a" .

v.
al.1-
$233.;
. A. «1C
V . . 13
is

 

 

4r... . [5.27.-

:wfiu
a ..

rd

0.
unit...

:—

11. ...l...
I. ’3.
1 1

 

In
C
:3ng

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

GrOwth of single crystal diamond

presented by
Murari Regmi

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Physics and Astronomy

W“

/‘

lfl;

“ Major Professor’s Signature

33 (BMW—f. 7-007
Q [

Date

MSU is an affirmative-action, equal-opportunity employer

 

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

 

DATEDUE

DATEDUE

DAIEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:IClRC/DateDue.indd-p.1

 

GROWTH OF SINGLE CRYSTAL DIAMOND
By

Murari Regmi

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY

Department of Physics and Astronomy

2007

ABSTRACT
GROWTH OF SINGLE CRYSTAL DIAMOND
By

Murari Regmi

The subject of the research presented in this dissertation is the growth of single
crystal diamond by microwave plasma chemical vapor deposition (CVD). Both
heteroepitaxial and homoepitaxial growth methods have been examined, with emphasis

on producing large diamond crystals of high structural and chemical perfection.

By heteroepitaxy, epitaxial growth on a foreign substrate, diamond was grown on
(001) Ir thin-film epitaxial buffer layers deposited on (001) strontium titanate (SITIO3) by
DC magnetron sputtering. To nucleate diamond on Ir, the Ir surface was bombarded with
low energy ions extracted from the hydrocarbon plasma, a process known as DC biasing
or bias-enhanced nucleation. Since this critical process is poorly understood, attention
was paid to the spatial and temporal evolution of the Ir surface during the bias treatment.
It was discovered that the biased Ir surface is etched on a surprisingly short time scale
during which highly correlated nanopillars, 3-4 nm in height with mean separation 15
nm, emerge. The etching process is spatially non-uniform, propagating from substrate
center to substrate edge in minutes. Diamond grew on Ir without an intervening phase.
Lattice images revealed that interfacial strain from the 7% Ir-diamond lattice mismatch is
largely relieved by misfit dislocations within 1 nm of the interface. It is suggested that the
high nucleation density obtained with specific bias conditions is associated with the

roughened Ir surface.

To grow heteroepitaxial diamond as thick films, a two—step growth method was
explored. This process involved the transfer of a thin heteroepitaxial diamond film, still
attached to a substrate, to a second reactor where high growth rate conditions were
possible. Characterization of films grown by this approach showed that the resulting
diamond had much lower levels of internal strain, suggesting that the process could be

used to grow diamond crystals of structural quality similar to natural diamond.

In homoepitaxy, epitaxial growth on a substrate of the same material, diamond
was deposited by CVD directly onto high-pressure, high-temperature Type Ib diamond
substrates. Methods for removing substrate surface damage, as well as other
imperfections, were devised by use of plasma etching. The characteristics and statistics
of pits formed during etching were studied. Diamond growth methods were developed
with the aim of minimizing the formation of various structural and chemical defects. To
accomplish this, several growth parameters were varied, including substrate temperature,
feed gas concentration, growth rate, substrate surface, microwave power, and reactor
geometry. Regions of parameter space were found in which the diamond (001) surface
remained smooth during growth, and complete suppression of instabilities that create
hillocks or non-epitaxial crystallites was demonstrated. A great deal of information was
obtained by interrupting growth, removing the crystal from the reactor for optical

inspection, and then resuming the process with no apparent negative effects. Diamond
crystals were grown on 3 x 3 mm2 substrates with thicknesses greater than 0.5 mm. The

chemical purity of the crystals was such that it was impossible to observe signatures of

substitutional nitrogen at the ppm level.

In loving memory of my grandmother Krishna Kumari Regmi

iv

ACKNOWLEDGEMENTS

I would like to thank my advisor, Dr. Brage Golding, for the opportunity to work
on this project, and his boundless optimism that I could complete it. He taught me how to
think as a researcher. He was always there for suggestions and help when I had problems
of any kind.

I acknowledge Dr. R. J. Anderson (Sandia) for the exciton luminescence
characterization and Dr. D. L. Medlin (Sandia) for the HRTEM images.

I would also like to thank Dr. Mark Dykman, Dr. Stephen Zepf, Dr. Kirsten
Tollefson and Dr. Chong-Yu Ruan for serving on my guidance committee.

I would also like to thank Drs. Baokang Bi, Reza Loloee and Katherine Severin
for their help and technical suggestions.

Tom Palazzolo, Jim Muns, Tom Hudson, and the rest of the Physics and
Astronomy machine shop staff patiently answered my questions and made the parts I
needed in a timely manner. Their excellent work helped me to finish the project on time.

Ann Kirchmeier, Darla Conley Shawna Prater and Debbie Simmons helped with
orders and administrative items I wanted in a timely manner.

I would like to thank my senior Connie Bednarski-Meinke for her previous work
on the heteroepitaxial diamond growth. Her occasional visits and the suggestions gave
me the boost I needed. I also like to thank Ryan Kruse and Mahdokht Behravan for their
suggestions and encouragement.

Finally, I would like to thank my grandfather, my parents, my wife and kids, my

brother, my sister-in-law and other relatives for their support and encouragement.

TABLE OF CONTENTS

 

 

 

 

LIST OF TABLES _ IX
LIST OF FIGURES X
1 CHAPTER 1 INTRODUCTION 1
2 CHAPTER 2 STRUCTURE AND PROPERTIES OF DIAMOND .................... 4
2.1 CRYSTAL STRUCTURE .......................................................................................... 4
2.2 BAND STRUCTURE ................................................................................................ 5
2.3 HISTORICAL CLASSIFICATION OF DIAMOND TYPES: IMPURITIES ............................ 6
2.4 PHYSICAL PROPERTIES ......................................................................................... 7
2.4.1 Electronic Properties ................................................................................... 7
2.4.2 Optical properties ...................................................................................... 10
2.4.3 Optical properties of impurity-related defects .......................................... 12

3 CHAPTER 3 GROWTH OF DIAMOND BY CVD 15
3.1.1 Epitaxy and epitaxial growth .................................................................... 15
3.1.2 Nucleation ................................................................................................. 16
3.1.3 Crystal growth modes ............................................................................... 17
3.1.4 Substrate criteria for diamond heteroepitaxy ............................................ 18
3.1.5 Epitaxial diamond grth on different substrates .................................... 19
3.1.6 Heteroepitaxial diamond growth on iridium ............................................. 21

3 .2 BIAS-ENHANCED NUCLEATION (BEN) ............................................................... 23
3.2.1 BEN mechanisms ...................................................................................... 23
3.2.2 Biasing experiment with Ir ........................................................................ 26

3.3 HOMOEPITAXY ................................................................................................... 28
3.3.1 Background ............................................................................................... 28
3.3.2 Growth of defects ...................................................................................... 31
3.3.3 Effect of growth parameters ...................................................................... 34
3.3.4 Growth anisotropy: the alpha parameter ................................................... 39

4 CHAPTER 4 MATERIALS CHARACTERIZATION METHODS ................ 41
4.1 OPTICAL MICROSCOPY ...................................................................................... 41
4.2 ATOMIC FORCE MICROSCOPY ............................................................................ 43

vi

4.3 SCANNING ELECTRON MICROSCOPY .................................................................. 44

4.4 TRANSMISSION ELECTRON MICROSCOPY ........................................................... 46
4.5 RAMAN SCATTERING AND PHOTOLUMINESCENCE .............................................. 48
4.6 X-RAY DIFFRACTION .......................................................................................... 50
4.6.1 (I) - 20 or 0 - 20 scan .................................................................................. 54
4.6.2 0 or co scan or rocking curve ..................................................................... 54

 

CHAPTER 5 HETEROEPITAXIAL GROWTH OF DIAMOND .................... 55
5.1 SUBSTRATES ...................................................................................................... 55
5.1.1 Iridium ....................................................................................................... 55
5.1.2 Strontium titanate (SrTiO3) ....................................................................... 56
5.1.3 Sapphire (A1203) ....................................................................................... 56
5.1.4 SrTi03 preparation .................................................................................... 57
5.2 DC MAGNETRON SPUTTERING OF IRIDIUM ......................................................... 58
5.2.1 Sputtering system ...................................................................................... 59
5.2.2 Sputtering procedure ................................................................................. 62
5.2.3 Ir growth on SrTi03 .................................................................................. 64
5.3 CVD REACTOR PROCESS .................................................................................... 69
5.3.1 Substrate stage and bias set up .................................................................. 70
5.3.2 Substrate holder ........................................................................................ 72
5.4 DIAMOND NUCLEATION ..................................................................................... 73
5.4.1 The bias process ........................................................................................ 73
5.4.2 Discussion ................................................................................................. 91
5.4.3 Summary ................................................................................................... 95
5.5 DIAMOND THIN FILM GROWTH ........................................................................... 95
5.6 DIAMOND THICK FILM GROWTH ......................................................................... 99
5.6.1 Raman spectroscopy ............................................................................... 103
5.6.2 Free exciton luminescence ...................................................................... 105
5.7 SUMMARY ........................................................................................................ 107
CHAPTER 6 HOMOEPITAXIAL GROWTH OF DIAMOND 108
6.1 EXPERIMENTAL METHODS ................................................................................ 109
6.1.1 AsTeX diamond CVD reactor ................................................................ 109
6.1.2 Operational procedures ........................................................................... 116
6.2 GROWTH PROCESS ........................................................................................... 117
6.2.1 Diamond substrate preparation ............................................................... 118
6.2.2 CVD homoepitaxial diamond grth process ........................................ 127
6.2.3 Long-term growth procedures ................................................................. 133
6.2.4 Nitrogen in CVD diamond ...................................................................... 138
6.3 HOMOEPITAXIAL DIAMOND CHARACTERIZATION ............................................. 147
6.3.1 High resolution x-ray diffraction ............................................................ 147
6.3.2 Raman spectroscopy ............................................................................... 149

vii

7

8

9

10

11

6.3.3 Fourier transform infrared spectrophotometry (FTIR) ........................... 151

 

 

 

 

6.4 DISCUSSION ..................................................................................................... 153
CHAPTER 7 DISCUSSION AND CONCLUSIONS 155
7.1 SUGGESTIONS FOR FUTURE WORK .................................................................... 156
APPENDIX A 159
8.1 SAMPLE INFORMATION ..................................................................................... 159
8. 1 . l Heteroepitaxial sample information ........................................................ 1 5 9
8.1.2 Homoepitaxal sample information .......................................................... 159
APPENDIX B ...... 160
9.1 PYROMETER CALIBRATION ............................................................................... 160
APPENDIX C 165

10.1 IR GROWTH ON AL203 ...................................................................................... 165
10.1.1 A1203 preparation .................................................................................... 165
10.1.2 11' growth on A1203 .................................................................................. 166
REFERENCES 170

 

viii

LIST OF TABLES

TABLE 2.1.DIAMOND CLASSIFICATION ACCORDING TO THE AMOUNT OF IMPURITY ............. 6
TABLE 2.2 ROOM TEMPERATURE CARRIER MOBILITIES OF DIAMOND ................................... 9

TABLE 3.1 Two GROWTH REGIMES IN MICROWAVE PLASMA CVD DIAMOND GROWTH ...... 31

TABLE 5.1 GROWTH CONDITIONS FOR 3.5 MM DIA DIAMOND FILM ..................................... 96
TABLE 5.2 GROWTH CONDITIONS FOR 5 x 5 MM2 DIAMOND FILM ....................................... 97
TABLE 5.3 GROWTH CONDITIONS FOR 8 MM DIA DIAMOND ON 10 x 10 MM2 IR/STO ......... 98
TABLE 5.4 HIGH GROWTH CONDITIONS OF HETEROEPITAXIAL DIAMOND FILM. ................ 100

TABLE 6.1 GROWTH CONDITIONS FOR 4 SAMPLES GROWN AT DIFFERENT TEMPERATURE 127

TABLE 6.2 GROWTH CONDITIONS FOR DIAMOND GROWN AT DIFFERENT R. ...................... 130
TABLE 6.3 GROWTH CONDITIONS FOR TWO DIAMOND SAMPLES. ..................................... 130
TABLE 6.4 GROWTH CONDITIONS FOR SAMPLES A6 AND A7 ............................................. 132
TABLE 6.5 GROWTH CONDITIONS FOR A 515 pM THICK DIAMOND .................................... 135
TABLE6.6 GROWTH CONDITIONS FOR SAMPLE A9 ............................................................ 137

TABLE 6.7 PL INTENSITIES FROM (N -V) CENTERS IN CVD DIAMOND. ............................. 144

ix

LIST OF FIGURES

FIGURE 2.1 CRYSTAL STRUCTURE OF DIAMOND ................................................................... 4
FIGURE 2.2 ELECTRONIC BAND STRUCTURE OF DIAMOND ................................................... 5
FIGURE 2.3 ELECTRICAL CONDUCTIVITY vs. RECIPROCAL TEMPERATURE FOR DIAMOND 8
FIGURE 2.4 TEMPERATURE DEPENDENCE OF ELECTRON MOBILITY IN CVD DIAMOND ........ 9
FIGURE 2.5 ABSORPTION SPECTRUM OF A HIGHLY PURE TYPE IIA NATURAL DIAMOND ..... 1 1

FIGURE 2.6 INFRARED ABSORPTION SPECTRUM OF A HIGHLY PURE NATURAL DIAMOND . .. 12

FIGURE 3.1 CRYSTAL GROWTH MODES AS A FUNCTION OF SURFACE COVERAGE ................ 18
FIGURE 3.2 MAP OF SURFACE MORPHOLOGIES AS A FUNCTION OF MISCUT ANGLE .......... 30
FIGURE 3.3 PYRAMIDAL HILLOCKS WITH NON-EPITAXIAL CRYSTALLITES .......................... 32
FIGURE 3.4 MECHANISM OF TWO DIMENSIONAL NUCLEATION ON A WIDE TERRACE. ......... 32
FIGURE 3.5 SCHEMATIC VIEW OF STEP FLOW GROWTH ....................................................... 33
FIGURE 3.6 MACRO-STEP BUNCHING IN HOMOEPITAXIAL DIAMOND .................................. 34
FIGURE 3.7 DIAMOND GROWTH RATE AS A FUNCTION OF SUBSTRATE TEMPERATURE ........ 35

FIGURE 3.8 GROWTH RATE AS A FUNCTION OF METHANE CONCENTRATION AT 850 °C . ..... 36
FIGURE 3.9 SHAPE OF DIAMOND CRYSTALS FOR DIFFERENT VALUES OF ALPHA ................. 40

FIGURE 4.1 SCHEMATIC OF A DIFFERENTIAL INTERFERENCE CONTRAST MICROSCOPE ....... 41

FIGURE 4.2 THE OLYMPUS BX—60 DIC OPTICAL MICROSCOPE .......................................... 42
FIGURE 4.3 SCHEMATIC OF ATOMIC FORCE MICROSCOPE ................................................... 44
FIGURE 4.4 SCHEMATIC OF A SCANNING ELECTRON MICROSCOPE ...................................... 45
FIGURE 4.5 SCHEMATIC OF A TRANSMISSION ELECTRON MICROSCOPE. .............................. 47
FIGURE 4.6 SCHEMATIC OF A RAMAN SPECTROMETER . ..................................................... 49

FIGURE 4.7 SCHEMATIC OF x-RAY DIFFRACTION FROM A CRYSTAL .................................... 50

FIGURE 4.8 SCHEMATIC SHOWING THE BRAGG"S LAW OF DIFFRACTION ............................ 51
FIGURE 4.9 EWALD SPHERE CONSTRUCTION ...................................................................... 52
FIGURE 4.10 SCHEMATIC OF BRUKER AXS x-RAY DIFFRACTROMETER ............................. 5 3
FIGURE 5.1 AFM IMAGE SHOWING THE TOPOGRAPHY OF THE SRTIO3 SUBSTRATE. ........... 58
FIGURE 5.2 SCHEMATIC OF THE THREE-GUN APX SPUTTERING SYSTEM ............................ 60
FIGURE 5.3 SCHEMATIC OF THE SAMPLE STAGE OF SPUTTERING SYSTEM. .......................... 62
FIGURE 5.4 TOPOGRAPHY OF 500 NM THICK IR FILMS ON SRTIO3 ...................................... 66
FIGURE 5.5 LINEWIDTH AND MEAN ROUGHNESS OF IR VS. TEMPERATURE .................. 66
FIGURE 5.6 X-RAY SCAN OF A 600 NM THICK IR FILM. ........................................................ 68
FIGURE 5 .7 X-RAY (1 1 l) POLE FIGURE OF IR AND SRTIO3. ................................................ 69
FIGURE 5.8.GENERAL SCHEMATIC OF THE MICROWAVE PLASMA CVD REACTOR. ............. 70

FIGURE 5.9 SCHEMATIC OF THE SAMPLE STAGE AND BIAS SET UP FOR DIAMOND GROWTH. 71

FIGURE 5.10 ELECTRICAL CIRCUIT USED FOR BIASING. ...................................................... 72
FIGURE 5.1 l SCHEMATIC OF SAMPLE POST AND CAP WITH SAMPLE STAGE ......................... 73
FIGURE 5.12 TIME PLOT OF THE PROCESS SEQUENCE ......................................................... 74
FIGURE 5.13 AFM IMAGES OF THE IR SURFACE BIASED FOR DIFFERENT DURATIONS. ........ 76
FIGURE 5.14 SEM IMAGES OF THE IR SURFACE BIASED FOR DIFFERENT DURATIONS .......... 78
FIGURE 5.15 HRTEM IMAGES OF THE IR SURFACE AFTER BIASING .................................... 81

FIGURE 5.16 SEM IMAGES OF SAMPLES WITH DIFFERENT TB FOLLOWED BY TG = 20 MIN.... 82

FIGURE 5.17 PLOT SHOWING % AREA COVERED BY EPITAXIAL DIAMOND. ......................... 84

FIGURE 5.18 SEM IMAGES OF 60 MIN BIASED SAMPLES FOLLOWED BY SHORT GROWTH. 86

FIGURE 5.19 HRTEM IMAGE OF IR SURFACE FOR A TB = 60 MIN, TG = 0 MIN SAMPLE ......... 87

xi

FIGURE 5.20 CROSS SECTION HRTEM IMAGES FOR A TB =60 MIN, TG = 90 S SAMPLE ........ 89

FIGURE 5.21 HRTEM IMAGE FOR A TB = 60 MIN TG = 150 s SAMPLE ................................. 90
FIGURE 5.22 HRTEM IMAGE OF A TE = 10 MIN AND TG = 20 MIN SAMPLE. ......................... 91
FIGURE 5.23 DIC MICROGRAPH AND SEM IMAGE OF 3.5 MM DIA DIAMOND FILM ............. 96
FIGURE 5.24 DIC MICROGRAPH AND SEM IMAGE OF A 5 x 5 MM2 DIAMOND FILM. ........... 98
FIGURE 5.25 DIC MICROGRAPH AND SEM IMAGE OF A 8 MM DIA DIAMOND FILM ............. 99

FIGURE 5.26 HETEROEPITAXIAL DIAMOND FILM BEFORE AND AFTER PLASMA ETCHING. .. 100
FIGURE 5.27 DIC OPTICAL MICROGRAPH OF A DIAMOND FILM AFTER 3 H GROWTH ......... 101

FIGURE 5.28 DIC MICROGRAPH OF A 6511M THICK HETEROEPITAXIAL DIAMOND ON IR. 101

FIGURE 5.29 RAMAN SPECTRUM OF HETEROEPITAXIAL DIAMOND ................................... 104
FIGURE 5.30 RAMAN SPECTRA OF TWO HETEROEPITAXIAL DIAMOND CRYSTALS. ............ 105
FIGURE 5.31 COMPARISION OF FREE EXCITON LUMINESCENCE OF DIAMOND. .................. 106

FIGURE 6.1 SCHEMATIC OF CVD REACTOR FOR HOMOEPITAXIAL DIAMOND GROWTH ..... 11 1

FIGURE 6.2 SCHEMATIC OF COOLING STAGE AND SUBSTRATE HOLDER ............................ 114
FIGURE 6.3 RGA SPECTRUM TAKEN DURING DIAMOND GROWTH AT 150 TORR. .............. l 16
FIGURE 6.4 OPTICAL PHOTOGRAPH OF A HPHT TYPE IB DIAMOND SUBSTRATE. ............. 1 18
FIGURE 6.5 DIC MICROGRAPH OF A PLASMA-ETCHED DIAMOND SUBSTRATE. .................. l 19
FIGURE 6.6 POLISHING SCRATCHES ON THE SURFACE OF DIAMOND SUBSTRATE ............... 120
FIGURE 6.7 DIAMOND SURFACE AFTER PLASMA ETCHING AT R = 2% P = 150 TORR ......... 122
FIGURE 6.8 DIAMOND SURFACE AFTER PLASMA ETCHING AT R = 2 % P = 100 TORR ........ 123

FIGURE 6.9 DIC IMAGES OF DIAMOND AFTER PLASMA ETCHING FOR DIFFERENT TIME ..... 124

FIGURE 6.10 STATISTICS OF ETCH PITs. ............................................................................ 125

FIGURE 6.1 1 DIAMOND SURFACE AFTER PLASMA ETCHING SHOWING INCLUSIONS ........... 126

xii

FIGURE 6.12 DIC MICROGRAPHS OF DIAMOND SURFACES GROWN AT DIFFERENT TG. ...... 128

FIGURE 6.13 EFFECT OF TEMPERATURE ON GROWTH RATE .............................................. 129
FIGURE 6.14 DIC IMAGES OF A DIAMOND SAMPLE GROWTH AT DIFFERENT R .................. 130
FIGURE 6.15 DIC MICROGRAPHS OF DIAMOND SURFACES GROWN AT DIFFERENT R ......... 131
FIGURE 6.16 GROWTH RATE VS. METHANE-HYDROGEN RATIO R. ..................................... 131
FIGURE 6.17 SEM IMAGE OF A 290 uM THICK DIAMOND CRYSTAL ON SUBSTRATE .......... 132
FIGURE 6.18 OPTICAL PHOTOGRAPH OF A 290 uM THICK FILM ON ITS SUBSTRATE ........... 133

FIGURE 6.19 DIC MICROGRAPHS OF CRYSTAL A8 AFTER DIFFERENT GROWTH PERIODS 134

FIGURE 6.20 HEIGHT PROFILE OF DIAMOND CRYSTAL A3 ALONG ITS SURFACE DIAGONAL135

FIGURE 6.21 OPTICAL PHOTOGRAPHS OF A 515 11M THICK DIAMOND CRYSTAL. ............... 136
FIGURE 6.22 OPTICAL PHOTOGRAPH OF A 550 uM THICK CRYSTAL .................................. 137
FIGURE 6.23 RAMAN SPECTRA OF TYPE IB DIAMOND SUBSTRATES .................................. 139
FIGURE 6.24 FTIR SPECTRA OF DIAMOND (A) NATURAL TYPE IIA (B) TYPE IB. ............... 140
FIGURE 6.25 X-RAY (004) ROCKING CURVE LINEWIDTH OF A DIAMOND SUBSTRATE ....... 141
FIGURE 6.26 RAMAN SPECTRUM TAKEN FROM CVD HOMOEPITAXIAL DIAMOND ............. 142
FIGURE 6.27 RAMAN SPECTRA OF CVD DIAMOND CRYSTALS ......................................... 143
FIGURE 6.28 GRAPH SHOWING THREE DIFFERENT REGIONS IN TG—R SPACE. ..................... 144
FIGURE 6.29 GROWTH RATE vs. NITROGEN CONCENTRATION (RELATIVE TO H2). ............ 145

FIGURE 6.30 ROOM TEMPERATURE RAMAN SPECTRA FOR DIAMOND GROWN WITH N2 146

FIGURE 6.31 .(N-V) PL PEAK HEIGHTS, AFTER BACKGROUND SUBTRACTION ................... 147
FIGURE 6.32 X-RAY ROCKING CURVE FOR THE (004) REFLECTION OF DIAMOND .............. 148
FIGURE 6.33 X-RAY (11 1) POLE FIGURE FROM HOMOEPITAXIAL CVD DIAMOND ............. 149

FIGURE 6.34 RAMAN SPECTRUM OF CVD HOMOEPITAXIAL DIAMOND SAMPLE ................ 150

xiii

FIGURE 6.35 RAMAN SPECTRA SHOWING RAMAN LINEWIDTH FOR DIAMOND ................... 151

FIGURE 6.36 FTIR SPECTRA OF DIAMOND CRYSTALS ....................................................... 152
FIGURE 6.37 MAGNIFIED FTIR ABSORPTION SPECTRA OF DIAMOND CRYSTALS ............... 153
FIGURE 9.1 PYROMETER TEMPERATURE vs R TYPE TC TEMPERATURE ........................... 161
FIGURE 9.2 PYROMETER AND TC TEMPERATURE WITH 2 KINDS OF QUARTZ PLATE .......... 162
FIGURE 9.3 PYROMETER TEMPERATURE WITH AND WITHOUT QUARTZ PLATE .................. 163
FIGURE 9.4 TWO-COLOR VS. SINGLE COLOR PYROMETER TEMPERATURE ......................... 164
FIGURE 10.1 TOPOGRAPHY OF VICINAL (1 120) AL203 SURFACE AFTER ANNEALING. ...... 166

FIGURE 10.2 TOPOGRAPHY OF 500 NM THICK (001) IR SURFACE GROWN ON A-AL203 167

FIGURE 10.3 PLOT OF (002) IR LINEWIDTH 1(1 1 1)/I(002) FOR DIFFERENT T ..................... 167
FIGURE 10.4 (002) IR LINE WIDTH AND 1(1 1 1)/I(002) VS. FILM THICKNESS. ..................... 168
FIGURE 10.5 (002) IR LINEWIDTH AND 1(1 1 1)/I(002) vs. AR PRESSURE. .......................... 169

Images in this thesis are presented in color.

xiv

1 Chapter1 Introduction

Diamond is a wide bandgap semiconductor with an unusual combination of
extreme physical properties. Many are well-known, such as unequalled mechanical
hardness, a thermal conductivity greater than copper at room temperature, negative
electron affinity, and high transparency in the visible and near-ir spectral region. Others,
such as high carrier mobilities, dielectric breakdown field, and resistance to radiation
damage, make diamond attractive for applications as energetic particle detectors. In the
future, detectors will be required for environments with orders of magnitude more
radiation intensity [1], leading to severely reduced time before replacement of the silicon
detectors presently in widespread use. Diamond can be a superior alternative, provided it
can be produced as high-purity, large-area Single crystals. Natural diamond cannot satisfy
these needs as its quality is highly variable and suitable material is far too expensive.
CVD polycrystalline diamond is generally unsuitable for high-performance detectors as
intn'nsic electronic properties are degraded by the presence of grain boundaries
containing graphitic carbon, dislocation networks, and impurity complexes, all of which
act as carrier traps [2]. A solution to these problems is provided by pure single crystalline
diamond grown by epitaxial CVD. A recent report that CVD homoepitaxial diamond has
electron and hole mobilities much higher than that of the best natural diamond [3] has
indicated that high-purity diamond offers improved performance over natural diamond,
as well as Si. However, it remains unclear Whether diamond can be grown by

homoepitaxy over sufficiently large areas to satisfy needs in many detector applications.

Heteroepitaxy may provide the means for producing diamond on the wafer scale.
The first efforts in CVD heteroepitaxial diamond grth began with the use of Si as a

l

substrate. However, the large lattice mismatch did not allow true epitaxy, only highly-
oriented polycrystalline growth was possible on Si [4,5]. With discovery of thin-film Ir as
a suitable buffer layer, grown on MgO, diamond heteroepitaxy became more practical
[6-8]. With SrTiO3 as a substrate for Ir, it was shown that the mosaic spread of
heteroepitaxial diamond could be reduced to less than 0.2 deg [9,10]. Discovery of a-

plane sapphire (A1203) as a substrate for (001) Ir led to predictions of wafer-scale size

diamond, as sapphire is available as highly perfect crystals up to 25 cm in diameter [1 1].

In a parallel development, a major advance occurred with the discovery of a
region of CVD parameter space in which homoepitaxial diamond could be grown at very
high rates [12,13]. Although, the process required the addition of nitrogen, at major
electronic impurity in diamond, it increased the possibilities for thick free-standing
diamond crystals. Subsequently, growth in the absence of nitrogen led to higher-purity
free—standing diamond crystals, albeit at lower rates. Thus, there are prospects that
homoepitaxy may yet lead to sufficiently large, detector grade diamond materials

[3,14,15].

This dissertation describes a broad investigation of heteroepitaxial and
homoepitaxial diamond growth methods. A great deal of attention has been paid to the

role of substrates, their growth, characterization, and preparation prior to diamond CVD,
including a Sputtering process for (001) Ir epitaxy on SrTi03 and A1203. The nucleation

of diamond on Ir, intimately connected to the DC bias process, has been examined, with
investigation of its temporal and spatial evolution. The early growth and coalescence of
diamond crystallites on Ir were also studied. By modification of the plasma-sample

environment, growth processes were developed that have led to 8 mm dia heteroepitaxial

2

diamond films. With a two-step process, such films were grown to thick films with

quality approaching that of natural diamond.

In homoepitaxial diamond growth, the role of several process parameters on

structural and chemical quality, and growth rate, was investigated. A process was
developed that led to 4 x 4 x 0.5 mm3 free-standing homoepitaxial diamond crystals with

low defect levels.

The dissertation contains seven chapters. Following this Introduction, Chapter 2
covers the basic properties of diamond, its historical classification schemes, and the role
of impurities. Chapter 3 provides a review and background for heteroepitaxial and
homoepitaxial diamond growth. The role of DC bias in enhancing diamond nucleation,
and associated physical models of the process, are presented. Homoepitaxial growth and
the influence of plasma CVD parameters are then discussed. Chapter 4 describes several
analytical tools used to measure the physical properties of CVD diamond. Chapter 5
presents the details of the present CVD heteroepitaxy research: Ir epitaxy, DC biasing,
and the two-step process. Chapter 6 describes homoepitaxial diamond growth: substrate
preparation, growth parameters, the role of nitrogen and subsequent nitrogen-related

electronic centers. Chapter 7 summarizes the major results and makes suggestions for

further work.

2 Chapter 2 Structure and properties of diamond

2.1 Crystal structure

Diamond has the face centered cubic (FCC) structure with two lattice points in the
primitive unit cell, one at (000) and the other at (V: M: ‘A). It can be viewed as the
superposition of two FCC lattices, one displaced relative to the other along the body
diagonal by one quarter of the lattice parameter from the origin. Its cubic lattice
parameter at room temperature is 0.357 nm. Every carbon atom in the lattice is bonded
with four other carbon atoms to form a tetrahedral structure. Four valence electrons in
each carbon atom form strong covalent bonds by sp3 hybridization with nearest-neighbor

bond length 0.154 nm. Figure 2.1 shows the unit cell of diamond. It is associated with

(Fd3 m) or 01.7 space group, which has a total of 48 symmetry operations.

   

0.154 nm 0.254 nm

Figure 2.1 Crystal structure of diamond

There are eight carbon atoms in the cubic unit cell with an effective packing

fraction of about 34 %. Diamond has the shortest bond length of any three-dimensional

crystal and a binding energy of 347 KJ/mol [16], which makes it the hardest material

known.

2.2 Band structure

Diamond and Si have quite similar electronic band structures. The energy
difference between an electron in the lowest level of the conduction band and the highest
level of the valence band, the electronic bandgap, is 5.5eV for diamond, almost five
times that of Si (1.12eV) [16]. The band structure of diamond, calculated using the

linearized augmented plane wave method (LAPW) [17], is shown in Figure 2.2.

 

25 #V w

 

 

 

 

 

 

 

 

 

 

Figure 2.2 Electronic band structure of diamond [17]

Conduction band minima occur along <100> direction at nearly 0.7 of the
distance to the edge of the Brillouin zone at X. The valence band maximum occurs at the

center of the Brilliouin zone, F. Thus, the minimum bandgap in diamond is indirect.

2.3 Historical classification of diamond types: impurities

Natural diamond has been classified historically by the nature and concentration
of impurities. Almost all diamond, natural or synthetic, always contains as impurities
small atoms such as nitrogen or boron. There are several classes of impurities: (1)
substitutional impurities in which a foreign atom replaces a carbon atom with minimal
distortion of the lattice, (2) interstitial impurities, (3) impurities associated with lattice
vacancies and (4) inclusions, in which aggregates of elements such as Al, Ni, and Mg are
non-crystallographically inside crystals. Table 2.1 shows historical classification of
diamond types, based primarily on included impurities. For example, Type Ib diamond is
defined on the basis of its substitutional N concentrations, which imparts a yellow color
to natural stones. Diamond grown by the high-pressure, high-temperature (HPHT)
method, not only contains N from its growth in air, but also may contain Al, Ni or Fe
impurities, since they are used as catalysts in the growth process. However, the very short
C-C bond in diamond does not allow many elements to be incorporated substitutionally.

The major impurity centers involve N, B, P, S, Si and Ni.

Table 2.1.Diamond classification according to the amount of lattice impurity [18]

 

 

 

 

 

 

 

Diamond Type Nitrogen Boron (ppm) Color
(ppm)

Ia 2000 - Clear to yellow

lb 102 — 103 - Green, brown, yellow

lb 1-100 - Yellow

Ila ~l - Colorless clear

IIb ~l ~100 Blue

 

 

 

 

 

2.4 Physical Properties

2.4.1 Electronic Properties

Because of the large bandgap, pure diamond is highly resistive at room
temperature but its resistivity decreases at higher temperatures. Common dopants are B,
N and P. B-doped diamond has p-type conductivity with an acceptor level that lies 0.36
eV above the valence band [19]. Heavily B-doped diamond was recently found to exhibit
type-II superconductivtiy below 4 K [20]. N and P-doped diamond have n-type
conductivity with donor levels that lie 1.7 and 0.6 eV below the conduction band
minimum, respectively. Although some studies have shown that S provides n-type
conductivity with an activation energy of 0.37 eV [21], more recent work has suggested
that these results are spurious and result from other dopants [22]. Recent conductivity
measurements on heteroepitaxially-grown single crystal diamond have shown that the
conductivity and carrier activation energy are comparable to natural Type IIa diamond

[23]. These data are shown in Figure 2.3.

1E-7

g ' I I T I l . .
5 ‘I. E
1E‘8 '5 O _.:1
: . q
d h d
—,- 15'9 a 'u. a
E 1 O 4
9 . I
is: 15-10 1: ‘ a
'5 i ‘. E
'4: . .
0

.g 1E-11 1 o ‘. 1::
fl 3 - :
O . - a
U ‘ ‘ -
. J .
15-12 1 o - -,

1E‘13 1 I I I V I 1

0.0010 0.0012 0.0014 0.0016 0.0018
-1 1
T (K' )

Figure 2.3 Electrical conductivity vs. reciprocal temperature for heteroepitaxial diamond
(u),natural Type Ila diamond (o) [23].

Chemical vapor deposition (CVD) diamond is grown in the presence of hydrogen
so that its surface is naturally hydrogen-terminated. Such diamond exhibits p-type surface
conductivity. It can be destroyed by oxygen-termination, which can be accomplished by

immersing the diamond in a strong acid or by short exposure to an oxygen plasma.

Electrons and holes in diamond have high mobilities with respect to other

semiconductors. Table 2.2 [16] compares several electronic properties of common

 

 

 

 

semiconductors with diamond.

0.0020

Table 2.2 Room temperature carrier mobilities of diamond and other semiconductors [3]

 

 

 

 

 

 

 

 

 

 

 

 

 

Semiconductor Electron Hole Breakdown Bandgap
mobility mobility electric field (eV)
( cmzN-S) (cmzN-s) (V/cm)
Natural diamond Ila 2000-2800 1800-2100 107 5.5
Homoepitaxial diamond 4500 3800 107 5.5
Silicon carbide 700 3x106 3.26
Gallium nitride 2000 2.0x106 3.0
Gallium arsenide 8500 4.0on3 1.42
Silicon 1500 450 3.7x105 1.12
Germanium 3900 1900 2.0x105 0.66

 

 

Note that the room temperature mobilities for homoepitaxial CVD diamond [3]
are surpassed only by values for GaAs. At high temperatures, semiconductor carrier
mobility decreases because of increased phonon scattering. Figure 2.4 shows the
temperature dependence of the electron mobility in un-doped CVD diamond. The
behavior in the higher temperature region has been attributed to scattering by optical

phonons whereas at lower temperatures acoustic phonon scattering is implicated [3].

 

 

 

Mobility (cmst)

 

 

 

 

 

 

 

 

 

300 400 500

T (K)
Figure 2.4: Temperature dependence of electron mobility in homoepitaxial CVD

diamond [3]

Other important room temperature electronic properties of diamond are its high

dielectric breakdown field (107 V-cm'l), high carrier saturation velocity (107 cm-s'l), and

high thermal conductivity (20 W-cm'l-KJ), almost five times greater than for Cu. Many

of these electronic properties position diamond as a favorable material for applications in

high power, high temperature electronics.

2.4.2 Optical properties

To the eye, pure diamond is clear and colorless. Type IIa diamond, with its low
impurity content, is highly transparent in the spectral region from 225 to 2500 nm [24].
Impurities give rise to absorption bands and the coloration given in Table 2.1. Figure 2.5
shows the intrinsic absorption spectrum of very pure diamond in the visible and uv
region. The rapid increase in absorption in the ultraviolet region just below 225 nm,
represents the onset of inter-band transitions just above 5.5 eV. Because of the indirect
bandgap, band-to-band luminescence is largely quenched since high-energy phonons

must participate to conserve momentum.

10

 

15 F'TIUVY'IVY-VTIUYTIfI'iTVr—vr

10[— a

 

 

 

 

'a
3
5
5 _ '1
0 [5% I - - - A 1+Lk I A L4 1 _2
200 400 600 800
Mnm)

Figure 2.5 Absorption spectrum of a highly pure Type IIa natural diamond [24]

Pure diamond exhibits considerable absorption in the infrared region between 0.2
and 0.5 eV as shown in Figure 2.6. This absorption arises fiom one, two and three-
phonon processes. The lowest ir-active optical vibrational frequency corresponds to 0.165
eV so this sets a cut-off for single-phonon absorption. The absorption at the highest
energies in figure 2.6 is due to the involvement of three phonons with a cut-off around

0.495 eV [24].

ll

 

15 V V f] V—‘ T—fij—T I 1f1 I I—T T—V V [1 fit

10

lfifI Y—' TW‘V
J
j—

a (cm'l)

I I I Ti l

 

 

 

0 41+l1 gl_A_l 1 I41 I+I_Ll £41 J—L I“

4000 3200 2400 2000 1600 1200 800 400

Wavenumber (cm‘l)

Figure 2.6 Infrared absorption spectrum of a highly pure Type IIa natural diamond [24].

2.4.3 Optical properties of impurity-related defects

Nitrogen is the most common impurity in natural, HPHT and CVD diamond. It
appears in many different configurations. Nitrogen in pair-wise aggregated form,
sometimes called an A center appear, when two nitrogen atoms replace adjacent carbon

atoms. It is found in Type Ia(A) diamond. Its infrared absorption signature consists of a

major peak at 1282 cm.1 and another at 1212 cm-1. In the uv region, a sharp absorption

peak appears at 320 nm (4 eV) [25]. When four nitrogen atoms replace four carbon atoms
from the lattice sites surrounding a vacancy, then a B optical center is formed. This is

found in Type Ia(B) diamond. It gives rise to broad absorption in the ir centered at 1185

cm'1 [26]. Three substitutional nitrogen atoms and a vacancy give rise to the N3 optical

center, with a sharp absorption band at 415 nm in the near-uv [24]. N-based platelets

have been identified in Type Ia(B) diamonds, consisting of the interstitials ejected from B

12

type aggregates, with absorption at 1365 cm.1 in the ir [27]. Nitrogen in interstitial form,
sometimes referred as Type Ia(H) diamond, shows a peak in the ir at 1450 cm'1 [28].

Nitrogen-vacancy complexes are another class of impurity centers found in

natural as well as CVD diamond. When a substitutional nitrogen atom is adjacent to a
. . 0 . . .
vacancy, It forms a nitrogen-vacancy center, or (N-V) , In Its neutral charge state. It grves

rise to a 2.156 eV (575 nm) zero phonon line (ZPL) luminescence. It can trap an electron

to form a (N-V)' center, with a ZPL at 1.945 eV (637 nm) [29].

Impurity-related defect populations are not independent of one another. For

example, if Type Ia(A) diamond is irradiated and annealed above 650 °C, an A aggregate

captures a vacancy to form an H3 optical center with a ZPL at 2.463 eV. B aggregates

irradiated and annealed under the same conditions may capture a vacancy to form the H4

optical center with a ZPL at 2.498 eV [26].

Type Ib diamond, by definition, contains predominantly dispersed, singly
substitutional nitrogen. This simple impurity is sometimes referred to as a C center. It

acts as a deep donor with activation energy 1.7 eV and gives rise to ir absorption at 1130

cm'1 and 1344 cm]. It also contributes to a rapid rise in absorption just below 2 eV
(600nm). An ionized nitrogen donor NI gives rise to an ir absorption line at 1332 cm-1
[26].

Boron in Type ITb diamond acts as an acceptor, with a series of hydrogenic levels

starting at 2460 cm.1 (0.304 eV) with other lines at 0.347, 0.349, and 0.362 eV [30].

13

Transitions to the valence band extend into the red, giving rise to a characteristic blue

color.

Silicon- and hydrogen-related defects have also been identified in CVD diamond,
with Si originating from glass window contamination and H from the feed gas. A Si-

vacancy ZPL iS observed at 1.681 eV (738 nm) [26]. C-H vibrational modes in the region
from 2750 to 3100 cm'I have been seen in polycrystalline CVD material [31] and

attributed to H-termination at surfaces or hydrogen bound at grain boundaries.

l4

3 Chapter 3 Growth of Diamond by CVD

In this chapter we review the basic principles of nucleation and growth which
apply to the deposition of diamond by the chemical vapor deposition (CVD) process.
CVD is a generic method whereby chemical reactions are induced at relatively low
pressures by energy supplied to reactive gases by heating, often by the creation of a
plasma discharge. The products of the reaction are collected on a substrate located in
close proximity to the focus of the chemical reactions. The process is continuous in that,
as reagents are depleted, they are exhausted and replenished by allowing for continuous
output and input flows. After chemical constituents are deposited, the formation of solid
structures can be further manipulated by diffusion rate control via substrate temperature.
Diamond growth by low-pressure CVD is particularly significant as it represents a
method for the formation of a metastable crystallographic phase. Since graphitic carbon
represents the ground state of carbon under typical terrestrial temperatures and pressures,

formation of diamond by CVD relies on the presence of molecular hydrogen, which alters

chemical kinetics to favor diamond (Sp3 bonded carbon) formation rather than that of

graphite (sp2 bonded carbon).

Before reviewing previous research on diamond growth by heteroepitaxy and

homoepitaxy, we first discuss general aspects of epitaxy and nucleation and growth.

3.1.1 Epitaxy and epitaxial growth

The ordered growth of a crystalline material on another crystalline substrate is

called epitaxial growth if at least one crystal plane of the deposited crystal and one of the

15

substrate are parallel. For example, growth of (001) diamond on (001) Ir such that a

<110> direction of both crystals are parallel to each other would satisfy this definition.

In heteroepitaxy, given similar crystallographic structures, the lattice parameter of

the deposited crystal, ad, and that of the substrate crystal, as, differ. We define the lattice

mismatch f by f = (a, — ad)/ad. If the deposited and the substrate crystal have different

crystal structures, then the mismatch in atomic spacing on the contacting substrate plane
and the deposited crystal plane is more relevant than the lattice parameter mismatch. For
epitaxial growth, small f is generally desirable so that better atom-to-atom registration
occurs. Systems with large f will have large strains induced at the interface (provided that
f is not an integer or a small rational fraction). Interfacial strain can be accommodated by
the formation of defects, in particular, by the formation of dislocations. The density of

such misfit dislocations is generally proportional to the degree of mismatch.

3.1.2 Nucleation

Nucleation is the process by which a substance undergoes a phase transition by
first creating small domains of the new phase within the parent phase. A first order phase
transition occurs when thermodynamic fluctuations initiate the continuous growth of the
new phase within a small temperature or pressure range. It is believed that fluctuations of
a thermodynamic order parameter must exceed a critical value before the phase transition
occurs. For example, the Spatial region over which the new phase forms must exceed a
certain value, the size of the critical nucleus. Once the fluctuation exceeds this value,
energy considerations will favor grth of the new phase fiom the nucleus, and the

volume of the new phase grows. However, additional kinetic considerations may

16

influence the growth rate of the new phase. Nonetheless, nucleation is the essential

beginning process that precedes crystal growth.

3.1.3 Crystal growth modes

It is generally accepted that there are three major modes possible in epitaxial
growth, when there is no interdiffusion between the substrate and the deposited phase
[32,33]. In three-dimensional island growth, or Volmer-Weber growth, small three-
dimensional clusters form directly on top of the bare substrate and then grow into islands.
This is most likely when the atoms of the deposited crystal are more strongly bound to

each other than to the substrate atoms, i.e., the surface energy of the deposited crystal is

much larger than that of the substrate crystal, yd + 71 > 75, where yd, 7i and y, are,

respectively, the surface energy of the deposited crystal, the interfacial energy, and the

substrate surface energy [32].

A second possible mode is two-dimensional layer-by-layer or Frank-van der
Merwe growth. Here, the deposited crystal wets the substrate surface, which means that
the deposited crystal atoms bind more strongly to the substrate atoms than to themselves.

At first, atoms that condense on the substrate form a complete monolayer then, the next

successive layers cover the preceding ones. In this case one has the relation, 7.] +Yi < 75.

The third common mode, layer plus island growth, or Stranski-Krastanov growth,
is a combination of the preceding two modes. Here, the grth of the first few
monolayers proceeds layer-by-layer, but eventually this mode becomes energetically
unfavorable and the mode changes to island growth. Although there may be several

reasons for this to occur, the most probable reason arises from an increase of the

17

interfacial energy as the thickness of the deposited crystal increases. The condition for
layered growth to take place is Yd + yr < ys, but if y, increases with the thickness of the

deposited layer, at a critical thickness this relation is no longer satisfied and the mode
switches to island growth. Figure 3.1 shows a schematic representation of the three types

of growth modes.
6<1ML %W/ Wfi/ iii/[77771555

! é % rm [— r-—1
1< 6<2 / 77777777777” 7////////////

9,2 - I / \ / ““
/ 77777777777” 7////////////
(a) (b) (c)
Figure 3.1 Crystal growth modes as a filnction of surface coverage 0 in terms of

monolayers (ML) (a) island or Volmer-Weber mode (b) layer plus island, or Stranski-
Krastanov mode (c) layer-by-layer or Frank-van der Merwe mode [32].

 

 

 

 

 

 

3.1.4 Substrate criteria for diamond heteroepitaxy

For heteroepitaxial growth of diamond the following substrate criteria are

desirable.

1. Crystal structure: diamond has a cubic structure so that substrates with cubic
structures are often advantageous, but not absolutely necessary.
2. Lattice parameter: a lattice parameter that is a close match to that of diamond,

0.357 nm, is helpful.

18

Surface energy: an equivalence of surface energy density between substrate and
diamond is advantageous. Otherwise 3D island growth will occur, which
ultimately requires coalescence of islands into a continuous film. This opens up

opportunities for defects to form at the coalescing surfaces. Surface energies of
diamond are quite large, e.g., for (111) and (100) planes they are 5.3 and 9.2 J/mz,

respectively [34].

Coefficient of thermal expansion: for growth at high temperatures, unequal
thermal expansion coefficients match lead to high interfacial stress after cooling
to room temperature. Large thermally-induced strains may lead to fracture of
substrate or epilayer.

Physical and chemical properties: substrate melting point should be much higher
than the growth temperature. Substrate Should have structural phase stability
between room temperature and growth temperature. Chemical stability at the
growth temperature is important: reactions that form carbides must be avoided.
For example, although Ni has only a 1.3% mismatch to diamond, it forms
carbides at diamond growth temperatures leading to an interface unsuitable for

epitaxy.

3.1.5 Epitaxial diamond growth on different substrates

As a working definition of diamond heteroepitaxy, we adopt the following criteria.

1.

2
Growth over areas of 1 mm or larger.

Structural confirmation by diffraction of epitaxial crystallographic registration. A

measure of mosaic spread, or average angular misalignment of grains, from

19

diffraction linewidths, preferably much less than 1 deg. For example, highly
oriented diamond (HOD), grown on Si substrates, shows some degree of epitaxial

alignment but presents an x-ray linewidth of several deg.

At present, many substrates have been evaluated for purposes of diamond epitaxy.
Silicon has been more extensively tested than any other material as it would allow
integration with much semiconductor technology [4,35-38]. Si has a lattice parameter
0.543 nm, 52% larger than that of diamond, but with an identical structure. Despite a
large mismatch, (001) HOD has been grown on Si. A 16 pm thick diamond film grown

with microwave plasma CVD yielded a (004) x-ray linewidth of 2.1 deg [39].

Cubic boron nitride (c-BN), with the zincblend structure and a 0.361 nm lattice
parameter, 1.3% greater than diamond, was explored several years ago as a substrate for
diamond epitaxial growth [40,41]. A 0.2 pm thick continuous film was grown over a
region of 200 pm [40]. An epitaxial relationship was confirmed by reflection high energy
electron diffraction (RHEED), although no linewidth were reported. However, the growth

of c-BN is difficult and is limited to small Sizes [42].

Silicon carbide (B-SiC), with the zincblend structure and a 0.436 nm lattice
parameter, 22 % larger than diamond, has been grown on Si as a film and used as a

substrate for subsequent HOD growth [43,44]. A diamond film with area greater than
1mm2 has been grown [44]. The epitaxial relation was established by x-ray diffraction

and an (004) linewidth of 0.6 deg measured.

(111)-oriented HOD was grown on (111) Pt, in bulk and thin film form [45-48].

Pt is FCC with lattice parameter 0.392 nm, almost 10 % larger than that of diamond. A

20

1.5 pm thick film was grown on a 10x10 mm2 area with a (111) linewidth of 1.1 deg
[49]. An epitaxial relationship was established by x-ray diffraction.

(111) and (001)-oriented diamond were grown on Ni substrates[50-52]. Ni is FCC
with lattice parameter 0.352 nm, just 1.3 % smaller than that of diamond. A 30 um
diamond film was grown on Ni With (100) and (111)-oriented grains. Ni has been

reported to form nickel hydrocarbide [50] or a nickel carbide [52,53] at a growth

temperature, of 900-950 °C.

(111)-oriented diamond films were grown on 10x10 mm2 Re buffer layers on

c-plane A1203 [54]. The (111) tilt and twist were reported to be 4.8 and 3 deg,

respectively, much larger than for diamond grown on B-SiC.

3.1.6 Heteroepitaxial diamond growth on iridium

Ir has the FCC structure with a lattice parameter 7.5% larger than that of diamond
and was first used as a substrate for diamond growth in 1996 [6]. It is the Pt-group
element with lowest carbon solubility [55] and shows no phase transition between room
temperature and its melting point of 2450 °C. Epitaxial Ir can be grown on oxide
substrates and used as a buffer layer for diamond growth. Several groups have already

demonstrated diamond epitaxy on Ir [6,8,56-5 8].

Ohtsuka et a1. [6,7,59] grew (001) diamond on 500 nm RF sputtered Ir on cleaved

(001) MgO using DC plasma CVD. A 8 pm thick freestanding film of about 2mm2 area

exhibited a Raman linewidth of 18 cm], an order of magnitude greater than the intrinsic

21

diamond linewidth at room temperature. No x-ray linewidth was reported for this crystal
[60].

The use of SrTiO3 as a underlying substrate for a (001)Ir buffer layer [9] led to an
epitaxial diamond film[10,56]. The 34 um freestanding diamond film had a (004) x-ray
linewidth of 0.17 deg and (311) linewidth of 0.38 deg [10], and led to the establishment
of a cube-on-cube epitaxial registration.

A freestanding 25 um single crystal diamond film was grown on electron-beam
evaporated Ir on SrTiO3 [58]. Improvements in nucleation density were shown to yield
coalescence of diamond grains after a shorter growth period. Epitaxial (001) diamond

was also grown on Ir evaporated on a-plane A1203, which has a complex hexagonal

crystal structure [61]. A1203 has a coefficient of thermal expansion closer to diamond

than SrTi03 leading to less thermal stress in the diamond fihn at room temperature.

Epitaxial (100) CaF2 and Ir were sequentially grown on (100) Si by e-beam

evaporation after which (100) diamond was grown on Ir. After 2 h growth, no diamond
coalescence was observed by SEM. No thickness and linewidth information was reported

for this film [62].

Recently, epitaxial diamond was grown on Ir, which was grown on thin-film
SrTi03 grown by molecular beam epitaxy on Si. Growth of Ir on yttria-stabilized zirconia

(YSZ) was also demonstrated [63,64].

22

3.2 Bias-enhanced nucleation (BEN)

BEN commonly refers to a process in which diamond nucleation density is
increased significantly by applying a negative bias (voltage) to the substrate at the
beginning of the CVD process. The terms bias treatment, bias process, ion-enhancement
or ion-irradiation are also used. With negative bias, positive ions from the plasma
bombard the substrate surface. The process was first used by Yugo et al. to grow
diamond on Si [65]. They reported a nucleation density 1010 cm'2 at -70 V bias. The
nucleation density without BEN was not reported in this study, but has subsequently been

reported to be 104 cm'2 [66]. It was proposed that, during BEN, positive ions accelerated

toward the substrate increased active hydrocarbon species at the sample surface due to
collisions between ions and substrate, which consequently led to increased nucleation
density [65 ,67]. When the DC bias is applied, an additional discharge appears, referred to

as the secondary plasma. This forms over the substrate above the dark space.

3.2.1 BEN mechanisms

Five different mechanisms that have been proposed to explain the increased
diamond nucleation due to BEN on Si are: (1) ion subplantation, (2) non-diamond carbon
layer deposition, (3) increased surface diffusion, (4) formation of a SiC buffer layer and
(5) electron emission.

Ion subplantation is the deposition of hyperthermal (1-103 eV) species by shallow
sub-surface implantation [68]. Robertson et al. [66,69] found the ion subplantation in Si
to be most efficient around 80-100eV ion energy [66]. They measured ion energies in

BEN at -250 V bias using a retarding field probe (inserted through a hole in the sample)

23

that had a distribution peaked at 80 eV. They proposed that subplantation creates
nanocrystalline graphitic carbon and that diamond nucleates where the graphitic planes
are locally oriented perpendicular to the surface. The graphitic carbon is able to
accommodate the 2:3 lattice mismatch between diamond and silicon. Molecular
dynamics (MD) simulations calculated an ion energy spread of 40-70 eV [70] and ion-
beam mass spectrometry measurements indicated average energies between [71,72] 50-
70 eV for -200 V bias. These results suggested ion subplantation to be the dominant
effect during BEN on Si.

Combining the results of MD simulations and experiments, Lifshitz et al. [68,73]
proposed a model of diamond nucleation on Si due to ion subplantation during BEN.

Here, a dense amorphous hydrogenated carbon (a-C1H) phase about 1-2 nm thick is
formed by subplanted ions. Spontaneous precipitation of a pure Sp3 carbon cluster,
containing tens of atoms of the a-C:H phase, is induced by a thermal Spike created by the

impinging energetic species. A few of the clusters are diamond (1 out of 104 to 106) and

they grow at the expense of the sp3 carbon in its Vicinity.

Other studies have suggested that ion subplantation is not able to nucleate
diamond [74,75]. The average energy of impinging ions was estimated to be only 15 eV
for -200 V bias. A threshold bias voltage, -100 V, was necessary for this process to be
effective [76]. Below the critical voltage, impinging ions are not energetic enough to
overcome the nucleation barrier.

Electron energy loss spectroscopy and TEM showed a 4mm thick amorphous

carbon layer after BEN at -200 V [74]. The ion energy was estimated as 20 eV, at most.

24

The negative bias led to the formation of a thin amorphous carbon layer with nucleation
sites formed as the amorphous carbon layer was deposited.
Another study, using optical emission spectroscopy (OES) at -150 V bias, [77]

showed an increased amount of atomic hydrogen at the sample surface during the BEN.

The atomic hydrogen preferentially etched the sz carbon, stabilizing the sp3 carbon

clusters. The thin layer of sp3 carbon provided the nucleation sites for diamond.

In other experiments, the formation of a nm thick SiC layer on top of the Si
surface was demonstrated [78-81]. The SiC layer is usually believed to be oriented but in
one case it has been claimed to be amorphous [82]. Subplantation of carbon into the SiC
layer causes carbon supersaturation, resulting in the subcutaneous formation of
epitaxially oriented nucleation centers in the SiC layer [80].

In other studies, BEN has been found to simply increase the atomic diffusion of C
on the Si surface [83,84]. Jiang et al. used -150 V bias to induce nucleation on silicon,
with 3D non-faceted diamond islands initially formed. The size of the islands increased
as biasing progressed. They claimed formation of a depletion zone between the islands,

attributed to the enhanced diffiision of carbon by ion bombardment during the biasing.

In addition to the nucleation enhanced by the ion bombardment, in some
experiments electron emission from diamond has been suggested to lead to nucleation
enhancement. Biasing experiments done under identical conditions with a bare Mo
holder, a diamond-coated Mo holder, and an uncoated sapphire holder yielded different
results. Much higher nucleation densities were obtained with the diamond-coated holder
and bias current increased almost five-fold. Experiments with the other holders showed

no increased nucleation [85].

25

Enhanced nucleation was also achieved when a tungsten ring inserted between the
substrate and a hot filament was negatively biased at -300 V relative to the hot filament
[86,87]. It was claimed that this result was due to electron emission from the diamond-

coated tungsten ring.

3.2.2 Biasing experiment with Ir

In comparison to Si, relatively few studies have examined the role of biasing on Ir
substrates. In some studies, it was reported that the Ir surface is roughened during BEN
[6,88]. X-ray photoelectron spectroscopy (XPS) and RHEED of an Ir sample ion-
irradiated at -260 V in a DC plasma CVD system showed that the Ir surface is roughened
during ion irradiation. XPS did not Show any compound formation between Ir and C or

H, but did indicate the presence of a thin layer of amorphous carbon [6].

Ir roughness was reported at the diamond-Ir interface by cross sectional HRTEM
after ion irradiation at -260 V for 10 min followed by 30 min of growth in a DC plasma
CVD [89]. No interlayer formation such as IrC or complex hydrocarbon compounds of Ir,

was found. The top surfaces of the Ir were proposed as diamond nucleation sites.

Diamond crystallites nucleated along [110] ridges of an Ir surface after biasing at
-120 V for 15 min in an antenna-edge type CVD system. Nucleation was viewed as more
likely on areas of Ir that protruded into the plasma. After growth for 2 rrrin diamond
crystallites smaller than 10 nm in size were etched away by a hydrogen plasma. It was

concluded that a critical size for diamond crystallites exists [90].

X-ray photoelectron Spectroscopy (XPS) and x-ray photoelectron diffraction

(XPD) studies on Ir surface after a bias treatment showed no layers that could be

26

associated with compounds of Ir, C and H, but almost 20 monolayers of amorphous
carbon on the Ir surface was detected. XPD measurements also detected nm-Sized
diamond crystallites after BEN with no growth; however, they were not seen in low

energy electron diffraction (LEED) [91,92].

The structural effects of biasing were studied with SEM, AFM and TEM. Two
different types of furrows and ridges along <100> and <110> were reported after 45 min
of bias at -250 V. Shape, dimensions and the crystallographic alignment of the furrows
and ridges changed with different methane concentrations [93-96]. In addition, 5-10 nm
structures were found with a typical distances of 100-300 nm. 30 min of growth resulted
in similar densities of diamond crystallites [93]. In another study after 45 min bias at -250
V, a smooth Ir surface with less than 1 nm average surface roughness was observed.
After 30 min of growth, diamond crystallites grew on the flat Ir surface as observed in

TEM. No evidence for any intermediate layer of amorphous C was detected [97].

In a HRTEM study, 5-10 nm nanoparticles were found to lie on top of diamond

grown on Ir. HRTEM revealed the lattice parameter of the nanoparticles to be that of Ir.
The particle density was close to 1011 cm‘z, 3 orders of magnitude higher than that of

diamond nuclei [94].

Domains, several pm in size, were observed by SEM to form after bias with
negligible topographic features. After the growth step, only the domains developed into
islands of epitaxial diamond. Outside these domains, either the nucleation density was
orders of magnitude lower or the diamond grains were misoriented [98]. In a combined
AFM and SEM study, in addition to 2-3 nm roughening of domains, the domains were

found to be 1 nm lower than the surrounding layer. It was suggested that domain

27

formation is a continuous process during the bias procedure, which starts from a local

nucleation event and subsequently grows [99].

More recently, in a HRTEM study a 0.75 nm carbon layer was seen on the Ir
surface after biasing for 50 minutes at -250 to -280 V. After 2 min of growth, diamond
crystallites up to 8 nm in height were observed. Amorphous carbon deposited during the
BEN was etched by the plasma during this short growth step [95]. It was suggested that
the surface roughening of the Ir that occurs in most cases is not essential for diamond
nucleation. During biasing, the initially deposited amorphous carbon layer is densified by
ion bombardment and diamond nucleates in a large fluctuation caused by a preferential

displacement process induced by an impinging ion.

In summary, how biasing leads to nucleation or to enhanced nucleation density on
Ir is still imperfectly understood. Most studies agree that the Ir surface is roughened
during biasing and that a thin layer of amorphous carbon is deposited; yet, one study
claimed that the Ir surface can remain smooth during biasing under the same conditions
and diamond can be grown on this surface. Thus, two basic questions that need resolution
are: how does roughening of the Ir surface evolve with biasing and is roughening

responsible for enhanced nucleation?

3.3 Homoepitaxy

3.3.1 Background

The first homoepitaxial diamond growth by the microwave plasma CVD method
was first reported by Kamo et al. in 1988 by growing on a polycrystalline diamond

substrate [112]. After this demonstration, homoepitaxial growth was investigated on

28

natural or HPHT Type Ib substrates. Growth rates were typically lower than 1 pm/h [100-
107]. Also, methane concentrations, usually defined as the ratio of methane to hydrogen
flow rates, had to be maintained at 0.025% to suppress growth of defects. Three kinds of
macroscopic defects are commonly observed in homoepitaxial diamond: nonepitaxial
crystallites (NC or UC), pyramidal hillocks (PH), and macro-step bunching (B) [108].
The first two defects are serious as they cannot be reversed, whereas the third type is less
problematic. Takeuchi et al. [108] grew homoepitaxial diamond films on substrates with
different misorientation angles from the (001). They produced a map of diamond surface
morphologies as a function of misorientation angle and methane concentration as shown

in Figure 3.2.

29

Atomically Flat

 

 

      

 

 

 

MISORIENTATION ANGLE [deg]
N

..................... . rrcikcr _ r.“ 55.
' . e
1 44/ J . 1‘0“"; V
(l) e . .
O c 0‘ tt‘,
‘ 'L l“
, ( I: (i (
0 I I x L n
0.0 0.5 1.0 1.5 2.0

CH4/H2 [%]
Figure 3.2 Map of surface morphologies as a function of misorientation angle from the

<001>and methane/hydrogen ratio. Three types of morphologies appear. Atomically flat
diamond is preferred, which only grows in region III [108].

However, more recent work has shown that a different region of parameter space
can be effective in growing diamond at much higher rates. Yan et a1. grew diamond at
rates approaching to 40 um/h in the presence of nitrogen [12] with growth temperature
1500 °C, pressure 150-200 Torr and methane concentrations from 1 to 6%. Later, grth
rates were increased to 150 um/h by introducing 1-5 % N2 relative to methane flow, and

methane flow up to 12 % of hydrogen flow [13].

More recently, by lowering the growth temperatures to 700-800 °C and by
increasing microwave power density at the substrate, fairly high growth rates were

achieved in the absence of nitrogen [14,15,109-111]. In addition, whereas gas pressures

30

previously used ranged from 20-50 Torr [112], for high rate grth pressures were

increased to 100-200 Torr. Former microwave power densities were around 5 W/cm3 but
for high rates they were increased to 30-100 W/cm3 [112]. To accommodate high

microwave power densities without changing growth temperatures, heat must be
extracted from the substrate by a cooled stage. Table 3.1 summarizes differences between

low and high growth rate regimes.

Table 3.1 Low and high growth rate regimes in microwave plasma CVD diamond grth

 

 

 

 

 

 

Parameter Low growth rate High growth rate

Power density(W/cm3) 5 30'100

Substrate temperature (°C) 600-900 700-1200

Gas pressure (Torr) 20-50 100-200

Growth rate (pm/h) <1 100 um/h with N2
20um/h withouth

 

 

 

 

3.3.2 Growth of defects

3.3.2.1 Pyramidal hillocks (PH) and non-epitaxial crystallites (NC)

On the (001) diamond surface, the most commonly occurring defect is the
pyramidal hillock, either with a flat top surface or with a sharp truncated top containing a
non-epitaxial crystallite [113,114]. The sides of this hillock are aligned with [110]. Under
favorable conditions, this hillock grows very rapidly compared to its surroundings.
Hillock growth is favored at high temperatures and high methane concentrations [14].

Figure 3.3 shows a homoepitaxial diamond surface covered with these defects [115].

31

 

figuf'e 3.3 Pyramidal hillocks with non-epitaxial crystallites at the top (dark particles)
115 .

According to Lee et a1. [116-118], the type of growth strongly depends on the
diffusion of the growth species on the surface, the rate at which growth species are
deposited on the surface from the vapor phase, and the width of the terraces on the
growth surface. If the terrace width is long compared to the diffusion length, then the
growth species tend to deposit on the middle of the terraces, instead of diffusing to the
steps. This results in two—dimensional nucleation, as shown in Figure 3.4. Then, the two-
dimensional nuclei provide shorter steps for further growth along the sides of the hillock.
Non-epitaxial crystallites are believed to grow on surface defects by enhanced secondary
nucleation [113,116]. At higher temperature these non-epitaxial crystallites grow very

fast and provide step edges for further growth.

ii 111
m
V.

Figure 3.4 Mechanism of two dimensional nucleation on a wide terrace. Beginning of
hillock growth [1 16].

32

3.3.2.2 Step bunching

If the diffusion length of the growth species is larger than the terrace width, the
growth species diffuse toward the steps [116] because of the lower potential there. In this
growth mode, the steps appear to flow across the growth surface. A schematic View of the
step flow growth mechanism is shown in Figure 3.5. Step flow grth in diamond is
favored at higher temperatures, low methane concentrations and high misorientation
angles of the substrate surface. Several reports have appeared on single crystal diamond

growth in step flow mode using vicinal substrate surfaces [115,119,120].

0

l i 1’ ° °
Figure 3.5 Schematic View of step flow growth showing the diffusion of growth species
to steps and subsequent flow of steps to the right [116].

During growth, impurities or defects may block the transport of growth species or
adsorbates to a particular step [121,122] so that flow ceases, but other steps will flow to
that position and become pinned. If this process continues, many steps will become
bunched at that position leading to a macroscopically high step. This is referred to as

macro-step bunching. Figure 3.6 shows macro-step bunching in homoepitaxial diamond

[123].

33

 

Figure 3.6 Macro-step bunching in homoepitaxial diamond [123]
3.3.3 Effect of growth parameters

3.3.3.1 Substrate temperature

Substrate temperature during diamond growth plays a most important role. It
significantly affects growth rate and the crystalline quality of the diamond. The growth
rate R3 of homoepitaxial diamond was found to obey an Arrhenius relationship between
800-1000 °C [112,124] with Rg = A exp(-Ea/kBT), where E3 is an activation energy and T
is the growth temperature. Diamond growth in this temperature region is controlled by
chemical reaction kinetics rather than by surface diffusion. Above 1000 °C, the growth
rate no longer obeys the above relation, but saturates. Here, growth changes to a
diffusion-controlled mode, with a weaker temperature dependence. Values of E, for
temperatures between 735-970 °C were reported as 8 :l: 3, 18 i 2 and 12 :I: 4 kcal mol'l,
respectively, for (100) and (110) and (111) diamond surfaces [125]. Different growth
conditions and reactor geometries will change 13,, so that the use of these parameters may
be sensitive to local conditions. Figure 3.7 shows a plot of growth rate vs. substrate
temperature.

34

Tern peratu re [°C] '

 

 

 

 

o o
5 \\Q' ‘00 ' 990 , $00
~ 45—
E o 0‘65; 5
5 ~. 9 a
3 .- ‘. If'0 1
3 : ‘.\ ‘9/ :
E .s‘IbO/‘I E
2 E. ‘ \ ‘1
I ‘0:
: Cme=4.0°/o :
1 - I l 1 l d
0.7 0.8 0.9
1000/T(K'1)

Figure 3.7 Diamond growth rate as a function of substrate temperature for 4% methane
concentration [1 12].

At temperatures above 1200 °C, graphite starts to form [112]. At temperatures
lower than 700 °C, diamond growth rate is very low and the substrate becomes highly
pitted [126]. On the other hand, the growth of smooth, ahnost atomically flat, diamond
also has been reported at 700 °C [15]. This apparent contradiction may not exist if growth
depends also on other parameters such as methane and nitrogen concentrations, and
reactor geometries. Usually there is a small temperature window in which defect-free

diamond can be grown with other conditions fixed.

3.3.3.2 Methane concentration

Methane concentration in the gas phase significantly affects the diamond growth
rate and the morphology of diamond. The growth rate is generally found to be linearly
proportional to methane concentration. Figure 3.8 shows a plot of grth rate vs.

methane concentration at 850 0C Without nitrogen from Achard et al. [126].

35

20

 

 

 

 

T=850°C
‘ MWPD=96 Wicm3
16 - N2 concentration = 0 ppm I
E
3 12 - II
.9
E
.C 8 ..
E
o I
4 _
I
0 I I ' ' fl 1 ' I '

Methane content in the gas phase (%)

Figure 3.8 Growth rate as a function of methane concentration at 850 °C without nitrogen
[126].

The above data indicate a slightly supra-linear relation between methane
concentration and growth rate. In contrast, Bogdan et al. [127] reported an sub-linear
result. Teraji et al. [112] reported a supra-linear relationship from 1 to 5% methane but
then a linear dependence up to 32 % methane at 890 °C. Whereas, the growth rate always
increases as methane concentration increases, the proportionality coefficient can be

sensitive to other growth parameters.

Bogdan et al. [127] reported a smooth diamond surface grown at 15% methane
concentration at 700 °C. In contrast, Achard et al. [126] reported macro-step bunching on
the diamond surface When they increased the methane concentration from 4 to 6%. At 2
% methane concentration, they found the diamond surface to be highly pitted. Teraji et al.
[112] observed the formation of large hillocks, different from pyramidal hillocks, at 16%

methane concentration and small faceted structures at 1% methane concentration. They

36

attributed it to different growth modes occurring at widely different methane

concentrations [1 12].

3.3.3.3 Process gas pressure

The total process gas pressure has an influence on the size and energetics of the
microwave plasma. As pressure increases, the size of the plasma ball becomes smaller,
since the mean free path of the ions becomes shorter. This influences the substrate
temperature, since the power density may become greater in the Vicinity of the substrate.
In a microwave plasma CVD system, the electrons in the plasma are accelerated by the
high frequency electric field, but the ions cannot follow the microwave field, leading to a
low ion temperature. The ion temperature can be increased by enhancing collisions with
hot electrons, brought about by a higher gas density. When the total gas pressure is low,
the ionization rate in the plasma is also low with only a relatively small number of
electrons available for creating atomic hydrogen and hydrocarbon radicals. At higher
pressures and power densities, the ionization rate is higher and the frequency of collisions
between the ions and electrons become larger, which increases the ion temperature. This
ultimately leads to an increase in the density of hydrocarbon grth species and atomic
hydrogen. Mokuno et al. [128] found the growth rate increasing almost linearly with
pressure in their microwave plasma CVD reactor in the presence of nitrogen. Yan et a1.
[12] reported a five-fold increase in growth rate, when the pressure was changed from 60

to 200 Torr.

37

3.3.3.4 Microwave power density

Microwave power density is found to strongly affect the grth rate. Tallaire et

al. [129] observed a three-fold increase, from 3 to 8.5 um/h in growth rate, with a change
of power density from 65 to 125 W/cm3. They also used a pulsed microwave source to

increase the peak power density and while maintaining a constant average power density.
They observed an increased growth rate of 40%, with the conclusion that peak power
density, not average power density, is responsible for the enhanced growth rate. Teraji et

al. [112] saw little difference in growth rate when the power density changed from 40 to

50 W/cm3.

3.3.3.5 Nitrogen

Nitrogen is the most common impurity in natural, HPHT, and CVD diamond. As
discussed earlier, nitrogen forms several defect centers in diamond, in conjunction with
other nitrogen atoms or with vacancies. In CVD diamond the most common nitrogen
center is singly substituted nitrogen. Substitutional nitrogen reduces the optical
transparency in the region above 1.7 eV and in the one-phonon region in the it beyond 9
pm. In addition, during CVD growth, nitrogen has beneficial effects on the surface
morphology, growth rate and the promotion of (001) growth. Addition of nitrogen to the
gas phase during diamond growth on Si, changes the grth habit from (111) to (001)
and increased the growth rate by a factor of two on the (001) surface at low nitrogen to
carbon ratios [130]. In homoepitaxial diamond growth, adding controlled amount of
nitrogen increased the growth rate to more than 100 um/h [13,131]. So far, the

mechanism by which nitrogen influences the (001) diamond grth is not well

38

understood. Mechanisms such as surface poisoning or the generation of additional growth
steps have been considered [132]. According to a theory based on defect-induced
stabilization of diamond grth [133], diamond growth is only thermodynamically
possible by the creation of vacancies, since a diamond vacancy can be created with lower
energy than a graphite vacancy. If nitrogen is doped into diamond, the Fermi level at the
growth surface is shifted towards the conduction band, which reduces the energy of
vacancy formation in diamond, but not in graphite, so that diamond growth is favored

over graphite.

3.3.4 Growth anisotropy: the alpha parameter

The diamond growth parameter a is defined as the ratio of the growth rate normal

to (001) with the growth rate perpendicular to a (111) surface, usually written as,

01 = 3 V(001)/V(111)

where W001) and V011) are the diamond growth rates on the (001) and the (111) diamond
surfaces, respectively [134]. The equilibrium shape of a diamond crystal depends upon
the growth rate anisotropy, thus on 0.. Shapes of diamond crystal for different values of a
are shown in Figure 3.9 [135]. If a = 1, then the fastest growth direction is normal to
(111) so that no (111) facets appear. For 01 larger than 1, (111) faces of diamond appear
because the fastest growth direction is no longer normal to (111) surface. For a = 3 the

fastest growth direction is normal to (001) and the crystal shape is an octahedron.

39

PEPPPPPW

a = l 1.25 2.75 3

Figure 3.9 shape of diamond crystals for different value of 01. The arrow shows the
fastest growth direction for that 01 [135].

Although the value of the parameter depends upon the overall growth conditions,
it is found to be larger for higher methane concentrations and lower temperatures in
general. In actual experiments, growth occurs on many crystal planes, not only on (001)
or (111), but also on (110) and higher index planes, so it is difficult to accurately predict

the overall crystal Shapes from a single parameter [136].

40

4 Chapter 4 Materials Characterization Methods

4.1 Optical Microscopy

Optical inspection and photography are valuable tools for the examination of
surface features of crystalline materials. With optically transparent crystals, several
methods exist for providing optical contrast. The differential interference contrast (DIC)
microscope converts the gradients in the specimen optical path length into amplitude that
can be visualized as enhanced contrast of the resulting image [137]. A general schematic

of such a microscope is shown in Figure 4.1

Light detector

Obi ective lens . .
Polarizrng filter

Wollaston prism

Condenser lens Sample

Polarizing filter
Wollaston filter

 

Light source
Figure 4.1 Schematic of a differential interference contrast (DIC) optical microscope

Unpolarized light from the source passes through a linear polarizer to be polarized
at 45 degrees, and then passes through the Nomarski-modified Wollaston prism which

splits it into two beams with 90 deg polarization differences. The two beams pass through

41

the condenser lens to be focused on two adjacent points on the sample surface. The
beams traverse two slightly different optical paths due to slight topographical variations
across the sample thickness. This creates a slight difference in their relative phases. The
rays then exit the sample, are focused by the objective, and pass through a second
Wollaston prism that combines them into a single beam. Their interference leads to a
brightening or darkening of the image at that point, dependent on their optical path

difference.

The DIC optical microscope used for this work is the Olympus BX-60, coupled to

a digital Spot camera. Digital images were acquired and processed on a computer.

     
 

 

 

 

 

 

Observation
tube
__-;’—_:- Collective lens
Eye piece unit
@ 1 in Housing for
(5‘ ...__ ‘ \ reflected light
Ma}.- ll. ......
‘6 b—. _A
w Ag; 1
Mirror cube . ' ' ' Ob' t'
housing 3.2; 2,; I jec Ive
- ‘ "I ‘ ...-e
Nose piece ' ,3/ "‘
  / iii
Stage i ' k . g
3. I. o Lamp housing
Condenser ‘ ‘ ‘ ’ ' "'" ‘1'" ransmitted light

Base

Figure 4.2 The Olympus BX-60 DIC optical microscope

42

4.2 Atomic Force Microscopy

The atomic force microscope (AFM) is a high-resolution scanning probe
microscope. It relies on a sharp tip, with typical radius 10 nm, to map the topography of a
surface on a scale of a few nm to 100 um, while the vertical sensitivity can be 0.1 nm or
less. The tip is attached to the end of a cantilever with force constant of order 1 N/m.
When the tip is brought to close to a sample surface, the force between the tip and the
sample surface deflects the cantilever. The nature of the force depends on their material
properties. It can be a magnetic force, electrostatic force or van der Waals force.
Deflection of the tip is tracked by a photodiode which detects a laser beam reflected from
the back of the cantilever. The AFM in this work was a Digital Instruments D1 3100
scanning probe micrOSCOpe used in tapping mode. In this mode, a Si tip near the sample
surface is driven at its resonant frequency by a piezoelectric tube [138]. As the vibrating
tip is rastered across the surface, variations in topography change the external forces
acting on the tip. The force changes lead to damping of the tip’s vibrational amplitude,
which is maintained at a constant level by application of a feedback loop. The system
controller interprets the changes in the feedback amplitude as a variation of the surface

topography. A schematic of an AF M is shown in Figure 4.3.

43

 

Amplitude System High Resoluu .
Detector Controller Oscillator

     

 
   
   
  
    

Laser source
Photo diode

Piezoelectric
Oscillator

 

Laser beam

Cantilever Sample

Figure 4.3 Schematic of atomic force microscope

 

4.3 Scanning Electron Microscopy

The scanning electron microscope (SEM) uses an electron beam to form an image
of a sample surface. The electron beam is focused to a nm spot using one or more
electromagnetic condenser lenses. The beam then passes through pairs of scan coils in the
objective lens assembly which raster the beam in x and y directions across the sample
surface. When the primary electron (PE) beam, with energies in the range of 10-30 keV,
impinges on the sample, it produces large numbers of radiated electrons and photons.
These include secondary electrons (SE), backseattered electrons (BSE), Auger electrons
(AE) and x-rays, characteristic of the elemental material [139]. Secondary electrons are
the low-energy electrons produced by the inelastic scattering of weakly-bound electrons
due to the primary beam. Secondary electrons have energies below 50 eV, with a

distribution peaked at 3-5 eV. These electrons are produced within a few nm below the

44

surface. The number of SE’s produced depends on the topographical features of the
surface. Steep surfaces and edges produce more SE’s than a flat surface. The SE’s are
detected by an Everhart-Thornley (E-T) scintillator-photomultiplier to form a two-
dimensional intensity distribution. The image of the surface is stored by a computer built

into the SEM for subsequent processing.

 

_] Variable hlgh voltage
Electron gun DOWN WY

 

 

 

 

 

 

 

 

 

Electron beam

 

   
 

 

Scan(magniflcatlon)
call control:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

I [151329-94 I!
_] Sample
To mount 4— li mfg!”
1::
Sample Chamber

 

 

 

Figure 4.4 Schematic of a scanning electron microscope [140]

Backseattered electrons are also produced in the SEM. The BSE’s have a wide
energy spectrum, varying from 50 eV to a maximum limited by the applied potential.

They also produce SE’s and Auger electrons as they propagate through the sample.

45

BSE’S tend to move in linear trajectories. A separate semiconductor detector is used to
detect BSE’S that gives a signal proportional to the BSE energy. The BSE image carries
the information about the atomic number (Z) of the sample, since a larger number of
BSE’S are produced at larger Z values. BSE images are especially useful for identifying
areas with different chemical composition. BSE images have lower resolution than SE
images, a result of their larger interaction volume. Finally, x-ray fluorescence is used in
energy dispersive spectroscopy (EDS) to identify elements present in a sample.

The SEM used in this research is the Hitachi 4700 field emission SEM (F ESEM).
This F ESEM has a sharp tungsten cathode, with radius of curvature less than 100 nm, as

a field emission electron source. The tip of the cathode is biased at a high potential so that
the electric field at the tip is of order 107 V cm'l. In this field, emission occurs when

electrons tunnel through the potential barrier into vacuum. The crossover beam diameter
in the FESEM is almost 1000 times smaller than for thermionic emission guns used in
conventional SEMS. A FESEM thus requires fewer magnetic lenses to focus the electron
beam to a small spot size at sample surface, yielding better resolution than thermionic
emission SEMs. The Hitachi 4700 II FESEM contains three sets of electromagnetic

objective lenses, two electron detectors and one x-ray detector.

4.4 Transmission Electron Microscopy

In a transmission electron microscope a high-energy electron beam interacts with
a thinned sample, much less than a pm, with some electrons being transmitted through it
to form an image. A simple schematic of a TEM is shown in Figure 4.5.

The nearly monochromatic electron beam emitted from an electron source, a

thermionic or field emission gun is focused by a pair of condenser lenses to a narrow

46

coherent beam. A condenser aperture is used to block electrons emitted at a high angle to
the optic axis. The beam of electrons transmitted through the sample passes through an
objective lens and then through the objective aperture, which again blocks electrons
scattered at high angles. A first image is formed at the image plane of the objective lens
before it passes through the intermediate lens. The diffraction information from the
sample appears at the back focal plane of the objective. By adjusting the position of the
intermediate lens, and with use of different apertures, a highly magnified bright field
image, a dark field image, or a diffiaction pattern of the sample can be projected on the

final fluorescence screen, a photographic plate, or a CCD camera

  
 
  
 
 
  
   
   
  
  

Electron source

First condenser lens

Electron beam

condenser lens
Condenser aperture

Sample

Objective lens
Objective aperture

Selected area aperture
lens

Projector lens

Image
Figure 4.5 Schematic of a transmission electron microscope.
A crystalline material interacts with the electron beam mostly by diffraction so

that a characteristic diffraction pattern is generated. A high contrast bright field image is

47

formed by blocking all the diffracted beams except the central transmitted beam. On the
other hand, a dark field image is formed when one of the diffracted beams is used to
obtain the final image. The bright field technique can be used to image extended defects

such as dislocations or stacking faults.

High Resolution Transmission Electron Microscopy (HRTEM) allows the
visualization of atomic-scale crystallographic structures. HRTEM, in contrast to the
conventional TEM, uses a phase—contrast imaging technique instead of amplitude-
contrast imaging. For phase-contrast imaging, the sample must be sufficiently thin so that
amplitude variations do not contribute to image contrast. Because of its very high

resolution, ~0.1nm, lattice points and point defects can be resolved.

4.5 Raman scattering and photoluminescence

When monochromatic light interacts weakly with molecules or condensed matter,
the scattered light has an inelastic component due to internal vibrational modes. This
phenomenon is called the Raman effect [141], shown experimentally by CV. Raman in
1928. During the scattering, phonon creation is referred to as Stokes scattering, whereas
phonon absorption is anti-Stokes scattering. The change in wavelength of the scattered
light from the incident light is called the Raman shift, usually expressed in terms of wave
numbers cm'1

In a Raman spectrometer, the laser light scattered by a sample is collected by a
lens, often the objective lens in a microscope, then sent through notch filter to remove the

elastically scattered light. The light then is dispersed by a holographic grating onto a

detector, usually a CCD camera.

48

 

Figure 4.6 Schematic of 3 Raman spectrometer set up.

FC: fiber optic coupler, FO: fiber optic cable, FPH: fiber optic probe head, L 1: focusing
and collection lens, SA: sample, L2: pre filter collimating lens, NF: holographic notch
filters, L3: post filter focusing lens, SL: spectrograph entrance slit L4: spectrograph
collimating lens, HG: holoplex grating, L5: spectrograph focusing lens, CCD: detecting
CCD camera.

For nominally unstrained diamond, there is only one Raman-active mode, which
appears at 1332 cm'1 with a linewidth of 1.6 cm'1 at room temperature [142]. The actual
linewidth will depend on temperature [143], the resolution of the spectrometer, and the
structural quality of the diamond. The Raman phonon is three-fold degenerate, but the
degeneracy can be lifted by internal strain. For small strain, this may appear as a line
broadening. For the Raman measurements described here, a Kaiser Optical Systems
Holoprobe Raman spectrometer coupled to an Olympus BX-60 optical microscope was
used, with 532 nm laser excitation. A 100x objective was used to focus the laser beam to

a diffraction-limited spot, and to collect the backseattered light. This system has available

two gratings: a low-resolution grating with a 0 to 4000 cm'1 spectral range and a high-

49

resolution grating with a 1000 to 1600 cm'1 range. The best Spectral resolution is about 1

cm". A schematic of the Raman spectrometer is shown in Figure 4.6.

4.6 X-ray diffraction

X-ray diffraction probes the microscopic atomic arrangement of condensed
matter. It is the basic method to analyze the structural organization of a solid’s
constituent atoms or molecules. For single crystals, the linewidths of diffraction peaks
(Bragg reflections) can be used as measures of the internal strain of the crystal. A sample
is placed in the path of a monochromatic x-ray beam at an angle a) to the sample surface;
the detector is rotated in the scattering plane to observe the diffiacted intensity as a
function of angle, as shown in Figure 4.7.

X-ray source X-ray detector

at

       

 

 

Sample ]

 

 

Figure 4.7 Schematic of x-ray diffraction from a crystal.

50

Diffraction intensity is observed only when the Bragg condition, 2(1th sin0 = nit,

is satisfied [144] ,where A is the wavelength of the incident x-ray, dhkl is the lattice

spacing between (hkl) planes of the crystal and n is an integer corresponding to the order

of the diffraction, as illustrated in Figure 4.8.

 

Figure 4.8 Schematic showing the Bragg"s law of diffraction

With incoming radiation of wavevector K, there will be an outgoing wavevector

K’ if and only if the relation K’- K = Q = thl is satisfied, where thl is a reciprocal

lattice vector with magnitude 2112/th This is an alternative statement of Bragg’s law,

which can be visualized by the Ewald construction. An Ewald sphere is drawn with a

radius equal to the magnitude of K in reciprocal space, Fig. 4.9.

51

2n

K-K'=—=2Ksin6l
hkl
2.2—”sin 6? = E
A dhk,
261W sin (9 = it

This is the Bragg equation for the first order diffraction maximum.

29

Q=thl

 

 

Figure 4.9 Ewald sphere construction

The x-ray diffractometer used for this work is a Bruker AXS D8 instrument, with
a 4-circle Eulerian cradle. A vacuum chuck is used to hold a sample on a stage, which
can be rotated around three mutually perpendicular axes for (0, x and (I) circles. To align
the sample at the center of the rotation circles, motion in the X, Y, and Z directions is

possible. A schematic drawing of this diffractometer is shown in Figure 4.10.

52

Chi rotation
X

Phi rotation //.\
/

*1

I

I

. a +4

Theta rotation

C,

        
  

 

 
 
    
  

 

 

 

    

7

_
=
-

1. X-ray tube 2. X-ray focussing mirror

3. X-ray attenuator 4. X-axis stage sliding knob
5. Y-axis stage sliding knob 6. X-ray detector

7. rotatable sample stage 8. Germanium monochromatc

Figure 4.10 Schematic of Bruker AXS x-ray diffractometer

The x circle is perpendicular to the X-Y plane, the (I) circle lies in the XY plane
and the (o, 0 or 20 circles are perpendicular to both x and (I). In the schematic, the x circle
is denoted by an “x”, which is perpendicular to the plane of the paper or sample. To align
the sample, initial conditions are set to x = to = 20 = 0 and (I) = 90, the incident x-ray
beam, the plane of the sample and the line joining the sample plane and detector all lie in
the XY plane and the x axis becomes parallel to the incident beam. X-rays emitted fi'om a
Cu target are focused by a curved multilayer mirror, and then (optionally) pass through a
Gee monochromator before they are incident on the sample. All rotations and translations

are controlled by a computer which also acquires the intensity data.

53

4.6.1 to - 20 or 6 - 20 scan

For a 0 - 20 scan, the sample and detector both are rotated such that the angular
speed of the detector is twice that of the sample. Diffracted intensity is observed when the
Bragg conditions are satisfied at particular value of 20; records of intensity vs. 20 are

recorded.

4.6.2 0 or w scan or rocking curve

For a rocking curve scan, the detector is fixed at the 20 value that satisfies a Bragg
condition, then, the crystal is rotated so that the lineshape of the diffiacted intensity can
be analyzed. This method is most useful for obtaining the mosaic spread of a crystal from

its linewidth.

54

5 Chapter 5 Heteroepitaxial Growth of Diamond

This chapter describes the processes developed for heteroepitaxial diamond
growth. A double epitaxial process has been used, in which an Ir buffer layer is first
grown as a thin single crystal film, 300-600 nm thick on an oxide substrate, either
strontium titanate (SrTiO3) or sapphire (A1203). We first discuss the oxide substrates and
their suitability for epitaxy. The Ir growth process on the oxide substrate is then
described, as well as methods used to characterize the thin films, particularly AF M and x-
ray diffraction. The CVD reactor, sample geometry and the role of the initial DC biasing
step to induce nucleation are explained in detail. The resultant products of diamond
nucleation and growth are analyzed by SEM, AF M, and HRTEM. The results are used to
reach a better understanding of the mechanisms that underline nucleation, growth, and
coalescence into thin films of diamond. Experiments to grow diamond with large lateral

Size are described.

5. 1 Substrates

This section describes the properties and structures of Ir, SrTiO3 and A1203 , used

as substrates for diamond growth. Previous, as well as current work of epitaxial Ir growth

on different substrates is first presented, followed by a discussion of the oxides.

5.1.1 Iridium

Ir is a metal with melting point 2454 °C, low vapor pressure (1.47 Pa at 2443 °C)
and high (001) surface energy (3.81 J/mz) surface[145]. It has a FCC unit cell with cubic

lattice parameter 0.386 nm. Thin films of Ir have been previously deposited by metal

55

organic chemical vapor deposition [146,147] pulsed laser deposition [148,149], electron
beam evaporation on SrTiO3 [56] [58,140], and radio frequency magnetron sputtering on
cleaved MgO [6,150]. In general, oxide substrates have been used as Ir substrates,
including MgO, SrTiO3, a-Ale3[ll], and yttrium stabilized zirconia (YSZ) [151].

Processing of the oxide surfaces is necessary to insure good epitaxy as described below.

5.1.2 Strontium titanate (SrTiOa)

SrTiO3 is a cubic perovskite structure with lattice parameter 0.391 nm, 1.7%

larger than Ir. Its melting point is 2080 0C. The (001) SrTiO3 and the (001) Ir surfaces

have sufficiently similar lattice parameters that epitaxial growth is easily accomplished.

5.1.3 Sapphire (A1203)

Sapphire has a hexagonal crystal structure with space group R3C. Its lattice

parameters are a = b = 0.476 nm and c = 1.299 nm and its melting point is 2040 °C.

Epitaxial growth of (001) Ir has been demonstrated on the a-A1203 (1120), or a-plane

surface[l 1] In the a-plane, two orthogonal directions lie along <0001> and <1100>, the

c-axis and m-axis, respectively. The distance between nearest-neighbor oxygen atoms is
0.433 run along <0001> and 0.275 nm along <1100>. If the (001) Ir plane is

superimposed on this surface with [110] Ir || [1100] A1203, there is only a 1.2%
mismatch. a-Aleg has two advantages over SrTiO3: the coefficient of thermal expansion

of A1203 is smaller than that of SrTiO3, so that thermal stresses in epilayers are smaller

56

and large wafers of A1203 are available so that there is no practical limit on the size of

diamond films that can be grown on sapphire.

5.1.4 SrTi03 preparation

Epitaxial grade SrTiO3 substrates with (001) surfaces, polished on one side, with
dimensions 10x10x0.5 mm3 were purchased from Mateck. In this investigation two
standard sizes were used: 10x10 and 5x5 mm2 with 0.5 mm thickness. To prepare 5x5

mm2 substrates, 10x10 mm2 substrates were coated with photoresist and cut into 4 equal
square pieces using a diamond-coated wire saw. Each substrate was ultrasonicated
sequentially in acetone, iso-propanol and DI (deionized) water to remove the photoresist.
If necessary, they were degreased in Alconox solution in DI water to remove surface
particulates. To create an atomically flat TiO2-terminated surface, water-cleaned
substrates were placed in buffered oxide etch (10 molar NHan HF: H2O = 12: l: 20, with

pH 5.5) for 10 to 30 seconds. It is believed that water, accelerated by the etch, dissolves
the SrO termination and leaves a TiO2-terminated surface [152]. This procedure was

followed by 3 h annealing in a tube furnace at 925 °C in air to allow reconstruction of the
surface [152,153]. After annealing, any remaining particulates on the surface were

removed by solvent processing.

Figure 5.1 shows AFM images of a (001) SrTiO3 surface after annealing and

cleaning. Figure 5.1(a) shows a terraced surface with terraces separated by steps of 1 unit

cell height (0.391 nm). Terraces appear because this crystal surface is miscut by 0.2 deg

57

from the (001). Figure 5.1(b) shows a perspective view of region (a). Figure 5.1(c) shows

a linear scan along the line shown in (a) showing the regular atomic-height steps.

 

5nm

(c1 <otlg>

Figure 5.1 (a) AFM image showing the topography of the SrTiO3 surface after annealing
and reconstruction; (b) perspective view of the first image; (0) scan along the line shown
in (a) showing the regular steps of atomic height corresponding to the lattice parameter of
SrTiO3.

<100>

0 250nm
_

5.2 DC magnetron sputtering of iridium

Sputtering is a process in which positive ions extracted from a plasma impinge on
a target, eroding small fragments of the target material which then deposit on an adjacent
substrate. If the sputtered material has sufficient mobility on the substrate, it can form a
polycrystalline or single crystalline film. To produce epitaxial (001) Ir, it is necessary to

sputter onto a heated substrate to suppress the formation of (1 1 1) Ir.

58

5.2.1 Sputtering system

5.2.1.1 Instrument design

Ir films were prepared with a three-gun magnetron sputtering system (APX
Instruments) that can be configured for DC or RF sputtering. Power supplies provide a
maximum of 1000W for DC and 500W for RF (13.56 MHZ) sputtering. The three guns
are arranged in confocal geometry, i.e., they are focused onto a central sample stage. This
alignment is advantageous for co-sputtering, or for sequential sputtering using multiple
targets. The sample holder stage can be rotated to improve homogeneity of the deposit
while sputtering. A contact-less SiC heater allows the substrate to be heated to
temperatures as high as 1000 °C. A separate load-lock is used for the insertion of samples
without the necessity of venting the main chamber to the atmosphere. The system is
operated and controlled by a LabView-based interface, developed for this research. A

schematic drawing of the system is shown in Figure 5.2.

59

 

 

 

 

 

 

 

11

 

 

 

 

 

 

 

 

 

 

 

 

 

1. DC magnetron gun 9. Ion gauge 17. Turbo-pump

2. Pyrometer 10.Vent valve 18. Turbo-backing valve
3. Ion gauge 11. Roughing valve 19. Movable bellow

4. Capacitance gauge 12. Gate valve 20. Feed through flange
5. RGA 13. Stage 21. Roughing pump

6. RF magnetron gun 14. Butterfly valve 22. Roughing pump

7. Target shutter 15. Turbo-pump 23. Gas tanks

8. Gas valve 16. Flow controllers

Figure 5.2 Schematic of the three-gun APX sputtering system

5.2.1.2 Vacuum and gas handling

The vacuum system consists of two turbo-molecular pumps (Pfeiffer) and
mechanical backing pumps (Alcatel) attached as pairs to load-lock and main chamber. A

large gate valve separates the load-lock from the main chamber. The minimum base
pressure in the main chamber is 5 x 10—9 Torr after baking for 24 h. A residual gas

analyzer (RGA, Ferran) measures the individual partial pressures of residual gases and

60

also serves as a leak detector. The gas manifold has three mass flow controllers (MKS
1249), which allow the introduction of gases for sputtering. Only pure Ar is used for Ir
sputtering. The gas lines are kept pressurized to avoid leaks fi'om outside. A butterfly
valve before the turbo pump is used to modulate the amount of gas exhausted through the
turbo-pump. The system pressure during the sputtering process is controlled by regulating

the Ar flow via a flow controller.

5.2.1.3 Sample stage, sample heating and temperature measurements

A schematic of the sample stage and heater assembly is shown in Figure 5.3. The
sample holder assembly is designed so that a carrier, consisting of sample holder base
plate and the holder, can be transferred from the load lock to the main chamber using a
transfer arm. This requires that the stage height be varied to accommodate the fixed
transfer arm. This is accomplished by a motorized stage with i 7.5 cm travel. A 12.5 cm
dia, 2 Q, SiC heater is used to heat a 7.5 cm dia stainless steel sample holder. Heating is
by radiation and the maximum temperature attainable is 1000 °C. The heater does not
contact the stage, since it must be free to rotate. Because the heater may draw as much as
15 A, water-cooled leads are required. Two radiation shields were built for thermal
radiation trapping to improve heating efficiency. A double-layer circumferential stainless
steel shield surrounds the heater assembly while a double-layer Mo shield sits below the
heater on the water-cooled stainless steel reflector plate. A K type (Alumel-Chromel)
thermocouple tip is attached to the inner side of the stainless steel radiation shield for
temperature measurement. Since the thermocouple does not contact the sample holder it
is not expected to represent accurately the sample temperature. Instead, sample

temperature is measured with a single-color pyrometer (Ircon) mounted outside the

61

vacuum chamber above the sample holder. The sample is viewed through the top quartz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

window.
> 5
8 ‘4 4
7
U T
L _
.3 L. h
E: a
:2 3
1. Substrates 5. Stainless steel radiation shield (1 of 2)
2. Stainless steel sample holder 6.Water-cooled stainless steel reflector
3. Hastalloy disc holder 7. Molybdenum radiation shields
4. Stage supporting rods 8. Silicon carbide heater

Figure 5.3 Schematic of the sample stage of sputtering system. Supporting rod can rotate
along with the sample holder. The heater and the cooling lines are fixed.

5.2.2 Sputtering procedure

An abbreviated procedure for sputtering Ir is given here. During sputtering, the
deposition rate, sputtering current, sputtering voltage and the thermocouple temperature
are monitored by a LabView program. The sputtering power, shutter and valve positions,

gas flow, and heater power, are also controlled and monitored by the software.

5.2.2.1 Substrate loading

A 7.5 cm dia stainless steel substrate holder, that can accormnodate 25 5 x 5 mm2

or 16 10 x 10 mm2 substrates, was designed. Substrates are placed in shallow depressions

62

recessed by 0.5 mm to prevent them from moving. A Si reference substrate, partially
covered by a stainless clip, is attached at an empty slot. It is used to record film thickness,

which can be measured by AF M or surface profiler.

While in the cleanroom, freshly cleaned substrates are loaded into the stainless
steel holder. The substrate holder is then wrapped in aluminum foil to avoid

contamination, taken to the sputtering system, and placed onto the load lock transfer arm.
When the load lock pressure falls below 2x10'7 Torr, the gate valve to the main chamber

opens, and the transfer arm inserted until the substrate holder assembly is positioned just
above the stage. The stage is then slowly raised using the stage motor. If the alignment is
correct, the stage pushes up the assembly and releases the transfer arm. The transfer arm
is then removed from the main chamber, the gate valve closed, and the main chamber is
further evacuated. The stage is then lowered to its sputtering position, where the target-

substrate holder distance is about 24 cm.

5.2.2.2 Sputtering process

For Ir sputtering, the substrate holder is heated to 850 °C, as measured near the
middle of the substrate holder. This is accomplished by increasing the heater power to
about 60% of its maximum power over a period of 1 h. Once this temperature is reached,
the shut-off valve is opened, the butterfly valve is set to its minimum aperture position,
and Ar gas introduced at a flow rate of 6-7 standard cubic centimeters per rrrinute (sccm).
When the pressure is 12-14 mTorr, the DC power supply is set to 100 W and turned on.
At this point, plasma should appear on the target. The flow rate is adjusted to maintain

the pressure setpoint. Once sputtering current and process pressure become stable, the

63

target shutter may be opened, beginning the deposition. The film thickness is monitored
by a calibrated quartz thickness monitor inside the main chamber. When the required
thickness is reached, the shutter is closed, sputtering power is turned off, the gas shut-off
valve is closed, Ar gas flow turned off, butterfly valve is set to maximum aperture
position, and the heater is slowly ramped down. After about 2-3 hours when the samples
have cooled to room temperature, the substrates are removed through the load lock and

stored in the cleanroom.

If the main chamber has been vented to atmosphere, it needs to be baked for 24 h
at 50 °C until the base pressure reaches 2x10'8 Torr, before sputtering is attempted. If the

system must be brought to atmospheric pressure, it is vented with nitrogen to minimize

the introduction of water vapor.

5.2.3 Ir growth on SrTiOa

The main goal of Ir growth is the deposition of an epitaxial (001) Ir film.
Departure from epitaxy is assessed by examining the ratio of Ir (002) to (111) x-ray
intensities. In addition, a small value of the (002) x-ray linewidth is beneficial for good
single crystal diamond growth. In general, the diamond linewidth can not exceed the

linewidth of its substrate.

The sputtering parameter ranges investigated here were temperatures between 730 to 930
°C and fihn thicknesses between 85 to 850 nm. Ar pressure and flow rate were always
held at 12 mTorr and 6 seem, respectively. During the sputtering process, the temperature
measured by the pyrometer (focused on the substrate) dropped by 100 °C after the

initiation of Ir deposition. Since it is unlikely that the substrate could cool this much, it is

64

surmised that the pyrometer temperature is representative of the substrate holder
temperature, and not the substrate, for the following reasons. Prior to sputtering, the
pyrometer would be more sensitive to radiation from the holder, which passes through the
transparent substrate, if the substrate were cooler than the holder. Once sputtering began,
the apparent temperature dropped because the intervening Ir fihn blocked radiation from
below so that the pyrometer read the Ir surface. The Ir and substrate were actually 100 0C
cooler. On the other hand, when the pyrometer was focused on the sample holder, the
temperature remained the same before and during sputtering. This is most easily
explained in terms of poor thermal contact between the holder and substrate with a

temperature difference of about 100 °C.

XRD analysis included a 0-20 and rocking curve scans of the Ir film measured
with a Bruker AXS x-ray diffractometer using Cu K, radiation (0.154 nm wavelength) at
40 kV accelerating voltage and 40 mA emission current. The (002) Ir linewidth was

found by fitting a Gaussian function to the rocking curve data. The Gaussian function

exp(-x2/2crz) has a full-width at half-maximum (FWHM) or linewidth equal to 2.3540.

The surface topography and average roughness of the Ir films were studied by

AF M in non-contact mode.

5.2.3.1 Topography and structure of the Ir films

Figure 5.4 shows the topographies of 500 nm thick Ir films grown at four different
temperatures between 730 and 930 °C. In each image, unusual stripe-like structures can
be seen aligned along [110]. The reason for this anisotropic morphology is not presently

clear.

65

500 nm

 

Figure 5.4 Topography of 500 nm thick Ir films on SrT103 grown at different
temperatures (a) 730 °C (b) 800 °C (c) 850 °C (d) 930 °C. The fuzziness in (d) is a
measurement artifact.

 

0.40 . . . . *fi , . . . 0.40
0.35- 0.353
A a
g 3
'0 3
:030- —030:
D
F a
E o
2 8
-_-, 0.25— 4253-
3.

020- l I l -020

 

 

 

§ .

700 I 750 300 ' 850 I 900
0
Temperature ( C)

Figure 5.5 Variation of (002) linewidth and mean roughness of the sputtered Ir films
with temperature

The 730 °C grown Ir film is quite rough. With higher growth temperatures, the Ir surface
becomes smoother. Figure 5.5 Shows the variation of average roughness with growth
temperature. The lowest roughness occurs at 850 °C. Figure 5.5 also shows a plot of the Ir
(002) x-ray linewidth versus substrate holder temperature. The linewidth is a minimum

for the film grown at 850 °C.

Figure 5.6(a) shows a 0-20 x-ray scan of a 600 nm thick Ir film grown at 850 °C.
The Ir (002) reflection, which occurs at 20 = 47.3 deg, is quite strong. Although the Ir
(111) reflection has twice the intensity of the Ir (002) reflection, this scan shows no
detectable Ir (111) intensity at 20 = 40.6 deg. Here, the ratio of background noise
intensity to (002) Ir intensity is 0.0002. Note the logarithmic intensity scale. Some Al
reflections, denoted as “Al”, come from scattering by the sample holder and not the

sample. Since no monochromator was used here, additional peaks due to Cu K3 radiation

appear for Ir and SrTiO3.

Figure 5.6 (b) shows (002) Ir and (002) SrTiO3 reflection regions taken in

identical conditions. Linewidths were found by fitting a Gaussian function to data for
SrTiO;; and Ir with values 0.030 i 0.002 and 0.16 i 0.01 deg, respectively. The larger Ir

linewidth is due to interfacial stress and dislocations. This linewidth is comparable to the

previously reported linewidth for evaporated Ir on SrTiO3 [9,56]

67

 

 

 

   

 

      

 

 

 

 

 

 

3‘ ' 5’ g “A E 1 Experiment 61‘ .
g i '5 ég I g - - Gaussianfit 8, -
E“ i A .9: 3 E‘ _ :
g ... 22 “8’ m
E i § 3 v 1 g - .
' V O 1 _ " .-
5 : o g g 5 : 'E . .
g r 5 9i - i 5 '
I N x ‘ . q
5 5 m8 gs? 5 E.
T: A W " -i
5 I - a ' = . .
E- B: I! 8 " «I
2 2 2 ”3- : E , ‘
20 40 60 80 100 23.2 23.4 23.6 23.8 24.0
200109) 0(deg)

Figure 5.6 (a) 0-20 scan of a 600 nm thick Ir film on SrTiO3 taken using Cu K,l radiation.

Note the log intensity scale (b) rocking curve of SrTiO3 and It (200) reflections with
linewidths 0.03 and 0.16 deg, respectively.

To confirm the in-plane epitaxial relation between Ir and SrTiO3 a (111) pole-

figure, out-of-plane scan of (001) 1r on (001) SrTiO3 was performed. The substrate was

mounted on the Bruker chuck with two lateral sides of the substrate vertical. The x angle
was set to 54.7 deg, the angle between (001) and (111) planes. The detector and the
sample were fixed at the (111) reflection position at 20 = 40.6 deg. The crystal was then
rotated about the (0 axis (the axis perpendicular to the (001) plane of the crystal) from 0 to
360 deg to bring a (111) plane to the Bragg condition. Four (111) reflections were

observed as shown in Figure 5.7. The Ir and SrTiO3 (111) reflections occur at the same

angles, demonstrating that their <110> directions are parallel, Thus, the epitaxial relation

between Ir and SrTiO3 can be stated as (001)1,<110>1,||(001)smo,<110> 5mg.

68

 

A
N

(111) (”'11) (1'13) (111)

Ir

W W

 

 

 

 

Intenslty (arbitrary units)

A
U'
v
I V V‘VU'

SfTiO3

 

 

 

 

 

lntenslty(arbltrary unlts)

 

AAAAAAAAAAAAAAAAAAAAAAAAAA J_I_A444A_L

0 50 100 150 200 250 300 350
Phl (degree)

Figure 5.7 X-ray (111) pole figure (a) Ir (b) SrTiO3. The (111) reflections fi'om Ir and
SrTiO3 are at the same angular positions confirrrring the in-plane epitaxial relation
between the two.

5.3 CVD reactor process

Heteroepitaxial diamond growth was carried out in the microwave plasma CVD
reactor shown in Figure 5.8. It is powered by a 2.45 GHz microwave source capable of
producing 6 kW. A waveguide connects the source to a cylindrical brass microwave
cavity via a tunable electrostatic coupler. The vacuum and gas handling systems include
mass flow controllers, a pressure controller, and a turbo-molecular pump backed by a

roughing pump. A LabView system monitors biasing current, pressure, and temperature

69

and controls the process sequences. It shuts the system down if pressure and reflected

microwave power exceed limiting values.

 

 

 

Power supply

 

 

Optical

    

l
Tiermocouple
readout

~—

 

Btcitation
Robe

  
 

 

PC
Control

 
  
 

Quartz
dome

Gas
valve

 

 

 

 

 

 

 

 

 

 

 

 
  
 

 

 

 

l :ffrrl Vacuum
l T""' ’ Chamber
L 0-300VDC
RJWOI'
amply

 

 

 

 

 

3 Ar (vent)

- Exhaust

 

 

 

 

u
Nitrogen purge

Figure 5.8 General schematic of the microwave plasma CVD reactor used for
heteroepitaxial diamond growth [140].

5.3.1 Substrate stage and bias set up

The substrate stage, sample holder, and bias ring, within the CVD reactor, are

shown in Figure 5.9. The stainless steel stage is electrically isolated from the lower stage

plate by a fused quartz tube. A bias ring is supported above the sample stage assembly by

Mo posts. During biasing, the lower part of the stage and the bias ring are grounded,

whereas the components in contact with the sample stage can be placed at an arbitrary

potential with respect to ground.

70

Molybdenum
SJbstrate B73 ring

G“

 
  
  

   
      

Basn'ng

 
  
       
 

Sandoff \* V ‘ ‘ a Sage
. 1.9) ;~ /
Molybdenum // EV;
fiQuartan
Sample cap Iva F forelectricgal
M inlation

 

Figure 5.9 Schematic of the sample and bias set up for heteroepitaxial diamond growth

A Kepko DC power supply, 0-300V/0-200mA, generates the bias voltage. Two 50
Q resistors are connected in series with the power supply, one at the positive and one at
the negative terminal. The voltage drop across one of the 50 Q resistors is measured by a
Fluke data logger and used to monitor the bias current. The circuit is shown in Figure
5.10. A circular Mo wire ring is supported by three Mo posts which are grounded at the
lower stage assembly. The bias ring is in contact with the microwave plasma and keeps

the overall resistance of the bias circuit near 3 m for typical operating conditions.

71

 

Isolated stage
\

 

 

 

300V/200mA DC

Figure 5.10 Electrical circuit used for biasing.

5.3.2 Substrate holder

The molybdenum substrate holder consists of a post and a cap, Shown in Figure
5.11. The post is attached to the sample platform with a threaded rod. The substrate sits
on the post and is covered by the cap. The standard design uses a 5 x 5 mm2 substrate
with 3.5 mm dia circular aperture and 8 mm dia Mo post. To grow diamond on a 5 x 5
mm2 square substrate, a 10 mm dia cap is used with 6 x 6 mm2 square aperture. The
sample is centered on a 9 mm dia post such that the edges have equal gaps. To prepare an
8 mm dia diamond sample on a 10 x 10 mm2 substrate, a 14 mm dia Mo post with a 15

mm dia cap having a 8.5 mm dia aperture is used.

72

 

A. Mo cap
B. Mo post
C. Ir/oxide substrate

Figure 5.11 Schematic of sample post and cap with sample stage

5.4 Diamond nucleation

5.4.1 The bias process

To induce diamond nucleation on Ir, a negative bias voltage is applied to the
substrate in the presence of the microwave plasma. The variables explored were: bias
voltage Vb (fixed at -150V here), bias duration tb, substrate temperature Ts (fixed here),
methane concentration r (fixed here) and the geometry (fixed here). The diamond
nucleation (and growth) process sequence is shown in Figure 5.12. To begin, pure
hydrogen is introduced into the reactor bell jar and, at 10 Torr pressure, a plasma ignites
at 700 W incident microwave power. At 15 Torr, the incident microwave power is
increased to 1100 W. The sample temperature Ts increases rapidly and after 6 minutes
reaches about 650 °C, where the pyrometer begins to register. Then, methane is

73

introduced and allowed to react with the sample; this step is referred to as carburization.
After 12 minutes of carburization, the bias voltage is switched on for a time tb. At tb, the
bias is switched off and growth proceeds for a time tg. At tg, the microwave power is
turned off and gases are pumped out of the system.

The initial ramp-up and carburization time were not varied: 6 minutes of heating

was followed by 12 minutes Of carburization. For the zero-min biased sample, the plasma

was turned off after carburization without turning on the bias voltage

 

 

9’ S
c ._ _=
-- or
§ E s E
c '= a e
3 _
.72 e m 0
9: a
5 o .- », . . ,.
..- 710°C -.Tsnpsraturs
650°C
I 2% I Methane concentration

 

 

o v I I I Time (min)

0 6 18 23-78

Figure 5.12 Time plot of the process sequence used in studies of diamond nucleation and
growth

Throughout the process, heating of the sample occurred only by contact with the
plasma; no external heater was used. In some cases, the bias and growth processes
utilized pressures and methane concentrations that slightly varied from the

aforementioned processing steps.

74

Two sets of experiments were carried out to understand the effect of the bias time

duration tb, on Ir morphology, nucleation rate, density and spatial distribution of nuclei.

Another experiment examined the effects of varying tg,, the growth time, at fixed tb,

In the first series of experiments, tb was varied from 0 to 60 min, followed by
abrupt turn-off of the microwave plasma and bias to minimize diamond growth
after biasing. These runs are referred to as 0+ growth runs, since cooling is not
instantaneous and some growth may occur. The main goal of this part of the study
was to observe how the Ir surface evolves after it interacts with the plasma for

different times.
In the second series of experiments, tb was again varied. At tb the bias voltage was

switched off, but then followed by 20 minutes of growth. The finite growth period
allowed diamond nuclei to grow to a size that could be imaged by SEM. This

allowed us to estimate the density and distribution of the nuclei.
In the third series, tb was fixed at 60 min but the growth time was varied from 7 to

150 s. The reason for these experiments was to observe the early stages of

diamond growth and coalescence.

The diagnostics used were AFM and SEM. We also have used HRTEM on thin

sectioned samples to study the diamond-Ir interface.

5.4.1.1 0+ growth experiments

In this series, tb was varied from 0 to 60 min then the plasma quickly turned off.

Figure 5.13(a) shows an AFM image for tb = 0 min, i.e., a carburized sample. It cannot be

75

distinguished from the as-grown Ir surface, Figure 5.4. The Ir domain structures are
similar. It should be noted that the domains are not evident by SEM, i.e., the Ir domains
do not produce electron contrast. Average roughness (Ra) measured over an area 1x1 umz

was 0.4 nm, comparable to 0.3 nm for fresh Ir. Thus, carburization has little effect on the

Ir surface.

 

      

-., ". "~. 500nm " . 500 nm 1

Figure 5.13 AFM images of the Ir surface after different biasing times (a) tb = 0 min.
little change is seen from the as-prepared surface. (b) tb = 5 min, the surface is covered
with dots and chain like structures. (c) tb = 20 min, middle, the Ir surface is covered by
dots separated by a mean nearest neighbor distance of 15 nm. About 1 % of the dots are
4-5 nm taller than average height. ((1) tb = 20 min, edge a two phase region appears. The
upper part of the image shows chain-like structures with mean distance between the
chains 15 nm. The lower part of the image has no chains or dots but the step edges are
corrugated on a scale of 15 nm.

76

Figure 5.13(b) shows an AFM image for tb = 5 min, taken in the middle of the

Ir/oxide substrate. In contrast to (a), the surface is now covered by small round dots. They
are preferentially aligned along the [110] Ir step edges. In some regions, instead of dots,
elongated islands connect to each other to form a chain-like structure. The large-scale
structure of the Ir step edges is still visible.

Figure 5.13(c) shows an AFM image for tb = 20 min taken at the middle of the
substrate. Now, the step edges are nearly erased and the surface is covered with dots,

separated by 15 nm on average. The dot density is ~1012 cm'z. The white dots are about

4-5 nm taller than the typical dot, with density ~1010 cm'z. Since the AFM image is

formed by the convolution of the tip and dot shapes, the dots are actually smaller than
they appear in the images.

The previous images were taken at the middle of the substrate. However, when
the edges of the sample were examined, a significant difference in topography was seen.
Figure 5.13(d) shows an AFM image taken near the edge of the biased region of the

sample where a two-phase topography occurs. In the upper part of the image, elongated
dot-like structures connect to form [110] chains, similar to, but more distinct than, the tb
= 5 min sample, Figure 5.13(b). The mean inter-chain distance is 15 nm, the same as the
mean distance between nearest neighbor dots in the middle of the sample. The surface in
the lower part of figure 5.13(d) is not decorated with chains, but is similar to the tb = 0
min sample. The Ir step edges show corrugated structures, separated by a mean distance

of 15 nm, the same as the mean distance between nearest neighbor dots at the middle, or

the mean inter-chain distance in the decorated regions at the edge.

77

tb = 5 min
middle

 

‘ x. :11,"-
_ 2,... “Nam.
Figure 5.14 SEM images of the Ir surface afier biasing for different durations. (a) tb = 5
min; a few bright patches on the surface have appeared (b) magnified image of one of the
bright patches shows internal structure(c) tb = 20 min; sample middle shows uniform
contrast.(d) magnified image of (c) shows the surface to be covered by dots with 15 nm
mean separation (e) tb = 60 min; sample middle shows uniform contrast (0 Magnified
image of (e) shows the surface covered by dots as in (d).with same separation (g) t1, = 20
min; sample edge shows isolated bright patches on um scale. (h) Magnified image of the

bright island shows structures similar to the chains observed in AFM image in Figure
5.18(d)

78

The chain-covered region is about 1-2 nm higher than the smooth region, and the
chains are also 1-2 nm higher than their surroundings. These observations suggest that
biasing initially leads to etching of Ir step edges in the higher regions of the substrate,
creating chains, which then break up into dots under prolonged exposure to the plasma.

Figure 5.14 shows SEM images of samples produced with different bias
durations, which varied from 5 to 60 minutes. Figure 5.14(a) shows an SEM image for tb
= 5 min at the middle of the sample. Small bright patches are observed on the surface of

the entire sample. Figure 5.14(b) shows the magnified image of one of the bright patches

seen in (a), which shows some internal structure.
Figure 5.14(c) shows the middle region of the tb = 20 min sample. At this scale,

the surface appears uniform, with little structure discemable. The magnified image of this
region, Figure 5.14(d), shows, however, that the surface is covered by dots with 15 nm
mean nearest-neighbor spacing. Both AFM, Figure 5.13(c), and SEM provide images of

the same dot distribution.

Figure 5.14(c) shows the middle of a tb = 60 min sample. This image is similar to

(c), the tb = 20 min sample. In 5.14(1), the magnified image from (e) shows a surface

covered by dots with 15 nm mean nearest neighbor spacing. Samples produced within a

tb = 20-80 min window show similar morphology.

Figure 5.14(g) shows an image taken near the edge of the biased region for tb = 20
min. This image shows bright isolated islands on the scale of um. Their contrast seems to
be similar to that of the bright patches observed for tb = 5 min as shown in Figure 5.14(a).
Figure 5.14(f) shows the internal structure of the bright islands. The morphology is

79

similar to the chain structures observed in the AF M image shown in the upper region of
Figure 5.13(d). The agreement between the AF M and the SEM images suggests that SEM
contrast arises from topographical differences, i.e., the bright islands seen in the SEM
image are locally high regions, and not from chemical differences. Since the undecorated
regions here are similar to the 0" min biased sample, it suggests that these regions are
etched less during the biasing. The etching of the Ir is thus not spatially uniform. Etching
starts at locally high regions then propagates across the surface.

Afier 5 minutes of biasing, it is observed that the secondary plasma, initially
centered over the circular cap, has become brighter at the middle of the sample. Because
of this higher secondary plasma density, the middle of the sample now etches faster than
the edges. As the biasing progresses, the secondary plasma gradually moves toward the

edge of the sample, increasing the etching rate there. This creates the difference in
surface topography seen, for example, at the middle and edge of the tb = 20 min sample.
Thus, the region at the edge is etched for an effectively shorter time than at the middle.

We infer that the edge regions are then similar to the central regions at an earlier time of

the biasing period.

HRTEM imaging was used to study the diamond-Ir interface on a scale as small
as 0.1 nm. This technique requires the preparation of thin sections perpendicular to the
interface using mechanical and ion milling. To protect the surface, an adhesive of Pt
nano-particles mixed with epoxy was deposited on the surface. This glue appears in the

HRTEM images and must be distinguished from the diamond-Ir sample.

Figure 5.15(a) shows a cross-sectional HRTEM image of the diamond-Ir interface

for tb = 10 min and tg = 20 min. Of interest here is the structure of the Ir. The Ir surface

80

has faceted pillars with mean nearest neighbor spacing 15 i 0.5 nm. These pillars are
clearly the dots observed in AFM and SEM. Lateral sizes of the pillars, measured at the
middle, vary from 5 to 12 nm, with a mean of 10 i 0.5 nm. The mean height of the pillar
is 4 :1: 0.5 nm. The dimensions and the errors are calculated from the HRTEM images

shown here.

Figure 5.15(b) shows an HRTEM image for tb = 60 min and tg = 0 min. The Ir
shows faceted pillars with a mean spacing 14.6 i 0.5 nm. Sizes of the pillars vary from 5
to 10 nm with an average of 8 i 0.5 nm. The mean height of the pillar is 4.5 i 0.5 nm.
This may be evidence that a longer tb leads to a slightly smaller pillar width. The results

for the pillar spacings are consistent with AF M and SEM.

Iridium

 

Figure 5.15 (a) HRTEM images of the Ir surface with (a) t1, = 10 min, tg =20 min (b) tb =
60, t8 = 0 min [154].

5.4.1.2 20 minutes growth after biasing

We now discuss experiments with variable tb followed by ti; = 20 min. SEM

images taken at the middle of tb = 5, 10 min and at the middle and edge of tb = 20 min

81

diamond films are discussed. This yields the spatial distribution of diamond on the Ir

surface for different tb, whose precursor is the distribution of diamond nucleation sites.

 

Figure 5.16 SEM images of samples with different tb followed by t8 = 20 min (a) tb = 5
min; film shows sparse diamond crystallites (b) tb = 10 min, middle; film is mostly
covered by epitaxial diamond except for some coalescence boundaries. (c) tb = 20 min,
middle; film is similar to (c) with better coalescence (d) tb = 20 min, edge; sample shows
diamond growth in mm size isolated islands

For tb = 5 min and tg = 20 min, the surface in the middle has a relative coverage of
about 20% epitaxial diamond, Figure 5.16(a). When the bias time is lengthened, for tb =
10 or 20 min, epitaxial diamond crystallites have grown and coalesced over most of the
surface with a relative coverage of over 90%. Some coalescence boundaries can be seen
in Figure 5.16(b) and (c), respectively. In contrast to the middle region, for tb = 20 min,

diamond grew as isolated islands at the edge of the film, Figure 5.16(d). These islands

82

apparently correspond to the bright areas seen in SEM after tb = 20 min and tg = 0",

Figure 5.14(g), since the area] coverage in both figures is about the same. These results
indicate that the etched regions provide the conditions for diamond nucleation. It is noted,
however, that the etched structures in Figure 5.14(d) and (1) lead to well-coalesced
diamond, whereas the structures in Figure 5.14(h) yields islands of diamond with fine

inter-granular boundaries for equivalent growth time.

Figure 5.17 combines these results in a plot of diamond areal coverage vs. radial
distance from the edge of the sample. The growth region is 3.5 mm dia on a 5 x 5 mm2
substrate. Each point in the plot is derived from a separate SEM image. The Image J
package [155] was used to analyze the areal coverage of 20 x 25 um2 SEM images. Since
the analysis was made from an image with low magnification where the coalescence
boundaries were not visible, the calculated value of the areal coverage is an overestimate.

For example, using SEM image in Figure 5.16(b), the areal coverage is 96 %. With the

lower magnification image, where the boundaries are not visible, the coverage is 100 %.

For tb = 5 min, only 20 % of the surface is covered by diamond at the middle. For tb = 10
min, most of the sample is covered with epitaxial diamond except for a narrow band near
the edge. Increasing tb to 20 or 60 min increased the coverage at the edges. These results

suggest that the modification of the Ir surface due to the biasing starts from the middle

and propagates toward the edge.

83

 

 

3

I
. Elf-lI-IIIII'III.II-./
I

O
'1 V U
1

E3

3100 - : i c 7 : ¢ -
C ’ .
o : l
E l. . ' . .
g 80 _- Blastlme (min) 1
_ . _._ 5 .
E I

a 60 __ -0—10 _,
'a : -‘- 20 4
o I —v-60 4
> 40 " -i
.o . .f'" 1
r - 4
° -
é

a

2

<

 

 

. I l l L I I I. I I I I . I I I I L I I I I l I I I I I I A.

A 0.0 0.2 0.4 0.6 0.8 1.0
Normalized distance from the edge

 

Figure 5.17 Plot showing % area covered by epitaxial diamond as a function of distance
measured from the edge of the film for different biasing time. Higher biasing time
increased the area covered by diamond at the edge.

5.4.1.3 60 minutes biasing followed by short growth

We now describe the experiments with tb = 60 min in which tg varied from 7 to
150 s, Figure 5.18(a) to (1). Here, we expect the Ir surface to be fairly uniformly covered
with dots after bias. As shown in Figure 5.18(a), after t8 = 7 s Ir dots continue to be

clearly visible. Note the simultaneous presence of diffuse patches. As the growth time
increases, the dots become less visible and the cloudy, diffuse regions become more
prominent and compact. For tg > 60 s, the small, bright regions have become evident.
Two reasons for the enhanced brightness are: more efficient emission of secondary
electrons arising from the topography, or the large secondary yield of hydrogenated

diamond, a consequence of its negative electron affinity. Since HRTEM images showed

84

that the topography of the Ir surface is invariant during growth, we ascribe the changing

contrast after 60 s to the grth of diamond regions.

The interfaces between diamond clusters preferentially align along [110].
Although initially ascribed to faceting, it now seems likely to occur because diamond
grows laterally in the channels between the dots, assuming that diamond nucleates in the

channels but not on the top of the dots. The density of the bright regions in Figure 5.18(d)
is ~2x10ll cm'z, which is somewhat lower than the dot density. If there is little

coalescence at this point, each region originates from a single diamond nucleus. This
number may yield a more reliable estimate of the nucleation density. After 120 s, the
bright regions have grown larger, yet many small dots are still visible. These could be the
underlying Ir dots or correspond to fieshly nucleated regions. The image in Figure 5.18(1)
shows more compact bright regions then in Figure 5.18(e), although growth is somewhat

longer; this could be due to sample to sample variation.

85

 

Figure 5.18 SEM images of 60 min biased samples followed by short growth (a) 7 s (b)
30 s(c) 60 s (d) 90 s (e) 120 s (f) 150s. Each image is from a different sample.

5.4.1.4 HRTEM study of the diamond-Ir interface

Here we discuss the results of HRTEM studies for samples with tb = 60 min
followed by t8 = 0 to 150 s. HRTEM images are formed by electrons transmitted through
the crystal lattice. Contrast depends on the atomic number (Z) of the specimen, its

86

thickness, and the crystallographic orientation of the sample with respect to the incident
electron beam. Specimens with larger Z scatter electrons better, producing higher
contrast; however, thicker specimens absorb more electrons, producing lower contrast.
Because of the ensuing large dynamic range, imaging of diamond and Ir simultaneously

is difficult. Strain at the interface will also change the contrast.

Figure 5.19 [154] shows an HRTEM image for a tb = 60, tg = 0 min sample. The
(110) plane is perpendicular to the incident beam. Ir pillars and (111) lattice planes can
be seen. The lattice planes of the pillars and substrate are continuous, both of which
exhibit the Ir (111) interplanar spacing, 0.222 nm. This confirms that Ir is etched by the
bias process, leaving dots or pillars, with little deformation of the Ir lattice. No features
are observed here that can be associated with diamond nuclei or crystals. The crystalline
structures seen in some places above the Ir pillars are Pt nanoparticles embedded in an
amorphous epoxy matrix used for specimen preparation. The bright region in the
channels between the pillars, which is not Pt or epoxy, could arise fi'om carbon deposited
during the biasing, as previously reported [95,156].

rm». 1 W- em w.-.-
‘ .'- I“. -u ..." ‘ . ..ILI
,_. . _ ~ . . x _ .' n
. *" .,./ ‘. ‘ ~.I,
' I
..

       
 

   

6 “I

Iridium

Figure 5.19 HRTEM image of Ir surface for a tb = 60 min, tg = 0 min sample showing the
Ir pillars [154].

87

Figure 5.20(a) shows a HRTEM image for tb = 60 min and tg = 903. Ir pillars can

be seen, comparable in size and distribution to the tb = 60 min and tg = 0 min sample,

discussed earlier. Here, the interfacial carbon layer is more clearly seen.

Figure 5.20(b) shows a magnified region from (a). Although diamond crystallites
are not resolved in (a), here, two diamond crystallites can be seen in the channels
between the pillars. The (111) interplanar spacing for Ir and diamond are 0.222 and 0.206
nm, respectively. Owing to the lattice mismatch, the planes cannot be aligned everywhere
at the interface, and misfit dislocations occur. With a mismatch of 7.5 %, there should be
misfit dislocations with an average period of 13-14 lattice spacings if the dislocation
Burgers vector is inter plane spacing. The distance of the misfit dislocations from the
interface is determined by a balance between the pseudomorphic interfacial strain energy
and the energy of dislocation formation [157]. Elasticity theory predicts that the critical
length for misfit dislocation formation should be 0.2-0.3 nm from the Ir-diamond

interface. The HRTEM observations are consistent with the theory.

Misfit dislocations are clearly visible at the Ir-diamond interface shown in Fig.
5.20(b). No intervening phase can be seen between diamond and Ir, which means that
diamond has directly nucleated on the Ir surface. No particles are observed on top of the
diamond crystallites, other than the amorphous carbon layer deposited during biasing.
The larger diamond crystallite is about 5 nm by 4 nm and the smaller crystallite about 3
by 3 nm. In figure 5.20(c), blue lines are drawn to show the corresponding (111) Ir and
diamond lattice planes. A misfit dislocation is evident in the middle of the particle.

Diamond crystallites of this size were only observed in the channels between Ir pillars or

88

on the sides of the pillars in images of this sample. Blue lines indicate corresponding Ir

and diamond (11 1) planes

‘ . \ “‘4 ‘..~") "1; ixi,"'1._-.'-.-‘.. '. '3 "'v‘.‘,,"‘c:. twivx “ - ‘~ . :1”;-~ ..

f , “.1 , ,-_ 1. 5111‘, *3". ...-ma.”
.z 1?) ,~.-: 5‘;‘§?71;;x..>>W‘>4‘..i . vi: M33555, -, . ”P ~ ... . "
' i . o a" ' ' '4 V ' I ‘ W '

. l

'I

.2 ‘ . l' . \ \ ‘
a I ‘ ‘N \
:. :\-‘-; . . '

\A 1‘: lrlqyi‘\\\\\\\

 

\\\
\\

Diamond"

   

     
 

\ :'t \ \
. \
\ 3‘?"

Figure 5.20 Cross section HRTEM images for a tb =60 min, tg = 90 s sample. (a) Carbon-
Ir interface showing a bright layer, possibly amorphous carbon. (b) Two diamond
crystallites in the channels between the Ir pillars. (c) Misfit dislocation at the diamond-Ir
Interface [154].

89

Isolated diamond crystallites were not observed on the top (001) Ir surfaces. This
observation lends credence to the idea that etching of the Ir surface generates nucleation

sites that are absent on Ir (001).

Figure 5.21 [154] shows a high-resolution image of the diamond-Ir interface for a
tb = 60 and tg = 150 5 sample. For this slightly longer growth time, a diamond crystallite
has now overgrown the top of an Ir pillar. The blue lines indicate the top boundary of the
diamond crystallite, whose size is about 15 nm laterally and extends 3-4 nm above the Ir
pillar. Two blue arrows show the positions of misfit dislocations about 14 lattice spacings
apart. The Ir lattice is distorted close to the dislocation core due to the large local stress.
Since there is no slip apparent at the interface, it suggests strong bonding between Ir and

diamond.

‘= . '
\h.‘.fi 10"
-* t ,

n
I.
\

 

Figure 5. 21 HRTEM image for tb = 60 min, tg= 150 s sample. This image shows a
diamond crystallite over-growing on top of an iridium pillar. Blue line shows the top
boundary of the diamond crystallite. Blue arrows show the positions of misfit
dislocations [154].

90

Figure 5.22 [154]is an HRTEM image for a tb = 10 and t3 = 20 min sample. This
image shows the diamond-Ir interface after longer growth. Faceted Ir pillars are visible
with typical spacing. Bright lines, which appear to radiate from the Ir pillars, are
observed. Some of these features, at a 60 deg angle from the surface, are most likely twin
boundaries. They may arise from different nucleation conditions on the opposite sides of

an Ir pillar or may originate at imperfections on the Ir surface.

 

Figure 5.22 HRTEM image of a tb = 10 min and tg= 20 min sample [154].

5.4.2 Discussion

The previous sections have explored the role of the bias process in modifying the
Ir surface. Under appropriate conditions, relatively dense, uniform, and highly correlated
Ir pillars are created. Subsequently, diamond nucleates and grows to cover most of the Ir
surface in about 20 min in the CVD reactor.

91

Under bias, ion bombardment etches the Ir surface initially along [110] step
edges, as suggested by AFM images. If the secondary plasma moves to the center of the
sample, strong etching of the Ir surface starts there after about 5 min, gradually
expanding outward, reaching the edges in another 5 min or so. The initial 5 min may be
associated with the time for the secondary plasma to move from the cap, where it first
forms, to the middle of the sample. A time-dependent secondary plasma front explains
how the etched structures at the sample center and edges are related: the edges are simply
etched for a shorter time, since the etching begins at the center. The movement of the

biasing front has been reported previously, for DC biased silicon [80,158]

We have also inferred that the etching of the Ir surface may be sensitive to the
height fluctuations of the surface. Higher regions are etched faster, possibly because of
the higher electric field at discontinuities or the higher temperatures closer to the plasma.
These effects can create the two-phase region, seen near the sample edges. At the middle,

the two-phase region is unlikely to persist more very long, owing to the faster etch rate.

The large increase in nucleation density that occurs after 5 min of bias suggests a
non-linear process. During the period from 5 to 10 min after the start of bias, the area
covered by diamond increased from 20 to 100 % in the middle and 15 to 20 % in the edge
of the sample. Again we infer that this reflects the dynamics of the secondary plasma as it

moves across the sample.

It has been proposed that Ir is sputtered during biasing, then redeposited, and
could be responsible for the Ir dots on the Ir surface [94]. However, if this were true, the
Ir particles would be randomly distributed on the surface and would not be expected to

self-organize. Also the surface would become more deeply etched as time progressed, in

92

contrast to the observations. The 15 nm length scale is apparently imprinted on the Ir

surface shortly afier 5 min bias, and varies little for bias times up to 80 min.

The bias process not only etches the Ir surface, but leaves a small amount of
amorphous carbon on Ir. XPS and XPD measurements estimated about 20 monolayers of
carbon after 60 min bias [91,92]. This carbon layer may serve a number of roles during

biasing and nucleation. As observed by HRTEM, the lateral size of the Ir pillars is
slightly larger for tb = 10 min than for tb = 60 min. However, in both cases the height and

the nearest neighbor spacing are comparable. This suggests that as biasing progresses
only the side walls of the Ir pillars are etched away. Two ideas may explain this
selectivity. As the pillared landscape is created, the biasing front may propagate to the
topographically higher un-etched regions. In addition, the deposited carbon layer may

protect the Ir from further etching.

If the carbon deposited during the biasing is predominantly sp3 bonded, it may
serve as a precursor for diamond nucleation. Since tetrahedral amorphous carbon ( ta-C)
is structurally close to crystalline diamond, only a small energy barrier separates the two
phases. The barrier could be overcome when a thermal spike is created when energy is
dissipated in the Ir lattice by a hydrocarbon ion extracted fi'om the plasma. Once a stable
diamond nucleus is created, any nearby ta-C would readily transform to diamond,

enhancing the grth of the diamond cluster.

From the HRTEM images presently available, diamond appears to nucleate preferentially
between, or on the sides, of the Ir pillars. No image shows nucleation on the top (001)

surface. Therefore a fimction of the bias process may be the creation of nucleation sites

93

on higher index planes or at special step edges. Alternatively, if the nucleation process is
due to ion subplantation, subplanted regions would need to organize into channels,
leaving intact Ir pillars. Subplantation models may invoke carbon saturation of the sub-
surface region during the bias process. Within the subplanted region, diamond nuclei may
form, effectively at high pressure due to confinement in the substrate. Subsequently, the
substrate surface is etched away, leaving the diamond nucleus [73]. This model is
consistent with observation, although it cannot be proven. Another possibility is that ion
bombardment leads to demethylation reactions and formation of special compounds at
the interface, i.e., an Ir hydrocarbide [159] as opposed to a monatomic ta-C layer. This
material could act to enhance nucleation or could serve as a protective layer to prevent
erosion of the Ir. At this time, no evidence exists that supports the presence of a chemical

species at the interface other than carbon.

XPD measurements have shown the presence of crystalline, epitaxial diamond on
the Ir surface after tb = 60 min and zero growth, but HRTEM does not reveal any
diamond nuclei [92]. The XPD result is not necessarily incompatible with our

observations, since the size of the nuclei may be too small at that stage to be resolved by

electron microscopy.

From the SEM images after 1508 growth and the HRTEM images after 908

growth, it seems that the nucleation density must be less than the density of the Ir pillars,
which is ~ 3 x 1011 cm'z' However, the HRTEM image region is a slice that cannot reveal

the 3-D environment of a pillar. Thus there may be additional diamond crystallites that

are not visible, leading to an underestimate of the nucleation density.

94

HRTEM images have reinforced x-ray data for the specific epitaxial relation
between diamond and Ir. The presence of the misfit dislocations at the diamond-Ir
interface confirms that diamond growth is not pseudomorphic. The critical distance for
misfit dislocation from the interface based on elasticity theory, which predicts 0.2 nm as

the critical distance, is in accordance with our observations.

5.4.3 Summary

Based on the findings presented in this section, we propose the following scenario

for the nucleation and the growth of diamond on Ir.

1. During biasing, ion bombardment etches the Ir surface creating first chains, then
pillared structures.
2. Strong etching begins at the center of the sample then propagates to its edges. The

dynamics are controlled by movement of the secondary plasma. This explains the

temporal and spatial patterns that are observed.

3. Diamond nucleation does not occur during biasing because of intense ion

bombardment. Nucleation is triggered at the end of the bias step.

4. Growth is largely determined by transport from the vapor phase. Epitaxial

crystallites start coalescing with one another after only 90 s of growth.

5.5 Diamond thin film growth

In this section we describe processes for producing diamond thin films. These

studies led to a focus on the role of sample holder geometry.

95

Typical growth conditions needed to produce a 500-600 nm diamond film, grown

over a 3.5 mm dia region on a 5 x 5 mm2 Ir/oxide substrate, are given in Table 5.1. The
variables are bias voltage Vb, bias ring height hb, bias current 11,, methane to hydrogen

ratio r, incident microwave power P, and growth pressure p.

Table 5.1 Growth conditions for 3.5 mm dia diamond film on 5 x 5 mm2 Ir/oxide

Time Pi T P vb 1 r hb
(min) (°C) (Torr) (“M

,5 pin

 

Figure 5.23 (a) DIC micrograph and SEM image of a 3.5 mm dia diamond film. (a) DIC
micrograph, the circular band at the edge of the film is the region below the cap.
Asymmetry in the left and right side of micrograph is due to sample tilt (b) SEM image
of the same film at its center. Bright White dots are locally high regions. Diamond
crystallites have undergone complete coalescence.

A typical maximum thickness at the middle of a diamond film grown under these
conditions with tg= 3h is about 550 nm. A DIC micrograph of such a film is shown in
Figure 5.23(a). The diamond film is transparent, so that interference occurs between the

light rays reflected from the front and the back surface of the film. The color differences

96

seen in the diamond film are interference fringes, arising from thickness variations across
the film. The thickness variation from the middle to edge of the film is estimated to be

150 nm.

Figure 5.23(b) shows an SEM image of the same film taken at the center.
Diamond crystallites have already coalesced with each other. No coalescence boundaries,
voids or bare regions can be seen. The small bright regions arise from small height

fluctuations on the surface.

Now we discuss growth in a square geometry with a 5 x 5 mm2 Ir/oxide substrate.

A different Mo cap and post were used. The height of the bias ring above the substrate
was found to influence the formation of the pillared Ir surface. As the bias ring is moved
closer to the surface it flattens the plasma, increasing the uniformity of the plasma over a

larger area.

Typical growth conditions for this geometry are given in Table 5.2. Bias current
increased from 40 to 50 mA at a bias voltage of -150 V. Growth temperature was 780 °C,

10 °C higher than for the smaller area diamond film.

Table 5.2 Growth conditions for 5 x 5 mm2 diamond film on 5 x 5 mm2 Ir/oxide

 

 

 

 

 

substrate.
Time P T(°C) P v 1 mA I h
(min) “w" (Torr) (v3 ”( ) (%) (min)
heating 6 0-1 100 25-660 0-18 0 0 0 25
carburization 12 l 100 660 1 8 0 0 2 25
biasing 60 l 100 715 18 -150 40-50 2 25
growth 180 1 100 680 18 0 0 l 25

 

 

 

 

 

 

 

 

 

 

 

Figure 5.24 (a) shows a DIC micrograph of the overall diamond film on a

Ir/SrTi03 substrate. Diamond has grown to the edges without any voids. An SEM image

97

taken at film center is shown in figure 5.24(b), which is similar to the image in Figure

5.23(b) for the smaller film. Thickness variation, estimated from the number of

interference fringes, is about 350 nm.

 

-_: ’5 um} 1.'
Figure 5.24 DIC micrograph and SEM image of a 5 x 5 m2, 3 hour diamond film. (a)
The color bands are due to film thickness variation. (b) SEM image of the same film

taken at the middle. Diamond crystallites are completely coalesced. Bright dots are due to
the local height variation of the fihn.

Diamond was also grown over an 8 mm dia region on 10 x 10 mm2 Ir/SrTi03
substrates. The substrate was supported by a 14 mm dia Mo post. A Mo cap, 15 mm dia
with an 8.5 mm aperture covered the substrate. The bias ring height and bias voltage are
listed in Table 5.3, which shows process parameters. The increased lateral extent of the
plasma led to better nucleation uniformity over the 8 mm region.

Table 5.3 Growth conditions for 8mm dia diamond on 10 x 10 mm2 Ir/STO substrate

Time Pi T p Vb Ib r hb
(min) (°C) (Torr) (%)

 

98

Figure 5.25(a) shows a DIC micrograph of a diamond film grown under these
conditions. It has grown without any apparent voids and with high uniformity near the
center. The thickness variation is about 250 run across the film, less than for the smaller
fihn. Figure 5.25(b) shows an SEM image taken near the middle of the film. The
microstructure is similar to the films shown in Figs 5.23(b) or 5.24(b) except that a few

coalescence boundaries are now visible.

 

DIC micrograph. (b) SEM image of the same diamond film shows the diamond
crystallites coalescing to each other. Residual coalescence boundary can be seen

5.6 Diamond thick film growth

To grow thicker films, a two-step process was used. In the first step, diamond was
grown for 3 h in the heteroepitaxial CVD reactor under the conditions described above.
In the second step, the film on its substrate was transferred to an AsTeX CVD reactor.
Here, process conditions appropriate for homoepitaxial growth were used, as discussed in
Ch. 6. When diamond is grown in the heteroepitaxial reactor, no substrate cooling is
available so that the microwave power and growth pressure must be low to maintain the

correct temperature. Diamond growth under these conditions also requires methane to

99

hydrogen ratios at or below 1% to suppress growth of nonepitaxial diamond hillocks. The
typical growth rate is 0.5pm/h. Furthermore, the diamond film may shatter during
cooling, due to differential thermal contraction. For homoepitaxial growth, using a cooled

substrate stage, growth rates of Sum/h were achieved.

A 5 x 5 mm2 diamond film grown on SrTiO3 for 3 hours with the heteroepitaxial

reactor was first cleaned by etching in a pure hydrogen plasma in the AsTeX reactor.
Process conditions for etching (and growth) are shown in Table 5.4. Figure 5.26 (a) and
(b) show SEM images of a diamond film before and afler hydrogen etching. Square and
rectangular pits have appeared on the surface with sides aligned along [110]. After
etching, diamond grth proceeded under the conditions in Table 5.4. For this film a

growth rate of 4.3um/h was measured. The higher growth rate is a consequence of the

higher microwave power density and methane concentration.

 

Figure 5.26 SEM image of a heteroepitaxial diamond film (a) before hydrogen plasma
etching (b) after hydrogen plasma etching for 15 min.

Table 5.4 Growth conditions of heteroepitaxial diamond film in high growth rate
conditions.

 

 

 

 

 

P (W) p (Torr) r (°C) Time h r 0%.)
H2 etchin 800 150 770 0.25
rowth 800 150 770 3 7

 

 

 

 

 

 

 

100

<010>

T—’<100>

 

Figure 5.27 DIC optical micrograph of a diamond film after 3 h growth under high rate
conditions. Macroscopic steps cover the growth surface.

A DIC micrograph of a diamond film grown for 3 h in each stage of the two-step
process is shown in Figure 5.27. The film is a single crystal, but steps of macroscopic
height and extent are aligned along [110]. This 5 x 5 mm2 diamond film cracked and
detached from the SrTiO; substrate during growth. The film surface was slightly curved
because interfacial stress between the diamond and the underlying substrate led to

bowing.

<110>

"—9 <110>

400 Em

 

Figure 5.28 DIC micrograph of a 65 pm thick heteroepitaxial diamond on Ir. Steps are
observed on the surface. Sample is slightly curved due to stress.

101

A part of the detached film with Ir remaining intact on the back surface was

grown for an additional 12 hours. Figure 5.28 shows a DIC micrograph; again macrosteps

are observed. The film thickness was 65 pm and its area 1.5 x 3 m2.

The two-step process always concluded with the substrate cracking. It was noticed
that after diamond growth the SrTiO; turned black. Also, an intact substrate would break
during cool down to room temperature even after growth for short times. This stress-
induced cracking can occur for several reasons. Owing to the different coefficients of
thermal expansion of diamond, Ir and SIT103, 1.5 x 10'6, 6.8 x 10'6 and 10.3 x 10'6 K'1 at
25 °C, respectively [61], cooling induces a large strain at the interfaces, especially
between diamond and Ir. However, since the Ir layer is relatively thin and its adhesion

with diamond is very good, it is believed that Ir will deform to relieve the interfacial

stress with diamond. Then, cooling leads to bending of the composite structure until a
critical stress is exceeded and the SrTiO3 fiactures. This can explain the cooling-induced
breakage. On the other hand, a different mechanism must be found to explain the
cracking of SrTiO3 during growth at constant temperature. The blackened condition of
SrTiO3 is due to its reduction by the hydrogen plasma at high temperature. As oxygen is
removed, the lattice parameter of SrTiO3.x becomes larger than that of SrTiO3. SrTiOzs,

which has the same structure as SrTiO3, is black and conductive with lattice parameter

0.400 nm[160]. The oxygen depletion zone could be uniform throughout the SrTiO3, if
the oxygen diffusion rate is very high, or it could be non-uniform, with greater oxygen
depletion at the exposed edges. In both cases, interfacial stress would develop, whereas in

the latter case, internal stress would occur within the SrTiO3-x a result of the lattice

102

parameter gradient. Since the substrate sits on a cooled stage in the AsTeX reactor, a
vertical thermal gradient is also present. The temperature gradient ~ 500 °C across the 0.5
mm substrate, would also lead to a gradient in the oxygen depletion rate, leading to a
lattice parameter gradient. It was observed that the substrate would break just after 3 h of

growth if the sides and the edges of the SrTiO3 substrate were exposed. When the edges

and the sides of the substrate were masked with a Mo cap, the time before breaking was

extended to 15 h.

5.6.1 Raman spectroscopy

The 65 pm thick heteroepitaxial diamond film discussed above was examined

with Raman spectroscopy. Figure 5.29(a) shows the Raman spectrum of this film, excited
at 532 nm. A strong first-order Raman peak is observed at 1332.6 cm], characteristic of

unstrained diamond. No evidence for graphitic phase that exhibit Raman intensity

between 1500-1600 cm.1 was seen.

In Figure 5.29(b) two small peaks appear at 1415 cm'1 (575 nm wavelength) and
at 3120 cm'1 (637 nm wavelength). These peaks are identified as photoluminescence (PL)
from substitutional nitrogen-vacancy (N-V) centers. The first peak is emission from the
(N -V)0 center, which is a substitutional nitrogen adjacent to a vacancy in a neutral charge
state. The second is emission from the (N -V)' center, a substitutional nitrogen adjacent to
a vacancy which has trapped an electron. Since either PL signal occurs only for a N-V
pair, it cannot be used to determine the total nitrogen concentration. To estimate the
nitrogen concentration, we note that homoepitaxial diamond grown with 2 ppm nitrogen

intentionally added to the feed gases, produced N-V photoluminescence intensities

103

relative to the Raman line, larger than that seen here [123]. If it is assumed that the
vacancy concentrations and their pairings with nitrogen are similar in both crystals the

present crystal should have a total nitrogen incorporation of about 1 ppm.

 

 

 

 

 

 

 

 

 

 

 

 

Energy (eV) Energy (eV)
2.33 2.21 2.08 1.96 1.84 1.71 2,33 2.21 2.08 Lee 1.84 1.71
A 1 x " A
a . (a) g . a
C C
C .. g . C
E‘ . E i g
a:
5 . or .
r s.
8 . .5; -
C C
3 ‘ 3
s . s
.. #Lfl’rv
”o‘uiooe "" zooo AAAA $006 AA aooo """ sooo A o-“iooouioocuiooéuiooe‘"sooo
Raman shin (em-1) Ramon shift (cm'1)

Figure 5.29 (a) Raman spectrum of heteroepitaxial CVD diamond (b) same spectrum
with 15x magnification. (N -V) photoluminescence peaks and the second order Raman
peak are visible. Two bad pixels in the CCD camera are responsible for the spikes seen in

the vicinity of the 2nd order Raman peak.

Figure 5.30 shows Raman linewidths of heteroepitaxial and natural Type IIa
diamond. Since the linewidth of Type IIa natural diamond is 1.65 cm'1 [143,161], we can
infer an instrumental linewidth of about 1 cm], which is also consistent with the known
grating resolution. Taking this into account, the linewidth of the 65 um heteroepitaxial
diamond, grown under high growth rate conditions, is estimated as 2.9 cm], whereas for
heteroepitaxial diamond grown under low growth-rate conditions by the single-step
process, the linewidth is about 7.0 cm]. Thus, the two-step process leads to a significant

decrease in diamond internal strain. With even thicker crystals, one would expect

heteroepitaxial diamond to approach the quality of high-purity natural diamond.

104

 

Lorentzian fit -

 

'IT‘f‘rITI'IVTT'Y'V'I‘
O

Intensity (arbitrary units)

 

 

 

 

i'nlnlalnlminl

ljllllllllllllllllllllllllllll
1300 1310 1320 1330 1340 1350 1300
Raman shift (cm'1)

I I 11

 

Figure 5.30 Raman spectra of two heteroepitaxial diamond crystals and a Type Ila
natural diamond crystal (line A). High growth-rate heteroepitaxial diamond (line B) and
low growth-rate heteroepitaxial diamond (line C). Linewidth values obtained after

Lorentzian fits are respectively 2.6, 3.9 and 8.5 cm], where an instrumental line width of
1.0 cm.1 has not been deconvolved from the spectra.

5.6.2 Free exciton luminescence

Observation of the free exciton luminescent signal from diamond has ofien been
used as a measure of crystal quality. Here, we compare exciton luminescent intensities of
heteroepitaxial diamond and natural Type 113 diamond. When a photon with above
bandgap energy excites an electron from the valence band, it leaves a hole which can
bind with the electron to form a bound state, or exciton. The exciton will decay by
emitting a photon, whose energy is less than the original photon. This is called exciton

luminescence.

105

 

600 a "
l

 

Type Ila
CVD

500 4

 

400 ,

300 -

Intensity (a.u)

 

 

200 '

100 I

 

 

 

230 240 250 260 270
Wavelen gthinm)

Figure 5.31 Comparison of free exciton luminescence of CVD heteroepitaxial diamond

with Type IIa natural diamond, excited with 213 nm laser radiation under identical
conditions at room temperature [163].

In diamond, the free exciton has a binding energy of 75 meV [162], which is large
compared to typical semiconductors such as Si or GaAs with smaller bandgaps. Decay at
room temperature is thermally induced so that lifetimes are relatively long. Figure 5.31
[163] shows a comparison between the free exciton luminescence from high growth rate
heteroepitaxial diamond and natural Type Ha diamond, both taken at room temperature
with 213 nm laser excitation under identical conditions. Both spectra are similar, but the
intensity, which reflects non-radiative recombination processes, is only slightly larger for

the natural diamond. This measurement suggests that the high growth rate heteroepitaxial

106

diamond compares favorably with natural Type Ila diamond, a conclusion which was

reached from analysis of the Raman scattering.

5.7 Summary

In this study, we used Ir/SrTiO3 substrates to understand the role of biasing in
modifying the Ir surface. By systematically changing the biasing time and examining the
spatial evolution of the Ir surface, we found that the Ir surface is etched by the
hydrocarbon ions during the biasing. The etching process is not uniform. It first starts
from the middle and then propagates toward the edge of a substrate. Faceted Ir pillars are
created after biasing for only 10 min. Diamond nucleates in the channels between two
faceted pillars but not, apparently, on the top (001) surface. Misfit dislocations are
observed at the diamond-Ir interface as expected from equilibrium elasticity theory. The
two-step process for growing thick heteroepitaxial diamond can be used to grow diamond

with quality approaching that of the natural type IIa diamond.

107

6 Chapter 6 Homoepitaxial Growth of Diamond

In this section, growth processes for homoepitaxial diamond are described.
Whereas in heteroepitaxial growth the goal was to increase the lateral dimensions of
diamond crystals, here we are concerned with the growth of much thicker diamond by
homoepitaxy. The growth of diamond on diamond does not rely on a difficult nucleation
step; however, preparation of the diamond substrate requires careful attention prior to
CVD growth. Since lattice mismatch is not an issue in homoepitaxy, the structural
quality of the epitaxial diamond can be very high, virtually indistinguishable from the
substrate. In diamond homoepitaxy, a common substrate is HPHT Type Ib diamond.
This material contains high concentrations of impurities, so it is of interest to see whether
higher purity diamond can be grown as a thick epilayer. The most common impurity in
diamond is associated with nitrogen, which creates several types of defects. These may
act as scattering centers for charge carriers in electronic devices, leading to sub-optimal
performance. Much of this work, therefore, focuses on ways the growth process affects

impurity levels, as well as the overall structural quality.

A new CVD reactor was used for this work. A significant part of the system was
modified and constructed as part of this research program. After a description of some of
these improvements, methods used for preparation of substrate surfaces are explained,
followed by results from experiments on homoepitaxial growth. As a result of this work,
free-standing diamond crystals with thicknesses over 0.5 mm and outstanding optical and
structural quality were grown, as demonstrated with a variety of optical and x-ray

characterization methods.

108

6.1 Experimental methods

This section describes the microwave plasma CVD reactor. The reactor, based on
an AsTeX design, was acquired from Seki Technotron (Japan). Since only the reactor
body and microwave system were supplied, the gas handling and vacuum system were
designed and constructed from individual components. A major improvement in the
reactor was made by designing, constructing, and installing a water-cooled sample stage,
that was retrofitted into the AsTeX system. Methods for temperature measurement,

residual gas analysis, and computer-control of the CVD process are described below.
6.1.1 AsTeX diamond CVD reactor

6.1.1.1 Instrument design overview

The AsTeX-style CVD reactor is shown in Figure 6.1. Only the reactor, the
central vertical structure in the figure, and the microwave components, were acquired
from Seki Technotron. A 1.5kW max, 2.45 GHz microwave source connects to a TMOl
mode cavity by rigid waveguide. A triple-stub tuner impedance matches the source to the
cavity. The cavity contains a water-cooled motorized substrate stage, which can move 6
cm vertically. A vacuum- and microwave-tight access door is used to load samples. Two
vacuum pumps are used to evacuate the chamber prior to growth and to maintain a
constant pressure during the growth process. Introduction of feed gases into the chamber
occurs by computer-interfaced mass-flow controllers. Temperature is monitored by two
pyrometers positioned outside the reactor chamber. A PC-based LabView interface was

developed to monitor and control gas flows, microwave power, substrate temperature and

109

system pressure. Multiple process steps can be programmed into the system, which is

designed to shut down in the event of malfunctions.

110

 

. Feed gas tanks

. Mass flow controllers
. Toggle valves

. Magnetron head

. 50 9 load

. Triple stub tuner

. Rectangular wave guide
. Two color pyrometer
. One color pyrometer
10. Optical fiber

11. Inline gas filter

12. Inlet gas valve (V4)

\OOOQQMAUJNp—a

33

13. Baratron pressure gauge
14. Vacuum pressure gauge
15. Vent valve (V5)

16. Needle valve (V7)

17. Toggle valve (V9)

18. Residual gas analyzer
l9. Shut-off valve (V3)

20. Turbo-molecular pump
21. Gate valve (V6)

22. Cooling water feedthrough
23. Stage control motor

24. Backing valve (V 3)

25. Isolation valve (V 10)
26. Pirani gauge

27. Roughing valve (V 1)
28. ThI'Otth valve (V11)
29. Isolation valve (V 2)
30. Roughing pump

31. Flushing valve (V12)
32. Flow meter

33. Heater power feedthrough
34. Loading door

35. Growth chamber

Figure 6.1 Schematic of CVD reactor used for homoepitaxial diamond growth.

111

6.1.1.2 Gas handling

The gas handling system is responsible for producing and maintaining the gaseous

environment in the reactor chamber. An air-cooled turbo-molecular pump (Pfeiffer)
backed by a roughing pump (Alcatel) reduces system pressure to a base level of 3 x 10'8

Torr. A low base pressure is necessary to insure that residual gas concentrations are
reduced below the ppm level. A high-throughput gate valve (VAT) isolates the main
chamber from the turbo-pump. Feed gases, from high-pressure cylinders, are introduced
by individual mass flow controllers (MKS 1249). They are interfaced to a control box
(MKS 427 type D), which is monitored and controlled by the PC. Each mass flow
controller is isolated from the main flow line by a shutoff valve to eliminate leakage
through the flow controller. Pressure is controlled by a throttle valve (MKS type 653)
interfaced to a pressure controller (MKS type 651C) that uses as pressure sensor a 0-1000
Torr range capacitance gauge (MKS type 626A). The pressure controller is interfaced to
the PC and system pressure is maintained within :1: 0.1 Torr of the set-point (typically in

the 30-200 Torr range).

Since the system does not have a load lock, it is necessary to expose the sample
chamber to atmosphere when loading. This introduces nitrogen into the system which
must be removed by the turbopump. A residual gas analyzer is used to evaluate vacuum
quality. When the system is brought to atmospheric pressure, Ar gas is used as a venting
gas since it does not affect crystal properties. During plasma processing, the dominant
gases used are H2 and CH4. At the roughing pump output, exhaust gases are mixed with
nitrogen to reduce the hydrogen concentrations below 4 %, the explosive lower limit,

before they are exhausted through a vent.

112

6.1.1.3 Sample stage and sample cooling

The as-received reactor was equipped with an inductively-coupled heating stage.
For homoepitaxial growth, it is desirable to maintain a relatively low substrate
temperature in the presence of high microwave power. Therefore, the original stage was
modified for water cooling, as shown in Figure 6.2. Water is introduced by one of two
0.25” dia stainless steel tubes into a hollow stainless steel stage clamped onto the top of
an existing platform. Since the reactor allows for application of a DC bias to the sample
stage, all connections, including the water lines, needed to be electrically isolated from
ground potential. Water is circulated by a water-to-water heat exchanger (Haskris) with
16 l/min flow capacity for cooling reactor walls and magnetron head. Part of this flow, 6
l/min, was used for stage cooling. Since the cooling is critical to reactor function, a low-

flow condition triggers an alarm and initiates a reactor shutdown.

In the microwave cavity, the substrate is placed on a Mo post threaded into the
cooled stage. A Mo cap is placed around the post so that plasma forms at the sample
position. The height of the substrate relative to the top of the cap is adjusted by rotating

the threaded post. A Mo plate on short stand-offs protects the cold plate from the plasma.

113

 

 

 

 

 

 

 

L
A
‘ c
x r
to
§;
§
1. Substrate 5. Water-cooled stage 9. Water lines
2. Mo cap 6. Graphite stage 10.Ceramic support plate
3. Mo plate 7. MW chokes 11. Mo post
4. Corona ring 8. Ceramic support tube

Figure 6.2 Schematic of cooling stage and substrate holder

6.1.1.4 Temperature measurement

Infiared pyrometers are used to measure substrate temperature. A two-color
pyrometer (Williamson) (sensitive to 2.4 and 2.2 pm ir radiation) is positioned directly
above the substrate with only a single, intervening fused quartz window. The focal point
can be adjusted with the help of a laser beam that passes through two focusing lenses

onto the emitting region. The field of view is 4 mm dia at a distance of 45 cm.

114

Differential absorption at the two ir wavelengths must be negligible if the pyrometer
calibration is to be maintained. A single-color pyrometer (Ircon) is focused on the sample
through a side window. This instrument has higher sensitivity and was used as a working
thermometer. The pyrometers were calibrated at the Au melting point (1065 °C) inside

the reactor. Details are given in Appendix B.

6.1.1.5 Residual gas analysis

The gas handling system included a residual gas analyzer (RGA) used to monitor
gas composition during reactor operation. Since the RGA only functions below 10'4 Torr,
a differentially-pumped RGA chamber was constructed, in which reactor gases were
introduced by a needle valve, Figure 6.1. When the process chamber is at low pressure
with the turbopump running, the gate valve and isolation valve are open so the RGA can
be operated below the upper-pressure limit. During growth, with the main chamber above
100 Torr, the gate valve is closed so that the RGA is connected to the main chamber
through the needle valve side. The turbopump then pumps on the RGA through the open

isolation valve without affecting the main chamber, so that the pressure remains below

10'4 Torr.

Figure 6.3 shows how partial pressures of several mass species evolve during a 4
h growth run. Only hydrogen and methane were introduced into the reactor here, at a 4%
methane to hydrogen ratio. Prior to methane introduction, all partial pressures were below
10'7 Torr, with the exception of hydrogen, which is always present at high concentrations.
With methane in the chamber, chemical reactions create hydrocarbon species,

predominantly 26 amu acetylene (Csz). Since the RGA is located far from the reaction

115

chamber, transient species will probably not be detected. However, the RGA itself can
create radicals, accounting for the observation of methyl species at 15 amu, derived from

methane at 16 amu. Unambiguous detection of N2 is difficult since it coincides with

 

 

 

 

 

 

 

ethylene (C2H4) at 28 amu.
l..11..J ' 1..' '. ' J .1. ' ' .
' i w" . .- '7 .
154 i g .5
: :r:' :
A _ Molecularmass(arnu)
E 155, ‘I’ __28 ,5
1% : ———16 ;
5 1515 1' —26 i
8 5 —27
95>- . ”if 15 .
1E-7 . , n-
r i 13 i
F1, : 2 .
0- 1E-8 .r g 1
E ~17 :
1E-9
r I r l A l I I A l I l n I;
0 1 2 3 4 5 6

Time(h)

Figure 6.3 RGA spectrum taken during diamond growth at 150 Torr.

6.1.2 Operational procedures

An abbreviated procedure for homoepitaxial diamond growth is given in this

section.

6.1.2.1 Substrate loading

The M0 post is first threaded into the cooling plate and the Mo cap inserted
around the post. The substrate is placed on the Mo post and the Mo plate installed around

the cap. The stage is translated vertically to its standard position and pyrometers are

116

focused onto the substrate. The reactor door is closed and the chamber turbopumped to l

x 10'7 Torr or lower.

6.1.2.2 System operation

With system pressure below 1 x 10'7 Torr, gas partial pressures are recorded.

Gases consist mostly of water vapor, 3 x10'8, N2, 3 x 10'9 and Ar, 1x10.9 Torr. With the

turbopump off and gate valve closed, H2 flow is initiated at 300 sccm until the pressure
reaches 100 Torr. The growth chamber is then slowly pumped out to flush the roughing
line with H2. At 10 Torr, application of 300 W microwave power ignites the plasma.
After ignition, pressure and power are slowly increased to the required level. After
stabilization of temperature and pressure, other gases are introduced as needed. After
completion of the run, all gases with the exception of H2 are turned off, and system power
and H2 pressure are gradually reduced to 100 W and 10 Torr, respectively. MW power is
turned off and the remaining gases are pumped out. After 10 min or so, when the sample
has cooled, the main chamber is vented with Ar and the sample is removed in flowing Ar.

The main chamber is then pumped out and sealed under vacuum.

6.2 Growth process

This section describes the details of diamond substrate preparation using chemical
cleaning and plasma etching. The rationale is to prepare the substrate surface so that no

defects propagate into the growing epitaxial film from the substrate.

117

6.2.1 Diamond substrate preparation

6.2.1.1 Type lb diamond substrates

In homoepitaxy, the surface of the substrate on which the epilayer is grown plays
a crucial role. Homoepitaxial diamond is grown on HPHT Type Ib diamond substrates.
The (001) surface was chosen since the probability of defect formation, principally
penetration twins, is lower relative to (110) or (111) surfaces [164]. Type Ib Sumicrystals
(Sumitomo Electric) are yellow since they normally contain 100-150 ppm substitutional

nitrogen and were supplied as 3 x 3 mm2 parallelopipeds with (001) polished major

surfaces. The crystals were 0.5 :t 0.05 mm thick, with a misorientation angle less than 1
deg from (001) as confirmed by x-ray diffraction. Figure 6.4 shows an optical photograph

of a HPHT crystal taken in white light.

<0>

<100>

         

. . . . ,, J31?
Figure 6.4 Optical photograph of a HPHT Type Ib diamond substrate taken in white
light. Left and the top sides were laser-cut with a slight bevel.

 

.13

HPHT diamond contains several types of imperfections that may influence

subsequent epitaxial growth, such as growth sectors, inclusions, and surface damage.

118

Growth sectors arise during HPHT growth when different crystal planes grow at different
rates and incorporate different concentrations of nitrogen [165]. When the crystal is cut
into small pieces, each piece may contain several growth sectors, usually representing

[001] and [l l 1] growth directions.

Fig. 6.5 shows a HPHT crystal after 1 h of plasma etching. The dark objects on
the surface are etch pits, to be described below. For this crystal, all 6 surfaces are {100}
planes. The two lines intersecting the top surface at 45 deg are {111} planes at the
boundaries of grth sectors. The change of refractive index at the boundary creates

sharp contrast in a DIC image.

<010>

<1 00>

 

Figure 6.5 DIC micrograph of a plasma-etched diamond substrate. Dark objects are etch
pits. The crystal contains three different growth sectors, separated by the dark lines along
<110>. They represent regions with different growth directions.

HPHT growth uses transition metals, predominantly Fe and Ni, as catalysts.
Inclusions arise when part of the catalytic melt is trapped inside a crystal. Inclusions are
generally quite small and not easily visible by optical microscopy of as-received

substrates. They may appear after plasma etching, as discussed below.

119

 

Figure 6.6 Polishing scratches on the surface of a type lb HPHT diamond substrate

Surface-related damage is generally associated with polishing of the surface.
Diamond grit leaves scratches, Figure 6.6, which can penetrate 3-4 1.1m below the surface.
A series of processing steps must be followed to remove scratches, embedded diamond
grit, and any shallow catalyst inclusions. The surface preparation must be sufficiently

reliable that not even one defective region is allowed to remain.

6.2.1.2 Chemical cleaning

As-received diamond substrates are cleaned sequentially in acetone, iso—propanol
and water using ultrasonic agitation for 10 min each to remove organic or water-soluble
contaminants. They are then ultrasonicated in Alconox solution in DI water for 0.5-1 h to
remove particulates. A solution of 25 ml of H202 (hydrogen peroxide) and 25 ml of
H2304 (sulphuric acid) is mixed in a round-bottom flask. The diamond substrate is placed
in the flask and the solution boiled at 300 °C for 0.5 h using a flask heater. This process
removes metallic and graphitic contaminants. After cooling, the substrate is rinsed in DI
water, N2 dried, and placed inside a clean substrate carrier box. The substrate is
sequentially ultrasonicated in DI water and iso—propanol and optical micrographs taken

120

made reference. Sample thickness is measured with a stand-mounted gauge (Mahr) with
1 pm resolution and maximum travel 5 cm. Since the gauge’s tip can transfer iron to the
diamond surface, each time a thickness measurement is made the crystal must be cleaned
with aqua-regia to remove contamination. This gauge is also used to measure changes in

crystal thickness so that vertical growth rates can be calculated.

6.2.1.3 Hydrogen -Oxygen plasma etching

A plasma etching process was developed to remove the damaged region fiom the
top surface of diamond substrates. The idea is to remove the minimum amount of
diamond necessary to produce a microscopically-undamaged surface suitable for
subsequent CVD growth. Typical etch chemistry has previously utilized pure
hydrogen[109] or a hydrogen-oxygen plasma [127,166,167]. In most cases, we used a
dilute oxygen-in-hydrogen plasma. Several variables were examined and optimized,

including system pressure, substrate temperature; etch time, and oxygen-hydrogen ratio.

Etching experiments were conducted inside the CVD reactor, with oxygen
concentrations 0.5-3.3 % relative to H2 concentration, total gas pressure 70-150 Torr, and
temperatures from 675-930 °C. The etch rate was determined by thickness measurements,

with uncertainty :1: 211m, made at the middle and four comers of the sample.

When diamond is plasma etched, material removal is not spatially uniform, but
small pits appear, often with crystallographic registration. At 675 °C, 150 Torr, and 2 %
oxygen-hydrogen ratio, etching was slow and scratch marks remained after 1.5 h. The
etch depth was about 1 11m and etch pits of average lateral size 1 pm appeared. At 725 °C

the surface was covered with shallow pits and an etch rate of 1.5 um/h was observed. The

121

typical lateral pit size was 15 pm. At 800 0C, the vertical etch rate increased to 3um/h

with lower pit density, but the average lateral pit size became larger, 25 pm.

It was realized that etch pits do not necessarily lead to defects that propagate into
the growing crystal. The types of pits that appeared were readily classified, and malicious
pits identified. Two benign types of rectangular/square pits were commonly observed:
one with (111) lateral facets (type A) and the other with a flat (001) bottom (type B) as

shown in Figure 6.7. The pit sides are aligned along [110].

 

Figure 6.7 Diamond surface after plasma etching at r0 = 2 %, p = 150 Torr (a) T = 725
°C, t = 1h (b) T = 800 °C t = 1h. Two types of etch pits are observed

The size of the pits could be reduced by operating at 100 Torr. Figure 6.8(a)
shows the surface of a diamond substrate etched at 2 % oxygen concentration, 800 °C and

100 Torr for 1.5 h to a depth of2 pm.

122

 

Figure 6.8 Diamond surface after plasma etching at p = 100 Torr, r0 = 2 % (a) T = 800
°C, t = 1.5 h (b) 930 °C, t = 1.5 h, deep holes and large pits are created on the surface

Etching at higher temperature and higher oxygen concentration created many
deep, large holes on the surface as seen in Figure 6.8(b) for etching at 930 °C and 1.5 h.
An etch time of 2 to 3 hours at 800-850 °C, 100 Torr and 2 % oxygen concentration was

found to be optimal for most substrates.

To study the effect of etch duration on the morphology of the pits, a diamond
sample was etched in four sequential steps at 100 Torr, 850 °C and 2 % oxygen
concentration for a total of 2.5 h. After each step, the sample was removed from the
reactor and examined optically. The first etching step was 1 h while the other steps lasted

0.5 h. After 2.5 h, the total etch depth was about 4 um.

123

 

Figure 6.9 DIC micrograph of diamond substrate after etching in plasma for a total time
of2.5 h at r0 = 2 %, T = 850 °C and p = 100 Torr; (a) 1h (b) 1.5 h (c) 2 h (d) 2.5 h. Dark
line is a growth sector.

Figure 6.9(a)—(d) shows micrographs taken after each step of the etching. Image J
software was used to analyze the size and the number of the etch pits in the figure.
Figure 6.10(a) shows the total number of etch pits per image vs. etch time. The number of
etch pits decreased after one hour. Figure 6.10(b) shows the average etch-pit size and
areal coverage vs. etch time. The average size increased fiom 6 pm to 13 um and the
coverage increased from 12 to 26%, when the time increased from 1 to 2.5 h. Figure
6.10(c) shows the number of type A and type B etch pits vs. etch time. Type A pits

decreased from 800 to 60, whereas type B etch pits increased from 250 to 360, with the

124

maximum at the third etching step with 380, when the etching time increased fiom 1 to

 

 

 

 

 

 

 

 

 

 

 

 

2.5h.

(3)1200 -. .............. . (hi4_fi-....-.. ......... ,...-.....‘3o
1100- - 2 €28
10001 .‘ E 12L gze

. « 3 I €24°\°
900- 4 ' 5 >

g . « fiw; 3223

'5 800’ ‘ . 52001

L ‘ ‘ g ' E O

3 70°“ ‘ °- I e183

5 0001 . 3 a: 4162

2 l E ’ E148

.. .. o 1
500' 1 > 3* 512“
Mr- - < t
...................... 1....1....r....1....r....-1o
1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Time of etching (h) 11m. of etching (h)
(C)
1000, ............................. .
: A :—380
3°“: 3.3.0
“ : :' g
u 4' U
C : 2-320 2
«E . :»3oo§
3200:- :r a:
z ‘ use 39
OI- '31-260
0.0 05 10 1.5 20 25 30
Time of etching(h)

Figure 6.10 Etch pit statistics. (a) Number of etch pits vs. time of etching; etch pit
number decreases with time. (b) Average pit size, % area coverage vs. time of etching (0)
number of type A and B pits vs. time of etching. Number of type A decreases but B
increases.

After 1 h, a few polishing scratches remained on the surface but all were removed
after 2.5 h. The decrease in the pit density after 1 h etching was most likely caused by the
merging of pits as their size increased. Clearly, the pit density must reach a maximum
before 1 h. The A and B pits are not independent. The A pits form initially with {111}

sidewalls. As etching continues, they become deeper. At a relatively shallow depth, the

125

anisotropy of the etching changes, with the {111} surfaces etching faster than {100}.

This leads to flat-bottomed etch pits.

 

Figure 6.11 Diamond surface after plasma etching at T = 850 °C, p = 100 Torr, r0 = 2 %
and t = 1h. The large dark region is a pit forming at inclusions.

In addition to these benign pits, a third type of etch pit was sometimes observed.
This pit had depths as large as 10 um and showed no specific geometrical shape. It is
believed that it forms at regions with foreign inclusions. Figure 6.11 shows a diamond
sample etched for 1 h at 850 °C 100 Torr and 2 % oxygen concentration. The dark region
in the micrograph is a group of malicious pits. Energy dispersive x-ray (EDX) showed
the signatures of Ni, Fe and Al in this region. Defects will emerge from these regions if
the inclusions are not removed prior to growth. Usually, flat-topped pyramidal hillocks

(without non-epitaxial crystallites) grew out of these pits

126

6.2.2 CVD homoepitaxial diamond growth process

6.2.2.1 Preliminary processing results

In this section we describe the first attempts at homoepitaxial growth using 1.5 x
1.5 mm2 substrates. The objective was to determine the region of parameter space where
growth was reasonably defect-free. Definitions of variables are: substrate (or growth)
temperature, Tg, methane to hydrogen ratio (based on flow rates), r, growth duration of
step i, t, total gas pressure, p, and microwave power, P. For the experiments discussed
immediately below, the pressure was held constant and the geometry was not varied.

Growth was canied out in steps, with the sample taken out of the reactor after each step

and examined by optical microscopy.

Diamond was grown on four different substrates at r = 4 % and T3 = 800-950 °C,

as given in Table 6.1. Substrate temperature was controlled by adjusting the microwave
power. Growth pressure was fixed at 150 Torr and growth duration was varied from 3-6

h.

Table 6.1 Growth conditions for 4 samples grown at different temperatures. The symbol
hi is the epilayer thickness and V001 is the <001>growth rate

 

 

 

 

 

 

sample p P Tg r t1 hi V001
“0") (W) (°C) W 00 (um) (um/h)
A. 150 1450 950 4 3 12 4
A2 150 1350 900 4 4 14 3.5
A3 150 1300 850 4 3 8 2.7
A4 1 50 1 100 800 4 3 7 2.3

 

 

 

 

 

 

 

 

 

Figure 6.12 shows diamond surfaces grown at different temperatures. With T8 =
950 °C, after 3 h the surface of A1 was almost fully covered by pyramidal hillocks (a).

The dark region on the top of the hillock is a non-epitaxial crystallite; hillock sides show

127

macro-steps parallel to [110]. The height of the highest hillock was 2.5 pm, as measured

by a Dektak profilometer. The growth rate under these conditions was 41.1m/h.

 

100 um

Figure 6.12 DIC micrographs of diamond surfaces grown at 4% methane concentration,
150 torr and (a) 950 °C (b) 900 °C (c) 850 °C (d) 800 °C temperature.

With Tg = 900 °C, hillock density is shown in Figure 6.12(b). As in (a), most
hillocks contained a non-epitaxial crystallite at their apex. Growth rate in this case
dropped to 3.511m/h.

Growing at Tg = 850 °C produced no large pyramidal hillocks, but the surface was

covered by tiny round hillocks without non-epitaxial crystallites, as shown in Figure 6.12

(c). Maximum height of these hillocks is 0.2 pm.

128

Growth at Tg = 800 °C produced the diamond surface shown in Fig. 6.12(d). Note

the shallow height fluctuations and absence of hillocks; however, the growth rate

decreased to 2.3um/h under these conditions.

 

 

h

' 1
C
1

b
r
I
9"
0|

' I
I
r

2.5 -

Growth rate (pm/h)
00
Growth rate (pm/h)

N
r

‘

 

 

 

 

 

 

7150‘800‘850A900"95041000 0.82"0.84‘0.86"0.88A0.90A0.92‘0.94
-1
Temperature(°C) 1000/79“ )

Figure 6.13 Effect of temperature on growth rate at 4 % methane concentration in gas
phase.

Figure 6.13 shows a plot of the growth rate vs. substrate temperature. Growth rate

obeys an Arrhenius relation given by V001 = A exp (-Ea/kBTg). The activation energy E,

for this temperature range is found to be 10 Kcal mol'l or 0.5 eV/atom, similar to values

previously reported [109,125].

To study the effect of methane concentration on defect formation and growth rate,
three different diamond crystals were grown at r = 2.6-8 %. First, an epilayer was grown
at r =2.7 % at 800 °C for 7 h. Then it was overgrown at r = 4 % and 800 °C for an
additional 3 b. Figure 6. 14(a) shows the diamond surface after the first growth. There are
several shallow pits visible; however, all the pits filled in after the second growth stage at

higher methane concentration. The resulting surface is quite smooth.

129

Table 6.2 Growth conditions for diamond grown in two stages at different methane
concentrations; r = 2.7 and r = 4% at 800 °C.

 

 

 

 

Sample p P Tg r T1 111 V001
(Torr) (W) cg ('70) a.) (lull) ("m/1.1
A5 150 1150 800 2.7 7 12 1.7
A5 150 1 100 800 4 3 7 2.3

 

 

 

 

 

 

 

 

 

   

Figure 6.14 DIC images of a diamond sample after grth at 800 °C, (a) at r = 2.7 %, t =
7hand(b)atr=4%,t=3h.

Table 6.3 Growth conditions for two diamond samples grown at different methane
concentrations at T= 770 °C and t = 6 h.

 

 

 

 

sample p P T g r t1 111 V001
(Torr) (W) (°C) ('70) Q2) (Em) (pm/h!
A6 150 1000 770 7 6 28 4.5
A7 150 1000 770 4 6 29 4.8

 

 

 

 

 

 

 

 

 

Next diamond films were grown on different substrates at 770 °C for 6 h at r = 7
and 8 %, Table 6.3. Figure 6.15 shows their surfaces, both of which are quite smooth and

defect-free.

130

 

Figure 6.15 DIC micrographs of diamond surfaces afler growth at T = 775 °C and t = 6h,
(a) r = 7 % (b) r = 8 %. Surfaces seem smoother than at r = 4%. Lines seen in (b) are due
to deep scratches on the substrate not removed by etching; however, they did not generate
defects.

 

5.5 . . . . . . .
5.0 -
4.5 - I
4.0 -
3.5 -
3.0 -
2.5 -
2.0 -
1.5 -

Growth rate (pmlh)

 

 

l I I r | I

 

2 3 4 5 6 7 8
Methane concentration In gas phase(%)

Figure 6.16 Growth rate vs. methane-hydrogen ratio r. The upper two points represent
growth at T3 = 770 °C; the lower two are for Tg = 800 °C.

Next, thicker diamond crystals were grown by a multi-step process to 310 pm.
Crystal A6 was grown at r =7 % and A7 was grown at r = 8 % under the conditions given
in Table 6.4. Smooth surfaces without hillocks resulted; only a few shallow pits appeared.

Figure 6.17 shows an SEM image of A6 tilted at 45 deg to the electron beam. The top

131

surface is (001); {111} facets appear in each corner. At the lower part of the image the

substrate can be seen. It is evident that appreciable lateral growth has taken place. The

major lateral surfaces are rounded, not faceted. An optical micrograph is shown in Fig.

6.18, taken in white light. The yellow color is due to the substrate, still attached. The

crystal is as-grown; only some graphitic deposits were removed from the back surface.

Table 6.4 Growth conditions for samples A6 and A7

 

 

 

 

 

 

 

 

 

 

sample p P T: r t h V001
(Torr) (W) (c) 0%) 00 (um) (um/h)
A6 150 1000 775 7 68 310 4.6
A7 150 1000 775 8 59 290 4.9

 

 

 

 

Figure 6.17 SEM image of a 290 pm thick diamond crystal on substrate. {111} faceting

at the corners occurs. Tilt angle is 45 deg.

132

 

Figure 6.18 Optical photograph of a 290 pm thick film on its substrate. The dark bands
are artifacts which occur in regions where total internal reflection occurs.

6.2.3 Long-term growth procedures

We have used multi-step processes to obtain diamond crystals as thick as 550 pm
on 3 x 3 mm2 substrates. A film was grown for a specified period, taken out of the
reactor, examined in an optical microscope, refined to the reactor, and growth resumed
As many as 8 stages were used, usually with different processing parameters at each
stage.

A film grown at T3 = 770 °C, r = 7%, and t1 = 6 h produced a smooth diamond

surface as shown in Figure 6.19(a) with V001 = 4.2 um/h, comparable to earlier

experiments on small substrates.

133

 

Figure 6.19 DIC micrographs of crystal A3 after different growth periods and conditions
(see Table 6.5); (a) 770 °C, 25 um, growth sector boundary is seen at the corner, (b) 755
°C, 117 um, faceted deep pits appeared, (c) 800 °C, 273 um, pits are healed at higher
temperature, ((1) 800 °C, 515 pm macro steps observed. Total thickness of epilayer at
each stage is given.

Growth at 755 °C for 22 h in the next step produced several deep pits as shown in
Figure 6.19(b). The sides of the pits are parallel to [100]. There is no apparent
correspondence between these pits and those formed during plasma etching on the
substrate. In the next step, growth at 800 °C for 24 h healed all pits, leaving a smooth
surface, Figure 6.19(c). In the next two steps, the crystal was grown for 48 h at 800 °C to

a cumulative thickness hcum = 515 pm. The total growth time for all steps was 110 h,

134

yielding an average growth rate V<100> = 4.7 um/h. The process is summarized in Table

6.5 where cumulative epilayer thicknesses are given.

Table 6.5 Growth conditions for a 515 pm thick film sample A3

 

P

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample p T r t1 h V
g cum 100
(1‘0") (W) (°C) (“/0 (10 (um) (um/h)
A8 150 950 770 7 6 25 4.2
A8 150 900 755 7 22 1 18 4.4
A8 150 950 800 4 4 132 3.5
A8 150 900 800 7 6 158 4.3
A8 150 900 800 7 24 273 4.8
A8 150 860 800 7 12 331 4.8
A8 150 850 800 7 36 515 5.1
1125 . , . 1 . I - r . r f
. -l
E 1100 - 1
a I
— 1075 - -
a
8 . '
o 1050 " '1
é . .
o
E 1025 ‘- \- u
.9 - J
0
:n 1000 . -
i
975 A 1 A l A l A l A L A
-3 -2 -1 0 1 2 3
Distance form the center (mm)

Figure 6.20 Height profile of diamond crystal A3 along its top surface diagonal

In contrast to the small substrates, the diamond growth rate here was faster at the
edges of the substrate. At the end of the growth stages, the film edges were 50 um higher

than in the middle of the top surface. Roughening was more pronounced, with macrosteps

135

appearing along the thickness gradient, as seen in Figure 6.19(d). The surface profile,

taken by scanning along the diagonal, is shown in Fig. 6.20.

Figure 6.21(a) is an optical photograph of crystal A3 on its substrate. Here,

surface steps, though present, are not visible. Significant (111) faceting appears at the
four comers. No sign of pits or growth defects can be seen. During growth, the lateral
dimensions increased from 3.0 to 3.8 mm. To examine the epitaxial diamond without the
influence of the substrate, the crystal was processed by a commercial diamond cutter.
The substrate was removed by grinding, resulting in the crystal shown in Figure 6.21(b).
The clarity and absence of optical distortion of the diamond is evident. However, it was

found that at least 35 pm of substrate was not removed, resulting in the slightly yellow

tinge in the image.

   

1 .
O
I
O
O
O

C
0

Figure 6.21 Optical photographs of a 515 pm thick diamond crystal taken in white light
(a) on its substrate. (b) after removal of most of the substrate. The crystal still contains at
least 35 pm of its substrate.

Another crystal, A9, was grown to a final thickness of 550 pm. In this case, four

stages were used, as shown in Table 6.6. To better understand the role of residual

136

impurities, purer feed gases were used, 99.9999% for H2 and 99.999% for CH4, compared
to 99.9995% for H2 and 99.99% for CH4 used before. The first stage was carried out at
low r to suppress defect formation at the substrate-epilayer interface. Alter growth, the
substrate was removed by mechanical grinding; the top surface was also flat ground and
polished, as shown in Figure 6.22. In contrast to most growth experiments, the substrate
used for A9 did not have (001) and (010) lateral edges. As a result, the {111} corner
facets of the homoepitaxial crystal were smaller and rotated with respect to the crystal’s

edges.

Table 6.6 Growth conditions for sample A9
sample p P r
(Torr) (W) (%)
l

 

l 3.6 mm
Figure 6.22 (a) Optical photograph of a 550 pm thick crystal on its substrate. For this

substrate <110> was 11 deg fi'om a lateral face. (b) CVD diamond crystal after polishing
off the substrate. This crystal’s corners were damaged during polishing.

137

A diamond crystal A10 was grown to a thickness 515 pm for a total of 120 hours

at r = 6% and Tg = 800 °C. This sample was characterized using x-ray diffraction and

Raman scattering discussed in section 6.3.

6.2.4 Nitrogen in CVD diamond

In this section we discuss the role of nitrogen in CVD diamond. As noted
previously, a manifestation of nitrogen in diamond at the 100 ppm level is the yellow
color imparted to crystals. There may be several types of nitrogen-related centers
introduced in the epitaxial growth process. The tools to study the impurity states were
Raman scattering, photoluminescence spectroscopy, and infrared absorption
spectroscopy. Fortuitously, the spectral range of the cleanroom Raman instrument, when

using 532 nm excitation light, included the PL wavelengths of the two dominant

nitrogen-vacancy centers. The zero-phonon line (ZPL) for the (N-V)' center occurs at
1.95 eV ( 637nm) and at 2.16 eV ( 575 nm) for the (‘N-V)0 center[29]. By assuming that

the 1332 cm"1 line for the single Raman-active phonon has the same intensity for all Type

Ib substrates and epilayers, it is possible to normalize the PL signals to the Raman line,
allowing comparison of PL intensities taken under different conditions. Justification for
this procedure is given by the identical Raman linewidths observed for all crystals
studied. The same experimental setup was used for all measurements, namely, a 100x
objective to focus the excitation light to a spot at the surface of the sample. Since the
system is not strictly confocal, backseattered Raman light and PL are collected from a
region approximately 50-75 pm below the surface. Additional evidence for the high

structural quality of CVD diamond is given by high-resolution XRD. The object of the

138

research presented in this section is to relate the concentrations of nitrogen and N-V

centers to growth conditions.

6.2.4.1 Type lb substrate

Since many of the studies were made with the epilayer attached to the substrate,
the properties of the HPHT diamond substrate were first examined, using the Raman
instrument, to study sample-to-sample variability of luminescence intensity and
distribution. Although the Raman lines were nearly indistinguishable for all samples,
some substrates showed a broad luminescence background that appeared simultaneously

with weak N-V PL peaks, as seen in Figure 6.23.

 

 

 

 

 

 

 

Energy(eV)
2.33 2.21 2.08 1.96 1.84 1.71
I . 18 v u . I . r .
n - g
.12 - c .
g _ g
2" n:
N - .
h
.e . .
1‘." . 4
...? .
10
g " ‘ -I
"’ 9
"- l A l A l A V—LVA (bl) A -
0 1000 2000 3000 4000 5000
Raman shift (cm1)

Figure 6.23 Spectra of Type Ib diamond substrates (a) broad background and weak
(N -V) peaks, (b) no background and no (N -V) peaks.

139

 

15 r v fi—I fi—Y r T—v Tfi l I 1—1 I Tfi' Tfi Y liw r—I

AL

I 1—'

10

—V_ 1 Tfi
A L l A- l 1 l J I
1—

0 (cm!)

 

 

Tfi' l

 

 

0 4 1+] 1 L_l_l_L| 1 1‘1 1 I44. 14' _L_.L l q

4000 3200 2400 2000 1600 1200 800 400

Wavenumber (cm-l)

 

16 Y r V V Y Y I r‘r I V I V V V I I V V Y Y T I V I V I V V V V I V V Y Y I V r

-1

Coefficient of absorption(cm )

12- -
101- .

 

 

O N & O Q
' l
l

500 1000 1500 2000 2500 _ 3000 3500
Wavenumber (cm )

 

Figure 6.24 FTIR spectrum of diamond (a) natural Type Ila (b) Type Tb. Substitutional
nitrogen absorption peaks appear at 1130 cm'1 and 1344 cm"1 in (b).

Substrates were also studied by Fourier Transform Infrared Spectrophotometry

(FTIR) using a Matson-type FTIR instrument with resolution 0.4 cm'1 between 600 and

4000 cm]. The measurement was made with the samples at room temperature. Figure

6.24 shows the FTIR spectrum of a HPHT substrate together with a natural type Ila

140

diamond with very low level of impurities. It shows two absorption peaks, a broad one at
1130 cm'1 and a sharp one at 1344 cm'1 which indicate the presence of substitutional
nitrogen. Above 2800 cm], two and three-phonon absorption bands characteristic of

diamond appear.

The (004) X-ray linewidth of several substrates was measured with the Bruker
instrument, using the 0.154 nm Cu K, line, and a Ge monochromator with instrumental

resolution 0.0014 deg. Figure 6.25 shows a (004) rocking curve for a Type lb crystal,

yielding a total linewidth of 1.8 i 0.1 mdeg, uncorrected for instrumental broadening

 

Intensity (arbitrary units)

I- II

D -

 

 

59.61? 159.620. "59.624" 4 59.628" A
ewes)

 

Figure 6.25 X-ray (004) rocking curve linewidth of a Type lb diamond substrate

6.2.4.2 CVD epilayers

Diamond epilayers grown at r = 4-8 % and Tg =750-880 °C were examined with

the Raman instrument. In the growth process, no nitrogen was added intentionally and the

grth chamber was evacuated for the same amount of time before starting. The base

141

pressure and the nitrogen partial pressure were monitored before growth to insure no

external leaks.

Diamond grown with r = 4 % showed no (N-V) PL for Tg = 770-880 °C. Only
one exception to this observation occurred, where a very weak (N-V)’ PL was observed
for a crystal grown at the initiation of homoepitaxial experiments when small leaks into

the grth chamber might have been present. Similarly, growth at r = 6 % and T8 = 800

°C showed no (N-V) emission as was as the case at r = 7 % with Tg = 880 °C. Figure 6.26

shows spectra taken from these three diamond crystals. No luminescence background or
(N-V) ZPL lines are evident even after 20x magnification (b) where the second-order
Raman spectrum is easily seen.

EMTQWGV) Energy(eV)

2.33 2-21 2-08 1-95 1-84 1-71 2.33 2.21 2.08 1.96 1.34 1.71

7’

 

 

fir f ‘f

4

x20

Raman peak‘

(I)
. (u)
_ (III)

 

Intensity (arbitrary units)

iii

Intensity (arbitrary units)

 

 

 

 

 

 

 

 

(iiuioo‘zohoiaoho‘w‘oo‘sooo 0100020004370040005000
Raman shift (cm‘1) Raman shift (cm'1)
Figure 6.26 (a) Raman spectra taken from CVD homoepitaxial diamond grown at

different conditions: (I) r = 4 %, T8 = 770 °C; (11) r = 6 %, T = 800 °C; and (111) r = 7 %,
T = 880 °C. (b) Data in (a) is magnified 20x. No (N-V) related PL is observed.

At higher methane concentrations, however, with r = 7-8%, (N -V)-related PL was

observed for Tg s 800 °C. Furthermore, evidence was discovered that the charge state of

142

the (N -V) center depends on growth conditions. With Tg = 750-770 °C, the intensity ratio,

I(N--V)0 to I(N-V)', was found to be greater than 1, whereas close to 800 °C the ratio was

unity or less than 1.

 

 

 

 

 

Energy(eV)

E 2.33 2.21 2.08 1.96 1.84 1.71
'E .- l I; I I l
a
a i
e a
a a:
a . (
-,,-, - (N-V)
5 o
0:- (MN)
3 - (a)
o
.5
3
E r
g L l l l l

0 1000 2000 3000 4000 5000

Raman shift (cm-1)

Figure 6.27 Raman spectra of CVD diamond crystals grown at 770 °C (a) r = 7 % (b) r =
8 %, PL luminescence from (N-V) centers are observed with 532 nm excitation.

Figure 6.28 summarizes the results of PL spectra from 22 different CVD diamond
crystals plotted in the Tg — r plane. Three regions in the parameter space are identified on
the basis of (N -V) PL intensity. There is a clear delineation between the region with no
(N-V) PL intensity at high Tg and low r, and the remaining part of the diagram. In the
region where PL intensity is seen, it is found that I(N-V)0 > I(N-V)' at the lower growth

temperatures.

143

 

      
      

 

 

 

P I
<8, No(N-V)
e 850 _ . detected
3
£9
a)
Q.
E
01
I— 800 1- 1:1
.C
E .
2 r I
(D
750 l” I
o .
. (.N'V) «NM: . . .(N-V)?I(N.V[>1
4 5 6 7 8
CH 4/H2Ratlo(%)

Figure 6.28 Graph showing three different regions in Tg—r space; no I(N-V),
I(N-V)0/I(N-V)'<1 and I(N-V)0/I(N-V)'>l.

Table 6.7 PL intensities from (N-V) centers in CVD diamond grown under different
conditions, relative to the diamond Raman peak height.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T2 (°C) r (%) I(N-V)°/In 1(N-V)‘/1._.
770 4 0 0
880 4 0 0
800 6 0 0
880 7 0 0
780 7 0.048 0.021
750 7 0.055 0.030
770 T 7 0.030 0.023
800 7 0.003 0.005
800 7 0.004 0.005
770 8 0.135 0.003
780 7 0.023 0.014

 

 

144

(N-V) PL intensities from a subset of the diamond crystals used in this study are
listed in Table 6.7. The ZPL intensities are given in units relative to the 1332 cm.1 Raman

peak intensity (IR).

6.2.4.3 Effect of nitrogen

To study how nitrogen influences growth rate and (N-V) PL intensities, diamond
was grown in stages with 0, 10 and 50 ppm nitrogen deliberately introduced into the H2

feed gas. As is well known [13,123,128,168], small amounts of nitrogen greatly enhance

diamond grth rates, as confirmed in Figure 6.29.

 

 

11YV'VVVVI'VVYTIVIII’TYYVIVYI’TIY"
p
I C I
n q
10- 4
I ‘
b i
A u
E ’
9'- q
E ,.
a. I
v 8»
h
3 > "
D
E .
p
7- .1
D
g h
0
<3 6' ° ‘
I
D
h
5' d
I O
.
4.‘A4JA.JiJlllJlllLLlll‘llllllll‘

 

 

0 10 20 30 40 50
Nitrogen concentration during growth (ppm)

Figure 6.29 Growth rate vs. nitrogen concentration (relative to H2).

145

Energy (eV)
2.33 2.21 - 208

r‘v 1 v v

1.96 1.84 1.71

‘r fj V

 

     
   

%- S .
3 E
E- r. .
z 8
'2 c
3 E
3'- m ‘
'3 a:
c
3- .
5
(b0
i- u:

 

 

 

A l A 144 l A

 

0 1000‘ 31000 5000-1000‘ 5000
Raman shift (cm'1)

Figure 6.30 Room temperature spectra excited at 532nm for diamond grown with N2 in
feed gases: rN = (a) 0 (b) 10 (c) 50 ppm.

Spectra taken at room temperature under identical conditions with 532 nm laser
excitation after each grth step are shown in Figure 6.30. With 10 ppm nitrogen added,
(N -V) and background luminescence noticeably increased above the levels for the
nominally undoped crystal. At 50 ppm, the luminescence background became almost as
intense as the diamond Raman peak. Figure 6.31 shows a plot of (N-V) peak intensities,
afier background subtraction, vs. feed gas nitrogen concentration. Although there are not
enough data to extract the functional form of the I(N-V) nitrogen concentration relation,

it is clear that there is a supra-linear dependence.

146

 

'c12000+ .
a a
r q

Emooo- .
.2 - (N-V)°
'3 3000 ’ (N'W -
g . 1
96000- -
O .
.C
x 4000- -
i

2000- .
$ .
a 0% I ' «

 

 

 

0 10 20 30 40 50
Nitrogen concentration (ppm)

Figure 6.31.(N-V) PL peak heights, after background subtraction for different nitrogen
concentrations during growth.

6.3 Homoepitaxial diamond characterization

6.3.1 High resolution x-ray diffraction

Figure 6.32(a) shows a (004) rocking curve for crystal A10. The linewidth was

extracted by performing an unweighted, least-squares fit of a Gaussian function to the
data, assuming a flat background. The linewidth is 2.1 :1: 0.1 mdeg. For comparison,
Figure 6.32(b) shows the same measurement for the Type Ib substrate. The measurements
were taken in consecutive scans with no adjustment of the monochromator between
scans. However, owing to the miscut angle of the substrate, the absolute line positions are
not accurate, since the goniometer position was changed for different crystals. The
linewidth for the substrate (004) reflection is 1.8 d: 0.1 mdeg. In general, XRD linewidths
this narrow are found only for strain-free, nearly perfect crystals, such as Si or Ge grown

by float zone methods, or molecular-beam epitaxy grown films such as GaAs. This

147

linewidth value is lower than previously reported for CVD homoepitaxial diamond [13,

169, 170].

 

CVDdiamond 1
. 0 Observed
Gaussian fit

 

   
 
 

2.1 :1 0.1 mdog

lntenslty (arbitrary units)

b

 

 

1 L J L L

59.465 '59.470 59.475' 59.460 59.465

 

 

 

0(de9)

3 ' r lb ‘
'E (b) y" 1
3 ' 0 Observed ‘
s ' Gaussian fit ‘
b . -
e _ .'
3

E - ~ .
g _ 1.810.1mdeg ‘
8 - -
E . .

 

 

 

59.615 A 59.620 A A 59.625 59.630
9(deg)

Figure 6.32 X-ray rocking curve for the (004) reflection of (a) CVD homoepitaxial
diamond crystal A10 (b) a HPHT Type Ib substrate.

To establish that the diamond grown was actually a single crystal, an XRD pole

figure scan was made. This measurement examines how similar reflections transform
under rotation. Figure 6.33 shows an x-ray (111) pole figure of sample A10. Four (111)

reflections appear at angular rotations (1p) of 45, 135, 225 and 315 deg, as expected for a

148

diamond single crystal. The difference in peak intensities arises from slight
misalignment. The logarithmic scale shows the dynamic range of the measurement and

the signal to noise ratio.

 

 

(111) (111) (1.1.1) (111)

Diamond

I I I l A I A l llll l AAAA I A I I I ILLA A4 I A I I

50100150200250300350
9(deg)

  

Intensity (arbitrary units)

 

 

 

 

 

 

 

Figure 6.33 X-ray (111) pole figure from homoepitaxial CVD diamond film. The
intensity axis is logarithmic.

6.3.2 Raman spectroscopy

Diamond sample A10 was studied by Raman spectroscopy using 532 nm laser

excitation at room temperature. The Raman spectrum for this sample is shown in Figure

6.34(a). The first-order Raman shift appears at 1332.6 cm]. The second order Raman

peak can be clearly seen after 20x magnification in Figure 6.34(b). No luminescence

background or nitrogen-related PL can be detected with the present sensitivity.

149

Energy (eV) Energy (eV)

39?: . . .233. . .2;°.3. . .19.“. JWW .1-7‘ 2.33 2.21 2.08 1.96 1.64 1.71

vvvvvvvvvvvvvvvvvvvvvvvvv

 

 

Intensity (arbitrary units)
intensity (arbitrary units)

 

 

 

 

 

AAAAAAAAAAAAAAAAAAAAAAAAAAA

0 1000 2000 3000 4000 5000 - '0' ' '1'0'05' 3505' 3505' '4'0'05' '5000

Raman Shift (cm'1) Raman shift (cm'1)

 

 

 

 

Figure 6.34 (a) Raman spectrum of sample A10 taken at room temperature using 532 nm
laser excitation (b) after 20x magnification. Note the complex line shape of the second-
order Raman peak near 2665 cm- 1.

The Raman linewidth could not be directly measured because of the finite

instrumental resolution of the spectrometer grating. It is known that the spectrometer
resolution is approximately 1 cm]. This was confirmed by measurements on a reference
sample of high-purity, high-quality Type IIa diamond, where the standard value of the

Raman linewidth at room temperature is 1.65 cm.1 [143]

Figure 6.35 shows the vicinity of the 1332 cm"1 Raman line for (a) CVD sample

A10, (b) HPHT substrate diamond, and (c) a Type Ha natural diamond crystal. A

Lorentzian function was fit to the experimental data. The fit results for the linewidths are:
(a) CVD sample A10, 2.55 cm.1 (b) Type Ib substrate, 2.64 cm], and (c) Type Ha, 2.60
cm'l. Since the actual resolution function is not known, it was not possible to deconvolve

the diamond spectra to obtain the true linewidth. However, the actual linewidth can be

estimated by subtracting the instrumental resolution from the values measured for the

150

three samples. It appears that all samples are nearly indistinguishable and very close to

the value of 1.65 cm.1

 

 

 

 

 

 

 

 

 

 

 

E Q ' (b) j Type lb ?
'E g . .
3 z. - ‘0 4
a ,- 1
g g : 1 11 ‘
fl :- —d (p— .1
a 3 2.64661" .
a a ‘
Q a ' 1
G g i: '1
5 g t

1300 1310 1320 1330 1340 1350 1360 1370 .130013101132013.301340035011360.1370

Raman shift (cm ) Raman shift (cm1)

g
C
3
5‘
3'!
'E
3
E
a
C
3
s

 

 

 

 

 

1300 1310 1320 1330 1340 1350 1360 1370
Raman shift (cm )

Figure 6.35 Raman spectra and Lorentzian fits for (a) CVD homoepitaxial diamond A10

(b) HPHT homoepitaxial substrate (0) Type IIa natural diamond. The Raman
spectrometer increases the linewidth by approximately 1.0 cm'l.

6.3.3 Fourier transform infrared spectrophotometry (FTIR)

CVD diamond samples A8 and A9 were examined by F TIR at room temperature.
The results for these CVD crystals can be compared to the HPHT substrates, which have
about 100 ppm of substitutional nitrogen. Absorption spectra are shown in Figure 6.36.

The substrate shows substitutional nitrogen peaks at 1130 cm"1 and 1344 cm]. Both
151

CVD crystals show weak absorption peaks at 1344 cm'l' as seen with higher
magnification in Figure 6.37(a). Weak hydrogen-related absorption can be observed at
2800, 2960 and 3020 cm], as shown in Figure 6.37(b). These peaks are due to C-H

stretching modes, indicating the presence of small quantity of hydrogen in the film

[31,171].

 

30 I I I I I I I I
7825-
g . .
_§2o- (a) .
a .
3
315- -
a - (b)
“510- -
u
5 .
g“ (c) ‘
w . .
O
U

 

 

I l I A I A I n I . I . I . I .
50010001500200025003000350040004500
-1
wave number(cm )

 

Figure 6.36.FTIR spectrum of diamond crystals (a) Type Ib (b) CVD diamond A3 (0)
CVD diamond A9. Type Ib diamond shows strong absorption at 1130 cm‘1 and 1344

cm'l, but for CVD diamond the absorption peaks are difficult to see on this scale.
Vertical scale is shifted by arbitrary amount for clarity

Now we use the results of Kiflawi et al. [172] to set an upper limit on the nitrogen
concentration in the CVD diamond. They reported that 25 i 2 ppm of single
substitutional nitrogen produced 1 cm'1 absorption at 1130 cm'l. Using this calibration

factor, the nitrogen concentration in a 0.5 mm thick HPHT Type Ib substrate is 120 ppm.

152

Using the same calibration factor for CVD diamond crystal A8 the concentration is 9 :i: 1

ppm, However, it is known that this crystal has at least 35 pm of substrate attached. The
residual substrate material contribution is thus 8 d: 1 ppm of nitrogen. This leads to an

upper limit for nitrogen in the CVD diamond crystal of 2 ppm. Diamond crystal A9 does
not show a peak at 1130 em1 but at 1160 cm]. The source of this weak absorption peak

is not clear, but there is evidently very little nitrogen.

 

 

 

 

a (b)
()3? ..2.5T-....4H.e....rm+
E ,‘7
% 5.
8 s
3’ i=3
u
n 8
3 a
0 N
h
E O
.2 '5
0 O
E E
o 3
-1.0 °

 

 

 

 

 

 

300 A 900 A1000‘1100‘1200‘1300A1400‘1500
wave number(cm°1)

 

Figure 6.37 Magnified FTIR absorption spectra of (I) Type Ib substrate (H) CVD
diamond sample A3 (III) CVD diamond A9. (a) Region near 1344 cm'1 (b) region with C-
H absorption bands at 2800, 2960 and 3020 cm". Vertical scale is shifted by arbitrary
amount.

6.4 Discussion

In summary, experiments on homoepitaxial grth have led to a better
understanding of the roles of various important growth parameters. Examination of the
effect of plasma etching of substrates prior to homoepitaxial growth, led to methods that

reduced, or eliminated, the propagation of defects from substrate to diamond epilayer.

153

Several highly perfect homoepitaxial crystals were grown, as evidenced by their narrow

x-ray and Raman linewidths.

By studying the (N-V) PL intensities for crystals grown under differing reactor
conditions, we were able to map regions in the growth temperature-methane
concentration space in which no N-V centers could be detected by optical emission.
Furthermore, there appeared to be regions in this space which were favorable to
formation of either neutral or negatively-charged N-V centers. These results suggest that
growth temperature and methane concentration are not independent variables in forming
nitrogen-vacancy complexes. The situation is quite complicated, but it is likely that
vacancy concentrations will be quite sensitive to growth temperature. On the other hand,
the vacancies must become associated with a single N impurity, also implicating vacancy
diffusion dynamics. It is also a possibility that the major effect of increased methane
concentration is to increase growth rate, thereby introducing greater vacancy
concentrations for the same grth temperature. Recent work showed lower N-V
concentrations for growth at 950 °C compared to 875 °C [173]. The interpretation given

was that less nitrogen was incorporated into diamond as growth temperatures increased.

The x-ray examination of CVD diamond crystals found extremely narrow x-ray
linewidths, comparable to the best crystals of any substance. Also, Raman widths for
CVD material were superior to those from high-purity natural diamond. With the use of
FTIR, it was possible to set limits on the total amount of nitrogen incorporated into CVD
diamond grown with high-purity gases to no more than 2 ppm. This result is encouraging
as it suggests that the present process should be able to produce detector-grade diamond

for use as high-speed, energetic particle detectors.

154

7 Chapter 7 Discussion and conclusions

During the course of this research, the growth of single crystals of diamond has
been investigated by heteroepitaxy and by homoepitaxy. The initial motivation for
heteroepitaxial diamond growth was based on the premise that large area, diamond
substrates could be produced by a CVD process. In this work, the diameter of (001)
diamond thin films has been increased from 3.5 mm to 8 mm. Although not yet at the
wafer scale, many of the insights derived fi'om this work are pointing the way to
achieving larger and better quality diamond. The rationale for pursuing homoepitaxial

growth has also been justified by the production of essentially defect-free, high-purity
diamond single crystals. It is worth noting that the 3.8 x 3.8 x 0.5 mm3 crystal represents

about 1/8 carat of diamond, where the carat (ct) is defined as 200 mg. The knowledge
obtained in developing the homoepitaxial process has been helpful in devising the two-
step process for heteroepitaxy, where the second step is essentially a variant of the

process used for high-rate, homoepitaxial growth.

Some of the major conclusions of this research work can be summarized as

follows:

1. The biasing process for enhancing diamond nucleation on Ir is a complex etching
process. The appearance of highly correlated Ir nanopillars originates in the self-
organization of facets on the [110] step edges of (001) Ir. As etching proceeds, the
facets develop into the isolated nanopillars, which protrude above the Ir surface 3-
4 nm. The pillars attract the plasma, which propagates across the Ir surface,

leaving etched regions of pillars in its path. An incubation period, lasting about 5

155

min, has been explained as the time interval required for the secondary plasma to
form at the center of the biased substrate. High densities of nanopillars are
associated with high densities of diamond nucleation sites.

2. HRTEM images have revealed atomic-scale details of the interface between
diamond and Ir. True epitaxy occurs, with misfit dislocations forming within 0.3
nm of the interface. No evidence for diamond nucleation occurring on (001) Ir
surfaces was seen. Thus, the role of biasing may be simply to expose higher index
planes which may be more favorable to diamond nucleus formation.

3. Earlier work with the same biasing process led to the suggestion that nucleation of
diamond does not occur while the bias process is operative [140]. This idea
contradicted the standard view that nucleation and growth proceeds throughout
the bias period. However, by viewing HRTEM images of Ir surfaces subsequent
to the bias process, for time intervals up to 150 s, it was seen that diamond
crystals do not emerge until roughly a minute after bias ceases.

4. In homoepitaxial diamond growth, diamond crystals exhibit x-ray linewidths
which are among the narrowest observed for any crystal, at the limit of resolution
of the x-ray spectrometer. With high-purity feed gases, the nitrogen-related
impurity centers can be limited to the ppm level or less. Again, the ability to

measure impurity concentrations is at the limit of the available instrumentation.

7. 1 Suggestions for future work

In heteroepitaxy, two-step growth was found to be a viable method for producing
diamond. By changing the sample holder geometry so that the secondary plasma

uniformity is increased, it should be possible to significantly increase the diamond

156

nucleation area. However, the cracking of the SrTiO3 substrate needs to be eliminated.

More attention should be paid to Ir growth on A1203. Since A1203 is stronger than

SrTiO3, it can survive greater stresses and is more resistant to reduction by the plasma.

In homoepitaxial growth, nitrogen-related defects can be further reduced by using
higher purity feed gases and by selecting the appropriate growth parameters from the Tg-r

map prepared dtu‘ing this work. Plasma uniformity should be improved so that a uniform
vertical grth rate occurs. This can be done with the present CVD reactor by
geometrical changes in the vicinity of the sample. Higher microwave power could be

advantageous if efficient cooling of the substrate is provided.

There are many applications for high-quality diamond. A particularly attractive
one is in particle detection. Crystals of heteroepitaxial diamond were used at the NSCL to
detect relativistic heavy ions. Sub-ns response times were observed [23]. Application to
beam tracking would be appropriate for thin diamond films since diamond is insensitive

to radiation damage.

Another application is for fission neutron detection. In the spectral range where
energy loss is by nuclear recoil, diamond possesses the greatest sensitivity to neutrons of

any semiconductor, with over 4% of the fast neutrons interacting per mm. Besides the
low carbon mass, this property originates in the short sp3 C-C bond length, only 0.154

nm. The resultant three-dimensional tetrahedral coordination of the diamond cubic
structure leads to the highest elemental packing density; the strong bonding produces
extremely high resistance to radiation damage. Furthermore, the low Z of carbon yields

the lowest photoelectric cross-section in this group of semiconductors. This is useful in

157

discriminating against gamma emission. The diamond crystals produced in this research

should be seriously considered for these, and similar, particle detection applications.

158

8 Appendix A

8. 1 Sample information

8.1.1 Heteroepitaxial sample information

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figrre no Sample no Date Comments
5.13 (a) 03-09-06-17-2 08-12-06 tb = 0 min
5.13 (b) 03-09-06-15-2 08-21-06 tb= 5 min, t,l = 0 min
5.13 (c) &(d) 08-10-06-1-2 08-20-06 tb=20 min, t,l = 0min
5.14 (e) &(f) 12-20-05-11-2 02-25-06 tb=60 min, tg = 0min
5.15 (a) 03-09-06-16-4 08-03-06 tb=10min, tg= 20min
5.16 (a) 08-10-06-1-3 08-28-06 tb=5 min, =20 min
5.16 (c) & (d) 03-09-06-16-1 08-02-06 tb=20 min, tg=20min
5.18 (a) 12-20-05-11-4 02-28-06 tb= 60 min, tfl = 7s
5.18(b) 12-20-05-11-3 02-26-06 tb= 60 min, t,I = 3s
5.18(c) 12-20-05-12-1 03-02-06 tb= 60 min, t,l = 603
5.18(d) 03-09-06-16-2 07-31-06 11,: 60 min, t, = 90s
5.18(e) 03-0-06-14-3 08-08-06 tb= 60 min, t, = 120s
5.18(1) 08-10-06-1-1 08-16-06 tb= 60 min, is: = 150s
5.23 09-13-06-10-1 09-21-06 tb= 60 min, t, = 3b
5.24 08-10—06-2-2 09-06-06 tb= 60 min, t1 = 3h
5.25 08-10-06-6 12-14-06 tb= 60 min, t; = 3b
5-27 & 5-28 03-09-06-13-3 07-06-06 tolO-20-06 65 pm thick hetero
diamond

8.1.2 Homoepitaxal sample information
Figure or N o. in Sample no Start date Last date

thesis

6.7 (a) S1-52-2 04-01-06 04-04-06
6.7 Q3) S1-52-1 03-22-06 03-26-06
6.8 (a) S1-53-4 05-29-06 06-06-06
6.8(b) Sl-53-1 04-30-06 05-02-06
6.9 S1-25-l 10-03-06 10-17-06
6.10 S1-25-2 02-15-07 03-05-07
A. S1-51-4 03-14-06 03-16-06
A; S1-52-1 03-22-06 03-26-06
A3 S 1 -52-2 04-01 -06 04-04-06
A; 81-52-4 04-12-06 04-15-06
A5 S1-53-3 05-10-06 05-19-06
A6 Sl-54-2 06-15-06- 06-21-06
A7 S1-53-04 05-29-06 06-06-06
A3 S 1-25-1 10-03-06 10-17-06
A9 S1-25-6 11-01-06 12-20-06
A10 S1-25-15 07-19-07 07-29-07

 

 

 

 

 

159

 

 

9 Appendix B

9.1 Pyrometer calibration

The Williamson type two-color pyrometer works by calculating the ratio of ir
intensity at two wavelengths, 2.4 and 2.2 pm. Temperature measurement based on two
wavelengths eliminates the need for an absolute emissivity measurement. In contrast, a

single color pyrometer requires knowledge of the surface emissivity coefficient.

It was first used in the sputtering system to measure the temperature of the Ir
surfaces and was compared with the single color pyrometer measurement. In this
measurement two-color pyrometer measurement was about 250 °C higher than the single
color measurement, which was quite puzzling. Then we tried to measure the temperature
of an alumina boat inside a tube furnace. It measured about 300 to 400 °C higher that an
R type (Pt-Rh) thermocouple (TC), calibrated against the gold melting point. This result
was very puzzling and similar to that of the sputtering system measurement. We
considered the following argument. The alumina tube in the furnace is transparent so that
the radiation from the heating elements, that are much hotter than the specimen, would
actually radiate inside the tube leading to the specimen reflecting this radiation to the
pyrometer. Thus, the pyrometer would be sensing the heating element temperature. This
also explained measurement inconsistencies in the sputtering system. Here, ss sample
holder contained small holes allowing radiation from the heater to have a direct path to
pyrometer. To address this problem, four layers of optically opaque furnace tubes were
used as inserts for the outer quartz tubes. Now, with the pyrometer focused on the closed

end of the innermost mullite tube the pyrometer read within i 10 °C of the thermocouple.

160

This confirmed that our previous explanation was indeed correct. Figure 9.1 shows the

plot of the pyrometer- and the thermocouple-based temperature scales.

In the CVD system the signal from inside the growth chamber passes through a
quartz window which could affect the pyrometer. By inserting a spare window before the

pyrometer, we found that the pyrometer measured more than 100 °C lower than without

the window.

 

I I I f I 1
1000- -
E .
2 900- 4
3 1 1
N
3 8m“ ct
Q
E .
3
._ 700- -
3
g 4
E. 600- .
500‘ -
. 4 . 1

 

 

 

500 A 600 I 700 ' 800 900 1000
R-type thermocouple temperature (‘C)

Figure 9.1 Pyrometer temperature vs R Type thermocouple temperature

This confirmed that the quartz plate has a strong effect in the temperature
measurement. Hydroxyl ions, if present in fused quartz, have absorption bands in the
near-ir. Figure 9.2 shows the pyrometer and the TC measurement in this case. A GB 124

quartz plate that contains less than 10 ppm OH was available. Figure 9.2 shows the

pyrometer temperature vs. the R Type TC temperature with these two different quartz
plates in place. Temperature measured with the GB 124 quartz plate was just about 10-15

°C lower than the TC temperature. In the pyrometer there is a correction method defined

161

as the E-slope offset correction, which changes the ratio of the wavelengths by a constant

factor.

‘

923233323

 

I ' I ' I ' I ' I ' I ' I

L—-- Temperature with as received quartz plate 1
. — 0 — Temperature with GE 124 quartz plate

P

did

Pyrometer temperature(° C)

 

 

 

 

500L600-700.800‘900A1000‘1100
R-type thermocouple temperature(°C)

Figure 9.2 Pyrometer and TC temperature with 2 kinds of quartz plate

E slope offset correction was applied to correct the temperature measurement with
the quartz plate in place. The factor was found to be -0.036 for the as received quartz
plate and -0.004 for the GB 124 quartz plate. Figure 9.3 shows the temperature measured
by the pyrometer after the correction factor was used. This measurement was consistent

with the temperature without the correction factor and the quartz plate.

After this correction, the temperature measurement was correct within :1: 10 °C
with the temperature measured without the quartz plate and the E-slope offset. This
measurement suggested that the E-slope offset works. Nevertheless, it was necessary to

calibrate the reactor pyrometer in the presence of the OH-containing quartz window.

162

1200 3 . . w I v I . u . :

 

I Temperature without E-slope offset and quartz plate
1 100 _ 0 Temperature with E- slope offset and quartz plate
3‘: 2 .-
§1ooo — . _
3 * - +
g 900 — a a
E - n l
O
t 800 '- . q
2 - .
s 700 - ' .
2 . a .
I
d? 600 - -

 

 

 

600 700 800 900 1000 1100 1200
R-type thermocouple temperature (°C)

Figure 9.3 Pyrometer temperature with and without quartz plate

To calibrate inside the reactor, the two-color pyrometer was focused on a gold foil
on the sample stage. The foil was then heated with a hydrogen plasma until it melted. The
temperature at which gold melted was recorded as 960 °C. The E-slope correction factor
required to bring it to 1065 °C, the gold melting point, was -0.022. This correction factor
is actually smaller than the previous correction factor. This might be because the two

quartz plates are different or the measurement in the presence of the plasma is different.

After this calibration, the two-color pyrometer was focused on a diamond sample
from the top. A single color pyrometer, which was already calibrated in the furnace with
a thin quartz window, was focused on the sample from the side window. In this case the
two-color pyrometer was fluctuating about :l: 50 °C, so that the temperature shown in
Figure 9.4 is the average temperature measured. In contrast, the temperature measured by

the single color pyrometer was very stable.

163

 

J
300i- /
750- .

700- .

Two color pyrometer temperature (C)

 

 

650 700 750 800 850 900 950

 

Single color pyrometer temperature (° C)
Figure 9.4 Two-color vs. single color pyrometer temperature on diamond sample

Figure 9.4 shows the temperature measured by the two pyrometers focused on a

diamond sample at different microwave power. This measurement agrees to within :L- 25

°C.

Since the two-color pyrometer was fluctuating about i 50 °C, temperature
measured by the single color pyrometer was considered to be the more accurate. Most of
the time the difference between the two and the single color pyrometer measurement is
within at: 50 °C, but in some growth runs the temperature measured by the two-color
pyrometer could be as much as 100 °C higher than the single-color measurement. This

could be attributed to small variations in sample geometry.

164

10 Appendix C

10.1 Ir growth on AI203

10.1.1 Al203 preparation

Two inch diameter 0.5 mm thick (11i0) A1203 substrates having 1.3 deg vicinal
angle rotated about the m—axis and epi-polished on one side were purchased from
Substrates Technology (Russia). Two standard sizes were prepared: 10x10 mm2 and 5x5
mmz. As-received wafers were coated with photoresist and cut with a dicing saw. Each

piece was solvent-cleaned and degreased to remove photoresist and particulates. Afier
optical inspection the substrates were annealed in a tube furnace at 1400 °C in air for 15

hours.

A typical surface structure afier annealing is shown in Figure 10.1. The

topography depends on the vicinal angle and miscut direction. In contrast to SrTi03, the

(11i0) A1203 surface for this vicinal angle does not show a simple reconstruction. In

fact, the terraces are separated by steps of 4-6 unit cells height. For other vicinal cuts,
different surface reconstructions have been observed [11]. The reconstruction depends on

annealing temperature and may also depend on the gaseous environment.

165

 

    

<1100>
T—v <0001>

Figure 10.1 Topography of vicinal (11 i0) A1203 surface after annealing in air at 1400
°C for 15 hours.

10.1.2 Ir growth on A1203

Ir growth was carried out on vicinal a-A1203 substrates. The temperature,
thickness, and Ar gas pressure were systematically varied to optimize growth. In general,
both (001) and (111) Ir grains grow on A1203, but it is possible to suppress (111) growth

considerably. Figure 10.2 shows the topography of a 500 nm thick Ir film grown on

A1203 at 800 °C, with 3 nm/min deposition rate and 12 mTorr Ar pressure. The
topography is quite different fiom Ir on SrTiO3. Alignment of the elongated structures in
this case is along the <1100>direction. Mean roughness is 0.8 nm, almost 3 times greater

than for Ir on SrTi03.

166

<1100>

T—v <0001>

250 nm

 

Figure 10.2 Topography of 500 nm thick (001) Ir surface grown on vicinal a-A1203
surface at 800 °C, 3 nm/min deposition rate and 12 mTorr Ar gas pressure.

 

A
a:
P

a:
on

 

 

12_ —o.eo

3
A -0.55o
.\° 9- 3
g fio-wg
§ 6- i
5 30.453
: E
=’ 3' -0408“

3

o- -o.35

 

4 6 8 1o 12 14 16 18
Growth pressure (mTorr)

Figure 10.3 Plot of (002) Ir linewidth and the ratio of (111) to (002) reflected x-ray
intensity vs deposition temperature. Substrate vicinal angle is 1.3 deg with respect to the
rotation about m-axis.

Figure 10.3 shows (002) Ir linewidths vs. deposition temperature for 500 nm thick

films. It also shows the ratio of (111) to (002) intensity, denoted as I(111)/I(002) vs.

167

deposition temperature. The linewidth minimum occurs at 930 °C, but I(111)/ 1(002) is

higher than its optimum value of 0.13% for 800 °C film.

To study the effect of film thickness on (002) Ir linewidth and I(111)/I(002), five
Ir films were grown at 800 °C, 3nm/min deposition rate and 12 mTorr gas pressure, with
thicknesses from 85 to 850 nm. Figure 10.4 shows the variation of linewidth and
I(111)/I(002) with film thickness. The (002) Ir linewidth is smaller for thicker films, but

I(111)/I(002) shows no systematic variation.

 

I I I T I I I r 0-55
0.30 -
0.25 - 1 0-50
’5
°\° 0 20 - 33
g ' - 0.45 =
3
E 0.15 - 1 CD
3 0 40 E.
z: - . a.
z 0.10 - 3
’5
0-05 - - 0.35 8
0.00 -
I l l l l l l 0.30

 

 

 

0100200300400500600700800900
Ir film thickness (nm)

Figure 10.4 Variation of (002) Ir line width and the ratio of (1 1 1) to (002) reflected x-ray
intensity with Ir film thickness.

Variation of linewidth and I(111)/I(002) was also studied by changing Ar pressure
from 5 to 17 mTorr. Four Ir films each 500 nm thick were grown at 800 °C and 3 nm/
min deposition rate at different pressures. Figure 10.5 shows the results of this study. The
(002) Ir linewidth is a minimum at 12 mTorr, whereas I(111)/I(002) is minimum at 8

mTorr.

168

The (002) Ir linewidth is much smaller for Ir on SrTiO3 than for the film grown

on A1203. However comparable linewidth values were found for electron-beam

evaporated Ir on both of these substrates in previous studies [11] The reason for this

difference is not clear presently and further investigation is required.

 

 

 

 

 

15 I I I I I I 0.65

12 _ - 0.60
8
A ~ 0.55 6
.\° 9 . :3
§ - 0.50 g
g 6 - E.
5 - 0.45 g
F 3'
F A
=' 3 ' . 0.40 3
E

0 - ._ .9 0.35

4 6 8 10 12 14 1e 10

Growth pressure (mTorr)

Figure 10.5 Variation of (002) Ir linewidth and the ratio of (111) to (002) reflected x-ray
intensities with Ar pressure.

169

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

11 References

W. Adam, B. Berdermann, P. Bergonzo, W. De Boer, and R D collaboration,
"The development of diamond tracking detectors for the LHC," Nuclear
instruments and methods in physics research A 514, 79-86, 2003.

W. Adam , B. Bellini, E. Berdermann, P. Bergonzo, and R D collaborations,
"status of the R & D activity on diamond particle detectors," Nuclear
instruments and methods in physics research A 511, 124-131, 2003.

J. Isberg, J. Hammersberg, E. Johansson, T. Wikstrom, D. J. Twitchen, A. J.
Whitehead, S. E. Coe, and G. A. Scarsbrook, "High carrier mobility in single-
crystal plasma-deposited diamond," Science 297, 1670-1672, 2002.

X. Jiang and C.-P. Klages, "Heteroepitaxial diamond growth on (100)
silicon," Diamond and Related Materials 2 l 1 12-1 1 13, 1993.

C. Wild, P. Koidl, W. Mullersebert, H. Walcher, R. Kohl, N. Herres, R.
Locher, R. Samlenski, and R. Brenn, "Chemical-Vapor-Deposition and
Characterization of Smooth (100)- Faceted Diamond F ilms," Diamond and
Related Materials 2, 158-168, 1993.

K. Ohtsuka, K. Suzuki, A. Sawabe, and T. Inuzuka, "Epitaxial growth of
diamond on iridium," Japanese Journal of Applied Physics Part 2 35, L1072,
1996.

K. Ohtsuka, H. Fukuda, K. Suzuki, and A. Sawabe, "Fabrication of epitaxial
Diamond Thin Film on iridium (Reprint)," Japanese Journal of Applied
Physics Part 2 36, L1214, 1997.

T. Tsubota, M. Ohta, K. Kusakabe, S. Morooka, M. Watanabe, and H. Maeda,
"Heteroepitaxial grth of diamond on an iridium (100) substrate using

microwave plasma-assisted chemical vapor deposition," Diamond and Related
Materials 9, 1380-1387, 2000.

F. Hormann, H. Roll, M. Schreck, and B. Stritzker, "Epitaxial Ir layers on
SrTiO3 as substrates for diamond nucleation: deposition of the films and

modification in the CVD environment," Diamond and Related Materials 9,
256-261, 2000.

M. Schreck, F. Hormann, H. Roll, J. K. N. Lindner, and B. Stritzker,
"Diamond nucleation on iridium buffer layers and subsequent textured

growth: A route for the realization of single-crystal diamond films," Applied
Physics Letters 78, 192-194, 2001.

Z. Dai, C. Bednarski-Meinke, R. Loloee, and B. Golding, "Epitaxial (100)
iridium on A-plane sapphire: A system for wafer-scale diamond
heteroepitaxy," Applied Physics Letters 82, 3847-3849, 2003.

170

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

C. S. Yan and Y. K. Vohra, "Multiple twinning and nitrogen defect center in

chemical vapor deposited homoepitaxial diamond," Diamond and Related
Materials 8, 2022-2031, 1999.

C. S. Yan, Y. K. Vohra, H. Mao, and R. J. Hemley, "Very high growth rate
chemical vapor deposition of single crystal diamond," PNAS 99, 12523-
12525, 2002.

A. Tallaire, J. Achard, F. Silva, R. S. Sussmann, and A. Gicquel,
"Homoepitaxial deposition of high-quality thick diamond films: effect of
growth parameters," Diamond and Related Materials 14, 249-254, 2004.

G. Bogdan, M. Nesladek, J. D. Haen, V. V. Moshchalkov, K. Haenen, and M.
D'Olieslaeger, "Growth and characterization of nearly-atomically flat, thick
homoepitaxial CVD diamond films," Physica Status Solidi 202, 2066-2072,
2005.

M. Behravan, "Electrical conductivity of heteroepitaxial diamond," Ph.D.
thesis, Physics and Astronomy. East Lansing, MI: Michigan State University,
2005.

C. Y. Fong and B. M. Klein, "Electronic and vibrational properties of bulk
diamond," in Diamond: Electronic properties and applications, L. S. Pan and
D. R. Kania, Eds. Boston: Kluwer Academic publisher, 1995, 1-29.

K. Miyoshi, "Structures and mechanical properties of natural and synthetic
diamond, report," NASA, Lewis Research Center, Cleveland, Ohio 107249,
Chapter 8, 1998.

T. Teraji, M. Hamada, H. Wada, and M. Yamamoto, "High rate growth and
electrical/optical properties of high-quality homoepitaxial (100) diamond
films," Diamond and Related Materials 14, 255-260, 2005.

A. K. Ekimov, V. A. Sidorov, E. D. Bauer, N. N. Melnik, N. J. Curro, J. D.
Thomson, and S. M. Stishov, "Superconductivity in diamond," Nature 428,
542-545, 2004.

I. Sakaguchi, M. N. Gamo, Y. Kikuchi, E. Yasu, H. Haneda, T. Suzuki, and T.
Ando, "Sulfur: A donor dopant for n-type diamond semiconductors," Physical
Review B 60, R2139-R2141, 1999.

R. Kalish, A. Reznik, C. Uzan-Saguy, and C. Cytermann, "Is sulfur a donor in
diamond?," Applied Physics Letters 76, 757-759, 2000.

A. Stolz, M. Behravan, M. Regmi, and B. Golding, "Heteroepitaxial diamond

detectors for heavy ion beam tracking," Diamond and Related Materials 15,
807-810, 2006.

J. Wilks and E. Wilks, Properties and applications of diamond, lst edition.
Oxford, UK: Butterworth Heinemann, 1991.

G. Davies, "The A nitrogen aggregate in diamond- its symmetry and possible
structure," Journal of Physics C: Solid State Physics 9, L537-L542, 1976.

171

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

A. Mainwood, "Points defects in natural and synthetic diamond: what they can
tell us about CVD diamond," Physica Status Solidi 172, 25-35, 1999.

G. S. Woods, "Platelets and the infiared absorption of type III diamond,"

Proceedings of the Royal society of London A: Mathematical and Physical
Sciences 407, 219-238, 1986.

I. Kiflawi, A. Mainwood, H. Kanda, and D. Fisher, "Nitrogen interstitials in
diamond," Physical Review B 54, 16719-16726, 1996.

A. T. Collins, "The fermi level in diamond," Journal of Physics: Condensed
Matter 14, 3743-3750, 2002.

A. T. Collins, "The Optical and Electronic-Properties of Semiconducting
Diamond," Philosophical Transactions of the Royal Society of London Series
a-Mathematical Physical and Engineering Sciences 342, 233-244, 1993.

F. Fuchs, C. Wild, K. Schwarz, and P. Koidl, "Hydrogen-Related IR
absorption in chemical vapor deposited diamond," Diamond and Related
Materials 4, 652-656, 1995.

J. Venables, G. Spiller, and M. Hanbuecken, "Nucleation and grth of thin
films," Report in Progress in Physics 47, 399, 1984.

J. A. Venables, "Nucleation calculations in a pair binding model," Physical
Review B 36, 4153-4162, 1986.

J. E. Field, "The properties of natural and synthetic diamond." London:
Academic Press, 1992.

X. Jiang, "Epitaxial diamond thin-films on (001) silicon substrates," Applied
Physics Letters 62, 3438-3440, 1993.

S. Wolter, B. Stoner, J. G. P. Ellis, D. Buhaenko, C. Jenkins, and P.
Southworth, "Textured growth of diamond on silicon via in situ carburization
and bias-enhanced nucleation," Appied Physics Letters 62, 1215, 1993.

N. Ishigaki and S. Yugo, "Mechanism of diamond epitaxial growth on
silicon," Diamond and Related Materials 9,. 1646-1649, 2000.

X. Jiang, "High quality heteroepitaxial diamond films on silicon: recent
progress," Diamond and Related Materials 9, 1640-1645, 2000.

X. Jiang, K. Schiffrnann, C. P. Klages, D. Wittorf, C. L. Jia, K. Urban, and W.
Jager, "Coalescence and overgrth of diamond grains for improved
heteroepitaxy on silicon(001)," Journal of Applied Physics 83, 2511-2518,
1998.

S. Koizumi, T. Murakami, T. Inuzuka, and K. Suzuki, "Epitaxial-growth of
diamond thin-films on cubic boron nitride(111) surfaces by DC plasma
chemical vapor-deposition," Applied Physics Letters 57, 563-565, 1990.

S. Koizumi, "Heteroepitaxy of diamond on cubic boron nitride," Israel
Journal of Chemistry 38, 33-40, 1998.

172

[42]

[43]

[441

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

[54]

[55]

W. Yarbrough and R. Messier, "Current issues and problems in the chemical
vapor deposition of diamond," Science 247, 688, 1990.

H. Kawarada, T. Suesada, and H. Nagasawa, "Heteroepitaxial growth of
smooth and continuous diamond thin films on silicon substrates via high-
quality silicon-carbide buffer layers," Applied Physics Letters 66, 583-585,
1995.

H. Kawarada, C. Wild, N. Herres, R. Locher, P. Koidl, and H. Nagasawa,
"Heteroepitaxial growth of highly oriented diamond on cubic silicon carbide,"
Journal of Applied Physics. 81, 3490, 1997.

T. Tachibana, Y. Yokota, K. Miyata, T. Onishi, K. Kobashi, M. Tarutani, Y.
Takai, R. Shimizu, and Y. Shintani, "Diamond films heteroepitaxially grown
on platinum (1 1 1)," Physical Review B 56, 15967-15981, 1997.

T. Tachibana, Y. Yokota, K. Kobashi, and M. Yoshimoto, "Heteroepitaxial
growth of {111}-oriented diamond films on platinum {1 ll}/sapphire {0001}
substrates," Journal of Crystal Growth 205, 163-168, 1999.

T. Tachibana, Y. Yokota, H. Hayashi, and K. Kobashi, "Growth of {111}-
oriented diamond on Pt/Ir/Pt/ substrate deposited on sapphire," Diamond and
Related Materials 10, 1633-1636, 2001.

T. Tachibana, Y. Yokota, K. Kobashi, and Y. Shintani, "X-ray diffraction pole
figure measurements of diamond films grown on platinum (111)," Journal of
Applied Physics 82, 4327-4330, 1997.

T. Tachibana, Y. Yokota, K. Nishimura, K. Miyata, K. Kobashi, and Y.
Shintani, "Heteroepitaxial diamond growth on platinum(111) by the Shintani
process," Diamond and Related Materials 5, 197-199, 1996.

W. Zhu, P. C. Yang, and G. J. T., "Oriented diamond films grown on nickel
substrates," Applied Physics Letters 63, 1640-1642, 1993.

P. C. Yang, R. Schlesser, C. A. Wolden, W. Liu, R. F. Davis, and Z. Sitar,
"Control of diamond heteroepitaxy on nickel by optical reflectance," Applied
Physics Letters 70, 2960-2962, 1997.

Z. Sitar, W. Liu, P. C. Yang, C. A. Wolden, R. Schlesser, and J. T. Prater,
"Heteroepitaxial nucleation of diamond on nickel," Diamond and Related
Materials 7, 276-282, 1998.

W. Liu, P. C. Yang, C. A. Wolden, R. F. Davis, J. T. Prater, and Z. Sitar,
"Transmission electron microscopy analysis of the oriented diamond growth
on nickel substrates," Journal of Applied Physics 83, 7658-7663, 1998.

T. Bauer, M. Schreck, S. Gsell, F. Hormann, and B. Stritzker, "Epitaxial
rhenium buffer layers on A1203 (0001): a substrate for the deposition of(111)-
oriented heteroepitaxial diamond films," Physica Status Solidi 199, 19-26,
2003.

M. Hansen, Constitution of Binary Alloys, Second edition New York, Toronto,
London: McGraw-Hill Book Co., 1958.

173

[56]

[57]

[58]

[59]

[60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

M. Schreck, H. Roll, and B. Stritzker, "Diamond/Ir/SrTiO3: A material
combination for improved heteroepitaxial diamond films," Applied Physics
Letters 74, 650, 1999.

T. Fujisaki, M. Tachiki, N. Taniyama, M. Kudo, and H. Kawarada,
"Fabrication of heteroepitaxial diamond thin films on 1r(001)/MgO(001)
substrates using antenna-edge-type microwave plasma-assisted chemical
vapor deposition," Diamond and Related Materials 11, 478-481, 2002.

C. Bednarski, Z. Dai, A.-P. Li, and B. Golding, "Studies of heteroepitaxial
growth of diamond," Diamond and Related Materials, 12, 241-245, 2003.

K. Ohtsuka, H. Fukuda, K. Suzuki, and A. Sawabe, "Fabrication of epitaxial
diamond thin film on iridium," Japnanese Journal of Applied Physics Part 2,
36, L1214,1997.

K. Suzuki, H. Fukuda, T. Yamada, and A. Sawabe, "Epitaxially grown free-
standing diamond platelet," Diamond and Related Materials, 10, 2153-2156,
2001.

Z. Dai, C. Bednarski, and B. Golding, "Heteroepitaxial diamond film growth:
the a-plane sapphire-iridium system," Diamond and Related Materials 13,
552-556, 2004.

C. H. Lee, J. Qi, L. S. T, and L. S. Hung, "Epitaxial diamond on a Si/Can/Ir
substrate," Diamond and Related Materials 12, 1335-1339, 2003.

T. Bauer, S. Gsell, M. Schreck, J. Goldfub, L. J, D. G. Schlom, and B.
Stritzker, "Growth of epitaxial diamond on silicon via iridium/SrTi03 buffer
layers," Diamond and Related Materials 14, 314-317, 2004.

S. Gsell, M. Fischer, T. Bauer, M. Schreck, and B. Stritzker, "Yttria-
stabilized zirconia films of different composition as buffer layers for the
deposition of epitaxial diamond/Ir layers on Si (001)," Diamond and Related
Materials 15, 479-485, 2006.

S. Yugo, T. Kanai, T. Kimura, and T. Muto, "Generation of diamond nuclei
by electric field in plasma chemical vapor depostion," Applied Physics Letters
58, 1036, 1991.

J. Robertson, "Mechanism of bias-enhanced nucleation and heteroepitaxy of
diamond on Si," Diamond and Related Materials 4, 549-552, 1995.

S. Yugo, N. Nobuyuki, and K. Tadamasa, "Analysis of heteroepitaxial
mechanism of diamond grown by chemical vapor deposition," Diamond and
Related Materials 7, 1017-1020, 1998.

Y. Lifshitz, S. R. Kasi, and J. W. Rabalais, "Subplantation model for film
growth fi'om hyperthermal species: Application to diamond," Physical Review
Letters 62, 1290-1293, 1989.

J. Robertson, J. Gerber, S. Sattel, M. Weiler, K. Jung, and H. Ehrhardt,
"Mechanism of bias-enhanced nucleation of diamond on Si," Applied Physics
Letters 66, 3287-3289, 1995.

174

[70]

[71]

[72]

[73]

[74]

[75]

[76]

[77]

[78]

[79]

[80]

[81]

H. P. Kaukonen and R. M. Nieminen, "Molecular-dyanamics simulation of the
growth of diamond like films by energetic carbon atom beams," Physical
Review Letters 68, 620-623, 1992

S. Katai, Z. Tass, G. Hars, and P. Deak, "Measurement of ion energy
distributions in the bias enhanced nucleation of chemical vapor deposited
diamond," Journal of Applied Physics 86, 5549, 1999.

S. Katai, A. Kovats, I. Maros, and P. Deak, "Ion energy distributions and their
evolution during bias-enhanced nucleation of chemical vapor deposition of
diamond," Diamond and Related Materials 9, 317-321, 2000.

Y. Lifshitz, T. Kohler, T. Frauenheim, I. Guzmann, A. Hoffman, R. Q. Zhang,
X. T. Zhou, and S. T. Lee, "The mechanism of diamond nucleation from
energetic species," Science 297, 1531-1533, 2002.

B. Sheldon, R. Csencsits, J. Rankin, R. Boekenhauer, and Y. Shigesato, "Bias-
enhanced nucleation of diamond during microwave-assisted chemical vapor
deposition," Journal of Applied Physics 75, 5001, 1994.

S. P. McGinnis, M. A. Kelly, S. B. Hagstrom, and R. L. Alvis, "Observation
of diamond nanocrystals in carbon films deposited during ion-assisted

microwave plasma nucleation pretreatments," Journal of Applied Physics 79,
170-174, 1996.

S. McGinnis, M. Kelly, and S. Hagstrom, "Evidence of an energetic ion
bombardment mechanism for bias enhanced nucleation of diamond," Applied
Physics Letters 66, 3117-3119, 1995.

Y. Shigesato, R. E. Boekenhauer, and B. W. Sheldon, "Emission-spectroscopy
during direct-current-biased, microwave- plasma chemical-vapor-deposition
of diamond," Applied Physics Letters 63, 314-316, 1993.

E. Maillard-Schaller, O. M. Kuttel, P. Groning, O. Groning, R. G. Agostino,
P. Aebi, L. Schlapbach, P. Wurzinger, and P. Pongratz, "Local heteroepitaxy
of diamond on silicon (100): A study of the interface structure," Physical
Review B 55, 15895-15904, 1997.

M. Stammler, R. Stockel, L. Ley, M. Albrecht, and H. P. Strunk, "Diamond
nucleation on silicon during bias treatment in chemical vapour deposition as

analysed by electron microscopy," Diamond and Related Materials, 6, 747-
751, 1997.

R. Stockel, M. Stammler, K. Janischowsky, L. Ley, M. Albrect, and H. P.
Strunk, "Diamond nucleation under bias conditions," Journal of Applied
Physics 83, 531, 1998.

S. Pecoraro, J. C. Amault, and J. Werkmann, "BEN-HFCVD diamond
nucleation on Si (111) investigated by HRTEM and nanodiffraction.,"
Diamond and Related Materials 14, 137-143, 2004.

175

[82]

[86]

[87]

[88]

[89]

[90]

[91]

[92]

[93]

B. R. Stoner, G. H. M. Ma, S. D. Wolter, and J. T. Glass, "Characterization of
bias-enhanced nucleation of diamond on silicon by in-vacuo surface-analysis

and transmission electron- microscopy," Physical Review B 45, 11067-11084,
1992.

X. Jiang, "Effects of ion bombardment on the nucleation and growth of
diamond films," Physical Review B, 58, 7064-7075, 1998.

N. Jiang, B. W. Sun, Z. Zhang, and Z. Lin, "Nucleation and initial grth of
diamond film on Si substrate," Journal of Materials Research 9, 2695-2702,
1994.

B. R. Stoner, G. H. Ma, S. D. Wolter, W. Zhu, Y. C. Wang, R. F. Davis, and J.
T. Glass, "Epitaxial nucleation of diamond on beta-SiC via bias-enhanced

microwave plasma chemical-vapor-deposition," Diamond and Related
Materials 2, 142-146, 1993.

W. L. Wang, K. J. Liao, J. Wang, L. Fang, P. D. Ding, J. Esteve, M. C. Polo,
and G. Sanchez, "Mechanism of diamond nucleation enhancement by electron

emission via hot filament chemical vapor deposition," Diamond and Related
Materials 8, 123-126, 1999.

W. L. Wang, G. Sanchez, M. C. Polo, R. Q. Zhang, and J. Esteve, "Nucleation
and initial growth of diamond by biased hot filament chemical vapour

deposition," Applied Physics a-Materials Science & Processing 65, 241-249,
1997.

M. Tachiki, T. Fujisaki, N. Taniyama, M. Kudo, and H. Kawarada,
"Heteroepitaxial diamond thin film growth on Ir (001)/ MgO(001) substrate
by antenna-edge plasma assisted chemical vapor deposition," Journal of
Crystal Growth 237-239, 1277-1280, 2002.

A. Sawabe, H. Fukuda, T. Suzuki, and Y. Ikuhara, "Interface between CVD
diamond and iridium films," Surface Science 467, L845-L849, 2000.

T. Fujisaki, M. Tachiki, N. Taniyama, M. Kudo, and H. Kawarada, "Initial
grth of heteroepitaxial diamond on Ir(001)/MgO(001) substrates using

antenna-edge-type microwave plasma assisted chemical vapor deposition,"
Diamond and Related Materials 12, 246-250, 2003.

S. Kono, T. Takano, T. Goto, Y. Ikejima, M. Shiraishi, T. Abukawa, T.
Yamada, and A. Sawabe, "Effect of bias treatment in the CVD diamond
growth on Ir(001)," Diamond and Related Materials 13, 2081-2087, 2004.

S. Kono, M. Shiraishi, N. I. Plusnin, T. Goto, Y. Ikejima, T. Abukawa, M.
Shimomura, Z. Dai, C. Bednarski-Meinke, and B. Golding, "X-ray
photoelectron study of the initial stages of CVD diamond heteroepitaxy on
Ir(001)/SrTiO3," New Diamond and Frontiers Carbon Technology 16, (363-
371), 2005.

F. Hormann, M. Schreck, and B. Stritzker, "First stages of diamond nucleation
on iridium buffer layers," Diamond and Related Materials 10, 1617-1621,
2001.

176

[94]

[95]

[96]

[97]

[98]

[99]

[100]

[101]

[102]

[103]

[104]

[105]

[106]

H. Hormann, T. Bauer, M. Schreck, S. Gsell, and B. Stritzker, "TEM analysis
of nanometer-size surface structures formed by bias enhanced nucleation of
diamon on iridium.," Diamond and Related Materials 12, 350-355, 2003.

M. Schreck, H. Hormann, S. Gsell, T. Bauer, and B. Stritzker, "Transmission
electron microscopy study of the diamond nucleation layer on iridium,"
Diamond and Related Materials 15, 460-464, 2005.

T. Bauer, S. Gsell, F. Hormann, M. Schreck, and B. Stritzker, "surface
modifications and first stages of heteroepitaxial diamond growth on iridium,"
Diamond and Related Materials 13, 335-341, 2004.

H. Hormann, H. Y. Peng, T. Bauer, Q. Li, M. Schreck, Y. Lifshitz, S. T. Lee,
and B. Stritzker, "Flat epitaxial diamond/ Ir (001) interface visualized by high

resolution transmission electron microscopy," Surface Science 513, 525-529,
2002.

M. Schreck, T. Bauer, G. S., H. Hormann, H. Bielefeldt, and B. Stritzker,
"Domain formation in diamond nucleation on iridium," Diamond and Related
Materials 12, 262-267, 2003.

S. Gsell, M. Schreck, G. Benstetter, E. Lodermeier, and B. Stritzker,
"Combined AFM-SEM study of the diamond nucleation layer on Ir(001),"
Diamond and Related Materials 16, 665-670, 2007.

K. Hayashi, S. Yamanaka, H. Okushi, and K. Kajimura, "Stepped growth and
etching of (001)diamond," Diamond and Related Materials, 5, 1002-1005,
1996.

K. Hayashi, S. Yamanaka, H. Okushi, and K. Kajimura, "Homoepitaxial
diamond films with large terraces," Applied Physics Letters 68, 1220-1222,
1996.

D. Takeuchi, S. Yamanaka, H. Watanabe, S. Sawada, H. Ichinose, H. Okushi,
and K. Kajimura, "High quality homoepitaxial diamond thin film synthesis
with high grth rate by a two-step growth method," Diamond and Related
Materials 8, 1046-1049, 1999.

H. Watanabe, D. Takeuchi, S. Yamanaka, H. Okushi, K. Kajimura, and T.
Sekiguchi, "Homoepitaxial diamond film with an atomically flat surface over
a large area," Diamond and Related Materials 8, 1272-1276, 1999.

C. L. Wang, M. Irie, and T. Ito, "Growth and characterization of hillock-free
high quality homoepitaxial diamond films," Diamond and Related Materials
9, 1650-1654, 2000.

H. Okushi, "High quality homoepitaxial CVD diamond for electronic
devices," Diamond and Related Materials 10, 281-288, 2001.

H. Okushi, H. Watanabe, S. Ri, S. Yamanaka, and D. Takeuchi, "Device-
grade homoepitaxial diamond film growth," Journal of Crystal Growth 237-
239, 1269-1276, 2002.

177

 

[107]

[108]

[109]

[110]

[111]

[112]

[113]

[114]

[115]

[116]

[117]

[118]

[119]

M. stammler, H. Eisenbei, J. Ristein, J. Neubauer, M. Gobbels, and L. Ley,
"Growth of high quality homoepitaxial diamond films by HF -CVD," Diamond
and Related Materials 11, 504-508, 2002.

D. Takeuchi, H. Watanabe, S. Yamanaka, H. Okushi, and K. Kajimura,
"Homoepitaxial diamond films grown by step-flow mode in various

misorientation angles of diamond substrates," Diamond and Related Materials
9, 231-235, 2000.

T. Teraji and T. Ito, "Homoepitaxial diamond growth by high power
microwave-plasma chemical vapor deposition," Journal of Crystal Growth
271, 409-419, 2004.

O. A. Williams and R. B. Jackman, "High growth rate MWPECVD of single
crystal diamond," Diamond and Related Materials 13, 557-560, 2004.

O. A. Williams and R. B. Jackman, "Homoepitaxial growth for surface
conductive device applications," Diamond and Related Materials 13, 325-328,
2004.

T. Teraji, "Chemical vapor deposition of homoepitaxial diamond films,"
Physica Status Solidi 203, 3324-3357, 2006.

W. J. P. Van Enckevort, G. Janssen, W. Vollenberg, and L. J. Giling,
"Anisotropy in monocrystalline CVD diamond growth. Surface morphology

of hot filament grown films deposited on planar substrates," Journal of
Crystal Growth 148, 365-382, 1994.

C. Wang, M. Irie, and T. Ito, "Growth and characterization of hillock-flee
high quality homoepitaxial diamond films," Diamond and Related Materials
9, 1650-1654, 2000.

T. Bauer, M. Schreck, H. Sternschulte, and B. Stritzker, "High growth rate
homoepitaxial diamond deposition on off-axis substrates," Diamond and
Related Materials 14, 266-271, 2004.

N. Lee and A. Badzian, "A study on surface morphologies of (001)
homoepitaxial diamond films," Diamond and Related Materials 6, 130-145,
1996.

N. S. Lee and A. Badzian, "Effect of misorientation angles on the surface
morphologies of (001) homoepitaxial diamond thin-films," Applied Physics
Letters 66, 2203-2205, 1995.

N. Lee and A. Badzian, "Effect of methane concentrations on surface
morphologies and surface-structures of (001) homoepitaxial diamond thin-
filrns," Applied Physics Letters 67, 2011-2013, 1995.

K. Hayashi, H. Watanabe, S. Yamanaka, H. Okushi, K. Kajimura, and T.
Sekiguchi, "Direct observation of hydrogen-related luminescent states in

subsurface region of homoepitaxial diamond films," Applied Physics Letters
69, 1122-1124, 1996.

178

[120]

[121]

[122]

[123]

[124]

[125]

[126]

[127]

[128]

[129]

[130]

[131]

[132]

K. Hayashi, S. Yamanaka, H. Watanabe, T. Sekiguchi, H. Okushi, and K.
Kajimura, "Diamond films epitaxially grown by step-flow mode," Journal of
Crystal Growth 183, 338-346, 1998.

F. K. de Theije, J. J. Schermer, and W. J. P. Van Enckevort, "Effetcs of
nitrogen impurities on the CVD growth of diamond: step bunching in theory
and experiment," Diamond and Related Materials 9, 1439-1449, 2000.

E. Van Veenendaal, F. K. de Theije, J. van Suchtelen, and W. J. P. van
Enckevort, "simulation of step patterns on natural diamond (111) surfaces,"
Diamond and Related Materials 11, 145-152, 2002.

A. Tallaire, A. T. Collins, D. Charles, J. Achard, R. S. Sussmann, A. Gicquel,
M. E. Newton, A. M. Edmonds, and R. J. Cruddace, "Characterization of
high-quality thick single-crystal diamond grown by CVD with a low nitrogen
addition," Diamond and Related Materials 15, 1700-1707, 2006.

T. Teraji, M. Hamada, H. Wada, M. Yamamoto, and T. Ito, "High-quality
homoepitaxial diamond (100) films grown under high-rate growth condition,"
Diamond and Related Materials 14, 1747-1752, 2005.

C. J. Chu, R. H. Hague, J. L. Margrave, and M. P. D'Evelyn, "Growth kinetics
of (100) (110) and (111) homoepitaxial diamond films," Applied Physics
letters, 61, 1393-1395, 1992.

J. Achard, A. Tallaire, R. S. Sussmann, F. Silva, and A. Gicquel, "The control
of growth parameters in the synthesis of high quality single crystalline
diamond by CVD," Journal of Crystal Growth 284, 396-405, 2005.

G. Bogdan, M. Nesladek, J. D. Haen, K. Haenen, and M. D'Olieslaeger,
"Freestanding (100) homoepitaxial CVD diamond," Diamond and Related
Materials 15, 508-512, 2006.

Y. Mokuno, A. Chayahara, Y. Soda, H. Yamada, Y. Horino, and N. Fujimori,
"High rate homoepitaxial grth of diamond by microwave plasma CVD with
nitrogen addition," Diamond and Related Materials 15, 455-459, 2006.

A. Tallaire, J. Achard, F. Silva, and A. Gicquel, "Effect of increasing the
microwave density in both continuous and pulsed wave mode on the growth
of monocrystalline diamond films," Physica Status Solidi 202, 2059-2065,
2005.

S. Jin and T. D. Moustakas, "Effect of nitrogen on the growth of diamond
films," Applied Physics letters 65, 403-405, 1994.

A. Chayahara, Y. Mokuno, Y. Horino, Y. Takasu, H. Kato, H. Yoshikawa,
and N. Fujimori, "The effect of nitrogen addition during high-rate
homoepitaxial growth of diaomd by microwave plasma CVD," Diamond and
Related Materials 13, 1954-1958, 2004.

R. Locher, C. Wild, N. Herres, D. Behr, and P. Koidl, "Nitrogen stabilized
(100) texture in chemical-vapor-deposited diamond films," Applied Physics
Letters 65, 34-36, 1994.

179

 

[133]
[134]

[135]

[136]

[137]

[138]

[139]

[140]

[141]
[142]
[143]
[144]
[145]

[146]

[147]

Y. Bar-Yam and T. D. Moustakas, "Defect-induced stabilization of diamonds
films," Nature 342, 786-787, 1989.

S. Ferro, "Synthesis of diamond," Journals of Materials Chemistry, 12, 2843-
2855, 2002.

P. Smereka, X. Li, G. Russo, and D. J. Srolovitz, "Simulation of faceted film

growth in three dimensions: microstructure, morphology and texture," Acta
Materialia 53, 1191-1204, 2004.

F. Silva, J. Achard, X. Bonnin, A. Michau, A. Tallaire, O. Brinza, and A.
Gicquel, "3D crystal growth model for understanding the role of plasma pre-
treatment on CVD diamond crystal shape," physica status solidi 203, 3049-
3055, 2006.

D. B. Murphy, E. D. Salmon, K. R. Spring, M. Abramowitz, and M. W.
Davidson, "Differential interference Contrast Microscopy," Olympus
micrscopy resource center, 2007.

C. B. Prater, P. G. Maivald, K. J. Kjoller, and M. G. Heaton, "Tapping Mode
Imaging: Applications and Technology," Digital Instruments,
http://www.di.com, Application note AN04.

S. Flegler, J. Heckman, and K. Klomparens, Scanning and Transmission
Electron Microscopy: An Introduction. New York: W.H. Freeman and
Company, 1993.

C. R. Bednarski-Meinke, "Nucleation and growth of heteroepitaxial
diamond," Ph.D. thesis, Physics and Astronomy, East Lansing, MI: Michigan
State University, 2002.

M. P. Marder, Condensed Matter Physics. New York: Wiley-interscience
Publication 2000.

D. S. Knight and W. B. White, "Characterization of diamond films by Raman-
spectroscopy," Journal of Materials Research, 4, 385-393, 1989.

S. A. Solin and A. K. Ramdas, "Raman spectrum of diamond," Physical
Revin 1, 1687-1698, 1970.

J. S. Blakemore, Solid State Physics, Second edition. New York: Cambridge
University Press, 1985.

H. L. Shriver and N. M. Rosengaard, "Surface energy and work function of
elemental metals," Physical Review B, 46, 7157-7168, 1992.

R. Vargas, T. Goto, W. Zhang, and T. Hirai, "Epitaxial-Growth of iridium and
platinum films on sapphire by metalorganic chemical-vapor-deposition,"
Applied Physics Letters 65, 1094-1096, 1994.

Y. M. Sun, J. P. Endle, K. Smith, S. Whaley, R. Mahaffy, J. G. Ekerdt, J. M.
White, and R. L. Hance, "Iridium film growth with iridium tris-
acetylacetonate: oxygen and substrate effects," Thin Solid Films 346, 100-107,
1999.

180

[148]

[149]

[150]

[151]

[152]

[153]

[154]
[155]
[156]

[157]

[158]

[159]

[160]

[161]

M. A. El Khakani, B. Le Drogoff, and M. Chaker, "Effect of the deposition
temperature on the properties of iridium thin films grown by means of pulsed
laser deposition," Journal of Materials Research 14, 3241-3246, 1999.

B. Le Drogoff, M. A. El Khakani, P. R. M. Silva, M. Chaker, and G. G. Ross,
"Surface properties of pulsed laser deposited Ir, Rh, and Ir (0.9) Rh (0.1) thin
films for use as microelectrode arrays in electroanalytical heavy metal trace
sensors," Applied Surface Science 152, 77-84, 1999.

M. A. El Khakani, M. Chaker, and B. Le Drogoff, "Iridium thin films
deposited by radio-frequency magnetron sputtering," Journal of Vacuum
Science & Technology a- Vacuum Surfaces and Films 16, 885-888, 1998.

S. Gsell, T. Bauer, J. Goldfuss, M. Schreck, and B. Stritzker, "A route to
diamond wafers by epitaxial deposition on silicon via iridium/yttria-stabilized
zirconia buffer layers," Applied Physics Letters 84, 4541-4543, 2004.

G. Koster, B. L. Kropman, G. Rijnders, D. H. A. Blank, and H. Rogalla,
"Quasi-ideal strontium titanate crystal surfaces through formation of strontium
hydroxide," Applied Physics Letters 73, 2920-2922, 1998.

M. Kawasaki, A. Ohtomo, T. Arakane, K. Takahashi, M. Yoshimoto, and H.
Koinuma, "Atomic control of SrTiO3 surface for perfect epitaxy of perovskite
oxides," Applied Surface Science 107, 102-106, 1996.

D. L. Medlin, (Private Communication)
W. Rasband, "Image/J," version 1.28, National Institutes of Health, 2006.

B. Golding, C. Bednarski-Meinke, and Z. Dai, "Diamond heteroepitaxy:
Pattern formation and mechanisms," Diamond and Related Materials 13, 545-
551, 2004.

J. W. Matthews and A. E. Blakeslee, "Defects in epitaxial multilayers, misfit
dislocations," Journal of Crystal Growth 27, 118-125, 1974.

S. Barrat, S. Saada, I. Dieguez, and E. Bauer-Grosse, "Diamond deposition by
chemical vapor deposition process: Study of the bias enhanced nucleation
step," Journal of Applied Physics 84, 1870, 1998.

K. K. Irilcura and J. L. Beauchamp, "Electronic structure considerations for

methane activation by third-row transition-metal ions," Journal of Physical
Chemistry 95, 8344-8351, 1991.

R. Perez-Casero, J. Perriere, A. Gutierrez-Llorente, and D. Defourneau, "Thin
films of oxygen-deficient perovskite phases by pulsed-laser ablation of
strontium titanate," Physical Review B 75, 165317-1 tol65317-7, 2007.

N. V. Surovtsev, I. N. Kupriyanov, V. K. Malinovsky, V. A. Gusev, and Y. N.
Pal'yanov, "Effect of nitrogen impurities on the Raman linewidth in
diamonds," Journal of Physics: Condensed Matter 11, 4767-4774, 1999.

181

 

[162]

[163]
[164]

[165]

[166]

[167]

[168]

[169]

[170]

[171]

[172]

[173]

F. Bogani, S. Miglio, S. Sciortino, A. Taschin, and A. Vinattieri, "Free-

exciton photoluminescence as a probe of CVD diamond," Physica status
solidi 181, 23-28, 2000.

R. J. Anderson, (Private communication)

R. Samlenski, J. Schmalzin, R. Brenn, C. Wild, W. Muller-Sebert, and P.
Koidl, "Characterization of homoepitaxial diamond films by nuclear
methods," Diamond and Related Materials 4, 503-507, 1995.

A. R. Lang, M. Moore, A. P. W. Makepeace, W. Wierzchowski, and C. M.
Welbomn, "On the dilation of synthetic type Ib diamond by substitutional

nitrogen," Philosophical Transactions: Physical Sciences and Engineering,
337, 497-520, 1991.

C. Tavares, S. Koizumi, and H. Kanda, "Effects of RIE treatments for (111)
diamond substrates on the growth of P-doped diamond thin films," Physica
status solidi 202, 2129-2133, 2005.

A. Tallaire, J. Achard, F. Silva, R. S. Sussmann, A. Gicquel, and E. Rzepka,
"Oxygen plasma pre-treatrnents for high quality homoepitaxial CVD diamond
deposition," Physica Status Solidi 201, 2419-2424, 2004.

W. Muller-Sebert, E. Worner, F. Fuchs, C. Wild, and P. Koidl, "Nitrogen
induced increase of growth rate in chemical vapor deposition," Applied
Physics Letters 68, 759-760, 1995.

T. Bauer, M. Schreck, J. Hartwig, X. H. Liu, S. P. Wong, and B. Stritzker,
"Structural defects in homoepitaxial diamond layers grown on off-axis Ib
HPHT substrates," Physica Status Solidi 203, 3056-3062, 2006.

M. G. Donato, G. Faggio, G. Messina, R. Potenza, S. Santangelo, M. scoccia,
C. Tuve, and V. R. G., "Characterisation of homoepitaxial CVD diamond
grown at moderate microwave power," Diamond and Related Materials 15,
517-521, 2006.

P. K. Bachmann and U. W. Detlef, "Optical characterization of diamond,"
Diamond and Related Materials 1, 422-433, 1992.

I. Kiflawi, A. E. Mayer, P. M. Spear, J. A. Vanwyk, and G. S. Woods,
"Infrared absorption by the single nitrogen defect center in diamond,"
Philosophical Magazine B 69, 1141-1147, 1994.

J. Achard, F. Silva, O. Brinza, A. Tallaire, and A. Gicquel, "Coupled effect of
nitrogen addition and surface temperature on the morphology and the kinetics
of thick CVD diamond single crystals," Diamond and Related Materials 16,
685-689, 2007.

182