PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DateDuepBS-sz

SCfl

SCANNING PROBE ELECTRICAL CONDUCTIVITY MEASUREMENTS USING
SUBMICRO-ELECTRODES FABRICATED BY
NONLITHOGRAPHIC TECHNIQUES
By

Akiko Kubota

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
Department of Electrical and Computer Engineering

2003

SC ANNL‘

The measurt
can be realii
2000 CO;

containing a
then pulled I
the second

diameters.

Others inclut
be COUSImm
and [DC SUbg
fabricated, I
Sl'Stem. Thi
entire 33mm
\mationS. St

ABSTRACT
SCANNING PROBE ELECTRICAL CONDUCTIVITY MEASUREMENTS USING
SUBMICRO-ELECTRODES FABRICATED BY
NONLITHOGRAPHIC TECHNIQUES
By

Akiko Kubota

The measurement of electrical conductivity of a substance at extremely localized areas

can be realized through utilization of micron-scale probes fabricated using the Sutter® P—
2000 C0; laser micropipette puller. These probes are composed of a glass tube
containing a metal wire. Using the micropipette puller, the glass/metal combination is
then pulled in two steps. The first step is to pull down the glass tube around the wire and
the second step is a final pull to achieve the tip dimension of micron or sub-micron
diameters. By manipulating the parameters on the micropipette puller, which among
others include the heat of the laser and the force of the pull, this type of “nano-probe” can
be constructed. The goal of this thesis was both the formulations of a fabrication process
and the subsequent application of the nano-probe. After the nano-probe was successfully
fabricated, it was employed in a scanning probe electrical conductivity measurement
system. This system is very useful for obtaining electrical conductivity variations over an
entire sample (up to 1” x 1”) and for investigating nonuniforrnities (in the form of doping
variations, surface cracking, etc.) that might exist in a sample. The design and testing of

this system will be presented, along with data from several samples.

Dedicated to

my family

iii

I We
excellent
throughout
patience. tl
have helpeu
technical ut
Aslam and
suggestions
Robert PClt
facility. Tl
other memt
and special
towards my

My
and Takuo
brought me
N'Iab’fihlro Q
Muhammad

SlUdy at th

ACKNOWLEDGMENTS

I would like to express my sincere thanks to my advisor Dr. Tim Hogan for all his
excellent guidance, encouragement, friendly manner, and professional example
throughout my master studies at Michigan State University. Without his support and
patience, this work would not have been possible. His inspiring efforts and knowledge
have helped resolve a number of project-related issues, and provided me with a better
technical understanding of certain key concepts. I would also like to sincerely thank Dr.
Aslam and Dr. Reinhard for serving as my graduate committee and giving me valuable
suggestions to both the thesis and experimental designs. I also thank Ungsik Kim and
Robert Pcionek for their kind help and instruction at the DER and Optics cleanroom
facility. The success of this work is also based on the support and friendship from all
other members in our research group: Fu, Chun-I, Luke, Hyun, Nishant, Nick, and Angelo
and special thanks goes to, Sim Loo and Sangeeta Lal for their support and inspiration
towards my research work.

My deepest thanks are extended to my family: Kazuhisa Kubota, Sachi Kubota,
and Takuo Kubota for their unconditional love and continuous support which have
brought me up to this point. And also I would like to deeply thank my friends: Ari Gajraj,
Masahiro Otani, Kathy Pui-Kwan Lai, Chigusa Ohbuchi, Paul Kabir, Mei Wang, Zarini
Muhammad, and Sanghoon Rhee for their friendship and support throughout my graduate

study at Michigan State University.

iv

TABLE
LIST OI

LIST 0]

TABLE OF CONTENTS

TABLE OF CONTENTS .................................................................................................... v
LIST OF FIGURES ........................................................................................................... vii
LIST OF TABLES ............................................................................................................. xi
1. INTRODUCTION ....................................................................................................... 1
1.1 Thermoelectric effect .......................................................................................... 1
1.1.1 Research objective ........................................................................................... 5

1.2 History of micropipette puller ............................................................................. 6

2. THEORY ..................................................................................................................... 8
2.1 Four point probe conductivity measurement ....................................................... 8
2.1.1 Bulk sample ..................................................................................................... 8
2.1.2 Thin sheet or wafer ........................................................................................ 10

2.2 Composite material ........................................................................................... 13
2.3 Force calculation on the circular cross-sectional cantilever beam .................... 16

3. MICROPIPE'I'I‘E APPLICATIONS ......................................................................... 22
3.1 Micropipettes with metal filaments ................................................................... 22
3.1.1 Micropipette-based submicrometer thermocouple ........................................ 22
3.1.2 Cantilevered glass multiple-wire force-sensing thermal probes. .................. 25

3.2 Micropipettes without metal filament ............................................................... 26
3.2.1 The tip of Near-Field Scanning Optical Microscopy .................................... 26
3.2.2 Scanning nanolithography using a material-filled nanopipette ..................... 32

4. FABRICATION AND APPLICATION OF MEMS MICRO CANT ILEVER ......... 39
4.1 MEMS microfour-point probe .......................................................................... 39
4.1.1 Fabrication of micro four point conductivity probes ..................................... 41

4.2 Scanning nano four-probes or nano tweezer ..................................................... 43

5. MATERIALS AND METHODS .............................................................................. 48
5.1 Electrode fabricated by micropipette puller ...................................................... 48
5.1.1 Material selection .......................................................................................... 48
5.1.2 Laser Specification and variables .................................................................. 53
5.1.3 Fabrication procedure using P-2000 .............................................................. 56
5.1.4 Pulling steps .................................................................................................. 58

5.2 Other Fabrication technique to make microelectrode ....................................... 61
5.2.1 Forming V-groove on the silicon substrate ................................................... 62
5.2.2 Preparing the glass fiber metal evaporated electrode probe .......................... 69
5.2.3 Results ........................................................................................................... 73

6. DAT
6.1
6. l .l
6.2
6.2.1
6.2.2

6.3

6.3.3
6.3.4
6.3.5
6.4
6.4.]
6.4.2
6.4.3

7. CO}
7.1
7.2

7.2 ‘

7.2..

8- APF

9- BIB:

6. DATA ANALYSIS ................................................................................................... 76

6.1 One-point probe scanning measurement method .............................................. 76
6.1.1 The probe mounting angle calculation .......................................................... 77
6.2 DC sweeping current source using Keithley 2400 source meter ....................... 79
6.2.1 B12T63 ............................................................................................................ 79
6.2.2 Aluminum foil ............................................................................................... 84
6.3 AC source using EG&G 7265 DSP (Digital Signal Processor) Lock-In
Amplifier ........................................................................................................... 87
6.3.1 Lock-in Amplifier .......................................................................................... 87
6.3.2 Setup procedure using EG&G 7265 DSP (Digital Signal Processor) Lock-In
Amplifier ....................................................................................................... 89
6.3.3 Copper strip ................................................................................................... 92
6.3.4 Graphite sample ........................................................................................... 101
6.3.5 B12TC3 .......................................................................................................... 106
6.4 Data acquisition Instrumentation ..................................................................... 119
6.4.1 Keithley Model 2400 SourceMeter® .......................................................... 119
6.4.2 Model 2182 Keithley Nanovoltmeter .......................................................... 120
6.4.3 EG&G 7265 DSP (Digital Signal Processor) Lock-In Amplifier ............... 120
7. CONCLUSION AND SUGGESTIONS FOR FUTURE WORK ........................... 122
7.1 One point probe scanning conductivity measurement system ......................... 122
7.2 Suggestions for future work ............................................................................ 122
7.2. 1 Heat treatment ............................................................................................. 122
7.2.2 Study of a compound semiconductor with the domain including doping
profile............ .......................................................................................... 123
7.2.3 Pulled pipette as a thermocouple ................................................................. 124
8. APPENDIX ............................................................................................................. 127
9. BIBLIOGRAPHY ................................................................................................... 129

vi

Figure 1.1
Figure 1.;
Figure 2.1

me2-

J

,
Figure 2.-

Flame 2.4

Figure 2.7
fingS
Figure 3.]
Figure 3.2
Figure 3.3
Hflm34
Ham3s

Figure 3.6

Figure 1.1
Figure 1.2
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 2.5
Figure 2.6
Figure 2.7
Figure 2.8
Figure 3.1
Figure 3.2
Figure 3.3
Figure 3.4
Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 3.10

LIST OF FIGURES

Thermoelectric couple to form a module ....................................................... 2
Principle of thermoelectric cooling ............................................................... 3
Schematic of 4-point probe configuration in bulk crystal ............................. 9
Schematic of 4-point probe configuration in thin wafer. ............................. ll
Strain ............................................................................................................ 14
Total force including force acting on the matrix and on the fiber. .............. 14
Cross sectional area of the composite, the matrix, and the fiber. ................ 16
Cantilever beam and its axis orientation ...................................................... 17
The moment of inertia of circular beam ...................................................... 19
Slope and deflection for cantilever beam .................................................... 21

Schematic presentation of the sensing area of the pipette thermocouple. 23

Overview of the fabrication of a double-wire thermal resistive probe. ....... 25
ATR illumination from the back of the transparent sample. ....................... 28
Various methods of NSOM image capturing technique. ............................ 29
Dye-embedded sol gel ion-sensing cantilever micropipette. ....................... 31

Schematic diagram of the photoresist-filled nanopipette controlled by shear
force position detection ............................................................................... 34

Principle of the utilized optical shear force detection applied to the
micropipette tip with a droplet near the apex . ............................................ 35

Resonant frequency shift of the liquid-filled nanopipette (curve 2) with
respsect to that of the empty pipette (curve 1). ............................................ 36

Two fabrication schemes: (a) point drawing and (b) line drawing . ............ 37

Three photoresist dots of 300 nm size fabricated on the gold-sputtered glass
substrate. (a) Topographic image of the dots, and (b) height profile across
the line in (a). ............................................................................................... 38

vii

F1 gure
Figure

Figure

Figure
Figure
Figure
Figure
Figure
Figure
Figure
FiguIe
Figure
Figure .
I”Figure .

Flgure .

Figure 4.1
Figure 4.2

Figure 4.3

Figure 4.4
Figure 4.5
Figure 5.1
Figure 5.2
Figure 5.3
Figure 5.4
Figure 5.5
Figure 5.6
Figure 5.7
Figure 5.8
Figure 5.9
Figure 5.10
Figure 5.11
Figure 5.12
Figure 5.13

Figure 5.14

Figure 5.15

Figure 5.16

(a) Schematic of the experimental set up. .................................................... 40
The major steps in the microfabrication process. ........................................ 42

(a) Schematic of the nanotip deposited in a plane perpendicular to the plane
of the microcantilevers. (b) SEM image of micfrofour-point probe. (c)
SEM image of nanotwo probe with 200 nm spacing. (d) Close up of 200
nm gap two-probe. (e) SEM image of nanofour point probe with an average

spacing of 350 nm ........................................................................................ 44
The nanofabrication process. ....................................................................... 46
The circuit set up of nanotweezer. ............................................................... 47
The metal breaking at the tip ....................................................................... 50
The metal boiling in the tube ....................................................................... 50
The pulled micropipette with platinum filament inside ............................... 51
The delay value less than 128 millisecond. ................................................. 55
The delay value greater than 128 millisecond. ............................................ 55
Left puller bar of P-2000 micropipette puller. ............................................. 57
Heat is applied by laser and pulling force is activate from both ends ......... 57
First pull for making composite material ..................................................... 58
Final pull ...................................................................................................... 59
SEM picture of gold nanoelectrode with the tip dimension of 300nm. ....... 60
Cut silicon in rectangular shape ................................................................... 64
Silicon substrate with photoresist coated ..................................................... 64
(a) UV lithography side view, (b) UV lithography top view ..................... 65
(a) Photoresist developing, (b) Patterning oxide layer, (c) Stripping

photoresist .................................................................................................... 66
Silicon anisotropic etching. ......................................................................... 66
V-grooves created by KOH etching. ............................................................ 67

viii

'J\

Figure

U\

Figure

(J\

Figure

Figure 5.
Figure 5.

Figure 5.
Figure 6.
Figure 6.
Figure 6.
Figure 6.

Figure 6.

FiSure 6.
Figure 6.
me6
Hflmox
FiSure 6.
PiSure 6_

fiflmo.

Figme 6.:

Hflmoi

Figure 5.17

Figure 5.18
Figure 5.19

Figure 5.20

Figure 5.21

Figure 5.22
Figure 6.1
Figure 6.2
Figure 6.3
Figure 6.4

Figure 6.5

Figure 6.6
Figure 6.7
Figure 6.8
Figure 6.9
Figure 6.10
Figure 6.11

Figure 6.12

Figure 6.13

Figure 6.14

Figure 6.15

(a) The top view of V-groove after removing oxide. (b) The side view of

V-groove after removing oxide mask. ......................................................... 67
Triangle relations on the V-groove .............................................................. 68
The glass fibers mounted on V-groove (a) Top View, (b) Side view ....... 69
Sample covered with Shadow mask for metal evaporation

(a) Top view, (b) Side view ...................................................................... 70
(a) Titanium deposition, (b) Gold deposition, (c) final look after metal
depositions ................................................................................................... 72
The circuit of fabricated four point probe .................................................... 73
Calculation of the approximate cantilever radius. ....................................... 78
DC conductivity measurement system set up. ............................................. 79
Sample set up. .............................................................................................. 80
Plot of scanned voltage difference on BizTe3 sample using DC technique. 82
Plot of the linear region of scanned voltage difference on BizTe3 sample
using DC technique ...................................................................................... 83
The electrical conduction test of submicron probe. ..................................... 86
The demodulation of two Signals ................................................................. 88
Lock-in amplifier set up to determine the current of the system ................. 90
Lock-in amplifier set up to measure the scanned voltage of the sample. 90
Lock-in amplifier set up picture ................................................................... 91
Oscilloscope connected to the output of lock-in amplifier. ......................... 93
The signals captured by Oscilloscope: A) channel 1 B) channel 2 (20
mV/div) ........................................................................................................ 94
Lock-in amplifier circuit analysis ................................................................ 95
Plot of scanned voltage difference of copper using AC technique and pin
probe and Pt wire probe. .............................................................................. 97
SEM picture of gold nanoelectrode ............................................................. 98

ix

1:: ure

(IQ

Figure

Figure
Figure
Figure

Figure

Figure

Figure

Figure I

Figure ‘

Figure 6.16

Figure 6.17

Figure 6.18
Figure 6.19
Figure 6.20

Figure 6.21

Figure 6.22

Figure 6.23

Figure 6.24

Figure 7.1

Plot of scanned voltage difference of copper using AC technique and gold

submicron electrode (d=700nm). ............................................................... 101
Plot of scanned voltage difference of graphite using AC technique and Pt

microelectrode. .......................................................................................... 105
Scanning location of each run .................................................................... 109

Plot of scanned voltage difference of BizTe3 using AC technique and pin. 109
Setup of the differential voltage measurement. ......................................... 1 11

The set up of differential voltage measurement connecting node B on the
top surface .................................................................................................. 112

Plot of scanned voltage difference of new BizTe3 using AC technique and
pin probe. ................................................................................................... 113

Plot of scanned voltage difference of new B12T63 using AC technique and
pin probe for contact resistance analysis. .................................................. 116

Plot of scanned voltage difference of new BIzTC3 using AC technique and
Pt microprobe. ........................................................................................... 118

Doping profile/ striation ............................................................................ 124

Table 2.1
Table 5.3
Table 5.4
Table 5.5
Table 6.1

Table 6.2

Table 6.3

Table 6.4

Table 6.5

Table 6.6

Table 6.7

Table 6.8

Table 2.1
Table 5.3
Table 5.4
Table 5.5
Table 6.1

Table 6.2

Table 6.3

Table 6.4

Table 6.5

Table 6.6

Table 6.7

Table 6.8

LIST OF TABLES

Correction factor k for the measurement of sheet resistance ....................... 12
Five programmable parameters on P-2000 .................................................. 53
FILAMENT scan pattern values .................................................................. 54
Dry and wet oxidation coefficients for silicon ............................................. 63

Data of scanned voltage difference on B12T63 sample using DC technique.8l

Data of scanned voltage difference of copper using AC technique and pin
probe and Pt wire probe. .............................................................................. 96

Data of scanned voltage difference of copper using AC technique and gold
nanoelectrode. ............................................................................................ 100

Data of scanned voltage difference of graphite using AC technique and Pt
microelectrode. .......................................................................................... 104

Data of scanned voltage difference of BizTe3 using AC technique and pin. 108

Data of scanned voltage difference of new BizTe3 using AC technique and
pin probe. ................................................................................................... 113

Data of scanned voltage difference of new BizTe3 using AC technique and
pin probe for contact resistance analysis. .................................................. 115

Data of scanned voltage difference of new B12T63 using AC technique and
Pt microprobe. ........................................................................................... 1 17

xi

mater
sensit
the de
used 1

IldITm

Cooler
Solid
effect
0f the
Cooler
applic
Empe
aPplic
COolin
akgp
gradie

SUPP“.

1. INTRODUCTION

1.1 Thermoelectric effect

Transport measurements are essential to the development of novel electronic
materials and electronic devices. In semiconductor materials, these measurements can be
sensitive to parts per million concentrations of dopants. They further give us insight into
the development of accurate models for materials and their structures which can then be
used for engineering new devices. Of particular interest to our laboratory has been new

narrow bandgap materials which exhibit promising thermoelectric transport properties.

A thermoelectric (TE) cooler, sometimes called a thermoelectric module or Peltier
cooler, is a semiconductor-based electronic component that functions as a heat pump.
Solid State thermoelectric coolers have been studied since the discovery of the Peltier
effect in 1834, however, the devices became practical only recently with the development
of more efficient semiconductor thermocouple materials [1]. Although thermoelectric
coolers are not the solution for every cooling system, they are suited to wide variety of
applications. Their advantages include small size, low cost, low weight, wide operating
temperature range, or precise temperature control, and high reliability. Traditional
applications of TE coolers have included temperature control of infrared detectors,
cooling of electronic components (IC’S), and small (personal) coolers/refrigerators. It is
also possible to utilize thermoelectric modules for power generation when a temperature
gradient is established across the module. This has found applications in satellite power

supplies, remote site power generations, and waste heat power recovery.

 

The It
often 3110."CC

material 1mm

The t
array of p at
them. The 1
optimization

strength. Or

Fig

The Iherrm
MultjCOup.
the Couple
COmain 01‘.
increaSe tl

aircrew“

The most common thermoelectric material is bismuth telluride, (BizTe3), which
often alloyed with selenium and antimony, is the highest efficiency thermoelectric

material known for near room temperature applications.

The typical thermoelectric module consists of two thin ceramic plates with an
array of p and n doped bismuth-telluride semiconductor materials sandwiched between
them. The ceramic materials on both Sides of the thermoelectric module provide
optimization of the electrical insulation and thermal conduction with high mechanical

strength. One p-type leg and one n-type leg make up a couple, as shown in Figure 1.1.

 

 

 

 

 

 

 

 

 

Heat Sink

Figure 1.1 Thermoelectric couple to form a module

The thermoelectric couple is electrically connected with metallic strips to form a module.
Multicouple modules can be fabricated for larger heat pumping capabilities by connecting
the couples electrically in series and thermally in parallel. A thermoelectric module can
contain one to several hundred couples or more. Modules can be mounted in parallel to
increase the heat transfer effect or can be stacked in multistage cascades to achieve high

differential temperatures.

Once a s“:

 

in Figure 1.2. I1
bottom of a mo
module to the I“
but they also car
energy from the
of the electrons

scattering event

P-t
set

Fig“ ['9

Once a small DC voltage is applied across the couple in the configuration shown
in Figure 1.2, the majority carriers of each leg will be transferring from the top to the
bottom of a module. In the case of n-type leg, electrons transfer from the top of the
module to the bottom of the module. These electrons not only carry charge as the flow,
but they also carry heat. This transfer of heat is the result of electrons absorbing vibration
energy from the surrounding crystal through scattering events. Since the average motion
of the electrons is from the top to the bottom, the thermal energy they carry to their next

scattering event also flows from the top to the bottom in this configuration.

