?.:_g.. In"

ll
1.523...-“

d .
. -- 'v a." . ,. ’ ‘. ' Q r u_ w. “MW qr... .. .,
. 4-2:. F ‘ ‘ ' ’. 1‘ _

-' .’ 411'. p... 1.. ' '- I

.x‘ 1;
mm .17. "Hr. ‘
1.334;. 1
’3.- fl .1,

 

tn 7.?
J" o.“ .v‘..\u " .01.?! 3:01.
L' 1'? a . ‘3‘“ t‘. ‘ ‘n‘ . duty... nu.»-
.4. L_.._,;_‘_3...--':..,.;.-:. '13- _ ‘ , ,1.- - 1-. - .13.“. n 1. ~.---.r§-3~.~, g“ ' .. ii '
p—Mu-4 t‘ "L'f { " " , 11 i a . . y 0 > . . ‘ -~4...;.,1 ~,
-. .u. a... .-, ,_ , . L . J y _ ‘ . - - m“? --"L
_. .4 V «T _ . ,. . (La. 13.
- w». .
n
b . -- .xa.
F. ‘ l - . . ~- \ q u
. A ,.,. L .- . __v 3.: _ H.
E n x .45 .3 4"“.77' fit. '_~.- ‘ _ nflz-u ‘»IB.Q}£§.}P‘ -’
u n “s 1.... - aim-34.. - ~.. I“
"' . - ~'.A 1: 7‘
”I.“ . ‘

‘1- g!
.....
325: E!"

 

 

 

 

 

 

 

1..
,. _v-. u .- .‘5‘ '1. “A“ h
. . ~41 :- -
~ " . . -.. . *5?“ -< “:7314“:.N“,_:’91-+Mg “.2“:
. ~ ”.4- x... “‘4. r... f. 1, , ' {21;}: :vfiiy r , .‘p,.v£“ qu- .
_ ‘ . 1 :‘Lii‘? .51.. . >3: w. a".
i, a: . _ an. "4.1"
‘ ' =31.“ fwu
’L"rLaa.1‘:Zf-$AI~‘
‘ ,
:9
. .
‘4: . 2:43.433”
. 1 . “ 1......“ 1:.
' ”Eff?“ 4:..:; $3351: fit? .nm'i‘anu
‘ ‘ 4—,..n \ , . _ ..‘ 1- an -91. U 5: L
.»1— \ Hath-w WW I _ , ‘
.(L a "‘ ’g. r. "m". int-3‘; JILL“.
, . . .1 .fi..« “‘.
-:5.a~fim& I '43:. .4 , : «w... 3'»
q “flu.” -.4 «on ‘W "
r: leu > K?”
2 W4
\ ‘ ,3...
. menu 'J’.‘:‘ 3va «Sh
» mu- 41:34 \‘r ._n 1
M . L“: “50!: ’l¢kll\‘w:|‘*
1 ' -
3 .
E
x -
'4 ~ v .
, V“ ' - Q‘
~ . 4 . .34? ~ « u t u- .
l _ n It. '4 a . .J ~15::..-n\_q'-w‘1'.~r«:-.ano'
#:«JS‘J‘I. 21"....” . ’ V‘r.’ ~ ”9;; «a.» - * .
.r ‘ 131.11%.3-3», I“... ‘ : «rt-12$ ‘ V “~ mm W
, . ~ . _ ..- . 1-
3 73E. 5.11:“:441‘53 1 .1 . ~ M ”,-...‘§.';’ n:- v.1 $4.qu
, ~ "“431... ‘ ¢ 4 .H- v. u, -. 5“; “4‘HK .,.‘.; on. w 1"”
I: 1 . > . \\ ::~ ' 5:3 '~ ‘ . 3:“ ‘4 ‘.{;:‘- «.4
J“ ‘ ' .a ” '“Q '7'9‘74’1‘“*"
' l I. J l
'1
i v
' . ' 2:44;";
. :I. ‘ . ‘ 1‘ v. > « kirk”; m1:
. - ' . v , " ’ ' .437“ ~ 4' ‘
. 1 ' . _' -‘ , _ . _ _
. , ' ‘«~ .1
r h: bk. 'v ~‘ ‘ v. . a: '0 w '
. is,
E; f 33.14
w ‘ ._ >77
n5. hard.
33‘. n!
34¢ 4
»
I ‘ .«4' w
. . ~71, fig”:-
, 1“ ‘J(""r
1 HR ’42;
r {3
= .. *1 7.
‘ ~ '3‘." : ';,
t . a1 ‘ s: n
’ t "s ‘ 1'
. , . w» v .
' K "5 v ,3" «5w;- 4;
' a ”A v 3" -~ -'-
, I... . v .
’ * 9!". ‘ »‘ u.
. _ r1
fl . 'U -
'l—I
- {u
: w
. 5
I r
..
.
' unawar-
4’ :>::rlr"1!};;d’hlp.
'3wa 2'“
' J
:. 3 '1": u “I."
‘ ' ‘..f‘ 1 "
.1 q .139, ‘ .
q n.
V'J‘ r
I n - I : . . ),.
, m 1
.
.

It .

3g».- '1

V'Kf...
Iv?

.7:

 

~ .
"v :3”. 4
‘, '3' . .. , ;
€ \
1'. n. ..._..,,,
’1 5:451“; hr. .11
3.; . H". ‘ 1»:
~\
.-

LIBRARIES

llllllll‘l’illllm \l lllll

3 1293 00909 1327

 

 

 

 

 

 

This is to certify that the

dissertation entitled
Electra‘cal P'WPCYtiCS “MK Phys-‘0‘! Characten‘yu‘cs
of Pol/cr‘jstalhng Dt‘auvmnd Ftlms Dcpbst‘t¢c\ Tn A

Mtcrowave. Plasma piSK Reactov’

\

presented by

BULK/”dram HUanj

has been accepted towards fulfillment
of the requirements for

PL“ A degreein Elec‘lflCeL/Enymceflrj

Major professor

Date and“ 8/. I??2_

MSU is an Affirmative Action/Equal Opportunity Institution 0-12771

 

 

”WM?
Mm? *c’i” "“
Universsi iv 3'

_l

. u. .-.--._ -.__ .,

P‘s—vi 0.. “-...

 

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
1mg .3 2" “ ll

“AL-CM; __ _____l

 

 

 

 

c‘v- _ l
N . ' ‘
_

“I

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU to An Affirmative Action/Equal Opportunity Institution
chnS-DJ

ELECTRICAL PROPERTIES AND PHYSICAL
CHARACTERISTICS OF POLYCRYSTALLINE
DIAMOND FILMS DEPOSITED IN A MICROWAVE
PLASMA DISK REACTOR

By

Bohr-ran Huang

A DISSERTATION

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

DOCTOR OF PHILOSOPHY

Department of Electrical Engineering

1992

672' 373/

Copyright by
Bohr-ran Huang
1992

ABSTRACT

ELECTRICAL PROPERTIES AND PHYSICAL
CHARACTERISTICS OF POLYCRYSTALLINE
DIAMOND FILMS DEPOSITED IN A MICROWAVE
PLASMA DISK REACTOR

By

Bohr-ran Huang

Diamond possesses many excellent properties which motivate investigations of
the potential of diamond for a wide range of electronic, mechanical, and optical
applications. This work experimentally investigates techniques for high quality dia-
mond synthesis and develops means for electrical and physical characterization of the
films. The films are deposited by plasma assisted chemical vapor deposition using a
methane/ hydrogen plasma in a microwave plasma disk reactor system.

Both a diamond paste nucleation method and a diamond powder nucleation
method are studied in this research. Although as indicated by Raman spectroscopy
both methods produced similar quality diamond films, the powder nucleation method
produced fine grain, sub-micron sized crystallite, films whereas the paste nucleation
method produced large grain, several-micrometer size crystallite, films. This differ-
ence is due to a higher density of surface nucleation sites in the powder polished
films. Grain size could also be varied by using microwave power, pressure, and the

methane/ hydrogen ratio as processing variables, however, the effect is smaller than

the differences caused by the two nucleation methods.

For electrical characterization, a new sample preparation method was developed
in cooperation with the University of Wuppertal which allows metallic access to
both sides of the diamond film. Using this technique, the properties of a variety
of metal/ diamond contacts were investigated. Although films were not intention-
ally doped, thermo-power measurements show all films to be p-type with activation
energies between 0.1 and 0.5 eV.

For powder polished films, all metallic contacts were ohmic. These samples were
used to explore the high electric field properties of diamond. It was discovered that for
fields larger than approximately 1x105 V / cm the electrical properties are dominated
by defects, where defect is used generically for either an impurity or a structural
defect. For low electric fields, the electrical conductivity was constant which resulted
in ohmic behavior. But for high fields, the conductivity was field activated according
to Poole’s law. This behavior was modeled as being due to ionizable defects and
indicates that there is approximately one ionizable defect per 10,000 host atoms. As
a result of such defects, the breakdown field for these films was somewhat less than
1x106 V/cm. A large concentration of defects is compatible with the observation
of ohmic contact behavior regardless of metallic work function since contact space
charge layers would be sufficiently thin to allow tunneling.

N on-ohmic, Schottky barrier contacts were achievable on the paste polished films.
For Al/diamond/silicon structures diode characteristics were observed. These I-V
characteristics were modeled as an ideal Schottky barrier diode in series with bulk
diamond, for which the property of the bulk diamond follows an I ocV'" relation-
ship, indicative of space charge limited current in an insulating material. The rec-
tifying behavior was determined to be at the Al / diamond surface rather than the

diamond/ silicon surface. The best rectification ratios were 2x105.

To my parents
Chien-Ching, Tsui-Kuan C. Huang
and my wife

Pey—Nan C. Huang

ACKNOWLEDGEMENTS

The author would like to thank Dr. Donnie K. Reinhard for his guidance , support,
and encouragement. Special thanks is extended to Dr. Jes Asmussen Jr. and Dr.
Timothy Grotjohn for their valuable discussions, and Dr. Peter A. Schroeder for his
comments. Additional thanks is given to Dr. Engemann and Mr. Hans Keller in
Wuppertal University (Germany) for their help on fabricating back-etched diamond
samples, Dr. Kevin Gray of Norton Company and Dr. Mark Holtz in the Physics
department of Michigan State University for providing the Raman spectra, and Dr.
Kevin Hook in the Composite and Structural Materials Center of Michigan State
University for performing the XPS analysis.

This work was supported in part by grants from Wavemat, Corp. and the Michigan
Research Excellence Fund.

vi

TABLE OF CONTENTS

LIST OF TABLES x
LIST OF FIGURES xi
1 Introduction 1
1.1 Motivation for Diamond Research .................... 1
1.2 Research Objectives ............................ 3
1.3 Dissertation Outline ........................... 4

2 Synthetic Diamond : Background Review 5
2.1 Introduction ................................ 5
2.2 Brief History of Diamond Synthesis ................... 6
2.2.1 Historical Background ...................... 6

2.2.2 Growth Mechanism of Diamond Films ............. 13

2.3 Review of Diamond Film Technologies ................. 17
2.3.1 Hot Filament Chemical Vapor Deposition Techniques ..... 18

2.3.2 Plasma Assisted Chemical Vapor Deposition Techniques . . . 20

2.3.3 Other Deposition Methods .................... 24

2.3.4 Nucleation Methods. ....................... 24

2.4 Physical Attributes of Diamond ..................... 27
2.4.1 Comparison of Bulk and Film Material Properties ....... 27

2.4.2 Comparison of Semiconductor Properties ............ 30

2.5 Review of Diamond Diodes and Transistors ............... 33

3 Film Deposition and Sample Preparation 38
3.1 Introduction ................................ 38
3.2 The MPDR System ............................ 38
3.2.1 The Microwave Cavity ...................... 38

3.2.2 The Deposition System ...................... 41

3.3 The Nucleation Methods .................. . ...... 45

vii

 

 

3.4 Diamond Film Deposition ........................

3.4.1 Operation of The System .....................
3.4.2 Experimental Parameters ....................
3.4.3 Substrate Temperature ......................
3.5 Fabrication of Electrical Samples ....................
3.5.1 Four-Point Probe Samples ....................
3.5.2 The Metal / Diamond/ Silicon Samples ..............
3.5.3 Back-Etched Samples .......................
Physical Characterization
4.1 Introduction ................................
4.2 Raman Spectroscopy ...........................
4.3 X-ray Photoelectron Spectroscopy ....................
4.4 Dek-Tak Analysis .............................
4.5 Scanning Electron. Microscope Analysis .................
4.6 Film Uniformity Analysis .........................

Electrical Characterization

5.1 Introduction ................................
5.2 Four Point Probe Characterization ...................
5.2.1 Experimental Method ......................
5.2.2 Theory of Conductivity vs. Temperature ............
5.2.3 Comparison with Physical Characterization ..........
5.2.4 Experimental Results and Match with Theory .........
5.3 Back-Etched Samples. ..........................
5.3.1 The Back-Etched Samples and IV Measurement Set-Up . . .
5.3.2 Contact Effects on Back-Etched Samples ............
5.3.3 High Field Effect .........................
5.3.4 Photo Efiect ............................
5.4 The Metal/ Diamond/ Silicon Samples ..................
5.4.1 The I-V Measurement Set-up ..................
5.4.2 Contact Effects on Metal/ Diamond/ Silicon Samples ......
5.4.3 Photo Effect ............................
5.4.4 Diamond Schottky Barrier Diode ................

Summary and Future Research
6.1 Summary of Important Results .....................
6.1.1 NucleationMethod ......

viii

48
48
53
54
61
61
65
65

72
72
73
79
89
94
107

113
113
113
113
116
118
122
130
130
132
132
145
147
147
151
158
165

183

184

6.1.2 Quality of Diamond Films ....................

6.1.3 Back-Etching Technique .....................
6.1.4 Diamond/ Silicon Interface ....................
6.1.5 Activation Energy of Diamond Films ..............
6.1.6 Electric Field Dependent Conductivity of Diamond Films . . .
6.1.7 Diamond Schottky Barrier Diode ................
6.2 Future Research ..............................
6.2.1 Improvement for Diamond Deposition ..............

6.2.2 The Techniques for Physical Characterization .........

6.2.3 The Role of SiC at the Diamond/ Silicon Interface .......

6.2.4 Defects States in Diamond Films

6.2.5 Diamond Devices .........................

A Deposited Diamond Films
B Electrical Samples

BIBLIOGRAPHY

ix

185
185
186
186
187
188
188
189
189
190
191
191

193

198

202

 

 

2.1
2.2
2.3
2.4
2.5

4.1

4.2

4.3

4.4

LIST OF TABLES

Physical properties of bulk natural diamond.[68] ............ 28
Properties of bulk natural diamond and synthetic diamond films.[13] . 29
Properties of diamond, i-C, and graphite.[21] .............. 31
Working definition for different carbon coatings.[69] .......... 32
Comparison of semiconductor properties for diamond, Si, GaAs, ,BSic.[70] 34

Relation of average grain size vs. methane concentration, microwave
power and plasma pressure by the paste-polished method. ...... 105
Relation of average grain size vs. microwave power and plasma pressure
by the powder-polished method ...................... 106
Relation of deposition rate vs. methane concentration, microwave
power and plasma pressure by the paste-polished method. ...... 108
Relation of deposition rate vs. methane concentration, microwave

power and plasma pressure by the powder-polished method. ..... 109

2.1
2.2
2.3
2.4
2.5

3.1

3.2
3.3
3.4
3.5
3.6
3.7
3.8

3.9

3.10

3.11

3.12

3.13

LIST OF FIGURES

The atomic structure of (a) diamond (b) graphite ............ 7
Phase diagram of carbon.[12] ....................... 8
Illustration of free energy difference between diamond and graphite. . 10
Low pressure CVD diamond synthesis in the phase diagram of carbon.[13] 12

Relationship between diamond properties and engineering applica-
tions.[70] .................................. 35

The cross section of the MPDR and processing chamber, and a
schematic display of the electric (E) and magnetic (H) field patterns

Of the TMou mode. ........................... 39
The microwave power source, circuit, and cavity applicator ....... 42
The diagram of the MPDR CVD system ................. 44

SEM pictures of the silicon substrate prepared by diamond paste method. 47
SEM pictures of the silicon surface prepared by diamond powder method. 49

The MPDR CVD system during the diamond film deposition. . . . . 52
Top view of the MPDR processing chamber. .............. 56
(a) Typical temperature profile and (b) better temperature uniformity

of the silicon substrate ........................... 58
Si substrate temperature vs. microwave input power for 1 ‘70 and 0.5
% methane concentration at 50 Torr. .................. 59
Si substrate temperature vs. microwave input power for 1 % and 0.5

% methane concentration at 60 Torr. .................. 60
Si substrate temperature vs. microwave input power for 50 and 60 Torr
at 0.5 % methane concentration ...................... 62

Si substrate temperature vs. microwave input power for 50 and 60 Torr
at l % methane concentration ....................... 63
Temperature difference between silicon and silicon nitride for 0.5 ‘70
methane concentration at 50 Torr ..................... 64

xi

3.14

3.15

3.16

3.17

4.1

4.2

4.3

4.4

Fabrication of the four-point probe samples. The actual surface was
more uneven than is indicated here. (a) The starting point is a silicon
nitride substrate which has received a nucleation procedure. (b) Since
the substrate is insulating, the sample is ready for the 4 point probe
measurement. ...............................
Fabrication of the metal/ diamond/ silicon samples. (a) The surface is
nucleated by diamond powder or diamond paste, (b) the diamond film
is deposited, and (c) top contacts are evaporated through a shadow
mask. Aluminum is shown here as an example ..............
Fabrication of back-etched diamond samples for surface analysis. (a)
Silicon is coated with diamond using conventional method. (b) The
sample is secured to an epoxy substrate. (c) The silicon is removed by
chemical etching. .............................
Fabrication of dual-sided metal contacts on isolated diamond films.
(a) The silicon is coated with diamond and (b) a metallic contact is
evaporated on the first diamond surface. (c) A wire lead is attached
to the evaporated metal and (d) the substrate is placed face down in
epoxy. (e) The silicon is removed by etching and (f) metal contacts
are evaporated on the second diamond surface ..............

Raman spectroscopy is based on a process of inelastic scattering of
photons by phonons. ............ . ..............
Raman spectra for diamond and silicon on (a) absolute wave number
scale and (b) relative wave number scale .................
Raman Spectroscopy set-up for the Raman spectrum analysis.(From
Dr. Gray) .................................
Raman spectra of 0.5% methane samples at (a) 936 °C (b) 1040 °C'

- and (c) 1090 °C. (From Dr. Gray) ...................

4.5

4.6

4.7

Raman spectra of samples with conditions : (a) 0.5% methane, 1000
°C’ (b) 0.5% methane, 1040 °C (c) 1% methane, 1000 °C and (d) 1%
methane, 1040 °C. (From Dr. Gray) ..................
Raman spectra of samples of 0.5% methane at 1000 °C’ for (a) 10 hours’
deposition and (b) 6 hours’ deposition. (From Dr. Gray) .......
Raman spectra of samples with 0.5% methane at 1040 °C’ by (a)
powder-polished method and (b) paste-polished method. (From Dr.
Gray) ...................................

xii

66

67

69

70

74

76

78

80

81

82

4.8 Raman spectra of (a) the as-deposited film and (b) the free standing
film of the sample with 1.5% methane at about 1040 °C. (From Dr.
Gray) ...................................

4.9 Principle of X-ray Photoelectron Spectroscopy. When a core level elec-
tron absorbs a photon of energy (hu) greater than its binding energy

(E3), the electron is ejected from the atom with kinetic energy (K.E.)

4.10 ESCA spectra of samples exposed to air for (a) a long time and (b) no
time before the analysis. (From Dr. Hook) ...............
4.111 (a) Elemental survey scan of the sample before the Ar sputtering (b)
higher resolution narrow scan of (a) in the carbon region. (From Dr.
Hook) ...................................
4.12 (a) Elemental survey scan of the sample after the Ar sputtering (b)
higher resolution narrow scan of (a) in the carbon region. (From Dr.
Hook) ...................................
4.13 SEM view of the paste-polished produced diamond film on a cleaved
sample showing the silicon substrate, the interface, and the side view
and top surface of the diamond film. ..................
4.14 The surface profile on (a) the top surface of an as-deposited paste-
polished produced diamond film and (b) the back surface of the film
after transfer to the epoxy substrate. ..................
4.15 SEM view of the powder-polished produced diamond film on a cleaved
sample showing the silicon substrate, the interface, and the side view
and top surface of the diamond film. ..................
4.16 The surface profile on (a) the top surface of an as-deposited powder-
polished produced diamond film and (b) the back surface of the film
after transfer to the epoxy substrate. ..................
4.17 The extreme top surface profiles of (a) large-grain-size and (b) small-
grain-size diamond films ..........................
4.18 Typical SEM photo of the top view of the diamond film prepared by
the paste—polished method. The triangular shapes indicate [111] crys-
tallite faces. (MW power: 600 W, plasma pressure: 60 Torr, methane
concentration: 0.5%.) ...........................
4.19 Typical SEM photo of the top view of the diamond film prepared by
the powder-polished method. (MW power: 600 W, plasma pressure:
60 Torr, methane concentration: 0.5%.) .................

xiii

84

86

88

90

91

93

95

98

101

102

 

 

 

 

 

 

4.20 Growth mechanism of diamond film prepared by (a) paste-polished

and (b) powder-polished method. The latter has a higher nucleation

density and therefore a finer grain film .................. 103
4.21 This SEM photo shows large and small grain size diamond growth on

the same substrate ............................. 104
4.22 (a) Cross section view of laser scan from plane A to D. (b) Thickness

(pm) uniformity analysis on an area of 1 cm X 2 cm of a sample. . . 112
5.1 The set-up of the four point probe method ................ 115
5.2 Raman spectra of (a) NDF-S3, (b) NDF—S4, and (c) NDF—S6. (From

Dr. Holtz) ........................ ' ......... 120
5.3 Raman spectrum of N DF-Sl. (From Dr. Gray) ............ 121
5.4 Conductivity vs. 1000/ T of N DF -Sl .................. 125
5.5 Conductivity vs. 1000/ T of N DF—S3 .................. 126
5.6 Conductivity vs. 1000/ T of N DF—S4 .................. 127
5.7 Conductivity vs. 1000/ T of N DF-S6 .................. 128
5.8 Raman spectrum of the diamond film as deposited on the silicon sub—

strate with substrate temperature at 1000 °C. (From Dr. Gray) . . . 131
5.9 The I-V measurement set-up of the back-etched samples. ....... 133

5.10 I-V characteristics of a small grain size sample with gold contacts top
and bottom. It shows linear and symmetric behavior at low voltages.
(Electrical sample : P47-Tl) ....................... 134
5.11 I-V characteristics of a small grain size sample with gold contacts top
and bottom. It shows nonlinear behavior at high voltages. (Electrical

sample : P47-T1) ............................. 136
5.12 The I — V3/2 characteristic and (b) the I - V2 characteristic of Figure
5.11. (Electrical sample : P47-Tl) .................... 137

5.13 I-V characteristic of diamond film with gold contacts top and bot-
tom. Experimental data is shown by data points. The solid line
represents the model results of Eq. (5.12) with (27¢)o=2.6x10'10 It“,
Go = 2.1x10‘11 0“, and a = 0.042 V". (Electrical sample : P47-T1) 139
5.14 I-V characteristic of diamond film with top indium contacts and
bottom gold contact. Experimental data is shown by data points.
The solid line represents the model results of Eq. (5.12) with
Goo=2.1><10‘1° fl“, Go = 2.0x10‘11 9“, and a = 0.045 V“. (Elec-
trical sample : P47-T3) .......................... 140

xiv

5.15

5.16

5.17

5.18

5.19

5.20

5.21

5.22

5.23

5.24

5.25

5.26

5.27

I-V characteristic of diamond film with silver contacts top and bot-
tom. Experimental data is shown by data points. The solid line
represents the model results of Eq. (5.12) with Goo=2.9x10'1° Q“,
Go = 4.0><IO"12 fl“, and a = 0.08 V”. (Electrical sample : P49-T2) 141
The photoconductance is nearly independent of the applied voltage.
Shown here are data for different photon fluxes corresponding to dif-
ferent optical density (O.D) filter values. (Electrical sample : P47-T1) 143
(a) The calibration factors corresponding to different optical wave-
length for Si and Ge photodetector. (b) The relative power intensity
factors for Si and Ge photodetector at different wavelength of the op-
tical monopass filters ............................ 148
Relation of photoconductance vs. photon energy for Au-Au contact
sample (a) before normalization and (b) after normalization. (Electri-
cal sample : P47-T1) ......................... i . . 149
Relation of photoconductance vs.- photon energy for In-Au contact
sample (a) before normalization and (b) after normalization. (Electri-

cal sample : P47-T3) ........................... 150
The I-V measurement set-up of the metal / diamond/ silicon samples. . 152
I-V characteristic of the small grain size Au/ diamond/ silicon samples.
(Electrical sample : P39) ......................... 154
I-V characteristic of the small grain size In/ diamond/ silicon samples.
(Electrical sample : P39) ......................... 155

I-V characteristic of the large grain size Al/diamond/silicon samples.
The average grain size is 2.1 pm. (Electrical sample : P6) ....... 156
I-V characteristic of the large grain size Al/diamond/silicon samples.
The average grain size is 1.4 pm. (Electrical sample : P7) ....... 157
Photoconductance versus photon energy (a) before normalization and
(b) after normalization for the small grain size Au/diamond/silicon
sample. (Electrical sample: P39) .................... 159
Photoconductance versus photon energy (a) before normalization and
(b) after normalization for the small grain size In / diamond / silicon sam-
ple. (Electrical sample: P39) ...................... 160
Photoconductance versus photon energy (a) before normalization and
(b) after normalization for the large grain size Al / diamond/ silicon sam-
ple. The average grain size is 2.1 pm. (Electrical sample: P6) . . . . 161

XV

 

 

 

5.28
5.29

5.30
5.31

5.32
5.33
5.34

5.35
5.36

5.37
5.38
5.39
5.40

5.41

5.42

 

—--- ~ «-‘:‘,— ...- A

Photoconductance versus photon energy (a) before normalization and
(b) after normalization for the large grain size Al / diamond / silicon sam-
ple. The average grain size is 1.4 pm. (Electrical sample : P7) . . . .
Photoconductance versus photon energy (a) before normalization and
(b) after normalization for the large grain size Al / diamond / silicon sam-
ple. The average grain size is 1.1 pm. Electrical sample: P20) . . . .
The energy band diagram at the Al/p-type diamond interface. . . . .
The I-V characteristic of the point contact / diamond/ silicon structure
showed ohmic behavior. The top contact is a tungsten point probe and
the bottom contact is the silicon wafer. (Electrical sample : P20)

The best Schottky barrier diode (SBD) characteristic of the large grain
size Al / diamond / silicon samples. The top contact is aluminum and the
bottom contact is the silicon wafer. (Electrical sample : P20) .....
The Raman spectrum of N DF-P20. The peak between 1550cm"1 and
1600cm’1 indicates that the existence of graphitic component in the

The SBD I-V characteristic of a film with coplanar surface contacts.
One contact is aluminum and the other contact is a tungsten point
probe. (Electrical sample : P20) .....................
The log(I)-V characteristic of the SBD. (Electrical sample : P20) . . .
The forward bias I-V characteristic of the SBD. The 0 represents the
experimental data, the solid line shows the ideal diode with n = 1,
10:5.2x 10‘16 A and the dashed line shows the ideal diode with 17 = 2,
10:2.9x10‘1‘ A. (Electrical sample : P20) ...............
Conductance-bulk voltage characteristic. It follows the field activated
model above 10 V. (Electrical sample : P20) ..............
The I-Vb3/2 characteristic of the SBD. (Electrical sample : P20)

The I-Vg,2 characteristic of the SBD. (Electrical sample : P20)

(a) The energy band diagram of the SBD. The barrier height 45,, equals
to the sum of built-in potential V.- and activation energy E... (b) The
equivalent circuit of the SBD. ......................
Capacitance versus frequency characteristic at different reverse bias.
(Electrical sample : P20) .........................
Capacitance versus reverse bias characteristic at 50 kHz. (Electrical
sample : P20) ...............................

xvi

162

163
164

166

167

169
171

173
174
176
177
178

180

182

CHAPTER 1

Introduction

1.1 Motivation for Diamond Research

Diamond is well known as one of the most valuable gems and possesses many excellent
properties, such as an extremely high degree of hardness, high thermal conductivity,
high electrical resistivity, chemical inertness, and good optical transparency. When
properly doped with impurities, diamond also has potential as a useful semiconducting
material.

Currently in industry, diamonds are mostly used in hard-surfacing and abrasive
applications. The general methods to get synthetic diamond are variations on the
high pressure and high temperature technique (HPHT), reported by Bundy et a1. [1]
at General Electric several decades ago. That process is a mature and economically
feasible technology, but it has some limitations. First, commercial HPHT diamond
synthesis needs a large, expensive system. Secondly, low impurity diamond is rather
difficult and costly by this method. Finally, and most important it is only suitable
for production of diamond in the form of small pieces, grit, and powder. It is not
practical for many technological applications where a continuous film is required.

Now there appears to be a high likelihood that these limitations can be overcome

by new diamond thin film deposition techniques which allow large area coating with

1

lower cost. These new methods are based on metastable synthesis of diamond at
pressures of an atmosphere or less. So far this new method of synthesis has moti-
vated a major international research effort aimed at developing techniques for high
quality, low cost, large area, and high deposition rate diamond thin film production.
This research raises the possibility of many new potential applications in mechanical,
electronics, and optical industries.

Following are some of the potential diamond film applications:

* Tribology and abrasive coating applications.

* Thermal heat sinks.

* Electronic packaging and passivation.

* Electrical isolation.

* High temperature, high speed electronic devices.
* Optical windows.

* Microwave and millimeter wave power devices.

* Corrosion resistant applications.

Significant issues still remain that hinder widespread commercial applications of
diamond films. These include reproducibility, film quality, film uniformity, and high
deposition rate of the diamond films.

The. initial steps of this research were to develop techniques on new apparatus for
producing high quality diamond films. Subsequent steps focused on the correlation
between processing techniques and diamond film physical properties and specifically

on studying electrical properties of the diamond films.

1.2 Research Objectives

One objective of this research was to develop techniques for high quality diamond film
synthesis using a microwave plasma disk reactor (MPDR) chemical vapor deposition
system developed by Asmussen et a1. [2, 3] which had previously been proven success-
ful in plasma assisted etching of semiconductors [4, 5, 6, 7] and in plasma oxidation
techniques [8, 9]. This objective, which was part of a group effort, was met, and in
fact during the past year a commercial version of the MPDR was placed in use on an
industrial processing line.

A second objective was to correlate diamond film properties with substrate prepa-
ration methods and synthesis conditions, providing insight into the factors affecting
diamond quality. Two different nucleation methods were shown in this research to
produce films of similar diamond quality as measured by Raman spectroscopy, but
considerably different morphology and different electrical characteristics. A variety
of physical characterization methods were used for film analysis.

The third and final objective was to specifically focus on the electrical properties of
the films in order to provide insight into the relationship between synthesis variables
and electrical properties, and to investigate electron device related film properties.
Toward this objective, new sample preparation methods were used to study the role
of metal contacts. New information about the role of defects in limiting the use of
diamond films at very high electrical fields, above 105 V/cm, was provided by this
research. Also, the importance of nucleation procedures on the properties of metal-
diamond junctions, specifically on the properties of diamond Schottky barrier diodes,

was demonstrated.

1.3 Dissertation Outline

This dissertation is divided into four subjects. These are (1) background literature
review, (2) diamond film deposition and sample fabrication, (3) physical characteri-
zation, and (4) electrical properties analysis.

Chapter 2 of this dissertation reviews the diamond synthesis theory, diamond film
production methods, and previous diamond electronic device results from the lit-
erature. Chapter 3 describes diamond film deposition by using the MPDR chemical
vapor deposition system and also describes the sample preparation techniques used in
this research. Chapter 4 reviews characterization results of the diamond films by Ra—
man spectroscopy, scanning electron microscope, X-ray photoelectron spectroscopy,
and surface profiling, The electrical conductivity versus temperature of the diamond
films as measured by the four point probe method is presented in chapter 5 (section
2). Chapter 5 (section 3 and 4) also analyzes the electrical contact effects on back
etched diamond samples and on diamond / silicon structures. Chapter 6 concludes this

work with a summary of important results and discussions for future research.

. 0..

CHAPTER 2

Synthetic Diamond : Background

Review

2.1 Introduction

This chapter begins by describing the historical background of diamond synthesis from
high pressure high temperature methods to low pressure low temperature metastable
production methods. Then the growth mechanisms of the metastable production
methods are covered. In chapter 2 (section 3), several categories of the diamond film
production technologies are discussed, such as chemical vapor deposition techniques,
plasma assisted chemical vapor deposition techniques, and ion beam deposition tech-
niques. Then several nucleation methods are also described. The physical attributes
of diamond, such as material properties and semiconductor properties, and some of
the potential applications are explained in chapter 2 (section 4). Finally, some of the

diamond device work is reviewed in chapter 2 (section 5).

h?

