close

Вход

Забыли?

вход по аккаунту

?

Parallel scanning microwave microscopy (SMM) with integrated neuromorphic electronics

код для вставкиСкачать
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films
the text directly from the original or copy submitted. Thus, som e thesis and
dissertation copies are in typewriter face, while others may be from any type of
com puter printer.
The quality of th is reproduction is d e p e n d e n t upon th e quality of th e
c o p y subm itted. Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleedthrough, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript
and there are missing pages, these will be noted.
Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and continuing
from left to right in equal sections with small overlaps.
Photographs included in the original manuscript have been reproduced
xerographicaliy in this copy.
Higher quality 6B x 9” black and white
photographic prints are available for any photographs or illustrations appearing
in this copy for an additional charge. Contact UMI directly to order.
Bell & Howell Information and Learning
300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA
800-521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Parallel Scanning Microwave M icroscopy(SM M ) with
Integrated Neuromorphic Electronics
by
DONGCHAN PARK
Submitted in partial fulfillment o f the requirements
for the degree o f Doctor o f Philosophy
Thesis Advisor: Dr. Massood Tabib-Azar
Department o f Electrical Engineering and Computer Science
CA SE WESTERN RESERVE U N IV ER SITY
August, 2000
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 9996320
___
®
UMI
UMI Microform 9996320
Copyright 2001 by Bell & Howell Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
Bell & Howell Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the thesis/dissertation of
P o A /< 5Q M A /
candidate for the
(signed)
(date)
iLl * C
F flR K
FLO
(chair of the committee)
degree *.
-
' ST / 3 * / <2°™
*We also certify that written approval has been obtained for any
proprietary material contained therein.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I giant to Case Western Reserve University the right to use this work, irrespective of
any copyright, for the University’s own purposes without cost to the University or to its
students, agents and employees. I further agree that the University may reproduce and
provide single copies of the work, in any format other than in or from microforms, to
the public for the cost of reproduction.
(sign)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table of Contents
List o f Tables
vi
List o f Figures
vii
Acknowledgem ents
x
Abstract
xi
Chapter 1. Introduction
1.1 Scanning Evanescent Microwave M icroscopy
1.1.1
Background o f Evanescent M icrowave Microscopy
I
1.1.2 Evanescent M icrowave Microscopy: an overview
5
1.1.3 Evanescent M icrowave M icroscopy Application in Semiconductor
15
1.2 Parallel Probe M icrowave Microscopy: an overview
1.2.1
Motivation o f Parallel Probe M icrowave M icroscopy
23
1.2.2 Signal Processing Algorithms in Parallel Probe M icrowave M icroscopy 25
1.2.3
Implementation o f Parallel Probes
1.3 Research Objectives
28
30
Chapter 2. S e lf Oscillating M icrowave Probe
2.1 S e lf Oscillating Probe System
31
2.1.1 Operation o f S elf Oscillating Resonator
33
2.2 D esign o f S e lf Oscillating Probe System
2.2.1 Planar Transmission Line Resonator in S elf Oscillating Probe System
39
2.2.2 Configuration o f S elf Oscillating Probe System
44
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 3. Single Probe Operation
3.1 Fabrication o f Single Self Oscillating Probe System
47
3.2 Characterization o f the Single S elf Oscillating Probe System
3.2.1 Sensitivity and Resolution in EMP
50
3.2.2 Z decay o f Single Self Oscillating Probe System
51
3.2.3 Performance o f Single Self Oscillating Probe System
54
3.3 Frequency Response o f S elf Oscillating Single Probe System
62
Chapter 4. M ultiple S elf Oscillating Probe System
4.1 Signal Processing in Multiple S elf Oscillating Probe System
67
4.1.1 M odeling o f Biological System
69
4.1.2 Correlator
71
4.1.3 Signal Processing Algorithms in Correlator
72
4.2 D esign o f Multiple Probe System
77
4.3 Performance o f Multiple Probe System
4.3.1 Independent Operation o f Multiple Probe System
79
4.3.2 Applying the Signal Processing Algorithms
81
Chapter 5. Conclusion and Future Work
89
Reference
92
Bibliography
96
V
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
Table 1-1 Comparison o f non destructive test method for material characterization
with evanescent microwave microscopy
4
Table 1-2 The equations for the parameter k[,k 2 ,k3 ,k4 in Eqs.( 1-4)
22
Table 3-1 The conductivity and surface resistance o f metal samples
66
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
Figure 1-L The resonant frequency shift o f the EMP in the presence o f a sample
6
Figure 1-2 Microstripline Resonator and Probe
7
Figure 1-3 The different resolution according to structure o f planar type resonator
8
Figure 1-4 Circuit Model o f Resonator
10
(a) Series o f lumped LCR model o f the evanescent microwave probe
(b) Circuit Model in presence o f an insulating sample
(c) Circuit M odel in presence o f conducting sample
Figure 1-5 The different coupling response according to gap size
12
Figure 1-6 A -B Curves vs Voltage
22
Figure 1-7 D oping concentration and EMP Curve
23
Figure 1-8 Application o f Parallel Probe M icrowave M icroscopy
25
Figure 1-9 Structure o f motion sensitive neuron having correlation type retina
to process an input signal
27
Figure 2-1 Schematic o f the self oscillating resonator and it’s circuit model
32
Figure 2-2 The change o f frequency o f self oscillating probe system due to
metallic sample(Brass). Left signal is the shifted one due to brass
35
Figure 2-3 The oscillation spectra o f self oscillating resonator
36
Figure 2-4 The previous experimental setup having equipment for single probe
37
Figure 2-5 Four different type o f planar transmission line
40
Figure 2-6 Electric field distribution in coplanar waveguide
41
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 2-7 The various configuration o f feedback loop for oscillator
42
(a) General configuration o f impedance element
(b) Hartley (c) Coplitts (d) Clapp
Figure 2-8 (a) Lossless transmission line with voltage generator
43
(c) Equivalent circuit model with lump element
Figure 2-9 The end o f tungsten tip for S e lf Oscillating Probe
44
Figure 2-10 Model o f Oscillator concept
46
Figure 2 -1 1 Schematic o f Single S e lf Oscillating Probe System
46
Figure 3-1 The Single S e lf Oscillating Probe System
48
Figure 3-2 The W aveform o f output signal coming from the mixer
49
Figure 3-3 The experiment setup, which include step motors for the three axis,
the probe holder, frequency counter, and oscilloscope
52
Figure 3-4 The front panel for operation o f data acquisition o f Labview Program
53
Figure 3-5 The z-decay o f Single S e lf Oscillating Probe System
54
Figure 3-6 The actual image o f line pattern
55
Figure 3-7 The EMP image o f line pattern
55
Figure 3-8 The 100pm diameter tungsten wire
56
Figure 3-9 The 3D EMP image o f 100pm tungsten wire
57
Figure 3-10 The 2D EMP data map o f 100pm tungsten wire
58
Figure 3-11 The SEM image o f the gap between feedline and resonator
59
Figure 3-12 The 2D EMP image o f the gap between feedline and resonator
60
Figure 3-13 The 3D EMP image o f the gap between feedline and resonator
61
Figure 3-14 The frequency response for the sample which has a different
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dielectric constant
Figure 3-15 The profile o f frequency change resulting from dielectric material
Figure 3-16 The frequency response for conductive material (Metal)
Figure 4-1 The im age o f fly ’s eye
Figure 4-2 M odel o f fly ’s retina
Figure 4-3 (a) Schematic o f correlator for both scanning direction
(b) Schematic o f correlator for the scanning direction moving
left to right
(d) Schematic o f correlator for the scanning direction moving
right to left, in case we use the same direction with (b),
the output is null.
Figure 4-4 Signal processing in a correlator
Figure 4-5 Top view o f multiple probe system
Figure 4-6 Side view o f multiple probe system
Figure 4-7 The image o f sample which has a different dielectric constant spot
Figure 4-8 The scanning signal from each probe o f multiple probe system
Figure 4-9 The image o f sample having the two holes
Figure 4-10 The exam ple data for correlation from four holes sample
Figure 4-11 The raw scanning data from the multiple probe system
Figure 4-12 The delayed signal for the data from one o f the multiple probe
Figure 4-13 The correlation between the signals from multiple probe system
Figure 4-14 The single output signal from the different sample
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
I am indebted to many people to complete this dissertation. Especially, I would like
to express m y thanks to my advisor, professor M assood Tabib-Azar, who had given me a
lot o f support and an opportunity to fulfil my dissertation work successfully, and whose
knowledge and dedication to the work have encouraged and guided me through the whole
dissertation work. I also appreciate sincerely for the com m ittee member’s valuable
comment and suggestion. Without their technical support and suggestion, it would have
been very hard to com plete this dissertation. I also thank all o f m y labmate for the useful
discussion and help.
Finally, it is also very thankful that my mom, w ife, sister, and all o f my family
member always support and help me too much. I can not forget their love and assistance
in many ways. Very sincerely, I deeply appreciate for their love and support.
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Parallel Scanning M icrowave M icroscopy(SM M ) with
Integrated Neuromorphic Electronics
Abstract
by
DONGCHAN PARK
During the last years, the usefulness o f m icrowave microscopy has been shown by
many researches. According to the research regarding evanescent microwave
microscopy, it is obvious that scanning microwave microscopy have a great potential for
variety o f applications while a remarkable high resolution could be achieved by
improvement in terms o f mechanical and a circuit’s point o f view.
M axim izing the area scanned at one time by using a multiple probe system rather than a
single probe, could turn the EMP into a more time efficient and more effective scanning
tool. In other words, to scan more area means more data result from more scanned area.
Moreover, the setup and data acquisition is substantially complicated.
The experimental setup o f multiple probe is going to be big in terms o f size o f equipment,
because each probe needs the equipment which make signal process possible, for
example, m icrowave signal generator, lock in amplifier and circulator. Therefore, to
make use o f the equipment like signal generator for each probe is not a good approach in
terms o f size and cost.
The w ay to solve this problem is to design compact probing system, which can be
run without a big size and high cost equipment. The se lf oscillating probe provide a good
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
substitution for multiple probe system. Without loss o f EMP operational principle, we
may use it for evanescent microwave imaging and scanning system.
Apparently, w e have more data to be processed using multiple probe rather than a single
probe in a certain time. In other words, w e need more efficient approach to process large
amount o f data. One o f the approaches we may imagine is to use the algorithms coming
from biological system. It is very w ell known that a biological system, for example,
insect, has a very fast signal processing mechanisms comparing to general electronic
systems. And the signal processing mechanism allows multiple input signal process to be
possible. The fly’s landing system could be seen as a good example o f a multiple input
signal process. The fly’s retina which consists o f a movement detector o f correlation
structure physiologically implement a multiple input signal process. To make use o f this
kind o f biological signal process in a microwave microscopy system, a good approach
could be to build a multiple parallel probe scanning microwave microscopy system.
B eing scanning a material with parallel probes, we get more information about a material
in less time. One o f example is to detect a defected area or an area which have a different
properties in a material. For example, a thin film deposited with a specific material need
to be checked out if the specific material is deposited completely. Otherwise, how much
amount o f area is defected by a deposition process is need to be verified. In this case, we
may determine if the scanned area is defected using the output signal produced by a
biological signal processing algorithm like fly ’s landing system proposed by many
researchers.
W e design the parallel probes and implement a signal processing algorithms based
on a biological system, fly’s landing system. W e integrate them and add up data
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
acquisition program and make it parallel scanning system . Then, w e determine if there is
a defect in a material using the single output data from scanning system.
In the research, w e design the self oscillating the probes for parallel scanning evanescent
microwave application. For multiple input system , the mentioned biological system is
implemented by electronic circuits which covers a high frequency microwave signal.
It is applicable to manufacturing line that scanning and analyzing a material moving on a
conveyer. And the proposed modeling regarding the relation between evanescent
microwave signal and doping concentration show the direct translation o f evanescent
m icrowave signal in terms o f doping concentration.
xiii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 1
Introduction
The follow ing chapter describes a parallel scanning m icrowave microscopy and the
motivation for developing a parallel probe system and how it can be implemented.
A lso, the advantages o f a parallel probe system are described. Furthermore, we review
the general principles o f evanescent microwave microscopy, historical background and
its applications.
1.1 Scanning Evanescent M icrowave M icroscopy
1.1.1
Background o f Evanescent M icrowave M icroscopy
The versatility o f Evanescent M icrowave M icroscopy is incredibly increased in
non-destructive characterization o f materials. In a number o f important applications, it is
very critical to characterize a material nondestructively. For instance, the detection o f
defects and variations in the carrier and doping concentration with submicron resolution
in semiconductors is a very important example the nondestructive characterization can
apply. In addition, the detection o f defects and variations in resistivity in conducting, and
superconducting thin films is also one o f the important applications. Other application
include testing the continuity o f embedded transmission lines in high density integrated
electronics (IC) and printed circuit boards, and mapping the bulk permittivity profile o f
dielectrics. Imaging the moisture and mineral content o f both biological tissue and
botanical samples such as plant leaves and wood are also good applications.
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
S y n g e[l] proposed Near-field microscopy in 1928 at optical frequencies and Ash
and his colleagues[2,3] demonstrated it in the 1970’s at m icrowave frequencies. They
could image m etallic gratings with periods o f A7200, and detect fatigue cracks less than
2pm in width on planar metallic surfaces[3]. This work initiate near-field evanescent
microwave m icroscopy research [4-9]. A lateral resolution o f 77500 to detect lateral
conductivity variation in semiconductor was published [4]. A t 7.5GHz, the quantitative
image o f sheet resistance o f thin films with a resolution o f 0.6Q /cm for 1000/cm could
be obtained using m icrowave microscopy [5]. And mapping the sheet resistance o f thin at
80GHz [6] and the highest resolution o f 7/5,000,000[7,8] in contact mode reported.
Dr.Tabib-Azar has used evanescent microwave to characterize a wide variety o f
material such as semiconductors, metals, insulators, and biological samples using the
planar type resonator as a probe. [9-13].