Heat absorption
Metallic electrode

/////////‘J

speweoonductor—S @613: §@g E gemndutoor

t (9 t t G t
I. J.
-——'” //// /////

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

@ .@
Heat generation
. l
I | Current

\/

Figure 1.2 Principle of thermoelectric cooling

The e
[Hype materi
determined
thermoelectri
operational It
contains thes

Zt'viith units

Where S is ti
of the semi
therefore the
0f merit Car
lhtrmal CQm
majority Cari
cold 10 Tim S
a higher Valt
R and thus

desired IO he
lhfi maleu]

The efficiency of this device is dependent on a combined effect of the n-type and
p-type materials used in making the module. The quality of a particular materials can be
determined through the materials properties of absolute Seebeck coefficient (or
thermoelectric power), electrical conductivity, and thermal conductivity over the
operational temperature range between the cold and hot sides [2]. The expression that
contains these material parameters is referred to as the figure-of—merit, and is denoted by

Z (with units of reciprocal Kelvin)

Sza

 

Z= (1.1)

where S is therrnopower, G is electrical conductivity, and 1C is thermal conductivity. Each
of the semiconductor material properties varies as a function of temperature, and
therefore the figure-of-merit for each material is also temperature dependent. The figure
of merit can be improved by increasing the electrical conductivity and decreasing the
thermal conductivity. By having higher value of electrical conductivity, it ensures more
majority carrier to transport from the top to bottom of the module, or pump heat from the
cold to hot side. In other words, a higher value of mobility, It, could also be the cause of
a higher value of electrical conductivity, 0', and also higher 0' does reduce the resistance,
R, and thus 12R losses are reduced. In the case of thermal conductivity, lower values are
desired to help insulate between large temperature gradients. Also it can be shown that
the maximum temperature differential that can be achieved by a single pair of n- and p-

type materials is directly proportional to the “temperature averaged” figure-of-merit of

9"”

each semicc

objective in

H.) Resea

At
Engineering
Department
researchers
semicondut
fabricated.
Characteriz:
prm'iOUS Se
Conductis'ii

Thi
CODductim
COmPOUnd:
SijCOndu
diSCuSSed i
dimenSiOni
small Slim]
fabTICarjOn
Pulling de'

Probe meg“

each semiconductor material. Therefore, maximizing the figure-of-merit is the major

objective in the selection and optimization of thermoelectric materials.

1.1.1 Research objective

At Michigan State University, members of the Electrical and Computer
Engineering Department, the Chemistry Department, and the Physics and Astronomy
Department are collaborating through research on new thermoelectric materials. The
researchers from the Chemistry group are studying the fabrication of compound
semiconductors, and have discovered many new materials. After a new sample is
fabricated, researchers from the Electrical and Computer Engineering Department
characterize the sample to determine the unit less figure of merit, ZT. AS discussed in the
previous section, ZT value includes the values of thermal conductivity, the electrical
conductivity and the therrnopower of the sample.

This thesis focuses on the development of scanning probes for electrical
conductivity characterization of new electronic materials such as the thermoelectric
compounds discussed above. The electrical conductivities of the compound
semiconductors can be measured by a four-point probe method, which will be further
discussed below. Because the sample often has dimensions of small single crystals with
dimensions near [10pm x 50pm x 1mm], in order to measure the conductivities of such a
small sample, voltage probes with submicron diameters have been developed. In the
fabrication process of these probes, a laser-based micropipette puller P-2000 is used as a
pulling devices. In this thesis, submicron probe fabrication procedures and scanning

probe measurement data are presented and discussed.

1.2 History

Micrc
beveller wen
manufacture
technical bar
micropipette
submicron ti]
to produce 51
cellular level
many of thc
fabrication t
analytical tee
in nanofabrit
fOCus beyonc
We used i
Immufaclurir
infertility Ire

1.2 History of micropipette puller

Micropipette fabrication devices, including micropipette puller and micropipette
beveller were first produced by the Sutter Instrument company in order to develop and
manufacture precision electromechanical devices that could be used to overcome
technical barriers associated with electrophysiology experimental techniques [3]. The
micropipette puller is capable of melting and pulling a glass (or quartz) tube down to
submicron tip dimensions while maintaining the tubular Shape, thus they have been used
to produce some of the fundamental tools necessary to perform biological research at a
cellular level. In addition, micropipettes pulled by a micropipette puller can be found in
many of the major scientific research organizations in the world. Micropipette
fabrication technology accelerated the pace of biomedical research and emerging
analytical techniques that utilize submicron probes and pipettes. The tremendous growth
in nanofabrication methods, optical research, and imaging techniques expanded market
focus beyond the boundaries of the biological research sciences. Micropipettes are now
being used in a broad range of applications including, but not limited to, optical probe
manufacturing, semiconductor research, electronics manufacturing, clinical diagnostics,
infertility treatment, atmospheric research, astronomical imaging, fusion research, and
material analysis.

In order to fabricate electrodes for two point, or four point probe conductivity
measurements, the model P-2000 micropipette puller was selected for the following

reasons:

 

il-

Th

CH

1111

C0

161

[ill

o P-2000 laser based Micropipette / fiber puller

The P-2000 integrates a C02 laser-based heat source with the technology derived from the
extensive experience with conventional pullers. This system offers capabilities
unmatched by other pullers. While the P-2000 is suitable for working with most
conventional glasses, its primary advantage is the ability to work with quartz glass (fused
silica). Quartz offers superior material properties for a variety of research applications. It
is stronger than other glasses and can facilitate penetration through tough tissues which
would normally break conventional pipettes [4]. Quartz can operate at higher
temperatures and has been found to yield lower noise probes. For this reason, quartz is
the best material to use in fabricating the nano-electrode cantilever beam used to measure
the electrical conductivity. A C02 laser was selected as the heat source for the P-2000 for
several reasons:

1) The nominal emission wavelength of the laser approximates the resonant
frequency of the SiOz lattice in glass. Thus quartz and other conventional glasses
can be melted when the appropriate laser power is supplied.

2) Laser heat is clean and leaves no metal residue on the pipette as do conventional
heating filaments.

3) Laser heat can be turned off instantly, leaving no residual filament heat.

4) The user can program the amount and distribution of heat supplied to the glass.

The additional versatility of quartz for other research projects was the main reason for

choosing the P-2000.

 

2.1 Four point 1‘
Four-poim
electrical pIOPEm

typical four-poi!“

 

consists of two 011
measuring the \‘01
be derived from t‘.
related to the thicl
of the tip dimens
created by the pro

For the w;
“*0 P0111! probe 1
not be easily me:
not be. accurately
Sllffiading resistai

calculation of the

T .. .
he reustititv n

Probe techniqu€

2. THEORY

2.1 Four point probe conductivity measurement

Four—point measurements have played an important role in understanding the
electrical properties of solid state bulk materials and films for many decades [5]. A
typical four-point configuration is a linear array of four equidistance electrodes. It
consists of two outer electrodes performing as source and drain, and the inner electrodes
measuring the voltage difference [6]. The conductivity, the reciprocal of resistivity, can
be derived from the voltage-to-current ratio, while compensating for geometrical effects
related to the thickness, size, and shape of the sample. Because of the infinitesimal size
of the tip dimension, compared to the whole sample dimension, the contact resistance
created by the probe tip can be safely neglected over the total impedance of the sample.

For the wafer conductivity measurement, the four point probe is preferable over a
two point probe because the contact resistances associated with the two point probe can
not be easily measured [7]. With a two point probe method, the true sheet resistance can
not be accurately separated from the measured resistance. Since very little contact and
spreading resistance is associated with the voltage probes, one can obtain a fairly accurate

calculation of the sheet resistance which is then used to calculate the resistivity.

2.1.1 Bulk sample

The resistivity measurements is made on the flat ends of the crystal by the four-point

probe technique [8].

Minnow-l.
a

it

 

 

 

 

 

 

 

 

 

Bulk crystal

 

 

\
a
‘\ ’I
‘~ -a'
‘~-—.._..'

    

Figure 2.1 Schematic of 4-point probe configuration in bulk crystal

The assumption made in this measurement is that, the metal tip is infinitesimal and
samples are infinite in lateral dimension. For bulk sample where the sample thickness t
>> s, the probe spacing, a spherical protrusion of current is emanated from the outer

probe tips. The differential resistance is [9]:

dx
AR = — 2.1
pl .1 t .
The differential resistance R is integrated between the inner probe tips, (where the

voltage is measured):
x2
dx ,0 1 ‘2 1
R : : —- ——
LO 27er 27r£ x)

=__p_ 22
x] 25275 ()

 

 

where probe spacing is uniformly 5. Thus the distances of x] and x2 are, s and 2s
respectively (Figure 2.1).

Since the surface area of the sphere is 41tr2, the hemisphere would be the half
value of it, which is 21tr2, and dotted lined hemisphere under the bulk substrate represents
the current flow. Thus the area integrated is 21th2 where x is the radius of the Sphere
originated from the I+ probe tip. Due to the superposition of current at the outer two tips,

R = V/ZI. Thus the expression for bulk resistivity is:

p = 27%;} (2.3)

In this case, the assumption is that the wafer is infinite in dimension, with respect to the
probe spacing, however in the reality, wafers are not infinite in extent. Therefore the

right hand side of the equation must be multiplied by a combination of correction factors.

2.1.2 Thin sheet or wafer

For a very thin layer (thickness t<<s), the current is emanated from the outer

probe tip in ring Shape as seen in the figure 2.2.

10

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 2.2 Schematic of 4-point probe configuration in thin wafer.

Therefore the expression for the area A = 21txt, where x is the radius of the circle and t is
the thickness of the sample. So the area integrated is the periphery of the dotted

cylindrical structure. The derivation is as follows:

x2 25 2s
dx ,0 dx p p
R: —= ———=—lnx =—ln2
JP 27er 5 2m x 2m ( i 3 2m (2'4)

Consequently, for R = V/ZI, the sheet resistivity for a thin sheet is:

m V
=— — 2.5
,0 ln2[l)( )

Here, it is noted that this expression is independent of the probe spacing s. In general, the

sheet resistance Rs (=p/ t) can be expressed as:

Rs = 4%] (2.6)

where V is the dc voltage across the voltage probes (in volts); 1 is the constant dc current

passing through the current probes (in amperes), and the factor k is a correction factor or

11

also often called the geometric factor. In the case of an infinite thin sheet, k=4.53, which
is rr/ln2 from the derivation. However, the factor k will differ for non-ideal samples that
have a finite dimension with respect to the probe spacing. The correction factors for a
circular sample with a diameter [d], and a rectangular sample with the side parallel to the

probe line as [a] and the side perpendicular to the probe line as [d] are given in Table 2.1

[10].

 

 

 

 

 

 

 

 

 

  

 

 

 

 

 

 

 

 

 

 

——-l r—a—eil

Table 2.1 Correction factor k for the measurement of sheet resistance

0 R O R O Q
5” b7 v v s? r
l' s"' S'I' S" d

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Circle Square Rectangle
a/d = 1

d/s a/d = 2 a/d = 3 aid 2 4
1.0 0.9988 0.9994
1.25 1.2467 1.2248
1.5 1.4788 1.4893 1.4893
1.75 1.7196 1.7238 1.7238
2.0 1.9475 1.9475 1.9475
2.5 2.3532 2.3541 2.3541
3.0 2.2662 2.4575 2.7000 2.7005 2.7005
4.0 2.9289 3.1 127 3.2246 3.2248 3.2248
5.0 3.3625 3.5098 3.5749 3.5750 3.5750
7.5 3.9273 4.0095 4.0361 4.0362 4.0362
10.0 4.1716 4.2209 4.2357 4.2357 4.2357
15.0 4.3646 4.3882 4.3947 4.3947 4.3947
20.0 4.4364 4.4516 4.4553 4.4553 4.4553
40.0 4.5076 4.5120 4.5129 4.5129 4.5129

00 4.5324 4.5324 4.5325 4.5325 4.5324

 

l2

 

2,2 Composite material

The word “composite” means “consisting of two or more distinct parts” [11]. In
the case of the probes fabricated by the micropipette puller, the composite consists of
glass and metal. Thus a material having two or more distinct constituent materials or
phases may be considered a composite material. Composite material has significantly
di f ferent physical properties than do the constituent materials.

Composites consist of one or more discontinuous phases embedded in a
continuous phase. The discontinuous phase is usually harder and stronger than the
continuous phase and is called the reinforcement or reinforcing material, whereas the
continuous phase is termed the matrix. In the probes described here, the glass tube acts as
the matrix and is more plastic than the metal wire, which acts as the reinforcement.
Composite materials can Show improve mechanical properties such as strength, Stiffness,
toughness, and high-temperature performance. Here it is the glass metal composite
material that allows the probes to be pulled down to small dimensions (a task that is very
diffi cult using a metal wire alone).

After the micropipette puller drew the glass tube down around the wire, it is
assurtned that a perfect bonding exists between the fibers and the matrix so that no
Slippage can occur at the interface. Thus the probe is stretched at the same rate during the
secOl'ld pull as shown in Figure 2.3. Therefore, the strains experienced by the fiber,

matri x, and composite are equal:

 

Al Al

.9, =3], e, =——’-, em = m (2.7)
L L L
8:86 =£f =8", (2.8)

13

 

Pull
H: —}

i: L taunt"

U(W W“)

I I l
l 7 1
Glass *IAW"

+ (matrix)

Vifire ’1 Altl"
(fiber)

 

 

 

 

 

 

Figure 2.3 Strain

Also the force applied to the composite material can be separated into the forces applied

to the fiber and to the matrix (Figure 2.4).

F6 = Fm +F, (2.9)

[Cc FC
{— H —)

 

 

 

 

 

 

 

deoorrpose Fc into
- Fm acted on glass
- Ft acted on wire
4— —>
I 1
Glass
Ft Ft
4— —>
Wire
Fc = Fm + Ff

 

 

 

Figure 2.4 Total force including force acting on the matrix and on the
fiber.

14

The elasticity fort

where o is the str
composite elastici

Composite elastic

 

Matrix elasticity i

Fiber elasticity is:

.\ow the total for:

[5ng (214) in ("

y .

The elasticity formula is:
E = 9- (2.10)
8

where 0’ is the Stress, which is simply the force divided by the area. This is applied to the

composite elasticity, the matrix elasticity, and the fiber elasticity.

C omposite elasticity is:

 

 

 

 

15,-_-‘7C=FC/Ac (2.11)
E 8
Matrix elasticity is:
E... :E1 = Fm IA",
8 8 (2-12)
:> Fm = EmAme
Fiber elasticity is:
E _—_J_f = Ff lAf
f e 8 (213)
:> F, = E, Afe

N ow the total force acting on matrix and fiber is:

F, = Fm +F, =(EmAm +E,A,)e (2.14)

Using (2.14) in (2.11) gives,

(Em/1m + EfAf )e/A, (Em/1m + 5,21,)

6 A

C

C

 

(2.15)
NOW the areas of the composite, the matrix, and the fiber are (Figure 2.5):

_7[ 2 _fl' 2 2 _]Z' 2
21,-de , AFR—(avg) , 24,-de (2.16)

15

 

Figure 2.

This equation Ca

desired YOUng‘S

The SttbmiCrOme

one direction_ T

glass (matrix)
Side View

 

F

wire

Cross sectional View

 

 

Figure 2.5 Cross sectional area of the composite, the matrix, and the
fiber.

Combining equation (2.15) and (2.16), yields the composite elasticity as:

_ Em(d3, —d})+E,d}

E d, (2.17)

C

This equation can be used to select a fiber dimension that will match the glass to yield a

desired Young’s modulus at the tip of the probe.

2-3 Force calculation on the circular cross-sectional cantilever beam

The submicron—electrode is a single-ended cantilever, which is designed to bend in only

one direction. The probe is mounted on the xyz translation stage and cantilevered at an

0 V)
angle of 30 -

. 3 1
contact betwetn

the cantilever bea

“(9

Figure 2

The dfiflection (

Similarly the 5

Mamie p.

M is [he mon.

dEIleCtion‘ the

angle of 30°. When the xyz translation stage controls the vertical movement to make
contact between the probe and the sample, a perpendicular load is applied at the end of

the cantilever beam (Figure 2.6)

Load (F)

/
/
/i

 

6)
z

 

 

 

/

 

L 0

Figure 2.6 Cantilever beam and its axis orientation

The deflection displacement y is and its first derivative with respect to x equals the slope

0.

dy
—=e 2.18
dx ( )

Similarly the second derivative of deflection is equal to the reciprocal of radius of

curvature p.

2
d 2y=i=£ (2.19)
dx p EI

 

M is the moment, E is the elasticity, and I is the inertia. In order to calculate the
deflection, the moment of the cantilever is required.

M = Fx (2.20)

17

Hefee F is the l

with the bound‘

There 18 Zero d‘

[sing the boun

B.) plussin2 1*“

Maximum dCfll

0‘9

Here, F is the load to the cantilever beam and x is the location of the moment measured
(Figure 2.6). Using (2.20) in (2.19) gives,

new

— — 2.21
dx2 EI ( )

with the boundary condition:

y(L) = o (2.22)
d
EU”: 0 (2.23)

There is zero deflection (2.22) and negligible angle of deflection (2.23) at the anchor.

Taking the first and second integral of differential equation (2.21), the solutions are

 

2
j—y = 5 121 + C, (2.22)
x
Fx3
y 3327+CIX+C2 (2.23)

Using the boundary conditions C 1 and C 2 can be obtained as:

 

 

FL2 3
1 — — , 2 = FL (2.24)
2E1 3E1
By plugging these coefficients into equation (2.23):
F , FL2 Fl.3
W) = — + (2.25)

 

x ——x

6E1 2E1 3E]

Maximum deflection occurs at x = 0. Thus, the maximum deflection of a cantilever is
given by [12]

FL3

=— 2.18
3E1 ( )

18

where y iS the ‘
the cantilever b

the beam.

In order
of the moment
is a moment of

oriented along

2.6.)

where y is the cantilever vertical displacement, F is the force applied, L is the length of
the cantilever beam, E is the elasticity of the cantilever, and I is the moment of inertia of

the beam.

In order to determine the moment of inertia I for a circular cross section, the axis
of the moment of inertia is required. Since the load is applied along the y direction, there
is a moment of inertia about the Xc axis (Figure 2.7). (Note: In Figure 2.6, the x axis is
oriented along the beam length but, here in Figure 2.7, the x axis is the z axis of Figure

2.6.)

 

 

 

 

 

Figure 2.7 the moment of inertia of circular beam

The moment of inertia is the product of a differential area (111 and the square of its
distance from an axis [13]. A more precise name is second moment, because, in contrast
to the first moment the second power of distance is involved instead of the first power.
Thus, in order to determine the moment of inertia of a circular beam, the cross sectional
area of its beam must be determined (Figure 2.7). Since the cross sectional area is

circular instead of rectangular, the expression /dA is called the polar moment of inertia

l9

of the differentia

 

inertia l; with re

element. the pola

First select an m-

origin. the polar ,

 

It is “idem that

inertia “a.

From this f‘dCt ll
the polaI mom

Cr

leEC
UV as J' 1
. _\' -(1‘

of the differential area with respect to origin. It might have been called the moment of

inertia IZ with respect to the z axis. For the entire area of which dA is a representative

element, the polar moment Jo = Jrsz, where r2 = r2 + yz. Hence
JO=Ir2dA=J(x2+y2)iA=Ixsz+Iy2dA=Iy+lx (2.19)

First select an area (M = rd6Ur. Since the differential element is at distance r from the

origin, the polar moment dJo = psz. For the circular area

21tr

Jo = [61.10 = Jrsz= Jr3dr d9
0 0
Zn 4 r 4 4 (2.20)
= [[L] d9=r_[9]3"=£
O 4 o 4 2

It is evident that the moment of inertia with respect to any diameter equals the moment of

inertia with respect to any other diameter. Thus, I,r = Iy and:
J0 = Ix + I, = 21x (2.21)

From this fact, the moment of inertia I, with respect to any diameter is equal to one-half

4

 

. . rtr . . . .
the polar moment of Inertia, or I x = . The above technique IS eaSIer than finding IJr

directly as Iysz. Thus, the moment of inertia (I) for a circular beam is given by

1:

 

(2.22)

Thus the maximum deflection of circular cantilever beam is:

20

B)’ trigonome

Combining (g;

In

FL3

3E1tr4

 

(2.23)

 

 

\\\\\\

 

 

 

Figure 2.8 Slope and deflection for cantilever beam

By trigonometry,
y = (tan 6)L (2.24)
Combining (2.23) and (2.24)

4FL2

tan6= 4
3Eflr

 

(2.25)

In this way the deflection angle formula is derived, and the formula will be used to verify

the probe mounting angle in chapter 6, Data and Analysis.

21

Noun
applications (
the pulled mi
inside a glas

following sur

3.1 Micropi]
Varioi
micr0pipcue

criteria neces:

3.I.I Micmp

The r.
StUdies of m;
{Emperalufe C

3. MICROPIPETTE APPLICATIONS

Nowadays, many researchers apply micropipettes in various field, and
applications of this can be found in numerous articles in scientific literature. Basically,
the pulled micropipette can be sorted into two categories: an electrode with a metal wire
inside a glass pipette or Simply a glass pipette pulled without any filament. In the

following subsection, the kinds of micropipettes application will be discussed.

