
From these results, it appears that substrate
modes exist from 100-110 GHz.
This represents a
significant loss of energy.
Figure 29. Transmission response of antenna.
Figure 30. S11 return loss response of antenna.
150 um LN, 95 GHz, CoPol
H- plane
St dHor n
- 38
- 50
- 40
- 30
- 43
- 48
- 53
- 58
- 63
- 68
- 73
- 78
- 83
A n gl e ( D e gr e e s)
Figure 31. Comparison of E- and H-plane radiation with
respect to the standard gain horn antenna.
Figure 32. Evidence for substrate modes.
By compensating for the loss of the transmission line
data by adding loss measured from comparable transmission
line structures to the radiation pattern measurements, the
efficiency of the antenna increased from roughly 14-15% to
approximately 30%.
Figure 33 compares the radiation pattern
of the antenna with and without the loss from the
transmission line considered.
From the figure, it is
apparent that the HPBM is approximately ±25 degrees.
red line represents the radiation pattern between two
standard gain horn antennas.
Since the losses of the
transmission line of the antenna were measured to be 5.03 dB
at 95 GHz, 5.03 dB was added to the response of the antenna
at 95 GHz at every angle measured.
At this particular
frequency, the radiation efficiency of the antenna is 29%.
Table 1 shows the radiation efficiencies of the antenna at
different frequencies and which plane the measurements were
The measurements were performed by finding the
transmission line loss at specific frequencies, adding the
loss to radiation pattern measurements for that frequency,
and then taking the integral of the total energy radiated.
It is clear from the table that the antenna operates better
in the E-plane.
From the various figures, an antenna with a
high gain, wide bandwidth, and a narrow beamwidth has been
Figure 33. Radiation pattern of antenna.
Table 1. Efficiency of antenna in E- and H-plane with
respect to frequency.
75 GHz
80 GHz
85 GHz
90 GHz
95 GHz
100 GHz
105 GHz
110 GHz
At the University of Delaware, two different antennas
have been designed, characterized, and fabricated.
Both of
these antennas utilize a stacked patch configuration to
increase the bandwidth and a waveguide-to-horn transition to
increase the gain.
The main difference between these two
antennas lies in the feed structure.
In the K-band, the
antenna feed structure adopts an aperture-coupled feed to
improve design flexibility and reduce the amount of spurious
back radiation.
At W-band, a CPW-fed square slot creates a
planar configuration and improves impedance matching.
performance of these antennas are determined using a VNA for
return loss and transmission measurements, as well as a
setup that determines the radiation pattern of the
fabricated antennas in the far field.
The designs of these antennas have been characterized
by determining the proper commercially available materials,
requisite fabrication processes, and the effect of different
design parameters on antenna performance.
Through this
description, the advantages and disadvantages of specific
design features have been shown.
Such advantages allow the
construction of a high gain, wide bandwidth, and narrow
beamwidth millimeter-wave antenna, which will be used in
millimeter-wave imaging systems.
There are many considerations for future work to
improve the operation and design of millimeter-wave antennas.
The first option could adopt a ridged SiO2 structure [19].
This structure keeps microwave propagation loss low and
enables a large interaction between RF and optical waves.
second option is to create a backside polishing method for
lithium niobate that will increase the yield and reduce
substrate modes.
The existence of substrate modes adversely
affects the performance of the antenna.
Reference [20]
illustrates the possibility of reducing substrate modes by
incorporating a 2D photonic bandgap structure into the
lithium niobate substrate.
Also, measuring the device
performance in an anechoic chamber will greatly reduce
spurious noise and improve antenna radiation pattern
A future design that incorporates the
aforementioned design improvements with the current
configuration will help to improve millimeter-wave antenna
performance and achieve an acceptable millimeter-wave
imaging system.
[1] Ulaby F.T, R.K. Moore, and A.K. Fung, Microwave Remote
Sensing – Active and Passive, vol. 1, 1 ed. Reading,
Massachusetts: Addison- Wesley, 1981.
[2] Schuetz, C.A. “High-Sensitivity Millimeter-wave
Detection via Optical Upconversion Techniques.” Masters
thesis, University of Delaware, 2005.