Various type o f resonator have been used for evanescent microwave microscopy
in Dr.Tabib-Azar’s group to prove the performance o f evanescent microwave
microscopy. For instance, open resonators, rectangular waveguide resonators with an
aperture in the end plane, the center conductor o f coaxial transmission line, a coaxial
transmission line resonator with a sharpened metal tip connected to and extending out
from the center conductor, microstrip, stripline, and coplanar waveguide resonators with
a thin cylindrical wire attached to the open end have been used to implement the
Evanescent M icrowave M icroscopy. These probes have a limitation in terms o f resolution
due to dim ensions o f the probe tip, the quality factor o f resonator and so on [14],
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
There are several different methods to image the properties o f material using
evanescent w ave[15-18]. For example, Scanning Tunneling M icroscopy(STM ), Atomic
Force M icroscopy(AFM ), and Near-field Scanning Optical M icroscopy(NSOM ) are
some example. STM uses evanescent electronic wave functions to image atoms at the
surface o f conducting materials. AFM uses interaction between near-field electronic
wave functions and atomic cores to image atoms at the surface o f insulators. NSOM uses
an evanescent optical field to image optical properties o f materials.
M eanwhile Evanescent M icrowave Microscopy or Evanescent Microwave Probe
(EMM or EMP) uses evanescent microwaves to image the microwave properties such as
conductivity, permittivity and permeability o f materials.
Evanescent microwave can be used in various applications with a resolution
range o f around 0.1pm -1cm . One o f the advantages is to image or scan sub-surface o f
conducting material due to skin depth effect o f microwave itself. And, since this method
is completely non destructive and non-contact to the scanning sample, the possibility to
extend the application is numerous.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
M ethod
Scanning
Tunneling
Microscope
(STM )15
Resolution
Atomic level.
Conductivity Lim its
Good electron or
ionic conductivity
required.
Atomic Force
Microscope
(AFM )16
Crystallized hard
material: atomic.
Biological: 2 nm.
No requirements on
conductivity.
Optical
M icroscopy17
1-10 nm
N o requirements on
conductivity.
Scanning
Electron
Microscope
(SEM)
100 nm.
Limited to materials
that are conductive.
X-ray18
5 pm (subpm with
synchrotron).
N o limitations.
Evanescent
Microwave
.4 pm at I GHz
Penetration depth
limits bulk probing
in metals.
Eddy
Current19
50 pm.
Sample should be
conducting.
Ultrasonic20
Order o f 1 mm.
No requirements.
Comments
No free electrons involved
so can be conducted in
air/liquid/vacuum. Field o f
view is only a few pm2. 300
by 300nm area scan takes 10
min.
Surface preparation
required.
Both contact and noncontact methods exist.
For high resolution the probe
needs to be a few nms from
the sample. A 250pm x 250
pm scan takes 30 s.
Vacuum sample preparation.
Charging in non-conductive
can be avoided by using thin
metal layer. Expensive
instruments.
Poor sensitivity to the
surface.
Sample preparation required.
Expensive and huge
equipment.
Good for large-scale
mapping.
Hot and moving samples can
be imaged. No sample
preparation. Can be used in
air/liquid/vacuum.
Cannot detect planar cracks
in the plane o f the eddy.
Complicated coil designs for
some applications.
Intimate coupling required
due to poor transmission
over boundaries. Not useful
at high temperature.
T a b l e 1-1: C o m p a r i s o n o f n on d e s t r u c t i v e t e s t m e t h o d for m a t e r i a l
c h a r a c t e r i z a t i o n w i t h th e e v a n e s c e n t m i c r o w a v e m i c r o s c o p e , (f r o m [ 1 1 ] )
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
1.1.2
Evanescent M icrowave Microscopy: an overview
Evanescent M icrowave M icroscopy has been researched to inspect a material
non-destructively using the property o f resonant frequency shifting according to the
material to be presented under the microwave probe.
The resonator frequency can be perturbed by the presence o f the material to be scanned.
The change in the resonator frequency mainly depends on the electric permittivity,
magnetic permeability, and conductivity. These parameters are com plex quantities that
depend on different properties such as the composition o f the material. Therefore, the
category o f material scanned is wide. The resonator in Evanescent M icrowave
M icroscopy (EMM) uses the microwave stripline as a probing tool, which is based on
microwave transmission line theory. In general, at microwave frequency, to use lump
element is nontrivial or unusual in practice, because the lumped elem ent w ill dissipate a
lot o f energy by radiation. Nevertheless, transmission line is a bounded media that can
contain and transfer electromagnetic field energy with little loss.
Therefore, transmission line can be a great tool to treat a microwave signal application.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
■4
■8
-12
-16
AS
-AS
00
-a
■3o
Metal
"5.
-28
-32
-36
Center 9.850 GHz
Span 20 MHz
Figure. 1-1. The resonant frequency shift o f the EMP in the presence o f a sample.
Usually, the length o f transmission line is an integer multiple o f A./2 for open circuit
transmission line or an odd multiple o f A./4 for short circuited line. A. is the wavelength of
the electromagnetic wave.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
Microstrip Feedline
Microstrip Resonator
Effect of
Cr=C3lKC1+C2 >
t
■AAIV
Optional
wire tip
Figure. 1-2. Microstripline Resonator and Probe
If w e can design a resonator that has a high quality factor, w e may use the resonator
to get a high resolution. T o get a high quality factor from a resonator, we need to have a
critical coupling or impedance matching between the feedline and resonator. If it is
uncoupled, it is hard to get a high peak resonance signal, which has a high quality factor
such as 1700 Q-factor.
In general, to use a planar type resonator is reasonable approach in EMP. In the
fam ily o f planar type resonators, there are several different types, which include
microstrip line, coplanar waveguide, and shielded coplanar w aveguide. Among these
types o f resonators, a coplanar waveguide type resonator produce better performance in
terms o f lateral resolution because coplanar type offers better confining o f
electromagnetic fields. T o increase resolution, there are several approaches. One is
tapering the end of the probe tip. To make sharp the end o f the tip is a good way to get a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
high resolution. T he others are to use thinner and shorter tip and to reduce the distance
between the sample and tip as much as possible w ithout touching the sample.
Spatial Resolution -> 20um - 65um
1.0 -
Shielded CPW
0 .8 -
>
CPW
«
E
h_
o
z
0 .4
-
microstrip
0 .2 -
0 .0
-
0
50
100
150
200
250
300
distance (um)
Figure 1-3. The different resolution according to structure o f planar type resonator [19]
Theoretically, w e can prove that the shift o f frequency is due to the properties o f the
material. In other words, i f w e have a change o f permittivity and permeability, then we
have a change o f frequency. Using quasi static perturbation analysis, w e may formulate
the field interaction due to the change o f property o f material. The Eqs.(l-l) show the
relationship betw een the change o f properties o f the material such as permittivity and
permeability, and a change o f frequency.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
Af _
f
Ei, Hi,
Jv(A££i* E o+ AfxH\- Ho)dv
! v(sEo * Eo)dv
( 1 - 1)
and Eo, Ho are the perturbed fields (fi) and unperturbed fields (fo)
respectively. In other words, w e may estimate the sensitivity o f EMP in terms o f change
o f material property by using the relationship. The change o f frequency is proportional to
the change o f material property. And, the other way to see this kind o f relationship is in
the E q s.(l-2).
This equation show how the parameters such as permittivity and conductivity
effects the sensitivity o f EMP. Because the input admittance (Y l) can be converted to the
impedance, the impedance is a parameter w hich determines reflection coefficient.
( 1- 2 )
l n ( T / 6 ) - l ^ P ; + ( l + y ) 2J
'
\ n ( b / c ) [ P 2 + (l + / ) \
g,„r l n ( 7 V 6 ) - 1
s
In( 6 / c )
cr
The parameters T, b, C are probe length, the distance between the probe and the
sample, and the radius o f the wire respectively, and the parameters o s, e, are the
conductivity o f the sample and the permittivity o f the insulator.
The planar type resonators such as microstrip, and coplanar waveguide in EMP do
not have a simple frequency response in EMP. Therefore, to predict the behavior o f the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
resonator in EMP, w e may make it model with a simple L,C,R circuit With an LCR
circuit, we may represent the probe as in Figure. 1-4.
R
L
—
vw-
‘
(a)
to ta l
R
Cc
~wv-
‘ to ta l
Cs
R
-W r-
Rs
Cb)
Cc
-lb
I Ls
*to ta l
(c)
i Rs
Figure. 1-4. Circuit M odel o f Resonator
(a) Series o f lumped LCR model o f the evanescent microwave probe.
(b) Circuit model in presence o f an insulating sample.
(c) Circuit model in presence o f a conducting sample.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
The Cc means the coupling capacitance resulted from the distance betw een the probe
tip and material sample to be scanned.
The Coupling Capacitance couples the microwave properties o f sam ple to the planar
type resonator that can be m odeled with an LCR circuit, w hile the Cs and Rs represent
the capacitance and resistance o f the sample material respectively.
The planar type m icrow ave resonator can be excited by coupling it to a feedline,
changing its input impedance by a small amount. It makes a change o f the feedline
characteristic impedance that is a kind o f load at input o f resonator.
The coupling can be m odeled w ith resistive and reactive elem ent such as a capacitive pi
network as shown in figure. 1-2. To get a high performance in terms o f quality factor, we
have to ensure setting a critical coupling. At the condition that the critical coupling
occurs, we can have a maximum power transfer from the feedline to the resonator.
Adjusting the distance o f gap betw een the feedline and resonator is a w a y to ensure the
critical coupling.
I f the gap is small, it w ill be overcoupled and i f the gap is too wide, it w ill be
undercoupled, as illustrated in figure. 1-5.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
-5 -1 0
-
-1 5 -2 0
-
Undercoupled
Overcoupled
-25 -3 0 -3 5 •40 -
Critically C o u p le d
-4 5 -50
1 0 .2 0
10.25
10.35
10 .3 0
1 0 .4 0
10.45
F r e q u e n c y ( GHz )
Figure. 1-5 The different coupling response according to gap size [19].
The E q s.(l-3) and (1-4) show that the intrinsic resonance frequency shift occur
from the presence o f sample material.
cc
co - co 1 - 0 . 5 - ------ £_£o
C0V
(C5 + C s ')J
f
6 ) '= CO
(
c 1
1 —0 .5 —^ 1
C V
V
0
= - 0 . 258co
for insulator (1-3)
C C
s c
)
C0 (C
+ C s ')J
v c
for metal (1-4)
C C
s c
cov(Cc + C s ') J
Clearly, the equations show that the shift o f resonance frequency depends on the
capacitance and resistance o f sample.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
As w e mentioned before, the quality factor is very important in EMP
performance, because a high Q factor represent the high resolution o f the EMP.
The equation for the Q factor in conceptual points o f view is as E qs.(l-5)
Q s
0 ( Energy Stored ) _ & QU
Average
Power
Loss
wL
(1-5)
To make it more useful, w e may modify the equation as E q s(l-6), this is a more
informative form.
Q = 2 AWW '
fo
.
( 1-6)
2 A f
where 2 A f is bandwidth. Therefore, it is clear that the Q-factor w ill be high if f is small
when the resonance frequency fo is fixed. The higher Q w ill provide a more sensitive
EMP. The total impedance is given by:
total
= R '+ j
1
coL '-
co C '
r
= R '
l + JQ '
\
\
co
CO
«
o
CO
(1-7)
In E q s(l-7), the R', L' and C' represent the sample capacitance, resistance and inductance.
These parameters depend on the properties o f the material such as permittivity and
conductivity. In other words, to see the relationship between these parameters and the
shift o f resonance is one approach to appreciate the EMP image technology.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
D ue to the presence o f material, w e have also a little bit o f a changing quality factor
represented by Q' in E qs.(l-8).
0' =
~
( 1- 8 )
qj ' C
R'
Since w e have the dielectric losses or conductor losses occurred from the presence
o f sample near the probe, the quality factor w ill be dropped.
Therefore, it is apparent that the shift o f resonance frequency and quality factor is very
clo sely related to the properties o f sample material such as permittivity and conductivity.
It means that this factor heavily effect on EMP output directly.
N ow , it is clear that the EMP output depends on the resonance frequency shift and
the resonator's quality factor. Then, i f w e can formulate the relationship between EMP
output and shift o f resonance or quality factor, it w ill provide us information how
evanescent m icrowave microscopy works. As a matter o f fact, the EMP output can be
equated as show n in E qs.(l-9), using scattering parameter o f its resonator
EMP = S „ V in
(1-9)
where V in is the microwave input to the probe and is constant. And, it is w ell known that
the scattering parameter can be expressed in terms o f the impedance o f the transmission
line like E qs(l-lO ).
And, w e may get the E q s ( l- ll) or (1-12) from combining E qs.(l-9) and E qs.(l-lO )
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in
or
50(V'.n + E M P )
Z,otal
(M2)
Vin —E M P
where Zo is the characteristic impedance o f the microstrip line (=50Q ).
In E q s.(l-9) and E q s.(l-lO ), w e may see that the EMP output can be changed due to the
properties o f materials, since Ztotai is related to the quality factor and scattering parameter
is directly related to EMP output.
1.1.3
Evanescent M icrowave M icroscopy Application in Semiconductor
EMP could have a various application on semiconductor characterization. One o f
them is to measure the carrier life tim e[9,10] using the reflection coefficient type EMP
output. Measuring the doping profile could also be one o f the EMP applications in the
area o f semiconductors.
If w e can setup the equation regarding the relationship between EMP and doping
concentration, w e can characterize the doping concentration at a spot in the
semiconductor material.
In order to derive the relation for the doping concentration, w e need to make a model
o f the microstripline probe that is coupled with the sample to figure out the total
impedance.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
The EMP probe which is shown in figure. 1-2 and the sample can be modeled by a
resistor, capacitor and inductor like figure. 1-3.