(“(15

I‘-
"Cd;

‘L

2.2 Brief History of Diamond Synthesis

2.2.1 Historical Background

Diamond was first known to be a form of carbon in 1797. Later it was discovered that
diamond is formed by carbon atoms which are arranged in a particular crystalline
structure. It has a cubic lattice which is built up from sp3-hybridized, tetrahedrally
arranged carbon atoms. A much more common form of solid state carbon is graphite,
in which carbon atoms are arranged in a hexagonal crystalline structure. The graphite
lattice consists of layers of condensed spz-hybridized rings. The atomic structure of
diamond and graphite are shown in Figure 2.1. Most recently, however, scientists
have just found the third form of carbon, the fullerenes [11]. With the knowledge of
the fact that diamond is formed by the sp3 bonding arrangement of carbon atoms, a
scientific approach was open for research on synthetic diamond technologies.
Thermodynamic calculations based on the Gibb’s free energy show the conditions
under which carbon atoms prefer to form diamond or graphite. At ordinary tem-
peratures and pressures, diamond is a metastable form of carbon. This means that
diamond is not the most stable form of carbon compared to graphite in nature, be-
cause the hexagonal structure of graphite has a lower thermodynamic potential than
the diamond crystal structure. Therefore, graphite is favored in terms of thermody-
namic stability. However, at very high pressures, greater than 100,000 atmospheres,
the phenomenon is reversed and diamond becomes the preferred state. The phase
diagram 0f carbon, shown in Figure 2.2, illustrates the regions of temperature and
pressure corresponding to diamond, graphite, liquid carbon, and vapor carbon [12].
Although diamond is metastable, it does not degrade to graphite spontaneously
under ordinary conditions. A significant potential energy barrier prevents such degra-

dation except at very high temperatures as shown in Figure 2.3. However, the phase

(a) Diamond (sp’bonding)

 
 

.
~~
---
‘5

  

‘~
~‘~~
~
~

I
I’

‘
---
~~
Q
‘~
~
~.

 

 

 

 

 

 

Figure 2.1. The atomic structure of (a) diamond (b) graphite.

A N HT: .A.‘ v 3.: 3.4.7..th ~

Pressure (P.S.I)

--—---—------—-—-----------
nun-un--------------------------
------------------I--------
---l-------I-------I------
I-ll-I-III-I-I-l-I-I-IIIII-l-I-I-l-I-IIIII-I-I-I-I-II-

— \
=—"' \ ~‘\

107 ————-————‘

106::

105.—

  
  

104‘I

 

 

 

I
II

 

103i

 

102 1

10

---------------
.............

uuuuuu
.....
—-.--
:-.-u-

 
  

lIlllllfllllllllllllllllllllllllllllllllllll
llllllllllllllllllllllllllllllllllllllll
IlllllllllllllllllllllllllllllllllfllllllllllII

lllllllfllllI lllllllll lllllll

   

 

 

 

 

 

 

 

HE
(III

   

 

   
 
   

  
 

 

 

 

llllI

 

 

0 2,000 4,000 6,000 8,000 10,000

Temperature (°F)

Cl

Vapor

Figure 2.2. Phase diagram of carbon.[12]

Ԥ
Liquid

Milli]

Graphite Diamond

E

diagram does show that at ordinary conditions graphite can’t transform to diamond
because graphite has the lower thermodynamic potential under such conditions.

Many experimenters in the 19th century and early 20th century tried unsuccess-
fully to make diamond by subjecting graphite and many other carbon compounds to
the conditions where diamond is the favored state of carbon, namely under high heat
and great pressure. Eventually, a group of researchers at GE successfully developed
the first technique to make ”synthetic diamond”, or ”man-made diamond” in 1955
[1].

This technique involves processing at high temperatures and high pressures, which
are in the range of 30,000-100,000 leg/cm2 and 1000-3000 °K respectively. It also in-
volves a metal, such as iron, nickel, cobalt, manganese, chromium, or tantalum to
serve as a solvent and catalyst. The GE work not only lead directly to present indus-
trial production of synthetic diamond for abrasive applications, but also to advanced
carbon research in general with more understanding of the carbon phase diagram.

There is another important high pressure method to synthesize diamond which is
the so-called shock wave technique. Diamond containing material was achieved first
by means of shock waves by Paul S. DeCarli and John C. Jamieson in 1961. This
technique, industrialized by Du Pont, utilizes the high pressure in the shock wave of
explosions to directly convert crystalline graphite into diamond. The particles are
then separated from the starting material by sedimentation [13].

With the success of these two techniques, over 30 tons of industrial diamond abra-
sive grain are made each year around the world. However, it has long been recognized
that there is a possibility that diamond may be formed in the metastable portion of
the phase diagram , i.e. in the graphite stable region, under the conditions where
”nascent” (un-combined) carbon atoms are liberated [12]. Such synthetic conditions

would be easier to achieve and also better suited for diamond film coating at relatively

MC

...

A
0.12

10

Energy A

 

0.02 eV ..................

I 1 , \ Atomic
7 Spacing

 

Diamond Graphite Vapor

A E( free carbon. graphite ) = 170 Kcal/g-atom

A E( diamond. graphite ) = 0.5 Kcal/g-atom

Figure 2.3. Illustration of free energy difference between diamond and graphite.

low cc
PO
1970's

I

31 {Eli

iment.
Ion be
Here

AQI.
'- ill.

11

low cost over larger areas.

Following this spirit of thought, attempts were made in the 1950’s, 1960’s, and
1970’s to form diamond films on non-diamond substrates by depositing carbon atoms
at relatively low pressures and low temperatures under a variety of conditions.

The first results of the diamond film work were the development of a few exper-
imental methods to form diamond-like-carbon (DLC) films by vapor deposition or
ion beam deposition. [14, 15, 16, 17] However, the atomic structure of DLC films is
different from that of diamond films. The difference will be explained in chapter 2
(section 4).

A significant breakthrough occurred when researchers at the Moscow Institute of
Physical Chemistry developed methods for depositing diamond films on both dia-
mond and, for the first time, non-diamond substrates. They suggested three different
approaches to achieve a ”super-equilibrium” with atomic hydrogen : catalysis, heated
A filament, and electric-discharge plasma [18]. However, their work did not get much
attention in the west until Japan’s National Institute for Research in Inorganic Ma-
terials (N IRIM) repeated the work and also reported true diamond films. The key
feature of this work is the addition of hydrogen gas associated with methane gas into
the deposition chamber [19]. This result confirmed the hypothesis that diamond syn-
thesis techniques can take place under the conditions in which diamond is metastable
with respect to graphite. The region for low pressure chemical vapor deposition
(CVD) technique of diamond synthesis is illustrated in Figure 2.4. Not only did this
development trigger many research efforts on diamond film technology but it also
pushed the synthetic diamond films towards industrial applications.

Since 1983, many groups successfully deposited diamond films or DLC films on
all kinds of non-diamond substrates by all kinds of techniques. Some representative

techniques and their experiment parameters will be described in details in chapter 2

Agatha! v aanawmmwhfiu

lit

NV

Pi

g 400

is 300
Q
X
a

i 200
e
O.

1 00

0

 

 

 

 

 

 

 

 

 

 

 

Shock wave Liquid carbon
synthesis
Diamond
Diamond a
metastable
graphite
High-pressure.
high-temperature
Catalytic -
high-pressure. symhes's
high-temperature
synthesis
Graphite a
metastable
. . diamond
Chemical vapor
,, . — . a . 3/ deposmon
0 1 000 2000 3000 4000 5000

Temperature ( 9C )

Figure 2.4. Low pressure CVD diamond synthesis in the phase diagram of carbon.[13]

13

(section 3).

2.2.2 Growth Mechanism of Diamond Films

In general, there are two ways to form metastable crystal:

1. If a stable crystal is brought into a new temperature or pressure range in which it
doesn’t transform into the more stable form;

2. A precipitate or transformation may form a new metastable phase. [20]

Metastable crystal will remain present because the high activation energy required
for the conversion into more stable phase causes a low rate of transition. This explains
why diamond doesn’t spontaneously degrade to graphite at ordinary temperature and
pressure.

The possibility of synthesis of diamond in the graphite region of the phase-diagram
is based on the fact that the free energy difference between diamond and graphite (0.5
Kcal/g-atom) is so small under ambient conditions. See Figure 2.3. There is some
probability that diamond and graphite can both nucleate and grow simultaneously
from nascent carbon atoms, especially under conditions in which kinetic factors can
dominate, e.g. high energy. If graphite can be prevented from forming, or if it can
be removed preferentially, diamond can be recovered. [21] A sampling of theories and
models for diamond growth mechanism from methane/ hydrogen mixture are briefly
described below.

In 1986, M. Tsuda, et al proposed the ”Epitaxial Growth Mechanism of Diamond
Crystal in CH4-H2 Plasma.” They believe that no phase transition of graphite or
amorphous carbon takes places under the conditions of low temperature plasma reac-
tion. Rather, the diamond formation in vacuum is considered to proceed by chemical
reactions which are different from the phase transition of graphite under ultrahigh

pressure at very high temperature. By a quantum chemical calculation a mechanism

14

was found to explain the process in two steps. First, the (111) plane of diamond
surface is covered by methyl groups (CH3) via methylene insertion or hydrogen ab-
straction followed by methyl radical addition. Second, epitaxial growth occurs when
three neighboring methyl groups standing on the ( 111) plane of the diamond surface
bind spontaneously to form the diamond structure via methyl cation intermediates
[22]. However, S. J. Harris also proposed a mechanism for homo-epitaxial growth on
an finite (100) diamond surface from the methyl radicals (CH3) [23].

In 1987, M. Frenklach and K. E. Spear proposed a different growth mechanism of
vapor-deposited diamond. This central postulate is that the monomer growth species
is ”acetylene” (CgHz). This mechanism basically consists of two alternating steps.
The first is surface activation by the H abstraction of a bonded hydrogen atom from
a surface carbon. Then the activated carbon radical acts as a site for the addition of
one or two acetylene molecules. During the addition reaction cycle a number of solid
C-C bonds are formed and hydrogen atoms migrate from a lower to a upper surface
layer. This paper also discusses the two possibilities for the production of graphite
on a growing diamond layer. First, graphite precursors may form in the gas phase
followed by surface deposition and further growth. Second, the initiation and growth
of graphite may happen directly on the surface [24].

Additional insight into growth mechanism is provided by the fact that methane
and hydrogen is not a unique recipe for diamond film growth. For example, Hirose
and Terasawa reported that diamond thin films with good crystallinity and quality
were quickly synthesized by thermal CVD using several organic compounds, some of
which include oxygen [25]. However, the specific roles of oxygen were not discussed
in that work.

T. Kawato and K-I Kondo were the first to specifically study the effects of oxygen.

With the addition of oxygen, the deposition of graphitic or amorphous carbon could be

15

suppressed, the growth rate of diamond was increased and the pressure range for the
synthesis of diamond was extended. It was suggested that these effects resulted from
the fact that the acetylene concentration is significantly reduced upon the addition
of oxygen [26].

S. J. Harris and A. M. Weiner also reported that 02 can reduce the effective initial
hydrocarbon mole fraction, which is important because higher quality diamond is
grown at a lower initial hydrocarbon mole fraction. Perhaps more importantly, it was
also noted that there is a possibility the formation of sufficient gas phase 0H will
remove non-diamond (pyrolytic) carbon from the film [27]. Y. Liou, et al. also found
that the addition of oxygen was critical for diamond growth at temperatures below
500 °C' [28].

It is likely that details of the growth mechanism depend on the synthesis method.
For example, with the acetylene flame technique, Y. Matsui, et a1. [29] proposed
that stable species such as C H.. and C2H4 are rapidly produced, followed by the
methyl radical formation. Then the C-radicals adsorbed on the diamond surface are
etched by H-atoms to form CH... The growth rate depends both on the substrate
temperature and on the CzHg/Oz ratio. This result is also confirmed by Y. Tzeng,
et a1. [30].

Although there are many postulates attempting to explain diamond synthesis,
the experimental evidence in support of any particular growth mechanism is limited.
However, there are several hypotheses most researchers believe that can explain parts
of the growth mechanism for the diamond synthesis. There are listed as follows.

First, the best quality diamond films have been generally obtained in a predom-
inantly hydrogen atmosphere, with only small amounts of hydrocarbon (typically
methane, 1 % or less) present in the reaction mixture. That is, the hydrogen atoms

should be in a super-equilibrium state for a successful diamond growth. Moreover, H

16

atoms play a very crucial role in enhancing the film growth. The effects of hydrogen
are described in the following [31].

1. Hydrogen acts as the etching agent for non-diamond carbon :

The formation of diamond kinetically competes with the formation of graphite. When
graphite and diamond are formed simultaneously, the atomic hydrogen created from
the hydrogen gas either by thermal energy or by electric energy is considered a pref-
erential etching agent of graphite over diamond.

2. Hydrogen provides stabilization of sp3 bonding :

Atomic hydrogen serves to satisfy the dangling bonds of surface carbon atoms, keep-
ing them in the sp3 configuration and thus preventing the diamond surface from
reconstruction into graphitic (spz) or carbonic (sp) structures.

3. Hydrogen promotes the main monomer of the gas-phase production for
diamond growth :

Either acetylene (CgHg) [24, 32] or methyl radical (CH3) [22, 23] is the main monomer
for the diamond growth and hydrogen is believed to support the formation of theses
species.

Secondly, the temperature dependence of the rate of diamond growth exhibits
a maximum. That is, it initially increases with temperature and then decreases.
Graphite formation rates also exhibit a temperature maximum, but at higher tem-
peratures than that of diamond.

Finally, the addition of oxygen not only increases the deposition rate but also
improves aspects of the film quality. The roles of oxygen are suggested as follows [32].
1. Oxygen may increase the H atom concentration which selectively etches
graphitic and amorphous carbons.

2. Oxygen may cause the reduction of pyrocarbon-forming species.

Consequently, it will increase the growth rate of diamond.

17

3. Oxygen may also act as a good selective etchant of non-diamond car-
bons.

Oxygen is better than hydrogen in the efficiency of etching graphite, however, the
addition of an excessive amount of oxygen may result in a decreased growth rate,
because the diamond crystals are also etched.

Besides the growth mechanism of the diamond synthesis, the nucleation mech-
anism still needs to be briefly considered. The initial nucleation mechanism of a
diamond crystallite is distinct from the mechanism for the extension of a preexisting
diamond lattice. Without special substrate preparation, the initial nucleus density
of diamond is extremely low, and a continuous film does not result. To improve the
amount of nucleation sites becomes necessary. Generally speaking, it is well recog-
nized that the nucleation rate can be enhanced prominently both by scratching the
substrate surface or by attaching small seed crystallites to the substrate before the
deposition. Nucleation methods will be described further in chapter 2 (section 3).

Until now, the understanding of the growth mechanism is only at the ice-breaking
stage. Many more experiments with better control of experiment parameters are

needed to verify the physics and chemistry of the diamond film synthesis.

2.3 Review of Diamond Film Technologies

In the past few years the potential applications of diamond films have stimulated a
great deal of research interest on various synthesis technologies, each of which has
advantages and disadvantages. A broad sampling of these technologies is briefly
presented in this section in which the synthesis methods are grouped into three cate-
gories. Theses are (1)hot filament assisted chemical-vapor-deposition (CVD);
(2)plasma assisted CVD; and (3)0ther deposition methods. Additionally, in

the last part of this section, several nucleation methods will be described.

18

2.3.1 Hot Filament Chemical Vapor Deposition Techniques

Generally, CVD is a reaction in which two types of gas, C(g) and D(g), react at a
high temperature to form a solid phase A(s) and a gas phase B(g). Sometimes, more
than two kinds of gases are introduced for certain applications. One way of enhancing
the reaction rate is to add thermal energy. In hot filament CVD the thermal energy
comes from a hot filament located in the reaction chamber into which the gases are
introduced.

In 1982, Matsumoto, et al. reported synthesis of microcrystal diamond using a hot
filament CVD process from a mixture of methane and hydrogen gas. This technique
involves the use of a hot tungsten filament in the vicinity of the substrate. The
hot filament causes the dissociation of the gas as well as heating of the substrate,
providing appropriate conditions for the growth of diamond [19].

A modified version of the thermal CVD method has been used by Sawabe, et al.
to improve the growth rate of diamond films. It involves the electron bombardment
of the substrate surface by biasing the substrate positively with respect to the hot
filament. It is suggested that the decomposition of C H4 and H; by electrons occurs
mainly at the substrate surface, because the scattering cross section for electrons
is fairly small in space under these conditions. It is also found that the number of
nucleation sites on the substrate surface is increased by electron bombardment. Thus,
growth rates as‘high as 3 - 5 pm/hour were achieved [33].

Hirose and Terasawa experimented with many organic gases other than methane
using the hot filament CVD method. These experiments were based on their hypothe-
sis that the methyl radical (C H3) with sp3 hybridized orbits and atomic hydrogen (H)
play important roles in the synthesis of diamond. Consequently, they choose methyl
alcohol (C H30H ), ethyl alcohol (CszOH), acetone (C H3C'OCH3), dimethyl ether

(CgHsOCsz) , and some other organic gases which can generate methyl radicals

19

more easily in order to increase the growth rate. They found that good quality dia-
mond films can be grown on silicon substrates with high growth rates (8 - 10 um / hour)
under the pressure range 1 - 800 Torr. The growth rate is about 10 times faster than
thermal CVD method using CH4 or CgHg gas [34].

H. Aikyo and K.-I. Kondo reported diamond synthesis by a new hot filament
CVD method in which methane is decomposed mainly by an abstraction reaction of
CH4 with atomic hydrogen. The film obtained by this method was compared with
that obtained by the method which involves the thermal decomposition of CH, near
a filament. They found that the film quality of the former method was better in
terms of the ratio of crystalline to amorphous carbon, but slightly lower in deposition
rate compared to that by the latter method. They hypothesize that production of
acetylene (CgHg) is suppressed by the separate introduction of the gases and that
there is consequently an improvement in the film quality [35].

Y. H. Lee, et al. used the bias-controlled thermal CVD method to do diamond
synthesis. They found that high quality films with highly facetted cubo—octahedral di-
amond growth was observed under low current bias conditions, in which the substrate
is negative with respect to the filament. In contrast, multiply twinned microcrystal
growth incorporating spz-bonded carbon was obtained under high current bias condi-
tions. They suggested that this is due to a decrease in electron and ion bombardment
of the sample during the low bias conditions which minimizes surface damage. Clearly
this observation is quite different from Sawabe’s point of view about the role of the
electrons [36].

Lately, M. Ihara, et al. reported that they successfully deposited good quality
diamond on silicon substrates at a temperature as low as 135 °C by controlling the
temperature of the substrate holder. A tantalum filament was typically used and

maintained at 2300 °C in this experiment [37].

20

There are some advantages and limitations associated with the hot filament CVD
process. The fact that the apparatus is inexpensive and easy to operate is the most
attractive advantage. However, film non-uniformity due to the high temperature
gradient of the filament and the difficulty to scale up the size limits the possibilities
for a production process. Also, the hot filament can be a source of impurities in the

films.

2.3.2 Plasma Assisted Chemical Vapor Deposition Tech-
niques

Plasma assisted chemical vapor deposition (PACVD) processes used for the synthesis
of diamond and diamond-like films involve the decomposition and dissociation of
hydrocarbon gas in a plasma by direct-current (dc), radio-frequency (rf), or microwave
excitation. Several variant techniques of the PACVD process as applied to diamond

are described in the following.

DC Plasma CVD Techniques

In 1987, K. Suzuki, et a1. first successfully deposited good quality diamond films by
the DC plasma parallel plate CVD method. Films with growth rates of about 20
pm/ hour were grown on silicon and a-alumina substrates from a methane/ hydrogen
mixture gas at a pressure of 200 Torr without surface scratching by diamond or c-
BN powder. By experiment they also observed that electrons could enhance the film
growth. [38] Later epitaxial growth of diamond thin films on cubic boron nitride
(111) surfaces [39] and deposition of diamond onto an aluminum substrates at low
temperature (below 480 °C) [40] by dc plasma CVD technique were also reported by
other groups.

K. Kurihara, et al. reported diamond synthesis with the use of ,a DC plasma jet

21

(chamber pressure: 100 - 400 Torr). A plasma jet, formed by the dc arc discharge of
CH. diluted with Hg, was sprayed onto a water—cooled substrate. Because of the high
density of radicals, a high growth rate of 80 um / hour was achieved by this technique
[41].

Diamond films synthesized by dc arc discharge plasma CVD in a hydro-
gen/argon/ethanol mixture gas was reported by F. Akatsuda, et al. (chamber pres-
sure: 200 - 400 Torr). The growth rate of diamond films was reported to be about
200 - 250 pm/ hour which is higher than that of other CVD techniques [42].

S. Matsumoto, et al. also studied the substrate bias effect on diamond deposition
by the dc plasma jet method. When they applied a positive bias voltage to substrates
in a dc plasma jet of the Ar-Hg-CH4 system, the deposition rate increased more
than twice, and a maximum rate of 15 pm/ min was obtained. The deposition area
also increased but the uniformity of film thickness did not improve. They deduced
that positive bias on the substrate enlarges the plasma region near the substrate
by inducing a secondary discharge between the arc plasma and bias electrode. As
the concentration of activated species increased near the substrate, the deposition
of diamond also increased. However, too much bias current changes the secondary
discharge from glow-like to an arc discharge, causing the over heating of the substrate

and deposition of graphite instead of diamond [43].

RF Plasma CVD Techniques

In 1987, S. Matsumoto, et al. developed a technique using an rf (4 MHz) induc-
tive heating of an argon-hydrogen-methane mixture plasma under 1 atm pressure.
Micro-crystals of diamond and continuous polycrystalline films were deposited on
molybdenum substrates with a 1 pm / min deposition rate, however the films were not
uniform in thickness and had poor adhesion [44]. _
D. E. Meyer, et al. studied the 13.56 MHz inductive coupled plasma CVD tech-

22

nique. They found that the diamond-like carbon scratching only enhanced the forma-
tion of diamond particles on silicon and niobium, but not on stainless-steel substrates

[45].

Microwave Plasma CVD Techniques

Based on the number of journal articles, the literature indicates that the microwave
plasma CVD process is the most often used technique by diamond film researchers.
Various techniques for this approach are described in the following.

For microwave (usually 2.45 GHz) plasma excitation, typically a quartz tube is
positioned inside a microwave waveguide, a gas flow is introduced through the tube,
and a plasma is produced and confined in the center of the tube. Substrates are
mounted in the tube and exposed directly to the plasma which is formed at a gas
pressure in the range of 10 to several hundred Torr.

In 1983, M. Kamo, et al. reported diamond synthesis from the gas phase in a
microwave plasma [46]. With a methane/ hydrogen plasma the deposition rates were
approximately 0.5 - 1 pm/ hr. A number of subsequent investigations used similar
apparatus for diamond synthesis or used a variation where a bell jar is placed in a stub-
tuned microwave reactor. For example, single-crystal growth rates over 20 um/ hour
were obtained from a methane/ hydrogen /oxygen gas mixture by C.-P. Chang, et al.
[47]. I. Watanabe and K. Sugata used a microwave plasma to synthesize diamond with
different organic gases. Films with deposition rates higher than methane/ hydrogen
plasmas and with a uniform film thickness over a 5 cm by 3.3 cm area were achieved
[48]. J. S. Ma, et al. studied the selective nucleation and growth of diamond on a
SiOg dot-patterned Si substrate exposed to a microwave CO/ C H4 / H2 plasma [49].
Epitaxial growth of high quality diamond film on diamond substrate [50] and local
epitaxial diamond growth on Si (100) substrate [51] have also been studied by different
microwave plasma CVD (MPCVD) techniques. I

23

In an alternative microwave approach, Y. Mitsuda, et al. developed a microwave
plasma torch (MW plasma jet) system, which can generate a pure H2 plasma jet
at atmospheric pressure from the end of a center electrode. The electric field has
maximum strength at the end of the electrodes, so the plasma is initially ignited by
the electric breakdown. Then the plasma jet was powered and sustained by the elec-
tromagnetic field generated between the electrodes and / or in the chamber. Diamond
films were successfully synthesized at a high rate (30 pm/ hour) on an area of 2.5x2.5
cm2 from a AT°H2~CH4 mixture gas [52].

Recently, low temperature (S 500 °C) CVD of diamond films has been addressed
as an important task for diamond synthesis because many materials will melt or be-
come unstable at high temperature deposition conditions. Microwave plasma systems
have played an important role in this effort. Usually, low temperature deposition is
achieved by lowering the gas pressure and the microwave power from normal depo-
Isition conditions. For example, a magneto-microwave plasma for deposition of wide
area diamond films was reported by Hiraki and his coworkers. In this case, the pres-
sure was reduced to 0.1 Torr and a magnetic field higher than that required for ECR
resonance was added to the plasma. The plasma, which is 16 cm diameter wide, filled
the reactor. High quality diamond films were formed on a positively biased silicon
substrate from a mixture gas of CH4/Hg or C'O/Hg [53]. Later, they also successfully
deposited low temperature (500 °C)diamond films on Al substrates at 0.1 Torr from
the CH4/002/H2 mixture gas [54].

Growth of diamond films on silicon at 500 °C has been accomplished in a MPCVD
system by W. L. Hsu, et al. [55]. Y. Liou, et al. also found that they could deposit
diamond films on silicon, MgO, and fused silica at about 400 °C in a bell jar MPCVD
system [56]. D. J. Pickrell, et a1. successfully deposited diamond both in the plasma

and downstream at about 400 °C by using the same system [57]. However, all low

24

temperature techniques have the disadvantage of extremely low growth rates because
of the reduction in the plasma density.

Recently, Norton Company has begun a unique microwave reactor approach, called
the microwave plasma disk reactor (MPDR), which was developed at Michigan State
University. This is the reactor type used for the research in this dissertation. It is

described in detail in chapter 3 (section 2).

2.3.3 Other Deposition Methods

In addition to hot filament and plasma approaches to diamond thin film deposition,
a number of other interesting methods have been reported. At Rice university, re-
searchers showed that diamond can also be deposited by using the mixture gas of
halogen, hydrogen and carbon atoms on a variety of substrates at temperatures as
low as 300 °C' [58]. In perhaps the simplest procedure in terms of experimental ap-
paratus, acetylene torches (02H; and 02) have been used for diamond synthesis. Ion
beam deposition, in which 0* ions are directed at a substrate, have been reported
by several groups [14, 15, 59, 60, 61, 62, 63, 64] for ”diamond-like” carbon films.
These films do not have the diamond crystal structure and the 1332 cm‘1 Raman
signal, but they do have some diamond-like properties in terms of transparency and
hardness, although generally to a lesser extent than diamond. The method appears
less successful at this time for true diamond synthesis. In all diamond film deposition
approaches, the challenge is to produce high quality diamond films over large areas

with useful deposition rates.

2.3.4 Nucleation Methods.

In chapter 2 (section 2) it was noted that enhancing the number of nucleation sites

on the surface of the substrates is a precursor to most diamond deposition processes.

25

Typically the substrates are either abraded with diamond powder or diamond paste,
in the size of approximately 0.1 pm to 40 pm, or seeded with small diamond crys-
tallites in order to increase the number of nucleation sites. After a surface abrasion
treatment, a cleaning procedure is always followed to remove residue and minimize
the contamination. Substrates are cleaned by one or more organic solvents, such as
acetone, methanol, ethyl alcohol, etc. Sometimes, ultrasonic treatment or acid
solutions are also used in order to clean more eflectively. There are many kinds of
nucleation methods [ll-70]. Only a sampling of nucleation methods will be described

in the following.

1. C. P. Chang, et al. described that they abraded polished silicon substrates with
diamond powder (0.5 pm), rinsed the substrate in deionized water and blew it dry
[47].

2. I. Watanabe and K. Sugata reported that the substrates were roughened for 40
min by diamond powder with a particle size of 40 um suspended in ethyl alcohol
where ultrasonic waves were passing [48]. I

3. H. Shiomi, et al. reported that the substrates were polished with a diamond paste
of 1/4 pm size for 5 min and were de-greased in acetone. Then the substrates were
treated in a 1:1 solution of HF, HN03 and in a 1:3 solution of HN03, HCl. Sub-
strates were finally treated with deionized water and acetone [50].

4. J .-I. Suzuki, et al. reported that the substrates received ultrasonic treatment for
30 min in a suspension of diamond powder (20 - 40 um) and ethyl alcohol. After
that, they were cleaned in organic solvents [53].

5. Dr. Richard Guarnier, senior engineer in IBM Watson research center, described
an electrophoresis seeding method [65]. Diamond powder (0.1 pm) is first put in an

ethyl alcohol solution and mixed homogeneously. The substrate is positively biased

26

at 40 V and placed face to face with a 2 cm gap away from a reference electrode for
about 30 sec. The substrate is gently taken out of the solution and dried on a spinner.
As a result, nuclei are homogeneously seeded on the substrate. In this case there is
no subsequent cleaning procedure since that would remove the seeds. In a similar
technique, the diamond powder is dispensed on the substrate by an aerosol mist.

6. Professor Aslam and colleagues have described a seeding method whereby diamond
powder is in solution with photo-resist, which allows patterned seeding and patterned
growth of diamond films [66].

For the nucleation techniques involving direct seeding by diamond powder, the nu-
cleation action is clear. The diamond particles serve as sites for homo-epitaxial growth
and the crystallites size increases. Eventually a continuous polycrystalline film results.
For nucleation methods which involve abrasion, the nucleation action is less obvious
since microscopy generally indicates a clear surface. However, recently S. Iijima, et
al. [67] reported that extremely high power transmission electron microscopy (TEM)
indicates that abrasion does, even after cleaning, leave very small plaques of diamond,
S 50A in size, that serve as homo-epitaxial growth sites for individual crystallites
in the polycrystalline film. Therefore a hypothesis is that nucleation methods work
by providing sites for homo-epitaxial growth. However, it should be noted that the
literature does contain some cases where nucleation of substrates is claimed without
the use of diamond powder at all. For example, D. E. Meyer, et al. reported that
all the substrates were polished or roughened by sandpaper, sand-blasting, or silicon
to silicon abrasion. Substrates were cleaned by sonicating in trichloroethylene and

acetone [45].

27

2.4 Physical Attributes of Diamond

2.4.1 Comparison of Bulk and Film Material Properties

The great interest in diamond is due to its several extreme properties. It has the
highest atom number density of any material. Because of its high atom number
density and the strong covalent bonding, diamond has the highest hardness and elastic
modulus of any material and is the least compressible substance known [64]. These
properties contribute to its wide use as an abrasive and cutting material.

But, if you look further into diamond’s physical properties as listed in Table 2.1
[68], you will find that the thermal conductivity of diamond is higher than that of
any other material, even 5 times that of copper at room temperature. Its thermal
expansion coefficient is very low. Electrical resistivity is extremely high and chemical
reactivity is extremely low. The combination of these excellent properties makes
diamond unique, special, and attractive for many applications.

However, these are the properties of bulk natural diamond. There are some dif-
ferences between natural diamond and synthetic diamond films which are synthesized
by the different kinds of techniques mentioned in chapter 2 (section 3). Those major
difference are shown in Table 2.2 [13]. The most prominent difference is that the
properties reported for natural diamond are for single crystalline material, but syn-
thetic diamond films on non-diamond substrates are polycrystalline. ”Diamond-like”
carbon is often amorphous. Consequently, the properties of synthetic diamond films
are often not as great as natural diamond. Thermal conductivity, electrical resistivity,
and hardness are smaller depending on the techniques used for the diamond synthesis
and their experimental parameters.

It should be noted that the ”diamond films” produced by those techniques may

exist as composites of crystalline diamond and /or amorphous diamond-like carbon

28

 

 

 

Property Diamond’s Value

Chemical Reactivity Extremely Low

Hardness 9000 kg / mm'2

Heat Conductivity 20 W / cm / degree K @ 30° C
6

Tensile Strength
Compressive Strength
Themal Expansion 'Coeff.
Refractive Index
Transmissivity

Friction Coefficient

Band Gap

Electrical insulation
Young’s Modulus

0 .5x10 psi (natural)
14x 106 psi (natural)
.8 x 10'6/ degree K

2.41 @590 nm

225 nm-farlR

0.05 (dry)

5.4 eV

16 ohm-cm (natural)

10
10.35 x 1012 dynes/cm2

 

Table 2.1. Physical properties of bulk natural diamond.[68]

 

29

 

 

 

 

 

 

 

 

 

 

 

 

 

Diamond and
P’°P‘"y 5"“ ”mm" Diamond-Ilka Films
Face-centered Amorphous,
Crystal stnicture cubic micrograined
polycrystalline
Densityigm I cm 3) 3.51 1,3 - 3,4
Hardness ( Mohs ) 10 7 - 9
Coefficient of friction
( against Steel ) 0.05 ' 0.15 0.002 " 0.2
Refractive Index , 2.42 1.5 - 3.0
Optical transparency Yes Yes, but shows
(visible region ) interference color
Optical transparency
(infrared rggion ) Yes Yes
. . . Type la, b, and
Eiectncal resrstiwty "a, 10,5 102_ 1014
( Ohm.cm ) ' 3
‘ Type lib: 10
Thennai conductivity
(W/cth) 1° 2° 5'18
. Most inorganic Many inorganic
Chemical Inertness acids and solvents

 

acids and solvents

 

 

Table 2.2. Properties of bulk natural diamond and synthetic diamond films.[13]

 

30

(i-C) and/or graphite. The relative concentration of these phases depend on the
synthetic processing conditions. Likewise, the properties of the film strongly depend
on the relative concentration of each respective phases. The characteristic properties
of these three phases are given in Table 2.3 [21].

So far, we have mentioned diamond films and diamond-like films very often. Some-
times in the literature, particularly for early papers, the terms ”diamond films” and
”diamond-like films” are used without a clear distinction. But most researchers are
trying to clearly specify these two as different categories. Table 2.4 carefully sets the
definitions for diamond films, diamond-like films, carbon and graphite coatings [69].

It should be noted that film properties can vary considerably for different deposi-
tion techniques. The chemical structure, crystallinity, and thickness have a significant
influence on the final film properties. Consequently, the physical, mechanical, elec-
trical, optical and thermal properties of the films are not solely determined by the
intrinsic properties of the bulk diamond. In some cases this may be advantageous
since film properties can be tailored to the specific requirements of certain applica-

tions.