3.1 Micropipettes with metal filaments

Various metals and types of glass can be used to make an electrode probe by the
micropipette puller technique. Each material is carefully chosen in order to meet the

criteria necessary for a specific research application.

3.1.] Micropipette-based submicrometer thermocouplefl 4]

The measurements of spatially localized temperature changes are required in
studies of many physical and biological processes and objects. To measure the rapid
temperature change at the cellular and subcellular level, a thermocouple with submicron
contact Size and a response time of a few microseconds is required. G. Fish [14], from
the Hebrew University of Jerusalem, demonstrated a fabrication process of micropipette-
based submicrometer thermocouple. The submicron thermocouple consists of a
micropipette containing a platinum core, and the unit is coated with gold to make a gold-

platinum point thermocouple at the tip of the micropipette.

22

Figure .‘

In order
the size of the
platinum Core 1
laser Pipette p

Plated inside

platinum glass

 

 

 

 

 

\ /
/
add.
coating
11"
hi ///
1 .111...

 

 

Figure 3.1 Schematic presentation of the sensing area of the pipette
thermocouple.

In order to obtain optimal parameters of a thermocouple, it is necessary to reduce
the size of the sensor d, the thickness of the gold coating hw, and the cone angle of the
platinum core ,6 (Figure 3.1). The reproducibility of the technique using a Sutter P-2000
laser pipette puller is approximately 90%. A platinum wire, 80 mm in diameter, was
placed inside a borosilicate tube with outer and inner diameters of 1.2 and 0.3 mm,
respectively. The pipette wire assembly was placed in a P-2000 pipette puller in which
five parameters could be Optimized: temperature of heating, the length of the segment
heated, delay time between turning the heat on and the beginning of pulling, the velocity
of the pull, and the strength of the pulling. To find the optimal combination of these
parameters, they treat their sample, consisting of a glass tube containing a platinum wire,
as a composite material (CM). In this material, the heated glass tube acts as a matrix

which is more plastic than the platinum wire which acts as the frame. During the pulling

23

Procedure: the
uniform [hinni
smaller dimem
First. I
platinum wire
further pulling
Subseq
in the formatit
shape of the
parameters at t
D, over the ex
and the platinu
when the glass
The sec
The pipette tip
like coatings w
the required thi
time of depositi
0f the microthei

abmk 801d~p1at

procedure, the surrounding glass acts to stabilize the metal and thus results in a more
uniform thinning of the metal core. Therefore, the metal core can be pulled down to a
smaller dimension before breaking occurs.

First, the glass tubes were pulled to an inner diameter that was equal to the
platinum wire diameter. At this point, the sample was heated for several seconds without
further pulling to ensure a firm connection between the glass and the platinum.

Subsequently, the pulling was performed Slowly, in four stages, and this resulted
in the formation of two glass pipettes that were filled with platinum to the end. The
shape of the cone and the outer diameter was determined by the pulling program
parameters at both steps. At 0.2<A<O.4, where A is the ratio of platinum wire diameter
Dw over the external diameter of the glass tube Dgl, it was possible to pull the glass tube
and the platinum core to dimensions as small as 50 nm. The Pt electrode is clearly seen
when the glass at the tip is etched back.

The second therrnoelectrode was made as a vacuum-evaporated thin gold film.
The pipette tip was facing the gold source. This geometry, succeeded in producing edge-
like coatings with a maximal thickness of the gold layer near the tip. The coatings with
the required thickness and electrical conductance were Obtained by varying the rate and
time of deposition. This improved the spatial resolution of thermocouple. The sensitivity
of the rnicrothermocouple produced was about 7 uV/°C, which is slightly less than that of

a bulk gold-platinum thermocouple.

24

3.1.2 Candler?
With the
their modified it
glass probe was
measurements.
with an outer di.

inserted into eacl"

 

 

 

a Sutter P-2000 p
produce a sensitii
the two thin Pt w
the wires is 200-

fused by heating

Figure 3"

3.1.2 Cantilevered glass multiple-wire force-sensing thermal probes.[15]

With the further study on thermocouple, Dr. Dekhter[14] published the article on
their modified thermocouple with force-sensor. A double platinum wire cantilevered
glass probe was produced for scanned probe microthermal resistivity, and topographic
measurements. For such a double-wire probe fabrication, borosilicate theta capillaries
with an outer diameter of 1 mm were used. Platinum wires, 25 11m in diameter, were
inserted into each of the two channels of the capillary, and the combination was pulled in
a Sutter P-2000 pipette puller to a small tip with a diameter of about 100 nm. In order to
produce a sensitive element, the glass on the tip was removed by etching in HF, so that
the two thin Pt wires were exposed to a length of about 3-10 pm. The distance between
the wires is 200-1000nm. To create a point of resistance, the two protruding wires were

fused by heating the tip (Figure 3.2).

Glass theta

 

thirefl
V
\Etchedtip

 

/

Fusedwires

 

Figure 3.2 Overview of the fabrication of a double-wire thermal resistive
probe.

The resis
the protruding w
operation. the st,
produce a high It
coated with gold
placement in a N

this probe. a vari

 

 

probe resistance.
this heated the se
the test sample at
the surface. the 1
resistance of the

the control svster

3.2 Micropipeu

The glass
"aileron of mat:
Waveguides Or
a13Plied to Prodt

Scan ‘ .
mug Optical

3.2.1 771 e tip of

Tearrficlt

”lg prObe I

The resistance of the fused wires, 20-60 9, depends on the length and thickness of
the protruding wires. The diameter of the fused wires is 100-200 nm. Subsequent to this
operation, the straight tip is cantilevered about 150-200 um from the end. In order to
produce a high refractive surface for the feedback control the cantilever of the probe was
coated with gold. The cantilevered probe was then mounted on a special tip mount for
placement in a Nanonic NSOM/AFM 100 confocal system scanned probe microscope. In
this probe, a variation of 1°C in temperature corresponded to a variation of 0.1 Q in the
probe resistance. A constant current of about 4 mA was passed through the probe and
this heated the sensitive resistive wire at the tip. The probe was brought into contact with
the test sample and the heat flow passed into the sample surface. During scanning across
the surface, the probe cooled faster in a region with higher thermal conductivity. The
resistance of the probe was changed by this cooling and these variation were detected by

the control system.

3.2 Micropipettes without metal filament

The glass micropipettes, pulled without any metal, can be used for controlled
transport of materials such as liquid or gas. Also, they have been used as laser/optical
wave-guides or as force sensors [1]. In addition, glass-pulled micropipettes are also
applied to produce probe tips made from a Single mode fiber for use in near-field

scanning optical microscopy (NSOM).

3.2.1 The tip of Near-Field Scanning Optical Microscopy

Near-field Scanning Optical Microscopy, also known as NSOM or SNOM, is a

scanning probe microscopy technique that allows Optical imaging with spatial resolution

26

beyond the diffr

sub-wavelength .
from the apertur
light that is em)
through space ht
light or forbiddc
than far-field (FF

exist at the samr

 

the wavelength-

apeiture size a <
position can be

aperture.

There an
method. the san‘
surface (being i
geometry, w'hic
iFigure 3.3). T1

beyond the diffraction limit. In this approach to Optics [16], light is transmitted through a
sub-wavelength aperture and the sample is placed within the near field at the distance
from the aperture that is few times closer than the dimension of the wavelength of the
light that is employed. Near-field (NF) is defined as the light that does not propagate
through space but is localized on the surface of objects [17]. It is also called evanescent
light or forbidden light. NF light contains more information, higher Spatial frequencies,
than far-field (FF) light which is the propagating normal light. Higher spatial frequencies
exist at the sample surface; however, it decays exponentially within a distance less than
the wavelength. By placing the sample within the NF of the aperture, the condition of
aperture Size a < A is satisfied, and a record of the optical interaction as a function of
position can be generated. In other words, the sample will be resolved at the Size of
aperture.

There are various approaches being used to achieve the near-field light. In one
method, the sample is illuminated with FF light. It can be completed either from the top
surface (being imaged) or by the back illumination ATR (attenuated total reflection)
geometry, which produces a near-field on the top surface of the transparent sample
(Figure 3.3). The dark arrows represent the light path to sample (FF light), and the lighter

arrows show the light from scattered near-field to detector.

27

Figure
Introducing a _~
and so can be
another metho
NF. A fiber 0
Placed close 1(
less than On e
Again this inf

the detector (F

 

 

 

 

Figure 3.3 ATR illumination from the back of the transparent sample.

Introducing a sharp probe into the NF of the object scatters far-field (FF) that propagates
and so can be detected by normal means such as a lens, CCD detector (Figure 3.4). In
another method, a probe with a very small aperture is used to illuminate the object in the
NF. A fiber optic terminates a FF light in a sub-wavelength aperture, and the probe tip is
placed close to the object surface. The distance between probe tip and the sample is much
less than one wavelength of FF light, so that the NF interacts with the sample surface.
Again this information is converted to FF scattered light and thus it can be collected by

the detector (Figure 3.4).

28

Fart

I
E

Figure

Recent
below 50 nm
l'niversity pro
probe that COI

The CQnVenu-O
of glass [apem
technique pr0(
appliCanOnS Of
the SUrfaCe “'11
small cone ang

height. In tip/

hard, but [hev 2

Apertureless Aperture

Far field

 
  
 

Scattered

l Far field: .. far-field

l .43}.

1

 

 

 

 

Figure 3.4 Various methods of N SOM image capturing technique.

Recently, a new form of NSOM has been developed with the optical resolutions
below 50 nm [18]. The device modified by the Dr. Lewis’s research group in Hebrew
University provide an integrated microscopic solution based on a near-field optical fiber
probe that combines the possibility of near-field optics with atomic force microscopy.
The conventional optical element is based on the introduction, by Hartoonian et a1. [19],
of glass tapering combined with metallic coating technology to near-field optics. The
technique produced straight near-field optical elements, which are used in most of the
applications of NSOM today. Nonetheless, the straight probes have limitation. To image
the surface with a considerable degree of roughness, it is desirable for the tip to have a
small cone angle over a length near its end which is larger than the surface corrugation
height. In tip/cantilever combinations diamond shards have been used as tips that are

hard, but they are not always sharp enough. In addition, such tips add considerable mass

29

to the spring ‘
combinations n'
or prohibitiveh

3mm
from glass m
forces not on
above. but tb
method of d
NSOM can
adaptable t-
force sensi
Sharp bent"
Thus. whii
NSOM_ 1
excitation
”MSmjni.

This [echr

to the spring, thus lowering the resonance frequency. Microfabricated tip/cantilever
combinations meet the second requirement, but the integrated tips are either short and fat,
or prohibitively expensive.

Shalon et al.[20] presents a simple technique to reliably produce force probes
from glass micropoipettes. Their modification of these micropipettes to sense surface
forces not only provides a reliable and inexpensive solution to the difficulties mentioned
above, but this modification is also of considerable importance for introducing a universal
method of distance feedback that is essential in NSOM. The glass micropipettes used for
NSOM can be bent near the tip to produce a structure that appeared to be readily
adaptable to the most generally applicable and simplest methods of normal and lateral
force sensing. The main drawback of such bent micropipettes for NSOM was that the
sharp bend would prevent the light from reaching the subwavelength aperture at the tip.
Thus, while the probe appeared interesting in its own right, it did not appear practical for
NSOM. This problem has been resolved by the development of a new method of external
excitation that produces a spot of light within the aperture at the tip, rather than
transmitting a beam from a remote source through the entire length of the pipette [21].

This technique and the proposed combination are shown schematically in Figure 3.5.

30

Figur

A dye embec
directed head
method‘ b€5i(
the Scanning

Cantilevered i

Oflhe Cantilet

glass capillaries by a micropipette puller.

Position
Sensitive
Detector

   
   

Metallized Pipette

Ion Sensing Dye

Epi-
Illumination

E
a
E
2.
o

Photomultiplier rL‘

Figure 3.5 Dye-embedded sol gel ion-sensing cantilever micropipette

[22].

A dye embedded plastic plug is formed inside the pipette tip and a laser beam is then
directed head on to this tip generating a very intense fluorescence from the dye. With this
method. besides receiving a very intense source of light, the excitation is uncoupled from
the scanning probe and is not affected by the geometry of the pipette. This allows for a
cantilevered pipette to be used for NSOM while simultaneously measuring the deflection
of the cantilever due to normal and lateral surface forces with the standard methodologies

used in force microscopy as depicted. The tips are fashioned from aluminum—silicate

original length of 1 cm, the falling lower chuck activates a microswitch, which turns off

31

After stretching by about 10 cm over an

the heating. 5
lengths of ml“
the force senSC
the tip/surface

bunsen flame

\‘lEWCd with a
pipette apenurt
using the reflec
cantilever. Sm;
glass cover sli]
cantilever. Wit

has achieved 0;

3.2.2 Scannim

Convent
including micrc
mUch Smaller
resolution beyo'
ScaHHlng nano“

PhotoresiSt in

the heating. Simultaneously, the capillary separates at its center, leaving two long, thin
lengths of micropipettes, with a small hole at each end. These are the tips used to create
the force sensors. To form the bend, which enables sensing of the force directed along
the tip/surface axis, the micropipette tips are passed briefly over the hot air from a small
bunsen flame which causes the end of the pipette to bend upwards. This process is
viewed with a microscope, and it is challenging to get a good bend without closing the
pipette aperture. Since the cantilevers were initially tested in a scanning force microscope
using the reflection technique [23] for detection it was necessary to attach a mirror to the
cantilever. Small pieces of mirror (50 um X 100 mm x 100 um) were formed by coating a
glass cover slip with aluminum and breaking it. These pieces were then glued to the
cantilever. With a further study on NSOM, by the group from Dr. Lewis, this technology

has achieved optical resolution below 50 nm, with the tip aperture as small as 50 nm.

3.2.2 Scanning nanolithography using a material-filled nanopipette

Conventional photolithography has been widely used in many applications
including microelectronics and nanotechnology. However, due to the potential need for
much smaller integrated electronics devices, nanolithographic methods for higher
resolution beyond the present limit are studied. Mun-Heon Hong et.al. [24] presented a
scanning nanolithography technique using a pulled micropipette which delivers liquid
photoresist in the vicinity of the substrate through a nanometric aperture formed at the
end of the pulled pipette [24]. In their scheme, the photoresist is deposited at the
predetermined position on the substrate by ejecting it through the small aperture of the

nanopipette. This approach has the following several advantages: (a) no spin coating of

32

photoresist required, (b) no need of a mask, (c) no alignment of the mask and the
substrate, (d) no need of UV light source, (e) no development process, (f) environment-
friendly instrumentation in ambient conditions, and (g) simple and inexpensive.

The micropipette used in the experiment, pulled by a commercial pipette puller
(P-2000, Sutter Instrum.Co.), is made of quartz with its inner (outer) diameter of 0.7 mm
(1.0 mm). The pulled micropipettes can have hollow diameter ranging from a few tens of
nanometer to a few hundred nanometers by adjusting the pulling parameters. A filament
made of metal thread is installed inside the pipette so that its pulled tapered region can be
mechanically strong. The filament is also designed to help the liquid photoresist wet the
inner wall of the pipette so that the liquid can fill spontaneously inside the pipette. When
the liquid, pushed by a syringe, is filled in a pipette without the filament, some trapped air
remains inside. As the photoresist is pushed out, the effective mass of the pipette is
changed, which results in a change of the resonance frequency of the pipette. Thus it
becomes difficult to maintain the resonance frequency constant for shear-force feedback
control. The xyz translation stage is controlled by shear-force detection. A shear-force
detection scheme was developed by using two optical fibers placed in the vicinity of the
pipette end (Figure 3.6): single-mode fibers for the focused light source as the incident
laser, and two adjacent multi-mode fibers for differential optical detection of the

diffracted light at the photodiode.

33

 

 
 
 
  
    
 

pipette pressure
Duncan)“
m —>I I

 

 

 

 

 

 

 
  
 

resist
[Substrate ———-—i
Laser (k633nm) Computer -
_ « Sin ;_le mode, (Digital Feedback)
uln- .- . . 7

 

 

 

x,y,z positioning
system

Figure 3.6 Schematic diagram of the photoresist-filled nanopipette
controlled by shear force position detection [24].

The concept of shear force distance control is described further by AD Muller
[25]. Shear force distance control between a probe tip and a sample works in the
following way. The whole tip is laterally excited by a dither piezo close to its resonance
frequency. It behaves like an oscillator with small damping. While the tip approaches
the sample surface, the shear force increases the damping. It follows a decreased
resonance frequency and a decreased vibration amplitude. Both result in a decrease of the
input signal for the feedback. The distance feedback maintains the percentage of
amplitude decrease relative to the free oscillation constant. Alternatively, the phase shift
between exciting and detected alternating signals can be utilized as an input signal for the
feedback. The resonance shift and the amplitude decrease due to the weight and the
damping of an evaporating drop of water over time. Their experimental configuration is
based on the commercial optical detection system in the AuroraTM microscope head that
was modified to carry micropipettes instead of optical fibers. The system provides an

adjustable laser that can be directed towards the final end of the tip (apex) with an

34

incident angle of about 30°. During operation, the tip scatters the laser beam that is

reflected on the sample surface, like shown in Figure 3.7.

laser ‘ til3

detector

    

drop -

sample

Figure 3.7 Principle of the utilized optical shear force detection applied to
the micropipette tip with a droplet near the apex [25].

In this way, the vibrating tip modulates the intensity of the reflected light with its
oscillation frequency fc. The detection system consists of a four quadrant photodiode. As
tips, pulled micropipettes are mounted in the Aurora head and fasten by a screw. The
pipettes were filled with distilled water that can be pumped and sucked in defined
amounts. For the following experiments, there are no further requirements concerning
the tip shape except that the liquid throughput has to be high enough to create a drop of
water at its end. Typical inner apex diameters for this are of the order of 520 um. A
water droplet is formed by pumping the liquid. Due to the sharp tip shape and the surface
tension of water on glass, this droplet becomes stable some 10 um above the apex.
Simultaneously, the laser beam points to a position above the water droplet so that the
detected signal is not influenced by the drop of water. The sample has to be close enough
to be able to reflect the laser beam. But sufficient distance from the apex is necessary for
the existing water not to touch the sample surface. When the liquid pump is switched off,

the drOp size decreases continuously because the laser heats the micropipette and the

35

water droplet evaporates. This evaporation is so fast (some seconds for a 10 um drop)
that the frequency spectrum, which takes about 1 min to obtained with a lock-in

amplifier, is difficult to correlate with a specific drop diameter. The resonance to, is given

by a), = (a): —I‘2 /2)”2, where (a, is the resonance of the nondamped system (here the

system without the drop). Therefore, the increase of the drop diameter is equal to an
increase of mass m and reflectivity F of the drop as well as a decrease of resonance
frequency a), and (4,. In other words, resonance frequency is lower in the filled pipette,
and intensity of the signal is larger in filled pipette due to the focusing effect on the

scattered light on the drop (Figure 3.8).

 

 

 

 

 

0.05
l
0.04 r
g .
S 0.03 .
d)
u b
E 0.02» 2
GI. O—n—vu—u—I—
E r 1
< 001 >
0.02 '
0 ._u. l - LAJ A A a ILAA

 

65 70 75 80 85 90 95100
Frequency (kHz)

J!
o
a .
05
c

Figure 3.8 Resonant frequency shift of the liquid-filled nanopipette
(curve 2) with respect to that of the empty pipette (curve 1)
[24].

Now, with the fundamental knowledge of shear-force control discussed above, the
article written on nanolithography can be revisited. Even when the pipette-substrate
distance is maintained constant, as the liquid is pushed out of the aperture (Figure 3.9 a),
the liquid-substrate separation becomes small and consequently the shear-force signal is

also decreased. The shear-force signal vanishes as in the case of the empty pipette. As

36

mentioned in the previous paragraph, intensity is larger in filled pipette because of the
focusing effect on the scattering light on the drop. When the pipette retracts after the
liquid is dropped, the signal returns to its original value (Figure 3.9 a). When the liquid
flows continuously inside the pipette toward the aperture opening without discontinuity in
mass distribution and if no liquid drop is formed at the end, the resonance frequency of

the pipette does not change (Figure 3.9 b).

   

 

 

1 . i 3:3
H AA
Substrate
(a)

unmet-4
fix ... ~—
I.
"‘ ,..

1" - ' "5:: tit} #
3e" /' resist
V 3 P
J

 

l
(b) Substrate

Figure 3.9 Two fabrication schemes: (a) point drawing and (b) line
drawing [24].