Cc is the coupling capacitance, Cs is the surface capacitance o f sample, Rs is surface
resistance o f sample. Co and Lo are the capacitance and the inductance o f the
microstripline per unit length o f the probe.
Total impedance is given by Eqs. (1-13) which is reported previously.
r
^ to ta l ~
+
'eff
- + j<o Lq ~ —
Geff +co Ceff
Ge/ + c o 2C„
(1-13)
where Lo is the unit-length inductance o f the stripline for probe , and co is the angular
frequency; CefrandGefr are given by:
C^r
C„ +
'eff -= ^0
C C
c ,+c.
c
s
C C
= C a (1 + —
-)
iC e + C , ) C 0
' CQ+ C c '
G e ff = G s
(1-14)
(1-15)
KC s + C cJ
Cc
Cair ~F Cw
Eo-A-eff/d "FSsAeff/w
Cs EsSs
(1-16)
(1-17)
where Aeff is area o f semiconductor that probe covers, £, is the smallest skin depth or the
decay length o f the electromagnetic waves at the tip o f the probe.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
In addition, Cair is the capacitance in air, d is the distance betw een the probe and sample,
Cw is the capacitance for the depletion region, and w is the depletion length o f sample.
The depletion length is defined by w = ^ 2 £ s. V I qn , where n is the surface doping
density. The skin depth is given by 5= 1/ ^7cffj.crs , d is the distance between the
microstripline probe and the sample.
To relate the reflectance o f the microstripline resonator to its impedance, w e have
r = s„
(i-is )
total
^ total
=
+
0
^_ p
(1-19)
where T (=Si 0 is the reflection co effic ien t, Zo is the characteristic impedance o f the
feed-line.
Experimentally, w e can get the reflection coefficient from the EMP measurement.
Then, from E qs.(l-18), w e can obtain the real value o f total impedance, ie, Z totai •
If w e know the total impedance from the experimental reflection coefficient, the total
impedance can be expressed in term o f Cs and Rs. These variables are dependent on the
doping concentrations because w e may describe Cs and Rs in terms o f it.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
18
Cs=esgs=esAeff/£ = SsAeff
= ssAcff -yj7f/jq/Jsn
Rs = l/G s= l/( g sa s) = l/5 a s= A/ ^ u / 4 qM s11
( 1- 20 )
G-21)
W e can m odify the E q s.(l-13 ) to express it in terms o f Cs and Rs.
’'ff
Z' total = R 0n +■
2
T_
G * +<a~Ceff
t
L0
~>r
'eff
r-r 2
v
2/’■'*
+ < a
y
—R0 + A. + j(coLo-B) — Ro "FjcoL0 ■+■A - jB
K —
+ co C eJf
= (Co + C c )2 G s 2 + 0 ) \ C o + C c) 2C s2 + 2 C oC c0 ) \ C o + Cc)C, + <o2C 02C 2
A=
G -22)
>/r
_
G s (C 0 + C c)(Cs + C c)
K
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(1-23)
(1-24)
19
B=
coC«ff
_ co(Cc + C „ ) C s- +co(2C0Cc + C c~)Cs +coC„CcK
G 'ff~ + 0 ) 2 c <ff
(1-25)
If we plug the E q s.(l-20) into E q s.(l-23) and (1-24), then, we can get the equation that
are expressed in term o f n, doping concentration.
A'= Geff = ai n + ai n'0.5
(1-26)
a.j — (C 0 + C c)(q|usEsi Aeff)
d -2 7 )
a2 = (C0 + Cc)Cc ( V ^ : / V ^ )
d -2 8 )
B ’= coCeff
''ff = bi n + b2 n 0 - 5 + b3
(1-29)
bi = co(C0 + Cc) (ssi 2 Aeff2 Tcfjj.q)is )
(1-30)
b2 = co(2CcC 0 + Cc2) (esi Aeff) ^rfjuqn,
(1-31)
b3 —©CoCc
(1-32)
K = G eff 2 + co1Ceff1 = ci n + C2 n + C3 n 0 5 + c 4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(1-33)
20
Ci = (C0 + Cc)2 (q(Os/7rf(j.)
(1 -34)
C2=CD-(C0 + Cc) - ( e si-Aeff-7tf|jq|as )
(1-35)
c3 = 2C0Ccco2(C o+Cc) (8si Aeff)
( 1-36)
c4 = arC0~Cc~
(1-37)
Therefore, the equation for the total impedance is
^ total
'off
R0 + „ 2 ,
2^ — +JO)
Geff + c o C ,off
L°
c
O ff
Ge/ + 0 ) 2C off
= Rq + A + j(coLo-B) = R q + jooLo + A - Jb
—Rq + j OJLq +
bxn + b2-Jn + b 3
a xn + a 2
c xn + c zn + cz 'Jn + c 4
-J
d -3 8 )
yfn c,« + c,n + c 3-v/«+c
To simplify the previous equations, we need to make some approximations to ease
the calculation needed for getting doping concentration.
W e drive the Eqs. (1-23), (1-24) in terms o f function o f voltage to evaluate which
equation is dominant in the total impedance.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
21
The value of Eqs. (1-24) is much bigger than that o f E q s.(l-2 3 ) as shown in figure 1-6.
Therefore, we can approximate the equation more simply.
Ztotal ~ JCOLq - j
b}n + b2-Jn + b3
b^n + b3
~ jcoL0 - j ------------------c,n + c 2n + c 4
d-39)
(1-40)
From the experiment, we can get the absolute value o f Zt0tai and the reflection coefficient,
The E qs.(l-38) can be rearranged with its magnitude..
In Eqs.(1-38), if w e evaluate the constant coefficient o f each term, we can obtain a more
simple equation. Through this approximation, we convert equation(l-38) to the equation
for the doping concentration.
(1-41)
^ 3 % total
The Z,otai is dependent on voltage, so we can get a doping profile and concentration that is
dependent on voltage. The constant ki,k 2 ,k3 ,k4 com e from the combination
o f the coefficient o f the E q s.(l-3 8 ),(l-4 0 )fo r real part and imaginary part o f ZtotaiEven if the equation is approximated by the reasonable assumption, the equation show us
the sturdy consequence overall about the dopant profile that is shown by Fig 1-7.
The span o f the dopant profile is ranged fromlO10 to 1016 approximately.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
The result is very similar to the general doping profiling.
k[
k3
oo2C02Cc2
-<aC0Cc2- 50 co2C02Cc2
-[(C0 -h Cc)2 (qps/rfp) + co2(C0+ Cc)2 (8s,- 2Aeff27rf'(4qps )]
k4
50[(Co + Cc)2 (qps/Ttfp) + co2(C0+ C c)2 (esi 2Aeff27rfpq|4s )]- 0)(Co + Cc) (eSI- 2Aeff27rfpqps )
Table 1-2. The equations for the parameter ki,k 2 ,k.3 ,k4 in Eqs.(l-41).
4 8 .4
4.00E -04
48.3
3.50E -04
3.00E -04
4 8 .2
e
2.50E -04
CJ
48.1
oo
U
48
<
47.9
2.00E -04
£
1.50E -04
ca
1.00E-04
47.8
5.00E -05
47.7
0
?
4
6
8
10
Bias V o lta g e (V)
Figure 1-6. A-B Curve vs Voltage
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
a
o
'co
I t
o
U
©0
a
• pN
CL
O
Q
1.0E+16
1.0E+15
1.0E+14
1.0E+13
1.0E+12
1.0E+11
1.0E+10
1.0E+09
1.0E+08
1.0E+07
1.0E+06
1.0E+05
1.0E+04
-
0.8
0.7
Inversion
VFB depletion
0.6
Doping Profile
f
A
0.5
0.4 >
EMP curve
0.3
\
/'
\
'
I
/
0.2
7
S
ed
0.1
y
< .
a.
0
p
-0.1
0.6
-
0.1
V ooltage
1.4
1.9
Figure. 1-7. Doping Concentration and EMP Curve
1.2 Parallel Probe Microwave Microscopy: an overview
1.2.1 Motivation o f Parallel Probe Microwave M icroscopy
The EMP can operate at very high frequencies (around 1GHz to 10GHz ), which
means that EMP has a high scanning speed. Furthermore if we can add up more probes to
the system, w e could achieve a better performance in terms o f scanning speed. For
instance, if w e have a thin film deposited immediately after deposited from a deposition
chamber and conveyers, then we could have a lot o f material to be inspected regardless o f
if the thin film is flawless or not. Ln general, we expect that most o f them are flawless, if
we use the m odem process facility, but it may be not perfect. In this circumstance, if we
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
have to inspect the material, adding up more probes could be a solution to scan a larger
area in less time. It is obvious that w e have more data to process or scan more area in a
limited time with multiple probe than that o f single probe.
In general, to save time or to scan larger area in less time can be a important factor in
terms o f efficiency.
However, an increasing data size is a kind o f disadvantage, because the amount o f data to
be processed will be larger. In other words, signal processing time will be longer.
For example one probe has a 1cm /sec scanning speed and a 10 pm resolution for a
scanning point, meaning that to scan 1cm, w e need lOOOpoints.If we implement 10
probes in parallel, w e’ll obtain 104points, if the resolution o f DAQ is 12 bit for
1 point, then total amount o f data w ill be 0.12M bit(=12bit x 104points).
If w e can find a solution to the signal processing issue, the parallel probe system is a
good alternative for scanning larger areas with EMP. As a signal processing solution,
one more thing to be considered is that the material to be inspected is possibly almost
flawless under the m odem fabrication process circumstance, it means that even if w e scan
a large area, it can not be necessary all the time.
Therefore, if w e can see the response only when the probes detect a defect or defected
area, then it will be very efficient signal processing algorithm.
£
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
Correlator
Parallel Probes
PLD(PuIsed Laser
Deposition) Chamber
YBCO F Im
Y
'
M o v in g d irec tio n
■
F igure.l-8. Application o f Parallel Probe Microwave M icroscopy
1.2.2 Signal Processing Algorithms in Parallel Probe M icrowave Microscopy
The signal processing algorithms w e can apply to parallel probe microwave
m icroscopy are based on the neuromorphic algorithms which are aspired by biological
system such as that o f an insect.
Usually, a biological system such as a fly has faster signal processing algorithms than
that o f general electronic system. [20]
Between the two systems, a biological system and a general electronic system, it is the
different primitives o f computation that make biological signal processing so efficient
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
and effective compared to the signal processing approaches used by the general electronic
system such as digital electronic system.
It is w ell known that biological signal processing systems operates on com pletely
different computational principles from the general electronic systems. [20]
It is based on relative grades inherent to fuzzy logic. This is the reason why humans may
take several seconds in performing a numerical computation, but require only a fraction
o f a second to recognize complex patterns. Biological system can routinely perform
computations, such as pattern recognition and visual motion.
This kind o f computation can be implemented by the parallel computational power o f the
neural networks and electronic circuits[2 1,22].
The representative example o f a biological system which has been used for a specific
purpose system such as vision system is a fly ’s retina and its signal processing
algorithms.
For a long time, the structure o f fly ’s retina has been researched, and, the
details o f the structure are well known. The attempt to implement a fly’s retina that is
applicable to vision system has been performed by researchers.[23,24]
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
Retina
Visual
ganglia
Central
brain
Cervical
conn ective
Thoracic
ganglion
Figure. 1- 9. Structure o f motion sensitive neuron having correlation type retina to
process an input signal
Basically, the fly ’s retina has a correlation type structure, which reduces signal
processing time and the amount o f data using the property o f correlation between
adjacent data. W e may distinguish the signal with correlation from others.
If we apply the parallel computation approach to a parallel probe system, the
efficiency in judging the output data as well as high speed computation is very good
point to Evanescent Microwave Microscopy in terms o f scanning speed and area.
High speed computation produces the output through highly efficient computational
algorithm. The smaller number o f output data make us judge about the material scanned
using biological computation algorithms, because w e may have a integrated signal to the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
multiple inputs as a output. It means the number o f output channel is less than that o f
input channel. Then, the size o f hardware as w ell as computation time could be reduced
to m eet the specification o f performance .
For exam ple, for 10 input signal to be processed, w e can determine or analyze the source
o f input signal using one output signal, it w ill be very efficient and powerful algorithms
and system s. Neuromorphic electronics make the com plex computation process
sim ple and concise. In general, in multiple input system such as parallel probe scanning
system , the processing o f the signal that com es from the probes can be performed using
biological computational algorithms.
1.2.3 Implementation o f Parallel Probes
To implement a multiple input microwave microscopy system, using the previous
EMP experiment setup is not an efficient approach in terms o f cost and size. Because, the
previous single probe evanescent microwave m icroscopy system depends on several
instrument to perform the signal process such as lock in amplifier, a microwave signal
generator, crystal detector and so on. For the parallel probe or multiple input probe
system , a new approach which eliminates the need o f equipment like a lock in amplifier
and that is more compact needs to be proposed.
The previous setup consisted o f a microwave resonator coupled to a feed line and
connected to a circulator. The circulator was also connected to a 0.01-1.05 GHz signal
generator, and to a crystal microwave detector. The detector output was proportional to
the magnitude o f the reflected wave. This voltage is fed to an amplifier and to a lock-in
amplifier. Essentially the lock-in amplifier measures an AC voltage and gives an output
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
in the form o f a DC voltage proportional to the value o f the AC signal being measured. In
other word, the lock in amplifer provide dc signal which is proportional to amplitude of
the reflected signal at specific frequency. The probe was mounted horizontally over a x-y
stage that contained the sample. Sinusoidal amplitude modulation o f the microwave
signal is used for performing synchronous measurement o f the EMP output using the
lock-in amplifier.
The signal fed into the computer is a dc type o f signal, which com e from the lock-in
amplifier.
The Signal Generator was used as a microwave signal generator which provide around
1GHz signal. The detector that was used to filter the reflected signal and produce a DC
type signal as an output, is operated between 0.1 GHz and 18 GHz .