2.4.2 V Comparison of Semiconductor Properties

Diamond is a very wide bandgap semiconductor (Eg=5.45 eV). Table 2.5 compares
diamond semiconductor properties to Si, GaAs, and SiC. Diamond’s thermal conduc-
tivity is nearly 20 times that of silicon. Also, diamond’s saturated carrier velocity,
the velocity at which electrons move in a high electric field, is nearly 3 times that of
Si and GaAs. The saturated carrier velocity of diamond is also greater than the peak
velocity of GaAs. Unlike GaAs, the saturated carrier velocity of diamond maintains
its high rate even in a large electric field. The Johnson’s figure of merit, of which

saturated carrier velocity and peak electric field at breakdown are factors, predicts

31

 

 

 

 

 

 

 

 

 

Property Diamond i-C Graphite
Amorphous with srnail
Crystal stnicture “3“ crystal regions mixed ”if?
( 5" l with spzand sp3 bonds
Densiy (gm/cm") 3.51 1.8-2.0 2.26
Inert. inorganic acids Inert. inorganic acids
stability and soivents and solvents Inert. inorganic acids
(Walks/M2) 7000-10000 900-3000
Thermaiczormctivlty 2° Kl/30-40.K.L1-2
(chm . K)
Optical properties:
Reiractivelndex 2.42 1.8-2.2 2.15(//C),1.8(.LC)
Transparency UV-VlS-iR VIS-IR Opaque
Electrical properties:
Resistivity(0hm.cm) 10" ‘lO"’-1013 0.04(/lc),0.2(.i.c)
Dielectncconstant 5.7 4-9 2.6(l/C),3.28(.LC)

 

 

 

 

Table 2.3. Properties of diamond, i-C, and graphite.[21]

 

32

-1. Diamond Film

' semi-transparent to transparent.

' hard.

' resistivity : 10" to 10’sohm-cm.
'chemically unreactive.
crystalline to X-ray diffraction.
shows Raman peak at 1332 cm"

2. Diamond-Like Film
' 0pague to semi-transparent.
' hard.
‘ insulating.
'chemically unreactive.

'non-crystalline to X-ray diffraction.
shows only a single broad Raman peak around 1550 cm1

3. Carbon

' black.
' hard.
'electrically conducting.
non-crystalline to X-ray diffraction.
' Raman shows broad peaks centered at 1350- 1360 cm"

and1580-1610cm‘.

4. Graphite
‘ black.
soft.

electrically conducting.
- ' X-ray diffraction showing 0 value of about 3. 3 to 3. 4 Angstroms.

" Raman shows well defined peak at 1590 cm I .may or may not
show well developed peak at 1355 cm" depending on crystal
size and ordering.

Table 2.4. Working definition for different carbon coatings.[69]

33

the suitability of a semiconductor for high power applications [70, 71]. The Keyes’
figure of merit, of which saturated carrier velocity, dielectric constant and thermal
conductivity are factors, indicates the thermal limitation of a material on its high
frequency electrical performance [70, 71]. From Table 2.5 the combination of high
Johnson’s figure of merit and high Keyes’ figure of merit make it an excellent poten-
tial material for high power microwave and millimeter wave devices. It is expected to
outperform GaAs and flSiC in microwave device application based on the available
data [70, 71]. Moreover, due to the large bandgap of diamond, it can also be used for
UV detectors. The combination of high bandgap and high resistivity would reduce
the contribution to the photocurrent of thermal and background radiation, thereby
make it possible to fabricate low noise UV detectors.

Diamond is also of interest in electronic systems for uses other than as an active
electronic material. For example, diamond heat sinks can be fabricated for high
power, high temperature, electronic and photonic devices. Diamond windows may be
used to permit viewing of an object over a wide optical range from infrared to UV
light.

In Figure 2.5 diamond’s physical properties are related to various potential en-
gineering applications. But can diamond films really be used to fabricate diamond
devices? Yes, however, the fabrication techniques for diamond devices are not quite
mature yet. Some of the preliminary diamond devices results will be reviewed in the

next section.

2.5 Review of Diamond Diodes and Transistors

In this section selective examples of diamond device work are briefly reviewed.
In 1982, J. F. Prins demonstrated a bipolar transistor action in ion implanted

diamond. P-type semiconducting natural diamond was used as the substrate. Then

34

 

PROPERTY

 

 

 

 

 

 

 

 

 

 

 

 

 

DIAMOND B-SIC SI GaAs
Bandgap (eV) 5.45 2.3 1.11 1.43
Saturated Electron Veloc'
7 “y 2.7 2.5 1.0 2.0
( 10 cm I 5) peak
velocrty
Carrier Mobility ( cm 2/ V- s )
elactron - 2200 1000 1350 8500
Dielectric Constant 5.7 9.7 11,9 12,5
Dielectric Strength ( MV I crn ) 10 4-0 0.5 0.6
Thermal Conductivity ( W I cm . C) 20 5.0 1.45 05
Johnson’ s Fig. Merit
( ratio to silicon ) 8206.0 1137.8 1.0 6.9
Keyes' Fig. Merit
( ratio to silicon ) 32-2 5-3 1-0 0.5

 

Table 2.5. Comparison of semiconductor properties for diamond, Si, GaAs, flSic.[70]

 

35

 

 

 

 

 

 

 

 

 

 

 

 

 

Properties Applications
I Hard I [ Abrasive coagrgs for cuttirflols
[ Low friction J [ Tribological applications
] Electric insulator ] [ Heat sinks for electronic devices

 

 

 

[ High thermal conductivity ]

 

 

 

[ Large band gap ] I Microwave power devices

 

 

 

I Low dielectric constant I I RF electronic devices

 

 

 

I I Higg hole mobility I I High speed electronic devices j

 

 

 

] Acid/base resistive |
' - \. Electronic devices for

l Radiation resistive ]———-—~ severe environments such as

 

 

 

 

 

 

 

in space or in nuclear reactor
] Transparent ]\
I I

Large refractive index [ EIectro-optical devices

 

Figure 2.5. Relationship between diamond properties and engineering applica-

tions. [70]

36

n-type layers were induced by implantation of carbon ions. However, very low current
gain was obtained by this configuration [72].

In 1987, M. W. Geis and coworkers fabricated high-temperature point-contact
transistors and Schottky diodes formed on synthetic boron-doped diamond. The
boron-doped single crystal diamond was produced by the high-temperature high-
pressure process. The transistors exhibited power gain at 510 °C’ and the Schottky
diodes were operational even at 700 °C [73].

Since 1986 several groups developed methods to synthesize boron-doped diamond
films by the microwave plasma CVD technique or the hot-filament CVD technique.
A mixture gas of BgHe/ C H, [74, 75, 76, 77] or a saturation solution of 3203 powder 1
in 0 H3011 mixed with acetone [78] are the most often used methods. Also, placing
boron powder [79] or boric acid [51] near the substrate during the diamond synthesis
is also used to get boron-doped films.

In 1989, H. Shiomi et al. fabricated metal-semiconductor field-effect transistors
(MESFET) using boron-doped diamond epitaxial films. This is the first report of a
planar type transistor using a diamond film. The single crystal diamond substrate
was prepared by the high-temperature and high-pressure process. A high-quality
boron-doped epitaxial diamond film was then obtained on the single crystal diamond
by a plasma assisted chemical vapor deposition method. Ti was evaporated for the
source and drain ohmic contacts by the electron beam method. Al was thermally
evaporated for the gate Schottky contact. Basic transistor operation was observed
[81].

In 1991, K. Okano et al. made a diamond p—n junction diode by the hot-filament
CVD method. Since silicon was used as the substrate, the synthetic diamond films
were polycrystalline. P2 05 and 8203 were used for the doping source. Rectifyihg

characteristics were observed [82].

37

G. Sh. Gildenblat and coworkers fabricated a high-temperature thin-film diamond
field-effect transistor by using a selective growth method. A natural single crystal
diamond was used as the substrate. Selective growth of boron-doped homo-epitaxial
diamond films was achieved using sputtered S 202 as a masking layer. The device was
operational at 300 °C. [83]

W. Tsai, et al. developed a diamond MESF ET using ultra-shallow rapid thermal
processing (RTP) boron doping. A novel doping technique was used to drive in and
activate boron in natural type Ila diamond. An ultra-shallow p-doped channel of less
than 500 A was created by RTP solid-state diffusion using cubic boron nitride as the
dopant source. A device pinch-off phenomenon was observed for the first time [84].

M. W. Geis and coworkers also found that the metal-SiOg-diamond structure
on (100)-oriented substrate is more acceptable for a depletion-mode metal-oxide-
semiconductor field-effect transistor (MOSF ET). Type IIb natural diamond was used
as the substrate in this work [85].

It is noted that commercial fabrication of diamond electronic devices is not a
reality, however, many researchers are devoted to optimize the deposition processes
and develop techniques for the fabrication of diamond devices. For example, the
technique of selective growth of diamond has been much improved in recent years
[86, 87]. N-type doping has been performed by incorporation with lithium [88] and
by using P305 as a doping source in the process [82]. In terms of achieving hetero-
epitaxial films, large—area mosaic diamond films approaching single crystal quality
has been obtained on the silicon substrate [89]. Consequently, significant progress is
being made on overcoming technological problems associated with the fabrication of

diamond devices.

CHAPTER 3

Film Deposition and Sample

Preparation

3.1 Introduction

In this chapter, the detailed configuration of the MPDR deposition system will be
explained in chapter 3 (section 2). Then the description of the nucleation methods
and the film deposition process will be presented in chapter 3, section 3 and section 4
respectively. Finally, the fabrication processes of the samples for physical characteri-
zation and electrical studies, such as four-point probe samples, metal-diamond-silicon

samples, and back-etched diamond samples, are described in chapter 3 (section 5).

3.2 The MPDR System

3.2.1 The Microwave Cavity

A tunable microwave plasma disk reactor (MPDR) system [2, 3] is used for microwave
plasma assisted CVD (MPACVD) diamond synthesis. The cross section of the MPDR

and quartz processing chamber are shown in Figure 3.1.

38

39

LJ Optical Pyrometer

...... Z Z

 

 

 

 

 

 

 

 

 

 

 

 
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

% Z Sliding Short
/
5 5 Microwave
Quartz . .. 1
Chamber\ W
*2:
O-rlng
\ Basepiate
Gaslnlet ‘ / Water
\ /Cooling
‘ Perforated *
Stainless Steel Sheet .Stalnless Steel
‘ ’ ’ ' Nacuum
l l l l * [Chamber
J to Pump ' * /

 

 

 

 

 

Figure 3.1. The cross section of the MPDR and processing chamber, and a schematic
display of the electric (E) and magnetic (H) field patterns of the TMou mode.

40

The body of the cavity is a 17.8 cm (7 inch) inner diameter double-wall brass
cylinder, with water cooling between the walls. The upper end of the cylinder is
terminated by a movable sliding short. The baseplate and quartz chamber are in the
lower end of this structure. Microwaves are introduced into the system via a short
antenna, which is a coaxial microwave input probe. The length of the probe inside
the cavity, Lp, is adjustable and serves as one of the two tuning parameters. The
other tuning parameter, the length of the cavity (L,), may be varied by adjusting
the sliding short. Using these two degrees of tuning , it is possible to match the
impedance of the cavity / plasma system to the microwave power source. A perforated
stainless steel sheet is affixed to the baseplate in the bottom part of the processing
chamber and serves to terminate the cavity and still allow gas to flow through.

The baseplate consists of an opening for the discharge chamber, gas inlets, and
water cooling. The plasma is centrally contained under a quartz disk which is vacuum
sealed to the baseplate by a viton O-ring. The height of the discharge chamber
measured from the bottom of the baseplate to the top surface of the quartz disk is
9.5 cm. It is 3.5 cm from the perforated stainless steel sheet to the stage where the
quartz disk sits on, and 2 cm from the stage to the top surface of the cavity bottom
respectively. The height of the quartz disk is 6 cm and its inner diameter is 9.25 cm.
Thus, the top of the quartz disk is 4 cm above the cavity bottom surface.

There are 16 small holes which serve as the gas inlets located in the inner surface
of the baseplate. Pre-mixed gas from multiple gas tanks is uniformly distributed to
each of these gas inlets by a common gas channel, which is located at the inner region
of the baseplate.

Water cooling of the system is accomplished through the double wall cylinder
and through a circumferential channel inside the baseplate. The latter cooling chan-

nel protects the viton O-ring. Also, the baseplate is designed to allow low pressure

41

electron cyclotron resonance (ECR) operating conditions. For such operation, eight
rare-earth magnets are placed inside the baseplate. [6] Since the magnets are sensi-
tive to high temperature, cooling is achieved through the water channel surrounding
the magnets. However in this work, plasma pressures were too high to allow ECR

operation, so the magnets were removed.

3.2.2 The Deposition System

The input microwave system consists of a 2.45 GHz Cheung power source (model
MPS2450-1200), a three port circulator, a matched dummy load, a cross waveguide
directional coupler, power meters and waveguides as shown in Figure 3.2 [90]. The
Cheung microwave power source can produce stable power from 200 to 1200 watts.
The output of the microwave source is attached to a three port circulator, which
protects the magnetron from reflected power by directing radiation toward the mi-
crowave source into a 50 fl matched dummy load. Incident and reflected power are
sampled by a cross waveguide direction coupler and are further diminished by 30 db
attenuators before being measured by the. power meters. Microwave power is then
coupled from a waveguide to the cavity input probe.

The HP 435A microwave power meters are used to measure incident power P,-
and reflected power P, at the cavity. The microwave power coupled into the plasma
loaded applicator, which is the microwave absorbed power, is given by P, = P,- - P,.
By tuning the wall probe and sliding short, Pr is generally less than or equal to 10
% of P5. Due to losses in the input system, waveguide and cavity, about 15 % of the
power is dissipated as heat. Consequently, the microwave absorbed power is about
75 % of the the power supplied from the microwave source.

Figure 3.3 gives a schematic diagram of the MPDR CVD system. Up to four
different gases may be mixed by a 4-channel (model MKS 1159A) mass flow controller.

42

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

/ I Incident
Power
2.45 GHz Cavity
Microwave Applicator
Power Source . 30 dB Directional
Circulator Coupler ’
—(o> P'
P’ Waveguide
Dummy Load

Reflected
Power

 

Figure 3.2. The microwave power source. circuit, and cavity applicator.

The fit
respect

aslnglt

A la
in some
not use
it perio:
stainless
I'aruum
It’ll] a
mun,
[01 Ton
region [1
the rang,
CODdlIlox
Same 0P!
pump {Ir

The 5
Pump is
355%. In
bfl'li’epn l

339- nit

43

The flow controllers have maximum operating flows of 10,000, 500, 100, and 10 sccm
respectively. The gas output of each mass flow controller is mixed simultaneously into

a single gas channel which feeds the quartz processing chamber through the baseplate.

A large stainless steel vacuum chamber is below the baseplate and quartz chamber.
In some cases, this could be used as a downstream processing chamber, although it was
not used as such in this research. [6] Pressure measurement within the vacuum system
is performed through four vacuum gauges. There are three gauges connected to the
stainless steel vacuum chamber; a baratron pressure gauge, a thermal conductivity
vacuum gauge, and an ion pressure gauge. The baratron pressure gauge (model MKS
122A) can determine pressure from 0.1 Torr to 1000 Torr. The thermal conductivity
vacuum gauge (TC2) (model MKS 286) can accurately measure pressure from 1 mTorr
to 1 Torr, and the ion gauge (model MKS 290) is able to measure in the high vacuum
[region (10"3 to 10‘9 Torr). These three different pressure gauges can therefore cover
the range from very good vacuum to above an atmosphere depending on the operating
conditions. In addition, a second thermal conductivity vacuum gauge (TCl) with the
same operating pressure range is located between the roughing valve and roughing
pump (mechanical pump).

The system is pumped down by a Alcatel 2033 roughing pump. The mechanical
pump is filled with Alcatel 200 oil which is safe for pumping hydrogen and methane
gases. In order to minimize the backstream oil from the pump, a baffle trap is placed
between the pump and the chamber. Because the hydrogen and methane are explosive

gases, nitrogen gas is also used to purge the gas exhaust for safety reasons.

 

 
  
   

Flow Controller
(“:8KD "DA;

 

 

 

Valve

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

fl-IIL‘ Melon mm mm
IonPresaureGaupe run-nun tau
(mutual
.l,
C m N,
WV“ 8.
T02 TC1
(FWM) (Oriana-r)
Thannd Doom Thermal 6mm
Vaaum Vaaium
Guac- Gm
(us-name-) (hi-Hurts-)
em

 

 

 

 

 

 

 

 

Figure 3.3. The diagram of the MPDR CVD system.

 

 

3.3 Ti

A wide variei
appwacbes‘
mfhesis. I]
mond p0WC
in detail as l

The dial
1. 0.25 W-
IEXMEI P
lletadi luhnI
diamond pas
rubbing hack
2. Step 1 pro
ishing. :1 him
substrate suri
3. Next the s
hnsdimmedl
water for ahof
l. The subst:

IHIOI I l.\'H

nation left ore
Iiii 1',

°- The substr.

”202: IHCI

Hints. Again, i‘

826“ and 5 fro

45

3.3 The Nucleation Methods

A wide variety of nucleation methods are described in chapter 2. In this research, two
approaches were used to do the sample polishing and cleaning before the diamond
synthesis. They are l. the diamond paste nucleation method, and 2. the dia-
mond powder nucleation method. The two nucleation methods will be described
in detail as follows.

The diamond paste nucleation method :
1. 0.25 pm Buehler Metadi synthetic diamond paste was placed on a Buehler
TEXMET polishing cloth sheet, which is water and oil resistant. Then Buehler
Metadi lubricant oil was sprayed onto the paste. The substrate was placed onto the
diamond paste position and was gently polished by hand for about 20 min by using
rubbing back and forth motions.
2. Step 1 produces a substrate surface which has a dirty appearance. After the pol-
ishing, a kimwipe with some acetone was used to get rid of the dirty stain on the
substrate surface as much as possible.
3. Next the substrate was placed in a boiling TCE solution for 3 min. Then it was
rinsed immediately in running acetone, running methanol, and running de-ionized(DI)
water for about 60 sec each. Finally it was dried with nitrogen gas (N2).
4. The substrate was then immersed in a freshly prepared degrease etch (5H20 :
1H202 : 1NH4OH) for 10 min at about 70 °C’ to remove residual organic contami-
nation left over from the solvent cleaning. Then it was rinsed in DI water and dried
with N2.
5. The substrate was next immersed in a freshly prepared de-metal etch (8H20 :
2H202 : 1H CI) for 10 min at about 70 °C' to remove ionic and metallic contami-
nants. Again, it was rinsed in DI water and dried with N2.

Step 4 and 5 from the RCA cleaning procedure are widely used in the silicon industry.

«I;

(l)

U-JJ

101

4-

Dir.

46

[91]

6. After the RCA clean, the substrate was placed in a 9:1 H20:H F diluted solution
in order to get rid of the thin oxide layer. This was again followed by a rinse in DI
water and a dry with N2.

The substrate at this point was ready for the diamond film deposition. If the
substrate was a silicon wafer polished to a mirror finish, then after step 6 one cannot
see scratches easily on the shining surface by eye unless it is observed under a strong
light. No diamond particle residue was found either by the optical microscope or by
the scanning electron microscope as shown in Figure 3.4.

The diamond powder nucleation method :

l. The substrate was cleaned by following step 3 to step 6 of the diamond paste
method.

2. Dry 1 pm natural diamond powder was spread on a 3-inch diameter sacrificial
silicon wafer which was used as a lapping surface. The substrate surface was then
prepared by gentle polishing by hand on the sacrificial wafer for about 2 min.

3. Following the polishing process, the substrate receives an ultrasonic ace-
tone/ methanol cleaning for 5 min respectively.

4. The substrate also received a diluted H F solution treatment, DI water rinse and
blow dry with nitrogen.

At this point the substrate was ready for the diamond film deposition. If the
substrate was originally mirror-polished silicon, after step 4 one can easily see a
marked difference of the surface. The surface is filled with a lot of scratches and is
now hazy compared to the previous mirror-like surface. Under the optical microscope
and the scanning electron microscope, the surface is seen to be rough and full of
scratches as compared to the diamond paste nucleation method. In Figure 3.5 one

can easily see the difference of the surface prepared by the powder nucleation method.

47

 

Figure 3.4. SEM pictures of the silicon substrate prepared by diamond paste method.

Inot

treat)

3.4

3.4.]

In the
are 50‘
a givel

by \‘irt

Single C
have be
mOde “
PFPSSuye
fined tt
Ton. b}.
the (War
Operatjn!

TM,“ m

48

What are the effects of the nucleation mechanism on the diamond film growth?
In other words, is there any difference on the film growth after different surface

treatments? The answer is yes! It will be described in the next section.

3.4 Diamond Film Deposition

3.4.1 Operation of The System

In the MPDR, a microwave (MW) plasma is generated within a resonant cavity. There
are several advantages over radio-frequency (rf) and direct current (dc) plasma. For
a given power input and plasma size, the electron density is higher in a MW plasma
by virtue of the higher frequency, so its reactivity is expected to be very high. Since
MW plasma can be electrodeless, electrode contamination can be eliminated.
Although the electromagnetic modes of a perfect cylindrical cavity are well known
[92], the introduction of a plasma into the cavity can significantly alter the field
geometry. However, the MPDR can create a microwave discharge when excited in a
single cavity electromagnetic mode. For example, in the past experimental discharges
have been sustained in the TEm {93], TE211 [3], TMou and TMou modes [94]. Each
mode was experimentally evaluated for its potential to deposit films at discharge
pressures of 30 - 90 Torr. The cavity applicator was first length (L,) and probe (LP)
tuned to a specific mode. Then a discharge was started at a gas pressure of 5 - 10
Torr, by applying microwave power which then ignited a discharge that entirely filled
the quartz processing chamber. The discharge pressure was then increased to an
operating condition of 30 - 90 Torr by manually adjusting the roughing valve while
length and probe tuning the cavity to a matched condition. It was found that the

TMou mode is superior to the others in this pressure range and set-up. [10]

49

\. '

 

15KU X18888 8813 l.8U CEU91

Figure 3.5. SEM pictures of the silicon surface prepared by diamond powder method.

50

The experimental parameter space is large, since the microwave power, CH4/H2
ratio, substrate material, shape of the substrate, position of the substrate, discharge
pressure (plasma pressure) all may contribute to the quality of the diamond depo-
sition. A systematical operation procedures should be conducted in order not to
inadvertently introduce new parameters. Consequently, a well defined procedure of
operation for ”turning on” and ”shutting off” the MPDR CVD diamond system was
developed as follows.

The procedures of turning on the system :

l. The substrate received the surface treatment either by diamond powder nucleation
method or diamond paste nucleation method before putting it into the quartz pro-
cessing chamber. The size of the substrate was always 2 cm x 2cm for silicon and
1.2 cm x 1.2 cm for silicon nitride.

2. The system stand-by condition is under vacuum. So the first step was to close the
baratron valve, open the system vent valve, and allow the nitrogen gas to flow into
the quartz chamber until the chamber pressure reaches atmosphere. Then the quartz
disk and the sliding short are removed from the system.

3. Silicon substrates were typically mounted on a 0.1 inch thick graphite holder with
a 2 cm x 2 cm recessed area. The graphite holder was placed on a quartz tube, typi-
cally 3.25 cm high, which stands on the perforated stainless steel sheet, contained in
the quartz processing chamber. The silicon nitride substrates were mounted directly
on the quartz tube without a graphite susceptor. In both cases, the substrates were
in a position along the cavity axis.

4. The quartz disk was positioned such that it smoothly fit the viton O-ring. The
system vent valve was closed and the processing chamber was evacuated by the rough-
ing pump with an ultimate pressure of 10" Torr. The baratron readout was then

adjusted to zero.

4..
5D!

PM

11:5

51

5. The sliding short was placed in the cavity, then the cavity length L. was set to
approximately 7 cm which is the position for the TMon mode. The probe position
L, was set at about 1.8 cm which corresponds to maximum absorbed power and min-
imum reflected power.
6. After the reading in both of the thermal conductivity vacuum gauges (TC 1, TC2)
was almost zero, gas was introduced into the discharge chamber. 99.999 % H2 and
99.99 % CH4 Airco gases are used in this research. The flow rate and ratio of the
gases were monitored through the 4-channel gas flow controller. The nitrogen purge
and water cooling were on from this point throughout the whole process .
7. The roughing valve was ”throttled” to increase the chamber pressure up to 5 to 10
Torr as measured by the baratron gauge, then MW power was applied. At ignition
the discharge completely filled the quartz chamber, but as the pressure increased to
30 - 90 Torr (It takes about 5 - 15 min), the discharge contracted and separated from
the surrounding quartz walls and formed a roughly semi-spherical shape when the
TMou mode was used. During the plasma pressure increase, cavity length and probe
tuning are still necessary to minimize the reflected power. Visual inspection showed
that the discharge uniformly covered the whole substrate under this particular mode.
Typically, the microwave input power and plasma pressure are in the range of 400 -
900 watts and 50 - 80 Torr respectively.
8. Typically the system ran for 5 - 10 hours in order to get the desired thickness of
the diamond films. For the duration of the run, the roughing valve was monitored to
provide the desired plasma pressure.

Operation of the MPDR CVD system during the film growth is shown in Fig-
ure 3.6. One can see the discharge glowing inside the cavity through a transparent
metal-screened side window.

The procedures of shutting off the system :

 

 

 

 

 

Figure 3.6. The MPDR CVD system during the diamond film deposition.

53

1. The MW power was reduced from several hundred watts down to 0 immediately and
the gas flow of both hydrogen and methane gases were turned off. Also the roughing
valve was fully open to rapidly eliminate the gases from the discharge chamber. At
this point, the T01, T02, and baratron pressure gauges all read almost zero.

2. A 20 min period was allowed to let the system cool down. Then the water cooling
was turned off and the system was vented to nitrogen and the sample was removed.

Unless otherwise noted, these were the procedures for starting and ending a de-
position process.

The procedure to terminate the deposition process is crucial to the question of
formation of a surface conducting layer on top of the diamond film. For example,
if reactive gases continue to flow while the substrate is cooling, a graphite layer
may be formed. Early in the research, another approach was used to turn off the
system. First, the methane flow was turned off to prevent graphite formation on the
surface of the diamond film but the hydrogen was left on. Second, the hydrogen flow
was gradually reduced,a.nd then the microwave power was turned down immediately.
However, Raman characterization showed. a graphite peak instead of the diamond
peak on one of the samples under this procedure. The energy of the hydrogen plasma

is high enough to convert diamond into carbonaceous conducting layer. [32, 79]

3.4.2 Experimental Parameters

The experimental deposition parameters varied in this research are summarized as
follows.

Methane Concentration : The percentage of methane in the methane/ hydrogen
mixture was varied from 0.5% to 1.5%.

Microwave Power : The microwave input power was varied from 400 W to 900 W.

Plasma Pressure : The plasma pressure range investigated was from 50 to 80 Torr.

54

Gas Flow : Total gas flow was varied from 100 seem to 250 sccm.
Substrate Position : The quartz pedestal length was varied from 2.5 cm to 3.5 cm.
Additional deposition parameters, as described elsewhere in this chapter, are the
choice of substrates (silicon or silicon nitride) and the nucleation procedures (dia-
mond paste method or diamond powder method). It should be noted that there is a
big difference between the deposited diamond films which are prepared by diamond
powder method and diamond paste method. Large grain (greater than 1 pm) size
diamond films were obtained by the diamond paste prepared method, and small grain
(sub-micron) size films were achieved by the diamond powder prepared method. A

more detailed study of the grain size difference will be discussed in chapter 4.

3.4.3 Substrate Temperature

In the last subsection, it was noted that during the film growth the position of the
substrate, flow rate, CH4/H2 concentration, microwave power, and plasma pressure
are all experimental parameters. Each of these may contribute to the film proper-
ties either directly or indirectly in that they affect the substrate temperature. In
this research, the substrate was self-heated by the plasma and microwave power.
Consequently, changing the deposition conditions generally changed the substrate
temperature.

Temperature is a very important parameter in any CVD process, including dia-
mond film deposition. Most often, good quality diamond films are deposited with
a substrate temperature between 800 °C and 1100 °C [10-60]. Usually, there are
two ways to determine the substrate temperature. One way is to determine the sub-
strate temperature by direct contact as for example by a thermocouple through the
substrate holder, which is on the down side of the substrate. However, microwave

coupling to the thermocouple wire is a potential problem in this approach. Another

55

way is to determine the temperature without direct contact as for example with an
optical or infrared pyrometer. For a microwave applicator one means of optical access
is through the system window, which is on the top side of the substrate. Generally
the windows are shielded by a metal screen which provides microwave power leakage
protection. In this case the measurement goes through the metal screen, transparent
window quartz wall, and discharge to the substrate. Pyrometer calibration is always a
questionable matter in such a situation. The temperatures reported for good quality
diamond films by direct contact with a thermocouple (800 °C to 950 °C) are often
lower than those reported by the pyrometer methods (900 °C to 1050 °C).

In our design, the original sliding short on the MPDR can be replaced with a short
constructed with an axial opening to allow vertical optical access to the substrate
as shown in Figure 3.7. A brass collar is used for shielding the MW leaks. The
advantage of this set-up is that one can measure the temperature through only quartz
and plasma to the substrate. Without a metal screen, calibration for the measuring
device is much easier and temperature measurement accuracy improved. Generally
these vertical pyrometer measurements produced higher (100 °C - 150 °C) readings
than pyrometer readings through the metal screened window.

An Ircon Ultimax optical pyrometer (model UX-20), with working temperatures
between 600 to 3000 °C, was used for measuring the substrate temperature. This
is a small spot size instrument which has a spot size approximately equal to D/ 100
mm where D is the focusing distance. This device has a focal range from 500 mm to
infinity, but close focus lenses can be used to reduce the focal range. There are four
different close focus lens with focal ranges of 100 - 130 mm, 130 - 180 mm, 180 - 290
mm, and 250 - 540 mm respectively. The unit was generally used with a 370 mm
focus distance, which corresponds to a spot size of 3.7 mm. This makes it possible to

measure the temperature distributions across the substrates.

56

 

tical
gyprometer

 

 

 

 

  
  

Plasma

Quartz Tube ______
Substrate and

Perforated Graphite Holder
Stainless Steel

Sheet

[III

to Vacuum Chamber

Figure 3.7. Top view of the MPDR processing chamber.

Eat
thu de
09‘! is
comes
respect
tempei

for
on eat]
min frc
Howev

Insom

higher :
Shfinks

Sllppbl

57

Each material has different emissivity, which is the main calibration factor for
this device. When the silicon emissivity of 0.89 and the silicon nitride emissivity of
0.92 is multiplied by the quartz correction factor of 0.95, the corrected emissivity
comes to 0.85 and 0.88 for measuring the temperatures of silicon and silicon nitride
respectively. Then the calibration is also confirmed using silicon wafers at comparable
temperatures in thermal diffusion furnaces.

For several runs at various deposition conditions, 9 spot measurements were made
on each substrate. The corner and side measurement points were approximately 2
mm from the edge of the substrate. Figure 3.8 (a) shows a typical temperature profile.
However the uniformity is quite dependent on substrate position and cavity tuning.
In some cases much better profiles were observed. A better uniformity case is shown
in Figure 3.8 (b). Generally, all points showed temperatures within plus/ minus 5 %
of the center temperature. However, better uniformity was sometime achieved within
less than 1 %.

Figure 3.9 and Figure 3.10 show the substrate temperature dependence on power
for 1 % and 0.5 % CH4/H2 concentrations for two different plasma pressures, 50 Torr
and 60 Torr. At both cases, a 2 cm x 2 cm silicon wafer was placed on a 0.1 inch thick
graphite holder which is supported by a 3.25 cm high quartz pedestal. There is no
obvious temperature difference for the 1 % and 0.5 % concentration depositions. The
temperature increases with increasing microwave power in a nearly linear fashion.

Figure 3.11 and Figure 3.12 show the the same data plotted so as to demonstrate
the pressure dependence of substrate temperature. With higher plasma pressure,
higher substrate temperatures, were observed. At higher pressures, the plasma ball
shrinks and the energy is concentrated nearer the substrate. In all the above cases, the
microwave input powers listed are shown as measured by the Cheung microwave power

supply. The actual power delivered into the cavity is about 75 % of the microwave

(a

“fire 3
lhe Sill“

58

(a)

 

 

 

 

 

905°C 930°C 960°C
Microwave
910°C 950°C 980°C Probe
915°C 950°C 985°C
Screened Window
( b )
1070°C 1070°C 1060°C
1 075. C 1 076° C 1 080° C Microwave
Probe
1080°C 1085°C 1090°C

 

 

 

Screened Window

Figure 3.8. (a) Typical temperature profile and (b) better temperature uniformity of
the silicon substrate.

59

 

 

 

 

1080 l l l I 1 T
: 0. 0-
. . 8395: : 1.8% o—

A Substrate matenal : Si

O 4 106° ' Plasma pressure : 50 Torr ‘
L

2 1040 - d
3

:3

q.) 1020 - r
2‘

E3 1000 e -
8

66

H 980 *- 'i
4.:

a:

.0

fl

U) 960 F .

940 I L l I l L
450 500 550 600 650 700 750 800

Microwave Input Power (W)

Figure 3.9. Si substrate temperature vs. microwave input power for 1 % and 0.5 ‘70
methane concentration at 50 Torr.

AO°~ ¢L~F€.quhcfih ~tn¥rr n\.‘ at! .‘..rua.n IIHU

60

 

1120 l T j r r l l

1100 - Substrate material : Si , . O .
Plasma pressure : 60 Torr

1080 -
1060 h
1040 -
1020 -
1000 -

980 P

Substrate Temperature (°C)

960 -

 

 

 

940 L 1 l 1 l l L L J

350 400 450 500 550 600 650 700 750 800 850

 

Microwave Input Power (W)

Figure 3.10. Si substrate temperature vs. microwave input power for l % and 0.5 ‘70
methane concentration at 60 Torr.

61

powers from the Cheung microwave source.

When silicon nitride substrates were used, a 1.2 cm x 1.2 cm x 0.5 cm substrate
was supported on a 2.75 cm high quartz pedestal without a graphite holder. F ig-
ure 3.13 shows the temperature difference between silicon and silicon nitride for 0.5
% methane concentration at 50 Torr. The substrate temperature of silicon nitride
increases almost linearly with microwave power and is higher than that of silicon at
the same deposition conditions. The reason for the higher temperature with silicon
nitride is that the surface area is smaller than the silicon. The plasma will then con-
tribute more energy per unit area on the silicon nitride samples as compared to the

silicon samples.