However, if there is some trapped air moving inside, there exists a non-uniform
mass distribution which results in the change of the resonance frequency depending on
the position of the trapped air. This phenomenon can be used to monitor the smooth
filling of the liquid in the entire pipette, by checking if the frequency changes while the
liquid is pushed.

Figure 3.10 (a) shows the shear-force topography image of the three resist drops
fabricated on the gold-sputtered substrate. The scanning image is obtained by using the

same nanopipette as an NSOM probe. The separation between the dots is about 1 um and

37

the height of each drop is about 80 nm. The full width at half maximum linewidth of the
dot is about 300 mm, which is just the inner diameter of the nanopipette. Therefore, the
rather large NSOM probe limits the resolution of the topographic image,[26] although the
fabricated patterns can be much smaller depending on the drawing parameters such as

liquid density, pressure, and approach/retract time.

 

 

 

Height (nm)

 

0 0.4 0.8 1.2 1.6 2
Position (um)

 

(a)

Figure 3.10 Three photoresist dots of 300 nm size fabricated on the gold-
sputtered glass substrate. (a) Topographic image of the dots,
and (b) height profile across the line in (a) [24].

38

4. FABRICATION AND APPLICATION OF MEMS

MICRO CANTILEVER

In the conventional probe fabrication method, cleanroom facility was required for
the probe fabrication process. The fabrication of a MEMS micro cantilever, requires
many individual steps/processes as well a use of hazardous chemicals to etch the
components. Although the micropipette puller was introduced to fabricate compatible or
even better probes, it is still useful to review the fabrication procedure of MEMS micro

cantilever and its application.

4.1 MEMS microfour-point probe

Conventional four-point probes are typically comprised of large, macroscopic tips
of millimeter-scale spacing [27] or stationary, planar electrodes deposited on the sample
using lithographic techniques [28]. The large contact forces exerted by macroscopic tips
easily result in damage to sensitive surfaces. To avoid this problem, researchers from Dr.
Boggild’s lab have designed a new scanning micro four-point probe, which can be used
non-destructively anywhere on a surface with a high spatial resolution [29]. The probe
consists of four silicon oxide cantilevers on top of a flat silicon support, coated with a
conducting electrode layer (25 nm Til 100 nm Au). The cantilevers are 1.5 pm thin, 7 um
wide, and extend 60 um over the edge of the support. Due to their thinness, the
cantilevers are flexible in the vertical direction and exert a contact force on the order of
just 10'7 N for an electrode deflection of 1 pm. The contact area is of the order 100 nm x

100 nm per electrode. Apart from reducing the surface damage, the advantage of using

39

soft, flexible microcantilevers is that the cantilevers adapt more easily to height
differences on the surface or misalignment of the probe. By measuring repeatedly over
the same area and obtaining a constant result, they verify that the probe does not alter the
electrical properties of the surface.

The probe is wire-bonded to gold film pads on the ceramic chip carrier, which is
then mounted directly in a compact current source and preamplifier unit. A video
microscope is located above the sample to monitor sample and probe during
measurement. The probe is tilted 30° with respect to the sample, and aligned to the focal
point of the microscope. The sample is mounted on a xyz-stage, which brings it into

contact with the electrodes (Figure 4.1 (a)).

   

Figure 4.1 (a) Schematic of the experimental set up.

(b) A scanning electron micrograph of the probe [29].

The resistance is determined from a linear fit to the V-I curve, where the bias current is
being swept from —100 nA to 100 nA. The maximum bias voltage is limited to 0.5 V in

order to avoid electrochemical reactions at the electrodes. To ensure consistent

40

mechanical conditions, they employ an optical force-feedback technique similar to that in
atomic force microscopy. A laser beam is directed towards the cantilever, so that the
reflection on the electrode can be detected in the microscope. When the cantilever
touches the surface, the four reflection spots are dimmed due to the deflection of the
cantilevers. By measuring the intensity of the reflection spots, the height differences of

the surface are detected simultaneous with the electrical measurement.

4.1.1 Fabrication of micro four point conductivity probes

The micro-four-point probe is fabricated by conventional silicon processing
techniques [30]. The process sequence is summarized in Figure 4.2. First a 1.0 um thick
layer of SiOz is grown by wet oxidation on a 340 ttm thick (100) silicon wafer (Figure 4.2
(a)). Then a 0.8 pm thick layer of photoresist (Hoechst AZ 5214B) is spun on top of the
SiOz layer, and is exposed to UV light through a chromium mask defining the electrode
pattern on the wafer front surface. The resist is then developed using the negative
development process for AZ5214E. A thin chromium layer (750 A) is deposited and
lifted off, serving as a mask for the subsequent reactive ion etching (RIE) process in
CHF3 plasma (Figure 4.2 (b)). The oxide is RIB-etched to form four microcantilevers.
After the chromium is removed a thin (1000 A) layer of silicon nitride is then deposited
on both sides of the wafer as a mask for silicon etching (Figure 4.2 (c)). A photoresist
layer is deposited on the wafer backside and is patterned to define the probe support
chips. The substrate is etched in a KOH solution at 80 °C until the wafer has been

completely etched through from the rear side (Figure 4.2 (d)). At this point, the silicon

41

oxide microcantilevers are suspended over the edge of the support chip, but still covered
with a thin nitride membrane. The silicon nitride membrane is removed by RIE in a
mixture of SF6 and 02. A low-stress nitride reduces the risk of damaging the
microcantilevers during the etching of the suspended membrane. Using an isotropic RIB
silicon etching process with SF6, the electrodes are undercut (Figure 4.2(e)). Because of
the undercutting of the oxide structures, the metallization is maskless. The undercut
structure allows silicon support chips to be coated individually or on a full wafer basis.
For measurements in ambient condition a typical layer sequence of 250 A Ti / 1000 A Au
is chosen, due to the excellent contact properties and low stress, resulting in a negligible

bending of the electrodes.

 

 

Figure 4.2 The major steps in the microfabrication process.

42

4.2 Scanning nano four-probes or nano tweezer

In this manner, four soft and flexible metallized SiOz microcantilevers, with
electrode spacings down to 1.5 pm, were fabricated. To reduce the gap even further, an
electron-beam induced deposition, which is a constructive three-dimensional
nanolithography technique has been used [31]. By focusing the electron beam of a
scanning electron microscope on the ends of the microcantilevers, hydrocarbon molecules
present in small concentration near the beam spot are cracked. This leads to the
formation of a narrow rod of carbon residues, growing in the direction of the electron
beam. Lengths of several microns are easily obtained by continuation of the process, and
such tip have successfully been used for high aspect ratio atomic force microscopy tips
[32], nanoparticle manipulation [33], and nanolithography [34,35].

To avoid charging of the oxide cantilevers by the electron beam, and thus a poor
growth rate and tip quality, the microcantilevers are coated with a thin metallic layer (100
A Ti / 800 A Au) prior to tip deposition (where 250 A Ti / 1000 A Au was deposited for
the microcantilevers). By keeping the electrodes at the ground potential of the scanning
electron microscope (SEM), a drain for the electron beam is provided.

The nanotips are grown at an acceleration voltage of 10 kV and beam current of 3-
6 pA in a JEOL 634OF field emission microscope operating at a base pressure of 10’8
Torr. Higher beam current reduces the growth rate, possibly due to the increased
charging and thus beam deflection. A growth rate of 250 nm/min was obtained by this
technique. The beam current is comparable to the 1.9 pA found to be optimal by Wendel
et a1. [34] in terms of growth rate and resulting sharpness of the tips. To obtain fast and

reproducible results, it is crucial to carefully adjust the focus and to correct for

43

astigmatism of the electron beam. The carbon tip resulting from this deposition technique
is mechanically strong and durable, as demonstrated in earlier works. It has been
suggested that the tip material is diamond-like, without being brittle, which would
explain the exceptional durability. By tilting the microprobe with respect to the beam, it

is possible to control the shape of the nanotips as shown in Figure 4.3.

100 n'm

 

Figure 4.3 (a) Schematic of the nanotip deposited in a plane
perpendicular to the plane of the microcantilevers. (b) SEM
image of micfrofour-point probe. (c) SEM image of nanotwo
probe with 200 nm spacing. (d) Close up of 200 nm gap two-
probe. (e) SEM image of nanofour point probe with an
average spacing of 350 nm [31].

In order to make the nanotips conducting they are metallized a second time with
100 A Ti / 600 A Au. Such cantilevers are used for not only scanning nano multiprobes

but also nano tweezers. For lifting and manipulating of free-hanging or floating

structures such as nanowires it is necessary to use opposing forces to seize and hold
efficiently. Kim and Lieber [36] reported the first nanoscale tweezers in operation, where
two carbon nanotubes, attached to metal electrodes on opposing sides of a micron thick
glass needle, were used to manipulate nanoscale clusters and wires. The tweezers were
closed electrostatically using a voltage difference between the nanotubes. There are,
however, situations where a large field across the object to be manipulated or measured is
not desired, such as biological systems and organic molecules. Even if the grabbed object
is electrically insulating, the strong fields near the object can induce structural changes,
charging and other unwanted side effects. To address this problem, the nano tweezers
have been further modified to include a set of independent electrodes used for Opening
and closing the tweezers. As shown in the SEM micrograph in Figure 4.3 (a), the
electron beam was focused at the ends of the cantilevers at a tilt angle of 45°. After 10-30
s, the instrument was switched to scanning mode, to reposition the beam target and
thereby compensate for drift. Here, an important consideration during the fabrication
process is that scanning at a high magnification will inevitably deposit a thin carbon film
over the entire image field. By alternatively depositing and viewing, the beam could be
targeted exactly at the tip of the rod. The length increase gave a measure of the growth
rate, which in their case was initially 400 nm min", but decreased to 200 nm min'1 as the
tip length increased. The second tip can be deposited at an angle of 90° with respect to
the first (Figure 4.4 (b)). By a continuation of this procedure, with smaller and smaller
exposure times, the gap size can eventually be brought down to about 100 nm. To
finalize the gap, parallel secondary tip are deposited at the ends of the converging tips.

This secondary parallel tips are deposited by short (20 ~ 100 s) depositions to facilitate

45

holding of elongated objects such as nanowires. By continually monitoring the growth
rate, and estimating the deposition times, the two tips can be fine-tuned to within 10 nm.
In Figure 4.4(c), two such secondary tips are shown. This unique feature makes it
possible to shape the tweezer to provide the best possible manipulation properties for a

given application.

 

Figure 4.4 The nanofabrication process.

The nanotweezer is opened by applying a potential difference between the inner and the
outer electrodes while maintaining a potential difference between the inner electrodes
near zero. A small bias voltage can be used for measuring the grabbed object (Figure
4.5). In other words, the outer cantilevers are used as actuating electrodes, and a

moderate voltage of around 10 V can open the gap without biasing the inner tweezer arms

46

with respect to each other. Thus, this results in avoiding electrical damage to the held

object.

 

Figure 4.5 The circuit set up of nanotweezer.

47

5. MATERIALS AND METHODS

5.1 Electrode fabricated by micropipette puller

The measurement of electrical conductivity of a substance at extremely localized

areas can be realized through utilization of micron-scale probes fabricated using the

Sutter® P-2000 C02 laser micropipette puller. The probes are produced in two steps.
First, a metal wire is inserted into a glass capillary tube. Using the micropipette puller,
the glass/metal combination is then pulled down to micron or sub-micron diameters. By
manipulating the parameters on the micropipette puller, which among others include the
heat of the laser and the force of the pull, this type of “micro-probe” can be achieved.
The goal of this section is to discover a reliable process that will create these micro-
probes with high repeatability and consistency. This relatively simple technique for the
fabrication of such cantilever probes can be extended to other sensing applications such

as temperature measurements through additional coating steps.

5.1.1 Material selection

The critical element in drawing glass with a metal core is to get both the glass and
metal to draw at the same time. In general this requires glass and metal with similar
melting temperatures, and close contact between the glass and metal at least in the final
draw.

In the best case (ideally), close contact is a complete metal core with no air gap
between the metal and the glass. This insures the best heat transfer from the glass to the

metal, and it also mechanically couples the metal to the glass. If a fine wire is simply

48

placed inside a capillary with a much larger inner diameter, the metal will not be
adequately connected to the glass mechanically, nor for purposes of heat transfer. In such
a case, the glass first must be drawn down until it tightly surrounds the metal wire, so it
forms one unit of glass and metal composite material. Then, in subsequent steps, the
glass-metal composite can be drawn down together.

The glass and metal used in combination must have appropriate melting points.
Through our experiments we have found that it appears the metal core should have a
slightly lower melting point than the surrounding glass. Heat is applied at the outside
surface of the glass and must conduct through the glass to reach the metal. Since the
metal is heated by the surrounding glass, it follows that the metal must be slightly cooler
than the glass. The metal has higher thermal conductivity than glass, so that the heat
reaching the metal core will be conducted away toward the cool unheated ends of the
metal wire more rapidly than in the surrounding glass. As the result, the metal doesn’t
get as hot as the glass during the heating phase. Thus, in theory, it is ideal that the metal
has slightly lower melting temperature than the glass softening temperature. This insures
metal and glass to fuse strongly and draw the composite further without any discontinuity.
If the metal has higher melting point than glass, the metal never melts and the metal is
more likely to break without any tapering (Figure 5.1). It is possible to utilize the higher
melting point metal, if the glass can withstand temperatures in excess of its softening
temperature. On the other hand, if the metal has far less melting point than the glass, the
metal boils and forms bubbles in the glass tube (Figure 5.2). Thus the metal core would

be discontinuous.

49

 

I

Figure 5.1 The metal breaking at the tip

 

Figure 5.2 The metal boiling in the tube

Therefore, each metal melting temperature and glass working temperature is carefully
checked in order to find the matched glass and metal combination (Table 5.1 and 5.2)

[37].

Table 5.1 Metal physical properties.

 

 

 

 

 

 

 

Metal Type Elasticity (E) Melting temperature
Copper 110 GPa 1083 °C
Nickel 270 GPA 1455 °C

Platinum 171 GPa 1770 °C

Constantan 162 GPa 1300 °C

Gold 77.2 Gpa 1064 °C

 

 

 

50

 

Table 5.2 Glass physical properties.

T
Borosilicate 64 GPa

70
Aluminosilicate —

 

Comparing the two tables above, it can be seen that the closest match between these
metals and glasses is platinum and quartz. Although the melting temperature of platinum
is a bit higher than that of quartz, this is still the best match because a full melt is not
required to pull the metal; a partial melt will suffice, in addition the quartz can exceed
1730 °C to achieve a “softer” condition while pulling.

In Figure 5.3, the quart glass (o.d.=1.0mm, i.d.=0.3mm) and a platinum wire

(diameter=80 um) were selected as the probe materials.

 

Figure 5.3 The pulled micropipette with platinum filament inside

After the final pull, the tip tapered and the wire decreased in diameter to 20 um. This
result would suggest that it is best for the metal and glass to have similar melting point

temperatures. It has been shown [38], however, that materials selected from largely

51

different melting point temperatures can also be used to draw the composite to make a
probe. As written in the article, a platinum wire with 80 um diameter and a borosilicate
tube with outer and inner diameters of 1.2 and 0.3 mm, respectively, are used as probe
materials. This would seem to run contrary to the argument stated mentioned above,
since platinum has much higher melting point (1770 °C) than borosilicate glass softening
point (800 °C), however, the properties of the composite for beyond the melting point for
the glass were not discussed. This selection shows the melting point temperatures of
metal and glass are not the only property for selecting materials. In order to stretch the
wire and glass together, mechanical properties of the wire must also be considered. If the
wire is soft, it stretches easily even without reaching full melt. Gold, with a hardness
value of 25 Vickers, is a very soft metal. Platinum is next with a value of 40 Vickers.
Silver and aluminum are also very soft metals but were not utilized with in this study.
Thus in the case of the platinum wire in the borosilicate glass, it is considered that even
though the melting point of platinum is higher than the borosilicate glass softening point,
platinum is soft enough to get pulled down in the borosilicate glass tube without reaching
full melt. It is also possible that the temperature exceeds the softening point of the
borosilicate glass, but does not reach the boiling point of the glass.

The compliance of gold is vital in pulling the wire down in the pipette and
achieving continuity to the very tip. Thus gold wire (d=50um) and borosilicate glass tube
(i.d.=0.5mm, o.d.=lmm) were selected as the electrode probe for electrical conductivity
measurements.

Selecting the dimensions of the glass tube and wire is also important for

successful pulling. In addition, five parameters are programmed and optimized for each

52

pair of tube/metal materials and dimensions by iterative testing. In the next section, those

variables are discussed.

5.1.2 Laser Specification and Variables

The P-2000 contains a 20W Class IV C02 laser with a 3.5 mm diameter beam
(spot size = 1.75 mm), and a 4 milliradian divergence.

The optimized structures of micropipettes with wires inside are determined by the
parameter values that are programmed on the P-2000 by the user. There are 100 separate
programs that can be saved for future use. Each of those programs consists of one or
more cycles. A cycle consists of five programmable parameters; heat, filament, velocity,
delay and pull. Those five parameters have some specific range of number to program in
respectively (Table 5.3) [39].

Table 5.3 Five programmable parameters on P-2000

 

 

 

 

 

 

 

Parameters Range
Heat 0 to 999
Filament 0 to 15
Velocity 0 to 255
Pull 0 to 255
Delay 0 to 255

 

 

 

The first parameter is heat which determines the power of the carbon dioxide laser
that reaches the micropipette. The second parameter is the filament, where the user can
select a range for the laser to scan over, or a scanning length. The following table lists the

examples of scanning length of 6 different scanning patterns (table 3.4).

53

Table 5.4 FILAMENT scan pattern values

 

 

 

 

 

 

 

 

Filament # Scan Length
0 1 mm
1 1.5 mm
2 1.9 mm
3 4.5 mm
4 6.5 mm
5 8 mm

 

 

 

The third parameter is the velocity. When the glass first begins to melt, the clamps
holding it will begin to move apart due to the decreasing viscosity of the heated glass.
This certain viscosity is a function of the glass temperature. As the temperature of the
glass continues to rise, and the viscosity of the glass correspondingly decrease, the pipette
pulls apart more rapidly until the velocity reaches the programmed value. When it reaches
this programmed velocity value, trip point, the hard pull is initiated. In other words, the
hard pull is initiated by the trip point which is defined by the velocity value. The fourth
parameter is the delay between when the laser turns off and the pull begins. If the delay
value is less than 128, the hard pull will be initiated before the deactivation of the laser
(Figure 5.4). At delay value = 128, the hard pull is initiated at the same time as the
deactivation of the laser. If the delay value is greater than 128, the hard pull will be

initiated after the deactivation of the laser (Figure 5.5).

54

Trip Point

 

 

 

 

 

 

 

 

 

 

/ Velocity Value
Velocity /
Heat value 3
Heat $3 EE 5
" \'\(delay value < 128) msec
_ 3 pull value
Full 1
sec

 

 

Figure 5.4 The delay value less than 128 millisecond.

 

 

 

 

 

 

Trip IPoint
/ Velocity Value
Velocity /
Heat valueilz | i
Heat i
/
(delay value > 128) msec/
a pull value
Pull ' 1 sec

 

 

 

 

Figure 5.5 The delay value greater than 128 millisecond.

Therefore a short delay value will pull the glass before there is any chance of cooling.

The final parameter is the pull parameter. This simply controls the force of the hard pull.

55

5.1.3 Fabrication procedure using P-2000

The metal wire selected for the electrode probe is inserted into the 10 cm long glass
capillary tube. This unit of capillary tube is loaded into the puller as follows [39] (Figure
5.6).

l) Loosen clamping knob.

2) Place glass in V-groove in puller bar, slide it beyond clamp about 2 cm and
tighten knob.

3) Depress the spring stop on each puller bar to release them from their catch
position. (note: catch position is when the capillary tube is pulled into 2 piece.)

4) Pull both bars towards each other using the finger bars. Hold bars in position
using the thumb and index finger from one hand. The hex head screw should be
touching the end of the slot in both puller bars.

5) Loosen both clamping knobs and carefully slide glass through the holes in the side
of the shroud and into V-groove of opposite puller bar.

6) Tighten down clamping knobs

56

Depress here to

  

@

  
 

clamp

\ finger bar

Pullerbar

Figure 5.6 Left puller bar of P-2000 micropipette puller.

After the glass tube is mounted on the puller, close the lid, and then press the PULL key
on the keypad. The laser should turn on and the glass should be pulled with your
designed program line (Figure 5.7). After the separation, loosen the clamping knobs and

remove the pipettes from the puller bars.