For a parallel probe system, the effective way to eliminate the need for equipment is
to design self oscillating resonator. If we design a se lf oscillating resonator, we may
remove signal generator from the experimental setup. It means that the probe system
itself can be a source o f signal and detector o f the signal without the equipment.
In addition, by using a mixer, the reflected microwave signal can be down-converted to a
lower frequency. This is a kind o f replacement for the lock in amplifier in the previous
experiment setup. If the evanescent microwave microscopy system become so compact,
to design a parallel microwave microscopy system would be a com er stone in evanescent
microwave microscopy.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
1.3 Research Objectives
To scan a larger area is desirable in some application, but taking more time is
disadvantage. If we com e up with a new approach to save time and to make a decision
regarding the property o f scanning material in a limited time very effectively, it will be a
good method to get an information about a material. To achieve this goal, we need a
compact type o f probing system and the algorithms to proceed with the signal processing
needed to analyze the data resulting from scanning a material.
To avoid a bulky experimental setup, using a microstrip resonator, the self
oscillating resonator circuit was designed and its performance is presented. The self
oscillating probe is the most crucial component in the parallel probe scanning system. By
using a mixer and a microwave detector, the self oscillating probing system becomes a
basic probe system unit in the parallel scanning system. The probing system present the
performance to scan a material in term o f the resolution and sensitivity. With single self
oscillating probing system, the parallel probing system is implemented by adding each
unit one by one. The signal processing algorithms to manipulate the data resulting from
the multiple probing system and the result present.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2
S elf Oscillating Microwave Probe
2.1 S e lf Oscillating Probe System
In order to implement a parallel probe system, the first thing w e have to consider is
designing a compact probe that can be operated without the need o f equipment.
I f w e kept the same schematic as the previous single probe system in order to implement
a parallel probe, w e would need huge experimental setup. I f w e have a schematic that
cover all the equipment w e need to execute the evanescent m icrow ave microscopy
experiment, the system w ill provide the possibility to implement as many as probes we
want to implement. First o f all, w e need to come up with the schem atic to replace signal
source and probing part. Instead o f a high frequency signal generator, w e need to replace
it with a new resonator having s e lf oscillation.
The s e lf oscillating resonator can be implemented by using h a lf wavelength transmission
line resonator. It is very w ell known that half wavelength the transmission line itself has
the property o f resonance. In general, to make a oscillator, w e need a feedback loop and
amplification to sustain the oscillation. Combining a radio frequency amplifier and half
wave transmission line resonator can be one solution for implementing s e lf oscillating
resonator, with an oscillation frequency around a few GHz.
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
32
Microstripline Waveguide
Samnle
Probe tip
RF AMPLIFIER
£ = X/2
S
'
RF AMPLIFIER
Cc
--VW-
Ro
Z **
Lo
Co
===>
“'VW—■
Ro
z tctta.
c
£>
Cs
-'WV—
Lo
R
C>
Figure.2-1 Schematic o f the s e lf oscillating resonator and it’s circuit model
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
In our case, the frequency range where w e need to execute evanescent microwave
microcopy is around I GHz. To make use o f a s e lf oscillating transmission line resonator
is a good approach to obtain a compact probe system.
2.1.1 Operation o f S elf Oscillating Resonator
In the S e lf Oscillating Probe, there are three major parts that make it work as an
oscillator. The first part is the microstripline resonator, which is the core o f the system
and can be modeled as in figure.2-1. When w e place a sample under the tip o f probe, as
w e discussed in previous chapter, w e have a shift o f frequency as in figure.2-2. For a
metal or dielectric material, the frequency is shifted to a low one. The amount o f shift
depends on the microwave properties o f material, and using this property, w e may
characterize the material. The second part is a radio frequency amplifier. The amplifier
prevents the oscillation o f the resonator from damping and keeps the resonant frequency
stable. W hen the high frequency signal is fedback through the loop, it may be attenuated
or scattered from the original signal without an amplifier. The mathematical condition for
the oscillation with an amplifier or loop gain equation, which is called Barkhausen
criterion, is stated as in Eqs.(2-1).
G,<oo)*G a ( co) = 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2-1)
34
Gf(a>) is gain o f feedback loop, GA(co) is gain o f amplifier. It means that the main criteria
for oscillation is that the loop gain should be larger than unity for the w ave phase shift o f
-180°. The amplifier w e use in this work operate at 0.5GHz - 2.5G Hz and has a gain
about 12dB. The last section o f the system is down-converts the signal from oscillator.
As a final output, we can use either the down-converted frequency w hich is more
convenient to deal with in general electronic circuits, otherwise w e can convert the
frequency properties to DC type signal such as voltage. Both o f them can be used as a
final output for a s e lf oscillating probe system. The downconversion can be performed
with a mixer, which produces the two different frequencies, one is the sum o f the two
input frequency, the other is the difference. Usually, w e take the difference o f two inputs
for our purpose. To serve as the local oscillator, w e use the regular type oscillator, VCO,
which generate 15MHz to 2.2GHz. To obtain voltage as a final output, w e may use a
diode detector. The sim ple diode detector rectifies the lower frequency and produces a dc
type signal. In our case, w e need to use a diode for high speed recovery such as a
schottky diode. The change in either frequency or voltage output by the system due to the
microwave properties o f the material being evaluated w ill make up the data to be
processed. To measure the oscillation spectra o f the s e lf oscillating EMP, w e use the
Tektronics Spectrum Analyzer. Figure.2-3 shows the spectra o f the oscillator.
W e may get a change o f frequency when w e place a metallic sample such as brass, near
the probe tip in figure.2-2. The change in frequency at the output o f m ixer is the useful
data for the mapping experiments.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
Figure.2-2 The change o f frequency o f s e lf oscillating probe system due to
metallic sam ple(B rass). Left signal is the shifted one due to brass.
The data in figure.2.2 shows the s e lf oscillating probe’s response to a sample in
terms o f a shift in oscillation frequency. The amount o f change is about 3M Hz. A large
amount o f change in frequency could be obtained with a high conductive material; the
brass is relatively low conductivity material compare to copper or silver. In case o f
copper or silver, w e may have a larger amount o f change in frequency than that o f brass.
W e need to show how the s e lf oscillating probe is responding to the microwave
properties o f the sample. The experimental setup is schematically shown in figure.2-4
which is much simpler than the previous set-up that required external rf signal sources, a
circulator and lock-in amplifier [25-33].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
Figure.2-3 The oscillation spectra o f self oscillating resonator
The reflection coefficient Sn is given by:
(2 - 2)
where Zo is the characteristic impedance o f the microstripline feed line and Zt0ta! is the
total impedance o f the resonator. A s w e can see in figure.2-1, the microstripline can be
modeled with a LCR passive lump element circuit, which facilitates the analysis from a
circuit’s point o f view. The total impedance can be represented by Z q+ Ro+j’cdLq + Z
where Zo is the characteristic impedance o f the line (50Q), Ro is the resistance
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
M otor
Controllers
KOLLMORGEN
501 PS
X -Y -
DAQ C ard AI 16XE
©
High Voltage Source
KEITHLEY 237
Semiconductor
nm pk
Stepper
Motor*
X -Y ax i*
Pre-Amplifier
EG & G
FRINCENTON
5187
Lock-in Amplifier
EG & G
INSTRUMENTS
7260 DSP
n ty tl
tefln
iq c t
R F Signal Generator
HP8648A
lOOKHz -1000 MHz
M OD
lapat/emtpax
RF
oatant
<>___ <j>
Figure.2-4 The previous experimental setup having equipment for single probe.
o f the metallic microstripline and it also includes the radiative losses, and Z = A+jB
where A and B are given by [33]:
A = RS
3
£ __________________
(C0 + Ccf +<y2i£ (C cCJ + CQCc + C0Csf
(2-3)
(Q + C r ) + ^ R ^ C , + C,XCrC. + C„Cr + Q C X a>[(C„ + Ct) ! +
(CtC, + C .C , + C0C , y }
where Co and Lo are respectively the capacitance and inductance o f the microstripline
resonator, Rs is the surface resistance o f the sample, Cs is the sample's capacitance, Cc is
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2-4)
38
the coupling capacitance and go is the radial frequency. A ll the circuit elements are
shown in Fig.2-1 [33]. W e note that for metallic samples, A =0 because Rs =0, and B «[(Co+Cc)®]'1. Thus, in this case the capacitance o f the combined probe and sample
system is just C=Co+Cc.
The resonant frequency f o o f the lumped parameter resonator in the presence o f the
sample is given by
r
0
0
=
=— P —
2 7c \ L ' q C q
(2-5)
where L'o and C’o are the m odified resonator inductance and capacitance due to the
sample. From figure 1 w e can see that L'o=Lo and C'o=C=-oo/B. The quality factor o f the
perturbed resonator (Q') is simply:
Q
=
2 q /n 4
R0 + R
( 2 - 6)
W hen Rs=0, R=0 and the Q o f the resonator does not change. But its resonant
frequency changes by Afo given by:
(2-7)
where fo is the resonant frequency o f the unperturbed resonator ( f 0 = — -------- ). From
»= T
7^>2 kV\ 7 L0C0
the above equation, w e can calculate C =
C=Co+Cc and
C
Q
. Moreover, for Rs=0 w e have
(l ~ 4 / o / / 0) ‘
w e can calculate the coupling capacitance once w e measure Afo/fo-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
2.2 D esign o f S e lf Oscillating Probe System
2.2.1 Planar Transmission Line Resonator in S e lf Oscillating Probe System
For the resonator part, a planar transmission line can be used. The resonator plays
a crucial role in determining the operational frequency o f the oscillator. In general, there
are four different type o f planar transmission lines; microstrip line, stripline, shield
coplanar waveguide, and coplanar waveguide as in figure.2-5. For the se lf oscillating
probe system , the coplanar type was adopted. The coplanar type waveguide has a signal
and ground on the same plane. It has several advantages over the other type transmission
line. It means that w e don’t need to put the via hole to connect the each side o f
transmission line plane and relatively easy to make a series or shunt configuration. Some
o f advantages are reducing the parasitic capacitance, less energy dissipation and better
confining o f field. To design the circuit using coplanar w aveguide, w e need to determine
the geometrical parameters such as the length o f the w aveguide. Depending on these
parameters, the frequency behavior o f the waveguide is different. The geometrical
parameters are determined by quasi-static analysis o f transmission lines using conformal
mapping method.[34,35]
Another w ay to design a planar type transmission line resonator is to implement the
microstripline on a silicon wafer. This approach provide a possibility to implement all o f
the probing system elements on a one single chip via m onolithic integration. The design
method is similar with that o f coplanar waveguide. U sing quasi electric field analysis, we
m ay determine the geometrical parameters such as the length o f resonator [36,38].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
For the s e lf oscillating probe system, w e simply use approximate method to get the
length o f waveguide, because all that w e want is to achieve an approximate frequency
around 1GHz. U sing Eqs.(2-8), the length o f waveguide for specific frequency can be
determined and its frequency is verified experimentally as in figure.2-3. The printed
circuit board used is etched on a duroid substrate. The duroid material has a dielectric
constant 2.2 and a thickness o f 1mm.
Microstrip Line
Front view o f Coplanar wavegide
K
W nWSN
Front view o f stripline
Front view o f shield coplanar waveguide
Figure.2-5 Four different type o f planar transmission line
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
Electric field line in the air
Electric field line in the dielectric material
Electric field line due to stray microstrip mode
Figure.2-6 Electric Field Distribution in Coplanar W aveguide
For example, in case o f the 1GHz operation frequency, w e may get a 30cm o f
wavelength.
For the dielectric material which is used for substrate, dielectric constant(s) is 2.2,
duroid material, half o f wavelength long(X72) is about 10cm. In general, the oscillator can
be implemented using various configurations such as the hartley type oscillator, the
colpitts type oscillator, these topologies depend on the choice o f the impedance value and
structure in the feed back loop as in figure.2-7[37]. Another crucial part to determine the
frequency o f oscillator is the value o f passive element in the feed back loop. For the
planar transmission type resonator, it is very w ell known that the transmission line can be
modeled by the passive elem ent such as in figure.2-8[38]. It means that the transmission
line may have an impedance element for feedback loop. It gives us the insight that the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
planar type transmission line resonator can be regarded as one o f the feedback loop line
configurations o f the oscillator. Instead o f areal passive element, w e can control the
impedance value o f the feedback loop line with the geometrical parameter such as the
length o f stripline when w e design the oscillator with a planar type transmission line
resonator.
Z3
C-.
l
J
______________
fa/!
Cl
=J=
:^
1
c2
Figure.2-7 The various configuration o f feedback loop for oscillator
(a) general configuration o f impedance element (b)Hartley (c)Coplitts
(d)Clapp
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
43
X<t)
-------- 1
-
-
I
Ca)
T
v ect)
I
_L
i_
_
_
_
_
_I_
C*0
Figure.2-8 (a) Lossless Transmission line with voltage generator, (b) Equivalent
circuit m odel with lump element [38]
A s the sensing part, the tip at the end o f probe is the important part to affect on the
performance o f the probing system. The tip o f probe that is made o f has a very sharp
end. The sharp end o f tip increase the resolution o f EMP image. We may get the very
sharp tip using chemical etching. Through the chemical process, the end o f tip is etched
away due to the chemical reaction . The diameter o f end o f tip is about 1p.m. The actual
shape o f tip is in figure.2-10. The details o f chemical reaction
are that a positive voltage
( 20 volts )is applied to the wire, which is a anode. Etching occurs at the air-electrolyte
interface.