[3] C.H. Tsao, Y.M. Hwang, F. Killburg, and F. Dietrich,
“Aperture coupled patch antenna with wide bandwidth and dual
polarization capabilities.“ IEEE Antennas Propagat. Soc.
Symp. Dig., Syracuse, NY, 1988 : 936-939.
[4] Chen, H.D. “Broadband CPW-fed Square Slot Antennas with
a Widened Tuning Stub.” IEEE Transactions on Antennas and
Propagation, No. 8, (August 2003):51, 1982-1986.
[5] Balanis, C.A. Advanced Engineering Electromagnetics.
Hoboken: Wiley, 1989.
[6] Rahmat-Samii Y, L.I. Williams, and R.G. Yoccarino, “The
UCLA Bi-polar Planar-Near-Field Antenna Measurement and
Diagnostics Range,” IEEE Antennas & Propagation Magazine,
Vol. 37, No. 6, December 1995.
[7] Balanis, C.A. Antenna Theory: Analysis and Design.
Hoboken: Wiley, 2005.
[8] Nakar, P.S. “Design of a Compact Microstrip Patch
Antenna for use in Wireless/Cellular Devices.” Masters
Thesis, Florida State University, 2004.
[9] Targonski, S.D. and R.B. Waterhouse, “An aperture
coupled stacked patch antenna with 50% bandwidth”, IEEE
International Symposium on Antennas and Propagation,
Baltimore, MD. 1996.
[10] Croq, F. and D.M. Pozar. “Millimeter-Wave Design of
Wide-Band Aperture-Coupled Stacked Microstrip Antennas.”
IEEE Trans. Antennas Propaga., No. 12, (Dec. 1991):39, 17701776.
[11] Horn, A. “Dielectric constant and loss of selected
grades of Rogers high frequency circuit substrates from 1-50
GHz.” Technical Report 5788. Rogers Corporation. 19 Sept.
[12] Quirk, Michael. Semiconductor Manufacturing Technology.
Columbus: Prentice Hall, 2001.
[13] Ding, X. and A. F. Jacob, “CPW-fed slot antenna with
wide radiating apertures,” Proc. Inst. Elect. Eng/ Microwave
Antennas Propagation, vol. 145, (1998), 104-108.
[14] Soliman, E.A., S. Brebels, P. Delmotte, and G.
Vandenbosch, E. Beyne “Bow-tie slot antenna fed by CPW,”
Electron. Lett., vol. 35, (1999), 514-515.
[15] Soliman, E.A., S. Brebels, E. Beyne, and G. Vandenbosch,
“CPW-fed cusp antenna,” Microwave Opt. Technol. Lett., vol.
22, (1999), 288-290.
[16] Bhobe, A.U., C.L. Holloway, M. Piket-May, and R. Hall,
“Coplanar waveguide fed wideband slot antenna,” Electron.
Lett., vol. 35, (2000), 1340-1342.
[17] Giauffret, L., J.M. Laheurte, and A. Papiernik, “Study
of various shapes of the coupling slot in CPW-fed Microstrip
Antennas.” IEEE Transactions on Antennas and Propagation,
vol. 45, (1997): 4, 642-646.
[18] Schuetz, C.A., C. Huang, R. Shireen, and T.H. Hwang,
“Electro-optic modulator optimization for optically-based
passive millimeter-wave detection.” Newark, DE: University
of Delaware, Dept. of Electrical and Computer Engineering,
[19] Noguchi, K., O. Mitomi, H. Miyazawa, and S. Seki, “A
broadband Ti:LiNbO3 optical modulator with a
ridgestructure.” Journal of Lightwave Technology. Vol. 13,
(1995):6, 1164-1168.
[20] Qian, Y., R. Coccioli, D. Sievenpiper, V. Radisic, E.
Yablonovitch, and T. Itoh, “A microstrip patch antenna using
novel photonic band-gap structures.” Microwave Journal. Vol.
42, (1999): 1, 66.
Без категории
Размер файла
1 347 Кб
Пожаловаться на содержимое документа