Heat applied
Pull Jr Pull

{of} l V

Figure 5.7 Heat is applied by laser and pulling force is activate from both
ends

 

 

 

57

5.1.4 Pulling steps

The pulling is normally divided into two steps. The first step is to pull the tube around
the wire (Figure 5.8). The dimensions of the borosilicate glass capillary tube are outer
diameter 1.0 mm and an inner diameter of 0.5 mm (the P-2000 is optimized for pulling
tubes having outer diameter of 1.0 mm). The diameter of the gold wire is 50 um, thus
the wire is 10 times smaller than the inner diameter of the tube. This gap between the
wire and the glass must be decreased in order to transfer sufficient heat to the metal for
the final pull, and thus the glass and metal form a composite material after the first

pulling step.

  

+3i1i7523532-1-2t‘2-tIgiaj~r_.~;-_'-_:je§:’ gold wire

...............
- mnu-M-“

[// / / / 71 Glass capillary tube

 

.”-\\

"""". .Afflflill

 

 

33'. " ‘:'*l:3;“§§~;"55‘7';'_.. ‘ 4W: @mm Vac ......... 'Z‘EJit' La...-
l'----. 7.
”III”. ”nu"

\\ /
_____

 

 
   

its...) e

/////////

 

 

Figure 5.8 First pull for making composite material.

The second step is the final pulling that separates the composite into two individual
microprobes (Figure 5.9). The objective of the pull is to have the wire run continuously
to the tip of micropipette. Yet, the tip of the pipette should be as small as possible. The

probe tips are tapered down to submicron size. More specifically, the probe tip less than

58

300 nm is appropriate for the electrode probe used for electrical conductivity

measurement.

Gold wire

   

' L“’2t@e:z-~T"Qiri:sifi%i4 '

 

U / / /1 Glass capillary tube

6— Pull Pull —>

 

 

 

 

 

 

 

 

 

Figure 5.9 Final pull

The tip of the microprobe is covered with the glass material, thus it should be etched with
hydrofluoric (HF) acid. One part of the 50% HF was mixed with six parts of H20 to
obtain a solution with a glass etching rate of 120 nm/min. The glass material is
borosilicate which is the amorphous form of SiOz containing: Si02: NaZO: B203: A1203
(~75:7:12:6 by weight). After the tip of nanoelectrode is etched, it is inspected using a
scanning electron microscope (SEM) (Figure 5.10), and the tip dimension is found to be

approximately equal to 300 nm. (1 unit=250 nm)

59

 

Figure 5.10 SEM picture of gold nanoelectrode with the tip dimension of
300nm.
Here the elasticity of the composite material at the glass end (indicated with an arrow A-
A in Figure 5.10) can be calculated using equation (2.17).

_ Em(d; -d})+ Efd}

 

E, d2

_ 640Pa((2.5,um)2 —(1pm)2)+ 77.ZGPa(l,um)2
(2.5m)2

= 66.1 lGPa

The value of dm and df are not the tip dimension but where the glass ends, thus (1! is 1 pm.

Thus the composite material elasticity was calculated at 66.1 GPa.

60

5.2 Other Fabrication technique to make microelectrode

In order to measure the electrical conductivity of small or fragile electronic
materials, a MEMS microelectrode cantilever would be an ideal solution to make the task
easier. Due to the fragility of both the sample and the probe tip, it is even better if
electrode probe is designed in the form of a cantilever beam. The cantilever probe can be
adjusted on the top surface of the sample gently while assuring the sample is not
damaged. Before the purchase of the micropipette puller P-2000, a MEMS
microelectrode cantilever fabrication technique using clean room facility was
investigated.

To fabricate MEMS microelectrode cantilever probes through this lithographic
technique, the following components were used:

a) intrinsic silicon substrate (100) orientation

b) positive photoresist (Shipley 1813)

c) developer (MP-319)

d) acetone (positive photoresist stripper)

e) 1 photolithography mask for V groove

f) HF glass etch solution (HF etch: 6:1 = H20 : 50% HF)

g) KOH etch solution to make V groove in the Silicon substrate.

h) Torr Seal epoxy to place the glass fiber in the V groove of silicon
substrate.

i) a solid glass fiber (with dimension of diameter = 50pm)

j) titanium source for metal evaporator

k) gold source for metal evaporator

61

1) 1 shadow mask
m) four bolts
11) sample holder for evaporator

o) circuit board (2 holes for bolts)

5.2.1 F arming V-groove on the silicon substrate

The silicon substrate should be intrinsic so that it is electrically insulating under
the room temperature environment. The silicon wafer was first oxidized in the oxidation
furnace, so there was SiOz layer on top of the silicon substrate, which will be used as a
mask for KOH etch. Before placing the wafer in the furnace, the wafer should be cleaned
by rinsing with acetone, methanol, and deionized water in that order. The oxidation was
followed by a dry technique in which the oxygen flows directly into the furnace tube, and
then completed by a wet technique in which the oxygen was “bubbled” through a heated
quartz flask containing deionized water. The reason for using those two oxidation
technique is in the following: 1) dry oxidation grows oxide layer with great quality but
the oxidation rate is very low compared to wet oxidation. 2) the oxide grown by wet
oxidation has poorer quality, with less density, however the oxidation growing rate is
excellent compared with dry oxidation. Thus in order to shorten the oxidation time, wet
oxidation was followed by dry oxidation. It is important to operate dry oxidation first to
create the thin oxide layer with good quality. Since it is the interface between oxide and
silicon wafer, the intermediate oxide layer should be in good quality to increase adhesion

between Si and SiOz. In this experiment, dry oxidation was executed for 15 minutes and

62

it was followed by a wet oxidation for 2 hours at the furnace temperature of 1100 °C.

The oxide thickness was calculated by the Deal-Grove Model formula [40]:

t; + Arm = B(t + r) (5,1)
t: +AtO
r = —— (5.2)
B

Where to, is the oxide thickness, B/A is the linear rate coefficient, B is the parabolic rate
coefficient (A and B have units of um and umzlhr), t is the oxidation time, to is the initial

oxide thickness, and T is the parameter to compensate for the rapid growth regime for thin

 

 

 

 

 

 

 

 

oxide.

Table 5.5 Dry and wet oxidation coefficients for silicon.

Dry Wet (640 torr)

Temp (°C) A(jJ.m) B(ttm2/hr) 2(hr) A(].tm) B(pm2/hr)
800 0.370 0.0011 9 - -
920 0.235 0.0049 1.4 0.50 0.203
1000 0.165 0.01 17 0.37 0.226 0.287
1 100 0.090 0.027 0.076 0.1 1 0.510
1200 0.040 0.045 0.027 0.05 0.720

 

 

 

 

 

 

 

According to the Deal-Grove coefficients table [41] above (Table 5.5), for dry oxidation
at 1100 °C the coefficient of A, B, and T, are 0.090 um, 0.027 umz/hr, and 0.076 hour
respectively. The dry oxidation time t is 0.25 hour (15 minutes). By using these values in
equation (5.1), oxide thickness to, is determined as 0.059 um after dry oxidation. The 'c
for wet oxidation can be found by the equation (5.2), and the coefficients of A = 0.11 um,
and B = 0.510 umzlhr for wet oxidation at 1100 °C and the initial oxide thickness to is
0.059 ttm which is the oxide created by a dry oxidation. With these parameters, the value

of 1' is determined as 0.0196 hour. The wet oxidation time, t = 2 hr, so using these values

63

to the equation (5.1), the final oxide thickness after the wet oxidation, to, = 0.96 um,
which is approximately 1 pm. The oxide layer with the thickness 1 pm was sufficient for
KOH etching mask in this procedure.

The silicon wafer was then broken into a rectangular shape with the dimensions of

1.5 cm x 2 cm (Figure 5.11).

 

Silicon wafer 1.5 cm

 

 

 

2cm

Figure 5.11 Cut silicon in rectangular shape

After breaking the wafer into small pieces, positive photoresist was spun coat on to the
substrate (Figure 5.12) using 3000 rpm for 30 seconds. The substrate was then baked, in
a “pre-bake” procedure (also called as a “soft bake” procedure) at 65~75 °C for 15

minutes to dry the photoresist.

 

 

Photoresist
_
A ; $102 -. > u . . .- 44
Si wafer Si wafer

 

 

 

 

.9

Figure 5.12 Silicon substrate with photoresist coated

UV lithography light was then exposed through the mask (Figure 5.13 (a)). The
following picture is the top view of how the mask was placed on top of the silicon wafer

(Figure 5.13 (b)).

shadow nmk
ll

(UV “@10me

    

 

 

L Si wafer T‘ Ij\
Silicon wafer covered with photoresist
(a) (b)

Figure 5.13 (a) UV lithography side view, (b) UV lithography top view

Next, the photoresist was patterned in a developer solution MF-319 (Figure 5.14 (a)).
The period of immersion was about 50 seconds. The developer acts as a solvent which
removes the u.v. exposed, positive resist. During this process, the wafer was checked to
avoid over developing. After developing (Figure 5.14 (a)), the wafer was rinsed with
running deionized water (DI) to terminate the developer stage of this procedure, and a
post-baking step is used to further harden the photoresist. The post-bake procedure or
also called as a hard bake procedure, was executed in the oven at 120 °C 1 5 °C for 15
minutes. Next the underlying silicon dioxide layer was etched using a hydrofluoric acid
(HF) (Figure 5.14 (b)). The HF solution was prepared by mixing 6 parts of H20 and 1
part of 50% HF, which gave an etching rate of typically about 1200 Almin (120 nm/min).
In order to etch 1 pm oxide layer, the HF etching time should be a little longer than 8
minutes. The photoresist mask was removed by acetone, leaving a mask of Si02 as

shown in Figure 5.14 (c).

65

a . . f 1 ' ‘4 i. | l ' ' . '4 I V
‘;, x. ...e t . ...“' .‘.-c.- , |,.' -‘,' .tlc ~,t..£ g... , ,. v .‘u'l ‘er' ..o ,‘v j _ ‘
A. at A t k ,3 4 t s A . .12

 

 

 

 

 

 

 

 

 

 

 

(a) (b) (C)

Figure 5.14 (a) Photoresist deve10ping, (b) Patterning oxide layer, (c)
Stripping photoresist.

The silicon substrate was then anisotropically etched (Figure 5.15) by a KOH solution
[42].

AN ISOTROPIC WET ETCHING: (100) SURFACE
(100) Surface Orientation

   
   

  

‘ 54.74° ’
fffff SILICQN

 

Figure 5.15 Silicon anisotropic etching [42].

The KOH is an anisotropic etchant for silicon such that it etches at different rates along
different crystal orientations. Silicon structure has three primary planes (100), (110), and
(111) in the Miller representation. Anisotropic silicon etching occurs because these
different planes represent a different density of silicon atoms to the etchant as each plane
is exposed and exhibit different surface bonding configurations for each plane. In (100)
plane, the atom has two dangling bonds while (111) has three. Thus it is much easier to
take a silicon atom in (100) plane orientation than (111) plane. In this reasoning, with

KOH anisotropic etchant, (100) surface etches faster than (111) surface. Therefore the V

66

grooves were created in the silicon wafer by KOH etching (Figure 3.16), and its top view

is shown in Figure 5.17.

 

 

 

 

 

Figure 5.16 V-grooves created by KOH etching.

The KOH solution can be prepared by mixing 50g of KOH, 100ml of H20, and 50ml of
methanol [43], and should be used at a temperature of 85 °C. This temperature was
maintained by keeping the solution on a hot plate. The etching time was determined as 1
hour 42 minutes, and the calculation is shown in the next subsection. After the KOH

etching, silicon oxide mask was removed by a HF solution (Figure 5.17).

 

 

V—V—W
Si wafer

 

 

 

 

 

 

(a) (b)

Figure 5.17 (a) The top view of V-groove after removing oxide. (b) The
side view of V-groove after removing oxide mask.

67

5.2.1 .1 Etching time calculation

The V groove in this experiment will have a width of 80 um, and the anisotropic etch of
silicon at a (100) orientation gives an angular profile of 54.74 ° (Figure 5.15). According
to the triangle relation’s as seen from the (Figure 5.18), the groove depth can be

calculated:

 

tan 54.74° = mm (5.3)
100m

x = 141.4,um
Thus, 141.4 pm is the depth that needs to be etched. The etching rate of silicon wafer by
the KOH solution is 1.38 pm / min along the (100), direction. Multiplying the etching
rate by the undetermined time of etching gives the etching depth which equals 141.4 pm.
The etching time follows from the calculation bellow:
1.38m/minx ymin =141.4,wn (5.4)
y = 102.5 min

Therefore, the KOH silicon etching time is 1 hour 42 minutes 30 seconds.

‘( groove width
/200 um\

100um

F141. 4 um

 

 

Figure 5.18 Triangle relations on the V-groove

68

5.2.2 Preparing the glass fiber metal evaporated electrode probe

A Pyrex glass fiber with diameter of 50 to 100 um was cut in pieces and glued
into V-groove traces of silicon wafer using Torr Seal epoxy. Torr Seal epoxy resin is
conventionally used to seal leaks on any type of vacuum system or components. It is
solvent-free and can be used at pressure of 10'9 Torr and below, and temperatures from 45
°C to 120 °C [44]. Since the wafer with glass fiber was placed in the evaporator where
the chamber was vacuumed, to be coated with some metals later, the Torr Seal was
suitable as a epoxy to hold the glass fiber into the V-groove created in silicon wafer.
Here the temperature allowance range for Torr Seal is up to 120 °C. In the metal
evaporator, the crucible containing metal sources requires an extremely high temperature
in order to boil/evaporate those designated metals; however, around the target/sample
area where the metal is coated does not need as high temperature as in the crucible. Due
to these reasons, Torr Seal epoxy was selected.

Small amount of the epoxy can be placed in the grooves, followed by the optical
fibers. The lengths of fiber cantilever beams were aligned by putting a ruler against the

end of the fibers before the epoxy dries/cures (Figure 5.19).

 

 

 

 

 

 

Place the optical fiber
Glass fiber
(20 (b)
Figure 5.19 The glass fibers mounted on V-groove (a) Top view, (b)

Side view

69

After the glass fibers were carefully mounted on the grooves of silicon wafer, the
unit was mounted and placed in the evaporator. The sample was covered with shadow
mask (Figure 5.20) and screwing down with bolts to the sample holder. The sample
holder was then mounted on the lid of the metal evaporator where was directly above the
metal boat. A titanium source and a gold source were placed into the boat and chamber

was closed to acquire a specific vacuum environment for each metal to evaporate.

Shadow
Mu

 

 

(a) (b)

Figure 5.20 Sample covered with shadow mask for metal evaporation
(a) Top view, (b) Side view

First titanium was deposited by metal evaporation technique. The goal here was
to deposit patterned gold metal on the grooves to make electrical connections for four-
point probe. Gold was preferred due to its non-oxidizing properties and its low resistivity
(p = 2.26 119cm at T = 298 K). However, gold does not exhibit good adhesion to SiO2,
therefore an intermediate layer such as chromium is often used which has significantly
better adhesion properties with SiO2. The coefficient of thermal expansion (CTE) of gold
(14.2 x 10‘6 /K) and silicon dioxide (0.4 x 10‘6 /K) has a significant difference, so that it
is recommended to introduce an intermediate layer before the gold deposition to enhance

the adhesion. There are several candidates for intermediate layers including titanium,

70

tungsten, nickel, and chromium. And titanium was selected as an intermediate layer due
to the availability at the Keck Microfabrication Facility located in the Physics and
Astronomy Department at Michigan State University. Moreover, the CT E of titanium is
8.6 x 10'6 /K, which is somewhat in between values for silicon dioxide and gold.
Coefficient of Thermal expansion Oi is defined as, oerzdexldT. Metal films deposited on
$102 expand at different rates as the temperature changes. Those films are deposited at
high temperature, and the sample are taken out and cooled down at room temperature.
When the substrate is cooled down, some stress develops in the thin film due to this
mismatch in expansion/contraction. Thus as temperature changes strain builds up in the

material, which can lead to bend (or curled) cantilevers.
e, (T) = e, (To) + AT (5.5)

Before the metal evaporation, the evaporator chamber needs to acquire certain
vacuum environments for specific metal evaporation run. Evaporation requires the
chamber environment to be 5 x 10’7 Torr. The evaporation can be done consecutively if
the evaporator allows more than one boat in the chamber with appropriate shutter. At the
Keck facility, the evaporation chamber has four boats with a shutter above, and thus there
are four options for metals at most. The vacuum environment was acquired with two
pumping systems called rough pump and turbo pump. The rough pump is capable of
pulling a vacuum down to about 1 x 10‘3 Torr. High vacuum pump such as turbo pump
can further reduce the pressure to approximately 1 X 10'6 Torr to l x 10’8 Torr. Thus the

rough pump was operated first until acquiring l mTorr, and after such an environment

71

was retained the turbo pump took over the rough pump for finalizing the vacuuming
process.

When the chamber acquired the certain vacuumed environment, the metal
evaporation can proceed. The evaporation was done by heating the boat which contained
metal inside. The both ends of boat were connected to electrodes and the current was
passed through the boat to resistively heat the boat by joule heating. A pre-deposition
heating of the metal source was used to remove contamination from the metal source and
the boat. After this step, the shutter between the metal source and the sample was opened
for the deposition step. The boiling point of titanium and gold are 3287 °C and 2856 °C
respectively [45], so it needs a higher value on the current control switch for titanium to
start evaporating it. The titanium layer of thickness = 10 A was first deposited (Figure
5.21 (a)), and the gold layer of thickness = 30A was deposited as a following step (Figure
5.21 (b)) in the evaporator. As a result, the gold metal was coated over the cantilever

beam with a intermediate layer of titanium (Figure 5.21 (c)).

Titanium Deposition

Gold deposition

 

 

 

(a) (b) (C)

Figure 5.21 (a) Titanium deposition, (b) Gold deposition, (c) final
look after metal depositions

72

This fabricated MEMS four-point probes can be then mounted on the xyz translation
stage, and the resistivity of the semiconductor sample can be measured by the probes

(Figure 5.22).

 

 

 

 

—-n

:lill—

 

 

1 1
Lou
{UL—HH—

 

 

 

 

Figure 5.22 The circuit of fabricated four point probe.

5.2.3 Results

Although this fabrication technique was planned originally, there were few difficulties
encountered during the fabrication process. During the preparation of making V-groove,
the oxidized silicon wafer should be coated with photoresist on both front and backsides.
Since the normal silicon wafer has a thickness of 500 pm or less, the KOH anisotropic
etching thinned silicon substrate and made it fragile if the back side of the wafer were not
covered with oxide layer. In the furnace, oxide layers were grown both front and
backside, and both layers should be a mask for KOH etching. Therefore the photoresist
should be coated on both sides. First, the entire top surface of substrate was covered with
photoresist, and then the substrate was placed in the oven for prebake. The wafer was

then turned over and mounted on spin coater. The backside of the silicon wafer was

73

coated with photoresist in advance. This proved to be challenging, since when the spin-
coater started to spin the substrate, the vacuum system activated and pulled vacuum on
the photoresist coated side of the wafer which damaged the photoresist coating on the
backside.

The front side is extremely important, since the v-groove pattern is done by
lithography on the front side. For a better photoresist film on the front side, the front
side was processed only after the backside was completed. Damage to the backside
photoresist layer continued to allow significant KOH etching of the substrate and resulted
in fragile substrates.

In the second trial after above problem had been encountered, the V-groove
creation was discontinued. Instead of coating the whole unit with metal sources, the
individual components, silicon substrate and the fiber, were coated with metals first and
glued (with electrically conducting adhesive) together in the end. This technique met
with limited success after significant effort was made in aligning the fiber to the metal
pattern created on silicon surface.

In addition to the difficulties of the fabrication procedure by clean-room method,
the size of the probe tip was limited. Since the normal glass fiber of diameter 50 um
without tapering was used as a probe, the tip dimension was limited at the micron size.
However, with the use of the P-2000 micropipette puller, the probe tip size can be drawn
in sub micron size (unfilled glass tubes have been pulled down to 17 nm in diameter and
wire filled tubes down to ~100 nm have been fabricated) extending our capabilities. This
enables to probe not only the small structured sample but also the small sample with

doping profiles within the extremely localized area. In addition, this technique results in

74

insulated probes such that thermocouple probes or multifunctional probes can be
envisioned. The reduction in exposure to, and use of hazardous chemicals in this

technique is also a desirable result.

75

6. DATA ANALYSIS

6.1 One-point probe scanning measurement method

To test the submicron-electrode, the electrical conductivity of a sample of Bi2Te3,
copper, and graphite was measured using the one-point probe method. Conventional
methods of measuring electrical conductivity include the four-point probe method and the
two-point probe method. We have developed a new technique for our lab that uses only
one probe to scan the length of the semiconductor material. This allows the
measurement, not only of the conductivity of the sample, but also of the contact resistance
at the interface of the semiconductor and the metal.