Cathode: 6H20 + 6e* -» 3H2(g) + 6 0 H '
Anode: W (s) + 8 Q H '^ W 0 42' + 4H2Q + 6e
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
Figure.2-9 The end o f tungsten tip for self oscillating probe
2.2.2 Configuration o f S e lf Oscillating Probe System
The main advantage o f the s e lf oscillating probe is to reduce the size o f the
experiment setup. In the previous experiment setup as in figure.2-4, the signal generator
generate the high frequency signal; the reflected signal was detected using crystal
detector. The lock in amplifier was used to demodulate signal and produce the dc type
signal. The s e lf oscillating probe system on the other hands implements the function o f
all the equipment with on board electronics such as the oscillator and mixer. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
45
oscillator composed o f planar type resonator, radio frequency amplifier and probe tip
generates the output signal. The oscillator is designed using the basic concept o f
oscillator as in figure 2-10. U sin g as the oscillator VCO, and m ixing the signal from the
resonator and the signal from VCO, the signal from the s e lf oscillating probe is
demodulated. We may have alternatives to take the output signal. One is to use the low
frequency that is output o f mixer, the other is to use the dc type signal that come from the
output o f detector.
The one more thing w e have to consider is the demodulator. The main performance o f
demodulator is to convert the frequency which is down converted. The natural o f diode
provide a insight to convert radio frequency to DC type o f signal. The diode equation
which be expanded in a Taylor series show how a diode convert the radio frequency to
DC signal[39].
Then, the equation can be rewritten as a sum o f the DC bias current and AC current when
the bias voltage have DC and A C component. The small signal approximation produce
the only DC signal. It show us that the diode convert radio frequency to DC type o f
signal. The high speed recovery diode, for instance, schottky diode, is good for this
application. Using schottky diode and capacitor, a signal detector which is same as a
demodulator is implemented.
The schematic o f s e lf oscillating probe system is in figure.2-11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
A(s)
Vout
Vin
AMP
►
Resonator
Feedback (3(s)
Figure.2-10. M odel o f Oscillator Concept
Schottky
Diode
z
1
Probe Tip
Figure.2-11. Schematic o f single s e lf oscillating probe system
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 3:
Single Probe Operation
W e discuss the construction and the performance o f a single self oscillating probe
system. In addition, we show the output data o f each part in the system. Using the data
from the single probe, we make an image o f the sample and determine the resolution o f
the probe. Dielectric and metallic samples are both used for the characterization o f the
probe system.
3.1 Fabrication o f the Single Self Oscillating Probe System
U sing a m illing machine, w e cut the circuit pattern o f the probe system. The circuit is
implemented on the duroid board which is the PCB substrate for microwave circuits. The
thickness used is about 1mm. The radio frequency amplifier is mounted on the board
using soldering machine. To connect the close feedback loop line, we make a via hole to
connect the bottom side line and top side line. The VCO is also mounted on the PCB
board. The VCO take a part as a reference frequency source as a LO in our case. The
VCO connect to the mixer to generate low frequency, about 20M Hz, even lower
frequency can be produced.
The mixer is also mounted on the PCB board. The signals from the resonator and the
VCO are m ixed together to produce the low frequency.
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
The mixer w e use has a four diode in ring configuration, which is a double balanced
mixer. This type o f mixer has an great performance in terms o f signal isolation.
The actual shape o f the probe system is shown in figure 3-1. The output o f mixer, which
is around 20M H z signal is shown in figure 3-2. The frequency o f the signal is a little bit
flexible and depends on the frequency o f VCO and the resonator.
M IX E R
Figure3-1. The single self oscillating probe system
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 3-2 The waveform o f output signal com ing from the mixer
When we put a sample under the tip o f the probe, the waveform o f the mixer change
in frequency. If w e connect the diode detector after mixer, the diode converts the amount
o f change to a D C type signal, which is our EMP output voltage. Conversely, if we don’t
use the diode detector, w e may track the amount o f change in the waveform in terms o f
frequency and use it as the EMP output.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
3.2 Characterization o f the Single S elf Oscillating Probe System
3.2.1 Sensitivity and Resolution in EMP
There are several parameters that affect the sensitivity o f EMP including both
mechanical and electrical aspects. The sensitivity and resolution o f EMP is compared
with the other kinds o f m icroscopy in Table 1-1. The way to estimate the sensitivity and
resolution in EMP is to calculate the MDS (minimum detectable signal), which is the
defined as the smallest amount o f change in the input that generate an output equal to the
rms value o f noise(V nrms)- It means that w e can not detect the signal if the output signal is
less than the noise.
M D S = Acr. =
V n r m s / V in
(3-1)
s xs ,
V2<2
f
COO \
A co
\
(3-2)
00,,
f Cc ) r 1- 2 —C
[cj V
Cc
3
s,
=
g j£ _
4 7C
c
f
r
LO 1 +
V
V
2\
(3-3)
Cc ^
c Oy
/
The Sxand Sfare the important parameter to determine the MDS. The equations for
these parameters is derived from previous work and used for the parameters for the
sensitivity in the various EMP application. Sx mainly depends on the quality factor of
resonator, Sf depends on the coupling capacitance between probe and sample. To get a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
high sensitivity, in other words, the smallest M DS, these two parameters should be as
large as possible. It means that a high quality factor is the dominant element for achieving
high resolution EMP. To get a high coupling capacitance, w e need to bring the probe tip
as close to the sample as we can. There is the alternative as a method to measure the
resolution in EMP. To measure the decay profile at the axis perpendicular to horizontal
axis is the another choice. If we measure the profile, w e may see the steepness in the
change o f profile in EMP output. At the onset where the profile is steep, we take a
derivative curve. Then, when the curve meet the axis which represent the distance, the
cross point where the two lines meet is a good estimate o f the resolution. This figure is
also regarded as a minimum detectable value, since it is the minimum length where the
output signal can be detected.
3.2.2 Z decay o f Single S elf Oscillating Probe System
The z_decay, which is the parameter that shows the length that we can get a 30 % of
maximum EMP at z=0. We may use this parameter as a kind o f resolution o f the
resonator. To measure the z decay, we need the station to mount the single probe system
as in figure 3-3.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
power supply
Figure 3-3. The experiment station, which include step motors for the three axis, the
probe holder, frequency counter, and oscilloscope
The stepper motors are used to control the movement o f each axis with I pm to 100
pm. as the range o f step increments possible. The step motor is controlled by a Labview
Program, which include a virtual instrumentation panel as in Figure 3-4.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
c o n tr o l w ith o u t Z
:
* « a
Jj gf c *- H«f r .
H j j13ptAppfcafaonFont
^ \ DI^NCESEHSPR INFQi
t.O£0
00
'
• . 1000 .
-.2 0 0 0
2 h 0.00. I moemelecsf
- W fflgA T O R S j
[aOLC&TQRSI
OirrentHefqhti
m
sa « -aj;. ;j.
V g H 307PM
Figure 3-4. The front panel for operation o f data acquisition o f the Labview Program
When we draw the derivative line over the EMP data as in Figure 3-5, the number on
the distance axis, 10pm, is the z decay constant o f the probe. It means that w e may take a
about 10pm as the resolution o f the single probe system.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
Z_Decay
3.90E -01
3.85E -01
3.80E -01
3.75E -01
3.70E -01
3.65E -01
0
10
20
30
40
Distance(um)
Figure 3-5 The z_decay o f single self oscillating probe system
3.2.3 Performance o f single self oscillating probe system
Line P attern
The output from the probe show if the probe system work properly or not. First of
all, the line pattern for a circuit connection is scanned. The width o f the line is 300pm
and the gap between the line is 200pm. Figure 3-6 is the actual image o f the pattern we
use. Figure 3-7 is the EMP image.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
Figure 3-6 The actual image o f line pattern
Two Line Pattern
□.□5 - i
(/Odra
Depth(um)
1000
Width(um)
500
Figure 3-7 The EMP image of line pattern
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission
56
In the line pattern scanning, the EMP response obtained for the conductive line
pattern was distinguishable from that o f the non-conductive gap. The difference between
them was about 30m V . For the scanning data, because o f the diode detector used, there is
a limit to the sensitivity that can be achieved. The sensitivity is defined as how much
voltage is output for a given input power level, with unit o f m illivolt per milliwatt
In other words, if w e use a more efficient or high RF characteristic diode in terms o f
sensitivity, w e may have a even better response or higher level output signal.
Wire
W e scan a single tungsten wire, which has a about a lOOpm diameter. The scanning
data would be mapped with the 3D and 2D im ages. The actual wire image is shown in
Figure 3-8.
Figure 3-8 The lOOum diameter tungsten wire
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
Wire-100 um
0 .08 —|
0 .0 7 -
0 .0 6 - -
s 0 .0 5 CL
s
0 .03
0.0 2 -
0.01
^
Depth(um)
100
200
300
400
W idth(um )
500
600
Figure 3-9 The 3D EMP image o f 100pm tungsten wire
The 2D map shows that the line width can be measured by measuring the full
width at h alf maximum o f the scan data. At the peak, the EMP output is about 68mV and
the baseline is on 35mV. Therefore, w e may take a two number from the distance axis
which correspond to about 52m V on EMP output.
The difference between the two number correspond to the wire diameter measured using
the EMP method. The measured diameter is about lOOpm.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
W ire-1 OOum
0.08
0.07
EMP(V)
0 .06
0 .0 5
0.04
0.03
0.02
0.01
0
100
200
300
W idth(um )
400
500
Figure 3-10. The 2D EMP data map o f lOOpm tungsten wire
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
600
Depth(um)
59
Gap in Microstripline
The microstirpline can be fabricated on a silicon wafer using general silicon process
technology. On the microstripline, there are two parts to make a signal flow on the
stripline properly, one is the feedline which can be connected to the signal source and
another is the resonator which is the core part to achieve a resonance. When the signal is
applied to the microstripline, there is a part which connects the feedline to the resonator.
Figure 3-11. The SEM image o f the gap between feedline and resonator
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
The distance o f the gap between the feedline and resonator determine the coupling
capacitance w hich make the resonator activate. Therefore, to determine the distance o f
gap is very important to the design o f microstripline resonator. The resonator on silicon
wafer has about 20pm distance. We scan the gap and image it to see the performance o f
the s e lf oscillating probe system. The Scanning Electron M icroscope(SEM ) image o f the
gap is show n in figure 3-11. The EMP image for the gap is shown in figure 3-12 and
figure 3-13, one is 2D image and another is 3D image for the gap.
gap in resonator
0 14
.
--
0.12
0.08
LU
0.06:-
0.04
0 . 02200
180
160
140
120
100
80
W idth(um )
Figure 3-12. The 2D EMP image o f the gap between feedline and resonator
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 1
gap in resonator
0 .1 4 -,
0. 12-
0.1
-
^ 0 .0 8 - '
LU
0 .0 6 -
0 .0 4 --1
150
100
Width(um)
Figure 3-13. The 3D EMP image o f the gap between feedline and resonator
As in figure 3-12, the waveform resulting from scanning the gap show the distance
o f the gap is about 20pm with relatively high accuracy. At the half of peaked waveform
amplitude, the width o f the peaked wave means the distance o f the gap. Therefore, it
shows that the EMP can image and detect the gap fabricated on the silicon wafer.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
3.3 Frequency Response o f S elf Oscillating Single Probe System
As we discussed in the previous section, there are alternatives to obtain the EMP
output such as the output o f diode detector, which is a voltage and the output o f mixer,
which is a frequency. Up to now, m ost o f EMP output reported have been in the form o f
voltage, however, it deserve a try to take frequency as the EMP output, if that kind o f
data is more sensitive.
D ielectric Material
U sing several kinds o f dielectric, we measure the frequency response to see how
much the frequency changes for a change in the dielectric constant.
The samples w e use in this measurement has a dielectric constants ranging from 2 to 10,
The dielectric constants for some o f them such as alumina, carbon material has been
approximated due to the com plexity o f material or the wide range o f dielectric constant.
However, w e have a stable response for the each sample when w e change the operational
frequency. The data shows that the self oscillating probe system produces a clear
response to the microwave properties o f material, a fact that is represented by the
dielectric constants. U sing this data, w e may prove the relationship shown in E q s.(l-l)
,that is, a change in the dielectric constant is proportional to the amount o f frequency
change in EMP.
The relationship between the change o f frequency and dielectric constant is shown in
figure 3-15. The behavior o f the two parameters is very close to each other and is a linear
as in E q s .(l-l)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
Frequecny R e sp o n se
2 .0 2 E + 0 7
2 .0 0 E + 0 7
Freq(MHz)
1.98E + 07
1 .96E + 07
1.94E + 07
1 .92E + 07
1.9 0 E + 0 7
23 samples which
have a different
dielectric constant
2.0 -1 0 Dielectric
constant
1 .88E + 07
1.8 6 E + 0 7
0
200
400
600
800
1000
1200
Tim e(sec)
Figure 3-14. The frequency response for the sample which has a different
dielectric constant
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
Profile of frequency change due to dielectric constant
0.8
1 .2 0 E + 0 6
0 .7
1.0 0 E + 0 6
(1-1/sqrt(e)
0.6
8 .0 0 E + 0 5
<
0 .5
6 .0 0 E + 0 5
0 .4
0.3
4 .0 0 E + 0 5
Af at 20MHz
2 .0 0 E + 0 5
0.2
Af at 16MHz
0.1
0 .0 0 E + 0 0
1
2
3
4
5
6
7
8
9
Dielectric Constant
Figure 3-15. The profile o f frequency change resulting from dielectric
material
When the dielectric constant is increased, the profile o f the frequency change is also
increased. As in figure 3-15, even if we change the operational frequency, the profile is
not changed. In other words, the profile of the dielectric constant follows up the profile o f
frequency change very closely. This data shows the frequency change has a linear
relationship with the dielectric constant as in E q s .(l-l).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
Conductive M ateriaUM etal)
To see the frequency response o f self oscillating probe system , w e use several kinds
o f metallic samples. The samples have a different conductivity. According to the
conductivity, a different response is produced for each o f the metallic samples.