3.5 Fabrication of Electrical Samples

3.5.1 Four-Point Probe Samples

Silicon nitride, which is an insulating composite ceramic material, was used as the
substrate for fabricating four-point probe samples. As described in chapter 3 (section
4), a 1.2 cm x 1.2 cm x 0.5 cm silicon nitride sample was placed directly on a 2.75
cm high quartz pedestal for deposition. Both diamond powder and diamond paste
nucleation methods were used to prepare the substrate. The as-received silicon nitride
substrates showed an uneven surface upon microscope examination, with multiple
trenches and ridges which were on the order of a few micrometers wide and deep.
However, the deposited diamond film followed the surface topography and successfully
covered this uneven surface.

As shown in Figure 3.14, as-deposited diamond films were ready for the four-point
probe measurements which determines the sheet resistance of the diamond films.

Results of the 4 point probe resistivity measurements will be discussed in chapter 5

1"- ‘i‘f‘llil'hk
/C°\ < .1

62

 

1120 T I I T l T T T l

1100

Substrate material : S i -
CH4/H2 concentration : 0.5%

1080 ~

1060 -

1040 -

1020 r

1000 -

980 P

Substrate Temperature (°C)

960 -

 

 

 

940 L l l l l l L J 4
350 400 450 500 550 600 650 700 750 800 850

 

Microwave Input Power (W)

Figure 3.11. Si substrate temperature vs. microwave input power for 50 and 60 Torr
at 0.5 ‘70 methane concentration.

63

 

 

11m T j T T l T T T
6 on 4“-
58 Torr ‘-
A 1080 - Substrate material : Si -
9 CH4/H2 concentration : 1%
v
q) 1060 .. “
H
3
“3 _ _
S-r 1040
Q)
Q.
a 1020 L ~
Q .
El
3 1000 - ‘*
:6
1'3
m 980 i- a
..Q
:3
U) 960 L -

 

 

 

350 400 450 500 550 600 650 700 750 800

Microwave Input Power (W)

Figure 3.12. Si substrate temperature vs. microwave input power for 50 and 60 Torr
at 1 7o methane concentration.

64

 

 

11m T T I I T T I T I
5i3N4 A-
S: 0-
A CH4/H2 concentration : 0.5%
0 Plasma pressure : 50 Torr
8.x 4
Q) 1050 r-
H
3
:6
i-
Q)
Q.
a 1000 - .
Q3
E-l
3
c6
33
m 950 i- "
.0
fi
U) .
900 l l L 1 l L 1 J l

 

 

 

250 300 350 400 450 500 550 600 650 700 750

Microwave Input Power (W)

Figure 3.13. Temperature difference between silicon and silicon nitride for 0.5 %
methane concentration at 50 Torr.

65

(section 2).

3.5.2 The Metal/ Diamond/ Silicon Samples

For the metal/diamond/silicon samples, both It and p—type silicon substrates with
(100) orientation and resistivity of 1-2 Q-m were used, with a area of 4 cm2. Some
samples were prepared with the diamond paste nucleation method and some by the
diamond powder nucleation method. After deposition, a shadow mask with 900 pm
diameter openings was used to form metal contacts on the top of the diamond films.
Typically, Al contacts were most often used in the diamond paste prepared samples
as shown in Figure 3.15. But on the diamond powder prepared samples, a shadow
mask with smaller openings (400 pm) was used. An, Ag, and In were used as the top
metal contacts in this case. The electrical properties will be described in chapter 5

(section 4).

3.5.3 Back-Etched Samples

For film characterization, it is useful to have access to both sides of the film. Towards
this purpose, a procedure suggested by Dr. J. Engemann from the University of
Wuppertal was used. Some of the preparation details were developed by H. Keller at
the University of Wuppertal. Specifically, after the diamond deposition on the silicon
substrate, a back-etched process was performed to transfer the diamond film from
the silicon substrate to an epoxy substrate.

For samples intended for back surface analysis, the coated silicon was cleaved into
approximately 5 mm x 5 mm samples which were placed diamond film down into
a Torr - SeaITM [96] mixture formed by 1 part of hardener and 2 parts of resin
on an aluminum oxide substrate. After a 70 °C' anneal for 30 min, the silicon was

removed by etching. Both NaOH (25 ‘70 by weight to water) and KOH (44 ‘70 by

66

-(a)

 

 

 

 

 

O. O .
...... Silicon Nitride
AAAAAAAAAAAA AAAAAAAAAAA AAAAAAAAAAAAAAAA o .
AAAAAAAAAAAAAAAAA AAAAAAA AAAAAAAAAAAAAAAA
KAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA.— [amon 'm

 

..__ Silicon Nitride
( S'aN 4)

 

 

 

Figure 3.14. Fabrication of the four-point probe samples. The actual surface was more
uneven than is indicated here. (a) The starting point is a silicon nitride substrate
which has received a nucleation procedure. (b) Since the substrate is insulating, the
sample is ready for the 4 point probe measurement.

67

 

 

 

 

 

 

8".
‘— IICOTI
AAAA“AKAAAnAKAAAAAAAAAKKKAAAAA“KI
AA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAA .
AAAAAAAAAAAAAA AAAAAAAA AAAAAA AAAAAAAAAA O O
AAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAN.‘
“AAA AAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAA‘ lm
AAAAAAAAAAAA‘AAA‘AAAA‘AAAAAAAAAAAAAAA
8..
IllCOTI

 

 

 

(c) . . .
W W W —A' Dot

AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

C .
RAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAA “Aflfllb‘ m m
AA AAAAAAAAAAAAAAAAAAAAAA “AAA AAAAA AAAA la on I
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

 

~—Si|icon

 

 

 

Figure 3.15. Fabrication of the metal/diamond/silicon samples. (a) The surface is
nucleated by diamond powder or diamond paste, (b) the diamond film is deposited,
and (c) top contacts are evaporated through a shadow mask. Aluminum is shown
here as an example.

68

weight to water) were successfully used with a bath temperature of 58 °C. For a
substrate thickness of 15 mils, it takes about 25 hours to etch the silicon. With
lower temperature, longer time is needed to finish the etching process. There is
sufficient adhesion between the top polycrystalline surface of the diamond film and
the epoxy that the intact film remains, attached to the hardened epoxy substrate,
thereby providing the back surface for analysis as shown in Figure 3.16. In some
cases, however, film peeling and wrinkling were observed from the epoxy, indicating
either films with higher internal strain or poor film-epoxy adhesion [95]. Some of the
diamond/ silicon interface (back-etched diamond surface) characterization results will
be discussed in chapter 4.

For electrical access to both sides of the film, the sample preparation procedure
was modified as follows. After diamond deposition on the silicon substrate, the first
(bottom) metal contact was evaporated onto the diamond surface through a shadow
mask. Then a wire lead is attached to the evaporated metal with Epo — TekTM
polymide silver conducting epoxy [97] which is cured first for 30 min at 50 °C and
then 60 min at 150 °C. The sample is placed top down in the Torr — SeaITM epoxy
and the silicon is removed by back etching. At this point, the evaporated metal
contact is visible through the transparent diamond film, and the second diamond
surface is available for electrical contacts as shown in Figure 3.17. The second (top)
metal contacts then can be evaporated onto the back-etched surface through another
shadow mask with much smaller openings. A careful study of diamond film samples

prepared by this method will be discussed in chapter 5 (section 3).

69

 

( a)
«— Diamond Film
‘\ Silicon
( P 0' N W99 )
( b )
Silicon

  
 

Diamond Film

°—Al203

(6)

T0" 393' Diamond Film

 

 

Figure 3.16. Fabrication of back-etched diamond samples for surface analysis. (a)
Silicon is coated with diamond using conventional method. (b) The sample is secured
to an epoxy substrate. (c) The silicon is removed by chemical etching.

70

(a)

.— —- Diamond Film

“~Sillcon
(PorNtype)

 
 

(b) , , First Metal Contact
.,
.— -— Diamond Film
‘- ~ Silicon

 

’ ,- First Metal Contact
_. .— — Diamond Film
' ‘— ~'Sillcon

  
  

/
Contact Wire

(d) ..... _.

Contact Vlfire ,- -’ ' T

2;, .._)- Diamond Film

§,‘

 

_______ _ First Metal Contact
A9 EPOXY / ( Bottom Metal Contact)
Torr Seal 3. , Silicon

 

Figure 3.17. Fabrication of dual-sided metal contacts on isolated diamond films. (a)
The silicon is coated with diamond and (b) a metallic contact is evaporated on the
first diamond surface. (c) A wire lead is attached to the evaporated metal and (d)
the substrate is placed face down in epoxy. (see also Engemann, et' al. [95])

71

Torr Seal

 

(f)

Second Metal Contact

 
  
  

.,....—-—...N_ /’ (Top Metal Contact)
Contact Wire ,-’ /V\_
\ W ..i _ Diamond Film
/ >4 """" "’i 5“ ‘ First Metal Contact
Ag Epoxy / 3' ' (Bottom Metal Contact)
Torr Seal
\
\
Contact Wire

 

\. - Diamond Film

Figure 3.17 (cont’d): Fabrication of dual-sided metal contacts on isolated diamond

films. (e) The silicon is removed by etching and (f) metal contacts are evaporated on
the second diamond surface.

CHAPTER 4

Physical Characterization

4.1 Introduction

Physical characterization of the deposited film is very important to determine whether
the film is in fact diamond and if it is indeed diamond, the film quality. There
are several methods which may be used for film characterization, such as Raman
spectroscopy, X-ray diffraction, reflection high-energy electron diffraction (RHEED),
scanning electron microscope (SEM), X-ray photoelectron spectroscopy (XPS), Auger
analysis, and others. In this chapter, we will use Raman spectroscopy to study the film
quality in terms of sp3 bonding, XPS to perform surface analysis, SEM to examine
the surface morphology, a Dektak profiler to evaluate the surface profile, and laser
scanning microscopy to observe the film uniformity. Detailed descriptions of these

methods are in the following sections.

72

73
4.2 Raman Spectroscopy

Raman spectroscopy is based on an inelastic light scattering process. The incident
photons, which are monochromatic, interact with the material either by creating or
annihilating one or several phonons and then emerge with energies different from
the incident photons. As in Figure 4.1, the strong line centered at hu is due to
elastic scattering of photons and is known as Rayleigh scattering. The weak lines
at 111/th originate from inelastic scattering of photons by phonons and constitutes
the Raman spectrum. The Raman bands at frequencies V — 6 are called Stokes lines,
corresponding to phonon generation, and those at frequencies V + 6 are known as
anti-Stokes lines, corresponding to phonon absorption. The intensity of the anti-
Stokes lines are usually considerably weaker than those of the Stokes lines [98] and
the Raman spectra reported in this chapter are based on the Stokes-line signals.
Figure 4.2 shows the Raman spectra for diamond and silicon on the absolute wave
number scale and on the relative wave number scale. Most traditionally Raman spec-
tra on the relative wave number scale were presented. For example, in this research,
a wavelength of 488 nm laser was used by the Raman operators. It corresponds to
the photon energy of 2.541 eV and wave number at 20492 cm‘l. Since only the
first-order Raman scattering, which involves the optical phonons with wave vector k
2 0, is considered, for diamond the energy of optical phonon involved is 0.165 eV.
Considering only the Stokes signal, the photon energy becomes 2.376 eV after the
phonon generation. Photon energy of 2.376 eV corresponds to wavelength of 522 nm
and wave number of 19160 cm“. A strong peak of diamond is then observed at 19160
cm"1 if an absolute wave number scale is used. It is 1332 cm“, which is called the
Roman shift for diamond, between 19160 cm‘1 and 20492 cm“. The reported natural
diamond peak at 1332 cm"1 is on the relative wave number scale [99]. Similarly, for

silicon the optical phonon energy is 0.063, eV. After calculation, the silicon peak will

74

Raman Spectroscopy '

(a)

 

 

 

m, Elastic Scattering
Argm h” /
...___’ ___) .
Laser , \ m, + ms >lnelastic Scattering
Sample h" - his

no : Optical phonon energy

(b) Rayleig'i Spectrum

 

 

 

. Raman Spectrun Raman Specmm
( Stokes line ) ( Natl-Stokes line )
hi! " h6 ‘ hi! bi! + 116

Figure 4.1. Raman spectroscopy is based on a process of inelastic scattering of photons
by phonons.

75

1 on the relative

be at 19984 cm‘1 on the absolute wave number scale and 508 cm”
wave number scale.

It is also known that graphite shows broad peaks at about 1360 cm‘1 , 1590 cm'1
[100] and amorphous carbon at about 1600 cm"1 [41]. Diamond-like carbon, which is
amorphous but has properties close to those of diamond, is reported to have a broad
peak at about 1500 - 1550 cm“. [101, 102, 103] The larger the Raman signal at
1332 cm"1 and the smaller the other peaks, the higher quality of the diamond films
in that the results indicate a preponderance of sp3 bonding (diamond) rather than
sp2 bonding (graphite). At the present time, Raman spectroscopy is considered to
be the best technique to distinguish between the diamond, diamond-like carbon and
graphite.

Dr. Kevin Gray in Norton Company (Northboro, Massachusetts) performed most
of the measurements for the Raman spectrum analysis. The principle set-up of the
instrumentation is illustrated in Figure 4.3 [104]. The monochromatic and polarized
light of the laser passes through an interference filter that rejects spurious lines and
background from the laser source. The light beam then enters the beam-splitter and
is focused by the lens to the samples. If micro-Raman analysis is performed, the
microscope shown in Figure 4.3 is added to reduce the beam size. Light scattered
from the sample through a variable beam-splitter then is focused by the lens onto the
entrance slit of the 1/8 meter double-grating monochromator. Before entering the
monochromator, a notch filter was used to reduce the laser light. The double-grating
monochromator acts as a tunable filter of extremely high contrast; its purpose is to
prevent the internally scattered intense Rayleigh light from overpowering the weak
Raman lines. Light leaving the final slit of the double-grating monochromator is
collected by a 1024 element silicon-array detector, whose output is processed with a

”photon counting” electronics. It includes a programmable detector controller and

76

(a) Laser

Diamond

Silicon

h

1 9160 1 9984 20492

Intensity

 

 

 

 

Absolute Wave Number( cm’1 )

 

 

0))
Diamond
- .é‘
in
§ Silicon
E A
508 1 332

Relative Wave Number( cm")

Figure 4.2. Raman spectra for diamond and silicon on (a) absolute wave number
scale and (b) relative wave number scale.

77

an oscilloscope, which is used for monitoring the real time signal synchronously with
the spectrometer. Finally, the data processing and data acquisition were collected
and processed with the personal computer.

In this work, typically a wavelength of 488 nm from an Argon laser and a laser
power of 400 mW was used to excite the Raman signals. In Figure 4.4 Raman
spectra of 0.5 % C H; / H2 powder-polished samples are compared at different substrate
deposition temperatures. All spectra show the characterization peak at 1332 cm“.
At 936 °C’ the 1332 cm‘1 peak is quite small and a broad shoulder around 1500
cm‘1 indicates amorphous diamond—like component in the film. Around 1040 °C' a
better defined diamond peak shows up. Then at 1090 °C a graphite peak at 1600
cm”1 becomes obvious. In all cases the peak at 508 cm“1 is the silicon Raman signal.
For depositing good quality diamond films, the depositing temperature must be in a
certain range. If the temperature is too low, an amorphous diamond-like component
is dominant in the film. If the temperature is too high, graphitic component will play
a major role in the film. It is observed that the preferential temperature range for
good quality diamond films is between 1030 °C and 1060 °C.

The Raman spectrum is also dependent on gas flow composition. In Figure 4.5
the Raman spectra for 0.5 ‘70 and 1 % gas mixture powder-polished samples are
compared at a fixed substrate temperature of 1000 °C and 1040 °C. It shows that
the 1 % samples have a larger graphitic component than the 0.5 % samples for both
substrate temperatures. It indicates that for the same temperature, a smaller methane
concentration of the gas flow produces better quality diamond fihns than a higher
methane concentration does.

Figure 4.6 shows that at the same growth condition a 10 hours deposition sample
has a similar Raman spectrum to a 6 hours deposition sample, but the sample with

longer deposition time has a stronger diamond signal and weaker silicon signal because

78

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Optical Silicon
Notch Filter Array
( ,\ a 433 nm ) Detector
, l
1:30 -1
Microscope \\ Aperture Aperture 1/8 Meter
eyepiece l \Van'able Monochromator
AT '0" Laser Beamsplitter
( Spam Physics 2on )
Detector I ]
Controller
A .488 nm :C Beamsplitter
Tunable Laser
Line Filter

Oscilloscope

 

 

32 X Microscope
Objective Lens

 

[] Computer

 

Figure 4.3. Raman Spectroscopy set-up for the Raman spectrum analysis.(From Dr.
Gray)

79

of the thicker diamond film.

In Figure 4.4 to Figure 4.6, only powder-polished samples were analyzed. In
Figure 4.7 Raman spectra of samples with the same deposition conditions but with
different nucleation methods were examined. It showed that the Raman spectra were
similar for samples with 0.5% methane at 1040 °C by both paste—polished and powder-
polished methods. Consequently, the quality of diamond films shows no obvious
difference between different kinds of preparation methods under the same deposition
conditions. Raman spectra are based on bulk properties of the sample rather than
on surface or interface characteristics. However, the method was used to indirectly
identify the location of the graphite signal in one case.

Figure 4.8 shows a good Raman spectrum with a small graphitic peak for a sample
prepared with 1.5% methane at about 1040 °C. The temperature was not measured
from the pyrometer directly but by estimation from the temperature measurement
on the similar deposition conditions. A part of the film peeled off when the sample
cleavage was performed. Absence of the graphitic peak was observed on the Raman
spectrum of the free standing film. This suggests that some graphitic component
exist in the diamond / silicon interface during the film deposition. When the diamond
film peeled off, most of that graphitic component stayed on the silicon surface.

Generally, Raman analysis is essential for characterizing the deposited diamond
films. However, it is not appropriate to do the surface or interface analysis. Another
characterization technique, X-ray photoelectron spectroscopy, is then introduced for

the chemical analysis in the surface and interface layers.

4.3 X-ray Photoelectron Spectroscopy

X-ray Photoelectron Spectroscopy (XPS) is also known as Electron Spectroscopy for

Chemical Analysis (ESCA). This technique is able to analyze a wide of variety of

(a)

(b)

(C)

”Half"

80

M "01’“
m "-000

1110.0

 

0N.” "
4

 

5‘4,” W-VOU-w-‘r-w-w—v—v—f—v—F-r v—r—w v-s-v-rP-v-w—m
4”.” 700.00 000.00 1100.0 1300.0 1010.0

 

7“.”

 

 

not,” W-c,c--e,---f,c--e

v ' ' ' ' l
“.00 700.00 use.” 1~.0 1340.0 1010.0

 

l”... -e---. -'-' Q-s-o-w—rm-' -Q-e~ o-v -r>v-v-o-v-
«0.00 700.00 ”.00 1200.0 1500.0 1010.0
mm CI

Figure 4.4. Raman spectra of 0.5% methane samples at (a) 936 °C (b) 1040 ac and
(C) 1090 °C. (From Dr. Gray) .

81

 

 

INIWIVV

 

 

     

 

 

 

451.00 I—' v v 1 v v u
no... 700.00 0“.“ 10..0 10‘.0 1010.0

“m0”

 

V a Y ' v v j
4”.“ 700-” ”a“ 10..0 1010.0 1010.0

(c) A (d)

 

 

 

8'10“ 1 I

 

 

 

V V v V ‘ v V v ' j ' V V V
as... 700.00 000.. 1000.0 1000.0 1010.0 400.00 700.“ ”o” 10..0 10‘.0 1010.0

Figure 4.5. Raman spectra of samples with conditions : (a) 0.5% methane, 1000 °C

(b) 0.5% methane, 1040 °C (c) 1% methane, 1000 °C and (d) 1% methane, 1040 °C.
(From Dr. Gray)

82

(3)

”AN 900m
SAMPLE: HOP-047

000.00 . -

 

T v ' v v—V v—j V v j v I V T W i I V Yfi V

700.00

 

702000

000.00

'V'WT'V'VY'WIYVU'

INTENSITY

 

010.00

 

I j’ v v’ v I fffi v v V V V V ‘

430.00 700.00 000.00 1200.0 1040.0 1010.0

0))

RANAN SPECTRUM
IANPLE: NDF-PJI

001.00 var v - . j—fi
731.00
001.00

001.00

INIENBIIY

flan-A

001.00

0000

 

45:.20.--v-.-- ----f----
430.00 700.00 000.00 sumo sumo tame
Home. on

Figure 4.6. Raman spectra of samples of 0.5% methane at 1000 °C for (a) 10 hours’
deposition and (b) 6 hours’ deposition. (From Dr. Gray)

83

 

   
     

 

 

 

( ) mm
was: III-P30
”a” fi V v ‘ v v v v I v v v v ' v I v v w v
r
. «l
I 4
. F 1
0..00 .. u
i- .
D 1
D G
D 1
"‘.oo - -
D I
f- » .
N D d
b
003.00 .-
I
on p
000.00
“t‘” [#7 v V ‘ V V W v ' v v v v ' v v v v ' v v v w
430.00 700.00 000.00 1200.0 1040.0 1010.0

(13)

RANAN SPECTRUM
SANPLE: NDF-N10

 

2100.0 '

1040.0

1030.0

1220.0

INTENSITY

 

010.00

 

YVIUTYVITYT'IVYYVIVTTY

U

 

 

 

430.00 700.00 000.00 1200.0 1040.0 ‘ 1010.0

Figure 4. 7. Raman spectra of samples with 0. 5% methane at 1040 °C' by (a) powder-
polished method and (b) paste-polished method. (From Dr. Gray)

84

RAMAN SPECTM
SAME: ”-05 01300111 SIDE

 

 

 

 

 

 

 

 

 

2w.o V V V V fl V v v ‘r v v v ‘fiv v v f I v v v
P 1
t 4
r
2410.0 L. l
: 1
L 4
19‘0-0 — d
> 4
>- . \\ 4
.-
H r- ‘ .
2 1470.0 .. ..
0- _ -‘
2
H > 4
1000.0 .—
p
v- 4
r 1
533.00 f---,.---,fi--.j--.-,-..-
430.00 ' 700.00 900.00 1200.0 1540.0 1010.0
mum». CI!
mm SPEC?”
ME: FIT-05 FKE STAN)!“ FILM
2‘”.o ' V ‘ 1“ V ' V V ' W v V V ' v v v v I v v v v
0 d
’ 1
h c1
1050.0 .. ...
D- d
" 4
’ 1
1510.0 _. _
>- " :
h b
H k d
2 L .
m 1170.0 ,. _.
p—
Z " <
04 L 4
y
b
030.00 ...
" <
‘93-” f fifi Yfifi ' ' I Y ' ' ' I ' ' ‘ fifi ' V V '
430.00 700.00 9”.00 1200.0 1540.0 1010.0

Figure 4.8. Raman spectra of (a) the as-deposited film and (b) the free standing film
of the sample with 1.5% methane at about 1040 °C. (From Dr. Gray)

85

samples in terms of elemental and chemical composition on the surface layer of a
sample, typically 5 to 50 )1 in depth.

XP S involves the energy analysis of electrons ejected from a surface under energetic
bombardment by X-rays as shown in Figure 4.9. The photoelectric process occurs
when a core level electron absorbs a photon of energy greater than its binding energy.
When this occurs, the electron is ejected from the atom with kinetic energy (K..E)

expressed by the following equation:
K.E. = hV—EB (4.1)

where hu is the X-ray photon energy and E3 is the photoelectron binding energy. E3
provides both elemental information as well as chemical bonding information. Far
example it is possible to distinguish carbon bonded as diamond from carbon bonded
to silicon as SiC. Typically, KO, X-ray emission from the light metals is the photon
source used by most XPS [105].

XPS characterization of the samples was performed in a Perkin-Elmer PHI 5400
X-ray Photoelectron Spectrometer in this research. Dr. Kevin Hook of MSU’s Com-
posite and Structural Materials Center performed the XPS analysis. Samples were
attached directly to the instrument analysis stubs and placed in the system pre-
pumping chamber. System pressure during XPS analysis was approximately 10'9
mbar. The spectrometer is equipped with both a M 9 KO, standard source and a Al
K a, monochromatic source. The Al K a monochromatic source was used mostly in this
study and was operated at 600 W (15 KV, 40 mA). A continuously variable stage was
used and set to 65° with respect to the sample surface. The portion of the sample
was analyzed through an initial lens system in this instrument. For all analyses, the
lens system was set in the large area, small solid angle mode. The size of the electron

beam was 3.3 mm diameter circle. Data points were collected 'in the fixed analyzer

86

H 3‘
_.____ 2p
H /e 25}L
— 15 K

A + hr ——> A“ + e' (ESCA)

Figure 4.9. Principle of X-ray Photoelectron Spectroscopy. When a core level electron
absorbs a photon of energy (hu) greater than its binding energy (E3), the electron is
ejected from the atom with kinetic energy (KME)

87

transmission mode utilizing a position sensitive detector and a 180° hemi-spherical
analyzer. There are two types of samples analyzed by this technique : as-deposited
samples as shown in Figure 3.16 (a) and back-etched samples as shown in Figure 3.16
(c). All the samples analyzed by XPS were treated by the paste-polished method.

For the as-deposited samples, the top surface of the diamond film was analyzed.
Figure 4.10 shows the spectra of two different samples; one was left in the air for
several weeks before the analysis and for the other the analysis was performed within
15 minutes after the deposition. Oxygen peaks were observed in addition to the
carbon peak on the sample exposed to the air for a long time. Since the sample
has some pinholes, silicon dioxide might in principle be a contributor to the oxygen
peaks. However, as the analyzing angle (the angle of the beam relative to the sur-
face) decreases the 0 peak increases. This suggested that the contamination came
from the diamond film surface rather than pinholes. This does not necessarily mean
that the diamond surface has formed carbon-oxygen bonds. The oxygen signal could
be simply a result of non-outgassing H20 on the surface, since hydrogen is not de-
tectable with XPS. Alternatively, oxidation of impurities in grain boundaries might
be a contributor, although no other elemental signal were observed.

Back-etched samples were used to analyze the diamond / silicon interface. In Fig-
ure 4.11 (a) C, 0, K and Si were observed on the surface of the back-etched sample
under the elemental survey scans. The K peak came from the K 0H which was in the
silicon etching solution. Si peaks may have come from the silicon residue or S iC (as
discussed below) which was not completely etched, or silicon from the etching bath.
An additional source of Si, as well as 0, may come from contamination from the
quartz disk. Such contamination has been previously reported for microwave CVD
diamond deposition. [106] However, it is noted that the Si signal was not observed

for the top surface XPS analysis. The big 0 peak, may also be due to contamination

88

 

 

 

 

 

 

 

 

 

 

 

 

10 v r L #t fi‘ .* i 4 c t ;
<p C «.-
a v ..
q» «-
3 «- ..
P qj.
7 ' #-
4»- 4»
6 .. 1
4» +
E 5 l1 <-
i .. 4,
4 1P «1-
'4" «>-
3 .. 0 1.
.. fl
2 "" O .0-
<)- 4,.
1 4M 7 A _... * - ‘ 0
.. 1,,
0 : : : % : ¢ ¢ ¢ ; : t 4. : : . r -1 x A
1110.0 9%.0 801.0 770.0 801.0 550.0 40.0 33.0 220.0 110.0 0.0
111mm DEM. 0V ‘
1° ; 4. 4 Y . # 4 ¢ c i 4 . :
qr- C Th
0 ‘P J»-
’ 4b
8 1P {b
.. 41
7 w ..
I. <1-
5 0 »
'"' «-
§ 5 + ..
i .. ,P
. 4 1f «-
.
3 1+- ..
4»
2 v ..
.. ..
1 4- .,
'"' 4)-
0 . ¢ ‘ ¢ 4 : g i - —A

770.0 600.0 3.0
Illllli ENERGY. 0V

A I A A
V I I
1100.0 ”.0 00.0

40.0

1
330.0

220.0 110.0

 

Figure 4.10. ESCA spectra of samples exposed to air for (a) a long time and (b) no

time before the analysis. (From Dr. Hook)

89

of the torr-seal during the back-etch process. However, another contributor might
be the oxidation of the silicon residue after the back-etch process. Figure 4.11 (b)
shows more detailed information about the C peak. The most intense contribution
to the overall peak shape is from the peak centered at 284.25 eV due to the deposited
diamond surface layer. The smaller peak present at the lower binding energy side of
the diamond peak is a contribution from the S iC interface layer (282.75 eV). It has
been suggested that the growth of a thin layer of SiC (20 - 100 A) in the early stage
of deposition is crucial to the formation of diamond films. [107, 108, 109, 110] In any
case, the observation of a SiC layer is consistent with several other groups’ studies.
[111,112,113]

Typically Argon sputtering is the most used method to clean the surface impurities
from an XPS sample. In this work, 3 KeV Argon sputtering at 15 mPa gun pressure
was performed for 30 seconds. After the surface cleaning, it was observed that the
0 peak is obvious reduced from Figure 4.12 (a) which indicates that part of the
contamination was removed. Also from Figure 4.12 (b) it is shown that the SiC
shoulder has completed disappeared. Based on the sputtering rate of SiC, it is
estimated that the carbide layer was approximately 30 A thick. However, the SiC
signal was not observed in some of the back etched samples. For these samples, it is
believed that the S iC layer was etched by the back-etch solution because the samples
were not removed from the etching solution right after the silicon was completed

etched.

4.4 Dek-Tak Analysis

In chapter 3 (section 3), it was mentioned that the diamond powder preparation
approach produced much finer grained films than did the diamond paste preparation

method. Microscope examination shows the powder-polished film to be made of fine,

90

 

 

 

 

 

 

 

 

 

 

 

 

10 * . ¢ ¢— : . : '1 a 1
4!- O C d!-
9 T 0
8 1» 4)-
<1- «L
7 4|- <>
0 4»- T-
< 0 11'
h: 5 4)- 1p
z «1- <1-
4 4p 4»-
.. ‘1) N K -
3 ~0- '1‘". . <-
w/W" -. lWM St Si ..
2 4) "W , 4
4)- 4)-
1 <1- 4»
0- 4L

0 : ¢ : a 4+ ¢ ¢ 4 * t . t : : ¢ : ¢ :
llmfi 931.0 880.0 770.0 650.0 550.0 40.0 330.0 220.0 110.0 0.0
01mm rm. 0"

lo ‘ 4 *7 * i 4 ¢ ¢ i i 41 4v f
9 4% C 284.25 CV 41-
11- o-
8 ~0- 4»
4r d?
7 4L 4)-
s 4r ..
s - \ -
:0 5 " «1-
2 <1- 4»
4)- r-
4 j 282.75 eV 1
3 4" ..
1h h’zP ‘F

r 3’?

«p 4)-
1 1* W «-
.mev \ 1

0 ¢ ¢ : t : t ‘ : ¢ : : : : - ¢ : : fl :
310.0 238.0 23.0 291.0 232.0 2W3 230.0 25.0 281.0 282.0 29.0

010110 DERBY. ev

Figure 4.11. (a) Elemental survey scan of the sample before the Ar sputtering (b)
higher resolution narrow scan of (a) in the carbon region. (From Dr. Hook)

91

 

A I
V V V I U

U T

MK

 

I([)/E
00

A A
j I 7 I

...
1

 

A
j

J
Y

 

Si - Si “

 

 

J l L 1 J I
‘1

ME "I
an

 

1100.0 09.0 “.0 770.0 09.0 550.0 40.0 330.0
010110 DERBY. 0V

 

L J
' Y 1

220.0 110.0 0.0

4—f-O-+-1-+- + —1-+~+-+— +——1—+—+—+—+——+—-

A A L
1 V j

284.0 282.0 200.0

Figure 4.12. (a) Elemental survey scan of the sample after the Ar sputtering (b)
higher resolution narrow scan of (a) in the carbon region. (FromtDr. Hook)

92

sub-micron crystallites whereas the paste-polished surface produced crystallites which
were larger than one micron for similar deposition conditions. The difference is also
obvious to the eyes. Prior to deposition, a silicon substrate has a mirror finish and
reflected images are easily observed. After deposition on a paste-polished silicon
substrate, the deposited diamond films show a highly specular surface with a gray
appearance and no observable reflected images. However, with the powder prepared
surfaces, the diamond film is flat and somewhat shiny with some ability to produce
reflected images.

There are also some other interesting visual observations. First, both types of
films appear glittery under the sun or strong light, but the film-produced by the
paste-polished method did show a darker background. Secondly, because the film
prepared by the powder method is less specular, some of the scratches on the silicon
substrate surface could be seen through the film. Third, if the grain sizes are small
enough, film produced by the powder-polished method showed colorful interference
rings due to thickness variation across the sample.

For the samples produced by the paste-polished method, optical and electron
microscopy of the transferred film shows a relatively flat back surface compared to
the original top surface of the film. This observation is consistent with the cross-
sectional view of the film and underlying substrate shown in Figure 4.13. It shows
the relative smoothness of the back surface relative to the top surface.

In order to quantify film smoothness, a Sloan DekTak II surface profiler was used.
The relative smoothness of the two surfaces of the paste-polished diamond film is
shown in Figure 4.14. Shown here are surface profiles measured over a 50 pm scan
across the samples. A standard deviation of 940 A is observed about the mean for
the top surface of the diamond film as deposited on the silicon substrate. When the

silicon is removed as shown in Figure 3.16 (c) and the back surface is scanned, a

93

<—— Top Surface of
Diamond

Side View of
*—'— Diamond

4—— Interface

4— Si Substrate

 

15w x7299 @063 1% ceoee

Figure 4.13. SEM view of the paste-polished produced diamond film on a cleaved
sample showing the silicon substrate, the interface, and the side view and top surface
of the diamond film.

94

standard deviation of 130 A is observed.

For the samples produced by the powder-polished method, SEM cross-sectional
view shows relative smoothness of the top and bottom surfaces in Figure 4.15. It is
clearly seen that the difference between these two surfaces is not as big as that of the
paste-polished produced samples. The quantified smoothness of the two surfaces is
shown in Figure 4.16. Typically, a standard deviation of 630 A is observed for the
top surface and 570 A is for the bottom surface of the diamond film. As shown in
Figure 3.4 and Figure 3.5, a rougher surface on the back side of the diamond film
produced by the powder-polished method is expected.

Figure 4.17 shows the top surface profiles of two extreme samples produced by the
paste-polished and the powder-polished nucleation methods respectively. Standard
deviations of 1600 A and 420 A are observed for the large-grain-size and small-grain-

size diamond film respectively.

4.5 Scanning Electron Microscope Analysis

The scanning electron microscope (SEM) has unique capabilities for analyzing sur-
face. It is analogous to the conventional optical microscope, but a different radiation
source serves to produce the require illumination. Whereas the optical microscope
forms an image from light reflected from a sample surface, the SEM uses electrons
for image formation. The different wavelengths of these radiation source result in
dramatically different resolution levels. Electrons have a much shorter wavelength
than light photons, and shorter wavelengths are capable of generating high-resolution
information. Enhanced resolution then permits higher magnification without loss of
details.

The maximum magnification of the optical microscope is about 2000X and the

theoretical resolution limit is about 0.17 pm. In practice it is difficult to clearly dis-

95

 

A
0°
v
:ol

 

 

 

 

    

 

 

 

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- "i 5.000
'3: i f = 4.000
3% g \ 4‘“ h 3 3,939
:3 ... 9 0 f l .i if 2.000
3 U l 3 ....
g 3 l‘ A U :
2 5 0
ei'é‘é—Th—Tfihjg—ih'fig'Te 4 :
scan length (pm)
(b) .1
3 {xi
0 L U fl— :1
'3 s 1" V v ,
1: ; .
s : 5 .6...
a E ~eee
ab—‘i—rs-Tfimirsf—zh‘fi 51‘9”

scan length (pm)

Figure 4.14. The surface profile on (a) the top surface of an as-deposited paste-

polished produced diamond film and (b) the back surface of the film after transfer to
the epoxy substrate. '

96

Q4; 4-. as «it 3:25:- j:-

H'i,‘ ‘}: - ‘n 1". . “7' fl
‘L‘q‘; 3&9; 4" . '..,.:_'_ f . ’ v ‘2’ 4—-—_ Top Surface of
31"!” (7': "-‘vl‘i' .."2 “ “‘O'VA Diamond

0 .. v «-‘ ‘ fig ,4 ’\ l

\ ‘37\:fi;~‘§¢_" *' ' _- u- -v'-4,-
, van-r- . ‘-— ~_ (‘52 «rc- .-

    

4—— Side View of
Diamond

+—— Interface

+—— Si Substrate

Figure 4.15. SEM view of the powder-polished produced diamond film on a cleaved
sample showing the silicon substrate, the interface, and the side view and top surface
of the diamond film.

97

surface deflection(A)

     

scan length (pm)

 

fl)
0|
0
'9

 

 

 

 

ht—
—-’—J
q'1 P'———._
Ffl—i r
H
in
0
G

 

my} “Mu-1H0...

I.

. UL” F. 1"qu .
* W v—“v v a

- 1 .- BEE!

 

 

”bl l-—..
.5?

,3

U!

3:

'3

 

 

surface deflection(A)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

- 7" '-"

 

5 00 §3* 40 4s 00

.“0
f.
p
G
....
I 0
J

..B .

h

scan length (pm)

Figure 4.16. The surface profile on (a) the top surface of an as-deposited powder-
polished produced diamond film and (b) the back surface of the film after transfer to
the epoxy substrate. '

98

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

‘ 1.. 000
1" [V1
.2 ' J ‘ I L fi L___1 El
3 i M JU will R A “Q q [i L.. 1 .. .333
g s l l l Ill [ l l
a: : [A I a '1] | 2.000
0 I
v s l l l
c L A . P3) 889
8 i . l| 7 I I
'13 l I r 4.000
a 5 ll!
' V #53868
0 s 10 1s. 20 2... 30 are... 42773—30
scan length (pm)
(b) ; .
I :‘i a
2 $500
‘5 3'1 L1 .000
.9. El \ z
8 I . 1 $1.500
“i: l ‘L I 1 ’ .1. 000
“U i Lani ' “I |' (1'. iv" mug WV J W lfifi .s. E 3 , 59g
8 : ' ‘l f V F." ii °
. cg ' ;-3.000
a : :
m ' '3.568
: L4.000
0 JET 10 157-‘50 2. 30+..«. 46%45 J0

 

 

scan length (pm)

Figure 4.17. The extreme top surface profiles of (a) large-grain-size and (b) small-
grain-size diamond films. .

99

tinguish features smaller than 1 pm. In comparison, the wavelength of the electrons
is less than 0.5 A, and theoretically the maximum magnification of electron beam
instruments is beyond 800,000X. However, because of the instrumental parameters,
practical magnification and resolution limits are about 75,000X and 40 71 in a conven-
tional SEM [114]. Since polycrystalline diamond films often have sub-micron feature
sizes, the high resolution capabilities of SEM are crucial for determining crystal mor-
phology.

In chapter 4 (section 4), it was noted that different grain-size polycrystalline di-
amond films result from different nucleation techniques. Relatively large-grain-size
diamond films, as in Figure 4.18, are formed by the paste-polished method and rel-
atively small-grain-size films, as in Figure 4.19, are formed by the powder-polished
method. The major reason for the difference is that the powder-polished method pro-
vides a much higher density of nucleation sites than did the paste-polished method.
Diamond particle residue and the scratches associated with polishing by diamond par-
ticles provide nucleation sites which enhance the local configuration for the diamond
growth. As shown in Figure 3.4 and Figure 3.5, more scratches and particle residue
were observed by the powder-polished method than the paste-polished method. As
the nucleation sites density increases, the space of lateral growth of diamond for each
site is limited. Correspondly, the grain size becomes much smaller as shown in Fig-
ure 4.20. This phenomenon also can be proved by a careful experiment described as
follows. The surface was intentionally polished with less nucleation sites in one place
than in the other places by the powder-polished method. After the deposition, the
difference of the crystal growth is obvious under the SEM in Figure 4.21. Large grain
size diamond particles show up in the area with less nucleation sites compared to
the small grain size diamond particles in the area with a higher density of nucleation

sites. This shows that the density of nucleation sites is a major factor in determining

grain

ll
surfac
surfac
of llllt

the a
press
them
plan
and
Ples

incre

00m
Pea;
59%

‘1

mm

100

grain size. This observation is also confirmed by J. F. DeNatale and coworkers [115].

The main objectives for using SEM analysis in this research are to examine the
surface morphology and grain size and to determine the film thickness. From the
surface micrograph the average grain size of a film can be estimated by the method

of linear intercepts, which is explained by the following equation.

total path length l -

. . . 4.2
no. of intercepts magmfzcatzon ( )

 

 

< grain size >=

The results in Table 4.1 and Table 4.2 show that for a given preparation method
the average grain size has a tendency to increase as the microwave power or plasma
pressure increases. The higher the microwave power, the higher the plasma energy,
then the higher the reaction rate, so the bigger the grain size. On the other hand,
plasma density increases as the plasma pressure increases, so does the reaction rate
and grain size. In Table 4.1 the data also indicates that in the paste-polished sam-
ples the CH4/H2 concentration plays an important role since the average grain size
increases significantly as methane concentration increases. It is understandable since
more methyl radicals can form the same crystal at the same time.

It was noted that the samples prepared by the powder-polished method were
deposited at lower microwave power and plasma pressure conditions than those pre-
pared by the paste-polished method since the experiment was performed sequentially
for these two groups. The samples were prepared by the paste-polished method in
the beginning. However, it was found from the Raman spectrum that a sample at the
condition of 700 W, 70 Torr, and l % methane concentration showed only graphite
peaks instead of a diamond peak since the deposition temperature is too high. Con-
sequently the deposition conditions for later samples prepared by powder-polished

method were lower.

Y'—

Dist.
WM]

101

I .

r a j
\

y ‘E\ 1

ISKU X48

 

Figure 4.18. Typical SEM photo of the top view of the diamond film prepared by the
Paste-polished method. The triangular shapes indicate [111] crystallite faces. (MW
power: 600 W, plasma pressure: 60 Torr, methane concentration: 0.5%.)

102

‘ 1% X "

~ “.2
1 V} ‘1‘ .
-. \‘€." 1

 

Figure 4.19. Typical SEM photo of the top view of the diamond film prepared by the
powder-polished method. (MW power: 600 W, plasma pressure: 60 Torr. methane
concentration: 0.5%.)

Voids

Nucleal
Sites

g”re 4.
(bl Powd.

103

 
  

 

 

 

 

 

  
  

 

 

 

 

 

 

 

 

 

 

 

(a)
Voids g ‘ I i .
\\_ ‘ \\ Diamond
. \ ‘—‘ Columns
Nucleation \.
Sites \ SiC Layer
\ /
\- 4
?--Silicon
(b)
Voids 1 10 \
\ \_ .3: Diamond
Nugeation\\ - - Columns
'tesix \ SiC Layer
\ ~ /
V '\. - v v fw ~
~—- Silicon

 

 

Figure 4.20. Growth mechanism of diamond film prepared by (a) paste-polished and
(b) powder-polished method. The latter has a higher nucleation density and therefore

a finer grain film.

104

3.3.3

.. 21*! ‘3 \ :3“.

 

' ISKU X1888 4285

Figure 4.21. This SEM photo shows large and small grain size diamond growth on
the same substrate.

3V5

(a)

(b)

(C)

Table 4.1. Relation of average grain size vs.

105

CH4/H2 Concentration : 0.5 %

Average grain sizes (pm) are listed below.

 

Plasma Pressure

 

 

 

 

 

Microwave
"'9‘“ Power 60 Torr 70 Torr 80 Torr
600 W 0.83 ..- ..-
700 W 0.89 1.18 1.28
800 W 1.01 ..- ..-

 

 

 

 

CH4/H2 Concentration : 1.25 %

600 W, 60 Torr : 1.41 pm

CH4/H2 Concentration : 1.5 %

600 W. 70 Torr : 2.14 pm

power and plasma pressure by the paste-polished method.

 

methane concentration, microwave

106

CH4/H2 Concentration : 0.5 ‘70

A verage grain sizes ( pm ) are listed below.

 

 

 

 

 

Microwave Plasma Pressure
Input Power 50 Torr 60 Torr 80 Torr
400 W 0.47 ---
500 W --- 0.57 ---
600 W 0.55 0. 63 0.76

 

 

 

 

 

 

Table 4.2. Relation of average grain size vs. microwave power and plasma pressure
by the powder-polished method.

Mi

n1

8"
...4.

. I?

107

The relationship of the growth rate or the deposition rate (pm/hour) versus
methane concentration, microwave power and plasma pressure by the two different
preparation methods are listed in Table 4.3 and Table 4.4. Methane concentration
is a major factor for the growth rate since the growth rate increases as the methane
concentration increases in both preparation methods. The higher the methane concen-
tration, the more methyl radicals (CH3) are available for deposition, so the growth
rate increases. Likewise, the growth rates also had a tendency to increase as the
microwave power and plasma pressure increase since the reaction rate increased as
plasma energy or plasma density increased , however, the effect is smaller than the

methane concentration effect.

4.6 Film Uniformity Analysis

So far, in this chapter Raman Spectroscopy was used for film quality analysis, XPS
was performed for diamond/ silicon interface studies, DekTak profiles were used for
the film surface smoothness test, and SEM were used for the grain size and growth
rate studies. But there is another important parameter, which is the film uniformity.

When depositing the diamond film by using the plasma CVD technology described
in chapter 3 (seCtion 4), usually an approximately hemispherical shape discharge is
formed right above the substrates. "This phenomenon becomes particularly evident
when the discharge pressure is above 50 Torr. The discharge size increases as the
pressure decreases because at about 10 Torr or less the discharge tends to fill the
quartz confinement chamber. The deposition rates of the film are dependent on the
positions correspond to the density of the discharge. If at low pressures, the discharge
volume is large enough to uniformly cover the sample, the deposition rate will be
fairly uniform over the sample. At high pressures as the discharge size decreases, the

deposition rates over the sample area will show more variation. The trade-off is that

(a)

(b)

(C)

Table 4.3. Relation of deposition rate vs. methane concentration, microwave power

108

CH4/H2 Concentration : 0.5 %

Deposition rates ( um/ hour) are listed belbw.

 

Microwave

Input Power

Plasma Pressure

 

60 Torr

70 Torr

80 Torr

 

600 W

0.34-0.38

 

700 W

0.35-0.4

0.36-0.41

0.37-0.42

 

 

800 W

 

0.39-0.44

 

 

 

CH4/H2 Concentration : 1.25 %

600 W. 60 Torr : 0.54 - 0.6 rim/hour

CH4/H2 Concentration : 1.5 %

600 w, 70 Torr :

and plasma pressure by the paste-polished method.

0.92 - 1.06 urn/hour

 

109

(a) CH4/H2 Concentration : 0.5 %

Deposition rates (um/ hour) are listed below.

 

Plasma Pressure
Microwave
"1901 Power so Torr so Torr so Torr

 

 

400 W 0.28-0.32

 

500 W ..- 0.33-0.38 «-

 

600 W 0.41-0.42 0.42-0.45 0.58-0.63

 

 

 

 

 

 

(b) CH4/H2 Concentration : 1 %

Deposition rates (um/ hour) are listed below.

 

Microwave Plasma Pressure

input Power so Torr so Torr

 

 

600 W 0.48-0.52 0.560. 62

 

 

 

 

 

Table 4.4. Relation of deposition rate vs. methane concentration, microwave power
and plasma pressure by the powder-polished method.

110

deposition rate goes down as the pressure and plasma density decreases.

To determine the thickness on different parts of the film, SEM is still the most
accurate measurement method. However, a gold coating of the sample is necessary to
obtain good image resolution and sample cleavage is required. Consequently, the SEM
technique is destructive. In order to determine a technique for measuring thickness
uniformity non-destructively, a laser scanning confocal microscope (LSM) was used
in this research.

The LSM is a light microscope, but one with a difference! In the microscope’s
name the words ”laser” and ”scanning” refer to the method of illumination, while
”confocal” refers to the method of image formation. The microscope has four light
sources: a fiber optic tungsten lamp, a mercury lamp, an argon-ion laser with 488
and 514 nm lines, and a helium-neon laser with a 543 nm line. So the LSM can be
operated both as a conventional microscope (using the tungsten or the mercury lamp)
and a laser scanning instrument.

The term ”confocal” indicates that the microscope is aligned so that the illumi-
nated spot and imaged spot coincide precisely, which is not the case in conventional
light microscopes. In a conventional microscope, light reaching the observer comes
from all parts of the specimen within the field of view - both from the narrow hori-
zontal plan which is in focus and from out-of-focus regions above and below it. The
light from out-of-focus areas, which severely degrades the focal-plane image, generally
limits to just a few microns the depth to which a specimen can be examined.

In a confocal instrument the specimen is scanned point-to-point with a finely
focused beam, most effectively by a laser, and a pinhole aperture is placed directly in
front of the detector at the focal point of light coming from the in-focus part of the
specimen. The effect of these modifications is to block light from out-of-focus regions

because the focal point of such light falls either in front of or behind the pinhole

111

aperture. The result of this instrument configuration on the final image is a dramatic
increase in resolution and contrast.

The theoretical limit of resolution for a conventional light microscope is about
0.17 pm. But much smaller objects can be easily seen in the LSM. There are multiple
reasons for this, chief among them being the elimination of out—of-focus light by the
confocal pinhole. In addition, monochromatic light (the laser) eliminates chromatic
aberration; scanning a specimen with a small spot, no matter what the light source,
always improve resolution; and the photomultiplier detector is more sensitive than
the human eye. Magnification in the LSM ranges from 200x to 16,000x [116]. This
upper limit is significantly higher than a conventional optical microscope.

In our case, however, the samples prepared by paste-polished method have a
better resolution in the diamond/silicon interface than those prepared by powder-
polished method since the interface was flatter by the former method. A sample with
the biggest average grain size and thickest diamond film was used to perform the
uniformity study. Under the SEM analysis, the thickness of the diamond film on an
area of 0.2 cm X 0.4 cm is between 6.5 and 7.5 pm. By the LSM technique, a scan
in the Z (or vertical) axis of the sample can be carefully observed since diamond is
transparent to visible light. A 488 nm laser scan was used for the best resolution in
our studies. The thickness of the film was then determined by the distance between
A plane and D plane as shown in Figure 4.22(a). Eight thickness measurements were
performed on an area of 1 cm x 2 cm of the same sample in different locations as in
Figure 4.22(b). A distribution of thickness between 5.8 and 8.0 pm were obtained.
This study did show that the LSM technique indeed can perform the uniformity
studies efficiently without significant loss of accuracy and also without any sample

destruction and contamination.

 

   
    
     

112

 

 

 

 

 

 

.................................................................................. A
................................................................... B
Voids .
\\ \s\
\ \ ,\ Diamond
NUCIGati0n\ — ’ Co'umns
Sites ‘\‘. ................................................... C
V“ i\ \ / SiC Layer
.1: ...... D
— —Silicon
(b)
8.0 6.9 6.1
6.5
7.8 7.4 5.8

 

 

 

Figure 4.22. (a) Cross section view of laser scan from plane A to D. (b) Thickness

(pm) uniformity analysis on an area of 1 cm X 2 cm of a sample.

CHAPTER 5

Electrical Characterization

5.1 Introduction

In this chapter, conductivity vs. temperature results for the as-deposited diamond
films will be described in chapter 5 (section 2). Then current-voltage characteristic of
' samples with dual-side metal contacts on the isolated diamond films will be discussed
in chapter 5 (section 3). Finally, current-voltage characterization of the samples with

metal-diamond-silicon structure will be discussed in chapter 5 (section 4).

5.2 Four Point Probe Characterization

5.2.1 Experimental Method

The measurement of bulk resistivity, particularly vs. temperature, provides a fun-
damental means of electrically characterizing materials and is essential to investigate
current transport in diamond films. In order to avoid contact effects, a method of
choice for measurement of electrical resistivity is the four point probe method. How-
ever, on high resistivity thin films (such as diamond films), the measurement technique

is nontrivial because of the requirement of accurately measuring small voltages and

113

9X

Spa

Tali

114

currents. A standard instrumentation approach such as is used on silicon wafers is not
adequate. Consequently, a sensitive four point probe apparatus, with the capability
of changing the measurement temperature, was constructed as in Figure 5.1.

The sheet resistance of diamond films on insulating silicon nitride substrates
(Si3N4) was measured with a Signatone four point probe station as follows. Four
equally spaced point probes are brought into contact with the sample surface. A
Hewlett Packard 4145 B Semiconductor Parameter Analyzer is used as the ammeter
to measure the current (I), which is supplied by a dc power supply, through the two
outer probes. The supply voltage is set at 40 V in this study. A high impedance
digital voltmeter (Fluke Model 8840 A) connected to the two inner-probes is used for
the voltage (V) measurement. This arrangement largely eliminates contact resistance
effects, since the voltage measurement probes draw negligible current. The probe sta-
tion is then placed in a Thermotron Environmental Control Chamber (Model 31.2)
which allows the temperature to be varied from —75°C to 175°C.

The sheet resistance Ra has a constant of proportionality between V/ I, and is
expressed as follows

Ro=€(

V
7) (5-1)

where 5 is a correction factor, the value of which is determined by the shape of the
test sample and the ratio between the size of the sample and the probe spacing. In
this experiment, the size of the Si3N4 substrate is 0.5” x 0.5” x 0.2” and the probe
spacing is 0.0625 inches. For a square substrate and a substrate size to probe spacing
ratio of 8, the correction factor 6 is determined to be approximately 4 [117].

The resistivity (p) of the diamond film is then obtained by

p = Ra - t (5.2)

115

Ammeter
(Model : HP414SB)

@—

 

 

 

 

 

 

 

 

 

Voltmeter
(Model : Fluke such Temperature
v Controlled Chamber
V (Model : Thermotron $1.2)
..... /

 

 

 

 

 

 

 

 

A
\AAAA AAAAAAAAAAAAAAAAAAAAAAAAAAA.— Diamond Film
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA

 

—- Silicon Nitride
( SiaN 4)

 

 

 

 

Figure 5.1. The set-up of the four point probe method.

116

where t is the film thickness, and the conductivity (0) of the film is

Eli-t

5.2.2 Theory of Conductivity vs. Temperature

The conductivity of the diamond film on insulating material (Si3N4) was studied at
different temperature levels in order to understand the conduction mechanism.

As is generally the case for semiconductors, the electrical properties of diamond
are dominated by the effect of dopants and other impurities or defects. Experimental
measurements of the electrical conductivity provide information about the concentra-
tion of electrically active impurities and defects as well as their energy levels.

Thermopower measurements on the diamond films used in this research show that
the films are p-type, indicating the presence of acceptor-type impurities or defects even
though the films were not intentionally doped with acceptors. For a semiconductor
with a single acceptor level at energy Ea in the energy gap, the hole concentration,

p, is given by [118, 119]

P = NueXP - [(Er - 190/le (5-4)

where N” is the effective density of states in the valence band, E f is Fermi level, E0
is top level of the valence band and k and T denote the Boltzmann constant and

absolute temperature, respectively. From charge neutrality,

p = n + N," (5.5)

where n is the electron concentration, No is the concentration of acceptors, and Na‘

Fr

It

prr

Val

117

is the concentration of ionized acceptor. Consequently,

Na
1+ 4exp[(Ea — Ef)/kT]

 

p = n + (5.6)
The term on the left side of the equation corresponds to positive charges, and those
on the right side correspond to negative charges. However, at the temperature of
interest here the electron density in p-type material is very small, so the first term

on the right hand side can be neglected and Eq. (5.6) becomes [118]

_ Na
1" ‘ 1 + 4exp[(E., — E,)/kT]

 

From Eq.(5.4) and (5.7), one obtains

 

1 Eu — Ea

p = §exp( kTE°)X[—N,,/4 + \/N,,2/16 + Nanezp( E”

 

 

[T )1 (5.8)

The hole concentration is directly related to the measured conductivity by the ex-

pression

0’ = (“11)? (5'9)

where q and up are electron charge and carrier mobility respectively. The reported
value of hole mobility pp is 1200 cmz/V-s. [120]

In this research, N” was calculate from
N. = 2(21rmp'kT/h2)3/2 (5.10)

where m,‘ is effective hole mass and h is Planck constant. [121] The reported value

of m,‘(= 0.75 mo) of natural diamond was used [120], which yields

N, = 3.137 - 1o15 - T3/2cm'3 ' (5.11)

118

From Eq.(5.8), (5.9) and (5.11), and the measured conductivities at different temper-
atures, (Ea-E”) and Na may be used as parameters to fit the experimental data. The
comparison of the experimental results and theoretical calculation will be discussed

in chapter 5 (section 2.4).

5.2.3 Comparison with Physical Characterization

There are four samples in this four point probe study, three of which are prepared by
the diamond powder nucleation method and one which is prepared by the diamond
paste nucleation method. The three powder-polished samples (N DF-S3,S4,S6) have
similar sub~micron crystallite morphology and the paste-polished sample (N DF ~Sl)
has a larger (greater than 1 pm) grain size diamond film based on optical microscope
analysis.

The deposition conditions for the samples are described as follows. The deposition
times are all 6 hours.
NDF-Sl, CH4/H2 : 0.5 %(1.25 sccm/250 sccm), M.W input power : 600 W,
plasma pressure : 60 Torr, temperature : 1060 °C.
NDF-S3, CH4/H2 : 0.5 %(0.75 seem/150 sccm), M.W input power : 500 W,
plasma pressure : 50 Torr, temperature : 1035 °C.
NDF-S4, CH4/H2 : 0.5 %(0.75 seem/150 sccm), M.W input power : 400 W,
plasma pressure : 50 Torr, temperature : 1000 °C.
NDF-SB, CH4/H2 : 0.5 %(0.75 seem/150 sccm), M.W input power : 320 W,
plasma pressure : 50 Torr, temperature : 950 °C.
Samples N DF-S3, S4 and S6 are as-deposited diamond films, for which the deposition
was terminated by turning off the microwave power while both methane and hydrogen
gas flows were on. However, the deposition for NDF-Sl was terminated by first

turning off the methane gas flow, exposing the sample to a hydrogen plasma and

119 .

then turning off the microwave power. This sample was subsequently treated in
a solution (2H20:16H3P04:1HN03:1 acetic acid) at 80 °C for 1 hour in order to
remove the conducting layer caused by the hydrogen plasma during the process.

The corresponding Raman spectrum for each sample is shown in Figure 5.2 and
Figure 5.3. In Figure 5.2, the Raman spectra, which were performed by Dr. Mark
Holtz in Physics department of Michigan State University, of NDF-S3, S4, 36 are
compared. Note that the horizontal wavenumber scale is in absolute wavenum-
bers with photon energy increasing from left to right and phonon energy increasing
from right to left. More typically, diamond Raman spectra are plotted vs. shifted
wavenumber. In terms of shifted wavenumber, the peak that is apparent in each spec-
trum is at 1331 cm’1 plus or minus 1 cm;1 and corresponds to the diamond peak.
At all temperatures shown in this figure a strong diamond peak is noted. There are
two indicators that the diamond films approach the Raman properties of natural di-
amond as temperature increases from 950 °C’ to 1035 °C. First, the height of the
diamond peak relative to the background level increases as the temperature increases.
Secondly, the width of the peak increases as the temperature decreases. For natu-
ral diamond, the peak is very sharp, with full-width-half-maximum (FWHM) of 2.1
cm”. For diamond films, FWHM values are substantially larger, in the range of 7 -
17 cm'1 [102].

In Figure 5.3, the Raman spectrum, which was performed by Dr. Kevin Gray in
Norton Company, of N DF -Sl is shown. The spectrum shows a very sharp diamond
peak along with a small graphitic peak on the right shoulder. This spectrum is similar
to the spectrum of N DF-S3, with a FWHM equal to 9 cm“.

 

1N7 “IVY lC/SI

 

 

r
l

.— .-. .. .

: 17700 17.40 am am guano um
. um i
... _ . -..- ..- --n _ - l

 

(b) a ;

 

INTENSITY (C/Sl

 

 

 

 

Figure 5.2. Raman spectra of (a) NDF-S3, (b) NDF-S4, and (c) NDF-S6. (From Dr.
Holtz) .

INTENSITY

121

RAMAN SPECTRUM
SAMPLE: NDF-Si

1010.0

890.00

770.00

850.00

530.00

415.20

 

 

l+llllllllllllllll

   
 

 

TITTITTII'TTTTIITTI'

  

 

1 T T T I I I V I ‘ I V V V I V I V I l fii V V V 1

430.00 700.00 980.00 1280.0 1540.0 1810.0

RECIPROCAL CM

Figure 5.3. Raman spectrum of N DF-Sl. (From Dr. Gray)

122

5.2.4 Experimental Results and Match with Theory

The films in this study were not intentionally doped with impurities. However, im-
purities may come from a variety of sources including impurities in the feed gases.
Although the methane used was the highest purity readily available (99.99 ‘76), this is
not high purity by semiconductor standards. A methane purity of 99.99 % essentially
corresponds to one impurity in 10,000 methane molecules. If, as a case in point, there
is one impurity per 10,000 carbon atoms in the diamond film, this would correspond
to an impurity density of 1x1019 cm'3 which is more than sufficient to dominate the
electrical properties. In conventional semiconductors like silicon , impurity densities
in active device regions are typically on the order of 1x1015 cm‘3. Other possible
sources of impurities include residual gases in the vacuum system, air leaks, and sam-
ple handling. Only a mechanical roughing pump was used to. evacuate the system.
Furthermore, structural defects in the crystallites and grain boundaries can introduce
states in the gap which may act as traps, acceptors, or donors. Consequently, even
though the films are not intentionally doped, it is anticipated that electrical properties
of the film will be controlled by defects and impurities.

Simple hot probe measurements based on the thermopower phenomenon showed
that all of the samples were p-type, indicating that defects and/or impurities play
the role of acceptors. By combining the theory of chapter 5 (section 2.2).with exper-
imental results, information is obtained about the concentration of these states, and
their energies.

Performing the four point probe measurement, one can get the (V/I) ratio. Then
following Eq. (5.1), it was found that the sheet resistances (R0) for NDF-Sl, S3,
S4 and 86 at room temperature are 1.2 x 108, 1.3 x 106, 5.6 x 105, and 2.1 x 105
Q/D respectively. It was also noted that the sheet resistance of N DF-Sl before the

cleaning procedure was 1 x 104 Q/D. All of these resistivities are low compared to

123

natural, un—doped, diamond. However the variation of resistivities among the samples
is consistent with the Raman indication that higher deposition temperature produced
properties closer to natural diamond since the diamond peak is more similar to that
of natural diamond.

Since the thickness of the samples were estimated to be about 3 pm thick for
6 hours’ deposition, from Eq. (5.2) the resistivities (p) at room temperature were
calculated to be 3.6 x 10‘, 3.9 x 102, 1.68 x 102 and 6.3 x 101 Q-cm for NDF-Sl, S3, S4
and 36 respectively. Taking the inverse of the resistivity (p), then the conductivity
(0‘) was determined and plotted.

Figure 5.4, Figure 5.5, Figure 5.6 and Figure 5.7 show the temperature dependence
of the conductivity for N DF-Sl, S3, S4 and 86 respectively. In all cases, points show
experimental results and lines show the results of theoretical calculation from Eq.
(5.8), (5.9) and (5.11) assuming a constant mobility (pp = 1200 cmz/V-s). The
values of Ea-Ev and Na are presented in the following.

NDF—Sl, Ea-E.,=0.51 eV and Na=2.5x1012 cm'3.

NDF-S3, E.-E.=o.31 eV and 1v..=2.0x101'3 cm-3.

NDF-S4, Ea-Eu=0.24 eV and Na=4.0x10‘3 cm‘3.

NDF—SG, Ea-E.,=0.22 eV and Na=1.0x10” cm'3.

However, in reality the carrier mobility for polycrystalline diamond films is much
smaller than that in single crystal diamond. If it is assumed that p, = 12 cmz/V-s,
the values of Ea-Ev and Na after the same curve fitting procedure are presented in
the following.

NDF—Sl, Ea-E.,=0.42 eV and Na=5.5x10“ cm'3.

NDF-S3, Ea-E.,=0.23 eV and N¢,=3.5x1015 cm‘3.

NDF-S4, E.-E,,=0.15 eV and N.,=4.0x1015 cm‘3.

NDF—SG, E.-E.,=0.13 eV and N..=1.2><1016 cm‘3.

124

Generally, the model calculated data for both )1, = 1200 cmz/Vos and p, = 12
cmz/V-s fit quite well to the experimental results. However, there are some deviations
from the experimental plots. This may indicate the existence of other levels in the
forbidden gap or non-monoenergetic states, also the mobility is general not constant
with temperature, but may either decrease or increase with temperature depending
on whether impurity scattering or lattice scattering dominates.

The paste-polished sample N DF-Sl has a much higher activation energy than the
powder-polished samples N DF-S3, S4, S6. Also, the concentration of acceptors is an
order of magnitude less than for the other samples. From chapter 4 (section 4), it is
known that paste-polished samples usually have bigger grain size diamond crystals
than the powder-polished samples. Comparatively, the concentration of grain bound-
aries for paste-polished samples are smaller than those of powder-polished samples.
This is consistent with a hypothesis that acceptor type states are associated with
grain boundaries. However, the situation is evidently more complex since the chang—
ing value of Ea-E'v indicates that the nature of acceptor level changes as well as the
concentration. For comparison purposes, it is interesting to note that type IIb di-
amonds which are naturally doped with boron are reported to have an activation
energy of 0.37 eV [120].

An annealing process was also performed on all the samples to study if the temper-
ature dependence of the conductivity would change. The samples first were cleaned
by acetone, methanol, DI water and dried with N2. Then they were annealed at 500
°C for 1 hour in a furnace atmosphere of nitrogen with a 500 sccm flow rate. After
the annealing, the sheet resistance of the N DF-SG is 1.6x109 Q/D at room temper-
ature. It increased by nearly 4 orders of magnitude. The sheet resistance of the
other samples could not be measured because the current was too small. For these

samples, the sheet resistance is beyond the limit of the measuring equipment used in

Conductivity (Q’l-cm'l)

0.001

0.0001

le—05:

H
0
l

142—07
3

125

 

06 .
p
p
I

 

 

l T l 1 I l I 1
Theory — :
Experiment 0

l LILJlll

 

l L L l l l l L

 

3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8

1000/T (°K-1)

Figure 5.4. Conductivity vs. 1000/T of NDF-Sl

Conductivity (Q‘l-cm’l)

126

 

0.01 _ I I I I I I I I
: Theory — 1
Experiment 0 .

0.001 _

 

 

 

 

0.0001 1 L l l l I
3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8

1000/T (°K-1)

Figure 5.5. Conductivity vs. 1000/T of NDF-S3

Conductivity (9'1 ~cm'l )

0.1

0.01

0.001
3

127

 

 

 

I l I I T I 1 r
Theory —
Experiment 0
0 o
O
O

l l l l l l L l

 

 

3.2 3.4 3.6 3.8 4 4.2 4.4 4.6

1000/T (ox-1)

Figure 5.6. Conductivity vs. 1000/T of N DFoS4

4.8

Conductivity (9‘1 -cm‘1 )

128

 

1 I r I - l I I I I

Theory(before anneal) - - - -
Theory(after anneal) —

 

 

0.1 Experiment(before anneal) D
Experiment(after anneal) O
, ..... a ....................
0.01 D D - [3 ..........................
0.001
0.0001
1e — 05
<> 0
1 — 06
e x
18 __ 07 1 1 l 1 l 1 l 1

 

3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6
1000/T (°K-1)

Figure 5.7. Conductivity vs. 1000/T of NDF-SG

129

this investigation, that is greater than 1010 Q/Cl.

The calculated Ea-E” and N, for N DF-S6-anneal are 0.44 eV, 1.9x 1010 cm‘3 when
p, = 1200 cmz/V-s and 0.36 eV, 2.2x1012 cm‘3 when p, = 12 cm2/V-s respectively.
No matter what the value of the mobility is, the significant decrease of the acceptor
concentration indicates that part of the defects or impurities are expelled from the
samples during the annealing process. The increase of the activation energy suggests
that there are multiple impurity energy levels existing in the forbidden gap. When
one level is annealed out, the previous minor level becomes dominant.

Landstrass and Ravi presented a specific explanation for such an annealing effect;
a model of hydrogen passivation of deep donor-like traps in the inter-band states. Hy-
drogenation electrically neutralizes the traps and the resistivity is governed by shallow
acceptor levels, the effective acceptor concentration increases and the conductivity in-
creases. The annealing process on the other hand causes dehydrogenation resulting in
electrical activation of deep traps. These donor-like traps cause the effective acceptor
concentration to decrease because of compensation and the conductivity decreases.
When the hydrogenated diamond films are heat treated in a neutral ambient, the hy-
drogen can be expelled from the crystals, restoring the high resistivity. For example,
with a 780 °C, 2 hour treatment in flowing nitrogen ambient, the resistivity increases
from ~105 to ~10“ fl-cm. [122, 123] Albin et al. [124] also had a similar observation
that the conductivity of the diamond films will increase several orders of magnitude
after hydrogenation by the hydrogen plasma. However these measurements were per-
formed by the two probe method instead of the four point probe method.

Defects states due to dangling bonds and grain boundaries and inter-band levels
due to impurities play major roles in the conduction mechanism of the diamond films.
Consequently, from the four point probe experiment, there are indeed multiple inter-

band energy levels existing in the forbidden gap and some of the impurities, such as

:0

di

hc

130

hydrogen, can be expelled from the samples by the annealing process. The effect of

defects states will be discussed in chapter 5 (section 3).

5.3 Back-Etched Samples.

5.3.1 The Back-Etched Samples and I-V Measurement Set-
Up

Typical deposition parameters and some physical characterization of diamond films
deposited for back-etched samples are described as follows. A mixture of methane
and hydrogen, with flow rates of 0.75 sccm and 150 sccm respectively, was allowed
to flow into the chamber and a gas discharge was maintained with 600 W microwave
input power. The plasma pressure was 50 Torr and the substrate temperature was
approximately 1000°C. Films were deposited on a sacrificial silicon substrate that
were diamond powder polished.

Scanning electron microscopy of the top surface of the resulting films showed well
defined crystallites with an average grain size of 0.55 pm as determined by the linear
intercepts method. A surface profile as measured by a stylus profileometer showed a
standard deviation of 630 A. Raman characterization of the films, as illustrated in
Figure 5.8, showed the characteristic diamond peak at 1332 crn’l and indicates an
absence of appreciable non-diamond carbonaceous material.

After deposition, dual side contacts were made to the top and bottom of the film
as described in chapter 3 (section 5.3). The top metal contacts are circular with a
diameter of 400 pm and approximately 20 contacts are on each transferred film. This
effectively provides 20 two-terminal devices for electrical characterization, with the

bottom contact being common to all devices.

131

RAMAN SPECTRUM
SAMPLE: NDF-P47

888.00 Y . . .

 

  

 

   

 

 

I . I 4* I—* I ”I r
I n 1
795.00 ._ ,; _
r i .
~ -1
702.00 _. _
~ 1
)- _ -4
1...
H _ .I
“2’ I “
.01 509.00 ._
H .-
2
H b—
516.00 L
423.11 . . . 1 :42 . . . 1 . . . . I I
430.00 700.00 990.00 1260.0 1540.0 1810.0

RECIPROCAL CM

Figure 5.8. Raman spectrum of the diamond film as deposited on the silicon substrate

with substrate temperature at 1000 °C. (From Dr. Gray)

132

The measurement set-up is illustrated in Figure 5.9. The sample was placed on the
Signatone microprobe station. Current-voltage (I-V) characteristics were measured
at room temperature, with and without photo-excitation. For voltages up to 100 V,
an HP 4145B semiconductor parameter analyzer was used to collect the data, with
a current resolution of approximately 100 pA. For higher voltages, a Tektronix 577

curve tracer was used.

5.3.2 Contact Effects on Back-Etched Samples

In this study, Au, Ag, and In contacts were used in order to investigate the contact
effect of different metal work functions. The work functions for Au, Ag, and In are
5.2 eV, 4.42 eV, 3.97 eV respectively [125].

For low voltages, less than 35 V, the I-V characteristic of powder-polished samples
were nearly linear and symmetric for gold, silver, and indium contacts, as shown
in Figure 5.10. For a film thickness of 3.5 pm, 35 V corresponds to an electric
field of 105 V/cm. This indicates that for electric fields below 105 V/cm, the films
exhibited predominantly ohmic behavior with a conductivity that was independent
of the applied voltage for the high work function metal contacts (Au) and low work
function metal contact (In). However, for large grain films described later in chapter
5 (section 4), results show that under certain conditions, metal-diamond contacts are

non-ohmic and dominate device I-V characteristics.

5.3.3 'High Field Effect

The combination of dual-side contacts and relatively thin films facilitates measure-
ment of electrical properties at higher electric fields than have been generally reported
for previous studies of diamond films with metallic contacts [126]. In the last section,

it is shown that ohmic behavior was predominant at low voltages for different metal

133

 

WWWSWI

0850c»

 

 

 

 

 

HP 4145 B
Semiconductor Parameter Analyzer

Tungsten
.' Probe ~.
/ \

IJq-o-a“.—._s—--o-aq-n—o—-—o—o_-—-_u---o-c—s-s-o—o—-_.d

/
/ ' Top Metal Contact ,°
/ .
/
Contact / / / Torr Seal /'

l
a/ . .
L , Diamond Film /
.
._ f _ Bottom Metal Contact ,

ammo. ./
' AI Coating l.’
Si Substrate ,'

 

 

       
    

 

 

 

ll
’ V .,

   
 

 

Signatone Microprobe Station

Figure 5.9. The I-V measurement set-up of the back-etched samples.

\
<
f

1“

Fig,

and

5am]

134

 

 

 

 

 

 

 

6e - 09 T I I I I I >
412 — 09 - “
2e — 09 )- ' “
s 0
a
t:
g —2e - 09 ‘- -
S
0
-4e — 09 - ‘
-6e - 09 "
-8e — 09 ‘ 1 1 ' l '
-20 -15 -10 -5 0 5 10 15 20
Voltage (V)

Figure 5.10. I-V characteristics of a small grain size sample with gold contacts top

and bottom. It shows linear and symmetric behavior at low voltages. (Electrical
sample : P47-T1)

C0

10

11C

V0

CO:

For

135

contacts. However, at high voltages, the current increases more rapidly than linearly,
as shown in Figure 5.11.

As shown in Figure 5.12 (a) and (b), current-voltage relationship did not follow
the space charge limited current-voltage power law where in (a) I ocV3/ 2 and in (b)
I ocV’.

However, the nonlinear behavior at high voltages is well modeled by a voltage
activated conductivity. The total dark conductance of the device may be expressed

as

G = Goo + Goexp(aV) (5.12)

where the first term represents ohmic behavior corresponding to low field conduc-
tion in the diamond film and second term represents the nonlinear behavior at high
voltages, with V being the applied voltage and a representing the slope of the curve.

Figure 5.13 and Figure 5.14 show that this model fits the data over the entire
measurement range for both Au-Au contacts on the top and bottom, and In contacts
on the top and Au contact on the bottom respectively. For both cases the diamond
film thickness was 3.5 pm. At voltages higher than 250 V, a breakdown regime is
entered in which a negative resistance state is followed by an irreversible breakdown.
For voltages less than 250 V, however, no hysteresis in the I-V characteristic is ob-
served. The experimental data is repeatable and the same for increasing voltage as for
decreasing voltage. Since both contact combinations gave the same low and high field
characteristics, the work function of the contact does not appear to contribute to the
I-V characteristics of the small grain size samples. This result is consistent with some
other reported work on metal-diamond contacts [127]. The slopes a of Figure 5.13
and Figure 5.14 are 0.042 V‘1 and 0.045 V‘1 respectively. Both Au-Au and In-Au
contacts gave rise to nearly symmetric I-V characteristics and exhibited a low-field,

constant conductance (ohmic) region with a conductivity ago. For Figure 5.13 and

136

 

 

 

 

 

 

 

 

-100 —50 0 50 100

Voltage (V)

Figure 5.11. I-V characteristics of a small grain size sample with gold contacts top and
bottom. It shows nonlinear behavior at high voltages. (Electrical sample : P47-T1)

137

1.8¢ - 07 T I T

 

1.6e -07

1.4e — 07

1.2e — 07

le—07

8e-08

Current (A)

6: -08

4e-08

2e-08

 

 

 

0..

 

1.88 - 07 T r

 

1.6e - 07

1.4e - 07

1.2e- 07

1e-07

8e-08

Current (A)

lie-08‘
4e-08

2e-08

 

 

0

 

 

Voltage2 (V2)

Figure 5.12. The I — V3/2 characteristic and (b) the I — V2 characteristic of Figure
5.11. (Electrical sample : P47-T1)

fi

sh

C0

41

Hz
of
pr

3P

V‘s

II,

10
d8:

“‘3

138

Figure 5.14, 000 is equal to 7.3x10“1 (fl-cm)”1 and 5.9x 10“1 (Q~crn)"1 respectively.

For samples deposited at similar conditions with a thickness of 2 pm. Figure 5.15
shows that this model also fits the data over the entire measurement range for Ag-
Ag contacts on the top and bottom. The breakdown regime happens at voltages
higher than 150 V. The slope a of Figure 5.15 is 0.08 V“. The Ag-Ag contacts also
gave‘rise to nearly symmetric I-V characteristics and exhibited a low-field, constant
conductance (ohmic) region with a conductivity 0'00. For Figure 5.15, 000 is equal to
4.6X10“l (fl-cm)“.

The slope a is approximately 0.045 V“ for a sample with a thickness of 3.5 pm.
However, the slope a is 0.08 V“ for a sample with a thickness of 2 pm. For samples
of varying thickness, the slope of the conductivity vs. voltage is found to be inversely
proportional to the sample thickness, indicating that this increase is related to the

applied electric field. Consequently, the conductivity may be expressed as

o = 000 + ooexp(aF) = 000 + ooexp(F/Fo) (5.13)

where F is the electric field and 0 represents the slope of the log of the conductivity
vs. electric field and F0 is the inverse of the a. The value of a and F0 for Figure 5.13,
Figure 5.14 and Figure 5.15 are 1.47><10'5 (V/cm)-1, 6.8)(104 V/cm; 1.58x10“5
(V/cm)"1, 6.3x10‘ V/cm; and 1.6X10'5 (V/cm)-1, 6.25x104 V/cm respectively.

In order to investigate Whether the high voltage increase in conductance is due
to an increase in dark carrier concentration or an increase in mobility, the voltage
dependence of the photoconductivity was also investigated. A tungsten light source

was used to illuminate the devices so that the current increased by an amount AIM.

139

 

 

1e — 05 : I I T I
: Experiment 0 .
. Model — 1
r J
1e—06 :
A
O
“E
v 16-07 _
v ' i
‘53
‘63
5 le — O8 :
"D I
1::
O
0
1e — 09 :
1e — 10 1 ' ' L

 

 

 

o 50 100 150 200 250
Voltage (V)

Figure 5.13. I-V characteristic of diamond film with gold contacts top and bottom.
Experimental data is shown by data points. The solid line represents the model
results of Eq. (5.12) with Goo=2.6x10“° 9“, Go = 2.1x10“l fl“, and a = 0.042
V“. (Electrical sample: P47-T1)

140

 

1e - 05 : I I I I
E Experiment 0
’ Model -— 1

112-06E
113—07:

1e-08:

Conductance (mho)

16-09:

 

 

 

 

1e — 10 1 1 1 1
0 50 100 150 200 250

Voltage (V)

Figure 5.14. I-V characteristic of diamond film with top indium contacts and bottom
gold contact. Experimental data is shown by data points. The solid line represents
the model results of Eq. (5.12) with Goo=2.lx10“° 9“, G0 = 2.0)(10’1l fl“, and
a = 0.045 V“. (Electrical sample : P47-T3)

141

 

 

 

0.0001 I I I I I I I I
' Experiment 0
_ Model —
le — 05 ,r
A
O I-
E le — 06 g .
v : 9
Q) .
U
5 1e — 07 y
+3 I: ’I
U .
=1 i
E 1 08 °
_ - Q
0 e E ,s’
O , ‘s‘
§
3
+ .s‘
\\
n—ogg .sy
: 5).)“. ‘
16 _ 10 1 1 1 1 1 1 1 l

o 20 40 60 80 100 120 140 160 180
Voltage (V)

Figure 5.15. I-V characteristic of diamond film with silver contacts top and bottom.
Experimental data is shown by data points. The solid line represents the model
results of Eq. (5.12) with Gm=2.9x10“° 0“, Go = 4.0)(10‘12 fl“, and a = 0.08
V“. (Electrical sample: P49-T2)

142

The photoconductance is defined here as
GP}. = AIPh/V. (5.14)

The results of the photoconductance are shown in Figure 5.16 for several photon flux
intensities. In all cases, the photocurrent was less than 10 % of the dark current, and
so may be considered as a small signal contribution to the total current. For a given

light intensity, the photoconductance is
GP}. = qAnphpA/t (5.15)

where A is the area of the top contact, t is the film thickness, )1 is the carrier mobility,
and An,”I is the increase in carrier concentration due to photo-excitation. Eq. (5.15)
is written generally for any carrier type, that is Anph may refer to photo-generated
electrons, or holes, or both. If the increase in dark conductivity at high voltages
is primarily due to a field activated mobility, then the photoconductance would be
expected to show a similar increase at high voltages. In fact, however, the data in
Figure 5.16 shows that GP], is essentially constant with respect to voltage. Over
the same voltage range, the dark conductance increased by approximately an order of
magnitude. Consequently, the field activated dark conductivity seems to be primarily
due to an increase in carrier concentration.

An electric field activated conductivity has been previously reported in a wide
variety of non-single crystal insulating and semiconductor films. A linear plot of
the logarithm of the conductivity versus the applied field is evidence of Poole’s
Law and is often interpreted as being a result of Poole-Frenkel reduction of the
ionization energy associated with Coulombic potentials surrounding ionizable sites

resulting from impurities, local non-stoichiometry, or defects. [128, 129, 130] The

143

 

16—08. I I I I I f I

999.
99.9
[00° —i
N“?

1e—09

‘I
l

 

1e-10 _

 

 

 

Photoconductance (Arb. Unit)

 

 

 

1e—11 1 1 1 l 1 L 1 1 1

0 10 20 30 40 50 60 70 80 90 100
Voltage (V)

Figure 5.16. The photoconductance is nearly independent of the applied voltage.
Shown here are data for different photon fluxes corresponding to different optical
density (O.D) filter values. (Electrical sample : P47-T1)

144

Poole-Frenkel mechanism causes an increase in conductivity due to an increase in
carrier concentration, resulting from ionization of Coulombic centers by the applied
electric field. This is consistent with our experimental evidence that it is a carrier
concentration increase, not a mobility increase, that gives rise to the field activated
conductivity.

When Coulombic potentials may be considered as non-overlapping, their contri-
bution to conductivity is proportional to exp(,BF1/2/kT) where 6 is the Poole-Frenkel
constant, equal to e3/2(1r£)“/2, and F is the electric field [130]. However, when the
Coulombic center density is sufficiently high that there is appreciable overlap of the
Coulombic potentials, then the contribution of the Coulombic centers to the conduc-
tivity is proportional to exp(ozF) where a is a function of temperature and distance
between centers [130]. The conductivity exhibited by the films in this study are ex-
amples of the latter case, as indicated by Eq. (5.13) and Figure 5.13 to Figure 5.15.

A direct indication of the density of the Coulombic centers may be found from
the slope of the log of conductivity vs. voltage in Figure 5.13 to Figure 5.15. From
Hill, the expected slope would be es / 210T t where s is the separation of Coulombic
centers (or defect centers) and t is the sample thickness. [130] For the experimental
slope of0.045 V“ for t = 3.5 pm, the corresponding value of s is 4x10“ cm. Taking
s to be equal to N “l 3 where N is the Coulombic center density, the value of N
is 1.6x1019 cm“. The overlap of Coulombic potentials is a matter of degree, and
setting a maximum separation for overlap is necessarily arbitrary. Following Hill, the
maximum separation which still produces appreciable overlap may be taken, as an
approximation, as twice the distance from a site to the maximum in the barrier. The

lower limitation of the density for overlapping is (efl’1F'1/2)3 [130]. Taking

($13 = N = (efl'1F1/2)3 (5.16)

145

this approximation leads to a relationship between 3 and the electric field according to
s = fl/eF1/2. As the electric field increases, this maximum separation decreases. For
3 = 4x10‘7 cm, the corresponding electric field is calculated to be 6x105 V/cm. This
value is within the range of fields at which we observed field activated conductivity
and, given the rough approximation used as an overlap criteria, is an indication that
overlap of Coulombic potentials is a reasonable hypothesis at these densities.

Since the lattice constant of diamond is 3.56721 and the lattice configuration is
face center-cubic structure, the diamond crystal has 1.76><1023 atoms/m3. The
calculated Coulombic center density corresponds to approximately one center per
10,000 host atoms. While this value is high by single crystal semiconductor standards,
it would not necessarily be unexpected for polycrystalline films deposited by chemical
vapor deposition. States at grain boundaries may be a contributor to these centers.
Additionally, thin film deposition system purity considerations are also a factor in
terms of impurity or defects states within individual crystallites.

Because of the field activated conductivity, the current is substantially larger at
high electric field than would otherwise be the case. The negative resistance observed
prior to breakdown is evidence of thermal effects which apparently cause thermal
runaway. For 3.5 pm and 2 pm thick films, this occurs at a voltage of 250 V and 150
V respectively. It corresponds to a breakdown field of 7.2x105 and 7.5x105 V/ cm.
Consequently, the dielectric strength of the polycrystalline diamond samples in this
study are substantially less than those reported for single crystal diamond (0.6 -
1x107 V/cm) [131].

5.3.4 Photo Efl'ect

If an incident photon has sufficient energy to excite a valence electron into the con-

duction band, then, with a certain probability, that photon will be absorbed in the

146

material creating a hole-electron pair. Alternatively, a valence electron may be excited
to a higher lying defect state, creating a free hole, or a trapped electron may be excited
to the conduction band, creating a free electron. In any case, these photo-generated
carriers, being in thermal equilibrium with their surroundings, recombine after some
time, generally via trapping dynamics. During the lifetime of the photo-generated
carriers, an increase in electrical conductivity will be observed.

Photon absorption in high purity, single crystal semiconductors and insulators is
generally characterized by an absorption edge which occurs at the minimum energy
required to free a valence electron and cause band-to—band excitation. Below this
energy there is little photon absorption, above it the absorption increases sharply.
The presence of defects and impurities allow different absorption phenomena which
may be exhibited by a tail on the absorption edge which extends into the energy gap,
or by structure in the absorption spectra within the energy gap. This investigation
studied photo-conductivity due to photon energies smaller than the energy gap, and
therefore corresponding to excitation involving defects or impurities.

In this research, two sets of optical filters, which are visible light filters and infrared
filters, were used to study the photo-effect. The monopass wavelengths of the visible
filters are between 412 nm and 714 nm and those of the infrared filters are between 775
nm and 1480 nm. Consequently, the covered range corresponds to photon energies
between 3 eV and 0.84 eV. Before the measurement were conducted on the back-
etched samples, the relative power density for each filter was measured. A Newport
power meter (model 815) with Si photodetector (model 818-SL) and Ge photodetector
(model 818-IR) were used to measure the power for photon energies above 1.2 eV and
below 1.6 eV respectively. As shown in Figure 5.17 (a), a calibration factor was
applied to the power meter for different wavelengths of light for each photodetector

in order to get the accurate readings. Since the photon flux passing through each

147

filter is different, the relative power intensities corresponding to each filter are shown
in Figure 5.17 (b).

The photoconductance versus photon energy at -20 V and 20 V for the Au-Au
contact sample is shown in Figure 5.18(a) before the normalization of the relative
power intensity factors and in Figure 5.18(b) the normalized photoconductance versus
photon energy is shown. Similarly, for the In-Au (In contact on the top) contact
sample, the relationship of the photoconductance versus energy before and after the
normalization are shown in Figure 5.19(a) and (b) respectively.

For both the Au—Au contact sample and the In—Au contact sample, the relationship
of photoconductance versus photon energy is nearly the same for positive and neg-
ative bias. The existence of appreciable photoconductance well below the band gap
indicates many defect states existing in the polycrystalline diamond films. The phe-
nomenon of increasing sub-bandgap photo-conduction with increasing photon energy
is also observed by a group with an international cooperation effort using a differ-
ent approach [132]. They found that all CVD diamond films exhibited an almost
monotonically increasing absorption withiincreasing photon energy. As the absorp-
tion increases, the generation of electron-hole pairs increases and the photo current

increases.

5.4 The Metal/ Diamond/ Silicon Samples

5.4.1 The I-V Measurement Set-up

The metal / diamond / silicon samples as described in chapter 3 (section 5.2) were placed
on the Signatone microprobe station as shown in Figure 5.20. Current-voltage (I-
V) characteristics were measured at room temperature, with and without photo-

excitation. As for the back-etched samples, an HP 4145B semiconductor parameter

148

 

 

 

 

(a)
m T I I T I
Si Detector 9-
A Ge Detector 4-
'2 700 " «A
D
.5 coo , «
l-
<
v .1
h 500 '
3
.2 .... .
=
.9. -
3 300
i-
:2
‘3 200 - -
D
100 1 1 1 1 l
400 ' 600 800 1000 1200 1400 1600

Wavelength of Optical Monopass Filter (nm)

(b)

 

 

Normalization Factor (Arb. Unit)

 

 

1

L 1

0 1 L
400 600 800 1000 1200 1400 1600

 

Wavelength of Optical Monopass Filter (nm)

Figure 5.17. (a) The calibration factors corresponding to different optical wavelength
for Si and Ge photodetector. (b) The relative power intensity factors for Si and Ge
photodetector at different wavelength of the optical'monopass filters.

149

 

 

 

 

 

 

 

 

18 - 08 _ I I I I r
bin: 20V 0-
? : hilt-20V '0- I
G . .
:D
I- E 1
< > .
v 4
8 4
a la - 10 F '5
o E 1
a ’ 1
'5 P I
t: » J
g le - 11 l:- 1
a t 1
O I .
.3 r 1
D- * 1
le — 12 * E 1 ‘ L
0.5 l 1.5 2 2.5 3 3.5
Photon Energy (eV)
16 - 08 ’ T I— I T r j
’ bias: cows; Detector) e-
A bias: 20V(Ge Detector) E—
.‘u': ' bi8:-20V(5i Detector) ~0—
5 * biu:o20V(Ge Detector) ---
. lc - 09 ;-
-° I
. h ’
< .
v ’
8
5 1e - 10 :- 1
U + .
z . l
'U .
8
0 1e - 11 f 1
§ E a
..r: I 1
0.. . 4
1c - l2 ' 4 i ‘ 1 1
0.5 l 1.5 2 . 2.5 3

Photon Energy (eV)

3.5

Figure 5.18. Relation of photoconductance vs. photon energy for Au-Au contact

sample (a) before normalization and (b) after normalization. (Electrical

P47-T1)

sample :

 

150

 

 

 

 

18 -08 ’ I I I r I

A E bias: 20v 9- i
.2 ’ biuz-20V -e— <
= ' ‘
D
‘9' 1c - 09 r 1
l- , I
< I I
v > 4
v ‘
o l
g lc - 10 .- 1
U ; .
= W
'5 l j
= . .
8
0 1c — ll :- 1
a : :
O . .
.2 l
G- r 4

r 4

1c - 1?. ‘ ‘ ‘ ‘ '
0.5 1 1.5 ‘2 2.5 3 3.5

Photon Energy (eV)

 

19 — 08 I I I f I
E bin: 20V(Si Detector) 0-
' bias: 20V(Ge Detector) B-
’ biu:-20V(Si Detector) o—
’ biu:o20V(Ge Detector) Q—
l
3
l
J
4

lc-OQF

Photoconductance (Arb. Unit)

 

 

 

1e - 10 )-
lc - ll :- 1
E 1
. 1
1c - l2 1 ‘ 1 '
0.5 l 1.5 2 2.5 :3 33.5

Photon Energy (eV)

Figure 5.19. Relation of photoconductance vs. photon energy for In—Au contact
sample (a) before normalization and (b) after normalization. (Electrical sample :

P47-T3)

151

analyzer was used to collect the data with a current resolution of approximately 100

pA. In all cases, p-type silicon was used as the substrate.

5.4.2 Contact Effects on Metal/Diamond/ Silicon Samples

There are two ways of fabricating the metal/ diamond/ silicon samples as previously
mentioned in chapter 3 (section 5.2). The I-V characteristic of the large grain size
diamond samples (paste-polished samples) with Al contacts and the small grain size
diamond samples (powder-polished samples) with Au, Ag, and In contacts will be
described and compared to the results reported in the last section for back-etched
samples. Silver paste was used to form contact on the back side of the silicon wafer.
For the p-type silicon used here, the paste contact was ohmic.

For low voltages, less than 25 V, the current-voltage (I-V) characteristics were
nearly linear and symmetric for all the Au, Ag, and In contacts as shown in Figure 5.21
and Figure 5.22 on fine grain, diamond-powder prepared samples. It is strongly
believed that in these small grain size diamond films, the defect state density is so
high that the tunneling process dominates the metal/ diamond contact I-V properties.
Consequently, ohmic behavior is observed for both high work function metal contacts
(Au) and low work function metal contacts (In) of the metal/ diamond contact. This
is consistent with the metal / diamond contacts on the back-etched samples.

The data in Figure 5.21 and Figure 5.22 also indicate that there is no rectifying
behavior at the diamond/silicon interface. On the diamond side of this interface,
it is reasonable to assume that, if a barrier did exist, a high defect density again
produces such a thin barrier that tunneling allows ohmic behavior. For the silicon,
the starting wafer which has been severely abraded as shown in Figure 3.5. Con-
sequently, the defect density in the powder abraded silicon is also expected to be

high, which would result in a silicon barrier that was also sufficiently thin to tun-

152

 

WMWWI

odors

 

 

 

 

 

HP 4145 B
Semiconductor Parameter Analyzer

Tungsten

/ Probe \

 

/_._._._._._...4.-._._._._._._._._._.-._._._.-.-.-.-..]
/AlDot ',
AAAAAAAAAAAAAAAAAAAAAAAAAA / Diamond Film _/

/P-typeSi /‘
/Ag Paste /’

    
  
 

 

 

 

 

 

b.-._.‘.—._u-.—._----._.---v—.-.—.—e-e—o—o_u_.-._.o

Signatone Mlcmprobe Station

Figure 5.20. The I-V measurement set-up of the metal/ diamond/ silicon samples.

153

nel through. In fact, several decades ago, ohmic contacts to semiconductors were
sometimes made by abrading the semiconductor with sandpaper in order to create a
defect-laden surface. Alternately, the lack of a rectifying barrier may also be due to
a lack of appreciable band bending in the silicon. As will be elaborated on later, the
results of photo-excitation experiments described in the next section indicate that
thereis no significant band bending in the silicon, either for powder-polished samples
or for paste-polished samples.

In contrast to the powder polished, small grain samples, for the large grain size
paste-polished diamond samples, the I-V characteristics of the Al/diamond/silioon
structure showed rectifying behavior as shown in Figure 5.23 and Figure 5.24. This
indicates that the defect density is not as high as in the small grain size diamond films.
It is reported that polycrystalline diamond films have a much higher defect density
than the single crystal, and the concentration of defect increases as the concentration
of methane increases [132]. This research indicates that different nucleation tech-
niques and/ or different grain sizes are also factors contributing to defect states in the
films.

Rectifying phenomena of metal / diamond/ silicon structures have been previously
reported by several groups [133, 134] in recent years, and similar behavior was also ob-
served for bulk boron-doped diamond synthesized under ultrahigh pressure conditions
[135]. The rectifying property has been variously attributed to the band bending at
the top-contact / diamond interface [133] and back-contact/diamond/silicon interface
[134]. It is important to know which interface in fact contributes to the rectifying
characteristics in our case. Further investigation using additional information from

photo-excitation will be discussed in the next subsection.

 

154

 

 

00m15 F I I I I I I 1
0.0001
A 5c — 05
:5.
‘5 o
2’.
I-
:3
0

-5e - 05

-0.0001

 

 

 

 

l l 1 1 i 1 1

-0.00015 L .
--25 -20 -15 -10 -5 0 5 10 15 20 25

Voltage (V)

 

Figure 5.21. I-V characteristic of the small grain size Au/diamond/silicon samples.
(Electrical sample : P39)

155

 

 

 

 

 

 

16-05 I 1* I I I I I I 7
8e—06 - ..
'6c-06 - -
A 4c—06 P ' "
< .
V
g 2e—06 -
t-
‘5 0
O
—2c-06 - ~
-4e-06 - ‘
—6e-06.~ . ‘
-86-06 I l J L L 1 l 1
-25 -20 -15 -10 -5 0 5 10 15 20 25
Voltage (V)

Figure 5.22. I-V characteristic of the small grain size In/diamond/silicon samples.
(Electrical sample : P39)

156

 

113-09

 

 

—1e—09

-2c-09

-3e-09

Current (A)

—4e-09

—5e-09

 

-6¢ _ 09< 1 1 L 1 1 1 1 1

-25 -20 -15 —10 -5 0 5 10 15 20 25

Voltage (V)

 

 

 

 

Figure 5.23. I-V characteristic of the large grain size Al/diamond/silicon samples.
The average grain size is 2.1 pm. (Electrical sample : P6)

4c-06

157

 

2e—06 -
0

 

-2¢ — 06
—4e — 06
-6e - 06
-8e - 06
—1e - 05
-1.2e - 05
—1.4e — 05
-l.6e — 05<
—1.8e - 05

Current (A)

 

 

l

 

 

l l 1 l l

 

—25

—20

—15

-10

—5 0 5 10 15 20 '25

Voltage (V)

Figure 5.24. I-V characteristic of the large grain size Al/diamond/silicon samples.
The average grain size is 1.4 pm. (Electrical sample : P7)

158

5.4.3 Photo Efl‘ect

The experiments of the photo effects on the metal / diamond / silicon samples used the
same method and technique as that described in chapter 5 (section 3.4).

Figure 5.25 and Figure 5.26 show the relationship of photoconductance versus
photon energy for the small grain size Au/ diamond/ silicon sample and the small
grain size In/ diamond/ silicon sample respectively. Similarly, the photoconductance
versus photon energy characteristics for different large grain size Al/ diamond/ silicon
samples are shown in Figure 5.27, Figure 5.28 and Figure 5.29. If the rectification
happens at the diamond/ silicon interface and is a result of silicon band bending,
then silicon should play a role in contributing to the photo current. Consequently,
the photo current should abruptly increase as the photon energy increases above 1.1
eV, which is the energy gap of the silicon. However, in both large and small grain
size metal/ diamond / silicon samples there are no indications that photoconductance
abruptly increases at photon energies above 1.1 eV. This showed that the rectifying
properties are not contributed by the diamond/ silicon interface but rather by the

Al/ diamond interface and that band bending happens at the Al/ diamond interface.

It is also observed that a negative bias did generate more photo current than a
positive bias in Figure 5.27, Figure 5.28, and Figure 5.29. Figure 5.30 shows a possible
energy—band diagram of the Al/p-type diamond interface. During the photo-excitation
the valence electrons are excited into the higher lying acceptor level creating more
holes in the valence band. Since the band bending happens at the Al / diamond inter-

face, that a negative bias drawing more photo current from the diamond is reasonable.

Generally, this research indicates that large grain size diamond films prepared

by the paste-polished method provided better metal/ diamond rectifying properties

 

159

 

 

 

 

13 - 06 I I I T T
’ bill: 15V 0- ‘
biee:-15V -- 1
A 1
.".'.’.

G ,
3 J
A. P
i .
g

Q

3
“U , 4

G

8
3

O
...:

G...

Ic - 07 ‘ 1 f ‘ ‘
0.5 l 1.5 2 2.5 3 3.5
Photon Energy (eV)
(b)
18 - 05 P I I I I T

 

: bias: 15V(Si Detector) 0-
. bin: 15V(Ge Detector) G-
. bile:-15V(Si Detector) -e-
, biuz-15V(Ge Detector) O-

I A A A AA

 

Photoconductance (Arb. Unit)

 

 

 

1c - 06 :- 1
' 1
1c - 07 4 ‘ i ' 4
0.5 l 1.5 2 2.5 3 3.5

Photon Energy (eV)

Figure 5.25. Photoconductance versus photon energy (a) before normalization and (b)

after normalization for the small grain size Au/ diamond/ silicon sample. (Electrical
sample: P39)

160

(a)

 

18-07 I I I I I

bin: 15V 0- ’
biuzolsv 4- t

P 4

Photoconductance (Arb. Unit)

 

 

 

1‘ - 080.5 i if: 2 2:5 3 3.5
Photon Energy (eV)
(b)
1e - 06 I r .

 

bite: 15V(Si Detector) @-
. bill: 15V(Ge Detector) B— .
. biee:-15V(Si Detector) «o—
biee:-15V(Ge Detector) o— 1

LA

A

Photoconductance (Arb. Unit)

 

 

 

1c — or r 1
i 1
P J
1‘ - 08 4 ' 1 1 1
0.5 1 1.5 2 2.5 3 3.5
Photon Energy (eV)

Figure 5.26. Photoconductance versus photon energy (a) before normalization and

(b) after normalization for the small grain size In / diamond / silicon sample. (Electrical
sample : P39)

161

 

 

 

 

 

 

 

 

13 - 06 I I I I I
I bin: 15V 0— 1
c ; him-15V .- j
5 le - 07 r 1
3?; . 1
1c - 08 r
$ ‘ i
o ’ I
U D
5 1c - 09 g- 1
a : .
U D
g D 1
5 1c - 10 )v 1
o » 1
O I ,
fl
.2 1c — 11 g- 1
D... E i
r 1
r J
1e - l2 1 l 1 1 4; '
0.5 l 1.5 2 2.5 3 3.5
Photon Energy (eV)
16 - 06 r I I I I
: his: 15V(Si Detector) O- :
c : bin: 15V(Ge Detector) €- 1
.E 1 07 bi.:-15V(Si Detector) -e-
D c ' 1’ hint-15V(Ge Detector) O- 1
e i 3
1c - 08
.3 1’ ‘ I
o P
Q P
5 1c - 09 g- 1
a P :
o ; 1
e . .
1 — 10
8 ‘ E 1
Q r I
o 1 .
a
2 1c - 11 g' 1
le - 12 1 l 1 l 1
0.5 l 1.5 2 2.5 3 3.5
Photon Energy (eV)

Figure 5.27. Photoconductance versus photon energy (a) before normalization and (b)
after normalization for the large grain size Al/ diamond/ silicon sample. The average
grain size is 2.1 pm. (Electrical sample : P6)

162

(a)

 

18-06 1 I I fl I

I
.11

le—07:

Photoconductance (Arb. Unit)

 

 

 

 

 

 

 

1c - 08 :- 1
lc - 09 ' ‘ 1 1 1
0.5 l 1.5 2 2.5 3 23.5
Photon Energy (eV)
13 "' 06 I I I I I 4
bill: 15V(Si Detector) o- 1
A ) bite: 15V(Ge Detector) e- .
f: F bieez-15V(Si Detector) ce— ‘
1: . bieez-15V(Ge Detector) e- *
ED .
'e lo - 07 r 1
< :
v >
8 :
E ,
U
5
E lc - 08 E- 1
c r
e:
O-
. 1e - 09 m 1 1 1 1
0.5 l 1.5 2 2.5 3 3.5

Photon Energy (eV)

Figure 5.28. Photoconductance versus photon energy (a) before normalization and (b)
after normalization for the large grain size Al / diamond/ silicon sample. The average
grain size is 1.4 pm. (Electrical sample : P7)

163

 

 

 

 

 

 

 

 

 

19 - 05 I I I I I
‘ bin: 15v 0- 3
E? : bias-15V -e- j
= ..
D 16 w E - - : 2 #2 t a
e; = r
< 1: - 07 g- 1
8 i .
U * I
Ila " 4
g M“ W 1,
u E :
o 4
‘5 i i ‘
_= 1: - 10 f
o. a 3
1e -— 11 ‘ 1 ‘ ‘ ‘
0.5 l 1.5 2 2.5 3 3.5
Photon Energy (eV)
0.001 r r l l r
I bill: l5V(Si Detector) O-
A ' bin: 15V(Ge Detector) Q- ‘
=3 °~°°°1 { munavm Detector) ..— 1
:5 E biu:-I5V(Ge Detector) O— :
, le - 05
.o F l
t- ; j
5. 1e - 06 F 1
q; I :
U ’ 4
g le - 07 r '1
a E i
g b .
“U 1: - 08 g- 1
s E 5
8 1c - 09 r 1
a ’ :
.2 . .
1e - 11 L l l L
0.5 l 1.5 2 2.5 3 3.5
Photon Energy (eV)

Figure 5.29. Photoconductance versus photon energy (a) before normalization and (b)
after normalization for the large grain size Al/ diamond/ silicon sample. The average

grain size is 1.1 pm. Electrical sample : P20)

164

 

 

 

A' Diamond
7 5°
7 ----- FF++4————EF
(')%7(__.£x {r o {g g, Ev
‘ //4 ...........................................

Figure 5.30. The energy band diagram in the Al/p-type diamond interface.

165

than the small grain size diamond films prepared by the powder-polished method.
However, it doesn’t mean that the larger the grain size the better the rectifying
properties, independent of other variable. In this work, the sample which has the
best rectifying properties is not the sample with biggest grain size diamond crystal.
The degree of rectifying is also dependent on other properties of the metal / diamond
interface. Details of studies on the device with the best rectifying properties in this

study will be discussed in the next subsection.

5.4.4 Diamond Schottky Barrier Diode

The best metal/diamond/silicon rectifying structures were prepared by the paste-
polished method. The best of these rectifying characteristics were formed on film
N DF -P20, deposited at conditions of 700 W microwave input power, 60 Torr pressure,
1 % methane/ hydrogen concentration (1.5 sccm : 150 sccm). Typical I-V characteris-
tics of the tungsten point contact / diamond / silicon and Al / diamond / silicon structures
on the film are shown in Figure 5.31 and Figure 5.32 respectively. The Raman spec-
trum of N DF -P20 is shown in Figure 5.33. The point contact structure shows the
ohmic behavior which again supports the mechanism that the rectifying property is
a result of the Al/diamond interface. When two coplanar contacts are made, with
one contact being the evaporated aluminum circle and the other being the tungsten
point probe, SBD characteristics are again observed as shown in Figure 5.34. In this
case the current passes via the rectifying diamond/Al contact through the diamond
film to the silicon, through the silicon, and back up through the diamond film to the
ohmic tungsten point contact. V

Figure 5.35 shows that the forward bias current (1;) to the reverse bias current
(In) ratio at 10 V is almost 2x105. However, the ratio drops to 2x102 when the

voltage increases up to 25 V. Similar phenomena have been reported by other groups

166

 

 

 

8e-O5
66-05
48—05
2? 2.3—05
+3
5 o
t-a
S
O -2e-05

—4e - 05

-63 - 05

 

 

 

 

—86-05 1 l l l l l l l
—25 —20 -15 —10 —5 0 5 10 15 20 25

Voltage (V)

 

Figure 5.31. The LV characteristic of the point contact/diamond/silicon structure
showed ohmic behavior. The top contact is a tungsten point probe and the bottom
contact is the silicon wafer. (Electrical sample : P20)

167

 

53‘05 I I I fl I I I I

 

-5e — 05

—0.0001

—0.00015

I Current (A)

-0.0002

 

 

 

 

 

_0.00025' m i 1 1 1 i 1 L 1
-25 -20 -15 -10 -5 0 5 10 15 20 25

Voltage (V)

Figure 5.32. The best Schottky barrier diode (SBD) characteristic of the large grain
size Al/ diamond/ silicon samples. The top contact is aluminum and the bottom con-
tact is the silicon wafer. (Electrical sample : P20)

168

 

100000 J ' I ' I ' I ' l ' l ‘ I u
5‘ 80000 - -
U!

C
0
fl
.5
60000 - ..
A L l L l l l 1 A L I L l

 

 

 

 

400 600 800 1000 1200 1400 1600
wavenumber (cm")

Figure 5.33. The Raman spectrum of N DF-P20. The peak between 1550cm‘1 and
1600cm'l indicates that the existence of graphitic component in the film.

169

 

58-06 I I I w I I I I

 

0

-5e—06
—le—05
—1.5e-05
—2e-05

-2.5e — 05

Current (A)

—3e — 05
-3.5e - 05

 

—4e - 05<

 

 

 

l l l

—25 -20 -15 -10 --5 0 5 10 15 20 25

 

—4.5e — 05 L ‘ 1 1

Voltage (V)

Figure 5.34. The SBD I-V characteristic of a film with coplanar surface contacts.

One contact is aluminum and the other contact is a tungsten point probe. (Electrical
sample : P20)

170

[76, 80, 137]. It is noted that the samples of [76, 80] are intentionally boron-doped
and the samples of [133, 136, 137] and of this research are self-p-typed-doped during
the deposition.

The current-voltage results of the diode are best analyzed by considering the
current as a function of both the voltage across the junction and the voltage across

the bulk region. For the former
IzIoexp(qu/nkT) (5.17)

where V,- is the junction voltage, 17 is the diode ideality factor (1 S 17 S 2), T is
temperature, I: is Boltzmann’s constant and 10 is the saturation current. Then for

any current, the junction voltage is found to be

__nk_T

Vj q

ln(I/Io) (5-18)
and the bulk voltage is next found from
ill) = VApplicd " V7 (519)

Consequently, the forward I-V characteristic of the total structure is modeled by

nkT

v = _q_1n(1/10)+ f(I) (5.20)

where Vb = f (I ) represents the I-V characteristics of the bulk material. In its simplest

form, the relationship between the bulk voltage and the current would be ohmic,

v. = IR (5.21)

171

0.001 r T
0.0001
le - 05

1e — 06 ‘3. 1
. 0 E
1) .
1e - 07 0
‘5

Q5

lc - 08 0

(5

Current (A)

le-09

 

le—lO

 

le_11 l 1 l L l l l l
—25 -20 —15 —10 —5 0 5 10 15 20 25

Bias Voltage (V)

Figure 5.35. The log(I)-V characteristic of the SBD. (Electrical sample : P20)

172

where R = pL/A is the resistance of the diamond film. However non-ohmic behavior
is often observed in diamond films, an example being the field activated conductance
reported in chapter 5 (section 3.3) of this research.

The data to be modeled by Eq. (5.20) is the forward l—V characteristic of the diode
shown in the semilogarithmic plot of Figure 5.36. The parameters used in fitting the
equation to the data are 10, 17, and the functional relationship f(I).

The data shown in Figure 5.36 doesn’t show any straight line region in which
the I-V characteristics are dominated by the diode portion of Eq. (5.20). For all
currents in the range of measurements, the bulk voltage plays a major role in the total
voltage. However, two straight lines representing the diode portion of the voltage are
superimposed on the plot of Figure 5.36 which are consistent with the measured data.
These straight lines correspond to quite different Io and n values. The solid line shows
the diode voltage versus current for 10:5.2x10‘16 A and for n = 1. The dashed line
shows the diode voltage versus current for 10:2.9x10‘“ A and for 17 = 2. As will be
shown below, both lead to similar conclusions about the bulk I-V relationship, f(I).

One might expect that the f(l) relationship would follow the field activated con-
ductivity model of Eq. (5.12). However Figure 5.37 shows that this is not the case.
Above 10 V the data is consistent with a field activated conductivity with Fo=6x104
V/cm, consistent with the field activated model. However, below 10 V the model
does not fit, since the conductivity decreases rapidly instead of approaching a con-
stant value.

The data is better fit by assuming a power law for f(I) where
loch'" (5.22)

which m is the degree of power. As shown in Figure 5.38 and Figure 5.39 both m=3/ 2

and m=2 give a reasonable fit to the data, with the m=3/2 fit being marginally

173

 

 

 

 

0.01 I I I I
: Experiment 9-
’ Ideal diode: 17:1 —
0-001 ,r Ideal diode: q=2
0.0001 g 5 ' '
A 1e —05 E 5
<1 , 3 .~,
:: le — 06 r 3 4:"
a E I 0
Q , : 4!
: le — 07 I ‘1’
5 E .0
0 1e — 08 r3
so
1e — 09 °
1c — 10 I
18 _ 11 D _L 1 L l
0 5 10 15 20 25

Bias Voltage (V)

Figure 5.36. The forward bias I-V characteristic of the SBD. The 0 represents the
experimental data, the solid line shows the ideal diode with 17 = 1, 10:5.2x10‘16 A
and the dashed line shows the ideal diode with 17 = 2, Io=2.9x10"“ A. (Electrical
sample : P20)

174

 

 

 

 

 

 

 

0.0001 E . , I I .
: Bulk Voltage(17=1) — 5
Bulk Voltaga(n=2) - - - - 4

A . lc — 05 :- _

o : — 1

F: I
E 1

V
8 le - 06 E _
E
g u ..

“c 1e -- 07 1
C: I - 1
O ’ 2

U h

16 — 08 _:
16 - 09 l 1 1 1
0 5 10 15 20 25

Bulk Voltage (V)

Figure 5.37. Conductance-bulk voltage characteristic. It follows the field activated
model above 10 V. (Electrical sample : P20)

175

better. The energy band diagram and the equivalent circuit of the SBD are shown in
Figure 5.40(a) and (b) respectively. The exact choice of Io and 17 is not critical to the
fits since the diode portion of the voltage drop is relatively negligible, especially at
the higher range of voltages. The range of Io values corresponds to a barrier height
range of between 1.1 eV and 1.2 eV.

The significance of the power law dependence may be considered further as fol-
lows. Lampert and Mark loosely defined materials with band gap E, S 2 eV as
semiconductors and those with E, > 2 eV as insulators [138]. By this definition,
un-doped diamond with E9 = 5.5 eV can be considered as an insulator. In a perfect
trap free solid state insulator, the space-charge limited current-voltage relation can

be expressed as

IocV2 (5.23)

rather than the V3/2 . In considering the difference it is noted that the derivation of

Eq. (5.23) is based on the assumption that carriers drift at a velocity, v, with
v = pE (5.24)

where p is carrier mobility and E is applied electric field. However, if the field strength
is too high, the drift velocity of the carriers may vary as the square root of the applied
electric field when above a certain critical field strength E, [138]. This is known as a
characteristic of ”warm” carriers when acoustic phonon scattering is dominant. Then

for

E > 15., v = p(EE,.)1/2 (5.25)

The current-voltage relation becomes

maxi/Wm , (5.26)

176

 

0.0003 T I I I I

Bulk Voltage 71:1; —
Bulk Volta; 71:2

0.00025 , . ‘

I

r

0.0002

I

0.00015

Current (A)

0.0001

5e—05

I

 

 

 

0 l 1 l l l -
0 20 40 60 80 100 120

Bulk Voltage3/2 (V3/2)

 

Figure 5.38. The I-Vb3/2 characteristic of the SBD. (Electrical sample : P20)

 

Current (A)

177

 

 

 

 

0.0003 1 I I I 1
Bulk Voltage(q=l) .—
Bulk Voltage(q=2)
0.00025 r 4
0.0002 - -
0.00015 - -
0.0001~ —
5e - 05 - '1
0 i l l L l
0 100 200 300 400 500 600

Bulk Voltage2 (V2)

Figure 5.39. The I-VI,2 characteristic of the SBD. (Electrical sample : P20)

178

Al Diamond

/

 

W..NFPW.N.—IP.N.-w.wr- — —— -—— ————— EA
*— 5‘1

0 .. .’ 'I . _. .' _. .' .1 ..1
... .. .. .. .. .. .. .. .. .
t'-.‘-.--.‘-.~-.--.‘-.'-.'-.'-
..........
.........
. .. .. .. .. .. .. .. .. .. .
...., .,_....._. _. . . .
‘. ..'I . . .° .".".".' e
I' 'I- .- '.0 ‘I- '.- ‘.- '.v '.- ‘.- I. .
..........
.. .. .. .. .. .I .. .. .. .. -
v.,-.,-.,-.,~._-._-.,-c.'-.'.
.' ‘.“.‘.',..'..I"hl',l.‘,'.'...‘..-'..
. , ......
.. .. .. .. .. .. .. .. .- .
‘ .,-o,-.,-. -._-.,-._e.,-I.'.
.........
.1, ... .__ ._‘ ... '.- .1, ._, ... b. *
. I. o. u 4.... g . .~ 0.. ........................................................
|~.,-..-.,-..'._~.,-.,-..-..-.

.........
..............

..................
.............................

 

f(I)

{<1 I

 

 

I zIoexp(qV,- / nkT) I 0: VI,”

Figure 5.40. (a) The energy band diagram of the SBD. The barrier height d), equals to
the sum of built-in potential V.- and activation energy Ea. (b) The equivalent circuit

of the SBD.

179

which has the same functional form as the observed data.

For single crystal diamond, the high field phenomenon happens above 1x10s
V/ cm. In this case the film thickness is about 4 pm, so a voltage of 10 V corresponds
to an electric field of 2.5x10‘ V/ cm. The high field effect might not contribute to
the phenomenon observed here.

An obvious point is that the powder-polished samples of chapter 5 (section 3.3) fol-
low a field activated conductivity and exhibit ohmic, non—Schottky behavior whereas
the paste-polished samples of this section follow a power law I-V and Schottky behav-
ior. Both of these observations are consistent with the hypothesis that the powder-
polished films have a larger defect concentration. Consequently, metal / diamond space
charge layers are sufficiently thin to allow tunneling and ohmic behavior and their
high field properties are dominated by ionizable defects. I

It must also be noted, here that some of the SBD exhibited a particular form of
I instability. The current becomes much smaller after several times of measurements.
However, the rectifying behavior still exist.

Capacitance versus frequency measurement were also performed on the
Al/ diamond/ silicon diodes with a HP4192A impedance analyzer and the data is pre-
sented in Figure 5.41 for different values of reverse bias.

The barrier height 45, can also be determined by the capacitance versus voltage
measurement. By plotting l/C'2 against reverse bias, one can get built-in potential V.-
from the extrapolated intercept on the horizontal axis. From the slope of the straight
line one can also calculate the doping concentration. As shown in Figure 5.40(a) the

barrier height (13,, can be found from
psi/,- + E, (5.27)

where E, is activation energy. However, from Figure 5.42 the data shown did not lie

180

 

 

 

1e—07 . . if...” T shun, f . . ”m, r . . ....
bias: -1V 0
bias: 0V ‘1'
bias: IV D ‘
_ bias: 2V X _
16 08 o bias: 3v A
bias: 4V * I
A + o bias: 5v o -
hut (>0 . "
Q) +%> :
g D + .
C]
if. '3 the" o
3 161-10 0
x ‘1' o
i . am We
0 044m 5 ER“
le-ll
16-12 1 . A..1LLL 1 r 1...”! A .4..Lui
1 10 100 1000

 

Frequency (KHz)

Figure 5.41. Capacitance versus frequency characteristic at different reverse bias.
(Electrical sample : P20)

181

in a straight line. This may indicate that the self-doping of the film is non-uniform.
Consequently, the barrier height can not be determined from Figure 5.42.

In Penn State University’s work [133, 136, 137], they determine the ideality factor
as 1.8, and the barrier height as 1.15 eV for Al and Au contacts to p-type CVD
diamond films. They were not able to determine barrier height from the I-V curve,
instead they use the technique of internal photoemission to determine the barrier
height. In H. Shiomi, et al.’s work [76], an ideality factor of 2.1 and barrier height
of 1.2 eV were all determined from the I-V measurement of the W contact to boron-
doped diamond films. They noted that barrier height could not be obtained from
C-V curve due to the high series resistance in the bulk. In D. G. Jeng, et al.’s work
[80], they noted that an ideality factor of 1.85 was determined from the low voltage
region (0.1-0.5 eV) slope for the Al/ diamond Schottky diode. It is obvious that the
understanding of the synthetic diamond film diodes is still preliminary. Many factors
contribute to the film properties and since the deposition conditions are so different

from group to group, more experiments are needed to further investigation.

182

 

 

 

 

 

 

5 I I —I— F I I
A
‘7’ 4'
in
‘1:
:L 3‘
“I'
8
5 2‘
:2
8‘ 1.
O
o l 1 l L 1 L
-3 -2 -1 0 l 2 3 4 5

Bias Voltage (V)

Figure 5.42. Capacitance versus reverse bias characteristic at 50 kHz. (Electrical
sample : P20)

CHAPTER 6

Summary and Future Research

6.1 Summary of Important Results

The application of the microwave plasma disk reactor (MPDR) chemical vapor depo-
sition system for diamond synthesis has been successfully developed. Good quality
polycrystalline diamond films have been successfully deposited and characterized on
silicon and silicon nitride substrates. Two sample preparation methods, which are
paste-polished and powder-polished nucleation techniques have been used for deposit-
ing large grain size diamond films and small grain size diamond films respectively. A
back-etching processing technique was also developed to isolate the deposited dia-
mond films from the silicon substrate in order to study the diamond/ silicon interface
and allow access to evaporate metal contacts on both sides of the diamond films in
order to study the electrical properties of the diamond films.

A goal of this research has been to provide an increased understanding about the
relationship between properties of thin film diamond and the preparation conditions.
Toward this purpose several important physical characterizations of the diamond
films, such as Raman spectroscopy, X-ray photoelectron spectroscopy, DekTak pro-

file measurement, scanning electron microscopy, and laser scanning: microscopy were

183

184

carefully studied (see chapter 4). A particular focus of this research has been on the
electrical properties of the films, and consequently several important electrical prop-
erties of the diamond films, such as activation energy, I-V characteristics, high field
effect, and contact phenomenon were also studied (see chapter 5). The importance
of preparation conditions on the nature of metal contacts is described in chapter 5
and new information about the relationship between defects and high electric field
properties is provided by this research. The following subsections provide specific

summary highlights.

6.1.1 Nucleation Method

The nucleation method is’important for the diamond synthesis since as was shown
in this research, the density of nucleation sites is a major factor in determining the
grain size of the diamond films.

The paste-polished nucleation method produced comparatively larger grain size di-
amond films than the powder-polished nucleation methods did since the latter method
provides many more nucleation sites than the former method. It was also observed
experimentally that the average grain size obviously increased as the methane con-
centration increased for the paste-polished nucleation method but not the powder-
polished nucleation method. For both nucleation methods, the average grain size of
diamond increased as the microwave power or plasma density increased, but the effect
was smaller than the effect of increasing methane concentration in the case for the
paste-polished nucleation method.

For both methods, the growth rate of diamond increased as methane concentration
increases. The growth rate also increased as the microwave power or plasma density
increased, however, the effect was smaller than the methane concentration. At the

same deposition conditions, the growth rate of diamond by the powder-polished nu-

185

cleation method is slightly higher than that by the paste-polished nucleation method.

6.1.2 Quality of Diamond Films

Raman spectroscopy is used as the major technique to distinguish the quality of the
diamond films. The reported natural diamond peak is at 1332 cm“. The larger the
Raman signal at 1332 cm“1 and the smaller the other peaks, the higher quality of
the diamond films. It is known that the temperatures for depositing good quality
diamond films is a critical variable for diamond synthesis. If the temperature is
too low, an amorphous diamond-like component can become dominant and graphite
becomes dominant instead of diamond if the temperature is too high.

Generally, from this research it was observed, based on Raman results, that at
lower methane concentration one can get good quality diamond films in a wider
temperature range than at higher methane concentration. It was found that for the
range of power and pressure investigated, the preferential temperature range for good
quality diamond films is between 1030 °C and 1060 °C. It was also noted that the
two nucleation methods, under the same deposition conditions, appears to make no

obvious difference in the quality of the diamond film.

6.1.3 Back-Etching Technique

For film characterization, it is advantageous to have access to both sides of the film. In
this research, a back-etching process was successfully developed in cooperation with
Dr. Engemann at the University of Wuppertal (Germany) to transfer the diamond
film from the silicon substrate to an epoxy substrate.

There are two kinds of back-etched samples fabricated by this technique. One
kind of samples were used for physical characterizations (specifically XPS) of the

back surface (diamond/silicon interface). The other kind of samples were processed

186

to have metal contacts on both sides of the film for electrical property studies of the

diamond film.

6.1.4 Diamond/ Silicon Interface

From the back-etched samples prepared by the paste-polished nucleation method, it
is found that a thin layer of SiC existed at the diamond/silicon interface by X-ray
photoelectron spectroscopy analysis. This supports the hypothesis that the formation
of a thin layer of SiC is an inherent part of the growth of diamond on the silicon

substrate.

6.1.5 Activation Energy of Diamond Films

A four-point probe technique was used for sheet resistance measurements of the di-
amond films. Diamond films were deposited on an insulating composite material,
silicon nitride (Si3N4), using both paste-polished and powder-polished nucleation
methods. All the diamond films were determined as p-type by the hot probe tech-
nique.

The sheet resistance for the as-deposited samples prepared by the powder-polished
nucleation method was on the order of 105 to 106 9/0. By varying the temperature
for the sheet resistance measurement, activation energies of 0.22 to 0.31 eV and 0.13 to
0.23 eV were obtained for carrier mobility equals to 1200 and 12 cm2/V-s respectively
(see chapter 5, section 2.4). The samples were then annealed in a nitrogen ambient
furnace at 500 °C' for one hour. This caused the sheet resistance to increase by at
least four orders of magnitude and the activation energy to increase as well.

However, the sheet resistance of the as-deposited samples prepared by the paste-
polished nucleation method was on the order of 10‘ fl/D. This occurred because a

different system shut off procedure caused formation of a conducting layer by the

187

hydrogen plasma. When the sample was cleaned by a solution (see chapter 5, section
2.4) in order to remove the conducting layer, the sheet resistance was determined
on the order of 108 0/0 with an activation energy of 0.51 eV and 0.42 eV for hole

mobility equals to 1200 and 12 cmz/V-s respectively (see chapter 5, section 2.4).

6.1.6 Electric Field Dependent Conductivity of Diamond
Films

The dc electrical conductivity of polycrystalline diamond films with submicron grain
sizes and ohmic contacts was studied as a function of the applied electric field up
to the point of electrical breakdown. For electric fields below approximately 105
V/ cm, the films exhibited predominantly ohmic behavior with a conductivity that
was independent of the applied voltage for both high work function metal contacts
(Au) and low work function metal contacts (In). For higher electric fields, however,
the conductivity was field activated according to Poole’s Law. The data is consistent
with a Poole-Frenkel reduction of the ionization energy associated with Coulombic
potentials surrounding ionizable centers, where the Coulombic potentials overlap. The
ionizable centers may result from impurities or defects. In this study, experimental
data showed that it is a carrier concentration increase, not a mobility increase, that
gave rise to the field activated conductivity which is consistent with the Poole-Frenkel
mechanism.

Because of the field activated conductivity, the current at high electric fields is
substantially larger than would otherwise be the case. For 3.5 pm and 2 pm thick
films, the breakdown happens-at voltages higher than 250 V and 150 V respectively,
corresponding to a breakdown field of 7.1 x105 and 7.5x 105 V/ cm respectively. Con-
sequently, the dielectric strength of the sub-micron grain size polycrystalline diamond

films in this study are substantially less than those reported for single crystal dia-

188

mond (0.6 -1x107 V / cm). The slope of the conductivity versus electric field indicates

approximate one ionizable impurity or defect per 10,000 host atoms in such films.

6.1.7 Diamond Schottky Barrier Diode

Polycrystalline diamond Schottky barrier diodes were fabricated by using the large
grain size diamond films with the A1 / diamond/ p—silicon structure. Rectifying behav-
ior was determined in the Al/diamond interface as opposed to the diamond/silicon
interface. The I p/ I 3 ratio is almost 2x105 at 10 V in the best device, but only 2x102
at 25 V. The I-V characteristic can be modeled as an ideal Schottky diode in series
with an insulator, for which the property (I ocV’" relationship), is indicative of a space
charge limited current effect in the bulk diamond. However, the metal/diamond/p.
silicon samples for the small grain size diamond films exhibited ohmic property since
the defect states in these films are so large that carriers can easily tunnel through the

barrier.

6.2 Future Research

This research has studied the synthesis of diamond films in the MPDR deposition
system; physical characterizations, and electrical characterization of the diamond
films and diamond/ silicon samples. However, the understanding of the relationship
between the physical characterization and electrical properties is still in the early
stage. More detailed investigation into the topics covered in this dissertation as
well as the effects of other parameters (such as the vacuum purity, the addition of
oxygen, and other gases, and lower plasma pressure) are still needed to complete our

understanding of this research. Recommendations for future research follow.

189

6.2.1 Improvement for Diamond Deposition

The deposition system described in this research can be considerably improved with
regard to system purity and with regard to deposition over larger area substrate.
With regard to the latter, considerable progress has been made by a colleague, J.
Zhang [139]. With regard to system purity, the vacuum system used in this research
was mechanically pumped with a ultimate pressure of 1x10" Torr. Residual gas
composition and leak/outgas rates were not specially quantified and monitored on
a run-to-run basis. In order to better control impurities it would be useful to up-
grade the system to a 1><10'7 Torr ultimate pressure and to perform residual gas
analysis prior to runs. By providing a lower pressure capability, this also offers the
opportunity to investigate ECR plasma deposition of diamond at lower pressures and
temperatures. If the ECR plasma technique can be successfully used in an MPDR
system, the deposition area could be quite large, 6 inches in diameter or larger.
Other things being equal, there is always a trade-off between a large deposition
area and a high deposition rate since the. amount of reaction species is fixed for a
given reaction mixture, gas flow and microwave power. Consequently, the larger the
deposition area, the lower the deposition rate. With the addition of oxygen, however,
it is known that the quality and deposition rate can be improved. The extent to
which oxygen affects the quality and deposition rate needs to be carefully studied on

different configurations of this system.

6.2.2 The Techniques for Physical Characterization

It is known that Raman spectroscopy is the best technique to distinguish diamond,
graphite, and amorphous carbon. However, Raman spectroscopy only provides lim-
ited information about the sample, specifically it determines if the film is similar

to natural diamond having a sharp peak at 1332 cm“. For example, Raman spec-

190

troscopy can’t show the difference between a sample with and without a very thin
layer of conducting graphite or chemical contamination on top of the films. Also as
shown in Figure 4.7, Raman spectroscopy can’t show the difference between a large
grain size diamond film and a small grain size diamond film deposited at the same
conditions.

Other characterization techniques are therefore also needed in future work to
provide the basic properties of the film to assist the Raman spectroscopy in order to
have a better understanding of the over-all physical properties of the film. Examples
include secondary ion mass spectroscopy (SIMS) and Auger electron spectroscopy
(ABS) to determine the elemental composition of the film; X-ray diffraction (XRD) to
determine the crystal structure and lattice spacing; electron energy loss spectroscopy
(EELS) to determine the localized bonding in the bulk film; X-ray photoelectron
spectroscopy (XPS) to examine the chemical composition at the surface of the film;

and scanning electron microscopy (SEM) to examine the surface morphology.

6.2.3 The Role of SiC at the Diamond/ Silicon Interface

It is known that a thin layer of SiC exists at the diamond/silicon interface from
this and other research. It may be hypothesized that SiC contributed to the ohmic
behavior of the diamond/p-silicon interface for both small and large grain size dia-
mond films in this research. However, it is not clear at this point how SiC affects the
electrical properties of the diamond / silicon hetero junction. It would be of interest in
future research to carefully prepare back-etched samples with and without the SiC
layer in order to study this further. It would also be interesting to investigate how
the interfacial electrical properties vary if the nucleation methods is changed from
abrasion to seeding since the abrasion nucleation method scratches the silicon surface

and the seeding nucleation methods does not.

191

6.2.4 Defects States in Diamond Films

There are clearly defect states in polycrystalline diamond films because of the exis-
tence of grain boundaries. However, defect states in each single crystal is also possible
since the methane and hydrogen gases used in this research are 99.99 % and 99.999 %
pure respectively. In this research, it was shown that defect states play an important
role contributing to the electrical properties. For small grain size metal/ diamond / p-
silicon samples, the number of defect states is so large that it contributes to the ohmic
contact behavior. Large grain size Al/diamond/p-silicon samples, however, did show
some rectifying properties since there are less defect states. Further knowledge of
the defect states in the diamond films under different plasma pressure and. different
deposition system becomes an important future task for understanding the electrical

properties of synthetic diamond films.

6.2.5 Diamond Devices

A major future goal of diamond synthesis research is to fabricate useful dia—
mond devices, such as diamond diodes, diamond transistors, diamond metal-oxide-
semiconductor structures, and diamond integrated circuits (IC) with performance
advantages based on the excellent properties of diamond (see chapter 2, section 4).
There is a need for considerable future work toward the development of electronic
grade diamond.

A difficult, but important, future task for diamond device research is the devel-
opment of the capability to synthesize epitaxial diamond films on non-diamond sub-
strates. The existing techniques only can deposit epitaxial diamond films on diamond
substrates and polycrystalline diamond films on non-diamond substrate. Further-
more, it is important to evaluate and improve the performance of the poly-diamond

devices. Future device related tasks also include the development of improved tech-

192

niques for diamond doping and etching.

The investigation of the electrical properties of synthetic diamond is still in the
preliminary stage and the fabrication of diamond devices is still far from mature.
Consequently, much more future research effort is needed in order to explore the

potential development of diamond devices.

 

APPENDICES

APPENDIX A

Deposited Diamond Films

( 1 ) Samples by Diamond-Paste Nucleation Method

 

 

 

 

 

 

 

 

 

Sample CH4/ H2 MiCrowave Plasma
Number Fl (39% ate PIQEVUGtI’ Pressure Time Date
(NDF-#) (sccm) (W) (Torr) (hour)
N6 "5 % 600 60 7 6/17/‘89
(3 : 200 )
P6 ( £1523, ) 600 60 7 6/18/‘89
0
N7 (2132230 ) 600 50 7 6/19/‘89
0
P7 (2132:5230 ) 600 50 7 6/20/‘69
0.5 % ‘
P8, N8 (125 : 250 ) 700 80 6 8/31/' 89
0.5 %
P9, N9 (125 , 250) 700 30 6 9/2l’89
P10, N10 (1 3557550 ) 700 80 6 9/3l'89
0.5 %
P11. N11 (125 : 250 ) 825 80 6 9/4l'89

 

 

 

 

 

 

 

 

193

.
_ L
I

   

194

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample CH4/ H2 Microwave PI
and In Ut asma °
In??? Flow Rate poem; Pressure Time Date
( ' ) (sccm) (W) (Torr) (hour)
0.5%
P12: N12 (1.25:250) 700 70 6 9/5f89
P13 N13 05% 700 70 6 9/6l’89
' (1.25:250)
0.5 %
P14, N14 (125150) 700 60 8 9/7/’89
P15 N15 05% 800 60 8 SIB/’89
' (1.25:250)
0.5%
P16, N16 (125150) 600 60 a 9/9l’89
0.57
61 (1.25: 2°50) 600 60 6 9/10/‘89
0.5%
82 (125150) 600 - 60 6 9/11/‘89
0.5%
P17, N17 (125150) 700 60 1o 9/12/‘89
2%
P18,N18 (2,100) 700 60 5 10/18/‘89
2%
P19,N19 (2:100) 700 50 5 10/19/‘89
1°/
P20, N20 (1.5350) 700 60 3 11/5/89
10/ .
P21,N21 (1.5350) 700 70 8 11/6/‘89
1%
P22, N22 800 60 10 11/7/‘89

 

(1.5:150)

 

 

 

 

 

 

195

 

Microwave

 

 

 

 

 

 

 

 

 

 

 

 

33mg: :Eéégie (43553, Ifr'easfs'iife Time Date

(NDF-#) (sccm) (W) (Torr) (hour)

P23, N23 (1.51150) 800 50 10 11/8/‘89
P24, N24 (1.51330) 700 70 10 11/9/39
P25, P26 (1.51:?50') 700 70 14 11/10/39
P27, N27 (1.51150) 700 70 10 1/11/90
P28, N28 (2255:0150) 700 70 6 1/22/90
P29, N29 (2313505) 700 70 5 1/23/90
P30, N30 (1.;51:?50) 700 70 6 1/24/'90
P31,N31 (1.; 7:50) 700 70 5 2/6/‘90
P32, N32 (113'? ,‘Vgo, 700 70 6 2/3r90
N25, N26 (1.857;?50) 700 70 6 2/13I'90

 

 

 

 

 

 

 

196

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

( 2 ) Samples by Diamond-Powder Nucleation Method
Sample CH 4/ H2 Microwave Plasma
Numb” Floiimiiato P311711 Pressure “me Date
(NDF'#) (scorn) (W) (Torr) I(hour)
0.5 %
P33 (0.75: 150) 600 30 6 379790
0.5 %
P34 (035, 150) 600 70 6 3/13790
P35 1 % 600 70 5 3/151'90
( 1.5: 150)
1.5 %
P36 (225,150) 600 70 5 3/16l'90
1 %
P37 (1.5: 150) 600 60 5 3/19/‘90
0.5 %
P38 (075: ,50) 600 60 6 3122790
. 0.5 % . .
P39 (0.75: 150) 600 50 6 3123/90
P40 ’ % 600 50 5 3/27/‘90
( 1.5 : 150)
P41 1 % 750 50 5 3/29/‘90
( 1.5 : 150) ‘
0.5 %
P42 (0.75: 150) 500 60 6 4/6/‘90
0.5 %
P43 (075, 150) 500 50 6 4/9/‘90
0.5 %
33 (075,150) 500 50 6 4/11/‘90
0.5 °/
34 (0.75, :50) 400 50 6 4/12/‘90

 

 

 

197

 

Microwave

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample CH4/ H2 Plasma
In ut
Number Floaviffaate peas, Pressure “"19. Date
(NDF-#) (scorn) (W) (Torr) (hour)
0.5 %
55 (0.75 : 150); 600 50 6 4/13/90
P44 0.5 %
((175,150) 500 60 10 4/14790
P45 05% 400 60 10 4715790
(0.75 : 150)
P46 0'5 % 600 60 10 4/16/‘90
(0.75 : 150)
P47 0'5 % 600 50 10 4/17790
(0.75 : 150)
0.5 %
P48 ( 0.75 ; 150) 700 60 10 4/18/‘90
$6 0'5 % 320 6 4/19/‘90
' (0.75 : 150) 5°
P49 0'5 % 400 50 6 4120790
(0.75 : 150)
0.5 %
P50 (075 : ,50) 800 60 6 4/23790
P 1% 400
51 (15,150) 60 5 4124/90
P52 1%
. (1.5 : 150) 500 60 5 4/257'90
P 1%
53 (15,150) 700 60 5 4/27r90
0
s7 1 P 50 5 4/29/90

 

(1.5:150)

 

400

 

 

 

 

 

APPENDIX B

Electrical Samples

( 1 ) Metal/Diamond/Silicon Samples

 

 

 

 

 

 

 

 

 

Film # Metal Contact
P6 Al
N6 Al
P7 Al
N7 Al
P8 Al
N8 Al
P9 Al
N9 Al

 

 

 

198

 

199

 

Film #

Metal Contact

 

 

 

 

 

 

 

 

 

 

 

P1 0 Al
N1 0 Al
P20 Al
N20 Al
P21 Al
N21 Al
P39 Au
P39 Ag
P39 In
P49 Au
P49 Ag

 

 

P49

 

In

 

 

 

(2) Metal/Diamond/Metal Back-Etched Samples

200

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Film # Top Bottom
Metal Contact Metal Contact
N14-T1 - - - Cr/Au
N14-T2 - - - Au
P33-T1 Ag A9
P33-T2 Ag Pt/Ag
P33-T3 In Pt/Ag
P44-T1 Au ' Ag
P44-T2 In In
P44-T3 In Pt/Ag
P46-Tl Au Ag
P46-T2 Au Ag
P46-T3 In In
P46-T4 Au In

 

 

 

201

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

an # Top Bottom
Metal Contact Metal Contact
P45-T1 - - - Ag
P45-T1 - - - A9
P45-T1 - - - Ag
P47-T1 Au AU
P47-T2 Au Au
P47-T3 In Au
P48-T1 - - - Au
P48-T2 - - - Au
P48-T3 - - - Au
P49-T1 Ag Ag
P49-T2 Ag Ag
P49-T3 Au Ag

 

 

 

 

BIBLIOGRAPHY

BIBLIOGRAPHY

[1] F. P. Bundy, H. T. Hall, H. M. Strong, R. H. Wentorf Jr., Nature, 176, pp.51,

1955.

[2] J. Asmusson and J. Root, ”Characteristics of a Microwave Plasma Disk Ion

[3]

[4]

[5]

[6]

[7]

[8]

[9]

Source”, Appl. Phys. Lett., 44, pp.396, 1984.

J. Root and J. Asmusson, ”Experimental Performance of a Microwave Cavity
Plasma Disk Ion Source”, Rev. of Sci. Inst., 56, pp.1511, 1985.

J. Hopwood, D. K. Reinhard, J. Asmusson, ”Experimental Conditions for Uni-
form Anisotropic Etching of Silicon with a Microwave Electron Cyclotron Reso-

nance Plasma System”, Journal of Vacuum Science and Technology, B, 6, pp.268,
1988.

Dahimene, M., ”Development of a Microwave Ion and Plasma Source Immersed

in a Multicusp Electron Cyclotron Resonant Magnetic Field”, Ph.D. dissertation,

Michigan State University, 1987.

J. Hopwood, ”Macroscopic Properties of a Multipolar Electron Cyclotron Reso-
nance Microwave-Cavity Plasma Source for Anisotropic Silicon Etching”, Michi-
gan State University, 1990.

J. Hopwood, R. Wagner, D. K. Reinhard, J. Asmussen, ”Electric Fields in
a Microwave-Cavity ECR Plasma Source”, J. Vac. Sci. Techno]. A, 8, pp. ,
May/June 1990.

G. T. Salbert, D. K. Reinhard, and J. Asmussen, ”Oxide Growth on Silicon
Using a Microwave Electron Cyclotron Resonant Oxygen Plasma”, J. Vac. Sci.
Techno]. A, 8, pp. May/June 1990.

T. Roppe], D. K. Reinhard, and J. Asmussen, ”Low Temperature Oxidation of
Silicon Using a Microwave Plasma Source”, J. Vac. Sci. Technol. B, 4, pp.295,
1986. ' '

202

 

203

[10] J. Zhang, B. Huang, D. K. Reinhard, and J. Asmussen, ”An Investigation of The
Effects of Electromagnetic Field Patterns on Microwave Plasma Diamond Thin
Film Deposition”, J. Vac. Sci. Technol. A, 8, pp.2124, 1990.

[11] R. F. Curl and R. E. Smalley, ”Fullerenes”, Scientific American, 265, pp.54,
1991.

[12] P. W. Bridgman, ”Synthetic Diamond”, Scientific American, 42, pp.42, 1955.

[13] P. K. Bachmann, R. Messier, ”Emerging Technology of Diamond Thin Films”,
C&EN, pp.24, 1989.

[14] S. Aisenberg, R. Chabot, ”Ion Beam Deposition of Thin Films of Diamondlike
carbon”, J. Appl. Phys., 42, pp.2953, 1971.

[15] E. G. Spencer, P. H. Schmidt, D. J. Joy, and F. J. Sanssalone, ”Appl. Phys.
Letter., 29, pp.118, 1976.

[16] C. Weissmantel, K. Bewilogua, D. Dietrich, H. Erler, H. Hinneberg, S. Klose, W.
Nowick, and G. Reisse, Thin Solid Films, 72, pp.19, 1980.

[17] T. J. Moravec, T. W. Orent, J. Vac. Sci. Technol. 18, pp.226, 1981.

[18] B. V. Derjaguin, B. V. Spitsyn, L. L. Bouilov, A. A. Klochkov, A. E. Gorodestskii,
and A. V. Smol’yanov, Sov Phys. Dokl, 21, pp.767, 1976.

[19] S. Matsumoto, Y. Sato, M. Kamo, and N. Setaka, ”Vapor Deposition of Diamond
Particles from Methane”, Japanese J. Appl. Phys., 21, pp.L183, 1982.

[20] W. D. Kingery, H. K. Bowen, D. R. Uhlmann, ”Introduction to Ceramics”,2nd
edition, pp.313, John Wiley & Sons, Inc. 1976.

[21] RC. DeVries, ”Synthesis of Diamond under Metastable Conditions.” Ann. Rev.
Mater. Sci., 17, pp.161, 1987.

[22] M. Tsuda, M. N akajima, S. Oikawa, ”Epitaxial Growth Mechanism of Diamond
Crystal in CH4-—2 plasma.” J. Am. Chem. Soc., 108, pp.5780, 1986.

[23] S. J. Harris, ”Mechanism For Diamond Growth From Methyl Radicals.” Appl.
Phys. Lett. 56(23), pp.2298, 1990.

[24] M. Frenklach, K. E. Spear, ”Growth Mechanism of Vapor-deposited Diamond.”
J. Mater. Res. 3(1), pp.133, 1987.

204

[25] Y. Hirose, Y. Terasawa, Jpn. J. Appl. Phys., 25, pp.L519, 1986.

[26] T. Kawato, K-I Kondo, ”Effects of Oxygen on CVD Diamond Synthesis.” Jpn.
J. Appl. Phys., 26(9), pp.1429, 1987.

[27] S. J. Harris and M. Weiner, ”Effects of Oxygen on Diamond Growth.” Appl.
Phys. Lett., 55(21), pp.2179, 1989.

[28] Y. Liou, R. Weimer, D. Knight, and R. Messier, ”Effect of Oxygen in Diamond
Deposition at Low Substrate Temperatures.” Appl. Phys. Lett., 56(5), pp.437,
1990.

[29] Y. Matsui, H. Yabe, Y. Hirose, ”The Growth of Diamond Crystals in Acetylene
Flames.” Jpn. J. Appl. Phys., 29(8), pp.1552, 1990.

[30] Y. Tzeng, C. Cutshaw, R. Phillips, T. Srivinyunon, ”Growth of Diamond Films
on Silicon From an Oxygen-acetylene Flame.” Appl. Phys. Lett. 56(2), pp.134,
1990.

[31] M. Frenklach, ”The role of Hydrogen in Vapor Deposition of Diamond.” J. Appl.
Phys. 65(12), pp.5142, 1989.

[32] J. A. Mucha, D.L. Flamm, D. E. Ibbotson, ”On the Role of Oxygen and Hydrogen
in Diamond-forming Discharge.” J. Appl. Phys., 65(9), pp.3448, 1990.

[33] A. Sawaba and T. Inuzuka, ”Growth of Diamond Thin Films by Electron Assisted
Chemical Vapor Deposition.” Appl. Phys. Lett., 46(2), pp.146, 1985.

[34] Y. Hirose and Y. Terasawa, ”Synthesis of Diamond Films by Thermal CVD
Using Organic Compounds.” Jpn. J. Appl. Phys., 25(6), pp.L159, 1986.

[35] H.‘ Aikyo and K. I. Kondo, ”Diamond Synthesis Suppressing the Thermal De-
composition of Methane in a Hot-Filament CVD.” Jpn. J. Appl. Phys., 28(9),
pp.L1631, 1989.

[36] Y. H. Lee, P. D. Richard, K. J. Bachmann, J. T. Glass, ”Bias-Controlled Chemi-
cal Vapor Deposition of Diamond Thin Films.” Appl. Phys. Lett., 56(7), pp.620,
1990.

[37] M. Ihara, H. Maeno, K. Miyamoto, and H. Komiyama, ”Diamond Deposition
on Silicon Surface Heated to Temperature as low as 135 °C.” Appl. Phys. Lett.,
59(12), pp.1473, 1991.

-/

205

[38] K. Suzuki, A. Sawabe, H. Yasuda, T. Inuzuka, ”Growth of Diamond Thin Films
by Dc Plasma Chemical Vapor Deposition.” Appl. Phys. Lett., 50(12), pp.728,
1987.

[39] S. Koizumi, T. Murakami, T. Inuzuka, ”Epitaxial Growth of Diamond Thin
Films on Cubic Boron Nitride (111) Surfaces by DC Plasma Chemical Vapor
Deposition.” Appl. Phys. Lett., 57(6), pp.563, 1990.

[40] S. Nakao, M. Noda, H. Kusakabe, H. Shimizu, ”Deposition of Diamond onto an
Aluminum Substrate by DC Plasma CVD.” Jpn. J. Appl. Phys., 29(8), pp.1511,
1990.

[41] K. Kurihara, K. Sasaki, M. Kawarada, N. Koshino, ”High Rate Synthesis of
Diamond by DC Plasma Jet Chemical Vapor Deposition.” Appl. Phys. Lett.,
52(6), pp.437, 1988.

[42] F. Akatsuka, Y. Hirose, K. Komaki, ”Rapid Growth of Diamond Films by Arc
Discharge Plasma CVD.” Jpn. J. Appl. Phys., 27(9), pp. L1600, 1988.

[43] S. Matsumoto, I. Hosoya, T. Chounan, ”Substrate Bias Effect on Diamond De-
position by DC Plasma Jet.” Jpn. J. Appl. Lett., 29(10), pp.2082, 1990.

[44] S. Matsumoto, M. Hino, T. Kobayashi, ”Synthesis of Diamond Films in a ti
induction Thermal Plasma.” Appl. Phys. Lett., 51 (10), pp.737, 1987.

[45] D. E. Meyer, R. O. Dillon, J. A. Woollam, ”Radio-frequency Plasma Chemical
Vapor Deposition Growth of Diamond.” J. Vac. Sci. Technol., A 7(3), pp.2325,
1989.

[46] M. Kamo, Y. Sato, S. Matsumoto, N. Setaka, ”Diamond synthesis from Gas
Phase in Microwave Plasma.” J. Crys. Growth, 62, pp.642, 1983.

[47] C. P. Chang, D. L. Flamm, D. E. Ibbotson, J. A. Mucha, ”Diamond Crystal
Growth by Plasma Chemical Vapor Deposition.” J. Appl. Phys., 63( 5), pp.1744,
1988.

[48] I. Watanabe and K. Sugata, ”Diamond Films Synthesized by Microwave Plasma
CVD of Ethyl A]coho],etc.” Jpn. J. Appl. Phys., 27(8), pp.1397, 1988.

[49] J. S. Ma, H. Kawarada, T. Yonehara, J. I. Suzuki, J. Wei, ”Selective Nucleation
and Growth of Diamond Particles by Plasma-assisted Chemical Vapor Deposi-
tion.” Appl. Phys. Lett., 55(11), pp.107l, 1989. I

 

206

[50] H. Shiomi, K. Tanabe, Y. Nishibayashi, N. Fujimori, ”Epitaxial Growth of High
Quality Diamond Film by the Microwave Plasma-assisted Chemical Vapor De-
position.” Jpn. J. Appl. Phys., 29(1), pp.34, 1990.

[51] D. G. Jeng, H. S. Tuan, R. F. Salat, G. J. Fricano, ”Oriented Cubic N ucleations
and Local Epitaxial During Diamond Growth on Silicon (100) Substrates.” Appl.
Phys. Lett., 56(20), pp.1968, 1990.

[52] Y. Mitsu, T. Yoshida, K. Akashi, ”Development of a New Microwave Plasma
Torch and Its Application to Diamond Synthesis.” Rev. Sci. Instrum., 60(2),
pp.249, 1989.

[53] J .-I. Suzuki, H. Kawarada, K. S. Mar, J. Wei, Y. Yokota, A. Hiraki, ”The Syn-
thesis of Diamond Films at Lower Pressure and at Lower Temperature Using
Magneto-Microwave Plasma CVD.” Jpn. J. Appl. Phys., 28(2), pp.L28l, 1989.

[54] J. Wei, H. Kawarada, J.I. Suzuki, A. Hiraki, ”Low Temperature Synthesis of
Diamond Films Using Magneto—Microwave Plasma CVD.” Jpn. J. Appl. Phys.,
29(8), pp.L1483, 1990.

[55] W.L. Hsu, D. M. Tung, E. A. Fuchs, K. F. McCarty, ”Low-Temperature Diamond
Growth in a Microwave Discharge.” Appl. Phys. Lett., 55(26), pp.2739, 1989.

[56] Y. Liou, A. Inspektor, R. Weimer, R. Messier, ”Low-temperature Diamond De-
position by Microwave Plasma-enhanced CVD.” Appl. Phys. Lett., 55(7), pp.631,
1989.

[57] D. J. Pickrell, W. Zhu, A. R. Badzian, R. Messier, ”Downstream Plasma-
enhanced Diamond Film Deposition.” Appl. Phys. Lett., 56(20), pp.2010, 1990.

[58] D. E. Patterson, B. J. Bai, C. J. Chu, R. H. Hauge, and J .L. Margrave, ”Halogen
Assisted CVD of Diamond”, pp.14, Abstract of 2nd international conference on
the New Diamond Science and Technology, Washington,DC, Sep.23—27, 1990.

[59] C. Weissmantel, Thin Solid Films, 35, pp.255, 1976.

[60] T. Mori and Y. Namba, ”Hard diamondlike carbon films deposited by ionized
deposition of Methane Gas.” J. Vac. Sci. Techno]. A 1(1), pp.23, 1983.

[61] T. Mori and Y. Namba, ”Crystal Structure of Diamondlike Carbon Films Pre-
pared by Ionized Deposition from Methane Gas.” J. Appl. Phys. 55(9), pp.3276,
1984. '

 

"I
—
I.

207

[62] M. Kitabatake and K. Wasa, ”Growth of Diamond at Room Temperature by A
Ion-beam Sputtering Deposition.” J. Appl. Phys., 58(4), pp.1693, 1985.

[63] J. W. Rabalais and S. Kasi, ”Growth of Thin Chemically Bonded Diamondlike
Films by Ion Beam Deposition.” Science, 239, pp.623, 1988.

[64] J. C. Angus and C. C. Hayman, ”Low-Pressure, Metastable Growth of Diamond
and Diamondlike Phases.” Science, 241, pp.913, 1988.

[65] personal discussion with Dr. Richard Guarnier, senior engineer in IBM Watson

research center.

[66] A. Masood, M. Aslam, M. A. Tamor, and T. J. Potter, ”Techniques for Pattern-
ing of CVD Diamond Films on Non-Diamond Substrates.” J. Electrochem. Soc.,
138(11), pp.L67, 1991.

[67] S. Iijima, Y. Aikawa, and K. Baba, ”Growth of Diamond Particles in Chemical
Vapor Deposition.” J. Mater. Res., 6(7), pp.1491, 1991.

[68] L. C. Conner, ”CVD Diamond : Beyond the Laboratory.” , Gorham Advanced
Material Institute at The Global Business and Technical Outlook for High Per-

formance Inorganic Thin Film Coatings, 1988.

[69] W. A. Yarbrough, ”Diamond and Diamond-Like Coatings by CVD.” , Gorham
Advanced Material Institute at The Global Business and Technical Outlook for
High Performance Inorganic Thin Film Coatings, 1988.

[70] R. Orr, ”The Impact of Thin Film Diamond on Advanced Engineering Systems.”
, Gorham Advanced Material Institute at The Global Business and Technical
Outlook for High Performance Inorganic Thin Film Coatings, 1988.

[71] K. Shenai, R. S. Scott and B. J. Baliga, ”Optimum Semiconductors for High-
Power Electronics.” IEEE transaction on electron devices, 36(9), pp.1811, 1989.

[72] J. F. Prins, ”Bipolar transistor action in ion implanted diamond.” Appl. Phys.
Lett., 41 (10), pp.950, 1982.

[73] M. W. Geis, D. D. Rathman, D. J. Ehrlich, R. A. Murphy, and W. T. Lindley,
”High-Temperature Point-Contact Transistor and Schottky Diodes Formed on
Synthetic Boron-Doped Diamond.” IEEE Electron Device letters, 8(8), pp.341,
1987.

l/_.

208

[74] N. Fujimori, T. Imai, and A. Doi, ”Characterization of Conducting Diamond
Films.” Vacuum, 36, pp.99, 1986.

[75] J. Mort, D. Kuhman, M. Machonkin, M. Morgan, F. Jansen, and K. Okumura,
”Boron Doping of Diamond Thin Films.” Appl. Phys. Lett., 55(11), pp.1121,
1989.

[76] H. Shiomi, H. N akahata, T. Imai, Y. Nishibayashi, and N. Fujimori, ”Electrical
Characteristics of Metal Contacts to Boron-Doped Diamond Epitaxial Film.”
Jpn. J. Appl. Phys., 28(5), pp.758, 1989.

[77] N. Fujimori, H. Nakahata, T. Imai, ”Properties of Boron-Doped Epitaxial Dia-
mond Films.” Jpn. J. Appl. Phys., 28(5), pp.824, 1990.

[78] K. Okana, H. N aruki, Y. Akiba, T. Kurosu, M. Iida, and Y. Hirose, ”Synthesis of
Diamond Thin Films Having Semiconductive Properties.” Jpn. J. Appl. Phys.,
27, pp.Ll73, 1988.

[79] S. A. Grot, G. SH. Gildenblat, C. W. Hatfield, C. R. Wronski, A. R. Badzian,
T. Badzian and R. Messier, ”The Effect of Surface Treatment on the Electrical
Properties of Metal Contacts to Boron-Doped Homoepitaxial Diamond Film.”,
IEEE Electron Device Letters, 11(2), pp.100, 1990.

[80] D. G. Jeng, H. S. Tuan, R. F. Salat, and G. J. Fricano, ”Thin-Film Al/Diamond
' Schottky Diode over 400V Breakdown Voltage.” J. Appl. Phys., 68(11), pp.5902,
1990.

[81] H. Shiomi, Y. Nishibayashi, and N. Fujimori, ”Field-Effect Transistor Using
Boron-Doped Diamond Epitaxial Films.” Jpn. J. Appl. Phys., 28(12), pp.L2153,
1989. ‘

[82] K. Okana, H. Kiyota, T. Iwasaki, T. Kurosu, and M. Iida, ”P-N Junction Diode
Made of Semiconducting Diamond Films.” Appl. Phys. Lett., 58(8), pp.840,
1991'.

[83] G. Sh. Gildenblat, S. A. Grot, C. W. Hatfield, and A. R. Badzian, ”High-
Temperature Thin-Film Diamond Field effect Transistor Fabricated Using a Se-
lective Growth Method.” IEEE Electron Device Letters, 12(2), pp.37, 1991. ,

[84] W. Tsai, M. Delfino, D. Hodul, M. Riaziat, L. Y. Ching, G. Reynold, and C.
B. Cooper, ”Diamond MESFET Using Ultrashallow RTP Boron Doping.” IEEE
Electron Device Letters, 12(4), pp.157, 1991. i

l/_.

209

[85] M. W. Geis, J. A. Gregory, and B. B. Pate, ”C-V Measurements on Metal—32°02-
Diamond Structures Fabricated with (100) and (111) Oriented Substrates.” IEEE
Transaction on Electron Devices. 38(3), pp.619, 1991.

[86] T. Inoue, H. Tachibana, K. Kumagai, K. Miyata, K. Nishimura, ”Selected-Area
Deposition of Diamond films.” J. Appl. Phys., 67(12), pp.7329, 1990.

[87] R. Ramesham, T. Roppel, C. Ellis, D. A. Jaworske, W. Bough, ”Selective and
Low Temperature Synthesis of Polycrystalline Diamond.” J. Mater. Res., 6(6),
pp.1278, 1991.

[88] K. Okumura, J. Mort, and M. Machonkin, ”Lithium Doping and Photoemission
of Diamond Thin Films.” Appl. Phys. Lett., 57(18), pp.1907, 1990.

[89] M. W. Geis, H. I. Smith, A. Argoitia, J. Angus, G.-H. M. Ma, J. T. Glass, J. But-
ler, C. J. Robinson, R. Pryor, ”Larg-area Mosaic Diamond Films Approaching
Single-Crystal Quality.” Appl. Phys. Lett., 58(22), pp.2485, 1991.

[90] J. Asmussen, H.H. Lin, B. Manring, and R. Fritz, Rev. of Sci. Inst., 58, pp.1497,
1987.

[91] W. Kern and D. A. Puotinen, RCA Rev., 31, pp.187, 1970.

[92] RF. Harrington, ”Time Harmonic Electromagnetic Fields.” pp. 216, McGraw-
Hill, New York, 1961.

[93] L. Mahoney, M. Dahimene, and J. Asmussen, Rev. Sci. Instrum. 59, pp.448,
1988.

[94] S. Whitehair, J. Asmussen, and S. Nakanishi, J. of Propu]. Power, 3, pp.136,
1987.

[95] J. Engemann and H. Keller, D. K. Reinhard, B. Huang, and J. Asmussen, ”Dual-
side Contact Formation on Isolated Diamond Films.” Appl. Phys. Lett. 57(23),
pp.2461, 1990.

[96] Torr — S ealT M is manufactured by Varian Industrial Components Division, 3100
Jay Street, Santa Clara, CA 95054.

[97] Epo — TckTM is manufactured by Epoxy Technology Inc., Billerica, MA 01821.

[98] M. Balkanski, editor, ”Light Scattering in Solids.” pp.65, Flammarion Sciences,
Paris, 1971. i '

 

A /—.

210

[99] S. A. Solin and A. K. Ramdas, ”Raman Spectrum of Diamond.” Physical Review
B, 1(4), pp.1687, 1970.

[100] M. Yoshikawa, G. Katagiri, H. Ishida, and A. Ishitani, ”Characterization of
Crystalline Quality of Diamond Films by Raman Spectroscopy.” Appl. Phys.
Lett. 55(25), pp.2608, 1989.

[101] M. Yoshikawa, G. Katagiri, H. Ishida, and A. Ishitani, ”Raman Spectra of
Diamondlike Amorphous Carbon Films.” J. Appl. Phys. 64 (1 1), pp. 6464, 1988.

[102] R. G. Buckley, T. D. Moustakas, Ling Ye, and J. Varon, ”Characterization
of F ilament-assisted Chemical Vapor Deposition Diamond Films Using Raman
Spectroscopy.” J. Appl. Phys. 66(8), pp.3595, 1989.

[103] Diane. K. Knight and William B. White, ”Characterization of Diamond F ilrns
by Raman Spectroscopy.” J. Mater. Res., 4(2), pp.385, 1989.

[104] The diagram of the set-up of the Raman spectroscopy is a courtesy from Dr.
Kevin Gray in Norton company.

[105] P. E. Sobol, ”A comparison of Techniques for Compositional and Chemical
Analysis of Surface.” technical reprint by Perkin Elmer, Physical Electronics

Division.

[106] .J. Ruan, W. J. Choyke, and W. D. Partlow, ”Si Impurity in Chemical Vapor
Deposited Diamond Films.” Appl. Phys. Lett., 58(3), pp.295, 1991.

[107] Y. Mizokawa, T. Miyasato, S. Nakamura, K. M. Geib and C. W. Wilrnsen, J.
Vac. Sci. Techno]., A 5(5), pp.2809, 1987.

[108] BE. Williams and J .T. Glass, ”Characterization of Diamond Films: Diamond
Phase Identification, Surface Morphology, and Defect Structures.” J. Mater, Res.,
4 (2), pp.373, 1989.

[109] D.N. Belton, S.J. Harris, S.J. Schmieg, A.M. Weiner, and T.A. Perry, ”In situ
Characterization of Diamond Nucleation and Growth.” Appl. Phys. Lett., 54(5),
pp.416, 1989.

[110] R. Meilunas, M.S. Wong, K.C. Sheng and R.P.H. Chang, ”Early stages of
plasma synthesis of diamond films.” Appl. Phys. Lett., 54 (22), pp.2204, 1989.

211

[111] T. Hartnett, R. Miller, D. Montanari, C. Willingham, and R. Tustison, ”Inter-
mediate Layers for the Deposition of Polycrystalline Diamond Films.” J. Vac.
Sci. Technol., A 8(3), pp.2129, 1990.

[112] D.-W. Kweon and J.-Y. Lee, ”The Interface Morphology of Diamond Film
Grown on Si.” J. Appl. Phy. 68(8), pp.4272, 1990.

[113] D.J. Pickrell, W. Zhu, A.R. Badzian, R.E. Newnham, and R. Messier, ”Near-
interface Characterization of Diamond Films on Silica and Silicon.” J. Mater.
Res., 6(6), pp.1264, 1991.

[114] B. L. Gabriel, ”SEM: A User’s Manual for Material Science.” pp.1-5, published
by American Society for Metals, Meta] park, Ohio 44073.

[115] J.F. Denatale, J .F. Flintoif, and AB. Harker, Microstructural Control of Di-
amond Thin Films by Microlithographic Patterning.” J. Appl. Phys., 68(8),
pp.4014, 1990.

[116] P. Baurschmidt, CLSM for biomedical research, Carl Zeiss, Inc., 1988.

[117] F. M. Smits, ”Measurement of Sheet Resistivities with the Four Point Probe.”
Bell System Technical Journal (May 1958), pp.711.

[118] S. M. Sze, ”Physics of Semiconductor Physics.” 2nd edition, Chapter 1, John
Wiley & Son, New York, 1981. .

[119] K. Okano, H. Naruki, Y. Akiba, T. Kurosu, M. Iida, Y. Hirose, and T. Naka-
mura, ”Characterization of Boron-Doped Diamond Film.” Jpn. J. Appl. Phys.,
28(6), pp.1066, 1989.

[120] J. E. Field, ”Properties of Diamond”, Academic Press, London, Chap. 3, pp.79,
1979.

[121] J. P. Mckelvey, ”Solid State and Semiconductor Physics.” , Robert E. Krieger
publishing company, Malabar, Florida, 1971.

[122] M. I. Landstrass and K. V. Ravi, ”Hydrogen Passivation of Electrically Active
Defects in Diamond.” Appl. Phys. Lett., 55(14), pp.1391, 1989.

[123] M. I. Landstrass and K. V. Ravi, ” Resistivity of Chemical Vapor Deposited
Diamond Films.” Appl. Phys. Lett., 55(10), pp.975, 1989.

.'l

212

[124] S. Albin and L. Walkins, ”Electrical Properties of Hydrogenated Diamond.”
Appl. Phys. Lett., 56(15), pp.1454, 1990.

[125] E. H. Rhoderick, ”Metal-Semiconductor Contacts.” Clarendon Press, Oxford,
pp.54, 1980.

[126] Y. Tzeng, C. C. Tin, R. Philllips, T. Srivinyunon, and Y. Chen, ”High Electrical
Resistivity Diamond Films Deposited from an Oxyacetylene Flame.” Appl. Phys.
Lett., 57(8), pp.789, 1990. i

[127] A. T. Collins, E. C. Lightowlers, and W. S. Williams, Diamond Res. (Suppl.
Ind. Diamond Rev.), pp.19, 1970.

[128] H. H. Poole, Loud. Edinb. Dub]. Phil. Mag., 33, pp.112, 1916.
[129] J. Frenkel, Phys. Rev., 54, pp.647, 1938.
[130] R. M. Hill, Phil. Mag, 23, pp.59, 1971.

[131] A. V. Bogdanov, I. M. Vikulin, and T. V. Bogdanova, Sov. Phys. Semicond.,
16, pp.720, 1982.

[132] A. T. Collins, M. Kamo and Y. Sato, ”A Spectroscopic Study of Optical Cen-
ters in Diamond Grown by Microwave-Assisted Chemical Vapor Deposition.” J.
Mater. Res., 5(11), pp.2507, 1990.

[133] G. Sh Gildenblat, S. A. Grot, C. R. Wronski, A. R. Badzian, T. Badzian, and
R. Messier, Appl. Phys. Lett., 53, pp. 586, 1988.

[134] C. L. Ellison, R. M. Cohen, and J. T. Hoggins, Materials Research Society
Conference Record, Boston, MA, pp.2l3, 1989.

[135] G. H. Glover, Solid State Electronics, 16, pp.973, 1973.

[136] M. C. Hicks, C. R. Wronski, S. A. Grot, and G. Sh. Gildenblat, ”The Barrier
Height of Schottky Diodes with a Chemical Vapor Deposition Diamond Base.”
J. Appl. Phys., 65(5), pp.2139, 1989.

[137] S. A. Grot, S. Lee, G. Sb Gildenblat, C. W. Hatfield, C. R. Wronski, A. R.
Badzian, T. Badzian and R. Messier, ”Rectification and Internal Photoemis-
sion in Metal/CVD Diamond and Metal/CVD Diamond/Silicon Structures.” J.
Mater. Res., 5(11), pp.2497, 1990.

213

[138] M. A. Lampert and P. Mark, ”Current Injection in Solids.” Academic Press,Inc.,
New York, pp.14-30, 1970.

[139] J. Zhang, ”Diamond Film Deposition By Microwave Plasma Assisted Chemical
Vapor Deposition” Ph.D dissertation, Michigan State University, (in progress).