In the four-point probe and the two-point probe methods a constant probe spacing
is employed. The constant probe spacing enables the simple calculation of conductivity
according to the formula shown below:

[I
a:—
VA

(6.1)
where 0'18 the conductivity, 1 is the current, I is the probe spacing, V is the voltage and A
is the cross-sectional area of the sample.

The four-point probe method has two outer probes where the current source is
applied to the sample, and two inner probes where the voltage difference is measured.
Similarly, the two—point probe method also features a set of inner probes that measure the
voltage difference. However, in this case the current is applied from the edges of the

semiconductor material through a metal contact. The thin metal contact is created either

by sputter coating, metal evaporation, or electrochemical deposition. In both probe

76

methods, since the voltage is measured between the two inner probes, the contact
resistance is not included. Thus, the one point probe method has the advantage of
detecting the voltage drop at the contact resistance. This allows the further study of the

contact resistance and helps to improve and make better contact with appropriate metals.

6.1.1 The probe mounting angle calculation

In order to verify the probe mounting angle, the maximum deflection formula was
used. In chapter 2, the maximum deflection formula was combined with a trigonometric
formula and, it was simplified as equation (2.25). In order to find the optimum angle, the
force applied, the tapered length of the nanopipette, L, and the average radius of the
tapered region, r, have to be estimated.

For a commercialized Hall measurement probe, the force on a 5 mil (127um)
radius of the probe tip is rated at 50 gram [46].

The pressure from the probe is:

Force _ 50g
TipArea Iz'(l27,ttm)2

 

(6.2)

In our case, we have a cantilever beam with a tip diameter of 250 nm. 80 the tip
radius is 125 nm; hence, for the same pressure the approximate force to be applied to our

cantilever is:

__598__ = __F_
”(127m)2 zz(12snm)2
F = 4.84x10'5 g

The unit of EC, the composite material elasticity is converted to SI units: from GPa to

103 g/m2

g/mzz EC: 66.1x109Pa(-9—§ )= 6.73x10'2g/m2

 

Pa

77

The tapered length of the nanopipette is approximately: L: 4 mm.

Since the cantilever radius is not consistent, the averaged radius of the tapered region is
estimated as follows. In the assumption, the tapering starts when the glass diameter is
approximately equal to lOum, and the tapering region ends at 250 nm (tip diameter).

Assuming the tapering occurs at linear rate (Figure 6.1), the formula is established as

follow:

y = slope X x + 125nm
_ (5,1071 —125nm)
— 4mm

yx=2mm = 256/9"

 

x+ 125nm (6.3)

Thus cantilever radius is: r = 3 pm.

Y slope=(5 125nm)/4mm

taperingstartat
(_ theradius=5um

'Iheprobetip
radius=125nm

 

 

 

 

4mm

Figure 6.1 Calculation of the approximate cantilever radius.

4FL2
tan 9 = 4
3E!”

19 - tan“l 4FL2
3Ec7t r4

=tan“[ 4(4.84X10'5gX4X10’3m)2 ]
3(

6.73x10'2g/m2):(3x10-6m)‘

 

 

=31.09°

78

These calculations verify a probe mounting/ deflection angle of 30° is appropriate.

6.2 DC sweeping current source using Keithley 2400 source meter

6.2.1 Bi2Te3

First, the edges of the Bi2Te3 sample were coated with nickel by electrochemical
deposition to form the electrical contacts. 1- (ground) and V- leads are connected to the
metal contact on the left, and the I+ lead is connected to the metal contact on the right
side of the semiconductor. The gold electrode probe with a tip dimension of 300 nm is
mounted on an xyz-translation stage at an angle of 30° with respect to the sample. The
sample is sourced by sweeping lOmA using 2400 source meter from Keithley Instruments

[47] (Figure 6.2).

Keithley 2182 nanovoltmeter

00.135268mV .02,
‘ - a

    
  

 

 

 

 

 

 

 

. 4:
Keithley 2400 source Eameter/ !.

xyz translation stage

 

 

 

Figure 6.2 DC conductivity measurement system set up[48].

The electrode scans the length of the sample from left to right. Both probe spacing, l, (in

this case, the distance of the electrode from the left metal plate) and the voltage

79

difference, V, increase as the electrode scans the surface (Figure 6.3). The voltage

difference is measured by 2182 nanovoltmeter [49] from Keithley Instruments.

Metal I+
—\

l

.<
s

 

 

 

 

 

\\\XALL\\\\
\\\ R \\\\\\\

 

 

 

 

 

 

 

 

 

 

V

Scan left to right

Figure 6.3 Sample set up.

Since the metal contact is included in the probe spacing area, the voltage drop (or contact
potential) at the metal contact is added to the voltage measured when the probe is
scanned. When I = 0, only the contact potential is measured, and then the voltage
increases linearly with l. The data taken from the measurement is listed in Table 6.1 and

shown graphically in Figure 6.4.

80

Table 6.1 Data of scanned voltage difference on Bi2Te3 sample using DC

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

technique.
Probe spacing Voltage drop Voltage drop Voltage drop
(1st run) (2“I run) (3rd run)
0 5.6 pV 5 EV 4.85 1.1V
1 129pV 140 [N 203 uV
2 136 uV 165 [N 228 [W
3 137 EV 156pV 225 pV
4 140 pV 173 uV 213 [N
5 141 [N l8lJlV 224 11V
6 144 pV 187 uV 224 11V
7 147 uV 188 uV 232 11V
8 150 [W 198 pV 235 11V
9 155 uV 200 [N 242 11V
10 153 11V 203 11V 245 uV
11 156 uV 205 uV 256 11V
12 167 [N 208 11V 250 uV
13 170 uV 211 [N 253 uV
14 175 uV 194 pV 258 11V
15 179 11V 186 11V 260 11V
16 183 11V 189 pV 282 uV
17 187 pV 191 uV 280pV
18 189 [W 212 [W 310pV
19 194 11V 213 uV 316 11V
20 194 [W 213 uV 300 11V
21 450 1.1V 574 11V 583 11V

 

Contact resistance of the Nickel plating is calculated as follows.

The differential voltage was taken from the data acquired at 1St run.

_ (0.129—0.0056)mv

 

 

contuctR - = 12'3 m9
left side IOmA

Wk = (0.45 -0.194)mV = 25.6 mg
right side 10mA

81

 

 

700

 

600

 

500

 

400

 

 

300

 

Voltage difference (UV)

 

 

 

200

 

100

 

 

 

 

 

 

 

 

0 5 10 15 20 25

probe spacing (1 = 240 um)

Figure 6.4 Plot of scanned voltage difference on BizTe3 sample using DC
technique.

To calculate the electrical conductivity, the slope of the linear region has to be
determined. The plot is linearized by KaleidaGraph software and the function including a
slope of the line is determined (Figure 6.5). The slope of the plot is inversely related to
the electrical conductivity of the sample, and the formula is shown below:

_ IA]

0 — — 6.4
AVA ( )

82

0.35

0.3
9
E

g 0.25
C
9
Q)
.a_=
‘O

8, 0.2
.0
a
>

0.15

0.1

-0.1

 

 

I I I I T l I I I I I I I fl I I I r I I I I I I I I I I I

y = 0.12756 + 0.14799x Fl: 0.99033

 

..... y 2 0.16578 + 0.1093x R: 0.77832 D D

 

ll IALI I ll I l l lLlllllll l l 111 I IL

I l 1 CL 1 l L l L 1 J L l l 1 l 1

L J L 1 l l l

 

 

0 0.1 0.2 0.3 0.4

Probe spacing (cm)

Figure 6.5 Plot of the linear region of scanned voltage difference on

BizTe3 sample using DC technique.

The slope of each linear fit gives the value of (%j for use in equation (6.4)

A1 is simply the difference between the final probe location and the initial location:

0.456cm.

The dimensions of the BizTe3 are: width (w) = 0.445cm, thickness (t) = 0.22 cm, length

(l) = 0.525cm, thus the electrical conductivity of the Bi2T63 is calculated as follows:

83

0.5

lOmA

 

 

 

= = 690S/
' (0.14799mvIcm)[(0.445cm)(0.22cm)] cm
lOmA
a = =934S /
2 (0.1093mv/cm)[(0.445cm)(0.22cm)] cm
mm
0' = = 483S/
3 (0.20356mV Icm)[(0.445cm)(0.22cm)] cm
_ 690+934+483 : 7025/0"

 

average _

Thus the electrical conductivity of the BizTe3 is: 702 S/cm

6.2.2 Aluminum foil

To test the gold electrode probe further, aluminum foil is used as the secondary
sample. Metal has much higher electrical conductivity than semiconductor, and this
makes it difficult to detect the increment in voltage with a scanning probe. In order to
increase the total resistance of the aluminum, the foil is cut in long strip with the smallest
width possible. The aluminum foil is cut with a width of 4.1 mm, a length of 20.9 mm,
and a thickness of 0.02 mm as measured by Vernier calipers. Total resistance of this

aluminum sample is calculated by the following:

 

l
Rtoral : p74—
2.09cm
= 2.67 uflcm
(0.41cmX0.002cm)
= 6.8m!)

If I = i50 mA is swept, the voltage across the aluminum is:
Va, = IR = 340/1V
Thus the voltage increment from I = 0 mm to l = 20.9 mm should be 340 11V, and the

value is large enough to be detected by a nanovoltmeter. It was discovered that these

84

small probes can be easily damaged on harder surfaces such as aluminum, however,
surface contaminants such as oxides require greater pressure on the probes to break
through the contaminant layer. Because of this, the data was not as stable as with the
previous sample. To strengthen the tips, a coating of silver paste at the tip was
investigated. This significantly enhanced the signal quality, by not only strengthening the
probe tips, but also by decreasing the contact resistance to the sample. However, in order
to achieve the original goal which was the fabrication of a submicro-electrode probe to
characterize the electrical conductivity of an extremely localized area, the probe was
tested without applying silver paste as an enhancement.

The problems listed above are all due to the extreme small dimension, (~300nm)
of the probe tip. To help avoid damage to the probes, a microscope was used during the
measurement; however, the data still showed instability even if the cantilever is deflecting
and contacting to the sample surface. In order to see this problem, the conduction is
tested by using multi-meter. The electrical conduction was tested by clipping the V+ wire
end, the wire connected to the nanoelectrode, and touching the aluminum surface while

the nano electrode made the contact to the sample (Figure 6.6).

85

 

CD

A?

 

 

 

clipping V+ wire
touching sample surface

 

U \I

 

 

 

sample

Figure 6.6 The electrical conduction test of submicron probe.

The conduction was not stable, and it alternated on and off rapidly. This resulted in
unacceptably noisy data. The contact is not consistently made, and therefore the
nanovoltmeter could not read a stable voltage value. One of the reasons might be
artifacts due to improper vibration isolation. Since the probe is extremely small, it is very
sensitive to even a small vibration. Furthermore, the nanoelectrode might have caught
some other noise signal (e.g. electromagnetic wave) from the environment. In order to
enable a more precise measurement, the electrode is employed using an AC source with a
noise filtering feature using EG&G 7265 DSP Lock-in amplifier. The set up of the

system is explained in the following section.

86

6.3 AC source using EG&G 7265 DSP (Digital Signal Processor) Lock-In Amplifier

In this section, it is described how we built an AC conductivity measurement
system using EG&G 7265 DSP lock-in amplifier. Before the system set up is explained

the basic concept of the lock-in Amplifier is reviewed.

6. 3.1 Lock-in Amplifier

In its most basic form, the lock—in amplifier is an instrument with dual capability.
It can recover signals in the presence of an overwhelming noise background or,
alternatively, it can provide high-resolution measurements of relatively clean signals over
several orders of magnitudes and frequency. A common problem has arisen while
detecting and measuring the amplitude of an electrical signal in the presence of much-
larger-amplitude, random, electrical noise. This electrical noise covers a wide band of
frequencies. The lock-in amplifier is a powerful device for filtering out the unwanted
noise by utilizing a phase-sensitive detection method.

A lock-in amplifier uses phase-sensitive detection to improve the signal-to-noise
ratio in a common experiment. The measured input signal will include an analytical
signal and the other noise signals. Using phase-sensitive detection requires the analytical
signal to be modulated at some reference frequency. The lock-in-amplifier then amplifies
only the component of the input signal at the reference signal, and filters out all other
frequencies (e. g. noise). Thus, the demodulator operates by multiplying these two signals
together to yield the “demodulator output”.

Input/measured signal x reference signal = demodulator output
Asin(wt+¢)* Bsin(wt)= -/%B:cos¢+l—4§Iicos(2wt +¢) (6.5)

87

Since there is no relative phase-shift between the input signal and reference phases (e.g.
0:0) the demodulator output takes the form of a sinusoid at twice the reference
frequency, but with a mean, or average, level which is positive (Figure 6.7). The mean
level is the DC component of the demodulator output, so it is a relatively simple task to
isolate it by using a low-pass filter. Since the reference signal amplitude is fixed, and the
reference phase is adjusted to ensure a relative phase-shift of zero degree, the input signal

amplitude can be determined by measuring the mean level.

\ A Signal In

Reference

\ A (internally
_ V U generated)

: Demodulator

   

 

mean level +ve

 

Figure 6.7 The demodulation of two signals [50]

The above discussion is based on the case of noise-free input signals, but in real
applications the signal will be accompanied with noise. This noise, which by definition
has no fixed frequency or phase relationship to the reference, is also multiplied by the
reference signal in the demodulator. However, this does not result in any significant
change to the mean DC level, due to its different frequency set. Noise components at
frequencies very close to that of the reference do result in demodulator outputs at very

low frequencies, but by setting the low-pass filter to a sufficiently low cut-off frequency

88

these can be rejected. Hence the combination of a demodulator and low-pass filter can be

used to monitor signals at a given frequency.

6.3.2 Setup procedure using EG&G 7265 DSP (Digital Signal Processor) Lock-In
Amplifier

The basic concept of lock-in amplifier was discussed in the previous section. In
our application, the internal reference signal was utilized to measure the signal having the
same frequency as the output oscillating sinusoidal signal.

The experimental setup with the lock-in amplifier is explained in the following.
First the sample was mounted on a circuit board, and the gold nanoelectrode probe was
mounted on the xyz translation stage at an angle of 30° with respect to the sample. The
wires were soldered to both ends of the copper strip and the oscillating sinusoidal signal
was sourced to the sample using EG&G 7265 DSP (Digital Signal Processor) Lock-In
Amplifier [51]. To monitor the value of current applied to the sample, a 1 Q resistor was
connected in series with the sample as shown in Figure 6.8. The current was determined
by measuring the voltage across the 1 (2 current sense resistor using the lock-in amplifier.
The voltage across this sense resistor was measured by the lock-in amplifier using the
differential measurement mode with both A and B channels. Since the resistor value is 1
Q, the current sourced to the sample is simply equal to the measured voltage across the

resistor.

89

7265 DSP Lock-In Amplifier
m “3"— 1 0 resistor Copper strip
I; II-

 

 

 

 

 

 

 

 

 

Figure 6.8 Lock-in amplifier set up to determine the current of the system

After measuring the value of current, the voltage difference between the probe (V+) and
the V- was measured using the lock-in amplifier with a signal channel input node A and
ground (Figure 6.9). The probe was then scanned from the side I- to 1+, and thus the

voltage should linearly increase from [=0 to l=L.

7265 DSP Lock-In Amplifier
(—
£8 “(3‘5 lflresistor /I

 

 

 

 

 

 

 

 

 

Figure 6.9 Lock-in amplifier set up to measure the scanned voltage of the
sample.

A photograph of the AC conductivity measurement system is shown in Figure 6.10.

90

 

 

 

 

Figure 6.10 Lock-in amplifier set up picture

As a first trial, a frequency of 50 Hz, and an amplitude of 2 V rms signal was
sourced to a copper strip. The voltage across the resistor was measured with the value,
34.19mV, and thus the source current with this setting is I = 34.19 mA. However the
voltage measured by the nanoelectrode probe was not stable. In order to improve the
signal, an oscilloscope was used to monitor the signal in the circuit. The signal was
inspected at the point A in the circuit using the oscilloscope. The signal captured by the
oscilloscope was not a pure sinusoidal wave, but exhibited cropping of the sinusoidal
extrema. It was speculated that the compliance current of the lock—in Amplifier exceeded
the limit, and thus the signal was not a pure sinusoidal curve. In order to get a more
stable signal, the amplitude of the oscillating voltage signal was reduced to 1 V rrns.

With the lowered rms voltage, the signal captured at the point A became sinusoidal. This

91

was found to be very important step that significantly improved the measured signal

quality and is recommended for each sample measured.

6.3.3 Copper strip

Utilizing a 1 V rrns oscillator signal at 150 Hz is the circuit shown in Figure 6.9,
the measured voltage across the 1 Q resistor was 19.28 mV, which indicated a source
current of 19.28 mA. To test the probe measurements, the total resistance of copper strip
and the entire circuit resistance were measured in an independent measurement system
using a Keithley 2400 source meter and 2182 nanovoltmeter. The sweep current of lmA
was sourced to the copper strip and the voltage drop was measured as 0.0217 mV by the
nanovoltmeter. The total resistance including the current lead wires [constantan wire (50
um diameter)] and the interface resistance of the solder to the copper strip was also
measured (0.816 mV for lmA source current). By taking ratio of VII, the total resistance
of the copper strip, and of the entire circuit was measured to be 21.7 m!) and 0.816 D
respectively. The large increase in impedance of the entire circuit is due to the constantan
wire with the small cross sectional area. The dimensions of copper strip are: thickness =
0.002 cm, width = 0.158 cm, and length = 3.763 cm. The resistivity of copper at the
room temperature is 1.72 uflcm [52]. Thus the total resistance of the copper is calculated
as:

3.763cm
(0.002cm)(0.158cm)

 

Rm, 2 1.72;:SIcm

= 20.48mQ

Thus the percentage error is:

92

 

%error = 100x

2100

R

measured

R

calculated

measured

x 21 .7mQ — 20.48mQ

 

21 .7m§2

= 5.62%

Now the circuit is analyzed as follows.

First the rms voltage value of the lock-in

amplifier was verified using the oscilloscope (Figure 6.11). A BNC cable was connected

to the oscilloscope and the output oscillating signal was captured.

 

1Vrms 509
(9 w»

 

inside lock—in

 

amplifier

 

oscilloscope

‘VVV——-

 

 

1MQ

 

Figure 6.11 Oscilloscope connected to the output of lock-in amplifier.

The peak of the signal captured was 1.5 V and the nns value of the signal was calculated

by:

93

 

The value is reasonable since the lock-in amplifier is sourcing 1 V rrns oscillating signal.
In the next step, the voltage of the nodes A and B in the circuit (across the resistor) are

captured by the oscilloscope as the channel 1 and 2 respectively (Figure 6.12).

 
  
  

hannel 1 (node A)

hannel 2 (node B)

Figure 6.12 The signals captured by Oscilloscope: A) channel 1 B)
channel 2 (20 mV/div)

The rms values of the nodes A and B are calculated and the voltage difference is obtained

as follows:

meu = 26m x20mV /div = 52mV
VP”,d9 =1.2d,.v X 20mV Idiv = 24mV
_ 52mV

Vm, = 3578me
Vm, = 24—m‘1 =16.97me
J2
Vacrosslfl = VrmJA — VrmsB = 35.78"!er —16-97me
= 1881me

94

The value measured here verifies that the voltage difference of 19.28 mV (A-B) taken by
lock-in amplifier is reasonable. The discrepancy is due to the approximated estimation of
the division at the peak of the real signal obtained by oscilloscope.

The total resistance of the circuit board is measured at 0.816!) by the
nanovoltmeter, so the voltage drop across the circuit would be (Figure 6.13): 19.28 mA x
0.81652 = 15.73 mV. Including the voltage drop across the resistor, the total voltage
difference is: 35.01mV. Since lock-in amplifier is sourcing 1 V rms voltage, there is
some voltage drop at the internal impedance of the oscillator: 1 V — 35.01 mV = 0.965
V. The internal impedance of the lock-in amplifier is: 0.965 V / 19.28 mA = 50.059.
This value is verified by the hardware manual of EG&G 7265 DSP (Digital Signal

Processor) Lock-In Amplifier: Output oscillator impedance = 50 Q.

 

 

 

 

1 V rim
®_5.Mm_ 1 0 resistor box
inside lock-in ‘
' V
mlphfier T c 1 1M0 0.8160
MEI (includingcoppersuip)

Figure 6.13 Lock-in amplifier circuit analysis

In order to test the one-probe scanning conductivity measurement system with
lock-in amplifier, bigger probes were employed. These are listed below: 1) a gold coated

pin with 500 um diameter (large dimensioned probe). 2) A Pt wire with 50 um diameter

95

(intermediate dimensioned probe). The data successfully acquired using these probes was

tabulated (Table 6.2) and plotted (Figure 6.14) in the following.

Table 6.2 Data of scanned voltage difference of copper using AC
technique and pin probe and Pt wire probe.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Probe spacing (l) Scanned voltage Scanned voltage
(500 um diameter pin) (50 um Pt wire)

0 11.31mV 11.63 mV
1mm 11.32 mV 11.63 mV
2 mm 11.33 mV 11.63 mV
3mm 11.33 mV 11.65 mV
4mm 11.34 mV 11.65 mV
5 mm 11.35 mV 11.68 mV
6mm 11.35 mV 11.70 mV
7 mm 11.36 mV 11.73 mV
8 mm 11.37 mV 11.74 mV
9mm 11.38 mV 11.76 mV
10mm 11.40 mV 11.78 mV
11 mm 11.42 mV 11.79mV
12mm 11.44mV 11.81 mV
13 mm 11.45 mV 11.83 mV
14mm 11.47 mV 11.79 mV
15 mm 11.48 mV 11.81mV
16mm 11.51mV 11.83 mV
17 mm 11.52 mV 11.85 mV
18mm 11.54mV 11.86mV
19 mm 11.56 mV 11.87 mV
20 mm 11.57 mV 11.90 mV
21 mm 11.58 mV 11.91 mV
22 mm 11.60 mV 11.93 mV
23 mm 11.61 mV 11.94 mV
24 mm 11.63 mV 11.96 mV
25 mm 11.64 mV 11.97 mV
26 mm 11.66 mV 11.98 mV
27 mm 11.67 mV 11.99 mV
28 mm 11.68 mV 12.0] mV

 

96

 

copper strip

 

 

 

 

 

 

 

 

13 I'II'UTleI'IIIUI'IIIFVUIIII‘1IUII

t y = 11.284 + 0.014721x R: 0.99487 ‘

r d

- ----- y = 11.612 + 0.015955x Fl: 0.9949 1
A 12.5 - —e— probe: a pin (diamter = 0.5 mm)
E ' - -O - - probe: Pt wire (diameter = 50 um)
v i-
d) n
0
C
e P
g 12 -
B I
Q)
C) .
E
6 I
> .

11.5 '-
" 1
11 l l l l I l l 1 l l l L 1 l l l l I l l l l l l I I l l l J l l l l
-5 0 5 10 15 20 25 30

probe spacing (mm)

Figure 6.14 Plot of scanned voltage difference of copper using AC
technique and pin probe and Pt wire probe.

The voltage difference across the sample is: AVl = 11.696 mV —11.284 mV = 0.412 mV.

AV; = 12.059 mV —1 1.612 mV = 0.447 mV.
The expected voltage difference across the sample is as following: 19.28 mA x 21.7 m!)

= 0.42 mV. The conductivity of the copper sample is calculated as follows:

97

a = (19.28mA)
P‘" (0.01472mvImmx10mm/cm)(0.002cm)(o.158cm)
a = (19.28mA)
W" (0.015955mV/mmxlOmm/cm)(0.002€m)(0.158cm)

 

=415 kS/cm

 

=382kS/cm

1

The electrical conductivity of copper is the reciprocal of the resistivity: 050“," = T
. mm

= 581 kS/cm. The measured conductivity by the one-point probe is lower than the
standard. The cross sectional area of the copper sample was measured under a
microscope; however, the thickness might vary in each location due to the surface
treatment. (The surface of the copper strip was polished to improve the smoothness.) The
measured conductivity seems to have a little offset due to this inconsistent cross sectional
area dimension.

After the system was verified with the larger probes, the small probe (tip
dimension = 700 nm) is now employed. First, the pulled quartz pipette coated with gold

is used as the nanoelectrode probe (Figure 6.15).

 

Figure 6.15 SEM picture of gold nanoelectrode

98

The gold material is sputter coated for 2 minutes (sputter coating rate = 7 nm/min), thus
the gold coating is approximately equal to 14 nm. It was extremely difficult to make a
contact to a sample with the nanoelectrode; and thus, the probe need to be deflected
significantly in order to take data. Fortunately, the nanoelectrode had a long tapered
neck, enabling it to deflect a large degree. The required deflection to acquire data
depends on the location of the sample measured. Therefore, it is speculated that, some
parts of the sample suffered topographical roughness or oxidation. However, even with a
large amount of deflection, the probe could not make a good contact at some locations on
the sample. In these cases, the probe was repositioned slightly in the x and y direction to
make a better contact. With this effort, data acquisition was completed (Table 6.3)
(Figure 6.16). Due to the extended length of this process, the data sometimes decreased
abruptly as an offset. This occurred at the transitions of l = 10mm ~ 11mm, and l =
17mm ~ 18mm. Therefore the conductivity of the copper is now taken by averaging

those 3 data sets (slopes) as follows:

 

 

0'1 2 (1928M) = 738kS/cm
(0.082727mV /cm)(0.002cm)(0. 158cm)
0'2 = 0928"“) = 356kS/cm
(0. 17143mV/cm)(0.002cm)(0. 158cm)
(19.28mA)

= 506kS / cm

 

03 Z (0. 12061mV /cm)(0.002cm)(0. 158cm)

0' = (G'+0_;’2+a3)=533kS/cm

l

 

99

Table 6.3 Data of scanned voltage difference of copper using AC

technique and gold nanoelectrode.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Probe spacing (1) Scanned voltage (mV)
0 mm 13.20 mV
1 mm 13.20 mV
2 mm 13.21 mV
3 mm 13.21 mV
4 mm 13.22 mV
5 m 13.23 mV
6 mm 13.24 mV
7 mm 13.26 mV
8 mm 13.26 mV
9 mm 13.26 mV
10 mm 13.28 mV
11 mm 13.19 mV
12 mm 13.21 mV
13 mm 13.22 mV
14 mm 13.24 mV
15 mm 13.26 mV
16 mm 13.28 mV
17 mm 13.29 mV
18 mm 12.83 mV
19 mm 12.85 mV
20 mm 12.86 mV
21 mm 12.88 mV
22 mm 12.89 mV
23 mm 12.91 mV
24 mm 12.92 mV
25 mm 12.92 mV
26 mm 12.93 mV
27 mm 12.94 mV

 

100

 

 

 

 

 

 

14 I I I Ij I I I I l I I I I 1 I I I I I I I I I I I I I I I I I I I
' y = 13.192 + 0.0082727x H: 0.979 ‘
- - — y = 13.001 + 0.017143x R: 0.99655 4
13 5 '_ ————— y = 12.622 + 0.012061x R: 0.98335 1
E 1- cl
V ’ 9.9 t
a: - WW0 o 96’6 e .
8 _ .
9
d) 13 r- _‘
3:
'6 i- 000 9 9
8, . o 90‘} e .
.9
B D I
> I I
12.5 - 4
- -l
i- It
- -l
12 LJ I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
-5 0 5 10 15 20 25 30

probe spacing (mm)

Figure 6.16 Plot of scanned voltage difference of copper using AC
technique and gold submicron electrode (d=700nm).

6. 3.4 Graphite sample

In order to protect the tip from damage, a soft sample such as graphite was used
for the measurement. Also a Pt microelectrode was used here to make a strong contact
without causing any damage.

The pulled gold nanoelectrode was tested on a copper sample as discussed
previously; however, the data was unacceptable due to problems of poor contact. It was
speculated that since gold is a soft material (when compared with other metals) it was not

able to tolerate the contact pressure imposed on tip of small dimensions (700 nm). Thus,

101

a platinum microelectrode was used instead. However, the tip of the Pt electrode
dimension was larger than the gold nanoelectrode. This was due to its material hardness,
that made it difficult to make a fine taper. The tip diameter of the Pt microelectrode
measured under a microscope in a cleanroom facility was determined to be 7.5 pm.

The total resistance of the Graphite sample and the entire circuit resistance were
measured using 2400 source meter and 2182 nanovoltmeter, and determined at 0.01224 (2

and 0.1 16 (2 respectively. The resistivity of the graphite was calculated by the following:

 

 

Across—sec iona
= 0.012249 (0-618cm)(0.633cm)
5.09cm

= 9.41x10-4Qcm

This value was compared to the standard graphite resistivity, from the material data sheet
[53], which is 0.006 Qcm. However, we measured a resistivity of approximately 0.001
Qcm. The difference between these values might come from the differences in
processing of the two graphite samples. The sample measured during this experiment
was denser than the standard graphite, and thus its resistivity is lower.

In the case of the graphite sample, the oscillator was again set at the l V rms and
150 Hz for output voltage and frequency respectively. Initially, a high frequency on the
order of 100k Hz, was used to achieve stable data, however, the high frequency
measurement was not as stable as the one at 150 Hz. The current measured by 1 Q
resistor is: 19.59 mA. And the circuit analysis for this sample is shown bellow:
The voltage drop, across the entire circuit, measured by the lock-in amplifier is: 3.62mV
Adding this to the voltage drop across the resistor box: 19.59mV + 3.62mV = 23.21 mV

102

The voltage across internal impedance of lock-in is: 1V (rms oscillating output voltage) -

23.21mV = 0.9768 V.

Thus the internal impedance is: M = 49.869 5 509.
19.59mA

Thus the internal resistance value was verified for this case as well.
The data taken by one probe scanning conductivity measurement system using Pt

microelectrode is tabulated (Table 6.4) and plotted (Figure 6.17) in the following:

103

or»-

Table 6.4 Data of scanned voltage difference of graphite using AC

technique and Pt microelectrode.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Probe spacing Voltage difference
0cm 236mV
Jcm 237mV
2cm 237mV
3cm 238mV
4cm 238mV
5cm 238mV
.6an 239mV
7cm 239mV
8cm 240mV
9cm 240mV
10cm 240mV
11cm 241mV
12cm 241mV
13cm 241mV
14cm 243mV
15cm 243mV
16cm 244mV
17cm 2A4mV
18cm 245mV
19cm 246mV
20cm 246mV
21cm 2A6mV
22cm 247mV
23cm 247mV
24cm 247mV
25cm 248mV
26cm 2A9mV

 

HM

 

 

2.5

2.45

2.4

Voltage difference (mV)

2.35

Graphite conductivity measurement
(Pt pulled pipette)

 

 

111111111111111111111111111111

0 0.5 1 1.5 2 2.5 3

Probe spacing (cm)

Figure 6.17 Plot of scanned voltage difference of graphite using AC

technique and Pt microelectrode.

The linear fit to the data in Figure 6.17 gives a slope of 0.048596 (mV/cm). Thus

resistivity is:

105

= RA = 93/1

1 [Al
(0.048596mV /cm)(.618cm)(.633cm)
(19.59mA)

= 9.7004 x10"4 Qcm

 

This result with the lock-in amplifier is very close to the resistivity value
determined by four-point probe method (p = 9.41x10“1 Qcm). The electrical
conductivity of the graphite sample is simply the reciprocal of resistivity: Ogmphnc = 1031

S/cm.

6.3.5 BizTe3

Although BizTe3 sample was measured by DC conductivity measurement
technique, the data analysis with its calculation was not ultimate. The voltage difference
should always increase strictly with a probe spacing; however, it decreased occasionally.
It is speculated that the decrease was due to the topographical problem such as cracks in
the sample and the surface oxidation. Although the sample surface was treated by
scraping using razor blade, the crack in sample is sometimes genuine and could not be
treated by the surface treatment. The expected range of the electrical conductivity of
BlzTC3 sample is about BOOS/cm to 1200 S/cm; however, with the DC analysis the
average of the conductivity was achieved at 702 S/cm.

In order to inspect the problem, the measurement of BizTe3 sample was revisited
with AC conductivity measurement technique using lock-in amplifier. First to verify the

samples topographical problem, the same kind of sample of B12T63, in which the crack

106

 

exists as the sample measured in DC technique, was used. The new good sample of
BizTe3 is then characterized consecutively to test the Pt microprobe capabilities.

In the case of the BlzTC3 sample, the lock-in amplifier was set to 0.5 V rrns, 1 kHz
as the oscillator output signal values. The amplitude of the voltage was decreased to
source the current approximately equal to the one used for DC measurement: IOmA. The
current of the system was calculated by measuring voltage difference across the 1 Q
resistance: I = 9.002 mA. The output frequency was chosen by running a frequency
sweep. The start frequency and the stop frequency was chosen as 1 Hz and 10 kHz
respectively. Thus, the frequency was swept linearly from 1 Hz to 10 kHz with the step
frequency of 100 Hz, and the optimizing frequency was observed by the stability plot.
Around 1 kHz, the input signal became stable, and thus, 1 kHz was selected as the output
frequency. Also, in order to see the micro range of the input voltage data, the input
sensitivity on the lock-in amplifier was decreased to 5 mV from 100 mV. The BizTe3
sample was run three times, and the data (Table 6.5) and plot (Figure 6.19) are in the

following.

107

Table 6.5 Data of scanned voltage difference of BizTe3 using AC

technique and pin.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Probe spacing Voltage difference Voltage difference Voltage difference
(cm) 1St run (mV) 2nd run (mV) 3rd run (mV)
0.000 cm 2.821 mV 4.110 mV 3.867 mV
0.025 cm 2.681 mV 4.140 mV 3.894 mV
0.050 cm 2.858 mV 4.128 mV 3.909 mV
0.075 cm 2.759 mV 4.116 mV 3.915 mV
0.100 cm 2.787 mV 4.116 mV 3.916 mV
0.125 cm 2.904 mV 4.112 mV 3.889 mV
0.150 cm 2.901 mV 4.114 mV 3.910 mV
0.175 cm 2.976 mV 4.112 mV 3.920 mV
0.200 cm 3.050 mV 4.120 mV 3.927 mV
0.225 cm 3.150 mV 4.122 mV 3.929 mV
0.250 cm 3.057 mV 4.120 mV 3.932 mV
0.275 cm 3.040 mV 4.130 mV 3.937 mV
0.300 cm 3.051 mV 4.130 mV 3.941 mV
0.325 cm 3.120 mV 4.132 mV 3.948 mV
0.350 cm 3.204 mV 3.940 mV 3.953 mV
0.375 cm 3.235 mV 4.143 mV 3.958 mV
0.400 cm 3.222 mV 4.150 mV 3.962 mV
0.425 cm 3.245 mV 4.155 mV 3.968 mV
0.450 cm 3.250 mV 4.151 mV 3.968 mV
0.475 cm 3.255 mV 4.153 mV 3.972 mV
0.500 cm 3.260 mV 4.154 mV 3.975 mV
0.525 cm 3.263 mV 4.146 mV 3.976 mV
0.550 cm 3.268 mV 4.136 mV 3.989 mV
0.575 cm 3.271 mV 4.143 mV 4.000 mV
0.600 cm 3.275 mV 4.136 mV 3.988 mV
0.625 cm 3.277 mV 4.141 mV 4.003 mV
0.650 cm 3.276 mV 4.174 mV 4.012 mV
0.675 cm 3.277 mV 4.180 mV 3.821 mV
0.700 cm 3.282 mV 4.185 mV 4.025 mV

 

 

In the case of the 2nd run data, the voltage difference decreased abruptly when the
probe spacing moved from 0.325 cm to 0.350 cm. The probe was again brought back to
the l = 0.325 cm location to see if this is the sudden offset or the topographical problem.
The voltage increase was seen at 0.325 cm, and thus it proves that the abrupt decrease

was due to cracks or surface contaminants on the sample. Also those three data sets were

108

scanned different location in width (Figure 6.18), and thus the data jump at different I for

each run, supporting the hypothesis that surface cracks exist.

 

scanningJocation of lst nm
scanning location of 2nd run
# scanning location of 3rd run

 

 

 

 

Bi2Te3 sample

 

 

 

Figure 6.18 Scanning location of each run

 

 

 

 

 

5 I I I l I I I I I I I I I I I l I I I
. —y=3.175+0.16018x R=0.93896 .
i — — y=4.1106+0.076827x R: 0.81311 :
4'5 1' ~--- y=3.8967+0.13797x R: 0.64444 '1
9 " I
g . 2ndrun DUDBDEDEBBBUBD—BEDUBGBDEUUQDD .
o .. . AA - -A-AAA~AA-AM‘AA
E [ 3rd run AKA AM A j
0
7»: - 1
D p I
8. 3.5 - _
S " II
6
> ' finnnnnm‘coe‘a "
. GUV VVVVVVV .
I O O i
O 000 .1
3 i- 1st run 000 ,
l- 0 000 It
1- O -1
25 h P I I I I I I I L I I I I I I I I I I ‘
-0.2 0 0.2 0.4 0.6 0.8
probe spacing (cm)

Figure 6.19 Plot of scanned voltage difference of BizTe3 using AC
technique and pin.

109

Since slope of the plot is AV/ Al

[Al
0 = —
AVA
__’___
slope x A

The dimensions of this BizTe3 sample are: width = 3.41 mm, thickness = 2.25 mm, and

length = 9.02 mm.

9.002mA
(0. l6018mV/cm)(0.341cmx0.225cm)

a = 9.002mA
2 (0.07683mV/cm)(0.341cmx0.225cm)

a? : 9.002mA
' (0.13797mV /cm)(0.341cm><0.225cm)
a _ a, + 0'2 + 03

average —

= 732S / cm

 

 

=1527S/cm

= 8505 / cm

 

=1036S/cm

 

Conductivity measurement on the 1St run clearly exhibited noise over the first 3
mm of the sample scanned. To investigate the influence of contact resistance, at the
correct leads, on the measurements, the input mode was changed from single node A
measurement (A-Ground) to the differential measurement (A-B). In this way, the input
voltage can be measured strictly between node A and B. The node A was connected to
the V+ scanning probe as it used to be; however, the input node B was connected to the 1-
side of the sample. A copper wire (diameter = 50 um ) was silver pasted to 1- side of the
sample, and it was connect to the BNC cable for input node B (Figure 6.20). Using
differential measurement the voltage difference across the 1- copper wiring was neglected.
The input impedance of the lock-in amplifier is 1M 9 and thus the resistance of the
copper wire (diameter = 50 um) connected to BNC cable was neglected. In this way, the

pure voltage difference of the BizTe3 sample can be measured. In this new configuration

110

the voltage difference did not decrease as much as we expected, however, the stability on
the datasets of 2"Id and 3rd run was tremendously improved after the change to the

differential measurement on the lock-in amplifier.

 

 

 

 

 

 

input node A
in the lock-in lifier
silver paste
3* ‘*
”L
input node B
in the lock-in amplifier

(acopperwireofSOrmndiameterisconnected
to the BNC cable connected to input B)

Figure 6.20 Setup of the differential voltage measurement.

0 New BizTe3 sample

After verifying the topographical problem of some BlzTC3 sample, a new BizTe3 was
acquired for the next measurement. Again the lock in amplifier was set to 0.5 V rms, 1
kHz as the oscillator output signal values. The current of the system was measured at
9.76 mA. In order to decrease the contact resistance of the silver paste, the small amount
of the silver paste was applied under the microscope to make a connection of 1+ and I-.
And also the copper wire is now pasted on the top surface of the BizTe3 sample (Figure

6.21).

111

input node A
in the lock-in lifier

 

_ \
Jr- 1 53L,

 

 

 

 

 

input node B
in the lock-in amplifier

Figure 6.21 The set up of differential voltage measurement connecting
node B on the top surface.

For this new sample, the ends of the sample were nickel plated before making the silver
paste contacts for the 1+ and 1- contacts. Thus the silver paste is pasted on to the thin
layer of the nickel layer for the connection of I- and 1+.

With these efforts, the voltage difference on the initial probe location decreased to
approximately 42 0V. The scanned voltage of this new BlzTC3 sample is tabulated (Table

6.6) and plotted (Figure 6.22).

112

Table 6.6 Data of scanned voltage difference of new BizTe3 using AC

technique and pin probe.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Probe spacing Voltage difference Voltage difference Voltage difference
(lSt run) (2nd run) (3rd run)
0.000 cm 0.042 mV 0.048 mV 0.049 mV
0.025 cm 0.045 mV 0.051 mV 0.051 mV
0.050 cm 0.048 mV 0.053 mV 0.053 mV
0.750 cm 0.049 mV 0.055 mV 0.055 mV
0.100 cm 0.052 mV 0.058 mV 0.058 mV
0.125 cm 0.055 mV 0.060 mV 0.060 mV
0.150 cm 0.058 mV 0.062 mV 0.063 mV
0.175 cm 0.059 mV 0.064 mV 0.065 mV
0.200 cm 0.061 mV 0.068 mV 0.068 mV
0.225 cm 0.066 mV 0.070 mV 0.069 mV
0.250 cm 0.070 mV 0.072 mV 0.073 mV
0.275 cm 0.072 mV 0.075 mV 0.075 mV
0.300 cm 0.073 mV
0.325 cm 0.074 mV

0.08 _....,. .44.... ....... . ,. ... '.

_ y= 0. 042057 + 0.10338x a: 0. 99479 2"d W" 1

0.075 }—---y= 0..048064+0096504x R=0.99861 - . O -

:---- y= 0..046015+0095804x R=0.99842 ,. :8 3

0.07 _- _.

E r a

8 0.065 E 1

9 E :

g 0.06 _ 1

'O I- d

g, I 2

g 0.055 _- 1

0 1- -1

> . .

0.05 _- 1

0.045 - J

0'04 :1 an l IIllllIl IAIIIILIIIIAJJLLJ:

-0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
probe spacing (cm)

Figure 6.22 Plot of scanned voltage difference of new BizTe3 using AC

technique and pin probe.

113

 

The voltage difference linearly increased with slopes indicated in Figure 6.22. The
dimensions of the BizTC3 were: w = 4.45 mm, t = 2.31 mm, and L = 4.45 mm. The

electrical conductivity was then calculated in each run as follows.

(9.76m)

 

 

 

 

a, = = 9185 /cm
(0.10338mV/cm)(0.445cm)(0.231cm)
0'2 = (9'76”) = 984S/cm
(0.096504mV /cm)(0.445cm)(0.231cm)
0'3 = (9.76mA) = 9915 /cm
(0.095804mV /cm)(0.445cm)(0.231cm)
M... = 918+9:4+991: 9648/0"

As mentioned earlier the range of the electrical conductivity of the B12T63 sample is about
800S/cm to 1200 S/cm, and thus the result achieved here met the criteria successfully.
Here in order to estimate the contact resistance from silver paste, the node B was again
brought back to the 1- side of the BizTe3 sample. The scanned voltage including contact

resistant is tabulated (Table 6.7) and plotted (Figure 6.23) in the following.

114

Table 6.7 Data of scanned voltage difference of new BlzTC3 using AC
technique and pin probe for contact resistance analysis.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Probe sgacing (cm) Voltage difference (mV)
0.000 cm 0.013 mV
0.025 cm 0.042 mV
0.050 cm 0.044 rnV
0.075 cm 0.046 mV
0.100 cm 0.048 mV
0.125 cm 0.050 mV
0.150 cm 0.052 mV
0.175 cm 0.054 mV
0.200 cm 0.056 mV
0.225 cm 0.059 mV
0.250 cm 0.061 mV
0.275 cm 0.063 mV
0.300 cm 0.065 mV
0.325 cm 0.068 mV
0.350 cm 0.070 mV
0.375 cm 0.073 mV
0.400 cm 0.075 mV
0.425 cm 0.077 mV
0.450 cm 0.080 mV
0.475 cm 0.082 mV
0.500 cm 0.098 rnV

 

 

The contact resistance of the silver paste is calculated as follows.

__ (o.042-o.013)»:v
16:233." _ 9.76m

_ (0.098—0.082)mV
contactR- 976mA

right side

21.23 m!)

 

=1.64mQ

 

115

 

 

 

 

 

0.1 I f I I I I I I r If III r I T ii I l I I I I r I 1 I I I I I I

r """ y = 0.041137 + 0.089684x R: 0.99918 ‘

L .

0.08 " ..

9 I

E. - «

8 0.06 F' -

C + .
9

:33 ’ -

'5 . u

8 0.04 - -

9 _ 4
B

> p I1

. 1

0.02 P .1

E 4

0 I I I I I l I I I I I I I I I I I I J 1 I I I I I I I I I I I I J J
-0.1 0 0.1 0.2 0.3 0.4 0.5 0.6

probe spacing (cm)

Figure 6.23 Plot of scanned voltage difference of new BizTe3 using AC
technique and pin probe for contact resistance analysis.

To see the voltage difference due to the contact resistance from the silver paste, the probe
was placed at the sides of the sample where the silver paste was coated. Thus the rapid
increase of voltage difference at the both ends (l=0.025 cm and l=0.500cm) was due to
the contact resistance created between the sample and the silver paste.

Only the linear region of the scanned voltage fit to determine the slope of 0.089684
mV/cm, thus the electrical conductivity was calculated as follows:

0' = (9.76mA) = 1059S / cm

(0.08968mV /cm)(o.445cm)(o.231cm)

 

Now, after the verification of the good B12T63 sample including the setting up of
the AC conductivity measurement system, the Pt microprobe (7.5 urn diameter) was

employed to the system to explore the feasibility of it. The scanned voltage difference

116

 

measured using Pt microprobe is tabulated (Table 6.8) and plotted (Figure 6.24) in the

following.

Table 6.8 Data of scanned voltage difference of new BizTe3 using AC
technique and Pt microprobe.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Probe spacing (cm) Voltage difference (mV)
0.000 cm 0.050 mV
0.025 cm 0.051 mV
0.050 cm 0.052 mV
0.750 cm 0.055 mV
0.100 cm 0.057 mV
0.125 cm 0.059 mV
0.150 cm 0.061 mV
0.175 cm 0.063 mV
0.200 cm 0.064 mV
0.225 cm 0.065 mV
0.250 cm 0.069 mV
0.275 cm 0.071 mV
0.300 cm 0.075 mV
0.325 cm 0.075 mV
0.350 cm 0.079 mV
0.375 cm 0.082 mV
0.400 cm 0.084 mV
0.425 cm 0.086 mV
0.450 cm 0.099 mV

 

 

The sudden voltage increase at the location 0.45 cm is due to the contact resistance of

nickel plating estimated as 922 11.0.

_ 0.095(mV) -0.086(mV)
$333122... 9.76(mA)
= 0.9221110

 

117

The Nickel plating of the I+ side was observed little thicker than the 1- side under the
microscope, and thus the Pt microprobe could detect the contact resistance of the nickel
plating here. The Pt microprobe scanned strictly on the top surface of the Bl2T63 sample,

so this voltage increase due to the contact resistance is purely from Nickel plating.

Bi2Te3 AC conductlvity measurement using
Pt microelectrode

 

 

 

 

 

0.1 - I I I I l I I I r l I I I r I fl I I I I I I I I l I I I I
- ----- y = 0.048099 + 0.086852x Fl: 0.99517 1
C 4
0.09 *- -i
+ 1
9 I. .1
E, 0.08 _ ‘
a) ' .
O I- d
c
a: - .
5 0.07 - ..
5:5 .. .
'O . .
m p I
a:
e - -
B 0.06 " '-
> " .1
0.05 L- _
0 04 h I I I I I I J I I I I I I I I l I I I I I I I I I I I I I
-0.1 0 0.1 0.2 0.3 0.4 0.5

probe spacing (cm)

Figure 6.24 Plot of scanned voltage difference of new B12T€3 using AC
technique and Pt microprobe.

118

The conductivity of the BizTe3 sample was measured using Pt microprobe and calculated
as follows:

a _—. (9'76“) = 10935 /cm

(0.086852mV/cm)(0.445cm)(0.231cm)

 

This is a reasonable result in good agreement with our previous measurements and

published values for B12T63.

6.4 Data acquisition Instrumentation

This section describes the specification of the instruments used to acquire the electrical

conductivity measurement.

6.4.1 Keithley Model 2400 SourceMeter®

The Model 2400 SourceMeter combines a precise, low-noise, highly stable DC power
supply with a low noise, highly repeatable, high-impedance multi-meter which
incorporates the following features:
0 0.012% basic measurement accuracy, 0.035% basic source accuracy with 5-1/2
digit resolution. 1000 readings/second at 4-1/2 digits via GPIB.
0 Voltage source/measure range: j: luV ~ 32 200V DC.
0 Current source/measure range: 1- lOpA ~ i 1A. Concurrent measurements of all
three functions over the remote interface.
0 Source-measure sweep capabilities (linear and logarithmic staircase sweeps).
0 22W, 4-quadrant source and sink operation.
0 Trigger-link interface to Keithley series 7000 switching hardware.

0 IEEE-488 and RS 232 interfaces.

119

This model 2400 SourceMeter was used to source current to the sample measured for

electrical conductivity.

6.4.2 Model 2182 Keithley Nanovoltmeter

Dual channels for measuring voltage and temperature.
Synchronization to line
Direct reading of ratio

Low noise at high speeds

6.4.3 EG&G 7265 DSP (Digital Signal Processor) Lock-In Amplifier

The EG&G lock-in amplifier has been employed to perform AC electrical conductivity

measurement for the noise improvement, with the following features:

Oscillator frequency range: 0.001 Hz ~ 250 kHz with absolute accuracy 225 ppm
i 30 qu.

Amplitude range: 1 uV~ 5 V with 1 1.1V ~ 1.25 mV resolution.

Signal channel voltage sensitivity: 2 nV ~ 1 V full-scale

Current input mode sensitivities: 2 fA to uA full-scale

Line frequency rejection filter

Dual phase demodulator with X-Y and R-0 outputs

Very low phase noise of < 0.0001° rms.

Output time constant: 10 us ~ 100 ks.

Output oscillator impedance: 50 9

Signal reference mode AC gain limit: 90 dB

120

 

o Auto-phase: In an Auto-Phase operation the value of the signal phase is computed
and an appropriate phase—shift is then introduced into the reference channel. So

the measured signal phase shift is zero to input signal.

121

7. CONCLUSION AND SUGGESTIONS FOR

FUTURE WORK

7 .1 One point probe scanning conductivity measurement system.

A new scanning probe electrical conductivity measurement system was designed,
implemented, and tested. The system is based on nanoprobes with tip dimensions of 300
nm ~ that were fabricated using a C02 laser based micropipette puller. Pulled
nanoelectrodes containing different types of metals (e.g. gold and Pt) were tested
successfully in the scanning probe electrical conductivity measurement system. The
measurement system was tested for both ac and dc electrical conductivity measurements.
The electrical conductivity of the semiconductor compound BizTe3, a copper strip, and a
graphite sample were measured by the system. With this electrode probe, the force
created while making contact is significantly reduced. Thus, it can be used for extremely

fragile samples such as small single crystals of new materials.

7 .2 Suggestions for Future work

7.2.1 Heat treatment

For one application of the pulled pipette, a quartz pulled pipette without any metal
filament was sputter coated with gold on the outside surface to make a nanoelectrode. It

was found that with this nanoelectrode it was difficult to make good electrical contact to

122

‘r—

the sample. It is speculated that the difficulty in making contact is due to the several
reasons:
0 The sample surface has problems caused by topographical variations or
oxidation.
0 The gold coating did not adhere well to the quartz, and was removed at the point

of contact.

In order to mitigate the second problem, the nanoelectrode was annealed at the
appropriate temperature to improve adhesion. Another alternative is to introduce an
intermediate layer such as chromium between the glass and the gold material, and
followed this with the heat treatment. This requires a good thin film deposition system

including multiple shutters for coating different metals consecutively.

7. 2.2 Study of a compound semiconductor with the domain including doping profile

Often new compound semiconductor samples contain domains in which the
electrical conductivity is different can vary. It is common that these domains will be on
the order of 10-100 um in dimension. Some of the domains have doping profiles /
striation within the domain with 40 nm spacing. In these samples it is speculated that the
spontaneous formation of these striations is key to improving the material properties of
interest for thermoelectric materials. In order to characterize such domains, probes with
tip dimensions smaller than 10 mm are essential. The nanoelectrodes fabricated in our lab
have tip dimensions less than 300 nm, and thus are well suited to studying the

characteristics of these domains and gaining insight into the further development of the

123

thermoelectric materials here at MSU. The nanoelectrode was successfully fabricated and
tested with different semiconductor compounds and metals. This probe could be used to

provide information about the electrical characteristics of each domain including their

doping profiles.

Domains on the surfaces

 

 

 

 

 

 

 

 

 

Figure 7.1 Doping profile/ striation

7.2.3 Pulled pipette as a thermocouple

Using P-2000 micropipette puller, thermocouples can be designed very easily.
First, pull the micropipette with metal, and then the outside of the probe can be coated

with another type of metal. This would allow for a localized temperature measurement

124

 

probe that could simultaneously be used for measuring the voltage at the point of contact.
With the addition of a separate localized heating source, then thermoelectric power
measurements could be made on a local scale. With such probes, doping profiles could

be imaged by scanning over the surface of a sample.

125

-h

APPENDIX

126

~.";

' ufi

 

o The pulling program code on P-2000 micropipette puller

8. APPENDIX

 

 

 

 

Selection A) Gold wire ((1 = 50pm), Borosilicate glass (o.d.=1.0mm, i.d.= 0.5mm)
Heat Filament Velocity Delay Pull Comment
Step 1 400 4 50 200 20 2&1118
Step 2 355 4 40 0 65 Final pull

 

 

 

 

 

 

 

Selection B) Platinum wire (d = 50pm) Quartz glass (o.d.=1.0mm, i.d.= 0.5mm)

 

 

 

 

 

 

Heat Filament Velocity Delay Pull Comment
Step 1 625 8 50 200 65 2 pulls
Step 2 645 2 80 0 75 Final pull

 

 

 

 

 

127

 

 

BIBLIOGRAPHY

128

10.

11.

12.

13.

14.

15.

9. BIBLIOGRAPHY

http://www.melcor.com/history.htm
D.M. Rowe, CRC Handbook of Therrnoelectrics, CRC Press, New York, 1994.
http://www.sutter.com/

Munoz, J .L. and Coles, J. “Quartz micropipettes for intracellular voltage micro-
electrodes and ion selective microelectrodes,” Journal of Neuroscience Methods,
22:57-64, 1987.

EM. Smits, Bell Sys. Tech. J., May (1958).

P. Boggild and TM. Hansen, “Scanning Nanoscale Multiprobes for Conductivity
Measurements”, Review of Scientific Instruments, Vol. 71, No. 7, pp. 2781-2783,
July 2000.

http://mitghmr.spd.louisville.edu/sops/sop45.html

S.M.Sze, VLSI Technology, 2“d edition, New York: pp 32, 1988.
http://microlab.berkeley.edu/~ee 143/Four-Point_Probe/

S.M.Sze, VLSI Technology, 2“d edition, New York: pp300, 1988.

Bhagwan D. Agarwal, flaIysis and Performance of Fiber Composites. Wiley, New
York, 1980.

Arges, K. Pete., Mechanics of Materials, New York, McGraw-Hill,l963.

Best Charles L., Analytical Mechanics for Engineers Statics, Scranton, International
Textbook Co., 1965.

G. Fish, “Ultrafast response micropipette-based submicrometer thermocouple”, Rev.
Sci. Instrum, 66 (5), May 1995.

Rimma Dekhter, A.Lewis, “Investigating material and functional properties of static
random access memories using cantilevered glass multiple-wire force-sensing
thermal probes”, Applied Physics Letters, Vol. 77, No. 26, pp 4425-4427, December
2000.

129

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

A. Lewis, M. Isaacson, A. Hartoonian, and A. Murray, “Scanning optical spectral
microscopy with 500 A spatial resolution”, Biophys. J., p.405a, 1983.

http://www.jasco.co.uk/nearfield.htm

Alina Strinkovski, “Chemical Application of Near-Field Scanning Optical
Microsc0py: Surface and Near-Surface Chemical Imaging with Conventional Near-
Field Optical Probes and Externally Illuminated Chemically Active Ion Sensors”,
Israel Journal of Chemistry, Vol. 41, pp.129-137, 2001.

Hartoonian, A., Shambrot, E., Radko, A., Lieberman, K., Ezekiel, S., Veinger, D.,
Yampolski, 6., Proc. IEEE, 88, pg. 1471, 2000.

Shmuel Shalom, “A micropipette force probe suitable for near-field scanning optical
microscopy”, Rev. Sci. Instrum, 63 (9), September 1992.

A. Lewis and K.Lieberman, Nature, 354, pg 214, 1991.

Shmuel Shalom, “A micropipette force probe suitable for near-field scanning optical
microscopy”, Rev. Sci. Instrum, 63 (9), September 1992.

G. Meyer and N. Amer, Appl. Phys. Lett., 53, pg. 1045, 1988.

Mun-Heon Hong, Ki Hyun Kim, “Scanning nanolithography using a material filled
nanopipette”, Applied Physics Letters, Vol. 77, No. 16, pg 2605, 2000.

AD. Muller and F. Muller, “Resonance frequency shift caused by the friction of a
drop of water in air: An approach to estimate shear forces in scanning probe
microscopies”, Applied Physics Letters, Vol. 78, No. 14, April 2001.

J. O. Tegenfeldt and L. Montelius, Appl. Phys.Lett., 66, pp 1068, 1995.
C. L. Petersen, F. Grey, M. Ono, Surf. Sci., 377-379, 676, 1997.

PL Mchen, N. G. Chopra, A. Zettl, A. Thess, R. E. Smalley, Science, Vol. 275,
pg.1922,1997.

Peter Boggild, Rancois Grey, Tue Hassenkam, Daniel R. Greve, and Thomas
Bjomholm, “Direct Measurement of the Microscale Conductivity of Conjugated
Polymer monolayers”, Advanced Materials, Vol. 12, no. 13, pp. 947-950, July 5‘h
2000.

130

30.

31.

32.

33.

34.

35.

36.
37.

38.

39.

40.

41.

42.

43.

44.

45.

CL. Petersen, T.M Hansen, P.Boggild, A. Boisen, O. Hansen, T. Hassenkam, F.
Grey, “Scanning microscopic four-point conductivity probes”, Sensors and Actuator
A. 96 (2002) 53-58.

P. Boggild, T.M. Hansen, O.Kuhn, and F. Grey, “Scanning nanoscale multiprobe for
conductivity measurements”, Review of Scientific Instruments, Vol. 71 , No.7, pp.
2781-2783, July 2000.

T.Fujii et al., J. Vac. Sci. Technol. B, Vol. 9, pg. 666, 1991.

T. Junno, S. B. Carlsson, H. Xu, L. Montelius, and L. Samuelson, Appl. Phys. Lett.,
Vol. 72, pg. 548, 1998.

M. Wendel, H. Lorenz, and J. P. Kotthaus, Appl. Phys. Lett., Vol. 67, pg. 3732, 1995.

M. Wendel, B. Irmer, J. Cortes, R. Kaiser, H. Lorenz, J .P. Kotthaus, A. Lorke, and E.
Williams, Superlattices Microstruc., Vol. 20, pg. 349, 1996.

Kim P and Lieber CM 1999 Science 386 2148
http://www.metals.com

G. Fish and O. Bouevitch, Ultrafast response micropipette-based submicrometer
thermocouple, Rev. Sci Instrum. Vol 66, No.5, May 1995.

Sutter Instrument Company, P-2000 micropipette puller operation manual

Stephen A. Campbell, The Science Jand Engineering of Microelectronic Fabrication,
Oxford University Press, New York, pg 70, 1996.

Stephen A. Campbell, The Science and Eggineering of Microelectronic Fabrication,
Oxford University press, New York, pg 72, 1996.

Etching profiles:
httpzl/www.ee.washington.edu/class/539/Lectures/lecture2/sld020.htm

Werner Kern, and Cheryl A. Deckert, Thin film processes- Chemical Etching,
Academic Press. Inc. pp 443, 1978.

Varian, “Product Catalog 2000”, Varian Vacuum Products. Inc. pg 579, 2000

David R. Lide, H_andbook of Chemistry and Physics 79th edition, CRC Press, pg. 12-
191-192,1999.

131

46.

47.

48.

49.

50.

51.

52.

53.

http://www.bridgetec.corrflanhall.html (June, 2001)

Keithley Instruments, “Model 2400 SourceMeter User’s Manual”, Fifth Printing,
June 1998.

K.N.Subramanian, and T.P.Hogan, “Internal Damage Accumulation Resulting from
Thermomechanical Fatigue in Lead-Free Solder Joints”, a proposal submitted to
National Science Foundation, p.7, November 2002.

Keithley Instruments, “Model 2182 Nanovoltmeter User’s Manual”, Second Printing,
June 1998.

PerkinElmer Instruments, “What is a lock-in Amplifier?”, Technical Note TN1000,
2000.

EG&G Instruments, “Model 7265 DSP Lock-in Amplifier Instruction Manual”,
1998.

David R. Lide, “CRC Handbook of Chemistry and Physics”, 79th edition 1998-1999.

http://www.matls.com (August, 2002)

132

   

niiiiiiiiinmigii11111111