A f vs Rs & Conductivity
7 .0 0 B O 7
2.00E+Q6
5.00E+07
1.50E+06
<
1.00E+06
2.00E+07
5.00E+05
O.OOEfOO
2.00E -07
2.50E -07
3.00E-07
3.50E-07
4.00E-07
4.50E -07
O.OOE+OO
5.00 E-07
Rs
Figure 3-16 The frequency Response for Conductive Material (Metal)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Conductivity(1/am )
2.50E+06
66
Sample
Conductivity
Rs (Surface Resistance)
Copper
6.30 x 107(l/Q m )
2.50 x 10'7(Q)
Aluminum
3.72 x 107(l/Q m )
3.25 x 10'7(Q)
Tungsten
1.85 x 107(l/Q m )
4.61 x I0'7(Q)
Brass
1.80 x 107(l/Q m )
4.69 x 10'7(Q)
Table 3-1. The Conductivity and Surface Resistance o f M etal Samples
For a given conductivity, the surface resistance is calculated. The profile o f
frequency change to surface resistance has a very close relationship to that o f
conductivity. As a m icrowave property o f material, the conductivity o f metal activate the
se lf oscillating probe system to produce the response as seen in the previous section.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 4
Multiple S elf Oscillating Probe System
The multiple probe system is implemented and tested using the single s e lf
oscillating probe system. To process the data com ing from each probe, w e use the
neuromorphic method which originate from signal processing algorithms based on
biological system . For the multiple probes system, w e need to eliminate any coupling
betw een the probes to guarantee an independent operation. Furthermore, the multiple
probe system w ill produce a single output, and its the scanning performance can be
evaluated.
4.1. Signal Processing in Multiple S e lf Oscillating Probes System
The scanning speed o f the EMP is very high due to its high operational frequency,
w hich ranges from 1 to 10GHz. The scanning speed can be increased even more by the
parallel probe system. It is clear that the parallel probe system can scan more area o f the
sam ple or scan the sample faster. D ue to the output signals from parallel probes, w e have
to have more data to be processed. For example, i f w e have one probe which has 1cm /sec
scanning speed. And i f w e suppose one scanning point covers 10pm, then to cover 1cm,
w e need 1000 scanning points. I f w e want to use 10 probes w e may have 10,000 points, if
one scanning point has 12 bit resolution in data acquisition, w e may get a 0.12 Mbit, and
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
W e need 0.12M bit/sec data processing speed orl20 kbps. This is just the case o f one
dimensional line scan, meaning that w e have more data for the case two dimensional
scanning map.
One more thing w e can discuss is the detection o f a defect or different property, for
example, an area w hich has a different conductivity or dielectric constant, in a sample. If
w e scan a material like a thin film, which should be inspected immediately after
deposition to detect any “defects”, w e also need to have a large amount o f data. In this
case, m ost o f the area o f the sample is free from defect, besides small portion o f area.
Therefore, if w e have the function to get a different signal from the signal o f the area that
is defect free, w e can figure out where the defect or defected area is with one single
output signal. To solve the two representative problems in a parallel probe system, it is
o f great importance to extend the usefulness o f parallel probes. One o f the solution we
com e up with is to adopt the signal processing algorithms from biological system like an
insect. It is the so called neuromorphic electronic system w hich is a electronic system that
implements a signal processing algorithm based on biological system. It is natural
reasoning that a biological system has better algorithms in terms o f efficiency in making
signal processing than a general electronic system.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
69
4.1.1 M odeling o f B iological System
B y taking a biological system, w hich is a different system from th o se most electrical
engineers are familiar with, w e m ay efficiently approach the solution to the problem o f t
implementing a multiple microwave probe system.
In general, behavioral models based on biological systems, for exam ple, fly ’s retina,
more easily describe a complicated system such as multiple input m icrow ave microscopy
system.
Figure. 4-1 The im age o f fly ’s eye
In the case o f the fly’s landing m odel, under a certain condition, i f the output o f
local directionally selective m ovem ent detectors are spatially pooled and subsequently
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
integrated in tim e, then the fly releases the leg to make a landing, h i other words,
whenever the level o f this integrated signal reaches a fixed threshold, landing is released.
Through the retina, the fly receives a visual signal like a light source, so the intensity o f
visual signal that is the integrated signal triggers a landing. So, i f w e look into the
structure o f the fly ’s retina, w e model it like a correlation type o f movement detector.
Retina Cell
Correlator
Temporal Integrator
£
OUT
Figure 4-2. Model o f fly ’s retina
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
The m ovem ent detector(correlator) collects a visual signal and produce a final
integrated output signal. In figure 4-2, the retina cells w hich get a visual signal produce
the correlated signal, and the signals then merge into the integrator. The process happens
on both the right and the left side. Finally, the one integrated signal produce, according to
intensity o f the signal, fly ’s landing system triggers.
4.1.2 Correlator
The correlator is the crucial part to take a charge o f signal processing using the
property o f calculation algorithms like the correlation between adjacent signals. In other
words, i f the tw o adjacent signals are highly correlated, then the output o f correlator to
the two signals is clearly different from the other signal which is not much correlated. In
our case, the signals w hich com e from the defected area having the different property
from the rest o f area o f the sample, then the signals are probably highly correlated
compare to the signal from the non defected area. The correlator in our system comes
from the m ovem ent detector in a biological system. The movement detector is usually the
core o f a biological system, for example, the fly’s retina. Therefore, the characteristic o f
the correlator is the same as that o f a movement detector. The general property o f a
movement detector in a fly’s retina has several aspects in terms o f structure. First o f all,
it needs two inputs.
Two inputs are necessary since motion is a vector that needs two points for its
representation. A single photoreceptor w hich is a kind o f sensor to detect light in fly’s
retina can not distinguish a dark bar that crossed its receptive field from the left to the
right form one that crossed from the right to the left, or from a momentary dimming o f
the light. It also has a nonlinear property. A nonlinear interaction between the input
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
signals is required. Otherwise, the time averaged output o f a detector would be equal to
the detector response to its time averaged input signals. In the averaged input signals,
however, ail information about the temporal sequence is lost. Thus, a movement detector
with a linear interaction can not be directionally selective. One more property is
asymmetry. The two input signals o f a movement detector have to be processed in a
slightly different way. I f the detector were symmetrical, its input channels could be
interchanged without affecting its output. It is then no longer possible to discriminate
which channel was excited first and which later. Accordingly, the detector would not be
directionally selective.
4.1.3 Signal Processing Algorithms in Correlator
The movement detector which has a correlator in a fly ’s retina produce an integrated
signal which triggers the fly’s instinct to land whenever the intensity o f signal m eet a
threshold value. But in our SMM(Scanning M icrowave Microscopy), using this detector
model, w e may decide the level o f defect by comparing the output signal with the regular
shape o f the signal in terms o f DC level amplitude, w hich is a standard reference. And
using the polarity o f output signal, w e may know the scanning direction.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
Scanning Direction
M------------------------CHI
CH2
Delay
Delay
Output
CHI
CH2
Delay
(b)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
<
CHI
CH2
Delay
Figure 4-3. (a) Schematic o f Correlator for both scanning direction,
(b) Schematic o f Correlator for the scanning direction
m oving left to right
(c) Schematic o f Correlator for the scanning direction
m oving right to left, in case w e use the same direction
with (b), the output is null.
The output signal o f the multiplier on the left branch side as in figure 4-3(b) is the
signal w hich is multiplied by the two signals. One is the signal which is passed through a
delay, the other is the input signal o f right branch.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
Input Signals
CHI
CH2
The output signal o f right side multiplier is the product o f the input signal o f channel 1
and delayed channel 2 signal.
Calculation in Left branch as in figure4-3(b),
X
CH2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
Calculation in Right branch under the same direction as left branch as in figure 4-3(c),
CHI
A
0
CH2
At
Figure 4-4. Signal Processing in a Correlator
For exam ple, i f the maximum value o f the multiplier output is “ 1”, the final output
o f correlator is 1 (1-0), if w e scan the other direction, the output o f correlator in the same
case is “-1 ”. W ith a polarity o f output, w e may know the scanning direction. Using the
amplitude o f the multiplier’s output, w e may know i f there is defect. It comes from
comparing the output with the reference. Ideally, i f w e get the same input signal, and
multiply them, then w e get the maximum amplitude o f signal, otherwise w e get a lower
amplitude o f signal than that o f a maximum signal. For example, i f the maximum
amplitude o f output signal is “ 1”, when w e get a defect, the multiplied signal is going to
be a less than “ 1”
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
4.2 D esign o f M ultiple Probe System
To make a multiple probe system , w e use the single probe system as a unit probe and
implements many a number o f probes as needed. When w e put single probes together, w e
need to consider the coupling problem that result from the high frequency characteristic
o f the probe. It can prevent the probe from achieving independent operation. In other
words, without considering the coupling problem, one probe could follow up the behavior
o f another probe. In this case, w e don’t get the advantage o f a multiple probe system.
Therefore, the m ost important aspect in the design o f multiple probe system is to cancel
out the coupling between the probes.
Figure 4-5. Top view o f Multiple Probe System.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
Figure 4-6 Side view o f M ultiple Probe System
To solve the coupling problem, w e build an extra ground plane on each single s e lf
oscillating probe system. In figure 4-5, the big pad in the center o f the board is the one o f
the extra ground plane. The extra ground plane provides better grounding that prevents
the signal from causing interference. To get better performance, the circuit board should
be shielded by an aluminum cover and w e need to make sure the mechanical part in the
multi probe system. The length o f the tip o f the probe also need to be considered. A long
tip is not a good w ay to confine the evanescent microwaves, meaning that a long tip has
less resolution in EMP. And, in a multi probe system, to make sure that the each tip
operates independently, w e need to isolate the tip from each other by shielding the area
where the tip extrudes. To have a different operational frequency is also a w ay to make
sure the probes operate independently.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
79
4.3 Performance o f M ultiple Probe System
4.3.1 Independent Operation o f Multiple Probe System
W e scan a material that has several spots with different property to show how
the neuromorphic signal processing algorithms are effective in detecting a defect or
different property o f the scanned material. For example, on the dielectric material,
assuming there are several holes, it means that the hole has a different property such
as dielectric constant o f one because the hole is in air. Therefore, the hole can be
assumed to be differentiated from the bulk o f dielectric material, which has a
different dielectric constant from the that o f the hole. An im age o f sample that has
several holes is show n in the figure 4-7.
Figure 4-7. The im age o f sample which has a different dielectric constant spot.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
80
W hen w e scan the sample from a vertical direction, the signals from each probe
should be generated independently with a constant tim e delay, because ,as seen in figure
4-7, the holes are located separately. W e may see th e signal from each probe in figure 4-8
and it can be seen that each probe operates independently.
Multiple P ro b e Data
2 .5 5 E + 0 7
2 .5 0 E + 0 7
£
2 .4 5 E + 0 7
S
O’
** 2 .4 0 E + 0 7
2 .3 5 E + 0 7
2 .3 0 E + 0 7
CO
T-
CO
TCM
CO
04
TCO
CO
CO
T-
CO
^
to
CO
LO
TCO
CO
CO
T im e(sec)
Figure 4-8. The scanning signal from each probe o f multiple probe system.
Each o f the probes w as operated at the different frequency to verify the independent
operation more clearly, meaning that the probes are alm ost free form the coupling
problem. With coupling problem between the probes, the operation o f each probe is
dependent on each other.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
4.3.2 A pplying the Signal Processing Algorithms
The main goal o f a multiple probe system is to achieve the detection o f any defect
w hich has a different property from that o f the bulk o f the scanned material. Usually, a
material resulting from a w ell controlled process does not have m uch defect or
nonuniformities. It means that w e don’t need to spend the capability o f the probe system
all the tim e during scanning. In other words, i f w e can use a single output instead o f
multiple outputs to detect defects, the system w ill becom e much more efficient because
w e don’t need individually check out the respective outputs o f all the probes in the
multiple probe system. In fact, the material w e have is almost defect free, sometimes, the
defect is detected, therefore, it is more convenient to monitor the defect detection with
one single output signal. The goal is to obtain a digital logic error signal o f either a “ 1” or
“0”, where 1 constitutes a defect, and 0 is no defect. Making the monitoring process
sim ple and potentially giving a possibility to save a lot o f resource in a manufacturing
line.
W e have a dielectric sample w hich has four holes on the dielectric material. The
four holes represent defects which have a different property. The material has a dielectric
constant 4.5, which the hole has a dielectric constant o f one because the holes is in air.
The im age o f four holes sample is show n in figure 4-7.
W e are going to apply the neuromorphic algorithms, and the so called ‘ correlator’
to implement the signal processing and to obtain a single output even i f w e have the four
signals corresponding to each o f four holes. Betw een holes there are tim e intervals, the
time intervals can be variable and depend on the location o f each hole. However, what w e
want is to find out about the presence o f a defected area in a large bulk area. Therefore,
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
82
w e can approximate the time interval as a fixed number, because i f w e want to control the
number adaptively, w e w ill have some complexity.
To avoid the complexity, w e simplifying it by averaging it. Even i f w e don’t apply the
time interval to the signal process in case by case, the single output does not change
much. In other words, in determining the presence or absence o f a defected area through
a yes or no signal applying the time interval case by case accurately is not too crucial.
What w e want to learn from the scanning process is i f there is a defect and approximate
its location. In any case, w e do have to investigate the time interval to get an average
number. The scanning direction used was from left to right, and the time interval between
the holes was measured at about 40 seconds. The correlation is achieved by multiplying
the probe signals; i f the signal have a high correlation, the output o f the correlator is
higher than that o f less correlated signals in terms o f frequency. I f the signal is not much
correlated, the output o f the correlator is probably around zero. In Figure 4-9, w e have
the sample w ith two holes. When w e scan this sample with multiple probe system, one o f
the multiple probe scan the two holes, the other scan the dielectric material. Therefore,
w e have a just tw o signals due to the two holes. These signal coming from the two holes
is relatively less correlated with signal coming from the dielectric material. It means that
the scanning the sample having two holes can be the example for less correlated scanning
comparing to the scanning the sample having four holes.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
r*
-,
r
Figure 4-9 The image o f sample having the two holes
W hen w e scan the sample having four holes, w e may get the four signals due to holes,
in this case, probably, all o f the signal from holes will be similar each other. In other
words, the signals are highly correlated. Figure 4-10 is the example to show how the
correlated signal is obtained.
