llllllllllllllllllllllllllllllllllll'llllllllllllllll

1293 00793 7653

This is to certify that the
dissertation entitled
Technology & Electronic Properties
of CVD Diamond Film Microsensors
for Thermal Signals

presented by
Ashraf Masood

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in EE

% . flflth

Major professor

Date ”/7415 /f2-

MS U is an Affirmative Action/Equal Opportunity Institution 0—12771

LEBMRY

Rubicon State
University

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE DATE DUE DATE DUE

 

 

751i» 1008055;

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU lo An Affirmative Action/Equal Opportunlty Institution
cm

 

TECHNOLOGY AND ELECTRONIC PROPERTIES OF
DIAMOND FILM MICROSENSORS FOR
THERMAL SIGNALS

By

Ashraf Masood

A DISSERTATION

Submitted to
Michigan State University
in partial fulfilment of the requirement
for the degree of

DOCTOR OF PHILOSOPHY

Department of Electrical Engineering

I992

ABSTRACT

TECHNOLOGY AND ELECTRONIC PROPERTIES OF
DIAMOND FILM MICROSENSORS FOR
THERMAL SIGNALS

By

Ashraf Masood

Microsensors based on boron doped polycrystalline CVD diamond films for ther-
mal signals in the temperature range of 77 K to 1273 K were developed. New tech-
niques for boron doping, nucleation, patterning and metallization of diamond films
were developed. A test chip was designed and fabricated for electrical characterization

of diamond films and acquisition of thermal sensing parameters.

The diamond films were synthesized by hot filament chemical vapor deposition
(HFCVD). In situ boron doping was implemented by using an accurately measured
quantity (0.1-2.6 mg) of high purity amorphous boron powder through a specially
designed holder. The quality of the diamond films as determined from Raman spectros-
copy and SEM was n0t affected by boron doping in the useful range of boron concen-
tration (up to 1019 cm‘3). SIMS depth profile showed uniform boron concentration
throughout the thickness of the films. The measured resistivity and Hall concentration
of diamond films doped by using 0.1-2.6 mg of boron powder were in the range of

().3-64 Q—cm and 2x10‘5-9x1018 cm‘3, respectively.

A test microchip containing several devices was designed and fabricated. Three
lC compatible techniques for diamond film patterning based on selective deposition
and selective etching were developed. The diamond photoresist (DPR) patterning tech-
nique based on mixing diamond powder in photoresist is especially advantageous in

terms of selectivity, resolution and ease and flexibility of implementation. In case of

Ashraf Masood

metallization, the stringent requirement of providing ohmic contacts on diamond films,
and of stability and good adhesion on diamond as well as Si02 surfaces simultaneously

accomplished by a two layer structure of Pt(8000°A)/Ti(100°A).

Resistivity, Current-voltage measurement, carrier type, Hall concentration and
Hall mobility were directly measured. The resistivity and Hall mobility were in the
range of 0.3-64 O—cm and 2-48 cmZV’ls’l respectively. The dopant activation ener-
gies, as computed from the resistivity versus temperature curves (up to 300°C), were in
the range of 0.38-0.11 eV corresponding to Hall concentration in the range of
2x10'5-9x10‘80m‘3 and boron concentration in the range of 1017-1023cm'3. The

annealing behavior of doped diamond films was investigated for the first time.

The static temperature response of diamond film thermal sensors was measured
over the temperature range of 77-1273 K for the first time. The change in resistivity
with temperature was monotonic with a sensitivity given by temperature coefficient at
~ 0.07 K"1 at all doping levels over the entire temperature range. This is the largest
temperature range ever covered by a single semiconductor temperature measuring dev-
ice either in absolute or relative terms. The dynamic temperature response time con-
stant for the diamond film sensors on oxidized silicon was determined experimentally
to be about 25 us. This is smaller than any known solid state temperature sensor. As
a heat flux sensor, diamond film thermistor is shown qualitatively to outperform con-

ventional Pt sensors for shock tunnel applications.

to

my parents
and

my family

ACKNOWLEDGEMENT

All praises be to Allah (SWT), the creator and sustainer of the universe, for pro-

viding me with the ability and opportunity to complete this research.

I am grateful for the financial support from Government of Pakistan during my

studies. Without this support, this research effort would not have been possible.

I would like to express my sincere gratitude to my advisor, Dr. Mohammad
Aslam for his inspiration, guidance and professional support throughout my graduate
studies. His help. able guidance and understanding were key to my academic and pro-
fessional growth. I owe thanks to Dr. Donnie Reinhard. Dr. Jes Assmussen, Dr. T. J.
Pinnaviah, Dr. Ronald Fredricks and especially Dr. M. A. Tamor for their valuable
guidance throughout this research. I am also grateful to the staff of Ford Scientific
Research Laboratory especially Mr. T. J. Potter for the help and cooperation they
extended during my experiments.

I wish to express my sincere thanks to my family and my wife’s family for their
continued support, prayers, patience and encouragement. My wife Perveen and
daughters Naveeda and Sadaf deserve special thanks for their understanding and sup-

port throughout this research effort.

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................................... ix

LIST OF FIGURES ....................................................................................................................... x
Chapter I. INTRODUCTION
1.0 Research Motivation .................................................................................................. l
1.1 Research Approach ..................................................................................................... 3
1.2 Dissertation Organization ......................................................................................... 5

Chapter 2. BACKGROUND

‘ 2.0 Introduction ................................................................................................................... 7
2.1 Development of Sensors ........................................................................................... X
2.2 Sensors for Thermal Signals ................................................................................... X

2.2.1 Thermocouples ........................................................................................... 1()
2.2.2 Resistance Temperature Detectors ...................................................... 14
2.2.3 Other Devices ............................................................................................. 15
2.2.4 Thermistors .................................................................................................. 16
2.2.5 CVD Diamond Films as Temperature Sensors .............................. 19
2.3 Theoretical Analysis of Thermistors .................................................................... 21
2.4 Properties of Diamond .............................................................................................. 27
2.5 Chemical Vapor Deposition of Diamond ........................................................... 28
2.6 Hot-Filament CVD ..................................................................................................... 31
2.6.1 The Deposition System ........................................... 31
2.6.2 Effect of Gas Composition .................................................................... 32
2.6.3 Effect of C-H-O Ratio ............................................................................ 32
2.6.4 Effect of Oxygen ....................................................................................... 34
2.6.5 Effect of Substrate Temperature .......................................................... 34
2.7 Analysis Tools ............................................................................................................. 35
2.7.1 Raman Spectroscopy ................................................................................ 35

vi

vii

 

 

 

2.7.2 Scanning Electron Microscope ............................................................. 36
2.7.3 Secondary [on Mass Spectroscopy ..................................................... 37
2.7.4 Surface Profiler ..................................................... 38
2.8 Electronic properties .............................................................. 39
2.9 Film and Bulk Properties ......................................................................................... 42
2.10 Device Fabrication technologies - - - -. ............................... 45

Chapter 3. SEMICONDUCTING DIAMOND FILM SYNTHESIS AND
CHARACTERIZATION

 

3.0 Introduction ................................................................................................................... 48
3.1 The Deposition System ............................................................................................. 51
3.2 The Nucleation Methods .......................................................................................... 52
3.2.1 Ultrasonic Treatment ............................................................................... 52

3.2.2 Diamond-Photoresist Seeding ............................................................... 52

3.2.3 Comparative Analysis .............................................................................. 56

3.3 Diamond Deposition Processing Parameters ..................................................... 57
3.3.1 Reactant Gas Composition .................................................................... 59

3.3.2 System Pressure ......................................................................................... 59

3.3.3 Substrate Temperature ............................................................................. 59

3.3.4 System Operation ...................................................................................... 63

3.4 Boron Doping ......... ' ...................................................................................................... 64
3.5 Silicon Dioxide Etching ............................................................................................ 75
3.6 Summary -. .................................................................................. 77

Chapter 4. TEST MICRO-CHIP FABRICATION

4.0 Introduction ................................................................................................................... 78
4.1 Test Microchip Design .............................................................................................. 78
4.2 Diamond Film Patterning ......................................................................................... 82
4.2.1 Selective Etching ....................................................................................... 84

4.2.2 Selective Deposition ................................................................................ 88

4.2.2.1 Ultrasonic Treatment Patterning ........................................ 90

4.2.2.2 Diamond-Photoresist Patterning ......................................... 93

4.3 Metallization ................................................................................................................. 106
4.3.1 Problem Analysis ...................................................................................... 106

4.3.2 Experimental Results ............................................................................... 108

4.4 Fabrication of Heat Flux Sensors .......................................................................... l 19
4.5 Summary ........................................................................................................................ 120

Chapter 5. ELECTRICAL CHARACTERIZATION
5.0 Introduction ................................................................................................................... 123
5.1 Preliminary Annealing study .................................................................................. 124
5.2 Electrical Measurements ............ I30

viii

5.2.1 Overview ...................................................................................................... 130
5.2.2 Current-Voltage Measurements ............................................................ 130
5.2.3 Conduction Type Measurements ......................................................... 134
5.2.3 Resistivity Measurements ....................................................................... 136
5.2.4 Hall Measurements ................................................................................... 141
5.3 Computed Parameters ................................................................................................ 144
5.5 Summary ........................................................................................................................ 151

Chapter 6. CVD DIAMOND FILM SENSORS FOR THERMAL SIG-
NALS

6.0 Introduction ................................................................................................................... 152
6.1 Static Response of Temperature Sensors ............................................................ 152
6.2 High Temperature Effects ........................................................................................ 160
6.3 Dynamic Response of Temperature Sensors ..................................................... 162
6.4 On—chip Signal Processing ....................................................................................... 173
6.5 Summary ........................................................................................................................ 176
Chapter 7. CONCLUSION
7.1 Summary of Present work ....................................................................................... 178
7.2 Future Research ........................................................................................................... 181
Appendix A. DIAMOND FILM DEPOSITIONS ............................................................ 183

BIBLIOGRAPHY ........................................................................................................................... I91

LIST OF TABLES

2.1 Comparison of principle parameters of main temperature sensing devices
[12] .......................................................................................................... 11

2.2 Comparison of semiconductor properties of diamond with SiC, Si, and
GaAs [86]. ............................................................................................... 43

4.1 Sample preparation for optimization of diamond film thickness and sur-
face smoothness study. (Photoresist quantity = 42 ml) .................................... 96

5.1 The organization of electrical measurements for the characterization of
p-type diamond films. ................................................................................ 131

6.1 Thermal properties of Si, SiOz, Pt and Cu [160.47.169l. ......................... 170

ix

LIST OF FIGURES

2.1 Temperature response curves of typical (a) thermocouples, (b) RTDs, (c)
NTC and PTC thermistors. (the Pt resistor curve has been given for sensi-
tivity comparison only) [3] ................................................................................. 12

 

2. 2 Static temperature response curves of thermistors made of semiconduct-
ing (a) natural and synthetic crystalline diamonds [38] (b) CVD diamond
films with various doping levels [39]. .................................................... 20

 

2.3 (a) Electrical resistivity of a semiconductor similar to Ge as a function

of temperature on logarithmic scales [44] (b) logc resistance of two commer-

cial thermistors plotted against reciprocal of temperature in region III of

curve given in (a). ............................................................................................................. 23

 

2.4 (a) Influence of C-H-O ratio on the morphology of CVD deposited dia-
mond films [71] (b) Atomic C-H-O diamond deposition phase diagram with
the diamond growth domain [72]. ............................................................................................. 33

2.5 Comparison of thermal conductivity of (a) natural diamond with several
other solids as a function of temperature [116] (b) CVD diamond films
using different methane concentrations with natural diamonds, Ag and Cu

[88]. ...................................................................................................................................................... 40
3.1 A simplified schematic diagram of diamond HFCVD system. .............................. 50
3.2 Raman spectra of diamond films on oxidized silicon substrates seeded

by UT and DPR nucleation methods. ...................................................................................... 53
3.3 SEM micrographs of diamond films on oxidized silicon substrates

seeded by (a) UT and (b) DPR nucleation methods. ................... _ ..................................... 55
3.4 Raman spectra of diamond films deposited by using methane. acetylene.

acetone and methanol as carbon source. ................................................................................. 58
3.5 Raman spectra of the diamond films deposited with and without CO. .............. 60

3.6 Effect of system pressure on the (a) quality (sp3/sp2) ratio and (b)

xi
growth rate of the diamond films. ............................................................................................. 61

3.7 Effect of substrate temperature on the (a) quality (sp3/sp2) ratio and (b)
growth rate of the diamond films. . .................................................. 62

 

3.8 SIMS depth profile of a diamond film deposited on oxidized silicon .................. 66

3.9 Measured Hall concentration at room temperature against the quantity of
boron powder used for the doping of diamond films during deposition. .................... 67

3.10 Raman spectra of the diamond films deposited on oxidized silicon sub-
strate using various amounts of boron powder quantity. .................................................. 68

3.11 Raman spectra of the diamond films deposited on undoped diamond
films deposited on oxidized silicon substrate using various amounts of boron
powder quantity. .............................................................................................................................. 69

3.12 Comparison of the effect of boron powder quantity on the purity of
diamond films deposited on (a) oxidized silicon and (b) undoped diamond
films. .................................................................................................................................................... 70

3.13 Raman spectra of the diamond films deposited on various surface
orientation of single crystal diamond and oxidized silicon substrates. ........................ 72

3.14 A short range high resolution Raman spectrum of a doped diamond
film deposited on undoped diamond films. ............................................................................ 73

3.15 SEM micrographs of (a) undoped and (b) doped diamond films. ...................... 74

3.16 SEM micrographs of presumably heavily doped films on oxidized sili-
con. ...................................................................................................................... 75

 

3.17 The effect of pre-deposition Si02 thickness on the etch rate of 8103
during diamond deposition. ......................................................................................................... 76

4.1 Composite mask layout of test micro-chip. .................................................................. 80

4.2 Test chip masks for (a) diamond, (b) Interconnect metallization, (c) pas-
sivation/ gate insulator and (d) gate metallization. ............................................................. 81

4.3 The hierarchal layout of diamond film patterning techniques. .............................. 83

4.4 Undoped diamond films on. Silicon etched in RTP at 700°C for (a) 40

seconds, (b) 80 seconds, (c) 120 seconds and (e) 220 seconds. The cross-

section view of films shown in in (c) and (e) are shown in (d) and (t) respec-

tively. ’ - - ...................................................................................................................... 85

 

 

4.5 The plot of etched thickness against etching time of diamond film thick-

xii

ness in RTP at 700°C under an oxygen flow of 40 sccm. The increasing size
of error bar is indicative of highly non-unifonn etching. ................................................. 86

4. 6 The schematic diagram of diamond film patterning technique through
selective etching 1n RTP. - __ _ ................. 87

 

4.7 Diamond films patterned by selective etching at 700°C in RTP. The
patterns in (b) is close up view of (a). - - ........................................................ 89

 

4.8 Schematic diagrams of selective diamond nucleation technique by UT
using (a) Pre-treatment masking and (b) Post-treatment masking. ................................ 91

4.9 SEM micrographs of diamond films patterned by'UT method using pre-
treatment masking. The masks used were (a) Al film (1.1 11m) and (b) ther-
mal 8102 (0.88 um). ....................................................................................................................... 92

4.10 SEM micrographs of diamond films patterned by UT method using
post-treatment masking. The masks used were (a) Ni film (40()0°A), (b) CVD
S102 (4400°A). ................................................................................................................................. 94

4.11 SEM micrographs of diamond films deposited on substrates coated with
DPR suspension of (a) 41.7 mg, (b) 122.6 mg and (c) 142.6 mg of diamond
powderin 42 m1 of photo-resist. ................................................................................................. 98

4.12 The cross-sectional view of a diamond film grown after thickness uni-

formity and surface smoothness optimization study. .......................................................... 99
4.13 The effect of diamond film thickness on the surface roughness. ......................... 99
4.14 Schematic diagrams of DPR methods of diamond film patterning. .................. 100

4.15. SEM micrographs of the diamond films patterned by DPR method.

 

 

 

.................................................................................................................................................................. 101
4.16 SEM micrograph of the diamond film line parallel to the main pattern
shown in Fig. 4.15(b). The image was taken in the early part of growth.

_ .............................................................................. 102
4.17 SEM micrographs of the diamond film scratched with knife edge. The
picture shown in (b) is the higher resolution image of upper portion of pat-
tern shown in (a). -- ................................................................................ 104
4.18 Micrographs of test micro-chip after diamond deposition using DPR
patterning at various resolution. ................................................................................................. 105

4.19 Test micro-chip metallization. The patterns on was generated by using
shadow mask. The pictures in (b) and (c) are higher magnification images of
(b). The diamond pattern in (c) on the left is 160 um. ..................................................... 109

xiii

4. 20 (a) Diamond film pattern on test chip with as deposited metal Pt
(3000°A)/ Ti (500°A). (b) The top view of metallized diamond film surface
in (a). ............................................................. 1 10

 

 

4. 21 SEM micrographs of the metallization (Pt (3000°A)/ Ti (500°A) on test
chip heat treated at 1000°C. .............................................................................. l 12

 

4.22. Resistance of a metallized diamond resistor with time during annealing
at 600°C. ..................................................................................................... 1 14

 

4.23 SEM micrographs of metallization Pt (1.2 pm] Ti (100°A) on test

micro-chip after annealing at 1000°C. (a) The top view with diamond pattern

in left half, (b) stray diamond crystals and (c) a 60° view around a metal

contact on diamond. ..... _ - _ ............................... 1 16

 

4.24 SEM micrographs of metallization Pt (8000°A/ Ti (100°A) on test
micro-chip after annealing at 1000°C. (a) a 60° view around a metal contact
on diamond and (b) stray diamond crystal. .- ........................................... 1 17

 

4.25 Gold wire bonding on the test chip bonding pad composed of Pt
(8000°A/ Ti (100°A). ..................................................................................................................... 1 18

4.26 Mask set for diamond film heat flux sensors. (:1) diamond and (b)
metallization mask. ......................................................................................................................... 121

4.27 SEM micrographs of diamond film heat flux sensors on silicon rod. (b)
is a high magnification images of (a). ..................................................................................... 122

5.1 Schematic diagram of the experimental set up for resistivity measure-
ments at high temperature. ................................ ................. 125

 

5.2 The temperature response of resistivity of diamond films sensor annealed
at different temperatures. The annealing temperature is also indicated. ...................... 125

5.3 The reproducibility of temperature response of resistivity of unannealed
diamond temperature sensor over a temperature range of 300-573 K. ........................ 127

5.4 The reproducibility of temperature response of resistivity after annealing
in temperature range (a) 300-1273 K and (b) 77-300 K. ................................................. 128

5.5 The time response of resistivity of two diamond film resistors during
annealing at 1000°C. The measured Hall concentration for each sample is
also indicated. ................................................................................................................................... 129

5.6 Schematic diagram of the high temperature characterization system. ................ 132

5.7 l-V characteristics of a test chip at selected temperatures in the tempera- ‘
, ture range 77-300 K (a) before and (b) after annealing at 1000°C. ............................. 133

xiv
5.8 I- V characteristics of (a) discrete and (b) test chip samples over tem-
perature range 300- 573 K and 3011-1273 K respectively. ................................................. 135

5.9 The contact designation at Van der Pauw pattern on the test chip. .................... 137

5.10 The room temperature resistivity vs quantity of boron powder used for
the doping of diamond films during deposition. .................................................................. 139

5.11 The temperature response of resistivity of discrete samples over tem-
perature range of 300-573 K. The room temperature Hall concentration of
each sample is also indicated. ..................................................................................................... 139

5.12 The temperature response of resistivity of test chips before and after
annealing over temperature range of (21) 3110-1273 K (p=2x1015 cm 3 and (b)
77- 300 K (p=1. 7x10 °.) - . ................................................................ 140

 

5.13 The room temperature (a) resistivity and (b) Hall mobility vs measured
Hall concentration of the two types of samples. .................................................................. 143

5.14 The temperature response of (a) Hall concentration and (b) Hall mobili-
tyin the temperature range of 77-300 K. - ............... 145

 

5.15 The plots of computed boron (a) activation energy and (b) concentra-
tion against measured Hall concentration for two types of samples. ........................... 148

5.16 The plots of computed boron activation energy of test chips before
(assuming no compensation) and after annealing (assuming compensation)
against Hall concentration measured at room temperature. ............................................. 150

6.1 The temperature response of the (a) resistivity and (b) sensitivity in
terms of Ap/AT, of diamond film thermal sensors. The Hall concentration of
each sensor is also indicated. ...................................................................................................... 153

6.2 The plots of temperature coefficient (1 versus temperature for diamond
film thermal sensors. The Hall concentration of each sensor is also indicated.

 

.................................................................................................... 155
6.3 The sensitivity of diamond film thermal sensors in terms of W as

defined by Eq. 6.2 against measured Hall concentration. ................................................. 157
6. 4 The plots of fitted curves along with measured data for (a) temperature

and (b) reciprocal of temperature against 1n resistivity. (p=2xl0 ) ............................ 159
6.5 The temperature response of the resistance of (a) 1.1 pm- thick thermal

SiOz layer and (b) metal conducting lines (see text for experimental set up).
.................................................................................................................................................................. 161

6.6. Raman spectra of diamond film thermal sensors before and after heat
treatment at 1000°C for a total of 14 minutes in four cycles at a pressure of

XV

10'7 torr at two scan ranges. ...............................................................................................

6.7 SEM micrographs of diamond film thermal sensor after heat treatment at
1000°C for a total of 14 minutes in four cycles at a pressure of 10‘7 torr. (a)
normal view and (b) view at 60° angle. .............. - _.

........ 163

 

 

6. 8 SEM micrographs of diamond 7film thermal sensors after two minutes of

heat treatment at (a) 1249°C at 10‘7 torr and (b) 1000°C at 100 mtorr. .............

6.9 Storage scope trace of voltage once a current step of 100 11A was
applied to the diamond film thermal sensor with room temperature resistance

of 165.5 k9. ....................................................................................................

 

6.10 The steady state resistance of diamond film thermal sensor against
current flowing through it. The rise time of the voltage across the respective

sensor with the applied current step is also shown. .....................................................

6.11 The one-dimensional heat conduction model for thin film heat flux sen-

sors on thermal insulator [124]. ........

 

6.12 (a) The implementation of NAND logic with two p-channel depletion
mode diamond MESFETs and a diamond resistor. (b) The block diagram of
the proposed analog multiplexer circuit using all diamond based NAND gates

shown in (a). -- .........................................

 

CHAPTER I

 

INTRODUCTION

1.0 RESEARCH MOTIVATION

With increasing automation of machinery and process control, the demands on the
accuracy, stability, range and dynamic response of the temperature sensors is also
becoming more and more stringent. Many applications require the operation of sensors
under physically and chemically hazardous environment and present day sensors are

unable to meet these demand successfully.

Among various kinds of temperature senors used today, thermistors. thennocou-
ples (TC), resistive temperature detectors (RTD), and integrated circuit (IC) sensors are
most popular. They all have one or more limitations among limited temperature range,
non-linearity, self heating, poor sensitivity, large response time, poor accuracy and
incompatibility with harsh environment which make them unsuitable for modern appli-
cation. While most kinds of sensors have intrinsic drawbacks, thermistor’s shortcom-
ings primarily come from the properties of material it is made from. Hence, a material
with desirable properties which can withstand harsh physical and chemical environ-
ment can alleviate its vital drawbacks such as limited temperature range and instability

under harsh environment, and fulfill today’s tough demands. Diamond with remarkable

combination of physical, chemical and thermal properties appears to be an ideal
choice.

The discovery of semiconduction in natural diamond revealed its excellent elec-
tronic properties in addition to its already known unrivaled physical properties. This
created a new opportunity for utilizing semiconducting diamond in thermistor applica-
tions. The successful manufacture of synthetic diamond further enhanced the possibil-
ity of semiconducting diamond devices with extraordinary characteristics. In 1969, a
patent was granted to General Electric for producing synthetic diamond thermistors.
However, due to the cost and several technological problems including reproducibility
in semiconducting properties and sensing parameters with synthetic diamond thermistor

stayed in the laboratory for over three decades.

Recent developments in the process of growing thin diamond films by various
chemical vapor deposition (CVD) techniques has led to a renewed interest in develop-
ing "low cost" diamond devices. However, the potential applications of thin diamond
films greatly depends on their crystalline structure which in turn depends on the
growth conditions. The CVD diamond films on non-diamond substrates are typically
polycrystalline that cannot be used for most high performance active electronic dev-
ices. However, it seems to be an excellent material for passive electronic components
and microsensors with performance that is significantly better than their counterparts

made from other semiconductors.

In case of a temperature sensor, diamond’s wide energy band gap (5.45eV) can
extend its measuring range from cryogenic (77 K) to beyond gold point (1337 K),
never possible before by any single temperature measuring device. Its high thermal
conductivity (20 W/cm/K at 300K) and possibility to fabricate extremely small size
sensors (feature size = few microns) can lead to unprecedented high speed (response
time constant 2 few usec) operation. Its chemical inertness (except to oxidizing agents

at temperatures >650°C) and radiation immunity can permit highly stable operation

under extremely hostile environment. The relatively high resistivity as compared to
other semiconductors at useful doping levels (1013—10‘8cm’3) can lead to a simple two

wire resistance measurement which is impracticable for RTDs.

Since the physical and electronic properties of thin polycrystalline diamond films
may considerably differ from single crystal bulk diamond, it is very interesting to look
into the feasibility of their utilization for sensing thermal signals. It is advantageous
since film properties can be tailored by adjusting deposition parameters and doping
concentration. The challenge, however, is to (i) synthesize high quality diamond films
on non—diamond substrates, (ii) develop a reproducible, IC technology compatible and
economical fabrication technology including doping, patterning. metallization and pas-
sivation processes, fabrication processes, (iii) gain an understanding the electronic
behavior of the thin diamond films and their correlation to the nucleation and deposi-

tion conditions and (iv) characterize the diamond sensors for the intended application.

The object of this research was primarily to meet these challenges and develop
diamond sensors for thermal signals with properties superior to all kinds of existing

temperature sensors.

1.! RESEARCH APPROACH

The work performed for the achievement of the objective is highly experimental
in nature. It consists of four main subjects. The deposition of high quality semicon-
ducting diamond films, development of technology for fabrication of a test microchip,
characterization of semiconducting diamond films and characterization of microsensors

for thermal signals.

Semiconducting diamond films were synthesized using hot-filament chemical
vapor deposition (HFCVD) in a reactor designed and built at Ford Scientific Research
Laboratory Dearbom, MI. Since the diamond deposition at low pressure is thermo-

dynamically metastable, it is crucial to control associated parameters such as gas

composition, gas flow rates, chamber pressure, filament and substrate temperatures
appropriately to ascertain the proper diamond phase. The experimental parameter such
as reactant gases and their flow rates, chamber pressure, substrate and filament tern-
peratures were optimized to synthesize high quality diamond films in terms of their
purity, morphology, homogeneity, uniformity, and growth rate. The deposited dia-
mond films were characterized by scanning electron microscope (SEM), Raman spec-
troscopy and surface profiler. A new nucleation technique was developed to achieve
optimum grain size, film thickness uniformity and surface smoothness. Boron is used
as p-type dopant in diamond. A method, using solid boron powder source, compatible
with diamond CVD process, was developed to incorporate boron into diamond films in
a controlled quantity without affecting the quality of diamond itself. The suitability of
this doping method was evaluated experimentally through electrical measurements and
secondary ion-mass spectroscopy (SIMS).

A test chip containing several resistors of varying sizes and orientation, a rec-
tangular pattern for Hall measurements, and a couple of MOSFET structures was
designed with four mask process. Fabrication process concentrated on film patterning
and metallization. Three methods of diamond film patterning were developed. In case
of metallization, the stringent requirement of providing ohmic contacts on diamond
films, stability and good adhesion on diamond as well as 810: surfaces simultaneously
was extremely demanding. This task was accomplished by two layer structure .of Pt/T i
based on experimental results of this research and previous work on high temperature

metallization of silicon ICs using silicides.

The electrical characterization of diamond films as a semiconductor was carried
out over a temperature range of 77-1273 K. Resistivity, Current-voltage measurement,
carrier type, concentration and mobility were directly measured. For these measure-
ments, a special vacuum chamber equipped with a high performance heater and a com-

puter controlled data acquisition system was designed and built. The information

obtained through the measurements were analyzed to estimate the impurity concentra-
tion, dopant and hopping conduction activation energies.

The diamond film sensors for thermals signals were characterized for their static
and dynamic temperature response. The static temperature response over a temperature
range of 77-1273 K was done in two parts. The measurements below room tempera-
ture were made in a commercial system equipped with miniature cryogenic refrigera-
tor. Measurements above room temperature were made in the above mention high tem-
perature characterization system. The dynamic response was acquired by applying a
temperature step function using a programmable current source and recording the sen-
sor response on a storage scope in a single trace mode. The experimental results were

compared with the existing sensors for temperature and heat flux measurements.

1.2 DISSERTATION ORGANIZATION

This dissertation is divided into five subjects. These are (1) literature review, (2)
synthesis of semiconducting diamond films and their physical characterization, (3)
development of diamond electronic device technology and fabrication of test micro-
chip, (4) electrical characterization of diamond films and (5) properties of diamond

film sensors for thermal signals.

These subjects have been covered in seven chapters. After an introduction in first
chapter, all the review material is placed in chapter 2. Chapter 3 describes the syn-
thesis of semiconductor diamond film with an emphasis on optimization of experimen-
tal parameters, boron doping and nucleation methods. Chapter 4 contains the develop-
ment of device technology including, patterning, metallization, and passivation. The
design and fabrication of test microchip and heat flux sensors is also included in this
chapter. The measurement of fundamental semiconducting properties such as resis-
tivity, carrier concentration are described in chapter 5. An analysis of these properties

leading to evaluation of activation energies and impurity concentration is also included.

Chapter 6 explains the properties of diamond film temperature and heat flux sensors. In
addition to their static and dynamic temperature response, an on chip analog multi-
plexer circuit is suggested. Chapter 7 concludes this research with a summary of

important results and suggestions for future research.

CHMWERZ

 

BACKGROUND

2.0 INTRODUCTION

In this chapter an overview of two distinct subjects is presented: sensors for ther-
mal signals, and synthesis and electronic properties of CVD diamond films. After a
brief review of sensor development, at the beginning. a comparative description of
major temperature sensing devices namely thermocouples, RTDs, therrnistors and le
is given. Available literature on semiconducting CVD diamond films as temperature
sensors is also reviewed. Following this, a generic semiconductor is theoretically
analyzed from the perspective of being able to understand and predict its performance
as a temperature sensor in terms of range, sensitivity and linearity. In the second major
part of this chapter, after brief mention of physical properties of diamond, various
approachs of its synthesis with an emphasis on hot-filament CVD with respect to depo-
sition parameters are discussed. This follows the discussion on characterization, elec-
tronic properties and fabrication technologies of diamond devices in comparison with
other conventional semiconductors with a view point of looking into the feasibility of

their application in active electronic devices and temperature sensors.

2.1 DEVELOPMENT OF SENSORS

Sensors are devices used to acquire physical informations and transform them into
usable form. In general most sensors are mechanical or electronic devices which pro-
duce analog electrical signals as their output. Most electronic sensors are based on ele-
ments of which one of the parameters such as resistivity, dielectric constant, Hall vol-
tage, junction voltage or signal frequency displays a small change in response to one
or more measurands [1]. The sensor output is processed by a signal conditioning cir-
cuit which performs one or more of the functions including amplification, linearization.
calibration, compensation, comparison and switching. This signal is then converted into
digital output for display, control. recording or digital signal processing purposes..The
sensors and signal conditioning interface circuits are usually separate units. While
more and more sensors are being fabricated from conventional semiconductors, it is
possible to integrate them with interface circuit on a monolithic or hybrid chip [2].
Such device is usually referred to as a smart sensor. The development of planar sem-
iconductor technology has made this integration of sensor elements and interface signal
conditioning economically feasible. This approach will eventually lead to a
microprocessor-compatible output which can be applied to a bus organized data
acquisition system [3]. A fully integrated sensor can offer significant advantage over
others by employing additional features such as self testing, auto-calibration, and digi-
tal compensation to improve system reliability and performance while reducing overall

cost [2].

2.2 SENSORS FOR THERMAL SIGNALS

All sensors employed in any application of detection and measuring thermal sig-
nals are primarily the temperature sensors. Most applications other than the basic tem-
perature measurements, in effect, are special configurations of a single or multiple sen-

sor elements along with transformation of their output signal to represent other

physical measurands. For example, the thermal vacuum sensor such as the Pirani gauge
is basically an array of thermistor elements [4]. The device operation is based on the
principle of measuring heat transfer between two adjacent sensor elements by the gas
molecules. Therefore, while looking for a new sensor for any application involving
measurement of thermal signals, it is sufficient to explore its fundamental characteris-
tics as a basic temperature sensor. This knowledge can then be used for any other
specific thermal measurement application through well known transformation functions
and sensor structures. Hence, following review will contain the discussion about the

basic kinds of temperature sensors and their typical temperature responses only.

Temperature sensors are frequently used to measure measurands other than tern-
perature itself. For example, an air flow velocity sensor based on measurement of
difference of temperature between individual thermistor elements of an array is
reported [5]. Also transducers for radiation, pressure, position, level, vacuum, radia-
tion and chemical reaction can be constructed on the basis of absolute temperature or
temperature difference [4,6,7]. They are also frequently used in the measurement of
velocity, thermodilution flow, power, level control and thermal conductivity [3]. Once
these sensors are fabricated from semiconducting material, they can be integrated into
the electronic circuit on a monolithic or hybrid chip. This arrangement is frequently
applied for the temperature compensation of electronic circuits to offset the errors

caused by the undesired temperature changes [3,8,9].

Lately, there is a trend towards integrated multi-functional sensors [10]. It
includes temperature sensors for several physical variables along with optional inter-
face circuitry. For instance, an integrated multi-function sensor for temperature.
vacuum and flow velocity based on heat transfer at a constant chip temperature has
been fabricated [5]. Another sensor capable of measuring the wall shear stress (the
friction force that fluid flow exerts on the surface of an object) has been reported [11].

The sensor is comprised of an integrated thermopile which measures the relation

1()

between flow rate and the heat loss as well as the temperature difference on the chip in

the direction of flow to acquire the wall shear stress.

Temperature sensors can be divided into two main categories: self generating and
modulating sensors [4]. The self generating sensors are thermocouples which do not
require any source of power for their operation. Temperature sensors based on thermal
modulation of electric energy supplied by an auxiliary source are metallic and sem-
iconductor thermoresistors, diodes and transistors. More recently, integrated circuit (IC)
temperature sensors have also been developed. Since it is difficult to compare the pro-
perties of all these devices descriptively, their relative merits with respect to some

common parameters have been summarized in Table 2.1 [12].

2.2.1 Thermocouples

When the junction of two wires of dissimilar metals is heated, a voltage com-
monly known as Seebeck voltage develops across the open terminals of the wires.
This phenomenon is known as the thermo-electric effect. The fact that Seebeck voltage
is a direct function of junction temperature, a voltage reading by a voltmeter represents
the temperature of the junction. This is precisely the principle of thermocouple opera-
tion. Although all dissimilar metals exhibit this effect to a varying degree of sensi-
tivity, only certain metals/metal alloy combinations are suitable for use as thennocou-
ples. Fig. 2.1(a) shows the temperature response curves of commonly used standard

thermocouple types [3].

The output voltage and voltage variations with temperature of all thermocouples
are in terms of mV and uV/°C respectively. Such small voltages definitely need very
sensitive and precision instruments. The output voltage of thermocouples is somewhat
non-linear and, therefore, several curve fitted equations and look up tables have been
developed and in some cases implemented in hardware of the readout instruments. In

addition, the cold junction (where thermocouple wire is connected to dissimilar metal

II

Table 2.1 Comparison of principle parameters of main temperature sensing
devices [12] ,

O

 

Pt RTDs

 

Parameter

100 Q

2000 Q

Thermis-
tors

Thenno-
couples

ICs

 

=—
Temperature range

 

 

 

wide
-240C to
+650C

short
-75C to
+200C

short-
medium
-75C to
+260C

very wide
-240C to
+2300C

very short
—50 C to
+135C

 

Interchangeability

Excellent

Poor~
Fair

Good

Fair

 

 

rig term stabiilty‘“

 

 

 
 

 

Good ’ Poor

Poor to

??*7?‘é:‘:I-:;-~:11;??-EIS’?‘ Pa”

 

Accuracy

 

Med1um

High '

Medium

Medium

 

Repeatability

Excellent

Fair to
Good

Poor to
Fair

Good

 

Ha

Low

High

 

 

Medium

Medium

Medium to
Fast

 

 

Excellent

Good

 

 

Low

 

 

Point end
sensitivity

 
 

Good

 

Lead Effect

   

Low

 

 

Physrcatmc/ . .. ..

  

* ~ :Medwm {5

 

 

 

Medium

 

 

 

Medium

 

 

12

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 
     
 

 

100 k
/4
10 k
‘ \
g t 1: \
\
i c ‘F\
i \ Platinum reeiator
b--‘---"--‘ .x“--de--—x--—l
1m \A
/ \‘."‘
_/
10
0 500 1000 1500 2000 2500 30 50 70 00 110 130 150 170
Temperature, '6 Temperature, 'C
(a) -,
(C)
Ill)
Ro
a _
Nickel
7 1-
5 ..
Copper
5 1—
Platinum
4 _
3 —
Manganin lO-WC), used for
2 /non-temperatureaenaitive resistors
1 1 r 1 1 1 1 1 m
-200 0 200 400 000 8(1) IMO
m, 0 '- Temperature, °C
platinum Nickel

 

(b)

Figure 2.1 Temperature response curves of typical (a) thermocouples, (b) RTDs, (c)
NTC and PTC thermistors. (the Pt resistor curve has been given for sensitivity com-

parison only) [3]

l3

connectors of the voltmeter or display instrument) voltage and voltage drop across
finite resistance of the lead wires cause undesirable error in measurements and which
is required to be compensated for. Both manual and automatic instrumental compensa-
tions are available but it is never precise over the entire temperature range. Hence, an
error in measurement is usually large especially at high temperatures. For example. an
error of greater of 22°C or 0.75% of the actual temperature is specified for type K

thermocouple (which is 11:7.5°C at 1000°C) and it is a typical of any thermocouple.

Another frequent source of error is drift with time. Even though high temperature
thermocouples are made of refractory metal alloys, their properties undergo a per-
manent change during long term exposure to high temperatures. Most thermocouples
are not suitable for all environments especially for oxidizing, reactive and high vacuum
applications. Such effects, in addition to precision, degrade the repeatability of thermo-
couples. The response time, by virtue of their size, generally exceeds several seconds
which is too large for certain applications. Due to all these drawbacks, thermocouples
are unsuitable for high speed precision temperature measurements. Despite this fact,
their enormous temperature range i.e., -240°C to 2300°C, largest of all kinds of tern-
perature measurement devices makes them popular for most conventional applications
such as in furnaces, heaters, industrial plants and controlled environment chambers.
However, the inherent design prohibits them from taking advantage of planar process-
ing technology or integration with sensors of other measurands or signal conditioning
circuitry. Lately, a unique thermocouple made of narrow bandgap compound semicon—
ductor pair, i.e., BizTe3 and szTe3, has shown tremendous improvement in sensitivity
over classical thermocouples [6]. This sensor has successfully been used in the meas-

urement of laser radiation and super high frequency (SHF) powers.

14

2.2.2 Resistance Temperature Detectors

The resistance of most materials change with temperature. Hence, passing a
known constant current will cause a voltage drop which is proportional to the tempera-
ture. Two types of modulating transducers use this phenomenon for temperature meas-
urement namely resistance temperature detectors (RTDs) and thermistors. The
difference between the two is the base material they are made from. RTDs use metallic
sensing elements whereas thermistors are usually semiconductors. Due to their simpli-
city, these devices are increasingly replacing thermocouples in many applications since
they do not need special wires or cold junction compensation.

RTDs are made from several metals or metal alloys such as Ni, Ni-Fe, Cu and Pt
[12]. Platinum is usually preferred over the others due to its relatively high resistivity
and linearity over the temperature range as shown in Fig. 2.1(b) [3]. Moreover, it is
malleable and resists corrosion. Regardless of the base material, all RTDs are positive
temperature coefficient of resistance (PTC) devices. Typically RTDs consists of a
wound metal wire inside a glass or stainless steel package. To achieve a reasonably
moderate resistance, long sensing wire is wrapped around a bobbin. In spite of this, the
ultimate resistance is so low that special arrangements (e.g., bridge circuits) must be
made to offset the lead effect [12]. A more rugged design is thin film type. The metal
slurry is screened onto a ceramic substrate. The film is sealed after curing and laser

trimming for calibration [13,14].

The packaging considerations of RTDs are very important for their operation. At
elevated temperatures (>40()°C) small amounts of iron impurities can leach out the
conventional stainless steel package and corrode the platinum [12]. Using other sheaths
such as quartz drastically increases the response time. Above 550°C, the hydrogen
poisoning can cause problem. Platinum elements breath through the seal and absorb
the hydrogen atoms provided by the dissociation of ambient water molecules. This

causes the platinum resistance to increase. In spite of these problems, RTDs are very

15

popular for their impressive stability, accuracy and speed compared to other tempera-
ture sensors. They generally have quoted accuracy of about 0.1% with typical drift of
<0.1°C per year. RTDs are particularly well suited for temperature measurements in

the laboratory due to their precision, wide temperature range and fast response.

2.2.3 Other Devices

Junction semiconductor devices such as diodes and transistors are frequently used
for temperature sensing applications. However, junction voltage. instead of resistance.
is temperature sensitive variable. PN junction diodes find useful applications as ther—
mometers in cryogenic range. Si diodes, in particular, are very popular for tempera-
tures below 30 K due to their high sensitivity and linearity [15]. Recently, sophisti-
cated modification of transistor sensors is provided as a two terminal monolithic inter-
grated circuit (AD590) [16]. They supply an output current or voltage that is linearly
proportional to absolute temperature. Typical values are luA/K and lllmV/K. How-
ever, the concept of integration of temperature sensors and accompanying electronics is
fairly innovative and provides a motivation for developing diamond based IC sensors
with superior characteristics.

A number of temperature sensors based on non-conventional physical principles
have been developed. For example a fiber optic temperature sensor and a quartz tem-
perature sensor have been reported [17]. The former is a non-contact device which
detects the intensity of emission of a particular wavelength from a hot surface whereas
the later utilizes the temperature dependence of the natural resonance frequency of a
quartz crystal for measuring temperature. Nuclear quadropole resonance (NQR) ther-
mometry has developed to the point where temperatures in the range of 90 K to 398 K
can be measured with a resolution of ~2mK [14]. At least one report describes a ther-
mometry utilizing the rather complex temperature dependence of ferrites [18]. How-

ever, the present interest is limited to semiconducting resistance thennometry and this

16

brief description of junction and other devices is given for completeness only.

2.2.4 Thermistors

All semiconductors exhibit some degree of resistivity variation with temperature
[8]. However, for temperature measurement purposes, special semiconductor resistors
known as thermistors are fabricated which show large variations in their resistance in a
predictable way as temperature is changed. They may be 103 to 10° times more sensi-
tive than conventional Pt RTDs [19]. However, thermistors are less stable and operate
in limited temperature range due to their high non-linearity (see Fig. 2.l(c) for curves
of typical thermistors [3]). This non-linearity has practically prevented the manufac-
turers from standardizing their response curves to the same extent as those of RTDs.
They are mostly employed for less precise temperature control, and for switching in
the temperature range from -100 to 500°C [9]. They are usually made from compound
semiconductors in ceramic composition in which solid solutions of transition metal
oxides are used [8,20]. The first commercial thermistors were produced in 1930’s from
U02, CuO or SiC. The elemental semiconductor based thermistors are, in particular,
widely used for temperature compensation of electronic circuits. They are very popu-
lar for temperature measurements in the cryogenic range [15,21]. However, due to
inherent material and electronic properties, they are unsuitable for high temperature
measurements. Though there are thermistors with positive temperature coefficient
(PTC) of resistance, most of the thermistors have negative temperature coefficient

(NTC) of resistance.

The advantage of semiconductors is relatively higher resistivity than metals. This
allows the resistance of thermistors to be in the range of several k!) to few M!) which
in turn boosts the temperature sensitivity. Such large changes in resistance are easily
measureable using a simple two wire resistance measurements. The problems associ-

ated with lead effect, contact resistance and cross sensitivity from stress in the sensing

17

elements, as faced by RTDs, are significantly reduced [19]. The real advantage of
thermistors results from their inherent structural difference. Thermocouples and RTDs
can be constructed from only a limited number of materials, and shape and size are
restricted. Thermistors on the other hand can be fabricated from a variety of semicon-
ductors materials and composition can be modified through introduction of impurities
to meet specific application requirements. Moreover, considerable latitude is possible
in sensor size and shape.

Thermistors are available in a variety of shapes including rods, films, flakes,
disks, washers etc. But beads and chips are more popular for common applications par-
ticularly where space is limited or where minimal disturbance of the monitored
medium by the sensor is required [9,12]. Bead thermistors have been particularly
effective as sensors in gas chromatographs and thermal conductivity analyzers.
Thermistor chips are especially well suited for direct surface mounting in hybrid and
other circuits. One important application is an on board sensor for temperature corn-
pensated quartz crystal oscillators in microwave equipment. They are usually fired on
Pt, Ag, or Au wires for external connections, and are encapsulated in glass, epoxy, or
a ceramic sheath [20]. They are available in a wide range of tolerances from 1% to
20% of nominal resistance at some reference temperature (usually 25°C). Individual
sensing elements, due to variety of reasons, cover only a narrow range of temperature
[9]. However, in recent years, they are evolving most effectively to meet the stringent
demands for improved stability, miniaturization, lower cost, close tolerance and inter-
changeability [14]. The largest emphasis has been to reduce non-linearity of the sen-
sors either by making structural changes (through doping [22], combining PTC and
NTC elements [23]) or through signal conditioning circuits [9,24].

There have been several reports about thermistors with unusual materials and pro-

perties. In 1971, a thennister made of graphite powder mixed in melted mixture of

paraphine and polythene was reported to have an anamalously large PTC in a limited

18

temperature range (20°C to 85°C) [25]. The reported resistivity change of about six
orders of magnitude was monotonic but highly non-linear. Another thermistor used
polysilicon as base material [26]. Its temperature response has been described to be
fairly stable and linear on reciprocal temperature scale in the range of 30()-9()0°C. Its
applications as multifunction sensor have also been investigated [10]. Similar studies
have been carried out on SiC thermistors in a temperature range below 300°C [27,28].
Furthermore, several non-conventional semiconductors including Bi [29,30], ruthenium
oxide [31], compound ceramics [32], InSb(Mn) [33] and (Ba, Sr)TiO; [34] have also

been investigated for varying applications involving temperature measurements.

The dynamic temperature response of thermal sensors is a parameter important in
certain applications where measurement of fast temperature changes is required
[21,35,36]. An improvement in the dynamic response of a sensor requires, among other
parameters, a reduction in its thermal mass. Doing this decreases the upper limit of
temperature range the sensor can physically withstand. Hence, widening the tempera-
ture range and reducing the dynamic response are contradictory requirements for a sen-
sor design. For example, type K thermocouple with response time of 1.8 seconds can
go up to 668°C. To reach the specified limit of this thermocouple i.e., 1200°C, the
wire size has to be increased which increases its response time up to 110 seconds [14].
Similar effect prevails in RTDs. A 1000 Pt RTD has 1.5 seconds response time with
upper temperature limit of 365°C. The same unit manufactured to reach 800°C has the
response time of 54 seconds [14]. Though thermistors also show similar effect, but it is
much less than thermocouples and RTDs. Moreover, their intrinsic response time is so
short (fraction of a second) that manufacturers usually do not specify it [14]. Addition-
ally, the inherent flexibility in their design is also, some times, used to improve their

dynamic response significantly [35].

19

2.2.5 CVD Diamond Temperature Sensors

When first diamond based thermistors were made of natural semiconducting dia-
monds, their remarkable features were pointed out [37]: corrosion and abrasion resis-
tance, low specific heat and excellent thermal conductivity. Additionally they could
operate at temperatures > 600 K. Later on, a patent [38] was granted to G. E. for pro-
ducing synthetic diamond thermistors. These thermistors were nominally doped with
0.001-0.15% B during synthesis and sealed into a glass envelop. They were said to be
superior to natural diamonds, since they did not exhibit the resistivity increase with
temperature between 700 and 800 K as shown in Fig. 2.2(a) [38]. These sensors were.
however, never produced commercially due to high production cost and several tech-
nological problems especially reproducibility of size, shape and doping concentration
which resulted in high tolerance in resistance (220%) and sensitivity (215%) [8].

Recently, a diamond thermistor based on homoepitaxial CVD diamond films has
been demonstrated [39]. The sensing properties as shown in Fig. 2.2(b) have been
described to be comparable to synthetic diamond. However, its characteristics have

been studied within a limited temperature range (ZS-500°C).

With this supportive background [40], it is expected that CVD diamond films
which are typically polycrystalline and cannot be used for most high performance
active electronic devices, will prove to be an excellent temperature sensing material.
superior to that of any other material in terms of temperature range, speed, stability
and sensitivity. Its chemical inertness except to oxidizing agents at high temperatures
(>650°C) [41] and radiation immunity can permit highly stable operation under
extremely hostile environment. The relatively high resistivity as compared to other
semiconductors at useful doping levels (1013—1018cm"3) can lead to a simple two wire
resistance measurement which is impracticable for RTDs. Such a temperature sensor
will possibly be able to replace Ge resistance thermometer in cryogenic aerospace sys-

tems [21] and platinum RTD in shock wave tunnels [36] with less constraint on the

20

 

 

 

   
  
 
  
  

 

 

10'
g mater emcee
3 acumen custom
”a
10
u m .
firderireferen
nave-mere 'ene"
(21)
I'm/C
000 400 200 100 3g
.0 ‘0")-
Nen-deped fem
0-70001£.-0.eOeV1
IO'P
I-deeed durum;
'5 e-seootz.-o.soov1
1 10'-
i B-deped lilml200wm
known-024.111
10“-
$th
10'1-
1 l l l
1 2 3
moonlit-'1 (b)

Figure 2.2 Static temperature response curves of thermistors made of semiconducting '
(a) natural and synthetic crystalline diamonds [38] (b) CVD diamond films with vari-

ous doping levels [39].

21

sensor design to reduce dynamic response time [35].

2.3 THEORETICAL ANALYSIS OF THERMISTORS

The electronic transport phenomenon in polycrystalline semiconductors including
polycrystalline diamond is not well understood at this time. However, if sufficient
allowance for effects unique to polycrystalline material is given, then most of the qual—
itative analysis for crystalline material can, in general, be applied to polycrystalline
materials as a first approximation. The most important aspect is the understanding of
the temperature dependence of the resistivity p. While the general shape of the resis-
tance versus temperature curve for semiconductors can be predicted, quantitative
theoretical calculations are not accurate enough for them to be adopted directly for
temperature sensing, especially in case of polycrystalline material. Nevertheless, they
provide the fundamental basis for the design and improvement of sensor properties

such as sensitivity and range.

In an intrinsic semiconductor, as the temperature approaches absolute zero (-
273°C), all the electrons are bound in the valence band i.e., energy E<Ev. With rising
temperature, electrons are thermally excited in increasing number in to the conduction
band i.e., E>Ec, leaving behind electron vacancies or holes. Since the current carrier
concentration increases with temperature leading to reduction in resistivity, intrinsic

semiconductors always have a negative temperature coefficient (NTC) of resistivity.

If the difference between the Fermi level, EF and conduction band or valence
band edge is large compared with kT, the expressions for electron and hole densities in

a semiconductor are given respectively by [42]

EF-EC
= N ——- 2.3.1
11 C exp kT ( )
E E
P = Nv eXp -——F_ V (2.3.2)

kT

22

rolw
to|w

21tmckT

where NC = 2 ——;—- and NV = 2

ha.

21tmhkT
hz
and mh and me are the effective masses of the holes and electrons respectively and h is
Planck’s constant.
The controlled addition of impurities to a semiconductor can make a large change
in its resistivity, both in its absolute value and in its rate of change with temperature.

However, the following charge neutrality condition is always maintained.

DJ

n+NA-=p+ND+ (2H3)

where n, p, N A’, and ND+ are electron. hole, ionized acceptor, and ionized donor den-
sities respectively. The resistivity of any semiconducting material, whether doped or

not, is given by [42]
—1
p = [q ( n(T) 11,,(T) + p(T) up 0)] (2.3.4)

where tin and up are electron and hole mobilities respectively. The T in parenthesis
indicates that the mobilities and carrier concentration are strong function of tempera-
ture. In effect, this temperature dependence is the basis of resistivity change with tem-
perature for semiconductor temperature sensors. The mobility due to impurity scatter-
ing 1,1,1 dominates at low temperature. The mobility due to lattice scattering follows a
power law of the type 11L oc T"8 and dominates at higher temperatures where S = 0.5

to 2.8 depending on the semiconducting material [43].

The variation of resistivity with temperature of a doped semiconductor is shown
qualitatively in Fig. 2.3(a) [44]. Four regimes of temperature can clearly be identified,
one conduction process being dominant in each. In the high temperature range (I), the
semiconductor behave as intrinsic material and the conduction is dominated by the
thermally excited carriers i.e., ni, because ni>>NA,ND. In this case n = p = ni, Eq.

(2.3.1) and (2.3.2) give

23

 

  

—-——_—————q

 
   

——————_———————1

Resistivity llogarithmicl

 

 

|
|
l
l
l
I
l
l
l

 

|
l
l
1
Temperature T (Kl (logarithmic)

(a)

 

12)-

10'—

ink
m
I

 

 

o i l i I n l i I i I
2.6 2.8 3.0 3.2 3.4 3.6

in x 103
(b)

Figure 2.3 (a) Electrical resistivity of a semiconductor similar to Ge as a function of

 

temperature on logarithmic scales [44] (b) loge resistance of two commercial thermis-

tors plotted against reciprocal of temperature in region II of curve given in (a).

24

2
E,

P 7

3 ..
21:1:T] (mcmh)/‘ exp

 

 

 

l‘li : (np)l/2 = 2[

When substituting into Eq. 2.3.4, the pre-exponential factor becomes independent of
temperature if S=1.5 or a weak function of temperature if S¢l.5 and, hence, the
exponential function will dominate the temperature response of resistivity. Then. it can

be written as

 

Eg 236
P CXP 2kT (.-. )

In the extrinsic region (11), all the impurities are still fully ionized but 11, is much less
than the impurity concentration. Then the resistivity, dominated by the temperature
dependence of the mobility, corresponds to that of the majority carriers. The resistivity
is again determined by Eq. 2.3.4 with n and p replaced with ND and N A respectively.
In range III, the ionization of the majority impurity atoms ceases to be complete.

The density of occupied acceptor states in a P-type semiconductor is given by

N
N A- = A (2.3.7)
1 + 2exp [(EA — rap/1m

 

However, almost all the donor states will be ionized, hence. NDaZND. Substituting this
expression into the charge neutrality equation (2.3.3) and assuming 11 << N A’ for a p-
type semiconductor yields

NA
_ NI)
1 + 2exp [(13A — Epym

 

p = (2.3.8)

Solving this equation in terms of p (using the expression from Eq. (2.3.2)), one gets

E

Nvexp {-fi] (2.3.9)

P(P+Np) : i
(NA-ND‘P) 2

 

Where E, = E A — Ev, is the acceptor activation energy. This equation can be simplified

by applying assumptions based on the compensation of impurities. For a highly

25

compensated P-type semiconductor, it can be assumed that N A>Np>>P in this tempera-
ture zone (freeze-out region) so the above equation simplifies to

N —N 13,
Li] NV exp [——‘] (2.3.111)

p : 2ND kT

 

and if no compensation is present, the assumption ND<<p<NA will apply. Then Eq.

2.3.9 will give

1
-q if... i. a...)
P 2 v A P 2k

Similar expressions can be derived for electrons in n-type semiconductors. The resis-
tivity for p-type semiconductor can then be found using Eq. (2.3.4) where the hole

density p will be substituted from Eq. (2.3.10) or (2.3.11).
-1
p = [qltp(T)p] Z constant . exp(el/kT) (2.3.12)

Hence, the slope of In p curve on l/T scale should be almost linearly proportional to
81 which is equal to either Ea or Ea/2 depending upon the situation whether the sem-
iconductor is compensated or not. This temperature dependence is shown in Fig. 2.3(b)
[9].

In region IV, the thermal energy is too low to excite electrons into the conduction
band and they are restricted to the impurity atoms. The conduction is believed to take
place by charge carriers jumping directly from one impurity atom to the next known as
hopping conduction. The activation energy 83 required for hopping conduction is, in
general, much smaller than normal activation energy E3 of the dopant. The resistivity

can approximately be written as [156],
p: p0 eF_r/kT (2.113)

where pO depend upon the material and density of donors and acceptor atoms.

26

Of particular interest in temperature sensing is the region 111, and to some extent
region IV, which occur in the temperature range (311-200 K) where resistance ther-
mometry is mostly employed. It is because of the linearity of the ln p curves on
reciprocal of temperature scale (see Fig. 2.3(b) [9] in region [11). This linearity is typi-
cal, contributed by Eq. (2.3.12), but other empirical equations can also be made to fit
the experimental curves. One equation uses a temperature-dependent pre-exponential

factor [9], and fits the experimental data more closely than Eq. (2.3.12). It is given by
P = Cf“ cl”T (2.3.14)

where C2 and c are constants and obtained by curve fitting to the experimental data.
The constant C2 is small and may be positive or negative. Neglecting the temperature
dependence of the pre-exponential factor, the resistivity from Eq. (2.3.14) at tempera-

ture T within region 111, can be represented by
p = 1).. 6‘3 (”r—“T3 (2.3.15)

where the resistivity p0 is measured at the absolute temperature T“.

Another way to represent the temperature response is to expand the exponent in

Eq. (2.3.15) to obtain

 

A A2 A,
P-p..CXPA..+—+ 2+—_,+--- (2.3.16)

This relationship lends itself to better computer curve fitting so that the resistance to

temperature relationship can be very accurately represented.

The temperature coefficient of resistance or is defined as

(I = — —- (2.3.17)

This is, in effect, the measure of sensitivity of the temperature sensor. For semicon-
ductor temperature sensors or is very sensitive to temperature. From Eq. (2.3.15) and

(2.3.17), we get

27

_ T2R(,efi(ln_l/T”)
a = _1_ a _ B

ith' ngemn

 

=—sr3 a3im

Following this brief theoretical analysis, the second major part of this review will deal
with the synthesis and characterization of diamond thin films and their application as a

sensor for thermal signals.

2.4 PROPERTIES OF DIAMOND

Diamond is one of the most technologically and scientifically valuable crystalline
solids found in nature, as it has a combination of properties effectively unrivaled by
any other known material. Among its unique physical properties are extreme hardness
(104 kg/mm3), high chemical inertness, optical transparency (225 nm to far-infrared),
radiation immunity, and high thermal conductivity (typically 20 W/cm-K). Additional
properties are extremely low friction coefficient (0.05), low thermal expansion
coefficient (l.lx1()"6 K‘l), high tensile and compressive strength (0.5><10° and 14x10“
psi respectively) and wide energy band gap (5.45 eV) [45]. It is a typical covalent
solid in which carbon atoms are joined in tetrahedral arrangement forming a diamond~
cubic lattice. Diamond crystals are most commonly found in octahedral and dode-
cahedral shapes with faces parallel to the [111] and [110] planes respectively. A sim-

ple cubic with faces parallel to [ 100] faces also exists but is less common [46].

Diamonds are classified according to their optical and electrical properties into
four types [47,48]:

Type Ia: These diamonds are optically transparent. They contain up to 0.1% nitro-

gen in small aggregates or platelets which induces infrared absorption and

limits thermal conductivity to 9 W/cm-K (at room temperature). Electrical

resistivity is > 101°Q-cm.

28

Type lb: These diamonds contain up to 0.2% paramagnetic nitrogen incorporated in
the lattice. They are typical high-pressure synthesized diamonds. Optical,

thermal, and electrical properties are similar to type Ia diamonds.

Type Ila: This very rare type is practically free of nitrogen and transparent to ultra-
violet above 225 nm. Thermal conductivity at room temperature ranges up
to 26 W/cm-K. Electrical resistivity is in general similar to type la dia-
monds.

Type llb: Extremely rare in nature and virtually nitrogen free, these diamonds con-
tain boron in small quantities. This produces a bluish color. They are p-
type semiconductors with electrical resistivity of only 10 to 1000 Q—cm.

Diamond is extremely inert chemically and is not affected by any acid or any
other chemicals, except those which act as oxidizing agents at high temperature. These
provide the only effective way to attack diamond at temperatures below ~1300 K and
at normal pressure [47]. Substances such as sodium nitride are known to attack dia-

mond in molten state as low as ~700K [47]. In oxygen itself, diamond starts to be

oxidized at about ~900 K [47].

The only other possible form of chemical attack is by two groups of metals. The
members of the first group are avid carbide formers, and include W, Ta, Ti and Zr. At
high temperatures (>550 °C) these will react chemically with diamond to form their
respective carbides. The second group includes Fe, Co, Mn, Ni, and Cr, and also the
platinum group of metals. ln molten state these metals are true solvents for carbon

[47].

2.5 CHEMICAL VAPOR DEPOSITION OF DIAMOND

The motivation and interest for the diamond synthesis started since the discovery
by Tennant in 1797 [46] that it is simply a crystalline form of carbon. Early attempts.

mostly based on unscientific approach, proved unsuccessful. Later on, with the

29

progress in understanding of chemical thermodynamics, the pressure-temperature range
of diamond stability was acquired. The real breakthrough came in 1955 at General
Electric where a high-temperature high-pressure (HPHT) process for diamond synthesis
was discovered [49]. In this process, diamond is synthesized in its thermodynamically
stable phase using a molten transition metal solvent-catalyst. After the formation syn-
thetic diamond remains kinetically stable as it is brought to ordinary atmospheric con-
ditions of pressure and temperature. This is due to the large activation energy required
for the conversion of diamond into graphite which is the thermodynamically stable

form of carbon at normal atmospheric conditions.

An entirely different approach has been directed towards growth of diamond from
the gas phase using hydro-carbon species at low pressure and moderate temperature,
that is, in the thermodynamically metastable region of diamond [46,50]. Although
these efforts were successful in growing diamond earlier than the HPHT method, they
did not attract much attention due to the low growth rates (<0.1t1m/h) achieved and the
simultaneous deposition of significant amount of graphite. Nevertheless, the critical
role of atomic hydrogen in achieving metastable diamond growth as a preferential
etchant for removing graphite against diamond was established. An important break-
through by Soviet scientists in early 1970’s indicated the conditions under which gas
activation techniques can greatly increase the growth rate of diamond and suppress the
graphite deposition [51]. The key feature of this work was the three approaches,
namely catalytic, electric discharge, and heated filament, which produced higher con-
centration of atomic hydrogen than that normally present in the thermal dissociation of
hydro-carbon gases. Since then, several techniques have been developed for rapid

chemical vapor deposition (CVD) of diamond films.
Of numerous methods presently used for CVD deposition of diamond using gas
activation, hot filament [52,53], d.c. plasma [54,55], r.f. plasma [56,57], microwave

discharge [58-61], thermal d.c. plasma [62] and combustion flames [63] are a few to

30

mention. Interestingly, all these techniques share several common features:

(i) High-energy densities are produced in the gas phase, sufficient to result in the

production of radical species, notably atomic hydrogen.
(ii) The nature of the hydrocarbon precursor used is relatively unimportant.
(iii) Very similar substrate temperatures are used (6()(’)°-l(')0(_)°C).

(iv) Deposits may vary from nanocrystalline to single crystal diamond with no or
little non-diamond C, depending upon proportions of C, H, O in the supply of

gases, but without reference to the type of gas phase activation used.

These common features strongly suggest that there is a common mechanism for the
growth process, and thus the choice of method is dependent upon considerations of
efficiency, convenience, cost and applicability to the problem at hand [64].
Semiconducting diamond, an attractive material for electronic devices, has been
grown homoepitaxially, and polycrystalline diamond films have been grown on many
substrates [46,53]. Hetroepitaxial growth of diamond films, a necessity for most elec-
tronic device applications, has not yet been accomplished except on cubic BN [65].
Recently, a novel technique named as "artificial epitaxy" has been reported [66]. In
that approach, diamond seed crystals were aligned by placing them in inverted pyramid
shaped craters micromachined on a silicon substrate surface. An oriented diamond

growth of limited size is also reported [67].

31

2.6 HOT-FILAMENT C VD

Since the hot-filament CVD method is employed for the synthesis of thin dia-
mond films in this research, an overview of the deposition system and influence of

associated growth parameters will be briefly discussed here.

2.6.1 The Deposition System

Thermally activated or hot-filament CVD (HFCVD) is one of the most common
techniques for diamond film growth. A conventional HFCVD system is composed of
a vacuum chamber equipped with a filament made of W, Re or Ta, a substrate holder.
a reactant gas inlet and an arrangement for the measurement of temperatures of I
filament and substrate [52]. Beyond these primary requirements, many additional
features have been added by many researchers to improve growth rate, film size and
reproducibility, to allow for in situ probes for gas thermodynamics and surface reac-
tions, and to have automatic fail safe operation. The filament at 2000-2400°C is used
to dissociate gas mixtures containing 0.5-5% of carbon carrying gas in hydrogen usu-
ally at sub-atmospheric pressures (10-700 tort). The dissociated products at these tern-
peratures consists mainly of hydrocarbon radicals, for example,
CH2, CZH, CH, CZHZ, CH3, and atomic hydrogen. These radicals lead to preferential
growth of diamond against graphite on the surface of an appropriately prepared sub-
strate placed typically 1 cm from the filament. The substrate to filament distance is
critical to meet the opposing requirements to minimize thennalization of the substrate

and radical recombination.

The main advantage of HFCVD method is its relative simplicity, inexpensiveness
and ease to scale up to large substrate areas and shapes. However, various components
in the system can add impurities into the films due to increased vapor pressure in the
presence of filament at very high temperature [64]. The role of these impurities in the

electrical properties of the films may be very critical. The brittleness and the

32

deformation of the filament due to carbodization at high temperature is another prob-
lem. However, these problems can be taken care of by using various alternative

materials and techniques.

2.6.2 Effect of Gas Composition

As mentioned earlier, the nature of the hydrocarbon precursor used for diamond
deposition is relatively unimportant [64]. The main requirement is the dilution of few
percent (0.5-5.0) hydrocarbon gas in hydrogen. Diamond has been shown to deposit,
with varying growth rate, from a variety of organic compounds including acetone
(CH3COCH3), ethanol (CZHSOH), methanol (CH3OH) and diethyle ether (C2H5OC2H5)
[68]. The variation in growth rate is possibly due to varying efficiency of each corn-
pound in supplying CZHZ and other radicals thought to be mainly responsible for dia-
mond growth [69]. Harris et a1. has also investigated the relative importance of various

species for diamond nucleation and growth [70].

2.6.3 Effect of C-H-O Ratio

To date, for most of the diamond CVD experiments, highly diluted mixture of a
carbon carrier with hydrogen, with and without oxygen are common. The carbon car-
rying species are usually less than 1%. A relative reduction in carbon component in
the gas mixture leads to an improvement in the quality but a reduction in the growth
rate of the deposited diamond [52]. It also has a pronounced influence on the morphol-
ogy of the deposited film [46] (see Fig. 2.4(a)) [71]. The best quality diamond films
are grown in a pre-dominantly hydrogen atmosphere. In any case, atomic hydrogen
should be in a super-equilibrium state for a successful diamond growth. Hydrogen
atoms play a crucial role in film growth since they have been shown to etch preferen-
tially non-diamond carbon, provide stabilization for sp3 bonding (diamond structure)

and promote the generation of the main pre-cursors (CZHQ and CH3) for the diamond

 

 

[Nucleation and Surlace Energy Dominate ] Growth Regime [Ontiamuai Demfirate
Kinetics Dominate

 

('ulto-i Idahcdrom
/ 1 l I I)
//_-> \\
/

(killer-chin

 

 

 

 

 

, Decreasing Hydrocarbon/Hydrogen (or Oxygen) Ratio

(21)

I... 0‘.
'3 I. Q mm

1
-' i ,o‘ worm
; do”

 

 

9.340
9‘0 .
'J. a.
C“ b‘
‘ ' ..
. ,.-,......:::::... ,. ..
“fl! filg'. . ', T .\;‘ ‘. .
" . .............................................. tw‘
.OIIIII I.‘jl’l"1130"11'I'IY"YII'I'I'I‘rIj'Y‘&VI‘I o‘ (b)
—-——.’
”WM “3

Figure 2.4 (a) Influence of C-H-O ratio on the morphology of CVD deposited dia-

mond films [71] (b) Atomic C-H-O diamond deposition phase diagram with the dia-
mond growth domain [72].

34

growth. Recently, an elegant carbon-hydrogen-oxygen (C-H-O) phase diagram as
shown in Fig. 2.4(b) for diamond deposition has been introduced [72]. Given the mag-
nitude of three components, the presence and quality of deposited diamond can be
evaluated. The region of phase diagram feasible for diamond growth has been defined

on the basis of experimental data collected from the literature and theoretical analysis.

2.6.4 Effect of Oxygen

Various studies [73-75] have shown that addition of oxygen in HFCVD growth
environment leads to higher quality films at increased growth rate. This has been attri-

buted to the following effects caused by oxygen:
(i) Formation of more reactive surface.
(ii) Formation of additional radicals necessary for diamond growth.
(iii) Destruction of gas phase pyrocarbon forming species and increase in atomic
carbon density.
(iv) Increase in atomic hydrogen density which tends to selectively etch the non-
diamond carbon from the surface.

(v) Oxygen may itself act as selective etchant of non-diamond carbon.

2.6.5 Effect of Substrate Temperature

The substrate temperature for diamond CVD, in almost all the techniques, has
been in the range of 800-1000°C. However, a small variation in the temperature may
greatly influence the growth rate and grain size [52]. On substrates hotter than 1300°C,
only graphitic carbon, if any, is deposited [52]. From C-H-O containing mixtures, dia-
mond may deposit as low as 400°C, but at very low rates [61]. The effect of substrate
temperature may well be visualized from Fig. 2.4(b) such that with increasing tempera-

ture, the width of region of diamond growth reduces and at T > 1300°C, it shrinks

35

down to almost zero [72]. Hence, the freedom of C-H-O composition fades away with

rising temperature.

2.7 ANALYSIS TOOLS

A number of analysis tools are used to monitor and characterize various stages of
CVD process of diamond. Most of them are, however, used either to study in situ gas
reaction and substrate surface chemistry or physical composition of the films. In this
research, diamond films were characterized after deposition. For this application
Raman spectroscopy, scanning electron microscope (SEM). secondary ion mass spec-
troscopy (SIMS) and surface profilers are the most appropriate tools and were used
extensively. A brief description of each in context with its specific use in diamond film

characterization is given here.

2.7.1 Raman Spectroscopy

Raman spectroscopy is based on the phenomenon of inelastic scattering of radia-
tion by a medium. When monochromatic radiation of wavenumber V0 is incident on a
material, the transmitted radiation contains, in addition to v”, a pair of new
wavenumbers of the type v’i = vuivM. They are caused by inelastic scattering due to
creation or annihilation of phonons and known as Raman scattering. In molecular sys-
tems, the wavenumber vM are found to lie principally in the range associated with
transitions between rotational, vibrational and electronic levels. In case of diamond,
they are dominated by the intramolecular vibrational transitions corresponding to opti-

calphonons

In the spectrum of scattered radiation, the new wavenumbers are termed as
R'aman lines or bands and collectively are said to constitute a Raman spectrum. Raman
bands at wavenumbers Vn-VM are referred to as Stokes bands and those at

wavenumber v0+vM as anti-stokes bands. The strong Raman line at v” i.e., without

36

change of frequency, arises from the scattering centers, like molecules, which are
much smaller than the wavelength of the incident radiation. It is known as Raleigh
scattering. The intensity of anti-stokes lines is usually much smaller than that of
Stokes-lines. Therefore, Raman spectra presented here is based on signals from

Stokes-lines.

Raman spectroscopy offers the advantage of sensitivity not only to crystalline
material, but also to the various possible non-crystalline phases. A diamond film shows
the characteristic Raman peak at relative wavenumber of 1332 cm’1 [76]. Well-
ordered graphite, similarly, has only one Raman peak at ~1600 cm“1 [77]. However,
the presence of disorder or small crystallite size gives rise to a peak at 1355 cm"1
[77,78]. Hence, most graphitic carbon phases produce two Raman peaks, sometimes
referred to as the "D" and "G" peaks. The breadth, position, and relative intensity of
these two peaks can vary significantly by the presence and relative density of sp3 to
sp2 hybridization. The peak at around 1550 cm’1 has been commonly attributed to the
presence of graphite i.e., sp2 hybridization. This is not an accurate assignment, since it
has been argued to arise from diamond-like carbon (DLC), which may not be graphitic
in nature [64]. In addition, films with nanocrystallites show a broad peak centered at
~1133 cm‘1 [79]. Despite some objections, the quality of diamond deposit is generally
judged from the relative intensity of the peaks at 1332 and 1550 cm”1 [64]. Robbins

et al. has published a comprehensive analysis of various attributes to the shape of

Raman Spectra [80].

2.7.2 Scanning Electron Microscope (SEM)

Another very useful tool for studying the morphology and visual analysis of dia-
mond films is scanning electron microscope (SEM). Since the CVD diamond films
generally have micron or sub-micron size grains, optical microscopes which has an

upper limit of magnification of about 2000x and, at that magnification, a small depth

37

of field, cannot resolve them effectively. SEM, on the other side, has a capability to
resolve objects in dimensions down to 100°A. The increased resolution arises from the
much smaller wavelength 1,, of electrons used in the illuminating beam, given approxi-

mately by [81]

MW = $ (2.7. 1)

where V is the electron acceleration voltage that is usually in terms of tens of kilo-
volts. SEM is helpful in analyzing the diamond films for voids, cracks, adhesion with
the substrate, thickness measurement, selectivity of patterns, and uniformity of surface.

and crystal orientations, etc.

In the present work SEM was mostly used for inspection of surface morphology
and to determine the thickness of diamond films. It was also used to analyze patterns,
nucleation density and growth selectivity. The need for a conducting specimens some-
what limited its utility for undoped films on electrically insulating substrates. More-
over, a prolonged exposure to high power electron beam caused some charge accurnu-

lation and blackening of the diamond surface.

2.7.3 Secondary Ion Mass Spectroscopy (SIMS)

Secondary ion mass spectroscopy, as the name implies, is a form of mass spec-
troscopy used to analyze secondary ions emitted from a surface after being sputtered
by an ion beam. By moving the sputtering beam of ions across the surface in a raster
fashion and setting the spectrometer to detect a particular mass value. topographic
scans of a given element’s concentration can be made. Using a static beam and
sputtering a hole into the surface, a depth profile normal to the plane of surface can be
generated. It was this mode in which SIMS was used to analyzed the boron doping

profile in semiconducting diamond films in this work.

38

An oxygen ion beam was used to sputter diamond films and underlying SiOz/Si
substrate. Oxygen sputtering has been shown to cause highly anisotropic etching of
diamond films and make holes along the grain boundaries down to the substrate in few
minutes [82]. This effect can cause a signal to be picked up from the substrate right
from the early phase of the scan while most of the diamond film is still intact and,
hence, generate an erroneous data. A further difficulty arises when the wall of the
holes, which are not exactly perpendicular, have some contribution to the signal.

Therefore a great deal of caution is necessary to evaluate and interpret the SIMS data.

2.7.4 Surface Profiler

SEM is by large the most effective tool for visual surface analysis and thickness
measurement using a edge of a conducting film. However, quantification of these
parameters is not easy. Now, electromechanical systems are commercially available
that can scan across a surface and readout vertical variations with a resolution better
that 5°A! [83]. In addition to sheer convenience, a major advantage over SEM is that
the step measurement is independent of optical properties of the film and substrate.
One such system, Sloan DekTak 1] surface profiler, was used in this work to measure

the diamond film thickness and surface uniformity.

In principle, the instrument uses a linear variable differential transformer (LVDT)
to convert vertical motion of the core to electrical signals. The transformer core is
attached to a diamond stylus that contacts the sample surface. The instrument shows
the scan data interactively on a CRT display which can be printed. A number of com-
putational and statistical functions are provided to manipulate the data to ascertain
specific informations. These functions were frequently employed to accurately deter-

mine the thickness and surface roughness of diamond and other films.

2.8 ELECTRONIC PROPERTIES

Diamond has been known as an excellent electrically insulating material. The
discovery of serniconduction in natural diamond [84] revealed its excellent electronic
properties [47,85] in addition to its already known unrivaled physical properties. It is
a covalent solid with diamond cubic lattice structure and energy band gap of 5.45 eV.
When doped, it exhibits semiconducting properties that are superior to those of corn-
monly used semiconductors. Table 2.2 gives a summary of principle properties of dia-
mond along with representative data for SiC. GaAs and Si [86]. It is obvious from
these data that, from an electronic perspective, diamond should be a good candidate for
high temperature (due to large band gap) high speed (due to carrier mobility) and high

power (due to high thermal conductivity and breakdown voltage) electronic devices.

For fast dynamic response time of temperature sensors, the thermal conductivity
of the base material is an important factor. Diamond, as shown in Fig. 2.5(a) [87], has
clearly an edge over other competing materials in this respect. However, the thermal
conductivity of diamond films is relatively less (3-17 W/cm-K) than that of natural dia-
mond depending on their quality. A diamond film grown with relatively higher concen-
tration of carbon carrying gas generally has poorer thermal conductivity and the quality
(see Fig. 2.5(b) [88]). At AT&T Bell Labs. it has been found that the thermal conduc-
tivity of polycrystalline diamond films is directly proportional to the quality, thickness
and the grain size [89]. An interesting finding was that the thermal conductivity of
poor quality diamond films improves on heating (possibly due to reduced role of grain
boundaries against phonon scattering) in contrast to high quality films and crystalline

diamonds.

Electrical transport measurements on B, Al, and Be doped synthetic diamond
have been reported in [90]. These specimens were found to be p-type semiconductors
with impurity activation energies of 0.17-0.18, 0.32, and 02-036. eV respectively.

Similar measurements by Wilson [91] in the temperature range of 88 K to 293 K on B

40

 

3
T

‘
3
r

.... a“, l

1111)

. . "l. Willi

1 :I re re in
mm (KI

3
1

menu». cououcrrvm m cm" 11")
U

u
i

 

 

 

 

 

 

 

‘ (a)
I i i I I

- at 100~ 130 '0

Po _
2 1 000 —— Nlttlll diamond
é ” \ (Typenb)
\ \
g .. \ _
2' _ 9 — Natural diamond
:3: *- ‘\ (Type 1a) 3

1 .

g 500 - ‘, —
D — A! -
i '- — on
E \ ..
2 " ‘
1- “ ..

" \

\“
)- ‘s‘-~ . d
0 1 1 1 --- q) -------- 'q
0.1 0.3 0.5 1.0 2.0 3.0
CH; {H3 IVOI %]
Methane concentration
' (b)

Figure 2.5 Comparison of thermal conductivity of (a) natural diamond with several
other solids as a function of temperature [116] (b) CVD diamond films using different

methane concentrations with natural diamonds, Ag and Cu [88].

41

doped specimens yielded nearly 30 discrete activation energies ranging between 2.9
and 87 meV. These data were interpreted as evidence of impurity conduction or hop-
ping transport which is commonly observed in heavily doped and highly compensated
semiconductors at low temperature [92-94]. In both natural and synthetic serniconduct-
ing diamonds acceptors are typically compensated by deep nitrogen donors with activa-
tion energy of about 4 eV above the valence band [94,95]. The acceptor activation
energy for boron in natural semiconductor diamonds has been recognized at 0.37eV

[47].

The electronic transport phenomenon in CVD diamond films is quite complex.
So far, there have been few efforts to evaluate the electrical properties of these .dia-
mond films [96-100]. This reported work has actually been limited to ascertain resis-
tivity, dopant concentration, Hall concentration and mobility, and dopant activation
energy in a limited temperature range. However, due to a large scatter in the reported
data, no comprehensive model could be developed so far. Furthermore, a few reports
about other kinds of measurements such as capacitance versus voltage (CV) measure-
ments on Schottky structure using diamond as a semiconductor [101], high field effect
in diamond thin films [102] and effect of annealing [103] and frequency [104] at vari-
ous temperatures on the resistivity of undoped diamond films have also been pub-
lished. At present, since most of the literature deals with the undoped films [99-103],
the conduction mechanism in doped diamond films is still not clear. Gildenblat has
recently published a comprehensive review of understanding of the subject to date
[105]. At present, there exists a need to measure all parameters and their inter-
dependence necessary to develop a comprehensive model of conduction mechanisms
over a wide range of doping and temperature in diamond films before any real high
performance high-temperature high-power electronic devices may be developed. In this
regard, previous work on natural and synthetic diamonds [95,106,107] and similar

semiconductors [156,157] can extend very useful support.

42

2.9 FILM AND BULK PROPERTIES

Since the main interest in diamond is spurred by its potential application in sem-
iconducting devices, the primary effort in CVD diamond deposition is focussed on
achieving device quality films. Although the excellent physical and electronic proper-
ties of diamond promise devices of extra-ordinary characteristics, semiconducting CVD
diamond films, so far, do not possess the electronic properties to the same degree of
excellence as single crystal bulk diamond [43]. For example, thermal conductivity.
electrical resistivity, hardness, carrier charge carrier mobility of polycrystalline dia-
mond films are relatively inferior than listed in Table 2.2. Some critical electronic
properties such as carrier mobility are usually orders of magnitude less than single cry-
stal bulk natural diamond. The major difference in these properties is contributed by
the high density of defects in single crystal and grain boundaries in the polycrystalline
films.

Thin films generally have high density of defects due to growth process [108],
lattice mismatch [109], differences in thermal expansion coefficients between adjacent
films or substrate and film [110], and stress [111]. For a variety of reasons, film
growth is generally done on materials other than itself, and the resulting semiconductor
thin films are usually polycrystalline. The conduction mechanism in polycrystalline
films is dominated by surface scattering and defect dominated properties [112,113].
The defects in polycrystalline materials are mainly the grain boundaries. The forego-
ing description is equally applicable to polycrystalline diamond films grown on non-
diamond substrates. The electrical properties of thin diamond films tend to vary
grossly from single crystal bulk diamond [43]. At present, no universal explanation of

transport characteristics exist for thin diamond films.

The surface of a thin film affects the electrical transport properties of a material
by limiting the transversal motion normal to surface plane. When the thickness of the

film becomes less than or comparable to the charge carrier mean free path, the

43

Table 2.2 Comparison of semiconductor properties of diamond with SiC. Si.

and GaAs [86].

 

Properties

GaAs

Silicon

 

 

Lattice Constant ("A)

5.65

5.43

 

 

ff-{jffaanal expansion (x1073. °C)

26

 

Densrty (g-cm 3)

 

Meltrng Point (°C)

 

 

.............. ,..L..i.-...‘..'...

 

Satureted electron velocrty (x107 cm-s

 

”Carrier mobility (cm 2V'ls'l)
Electrons
- Holes

 
 

 

   

 

 

Dielectric constant

  

 

 

 

Resistivity (O cm) . . ,

 

 

Refractive index

 

 

 

1000

 

Hardness (kg m 3)

 

 

 

Keyes figure of merit (x100 W cm 1’s 1 °C)

 

 

 

13.8

 

 

 

44

scattering of charge carriers from the surface has measurable effects on the transport
properties and can dominate the electrical characteristics of the film. The extent of
influence of surface scattering depends upon the nature of scattering mechanisms
involved, i.e., either totally elastic (specular reflection) or totally inelastic (diffuse

reflection), or combination of both [114].

Measurements of resistivity and Hall coefficient are necessary for determining
both the mobility and carrier concentration which are generally dependent on surface
scattering and thickness of thin films. The film Hall coefficient is usually less than that
of bulk material [115]. Hence, errors in measurements can arise from specimen con-
tours; electrode size, geometry, position, and symmetry; and spatial and thickness

homogeneity [l 16].

Two models have been widely used to interpret the electrical properties of ele-

mental polycrystalline thin film semiconductors.

(l) The Segregation Model, in which the impurity atoms segregate to the grain
boundaries and become electrically inactive [117,118]. The films show enhanced

resistivity that they actually have.

(2) Grain Boundary Trapping Model, It is commonly characterized by the presence
of a potential barrier ¢b created by capturing of free carriers [119,120] by the
active trapping sites at the grain boundaries. These barriers present obstacles to the
transport of carriers between grains, and the conduction is dominated primarily by
the thermionic emission. This model was originally developed for polysilicon [119]
and was refined later by others [121-123].

The application of these models to polycrystalline diamond films may be useful in

interpreting the relevant experimental data and development of its own model for con-

duction mechanism.

45

2.10 DEVICE FABRICATION TECHNOLOGIES

For electronic device fabrication, in addition to evaluation of electronic properties
of the diamond films, many technological problems are required to be resolved. The
main issues are controlled doping of the films, film patterning, and formation of physi-
cally stable and ohmic contacts. The diamond films grown without intentional doping
are usually good insulators. Semiconductivity may be induced in synthetic diamonds
in a variety of ways. However, doping during growth of synthetic single crystal dia-
monds and CVD diamond films has so far produced only p-type conductivity. The
most common method is the introduction of specific impurities such as Be, A] or B for
p-type doping to suitable growth mixtures of carbon and solvents [90]. It is also a
common method in case of CVD diamond films in which Boron is commonly added
into the growth mixture in solid (pure boron powder [58], 8203 [124,98]) or gaseous
phase (821-16 [125], Boric acid vapor [126]). Introduction of active impurities by
means of ion implantation has also received considerable attention [127-130]. Boron
doping through diffusion using a rapid thermal processor (RTP) was demonstrated in a
metal-semiconductor field effect transistor (MESFET) with ultra-shallow channel
(~500°A) [131]. The few reports on P doping to induce n-type conductivity
[91,127,132] are not very encouraging. The defect sites created by ion implantation of
dopants such as C+ [130], P+ [127], were found to be responsible for the observed n—
type conductivity.

Film patterning is another important issue in the absence of any conventional
chemical etchant of diamond. Several reports to pattern diamond films have been pub-
lished [133,138-141]. Although several distinct techniques have been used but they
can be grouped into two broad categories: selective deposition by enhancement
[134,135,138] or burying [l33,136,l40,14l] of the diamond nucleation sites and selec-
tive etching through oxidation [137,141]. However, most of these techniques have

several disadvantages and are not compatible with standard planar 1C processes.

 

46

Hence, they cannot be adopted directly and a considerable amount of refinement or
modification required in each case for actual device fabrication.

Anticipating the high temperature operation of diamond devices, ohmic contacts,
which are physically stable on the diamond as well as the substrate surface on which
the diamond film are deposited, are an essential requirement. Electrical characterization
of contacts of a number of metals with homoepitaxial diamond films has been investi-
gated [142]. Moazed et al. [143] have demonstrated the formation of ohmic contacts
using W or Mo that react with diamond at high temperature (~9()()°C). Indium was
found to form ohmic contact without high temperature anneal [144]. Heavily boron
doped diamonds (>1019 cm’3) also form ohmic contacts with most evaporated metals
without post deposition processing [144]. A novel method of making dual side con-
tacts to a diamond thin film has been reported [145]. Another report dealing with
conversion of rectifying contacts to ohmic by damaging the metal diamond interface
by ion implantation [146] seems to agree with the argument that every metal contact
on the rough semiconducting surface will behave pseudo-ohmic [147]. Homoepitaxial
films need special considerations since the same metal contact which is ohmic on
polycrystalline diamond surface may turn into a rectifying one [148]. Surface treat-
ment before metalization may have significant effect on the behavior of contact [149].
For interconnects running between diamond devices and to bonding pads in an IC on
non-diamond substrate (which is silicon in most cases) pose new challenge. In this
regard, the numerous metallization studies involving refractory metals [150,151] for

conventional monolithic Si le may prove very useful.

Using the technology developed to date, a number of rudimentary devices using
natural and synthetic diamonds have been developed. They include a thermistor [37-
39], bipolar junction transistor (BJT) [152], Metal-Oxide-Semiconductor field Effect
Transistor (MOSFET) [58,153], point contact transistor [154], Schottky diode

[154,155], resistors [124] and light emitting diodes [130,155]. However, most of these

47

devices were barely working models of respective device principles but never demon-
strated the clear advantage of diamond properties over the existing devices from con-
ventional semiconductors. This aspect, therefore, emphasizes the need to develop and
refine the device fabrication technology preferably in line with the modern state-of-

the-art 1C technologies.

CHAPTER 3

 

SEMICONDUCTING DIAMOND FILM
SYNTHESIS AND CHARACTERIZATION

3.0 INTRODUCTION

This chapter deals with the synthesis and physical characterization of boron doped
semiconducting diamond films produced as part of this dissertation research. After a
brief description of the HFCVD system for diamond deposition, two methods used to
enhance the low intrinsic nucleation density of diamond on non-diamond substrates
namely ultrasonic treatment and diamond-photoresist seeding are discussed. Since the
diamond deposition depends upon a number of experimental parameters, section 3.3
covers the influence of various parameters on the diamond growth and their optimiza-
tion work for the good quality diamond films. Next, the implementation of. in situ
boron doping and its effect on the quality of diamond films using solid dopants is dis-
cussed. Lastly, a critical effect for device fabrication i.e., etching of SiOz during dia-

mond growth is analyzed.

3.1 THE DEPOSITION SYSTEM

The diamond films for this research were deposited in a hot-filament CVD sys-
tem. The system was originally designed and built at Ford Scientific Research Labora-

tory. Several modifications and improvements were incorporated during the course of

48

49

this research to perform certain specific experiments or to improve the reproducibility.
growth rate and quality of the films, and ease of system operation. Some salient
features of the system are described here to enhance the explanation of the experimen-

tal parameters explained later in the present and following chapters.

This HFCVD system is basically an improved version of the system used for ini-
tial diamond depositions using this technique by Matsumoto et a1. [52]. The improve-
ments are mainly in terms of control and stability of growth parameters, boron doping
and flexibility of operation. A simplified schematic diagram of the system is shown in
Fig. 3.1. The main vacuum chamber is a 6" hollow cube made of stainless steel pro-
vided with openings for flanges on its each face. CH4 or CZHQ, H3 and CO were used
as reactant gases. The flow rate of all three gases is precisely adjustable independently
by mass flow controllers. In addition, nitrogen is used to purge and backfill the
chamber. A glass view port provides an in situ visual access to the substrate and vari-
ous components inside the chamber. The pressure of the system is monitored by a
baratron pressure sensor which supplies suitable signals through a controller to a down
stream valve which maintains the chamber pressure to be within 0.1% of the preset

value (up to 100 torr).

The filament is a horizontal array of seven 2" long parallel Ta wires (0.005"
diameter, 99.9% pure) spaced at 8 mm. The filament wires are prevented from sagging
down on heating by using spring force loaded Mo clips. Ta, in stead of conventional
materials such as W or Re, was selected because it was found to be less brittle and
more resistant to shocks and vibrations after heating and thus prolonging its opera-
tional life time. It has also been found to cause no or vary small metal impurities in
the deposited diamond films. The temperature of the filament is monitored by a two
wavelength pyrometer and controlled automatically to within i2°C by a controller
using an SCR based power supply. The filament typically draws 25-30 A current at
25-35 VAC at nominal temperature of 2400°C.

50

CWan
CO

    
  
      
      
   
   

TVpeK

Thermocouple
Substrate

mm
mm diameter

J.

1mm

T

N2

Figure 3.1 A simplified schematic diagram of diamond HFCVD system.

51

The substrates are placed on a 1%" diameter Boralectric heating plate (made of
BN with graphite heating element embedded inside) below the filament. The heater is
supported by a frame with a provision of controlled vertical motion up to 0.75". The
controlled vertical movement is important in view of dual requirement of very short
and precise (3/16") spacing between filament and substrate during deposition and
accessibility of the top surface of heater for handling of the substrates and related com-
ponents. The temperature of the substrate, an extremely important parameter for CVD
diamond growth, is monitored by a type K thermocouple at its top surface. The sub-
strate temperature is controlled to within 1°C of the set temperature by controlling the
heater current in a feed back control system through a microprocessor based digital
temperature controller. The top surface temperature monitoring is unique in the sense
that in almost all cases reported in the literature, the temperature of the back surface is
monitored which obviously causes an indefinite error in temperature measurement. The
heater typically draws 5—8 A current at about 50-70 Vac to maintain a temperature of
890°C in the presence of filament at 2400°C. It is important to point out here that all
the components in the proximity of substrate are made of high temperature compatible
refractory metals/ceramics which cause minimal impurities in the deposited diamond

films.

The overall system performance is highly reproducible. Though the deposition
process is manually started and terminated, yet once started, it can virtually operate in

self supervisory mode indefinitely which is important for long duration deposition.

3.2 THE NUCLEATION METHODS

The intrinsic nucleation density for diamond growth on most non-diamond sub-
strate materials is generally very low i.e., 104 cm'z. To synthesize a diamond film
without voids in a reasonable time duration, there is a need to enhance the existing

nucleation density. In this research, the methods of ultrasonic treatment (UT) and

52

diamond-photoresist (DPR) seeding were used. A brief description of these methods is

given here.

3.2.1 Ultrasonic Treatment (UT) Method

It is a widely used method to promote nucleation density for diamond growth on
non-diamond substrates [133-135,138]. During this work, a suspension of diamond
powder with 0.1 pm size particles in a solution of organic solvents is prepared in a
beaker. With substrate placed inside. the beaker is suspended in an ultrasonic bath and
agitated for 30 minutes. Following the treatment, the substrate is rinsed with acetone
and methanol. The surface of the substrate following the treatment does not look .any
different than the one without treatment visually under optical microscope (12(.)()X

magnification).

3.2.2 Diamond-Photoresist Seeding (DPR) Method

This method to enhance nucleation density for diamond growth is based on the
idea of spreading diamond seed crystals, suspended in photoresist, on the substrate sur-
face. During CVD diamond deposition, the photo-resist evaporates in initial stages and
the diamond particles act as nucleation sites for the diamond growth. All the develop-
mental work on this method to achieve uniform diamond films with optimally smooth

surface was done during this research work.

3.2.3 Comparative Analysis

Both nucleation methods were quite successful in improving the nucleation den-

sity. Raman spectra of the diamond films deposited on the substrates nucleated by the

1

two methods, as shown in Fig. 3.2, have strong peak at 1332cm’ . This similarity is

indicative of absence of any role of residues of photo-resist in the growth of diamond.

53

 

 

 

 

 

 

     

   

 

 

 

 

 

 

~27
D
g
E.“
t
(a
E O UT SUBSTRATB'
. E I DPR SUBSTRATE _
800 1200 w 1600 2000

WAVE NUMBER (cm'l)

Figure 3.2 Raman spectra of diamond films on oxidized silicon substrates seeded by
UT and DPR nucleation methods.

54

The nucleation density was determined by counting the number of diamond grains
in an area of 3cm X 3cm on SEM micrographs taken at various parts of a specimen.
SEM micrographs of the diamond films deposited on the substrates pre-treated by the
two methods are given in Fig. 3.3. Two main differences between the two films are
visible. First, the nucleation density of UT specimens was relatively 5-8 times higher
than that of DPR specimens which were found to have nucleation density of about
108 cm“? Second, a relatively large variation in the grain size is observed in case of

UT specimens. DPR specimen has fairly uniform grain size (~l um).

Considering the two methods from technological view point, DPR seeding method

has following advantages over ultrasonic treatment method:

(1) Since only the diamond seed particles form the active nucleation sites, the nuclea-
tion density and the grain size in the deposited diamond film is easily controllable
through the size and density of diamond particles to be suspended in the photo-

resist and the spin speed of substrate during photo-resist coating.

(2) The substrate surface is not scratched or damaged by the seeds or photo-resist.

This allows a clean interface between the diamond film and the substrate surface.
(3) The method is compatible with the lithographic process used in IC technology.

(4) Since the diamond seed particles do not need to scratch the substrate surface, this
method is not dependent on the hardness of the substrate material and equally

effective on any substrate.
A major advantage of UT is that it can produce relatively high nucleation density

which is, however, not desired in the present case. Owing to these differences, most

of the films deposited for electrical measurements were seeded with DPR method.

55

 

Figure 3.3 SEM micrographs of diamond films on oxidized silicon substrates seeded
by (a) UT and (b) DPR nucleation methods.

56

3.3 DIAMOND DEPOSITION PROCESSING PARAMETERS

Since CVD growth of diamond is essentially in metastable phase, the selection of
experimental parameters and their stability throughout the growth process is very
important to achieve reproducible high quality diamond films. Specifically, they
include reactant gas composition, absolute and relative flow rates of gases, system
pressure, substrate to filament spacing, and substrate and filament temperatures. These
parameters were varied or fixed as follows to optimize the quality and growth rate of

diamond films:

(a) Filament temperature: 2200 - 2400°C (2400°C)

(b) Gas composition: H2, along with one of the carbon sources from methanol,
acetone, CH4 and Csz in the presence or absence of CO. (H2, CQHZ and CO)

(c) Substrate temperature: 830 - 950°C (890°C)

(d) Gas flow rates: CH4 and Cszz 0.5 - 1.0 sccm; CO: 0 or 12 sccm; H2: 100 sccm
(szCszzCO 100:0.5:12.0 sccm)

(e) Chamber Pressure: 50 - 100 torr. (50 torr)

(f) Substrate to filament distance: 3/16"

(g) Boron doping source: B203 or pure boron powder (0.1 - 2.6 mg).

The typical values of parameters used for diamond deposition used in this work are
given in the parenthesis. Additional parameters are the choice of substrate material.

size and shape and pre-treatment technique.

The aim of varying these parameters is to obtain or verify an optimum condition
suitable for the high purity diamond films at fast growth rate with controlled amounts
of uniformly distributed boron impurity. The deposited films were characterized by
SEM, Raman spectroscopy and surface profiler. The purity of the deposited diamond
films were ascertained in terms of absolute ratio of peaks corresponding to diamond

(1332 cm‘l) and graphitic carbon (around 1550 cm‘l) commonly known as spf‘lsp2

57

ratio. The height of each peak was determined by subtracting the background signal
from the Raman spectrum using computer routine. The influence of major parameter

on the purity and growth rate of the diamond films is described below.

3.3.1 Reactant Gas Composition

Methane and acetylene are commonly used as carbon sources for CVD diamond
growth. The role of two gases in providing principle growth species such as CH3 and
C2H2 is still not understood and has been discussed in the literature [69,70]. In addi-
tion, good quality diamond films have been deposited using other organic solvents
[68]. In this research, a number of films were deposited using Methane, acetylene,
acetone and methanol, and their effect on the growth rate and the purity was ascer-
tained.

Typical Raman spectrum of the films deposited using each of the four carbon
sources is shown in Fig. 3.4. An inspection of these Raman spectra reveal no
significant difference in terms of sp3/sp2 ratio between films deposited using methane
and acetylene. However, the intensity of the Raman signal was generally weaker for
the films growth with methane. The films grown with acetone and methanol showed
poor diamond quality. Especially in case of methanol, diamond peak was totally absent
in most samples. Therefore, acetone and methanol were dropped out of this study in

favor of methane and acetylene.

While considering the flow rate of these carbon gases, both methane and ace-
tylene produced good films at lower concentration (0.5 sccm in 100 sccm of hydro-
gen). However, a sharp deterioration in purity was observed with increase in flow rate
in each case.

Oxygen is also frequently used in diamond CVD reaction and has been shown to

improve its purity and growth rate [73-75]. However, adding even a small amounts of

oxygen resulted in failure of filament. Therefore, an alternative way was adopted and

58

 

 

 

r 1 t j I r r l I 7 I
— ' 02H: _
I CH4
. ‘ ACETONE
O MBTHANOL

 

 

 

 

 

 

 

 

INTENSITY (Arbitrary Units)

 

 

i

 

800 l 200 l 600 2000
WAVE NUMBER (0111“)

Figure 3.4 Raman spectra of diamond films deposited by using methane, acetylene,

acetone and methanol as carbon source.

59

CO in stead of pure oxygen was used. The effectiveness of CO in supplying oxygen
or modifying the character of the film can be observed from the Raman spectra, as
shown in Fig. 3.5, of films deposited with and without CO. A degradation in purity in
the absence of CO is clearly visible from the presence of large graphitic peak. How-
ever, an SEM inspection of the film without CO did not reveal any significant

difference.

3.3.2 System Pressure

Generally diamond films are deposited at sub-atmospheric pressures. Though there
are reports of diamond depositions over a wide pressure range i.e., 5-700 torr, a pres-
sure <100 torr is commonly used [64]. This indicates that pressure is not a critical
parameter. 1n an effort to increase the growth rate, diamond deposition at various
selected pressures in the range of 50-100 torr was done. In SEM inspection, all the
films showed well faceted grains with no distinguishable features from film to film.
Raman spectra of the films showed more pronounced effect (see Fig. 3.6(a) for
sp3/sp2 ratio). The purity seems to be largely unaffected by pressure except at 70 torr.
A negligible effect with a downward trend with increasing pressure on the growth rate

as shown in Fig. 3.6(b) was observed.

3.3.3 Substrate Temperature

Among many deposition parameters, the substrate temperature has a unique
significance. Given all the parameters optimized for the good quality of diamond
growth, only a small window of substrate temperature (i20°C) supports the optimized
quality.

To investigate these effects, a number of diamond films were deposited at
different temperatures. The influence of temperature variation on growth rate and the

purity of the film as determined by surface profiler and Raman spectroscopy

60

 

 

 

 

 

 

 

 

 

 

 

 

0 co: 12 SCCM

_ I co: 0 SCCM 1
g __ -1
D
>-
E:
E-
a - -

800 1200 1600 ' 2000

WAVE NUMBER tern")

Figure 3.5 Raman spectra of the diamond films deposited with and without CO.

61

 

sp’lsp’RATIO
o t-l. [9 03 h M OK 4 W ‘O O
m

 

 

CL 1 1 1 4 I

40 50 60 70 80 90 1'00 110
PRESSURE (Torr) (a)

1.0 I I I I I I

 

 

P
\O
r
1

.°
\I
l
l

HRA
o
u

 

 

 

 

5 0 4 ~ “E g
a 0.3 - q
o 0.2 ~ -
0.1 ~ _
0.0 l I l l 1 l
40 50 60 70 80 90 l 00 1 10
PRESSURE (Torr)

(b)
Figure 3.6 Effect of system pressure on the (a) quality (sp3/sp2) ratio and (b) growth
rate of the diamond films.

62

 

 

 

 

 

860 880
TEMPERATURE (°C)

900 920
(a)

 

1.5 . r

y—s
.
O

T I V

p
M
r 1 1

GROWTH RATE arm/hr)

 

l 4 l

 

1

 

0.0 ‘ 1
820 840

860 880
TEMPERATURE (°C)

900 920
(b)

Figure 3.7 Effect of substrate temperature on the (a) quality (sp3/sp2) ratio and (b)

growth rate of the diamond. films.

63

respectively is given in Fig. 3.7. A sharp degradation of purity while moving away
from the 900°C is clearly visible from the curve in Fig. 3.7(a). However, it should be
recognized that this temperature is optimal for a given set of other parameters. The
growth rate, as shown in Fig. 3.7(b), despite large error in measurement, has a small

upward trend with temperature.

3.3.4 System Operation

The operation of the deposition system primarily consists of sequence of actions
to bring up the processing parameters namely gas flow rates, chamber pressure,
filament and substrate temperatures to the desired level. However, it was found that
the order and time rate of changing a parameter has great influence on the purity and
nucleation density of the deposited diamond films. Therefore, a few critical steps in

this respect are mentioned here.

(1) After setting the substrate(s), boron powder holder (if any) and adjusting the
filament to substrate surface spacing to 3/16", the chamber is evacuated to <10

mtorr.

(2) Filament and heater temperatures are brought up slowly in hydrogen environment
(>20 ton) in the respective order manually to their desired temperatures. It is
important to avoid heavy in rush current or sudden temperature changes espe-
cially beyond the desired temperature which may lead to any damage to heater.

filament or substrate.

(3) Carbon sources are switched on only after the substrate temperature is reached
beyond 800°C. Below this temperature mostly non-diamond carbon deposits are

formed which lead to poor quality films.

(4) For UT substrates, the substrate temperature is maintained at 850°C for first 15
minutes. A higher temperature at the beginning leads to poor nucleation density

apparently by destroying potential nucleation sites. After this period, the

64

temperature is raised to desired temperature. In case of DPR substrates, this prac-

tice was unnecessary since nucleation sites are well defined seed crystals.

(5) The presence of a very thin conducting carboneous layer on the surface of CVD
diamond film surface has been suggested [158]. A simple method was used to
remove this layer. In that, only carbon containing gases are shut down and all
other parameters are left unchanged, including hydrogen, gas for 3 minutes before
the termination of the deposition process. In this period hydrogen, being the pre-
ferential etchant, is assumed to etch away the carboneous layer. The effectiveness
of this method was verified by conductivity measurement of the patterned films.
The films cleaned by this method were treated in the Cqu+HzSO4 solution for 5
minutes. An insignificant change in the film.conductivity was observed contrary
to indicated in [158]. This fact indicates that films were clean prior to this treat-
ment. If this cleaning procedure is not followed before terminating the deposition
process, relatively higher conductivity in otherwise similar samples was observed

which is indicative of the presence of this carboneous layer.

After the evaluation of all the experimental parameters individually, a set of values for
these parameters is available which could produce high purity films with optimum
growth rate. Some of the value individually or in combination are reactor dependent.
Therefore, this formulation should be considered more specific to the present deposi-

tion system than a general one.

3.4 BORON DOPIN G

CVD diamond films deposited without intentional doping are usually good insula-
tors. Semiconductivity may be induced by incorporating controlled amounts of suitable
impurities. B, A], and Li are known to act as p-type dopants in diamond [47]. In
natural semiconducting diamonds (type IIb) boron has been found to be responsible for

low resistivity. Therefore, a number of techniques involving gaseous

65

(BZH6, B(OCH3)3) and solid (3203‘ B powder) sources to incorporate boron into CVD
diamond films have been devised. Since gaseous doping sources are highly poisonous,

only solid sources were considered in this research.

Two solid sources i.e., B203 and pure B powder through separate techniques were
investigated. First, the vapor from saturated solution of B203 in organic solvents
(acetone or methanol) were introduced but this method did not produce good quality
diamond films. In the second method high (5N8) purity amorphous boron powder was
introduced directly into the chamber. The powder is placed on the substrate holder
(heating plate) using a specially designed holder as shown in Fig. 3.1. The holder con-
sists of Mo plate (1 mm thick) with a number of 1 mm diameter holes drilled in it.
The plate is inserted in a clip which covers its one face and a selected number of holes
are filled with boron powder. One holes takes on an average 0.2 mg of boron powder.
The boron holder is placed very close to the substrate (within 1 cm.) at a fixed place on
the substrate holder for better reproducibility of temperature and, hence, of doping con-
centration. The temperature of the boron powder is assumed to be equal to the sub-
strate temperature as measured by thermocouple and is not measured independently.
The vapor pressure of boron at growth temperature appears to be sufficient to evolve

boron vapor for controlled doping the diamond films.

The effectiveness of using solid boron as dopant source is evaluated through
SIMS and Van der Pauw [159] measurements. SIMS depth profile of a 311m thick dia-
mond film is shown in Fig. 3.8. The boron concentration is fairly uniform throughout
the thickness of the film. Since oxygen plasma was used to sputter diamond films dur-
ing SIMS scan, the large oxygen signal is assumed to be mainly contributed by the
plasma. The source of large signal corresponding to silicon is not clearly known. It is
assumed to be contributed by the silicon substrate which was possibly exposed right
from the early stages of the scan due to creation of holes by oxygen plasma along the

diamond grain boundaries [82]. SIMS of diamond films also confirmed the absence of

66

any other impurity which can interfere with the electrical measurements or caused by

the metallic components of the chamber itself.

 

 

 

 

 

 

 

 

 

 

 

 

EAAA“AAAAAAAA_A AAAAAAAAAAAAAAAA
A -vvwvv'-vvvvvvvvv---YVLVvCCCCCCVV$33$CCCCCCCVCCCQVAA
.-
.§
a — A -_ A-A.A__ ‘_._‘ A ._._ A; H AAA #
1 3 — _ — _ - — —- —
_ ‘A-A-A AA “ A--A ‘v-A 'vvv
' v 7'- "V'"'- ' v :
w- I d
O I
U I
‘
ton I
_
-
I Carbon ‘
101 A Sili i
100 r 4 r 1 I I 1 1 1 I 1 r 1

DEPTH (pm)

Figure 3.8 SIMS depth profile of a diamond film deposited on oxidized silicon

The Hall concentration measured by Van der Pauw method against quantity of
boron powder used for doping the diamond films is shown in Fig. 3.9. The two
separate curves in the figure correspond to diamond films deposited directly on oxi-
dized silicon and the films deposited on undoped diamond films as buffer layer. The
later samples were annealed at 600°C and 1000°C for 35 and 8 minutes respectively.
The Hall concentration varies in the range of 1015—1019 corresponding to 0.1-2.6 mg of

boron powder used to dope diamond films. Though quite non-linear, the change in

67

Hall concentration with boron powder quantity is monotonic in each curve. From the
two curves one can see that the Hall concentration is comparable in the two types of
samples except in one case. Although there is a some scatter in the data, the control-

ling of doping through quantity of boron powder appears to hold sufficiently.

 

   

 

19
10 t IIIIIIIIIIIIIIIIII T I 3
E 3
«.7‘ i ‘
E 10"; E
z t E
2 t .
a . .
E 1017:? E
in 3 :
U I 1
6 _ _..s.___.___4.._. ___.. —4
_, 10'
3 . TEST CHIP

 

1 0‘5 111111111111111111 é 11111111
1
BORON POWDER QUANTITY (mg)

 

Figure 3.9 Measured Hall concentration at room temperature against the quantity of

boron powder used for the doping of diamond films during deposition.

Raman spectra of doped diamond films is shown in Fig. 3.10 and 3.11. The two,
figures correspond to diamond films deposited directly on oxidized silicon and the
films deposited on undoped diamond films as buffer layer respectively. All the films
display typical Raman spectra of diamond with a strong peak at 1332 cm"1 To esti-

mate the effect of boron on the purity of diamond films directly on oxidized silicon

68

 

 

 

 

 

: Boronpowderquantity

” A 01113

” V 0.2mg

’ I 0.7mg {

” 9 2.01113 (

7 C3 13mg
3 _
0a .—
D r—
E
s F .
,. r
g: _
m
E _

huh; ‘ l I 1 1 1 1 1

800 1200 1600 2000

WAVE NUMBER (cm“)

Figure 3.10 Raman spectra of the diamond films deposited on oxidized silicon sub-

strate using various amounts of boron powder quantity.

69

 

 

 

 

 

 

 

 

: Boron powder quantity
" A Drug
_ A 0.6 mg
_ I 0.8 mg
_ O 1.2 mg .
~ D ' 2.6 mg I
— l
'- l
l
.9 - .4
3 r
tn .—
.§ —
>.
E -
E3 _ _
hr 1 1 l 4— 1 1 I ‘ 1 ‘
800 1200 1600 2000 '

WAVE NUMBER (cm'l)

Figure 3.11 Raman spectra of the diamond films deposited on undoped diamond films

deposited on oxidized silicon substrate using various amounts of boron powder quan-

tity.

 

70

and on undoped diamond film buffer layer, measurements of diamond and graphitic
peaks from the Raman spectrum were made. Plots of the sp3/sp2 ratio corresponding to
two types of films, as given in Fig. 3.12, show an interesting trend. The increasing
quantity of boron powder initially improves the purity of diamond, in agreement with
others [96], but leads to degradation beyond a certain limit. This limits is reached ear-
lier (at about 0.6 mg of boron powder) and the purity degrades very sharply for the
diamond films deposited directly on oxidized silicon substrates. In case of doped films
deposited on undoped diamond films, the degradation is much slower and starts at

higher boron powder quantity (at about 1.2 mg).

 

 

 

 

 

 

12 II I
11_ E
10; i
9: E
O : :
[:8_ :
:7 3
.,\6_ :
8': :
5_ 1
4: 3
3' E
2’ .w.
0 1 2 3

BORON POWDER WEIGHT (mg)

Fig. 3.12 Comparison of the effect of boron powder quantity on the purity of diamond
films deposited on (a) oxidized silicon and (b) und0ped diamond films.

71

The purity of doped diamond films on undoped buffer layer is better for comparable
amounts of boron powder. Therefore, it is suggested that heavy boron doping is more
deliterious in the early phase of diamond growth and does not degrade much the
growth on an existing high purity diamond film. As a result of these observations, all

the films for the electrical measurements were deposited on undoped buffer layer.

The Raman spectra of doped homoepitaxial films deposited on thin diamond
wafers along with a co-deposited polycrystalline film are shown in Fig. 3.13. All the

spectra show strong peak at 1332 cm".

Since the diamond substrates were transparent
to illuminating laser light, any clutter around diamond peak is believed to be the signal
coming from the specimen holder. Comparing these spectra, two major differences are
evident. The intensity of Raman signal (counts/sec.) for homoepitaxial films is much
larger than polycrystalline films. The full width at half maxima (FWHM) of diamond
peak is much smaller for homoepitaxial films (4 cm"‘) than polycrystalline films (12
cm“). From a short range high resolution scan of the polycrystalline film as shown in
Fig. 3.14, an interesting point was observed. The diamond peak was observed to shift

by about 2 cm".

Such a shift is usually attributed to strong internal strain in the
material [64]. An important observation to note is that a peak corresponding to
Diamond-like Carbon (DLC) or disordered carbon at 1355 cm‘1 is totally absent which

is another evidence of the superior quality of these films.

The doped diamond films were also inspected by SEM for any changes in surface
morphology. SEM images of typical undoped and doped films are given in Fig. 3.15.
Although the Raman spectra of the two films are similar, a major difference in the
grain shape was detected. A sharp edged channels along the edges of most of the
grains in doped films can be clearly visible. These are caused by defects in the lattice
known as twinning. Also the crystal size tends to increase is size marginally. Beside
these observations, both films appear similar. The degradation of film purity while

doped by using higher quantity of boron powder can be seen in SEM image, as shown

72'

 

 

 

 

 

 

180 12
. Homoepitaxial (110)
170 ~ S ' Homoepitaxial (100)
i A DiamondFilmonSiO/Si ' 1°

. i
2 160 E
" n
3 - 8 A
E 6—- g
D’ ' D
E 150 - ~ E
.8
a 4.3
E 140 - t
E as
E *4 E

130 -

120 1 . 1 1 l 1 1 1 I 1 1 4 2

800 1200 1600 2000
WAVE NUMBm (cm'l)

Figure 3.13 Raman spectra of the diamond films deposited on various surface orienta-

tion of single crystal diamond and oxidized silicon substrates.

 

 

 

 

E

‘E

D I

5‘ .

3.1 I

' l

i . -
V I

=~ :

1: .

U! I

Z .

E .

1332 cm-1
1310 1320 1330 1340 1350 1360
WAVE NUMBER (cm-1)

Figure 3.14 A short range high resolution Raman spectrum of a doped diamond film

deposited on undoped diamond films.

74

 

Lb)

Figure 3.15 SEM micrographs of (a) undoped and (b) doped diamond films.

75

in Fig. 3.16. of presumably heavily doped film deposited on SiOZ. The diamond cry-
stals in the film deposited has poorly faceted cauliflower type structure. This kind of

crystal shape is usually attributed to poor quality diamond films.

7, ' 1‘s i“ I.
,- " M 3":
,8: "g .1" 2.9-17* I»;

'1'

9:.

 

Figure 3.16 SEM micrographs of presumably heavily doped films on oxidized silicon.

3.6 SILICON DIOXIDE ETCHING

Most of the diamond films deposited for device applications were required to be
insulated from the substrate materials. Since silicon was used as a substrate material
and it is a semiconductor with low resistivity, an insulating overlay of Si02 was neces-
sary. It was detected that the oxide was being etched during diamond deposition.
Although there are indications in the literature about this effect [54] but no quantitative

study has appeared so far. Since the presence of insulating layer is important for the

76

devices, Si02 etching was critically observed. A few observations in this respect are
given below.

(1) The etch rate of Si3N4 was found to be about three times that of SiOz.

(2) A comparatively thicker layer of SiOz was observed to be etched slower. This
effect is shown quantitatively in the Fig. 3.17. The oxide etch rate was deter-
mined by measuring oxide thickness before and after diamond deposition. The
thickness of the SiOz before diamond deposition (at abscissa in Fig. 3.17) appears
to have profound effect on the etch rate. Si02 layer on so called "dummy wafers"
which had spent several months in the oxide furnace in nitrogen ambient. were
found to be more resistant to etching than the SiOz layer which received no

annealing in nitrogen ambient.

 

25m I I I I l I I I I l I I I I l r I I F l I I I 1

N
p—A
8
1 I I

 

H
‘1
8
r I

 

Sio2 ETCH RATE (mi/hr)
3 .
8

\O
8
I l I I

111111111111111111

 

 

 

5m 1 g 1 1 l J 1 1 1 l 1 1 1 1 1 1 1 1 1 l 1 1 1 1
0.0 0.5 l .0 l .5 2.0. 2.5
' PRB-DEPOSIT ION SiO2 THICKNESS (tun)

 

Figure 3.17 The effect of pre-deposition Si02 thickness on the etch rate of Si02 dur-

ing diamond deposition.

77

(3) Higher filament temperature leads to higher Si02 etch rate.

(4) The etch rate was observed to increase during depositions accompanied with dop-
ing.

(5) Higher flow rates of carbon containing gases lead to higher etch rates of SiOz.
In addition to these regular patterns of 3102 etching, some non-uniform etching at

random was also detected which is believed to be the cause of hot spots created by the

irregular combination of the above factors.

3.6 SUMMARY

In this chapter, the study to synthesize and characterize semiconducting diamond
film for device fabrication has been presented. The new method of diamond pre—
treatment (DPR seeding method) to enhance the low intrinsic nucleation density of dia-
mond on non-diamond substrates was found to be advantageous over conventional UT
method. Experimental parameters such as reactant gas composition. relative flow rate
of gases, substrate and filament temperatures. pressure and system operation influence
the diamond deposition process. Each parameter was optimized for high purity dia-
mond growth at highest rate. Doping of diamond films to induce semiconductivity,
essential for electronic devices, was implemented by unique method in HFCVD using
solid boron powder as dopant. Si02 etching, an undesirable effect for device fabrica-

tion, was critically analyzed.

CHAPTER 4

 

TEST MICROCHIP FABRICATION

4.0 INTRODUCTION

In this chapter, the design and fabrication of a test microchip and heat flux sensor
array will be described. Appropriate techniques for patterning and metallization of dia-
mond devices on oxidized silicon were developed before an actual chip fabrication
could take place. Three major patterning techniques of diamond films through selective
growth and selective etching were developed. An experimental study was conducted to
find an appropriate metallization scheme which not only provided ohmic contact with
diamond films but also remained physically stable at high temperatures. The heat flux
sensors were fabricated on the cylindrical surfaces of a rods made of various materials.
Due to non-planar surface, sensor fabrication was done through suitable modifications

in planar technology.

4.] TEST MICROCHIP DESIGN

The test microchip is a multipurpose tool to characterize the semiconducting dia-
mond films and analyze the performance of active and passive electronic devices

including MOSFETs and sensors for thermal signals. The size of the chip is lcmxlcm

78

79

and contains four mask fabrication process. These masks are for diamond film pattern-
ing, metallization for contacts and interconnects, passivation layer/ gate insulator depo-
sition and metallization for gate contact in the respective order. Fig 4.1 shows a corn-
posite overview whereas the individual mask layouts are shown in Fig. 4.2. The smal-
lest feature size is about 160 um. Although this feature size looks large when corn-
pared to modern planar CMOS technology of 0.8 pm, it was sufficient for the proof of
concept rudimentary devices built for preliminary research using a new semiconductor
material. Each chip was separated by cleaving or sawing (known as dicing) after the
diamond deposition. The masks were designed using CAD software and fabricated on
i 2inchx2inch high resolution photographic glass plates (manufactured by Kodak) at
Ford’s photographic facility.

The design of the test microchip is based on optimizing the utility of devices for
multiple applications. As shown in Fig. 4.1. it contains several devices including resis-
tors of various size, geometry and orientation, two MOSFETs and a pattern for Hall
measurements. Owing to immediate applications of resistive sensors for thermal sig-
nals, most of the devices are primarily resistors. There are three sets of identical resis-
tors of two each i.e., R1, R2, and R3. One element of first two sets has been grouped
and placed at the opposite sides of the chip to ascertain the spatial doping profile
across the chip. the set R3 consists of two single square resistors to directly measure
the RD variations with strain, temperature, and doping profile in close proximity. The
contact resistance was measured by utilizing the resistors R1, R5 and R6. Note that all
these resistors have the same contact size and width but vary in length. Thus measur-
ing their resistance can be considered as an alternative to the sliding contact technique
[81] for the measurement of contact resistance. The resistor R7 and R6 have been put
in the design as extreme values to study the effect of geometrical aspect ratio on the
observation of physical measurands. The large square pattern in the middle of the chip

IS meant for van der Pauw measurements.

80

Passivation
Film Insulator

Metal #1 - Metal #2

 

Figure 4.1 Composite mask layout of test micro-chip.

 

-
+

(a)

 

(C)

Figure 4.2 Test chip masks for (a) diamond, (b) Interconnect metallization, (c) pas-

 

DC
._1

 

+-

(d)

   

II-

   

 

sivation/ gate insulator and (d) gate metallization.

X2

The width of most of the interconnection lines was 160 um. There are no inter—
connections provided between devices on the chip itself. This was done to be able to
test all the devices in isolation. The size of bonding pads was kept sufficiently large

(lmmxlmm) so that they are visible for microprobe access without magnifying aids.

The third mask is to establish a passivation layer and a gate insulator. The insu-
lator layer covers the entire area of the chip leaving out the bonding pads only. The
purpose was to protect the devices during high temperature operation in an oxidizing
or reactive ambient. The same insulation layer was also meant to act as gate insulator
for MOS structures as well. The fourth and final mask. as shown in Fig. 4.2(d) is for

metallization of gate contacts.

The following sections describe the development of appropriate techniques for
patterning and metallization of diamond films on oxidized silicon for the fabrication of

test chips.

4.2 DIAMOND FILM PATTERNING

The primary source of difficulty in direct patterning of diamond films on non-
diamond substrates is the extreme resistance of diamond to chemical attack. An alter-
native approach to etching an existing film is to actually grow a patterned film through
selective nucleation. Although many attempts employing varying methods of pattem-
ing of diamond films on non-diamond substrates through selective nucleation have
been reported [133-136,138], these have not achieved sufficient selectivity. In most
cases, a focused ion beam has been used to selectively enhance [136] or suppress
[133,134] diamond nucleation sites which resulted in badly damaging the substrate sur-
face. It is clearly undesirable for IC processing. Moreover, some of them were com-
plex to implement [135]. None of these methods was developed to a point that it
could be adopted directly. Therefore, there was a need to develop techniques for pat—

terning the diamond films which should be simple in implementation, achieve good

X3

selectivity and resolution, do not deteriorate the film quality and thickness uniformity
and above all are compatible with the existing integrated circuit fabrication technology.
Consequently, three major techniques which involved both selective nucleation and
selective etching of diamond film were developed. These techniques not only preserve
the high quality of the substrate surface but are also compatible with the existing
integrated circuit fabrication technology.

The patterning techniques developed during this work can be subdivided into two
major groups as shown in the tree structure in Fig. 4.3. that is, selective etching and
selective deposition. The selective etching of diamond films on non-diamond substrates
was implemented for the first time during this research. The specific details of each

method are illustrated below.

Diamond Film Patterning

. l
l |
Selective Etching Selective Nucleation
( in RTP ) I

 

Ultrasonic Treatment DPR Patterning

| - |
Pre-treatment Post- treatment
_ Masking Masking

 

 

Figure 4.3 The hierarchal layout of diamond film patterning techniques.

84

4.2.1 Selective etching

All of the techniques of patterning CVD diamond films on non-diamond substrate
reported so far are limited to the selective nucleation based on suppressing [135] or
simply burying nucleation sites [134]. The only known method of patterning through
selective etching is for homoepitaxial diamond films [137]. For polycrystalline dia-
mond films on non-diamond substrates, a simple patterning technique based on selec-
tive etching of diamond films through oxidation was developed. However, the chal-
lenge was to control the etching at high temperature. A rapid thermal processor (RTP)
was used for this purpose. High purity oxygen was passed over the diamond films dur-
ing the etching process.

The onset of etching was found to be at approximately 650°C, with etch rates
increasing at higher temperature. However, the etching, as observed through SEM, was
highly non-uniform and this non-uniformity increased with etching time. SEM micro-
graphs of the diamond films etched for various length of time have been shown in Fig.
4.4. One can observe that the etching starts with erosion of sharp edges and along the
grain boundaries (Fig. 4.4(a)). Then holes start to emerge through the films mostly
along grain boundaries (Fig. 4.4(b)). These holes reach the underlying substrate while
most of the film is still intact (Fig. 4.4(c)). The cross sectional view of Fig. 4.4(c), as
shown in Fig. 4.4(d), shows this effect more clearly. These holes keep widening with
time, as shown in Fig. 4.4(e) and its cross sectional view in Fig. 4.4(f) where only
skeleton of film is left, and ultimately the whole film disappears. The thickness of the
films was measured through cross sectional views of SEM micrographs and surface
profiler. Fig. 4.5 shows etched thickness against etching time of diamond films oxi-
dized at 700°C under flowing oxygen. From the linear part of the curve, the etch rate
was computed to be ~5 nm/sec. It is also easy to identify that the fast etch rate and the
small thickness of the diamond films (=2um) dictates the use of an RTP. Conventional

heat treatment time cannot be controlled with sufficient precision.

" l3 ‘3 J"
_

 

:1 .

. (e). m
‘ Figure 4.4 Undoped diamond films on silicon etched in RTP at 700°C for (a) 30
seconds, (b) 80 seconds, (0) 120 seconds and (e) 220 seconds. The cross-section view

of films shown in in (c) and (e) are shown in (d) and (f) respectively.

86

 

 

 

 

 

2000 d
@1600 s 1
@1200 r j
o C _
a - .
a 800: ‘
E 400 - :

o 1 1 1 4 1 1 1 ‘1 I 1 1 1 1 I 4 1 1 1 I g 1 1 1
0 50 100 150 200 250

ETCHIN G TIME (seconds)

Figure 4.5 The plot of etched thickness against etching time of diamond film thick—
ness in RTP at 700°C under an oxygen flow of 40 sccrn. The increasing size of error

bar is indicative of highly non-uniform etching.

A number of material including Ni, Pt, Ti, Si02 and Si3N4 were tried as masks.
However, even those metals possessing a high melting point did not withstand the
rapid thermal treatment and delaminated from the diamond film surface or formed their
own oxides (e.g., Ni and Ti) which were hard to remove after the etching. Only 813N4
deposited through low temperature plasma CVD was found suitable for this applica-
tion. After establishing the etch rate, the diamond films, selectively masked with
Si3N4, were etched at 700°C. The sequence of steps followed in this patterning tech-
nique is shown schematically in Fig. 4.6. The Si3N4 masking layer with a thickness of

0.88pm was deposited at 150°C by low pressure chemical vapor deposition method,

Diamond film
deposition

SI3N4 —>
Diamond —>

Silicon nitride
deposition and
patterning

Silicon _>

 

Diamond film
etching in
RTP

Removal of
silicon nitride
mask

 

Figure 4.6 The schematic diagram of diamond film patterning technique through

selective etching in RTP. .

XX

patterned by standard photo-lithography, and annealed at 500°C in nitrogen ambient
for 30 minutes. To prevent the substrate and the masking layer from suffering thermal
shock, the samples were heated to 450°C, still well below the onset of etching, before
raising the temperature quickly to 700°C. The masking layer remained stable during
oxygen etching and was later removed by dissolution in 50% HF. Fig. 4.7 shows
micrographs of a selectively etched diamond film. As seen in the figure, the diamond
film is completely removed from the etched locations without any damage to the film

under the masking layer. An insignificant amount of underetching was observed.

4.2.2 Selective Deposition

The selective deposition was achieved through two main techniques namely ultra-
sonic treatment (UT) patterning and DPR patterning. The patterning through selective
nucleation depends on generation of two areas on the substrate surface which have
many order of magnitude difference in nucleation density. Since the conventional
nucleation techniques lead to the generation of random nucleation sites and wide varia-
tion in grain size on non-diamond substrates, the effects of substrate material, surface
condition and crystal orientation on the diamond nucleation was first explore through

the following two experiments.

In the first experiment, silicon was used as a substrate material and its intrinsic
nucleation density for diamond growth on several crystal orientation was acquired. For
this purpose, grooves and craters of various geometrical shapes ranging from 1 to
lOum2 were created on the silicon (100) surface by micromachining through anisotro-
pic etching using diluted solution of 30%KOH in water. This process exposed many
additional well defined crystal orientations. These substrates were used for diamond
film deposition both with and without pre-treatment by fine diamond powder in an
ultrasonic cleaner. No substantial orientation dependent growth was observed on these

samples.

89

25kt) .267kx

 

Lb) .
Figure 4.7 Diamond films patterned by selective etching at 700°C in RTP. The pat-

terns in (b) is close up view of (a).

90

In the second experiment, diamond growth on the surface of various materials
like polycrystalline silicon, Pt, Ni, and Si3N4 was investigated. No worthwhile
nucleation on untreated substrates was observed. All substrates showed very low
intrinsic nucleation density which substantially increased after UT treatment. Pt and
Ni had relatively higher nucleation density than Si3N4. All the thin films such as Pt.

polycrystalline silicon and Si3N4 were unstable during diamond deposition.

4.2.2.1 Ultrasonic Treatment Patterning

In the ultrasonic treatment (UT) patterning method, the specimens were treated
ultrasonically by fine diamond particles in order to provide a high density of nucleation
sites for diamond growth. The associated patterning, as shown in fig. 4.8, was accom-
plished in two ways. In the first method, the surface was masked with a patterned layer
of Si02 or Si3N4 that shielded the masked regions from ultrasonic bombardment by the
diamond particles. This secondary layer was then removed before diamond growth. In
post-treatment masking, the entire specimen was treated and then selectively masked
just before depositing the diamond film. In the latter case, the masking layer

suppressed the nucleation by burying the nucleation sites during the diamond growth.

In UT patterning, the treatment time depends on the desired nucleation density.
However, in the pre-treatment masking technique, the treatment time along with the
particle size in the diamond powder determines the required thickness of the masking
layer. Electron beam evaporated Ni film (40()0°A in thickness) failed as an effective
mask. Despite being a relatively soft metal, an aluminum film (1.1 pm in thickness)
deposited by thermal evaporation was partially successful such that the masked area
exhibited a considerably lower nucleation density (see Fig. 4.9(a)). After several
experiments with SiOz and Si3N4, a minimum thickness of 0.6um of Si02 or 0.4ttm of
Si3N4 were found to be effective masks for 30 minutes of ultrasonic bombardment

with 0.1um diamond powder. Fig. 4.9(b) shows the diamond film patterned using 0.81

91

m Deposited diamond film.

Surface treated by diamond
powder.

 

8102

   

 

(a) (b)

Figure 4.8 Schematic diagrams of selective diamond nucleation technique by UT

using (a) Pre-treatment masking and (b) Post—treatment masking.

92

 

tb)
Figure 4.9 SEM micrographs of diamond films patterned by UT method using pre-

treatment masking. The masks used were (a) Al film (1.1 um) and (b) thermal Si02
(0.88 um).

93

um thick Si02 masking layer. The ultrasonic treatment leaves a large number of dia-
mond particles sticking to the surface and some of them transfer to the masked area
following the removal of masking layer. Most of them could be removed by ultra-
sonic cleaning in acetone.

In case of post-treatment masking method, both an annealed Ni film (1 pm in
thickness) and a Si02 layer (0.44 um in thickness) were found suitable as masks.
However, the handling associated with the deposition and patterning of the masking
material on the specimen surface after UT degraded the uniformity and the density of
nucleation. Fig. 4.10 shows two SEM micrographs of diamond films patterned through
this technique. A degradation in uniformity of nucleation density is clearly visible in
Fig. 4.10(a). Also the effectiveness of the masking layer was in question for long
duration deposition. Due to these difficulties and the complexity of the process, it was
rarely used for patterning of diamond films on non-diamond substrates. However, it

seems to be the only viable method for selective homoepitaxial deposition of diamond

films [58,139].

4.2.2.2 Diamond-Photoresist Patterning

Patterning by photo-resist mixed with fine diamond powder using standard
photo-lithographic process is referred to as diamond photo-resist (DPR) patterning
method. In this method, fine diamond particles act as seeds for the growth of diamond
only in the patterned areas. However, in the initial experiments, the density of diamond
nucleation and the thickness of diamond film was found to be grossly varying across
the specimen surface. This effect was attributed to uneven spread of diamond seed cry-
stals. Since the grain size and the density of the associated grain boundaries have a
great influence on the properties of diamond films [164] it was necessary to control the
nucleation density itself and its uniformity over the substrate surface. The film proper-

ties such as carrier mobility, thermal conductivity have been shown to degrade with

94

 

 

ing post-

d films patterned by UT method us'

1amon

hs-of d

micrograp

Figure 4.10 SEM

treatment masking. The masks used were (a) Ni film (4000°A), (b) CVD Si02

(4400°A).

95

decreasing grain size [164]. The homogeneity and surface smoothness is found to be
better for small grain size films especially of small thickness. To meet the opposing

demands, a grain size of ~1 um was considered to a good compromise.

When diamond powder was initially mixed in the photo-resist, it formed large
clusters and when seeded by this DPR, these clusters led to highly inhomogeneous dia-
mond film with a very uneven surface. To break these clusters, the diamond powder
was initially dried at 60°C for 2 hours on a heating plate and then mixed in the photo-
resist thinner (because of its low viscosity it forms more homogeneous suspension than
photo-resist itself) and stirred magnetically for 15 minutes followed by 15 minuted
ultrasonic agitation. This suspension was then mixed in the photo-resist and further
stirred and ultrasonically agitated for 15 minutes each. This photo-resist was then
coated on the Si substrate, inspected under SEM, deposited diamond on it, and re-

inspected by SEM. The density of clusters was found to be negligible.

To optimized the deposited film thickness and surface uniformity and control the
grain size, a number of samples, as listed in Table 4.1, using DPR seeding were
prepared. These samples are grouped according to the quantity of photo-resist thinner
and diamond powder mixed in 42 ml of the photo-resist. Following observations were
recorded during SEM inspection of diamond films deposited on samples indicated in

parenthesis.

(1) Higher spin rate leads to lower nucleation density but higher uniformity (group
A).

(2) A photo-resist adhesion promoter such as (HMDS) improved the uniformity of
nucleation density (group A). To improve PR adhesion, HMDS was applied by
spin coating at 3500 rpm followed by baking at 80°C for 20 minutes prior to the
application of DPR.

(3) A higher initial spin speed leads to non-unifonnity and clustering. The initial

ramp up instead of sudden switching to ultimate spin speed appears to give

96

relatively more uniformity in nucleation distribution (group B)

(4) Using a diamond photoresist which has rested for several hours leads to less clus-

ters than a freshly agitated one (group C as compared to group B).

The

difference between the samples of group B and C is that the latter were spin

Table 4.1 Sample preparation for optimization of diamond film thickness and

surface smoothness study. (Photoresist quantity = 42 ml)

 

 

 

 

 

 

 

 

 

 

Gro Diamond Photo-resist S in s ee d ( m)
u powder thinner No. 210,530,751?)
" (mg) (ml)

A 20.2 4 I 500/2000/500
II 500/3000/500
III 500/4000/500

B 41 .7 6 IV 500/4000/500
V 1000/4000/500
VI 0/4000/500
VII 0/3500/500

C 41.7 6 VII] 500/4000/500
IX 0/4000/500

D 122.6 14 X 500/4000/500
XI 500/4500/500
XII 0/4000/500
XII] ‘ 300/4500/500

B 142.6 16 XIV 0/4000/500

 

 

coated with DPR which was not stirred or agitated for about 24 hours.

(5) The spin schedule of R/4000/500 rpm for lO"/30"/5" corresponding to sample XII

is optimum in terms of uniformity of nucleation (group D). The initial spin speed

R represents the ramp up.

97

(6) A larger quantity of diamond powder in a given solution of photo-resist and
thinner gives a higher nucleation density (see Fig. 4.“) 142.6 mg of diamond
powder and 16 ml of photo-resist thinner are sufficient to give optimum nuclea-

tion density i.e., 1 [rm-2 (XIV).

From this study, an optimum quantities of diamond powder, photo-resist and photo-
resist thinner in DPR suspension which leads to the diamond films with a nucleation
density of l um’2 and an optimally smooth surface. was achieved. The film surface
roughness, as measured by a surface profiler, decreased sharply after this experiment.
An SEM image of cross sectional view of a film grown after the above study has been
shown in Fig. 4.12. An optimally smooth surface is clearly visible. However, as the
film thickness grows, the surface roughness was also observed to grow proportionally.
The plot shown in Fig. 4.13 shows this trend. A larger surface roughness of the

thicker films results from the lager grains on the surface.

After an optimum spin schedule for DPR was achieved, the patterning of diamond
films was achieved by patterning the DPR by standard lithography before diamond
deposition. However, some scattered diamond seed particles were observed on the sub-
strate which resulted in growth of diamond in undesired locations. To remove these
residues, the substrates were developed by spraying the photo-developer and then
slightly etching by a method depending upon the substrate surface. An Si02 surface
was etched with buffered hydrofluoric acid (BHF) while DPR served as an automatic
mask. This etching step removed most of the diamond particles left on the surface
after spray development. Schematics of this patterning method are given in Fig. 4.14.
SEM micrographs of diamond films patterned by this method are given in Fig. 4.15.
The thin line parallel to the main pattern in Fig. 4.15(b) in its early phase of growth is
shown in Fig. 4.16. One can clearly see that the line is basically a string of single
grains. This gives a boost to the possibility of fabrication of devices with feature size

near one micron.

98

 

Figure 4.11 SEM micrographs of diamond films deposited on substrates coated with

DPR suspension of (a) 41.7 mg, (b) 122.6 mg and (c) 142.6 mg of diamond powder in

42 ml of photo-resist.

 

99

.q .,p

p awwa-av r1111 «yum ~33.

 

Figure 4.12 The cross-sectional view of a diamond film grown after thickness unifor-

mity and surface smoothness optimization study.

 

§
I

‘1

8
I
l—O—I
I I

a
8

i

§

 

 

AVERAGE SURFACE ROUGHNBSS 0A)
u
8

I 1 I 1 I

 

1%

0 2.5 3.0 3.5
DIAMOND FILM THICKNESS (pm)

Figure .4.13 The effect of diamond film thickness on the surface roughness.

l ()0

Diamond-photoresist (DPR)

_ , v
'. ' . A. . .. ...'....‘...'....i.. 1 :“°
' ‘ ‘ :1‘ ‘ ‘ ’5‘)" ‘ i ' ' ‘52:. ‘ ‘n '3 u\ 'u‘l' :ic'h'n’ ‘i‘v'h‘f '0 Y.:\‘ 'IN‘ Y. o: ’- . 2
u

. a .-

 

Silicon

 

 

II4iIIIIIImunImmm IIIIIIIIIIIIIIIHIHIHI

. ..-‘... nun-u non-n- o

a 1 d .
.
«

 

n....1---.

 

  

Diamond deposition

Fig. 4.14 Schematic diagram of DPR methods of patterning diamond films.

I 3.1. 1
131‘

x !

., . Ink?-

 

 

Figure 4.15 SEM rrricrographs of the diamond films patterned by

DPR method.

102

 

Figure 4.16 SEM micrograph of the diamond film line parallel to the main pattern '
shown in Figure. 4.15(b). The image was taken in the early part of growth.

103

Since the as grown diamond films have large internal stress, their delarnination
from the substrate surface especially after patterning was of great concern. Therefore,
the films were subject to the well known scotch tape test. All the films showed good
adherence with the silicon or SiOZ substrate. Some of the films were scratched with a
sharp knife edge. Fig. 4.17 shows SEM micrographs of a patterned film which was
subjected to scratch test. As visible in the figure, the continuous film is strongly held
with the substrate and only scattered diamond particles could be dislodged by scratch-
ing.

As shown by the micrographs, all the patterning techniques give good selectivity.
The selectivity of DPR patterned films is practically limited by lithography and grain
size in the deposited diamond films. In the case of RTP etched films, it is controlled
by the stability of the deposited masking layer and underetching. Thus, the latter
method gives excellent selectivity on relatively thin films. Looking at them from ease
of implementation, DPR is the best of all. All the patterning methods presented here
are adopted from [C fabrication technology, and so are compatible with it. The choice
of a particular technique for an application depends primarily upon the substrate
material. In general, the DPR method may be employed on nearly all substrates found
in electronic device manufacturing. We note that the post-treatment masking method,
without the ultrasonic treatment, may be used for selective deposition of diamond on a
pre-existing diamond film or even single crystal diamond substrates. An appropriate
combination of these techniques can realize three dimensional structures in diamond

leading to many novel applications.

While all these techniques were evaluated for patterning diamond films on oxi-
dized silicon for test chip fabrication, the DPR was a clear choice by virtue of its ease
of implementation. Micrographs of test chips after diamond deposition at various

magnification are presented in Fig. 4.18.

 

Lb)

Figure 4.17 SEM micrographs of the diamond film scratched with knife edge. The

picture shown in (b) is the higher resolution image of upper portion of pattern shown

in (a).

 

105

 

Figure 4.18 Micrographs oftest micro-chip after diamond deposition using DPR pat-

terning at-various resolution.

106

4.3 METALIZATION

4.3.1 Problem Analysis

Metallization is an important process in the fabrication of an IC. During test chip
fabrication, diamond films are deposited on SiOz. Hence, the contacts are required to
behave ohmically with diamond film and adhere well on both the diamond and $0:
surfaces. The projected testing of these test chips at high temperatures in chemically
and physically harsh environment makes the problem tough. Most of the previous
metallization efforts [142,144,146.147] on CVD diamond films and single crystal dia-
monds remained focussed on to form good ohmic contacts only and very littlewas
said about technological issues such as mechanical stability, adhesion on the diamond
surface, pattern generation and high temperature behavior. Additional requirements of
a good metallization scheme may include but not limited to low resistivity, small con-
‘tact resistance, stability in oxidizing ambients, surface smoothness, ease of formation
and etching/patteming, stability during processing at high temperatures i.e., minimal
inter-metallic and metal to substrate chemical reactions, minimal junction penetration,

and low electromigration, eutectic and surface diffusion.

In view of the requirement for high temperature operation under an oxidizing
environment, high melting point and high oxidation temperature refractory metals
appeared to be the natural choice. However, the choice from these metals for simul-
taneously good adhesion on diamond and Si02 surfaces was difficult. In this regard,
the information available from conventional metallization schemes for relatively high
temperature on silicon ICs [151,160,161] and metallization studies on CVD diamond

films [142,143] were comparatively analyzed.

It is well known that most refractory metals react with SiOz [161]. This property
was seen as an advantage here since these metals could form a layer with good physi-

cal stability. However, their high oxidation rates are undesirable. This aspect is

107

primarily determined by the heat of oxide formation. Using heat of oxide formation
data, it was shown that group IV.A and V.A elements from the periodic table such as
Ti, Ta, Zr, Nb and Hf, by virtue of their lower heat of formation as compared to SiOz,
will reduce SiOz and form a metal oxide [151]. This process leads to a strong chemi-
cal bond with SiOz at relatively lower temperature. It has been shown from the T i-O-
Si phase diagram constructed from empirical data that Ti is not stable in contact with
SiOz and reacts with it at temperature around 550°C to form TiO and Ti3Si3 which are
in equilibrium with SIC; even at 950°C [160]. On the other hand, metals like Au, Pt,
W and Mo that form an oxide with a heat of formation higher than SiOz do not reduce
Si02 and, therefore, have poor adhesion on it. Therefore, refractory metals such as Ti
and Ta, with their relatively higher melting point, are most suited for a base metal
layer on SIOZ in an envisioned multi-layer metallization structure. However, since the
SiOz reduction process continues until the supply of either of the reactants lasts, this
base layer should be thin enough to form only the required "glue" layer between
metallization and SIOZ and not deplete the 8102 layer completely. In addition, the
environment during sintering process should, as far as possible, be oxygen free so as to
prevent the oxidation of the metal layer by interaction with the ambient oxygen. Thus
there is a need to protect the base layer by another protective passivation layer. An
easy way to accomplish this is to lay the top metal layer before sintering instead of
laying just a sacrificial layer.

Although most metals form pseudo-ohmic contact with small grain (Slum)
polycrystalline diamond films, it may not be true in case of large grain polycrystalline
diamond films [162] and single crystal diamond surface. From this perspective, among
the metals being considered, Ti has been shown to make ohmic contact with both sin-
gle crystal bulk diamond [152] and homoepitaxial diamond films [58,142]. Both Ti
and Ta are known to react with diamond at temperature <800°C to form their respec-

tive carbides [47]. Therefore, they can be considered appropriate for metallization.

108

The requirements of the top layer metal should include low resistivity, high melt-
ing point, oxidation resistant and high melting point of any alloys it forms with the
base layer metal. There exist a number of choices such as Ni, Cu, Au, Pd, Mo, W,
and Co. Out of all these metals, Pt stands the best chance due to its best overall com-
bination of desired properties. These include high resistance to oxidation, high melting

point, and fairly high melting point of alloys formed with Ti and Ta.

Pt has been reported to form three compounds with Ta namely
Pt4Ta, Pt3Ta, and PtzTa [163]. Although the melting points of these compounds are
not precisely available, they are well above 1000°C. Pt4Ta is found only above
1500°C. Pt forms three compounds with Ti: TiPt3, TiPt, and Ti3Pt, with melting points
of l950°, 1830°, and 1370°C respectively [163]. Although, Ti3Pt has a relatively low

melting point, it is expected to remain mechanically stable at 1000°C.

4.3.2 Experimental Results

In general, for two layer metallization, the thicknesses of lower and upper layer is
in the range of 300—600°A and 2000-5000°A respectively. As a starting point, a combi-
nation of Pt(3000°A)/Ti(500°A) was selected for the test chips. Both metals were
deposited using RF sputtering system. The pattern was generated by shadow mask. The
mask was fabricated on thin Cu plate at the Ford photographic facility. The test chip
after metallization is shown in Fig. 4.19. SEM micrographs of the as deposited metal
film on the diamond film surface and around the edge of the diamond film are shown
in Fig. 4.20. The diamond film morphology appears to be completely preserved. Also
the metal film appears to be featureless and very smooth. Due to deposition of metal

normal to the chip surface, the diamond film step coverage is not good.

These test chips were again tested at 1000°C for the physical stability of metal

)--3

film. The were heated for 15 minutes in vacuum (~11 torr). The observations of

optical microscope and SEM inspection after heat treatment are as follows:

 

109

 

 

Figure 4.19 Test micro-chip metallization. The patterns on was generated by using
shadow mask. The pictures in (b) and (c) are higher magnification images of (b). The

diamond pattern in (c) on the left is 160 tun.

110

 

Figure 4.20 (a) Diamond film pattern on test chip with as deposited metal Pt
(3000°A)/ Ti (500°A). (b) The top view of metallized diamond film surface in (a).

(a)

(b)

(C)

(d)

111

The metal film appeared to be either melted or at least softened up during heat

treatment.

The metal may have reacted strongly with diamond and dissolved it partly (see
Fig. 4.21(a) and (b)) to an extent that the diamond film almost lost its morphol-
ogy altogether. Possibly Ti reacted with the diamond film to form titanium car-

bide.

The film formed small islands which were completely dissociated from each other
in locations where the metal film thickness was relatively small i.e., in the boun-
dary regions (see Fig. 4.21(c)). In the areas with a thick metal film, a similar
behavior was observed but the islands were barely joining together (see ‘Fig.
4.21(d) for high resolution view). Also the surface of the metal film broke open

and left void spaces.

The metal appeared to pull away from the diamond film surface. This is truly
visible along the diamond film edge and around scattered diamond crystals on the

substrate surface (see Fig. 4.21(e) and (f)).

From these observations the following deductions were drawn.

(a)

0))

Ti consumed diamond excessively during its reaction with diamond in forming its
carbide and since this reaction continues (at temperature higher than 550°C) until
the Ti layer is completely depleted, the thickness of the Ti layer used was more
than desired. The same was true on the Si02 surface. There Ti reduces SiOZ in
the process of forming its own oxide. This is clearly not desired because the SiOz
thickness must be saved to provide sufficient insulation from the underlying sem-
iconducting silicon. Since Ti is used here only to act as a ’glue’ layer, a thinner
layer may be sufficient.

The initial thickness of the Pt layer was not sufficient. A relatively thick Pt film

is more stable at high temperature as can be seen from Fig. 4.21(c) and (a) in

case of Si02 substrate and a small patch of thick Pt film on top of diamond film

112

 

Figure 4.21 SEM micrographs of the metallization (Pt (3000°A)/ Ti (500°A) on test

chip heat treated at 1000°C.

respectively.

(c) The annealing should be done under better vacuum (<1 mtorr range) so that there
will be least amount of water possible to be absorbed by Ti to form its oxide.
Also, a fast temperature rise to 1000°C leads to dissociation of ambient water
molecules. Pt is known to absorb hydrogen at high temperature (>600°C) and

become unstable.

(d) Since all the required reactions namely formation of titanium carbide, oxide and
silicide occur at relatively low temperature ($550°C), the annealing at 1000°C is

unnecessary for metal films.

On the basis of above observations, the metallization scheme was changed as folloWs:

(l) Decrease of Ti layer thickness from 500°A to 100-150°A.
(2) Increase of Pt layer thickness from 3000°A to 1.2 pm.

(3) Reduce annealing temperature from 1000°C to 600°C.
The test chips, after metal deposition, were annealed at 600°C for 45 minutes. The
pressure was estimated to be in the range of 10'7 torr. An SEM inspection of this
metal combination showed no sign of degradation. The annealing time was adjudged
by in situ monitoring of the resistance of the metal film. The metal film resistance
decreased during annealing and stabilized at one point. An example of the change of

resistance with time is shown in Fig. 4.22. The resistance closely followed the relation
R(t) = [R(()) — R(45)] e’ "T + R(45) (4.1)

where R(t) is resistance after annealing of time t measured in minutes. The time con-
stant 1: of 7 minutes was determined through curve fitting. Thus, an annealing of 35
minutes (five time constants) was considered to be sufficient for all later samples. It is
recognized that part of the resistance change might be contributed by the diamond
resistor itself but a stabilization of the total resistance is a clear indication of stabiliza-

tion of the resistances of both the diamond film and the metal lines.

114

 

 

1111111111111111111111

 

 

 

Figure 4.22 Resistance of a metallized diamond resistor with time during annealing at

600°C.

115

After annealing at 600°C, no appreciable change in metal film appearance was
observed through SEM. After this the test chips were heated to 1000°C for 6 minutes
under vacuum and again inspected through SEM. SEM micrographs of this metalliza—
tion experiment are shown in Fig. 4.23. The metal appeared to be partially melted or
undergone a solid phase crystallization. The metal on diamond film surface showed
complete continuity. The metal was physically stable and erosion of the diamond film
was minimal. The morphology of the grains appears to be preserved. Also, no island
formation or breakage in the metal film was observed. The tendency of the metal film

to pull away from the diamond film also appears to be reduced significantly.

The physical adhesion of the film was tested by the scotch tape test. Also itwas
scratched by a knife edge. The metal film showed excellent adhesion on both the SiOz
and diamond film surface. The contact resistance was also measured before and after
annealing. lt decreased by a factor of 3.85 on the average. The formation of titanium
carbide and diffusion of Pt into the diamond film are possibly responsible for this
behavior. Overall this scheme appears to be successful for metallization of diamond

films on oxidized silicon substrate.

Although suitable for the present purpose, the thick platinum layer was not
appropriate for the lithographically masked wet etching process. The photo-resist mask
was unable to withstand the etching process long enough for the pattern to form and
either dissolved or delaminated. To make the metallization scheme universal, the plati-
num film thickness was reduced to 8000°A. This thickness could be etched with wet
chemical etching solution with photo-resist masking. At the same time any change in
its operational behavior was not detected. SEM micrographs on the test chip after this
metallization and heat treatment at 1000°C for 8 minutes are shown in Fig. 4.24. The
erosion of diamond is still minimal. The cross sectional step coverage is still lacking
by the same amount. However, it provided the necessary contact and no degradation in

electrical measurements or through visual inspection was observed. To verify the

116

 

LC»)

Figure 4.23 SEM micrographs of metallization Pt (1.2 um/ Ti (100°A) on test micro-

chip after annealing at 1000°C. (a) The top view with diamond pattern in left half, (b)

stray diamond crystals and (c) a 60° view around a metal contact on diamond.

117

 

\

Figure 4.24 SEM micrographs of metallization Pt (8000°A/ Ti (100°A) on test micro-
chip after annealing at 1000°C. (a) a 60° view around a metal contact on diamond and

(b) stray diamond crystal.

118

compatibility with IC technology, the gold wire bonding was made on the metal bond—
ing pads before and after heat treatment. The gold wire adhesion was good and bond-
ing formed a proper connection. A micrograph in Fig. 4.25 shows a wire bonded con-

nection after heat treatment at 1000°C.

;.rars.u:.»-..§g

;.
,
:1
,,
‘..

I

 

Figure 4.25 Gold wire bonding on the test chip bonding pad composed of Pt (8000°A/
. Ti (100°A).

119

4.4 FABRICATION OF HEAT FLUX SENSORS

In order to measure abnormalities in the flow of heat associated with the shock
wave in the wind tunnel, special temperature sensors at very high spatial resolution
(few um) operating with ultrafast dynamic response time (few us) may be required.
For this purpose, the sensors should be attached to the surface of a highly thermal
insulating body which is small enough and of such shape as not to interfere with the
aerodynamics. The sensors should be able to withstand physically harsh environment
in the wind tunnel. Due to these stringent technological and operational demands, most
conventional temperature sensors are rendered unsuitable. The remarkable combina-
tion of physical and electronic properties of chemical vapor deposited (CVD) diamond
films especially high thermal conductivity, chemical inertness, radiation immunity, and
wide energy band gap propose their possible application as high speed temperature
sensors operating at elevated temperatures under such an environment. The present
developments in the deposition, doping, patterning, metallization and passivation pro-
vide the basic fabrication technology for a micro-miniaturized temperature sensor
array.

For this application, a cylindrical shaped substrate about 7.6 cm long with a 1 cm
diameter was considered appropriate. However, placing the sensors on the curved sur-
face of a cylindrical substrate made of high thermal and electrical resistivity material
presents a major technological problem. A three dimensional lithographic technique
was required to implement the DPR diamond and metal pattern generation. As a first
attempt to solve this problem, the conventional two dimensional lithographic tech-
niques were suitably modified to implement a close approximation to this three dimen-
sional problem.

Of the many problems encountered, the uniform thickness of photo-resist (and

DPR) and transfer of the pattern on the curved surface of the rod substrate were most

critical. Solving the problem of holding of rod at high spin speed and controlling the

120

fluid dynamics of the liquid photo-resist for a uniform coating over a small angle
(about 35°) required extensive experimentation. To transfer the pattern onto the
curved surface of the substrate, flexible masks were prepared. The masks for diamond
film and metal patterning are shown in Fig. 4.26. The feature size is 150 um. Even
though farly large, the sensor size is sufficiently small for initial feasibility study. The

alignment was done manually using a low magnification (15X) microscope.

A requirement of heat flux sensors is that they should be placed on good thermal
insulators. Hence, diamond deposition on several materials including quartz, alumina,
vycor glass and thermally oxidized silicon was attempted. Diamond films showed
acceptable adhesion only on the thermally oxidized silicon rod and this was therefore
selected as the substrate material for present application. SEM micrographs of a 12
diamond film sensor array on silicon rod are shown in Fig. 4.27. A high density of
scattered diamond crystals, as visible in the figure, resulted from the failure of their
removal due to non-polished surface of the rod substrate. The seed crystals were held
in the channels. present on the substrate surface during the lithographic process.
Nevertheless, the density of such particles is not sufficient to form a continuous path

between adjacent sensors and will not interfere with electrical measurements.

4.5 SUMMARY

The fabrication process of test micro-chip and heat flux sensors has been
described. Appropriate patterning techniques for diamond films and high temperature
compatible metallization scheme were developed. Of the three patterning techniques,
DPR method was considered to be the most suitable for test chip fabrication from the
perspective of selectivity, resolution and the ease of implementation. The metallization
of test chip was done with thick Pt metal film (0.8-1.0 pm) on top of a thin Ti layer
(100-150°A). This metallization scheme is not only high temperature compatible but

also compatible with conventional chemical etching and wire bonding processes.

450um>1 |<

.. rrrrr .

)

    

(

Z‘I 0+

 

 

 

 

 

 

 

 

 

 

 

 

1 ”I 01 1. 1 .1 91 .1 VI .1

Figure 4.26 Mask set for diamond film heat flux sensors. (a) diamond and (b) metalli-

zation mask.

 

 

 

 

 

 

(b)

Figure 4.27 SEM micrographs of diamond film heat flux sensors on silicon rod. (b) is

a high magnification images of (a).

CHAPTER 5

 

ELECTRICAL CHARACTERIZATION

5.0 INTRODUCTION

For well known and widely used semiconductors, the fundamental material pro-
perties are generally well characterized. However, for a new semiconductor such as
diamond, the properties such as carrier and impurity concentration, resistivity and car-
rier mobility are required to be measured and relations among various parameters
developed before using it in any device. At present, these properties are either not
measured or there is a large scatter in the reported values. It is possibly due to experi-
mental difficulties involved in making these measurements as well as polycrystalline
nature of CVD diamond films, the electrical properties of whom are sometimes depen-
dent on deposition conditions. Therefore, fundamental electrical properties such as
resistivity, carrier mobility, carrier and impurity concentration were quantitatively
measured or evaluated and is the subject of this chapter. Since this research is pri-
marily aimed at developing temperature sensors, these measurements were made over a
wide temperature range. For this purpose, a special high temperature characterization
system was designed and built. In addition, many technological issues such as control
of doping and its uniformity and high temperature effects on the diamond films have

also been addressed.

123

124

5.1 PRELIMINARY ANNEALING STUDY

To estimate the effect of high temperature treatment, test chips were heated at

high temperatures and temperature response of resistivity was monitored. The tempera-

ture response was acquired by measuring the resistance of a selected resistor in situ

using the experimental arrangement shown in Fig. 5.1. Fig. 5.2 shows the temperature

response of the resistor after annealing at the indicated temperatures at a pressure of

10'7 torr. All the curves correspond to a single sample. The data for this figure was

collected through an annealing experiment as follows:

(i)

(ii)

The test chip was heated to 300°C. While at 300°C, the resistivity was monitored
for any change with time. No change in resistivity occurred and sample-was
cooled down to room temperature. Now the temperature response of resistivity
was measured up to 300°C. The measurements were repeated both with increasing
and decreasing temperature. This data generated the curve corresponding to the

annealing temperature of 300°C.

The chip was heated to the 500°C. While at 500°C, the resistivity was monitored
which was found to decrease gradually and stabilize slowly to a constant value in
~12 minutes. The sample was kept at the annealing temperature until the resis-
tance stabilized. Then the sample was cooled down to room temperature. Now
the temperature response of resistivity was measured up to 500°C. The measure-
ments were repeated both with increasing and decreasing temperature. This data

generated the curve corresponding to the annealing temperature of 500°C.

(iii) The same procedure was adopted for 600 and 1000°C. While at these tempera-

tures, the resistivity was found to change gradually but stabilized after 10 and 8

minutes, respectively.

The observations for the foregoing annealing experiments are as follows:

125

 

    
  
   
 

 

 

Nano Programm-

voltmeter able
current
source

 

 

 

 

 

HPIB / [BEE-488 bus

Computer

 

 

 

Fig. 5.1 Schematic diagram of experimental set up for I-V measurements.
The same arragement was used for two wire resistance

 

 

 

 

 

 

 

 

measurements.

“FEW...“,...,...fir..,...,...,.,,,...,,j,3
’ . 0 300°C ‘
10‘; :1 500°C 1.
A i \ . I 600°C 1
g t ‘ _‘ ’ O 1000°C ]
v 10° r \ —.
>1 \ = 3
E 10"? —
a : ' . ‘
a: - _
10"? 2
i i
‘ -1
10-3 1 1 111 1 1 1 1 1 111 11 1 1 1 111 1 1 1 1 1 11. 1_1 1 1 1 1141 1

0 l 00 200 3 00 400 500 600 700 800 900 l 000

TBMPERATURE°C

Figure 5.2 The temperature response of resistivity of diamond films sensor annealed at

different temperatures. The annealing temperature is also indicated.

(1)

(2)

(3)

126

There is no effect of annealing up to 300°C on the resistivity. Fig. 5.3 shows the
temperature response of resistance for two cycles up to 300°C. Both curves are

reproducible.

Once sample is heated to a temperature T>300°C for the first time, the resistivity
decreases gradually and stabilizes slowly. Once cooled down to room tempera-
ture, the room temperature resistivity is found to be lower than the value prior to

heating.

Once a sample undergoes annealing at a specific temperature, the temperature
response of resistivity up to that annealing temperature is found to be reproduci-
ble. No further change in resistivity was observed if the sample is kept below- that
temperature. Fig. 5.4 shows this reproducibility of temperature response in the
temperature range of BOO-1273 K and 77-370 K after annealing for 35 minutes at

600°C and 8 minutes at 1000°C.

Based on these observations, following deductions were drawn.

(1)

(2)

(3)

No annealing is required for temperature sensors if the maximum operational tem-

perature is not to exceed 300°C.

For a temperature sensor to operate beyond 300°C, an annealing at its maximum
operational temperature is required. The time duration of annealing will be deter-
mined by the time required for the resistivity of the sensor to stabilize at the

annealing temperature.

The room temperature resistivity of doped diamond films decreases after heating

at temperature >300°C. This effect is clearly visible in Fig. 5.2.

The decrease of room temperature resistivity due to annealing at temperature T>300°C,

as shown in Fig. 5.2 is contrary to what has been found for undoped films [99,159].

For undoped diamond films the room temperature resistivity was found to increase

 

127

 

 

 

 

 

 

 

 

 

1200 ' ' Y F T Y T T T T ' I I r r r r r v rfi' T r r l
0 CYCLE] ‘
1000 ~ I CYCLEZ —
J
A J
g 800 - 4
m ’ a
U L
E 600E
E-
2 »
U)
a 400 ~
200 r
01111114111111L1L111IJLILIJAIJ
0 50 100 150 200 250 300

TEMPERATURE ( C)

Figure 5.3 The reproducibility of temperature response of resistivity of unannealed

diamond temperature sensor over a temperature range of 300-573 K.

 

 

 

 

 

 

 

 

 

1 02 E T 1 7 1 T 1 ‘ 1 f f r l V 1 Y 1 r 1
E
” I ' CYCLE l ‘
10l 1 , , . CYCLE 2
a F ‘
fi 1*
[:1 10° :— j
0 E .
5 .
[— _
2 10'1 : _
E E
M i
1 0-2 1; I _,
1 0-3 1 1 1 1 1 g 1 1 1 1 1 1 1 1 1 1 1 1 1
0 l 00 200 300 400 500 600 700 800 900 l 000
TEMPERATURE (°C)

 

100 I l T j I Y T T V I 1 V V 1 I Y I I l I I Y I 7 I l T l T
.
_

 

 

 

 

 

 

 

 

>-
E
[-
E
a”:
M

0 CYCLE l

f I CYCLE 2

l 1 1 1 1 L 1 1 1 1 l 1 1 1 1 1 1 1 1 1 l 1 4 1 1 l 1 1 1 1
70 120 170 220 270 320 370

TEMPERATURE (K)

Figure 5.4 The reproducibility of temperature response of resistivity after annealing in
temperature range (a) 300-1273 K and (b) 77-300 K.

129

with annealing beyond 300°C. Any report regarding change of room temperature

resistivity of doped diamond films, whether increase or decrease. was not found in the
literature.

The interest of present research is using diamond film temperature sensors up to
1000°C. For this purpose, the test chips were annealed at 1000°C and their resistivity
was monitored in situ. The change of resistance of selected resistors with time, while
at 1000°C, of two test chips with measured Hall hole concentration of 1.02 ><1016 cm—3
and 2.65 ><10l7 cm’3has been plotted in Fig. 5.5. From the figure, it is clear that the
change in resistance became negligible after 8 minutes. Also, the time taken by the
resistivity to stabilize appears to be independent of Hall concentration. Hence, it was

decided to anneal all the sensors at 1000°C for 8 minutes before recording their tem-

perature response.

 

10.8 6.25

 

I . 1.02111016 -
‘ I 2.651110” 36.20

 

 

1 0.7

 

: 6.15

: 6.10

RESISTANCE (k9)
S
0‘

RESISTANCE (k9)

10.5 1
’ _ 6.05

 

 

 

 

l 0.4 6.00
0 l 2 3 4 5 6 7 8 9 l 0
TIME (Minutes)

Figure 5.5 The time response of resistivity of two diamond film resistors during

annealing at 1000°C. The measured Hall concentration for each sample is also indi-

cated.

5.2 ELECTRICAL MEASUREMENTS

5.2.1 Overview

Most of the electrical measurements were aimed at characterization of diamond
films for applications in temperature sensors. Primarily, they consisted of resistivity,
current-voltage (l-V) and Hall measurements at various temperatures and in different
ambients. The purpose was to acquire fundamental properties of p-type diamond films
such as resistivity, carrier concentration and carrier mobility as a function of tempera-
ture. Two groups of samples are used in electrical measurements. Group one consists
of test chips which were annealed at 600°C and 1000°C for 35 and 8 minutes, respec-
tively. Moreover, boron doped diamond film for test chips were deposited on oxidized
silicon with a buffer layer of undoped diamond film. Group two samples were not
annealed and consisted of rectangular resistors or square pattern. The doped diamond
films for these samples were deposited directly on oxidized silicon. They will be
referred to as ’discrete’ samples in the following discussion. The measurements on
these samples have been summarized in Table 5.1. All the measurements in the tem-
perature range of 77-575 K were made in commercially manufactured systems. A dedi-
cated system was designed and built, during this research, to perform annealing and
measure resistivity of diamond sensors in the range of 300-1273 K. A schematic

diagram of the system is shown in Fig. 5.6.

5.2.2 Current-Voltage Measurements

Current-voltage (l-V) measurements were made to observe the behavior of electri—
cal contacts. A programmable voltage source and an electrometer were placed in series
with the resistor for these measurements. The measurements were made over the tern-
perature range of 77-1300 K. Fig. 5.7 shows the IN measurements at selected tem-

)-3

peratures in the temperature range of 77-300 K at a pressure of 1( torr made on

131

Table 5.1 The organization of electrical measurements for the characteriza-

tion of p-type diamond films.

 

 

 

 

 

 

 

 

 

Temperature range Measure- Sample Metalliza- .
. Ambient
(measurement) ment type type tion
1 77-350 K l—V, p, Test chips Pt / Ti Vacquum
10" t
1411.1) Square Hall Al orr
pattern
I] 300-575 K l-V, p Rectangular A] Static air
resistors
1H 300-1300 K I-V, p Test chips Pt / T1 Vacuum
10'3 torr
300- 1300 K l-V, p Test chips Pt / T1 Vacuum
. . 10'7 torr
Homoepitax1al No metal

 

 

 

diamond film

 

 

 

 

132

 

. —> To vacuum
Vacuum PumP

 

 

 

‘—
Gas inlet

 

Heater
\ /

Test chip . l l" ’ ocouple

' 4—Heat shield

 

 

 

 

. 'cro-
- . ulators

  

 

 

1‘
\fl

 

Instrumentation port
(coaxial connectors)

Fig. 5.6 Schematic diagram of the high temperature characterization
sys em.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

15 T 1 T 1 j 1 T T T T 1 T T T 1 T 1 7 4
" A
A
12*— A 1
» 0 86K ‘1‘“...
9*— . 180K “‘.-. T
, A 250K ‘A .. ..O:
2 6: 1‘:-.-o‘...
. ..
3 3~ .‘E:.o’ *1
E ’ .21. '
0
m . .o!. 1
gab ..o‘::! E
U P 0". I-:‘ 1
'6’ .0 . .I “ "‘
1. ..I“ _‘
.9 E-.-I“A
I A
-12» A q
-lsi‘f 1 1 1 1 1 ;L 1 ¥ 1 1 1 1 1 1 1 ¥
-10 -8 -6 -4 -2 0 2 4 6 8 10
(Q ) VOLTAGE (Volts)
30 fl F l I 7 r T I I 1
24E ‘1-
’ 0 93K “‘ -
18" . 180K “ -.{
1 1 250K 1‘ .1-
A12_ ‘A‘- I. ..1
. A .0 .
_ A I O
§6* 0.35::00... '1
O
0 j'.‘
I111
F 00 .::§,. 4
‘6’ ... .0 .:“ .4
U L... .I ‘ ..
-12. '--.I“‘A _
I. A
~181 “A 1
44:1“ _
’ 1
_30 1 1 1 1 1 1 1 1 1 a 1 1 1 1 1 1 1 1
-10 -8 -6 -4 -2 0 2 4 6 8 10
VOLTAGE (Volts)

°(b

Figure 5.7 l-V characteristics of a test chip at selected temperatures in the temperature

range 77-300 K (a) before and (b) after annealing at 1000°C.

134

same test chip before and after annealing at 1000°C for 8 minutes. All the curves
appear fairly linear. The only difference between the curves at same temperature is the
change in slope that is due to change in resistivity during annealing process. Any
deviation from the linearity at higher voltage (and current) is due to heating effect

which will be discussed in next chapter.

A similar behavior was observed in case of l-V measurements at selected tern-
peratures in other temperature ranges as shown in Fig. 5.8. The measurements shown
in Fig. 5.8(a) and (b) were made at selected temperatures on discrete and test chip
samples in the range of 300-573 K (in static air) and 300-1273 K (at a pressure of
10’7 torr), respectively. Though metallized with different metals, i.e., Al and Pt/Ti
respectively, both type of samples showed fairly linear behavior over the respective

temperature ranges.

5.2.3 Conduction Type Measurement

The type of conduction was determined by two methods. First, by observing the
sign of the thermal EMF i.e., the Seebeck effect, using the hot probe method [81].
Second, by observing the sign of the Hall voltage. P-type conductivity was found
irrespective of doping level, temperature, diamond film thickness or type of sample
(annealed or unannealed). Even the undoped diamond films showed a p-type conduc-
tivity which is indicative of the absence of substantial amounts of n-type impurities or
defect states. The sign of Hall voltage consistently showed this behavior over the tem-

perature range of 77-300 K.

135

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

200 I I I I l I I I I I I I I I I I I I
160 ' —
0 ROOM TEMP. -
120 ' O 150‘C -
’ I 250‘C

A 30 j 1 300'C ]

3 40 ~ -1

I" r '1

0
-40 r -

U -80 - -
-120 - ..
-l60 _
.200 l 1 I 4

-20 -10 0 10 20
VOLTAGE (Volts)
200 I I I I 1 4 l I 01 T I 1 T f I .
.. I . —1
Cl 20°C 0 I. ’3’, ‘
A 360°C 0 .' .0 .
. 500°C 0 I. .. 7
A 100 f - 598 °C 0 11 .° ‘
< . F V 658 0C O .l Q. j
E _ 0 880°C .0. 1
5 . A A A A ‘ ‘ ‘ ‘ -
O ‘ s 3 = = = = = = : 2 .-. ". _ = 2 2 2 2 2 2 2 2 I

g 1 A ‘ ‘ o.

I ‘ ‘

O ‘ ..

» o. I O —

U -100 '— ... .- O '—
~ . . O 4

L .. .- -I

- .9 .I O -

. .2m 1 l 1 l J A 4A 1 l 1 l 1 l 1 1
-IO -8 , -6 -4 -2 0 2 4 6 8 10
VOLTAGE (Volts) .

Figure 5.8 I-V characteristics of (a) discrete and (b) test chip samples over temperature

range 300-573 K and 3100-1273 K respectively.

5.2.4 Resistivity Measurements

The importance of resistivity, in the present case, is due to the fact that the opera-
tion of diamond sensor for thermal signals will depend on the resistivity change with
temperature. The resistivity measurements were also employed to characterize the in
situ annealing behavior of diamond films as well as metal contacts and doping unifor-
mity. The resistivity p of the p-type diamond films was determined by two methods.
First, it was computed from the two terminal resistance measurements using geometri-

cal dimensions as follows:

_ RxWxD
L

(5.1)
where R, W, D and L are resistance, width, mean thickness and the length of the dia-
mond film resistor respectively. This method was employed mostly for high tempera-

ture measurements using the experimental arrangement shown in Fig. 5.1.

The second method used for resistivity measurement was the Van der Pauw
method [167]. It was widely used due to several distinct advantages. A single method
and experimental set up is sufficient for the measurement of resistivity and Hall effect.
1t is particularly useful for the thin samples of arbitrary shape and size. Unlike other
methods, it does not depend on various correction factors in measurements for finite
sized samples. This property is especially valuable for diamond film patterns of very
small dimensions. Furthermore, the contact resistance is automatically taken care of,
as in the case of four point probe measurement. The Van der Pauw [167] method was
mostly employed for measurements over a temperature range of 77-573 K. In this

method, the resistivity p of sample with thickness d is expressed as follows:

RAB,CD

1rd
P = Emmet) + RECDA) f (5.2)

RECDA

where the resistance R AB,CD is defined as the potential difference VD - VC between

the contacts D and C per unit current through the contacts A and B as per geometry

137

shown in Fig. 5.9. The current enters the sample through the contact A and leaves it
through the contact B. RBCDA is similarly defined. As indicated in the above equa-
tion,fis a function of the ratio RAB‘CD/RBCDA only and if this ratio is. close to unity. f
can be approximated by the formula

= [RABCD ' RBC.DA]2M _ [RARCD “ RBC.DA]4{(l_nl)i _ M} (S 7)

RARCD + RBC.DA 2 RAB.CD + RBCDA 4 12 ~ ._

It is a slowly varying function of the ratio RAR‘CD/RBCDA. In practical measurement, it
was evaluated to better than 99% precision by the following algorithm [165].

= —2 log 2
log(A) — log(1—A)

A
'Jt
l.»
V

f
where A=A20 which is found by the following recursive relation
A,=(1—A,_l)k fori= 1.20

with A0 = 0.5 and k = RAB.CD/RBC.DA or RBC.DA/RAB,CD whichever is smaller than 1.

Q9

 

Fig. 5.9 The contact designation of the Hall Pattern on the test chip for
Van der pauw measurements.

138

1f the contacts are placed so that they are symmetrical about a line through any pair of
non-adjacent contacts, RAB‘CD/RBC‘DA= 1. When symmetrical contacts are used, any
deviation in the ratio is a measure of homogeneity in the resistivity. This is one reason

for making the contacts symmetrical on the test chip.

The room temperature resistivity of a number of test chip and discrete samples
doped by using various amounts of boron powder is given in Fig. 5.10. The two
separate curves in the figure correspond to test chips and discrete samples. The
difference between two types of samples has been described in section 5.2.1. As
expected, the resistivity decreases with increasing quantity of boron powder used to
dope the films. There is an almost constant difference between the two resistivity
curves. The resistivity of test chip samples vary from 64 to 0.3 Q—cm corresponding to
O-2.6 mg of boron powder quantity used for doping these samples. In case of discrete
samples, the resistivity change is in the range of 98-20 Q—cm corresponding to 0.2-1.2
mg of boron powder. Although there is some scatter in the data, the controlling of

doping through quantity of boron powder appears to hold sufficiently.

The resistivity measurements (on In p scale) over the temperature range of 300-
573 K (on 1000/F scale) for four discrete samples with different Hall concentrations
are shown in Fig. 5.11. The reason for selecting a 1n p vs 1000/T axis system is to be
able to extract information on boron activation energy from the slope of the resistivity
curves. In this figure, each resistivity curve displays a single slope over the entire
temperature range. The slope of the curves decrease with increasing Hall concentra-
tion. The difference between two intermediate curves is not appreciable due to the
small difference in Hall concentration. In case of the test chips, the resistivity was
measured before and after annealing over a temperature range of 77-573 K and 77-
1273 K respectively. Typical resistivity measurements on a test chip are shown in Fig.
5.12. The resistivity measurements corresponding to temperature range of 300-1273 K,

as shown in Fig. 5.12(a), show two distinct slopes. The first slope exists from room

102

 

.1
1111

 

0 TEST CHIP
DISCRETE

   

N
I

 

u—o
<2.
1

1111

1—1

G
O
1

RESIST IVITY (Q-cm)

2 1.

 

 

l 0-1 lllllllll l ““““““ 1%, L J 1 1 1 1 1 1
0 l 2 3
BORON POWDER QUANTITY (mg)

 

Figure 5.10 The room temperature resistivity vs quantity of boron powder used for the

doping of diamond films during deposition.

TEMPERATURE (K)
573 475 373 500
1 l ' ‘

 

102_

 

L411

 

 

 

 

 

 

 

l .5 2.0 2.5
1000” (K")

Figure 5.11 The temperature response of resistivity of discrete samples over tempera-
ture range of 300-573 K. The room temperature Hall concentration of each sample is

also indicated.

 

140

TEMPERATURE (K)
1273 873 573- 300
I

 

§

fl
II'YTT I '7‘!

1.
1.
—
1-
p
.-

 

RESISTIVITY (II-cm)
8

 

 

 

 

 

 

 

10‘1 y K 7:.
E x‘ a
102; ’ . POST ANNEAL _j
/ - PRE ANNEAL ;
C /
10.3 1 /1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 n 4 4 1 1 1 1 1
0.5 0.9 1.4 1.8 2.2 2.6 3.1 3.5
(0) 1000/1 (K-l)

 

100,1

 

 

RESISTIVITY (SI-cm)
8

I PRE-ANN EAL
I POST-ANNEAL

 

 

 

 

 

1 1 1 1 1 I 1 1 1 1 l 1 1 1 1 1

2.5 4.5 6.150 $1,115 10.5 12.5
( b)

Figure 5.12 The temperature response of resistivity of test chips before and after

 

l

annealing over temperature range of (a) 300-1273 K (p=2><1015 cm“3 and (b) 77-300 K
(p=1.7><1016). The solid and dash lines indicate the exponential curve fit for tempera—

ture response before and after annealing.

141

temperature to approximately 573°C and corresponds to band conduction in freeze-out
region [107]. The second slope, that is for temperatures beyond 573°C, corresponds to
the extrinsic and intrinsic region [8]. However, no distinction in these two regions is
visible from this part of the curve. The slope of the curves in the temperature range of
300-573 K, for test chips, decrease with increasing resistivity (and Hall concentration)
but the slope for temperatures beyond 573°C is largely unaffected by resistivity values.
The resistivity measurements in the temperature range of 77-300 K, as shown in Fig.
5.12(b) before and after annealing, also show two slopes. The slope at temperature
below approximately 160 K corresponds to hopping conduction within the freeze-out
region whereas the slope at temperatures beyond approximately 160 K corresponds to
band conduction within the freeze-out region [93]. This behavior is typical of all the
test chip resistivity measurements. The evaluation of various activation activation ener-

gies will be discussed under section 5.3.

5.2.5 Hall Measurements

The Hall measurements were also made by the Van der Pauw method. The Hall
mobility was determined by measuring the change in resistance RBDAC namely
ARBD‘AC, once a magnetic field B is applied perpendicular to the sample with mean
thickness d. RBD.AC is defined as the potential difference VC -— V A between the con-
tacts C and A per unit current through the contacts B and D as per geometry shown in

Fig. 5.9. The Hall mobility “H is given by [167]

d ARBDAC
11H = —— (5.3)
B P
However, a number of spurious signals are included in this measurements. To elim-
inate them, a series of readings were taken with all possible combinations of current

and magnetic field and ARBDAC was evaluated as follows:

1 V 3.1+V—1,—B ‘V B.—I)_v—B. 1)
ARBD,AC=Z (+ +) ( )I (+ ( + (5.4)

 

142

From the mobility values, the Hall concentration (free holes) was evaluated using the

following relation for p-type semiconductor

l
9110p

 

p assuming “H = 11L. (5.5)

Though the two types of mobility are different by a factor of 3108, the Hall mobility is
usually approximated with conductivity mobility 11C for perfect single crystal bulk
material. By making measurements at high magnetic fields, this factor can be reduced
to 1, regardless of the conduction mechanism [81]. In this research, a magnetic field
of about 3kG was used which is considered high enough to allow the reduction of this

factor to unity.

The Hall concentration measured against various resistivity values for two type of
samples has been plotted in Fig. 5.13(a). The corresponding Hall mobilities are also
shown in Fig. 5.13(b). As expected the resistivity and Hall mobility decreased with
increasing carrier concentration. The range of Hall concentration i.e., 1015-1019 cm’3
covers fairly well the whole practical range of interest. Comparing the two types of
samples, the resistivity of discrete samples is larger and the Hall mobility is smaller
for comparable Hall concentration. The Hall mobility of the discrete samples is rela-
tively very small especially at higher Hall concentration. Nevertheless. both of these
parameters agree with values found in the literature on polycrystalline diamond films
[96-98]. However, the mobility is about two orders of magnitude less than reported
value of 1200 cmZV'1s’1 for crystalline diamond [47]. This may be due to the strong
influence of scattering at the grain boundaries present in the polycrystalline films. With
comparable Hall concentrations, the lower Hall mobility for discrete samples can be
attributed to the relatively smaller grain size (Hence more grain boundaries and carrier
scattering) due to smaller thickness (2-3ttm versus 3.4-4.5pm) or the relative purity of

the film as discussed in chapter 3.

        
  

 

fv‘vyvl IIIIIII'TVIYY'T TTTIII

. TEST CHIP
- DISCRETE

 

RESIST IVITY (SI-cm)

 

 

 

 

 

 

 

 

.1
10-1 1111111 111111111 11111 111111111111111 11111111111111 11111.1
1015 1016 1017 1018 1019
(o) HALL CONCENTRATION (cm-3)

102__ vvrfiv1111111l-1-1-1-1 rvrrrqfifi—rTn-y 1111;.,~1-.v.-1 11.,H‘
:2- 1 1
'> 1
"s
:1“
>- 10l : .3
E- _
3 1
E
O
2 _ __ _. ._1_...__,._
'4 . TESTCHIP
:11: I DISCRETE

100 “11211111111111.112111111111111111.1.111.111.1111 111111

1015 1016 1017 1018 1019
< b) HALL CONCENTRATION (cm'3)

Figure 5.13 The room temperature (a) resistivity and (b) Hall mobility vs measured

Hall concentration of the two types of samples.

144

The measurement of Hall concentration and the corresponding Hall mobilities
over the temperature range 77-300 K have been shown in Fig. 5.14. The Hall concen—
tration, as expected, increase with temperature. The Hall mobility plotted against 1-3/2,
showed a maxima around 160 K. The part of the mobility curve below and above this
temperature is usually controlled by lattice and impurity scattering respectively. The
mobility due to lattice scattering and impurity scattering are expected to follow
ttL oc T”?V2 and p10: T3/2 respectively in crystalline semiconductors. However, the
curve in Fig. 5.14(b), plotted against 1“”2 does not appear to follow these relations

which is indicative of additional scattering mechanism.

5.3 COMPUTED PARAMETERS

For all test chips, ln p vs l/T curves corresponding to resistivity measurements
over the temperature range of 300-1273 K show two distinct slopes (see Fig. 5.12(a),).
The slope in temperature range 300-573 K, corresponds to the band conduction zone
of the freeze-out region of diamond where boron impurities are partially ionized
[107,166]. The slope for temperature >300°C corresponds to the extrinsic and intrinsic
regions of natural semiconducting diamonds [47,107]. However, there is a possibility
that the freeze-out region extends beyond 300°C with a different boron activation

energy. In the intrinsic region the resistivity usually follows the relation

E

——g—] (5.6)

9 “up 2kT

 

Hence the slope of In p vs l/T curves should be representative of Eg/Z. Moreover, the
starting point of this region is determined by impurity concentration since they equal
the intrinsic carrier concentration at this temperature.

In actual measurements, the slope of the curves beyond 300°C correspond to 0.58.

0.421 and 0.375 eV for the samples with room temperature Hall concentration of

2x10”, 1.07X1016, and 9.8)(1018 cm“3 respectively. Clearly, these values are much less

145

 

 

 

 

10” m r . . 1 w . v v r v . r , I I I . r T . x I I . I 1
6‘ i 1
Z
2 -
[-
32
[- 10": a
Z L
DJ ~ -
U .4
Z
O
U -I
._1
3 r .

1015 I 1 l l l I 1 l 4 I #1 I I l I I I l 44 I l I l i l I l

100 150 200 250 300 350 400
TEMPERATURE (K)
(O)
103 L I I I I I I I I I I I I I I I I

 

fl
‘2.
VII

 

HALL MOBILITY It“. (cm’v-ls")

 

 

l 1 I r l I 1 L l

0.8 l .0

E

 

0.2

S3
o

1%30/1‘3’1 (re-"Iii6
(13}

K

Figure 5.14 The temperature response of (a) Hall concentration and (b) Hall mobility

in the temperature range of 77-300 K.

146

than half the band gap of diamond (that is 5.45 eV). Also, these values are somewhat
impurity concentration dependent. In addition, the starting point of this slope was
found to have no clear dependence on the Hall concentration. These observations sug-
gest that the temperature response of diamond films at high temperature has no correla-

tion with the intrinsic behavior of crystalline semiconducting diamond.

In the temperature range 77-300 K, the ln p vs l/T curves, as shown in Fig.
5.12(b), showed two distinct slopes. The slope at temperatures beyond approximately
160 K corresponds to band conduction in freeze-out region whereas the slope at tem-
peratures <160 K, corresponds to hopping conduction region as stated in Eq. 2.3.13.

Taking In in this equation we obtain
lnp = e3/kT + lnpO (5.7)

where 83 is defined as the hopping conduction activation energy. The slope of resis-
tivity curve below 160 K in Fig. 5.12(b) is essentially constant. Using Eq. 5.7, the
slope of resistivity curve equals 83/1000Xk. The computed values of 83 in this case is
11.72 meV. This value of 83, as shown in Fig. 5.12(b), did not show any sign of varia-
tion with annealing at 1000°C. Since hopping conduction takes place by jumping of
charge carriers from impurity atom to impurity atom, it requires some donor states to
be present. Hence, the presence of a hopping regime within freeze out region may be

due to compensation of impurities.

It was stated in chapter 2 that for a partially compensated p-type semiconductor
with a single impurity activation energy level at BA in the band gap, the hole concen-
tration can be extracted from Eq. (2.3.9). If little or no compensation is assumed. the
hole concentration is given by Eq. (2.3.11). Since no donor impurities were added
intentionally or found through SIMS analysis of diamond films used in this research, it
can safely be assumed that little or no compensation of impurities took place. This
assumption is further supported by the fact that any defect states present in the as

grown diamond films are passivated by hydrogen present in the diamond films

147

[99,100,103,159]. This was shown on the basis of Observing irreversible changes in the
l-V curves of undOped diamond films annealed at various temperatures. Since no
irreversible change in the temperature response of the resistivity was noticed up to
300°C, we can extend this assumption up to this temperature. Hence, all the discrete
and unannealed test chip samples can be treated under this assumption. Substituting

the value of p from Eq. (2.3.11) into the relation p = (qpup)’l, we get

1
MN, ‘3 E3 5 x
2 6"" ZkT ( ' )

 

p“KIT

 

If it is assumed that “OCT—3’4, the pre-exponential factor becomes independent of tem-
perature. In the ln p vs l/T curves shown in Fig. 5.12(a), slope of curve for tempera-
ture below 300°C for unannealed test chip is fairly constant. Using Eq. 5.8, this slope
equals Ea/2000Xk. The activation energy Ea, computed by this method for a number
of unannealed test chips and discrete samples, is shown against the measuredHall con-
centration in Fig. 5.15(a). The value of E3 decreases with increasing Hall concentration
in both the discrete and the test chip samples. Ea values for the discrete samples are
relatively large for comparable Hall concentration. The decrease of Ea with increasing
Hall concentration may be due to formation of an impurity band within the energy
gap. 138 values for most lightly doped test chips (p=2><1015 cm‘3) and discrete
(p=2.3><1015 cm’3) samples are 0.36 and 0.38 respectively which are in close agree-
ment with the reported value 0.37 eV for single crystal diamonds [47]. Moreover, ear-
lier measurements of Ba of CVD diamond films [97] showed values in the range of

0.26 to 0.38 eV which are consistent with 13a values given in Fig. 5.15(a).

The impurity concentration was computed by substituting values of E3 into Eq.

(2.3.11) which is re-arranged as follows:

2

Ba
NA = %exp[ [a] (5.9)

 

148

 

0.40

.0
La)
u.

BORON ACTIVATION ENERGY (eV)
3 S
LII o

 

 

TEST CHIP
DISCRETE

 

 

 

 

 

 

 

 

 

 

 

0.204;» . 11-2w-.- E-
1015 10'6 10” 10'8 1019
HALL CONCENTRATION (cm'3)
102‘» a #7 fl .
E 1023: . TESTCHIP
2’ . I DISCRETE .
O 1022i -.
r: I .
< A
E 102'} i
8 I.
z 102% 33
o .
u » .
>- 10'9 1
E: ' 1
a: .
18,. 1
E3 10 . 3
E .
1017 I I -EIAII lulzl 111.1 i will 1 ALL
1015 1016 1017 1018 1019

HALL CONCENTRATION (cm‘3)

Figure 5.15 The plots of computed boron (a) activation energy and (b) concentration

against measured Hall concentration for two types of samples.

149

Nv was computed using mp = m0 of natural semiconducting diamond [107] to yield
Nv = 3.137x1015 T3”- cm’3 (5.10)
At T = 300 K,
Nv = 16.3x101" cm'3 (5.11)

The resulting impurity concentrations are plotted against measured Hall concentration
in Fig. 5.15(b). As expected, the impurity concentration increases with Hall concentra-
tion. The fact that 0.04 to 0.75% of impurities are ionized at 300K over the whole
range, agrees with the observation that at room temperature 31% impurities are ionized

[166]. This supports the initial assumption of little or no compensation in these films.

In the case of the annealed test chips, the assumption of passivation of the defect
states is no longer valid due to dehydrogenation of diamond films during annealing
[99,100,103,158]. Assuming these samples to be partially compensated and the
assumption of N A>ND>>p holds, the hole concentration in the band conduction zone of
the freeze-out region is given by Eq. (2.3.10). Substituting the value of p into the rela-

tion p = (qpqul, we get

2 K
thN" K—l

 

Ea
p = exp(fi) (5.12)

where K = ND/NA is the compensation ratio. Now assuming that “OCT—3’3, the pre-
exponential factor becomes independent of temperature. Using Eq. 5.12, the slope of
annealed test chips in the temperature <300°C equals Ea/IOOOXk. Note that the com-
pensation assumption reduces the computed activation energy by one half for the same

‘ slope of In p vs l/T curve.

The activation energy computed by this method for a number of annealed test
chip samples is shown against measured Hall concentration in Fig. 5.16. The values
of Ba found earlier for the same (unannealed) samples with no compensation assump-

tion are also included in this figure for comparison. Note that the scale on the right

150

 

 

 

 

 

0.40 . .I - - . .- . 0200
E .
; 0 No compensation (Left scale)
0 0.35 ‘ I With compensation (Right scale) 0.175
g
5
g 0.30 0.150
E:
<
z 0.25 0.125
I"
U
<
Z
2 0.20 0.100
O
m

0.15 " " ‘ ' ‘ "‘“* 0.075

1015 1016 1017 1018 1019

HALL CONCENTRATION (cm'3)

C

Fig. 5.16 The plots of computed boron activation energy of test chips before
(assuming no compensation) and after annealing (assuming compensation)
against Hall concentration measured at room temperature.

BORON ACTIVATION ENERGY (eV)

151

vertical axis is exactly half of what is on the left vertical axis. The present values of
Ea are roughly one half of the earlier values. Beside this difference due to the compu-
tational method used, 138 increases due to annealing at lower Hall concentrations
whereas this effect is reverse at higher Hall concentrations. The reason for this
behavior is not understood.

The impurity concentration, in this case, can be computed by substituting the
values of Ba into Eq. (2.3.10). However, in this case, ND should be known for the
evaluation of N A. The equation can be solved for the compensation ratio K as follows:

-I
K: 1+-§1Pv-exp% (5.13)
The computed value of compensation ratio K is 79.6% for Hall concentration of

2x1015 em-3. This value appears fairly unreasonable. This may be due to the basic

assumption of N A>ND:>p, taken for the computation of K value.

5.5 SUMMARY

In this chapter, fundamental properties such as resistivity, mobility, carrier and
impurity concentration and impurity activation energy of the diamond films as a sem-
iconducting material were acquired. The annealing behavior of diamond films was
investigated. The resistivity of test chips was measured over the widest temperature
range (77-1273 K) ever reported. The control of doping through the quantity of boron
powder was measured. A characteristic curve of Hall concentration versus boron
powder was prepared. Finally, the conduction mechanism in CVD diamond films in
various regions was discussed in length. The knowledge acquired about the electrical
properties of diamond films will be useful in understanding and predicting the behavior
of diamond film based electronic devices in general and sensor for thermal signals in

particular.

CHAPTER 6

 

CVD DIAMOND FILM *
SENSORS FOR THERMAL SIGNALS

6.0 INTRODUCTION

In this chapter, the properties Of diamond film senors for temperature and heat
flux will be discussed. The static temperature response over the temperature range of
77-1273 K is acquired for various doping levels. High temperature effects on the qual-
ity of diamond films are presented in section 6.2. The measurement of dynamic tem-
perature response of the sensor and its implications on its application as a heat flux
sensor is described in section 6.3. An on-chip signal processing circuit made

exclusively of diamond film devices is proposed in section 6.4.

6.1 STATIC TEMPERATURE RESPONSE

The temperature response of a semiconductor is basically its resistivity change
with temperature. Hence, the resistivity curves against temperature presented in the last
chapter provide this information for CVD diamond films. Fig. 6.1 shows the tempera-
ture response for three doping levels (including two extreme measured values) for dia-

mond films over the temperature range of about 77-1273 K. All the samples were

152

153

 

 

102

101

 

 

 

 

10°:

104.;—

 

RESISTIVITY (Q-cm)

10'2 g‘

10-3E

'1
llillilll I Illlllll I

 

 

 

‘10-4 l l l L l ‘ ‘ 1 l 1 1 l A n 1 l 1 .1 r l r r r l r 1
_ 0 200 400 1315130 T 800 (K) 1000 1200 1400
(0) Elm “RE

101 I I I I I I I I I I I I I I 1 I I I T I I I 1

10p

 

10-l
10-2
10-3

10‘

 

 

ApflYF(flHmnl{4)

   
 

l 0'5 o 21.1 0I 5 cm'3
A 1 .72x1 01 5 cm"3
10-6 I 9.6x] 018 cm'3 ~

 

 

 

I 1 L

10-771111111111111111111 r. .
12001400

0 200 400 600 800 1000
(la) TEnAPquynURmsaK)

Figure 6.1 The temperature response of the (a) resistivity and (b) sensitivity in terms

of Ap/AT, of diamond film thermal sensors. The Hall concentration of each sensor is

also indicated.

154

annealed at temperatures of 600°C and 1000°C for 35 and 8 minutes respectively at a
pressure of 10'7 torr before taking measurements for temperature response of resis-
tivity. It can be seen that that the sensors have a negative temperature coefficient
(NTC) over the entire temperature range. This is unique to diamond films because all
conventional semiconductors, monocrystalline or polycrystalline, normally exhibit a
NTC only in the freeze-out and intrinsic regions and positive temperature coefficient
(PT C) in the extrinsic region. This is an essential property of a sensor in order to Show
a monotonic change of observable with a change in measurand to avoid showing a sin-
gle value of observable for multiple values of measurand. This property makes dia-

mond films useful as a temperature sensor over a wide temperature range i.e., 77—1300

K, a range never covered by a single semiconductor temperature sensing device before.

From the curves in Fig. 6.1(a), one can see that the lower the doping level, the
more rapid the change in resistivity. This effect is obvious from the plots of
pA = Ap/AT versus temperature as shown in Fig. 6.1(b). While comparing the two
extreme doping levels, the values of p A are about two order of magnitude larger at the
lower temperature limit and about one order of magnitude larger at higher temperature
limit for the lower doping level. This observation suggests that a lower doping level is
better in terms of sensitivity to the temperature changes. A criterion frequently used
for comparison of sensitivity of temperature sensors is the temperature coefficient Ot

defined as

(pi £91: (6.1)

The value for alpha for the same three doping levels, computed from Fig. 6.1, are
shown in Fig. 6.2. The large scatter in the data resulted from the large temperature
intervals between data points and the scatter in the original p versus temperature data.
Locally weighted average curves of (1 Show the existance of a minima and a maxima

at about 200 and 600 K, respectively. Since the curves are highly non-linear and

 

 

 

 

  

 

 

 

0.05 a- .
5.7 l“ ' r l
>4 .
V I 0 2x10” cm'3 I
E 0'04 5 A 1.72xl 016 cm‘3 i
Z I I 9.6)(1018 cm'3 I
u.) _ ________ __l
6 A
E 0.03 ' .
u‘ l
m . .
O
U . .
III-I 0 02 A /-—\ I O
a: ‘ / .'.\
a ‘ ‘ A A / j I \ I \ .
35001 WNW/ V - ' ‘\'- \ *
0 . [V " ‘ “353-4
E KW 47/ - \
0 A A
2 9f .
m - A ‘ A \f
E- 0.00 r r . A L l i r L I . r .
0 200 400 600 800 1 000 I 200 1 400
TEMPERATURE (K)

Figure 6.2 The plots of temperature coefficient at versus temperature for diamond film

thermal sensors. The Hall concentration of each sensor is also indicated.

156

overlapping each other, it is difficult to compare the effect of doping in terms of (1.
The mean value of Ot remains about 0006-0008 in all cases. For comparison, the aver-
age value Of Ot for Pt RTD is 0.00385 [9]. Hence, the sensitivity of diamond sensors is
better than Pt RTD over most of the temperature range. However, the thermistors Show
a typical value of Ot in the range of 0.2-0.4 but only over a limited net temperature
range i.e., typically 100 K [9].

Although at is widely used to represent the sensitivity of temperature sensors, but
it was basically devised for RTDs and is not suitable for comparison of sensitivity for
semiconductor sensors. To emphasize this point, a 100 k!) thermistor and a 259 RTD
both with on = 0.001 will Show a resistance change of 10052 and 25111.9, respectively,
with a temperature change of 1 K. Though rated at the same sensitivity, the tempera-
ture measurement with the semiconducting sensor will obviously be more accurate, can
achieve higher resolution and is much easier to detect. Therefore, 0t will not be used
for comparison of semiconducting temperature sensors in this discussion anymore.
Instead, another criterion, W, defined as ratio of resistance at two temperatures T1 and
T2 is frequently used for the comparison of temperature sensors [9] has proved better

for comparison of semiconducting temperature sensors.

In case of test chip, all the sensors spanned the temperature range of at least

100-1250 K, W will be defined as

P100

 

W = (6.2)

P 1250

The value of W for different doping levels is shown in Fig. 6.3. It is almost linearly
decreasing function of Hall concentration. From this curve it is obvious that a lower

doping value will have a greater sensitivity to temperature changes.

From an application point of view, the temperature is a dependent variable which
is determined from the value of resistivity (or resistance) using a calibrated curve. This

calibrated curve is either incorporated into a look up table or implemented into

 

157

 

1m I I IITITII I Y TTITITT I fir YIIYYI' I I ITrIIT
fi
4

f7

 

 

 

 

‘8
§
3
t
103 . rarrrrrl . rr.rr.rl r rirrlrrl r lirrrr.
10” 10" 10'7 10" 10"

HALL CONCENTRATION (cm’3)

Figure 6.3 The sensitivity of diamond film thermal sensors in terms of W as defined

by Eq. 6.2 against measured Hall concentration.

158

hardware. To estimate the diamond sensor suitability from this point of view, polyno-
mial curve fitting was done for several sensors. In the case of a sample with Hall con-

centration Of 2X1015cm‘3 the following fourth order polynomial equation
T = 606.906 - 54.745x — 5.613x2 - 1.633x3 +0.173x4 with x = ln p (6.3)

fitted the observed resistivity to the test temperature with a correlation of 99.89%.
Both the measured and estimated curves are Shown in Fig. 6.4(a). In spite of the large
scatter in the data the maximum error of estimation was less than 9 K. It is believed
that the measured data will be much smoother if it was acquired at a smaller tempera-
ture interval using a better calibrated reference than the type K thermocouple
employed. In that case, a third order equation may be sufficient for good curve fitting.
It is customary with manufacturers [14] to do curve fitting in terms of reciprocal tem-
perature. It is because most thermistors show excellent linearity in the freeze-out range
they are usually operated in. These curves are typically implemented in hardware using
logarithmic amplifiers. Therefore, it will be suitable to do curve fitting in terms of VT
for diamond sensors too. For the same sensor above (with Hall concentration of

2x1015cm’3) the following equation also gave a correlation of 99.88%.

% = 1.687 + 0.262x + 0.013x2 — 0.019x3 + 0.000075x4 with x = ln p (6.4)

Both the measured and estimated curves are shown in Fig. 6.4(b). Hence, an equation
of same order i.e., 4, gives sufficiently close fitted curves. For conventional thermis-
tors, which operate in a limited range, a third order polynomial equation is generally
sufficient. For diamond sensors covering a total range of over 1300 K, a fourth order

curve fit will give excellent performance.

Cross sensitivity to physical effects other than the primary measurand can always
plague the sensor output yielding random errors. Strain in the metallic temperature sen-
sors have shown pronounced effects on the calibration coefficients [12]. In case of dia-

mond film temperature sensors, the change of resistivity with strain has been Shown to

159

14m I I IIIIIII I I IITITTI I FITIIIII I I IIIIIII I I IIIIIII I I IIT

 

 

H
N
8

I I

 

 

 

fl
8 00 O
o 8 8
I IIIIIIII

TEMPERATURE (K)

A
8
.,...

llIlllJlllllLlllLl1lll1111

 

 

 

 

0 I 1 11111111 1 1 1111111 1 1 1111111 1 1 1111111 1 1 1111111 1 1111111
1 0'3 l 0’2 10'1 1 0° 101 102 103
(a) RESISTIVITY (SJ-cm)

 

lo I I IIIIIII I I IIIIIII I IIIIIIII I ITIIIIII I I IIIIIII I I IIIIII
I
b

 

 

 

 

lOOO/T (K4)

11111111111111111111111

 

 

 

0 b 1 1 1111111 1 1 1111111 1 11111111 1 11_14111L 1 1 1111111 L 1111111
10‘3 10'2 10" 10° 10l 102 103
(b) RESISTIVITY (D-crn)

 

Figure 6.4 The plots of fitted curves along with measured data for (a) temperature and

(b) reciprocal of temperature against 1n resistivity. (P=2X1015)

1 60

be several orders of magnitude smaller than because of temperature [168]. Hence. any

effect of strain on the resistivity of diamond film temperature sensors can be neglected.

In addition to cross sensitivity, the integrity of measurements are subject to
changes occurring in the properties of metal interconnects and substrate with tempera-
ture. The conductivity of the SiOz layer under the diamond film and the resistance of
the metal conducting lines increase with temperature. These effects are shown in Fig.
6.5. The resistance of the Si02 was measured between two adjacent metal bonding
pads (0.5 mm Spacing between them) with no other conducting path between them.
The resistance was beyond the measuring range of the instrument (10 M9) for most of
the temperature range but decreased sharply after 875°C. This provides a conducting
path parallel to sensors. The average SiOz thickness in this case was 1.129 11111.
Though the oxide resistance in this case was still much higher than sensor itself (a fac-
tor of 200 at 1000°C), it should be recognized that a smaller oxide thickness may have

higher conductivity than this sample.

The resistance of metal conducting lines was measured between two metal bond-
ing pads connected with 6.75 mm long and 160 um wide metal line. Though the resis-
tance of the metal line, as shown in Fig. 6.5(b), increases almost linearly with tempera-
ture, it is significantly smaller than that of sensors. Hence, any effect due to variation

of metal conductor resistance can be ignored.

6.2 HIGH TEMPERATURE EFFECTS

There are several effects associated with high temperature treatment of diamond
films. In addition to the irreversible change in resistivity discussed in the last chapter,
the oxidation of diamond is prominent. In chapter 3, diamond films were shown to oxi-
dize in flowing oxygen at temperature >650°C. Oxidation of diamond by an oxygen
plasma was observed even at lower temperatures [82]. To avoid oxidation, the dia-
)-7

mond sensors were always tested at pressures of about 1( torr for the acquisition of

161

 

IO

RESISTANCE (MD)
as

 

 

 

 

2 1 1 1 1 1 1 1 J 1 1 1 1 1 1 1 1 1 1
850 900 950 1000 1 050
TEMPERATURE (°C)

 

30 I I T I I I T —T I I I T I I Ti I I I r

N
LII
I

 

 

 

 

é , i '2
m . _
U . .
EzoL _
£3. : :
15 ~ 4
: 1
r- '1
10 . . - , . 1 ' . .
0 200 - 400 600 800 1000

Figure 6.5 The temperature response of the resistance of (a) 1.1 pm thick thermal

SiOz layer and (b) metal conducting lines (see text for experimental set up).

' 5
IL
.1.

162

their temperature response. The similarity of Raman spectra, before and after heat
treatment, as shown in Fig. 6.6(a), confirm the preservation of quality of diamond films
during heat treatment. A high resolution Raman spectra shown in Fig. 6.6(b) of the
same film, shows the absence of any peak at 1355 cm‘1 corresponding to amorphous
carbon. SEM micrographs of the diamond film after heat treatment at 1000°C for a
total of 14 minutes in 4 cycles are shown in Fig. 6.7. Any change in the surface mor-
phology is not visible. However, if the sample is heated at still higher temperatures,
some effects start to appear. For example, a sample heated at 1249°C for 2 minutes
was found to have somewhat darker grain boundaries (see Fig. 6.8(a)). However, no
sign of deterioration of quality was observed from the Raman spectrum of this sample
(not shown). To observe the effectiveness of high vacuum in protecting the diamond
film, a sample was heat treated at 1000°C for 2 minutes at a pressure of 100 mtorr.
SEM micrograph of this sample is shown in Fig. 6.8(b). Here a significant amount of
erosion is clearly visible. From these SEM micrographs and Raman spectra, it may be
assume that no structural change in the diamond film occurred during the high treat-
ment of diamond films at pressure of 10’7 torr. However, there is a clear need to pro-
tect the diamond films in the oxidizing ambient while at temperatures > 650°C if this
high temperature should persist beyond a few seconds. Possible protection layers are

undoped diamond, SiOz or Si3N4.

6.3 DYNAMIC TEMPERATURE RESPONSE

The dynamic thermal response time constant is defined as the time required by
the sensor to reach 63.2% of a step change in temperature under a specific set of con-
ditions (pressure, humidity, air flow, etc.,). However, generating a true step function of
temperature is quite complex. Therefore, a gradual change in temperature faster than
the sensor’s expected time constant is usually considered an acceptable step function.

Due to the small thermal time constant expected with diamond film sensors,

163

 

 

 

 

 

 

 

 

 

 

 

  

 

 

 

E _
ca .-
D _
E l-
1»
l:
0)
E
1 PRE HEAT TREATMENT
v POST HEAT TREATMENT
800 1200 1600 2000
WAVE NUMBER (cm’l)
,3 _ 1 PRE HEAT TREATMENT
:5 _ v POST HEAT TREATMENT
:3 -
E :
g»
t:
E
y .
h — 1 1 1 1 1 1 __ .4- ~
1310' 1320 1330 1340 1350 1360

WAVE NUMBER (cm-1)
Figure 6.6. Raman spectra of diamond film thermal sensors before and after heat treat-

ment at 1000°C for a total of 14 minutes in four cycles at a pressure of 10‘7 torr at

(W0 scan ranges.

164

 

Figure 6.7 SEM micrographs of diamond film thermal sensor after heat treatment at
1000°C for a total of 14 minutes in four cycles at a pressure of 10‘7 tan. (a) normal

view and (b) view at 60° angle.

165

.|‘

. - .5113? .940;

 

(b)

Figure 6.8 SEM micrographs of diamond film thermal sensors after two minutes of

heat treatment at (a) 1249°C at 10"7 ton and (b) 1000°C at 100 mtorr.

 

166

conventional method of generating a temperature step function by physically moving
the sensor from a cold medium to hot medium or vice versa are unsuitable. Hence. it
was decided to generate a temperature step by a step of electric current through the
sensor. For this purpose, the experimental set up shown in Fig. 6.9(a) was used. The
voltage across the sensor (IR voltage drop) was used as a trigger for the storage scope

to start its scan in a single trace mode.

In order to establish the validity of these measurements, a number of commercial
resistors ranging from 15 Q to 1 M9 in value were tested for their response time.
They exhibited a response time in the range of 0.2-1.1 ms without any clear depen-
dence on excitation current. The important observation was that these measurements
were not accompanied with any resistance change with current, hence, indicating that
voltage response time measurements across resistors are not dependent on the level of
excitation current. Once diamond sensors were connected in the circuit, a change in
resistance with excitation current was observed. A typical trace of storage scope, when
a current step function of 100 11A was passed through the sensor, is shown in Fig. 6.9
(b). The voltage across the sensor changes at the beginning of the scan and stabilizes
after some time. The rise time, as measured by the system shown at the bottom of Fig.
6.9(b) (219 11s) is representative of the time required for the current source to apply
the programmed current value and for the sensor to heat up and reach a steady state
value of resistance. The stabilized resistance (137.5 k9.) is less than the room tem-
perature resistance (165.5 kQ in this case). From the static temperature response of
the sensor (AR/AT = 5663QI°C at room temperature) a change ,in temperature of
523°C was computed. The rise time in this case was 219.56 11s and was found to be
independent of excitation current. As shown in Fig. 6.10, the rise time was between
ZOO-300 us for all of excitation current steps considered. The steady state resistance
of the sensor versus excitation current is also shown in Fig. 6.10. A decrease in resis-

tance with increasing current indicates an increasing heating effect. It can be assumed

167

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Pro ramm-
Storage ableg
sco e
p , current
source
TRIGGRER
(Cl)
IF U‘ ‘ r CHANNEL ‘___
w—m- WWWE, l 2 3 4 '
---—-~-—~«-~-~1_ . o o a 1
...... _) 1
, l -.
.1 .. v - 1
‘1" ‘ ‘ — .. w! ——‘
,1 ; _1
:‘ g;
1' NS,- Offset
1 _1
.: - T “m...
1 1‘- ac
lit - ~— ”fl I” J
1 i
a 2;: O :~.?§ a 'CéQC ‘5
‘y'md. ' 2 ' F1131. found ”3.1;le 1 , 17 q:
3”“1'“ 1 ' 51390-31: 110311 4‘ . 5,310 V 1 preset '
j «+3‘% 1 '25: $305: $55341 , v ~;g, V 'or9 Drone 1
v .- ‘ll': 1 4.!“ 'DtJ‘a « -.Jk.-

(b)

Fig. 6.9 (a) The experimental set up for dynamic response measurement of diamond
film temperature sensors and (b) Storage scope trace of voltage once a current step of

100 11A was applied to the diamond film temperature sensor with room temperature

resistance of 165.5 kQ.

168

 

 

 

 

 

200 300
ll ' ‘ ‘ " * —""‘—" “""‘ ' l
1 o RESISTANCE ‘
l I RISE TIME ~
A175 _ I f- u « 275
a .1
'31 17,
m a»
U .
$150 -250 E
E_. E—
a m
‘0 E
:3 a:
125 225
_ ..
I
100 .. ._.200
10° 10' 102 103

CURRENT (11A)

Figure 6.10 The steady state resistance of diamond film thermal sensor against current

flowing through it. The rise time of the voltage across the respective sensor with the

applied current step is also shown.

169

that the resistance change in the case of diamond sensors is due to the Joule’s heating

effect.

The time constant can be considered as 1/5 of the rise time. Thus, for a rise time
of 219 pm, the time constant will be 43.8 us. Some part of this time constant is due to
the measurement system itself. The sensor and the measurement system are in parallel
and both the measurement system and the diamond sensor are considered parallel com-
bination of a resistor and a capacitor individually. The time constant for the diamond

sensors was evaluated from the relation
r. = (C, + (3.002..—1 + Rd-lr‘ (6.5)

where R, C and I stand for resistance, capacitance and time constant whereas sub-
scripts s, d and t stand for measurement system only, diamond sensor only and total
system including sensor respectively. For this purpose, the time constant of the meas-
urement system was determined first. The mean value of this time constant was 135
us. After measuring the resistance of measurement system to be 985 k9, the value of
Cs was computed to be 137 pF using the relation 18 = RSCS. Once the diamond sensor
was connected in the system, the value of Rd was obtained from the storage scope
trace (137.5 k0). Substituting these values into Eq. (6.5), the value of Cd was com-
puted to be 182 pF. Using the relation Id = RdCd, the adjusted time constant of dia-
mond was found to be about 25 11s. This is fairly small time constant as compared to
commercial temperature sensing devices [14,931] which is in the range of ms to several

tens of seconds.

The response time measured in the present case is for the complete device includ-
ing the substrate. Hence, the thermal properties of the substrate material also contribute
to the overall dynamic response of the sensor. To achieve high sensitivity to heat
transfer rate, the sensors are usually mounted on thermal insulators such as A1203 [14]
and fused quartz [169]. This prevents the heat from escaping from the sensing element.

Quantitatively, the sensitivity ratio of two sensors with exactly similar sensing

 

170

elements but mounted on different substrate materials is given by [169]

S : A\/ (R DM Cp)suhstratel
(k pM Cp)suhstralc2

k = thermal conductivity (Wm—1K4)

 

(6.6)

where

pM = density (kg m“)

C], = specific heat (J—kg‘lK"1)

Table 6.1. Thermal properties of Si, SiOz, diamond, Pt and Cu [l6(),47,169]

 

 

 

 

 

 

 

 

 

 

 

 

 

Thermal . Specific
. Th .1 \lk C
Material Conducti- Density, thermal . erma M“ P
vity k pM cpacity CD d1ffus12\'1ty, K 2 4 2
. y 3 . ‘ K .'
W/m-K kg/m J/kg-K m /s .1 /m s
=~ :f -6 8
SI 157 2328 752 89.7 x 10 2.74 x 10
5102 12.7 2653 100 61.67 x 10'6 3.1711 x 10“
(max.)
Diamond 2000 3515 442+1.07 1.211 x 10'3 3.11 x 10"
(max.) (T-25 C)
Pt 73 21450 135.6 25.1 x 10'6 2.12 x 108
at 18-100 C
Cu 385 8960 3110 11.3 x 10'5 1.31 x 10"

 

The value of these parameters for various materials of interest are given in Table 6.1.

Considering Si02 and pure Si as two substrates, 8 = 86.22. Hence, a sensing element

 

171

placed on SiOz will respond 86.22 times faster than the one on Si substrate. In case of
the test chips, although the sensing elements are placed on SiOz, they can be con-
sidered on Si due to small (~1 11m) thickness of SiOz. The dynamic heat calculations
show that such a thin Si02 layer only retards the heat flow marginally in flowing from
diamond into the Si substrate. If it were placed on bulk SiOz, the mean time constant
of diamond sensors would reduce to about 290 ns. However, for temperature sensors
requiring only fractional ms response times, there is no need to increase the thickness

of 8102 layer.

As a heat flux sensor, the temperature sensor measures the surface temperature of
the body on which it is placed. The theory of transient heat conduction in a non-
homogeneous body is used to relate the surface temperature history to the rate of heat
transfer. In this way, it is possible to measure the surface temperature history of the
body and calculate the heat transfer rate. The present interest in diamond film sensors
as heat flux sensors is to compare their performance with conventional heat flux sen-

sors made of Pt or Cu.

The heat conduction analysis of a hetrogeneous sensor body of test chip or rod
can be modeled as a one dimensional problem as shown in Figure 6.11. Regions 1 and
2 represent the sensor and substrate materials with dissimilar thermaland physical pro-
perties. It will be assumed that lateral heat flow is relatively much smaller compared to
the vertical heat flow in Fig. 6.11 and can be neglected because of the extremely small

thickness of region 1 in comparison with region 2.

Considering this model, the surface temperature is given approximately by the

relation [169]

 

t l k1PM ICpl . .
T(t) = 2q \j — q — —’—— — 1 With q = constant (6.7)
“k2PM2Cp2 k1 k29M2sz

 

where

T = temperature increment above T(tSO) °C

 

 

172

t = time (seconds)
= thickness of sensor film (111)
q = rate of heat transfer to the surface

and subscripts l and 2 refer to the conditions in region 1 and 2 from the Fig. 6.11

 

 

<2) substrate

 

 

 

Figure 6.11 The one-dimensional heat conduction model for thin film heat flux sen-

sors on thermal insulator [124].

respectively. In this equation, the first term correspond to the temperature of the sub-
strate material if the sensor was not present, whereas the second term corresponds to a
temperature offset because of the presence of sensor. Hence, the temperature measured
by the sensor is actually the substrate temperature but offset by the second term in the

equation. At a given time, the temperature measured is less than the temperature of

173

the substrate (the measurement of which is actually desired) if the sensor was not
present. Therefore, second term should be as small as possible. In addition to decreas-
ing the sensor thickness, a material of high thermal conductivity can help reduce the
second term. Several materials such as Al, Cu, Au, Ni and Ag were tried as heat flux
sensing materials [14]. However, no significant improvement could be achieved. In
case of diamond film sensor, its high thermal conductivity along with its other thermal
properties, as mentioned in Table 6.1, can reduce this temperature lag by a factor of

more than 2 for same size sensor made of Pt.

The characteristic time defined as IZ/K (seconds), is another criterion for compar-
ing heat flux sensors. For the same thickness of sensors, diamond film sensor will be
51 and 11.3 times faster than Pt and Cu sensors respectively. Comparing diamond with
other sensing materials such as Ni, Ag, and Fe, the characteristic time of diamond film
sensors is at least 20 times smaller. Thus, it can be concluded that diamond heat flux
sensor with their added advantage in terms of temperature range of sensing and resis-

tance to harsh environment, are much superior to any other known sensor.

6.4 ON-CI-IIP SIGNAL PROCESSING

The concept of integration of sensors and signal processing circuit on a single
chip can enhance the system reliability, automation, precision and interface
simplification. This is now routinely done for pressure sensors [2] but is ralely found
for temperature sensors [9]. The reason is that unlike pressure sensor, the complete
chip is exposed to the temperature being measured. This poses the demand of an
extensive temperature compensation schemes for on-chip signal processing circuit. In
the present case, this problem becomes even harder due to the wide temperature range.
The diamond film technology is still in infancy. Currently, no active device with stable
operation over even part of this temperature range is available. The non-availability of

a suitable n-type dopant is another issue. The diamond based components currently

 

174

available are MOSFET [58,153], MESFET [131], Schottky diodes [155,148,154] and
resistors [124]. The operation of all of these devices is highly temperature dependent.
In addition, most of these devices are fabricated on homoepitaxial diamond films.
Nevertheless, the availability of "almost crystalline" diamond film synthesis technology
[66] on silicon substrates has made the fabrication of these devices possible on sub-

strate other than diamond.

From the foregoing description of the present state of diamond technology the
design of diamond signal processing circuit is precluded, at least in analog mode.
However, the inherent noise immunity encourages the design of digital circuits. For a
digital system, an electronic switch is the key element. From the available diamond
devices, the p-channel depletion-mode MESFET [131] is the only device which can
perform this role. It has been characterize as completely cut off under reverse bias
condition. A circuit using this device, as shown in Fig. 6.12(a). is proposed for imple-
menting NAND logic. For proper operation of the circuit RUFF/RON should be a large.
Since this ratio is not available for this device, a typical number of 105 is assumed.
Since any logic system can be built using a NAND gate, this circuit can act as build-
ing block for the proposed thermistor array multiplexer circuit shown in Fig. 6.12(c).
It is assumed that the counter and decoder circuits will be fabricated by using NAND
gates only. Due to device characteristics, logic 0 and 1 correspond to V+ and V’
volts. In absolute terms, these voltages are depend on the device behavior.

The primary function of this circuit is to reduce the output connections from the
sensor chip. The operation of the circuit is based on turning the MESFETS ’ON’
sequentially and making the current through the sensor, available at the output line for
read out by an ammeter. The following assumptions are made for the operation of this
circuit.

(1) RS/RONZ 100 and RS/ROFFSIO‘3 (RS: sensor resistance at reference tem-

perature).

 

175

 

 

 

 

 

 

 

I

I

I

I

. I

I

I

: sensor
: ..................... “lb
1 elements
I

I .
I

I

I

I

I

I

I

I

I .

I

I

I

I

I

I

I

I

I

I

I

I

I

 

 

 

output

 

 

 

 

 

 

 

N bit N-to-2N
counter decoder 2N

lines

 

 

 

 

 

 

MESFET
< switch

 

 

 

 

 

 

 

 

........................................................................

(b)
Fig. 6.12 (a) The implementation of NAND logic with two p-channel depletion mode
diamond MESFETs and a diamond resistor. (b) The block diagram of the proposed

analog multiplexer circuit using all diamond based NAND gates shown in (a).

176

(2) The doping level is the same for the sensor and the MESFET channel,

thereby both follow the same temperature dependence.

(3) All the MESFETs and sensors are identical respectively in terms of physi-

cal structure e.g., size, doping, etc.

(4) The whole chip is subject to almost the same temperature such that
assumption (1) holds at all times.

The whole circuit is primarily a bank of parallel circuits, each of which is consists of a

MESFET switch and a sensor in series. At one time only one of the switches will be

ON (low resistance). The resistance of the circuit is equivalent to the resistance of all

parallel branches. Since only one of the switches will be ON (low resistance) at one

time. the resistance of the circuit seen at the output line at any given time will be

V 2N-1 “1
D” = -—'— + z ‘ (6.8)

[our R5 + RoN 1:1 Rs.1 + ROFFJ

 

 

where [OUT is the measured current at the output. By giving an allowance for finite
leakage current during the OFF state (typically 10 nA), the error in the measured resis-
tance and true RS will be less than 1%. Even this error can be taken care of by cali-

brating individual sensors.

6.5 SUMMARY

In this chapter the temperature response of the diamond microsensor is discussed.
This class of sensor is shown to have reasonable (01 = (1006-0008) sensitivity over a
temperature range of 77-1300 K. At this time, no other single temperature measuring
device is known to operate over this temperature range either in absolute or relative
terms. By exhibiting a monotonic change of resistivity with temperature over the entire
range, the polycrystalline diamond film thermistors have outperformed even natural
semiconducting diamond thermistors. Moreover, the dynamic temperature response

time constant for the diamond film sensor has been measured to be about 25 11s. This

 

177

is smaller than any known solid state temperature sensor. As a heat flux sensor, dia-
mond film thermistor is shown qualitatively to outperform conventional Pt sensors for
shock tunnel applications. Finally a digital circuit made exclusively of diamond dev-
ices has been proposed to function as an multiplexer to cut down the number of output

connection from the chip.

 

 

CHAPTER 7

 

SUMMARY & CONCLUSIONS

The primary objective of this research was to develop CVD diamond film based
microsensors for thermal signals in the range of from approximately 77 K to 1273 K.
To achieve this objective new techniques for boron doping, nucleation, patterning and
metallization of diamond films were developed. The electrical characterization of the
diamond films as a semiconductor was carried out on a specially designed and fabri-
cated test microchip. Static temperature response of the sensors was acquired over a
temperature range of 77-1273 K for the first time. The dynamic temperature response '

of temperature sensors was also measured.
7.1 SUMMARY AND CONTRIBUTION

7.1.1 Synthesis of p-type Diamond Films

Using hot filament chemical vapor deposition (HFCVD) experimental parameters
such as reactant gases, gas flow rates, pressure and substrate and filament temperatures
were optimized to achieve high quality diamond films. In situ boron doping was

implemented using an accurately measured quantity (0.1-2.6 mg) of high purity

178

 

179

amorphous boron powder (5N8) through a specially designed holder placed near the
substrate. The quality of the diamond films as determined from Raman spectroscopy
and SEM was not affected by boron doping in the useful range of boron concentration
(up to 1019 cm‘3). SIMS depth profile and Hall concentration measurements showed
uniform boron concentration throughout the thickness of the films. Significant amounts
of impurities which might act as donor or compensators (such as P or N) were not
found. The SIMS data also confirm the increase in boron incorporation with the weight
of powder in the evaporator. The measured resistivity and Hall concentration of dia-
mond films doped by using 0.1—2.6 mg of boron powder were in the range of 0.3-64

Q—cm and 2x1015—9x1018 cm’3, respectively.

7.1.2 Patterning and Metallization

A test microchip was designed and fabricated. For the fabrication of the devices
the diamond film patterning and metallization were two crucial processes. Three IC
compatible techniques for diamond film patterning based on selective deposition and
selective etching were developed. Patterning of diamond films on non-diamond sub-
strates through selective etching was achieved for the first time. The diamond pho-
toresist (DPR) patterning technique based on mixing diamond powder in photoresist is
especially advantageous in terms of selectivity, resolution and ease and flexibility of
implementation. This method was developed to achieve optimum grain size (nucleation
density ~108 cmz, film thickness uniformity and surface smoothness. In case of metall-
ization, the stringent requirement of providing ohmic contacts on diamond films, and
of stability and good adhesion on diamond as well as SiOz surfaces simultaneously
was extremely demanding. This task was best accomplished by a two layer structure of
Pt(8000°A)/Ti(100°A) based on experimental results of this research and previous

work on high temperature metallization of Si ICs using silicide.

180

7.1.3 Electrical Characterization

Resistivity (77-1273 K), Current-voltage measurement, carrier type (77-300 K),
Hall concentration (77-300 K) and Hall mobility (77-300 K) were directly measured.
The resistivity and Hall mobility were in the range of 0.3-64 Q-cm and 2-48
cmZV‘ls’l respectively. These parameters provided informations to evaluate the impur-
ity concentration and various activation energies such as corresponding to dopant
(boron) and hopping conduction. The dopant activation energies, as computed from
the resistivity versus temperature curves (up to 300°C), were in the range of 0.38-0.11
eV corresponding to Hall concentration in the range of 2x1015—9x1018cm‘3 and boron
concentration in the range of 1017—1023cm'3. The compensation of impuritieswas
estimated to be in the range of 0.2 to 80% over the entire range of measured Hall con-
centration. The activation energies corresponding to high temperature conduction and
low temperature hopping conduction was computed to be 0.375-().58 eV and 11.72
meV respectively. Due to the change of properties of diamond during heating at high

temperature, the annealing behavior of doped diamond films was investigated for the

first time and an optimum annealing schedule (1000°C for 8 minutes) was acquired.

7.1.4 Diamond Film Sensor for Thermal Signals

The characterization of diamond film thermal sensors was carried out over the
temperature range of 77-1273 K. The change in resistivity with temperature is mono-
tonic with reasonable sensitivity (temperature coefficient ~ 0.07 K'l) at all doping lev-
els over the entire temperature range. At this time, no other single temperature
measuring device is known to operate over this temperature range either in absolute or
relative terms. In this range, the polycrystalline diamond film have outperformed even
natural semiconducting diamond since natural semiconducting diamonds show a nega-
tive temperature coefficient (NTC) of resistivity in the freeze-out and intrinsic region

and positive temperature coefficient (PTC) of resistivity in the extrinsic region. This

 

I81

behavior presents multiple values of temperatures for a single value of resistivity and,
hence, make them unsuitable for temperature measuring application over more than

one temperature region.

The dynamic temperature response of the temperature sensors was measured
experimentally. The typical response time of diamond temperature sensor on oxidized
silicon substrate was about 25 us, which is smaller than any known temperature sens-
ing device (for comparison a Pt RTD has a response time of 115 ms). An analytical
comparative analysis proved that diamond sensors will give better performance than
conventional Pt sensors. As a heat flux sensor, the characteristic response time of dia-
mond sensor was estimated to be 51 and 11.3 times smaller than Pt and Cu sensors
respectively. Comparing with other sensing materials such as Ag, Ni, Fe and certain
alloys, the characteristic time of diamond film sensors is at least 20 times smaller. The
success in fabrication of an array of sensors on the cylindrical surface of substrate
brings this application closer to practical realization. To minimize the number of pin
out connections from the sensing device, an on-chip analog multiplexer circuit

designed exclusively with diamond devices has also been suggested.

7.2 FUTURE RESEARCH

Since the diamond microelectronics is in its infancy, there is an abundance of
research work ahead in this area. Some of the key topics which need immediate atten-

tion for continuation of the present research are described as follows.

(a) The temperature response of resistivity of temperature sensors at the extremes of
the temperature range of 77-1273 K showed good sensitivity. This is an indication
that their operational temperature range can be extended on both sides. An
immediate area of interest could be to procure an appropriate apparatus and
characterize them below 77 K (to possibly 10 K) and beyond 1273 K (to possibly
1900 K).

(b)

(C)

(d)

(e)

(0

182

At present, the biggest obstacle in the operation of diamond sensors at high tem-
perature is its oxidation. Passivation of diamond sensors with other materials such
as SiOZ and Si3N4 tend to degrade their dynamic temperature response. A quanti-
tative analysis to this effect can give accurate informations about the suitability of
a particular material for specific applications. A natural choice of passivation
layer material for response time sensitive applications such as heat flux measure-
ment in shock tunnels would be diamond itself. These sensors will be disposable

since they will last until the passivation layer oxidizes completely.

The feasibility of homoepitaxial diamond film temperature sensors should be

investigated for potential improvement in the temperature sensor parameters. .

All the diamond film based active devices developed so far are highly tempera-
ture sensitive. This is understandable because they are operating in the freeze-out
region (up to approximately 700 K). An immediate application of the temperature
response characterization of the diamond films is to design a temperature compen-
sation scheme so that these devices could exhibit stable performance over a wide
temperature range. This will also help towards integration of signal processing

circuit on the sensor chips.

Since the properties of polycrystalline diamond films on non-diamond substrates
are dependent on deposition conditions, an effort to optimize the growth parame-
ters, especially the nucleation method, can improve the electrical properties
significantly and creat an opportunity for their application in high performance
electronic devices.

For the application of diamond films in active electronic devices, knowledge of
additional electronic properties beyond what were obtained during this research
can supply crucial information. Specifically, an investigation into the extent of
impurity compensation present and the thickness dependence of the mobility are

very important.

APPENDICES

APPENDIX A

 

DIAMOND FILM DEPOSITIONS

This appendix is a collection of information regarding the diamond depositions
made during the course of this research. A number of abbreviations and acronyms are

used in the tabulated format and are explained first.

A.I Process

The number appearing in this correspond to the sequence of the log book entries.
The numbers missing in the sequence correspond to the depositions made by others or

for other experiments.

A.2 Sample Description

This column describes the type of substrate, its pre-treatment and type of pattern
(if any). The substrate material is oxidized p-type doped silicon (100) polished wafer
unless otherwise mentioned. Following acronyms have been used to specify the nuclea-
tion method:

UT(*) - Ultrasonic treatment
DPR(*) - Diamond photoresist seeding

where * describes the pattern type as follows:

183

184

U - No pattern
D - Diffusion mask from MSU device fabrication lab.

L - Parallel lines, the width ranging from 75 11 m to 575 1.1 m. The length

of all the lines is 5 mm.
S - Square pattern (7.5 mm x 7.5 mm for Hall measurement)
TC4 - Test chip ( 4 patterns)
TCl - Test chip (1 pattern)

O - Others (unspecified)

A.3 Deposition Conditions

Following standard deposition conditions abbreviated as std has been used unless
otherwise indicated.

- Gas Flow Rates - C2“: : H2 : CO 0.5 : 100 : l2 sccm

Pressure - 50 torr

Filament Temperature (f. t.) - 2400 °C

Substrate Temperature (s. t.) - 890 °C

Boron powder quantity has been shown either by weight (for example B =
0.4 mg) or by number of holes filled in the boron holder (for example Bx4 to
indicate 4 holes filled). The weight of boron powder in each hole has been

measured to about 0.2 mg.

- std* stands for standard deposition conditions except the substrate tempera-

ture was held at 850 °C for first 15 minutes of the deposition.

 

185

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“0:688 Samples Deposition Conditions (EEC)
016 2xSi(micoromachined), s.t.=890 2.0
1x Ni/Si, 1xUT(U)
017 same as #016 s.t.=9l0 1.5
018 same as #016 s.t.=870 1.75
019 same as #016 (UT) s.t.=830 1.75
025 same as #016 s.t.=860 1.25
026 1 sample on Mo plate s.t.=830, Bias=60V 2.0
027 same as #106, f.t.=2200 2.0
Ni film (annealed) on
Si
031 4xUT(U) s.t.=870, f.t.=2475 1.5
053 same as #031 s.t.=830, f.t.= 2300, Bias=15V(0.12A) 1.5
057 4xDPR(D), 1xUT(D) std 1.0
058 6xDPR(D), 1xUT(D) std 3.0
060 1xDPR(D), 3xUT(D) std 1.0
061 3xDPR(D), 1xUT(U), std 2.5
1xDPR(D) on Ni/Si
062 3xDPR(D), 1xUT(D), std 2.0
1xUT(U)
063 3x DPR(D), 1xUT(D) std 1.5
065 3xUT(D), 1xUT(O) std 2.0
073 2xUT(O) std 1 .5
078 1xDPR(O) std 1.75
082 1xDPR(U), 1xUT(U) std* 4.0
083 4xDPR(D), 2xUT(D) std* 2.25
084 2xDPR(D), 2xUT(D) std* 3.0
088 2xDPR(D), 2xUT(D) std* 4.0
089 2xDPR(D), 2xUT(D) std* 4.0

 

 

 

 

I86

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1x DPR(U), 1x Si3N4

 

 

Process . . . . Time
# Samples Deposrtron Conditrons (Hrs.)
= E j 1
091 2x #089, 1x #088 std* 6.25
092 2x #89 std* 7.5
094 2xDPR(L), 1xUT(L) std* 3.25
095 lePR(L), 1xUT(L) std* 3.25
102 1xUT(U) H2=98, H2(bubbler)=5.0 1.5
CO=0.0, std*,
bubbler (Methanol) t.=23.9
103 leT (U) H2=95, H2(bubbler)=5.0 1.5
CO=10, std*,
Bubbler temp.=23.4
104 Ix UT(U) H2=92, H2(bubbler)=2.0 1.5
CO=0.0, std*,
Bubbler temp.=23.8
105 1xUT(U) H2=95, H2(bubbler)=0, 2.0
Bubbler line openCO=0.0, std*,
Bubbler temp.=23.4
110 1x UT(U) H2=95, H2(bubbler)=2.0, CO=0.0, 2.0
std*, Bubbler temp.=-78.2
115 Ix UT(U) std*, B=l3 mg 2.25
116 1xUT(D) std*, B=2 mg 3.5
123 1xDPR(U), 1xDPR(L), std*, B=1 mg 3.0
1x UT(U)
131 1x DPR(S), 1x DPR(L) std*, B=l4.5 mg 6.0
1x DPR(U)
132 1x DPR(S), lx DPR(L) std*, pre-cleaning run 6.0
lx DPR(U) without sample = 4 hrs
138 1xDPR(L), 1xUT(U) std*, b= 0.7 mg 6.0
142 lx DPR(S), 1x DPR(L) std*, B: 0.1 mg 6.0
1x DPR(U)
143 1x DPR(S), 1x DPR(L) std, B=0 6.0

 

 

187

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Process . . . . Time
# Samples DeposItIon CondItIons (Hrs. )
I44 Ix DPR(S), Ix DPR(L) std, B=0.2 mg 4
Ix DPR(U), Ix Si3N4
I45 1xPR(L), Ix UT(U) CO=0.0, B: 0.5 mg 3
I46 1x DPR(S), Ix DPR(L) s.t.= 950, CO=0.0, 8
Ix DPR(U), 1x Si(ch.) B: 0.6 mg
154 1x DPR(S), Ix DPR(L) std, B: 1 mg 6
2x DPR(U),
156 Ix DPR(S), 1x DPR(L) std, B: 1.2 mg 6
2x DPR(U),
158 same as #156 std, B: 0.7 mg 6
I62 5x DPR(U) std 6
164 Cu plate with diamond std I
grit sprayed
I65 lei3N4/Si, Ix Si02/Si std I
166 lei3N4/Si x #165 std 2
1x SiOz/Si x #165
1xSi3N4/Si, 1x SiOz/Si
167 Si with diamond grit std 4
sprayed
I68 1x UT(U), Ix DPR(U) No carbon gases, 3.5
bubbler with Acetone
169 1x #158, 1x DPR(L) on std, B=I.l mg 4.5
Ni coated oxidized Si
170 2x UT(U) Methanol = 0.5, No carbon gases 1
171 1x UT(U), Ix Si filmant distance = 5/32”, 1
Cu tube (1”) installed
I72 Ix UT(U) std 1
I73 same as #172 f.t.= 2348 l
174 same as #172 Cu tube distance=l .25” 2.0
from the flange plane, std
I75 same as #172 f.t. 2475 2.0
176 same as #172 Pressure = 60 torr 3.0

 

 

 

 

 

188

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Progess Samples Deposition Conditions (EEC)
I77 Ix UT(U), Ix DPR(U) Pressure = 70 torr 3.0
178 same as #177 Pressure = 100 torr 3.0
179 Ix DPR(U) CH4 = 0.5, Cu tube 1.6” 3.0
from the flange plane.

180 3x DPR(U) CH4 = 0.5, 3.0
181 same as # 180 CH4 = 0.5, 3.0
182 4x DPR(U) CH4 = 0.5, 0.25
183 2x DPR(U) CH4 = 0.5, 0.25
184 2x DPR(U) x #182 CH4 = 0.5, 3

2x DPR(U) x #183
185 4x DPR(L) CH4 = 0.75 3
186 2x #185 CH4 = 0.75 4
187 2x PR(S) CH4 = 0.75 0.25
188 2x DPR(L) CH4 = 0.75 3
189 Ix DPR(L) CH4 = 0.75 3
I90 1x DPR(L) CH4 = 1.0 3
191 Ix DPR(L), 1x DPR(S) CH4 = 0.5 4
192 1x DPR(TC4) CH4 = 0.5 4
I93 Ix DPR(TC4) CH4 = 0.5 4
I94 Ix DPR(TC4) CH4 = 0.5 4
195 1x DPR(TC1)x #194 CH4 = 0.5, B: 0.6 mg 4
I96 Ix DPR(TC4) std 4
197 1x DPR(TC1)x #196 std, B: 0.6 mg 4
198 Ix DPR(TC4) std 4
199 1x quartz rod UT(U), std 4

1x quartz rod DPR(P)
200 1x DPR(TCI) x #196 std, Bx3 4
203 Ix DPR(TCI) x #196 std, Bx4 4

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

189
“0:688 Samples Deposition Conditions ('32:
204 1x DPR(TCI) x #198, std, Bx5 4
2x crystalline (100) and
(I 10) diamonds
206 quartz rod DPR(A) std 4
208 qurtz rod x #206 std, Bx3 4
209 Ix DPR(TCI) x #198 std, Ex] 4
210 1x DPR(TCI)x#198 std, Ex] 5
214 flat quatz plate std 6
216 Ix UT(U), Ix (U), std, substrate=polysilicon 3
1x DPR(U)
219 1x DPR(TCI)x #198 std, Bx3 6
Ix crystalline (100)
diamond, 4x DPR(U)
221 Alumina rod DPR(A) std, Bx2.5 5
225 4x DPR(U) x #219 std, Bx3 3
226 Vycor glass rod std 4
DPR(A)
227 oxidized polysilicon std, Bx6 5
rod DPR(A)
228 1x DPR(TC4) std 4
229 1x DPR(TC4) std 4
230 DPR(TCI) x #229 std, Bx8 4
231 DPR(TCI) x #229 std, Bxl3 4
232 2x DPR(TCI) x #229 std, Bx10 3.5
233 Si rod DPR(A) std 4.0
234A 1x #233 s.t.=865 1.0
234B Ix DPR(TCI) x #228 std, Bx8 4.0
235A Ix DPR(T C I) x #228 s.t.=850, Bx6 4.0
235B 1x Si rod DPR(A) std 4.0
236 1x #235B std, s.t.=870 4.0

 

 

 

 

 

 

I 90

 

 

 

 

 

 

 

 

”0:688 Samples Deposition Conditions (1:26)
237 Ix PMMA(U) std 3.5
238 2x Si rod DPR(A) std 4.0
239 2x Si rod DPR(A) std, Bx8 4.0
240 1x DPR(TC2) x MM2, std, Bx8 4.0

1x MDPR(O)
242 1x MDPR(TC4) std 4.0

 

 

 

 

 

BIBIOGRAPHY

|3|

I41

[51

MI

I71

[8!

WI

IIOI

l11|

IIZI.

|13l

BIBLIOGRAPHY

S. Middelhoek, P. 1. French, I. H. Huijsing and W. J. Lian, "Sensors with
Digital or Frequency Output", Sensors and Actuators, 15, 119-133 (1988).

K. D. Wise and N. Najafi. "The Coming Opportunities in Microsensor Sys-
tems", Transducers '9]: The Next Decade, Tokyo, June 24, 1991.

W. l. Tompkins and J. G. Webster, "Interfacing Sensors to the IBM PC",
Prentice Hall. Englewood Cliffs, NJ (1988).

S. Middelhoek and S. A. Audit, "Silicon Sensors", Academic Press, London,
1989.

J. B. Huang and Q. Y. Tong, "Integrated Multi-FunCtion Sensor for Flow
Velocity, Temperature, and Vacuum Measurement", Sensors and Actuators,
19, 3-11 (1989).

A. Boyer, E. Cisse, Y. Azzouz and J. P. Cheron, "Narrow-bandgap Semicon-
ductors Based Thermal Sensors", Sensors and Acruators A, 25-27, 637-640
(1991).

B. Xie and S. Ren, "A Versatile Thermal Biosensor", Sensors and Actuators,
19, 53-59 (1989).

H. B. Sachse, "Semiconductor Temperature Sensors and Their Applications".
John Wiley and C0. (I975).

T. D. McGee. "Principles and Methods of Temperature Measurements", John
Wiley & Sons, 1988.

O. Zucker, W. Langheinrich and J. Mayer, "Investigation of Polysilicon for its
Applications to Multi-function Sensors", Sensors and Actuators A, 25-27,
647-651 (1991).

B. W. Van Oudheusden and J. H. Huijsing, "Integrated Flow Friction Sensor",
Sensors and Actuators, 15, 135-144 (1988).

S. Middelhoek, P. J. French, J. H. Huijsing and W. l. Lian, "Sensors with
Digital or Frequency Output", Sensors and Actuators, 15, 119-133 (1988).

K. Nelson, " Calibration of Sensor Resistance by Laser Trimming", Transduc-
ers '91: The Next Decade, Tokyo, June 24, I991.

191

1141

1151

[ml

[WI

1181

l19|

l20|

|21|

IZZI

l23|

124|

125l

126l

l27l

|28|

192

"The Temperature Handbook", Omega Engineering Inc., Stamford CA (I992).

J. M. Swartz and D. L. Swartz, "Recent Advances in Resistance, Diode and
Capacitance Thermometry for Use at Cryogenic Temperatures", Advances in
Cryogenic Engineering, 20, 389-401 (1975).

M. P. Timpo, "A two Terminal IC Temperature Transducer", IEEE J. Solid
State Circuits, SC-I I, 784-788 (1976).

W. Schafer, "Temperature Sensors: New Technologies on Their Way to
Industrial Applications", Sensors and Acruators, 17, 27-37 (1989).

D. G. Blair, C. C. Matheson abd B. J. Saunders, Cryogenics, 12, 561 (1972).

D. Hibberson. "Modern Thermistors Meet Rugged Demands", Machine
Design, 66-67, July I990.

E. D. Macklin, "Thermistors", Electrochem. Publications, Ltd., Ayn, Scotland,
(1979).

C. E. Woodhouse, "High Precision Rapid Readout of Cryogenic Temperature
Sensors in the Space Shuttle Environment", IEEE Trans. Instrurn. and Meas.,
39(1), 279-284 (1990).

H. V. Esrrade, "Linearization of the Temperature Dependence of the Silicon
Resistivity for its use as a Thermistor". Proc. SOUT HEAST CON ’90, New
Orleans, 3. 1095-1099, l-4 Apr. 1990.

D. S. Schwartz, "Narrow Range High Output Temperature Sensors Based on
Combined PTC/NTC Schemes", Proc. 37th Intl. lnstrum. Symp., CA, 581-601,
5-9 May 1991.

H. W. Trolander, D. A. Case and R. W. Harrugg, "Reproducibility, Stability
and Linearization of ThermiStor Resistance Thermometers", Temperature, 4
part 2, 997-1009 (1972).

K. Ohe and Y. Naito, "A new resistor having an anomalously large positive
temperature coefficient", Jpn. J. Appl. Phys, 10(1), 99-107 (1971).

E. Ling, H. S. Gamble, B. M. Armstrong and J. Wakefield, "A Novel Polycry-
stalline Silicon Temperature Sensor", Sensors and Acruators, 16, 225-234
(1989).

T. Nugai and M. Itoh, "SiC thin film thermistors", IEEE Trans. Industry
Applications. 26(6), 1139-1143 (1990).

E. W. Kiewra and P. C. Wayner Jr., "The Development of a Thin Film Sili-
con Carbide Thermistors Array for Determining Temperature Profile in an
Evaporating Liquid Film". J. EleCtrochem. Soc., 136(3), 740-744 (1989).

 

[291
UN

l31|

|32|

|33I

I341

I351

I361
[371

1381

I39!
1401

I411

l42l
I431

I44!

193

N. O. Ivanova, A. l. Chalyi, "Thin Film Bismuth Thermistors", Instrum. Exp.
Tech., 33(5) part 2, 1236-1237 (1990).

R. M. Nigamatullin, M, G. Shibanov and A. A. Kumanbeava, "Bismuth Based
Thin Film Thermistors", Instrum. Exp. Tech., 33(5) part 2, 1444-1446 (1990).

R. Dambra, G, Gallinaro. F. Gatti and S. Vitale, "Properties of Ruthenium
Oxide Resistor Elements of Composite Bolometers", Nucl. Instrum. Methods
Phys. Res. A, 302(1), 83-88 (1991).

E. Roess, "Ceramic Electronic Components", Siemens Components, 25(6),
213-216 (1990).

S. A. Obukhov, B. S. Neganov, Y. F. Kiselov, A. N. Chemikov, V. S.
Vekshina and and N. I. Pepkik and A. N. P0pkov, "Low temperature Resis-
tance of p-lnSb(Mn)", Cryogenics (UK), 31(10), 874-877 (1991).

T. Yamamoto and S. Takao, "Complex Impedance Analysis of Nb-doped (Ba,
Sr)TiO3 PTC Thermistors", Sumitomo Search, 45, 56-64 (1991).

C. J. Murray, "Design Cuts Sensor’s Response Time of Resistance Tempera-
ture Detectors for Supersonic Fighter Aircrafts", Design News, 47(6), 176-177
(I991).

C. lessen and H. Gronig, "A New Method of Manufacture of Thin Film Heat
Flux Gauges", Schoclt Waves, 1, 161-164 (1991).

G. B. Rogers and F. A. Raal, "Semiconducting Diamonds as Thermistors",
Rev. Sci. Inst, 31, 663-664 (1960)

US. Patent, 3,435,399, 25 March 1969.

N. Fujimori and H. Nakahata, "Thermistor made of CVD Diamond Films"
New Diamonds, 98-101 (1990).

"Diamond Thin Films: Hot New Material for the 90’s", Design News, 70-75,
Apr. 1989.

K. Tankala, T. DebRoy and M. Alam, "Oxidation od Diamond Films Syn-
thesized by Hot-Filament Assiswd Chemical Vapor Deposition", J. Mat. Res..
5(l I), 2483-2489 (I990).

S. M. Sze, "Physics of Semiconductor Devices", 2nd ed., John Wiley & Sons,
(1981).

A. T. Collin, "Diamond Electronic Devices ---. A Critical Appraisal", Sem-
icond. Sci. Technol., 4, 605-611 (1989).

I. S. Blackmore, "Design of Germanium Resistance Thermometers for Ther-
mometric Applications", Rev. Sci. Instrum., 33, 106-112 (1962).

I451

[46]

MN

1481

I491

[50]

IS”

[52!

l53|

154!

l55|

|56l

l57l

l58|

l59|

I601

194

L. C. Conner, "CVD Diamond: Beyond the Laboratory", Gorham Advanced
Materials Institute at the Global Business and Technical Outlook for High
Performance Inorganic Thin Film Coatings, (1988).

J. C. Angus, Y. Wang and M. Sunkara, "Metastable Growth of Diamond and
Diamond-like Phases", Annual Rev. Mat. Sci., (1990); J. C. Angus, and C. C.
Hayman, "Low-Pressure, MetaStable Growth of Diamond and Diamond-like
Phases", Science, 241, 913 (1988).

J. E. Field, The Properties of Diamond. Academic Press, (1971).

J. C. Angus, H. A. Will and W. S. Stanko, "Growth of Diamond Seed Cry-
stals by Vapor Deposition", J. Appl. Phys, 39, 2915 (1968).

F. P. Bundy, H. T. Hall, H. M. Strong and R. H. Wentroff. Nature. 176. 51-
55 (1955).

J. C. Angus, H. A. Will and W. S. Stanko, "Growth of Diamond Seed Cry-
stals by Vapor Deposition", J. Appl. Phys, 39, 2915 (1968).

B. V. Spitsyn, L. L. Bouilov and B. V. Deryagin, J. Crystal Growth, 52,
219-226(1981).

S. Matsumoto, Y. Sato, M. Kamo and N. Setaka, "Vapor Deposition of Dia-
mond Particles from Methane", Jpn. J. Appl. Phys, 21, L183 (1982); S.
Matsumoto, J. Mat. Sci. Lett., 4, 600 (I985).

Okano et. al., Jpn. J. Appl. Phys. 27, L173 (1988);

K. Suzuki, "Growth of Diamond Thin Films by DC Plasma Chemical Vapor
Deposition", Appl. Phys. Lett., 50 (12), 728-729 (1987).

K. Suzuki, "Characterization of the DC Discharge Plasma During Chemical
Vapor Deposition of Diamond Growth", Appl. Phys. Lett., 53 (19), 1818-1819,
Nov. 1988.

D. E. Meyer et. al., J. Vac. Sci. Technol. A7 (3), 2325-2327 (1989).
S. Matsumoto et. al., MRS Symposia Proceedings, EA-IS, 119-122 (1988).

G. S. Gildenblat, S. A. Grot, C. W. Hatfield, and A. R. Badzian, "High Tem-
perature Thin Film Diamond Field-Effect Transistor Fabricated using a
Selected Growth Method", IEEE Electron. Dev. Lett., EDL-IZ (2), Feb. 1991.

M. Kamo, Y. Sato, S. Matsumoto and N. Setaka, "Diamond Synthesis from
Gas Phase in Microwave Plasma", J. Crystal Growth, 62(3), 642 (1983).

Y. Mitsuda et. al., Rev. Sci. Instrum., 60 (2), 249-252 (1989); Y. Mitsuda, Y.
Kojima, T. Yoshida, and K. Akashi, "The Growth of Diamond in Microwave
Plasma Under Low Pressure", J. Mat. Sci., 22. 1557-1562 (1987).

[611

1621

I631

164!

l65|

I661

I671

168!

1691

I701

I711

I721

1731

W”

195

Y. Liou, A. lnspeckton, R. Weimer and R. Messier, "Low Temperature Dia-
mond Deposition by Microwave Plasma Enhanced Chemical Vapor Deposi-
tion", Appl. Phys. Lett.. 55(7), 631-633, Aug. 1989.

K. Kurihara, K. Sasaki, M. Kawarada and N. Koshino, "High Rate Synthesis
of Diamond by DC Plasma Jet CVD", Appl. Phys. Lett., 52(6), 437-438
(1988).

Y. Tzang, C. Cutshaw, R. Phillip, T. Srivinyunon, A. Ibrahim and B. H. Loo,
"Growth of Diamond Films on Silicon from an Oxygen-Acetylene Flame",
Appl. Phys. Lett.. 53(19). 1818-1819. Nov. 1988.

W. A. Yarbough and R. Messier, "Current Issues and Problems in the Chemi-
cal Vapor Deposition of Diamond", Science, 247, 688-695 (1990).

S. Koizumi, T. Murakami, T. Inuzuka and K. Suzuki, "Epitaxial Growth of
Diamond Thin Films on Cubic Boron Nitride {111} Surfaces by DC Plasma
Chemical Vapor Deposition", Appl. Phys. Lett., 57 (6), Aug. 1990.

M. W. Gies, H. 1. Smith, A. Argoit, J. Angus, G. H. M. Ma, J. T. Glass, J.
Butler, C. T. Robinsin, and R. Proyer, "Large Area Mosaic Diamond Films
Approaching Single Crystal Quality",

D. G. Jeng and H. S. Tuan, "Oriented Cubic Nucleations and Local Epitaxy
During Diamond Growth on Silicon {100} Substrates", Appl. Phys. Lett., 56
(20), May 1990.

Y. Hirose and Y. Terasawa, "Synthesis of Diamond Thin Films by Thermal
CVD Using Organic Compounds", Jpn. J. Appl. Phys., 25(6), L519-L521,
June 1986.

F. P. Doty and W. A. Jesser, "Investigation of the Role of Charged Species in
Hot Filament Assisted CVD of Diamond", J. Electron. Mat., 20(2), (1991).

S. J. Harris and L. R. Martin "Methyl Versus Acetylene as Diamond Growth
Species". J. Mat. Res.. 5(1 1), 2313-1319 (1990).

Diamond Deposition: Science and Technology, 2(6), 12-13, Aug. 1991.

P. K. Bachmann, D. Leers and H. Lydtin, " Towards a general concept of
Diamond Chemical Vapor Deposition", Diamond and Related Materials, 1(1),
l-IZ (1991).

S. J. Han'is and A. M. Weiner, "Effect of Oxygen on Diamond Growth",
Appl. Phys. Lett.. 55(21), 2179-2181, Nov. 1989.

J. A. Mucha, D. L. Flamm and D. E. Ibbotson, "On the Role of Oxygen in
Diamond Forming Discharges", J. Appl. Phys., 65(9), 3448-3452, May 1989.

I751

l76|

I771

[781

I791

I801

I811

I821

, 1831

l84|
1851
I861

I871

l83|

I89!

I901

I911

196

Y. Liou, R. Weimer. D. Knight and R. Messier, "Effects of Oxygen in Dia-
mond Deposition at Low Substrate Temperature".AppI. Phys. Lett.. 56(5).

S. A. Solin and A. K. Ramdas, "Raman Spectrum of Diamond", Phys. Rev.
B. l, 1687-l698 (1970).

M. Yoshikawa, G.Katagiri, H. Ishida and A. lshitani, "Resonant Raman
Scattering of Diamond-like Amorphous Carbon Films", Appl. Phys. Lett.,
52(19), May 1988.

M. Yoshikawa, G. Katagiri, H. Ishida, A. lshitani and T. Akamatsu, "Raman
Spectra of Diamond-like Amorphous Carbon Films". J. Appl. Phys., 64(11),
6464. Dec. 1988. ' ‘

D. S. Knight and W. B. White, "Characterization of Diamond Films by
Raman Spectroscopy". J. Mat. Res., 4, 385-393 (1989).

L. H. Robins, E. N. Farabangh and A. Feldman, "Line Shape Analysis of
Raman Spectrum of Diamond Films Grown by Hot-Filament and Microwave
Chemical Vapor Deposition", J. Mat. Res., 5(11), 2456-2468 (1990).

W. R. Runyan, "Semiconductor Measurements and Instrumentation", McGraw
Hill Book C0.. 1975.

H. Chiemi and Y. Sato, "Plasma Etching of Diamond Films", New Diamond.

‘ 6, 32 (1990).

G. E. Anner, "Planar Processing Primer", Van Nostrand Reinhold, New York,
110 (1990).

J. F. H. Clusters, Nature, 176, 360 (1955).
J. J. Brophy. Phys. Rev., 99, I336 (1955).
l. M. Buckley et. al., Diamond and Related Materials, 1(1), 43 (1991).

M. W. Gies, N. N. Efremow, and D. P. Rathman, "Summary Abstract: Device
Applications of Diamonds", J. Vac. Sci Technol. A, 6(3), 1953-1954,
May/June 1989.

A. Nishikawa, "Measurement of Thermal Conductivity of Diamond Films",
New Diamond. 6, 70-72 ( 1990).

"AT&T Reports Progress in Their Thermal Conductivity Studies", Diamond
Deposition: Science and Technology. 3(1), I (1992).

R. H. Wentrof and H. P. Bovenkerk, J. Chem. Phys. 36, 1987 (1962).

W. B. Wilson, Phy. Rev., 127, 1549 (1962).

192!
I931

I941

l95|

I961

I971

I981

1991

11001

11011

|102|

|I()3l

|I()4|

[105]

[106]

197

N. F. Mott, and W. D. Twose, Adv. Phys., 10, 107 (1961).
Y. F. Tsay et al., J. Appl. Phys., 43(9), 3677 (1972).

B. Massarani and J. C. Bourgoin "Hopping Conduction in Semiconducting
diamond",
Phys. Rev. 8., 17(4), 1758-1769 (1978).

l. A. Parfianovich, Y. S. Mukhachev and S. Y. Borzenko, "Electrical Conduc-
tivity and Thermoelectric Power of natural diamonds",

K. Nishimura. K. Das and J. T. Glass, "Material and Electrical Characteriza-
tion of Polycrystalline Boron-doped Diamond Films Grown by Microwave
Plasma Chemical Vapor Deposition", J. Appl. Phys., 69 (5), Mar. 1991.

K. Okano, H. Naruka, Y. Akiba. T. Kurosu. M. Iida, Y. Hirise. and T.
Nahamura, "Characterization of Boron-Doped Diamond Films", Jpn. J. Appl.
Phys., vol. 28, no. 6, pp. 1066-1071. June 1989.

R. Ramesham, T. Roppel, C. Ellis, B. H. Loo, "Electrical Characterization of
Undoped and Boron Doped Polycrystalline Diamond Thin Films", J. Electro-
chem. Soc., 138 (10), Oct. 1991.

L. S. Pan, D. R. Kania, S. Han, J. W. Ager III, M. Landstrass, O. L. Landen
and P. Pianetta, "Electrical transport Properties of Undoped CVD Diamond
Films". Science, 255, 830-833 (1992).

Y. Muto, T. Sugino, J. Shirafugi and K. Kobashi, "Electrical Conduction in
Undoped Diamond Films Prepared by Chemical Vapor deposition", Appl.
Phys. Lett., 59(7). 843-845 (1991).

G. H. Glover, "The C-V Characteristics of Schottkey Barrier on Laboratory
Grown Semiconducting Diamonds", Solid State Electron., 16, 973-983 (1973).

B. Huang and D. K. Reinhard. "Electric Field-dependent Conductivity of
Polycrystalline Diamond Thin Films",Appl. Phys. Lett;, 59 (12), Sep. 1991.

M. I. Landstrass and K. V. Ravi, "Hydrogen Passivation of Electrically Active
DefeCIs in Diamond". Appl. Phys. Lett., 55 (14), May 1990.

G. A. Sokolina, A. A. Botev, S. V. Bantsekov, O. I. Lazareva and A. F.
Belyamin, "temperature and Frequency Dependence of the Electrical Conduc-
tivity of Diamond Films", Sov. Phys. Semicond, 24(1), 105-108 (1990).

G. S. Gildenblat, A. Grot and A. Badzian, "The Electrical Properties and Dev-
ice Applications of Homoepitaxial and Polycrystalline Diamond Films", IEEE
Proc.,

A. S. Vishnevskil, A. G. Gontar, V. I. Torishnil, and A. A. Shul’zhenko,
"Electrical Conductivity of Heavily Doped P-Type Diamonds",

11071

[108]

009!
11101
11111

[112]

[113]
[114]
[115]
[116]
[117|

III8I
III9|

|120|

[121]

[122]

[123]

l124|

‘ 11251

198

A. W. S. Williams, E. C. Lightowlers, A. T. Collins, "Impurity Conduction in
Synthetic Semiconducting Diamond," J. Phys, C: Solid St. Phys., vol. 3.
1727-1735 (1970).

S. Amelinckx, The Direct Observation of Dislocations, Academic Press, New
York, (1964).

J. L. Shay et. al., Appl. Phys. Lett., 28, 31 (1976).
S. Wagner, "Epitaxy in Solar Cells", J. Crystal Growth, 31, 113 (1975).
P. Choudhari, J. Vac. Sci. Technol., 9, 520 (1972).

N. F. Mott and E. A. Davis, "Electronic Processes in non-crystalline materi-
als", John Wiley & Sons, (1971).

L. L. Kazmerski, "Polycrystalline and Amorphous Thin Films and Devices",
R. F. Greene, Phys. Rev., 141, 687 (1966).

A. Amith, J. Phys. Chem. Solids, 14, 271 (1960).

H. H. Weilder, Thin Solid Films, 31, 123 (1976).

M. E. Crowder and T. O. Sedgewick, J. Electrochem. Soc., 119, 1565 (1972).

A. L. Flipp, "Dependence of Resistivity on the Doping Level of Polycrystal-
line Silicon", J. Appl. Phys., 46, 1240 (1975).

T. l. Kamins, "Hall Mobility in Chemically Deposited Polycrystalline
Silicon"J. Appl. Phys., 42(11), 4357-4365 (1971).

J. Y. W. Seto, J. Electrochem. Soc., 122, 701 (1975).

J. Y. W. Seto, "The Electrical Properties of Polycrystalline Silicon", J. Appl.
Phys., 46(12), 5247-5254 (1975).

G. Baccarani et. al., J. Appl. Phys., 49, 5565 (1978).

C. H. Seager and T. G. Castner, "Zero Bias Resistance of Graon Boundaries
in Neutron ‘Transmutation Doped Polycrystalline Silicon", J. Appl. Phys.,
49(7), 3879-3889 (1978).

L. M. Edwards and. J. L. Davidson, "Fabrication Technology Development of
Polycrystalline Diamond Resistors", AbStract accepted to Intl. Symposium on
Microelectronics, 19-21 Oct. 1992.

J. Mort, D. Kuhman, M. Machonikin, M. Morgan. F. Jansen, K. Okumura, Y.
M. LeGrice and R. J. Nemanich, "Boron Doping of Diamond Thin Films".

[126]

[127|

[128]

[129]

[130]

[1311

I132]

[133]

[134]

[135]

[136]

|l37|

|138|

 

199

Appl. Phys. Lett.. 55(11). 1121-1123, Sep 1989.

K. Okano, Y. Akiba. T. Kurosu. M. lida and T. Nakamura. "Synthesis of B
Doped Diamond Films", J. Crystal Growth, 99, 1192-1195 (1990).

V. S. Vavilov, M. A. Gukasyan, M. l. Guseva and E. A. Konorova, "Conduc-
tivity of Diamond Doped by Implantation of Phosphorous Ions", Sov. Phys.
Semicond., 9(8), 962-964 (1976).

R. H. Wentrof and K. A. Darrow, Phy. Rev., 137, 1614 (1965).

V. V. Vavilov, G. E. A. Konorova, V. F. Sergienko and M. 1. Guseva, Sov.
Phys. Semicond., 4(1),12-16 (1970).

P. A. Sullivan, J. W. Boring, and R. A. Baragiola, "lono-luminescence of Dia-
mond and [on Implanted Light Emitting Diodes", Proc. 1991 Intl. Semicond.
Device Res. Symp., Charlottesville, Dec. 1991.

"Varian: Diamond Circuits are Feasible", Diamond Deposition: Science and
Technology, 2(5), 6, Aug. 1991.

K. Okano, H. Kiyota, T. lwasaki, Y. Naksmura, Y. Akiba. T. Kurosu, M. Iida
and T. Naksmura, "Synthesis of N-Type Semiconducting Diamond Films
Using Diphosphorous Pentaoxide as the Doping Source", J. Appl. Phys., A 51,
344-346 ( 1990).

K. Hirabayashi and Y. Taniguchi, "Selective Deposition of Diamond Crystals
by Chemical Vapor Deposition using a Tungsten-filament Method", Appl.
Phys. Lett., 53, 19 (1988).

J. S. Ma, H. Kawarada, T. Yonehara, J. Suzuki, J. Wei, Y. Yokota, and A.
Hiraki, "SeleCtive Nucleation and Grthh of Diamond Particles by Plasma-
assisted Chemical Vapor Deposition", Appl. Phys. Lett., 55, 1071 (1989).

Moeljanto W. Leksono, and H. R. Shanks, "Patterning of Microwave Plasma
Deposited Diamond Films", to be published in Microwave Processing of
Materials 11 (MRS 1990).

A. R. Kirkpatric, B. W. Ward, and N. P. Economou, "Focused Ion-beam
Crater Array for Induced Nucleation of Diamond Films", J. Vac. Sci. Tech-
nol., 7, 60 (1989).

N. N. Efremow, M. W. Geis, D. C. Flanders, G. A. Lincoln, and N. P.
Economou, "Ion beam assisted etching of diamond", J. Vac. Sci. Technol. B.
3(1). 416-418 (1985).

J. L. Davison, C.‘ Ellis and Ramesham, "SeleCtive Deposition of Diamond
Films", J. Electron. Mater., 18, 711-715 (1985).

[139]

[140]

I141]

[142]

[143]

1144l

[145]

[146]

[147]

[148]

[149]

[150]

[151]

[152]

 

200

S. S. Grot, C. W. Hatfield, G. S. Gildenblat, A. R. Badzian, and T. Badzian,
"Selective deposition of Homoepitaxial diamond Films",

J. L. Valdes, J. W. Mitchel, J. A. Mucha, L. Seibles and H. Huggins,
"Selected-Area Nucleation and Patterning of Diamond Thin Films by Electro-
phoretic Seeding",

L. M. Edwards and J. L. Davidson, "Delineation Methods of Polycrystalline
Diamond Films", Abstract accepted to 6th Intl. Electronic Materials and Pro-
cessing Conf, Baltimore MD, 23-24 June 1992.

J. F. Prins, "Bipolar Transistor Action in Ion-Implantation Diamond", Appl.
Phys. Lett., 41(10), 950-952 (1982).

K. L. Moazed, R. Ngnyen and J. R. Zeidler, "Ohmic Contacts to Semicon-
ducting Diamonds", IEEE Trans. Elecrron Devices, 9, 350-351 (1988).

M. W. Geis, "Diamond Transistor Performance and Fabrication", IEEE
Proceedings, 79(5), 669-675 (1991).

J. Engerrnann, H. Keller, D. K. Reinhard, B. Huang, and J. Assmussen,
"Dual-side Contact Formation on Isolated Diamond Films",

F. Fang, C. A. Hewett, M. G. Femandes and S. S. Lau, "Ohmic Contacts
Formed by Ion Mixing in Si-Diamond System", IEEE Trans. Electron Dev-
ices, 36, 1783-1786 (1989). .

A. T. Collins, E. C. Lightowlers, and A. W. S. Williams, "Electrical Contacts
to Semiconducting Diamonds," Diamond Res., pp. 19-22, (Suppl. Ind. Dia-
mond Rev.) (1970).

M. C. Hicks, C. R. Wronski, S. A. Grot and G. S. Gildenblat, "The Barrier
Height of Schottky Diodes with a Chemical Vapor Deposited Diamond Base",
J. Appl. Phys., 65(5), 2139-2141 (1989).

S. A. Grot, G. S. Gildenblat, C. W. Hatfileld, C. R. Wronski, A. R. Badzian,
T. Badzian and R. Messier, "The Effect of Surface Treatment on the Electrical
Properties of Metal Contacts to the Boron Doped Homo-epitaxial Diamond
Films", IEEE Elecrro. Dev. Lett., EDL-l 1, 100-102 (1990).

L. R. Zhang, L. S. Hung, S. Q. Feng, P. Revesz and J. W. Mayer, "Interaction
of Ti512 layer with Polycrystalline Si", Appl. Phys. Lett., 48(12), 767-769
(1986).

S. P. Murarka, "Refractory Silicides for Integrated Circuits", J. Vac. Sci.
Technol., 17(4), (1980)

J. F. Prins, "Bipolar Transistor Action in Ion-Implantation Diamond", Appl.
Phys. Lett., 41(10), 950-952 (1982).

[153]

[154]

[155]

[156]

[157]

[158]

[159]

[160]

[161]

[162]

[163]

[1641

[165]

[166]

[167]

201

H. Shiomi, Y. N. Yashi. and N. Fujimori, "Field Effect Transistors Using
Boron Doped Diamond Epitaxial Films", Jpn. J. Appl. Phys., 28, L2153-2154,
1989.

M. W. Geis, D. D. Rathman, D. J. Ehrlich, R. A. Murphy and W. T. Lindley,
"HighTemperature Point Contact Transistors and Schonkey Diodes Formed on
Synthetic Boron Doped Diamonds", IEEE Trans. Electron Device Lett., ELD-
8(8), 341-343 (1987).

G. S. Gildenblat "High Temperature Schottky Diodes with Thin Film Dia-
mond Base", IEEE Electron. Dev. Lett., EDL-l 1, 371-372, Sep. 1990.

B. 1. Shklovskii and A. L. Eferos, "Electronic Properties of Doped Semicon-
ductors", Springer-Verlag, (1984).

L. L. Kazmerski, P. J. Ireland and T. F. Ciszek, "Summary Abstract: Electri-
cal and Compositional Properties of Grain Boundaries in Multigrain Silicon
Using Surface Analysis Techniques", J. Vac. Sci. Technol., 17(1), (1980).

S. A. Grot, G. S. Gildenblat, C. W. Hatfield, C. R. Wronski, A. R. Badzian,
T. Badzian and R. Messier, "The Effect of Surface Treatment on the Electri-
cal Properties of Metal Contacts to Boron-Doped Homoepitaxial Diamond
Films", IEEE Electron Dev. Lett., EDL-l 1(2), 100-102 (1990).

S. Albin and L. Watkins ,"Electrical Properties of Hydrogenated Diamond",
Appl. Phys. Lett., 56(15), 1454-1456 (1990).

J. W. Mayer and S. S. Lau, "Elecu'onic Materials Science: For Integrated Cir-
cuits in Si and GaAs", Macmillan Publishing Company, New York, 1990.

S. P. Mrarka and M. C. Peckerar, "Electronic Materials: Science and Technol-
ogy", Academic Press Inc. 1989.

B. Huang, "Electrical Properties and Physical Characterization of Polycrystal-
line Diamond Films Deposited in Microwave Plasma Disk Reactor", PhD
Dissertation, Michigan State University, 1992.

M. Hansen, "Constitution of Binary Alloys, First Supplement", McGraw-Hill,
New York. 1965.

L. S. Pan, D. K. Kania, S. Han, J. W. Ager III, M. Landstrass, O. L. Landen
and P. Piaetta, "Electrical Transport Properties of Undoped CVD Diamond
Films". Science. 255, 830-833 (1992).

"H-50 Hall and Van der Pauw Controller", MMR Technologies Inc., 4 (1986).
A. T. Collins, Mater. Res. Soc. Symp. Proc., 162, 3 (1989).

L. J. Van der PAuw, "A method of Measuring Specific Resistivity and Hall
EffeCt of Discs of Arbitrary Shapes", Philips Res. Repts., 13, 1-9 (1958).

 

202

[168] M. Aslam, I. Taher, A. Masood, "Piezoresistivity in Vapor-deposited Diamond
Films". Appl. Phys. Lett.. 60(23), 2923-2925 (1992).

[169] R. J. Vidal, "Model Instrumentation Techniques for Heat transfer and Force
Measurements in a Hypersonic Schock Tunnels", Cornell Aeronautical
Laboratory Bufifalo NY rept., AD-9I7-A-l, (1956).

1110-11an smt: UNIV. LtaRnRtEs
[I1[I[[11]][I][I[[1]1]][11]][1]]|]]||]]|]][[|
31293007937653