Correlation
5 8 5 B -1 4
CbrelctiQn of ch1,(*2.
2.42EKJ7- Q *
575B -14
-2 3 8 E K J7 ------
iL
56SB-14
555B -14
5 50& 14
-I- O 03
0) 0 0
O O) CO N
03 O
T^ ©
NN
N x rto ero f Data
Figure 4-10. The example data for correlation from four holes sample
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
To show the signal process flow in detail, w e scan the material w hich has a different
type o f the scanning data such as the data from four holes, from tw o holes and from no
holes as show n in figure4-l 1. Each o f them represents a case o f either high correlated,
less correlated and non correlated signal processing. The first two peaked signals from
the top signal and bottom signal is for the four holes sample and the next two signal from
top are for the two holes sa m p le , rest o f them is the signal from just o n ly dielectric
material w ith no hole.
S ig n a ls of Multiple P ro b e
2 .6 5 E + 0 7
Ch1.
2 .6 0 E + 0 7
2 .5 5 E + 0 7
X
2 .5 0 E + 0 7
g 2 .4 5 E + 0 7
u.
2 .4 0 E + 0 7
p Q u r H Q ieg
2 .3 5 E + 0 7
2 .3 0 E + 0 7
CO
CM
lO
T f
NCO
05
CO
Two Holes
No
Hole
f i N Q r - C O l f l S O )
cn oi w
r * » 0 5 CMTr c o c o o
Time(sec)
Figure 4-11. The raw scanning data from the multiple probe system
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
85
According to the algorithms o f the correlator, w e need to make a delay for the data
from one o f the multiple probe, in this case, w e assume it as channel 1 data.
D elayed S ig n a l
2.65E+07
2.60E+07
2.55E+07
N
X 2.50E+07
2
& 2.45E+07
u_
2.40E+07
s
2.35E+07
2.30E+07
o
02 00
r— co in
^ t - c o c \ i T — o o j o o i '^.
C O l O O O T - T f t ^ - O O O C O O ) C M ' ^ - t ^ O
t - t - t - C M C M C M C M C O C O C O t J -
Time (sec)
Figure 4-12. The delayed signal for the data from one o f the multiple probe
In this case, the delay time is 40 seconds, which com e from the intervals between
the probes. The top one scan the hole first and bottom one scan it next. As much as the
delay, w e shift the data to right side to match up the time axis with the bottom one.
The reason to consider the time delay is to keep the temporal sequence o f the data as we
mentioned before.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
Correlation of Signals from
Multiple Probe System
^
1
er
£
3.50E+13
3.00E+13
2.50E+13
2.00E+13
1.50E+13
1.00E+13
5.00E+12
0.00E+00
T - s c o o ) i n T - N ( 0 0 ) U ) r - s n o ) i r ) T - s
c s i m i ^ o c o m c x D o c o c o o o i — cocooi-tr - r - T - r - C M C M C M C M n n c O n ^
Time(sec)
Figure 4-13 The correlation between the signals from multiple probe system.
A s w e mentioned earlier, the correlation between the signals is obtained from the
multiplication o f the related signals. The first two signals from the left side o f the figure
4-13 com e from the four holes samples. Comparing to the signal from the non defect area
o f the dielectric material, the signals from the four holes are highly correlated, because
the four signals com e from the hole, which is a different property to compare it with the
rest o f them. In this case, the hole looks like the defect in the scanning dielectric material.
The reference o f correlated signal is relative point o f v iew to compare it with the signal
from dielectric material, which constitutes most o f the scanned area in this case.
Therefore, as w e see in figure 4-13. The first two output signals left m ost from correlator
are higher than any other signals in terms o f frequency. The next two signals, which
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
87
com e from the two holes are less correlated w hile rest o f them is non correlated, because
measuring correlation is based on the signals com ing from the holes. Finally, the
processed signal to get a correlation from each case should be summed to produce a
single output as a final output. For each case such as the four holes sample, the two holes
sam ple and non hole samples, w e m ay have one single output. A s mentioned earlier, to
use a single output is a much more convenient w ay to monitor the scanning data from the
m ultiple probe system.
S in g le O u tp u t o f Multiple P ro b e S y stem
7.00E+13
Integrated output for multiple probe (1)
6.00E+13
5.00E+13
N
z 4.00E+13
Integrated output for multiple probe(2)
s
O’
S 3.00E+13
u.
Integrated output for
multiple probe (3)
2.00E+13
1.00E+13
0.00E+00
Time(sec)
Figure 4-14 The single output signal from the different sample.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
The sum o f multiple probe output(l),(2),(3) in figure 4-14 are for the four holes sample,
the two holes samples and non holes samples respectively. As w e can see in figure 4-14,
i f w e detect a defect or holes, the amplitude o f the single output signal is higher than that
generated due to a non hole sample. I f w e get the signal from all probes in the multiple
signal, w e m ay have much more higher peaked signal in terms o f frequency. In other
words, in a highly defected area, w e w ill have a very high peak signal. Therefore, the
amplitude o f the peak signal for a single output is the important parameter to measure the
density o f defect using our system.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 5
Conclusion and Future Work
In this work, a new schematic that implements EMP in a much more compact
experimental setup was presented and its performance demonstrated with a single probe
and a multiple probe system setup. On the duroid board, the previous experimental setup
is implemented with the RF electronics. The proposed probe is configured as a self
oscillating circuit o f frequency ranging a few gigahertz, avoiding the need for an external
signal source. The probe also is implement to make an independent probing system with
a high frequency source. Combining the probing part and the oscillator as a signal source
is a kind o f solution to make adaptive se lf oscillating probe system. From the probing
part, w e may obtain the change in the EMP signal due to the properties o f the material,
mixing the signal with a local oscillator achieved the concept o f s e lf oscillating probe
system. The proposed schematic implemented on board shows not only the possibility to
eliminate a bulky experimental setup, but also its modulating facilitates implementing a
multiple probe system. To prove this possibility, w e represented how the single probe
works for the scanning materials such as a wire, a line pattern on a circuit connector, and
the coupling gap fabricated on a silicon wafer.
When w e use a diode detector as a final stage for converting the RF signal to a DC signal,
the possibility o f enhancing the sensitivity by using a better quality diode such as a
schottky diode and achieving a high recovery time remains.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
Instead o f using voltage as the final EMP output, frequency can also be a good
substitution. Before feeding the m ixed signal to the diode detector, w e can ju st take this
signal and use the change o f frequency as an EMP output. W e scanned several different
kinds o f dielectric material, each o f them has a different dielectric constant. The dielectric
constant is one o f the m icrowave properties o f a material. As expected, the frequency o f
output o f m ixer changed depending on the dielectric constant o f the samples.
The basic idea regarding EMP measurement and the possibility to detect the presence o f a
material or variation in the properties o f a sample was validated for the s e lf oscillating
probe system.
In case o f frequency EMP output, w e have a more sensitive response to the sam ple
having a different dielectric constant. The possibility to get a high resolution w ith a se lf
oscillating probe system was shown. U sing the single self oscillating probe generating the
stable response in terms o f a change o f frequency, w e may build the multiple probe
system w hich has an advantage to scan more area in less time. To build the multiple
probe, w e need to solve the coupling probe causing’from the high frequency
characteristic o f the system . The strong grounding pad solve the most o f coupling
problem. W ith the coupling problem, w e don’t get the single probe operating
independently. Independent behavior o f the probes o f the multiple probe system at the
different operational frequency was demonstrated and the coupling was m inim ized.
The performance o f the multiple probe was shown by scanning the several kinds o f
samples to show probe behavior for where w e have a different dense o f defect.
W e obtained a different intensity o f single output from the neuromorphic signal
processing algorithms. For a very defect dense sample, we obtained higher level single
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
integrated signal that that o f less defect dense sam ple. The final single integrated signal
for each o f the samples was demonstrated. Therefore, the final single output signal show
that w e m ay estimate how much the sample is defected or what is going on the sample
using the final output.
It was show n that implementing more probes is a good w ay to build more efficient
system in terms o f a scanning time, a scanning area, and a scanning efficiency.
A ll o f the components o f a s e lf oscillating probe system were mounted on the duriod
board. Building on board hints that we may fabricate and integrate the all o f the
component o f the system on the silicon wafer using a silicon fabrication process
technology. For example, the resonator having the microstrip line resonator could be
fabricated b y a silicon process technology such as low pressure chemical vapor
deposited(LPCVD) and deep reactive ion etching(DRIE).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
References
1. Ash, E. A .; N icholls, G. "Super-resolution Aperture Scanning M icroscope," Nature
vol. 237, 1972, pp. 510-512.
2. Borst, A.; Bahde, S. “Visual information processing in the fly ’s landing system ”,
Journal o f Comp. Physiol A. 163,1988, pp.167-173
3. Borst, A . “H ow do flies land?”, BioScience, V ol 40, 1990, pp. 292-299.
4. B orst, A.; E gelhaaf, M. “Principles o f visual motion detection”, Trends N eu rosci.,
Vol 12, 1989, pp.297-306
5. Collin, R. “Foundations for M icrowave Engineering”, McGraw Hill, 1992.
6. Dekker ,A.J. Den.; Bos, A. van den "Resolution: A Survey," Journal o f the Opt. Soc.
O f America, vol. 14,March 1997, pp. 547-557.
7. Durig, U . ; Pohl, D.; Rohner, F. "Near-field Scanning O ptical M icroscopy", J. Appl.
Phys., vol. 59, 1986, pp. 3318-27.
8. Edwards, T. C. “Foundations for Microstrip Circuit D esign”. John W iley & Sons,
1992.
9. Etienne-Cummings, R.; Der Spiegel J Van, “Neuromorphic V ision Sensors”, Sensor
and Actuator, A 56, 1996,pp.19-29
10. Gao, C.; Xiang, X.-D." Quantitative m icrowave near-field m icroscopy o f dielectric
properties," Rev. Sci. Instruments, vol. 69, Nov. 1998, pp. 3846-3851.
11. Ghione G. ; Naldi, C. "Coplanar waveguides f o r M M IC applications: Effect o f upper
shielding, conductor backing, fm ite-extent ground planes, and line-to-line coupling,"
EEEE Trans. M icrowave Theory Tech., vol. 35, Mar. 1987, pp. 260-267.
12. G olosovksy, M.; Lann, A.; Davidov, D. "A M illimeter-W ave Near-Field Scanning
Probe with an Optical Distance Control." Ultramicroscopy 71 ,1998, pp. 133-141.
13. Gupta, K.C.; Bahl, I.; Garg, R.; Bhartia, P. “Microstrip lines and slotlines” Artech
House, 1996.
14. Gutmann, R. J.; Borrego ,J.; Chakrabarti, P . ; Wang, M "Microwave scanning
m icroscopy f o r p la n a r structure diagnostics," IEEE M TT-S,1987, pp. 281-283.
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
15. Harrison, R. R.; Koch, C. “A Robost A nalog VLSI M otion Sensor Based on the
Visual System o f the Fly.” Journal o f Autonomous Robots, V ol. 7 ,1 9 9 9 ,pp. 211-224
16. Husain, A.; and Ash, E. A."M icrow ave Scanning M icro sco p y f o r Non-Destructive
Testing," 5th European Microwave Conference, 1975.
17. Larin, A.; Golosovsky, M.; Davidov, D.; Frenkel, A ."M apping the thickness o f
conducting layers b y a mm-wave near-field m icroscope," IEEE M TT-S, 1998, pp.
1337-1340.
18. Ludwig, R.; Bretchko, P. “RF Circuit D esign”, Prentice Hall, 2000
19. Mead, C “Neuromorphic Electronic Systems”, Proc. IEEE,Vol 78, 1990, pp. 16291636.
20. Memmert, U.; Hodel, U.; Hartmann,U. "Com bined U ltrahigh Vacuum Scanning
Tunneling M icroscope Scanning Electron M icroscope System", Rev. o f Sci.
Instrument, vol. 67, June 1996, pp. 2269-73.
21. Mou, J.; Yang, J.; Shao, Z. "An O ptical detection L ow Tem perature A tom ic Force
M icroscope a t Ambient Pressure f o r B iological research", R.ev. Sci. Instrument, vol.
64, June 1993, pp. 1483-9.
22. Noma, T.; Iida, A."Surface Analysis o f L ayered Thin film s U sing a Synchrotron x-ray
M icrobeam com bined with a Grazing-exit Condition", R ev. Sci. Instrument, vol. 65,
April 1994, pp. 837-843.
23. Ponchak,G.; Akinwande, D.; Ciocan, R.; LeClair, S.; Tabib-Azar,M . "Evanescent
m icrowave p ro b e using coplanar and stripline f o r super-resolution im aging o f
integrated circuits," IEEE MTT-S Special Edition, Decem ber 1999.
23. Pozar, D. “M icrowave Engineering”, Addison W esley, 1990.
24. Steinhauer, D . E.; Vlahacos, C. P.; Dutta, S. K.; Feenstra B. J.; W ellstood ,F. C.
"Quantitative im aging o f sheet resistance with a scan ning near-field m icrowave
microscope," Appl. Phys. Letters, vol. 72, 1998, pp.861-863.
25. Synge,E. H.;Philosophy Magazine, vol. 6, 1928.
26. Tabib-Azar, M.; Shoemaker, N.; Harris, S. "Non-destructive characterization o f
m aterials by evanescent microwaves," Meas. Sci. Techn. 4, 1993, pp. 583-590.
27. Tabib-Azar, M.; Su,D. P . ; LeCIair,S.; Ponchak, G. "The ultim ate resolution o f super­
resolution im aging using evanescent m icrow ave probes," Rev. o f Sci. Instrument,
1999.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
94
28. Tabib-Azar, M.; Pathak, P . ; Ponchak, G. "Non-destructive super-resolution imaging
o f defects an d non-uniformities in metals, sem iconductors, dielectrics, composites,
an d p la n ts using evanescent microwaves" Rev. o f Sci. Instruments, vol. 70, June
1999, pp. 1-10.
29. Tabib-Azar, M.; Su, D. -P.; Pohar,A.; LeClair, S.; Ponchak,G. "0.4pm spatial
resolution with 1GHz evanescent m icrow ave p ro b e " Rev. o f Sci. Instruments, vol.
70, March 1999, pp. 1725-1729.
30. Tabib-Azar, M.; Akinwande, D.; Ponchak, G . ; LeClair, S. R .“N o v el Physical Sensors
U sing Evanescent M icrowave Probes.” R eview o f Scientific Instruments, V ol. 70
(8), 1999, pp. 3381-3386.
31. Tabib-Azar, M.; Garcia-Valenzuela, A.; Ponchak, G. “Evanescent Microwave
M icroscopy for High Resolution Characterization o f Materials”. Kluwer Academic
Publishing, Boston, M A (2000).
32. Tabib-Azar, M.; Su, D.P.; Pohar, A.; LeClair, S. R.; Ponchak, G. “0.4 pm Spatial
R esolution W ith 1 GHz (A.=30 cm) Evanescent M icrowave Probe.” R eview o f
Scientific Instruments, V ol. 70 (3),1999, pp.1725-1729.
33. Tabib-Azar, M.; Pathak, P. S.; Ponchak,G.; LeClair, S. R. “Nondestructive
Superresolution Imaging o f Defects and Nonuniformities in M etals, Semiconductors,
Dielectrics, Composites, and Plants U sing Evanescent M icrow aves.” Review o f
Scientific Instruments, V ol. 70 (6) ,1999, pp. 2783-2792.
34. Tabib-Azar, M.; Akinwande, D.; Ponchak, G.; LeClair, S. R. “Evanescent
M icrowave Probes on High-Resistivity Silicon and Its Application in
Characterization o f Semiconductors.” R eview o f Scientific Instruments, Vol. 70 (7),
1999, pp. 3083-3086
35. Tabib-Azar, M.; Akinwande, D.; Ponchak, G.; LeClair, S. R. “N o v e l Physical Sensors
U sing Evanescent Microwave Probes.” R eview o f Scientific Instruments, Vol. 70 (8)
1999, pp. 3381-3386.
36. Tabib-Azar, M.; Ciocan, R.; Ponchak G.; LeClair, S. R.“Transient Thermography
U sing M icrowave M icroscope.” R eview o f Scientific Instruments, Vol. 70 (8) 1999,
pp. 3387-3391.
37. Tabib-Azar, M.; Sutapun, B .“N ovel Hydrogen Sensors U sing Evanescent
M icrowave Probes.” R eview o f Scientific Instruments, Vol. 70 (9) ,1999, pp. 37073714.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95
38. Thanawalla, S.; Dutta, S. K.; Vlahacos, C. P.; Steinhauer, D. E.; Feenstra, B. J.;
Anlage S. M.;W elIstood, F. C. "Microwave Near-Field Imaging o f Electric Fields in a
Superconducting Microstrip Resonator." Applied Physics Letters, Vol. 73 (17), 1998
39. Wei ,T.; Xiang, X.-D.; Wallace-Freedman, W. G.; Schultz, P. G. "Scanning tip
m icrowave n ear-field microscope," Appl. Phys. Letters, vol. 68, June 1996, pp. 35063508.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bibliography
Acher O.; Baclet P.; Perin G. “Investigation o f the microwave permeability o f
ferromagnetic film laminations under a magnetic field” Journal o f Magnetism and
Magnetic Materials 157/158, pp. 480-481 (1996).
Aimoto, A . ; Masumoto,T. "Noninvasive method for measuring the electrical properties
o f deep tissues using an open-ended coaxial probe", Med. Eng. Phys., V ol. 18 pp641-646,
1996.
Alexander, J. S. "A Scanning M icrowave Microscope", MS Thesis, Princeton University,
1998.
Barett C. R.; N ix W.D.; Tetelman A.S. “The Principles o f Engineering materials”,
Pretince-H all, 1973
Bethe, H. A. "Theory o f Diffraction by Small Holes." Phys. Rev., 66(7), pp. 163-182
1944.
Binning G. “Atomic force microscopy” Physics Revue Letters 56, pp 930-933, 1986
Binning G. “Tunneling through a controllable vacuum gap” Appl. Phys. Lett. 40, pp.
178-180, 1982
Blenford, Don. "Reformulating for microwaves", Food Ingred., Process, 24. 1993.
Bradbury, S. “The Evolution o f the M icroscopy”, N ew York: Pergamon Press, 1967.
Burdette, E. C . ; Cain, F .L .; Seals, J. "In vivo Probe Measurement Technique for
Determining Dielectric Properties at VHF through Microwave Frequencies", IEEE Vol.
MTT-28, N o 4, April 1980.
Ciocan R . ; Tabib-Azar M. “Transient Thermography o f semiconductors using the
evanescent microwave microscope”, M icroscale thermophysical engineering pp 253262,1999
CO-NETIC & NETIC M agnetic shielding A lloys obtained from Magnetic Shield
Corporation, 740 North Thomas Drive, Bensenville, IL 60106, 1999.
Dekker A.J. Den. ; Bos, A.van Den."Resolution: A survey", J. Opt. Soc. Am. A 14, 3,
March 1997.
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
D ongpei S. "A Superresolution Multifrequency Evanescent M icrowave Probe", MS
Thesis, Case Western Reserve University, Cleveland, OH, 1999.
D uew er F . ; Gao C. “Tip sample distance feedback control in a scanning evanescent
microwave m icroscope” Applied Physics Letters 74 (18), pp. 2696-2698, 1999
Farrell, T.H. “A Preliminary Investigation into the Electrical Properties o f Human
Teeth”, Brit. Dent. J., May, 20, pp353-359,1958.
Gao, C . ; W ei T. “H igh spatial resolution quantitative microwave impedance m icroscopy
by a scanning tip microwave near-field m icroscope” Applied Physics Letters 71 pp.
1872-1874, 1999
Garcia-Valenzuela, A . ; Tabib-Azar, M. “Evanescent microwave probes for
nondestructive evaluation o f materials”, 1999
Hansma, P. “The scanning ion-conductance microscope” Science 243, pp 641-643,1989
H ew lett Packard “Operation and service guide HP 8648A ” , 1998
Jain, R.C.; V oss, W .A.G. "Dielectric measurement methods for industrial scintific and
medical applications in microwave frequency range", IETE Tech. Rev., 1 1 2 9 7 , 1994.
Ju, Y.; Saka, M. “M icrowave nondestructive detection o f delaminations in IC packages
utilizing open-ended coaxial line sensor”, NDT&E International 32 pp 259-264,1999
Kaatze, U . "Fundamentals o f Microwaves", Radiat. Phys. Chem., Vol. 45, N o 4, pp 539548, 1995.
ECrauss J. D . “Electromagnetism” Mc-Graw Hill, N ew York ,1984.
McCormick, J.B. “McCormick, McCormick, The atlas catalogue o f replica Rara Ltd.
Antique M icroscopes (1675-1840)”, Replica Rara limited, 1975.
Meyer, W . ; Schilz, W. ‘T easibility Study o f Density-Independent Moisture
Measurement with M icrowaves”, IEEE Transactions on MTT, Vol. M TT-29, N o. 7,
pp732-739, 1981.
Muller, R. A . "Applications o f M icrowave Imaging in Very High Resolution
Characterization o f Semiconductors and Superconductors.”, MS Thesis, Case Western
Reserve University, Cleveland, OH, 1999.
Mumford, J.M. "Resistivity o f human enamel and dentin", Arch Oral B iol., Vol. 12,
pp925-927, 1967.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
MWOFFICE™, A pplied W ave Research Inc., 1960 E. Grand A venue, Suite 530, El
Segundo, CA 90245.
Naoya, H . ; Y oshio, N . ; Keiji, K . ; Shigeyuki, E. " Application o f M icrowaves and
M illimeter Waves for the Characterization o f Teeth for dental D iagnosis and treatment",
IEEE MTT-Vol. 46, N o. 6, pp834-838, 1998.
N icollian, E.H.; B rew s J.R “M OS Physics and Technology”, A W iley-Interscience
publication 1990
Olympus Optical C o. Ltd, “Olympus Scanning Acoustic M icroscope UH3 instruction
manual”. p p l720 — 1731,1996.
Ramo, S . ; Whinnery, John R . ; Duzer, T V. “ Fields and W aves in communication
electronics”, 3rd edition, John W iley & Sons, Inc., pp256-259.
Rogers B. M .; D ylan F. W. "A general Waveguide Circuit Theory", J. Res. Natl. Inst.
Stand. Technol., V ol. 97, pp533-561, 1992.
Rogers Corp., Chandler, Arizona.
Romanofsky, R."Analytical and experimental procedures for determining propagation
characteristics o f m m -w ave GaAs microstrip lines", N A SA TP-2899, 1989.
Schroder, D.K. “Sem iconductor Material and D evice Characterization” , John W iley
N ew York, 1990
Schulte, A . ; Gente, M . K. “Posteruptive development o f electrical resistance o f human
premolars”, Caries Research, pp272-273, 1996.
Schulte, A. ;Gente, M . ; Pieper, K . ; Arends, J. “The electrical resistance o f enameldentin cylinders. Influence o f N aC l content in storage solutions”, J. Dent., Vol. 26,
p p ll3 -1 1 8 , 1998.
Sim onyi, K. Foundations o f Electrical Engineering: Fields-Networks-W aves, A
Pergamon Press B ook (the M acm illan Company, N ew York), pp526-536.
Soohoo R. F. “ A M icrow ave M agnetic Microscope” Journal o f A pplied Physics 33, 127
pp 510-512, 1962.
Soohoo, R.F. “A M icrow ave M agnetic Microscope”, J. Appl. Phys., V ol. 33, ppl2761277, 1962.
Sparrow, C. M. "On spectroscopic R esolving Power", Astrophys. J. V ol. 44, pp76-86,
1916.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
Stamm, J. W., "Is there a need for dental sealants? Epidemiological indications in 1980s",
J. Dent. Educ., 48, pp9-17,1984.
Steinhauer D . E . ; Vlahacos C.P. “ Quantitative imaging o f sheet resistance with a
scanning near fiel microwave microscope”, Applied Physics Letters 72 (7) pp. 861-863,
1998
Stephenson, R. J. "A Near-field Optical M icroscope with Normal Force Distance
Regulation", R ev. Sci. Instrument., Vol. 67(11), November 1996, pp3891-3897.
Sze, S. M. “Physics o f Semiconductor D evices”, W iley, 1981
Tabib-Azar M. “Integrated Optics and Microstructure Sensors” Kluwer Academic
Publishings, Boston 1995.
Tabib-Azar M. “Theoretical resolution o f Imaging with Evanescent Microwave Probes.”
IEEE Trans. On Instrumentation and Measurement, D ec. 1999.
Tabib-Azar M.; Akinwande D.; Ponchak, G . ; LeClair, S. R. “Evanescent microwave
probes on high-resistivity silicon and its application in characterization o f
semiconductors”Review o f Scientific Instruments J u ly 1999, Volume 70, Issue 7, pp.
3083-3086, 1999.
Tabib-Azar M.; Ciocan R.; Ponchak G . ; LeClair, S. R.“Transient thermography using
evanescent microwave microscope” R eview o f Scientific Instruments —August 1999 —
V olum e 70, Issue 8, pp3387 3390, 1999
Tabib-Azar M.; Ciocan R.;Ponchak G . ; LeClair, S. R.“Evanescent microwave imagining
in transmission and reflection modes” submitted to R eview o f Scientific Instruments,
1998
Tabib-Azar M.; Garcia-Valenzuela A.; Ponchak G. “Evanescent M icrowave M icroscopy
for High Resolution Characterization o f Materials” Kluwer Academic Publishing,
Boston, M A ,1999.
Tabib-Azar, M . "Applications o f Ultra-High Resolution Evanescent Microwave Imaging
Probe in Nondestructive Evaluation o f Materials", A SNT, to be published.
Tabib-Azar, M . ; Bumrerraj, S . ; Katz., J .L .; You, S. H. "Evanescent Electromagnetics:
A N ovel, Super-resolution, and Non-Intrusive, Imaging Technique for Biological
Applications.", Biomedical M icrodevices, (In Press).
Tabib-Azar, M . ; Sutapun, B .“N ovel hydrogen sensors using evanescent microwave
probes”, Rev. Sci. Instrum., Vol 70, N o 9, September 1999.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
Takeuchi T . ; W ei T. “Low temperature scanning-tip microwave near-field microscopy o f
Y B a 2 Cu 3 0 7 -x film s” Applied Physics Letters 68(24). Pp2026-2028, 1997
U stinov, V .V .; Rinkevich, A. B . ; Romashev, L.N .”M icrowave magnetoresistance o f
Fe/Cr multilayers in current-perpendicular to plane geometry” Journal o f Magnetism and
M agnetic Materials 198/199, pp. 82-84 (1999).
V lachos C.P.; Black R.C. “Near-field scanning m icroscope with 100mm resolution”
Applied Physics Letters 69(21) 1996
Volkmann, H. "Ernst Abbe and his Work", Applied Optics, vol.5, N o .l 1, N ovem ber,
W ei, T.; Xiang, X. “ Scanning tip m icrowave near-field microscope” Applied Physics
Letters 68(24). 1996
W illiam s, C. C. Proc SPEE 897, 129 1988
Y ogi, R .A .; Gangal, S.A ; . Karekar, R.N. ; Aiyer, R.C. "Study o f bi-coloured leaves by
m icrowave microstripline moisture sensor", Sensors and Actuators B, 55, pp 186-190,
1999.
Y ogi, R . A . ; Gangal, S.A ; Aiyer, R.C.; Karekar, R.N. "Microwave ring resonator as a
novel bio-material moisture sensor", Sensors and Actuators B, 50, pp38-44, 1998.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
4 027 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа