вход по аккаунту


Microwave plasma-enhanced chemical vapor deposition and structural characterization of diamond films

код для вставкиСкачать
T h e m o st ad v an ced technology has b e e n u sed to p h o to g ra p h an d
reproduce this m anuscript from the microfilm m aster. U M I films the
text directly from the original or copy submitted. Thus, some thesis and
dissertation copies are in typewriter face, while others may be from any
type o f com puter printer.
The q u ality o f th is reproduction is dependent upon th e quality o f the
copy su b m itted . B roken o r indistinct print, co lo red or p o o r quality
illustrations and photographs, print bleedthrough, substandard margins,
and im proper alignm ent can adversely affect reproduction.
In th e unlikely ev en t th a t th e author did not sen d U M I a com plete
m anuscript and th ere are missing pages, these will be noted. Also, if
unauthorized copyright m aterial had to be removed, a note will indicate
the deletion.
O versize m aterials (e.g., maps, drawings, charts) are reproduced by
sectioning th e original, beginning a t th e upper left-hand co rn er and
continuing from left to right in equal sections with small overlaps. Each
o rig in al is also p h o to g ra p h e d in o n e exposure a n d is in c lu d ed in
reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. H igher quality 6 " x 9" black an d white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. C ontact U M I directly
to order.
U niversity Microfilms International
A Bell & Howell Inform ation C o m p an y
3 0 0 N orth Z e e b R o a d , Ann Arbor, Ml 48106-1346 USA
3 1 3 /7 6 1 -4 7 0 0 8 0 0 /5 2 1 -0 6 0 0
Order Number 9105001
Microwave plasma-enhanced chemical vapor deposition and
structural characterization of diamond films
Zhu, Wei, Ph.D .
The Pennsylvania State University, 1990
300 N. Zeeb Rd.
Ann Aibor, MI 48106
The Pennsylvania State University
The Graduate School
A Thesis in
Solid State Science
Submitted in Partial Fulfillment
of the Requirements
for the Degree of
Doctor o f Philosophy
August 1990
We approve the thesis of Wei Zhu.
Date of signature
5 (lsh °
(^R ussell Messier
Professor of Engineering Science and Mechanics
Thesis Co-Advisor
Chair of Committee
Andrzej R. Badzian
Senior Research Associate and Associate
Professor of Solid State Science
Thesis Co-Advisor
s l is 170
Professor of Ceramic Science
Brace E. Knox /
Asstoeiate Professor of Materials Science
and Science, Technology, and Society
rjifo b
Paul k .. Howell
Associate Professor of Metallurgy
Robert E. Newnham
Alcoa Professor o f Solid State Science
In Charge of Graduate Programs in
Solid State Science
The theme o f this thesis is the internal structure o f diamond thin films prepared by
the low pressure chemical vapor deposition (CVD) which is a critical link between the film
deposition and the resulting properties. Various internal structural features including bulk
graphite inclusions, interfaces and structural defects, the parametric deposition processes,
the noble gas-involved plasma chemistry, and the oxidation properties o f diamond films, as
well as the relationships among them, are investigated. The results contribute to the general
understanding of the nature of the microwave plasma enhanced CVD processes o f diamond
film preparation and the resulting film structure.
A systematic study of the diamond growth behavior and its correlations with
controllable experimental parameters indicates that the diamond growth under CVD
conditions is a highly selective process w ith narrow and various sets o f optimum
conditions depending on which film characteristic or property is being optimized. The
conditions for the optimal growth o f diamond films with maximum growth rates, minimum
structural defects and large area uniform coverages can only be obtained by carefully
considering and combining each o f these parameters and, in addition, are system specific.
Compromises in choosing the experimental parameters are often necessary. The parametric
investigation provides an experimental basis on which predictive thermochemical models
can be established.
The optical emission spectroscopy o f the methane-hydrogen-noble gas plasmas
indicates that, although hydrogen is the main dilution gas reacting actively in the diamond
deposition process, the noble gases also are active in deposition by creating additional
excited atom-molecule or ion-molecule reactions. As a result, the dissociations o f hydrogen
and hydrocarbon molecules are enhanced, the film growth rate is increased, and the
parametric range of the diamond formation is extended. The results contribute to the global
data base of the diamond-forming plasma chemistry and fully demonstrate that a high
degree of molecular excitation and dissociation in plasmas is critical for obtaining high
growth rates of diamond films.
The internal structure of CVD diamond films is quite complex. They are
polycrystalline in nature with varying grain sizes and orientations. Typically, the {111} or
{100} faces dominate the surface morphology. Non-diamond carbon phases can be
introduced when the deposition conditions are not optimized. Interfaces between diamond
films and foreign substrates are present. Furthermore, CVD diamond contains extensive
structural defects. The unique bulk structure of graphite inclusions in the diamond matrix,
the interfaces of diamond films grown on silicon and silica substrates, and the various
crystalline defects are characterized and analyzed extensively in this thesis. The relations
between these structural features and experimental parameters and the relations between the
internal structural defects and the external surface morphologies are explored. The results
achieved represent a significant step toward a full understanding of the structural properties
of CVD diamond films.
The oxidation experiments with CVD diamond provide an effective means to study
the defects of diamond crystals and their distribution, as well as the possible silicon carbide
interfacial layers formed between diamond films and silicon substrates. Besides the results
on oxidation kinetics, the structural and morphological consequences o f oxidation are
particularly interesting. The results are important for high temperature applications o f
diamond films under ambient conditions.
F IG U R E S ......................................................................................................... x
T A B L E S .........................................................................................................xv
P R E F A C E ........................................................................................................................... xvi
A C K N O W L E D G M E N T S ..............................................................................................xviii
C hapter
BA CK G RO U N D ...........................................................................................1
The hierarchy o f carbon ..............................................................................1
C lassifications o f diam ond............................................................................ 3
P ro p erties o f diam ond....................................................................................4
D iam ond fo rm atio n .......................................................................................... 5
1.4.1. Thermodynamic and kinetic considerations........................................ 6
1.4.2. Form ation o f natural diam ond............................................................. 8
1.4.3. Solvent-catalyst synthesis o f diamond................................................9 B rief historic review ..............................................................9 S ynthesis p ro cess...................................................................10
1.4.4. Direct conversion o f graphite to diamond.........................................10 Shock w ave pro cess............................................................. 11 Static pressure synthesis......................................................11
1.4.5. Diam ond growth at low pressures.................................................... 12
1.5. T h esis
Theoretical considerations....................................................12
H is to ry ........................................................................................ 13
A pparatuses and techniques.................................................16
A chievem ents to date .........................................................20
Applications o f CVD diamond film s.................................. 27
C urrent p ro b lem s...................................................................28
p la n .........................................................................................................30
1.5.1. Statem ent o f the problem ...................................................................30
1.5.2. Thesis goals and o u tline.................................................................... 31
In tro d u c tio n ........................................................................................................33
General approaches for preparation o f diamond films at low pressures
2.2.1. PV D
2.2.2. CV D
p ro cess........................................................................................... 34
p ro c e ss...........................................................................................34
2.3. M PECVD o f diam ond film s.........................................................................37
2.3.1. Characteristics o f m icrowave plasm as.............................................. 37
2.3.2. M PECVD sy stem ...................................................................................38 Microwave generation and transmission.............................43 Reaction chamber and plasma generation........................... 44 Gas flow and control system ..............................................46
2.3.3. D eposition
p ro ced u res..........................................................................47
2.4. Characterization techniques for diamond film s...........................................50
2.4.1. F ilm growth rate m easurem ent..........................................................50
2.4.2. Transmission electron microscopy (TEM)
and the sam ple preparations.............................................................. 52
2.4.3. Scanning electron microscopy (SEM )...............................................57
2.4.4. R am an sp ectro sco p y............................................................................. 58
2.4.5. Optical emission spectroscopy (OES)............................................... 60
2.4.6. Therm ogravim etric analysis (TG A )................................................... 63
2.4.7. Other structural and compositional analysis techniques................... 65
Infrared spectroscopy (IR ).................... .'............................ 65
Laser excited luminescence spectroscopy........................... 66
Secondary ion mass spectrometry (SIMS)..........................67
Neutron activation analysis (NAA).....................................67
Chapter 3. DIAMOND FILM DEPOSITIO N ............................................................69
In tro d u c tio n ......................................................................................................... 69
L iteratu re review ........................................................................................... 69
E xperim ental ap p ro ach ....................................................................................73
Effect o f methane concentrations in hydrogen (CH4 %)...............................76
Effect o f substrate temperatures (Ts)........................................................... 79
3.5.1. Film growth rate as a function o f substrate temperatures (Ts)............. 79
3.5.2. Activation energies and growth rate-controlling mechanisms..............82
Effect o f gas pressures (P).......................................................................... 85
Effect o f gas flow rates (F )......................................................................... 88
Effect o f substrate positions relative to the plasma (D)............................... 92
D is c u s sio n .............................................................................................................94
S u m m ary ............................................................................................................... 98
C hapter 4. PLA SM A CH EM ISTRY .......................................................................... 99
4.1. In tro d u c tio n ......................................................................................................... 99
4.2. L iteratu re rev iew .....................................
Vl l
4 .2 .1 . P lasm a in g en eral................................................................................ 100
4.2 .2 . Spectroscopic techniques and identification o f plasma species...........101
4.2 .3 . Characterization o f diamond-forming plasm as................................. 105
4.3. E x p erim en tal ap p ro ach................................................................................... 109
4.4. R e s u lts .................................................................................................................. 110
4.4 .1 . Bulk plasm a optical em issions........................................................ 110
4.4.2. F ilm grow th ra te ................................................................................ 118
4.4.3. Film structure and com position.......................................................120
4 .5 . D is c u s sio n ............................................................................................................120
4 .5 .1 .
4.5 .2 .
4.5 .3 .
4.5 .4 .
4 .5 .5 .
G eneral functions o f noble gases...................................................... 120
Effect o f noble gases on the production o f atomic hydrogen (H)...... 125
Effect o f noble gases on the dissociation o f methane (CH 4 )............. 127
Effect o f oxygen on the film structure.............................................129
L im itations o f the study............................................
S u m m a ry .............................................................................................................. 131
Chapter 5. DIAMOND FILM STRUCTURE AND DEFECTS............................ 132
In tro d u c tio n ........................................................................................................132
Internal bu lk stru ctu re....................................................................................132
S tatem ent o f the problem .................................................................. 132
L iteratu re rev iew ...................................................................................133
E x perim ental ap p ro ach........................................................................136
R e s u lts ........................................................................................................137
E ffect
methane concentrations (CH 4 % )........................137
substrate temperatures (Ts) ...............................143
plasm a etching.................................................... 145
substrate m aterials..............................................149
film thickness......................................................149
argon ion bombardm ent.....................................153
5.2.5. D is c u s s io n .........................................................
153 M echanisms o f graphite form ation.................................... 153 Relationship between crystalline defects
and graphite inclusions....................................................... 157
5.2.6. S u m m a ry .................................................................................................... 158
5.3. In terfacial
stru ctu re......................................................................................... 159
5.3.1. L ite ra tu re rev ie w ...................................................................................160
5.3.2. E xperim ental ap p ro ach ........................................................................ 162
5.3.3. R e s u lts ........................................................................................................ 163
viii D iam ond/silicon in terface................................................... 163
5 .3.3.2. D iam ond/silica in terface...................................................... 169
5;3.4. D is c u s s io n .................................................................................................179
5.3.5. S u m m a ry .................................................................................................. 182
5.4. S tru ctu ral im p erfectio n s................................................................................183
5.4.1. L iteratu re
rev ie w .................................................................................. 183 Defects in natural and HP/HT synthesized diamonds
183 Defects in low pressure CVD diam ond............................. 186
5.4.2. E x perim ental ap p ro ach........................................................................ 188
5.4.3. P la n a r d e fe c ts........................................................................................ 188 TEM re s u lts ............................................................................188 Formation mechanisms o f planar defects...........................191
5.4.4.* D is lo c a tio n s..............................................................................................197
5.4.5. P o in t d e fe c ts..........................................................................................200 IR sp ectro sco p y .................................................................... 200 SIM S re s u lts .......................................................................... 202 Laser induced luminescence spectroscopy.........................202
5.4.6. Relationship between crystalline defects and surface
m o rp h o lo g ie s...........................................................................................206
5.4.7. D is c u s s io n ................................................................................................ 210
5.4.8. S u m m a ry ..............................................................
Chapter 6 . OXIDATION PROPERTIES OF DIAM OND........................................212
In tro d u c tio n .........................................................................................................212
L ite ra tu re rev ie w .............................................................................................212
E x p erim en tal ap p ro ach.................................................................................. 215
R e s u lts ...................................................................................................................216
O x id atio n k in e tic s................................................................................216
Structural and m orphological consequences.................................... 219
Characterization o f oxidation residues.............................................230
A nnealing in oxygen-free gases.......................................................235
6.5. D is c u s s io n ........................................................................................................... 235
6 . 6 . S u m m a ry ...............................................................................................................239
Chapter 7. GENERAL DISCUSSIONS AND CONCLUSIONS.............................241
7.1. E valuation o f the p resen t w ork.................................................................. 241
7 .2 . G eneral concluding rem ark s........................................................................ 243
MPECVD diam ond growth behavior..............................................243
Noble gas-involved methane-hydrogen plasma chemistry................243
Structure and defects o f CVD diamond.......................................... 244
Oxidation properties o f CVD diam ond........................................... 245
7 .3 . D irections o f fu tu re w ork.......................................
B IB L IO G R A P H Y .........................................
C arbon
phase d ia g ra m ............................................................................................ 2
The tubular microwave plasma enhanced chemical vapor deposition system.
Calibration curves o f the gas flow meter.......................................................... 48
The experimental set-up for plasma optical emission spectroscopy.................... 61
Schematic diagram o f the thermogravimetric analyzer...................
Film growth rate as a function o f methane concentrations.................................77
Raman spectra o f diamond films deposited at different
m ethane co n cen tratio n s...............................................................................78
An Arrhenius plot o f growth rate vs. reciprocal substrate temperature................ 80
Raman spectra of diamond films deposited at different
su b strate tem p eratu res.................................................................................81
Film growth rate as a function o f total gas pressures........................................86
3 .6
Raman spectra o f diamond films deposited at different
gas p re s s u re s ..................................................................................................87
Film growth rate as a function o f total gas flow rates.......................................89
Raman spectra o f diamond films deposited at different
gas flow ra te s ...............................................................................................90
sam ple p rep aratio n s.................................................................................... 54
3.9 Schematic diagrams of the substrate positions relative to the plasma................. 93
3.10 Film growth rate as a function of substrate positions relative to the plasma
3.11 Raman spectra of diamond films deposited at different
substrate positions relative to the plasma............................................. 96
4.1 A typical optical emission spectrum of a microwave plasma (CH 4+H2)............I l l
4.2 The emission intensities of C2 and H from the plasma (CH4 +R+H 2)...............114
4.3 The emission intensities of C2, Ha and H r radicals from the plasma
(C H 4 + A r+ H 2)........................................................................................116
4.4 The emission intensities of C2, Ha and Hr radicals and the ratio
of Ha/Hp from the plasma (4.4%CH4+16% R+79.6% H 2)
vs. the atomic weights of the added noble gases........................................ 117
4.5 Growth rates of diamond films deposited from plasmas (CH4+R+H 2).............119
4.6 Raman spectra of diamond films deposited from plasmas (CH4+R+H 2)...........121
5.1 Electron diffraction patterns, dark field images and Raman spectra
of diamond films deposited at different methane concentrations.............138
5.2 Electron diffraction patterns showing that the graphite {00.2} diffraction
arcs are aligned with the diamond {111 } diffraction spots
with a spread of orientation of 40°..............................................................141
A diamond <111> zone axis electron diffraction pattern showing that
no visible {10.0 }-type diffractions of graphite are present........................142
5.4 An electron diffraction pattern, bright and dark field images showing the
domain distribution of graphite crystallites in the diamond matrix.............144
Electron diffraction patterns, a dark field image and Raman spectra of
diamond films deposited at different substrate temperatures..................146
Electron diffraction patterns and Raman spectra of a diamond film
which was etched by air and hydrogen plasmas................................... 148
An electron diffraction pattern, a dark field image and a Raman spectrum
of a diamond film deposited on a Si 0 2 substrate................................... 150
5 .8
Electron diffraction patterns, dark field images and Raman spectra of
a diamond film deposited on a Co bonded WC material........................ 151
SEM micrographs and corresponding Raman patterns of diamond
films deposited on Si substrates........................
5.10 Cross-sectional TEM micrographs o f diamond film s....................................167
TEM bright field images and electron diffraction patterns of diamond
films deposited on Si substrates........................................................... 170
5.12 TEM micrographs at high magnifications of the interfaces of diamond
films deposited on Si substrates........................................................... 171
A distinct intermediate layer exists in the interface of a diamond
film deposited on a SiC>2 substrate......................................................... 172
SEM micrographs and corresponding Raman patterns o f diamond
films deposited on Si 0 2 substrates......................................................... 173
TEM bright field images and electron diffraction patterns of diamond
films deposited on Si0 2 substrates........................................................176
Schematic diagram of the diffraction pattern shown in Figure 5.15d.............. 177
TEM micrographs o f stacking faults and twins............................................ 189
TEM micrographs of planar defects in diamond films deposited at
different methane concentrations........................................................... 190
An electron diffraction pattern and TEM micrographs o f a microtwin.............192
5.20 The diamond structure and its {111} stacking sequences.............................. 193
5.21 A projection o f a twinned diamond structure on the (110) plane.....................196
5.22 TEM micrographs o f dislocations in a diamond film .....................................198
5.23 Atomic configurations o f dislocations in the diamond structure......................199
5.24 Infrared transmission and reflection spectra of polished diamond Elms...........201
5.25 SIMS spectra o f diamond film s............................................................................... 203
5.26 Lum inescence spectra o f diam ond film s....................................................... 205
SEM and TEM micrographs of surface morphologies and corresponding
internal structures o f diam ond film s......................................................208
6 .1
W eight loss vs. temperature of diamond powders, graphite powders
and diam ond film s.................................................................................... 217
The oxidation rate o f the diamond film as a function o f temperatures..............218
The oxidation rate o f the diamond film as a function o f time......................... 220
Optical images of a diamond film before being oxidized and after
being p artially o x idized ..........................
Raman spectra of diamond films oxidized at different temperatures............... 222
6 .6
SEM micrographs o f surface morphologies of the diamond film
oxidized at different tem peratures......................................................... 225
Oxidized diamond crystals showing etch pits concentrated on the
{111} fa c e s ...................................................................................................227
6 .8
Oxidation occurs on the {100} diamond surfaces........................................ 228
6.9 - Oxidation products attached on the diamond surface..................................... 229
M aterial transport during oxidation............................................................... 231
6 .11
A transmission optical image o f the oxidation residues........................................232
6 .12 SEM micrographs and Raman spectra o f the diamond film
and its oxidation residues....................................................................... 233
6.13 An electron diffraction pattern and a TEM bright field image o f the
o x id atio n re s id u e s.......................................................................................234
Schematic diagrams o f the selective oxidation around dislocation lines...........238
Activation energies (E) for various deposition processes..................................83
Plasm a species and their characteristics.......................................................... 104
Ionization potentials, excited states and corresponding energy levels
o f noble gases and their effects on the major species
in the diam ond-form ing plasm as........................................................... 112
A r contents in diamond film s measured by NAA........................................... 123
Interplanar spacings and lattice constants of diamond, graphite
and P -S iC ....................................................................................................... 178
Some physical properties o f diamond, graphite, Si, SiQz and p-SiC................ 181
This thesis is the third thesis on diamond thin films from the Thin Film Group at the
Materials Research Laboratory, The Pennsylvania State University. The work has been
carried out during the period of 1986-1990, under the guidance o f my advisors, Professors
M essier and Badzian. The two previous M.S. theses by T. M. Hartnett and T. A.
McKenna have dealt mainly with the characterization of the diamond film deposition
processes and the behavior o f the microwave plasmas. This thesis continues along that line
but is concerned in more detail with the internal structures of diamond films which is a key
link between the preparation and the properties. The thesis is based in part on five papers
listed below, some of which have been available in the open literature.
1. W. Zhu, A. R. Badzian and R. Messier, "Structural imperfections in CVD diamond
films," J. Mater. Res., 4 ,6 5 9 (1989)
2. W. Zhu, C. A. Randall, A. R. Badzian and R. Messier, "Graphite formation in
diamond film deposition," J. Vac. Sci. Technol., A2> 2315 (1989)
3. W. Zhu, R. Messier and A. R. Badzian, "Effects of process parameters on CVD
diamond films," in Proceedings of the First International Symposium on Diamond and.
Diamond-like Films, edited by J. P. Dismukes, A. J. Purdes, K. E. Spear, B. S.
Meyerson, K. V. Ravi, T. D. Moustakas and M. Yoder, The Electrochemical Society,
Vol. 89-12,61 (1989)
4. W. Zhu, A. Inspektor, A. R. Badzian, T. McKenna and R. Messier, "Effects of noble
gases on diamond deposition from methane-hydrogen plasma," J. Appl. Phys., (1990)
(in press)
5. W. Zhu, X. H. Wang, D. J. Pickrell, A. R. Badzian and R. Messier, "Oxidation
properties o f CVD diamond films," The Second International Conference on New
Diamond Science and Technology, (1990) (submitted)
In addition, papers written during this period due to collabradons with other people,
and which are strongly connected to those works outlined above but not included in this
thesis, are:
1. W. Zhu, A. R. Badzian and R. Messier, "Morphological phenomena of CVD
diamond," Proceedings SPEE, (1990) (submitted)
2. D. J. Pickrell, W. Zhu, A. R. Badzian, R. Messier and R. Newnham, "Downstream
plasma enhanced deposition of diamond films," Appl. Phys. Lett., (1990) (in press)
3. K. Srikanth, S. Ashok, W. Zhu, A. R. Badzian and R. Messier, "Deep level transient
spectroscopy study of thin film diamond," Proceedings of the Materials Research
Society Fall Meeting Symposium F, (1989) (in press)
4. D. J. Pickrell, W. Zhu, A. R. Badzian, R. Messier and R. Newnham, "Near-interfaces
of diamond films on SiC>2 and Si substrates," J. Mater. Res., (1990) (in press)
5. X. H. Wang, L. Pilione, W. Zhu, W. Yarbrough, W. Drawl and R. Messier, "Infrared
optical properties of CVD diamond films," J. Mater. Res., (1990) (in p ress). . „
I would like to take this opportunity to express my deepest gratitude to my co­
advisors, Dr. Russell Messier and Dr. Andrzej Badzian, for their guidance, support and
encouragement throughout this study.
To the members of my thesis committee, Dr. Karl Spear, Dr. Bruce Knox and Dr.
Paul Howell, I extend my appreciation for their time and invaluable input concerning this
I am grateful for the financial support provided by the K. C. Wong Education
Foundation Ltd. in Hong Kong that made my study at Penn State possible. This research is
possible because of the partial support from the Office of Naval Research (with funding
from the Strategic Defense Initiative Organization's Office of Innovative Science and
Technology) under Contract No. N00014-86-K-0443 and the Diamond and Related
Materials Consortium at The Pennsylvania State University.
There are many people who helped me during the course of this thesis research.
Thanks go to: Mr. David Pickrell and Dr. Clive Randall for their stimulating collabrations;
Mr. William Drawl, Mr. Chris Engdahl, Mr. Scott Woodrow and Mr. Tom Hartnett for
their skillful technical assistance; Mrs. Diane Knight and Mr. Eric Plesko fo r their
instructions on how to use the Raman spectrometer, Dr. Aharon Inspektor and Mr. Tom
McKenna for their help in conducting experiments on plasma diagnostics; Dr. Else Breval,
Dr. Peter Arapad and Mrs. Kristyna Okoniewski for their help in performing transmission
electron microscopy; Mr. Gerry Zimmermann for his assistance in operating the
thermogravimetric analyzer; and Dr. Joe Yehoda for his help in obtaining data from neutron
activation analysis measurements.
I thank my colleagues in the Thin Film Group for their kindness and providing a
scientific environment that fosters creativity.
I am indebted to all my friends who have made my life beside the work interesting
and pleasant.
My parents and brothers deserve my profound gratitude because o f their neverending moral support, encouragement and patience over the past years.
O f course, beyond question, my special thanks and deepest appreciation go to Xiao
Hong. W ithout her love, understanding, support and help, my goals would never have
been accomplished. Everything belongs to you, Xiao Hong; I love you.
Chapter 1
1.1. The hierarchy of carbon
Carbon occurs widely in its elemental form as crystalline and disordered solids.
Diamond and graphite are the two main crystalline allotropes o f carbon. As shown in the
phase diagram of carbon in Figure 1.1, diamond, with a covalently sp 3 tetrahedral bonded
cubic structure, is the high pressure form of carbon, and graphite, composed of hexagonal
layers o f carbon atoms with strong sp2 trigonal bonds in the basal plane and weak van der
Waals bonds between the planes, is the high temperature form o f carbon. Diamond is
metastable at room temperature and atmospheric pressure with respect to graphite. Graphite
has good electrical conductivity, lubricity, lower density, a grayish-black appearance and
softness, which are in contrast to those of diamond. It should be noted that both diamond
and graphite have their modified forms: hexagonal diamond and rhombohedral graphite.
Both of these modified forms are believed to be less stable than their counterparts, and their
thermodynamic properties are not fully understood; therefore, they are not associated with
any stability field in the carbon phase diagram. In the phase diagram, the liquid carbon has
two states (Rapaport, 1967), the separation parameter of which is the density, one having a
higher and the other a low er density than graphite. This is the reason why there is a
maximum temperature in the graphite melting curve. There is also predicted a metallic liquid
carbon phase at extremely high pressures (at least above 600 kbar), although no definite
experimental evidence has been found (Alder and Christian, 1961; Libby, 1963).
Another allotrope of carbon is called carbyne which has a sp 1 hybridized crosslinked chain-like structure. There are at least six forms of carbyne (Heimann et al., 1984).
They appear to be not stable at any pressure and temperature, and little is known about their
Metallic carbon
Liquid carbon
400 -
-6 0
-4 0
Bundy s.
(static .
Shockwave synthesis
—r^ryn*Graphite v
p ro c e s s ^ -^ ^ tJ p re s s u re w
v/ / / / * a
Figure 1.1. Carbon phase diagram (Sources: De Carli, 1966; Muncke, 1979;
Bundy, 1980)
Pressure ,GPa
-8 0
800 r
properties. The role o f carbyne as an allotrope o f carbon has been reviewed by Wedlake
(1974). •
There are numerous disordered forms o f carbon. They are not allotropes o f carbon,
rather only transitional forms of carbon. As graphite is the most stable crystalline allotrope
o f carbon, the structures of many disordered forms are based on the graphitic structure.
Various forms o f disordered carbon include carbon fibers which are important for their
strength in composites, chars and cokes which are intermediate species in pyrolysis of
carbonaceous materials into graphite, glassy carbon which is obtained by heating organic
polymers, m icrocrystalline carbon (|i-C) which is produced by irradiating graphite,
amorphous carbon (a-C) which is deposited by electron beam evaporation or by sputtering,
and hydrogenated amorphous carbon (a-C:H) which is grown by plasma deposition or ion
beam deposition from hydrocarbon gases (Robertson, 1986). The a-C:H is also called
diamond-like carbon (DLC). There have been comprehensive reviews available in the field
o f a-C, H-C and DLC (Robertson, 1986; Angus et al., 1986; Tsai and Bogy, 1987).
Recently, a new form of carbon was found in the process o f laser vaporization of
graphite (Baum, 1985). It is a stable, spheroidal, aromatic, Ceo cluster in the form of a
truncated icosahedron which has a sp 2 bonded structure with a central cavity. It is found to
be remarkably inert Theoretical background and properties o f this new carbon form as well
as other carbon clusters are being further explored (Haymet, 1985).
1.2. Classifications o f diamond
Diamond crystals are classified into two major groups based on their impurities and
related optical and magnetic properties (Davies, 1977). Type I diamond contains a
substantial amount o f nitrogen, and type II diamond essentially does not contain nitrogen.
In addition, there is a type III diamond which is o f a hexagonal structure instead of the
normally cubic structure.
Type la diamond contains nitrogen in substantial amounts (of the order of 0.1 at.%)
which appears as non-paramagnetic aggregates or platelets. Most natural diamonds are of
this type.
Type lb diamond contains nitrogen up to 0.2 at.% but in dispersed paramagnetic
subsdtutional form. Almost all synthetic diamonds are of this type.
Type Ha diamond is effectively free of nitrogen. It is very rare in nature and has
enhanced optical and thermal properties.
Type lib diamond is a very pure type of diamond, generally blue in color, which
has semiconducting properties. It is extremely rare in nature. Boron doping can make
synthetic semiconducting diamond.
Type III diamond, which also is called Lonsdaleite after Kathleen Lonsdale
(Hannemann et al., 1967; Frondel and Marvin, 1967), has a hexagonal structure instead of
the normally cubic structure of type I and type II diamonds. It has been found in
meterorites and has been made synthetically (Bundy and Kasper, 1967; Frondel and
Marvin, 1967). However, many of its physical properties are not known, including its
stability relative to cubic diamond.
If not specifically indicated, all diamond dealt with in this thesis refers to cubic
1.3. Properties of diamond
Diamond belongs to the cubic Fd 3 m-07 h space group. Its structure consists of two
face centered cubic lattices displaced from each other in one quarter of the cube diagonal.
The number of atoms in a unit cell is 8 . The lattice constant o f the unit cell has been
measured between 0.356683 ± 0.000001 nm and 0.356725 ± 0.000003 nm at 298K, and
the nearest neighbor distance is 0.154450 ± 0.000005 nm at 298K (Lonsdale, 1947;
Skinner, 1957; Kaiser and Bond, 1959). The carbon atoms in diamond are covalently
bonded with sp3 hybridization.
Diamond has many unique properties. Diamond is the most valued o f all gems
because of its brilliance and fire due to the high refractive index and high opdcal dispersion
and also because o f its durability and controlled scarcity. Diamond is the hardest and
strongest known material. It has the highest thermal conductivity at room temperature and
is optically transparent from 225 nm in UV to IR for type II diamond. It has a very low
coefficient of friction and is extremely inert chemically. It is an excellent electrical insulator.
When doped, it is a wide band gap semiconductor with high breakdown voltage and
saturation velocity. This unusual combination o f extreme properties makes diamond a
valuable and strategic industrial material which is very important both scientifically and
technologically. A detailed compilation o f the physical and chemical properties o f diamond
has been conducted by Field (1979).
1.4. Diamond formation
This section will briefly discuss the theoretical aspects and natural background o f
the formation o f diamond. The history o f man's efforts in making diamond will be
reviewed with primary interest in that o f the metastable synthesis. A summary o f prominent
results achieved in low pressure chemical vapor deposition (CVD) o f diamond films,
potential applications, and existing m ajor problems is presented in the later half o f this
1.4.1. Thermodynamic and kinetic considerations
Thermodynamics, which considers the relative energies of the reactants and the
products, states that when the free energy of the reactants is greater than the free energy of
the products, the relative energy, AG, is negative, and the reaction has thermodynamic
permission to proceed. For the reaction
C (graphite) —> C (diamond)
because Gd is always greater than Gg at zero or atmospheric pressure, AG° = Gd° -Gg° is
positive at all temperatures. F or example, A G °= +692 cal/(g.atom) at 25°C and latm
(Barin and Knacke, 1973), which is about 0.016 eV compared to the thermal energy, kT =
0.025 eV, at room temperature (25°C). This means that the reaction will not proceed to
transform graphite to diamond at these conditions. The application of pressure changes AG
according to the relation
0AG/9P) t = AV = Vd - Vg
AGP - AG° = /0pAVdP
------ (1.3)
Since diamond is much denser than graphite (Vd = 3.41 cm 3/mol, Vg = 5.34 cm 3/mol), AV
is negative, and at some pressures which depend on the temperature, AGP is reduced to
zero, reaching the equilibrium condition. The equilibrium pressure
(P e q )
can be calculated
AG° + /0Pe4 AVdP = 0
------ (1.4)
The free energy difference at any pressure and temperature will be
AGTp = A H t° - TA St 0 + J0pA V idP
------- (1.5)
The heats o f combustion, the specific heats, molar volumes and bulk moduli over
the range o f T and P need to be known or estimated before getting accurate and reliable
equilibrium pressures from equation (1.4). Data have been available for evaluating A H i°
and A S t° to 1200 K. It is now generally agreed that the equilibrium line between the
graphite and diamond phases increases linearly with temperature between T = 500K at 20
Kbar and 2250K at 70 Kbar. Diamond is stable above the line. However, experimental
confirmations are difficult due to a kinetic barrier for the phase transformation between
diamond and graphite that obscures the thermodynamic equilibrium line (Bundy et al.,
1961). Uncertainties still remain for the graphite-diamond equilibrium conditions,
especially for temperatures above 2000K due to the shortage of reliable data. Nevertheless,
the diamond-graphite equilibrium line has proved to be a useful guide for practical diamond
Thermodynamics does not give any information concerning the time required for a
reaction to take place, however. There are many instances in which the AG has a large
negative value, yet the reaction proceeds at an imperceptible rate. Again for the reaction
(1.1), since Gd is greater than Gg at atmospheric pressure and room temperature, diamond
is thermodynamically unstable with respect to graphite, and the reaction should be expected
to proceed in the opposite direction, that is, converting diamond to graphite. In fact,
diamond has not been observed to transform into graphite by any measurable amount over
periods o f hundreds o f years under ordinary conditions. On the other hand, it has been
found that subjecting graphite at some temperatures to a pressure marginally above the
equilibrium pressure by no means ensures that it will transform to diamond. These facts
suggest that both diamond and graphite are in deep energy potential wells, and there is a
very large activation energy barrier separating the two phases to prevent mutual
transformations from occurring. Practically, in order to overcome the activation threshold
and promote the transformation, elevated temperatures are necessary for diamond-tographite conversion to proceed at an observable rate (e.g., 1200°C at 1 atm), and for
diamond synthesis, the experimental conditions need to be noticeably above the graphite-
diamond equilibrium line, and various solvent-catalyst metals or alloys are used. A
possibility arising from these facts, however, is that a metastable phase (diamond) can form
if the kinetic barrier to a more stable phase (graphite) is sufficiently high. Once diamond is
formed, it remains as a metastable phase with a negligible rate of transformation to graphite
unless heated to a high temperature. It is these kinetic factors that make diamond synthesis
possible in a pressure-temperature field where diamond is thermodynamically unstable with
respect to graphite. They are, o f course, only necessary but not sufficient conditions for
metastable synthesis of diamond. Competing processes for the deposition o f the undesired
stable graphite phase have to be suppressed.
1.4.2. Formation of natural diamond
Historically, India and Borneo were the earliest sources of natural diamond. Later,
diamond was found in many places like Brazil, South Africa and Soviet Union. At present,
the Argyle Mines in Australia produce the majority of the world's diamond.
The genesis o f natural diamond, which ranges widely in purity, size, color and
shape, has long been an interesting but not well established topic in diamond geology. It is
now generally presumed that, although some diamond were alluvial, the original source of
diamond was volcanic in nature. Diamond was formed in plutonic reservoirs before
eruption. The volcanic pipes were first filled with kimberlite magma, and while still in a
plastic state, the deeper seated, more fluid magma carrying diamond and other mineral
crystals continued to rise up through the pipes to the earth's surface over geological ages.
The resulting rock became the present kimberlite which is the most important primary
source of terrestrial diamond. However, the manner in which diamond was formed is not
known. Some proposed that diamond crystallized from natural melts having a wide range
in chemical composition (peridotites, edogite xenoliths and kimberlites) at depths not less
than 150 km below the earth surface, even down to 1500 km. The diamond formation was
accomplished by means of a solvent which could dissolve carbon atoms from graphite,
and, in turn, the dissolved carbon atoms could migrate through the solvent and precipitated
in cooler regions as diamond. Kimberlite is such a solvent of carbon. However, mere
solvency is not enough. Other forces or mechanisms must be required that are probably
catalytic in nature.
Diamond was formed also in meteorites (Carter and Kennedy, 1964; Lewis et al.,
1987). Whether it was formed by a shock compression process or through a static pressure
process is still in debate (Kennedy, 1966; De Carli, 1966).
1.4.3. Solvent-catalvst synthesis of diamond Brief historic review
In 1797, Tennant discovered that diamond is simply carbon (Smithson-Tennant,
1797). This stimulated the interest in and motivated many attempts at synthesis of diamond.
However, early attempts to make diamond were not scientific; instead, they were full of
fascination, intrigue and occasional fraud. Most claims of successful diamond synthesis
failed in duplication and were later proved to be not true or unlikely (Hannay, 1880;
Moisson, 1894; Parsons, 1907,1920). By the mid-1800s, it was gradually realized that the
transformation o f graphite to diamond requires high pressures. It was not until the mid1900s that a reasonable idea of the pressures required was established (Wedlake, 1979). At
that time, chemical thermodynamics of the carbon system had been understood, and the
diamond stabilization field was established. O f the same importance as many successful
achievements in the history of making diamond was the failure of Bridgman's experiments
(Bridgman, 1947). He compressed graphite to pressures well within the diamond stable
region for a few seconds without producing diamond. This work led to our current
understanding of the large kinetic barrier between the diamond and graphite phases. The
real breakthrough came in 1955 at General Electric when Bundy and co-workers succeeded
in the reproducible synthesis of diamond (Bundy et al., 1955). They developed necessary
high pressure equipments and discovered and elucidated the solvent-catalytic processes by
which ordinary forms of carbon could be changed into diamond. The process is now
termed as the molten transition metal solvent-catalyst diamond synthesis. Synthesis process
In this process, a mixture of carbon (graphite) and catalyst metal is heated high
enough for the metal to be melted while the system is at a pressure high enough for
diamond to be stable (e.g., 2000K, 50-100 kbar). Diamond is believed to form by a three
step mechanism: 1) carbon (if not graphite) is catalytically graphitized; 2 ) the formed
graphite is dissolved by metal to form a metal-carbon solution; and 3) diamond nucleation
and growth occur in the solution. The most effective catalysts are transition metals of iron,
cobalt and nickel. By this method, a large scale production of synthetic diamond (several
tons per year) has become a commercial reality for the past thirty-five years.
A number o f apparatuses have been built for producing the high pressures and
temperatures necessary for diamond synthesis. Comprehensive descriptions of the
techniques and phenomena involved in the solvent-catalyst diamond synthesis can be found
in the literature (Bundy et al., 1973; Wentorf, Jr., 1974).
1.4.4. Direct conversion of graphite to diamond
Diamond also can be produced directly from graphite without a solvent-catalyst
Again a high temperature for sufficient atomic mobility for the transformation and a high
pressure for diamond stability are required. This combination o f high temperature and
pressure can be achieved either dynamically or statically. Shock wave process
When graphite is strongly compressed and heated by a shock wave produced by an
explosive charge (300 kbar and 1500K is estimated to be produced and maintained for a
few microseconds), up to 10% diamond may form (De Carli and Jamieson, 1961). The
main problem is to recover the small sized diamond (a few tenths o f a micrometer). This
quick method o f producing diamond powders has been commercialized by the Du Pont
shock-made diamond patent (Du Pont, 1965). Static pressure synthesis
Diamond (about 0.01 to 0.1 mm size) can form directly, either from well
crystallized highly oriented graphite at pressures o f about 130 kbar and higher and at
temperatures o f about 3000K (Bundy, 1963) or from carbonaceous materials at such
pressures by pyrolysis (Wentorf, Jr., 1965) followed by a very rapid cooling o f the melt.
N o catalyst is needed. The transformation can be carried out in a static high pressure
apparatus with flash heating (Bundy, 1963; Bundy and Kasper, 1967). A new form of
diamond with a hexagonal structure was discovered in this experimental process o f direct
conversion o f graphite to diamond.
1.4.5. Diamond growth at low pressures Theoretical considerations
Since diamond is unstable from the thermodynamic point o f view at atmospheric
pressure and room temperature, synthetic as well as natural diamonds are typically formed
under extreme conditions o f high pressures and temperatures. However, this does not
mean that diamond crystallization cannot occur under normal conditions. Interest in
metastable growth of diamond at low pressures where graphite is thermodynamically stable
has come a long way from Von Bolton's early attempt made in 1911 (Von Bolton, 1911) to
our present widespread use of activated CVD processes to grow diamond.
From the thermodynamic point o f view, in a system in which stable and metastable
phases o f various energy levels exist, the least stable phase will form first, with the others
occurring step by step, until the minimum energy level is reached. The free energy o f
carbon atoms in some compounds, such as methane (CH4 ) and monoxide (CO), may be
higher than that o f carbon in the diamond lattice (Bundy et al., 1973). During the
decomposition o f such compounds by suitable means, the carbon atoms could stay at the
energy level of diamond in the course o f falling from a state o f higher free energy all the
way down to a state of lower free energy of graphite. Such general phenomena occur
widely in many other chemical or physical systems, and there are no fundamental rules to
prohibit such a behavior for carbon as well. Therefore, diamond can be crystallized from
carbon atoms or carbon containing molecular species with high free energy under suitable
experimental conditions. The metastability o f diamond alone is not a deterrant factor for
obtaining diamond at reduced pressures. The possibility of diamond crystallization in a P-T
regime where diamond is thermodynamically unstable with respect to graphite is attributed
to a combination o f thermodynamic and kinetic factors, namely, graphite being more stable
than diamond only by a small free energy difference (0.692 kcal/mol) at normal pressures,
a significant activation barrier impeding spontaneous transformation o f diamond to graphite
(in fact, the rate o f spontaneous transformation does not become significant until 1300°C),
a high mobility o f carbon atoms on clean diamond surfaces at about 1000°C, and the
critical role o f atomic hydrogen in stabilizing diamond with respect to graphite.
1.4,5,?, History.
Excellent reviews have been available about the history o f the development o f
diamond growth at low pressures (Technical Insights Inc., 1987; International Resource
Development Inc., 1987; Angus, 1989), and therefore, only a simplified version will be
given here to illustrate certain historic events which are worth recording and understanding.
Thirty-five years ago, Bridgman, the father of modem high pressure technology, who
him self failed in attempting diamond growth at high pressures, predicted in a Scientific
American article that diamond growth at low pressures should be equally achievable as at
high pressures (Bridgman, 195,5). However, achieving the appropriate conditions for low
pressure diamond growth has taken decades o f research. It was initiated in 1949, and
Eversole o f Union Carbide was the first to succeed in 1953, growing diamond from carbon
monoxide and hydrocarbon gases on diamond powder at low pressures by pyrolysis,
although it w asn't publically reported until about a decade later (Eversole, 1962).
However, the growth rate was very low, the carbon deposit contained only about 1%
diamond, and the remainder was graphite. The process needed many cycles o f growth
followed by atomic hydrogen etching at high pressures and temperatures in order to remove
the graphite portion of the deposit.
Following the invention of Eversole, Deijaguin's group at the Academy o f Sciences
in Moscow and Angus and his co-workers at the Case Western Reserve University were
active in the field o f low pressure synthesis o f diamond. Deijaguin et al. (1968) used
various methods such as chemical vapor transport reactions to grow diam ond from
hydrocarbons and hydrocarbon-hydrogen mixtures. They also conducted theoretical
investigations o f kinetics o f pyrolysis o f various hydrocarbon-hydrogen mixtures and
nucleation and growth rates o f diamond on diamond seed crystals. A paper by Deijaguin
and Fedoseev in 1975 summarized their two decades o f research efforts in the vapor
deposition o f diamond. During their work, the Soviet researchers gradually realized the
importance o f atomic hydrogen for achieving metastable diamond growth. Angus et al.
(1968), Poferl et al. (1973) and Chauhan et al. (1976) used chemical vapor deposition
method to study the growth rates o f diamond and graphite from methane and hydrogen
mixtures or ethylene. They also recognized the crucial role o f atomic hydrogen by reporting
that atomic hydrogen could preferentially etch graphite compared to diamond. Low energy
electron diffraction (LEED) studies conducted by Lander and Morrison (1963,1964) might
be the most significant work at that time in elucidating the functions of hydrogen atoms on
the diamond surface and predicting the feasibility of diamond growth. The establishment of
the critical role of atomic hydrogen was the most important achievement, by today's
judgement, in all the early work o f low pressure diamond deposition.
By the mid- 1970s, although the low pressure diamond growth was apparently
achieved, it had not generally been accepted by the scientific community. Only thermal
decomposition methods were used in those early reports without additional activation
processes. The key problems were the competitive nucleation and growth o f graphite and
the low growth rate (<0.1pm/h). Nevertheless, these efforts became the intellectual
groundwork of later breakthroughs in the low pressure synthesis of diamond.
In a parallel development, efforts were made also in another direction to achieve the
metastable diamond crystallization in the 1970s, i.e., through various physical vapor
deposition (PVD) techniques to grow diamond films. In 1971, Aisenberg and Chabot
deposited carbon using a beam o f carbon ions in an argon plasma. Although the films
showed diamond properties, no definite diffraction patterns were given. They named these
film s "diamond-like carbon" (DLC). Spencer et al. (1976) also reported growing very
small sized diamond films (actually DLC) using a beam o f carbon ions with energies
between 50-100eV. Following these works, various PVD methods were introduced to
grow DLC films (Weissmantel et al., 1979; Holland and Ojha, 1979; Anderson, 1981),
including sputter deposition with or without bombardment by an intense flux o f ions with
energies of the order of IkeV, RF plasma deposition in hydrocarbon gases on to negatively
biased (lOOeV) substrates, and ion beam plating of benzene or other hydrocarbons at
acceleration voltages of 100-1000V. A generic term, i-C, has been proposed for these DLC
films to stress the essential role of energetic ions in all the preparation (Weissmantel et al.,
1980). These films were smooth, transparent, insulating, exceptionally hard and chemically
inert. They have found applications as anti-reflection coatings for IR optical elements and
solar cells and as protective coatings on nuclear reactor walls and magnetic and optical
Present interests in the low pressure vapor deposition of diamond is largely due to
the findings by Soviet scientists in the late 1970s and early 1980s which indicated that gas
activation techniques can greatly increase the growth rate o f diamond while suppressing the
graphite deposition and the large scale efforts in Japan since the early 1980s which resulted
in a variety o f new techniques for gas activation and diamond deposition. In 1977,
D eijaguin and Fedoseev outlined three approaches to produce higher concentrations of
atomic hydrogen than that resulting from the thermal dissociation of hydrocarbon-hydrogen
gases: catalytic, electric discharge and heated tungsten filament approaches. These
significant results were not published in English until 1981 (Spitsyn et al.) and more
detailed in 1984 (Vamin et al.). Since then, various techniques for low pressure growth of
diamond,- such as AC discharge, RF or microwave plasma enhanced CVD, DC glow
discharge assisted CVD and hot filament assisted CVD, at a rate of 1-100 pm/h on various
foreign substrates, have been described (Mania et al., 1981; Matsumoto et al., 1982a,
1982b; Kamo et al., 1983; Matsumoto and Matsui, 1983; Matsumoto, 1985; Sawabe and
Inuzuka, 1985,1986). All the techniques are based on the generation of atomic hydrogen
near the growing surface. Activities in the U.S. were started in earnest in 1984 after
Professor Rustum Roy of The Pennsylvania State University visited Japan, and saw the
results of the diamond CVD work at NIRIM (National Institute for Research in Inorganic
Materials, Tsukuba, Japan) first hand. A large cooperative program among universities,
industries and government has been established since then. On September 14, 1986, the
front page of the Sunday New York Times announced a "New Era of Technology Seen in
Diamond Coating Process". Indeed, it was the optimism of this headline that prompted me
to pursue my career in this area. Apparatuses and techniques
Within the last decade a number of thin film CVD apparatuses and techniques have
been developed for preparation of diamond at practical rates (Badzian et al., 1987), mainly
owing to the Japanese efforts. The techniques currently used can be grouped into five
major categories: 1) thermally activated CVD; 2) high frequency plasma enhanced CVD; 3)
DC discharges assisted CVD; 4) combustion flame; and 5) hybrids of these and others.
In the thermally activated CVD or hot filament CVD (HFCVD), a tungsten (or
tantalum, molybdenum, rhenium) filament heated at ~2000°C is used to dissociate
hydrocarbon-hydrogen mixtures at typical conditions of 0 . 1-2 % methane concentrations in
hydrogen at pressures of 50-100 Torr. A substrate is usually mounted close to the filament
(~1 cm) and heated to temperatures between 800-1000°C. Diamond films can be deposited
at a rate o f l-10pm /h. The substrate can be biased positively to induce electron
bombardment on the growing surface, thereby enhancing the dissociation o f gases at or
near the substrate surface and increasing the nucleation density and growth rate (Sawabe
and Inuzuka, 1985,1986). The HFCVD method is relatively simple, inexpensive and easy
to scale up to cover a large substrate area. But the brittleness and deformation of the
filament due to carburization at high temperatures is a major problem.
In the high frequency plasma enhanced CVD, the use o f 2.45 GHz microwave
plasm a for diamond deposition seems to be the most popular technique for practical
applications. Plasmas of hydrocarbon-hydrogen mixtures are sustained in a microwave
cavity where diamond is deposited on electrically floating substrates under conditions
similar to HFCVD, that is, a pressure o f 10-100 Torr, a substrate temperature o f 8001000°C, and methane concentrations of 0.2-5% in hydrogen. The typical deposition rate is
around 1-5 |im /h. The substrates are immersed usually in the plasmas which are heated by
both plasmas and microwave energy. However, "remote" diamond deposition also has
been reported in which the substrates are placed downstream outside o f the luminous
plasmas and an external heater is used to achieve the appropriate temperature (Pickrell et
al., 1990a). Diamond is grown also by microwave plasmas in a magnetic field to achieve
high plasma densities and electron energies and to deposit large area substrates at low
temperatures (Kawarada et al., 1987). Microwave plasma enhanced CVD (MPECVD)
appears to be the most promising technique for growing uniform, pure and high quality
diamond films.
The low pressure RF plasmas at 13.5MHz were first used by Matsumoto (1985) to
grow diamond films. Later, Matsumoto et al. (1987) developed a high power technique to
generate RF plasm as at atmospheric pressure. High growth rates (60-180 pm /h) were
achieved on Mo substrates by this m ethod In addition, a remote RF plasma enhanced CVD
system has been reported (Rudder et al., 1986). The RF plasmas require a large power
consumption, but the plasma density is usually lower than that o f a microwave plasma. In
general, the growth rate, uniformity and structure o f diamond films are inferior to those
deposited by HFCVD or MPECVD.
There are several modifications to using DC glow dischaiges to deposit diamond.
By applying a high DC voltage and a high current density (1 kV and 4 A/cm2) between
parallel electrodes in a relatively high gas pressure (200 Torr), Suzuki et al. (1987)
obtained a high nucleation density ( 108/cm2) and a high growth rate (20-100 jim/h) at the
anode. The stabilized DC discharge was believed to be in a transition state between glow
and arc, and the gas temperature was very high. However, the film quality deteriorated
when the film grew above 100 (im. Sawabe and Inuzuka (1986) applied a DC voltage and
weak current (150 V and 10 mA/cm2) between a filament and a substrate in HFCVD
system, naming it electron assisted CVD (EACVD). The major effect was to increase the
nucleation density. Singh et al. (1988) used a tubular hollow cathode instead o f a hot
filament to produce a stable discharge with a low DC voltage (95 V, 8 A). Although a
simple DC arc between two rod electrodes could be used for diamond synthesis (Akatsuka
et al., 1988), a DC plasma jet has advantages in ease of handling, reactant gas injection and
plasma stability, and in addition, this jet method can produce diamond at rates as high as 80
(im/h (Kurihara et al., 1988). Recently, a much higher growth rate, about 930 |im /h, has
been reported using an atmospheric DC arc discharge plasma je t (Ohtake et al., 1989). In
general, the most important advantage o f DC thermal plasmas is the high deposition rate.
Diamond films can be deposited on substrate surfaces without mechanical pretreatments.
Problems are non-uniformity, small area and poor temperature control due to the extremely
high gas temperatures.
The combustion flame method for diamond growth was discovered first by Hirose
and Kondo (1988) in Japan and was confirmed later in the U.S. by Hanssen et al. (1989)
and Yarbrough et al. (1989). This method can be used to produce diamond at atmospheric
pressure using an oxygen-acetylene torch. Because of its high gas pressure and high gas
temperature, the flame can be regarded as a thermal plasma, though the degree o f ionization
m ay be low. The ratio o f C 2H 2/O 2 , which is very critical for the formation o f the
oxyacetylene flame and for the diamond deposition, is usually close to unity. The gas
temperature may reach 3000°C, and the substrate temperature is kept around 800-1000°C.
High growth rates (30 (im/h) can be obtained with this simple method. However, substrate
temperature, film uniformity and process stability are difficult to control.
There are hybrid methods being developed. Some examples are hot filament
immersed in a microwave plasma (Anthony, 1987; W ild et al., 1989); hot filament CVD
assisted by RF plasma (Komatsu et al., 1987); and hot filam ent and DC plasma co­
enhancement (Fujimori et al., 1989). The objective of developing these hybrid techniques
is to take full advantage o f each method aiming at large area deposition and rapid growth.
It is seen that there is a strikingly large number o f methods to grow diamond.
Commonly in all techniques, a high supersaturation of atomic hydrogen is created along
with a supersaturation o f carbonic species and a substrate temperature in the range o f 7001000°C. Other similarities include process parameters o f gas pressure (~1/10 atmosphere),
the percentage o f hydrogen in the gas phase (~95-99.9% ) and the resulting film
morphology. However, there are a number o f other specific process parameters which are
considerably different from technique to technique. For instance, the conditions for and
modes o f gas activation are different for cold plasma and thermal plasma CVD, and, in
turn, both are considerably different from combustion flame deposition. These process
parameters lead to differences in the energy partitioning in the deposition process, the
deposition efficiency, deposition rate, and film uniformity to name just some o f the more
important ones. In thermal methods (HFCVD, atmospheric RF plasma and DC plasma jet,
combustion torch), the gas ionization is rare, and the temperature o f neutrals and electrons
are equal and much greater than the substrate temperature. In cold plasmas (low pressure
microwave, RF and DC glow discharges), there is a significant ionization in the gas phase,
and the electron temperature (10,000-20,000°C) is much greater than neutral and substrate
temperature (~1000°C). Both thermal and non-thermal methods can produce diamond with
higher growth rates occurring in thermal plasmas. Achievements to date
In the past several years, a large number o f papers appeared in the field of low
pressure diamond synthesis. There have been several excellent reviews on this subject
(DeVries, 1987; Badzian and DeVries, 1988; Angus andHayman, 1988; Spear, 1989). No
attempt will be made to duplicate these efforts by reviewing results from each significant
paper; rather the prominent results achieved will be summarized from an overall point of
1) Whereas PVD methods based on ion bombardment are the dominant general
deposition process for preparing the class of diamond-like materials which include dense
carbons and dense hydrocarbons, it is highly selective CVD processes which are capable of
controlling the growth of essentially single phase, polycrystalline diamond films.
2) There are many structurally and compositionally different types of carbon films
which can be deposited from the low pressure CVD processes. They often exhibit similar
apparent properties such as hardness and chemical inertness. Even within the so called
diamond films in the literature, there are a great many structures and compositions with
varying amount o f graphitic carbon, from amorphous (a-C) to microcrystalline (|i-C) to
large grain polycrystalline to semi-continuous multiple large crystals and to single crystal,
and containing hydrogen, silicon and various other impurities. Therefore, a clear working
definition of diamond films is necessary and has been proposed to distinguish real diamond
films from various other carbon films, including (i-C, a-C, DLC and graphite (Messier et
al., 1987). It is: 1) a crystalline morphology discernible by electron microscopy; 2) a single
phase crystalline structure identified by X-ray or electron diffraction; and 3) a Raman
spectrum typical for diamond, that is, a single, narrow line at 1332 cm*1.
3) It has been generally recognized that the gas mixtures, usually hydrocarbonhydrogen gases, have to be activated to achieve reasonable growth rates o f diamond under
low pressure CVD conditions. Dissociation but not ionization of reactant molecules is
absolutely necessary in the gas phase. High quality diamond films can be prepared by a
variety o f methods using various means of activation, for example, microwave or RF
plasmas, DC discharges, hot filament, etc., the choice of which largely depends on
considerations of efficiency, convenience, cost and applicability to the problems.
4) The process is kinetically competitive between the nucleation and growth of
diamond and graphite. A number of deposition parameters are of great importance in
promoting diamond growth while suppressing graphite deposition. These controllable
parameters include substrate material and its surface pretreatment, substrate temperature,
gas phase composition, gas pressure, gas flow rate, substrate position and power density.
Typical optimized deposition conditions are similar among the different methods. For
example, the substrate temperature is usually around 800-1000°C, the pressure is about a
tenth of atmosphere, and the concentration of carbon precursors such as methane in the gas
phase is from 0.1-5% in hydrogen.
5) A large amount of atomic hydrogen is a key factor in achieving diamond growth.
Actually, all the equipment design is based on the generation of a sufficient amount of
atomic hydrogen in the system. The role of hydrogen can be understood from two aspects.
Kinetically, hydrogen suppresses the co-deposition o f graphitic phases by etching graphitic
carbon preferentially and generating critical hydrocarbon diamond growth species.
Energetically, hydrogen stabilizes sp3 bonding sites on the growing diamond surface by
saturating dangling bonds which prevent surface reconstruction. It also can abstract
hydrogen from C-H bonds on the growing diamond surface providing energetic radical
sites for diamond species insertion. As a result, graphitic nuclei are hindered at the early
nucleation stage.
6 ) The beneficial role of oxygen has also been realized and has been introduced into
the hydrogen-hydrocarbon systems in various forms such as oxygen (O 2) (Kawato and
Kondo, 1987; Chang et al., 1988), water (H 2 O) (Saito et al., 1988), carbon monoxide
(CO) (Inspektor et al., 1989a), and with a number o f oxygen-containing organic
compounds such as acetone and methanol (Hirose and Terasawa, 1986). More recently,
atmospheric oxygen-acetylene flames have been used to grow diamond (Hirose and
Kondo, 1988). Oxygen, which itself is an effective etchant of graphitic phases, opens new
reaction paths in the plasmas and near the surface. Consequently, the dissociation rate of
reactant molecules is increased, more active diamond nucleation and growth sites are
produced, and the diamond growth rate can be enhanced. The oxygen introduction into the
gas mixtures expands the parametric range of diamond formation and can lead to an
effective way of rapidly growing diamond films at high hydrocarbon concentrations while
still keeping to a minimum the graphitic carbon content in the films.
7) The chemical nature of the hydrocarbon precursors does not appear to have a
significant effect on the deposition behavior of diamond, because common critical growth
species or radicals are produced under gas activation (Sato et al., 1987). Various saturated
or unsaturated aliphatic hydrocarbons (CH 4, C 2H 4 , C 2H 2), alcohols (CH 3OH, C 2H 5 OH)
and ketones (CH3COCH3) have been used satisfactorily to deposit diamond films.
8 ) A combustion flame method was invented to produce diamond from an
atmospheric pressure oxyacetylene torch (C2 H 2/O 2). This is a remarkable achievement
which fully demonstrates the diversity of methods that can be used to grow diamond. The
use o f combustion to grow diamond also opens a new experimental field to study
mechanisms and kinetics o f the process. High growth rates (50 -100 pm/h) can be achieved
using this method, but non-uniformity is a serious problem. Since the roles o f oxygen are
the same as some o f those given to atomic hydrogen, this partially explains why diamond
can be grown in an oxy-acetylene flames.
9) The growth rate has been steadily increased with the development o f new
techniques. The growth rate in early thermal decomposition work without gas phase
excitation was very low (<0.1|im/h). Using microwave plasma enhanced or hot filament
assisted chemical vapor deposition techniques, growth rates o f 0 . 1-2 pm/h can be easily
achieved. Electron assisted CVD increases the growth rate to 3-5 pm/h and further to 8-10
pm /h with the use of oxygenated organic compounds such as acetone, methanol and
ethanol in HFCVD (Hirose and Terasawa, 1986). A microwave plasma je t yielded growth
rates o f 30 pm/h (Mitsuda et al., 1989). Growth rates of 60 (im/h in atmospheric RFCVD
(Matsumoto et al., 1987) and 80 |im/h in a DC plasma jet (Kurihara et al., 1988) have been
reported. Recently growth rates as high as 930 pm/h using a high speed and high density
plasma jet generated by a DC arc discharge plasma torch have been claimed (Ohtake et al.,
1989). The growth rate mainly depends on the type of CVD process and to a lesser degree
on the gas composition, gas pressure and substrate temperature. It appears to scale with
power and gas phase species density with higher rates achieved in atmospheric DC o r RF
thermal plasmas and combustion flames, indicating the importance o f high electron
densities and high plasma temperatures.
10) The pressure range has been extended from tens of torr in microwave or hot
filament assisted CVD up to atmospheric pressure with the development of high power
thermal plasmas, chemical transport reactions and combustion flame techniques. In these
thermal plasmas, the temperature of free electrons, ions, radicals and neutral species are the
same as the gas temperature (5000-8000°C). Under such high temperatures and energy
levels, higher deposition rates are expected and indeed have been obtained.
11) There is a maximum in the growth rate as a function o f substrate temperature at
about 950-1000°C. Low temperature diamond deposition near 350-400°C has been
realized either by reducing the gas pressure and power density or by "remote" deposition
(Liou et al., 1989; Badzian et al., 1990), but it suffers from extremely low growth rates,
typically only around 50 nm/h. Oxygen is an important ingredient for ensuring diamond
deposition at low substrate temperatures.
12) A variety o f substrate materials have been used for the deposition o f diamond
films, although most o f the reports concern diamond deposition on silicon. The various
substrates can be classified into three groups: a) diamond, graphite and amorphous carbon;
b) stable carbides and carbide-forming materials such as SiC, TiC, W C, Si, Mo, Ti, W and
Ta as well as various oxides (Si0 2 , AI2 O 3), nitride (BN, Si3N4 ) and alloyed steels; and c)
non-carbide forming materials such as Cu, Au and Ni (Spitsyn et al., 1981; Sawabe and
Inuzuka, 1986; Chang et al., 1988; Joffreau et al., 1988; Rudder et al., 1988). Depositions
on the materials in the first two groups are generally successful, although there are
adhesion problems for some materials like graphite and Si0 2 due to lattice or thermal
expansion mismatch. Depositions on the third group o f materials are unsatisfactory. There
have been independent reports indicating the formation o f an interfacial carbide layer is a
necessary step for the initial diamond growth on carbide forming foreign substrates (Belton
et al., 1989; M eilunas et al., 1989; Williams and Glass, 1989). Non-carbide forming
materials are believed to have a bulk diffusion coefficient for carbon so large that nucleation
on the surface is difficult (Joffreau et al., 1988). It is interesting to note that the best
catalysts for HP/HT diamond synthesis, namely, Fe, Co, and Ni metals, are not good
substrates for CVD growth of diamond films.
13) Pretreatments o f the foreign substrate surface are very important for increasing
the nucleation rate. Polishing, scratching o r seeding using diam ond powders and
ultrasonificadon in diamond suspensions are effective means for achieving high nucleation
densities. Polishing or seeding with other powders o f hard substances such as BN and SiC
also is useful but to a lesser efficiency (Bachmann et al., 1988). The high energy sites,
such as scratches on the surface and some particles adherent to or embedded in the surface,
are believed to act as favorable nucleation sites. A DLC coating on foreign substrates can
also increase the nucleation rate. Ion beam bombardment poisons the surface for
14) Structurally perfect diamond can be grown homoepitaxially on the {100}
diamond planes but with difficulty on the {111} planes. Boron-doped p-type homoepitaxial
diamond films on the {100} diamond surface have been successfully prepared (Fujimori et
al., 1986). Active electronic elements such as Schottky diodes and metal-semiconductor
field-effect transistors (MESFETs) using these semiconducting CVD diamond films have
been fabricated (Gildenblat et al., 1988,1990; Shiomi et al., 1989).
15) Plasma diagnostics using emission, absorption and mass spectroscopies have
shown that, among a variety of hydrogen, carbon and hydrocarbon species, the primary
species in the diam ond forming plasma are methyl and acetylene radicals. Growth
mechanisms according to these two species have been proposed (Tsuda et al., 1986;
Frenklach and Spear, 1988). It is likely that the relative importance of each species may
change depending on the growing surface and specific techniques.
16) The dominant moiphology o f CVD diamond is {111} and {100} surfaces with
cubooctahedral-shaped crystals commonly occurring. However, many other morphologies
have been reported including pentagonal five-fold twinned crystals (Matsumoto and
Matsui, 1983). Twinning and stacking faults in diamond {111} planes are overwhelmingly
present because o f the small energy difference between the perfect and defective crystals.
As a result, multiple-twinned crystals, such as decahedral and icosahedral, are often found.
Among many other parameters, the internal structure of diamond films, such as graphitic
second phase content and planar defects density, is critically dependent upon the
proportions o f carbon, hydrogen and/or oxygen in gaseous mixtures. It is relatively
independent o f the activation types o r deposition techniques. The occurrence o f these
structural defects is a major obstacle towards achieving CVD diamond homoepitaxial or
heteroepitaxial growth for electronic applications.
17) The purity o f CVD diamond, which can be effectively controlled by the supply
gases and growth environment, is higher than most HP/HT synthetic diamonds. The most
common impurities in CVD diamond films are hydrogen and silicon. The concentration of
hydrogen is typically about 500 ppm (Hartnett, 1988), and that o f silicon is around 0.2 at%
(Badzian et al., 1988a). Silicon typically comes from the etching o f the substrate (Si) and
the reaction cham ber (fused Si0 2 ) and can obviously be eliminated. However, the
replacement materials could lead to similar problems. These impurities, together with
crystalline defects, degrade many physical properties o f CVD diamond films.
18) As for the characterization o f diamond films, Raman spectroscopy is the most
popular and sensitive technique for identifying diamond and graphite phases among various
carbon forms. Small amounts of graphitic phases can easily be detected, because the
scattering efficiency for graphite is much greater than that for diamond (Wada et al., 1980).
Both position and width o f the Raman peaks provide information on the type o f carbon
present, the structural perfection and the internal stress state. Electron and X-ray diffraction
techniques also are used widely, but they are only sensitive to crystalline phases, giving
little inform ation about the non-crystalline carbon phases. A variety o f other
characterization tools, such as infrared spectroscopy and cathodoluminescence, have been
employed to reveal structures and properties o f diamond films in different specific aspects. Applications o f CVD diamond films
There is now great motivation to scale up and commercialize chemical vapor
deposition o f diamond films, not only because less expensive equipment, low operating
costs and high purity diamond are involved in the low pressure CVD techniques, but more
importantly because the unique combinations o f properties o f diamond coupled with the
new ability to make it in thin film forms on a variety o f materials herald a new era in
diamond technology and open up more opportunities for potential applications. With the
traditional HP/HT methods, it is not possible to take full advantages o f the properties o f
diamond by coating other materials o f different sizes and shapes with diamond films.
Diamond films for abrasive, bearing and wear resistant surfaces, tool coatings and
corrosion protection are obvious because of the high hardness and strength, wear resistance
and low friction. The biocompatibility of carbon leads to diamond coated joint materials,
teeth and biosenser applications. The high thermal conductivity makes diamond films an
ideal heat diffuser material for high temperature, high power semiconductor electronic
devices, thus allowing a high degree of circuit integration and a more dense packaging of
the devices without thermal problems. Diamond films can be doped during the deposition
process, and the resulting semiconducting diamond has a high hole mobility and a high
breakdown voltage. This may ultimately lead to active semiconductor diamond elements
that complement the current silicon and gallium arsenide technologies. The high elastic
modulus and low thermal expansion of diamond films result in high quality loudspeaker
diaphragms. UV-visible-IR transparency makes diamond films an interesting material for
optical applications, such as windows, lens coatings, floppy and compact disk coatings and
X-ray lithography masks. Chemical inertness allows applications as protective coatings for
chemical and nuclear reactors. It even appears possible to deposit diamond films on plastics
with the development of low temperature techniques.
However, many o f these applications have not been realized and await further
progress in improving diamond crystalline quality, adhesion strength, surface smoothness,
heteroepitaxy, low structural defect density, etc. Nevertheless, in some areas, such as wear
resistant coatings on cutting tools, free-standing films for X-ray windows and high
frequency loudspeaker diaphragms, diamond films are on the threshold of large scale
industrial applications.
. Current problems
Despite the fact that the level of research activities in the field o f low pressure CVD
of diamond has grown tremendously in the past decade, metastable diamond growth
technology is still far from being capable o f large scale commercialization to realize those
attractive potential applications outlined in the above section. Many problems remain to be
solved before a realistic assessment o f the potential and impact o f this new synthetic
method can be made. A recent Science article by Yarbrough and Messier (1990) critically
examined many o f the major issues and problems currently attendant to the CVD o f
diamond. In the following, a brief list o f the problems facing the scientific community is
presented, which is by no means complete.
1) Although a common basic mechanism for diamond formation is very likely,
because there are a variety of methods capable of producing diamond films, and there are
so many common features involved in the diamond deposition processes, a sound
understanding o f the nature o f the diamond CVD process has not been established.
Experimental data on deposition kinetics, other than the Soviet scientists' early work
(Fedoseev et al., 1984), are lacking. M uch better understanding in the fundamental
chemistry o f the process, identification of dominant growth species, both in the vapor and
at the growing film surface, and specification of reactions involving them are required to
further its development whether for optimization or for scale-up.
2) The understanding o f nucleation and interface phenom ena is poor. Such
information is needed to solve problems related to the poor film adhesion to substrates, low
nucleation rates on certain substrates and non-uniformity across a substrate surface. High
nucleation densities without polishing or seeding pretreatments for coating smooth
surfaces, such as bearing and optical components, are yet to achieve. The nucleation issue,
if understood in detail, will be very helpful in achieving better control o f homoepitaxial
deposition and ultimately of heteroepitaxial growth of diamond.
3) Many o f the physical properties o f CVD diamond are inferior to those of natural
diamond, including electrical resistivity, optical transparency and thermal conductivity.
Structural imperfections such as extensive planar defects, grain boundaries, surface
roughness and impurities are the main factors affecting these properties. Effective control
or elimination of these imperfect structural features has not been achieved y e t
4) The general needs for industrial applications o f diamond films, i.e., large area
and high rates, particularly at low temperatures, have not been m e t Although high growth
rate techniques have been reported, they apparently suffer from non-uniformity, small
coverage, poor adhesion and poor reproducibility. The scale-up o f systems for large area,
uniform deposition is difficult.
5) Specifically, for tribological uses, adhesion o f diamond films on "hostile"
surfaces, such as ferrous alloys, Ni or Cr based alloys, silica, and silica based glasses
needs to be improved. The surface o f diamond films needs to be passivated to resist
oxidation, graphitization o r chemical reactions. Experimental techniques capable of
handling irregular-shaped substrates also need to be developed for conformal coatings. For
electronics applications, heteroepitaxial and doping technologies (n-type) must be
developed. Effective n-type dopants are needed to produce n-type semoconductor diamond.
For optical applications, the major problems are the adhesion o f diamond films on optical
or electrooptical materials and the film surface roughness or scattering. Diamond films with
small crystallites and smooth surfaces are desired.
6 ) Low temperature (<600°C) diamond deposition rates and process controls need
to be improved to enhance its applicability in coating heat sensitive materials such as
glasses, plastics, and steels.
1.5. Thesis plan
1.5.1. Statement o f the problem
Although there have been a number of papers on many aspects o f CVD diamond
preparation, characterization and properties over the last decade, there have been very few
detailed and systematic studies which cover all three areas. Interest in exploring new
techniques is much more advanced than our understanding of the basic processes involved
in the diamond deposition and the resulting film structure and properties. As discussed
above, m^ny basic issues still remain to be answered. Furthermore, polycrystalline
diamond films are not well defined but cover a continuum o f morphologies and internal
structures'with different levels o f contaminants. Many experimental variables and process
parameters are considerably different from technique to technique. Even within a specific
plasma-based deposition method, the exact geometry and materials used in a particular
plasma reactor are important. Thus, it is extremely difficult to compare partial results from a
large number o f studies to get a coherent picture of the diamond thin film deposition. The
limited understanding o f the nature of diamond growth under low pressure CVD conditions
is a major factor retarding the progress o f development in this field.
W hat is needed are studies carried out systematically in sufficient detail so that
general conclusions can be established which contribute to our current body of knowledge.
Such experimentally-based studies are extremely important in order to test theoretical work
already published or in progress and to perhaps inspire other studies. The general problem
of achieving a complete understanding o f the diamond processes involves many aspects
ranging from the basic theories o f kinetics and thermodynamics, to the chemistry o f plasma
and surface species and reactions, to the different routes to CVD growth, and to the film
structure and resulting properties. Certainly a complete understanding of the diamond
deposition process is the ultimate goal. Like any emerging materials the solutions to the
different parts o f the problem are never easy, and even a partial understanding takes many
years of study. That is the nature and history of materials research. This thesis will be
concerned in detail with the internal structural features o f CVD diamond films which is a
critical link between the film preparation and the resulting film practical performance.
1.5.2. Thesis goals and outline
The primary goal of this work is to achieve a comprehensive understanding of the
internal structure of CVD diamond films with the deposition-structure-property relations as
a thread. The means to this end is through a systematic approach to investigate the major
experimental parameters and the plasma chemistry o f the MPECVD process, to study the
internal structure o f diamond films which includes graphite inclusions, interfaces and
crystalline defects, and to correlate these internal structural features with the parametric
deposition process and the oxidation properties of diamond films. This will naturally divide
the major content o f the work into four different chapters. Chapter 3 deals with a parametric
investigation o f the deposition process and the resulting film structure; Chapter 4 describes
the investigation o f plasma behavior and the influence of the noble gas-involved plasma
chemistry on the film structure; Chapter 5 discusses in detail the internal bulk and interfacial
structures as well as the crystalline defects of diamond films; and Chapter 6 focuses on the
results o f oxidation experiments which are directly related to the structural defects and
interfacial phenomena o f diamond films. By the end of this work, correlations among
deposition parameters, film growth rate, plasma chemistry, various structural features, and
oxidation properties of diamond films are established. In particular, a clear image of the
overall internal structural properties o f CVD diamond films and their correlations with the
deposition process will emerge. The results will lead to a better understanding of the
preparation-structure-property relations in the diamond film deposition. It is the author's
belief that the efforts made in this thesis will be o f significance in promoting the
understanding of the fundamental nature of the MPECVD process and providing guidelines
for designing experiments to deposit tailored structures o f diamond films required for
particular applications.
Chapter 2
In this chapter, experimental details will be described for preparation and
characterization o f diamond films. Following a general introduction of the low pressure
vapor synthesis o f diamond, the tubular system of microwave plasma enhanced CVD
(MPECVD) used in this work is presented in detail. The main characterization techniques
o f electron microscopy, Raman spectroscopy, optical emission spectroscopy and
thermogravimetric analysis are described. Other techniques such as IR spectroscopy, laser
induced luminescence, secondary ion mass spectrometry and neutron activation analysis are
mentioned briefly, as well.
2.2. General approaches for preparation of diamond films at low pressures
The growth of diamond from a vapor phase is probably the most popular method in
attempting low pressure diamond growth, although other techniques such as growth from
molten liquid phases have been reported (Deijaguin et al., 1970). The basis o f vapor
growth is that atoms (or molecules, ions, and excited state species) impinge on the
substrate surface from the vapor phase and are adsorbed. These adatoms can diffuse a
considerable distance before desorption. When the vapor pressure is greater than the
equilibrium pressure in the system, a net deposition or growth occurs.
Historically, both physical vapor deposition (PVD) and chemical vapor deposition
(PVD) processes have been used for achieving metastable diamond growth as reviewed in
section Some general principles involved in these processes are briefly discussed
as follows.
2.2.1, PVD process
PVD techniques basically include evaporation, sputtering, and ion plating. Various
modified methods based on these techniques have been introduced in attempts to grow
diamond films (Weissmantel et al., 1979; Holland and Ojha, 1979; Anderson, 1981). They
essentially rely on the energy and momentum of impinging ionized species to stabilize a
metastable structure and can produce carbon films called i-C which exhibit apparent
diamond properties. In these PVD processes, the formation o f metastable structure
(microcrystalline diamond) in i-C has been interpreted as due to the occurrence of
temperature and pressure spikes at the instant of ion bombardment and rapid collapse o f the
spikes (Weissmantel et al., 1980; Namba and Mori, 1985). Other mechanisms also have
been suggested and reviewed in detail elsewhere (Angus et al., 1986). Both atomic
agitation and rapid quenching are important to the understanding of the non-equilibrium
process o f i-C deposition. Later, these i-C films were referred to generally as diamond-like
carbon (DLC) which is now used to describe a variety of carbon materials ranging from
amorphous to microcrystalline, typically containing an appreciable amount o f hydrogen
(10-50%) (Sokolowski et al., 1979; Mori and Namba, 1983).
2.2.2. _CVD process
CVD methods, which can control the surface chemistry of growing films by
manipulating a large number o f variables, are generally successful in depositing real
diamond films. Most o f the CVD diamond processes are plasma enhanced or thermally
activated. Therefore, they are more complex to control and understand than corresponding
PVD processes.
CVD, a materials synthesis process o f forming solid phases by decomposition
and/or chemical reaction o f vapor phases near or on a substrate surface using heat, plasma,
or other energy sources, is not a new technique. It was used to refine refractory metals in
the 1800$, to produce filaments for Edison's incandescent carbon filament lamps in the
early 1900s, for hard metal coatings in the 1950s and for semiconductor material
preparation beginning in the 1960s (Hammond, 1988). It has now become one of the most
important means for creating thin films o f a wide variety of materials essential to advanced
technology, particularly solid state electronics. The main feature is its versatility for
synthesizing materials, both simple and complex compounds. Film structures and chemical
compositions can be tailored by the control o f reaction chemistry and deposition conditions.
Fundamentals principles encompass an interdisciplinary range o f gas-phase reaction
chemistry, thermodynamics, kinetics, transport mechanisms, film growth phenomena and
reactor engineering.
Using plasmas to promote a gaseous chemical reaction has been known for over a
century. However, prior to the last decade, the plasma process had little commercial
success because o f the difficulty in controlling it and the complexity o f the phenomena
involved. In recent years, however, plasma enhanced chemical vapor deposition (PECVD)
techniques have become increasingly important. PECVD of silicon insulator dielectrics and
semiconducting materials, such as Si0 2 , Si3N 4 , and amorphous and polycrystalline Si, are
the most important commercial uses o f film formation for microelectronic and photovoltaic
applications. However, PECVD is not understood nearly as well as either conventional
thermal CVD or related PVD techniques such as plasma sputtering.
There are three general categories of plasmas, depending on the temperature range
o f their operation and thermodynamics o f the plasma: nuclear fusion, thermal plasmas and
non-equilibrium, or cold, plasmas. The theoretical understanding and experimental use of
these plasmas are very different and have become quite distinctive technologies. Nuclear
fusion (>10 6 K, <100 MW) is a specialty field aiming at generation o f power. In thermal
plasmas such as arcs at atmospheric pressure, the electrons, ions and neutral gas molecules
are in local thermodynamic equilibrium (10 3 -10 4 K, 0.1-10 MW). They are used in
chemical reactions, heating gases, welding and melting. Non-equilibrium or cold plasmas
are usually generated by the application o f electric field, such as DC, RF or microwave
forces (1-103W), across a body o f low pressure gas. They are electrically quasi-neutral (Ne
~ Ni), although this quasi-neutrality can be violated in the close vicinity o f surfaces. An
important characteristic o f a low pressure plasma is that the electron temperature is typically
30-1000 times greater than the average gas molecules temperature (10 4 -105oC vs. 25300°C). The translational and rotational excited species have temperatures between 100700°C, and vibrational and electronic excited species have temperatures around ~1000°C.
Therefore, the particles composing the plasma vary widely in temperatures (Te > Tj ~ Tn),
and the plasma is far from thermal equilibrium. Low pressure plasmas are utilized widely
for film deposition, VLSI technology, ion implantation and chemical reactions.
The primary role o f a plasma in chemical vapor deposition is to produce chemically
active species that subsequently react with each other to form solid films. A key advantage
over conventional thermal CVD is that substitution of electron kinetic energy for thermal
energy avoids excessive heating and possible consequent degradation o f substrates. It is the
highly energetic electrons in the non-equilibrium plasmas that enable many reactions to take
place at low temperatures. Elevated temperatures necessary to initiate classical chemical
reactions are not needed. Another beneficial aspect of PECVD is that plasma species are
constantly bombarding the growing film. This can influence significantly the film growth,
its composition, structure and stress. Therefore, the possibilities fo r producing films o f
various materials and for tailoring their structures and properties by judicious manipulations
o f reactant gases and plasma parameters are very extensive in PECVD process. It is widely
used to produce films at lower substrate temperatures and in a m ore energy-efficient
manner than other techniques and is particularly suitable for stabilizing metastable phases
such as diamond.
2.3. MPECVD o f diamond films
In this section, the unique characteristics of microwave plasmas for film deposition
will be emphasized. Experimental details including the deposition system and procedures
for the growth of diamond films are described.
2.3.1. Characteristics of microwave plasmas
Microwave frequency-excited plasmas differ essentially in many ways from lower
frequency and DC plasmas: 1) electrons in microwave plasmas oscillate within the volume
o f gas and constantly collide with gas atoms to cause excitation, dissociation or ionization.
The energy transfer process is very efficient, and the plasma density is high. In lower
frequency plasmas, the number o f electrons lost by colliding with walls is significant; 2 )
the electron energy distribution function (EEDF) depends on o/co, where 1) is the electronneutral collision frequency and co is the angular frequency of the applied electromagnetic
field. Generally, the average electron temperature decreases with CO. However, for
microwave plasmas, o/co <1, EEDF approaches a Maxwellian distribution, and there is a
high population o f energetic electrons in the tail o f the Maxwellian EEDF. For lower
frequency plasmas, o/co >1, EEDF is non-Maxwellian, and there is a large population of
low energy electrons; and 3) microwave plasmas are a rich source o f intense emissions in
the UV and visible region. These emissions can stimulate various photo-chemical reactions.
For lower frequency plasmas, this photo-chemical effect is relatively weak. Therefore, the
difference between microwave and lower frequency plasmas with regard to deposition is
due chiefly to the high energy levels o f the plasmas and high electron and ion densities.
More efficient chemical reactions can be expected from a MPECVD process.
2.3.2. MPECVD system
Alm ost all types o f PECVD systems (DC, AC, RF, microwave) have three
elements in common: 1) they are sustained by a source of electrical power; 2 ) the electrical
power is delivered by means of a coupling mechanism; and 3) it is delivered to a plasma
environment associated with a particular design. The design o f a reactor system (reactor
chamber and all associated equipments) for carrying out CVD processes must provide
several basic functions. It must allow transport o f reactant and diluent gases to the reaction
site, provide activation energy to the reactants (heat, radiation, plasma), maintain a desired
pressure and temperature, allow chem ical reactions for film deposition to proceed
optimally, and remove by-product gases. These functions must be implemented with
adequate control, maximal effectiveness and complete safety.
A t Penn State, work on microwave activation o f hydrocarbon gases at low
pressures began in the early 1960s with the main purpose o f understanding the gas phase
species (Vastola and Wightman, 1964). A MPECVD system was constructed in the early
1970s by Knox and co-workers for investigation of hard deposits produced on the walls of
the apparatus from hydrocarbon discharges (Knox and Vedam, 1976). Small sized
diamonds were believed to have been crystallized.
The first practical MPECVD system for low pressure diamond synthesis was
described by Kamo et al. in 1983. The experimental set-up for this work was based on the
reported Japanese system, which consisted o f a microwave generator and waveguide
system, a gas-handling system, a tubular silica reactor and a pumping system as illustrated
in Figure 2.1a. A different design o f MPECVD system for diamond synthesis has been
established also at Penn State, in which a large area bell jar contains the plasma (Bachmann
Figure 2.1. The tubular microwave plasma enhanced chemical vapor deposition system:
(a) schematic diagram o f the MPECVD system; (b) the waveguide for
microwave transmission; (c) the substrate/susceptor assembly; (d) the reaction
chamber; and (e) the gas flow and control system.
Screw Tuner
Sliding Short
Guide f"
2 .4 5G H z
O . l - 1.5 kw
Power M onitor/
Directional Coupler
Silicon Single Crystal
Graphite Susceptor
Silica Glass Rod
Stainless Steel Holder
Figure 2.1. (cont.)
+- SILICATUBE (<f)=46mm)
2450 MHz
^~-Zzjx==--- ► TO PUMP
Figure 2.1. (cont.)
Electronics for Moss Flow and Control Meters
K- I f| OBarotro
Vacuum MKS Pressure
Control Valve
Traps v
Rotary Vane Pump
E xhaust
F ilter
No Dilution
Figure 2.1. (cont.)
et al., 1988). A detailed account of the tubular MPECVD system is as follows. Microwave generation and transmission
A Toshiba microwave generator (TMG-132F/US) which could be operated at
powers between 0.1-1.5 kW at 2.45 GHz frequency was used. The electrical power supply
(220V, 60Hz, 2.7kW) to the generator was stabilized by a line voltage conditioner and
controlled by a Variac. The microwaves generated by the magnetron were transmitted to the
reaction chamber through a set o f aluminum rectangular waveguides (109.2mm x 54.6
mm) (Figure 2.1b). A n air cooled isolator allowed the microwaves to pass through from
the generator but protected the magnetron by absorbing reflected microwaves from the
applicator. It also stabilized the output power to the load (plasma) in a resonant cavity. The
forward microwave pow er was monitored by an anode voltage monitor, and the reflected
power was monitored by a current indicator mounted at the directional coupler in the
waveguide. Both the forward and reflected power calibrations were made by Toshiba and
were provided with the equipment. Minimization o f the reflected power was required and
achieved by adjusting three screw tuners built in the waveguide. The function o f these
tuners was to improve the energy efficiency by matching the impedance o f the cavity load
(plasma) to that of the waveguide so that the resonant frequency was tuned in, and the
power reflected from the cavity was minimized. The impedance o f the load and the resonant
frequency were functions o f the type of gas, pressure, power input, the substrate material
and its position relative to the plasma. Therefore, tuning and matching adjustments were
necessary to obtain efficient operations over a wide range o f plasma conditions. At the end
of the waveguide there was a movable plunger which could adjust the antinode o f the
electric standing wave (TEio mode) to the center o f the resonant cavity by its shorting plate,
therefore, optimizing the plasma shape and position. It could also be used to reduce the
reflected power. The part between the three tuners and the plunger was a water cooled
microwave applicator for plasma generation. Reaction chamber and plasma generation
The reaction chamber was a vertically-mounted fused quartz tube (Quartz Scientific
Co.), 5 cm O.D. and 91 cm long, which was placed within the water cooled applicator
(Figure 2. Id). Both sides o f the tube were sealed to stainless steel cross pipes with O-rings
(Buna-N). Located at the very top of the tube was a glass viewport which was used to
observe the progress of the deposition and to allow the measurement of the substrate
temperature with an optical pyrometer (Leeds and Northrup Co.) or an infrared radiation
thermometer (Ircon). The reactant gases (CH4+H 2) entered the tube through the gas inlet at
the top. In the reaction tube, the gas mixture absorbed microwave energy and created a
luminous plasma in the cavity. A substrate, which was placed usually on a susceptor made
o f graphite to avoid contaminating the growth environment, was mounted on a stainless
steel holder with a silica glass extension rod (Figure 2.1.c). This stainless steel substrate
holder was actually a motion feedthrough. Therefore, the substrate position could be
adjusted or rotated readily in the tube if necessary without destroying the vacuum. The
substrate was immersed usually in the plasma and heated by interactions with the
microwaves and the plasma. The silica glass extension rod stayed relatively cold as long as
it was clean. The potential o f the substrate floated at the potential o f the plasma due to the
insulating silica extension rod. The diamond film was deposited on the substrate surface as
a result of direct contact with the highly reactive hydiocarbon-hydrogen plasma.
Some aspects need to be further elaborated regarding the interactions between the
plasma and the substrate and the measurement o f substrate temperature. The plasma volume
was a function o f pressure and absorbed microwave power. The plasma should be adjusted
to the center o f the reaction tube so that it did not touch the tube wall and uniformly covered
the whole substrate surface. Otherwise, the tube might be damaged easily by local carbon
deposition, etching or even melting on its wall which could severely distort the plasma
formation in the cavity and contaminate the growth environment. Uniform films could be
deposited on 1 inch (~25 mm) diameter substrates in this system. If the substrate was too
large, temperature variations across the surface were great, and the plasma might touch the
wall, leading to a non-uniform film. The substrate could be heated sufficiently (>900°C)
without the graphite susceptor, at least for silicon. However, for the sake of better surface
temperature control, convenient substrate mounting and reduction o f contamination from
the substrate holder material (silica), the graphite part always was used. The geometry of
the graphite susceptor was important, since it had to ensure the least disturbance o f the
plasma, minimal absorption of microwave energy, sufficient protection o f the silica
extension rod from plasma etching, uniform radiation heating for the substrate and minimal
plasm a etching of itself. The present geometry, a thin, circular shape with a stem
underneath, was assumed to be optimum after many trial and error designs.
The substrate temperature critically depended on the substrate position relative to the
plasma, microwave power input and pressure. A variety o f combinations of these three
factors could fulfill the same temperature requirement The optical pyrometer had a nominal
operating wavelength near 650 nm and a useful temperature range o f 750-1250°C. It was
calibrated by the melting o f silver. There were errors introduced by emissivity variations
during the deposition, but these errors were believed to be minor and could be eliminated
by using a two color pyrometer (Hartnett 1988). The infrared radiation thermometer was
operated at a wavelength band between 2.0-2.6 p.m and a temperature range o f 4001100°C. The emissivity calibration was referenced to the optical pyrometer. During the
measurement o f the substrate temperature, the contribution from the plasma was assumed
to be negligible because its emission intensity was not significant enough within the
bandwidth o f the pyrometer or thermometer, and the transmittance of the viewport glass
was assumed to be unity. The infrared radiation thermometer was convenient for use in
setdng and monitoring temperature because of its digital readout. But it was more sensitive
to emissivity changes, and the temperature readings were considered to be less accurate
than the readings from the optical pyrometer.
2.3.2,3, Gas flow and control system
The gas handling system was composed o f mass flow meters (MKS), a Baratron
pressure gauge (MKS), a pressure control valve (MKS), respective electronic control units
(MKS), stainless steel tubing, hand operated valves (Nupro) and a rotary vacuum pump
(Leybold-Heraeus) (Figure 2.1e). The whole flow system used arc welded "VCR" fittings
and was checked for integrity with a helium leak detector. Methane and hydrogen were the
main reactant gases. They were used as-supplied, UHP grade from Linde-Union Carbide
Co. without further analysis and were mixed before entering the reaction tube. The mixed
gas was injected at the top of the tube, about 60 cm from the substrate position, flowing
downstream against the substrate surface, and pumped out the bottom o f the tube by a
vacuum pump with a speed o f 760 liter/min at atmospheric pressure and an ultimate
pressure o f 3 xlO *4 Torr. Its inlet could be shut off with a throttle valve, and in order to
control the pressure in the reaction tube, its conductance was effectively controlled by
means of feedback actions of a solenoid actuated butterfly valve, a capacitance manometer
and pressure control units (MKS). Nitrogen purging gas was connected to the exhaust line
in order to dilute the hydrocarbon-hydrogen gases before release.
The reaction tube needed to be evacuated prior to deposition. An increasing
pressure rate of 0.1 Torr/h or less due to a system leak when the gas inlet and outlet were
closed was considered reasonable to proceed with the deposition. The calibration of the gas
flow meters was conducted by using a soap bubble meter and measuring the rising speed of
the soap film. A major factor influencing the accurate control o f gas flow was the drifting
o f mass flow meters and controllers due to ambient temperature variations in the laboratory,
especially when very low flow rates were involved. The necessity of calibration of these
flow meters and controllers is made clear by the deviations between the set points and the
measured actual flows as shown in Figure 2.2. This kind of flow deviations might lead to
misinterpretations o f growth rates and quality o f diamond films. Therefore, calibration, or
at least determination o f the actual gas flow, was im portant for obtaining reliable
experimental data.
2.3.3. Deposition procedures
The whole deposition procedure on this tubular MPECVD system is described as
1) Insert the substrate (single crystal Si) into the reaction tube from the bottom
motion feedthrough. The substrate surface usually is pretreated by polishing with 1 |rm
diamond paste. Adjust the position of the substrate 1.5-2 cm below the center o f the cavity.
2) Turn on the pump to evacuate the tube. Set the desired pressure and flow rate (90
Torr, H 2 : 80 seem, 5 %CH 4 + H 2 : 20 seem). Turn on the cooling water for the applicator.
3) Once the pressure reaches 10' 1 T orr in the tube, switch on an argon gas flow to
purge the tube. A fter two cycles of purging, ignite a hydrogen plasma at a pressure
between 10-20 Torr. (The operation of microwave generator is omitted here.)
4) Center the plasma ball and minimize the reflected power by adjusting the plunger
and three tuners as the pressure in the tube continues to rise to the desired set point (90
Actual Flow (SCCM)
Actual Flow
Ideal Flow
Set Point (SCCM)
Figure 2.2. Calibration curves of the gas flow meter
5) Make sure the appropriate substrate temperature is reached by controlling the
microwave power in p u t The substrate temperature is measured by an optical pyrometer or
an IR thermometer.
6 ) Once the temperature is in the appropriate region (950-1000°C), switch on
methane flow. Monitor any temperature fluctuations on the substrate surface and the plasma
stability, and make corresponding adjustments if necessary.
7) The deposition process usually lasts about 8 hours. The diamond film can be
deposited at a rate of about 1 |im/h.
8 ) When the deposition is complete, shut off the microwave generator. Allow a
continuous flow of hydrogen for another 0.5 hour for cooling. Switch off all gas flows and
stop pumping. Vent the system, and the sample is ready for analysis.
The process for making diamond can be run successfully only after the optimization
o f process variables. A detailed parametric investigation was conducted and will be
presented in Chapter 3. Briefly, the plasma should be as large as possible without touching
the wall to cover and coat the whole substrate uniformly. The most suitable position of the
substrate relative to the plasma should be that which produces the most efficient heating
(the highest temperature) at a given power and pressure with no partial plasma formed
underneath the graphite susceptor. This is about 1.5-2 cm below the geometric center of the
cavity. Other standard conditions for diamond deposition in this tubular MPECVD system
include: forward power = 500-800 watts; reflected power = 10-20 watts; pressure = 70100 Torr; substrate material = <100> oriented single crystal silicon; substrate temperature =
950-1000°C; methane concentration in hydrogen = 0.2-1%; and total gas flow = 100-200
seem. The typical conditions for diamond film deposition used throughout the thesis will
refer to the above parameters. However, the optimum deposition parameters are system
specific. Thus, precise comparisons o f parameters from one deposition system to another
are difficult.
2.4. Characterization techniques for diamond films
In this work, characterization o f diamond films focused on extensive structural
investigations by transmission and scanning electron microscopies (TEM and SEM).
Raman spectroscopy was used to monitor the crystallinity and relative structural changes of
diamond films. The deposition process was characterized by optical emission spectroscopy
(OES) to gain insight into the plasma chemistry in systems of methane-hydrogen-noble
gases. The film growth rate was measured as a means for comparison o f deposition
efficiency under different conditions. Thermogravimetric analysis (TGA) was performed to
study the oxidation properties o f diamond films. Other techniques included secondary ion
mass spectrometry (SIMS) and neutron activation reactions (NAA) for compositional
analysis, infrared spectroscopy (IR) for the study o f chemical bonding, and laser excited
luminescence for the study of point defects.
2.4.1. Film growth rate measurement
The growth rate of diamond films is the most popular criterion used for comparing
results obtained from different experiments and different systems. The measurement is the
first and one of the most important steps in determining whether a deposition process is
successful, efficient or practical.
The measurement o f film growth rate can be fairly simple or quite complex. A direct
method is to measure the step height when a portion o f the deposited film is etched away or
a part o f the substrate surface is masked during the deposition process. This is usually done
on a surface profilometer that electronically tracking the position o f a mechanical stylus.
However, diamond film etching without affecting the substrate is not possible and substrate
masking will disturb the deposition process. Other methods for measuring the film
thickness-include using optical techniques such as ellipsometry and spectrophotometry.
These measurements are all veiy precise but are neither suitable for rapid evaluation nor for
judgement o f film uniformity. Furthermore, they are not applicable practically for diamond
film measurement due to the rough and non-transparent nature o f the films.
In this work, a simple method was adopted by expressing the growth rate of
diamond films in terms o f either the film thickness or the amount of deposited carbon. The
values obtained from these two expressions are proportional to each other if the film
density remains constant for all the experiments, which is a reasonable assumption for most
films according to the results o f density measurement conducted by Sato and Kamo (1989).
The method o f measuring the growth rat'’ by determining the film thickness was chosen in
this work. It was measured in two ways: 1) by breaking the sample and measuring the film
thickness directly under an optical microscope; and 2 ) by weighing the sample before and
after deposition, then using the following formula to obtain the average growth rate o f film
thickness per unit time:
Growth rate (pm/h) = {(Aw • A)/(w • t • d)} x 104
where Aw (g) is the weight gain after deposition, A (g/cm2) is the specific area weight of
the substrate, w (g) is the weight o f the substrate before deposition, t (hour) is the
deposition time, and d (g/cm3) is the density o f diamond (3.52 is assumed). Growth rate
results from these two methods agree within 12%. One of the sources of this discrepancy is
that the density of the films was assumed to be equal to that o f single crystal diamond.
Diamond films deposited at non-optimum conditions actually contain some graphite and/or
other non-diamond phases, and since such phases have a considerably lower density than
diamond, the films will have densities lower than 3.52. In order to maintain consistency
and to compare the results, the growth rate data presented in Chapter 3 all have been
determined by the first method, i.e., by the optical microscope measurement
2.4.2. Transmission electron microscopy (TEM1 and the sample preparations
TEM is a powerful technique for thin film structure characterization. It has very
high magnifications (normally in the range o f 8000-300,000X) associated with a high
resolution (0.2-0.3 nm using 100 kV acceleration voltage), allowing very detailed studies
of localized regions o f the sample. Selected area diffractions (SAD) allow detailed
correlations between images and electron diffraction patterns. The intensities and
distribution of diffraction spots provide information such as lattice types, lattice constants
and crystallographic orientations. Particularly, TEM is capable of detecting and analyzing
various internal structural defects, such as lattice imperfections o f stacking faults, twins and
dislocations, second phase inclusions, grain boundaries, interfaces, etc., because they
produce variations in the lattice orientation and are, therefore, distinguishable by their
unique image contrasts. In addition, the microscope can be operated in a dark field imaging
mode which is useful for determining the distribution o f regions of a particular orientation
in great detail and, therefore, in differentiating certain crystal types or orientations from the
remainder of the crystals, even when they are not so clearly distinguishable in the image
appearance. For quantitative details o f contrast theories and their applications in analyzing
structural defects, books o f Heidenreich (1964), Loretto and Smallman (1975), Grundy
and Jones (1976), Hirsch et al. (1965), Thomas and Goringe (1979) and Von Heimendahl
(1980) should be consulted. The TEM technique was used effectively in this thesis to
reveal the various structural features o f CVD diamond films which will be presented in
Chapter 5. All the TEM observations in this work were carried out on a Philips 420
microscope operating with an accelerating voltage o f 120kV.
The basic requirement for a TEM sample is that it is thin enough to allow the
electron beam to penetrate through to form images. Usually samples approximately 50-300
nm thick-, depending upon their atomic weights, are transparent to electrons. The
preparation of such thin samples is the most time-consuming part o f using a TEM for
studying film structures. There have been many special apparatuses and techniques
developed for preparation of TEM thin foils. Electrolytic polishing, such as the window
method, the Bollman technique and the jet-polishing method are often used for preparing
metallic materials. Other methods, such as mechanical procedures including forging,
cleaving, cutting with a microtome, grinding and ion milling, and chemical methods using
suitable acids to chemically thin materials, are of great importance for nonmetallic materials.
Replica and powder techniques also have been used.
For the TEM study of natural diamond crystals, diamond samples were thinned by
oxidation with carbon dioxide (CO 2 ) at about 1350°C (Evans, 1965). However, this
method suffered from disadvantages that a surface carbon layer had to be removed by
boiling in an acid mixture, and there was also the possibility that defects present in the
diamond were being thermally annealed at the high thinning temperature. Therefore, most
diamonds were thinned by passing a slow stream o f dry oxygen (O 2) at atmospheric
pressure over diamonds at 750°C. No surface carbon layer was formed, and the danger o f
changing the defect distribution was reduced at this lower thinning temperature.
In the present work, TEM samples of diamond films were prepared by mechanical
polishing and ion milling. The experimental procedures for preparing a diamond TEM planview sample are illustrated in Figure 2.3a. A 3 mm disc sample was ultrasonically cut from
an as-grown film. This sample was then mounted on a specimen lapping holder with
precise scales and mechanically polished to a thickness of 100 (im from the substrate side
using 600 grit sandpaper. Sometimes it was further thinned to 20 Jim with a dimpling
machine to facilitate the subsequent ion thinning. Then it was put into an ion milling
Diamond Film
y 20fj.m) optional
Figure 2.3. TEM sample preparations: (a) plan-view; and (b) cross-section.
D iam ond Film
3 mm
s id e view
( toci S e f
lO O /im
^ 2 0 / i m ) o p tio n al
Figure 2.3. (cont.)
machine for final thinning. The ion miller (manufactured by the Research Institute For
Technical Physics, Budapest, Hungary), has two water cooled argon ion guns which can
be tilted freely to any desired angle with respect to the sample. The argon ion beams were
accelerated with a voltage o f 6 kV and a current of 3mA, bombarding the sample at an angle
o f about 10°. The sputtering rate was about 7 p.m/h for silicon and 0.8 (im/h for diamond
film. All samples were polished and ion milled from the substrate side until perforation
occurred; thus it was the growing surface of the films that was examined by TEM in this
A method for preparation o f TEM samples, without any thinning procedure
required after film deposition, was developed as follows: a 75 Jim thick silicon wafer was
dimpled and ion milled until a tiny hole appeared; the perforated substrate wafer was then
put into the deposition system for 2 hours to obtain a film with a thickness o f 1-2 (im; the
diamond film formed near the edge of the hole was thin enough for TEM observations.
This simple method was used to test for possible artifacts introduced by argon ion
bombardment during TEM sample preparation which will be described in section
Care must be taken when making holes in the substrates to assure that the holes are very
small and have a low angle wedge around them.
Cross-section TEM has proven itself to be very useful for the study o f the evolution
of thin film structures. The essential part of the cross section TEM technique, however, is
the preparation o f cross section samples containing the desired interfacial structure. It was
more difficult and tedious than the preparation o f plan-view samples, consisting o f the
following steps (Figure 2.3b): two pieces (0.5x2x0.5 mm3) were sliced from a bulk
sample with a wire saw; they were bonded together face-to-face; this bonded piece was put
on a 3 mm titanium disc support which had a slot for its insertion; then it was thinned in the
same way as that described above for the plan-view TEM sample preparation, that is,
mechanical polishing and ion beam thinning but from both sides. Because o f large
differences in sputtering rates among the diamond film, silicon substrate and bonding glue,
special precautions must be taken before and during the ion milting process for the crosssection TEM sample to achieve homogeneous sputtering. For example, fine carbon powder
was added to the bonding glue to lower its sputtering rate, and a very small sputtering angle
was used.
2.4.3. Scanning electron microscopy (SEMI
SEM is a technique available to examine thin film surfaces with sub-|j.m size
features. It has many advantages over optical microscopes in its high resolution (2.5-10
nm), its extraordinary depth of field (500 times greater), high magnifications (normally in
the range o f 50-40,000X), and the three dimensional appearance o f pictures. One o f the
differences between SEM and TEM is that the sample for TEM must be so thin that it is
transparent to electrons, while for SEM, solid bulk samples can be used. A conducting
sample can be investigated immediately in its original condition without any preparation.
Only a non-conducting sample has to be pre-coated with a thin layer of conducting
materials (gold or carbon). In this work, free-standing diamond films and diamond films
grown on Si0 2 were coated with gold prior to the SEM observation, while diamond films
grown on Si were directly observed by SEM without any conducting coatings being
deposited on the surface.
It should be mentioned that backscattered electrons with approximately the same
energy as the primary electrons can also be used for imaging the topography. However, the
resolution is worse than with secondary electrons. In this work, all the SEM micrographs
revealing the surface morphology of diamond films were made with secondary electrons
for better resolutions on an ISID S 130 microscope at an acceleration voltage of 10 kV.
2.4.4. Raman spectroscopy
Raman spectroscopy is based on inelastic light scattering arising from the
interactions o f photons with lattice vibrations or phonons. The phonons are coupled to the
photons through the polarization induced in the crystal by the electric field of the intense
light beam. They are related by the polarizability as:
P = aE
------- (2.2)
where P is the induced electric moment, E is the electric field and a is the polarizability. A
lattice vibration is Raman active when the vibration changes the polarizability. Some
photons will be emitted from the oscillating induced dipoles which are either of the same
frequency as the incident beam (elastic Rayleigh scattering) or have been frequency shifted
by an amount equal to the vibrational frequency of the lattice (inelastic Stokes or antiStokes scattering):
P = EotXoCos(2rcvo0 + l/2E02<xn{cos27E(v0 - vn)t + cos27t(v0 + v„)t)
where E 0 is the electric field of incident light, v 0 is the frequency of the incident light, v„ is
the phonon frequency, a n represents the polarizability as a function of the vibration mode
and t is the time. The shifted frequency is the Raman scattering which was named for Sir
C. V. Raman, an Indian Nobel laurate, who discovered the effect in 1928 (Raman and
Krishnan, 1928). The stokes scattering is produced when energy is extracted from the light
beam to the crystal and sets it into vibration, whereas the anti-Stokes scattering arises from
the annihilation of the existing thermally excited vibration.
Raman scattering has been very useful for studying the chemistry and physics of
carbon, particularly of CVD diamond films (Nemanich et al., 1988; Knight and White,
1989). It is very sensitive to the bonding nature o f carbon, therefore, being able to
distinguish various types o f carbon: diam ond, graphite, am orphous carbon or
hydrogenated carbon. The intense first order Raman peak for diamond is located at 1332.5
cm -1 and a second order centered at 2458 cm '1. For highly oriented pyrolytic graphite, the
first and second order peaks are at 1580 cm *1 and 2710 cm*1, respectively. An additional
peak at 1355 cm *1 is observed for microcrystalline graphite. For amorphous carbon, there
are two broad band features around 1550 cm *1 and 1355 cm*1. If the amorphous carbon is
hydrogenated, the broad band at 1355 cm *1 becomes a shoulder o f the 1550 cm *1 band.
Since the Raman scattering efficiency for graphite is much greater than for diamond (Wada
et al., 1980), small amounts o f graphitic carbon co-deposited in CVD diamond films can be
detected readily by Raman spectroscopy. Furthermore, information about impurities,
structural defects and stress state can be obtained, because they cause broadening and shifts
o f the Raman peaks. Therefore, it has been widely used as a probe or indicator of the
quality o f CVD diamond films because of its sensitivity and capability o f rapid evaluation
with non-destruction.
Raman spectra were obtained on an Instruments SA Microfocus Ramanour U1000
spectrometer using the 514.53 nm green line from an argon ion laser with an output power
in the range o f 100-500 mW. The instrument was equipped with a microscope (40X
objective) with a focal laser spot size of a few micrometers. Power at the sample surface
was about 10% of the initial laser power, that is, 10-50 mW. The monochromator slit
width was set at 200 |xm. Raman spectra were recorded directly as a difference spectrum
with zero being the wavenumber o f the incident laser line. This allows the Raman peaks to
be read off directly as wavenumbers o f vibrational modes. The spectrometer was calibrated
using the 546.07 nm mercury emission from the fluorescent room lights.
In this thesis, Raman spectroscopy was used only as a qualitative measure of the
degree of crystallization o f the diamond phase in comparison with the graphite phase and
other non-diamond phases. The Raman peak position was o f primary interest, while the
peak width was not investigated in detail. Also, relative intensities o f Raman peaks were
assumed for interpretations o f Raman spectra. The absolute intensities o f the Raman peaks
were not taken seriously, because they were extremely sensitive to the power level of the
laser beam, its focusing and its polarization changes upon scattering which were sometimes
difficult to control or simply neglected.
2.4.5. Optical emission spectroscopy (OES)
W hen an atom or molecule undergoes a collision with an energetic free electron,
another atom or molecule, its bound electrons will be excited from a lower energy state to a
higher energy state. When these excited electrons return to lower energy levels, they will
emit photons with an energy equal to the difference between the two transition energy
levels. The wavelength o f these emissions is characteristic o f the atom, molecule or species
being excited, since different species correspond to different energy levels. Plasmas are a
rich source o f such em issions. Therefore, OES has been widely used as a plasma
diagnostic technique. W ith this technique, various types o f excited species and their
concentrations can be qualitatively determined without disturbing the plasma. In addition,
electron and gas temperatures (Te and Tg) and plasm a density can be estimated. An
assumption for quantitative interpretations o f plasma emission spectra is that the plasma
exhibits complete or partial local thermodynamic equilibrium (LTE). Also, it is assumed
that the plasma is optically thin so that the emission spectroscopy probes the properties of
the entire plasma. The first assumption (LTE) is generally not true for laboratory cold
plasmas, while the second assumption (optically thin) is reasonably good.
In the present work, OES was employed to study the plasm a chemistry in the .
system o f methane-hydrogen-noble gas mixtures, the results o f which will be presented in
Chapter 4. A schematic measurement set-up is shown in Figure 2.4. The spectra were
measured on an optical multichannel analyzer (OMA) system. The OMA 2 emission
Water Jacket
Fused Silica Tube
Fiber Cable
to Spectrometer
M ultichannel
Silicon Detector
X -Y -Z
Figure 2.4. The experimental set-up for plasma optical emission spectroscopy
spectrometer from EG&G Princeton Applied Research Corp. consists o f a system
processor and program diskettes, a silicon detector and a multichannel detector controller.
Other components include a HR-320 grating monochrometer with 300 grating per
millimeter blazed at 500 nm from Instruments SA, Inc., a waveform monitor (model 608
from Tektronix, Inc.) and optical parts from Oriel Corporation (a glass fiber optic bundle
which is 36 inches long with a 60° acceptance angle, X-Y-Z lens-filter-shutter, and Hg-Ar
and Ne calibration lamps). The spectra were recorded in the visible range from 430 nm to
670 nm with a slit opening of 50 |im. The calibration of the wavelength axis was done with
the mercury-argon lamp and the neon lamp using characteristic known Hg and Ne lines
(McKenna, 1990). This calibration enables a direct assignment o f emission peaks to a
plasma species by comparing the emission wavelengths with published possible transitions
of that species. The intensity axis was not calibrated, because it varied with the sensitivity
of the detector and grating, the transmittance o f the fiber optics, the focusing lens assembly
and the slit width. The optical fiber cable used for collecting optical emissions was placed
in a hole drilled in the microwave applicator. The sampled plasma was in the center o f the
cavity above the substrate surface where both the electron temperature and the plasma
density were the highest (Manory et al., 1988; Inspektor et al., 1989b). No attempt was
made to determine the spatial resolution of the species emissions.
By this technique, species o f atomic hydrogen in the Balmer series, carbon clusters
(C 2 ) and hydrocarbon species (CH) were detected in diamond-forming plasmas. The
variations of the relative concentrations o f the excited plasm a species and plasma
temperatures with different deposition conditions could be qualitatively determined.
However, several important species for diamond growth (CH 3 , C 2H 2 , C 2H) could not be
detected in the visible range o f emissions. Therefore, information from the emission
measurements was limited, and reaction mechanisms could not be determined by the
em ission spectroscopy alone. A combined use o f emission, absorption and mass
spectroscopies is desirable for complete plasma diagnostics.
2.4.6. Thermogravimetric analysis (TGA)
Thermal analysis, including TGA and DTA (differential thermal analysis), is an
effective means for investigations of phase transitions, chemical reactions, adsorption and
desorption at elevated temperatures. TGA relies on any weight changes o f a sample upon
heating to reflect those chemical and physical phenomena. In this thesis, TGA was
performed in an oxidizing atmosphere to investigate the oxidation properties o f CVD
diamond films. The TGA instrument consisted of two units: a TG 716 thermogravimetric
analyzer as shown in Figure 2.5 and a TA 700 thermal analysis console. The TG 716
analyzer consisted of a Cahn RG beam balance (resolution o f change in weight: 2x1 O' 7 g),
a platinum wound furnace with a platinum temperature sensing thermocouple, an alumina
hangdown tube, inert and reactive gas flow meters, the associate electronics, a vacuum
enclosure and a vacuum gauge. The operating temperature range was from 0-1600°C. A
TA 700 console controlled the heating and cooling rates o f the furnace (variable between
0.1 and 75°/min).
For TGA measurements, a hangdown wire was attached to part of the balance
beam. The wire extended into the atmospheric hangdown tube and into the furnace. A
sample bucket was attached to the end o f the hangdown wire. As the temperature of the
sample was raised, its weight change was recorded in grams or milligrams on the X-Y
recorder in the TA 700 console. Control and programming o f TGA was supplied by a
TEM-PRES TA 700 control console. The X-Y recorder indicated the weight change as a
function of temperature.
The reactive gases used in this study were 1 % 02 in A r and dry air. Free-standing
Cohn RG beam balance
Glass cover
Seal ring
Gas outlet
Water jacket
Pt hang down
Al20 3 hang
down tube
Pt sample
Furnace tube
Furnace shell
Seal plate
Reactive gas (O2 ) inlet
Figure 2.5. Schematic diagram of the thermogravimetric analyzer
diamond films were used for the TGA oxidation study. The weight changes as functions of
temperature and oxidation time were recorded. The heating rate was set at 5°/min. The
results will be presented in Chapter 6 .
2.4.7. Other structural and compositional analysis techniques
In addition to the above discussed techniques which w ere used as main
characterization tools in this thesis, there were a few more characterization methods used.
They provided useful and interesting information about the structures and compositions of
CVD diamond films. However, since each of these techniques and the interpretation o f
results constitute a field of specialty by itself, they were used only with great interest, not in
an exploratory manner. Therefore, they will be described briefly. Infrared spectroscopy fIRl
IR is a complementary technique with Raman spectroscopy. The selection rule for
IR active vibration modes is that the vibration produces a finite change in the existing dipole
moment, as opposed to the change of polarizability for Raman active vibration modes. IR is
useful for detecting various carbon phases and impurities in diamond films, since it is very
sensitive to the valence, coordination number, bond distance, strength and angle.
The present middle IR optical spectra (from 400 to 4000 cm '1, or 2.5 to 25 Jim),
both transm ission and reflection, were m easured with a Perkin-Elm er 283B IR
spectrometer (section An IR source condenser attachment (2x2 mm2) was used
for small sample transmission measurement. The polycrystalline nature and rough surface
o f diamond films present a formidable problem for obtaining accurate IR data because of
the severe scattering phenomenon. Various methods have been used to reduce or eliminate
this problem, including polishing and using KRS-5 material (Wang et al., 1990).
66 Laser excited luminescence spectroscopy
Laser-induced luminescence can be viewed simply as absorption followed by
emission. The process involves incident photon excitation o f electrons to higher energy
levels and subsequent radiative decay o f electrons from the excited high energy states to
some lower energy states, resulting in the emission of photons. The frequency o f these
emitted photons is determined by the particular electronic excitation process. There is a
characteristic decay time associated with the excitation and emission, while the Raman
scattering is an instantaneous process.
Photoluminescence o f diamond films in this work was excited by the same laser as
that used to stimulate Raman scattering. The luminescence features can be distinguished
easily from Raman scattering by using two different wavelengths o f incident laser beams
(Hartnett, 1988). The accurate wavenumber o f a luminescence peak measured from the
Raman spectrometer is obtained by subtracting the wavenumber shift which is read off
from the usual Raman spectrum from the incident laser wavenumber.
The luminescence was explored to examine point defects (impurities and vacancies)
in diam ond film s. As w ill be shown in section, there are characteristic
luminescence peaks associated with CVD diamond films, possibly related to various
im purities introduced during deposition, such as silicon and hydrogen. The correct
assignment o f these features to sources requires improved luminescence spectroscopic
techniques such as the low temperature (liquid nitrogen) excitation and a better
understanding o f the relation between deposition and point defect structures of diamond
67 Secondary ion mass spectrometry (SIMS)
SIMS is based on high energy (1-20 keV) ion beam sputtering (usually O', 02+ or
Cs+) o f a material resulting in the ejection o f neutral atoms and molecules along with a
small fraction o f ionized atoms (about 0.1-10%). These secondary ions, either positive or
negative, can be detected and analyzed by a mass spectrometer to provide elemental
identification and chemical concentration analysis. Typical SIMS detection limits range
from 1 ppm to 1 bbm. It can detect hydrogen and helium, although the evaluations may not
be very reliable, and is inherently capable of depth profiling.
In this thesis, SIMS was performed with a Cameca ion microscope to analyze
various contaminants in diamond films (section A n oxygen ion beam (7 keV, 50
pm in diameter) was rastered across an area o f 250 mm2, and signals o f mass spectra were
collected from a 50 mm 2 area within the scanned area to eliminate crator edge effects. Depth
profiles also were measured to 60 nm deep normal into the surface. Neutron activation analysis (NAA)
NAA is a technique capable o f both qualitative identification and quantitative
determination o f multi-elements o f sub-nanogram quantities o f impurities in materials. It is
based on selectively inducing radioactivity in some elements making up the sample and then
selectively measuring the radiations emitted by the radioactive elements. The radioactivation
is usually accomplished by bombardment with thermal neutrons (~0.025 eV) in a low
power nuclear research reactor. The half-life, energy and intensity of the radiation (y-ray)
are measured with a thallium-activated sodium iodide scintillation detector (NaI(Tl)) or a
lithium-drifted germanium semiconductor detector (Ge(Li)) coupled to a multichannel
analyzer. NAA is a mature, non-destructive technique which can detect a wide variety of
elements, but the detection limit varies across the periodic table. Carbon is an excellent low
absorber o f neutrons and thus acts as an excellent host matrix for NAA.
In the present work, NAA was used to determine the concentrations o f argon atoms
in the diamond films (Chapter 4). The reaction that was used for the measurement is as
Ar40 + n° = Ar41 + y (1293.6 keV)
------ (2.4)
with half time tj /2 = 1.83 hour. Work was done in the Penn State Breazeale nuclear reactor.
All the samples were irradiated at a neutron flux o f 1.6 x 10 13 n°/cm2-sec at 1 mW for 15
minutes. The detector was an ORTEC n-type intrinsic Go detector with an efficiency of
0.51% at 1300 keV. The resolution o f the multichannel analyzer was 0.4665 keV/CHNL
for an energy window o f 0 to 1911.66 keV.
Chapter 3
The theme and scope of this chapter concern the growth phenomena of diamond
films under CVD conditions. A systematic investigation o f the diamond growth rate and the
resulting film structure and their correlation with controllable experimental parameters was
conducted. The deposition rate o f diamond films as functions of methane concentration in
hydrogen, substrate temperature, gas pressure, gas flow rate and substrate position relative
to the plasma was determined. The effect o f additive diluent noble gases was also
investigated, the results of which will be presented in Chapter 4. Raman spectra are
presented as a measure of the relative structural changes of the films under different
deposition conditions. In addition, the activation energy o f diamond growth is determined,
and rate-controlling mechanisms are discussed. The goal o f this study was not simply to
establish the optimum sets of conditions for high rate, large area, uniform deposition with
minimum structural defects in the resulting diamond films, but rather to generate data with
which we could begin to understand the more fundamental processes o f the kinetics and
mechanisms of the diamond growth.
3.2. Literature review
In general, the overall growth process can be separated into two steps: a series of
reactions which occur on or near the substrate surface which is controlled by the reaction
rate, and additions o f the resultant adatoms to proper growth sites on the substrate surface
which is controlled by the growth rate. Some researchers have emphasized the importance
o f reaction rate by suggesting the use o f a source gas with sp3 bonding to add on to the sp3
carbon bonds required in diamond (Hirose and Terasawa, 1986), while others have
emphasized the role of growth rate, indicating that the low activation energy path to growth
of diamond depends on the initial diamond nuclei (Spitsyn et al., 1981; Deijaguin and
Fedoseev, 1973).
The post-nucleation growth o f diamond is possible only when there is a
supersaturated partial pressure of carbon in the gas phase and sufficient diamond growth
species in the plasma near the surface, and the diamond phase is stabilized. These
conditions are met by the presence of a super-equilibrium amount o f atomic hydrogen.
Atomic hydrogen generates critical hydrocarbon species and selectively gasifies graphitic
carbon, thereby permitting near-exclusive diamond growth. Setaka (1984) offered an
overall picture of diamond nucleation and growth from a plasma o f methane-hydrogen
gaseous mixture. Methane and hydrogen are dissociated in the plasma, and chemically
active hydrocarbon ions and radicals, as well as atomic hydrogen, are generated. The
hydrocarbon species diffuse and are adsorbed on the substrate. At the earliest stage, carbon
clusters are formed on the substrate surface as a result of chemical reactions releasing
hydrogen. The majority of the clusters have thermodynamically stable graphitic and
amorphous structures, but the metastable diamond structure also is formed due to
thermodynamic fluctuations. Since the chemical reaction rate of graphite with atomic
hydrogen is faster than that of diamond, graphitic carbon clusters are removed rapidly from
the surface, and only clusters with the diamond structure stay and grow. Furthermore,
atomic hydrogen attacks unsaturated sp 1 and sp 2 bonds to convert them into sp3 bonds.
Thus, atomic hydrogen facilitates the formation of sp3 diamond clusters and suppresses
other forms o f clusters with unsaturated bonds. It has been supposed that chemically active
species such as methyl radicals or acetylene arriving at the surface release hydrogen to form
covalent C-C bonds with carbon atoms of the clusters (Tsuda et al., 1986; Frenklach and
Spear, 1988). Thus, the surface o f CVD diamond is terminated by C-H bonds.
In 1982, Matsumoto et al. (1982b) reported that the growth rates o f diamond and
graphite in a HFCVD system depended on the rates of three reactions: the formation of C-C
bonds (sp 3 vs. sp2) by the decomposition o f hydrocarbons on substrate surfaces and
subsequent deposition o f active species, the transformation o f diamond into graphite in a
hydrogen atmosphere, and the reaction of atomic hydrogen with graphite. These three rates
were affected by methane concentrations, active hydrocarbon radicals and atomic hydrogen
as well as by substrate temperature and gas pressure. The substrate temperature controlled
the thermal stability and reactivity of the growing surface and the supersaturation of gas
phase near the surface. Later, a comprehensive investigation of the parameters o f the
HFCVD process was conducted by Singh et al. (1988). The growth rate was found to be a
function of gas pressure and flow rate, substrate material and temperature, filament material
and temperature, filament-substrate distance, and reactor wall material.
Badzian and Badzian (1988) outlined two ways to increase the growth rate. One
method is to intensify the plasma, which leads to a higher concentration o f active growth
species. This can be achieved by manipulating the exciting power and the corresponding
pressure and temperature o f the plasmas. This method has been proved to be extremely
effective by the high growth rate achieved in a series o f newly developed techniques
employing high density thermal plasmas such as atmospheric RF plasma (Matsumoto et al.,
1987), DC plasma jet (Kurihara et al., 1988) and combustion flame (Hirose and Kondo,
1988). The other approach is through the introduction o f a catalytic agent into the plasmas.
The presence o f some specific foreign atoms or species in the growth environment
participating in surface reactions increases the growth rate. These catalysts are active in the
growth process (unlike the catalyst in a conventional catalytic chemical reaction), and some
fractions of them (0.1% or less) are incorporated into the diamond lattice. The atoms which
have been found to exhibit catalydc functions in the diamond growth process include boron
(Poferl et al., 1973), silicon (Badzian et al., 1988a) and titanium (Badzian and Badzian,
The growth rate can be increased by inducing electron bombardment o f the growing
surface to accelerate the decomposition o f hydrocarbons and hydrogen and increase the
nucleation sites on the substrate surface (Sawabe and Inuzuka, 1985,1986). The growth
rate also can be enhanced by using oxygen containing organic compounds such as
CH 3OH, C 2H 5 OH, CH 3COCH 3 , and C2H 5 OC 2H 5 (Hirose and Terasawa, 1986). The
growth rate was ten or several tens times faster than the CVD method using hydrocarbons
such as CH 4 and C2 H 2 . The effect was attributed to the favorable sp 3 type bonding
arrangements in the precursor gases. The relatively easy generation of methyl radicals
(CH 3) from the organic compounds was thought to be responsible for the high growth rate.
However, Chang et al. (1988) indicated that the favorable results should be mainly
attributable to the presence of oxygen in these compounds, and the H/C/O atomic ratios in
the supplying gases are more important than the kind o f hydrocarbons. Later, the beneficial
roles o f oxygen were explored by a number of researchers. Small additions o f oxygen
(Kawato and Kondo, 1987; Chang et al., 1988) or water vapor (Saito et al., 1988) have
been reported to increase the growth rate several times. However, if the concentration o f
oxygen was too high in the gas phase (O/C >0.5), the growth rate decreased due to the
strong etching effect of oxygen with carbon phases (Inspektor et al., 1989a).
The substrate temperature is perhaps one o f the most important variables in
determining the film growth rate, since it closely relates to many growth phenomena such
as the supersaturation o f the gas phase on or near the surface and the mobility and residence
time o f adatoms. For MPECVD, the growth rate is not appreciable until the substrate
temperature is around 800-900°C, although the diamond growth at temperatures as low as
350°C was found possible (Liou et al., 1989). As the temperature is increased further, the
growth rate increases and reaches a maximum at about 1000°C (Spitsyn and Bouilov,
1988). Further increases in temperature result in a reduction o f the growth rate. There are
different activation energies associated with different deposition processes. Early studies on
methane pyrolysis for diamond growth on diamond seed crystals yielded an activation
energy o f 60 kcal/mol (Deijaguin et al., 1973). Thermal decomposition of methane and
hydrogen gaseous mixture had an activation energy of 55 kcal/mol (Chauhan et al., 1976).
A much lower activation energy, 25 kcal/mol, was associated with an electrical discharge
activated chemical transport reaction (CTR) processes (Spitsyn et al., 1981). As will be
presented below, the present MPECVD system produces diamond films with activation
energies o f 20-30 kcal/mol.
3.3. Experimental approach
As a first step toward understanding the basic growth kinetics, the film deposition
rate and structure were measured for a range of critical deposition parameters including
methane concentration in hydrogen (CH 4%), substrate temperature (Ts), total gas flow rate
(V), total gas pressure (P) and substrate position (D). Single crystal silicon wafers with
{100} orientation were used as substrates. The surface of the substrate was scratched with
1 Jim diamond paste before deposition to enhance the nucleation density.
The substrate temperature was varied from 850°C to 1150°C, as measured by an
optical pyrometer, by changing the microwave power. The low temperature deposition
(400-700°C) which was recently found possible (Liou et al., 1989) was not investigated
here. The methane concentration in hydrogen was in the range of 0.5% to 5%. At methane
concentrations higher than 5%, deposits on the wall o f the reaction tube would occur
which, in turn, would have a significant influence on the normal film deposition onto the
substrate. The pressure was varied from 50 to 150 Torr. W hen the pressure was lower than
50 Torr, the plasma expanded to the point of touching the tube wall, leading to deposits on
it. Tube etching could also occur. On the other hand, when the pressure was higher than
150 Torr, the plasma would shrink to a very small volume, leading to non-uniform films
over a small area on the substrate. The total gas flow rate was varied from 0 to 400
standard cubic centimeters per minute (seem). For the zero flow rate experiment, the gas
inlet valves were closed once the pressure in the reaction tube reached the designated value.
A t the other extreme, gas flow rates higher than 400 seem caused large pressure
fluctuations. The substrate was positioned 1.5 cm below the geometric center o f the
microwave cavity for all experiments except for the investigation of the substrate position
effect. The substrate position relative to the plasma was measured by the distance between
the substrate and the center of the cavity. Thus, 0 cm means that the substrate was in the
very center o f the cavity, and 3 or 4 cm indicates that the substrate was actually
dow nstream out o f the luminous plasma. A t these rem ote positions, an external
molybdenum heater was used to achieve the desired substrate temperature, since the heating
by the microwave energy and the plasma was not as efficient as immersing the substrate in
the plasma.
Each o f the process parameters was investigated individually by keeping other
parameters constant. However, the substrate temperature, which was controlled effectively
by the microwave power input, also was affected strongly by the methane concentration in
the gas phase, the gas pressure and the substrate position. The gas flow rate had no
substantial influence on the substrate temperature. It w as thus difficult to carry out
experiments to study effects o f the methane concentration, the gas pressure, and the
substrate position w ithout affecting the substrate temperature o r the pow er input.
Therefore, one had to decide what should be held constant when there was a conflict in
holding both the substrate temperature and the power input constant. It was reasonable to
assume that the substrate temperature was very critical for diamond growth, since it
determined the surface activity o f plasma species and the reaction kinetics for the diamond
nucleation and growth. Therefore, the highest priority was to ensure that a constant
substrate temperature was maintained during the experiments rather than a constant power
The film growth rate was obtained by measuring the thickness of a continuous film
under an optical microscope as described in section 2.4.1. The growth rate measured on an
individual crystal normally would be 3-5 times that o f a continuous film (Anthony, 1988).
It should be pointed out that the nucleation was considered as the initial stage of growth
here, since both nucleation and growth frequently occurred simultaneously during the film
formation, particularly for the deposition o f polycrystalline diamond films where secondary
nucleation constantly occurred. Thus, the measured deposition rate, which is termed
indiscriminately as growth rate throughout the context, actually was an average value over
the deposition time (usually about 8 hours) and included the contribution from the
nucleation rate, although it was minor due to the short nucleation period relative to the
whole deposition time. The reproducibility of the growth rate data was good as long as the
experiments were performed sequentially without interruption. Many factors, like the
microwave power stability, the repeatability of the growth environment (silica reaction
tube) and the consistency of the substrate surface treatment, could cause errors in the
growth rate results. Although these factors inevitably were present to some extent, efforts
to minimize their influence were essential and were taken.
3.4. Effect o f methane concentrations in hydrogen (CHa%)
The dependence of film growth rate on methane concentrations is shown in Figure
3.1. As might be expected, the growth rate increases with the methane concentration over
the range investigated apparently because o f an increasing amount of carbonic reactant
species in the plasma and an associated increasing degree o f supersaturation (Inspektor et
al., 1989a). The curve seems to level off or even possibly to decrease at high methane
concentrations (>5%) as recent results show (McKenna, 1990). The reason is that loosely
bonded non-diamond carbon films are produced at high methane concentrations. These
deposits exhibit poor adhesion to the substrates and are easily swept off the surface in the
open reactor. Therefore, although the amount o f solid carbon deposited from the gas phase
is expected to be high at high methane concentrations, the actual deposits adhered to the
substrate surface were reduced. The Raman spectra in Figure 3.2 indicate that the films
deposited at high methane concentrations contain an increasing amount of non-diamond
phases, which is consistent with the above analysis. The morphology of the films was
found to change from triangular faces prevailing to square faces dominating with an
increase of the methane concentration from 0.5% to 2%. Further increases in the methane
concentration to 5% result in a polycrystalline film with sub-ftm size grains (Zhu et al.,
1990). This morphology behavior at different methane concentrations is consistent with the
results o f Kobashi et al. (1988). It will be shown later in Chapter 5 by transmission
electron microscopy (TEM) that diamond films deposited at higher methane concentrations
also have a higher density of planar defects.
960°C, 100 seem, 90 Torr
Figure 3.1. Film growth rate as a function of methane concentrations in hydrogen
Figure 3.2. Raman spectra o f diamond films deposited at different methane
concentrations: (a) 0.5%; (b) 2%; and (c) 5%. The peaks at ~520 cm *1
result from the silicon substrate.
3.5. Effect of substrate temperatures HY)
3.5.1. Film growth rate as a function o f substrate temperatures (TV)
As a function o f substrate temperature (also microwave power), the growth rate
reaches a maximum at a temperature o f about 975°C for both 1% and 5% methane
concentrations as shown in Figure 3.3. Growth rates for 0.5% and 2% methane
concentrations also were obtained, and the results were similar. The maximum growth rates
achieved in this particular system were 0.88 |im /h for 1% methane concentration and 1.90
(im/h for 5% methane concentration. Outside the temperature range o f 950-1000°C, the
growth rate dropped o ff very rapidly. The relationship between the growth rate and
substrate temperature was similar to the work by Deijaguin et al. (1986) for homoepitaxial
diamond films deposited by the chemical transport reaction (CTR) method. This kind of
growth behavior at different substrate temperatures also has been reported for other
materials, such as aluminium nitride (AIN) and silicon carbide (SiC) grown by low
pressure CVD (Suzuki and Tanji, 1987; Sibiende and Benezech, 1988). An excellent
general discussion on this CVD behavior was given by Spear (1984).
The Raman spectra (Figure 3.4) o f diamond films deposited at a low methane
concentration (0.5%) show that the broad, non-diamond Raman peak at -1500-1600 cm ' 1
as well as the luminescence background decrease with increasing temperature, while the
diamond Raman peak near 1333 cm -1 increases relatively. This qualitatively indicates an
increasing degree o f structural perfection o f diamond with an increase of substrate
temperature. The film morphology was found to have diverse characteristics in this
temperature range (Zhu et al., 1990).
RATE) (A /h)
I00sccm,90 Torr
5% CH4
Es 30.6(kcal/mol) _
1% CH4
Es 20.4(kcal/mol)
Figure 3.3. An Arrhenius plot of growth rate vs. reciprocal substrate temperature
Figure 3.4. Raman spectra o f diamond films deposited at different substrate
temperatures: (a) 890°C; (b) 950°C; and (c) 1035°C.
3.5.2. Activation energies and growth rate-controlling mechanisms
A s shown in Figure 3.3, the diamond film deposition is a highly selective process
which has a narrow temperature range for the maximum nucleation and growth of
essentially single phase, micrometer-size, polycrystalline diamond films, and the nature of
the rate-controlling mechanism changes with substrate temperature. At temperatures below
approxim ately 1000°C, the reactivity o f the species in the plasma is relatively low
compared with that at high temperatures above 1000°C. The rate-controlling step is
probably a reaction process on the surface or in the plasma, which can be expressed by the
Arrhenius equation:
K = A exp(-E/RT)
where K is the growth rate, A is a constant, E is the activation energy, R is the gas
constant, and T is the absolute temperature (Latham and Burgess, 1977). The activation
energy which is characteristic of the reaction process mechanism is calculated from the
slope o f the lines in the temperature range of 850-1000°C by the method o f least squares,
as shown in Figure 3.3. The obtained values are 20.4 (kcal/mole) at 1% methane
concentration and 30.6 (kcal/mole) at 5% methane concentration. The values obtained in
this system at other methane concentrations are 21.9 (kcal/mole) for 0.5% methane
concentration and 29.8 (kcal/mole) for 2% methane concentration For comparison, Table
3.1 lists the values o f activation energy reported in the literature for various diamond
deposition processes as well as the values obtained in this study under different methane
concentration conditions. It is likely that the activation energy for the diamond growth is
closely related to the presence of atomic hydrogen in the process, and that the higher the
amount o f atomic hydrogen, the low er the activation energy. However, it should be
emphasized that the activation energy obtained here is believed to be an overall value
applied to a complex reaction process which is a combination o f many elementary reactions
Table 3.1. Activation energies (E) for various deposition processes
E (kcal/mol)
pyrolysis o f CH 4
Deijaguin et al.
atomic hydrogen
thermal decomposition
Chauhan et al.
o f CH4 +H 2
activated CTR
Spitsyn et al. (1981)
this thesis
occurring on the surface and in the plasma. Further complications in dealing with this
activation energy arise when one considers that the substrate temperature is actually
controlled by the microwave pow er input in our study. The substrate param eter
(temperature) is thus closely coupled with the plasma parameters such as the electron
temperature, the degree of molecular dissociation or excitation, the recombination rate of
species, etc., and variations of the substrate temperature would induce these changes in the
plasma. Therefore, attempts at accurate assignment o f the activation energy to a specific
reaction process are extremely difficult and risky because of the poor understanding o f the
deposition process. Diamond deposition downstream o f the plasma decouples the plasma
and substrate parameters, facilitating such systematic parametric studies (Pickrell et al.,
A t temperatures higher than 1000°C, thermal desorption o f both hydrogen and
hydrocarbon species at the growing film surface plays an active role, which may cause the
growth rate to decrease. Another possible reason accounting for the low growth rate at high
temperatures is that the homogeneous nucleation o f diamond in the gas phase may occur,
so that the nucleation and growth on the substrate surface is suppressed. The homogeneous
nucleation o f diamond in the gas phase has been studied by Fedoseev et al. (1984) in a
process o f laser heating liquid hydrocarbons droplets and by Mitura (1987) in a high power
RF methane plasma. However, there is no direct demonstration o f this phenomenon in our
experim ents. A t extrem ely high temperatures (above 1100°C), deposits would be
essentially composed o f graphitic phases.
In the temperature range o f maximum growth around 975°C, many peculiar and
important phenomena actively take place during the deposition process, such as adsorption
and desorption o f hydrogen on the surface of diamond, reconstruction o f the diamond
surface, etc. (Badzian and DeVries, 1988), which are unlikely simply to be coincidental
with the maxim um growth rate in this temperature range. Because low substrate
temperatures (<900°C) produce disordered structures, and high substrate temperatures
(>1000°C) introduce graphite in the films (section 5.2), it is believed that polycrystalline
diamond films deposited in this intermediate temperature range (950-1000°C) have a
minimal amount o f defects. A convincing explanation o f this growth behavior at different
temperatures can only be approached from thermodynamic analysis and vigorous kinetic
modeling o f the deposition process but is not yet available. Kinetic models proposed by
Fedoseev et al. (1984) supply some insights into the deposition process. Thermodynamic
analysis (Piekarczyk et al., 1989; Sommer et al., 1989) yields some useful information,
despite the limitations resulting from the facts that diamond is a thermodynamically
metastable phase under the CVD conditions and that the deposition technique is not an
equilibrium process. Mechanisms of the diamond formation are being pursued (Tsuda et
al., 1986; Frenklach and Spear, 1988). An extremely helpful experiment was conducted by
Vakil et al. (1989) to elucidate the growth mechanisms. They carried out an investigation of
the effect o f isotopic substitution of hydrogen by deuterium to determine whether the ratecontrolling step involves a C-H o r C-C bond. Their results strongly suggested that the rate
limiting step for the diamond growth involves the C-H bond.
3.6. Effect o f gas pressures (PI
The effect o f pressure on the growth rate is shown in Figure 3.5. A maximum was
found at 110 Torr. This kind o f parabolic pressure effect also has been seen in other
deposition processes such as RF plasma assisted deposition o f pyrolytic carbon (Inspektor
et al., 1986) and silicon carbide (Katz et al., 1980). As indicated by Raman spectra in
Figure 3.6, the structure o f diamond films is not so sensitive to the pressure as to other
parameters. The same insensitiveness was observed for the morphology o f diamond films
(Zhu et al., 1990).
Pressure in the deposition tube plays an important role in the plasma formation and
diamond growth. It is obvious that more reactant species are introduced in the higher
pressure plasma than in the low er pressure plasma. However, the electron temperature
decreases with increasing pressure because o f the decreasing mean free path (Thornton,
1982). Since electrons contribute activation energy for the dissociation o f molecules, the
degree o f excitation in the plasm a is expected to decrease with increasing pressure at
constant power in p u t The majority species in the high pressure plasma are undissociated or
unexcited molecules which are not the principal species responsible for the diamond growth
1% CH4 ,975oC ,l00sccm
20 4 0
60 80 100 120 140 160 180
Figure 3.5. Film growth rate as a function of total gas pressures
50 Torr
Figure 3.6. Raman spectra of diamond films deposited at different gas pressures:
(a) 50 Torr, (b) 110 Torr; and (c) 150 Torr.
according to the present proposed mechanistic models (Tsuda et al., 1986; Frenklach and
Spear, 1988). The fact that less microwave power was used at high pressures to maintain
the same substrate temperature also may contribute to the low degree of dissociation and
excitation o f the high pressure plasma. Thus, the decrease in the amount o f dissociated or
excited species with increasing pressure can be a significant factor in the reduction o f
deposition rate. The growth also could be limited by the mass transport under the high
pressure condition, which could be a second factor contributing to the growth behavior at
different pressures. As already indicated in section 3.3.1, the lower pressure plasma
occupies a larger volumetric space than the higher pressure plasma. Therefore, more
uniform films can be produced on a larger substrate area under a lower pressure condition
but at the sacrifice o f the growth rate.
3.7. Effect of gas flow rates (F)
Figure 3.7 presents the dependence of growth rate on the total gas flow rate. In
order to provide a normalization factor, the scale at the top of the figure gives the average
linear gas flow velocity in the reaction tube at room temperature. The linear gas flow
velocity is simply the input gas flow rate divided by the cross-sectional area o f the tube,
correcting for the factors o f temperature and pressure in the tube. It should be pointed out
that the linear gas flow velocity must be taken only as an approximate value, since it is
derived by neglecting the interference o f the plasma and the geometry effect o f the sample
holder. In addition, the actual velocity is not uniform in the tube, because temperature
gradients exist. It is seen that the growth rate decreases to a steady state value with an
increase o f gas flow rate. The maximum growth rate was found at zero flow rate, i.e., in a
closed system. The Raman spectra in Figure 3.8 indicate that the quality of diamond films
improves with increasing gas flow rate, i.e., the non-diamond peak is reduced. The strong
l% CH4 ,9 7 5 °C t 90T orr
RATE (/im /h )
Figure 3.7. Film growth rate as a function o f total gas flow rates
400 seem
50 seem
12.5 socm
Figure 3.8. Raman spectra o f diamond films deposited at different gas flow rates
(a) 0 seem; (b) 12.5 seem; (c) 50 seem; and (d) 400 seem.
spectral feature seen at ~1140 cm-1 for the film deposited under the zero gas flow rate
condition-, which also can be observed in the films deposited at the high methane
concentration o f 5% (Figure 3.2) and at the low substrate temperature of 890°C (Figure
3.4), has been related previously to the presence o f small crystallite size diamond (<0.1 ^m)
(Nemanich et al., 1988). The morphology o f diamond films shows a distinctive change
with the gas flow rate (Zhu et al., 1990).
There are two types o f limitations on the growth rate of films associated with the
gas flow rate. The first is the rate o f introduction o f reactant species into the tube. The
second is the transport o f these species into the substrate surface, which is dependent upon
geometry. It is this second phenomenon that appears to limit the growth rate o f diamond
films. It can be shown easily with simple calculations that the efficiency o f deposition,
which is the ratio o f the amount of deposited carbon to that of introduced carbon, is very
low under typical deposition conditions. For example, the deposition efficiency is only
about 0.2% at a temperature o f 1000°C, a total gas flow rate of 100 seem, a methane
concentration o f 1% in hydrogen, and a growth rate o f lfim/h. This implies that the
reaction rate for diamond formation is rather low. Because of the low deposition rate, the
residency time of reactant species on the substrate surface and/or in the plasma is critical to
the film growth. A t low gas flow rates, the introduced reactant species have sufficient
residency time to come into equilibrium with the substrate surface. Consequently, the
growth rate will be high as long as there is a sufficient number o f reactant species to be
depleted by the deposition process, which was indeed true in our experiments. When the
flow rate is increased, the residency time o f the reactant species in the plasma and on the
substrate surface becomes short in comparison to the time needed for mass transport in the
gas phase or surface reactions, which results in the decrease of the growth rate. Since not
all o f the introduced reactant species are able to participate in the reaction process under the
condition o f high gas flow rates, the deposition efficiency should decrease also.
Furthermore, in our experiments, the surface of the substrate was perpendicular to the gas
flow direction. It is evident that the high gas flow may disturb the species distribution in the
boundary plasm a layer (Venugopalan and Avni, 1985) in the vicinity o f the substrate
surface and sweep away those species which are vital for diamond growth. Therefore, it is
expected that the growth rate will be reduced. At flow rates higher than 100 seem, a steady
state is gradually reached in which the growth rate is not dependent on the input gas flow
rate. The effect o f high input rate of reactant species may overwhelm the effect o f the short
residency time o f these species on the diamond formation under this condition.
The actual plasma structure and concentration profile inside the tube were very
complicated, especially at low gas flow rates. No attempts have been made to predict them.
High gas flow rates could promote the degree of mixing of the gases in the tube and result
in more uniform films to be deposited. This has been experimentally verified. The
deterioration o f the film structure at low flow rates may be caused by local variations of
carbon supersaturation which may arise from the fluid dynamics o f the process as well as
the large amount o f species from the plasma etching o f the graphite susceptor.
3.8. Effect of substrate positions relative to the plasma (D)
The substrate position was changed from the geometric center of the microwave
cavity (D = 0 cm) to 4 cm downstream o f the plasma in this study, as schematically shown
in Figure 3.9. When the substrate was placed in the center of the microwave cavity (0 cm),
part o f the plasma resided above the substrate surface while the remainder appeared below
the substrate. Lowering the substrate 1 or 2 cm below caused more o f the plasma to reside
above the substrate. However, further lowering the substrate to 3 or 4 cm below resulted in
the substrate being outside o f the luminous plasma region.
6.0 cm
1 _________
6.0 cm
(P la sm a -
Mo HEATER-----
Figure 3.9. Schematic diagrams of the substrate positions relative to the plasma:
(a) the substrate is immersed in the plasma; and (b) the substrate is located
at a remote downstream position where an external heater is used.
As seen in Figure 3.10, a maximum growth rate was found at a position about 1-2
cm below, the center o f the cavity. Since the interaction between the plasma and the
substrate surface occurs most efficiently which results in a maximum substrate temperature
when the substrate is positioned 1 or 2 cm below the center o f the microwave cavity, a
corresponding maximum growth rate is not surprising. The growth rate drops as the
distance o f the substrate from the plasma increases due to a decay o f available active and
critical species to the diamond growth (atomic hydrogen and hydrocarbon radicals and/or
ions). However, the Raman spectra in Figure 3.11 show that the films with lower growth
rates have better quality (i.e., less non-diamond features). This also has been confirmed for
films deposited on Mo substrates. The improvement of the structural quality of diamond
films by depositing downstream of the plasma is probably due to a reduction o f severe
bombardment o f films by various species in the plasma. The film morphology does not
show marked changes with the substrate position (Zhu et al., 1990). As reported by
Pickrell et al. (1990a), the diamond deposition below the plasma appears to offer several
merits over immersion o f substrates within the plasma, including a reduction in substrate
etching and an improvement in film uniformity, adhesion and transparency for coating
certain oxide substrates. Another important advantage is that the plasma and substrate
parameters can be effectively decoupled, allowing them to be systematically studied.
3.9. Discussion
Most of the important deposition parameters for MPECVD o f diamond films have
been examined for their effects on the growth rate o f diamond film s from CH 4 +H 2
plasmas. Other important parameters which were effectively held constant include the
system geometry, the gas composition (CH 4 +H2 ), the substrate and its preparation.
Specific discussions on the effect o f each parameter have been conducted above
GROWTH RATE (//.m/h)
■ 0.5% C H 4,990°C ,
100 seem, 90 Torr
0.1 ~
Figure 3.10. Film growth rate as a function o f substrate positions relative to the plasma
4 cm
1.5 cm
Figure 3.11. Raman spectra o f diamond films deposited at different substrate positions
relative to the plasma: (a) 0 cm; (b) 1.5 cm; (c) 3 cm; and (d) 4 cm.
following the experimental results. It is seen that optimum conditions for growing diamond
films with the best quality, i.e., maximum growth rates, minimum structural defects and
large area uniform coverages, can only be obtained by carefully considering and combining
each o f these parameters. Compromises in choosing deposition parameters to meet the
above requirements for optimal diamond film deposition are necessary in most cases. For
example, large area uniform films can be obtained at a low pressure, but their growth rates
decrease. The growth rate can be increased by depositing films at high methane
concentrations, but the structure o f these films becomes more defective, and the graphite
second phase is introduced. Furthermore, these investigated parameters o f methane
concentration, substrate temperature or microwave power, gas flow rate, pressure and
substrate position are dependent upon each other and closely interconnected. The results of
this study by systematically varying each individual parameter make possible some useful
predictions for the growth behavior o f diamond films, provide experimental insight o f the
fundamental kinetic and mechanistic processes, and may eventually help in designing
experiments for particular applications of diamond films.
It should be pointed out that, although the results from this parametric study have
significance in developing an understanding o f the nature o f the MPECVD diamond
process in general, the deposition conditions described in this chapter are only applicable
for this particular MPECVD tubular system and the CH 4 +H 2 gas phase chemistry.
Different equipment designs and gas compositions will certainly yield specific, different,
optimum conditions for the diamond growth. A more profound fundamental understanding
o f deposition processes (e.g., type and energy o f depositing species, nucleation and
growth phenom ena, and surface kinetic and mechanistic processes) is needed for
maintaining reproducibility of experimental results and allowing the transfer o f a process
from one system to another.
3.10. Summary
A systematic investigation in the effects o f deposition parameters on the growth rate
and structure of MPECVD diamond films was conducted. Optimum conditions for growing
diamond films with maximum growth rates, minimum structural defects and uniform
coverage were obtained for this particular system. A maximum in the growth rate was
found at a substrate temperature around 975°C and a pressure about 110 Torr for methane
concentrations from 0.5% to 5% in hydrogen. The growth rate decreases to a steady state
value with an increase o f the total gas flow rate. Diamond films with minimum structural
defects can be obtained in the temperature range of 950-1000°C and low methane
concentrations such as 0.5-1%. Uniform films with large area coverage can be achieved at
low pressures (around 50 Torr, for example) and high gas flow rates such as 400 seem.
The substrate position relative to the plasma is also an important factor influencing the
growth and structure of diamond films. The activation energies for the MPECVD diamond
film growth are 20-30 (kcal/mol) in the substrate temperature range o f 850-1000°C. The
results provide experimental background for kinetic and mechanistic modeling for CVD
diamond growth processes.
Chapter 4
4.1. Introduction
The purpose of this chapter is to present the bulk plasma chemistry of gaseous
mixtures o f methane, hydrogen and noble gases as a further step toward understanding the
deposition-structure relations. In addition, a comprehensive knowledge o f the plasm a
chemistry (species, reaction kinetics, diffusion and transport, etc.) is one of the keys to
develop a mechanistic model o f the growth of diamond.
In this work, noble gases were used as agents for characterizing various plasma
species, for controlling their relative changes of concentrations under different process
conditions, and for understanding their effects on the growth rate and structure of diamond
films. Plasma diagnostic results by emission spectroscopy are compared with the growth
rate and Raman spectra of the films. Noble gases, which enhance the degree of excitation
of hydrogen and hydrocarbon molecules by energy transfer or charge transfer from their
highly excited and ionic states, are active in the deposition process by inducing additional
ion-molecule and excited atom-molecule reactions. As a result, the film deposition rate was
increased as was demonstrated in systems of CH4 +Ar+H 2 . It was also found that small
oxygen additions along with the noble gases can greatly suppress the formation o f non­
diamond carbon phases, leading to an effective way to rapidly deposit diamond films at
high m ethane concentrations while still retaining minimum non-diamond carbon
components in the films.
4.2. Literature review
4.2.1. Plasma in general
A plasm a is a quasineutral gas o f charged and neutral particles which exhibits
collective behavior. The plasma state can be considered the fourth state o f matter, and
although we are accustomed to solids, liquids and gases, most o f the materials in the
universe is in the plasma state. In laboratories, plasmas are usually generated and sustained
by electrical energy. Commonly in all laboratory plasmas, energy transfer from the electric
Held to electrons is significantly greater than to ions and neutrals because of the low mass
o f electrons. A characteristic feature o f these cool plasmas is, therefore, that the Boltzmann
temperature o f ions and neutrals is roughly ambient, while that o f electrons is several
orders o f magnitude greater, typically at lOMO 5 K. A plasma can thus be characterized in
terms of the average electron temperature and the charge density within the system.
Formation of a plasma is initiated by a few ever-present free electrons from cosmic
radiation in an ordinary gas. These electrons are accelerated by the applied electric field.
They collide with molecules, and occasionally impart enough energy to ionize these
molecules. Ionization, in turn, produces more electrons, so the process avalanches until a
steady plasma state is reached. The plasma is quite conductive, and electrical energy can be
continuously transmitted to the molecules by the impact o f accelerating electrons. The
number of electrons and the distribution of electron energy are o f primary importance in
determining reaction rates. The exact distribution is not known, but it can be crudely
conceptualized in terms o f a Maxwellian distribution where the mean electron energy is a
few electron volts. This varies with the composition o f the plasma, the pressure and the
applied power. In general, low pressure and/or high pow er will give more energetic
A plasma is a unique reaction medium in which reactants and intermediates find
themselves in a sea o f fast moving electrons. A plasma has two basic functions in
influencing chemical reactions. One is that high energy electrons collide with neutral gas
molecules, break chemical bonds, excite and activate the working gas and so initiate
chemical reactions at or near room temperature. The other is that plasma ions are accelerated
to the substrate and bombard the growing film which, in turn, significantly influence the
film growth, its composition, microstructure and stress.
The microwave plasma is excited in consequence o f an absorption o f microwave
energy in an ionized gas through three basic absorption mechanisms: collisional absorption,
collisionless absorption, and non-linear absorption (Musil, 1986). The properties o f a
microwave plasma significantly depend on the ratio o f v/co, where v (collision frequency) =
v ei + Ven (Vei for collisions between electrons and ions, ven for collisions between electrons
and neutrals) and co is the frequency o f the microwave radiation. Collisions between
electrons and neutrals prevail at high pressures and collisions between electrons and ions at
low pressures. The magnitude o f v/to determines the electron distribution function and the
degree o f activation of plasma species. Microwave plasmas differ significantly from RF
plasmas in their higher electron density (Ne) and temperature (Te) as discussed previously
in section 2.3.1. This results in an efficient generation of active species and higher radical
concentrations, thus exerting considerably larger influence over the surface reactions.
4.2.2. Spectroscopic techniques and identification o f plasma species
Spectroscopic techniques for plasma diagnostics represent a critical starting point
for plasma chemistry study. Techniques which are in-situ, non-intrusive (non-perturbing),
species and quantum level specific, and with excellent spatial and temporal resolutions are
very useful for providing information about the identity, temperature, concentration and
distribution o f active species in plasmas. Optical diagnostic techniques such as emission,
absorption, laser induced fluorescence, Raman scattering and coherent anti-stokes Raman
scattering (CARS) are ideal for these purposes. Emission and absorption spectroscopies are
briefly examined below, together with mass spectroscopy.
Emission spectroscopy relies on detection o f emission from plasma species in
excited electronic states. It is simple and requires only a single port access. It can provide
valuable information on the presence o f excited species. But this technique suffers from
numerous disadvantages. The spatial resolution is poor. Only the excited electronic state
can be seen, and this may not be proportional to the ground state density o f species,
particularly in non-LTE (local thermodynamic equilibrium) situations. Thus, quantification
is difficult. Also it is limited to certain species, since there are a number o f species that may
photodissociate or may be heavily quenched or may emit in a region where they simply can
not be observed by the spectrometer. In some cases, the concentrations o f ground state
species can be inferred from emission measurements with the use o f actinometry, which
involves the use o f emission intensity ratios to provide an estimate o f the concentrations o f
ground state species.
Absorption spectroscopy is a relatively simple and direct technique for the detection
o f ground state atoms, molecules or ions in a plasma. It is a ground state observation and
can be quantitative based on intensity, line width, wavelengths and spectral line shapes. In
addition, species o f interest which do not fluoresce can still be detected. This technique has
relatively poor spatial resolution. It is limited to species that absorb based on available laser
or other lamp sources. The sensitivity is limited also unless more sophisticated approaches
such as FM (frequency modulated) laser absorption spectroscopy or intracavity laser
absorption spectroscopy are used. Therefore, it is best suited for detecting abundant species
rather than chemically transient species (ions or free radicals) which occur in much lower
M ass spectroscopy o f stable, long lived gas phase species is possible using
conventional quadruple mass spectrometers. However, detection and quantification o f
metastable, short lived species require sophisticated instrumentation and gas sampling
Each diagnostic technique has its own inherent advantages and limitations. A
complementary use o f these techniques is needed for a fundamental understanding o f the
plasma properties. Only after a broad data base is established on the plasma chemistry in
diamond forming plasmas can we expect to develop a detailed understanding o f the
diam ond growth mechanisms. This knowledge w ill need to be coupled with an
understanding o f the detailed surface chemistry, a problem even more formidable than just
the plasma chemistry.
In a diamond-forming plasma o f hydrogen-hydrocarbon gaseous mixture, various
molecules, atoms, radicals and ions form, and they can be further excited to higher energy
levels by collisions. By emission spectroscopic measurements, information on the degree
o f molecular dissociation and the relative concentration changes o f excited plasma species
under different conditions can be obtained.
The existing literature regarding atomic and molecular spectra is o f great help in
identifying the hydrocarbon and hydrogen species in the present study. The particular
useful references on this subject are by Moore (1949-1958), Pearse and Gaydon (1963)
and Herzberg (1971). The observed species and their corresponding wavelengths as well
as excitation energies o f plasma species in a methane-hydrogen mixture are listed in Table
4.1. It should be noted that the instrument range for detectable wavelengths is from 430 to
Table 4.1. Plasma species and their characteristics1
Wavelength (nm)
Transition States
Excitation Energy (eV)
A3ng ->X3nu
2 .2
(Swan Bands)
2 .6
a 2a -> x 2n
Balmer Series
623 - 573
n 3r<+
A l g -> B
1 Sources: Moore (1949-1958); Pearse and Gaydon (1963); Herzberg (1971); Hartnett (1988)
670 nm. Because o f this small window in the visible regim e, several important
hydrocarbon radicals such as CH 3 , C2H 2 and C 2H could not be observed. Furthermore,
the densities of all ground state species are not determined. Therefore, growth mechanisms
can not be formulated by emission spectroscopy alone.
4.2.3. Characterization of diamond-forming plasmas
There have been a number o f reports on the spectroscopic characterization of
diamond-forming plasmas o f hydrogen and hydrocarbon mixtures using mass, emission
and absorption spectroscopies. Matsumoto et al. (1985) were the first to use emission and
mass spectroscopies to identify plasma species and to correlate the deposits with the plasma
characteristics. Plasma species o f C 2 , CH, C and H (Ha , Hp, Hy) were identified from the
emission spectra. The C2 emission was found to be increased by Ar or He additions, while
the CH emission remained constant. From mass spectroscopic measurements, it was found
that the degree of ionization of the plasma was generally very low. The hydrogen dilution
increased the formation of CH 3, while inert gas additions (Ar or He) enhanced C 2H 2. In a
subsequent work, Matsumoto and Katagiri (1987) examined the effect o f carbon specimen
(graphite) immersion in a pure hydrogen plasma on the plasma species. They observed CH
and C2 radicals by emission spectroscopy and C 2H 2 by mass spectroscopy resulting from
the chemical etching o f graphite.
Saito et al. (1986) measured the emission spectra o f methane-hydrogen microwave
plasmas in the range from 200 to 500 nm. Apart from the species detected by Matsumoto et
al. (1985), they identified H 5 and a hydrogen recombination continuum. They also noted
additional plasma emissions from CN and NH caused by an air leak.
Mitsuda et al. (1987) attempted to correlate the emission intensity ratio o f CH/Hp to
diamond formation and the ratio o f C2/Hp to graphite formation in a microwave plasma of
methane-hydrogen. F or diamond formation, the ratio o f CH/Hp had to be in the range of
0.005-2.0, C 2/Hp less than ~0.05, and C2/CH less than 1.
Park et al. (1987) reported emission spectroscopic results from RF plasmas of
CH 4, C 3H 8 , C2H 4 and C2H 2 at 0.1-0.35 Torr. They found that C 2 emission increased and
CH emission decreased as the plasma composition was varied from CH 4 to C 3H 8 to C2H 4
and to C2H 2 . However, the optical and electronic properties o f deposited a-C:H films were
relatively independent of the chemical nature of precursor gases.
Celii et al. (1988) employed IR diode laser absorption spectroscopy to examine the
gas phase species in a HFCVD process. In a gas mixture of 0.5% methane concentration in
hydrogen, acetylene (C 2H 2 ), methyl radical (CH 3 ) and ethylene (C2H 4 ) were detected
above the growth surface. In a later paper (Butler and Celii, 1989), they additionally used
resonance enhanced multiphoton ionization (REMPI) spectroscopy to determine the
variations of CH3, C2H 2 and H with filament temperatures and methane concentrations and
conducted a simple thermodynamic modeling calculation about the chemical reactions in the
gas phase.
Harris et al. (1988) used mass spectroscopy to measure the species in a HFCVD
process as a function o f filament-to-substrate distance. They reported that diamond growth
resulted mainly from reactions of C2H 2 and/or CH 3 radicals, but contributions from CH4
and C2H 4 could not be ruled out.
Kawato and Kondo (1987) used chromatography to examine the exhaust gas of
CH 4+H 2 and CH4+H 2+O 2 gaseous mixtures. They found that additions o f O2 caused the
concentration of atomic hydrogen to increase over the oxygen-free system.
Hartnett (1988) performed emission spectroscopic analysis o f microwave plasmas
in his thesis research. Identifications of various species and their relative concentration
changes with deposition parameters (mainly pressure) are reported. He also estimated the
plasm a temperature and plasma density from emission intensity measurements and
discussed-the applicability o f the state o f LTE (local thermodynamic equilibrium) in
microwave plasmas.
Inspektor et al. (1989b) exam ined spatial variations o f plasm a species in a
microwave plasma o f H 2+CH 4 . They suggested that the first step in diamond formation
was the dissociation of the starting monomer into various intermediate fragments and by­
products (CHXand C2) and the depletion o f atomic hydrogen in the plasma.
Roman et al. (1989) obtained coherent anti-stokes Raman spectroscopy (CARS)
spectra o f molecular species (CH 4 and C 2H 2) in a RF PACVD diamond process. The
temperature and concentrations o f these species were measured.
Wei et al. (1989) used a double probe technique to measure the plasma density in a
magneto-microwave plasma system. The plasma density was about 1011 cm -3 around the
ECR conditions at 0.1-0.01 Torr. They speculated that the OH radical was necessary for
low temperature diamond deposition.
Spectroscopic measurements have contributed greatly in elucidating the reaction
mechanisms o f diamond formation in an oxygen-containing plasma. Mucha et al. (1989)
used plasma emission actinometry to study the mechanism by which small additions o f
oxygen (-0.5% ) enhanced the deposition rate of diamond in a plasma (CH 4+H 2). They
found that increasing the oxygen concentration up to 5% produced a fivefold increase in
atomic hydrogen. Hirose and Terasawa (1986) attributed the enhancement o f growth rate
by oxygen (O 2) additions to the greater ease of CH 3 radical formation with the oxygencontaining precursors and implied that CH 3 favors diamond growth over other forms o f
hydrocarbon. Inspektor et al. (1989a) indicated that the main feature o f a C/H/O system
was the formation of new active intermediates in the plasma (OH, CO, O, H 2O). These
species opened new reaction paths in the plasma bulk and the plasma-surface boundary to
increase the dissociation rate o f molecules and accelerate preferential etching o f graphitic
Emission spectroscopic results also have been used to calculate the DC plasma gas
temperature by Suzuki et al. (1988). They used the emission intensities o f hydrogen (Ha ,
Hp, Hy, Hg) to calculate the statistical temperatures o f hydrogen atoms and electrons which
were 4.8-5.3 x 103 K and 1.0-1.1 x 10s K, respectively, in plasmas o f 0.5-4% methane
concentrations in hydrogen at 195 Torr.
It appears from the various spectroscopic measurements that methyl radicals (CH3)
and acetylene (C 2H 2) are two primary species responsible for diamond growth. Based on
these results, two models have been put forward for the mechanism of diamond growth.
Tsuda et al. (1986) carried out quantum chemical computations to determine a low energy
path for diamond formation as being the methyl radical insertion and desorption of
hydrogen (H 2 ). Frenklach and Spear (1988) proposed an alternative route for diamond
growth. The surface is first activated by the hydrogen atom removal from a hydrogen
covered surface, and then the activated surface carbon radicals (atoms) react with acetylene
or other carbon-hydrogen species. Huang et al. (1988) also carried out a quantum chemical
calculation for this acetylene mechanistic process and showed an even more favorable
energy path in terms o f lowering the free energy o f the system. However, neither of these
two models can conclusively combine all the experimental information unambiguously. On
the other hand, the aspects o f plasm a gas temperature, species concentration and
distribution (molecular, atomic, or excited) and associated synergistic effects are still
fragmentary and unclear. M ost papers have simply duplicated old results, and few new
findings have emerged.
F o r the effects o f noble gases on the chemical vapor deposition o f various
materials, Knights et al. (1981) studied the effects o f noble gas additions to silane for the
deposition o f a-Si:H films. It was found that higher deposition rates could be achieved
when noble gases were added, with a general trend o f increasing defect density with the
atomic weight o f the noble gas. For the deposition o f diamond-like carbon (DLC),
Matsumoto et al. (1985) and Matsumoto and Katagiri (1987) found that only graphite or
DLC films were produced from gas mixtures of CH 4+Ar and CH4 +He. However, the
mechanisms by which He and A r affect the plasma species, especially in the diamondforming plasmas where a large amount of hydrogen is present, are still not clear.
4.3. Experimental approach
A more detailed study was conducted here for the roles o f noble gases (He, Ne, Ar,
Kr, Xe) in the diamond deposition from CH 4+H 2 microwave plasmas. Since hydrogen is
such an important component in the plasmas to grow diamond, it is interesting to see if
noble gases behave differently in the diamond-forming plasmas o f CH 4 +R+H 2 (R is a
noble gas atom) than in the CH 4 +R plasm as which Matsumoto et al. (1985) and
Matsumoto and Katagiri (1987) have studied for the DLC formation.
Five different noble gases (He, Ne, Ar, Kr, Xe) were added to the CH 4 +H 2
plasmas separately. The noble gases, together with hydrogen, acted as dilution gases in
methane. With additions of each noble gas, an equal amount o f hydrogen gas was reduced.
Therefore, the methane concentration (4.4%) and the total gas flow rate (100 seem) could
be kept constant for all the plasma measurements. An emission spectrometer was used to
analyze and identify major species present in the plasmas which had emission lines within
the range o f the instrument from 430 nm to 670 nm. The microwave power input was
maintained constant (310 watt) during the optical emission measurements o f the plasmas. A
gas pressure o f 90 Torr also was kept constant. The details of the spectrometer and the
experimental set up have been described in section 2.4.5.
The effects o f noble gases on the growth and structure o f diamond films were
studied by growth rate measurements, Raman spectroscopy and neutron activation
4.4. Results
The results here will be presented in three parts: first, the bulk plasma emission
spectra; second, the effects o f noble gas additions on the film growth rate; and finally, the
structure and compositions o f diamond films deposited from noble gas containing plasmas.
4.4.1. Bulk plasma optical emissions
The major species detected in this study are C2 radicals (Swan band, 563.6,516.5
and 473.7 nm), H atoms (Balmer series, Ha : 656.2 nm, Hp: 486.1 nm and Hy: 434.0
nm), CH radicals (431.4 nm) and excited H 2 molecules (581.0 nm) as shown in Figure 4.1
for a typical emission spectrum from the microwave plasma of CH4+H2. CHXradicals
(x> l) could not be detected in the visible range of the emission spectrum. The emission of
C2 radical hereafter refers to the strongest line at 516.5 nm. The emissions from Hy, CH
and excited H 2 were weak and were not measured in this study.
Table 4.2 presents all the effects with noble gas additions that were experimentally
observed. The ionization potentials (I.P.s) and the lowest energy levels o f the excited states
of noble gases are given also. The first observed effect when the noble gases were admixed
into the CH 4+H 2 plasmas was the changes o f the substrate temperature. All five noble
gases increased the substrate temperature. The extent o f this temperature increase varied
with the atomic weight o f the gas. Heavier atomic weight gases, K r and Xe, resulted in
larger temperature increases than lighter atomic weight gases, He, Ne and Ar. The changes
in emission intensities o f the C2 radical and atomic hydrogen (Ha , Hp) at different
I0 6
W a v e le n g th (n m )
Figure 4.1. A typical optical emission spectrum of a microwave plasma (CH 4+H 2)
Table 4.2
Ionization potentials (I.P.s), excited states and corresponding energy levels (E) of
noble gases (R) and their effects on the major species in the diamond-forming plasmas
(CH 4 +H 2 ). The listed value o f emission intensity has been normalized to that from the
plasma (4 .4 %CH 4+9 5 .6 %H 2 ) with H 2 dilution only. Plasma parameters: power = 310 W;
pressure = 90 Torr; gas composition: 4.4%CH4+16%R+79.6%H2; gas flow rate = 100
seem; Si substrate with graphite susceptor immersed in the plasmas.
(V )2
1 .0 0
1 .1 2
3p 2
2 1 .6
1 Oxygen was introduced with argon for a plasma composition of 4.4%CH4+1.6%02
2 Source: Spinks and Woods (1976)
3 Source: Moore (1949-1958)
concentrations o f noble gases are shown in Figure 4.2. The C2 radical had a large increase
in its emission intensity with additions o f all five noble gases (Figure 4.2a). The emission
of atomic hydrogen had different behavior from that o f the C 2 radical (Figures 4.2b and
4.2c). H e and N e had no appreciable effect on the emission intensities o f Ha and Hp, while
Ar enhanced the emissions of H a and Hp, and Kr and Xe greatly reduced the intensities of
H a and Hp emissions. Figure 4.3 shows the emission behavior o f C2 , Ha and Hp at
different A r concentrations. It is clearly seen that Ar promoted the generation of the excited
C2 radical and atomic hydrogen. For example, the peak intensity o f emission from C 2 , Ha
and Hp increased by a factor o f 1.55, 1.14 and 1.17, respectively, when 16%Ar was
present in the CH 4 +H 2 plasmas. It should be noted that in the CH 4 +H 2 system without
noble gases, the changes of the amount o f hydrogen up to 16% had no appreciable effect
on the emissions o f C2 , Ha and Hp.
The emission intensities o f C2 and atomic hydrogen also are plotted vs. the atomic
weights o f the noble gases, as shown in Figure 4.4, in order to compare the effects o f these
different noble gases. The intensity o f the C2 emission increased with the atomic weight of
the noble gas. However, regarding the atomic hydrogen emission, although slight increases
in H a and Hp emissions were found with additions o f light gases, large decreases were
seen with additions o f heavy gases. Another interesting result is the behavior of the ratio of
Ha/Hp, which is an indication o f the plasma temperature (Lochte-Holtgreven, 1968), for
different noble gas additions (Table 4.2 and Figure 4.4). As can be seen, higher Ha/Hp
ratios, which represent lower plasma temperatures, were found when heavier noble gas
atoms were added to the plasmas, although the substrate temperature was increased in these
cases. It is believed that the substrate temperature is mainly controlled by the plasma
density, whereas the plasma temperature is governed by the electron energy. The different
K 2.0 r-
R % inCH4 + R + H2 plasma
Figure 4.2. The emission intensities o f C 2 and H from the plasma (CH 4+R+H 2) at
different concentrations o f noble gases (R): (a) C2 emission; (b) H a
emission; and (c) Hp emission. The intensities have been normalized to
those from a plasma (4.4% CH 4 + 95.6% H 2 ).
Emission intensity of Ho (normalized)
Emission intensity of HQ(normalized)
Figure 4.2. (cont.)
x z>
Ar % in CH4+ Ar + H2 plasma
Figure 4.3. The emission intensities o f C2 , Ha and Hp radicals from the plasma
(CH 4+Ar+H 2) at different A r concentrations. The intensities are
normalized values, the same as in Figure 4.2.
2 -i
Of £
■ C2
0 Hq
w I' -
f t
Hq / H ^
■o c
<u a
0 -T
Atomic weight of noble gas
Figure 4.4. The emission intensities o f C2 , H a and Hp radicals and the ratio o f Ho/Hp
from the plasma (4.4% CH 4 +16% R +79.6% H 2 ) vs. the atomic weights o f
the added noble gases (R). The emission intensities are normalized values,
the same as in Figure 4.2.
effects associated with the atomic weight o f the noble gas are believed to be closely related
to the noble gas chemistry, as will be discussed later.
Emission spectra also were recorded for a CH 4 +H 2 plasma in which a gaseous
mixture o f 10 %O2 +Ar was added (a CH4/ 0 2 /Ar/H 2 system). As indicated in Table 4.2, the
emission intensity of the C2 radical did not change substantially, while those o f H a and Hp
increased. Compared with the plasmas in which only A r gas was added (CH 4/A r/H 2
systems), the C 2 emission was weaker, indicating that the generation o f the excited C 2
radical was suppressed by the introduction o f oxygen into the system, and the intensities of
Ha and Hp remained constant. The substrate temperature was increased with additions of
the gaseous mixture of 10 %C>2 +Ar at the constant level of microwave power input
4.4.2. Film growth rate
Diamond film deposition experiments were conducted in CH 4 +Ar+H 2 plasmas.
The growth rate o f diamond films was increased by argon additions at various methane
concentrations as shown in Figure 4.5a. For example, the growth rate was increased by
60% when 35%Ar was added at a methane concentration o f 4.4% (Figure 4.5b). Growth
rate data for diamond film s deposited from plasmas o f 4.4% C H 4+16% H e+H 2and
4.4% C H 4+1.6% 02+14.4% A r+H 2 also are included in Figure 4.5b. Comparing the
growth rates obtained from the plasmas o f 4.4%CH4+16%Ar+H2 and 4.4%CH4+16%He
+H 2 , it is seen that Ar increased the growth rate more efficiently than He. For the
4.4%CH4+1.6%02+14.4%Ar+H2 gas mixture, the growth rate was decreased by 40%
with the oxygen addition, consistent with the work by McKenna (1990) which showed that
the growth rate of diamond films drops linearly with the atomic ratio of oxygen to carbon.
90Torr 980°C
100 Seem
35% Ar
16% Ar
0 % Ar
1 2
CH4 % inCH4 + A r + H2 plasma
4 . 4 % CH4
o 10% 0p + Ar
o 1.2
1.0 .
•v 0
11 1 1 1 1 1 1 1 1 1 11 1 1 1 1 ■ » 1 1 1 1
Percentage of Ar.He or 10% 0 2 + Ar
in CH4 + R + H2 plasm a
» 1 1 1 1 1 1 1 1 1
Figure 4.5. Growth rates o f diamond films deposited from plasmas (CH 4+R+H 2):
(a) growth rates at various methane and argon concentrations in plasmas of
CH 4+Ar+H 2 and (b) growth rates at various A r concentrations (0-35%) in
plasmas o f 4 .4 %CH4 +Ar+H 2 , at 16%He in a plasma of
4 .4 %CH 4+He+H 2 , and at O 2 and A r concentrations o f 1.6% and 14.4%,
respectively, in a plasma o f 4 .4 %CH 4+ 0 2 +Ar+H 2 .
4.4.3. Film structure and composition
The Raman spectra o f diamond films are shown in Figure 4.6. The peak at ~519
cm *1 is caused by the silicon substrate. This Si peak may disappear, depending on the
thickness and the structural quality o f diamond films as shown in Figure 4.6b. It is seen
that little amount o f additional non-diamond carbon components actually was introduced in
diamond films by additions o f up to 35%Ar or 16%He into the CH 4 +H 2 system. Further
increases in A r concentrations leaded to the formation o f DLC, or even graphite films, as
shown in an extreme case in Figure 4.6b in which 95.6%Ar was added to the system
producing a similar film to that reported by Matsumoto et al. (1985) and Matsumoto and
Katagiri (1987). Also it is seen that the addition o f a small amount of oxygen together with
A r gas greatly reduced the non-diamond carbon content in the film.
Neutron activation analysis shows that noble gas atoms can be incorporated in
diamond films when they are deposited in CH 4/R/H 2 systems. Table 4.3 shows that the Ar
concentrations in the diamond films are in the range of 23-60 atomic ppm when A r content
in the gas phase was varied between 16% and 35%. These incorporated noble gases atoms
may cause anomalous electron diffractions as indicated by Spencer et al. (1976).
4.5. Discussion
4.5.1. General functions o f noble gases
The plasma assisted chemical vapor deposition is based on the gas phase generation
o f reactive precursor species through plasm a excitation. The m olecular excitation,
dissociation and ionization in a plasma are accomplished by various processes involving
collisions with free electrons, ions and excited neutral species or radicals. Undoubtedly,
free electrons play a decisive role in sustaining the plasma and initiating plasma chemical
3 5 % Ar
16% Ar
0% A r
Wovenumber (cm-1)
Figure 4.6. Raman spectra of diamond films deposited from plasmas of:
(a) l%CH 4 +(0 -3 5 %)Ar+H2; and (b) 4.4%CH4+(0-16%)Ar+H2,
4.4% CH 4 + 16 %He+H 2 , 4 .4 %CH 4+ 14 .4 % Ar+ 1.6 % 0 2+H 2 and
4 .4 %CH 4 + 9 5 . 6 %Ar.
Relative Intensity
O’ 2 .
Table 4.3. Ar contents in diamond films measured by NAA 1
Diamond film
4 .5 %CH 4+3 3 %Ar+
79.5%H 2
0 .2 0
N (area)
tj (initial time)
tf (finish time)
Argon atoms (xlO14)
Ar content (ppm)
Error (%)
Plasma composition 4.5%CH4+16%Ar+
Initial weight (mg)
(firee-standing film)
Carbon atoms
Ar (atom)
I a (1 - e
where x = 900 sec; X= 1.052 x 10-4 /sec; £1= 5.1 x 10'^ (at 1300 keV); I = 0.996;
a = 6.6 x 10-25 cm2; q _ 0.991; Q = 1.6 x 10^ /cm2-sec.
reactions in which ionic, electronic or vibrational excited species are formed (Inspektor,
1987). These species interact with each other and with the matrix gas through various
chemical reactions and are directly involved in the deposition process. The active species
may be de-excited by radiative emissions or by various collisional processes such as charge
transfer or energy transfer. The consequence is that energy is redistributed among various
species, and physical and chemical properties of the gases are strongly influenced.
Dilution gases generally have two basic functions in a plasma system. The first is
simply a dilution effect on reactant species in the plasma. The second, and also the more
important, is a chemical effect related to the chemical activities of different dilution gases.
In our case, the noble gas, being a part o f the gaseous mixture exposed to the microwave
field, absorbs its share o f energy. Whether or how this energy absorbed by the noble gas
can be transmitted to a reactant molecule by energy or charge transfer, and thereby
influence the chemical reactions, is a question of great interest Noble gas chemistry, such
as dissociation and ionization energies and cross-sections for various collisional processes,
is thus of fundamental importance in understanding underlying mechanisms for the plasma
On the basis of the above spectroscopic measurements of the plasmas, it is clear that
the main difference between the gas mixture systems in which only hydrogen is the dilution
gas and those in which noble gases are added lies in the amount o f active species present in
the plasmas. A high emission intensity, which represents a high density o f the excited state
species, can result from either a high concentration o f the ground state species or a more
efficient excitation of that species. Since the observed changes o f the emission intensities
were associated with a constant E/P (E is the electrical field strength related to the
microwave power input, and P is the gas pressure in the system), it is proposed that the
noble gas atoms in their ionic or excited states behave as energy catalysts (Venugopalan and
Veprek, 1983) which influence the density and energy distribution of electrons and various
active speeies and accelerate reaction processes in the plasmas.
4.5.2. Effect o f noble eases on the production o f atomic hydrogen (HI
Let us first look at the process o f atomic hydrogen production. In the CH 4 +H 2
plasmas without noble gas additions, a direct electronic mechanism (DEM) or a joint
vibroelectionic (JVE) mechanism is often considered as the major mechanism for plasma
enhanced dissociation of hydrogen (Capitell and Molinari, 1980):
H 2 + e- = 2H
------ (4.1)
In addition, the following reactions also are expected to occur by ionic processes as well as
through electronic excitation (Firestone and Dorfman, 1971):
H2+ + H 2 = H3+ + H
H 3+ + e- = H 2 + H
------- (4.2)
------- (4.3)
The generated ground state hydrogen atoms can be excited subsequently by electron
collisional reactions (excitation energy: Ha =12.09eV, Hp=12.76eV), and subsequent
radiative decay will be observed by emission spectroscopy. When noble gases are
introduced into the systems, the plasmas contain a large amount o f excited noble gas
species such as He*, Ne*, A r*, Kr* and Xe*, as well as corresponding ionic species
which are produced by collision processes with high energy electrons. Since heavier noble
gas atoms have larger electron collision cross-sections, a greater portion o f electrons in the
plasmas is cooled by the heavier noble gas atoms. The hydrogen (H2 ) dissociation and
subsequent excitation of H atoms by the DEM, JVE and ionic mechanisms which favor
high energy electrons will, therefore, in general be less efficient when heavier noble gas
atoms are involved. Indeed, with the exception o f Ar, this is just the behavior described in
Figures 4.2b, 4.2c and 4.4.
On the other hand, the molecular excitation and dissociation o f hydrogen (H 2) by
the Penning effect (an energy transfer process), with species in a metastable state, occur
R* + H 2 = R + H 2 * = R + 2H (R is any noble gas atom)
------ (4.4)
By this Penning energy transfer process all the excited noble gas atoms R* could generate
ground state H atoms, since only 4.48eV is involved for the dissociation o f H 2 , although
the threshold energy may be higher than that (Boenig, 1982). However, Ar* can effectively
excite H atoms to emit, because there is an appropriate match between the energy available
in the Ar excited state and the energy needed by the receiving ground state hydrogen atoms.
W ith Kr* and Xe* the excitation probability of H atoms is considerably lower because of
their lower metastable energies. For He and Ne the numbers o f excited atoms, He* and
Ne*, are relatively small due to their higher excitation energies. Since the energy levels of
He* and Ne* are high enough to ionize hydrogen molecules and atoms, the energy transfer
process is believed to be fundamentally changed, and the Penning effect is distorted (Krogh
et al., 1986).
In addition, the ionic noble gas species may also play a role in the production o f
atomic hydrogen. Two oppositely directed effects could occur depending upon the
ionization potentials o f the noble gases relative to the ionization potential o f hydrogen:
enhancement and inhibition o f the atomic hydrogen production. The relevant reactions are:
RH+ + e* = R + H (R is any noble gas atom)
------- (4.5)
R+ + H 2 = RH+ + H (R is He, Ne or Ar)
RH+ + H 2 = R + H 3+ (R is He, Ne or Ar)
------- (4.7)
H 2+ + R = R+ + H 2 (R is Kr or Xe)
H 3+ + R = RH+ + H 2 (R is K r o r Xe)
------- (4.8)
------- (4.9)
The reaction (4.5) is a neutralization process applied to any noble gas. The reactions (4.6)
and (4.7) enhance the production of atomic hydrogen, since the ionization potentials of He,
Ne and A r are higher than the ionization potential o f H 2. The reactions (4.8) and (4.9)
inhibit the production o f hydrogen atoms due to the lower ionization potentials of K r and
Xe than that o f H 2 . They cut off the ionic reactions (4.2) and (4.3) which are important
channels for producing atomic hydrogen by converting ionic hydrogen (H2+ and H 3+) back
to hydrogen molecules (H 2). The latter two reactions, (4.8) and (4.9), together with the
low electron temperature and the small Penning effect for generating and exciting H atoms,
can account for the dramatic decreases in the emissions of atomic hydrogen with additions
of K r and Xe. He and Ne essentially have no effect on the atomic hydrogen generation,
because there are neither significant collisional processes to cool electrons, and thus
suppress atomic hydrogen, nor effective Penning processes to increase and excite atomic
hydrogen in the He and Ne-containing plasmas. The reactions (4.5) to (4.7) for He and Ne
also are not expected to be very efficient because o f their extremely high ionization
potentials. A r enhances the atomic hydrogen generation and excitation substantially due
mainly to the efficient Penning reactions for generating and subsequently exciting H atoms.
4.5.3. Effect of noble eases on the dissociation of methane (CH4 )
For the polyatomic molecule CH 4 , the dissociation is usually associated with its
excitation (Prince et al., 1964). The noble gases can greatly accelerate the dissociation rate
of CH 4 . The excitation and dissociation o f CH 4 by electron im pact and hydrogen
abstraction are certainly important functioning processes (Yarbrough et al., 1990):
CH 4 + e -= C H 3 + H
------ (4.10)
CH 4 + H = CH 3 + H 2
------ (4.11)
However, these two mechanisms can not account for the fact that K r and Xe, which
correspond to the lowest electron temperatures and possibly the lowest amount of atomic
hydrogen among the CH 4+R+H 2 plasma systems, promote the molecular dissociation of
CH 4 the greatest. CH 4 has an ionization potential o f 13.02V, which is lower than those of
He, Ne, A r and Kr. Ionic processes can supply enough energy to produce CH** or, more
probably, CH 3+ (Lind et al., 1961):
R+ + CH 4 = R + CH 3+ + H (R is He, Ne, Ar or Kr)
------- (4.12)
However, this ionic process is expected to be slow, since it involves R+ with high
ionization potentials, and in addition, it can not explain the Xe effect, either, because Xe
has a lower ionization potential than that of CH 4 . The phenomenon is, therefore, attributed
to an energy transfer process, i.e., the Penning effect, similar to the reaction (4.4)
associated with the metastable states of noble gases. Energy is transferred to the reactant
molecules from the excited noble gas atoms which take little part in the subsequent energy
disposal. The dissociation of the energized molecules is believed to occur subsequently.
Noble gas atoms with higher excited energies can induce more complete dissociation of
CH 4 molecules. For example,
Ar* + CH 4 = Ar + CH 4 * = Ar + CH2 + 2H
Xe* + CH 4 = Xe + CH4 * = Xe + CH 3 + H
The threshold energy is 9.3eV for the reaction (4.13) and 4.5eV for the reaction (4.14)
(Balamuta et al., 1983). If energetically accessible, the dissociative ionization also could
happen. Subsequent collisions involving CHX(x=l, 2 ,3 ) with R* can yield the C 2 radical.
Since a low excitation energy (~2.4eV) is required to excite the C2 emission (Table 5.1), all
the excited noble gas atoms R* can enhance the C2 emission by the above energy transfer
process and subsequent excitation. The differences among these different noble gases are
their atomic weights, the concentrations of the excited atoms (R*) in the plasmas which
result from the different excitation energy levels, and the degree o f match between the
metastable energies of noble gas atoms and the excitation energy needed by the ground state
C 2 to generate emissions. For particles o f constant energy, the momentum per impact
increases with the square root of the mass. The numbers of Kr* and Xe* are also larger
than those o f He*, Ne* and Ar* due to the relatively lower excitation energies of K r and
Xe. Furthermore, the lower metastable energies o f Kr* and Xe* more closely match the
energy required to excite C 2 emissions. As a result, the heavier gases could promote the
dissociation o f CH 4 molecules more efficiently, and the C2 emission intensity is inversely
proportional to the difference in the excitation energies between C2 and R*. Strong
emissions from the C2 radical due to the Kr and Xe additions to the CH4+H 2 plasmas are
thus interpreted.
4.5.4. Effefct o f oxygen on the film structure
Introduction of a small amount o f oxygen into the system along with Ar additions
yields some very promising results. Oxygen, which can itself act as a selective etchant of
non-diam ond carbon, opens additional channels to reduce non-diamond carbon
components in diamond films by suppressing the production of the C2 radical in the plasma
as observed in our emission measurements. Based on this result, it is possible to grow
diamond film s with high deposition rates at high methane concentrations in oxygencontaining plasmas, because, although a high growth rate can be achieved easily in the
plasmas with high methane concentrations, the films produced are usually highly defective
deposits containing a large amount o f non-diamond carbon phases. Oxygen introduced
with the noble gases can effectively suppress the simultaneous deposition o f non-diamond
carbon components at the sacrifice of a relatively slight reduction of the growth rate, and,
therefore, greatly improve the quality o f the deposited films. This should be an effective
way to deposit diamond films at high growth rates while still keeping non-diamond carbon
components in the films to a minimum.
4.5.5. Limitations of thestudv
It is seen from the above discussion that noble gases can strongly influence the
generation and excitation o f active plasma species through energy transfer (Penning effect)
and charge transfer (ionic reaction) processes. Since a large amount o f excited, dissociated
or ionized carbon-containing species, such as di-carbon radicals and atomic carbon (Mucha
et al., 1989), is a prerequisite for a high deposition rate, it can be understood that additions
o f noble gases into the CH4 +H 2 systems can increase the growth rate of diamond films.
Because atomic hydrogen is an important species promoting diamond growth while
suppressing non-diamond carbon deposition, and the C 2 radical mainly results in non­
diamond carbon components (Matsumoto et al., 1985; Matsumoto and Katagiri, 1987;
Mitsuda et al., 1987), there is a possible trend o f increasing non-diamond phases in the
films with increasing atomic weight of the noble gas. Furthermore, bombardment of the
growing films by the noble gas atoms in the microwave plasmas also possibly induce the
formation of non-diamond carbon components (Venugopalan and Veprek, 1983).
It should be noted that the investigation reported here is concerned mainly with the
bulk properties o f the microwave plasmas. How the noble gas additions affect the surface
processes, thus directly influencing the film growth and modifying the film structure and
related properties, is still an open question. Furthermore, many critical hydrocarbon species
such as CH 3 and C2H 2, which may be the responsible precursors for diamond growth
(Tsuda et al., 1986; Frenklach and Spear, 1988), could not be detected in this study due to
the small instrumental range o f emission observations. In order to have a more profound
understanding o f the mechanistic processes for the diamond deposition, a relationship
between the relative concentrations o f these hydrocarbon species and the emission
intensities o f C 2 , Ha and Hp is necessary. However, some correlations have been
observed between the bulk plasma state and the growth rate and structure o f the films. The
results demonstrate that a high degree of molecular excitation and dissociation in the
plasmas is a critical factor in obtaining enhanced growth rates o f diamond films.
4.6. Summary
In CH 4+R+H 2 systems (R is any noble gas atom), although hydrogen is the main
dilution gas reacting actively in the diamond deposition process, noble gases are also active
in deposition by creating additional excited atom-molecule or ion-molecule reactions. R*
and R+ can be generated and, in turn, create more active carbon-containing species and
atomic hydrogen, although the excitation o f hydrogen atoms is inhibited with additions of
K r or Xe. The presence o f noble gas excited atoms and ions is an important channel for the
dissociation and excitation o f hydrogen and hydrocarbon molecules and, therefore, for the
increase of the deposition rate. Such a catalytic effect can be attributed to energy transfer or
charge transfer from highly excited and ionic species o f the noble gases. Additions of a
small amount o f oxygen along with the noble gases can greatly reduce non-diamond carbon
components in the diamond films.
C hapters
5.1. Introduction
This chapter contains the core work o f the thesis research. It is devoted to the study
o f structural properties o f diamond films which include the internal bulk structure,
interfacial structure and crystalline defects. A series of self-consistent experiments were
performed in which various structural features were measured for a number of important
deposition parameters. Since the structural features are either directly responsible for or
related to the desired film properties, a comprehensive study and a thorough understanding
of the structure of CVD diamond films and its correlations with both the process parameters
and resulting properties are o f great significance. The results presented below represent a
great stride toward achieving such an understanding.
5.2. Internal bulk structure
In this section, we are interested in the unique internal structural feature of diamond
films deposited under non-optimum conditions, that is, a diphasic mixture of diamond and
graphite phases. The existence, distribution, crystallographic orientation, size and
morphology o f graphite crystallites within diamond films will be reported. The origin o f
graphite, the mechanism o f formation and effects o f various deposition parameters were
5.2.1. Statement of the problem
The MPECVD o f diamond films is a competitive process o f nucleation and growth
among different forms o f carbon, in particular between the crystalline phases, diamond and
graphite. Films formed from such a plasma environment grow under conditions that are
distinctly different from those for usual physical and chemical deposition processes. At
present, although pure diamond films with an undetectable amount of graphite phase and
other non-diamond components can be produced, most o f the so called "good quality"
diamond films contain some non-diamond carbon phases. Non-diamond carbon refers to
the small amount o f second phase carbon which is seen in typical Raman spectra of
diamond films (Knight and White, 1989) and can include microcrystals o f graphite,
carbynes and amorphous carbon. Mechanisms for the formation o f these non-diamond
components, particularly graphite in diamond films, are still to be determined. More
importantly, these non-diamond inclusions directly affect the optical and electrical
properties of diamond films.
In an effort to understand and control the formation o f these non-diamond phases, a
careful study of the graphite inclusion in diamond films was carried out. The TEM results
are compared with Raman spectra in order to get meaningful information on the origin of
the graphite second phase. Possible mechanisms of the graphite formation are proposed.
5.2.2. Literature review
Although graphite formation is probably due to a co-deposition process in CVD of
diamond films, literature information on the mechanism of diamond graphitization at high
temperatures also will be included, which may help in understanding the mechanism of the
graphite formation in the CVD process.
Evans (1979) reported that diamond could sustain temperatures up to 1500°C in
vacuum or an inert gas before converting to graphite. Graphitization did not start over all
the surface but was initiated at discrete sites. Preferred orientations could develop that the
hexagonal graphite axis was associated with the trigonal diamond axis when partial
graphitization occurred. However, for a complete conversion, graphite was polycrystalline,
and no preferential orientation could be observed. The graphitization rate depended on the
crystallographic orientation of diamond, with the {110 } faces having the highest rate, the
{111} faces intermediate, and the {100} faces the lowest in a temperature range o f 19002050°C. The activation energy for graphitization of the {110} faces was 175 kcal/mol
through a mechanism of detachment of a single atom from the diamond lattice by the
breaking o f two carbon-carbon bonds. For the {111} faces, the activation energy of
graphitization was 254 kcal/mol by detachment of surface atoms, half by breaking three
bonds per atom and half by breaking one bond per atom with the rate controlled by the
slowest step. Earlier in 1965, Evans used TEM to reveal that graphitization of diamond
consisted of at least two stages. The surface carbon atoms initially were transferred to a
surface carbon layer, and then grain growth rapidly occurred in this layer to give the
observed graphite crystallites with dimensions ranging from 100-150 A. On heating o f thin
fragments at 1500-2000°C in an inert atmosphere, there was no evidence for any selective
graphitization at defects in the diamond, although severe strains and probably dislocation
nucleation were observed near the diamond-graphite boundaries.
Goma and Oberlin (1980) and Rouzaud et al. (1983) found that pure carbon films
graphitized in five stages under heat treatments in an inert gas flow in the temperature range
of 1000-3000°C, and each stage was characterized by the release of a given type of defect.
This kind of progressive graphitization by a sudden disappearance o f certain types of
defects also has been reported by Fischbach (1971). Mrozowski (1980) once proposed that
the driving force for graphitization was the strain developed at the crystallite boundaries,
which were responsible for the wrinkling of the aromatic layers. When they were removed
by heat treatments, a crystallite rearrangement was induced. Actually there was much
controversy in the literature regarding the true mechanisms o f graphitization, as reviewed
by McLintock and Orr (1973).
For the low pressure diamond synthesis, Gonzalez-Hernandez et al. (1986)
reported that dehydrogenation of diamond-like carbon (DLC) films would lead to the
transformation o f the diamond-like phase to the graphitic-like phase. Soviet researchers
pointed out the possibility of graphite formation in the gas phase followed by surface
deposition and further growth. Recently, two schools of thought on the mechanisms of
graphite formation and, of course, also diamond formation, are emerging. One emphasizes
the kinetics of graphite and diamond deposition by suggesting that graphite does form in
the process but is preferentially etched or gasified by atomic hydrogen (Spitsyn and
Bouilov, 1988; Fedoseev et al., 1984; Setaka, 1989). The growth of graphite is kinetically
unstable with respect to the growth of diamond.
The other focuses on the roles of atomic hydrogen in surfaces by arguing that the
bulk stability of graphite relative to diamond actually is irrelevant, because it is small, and
the growth occurs at surfaces (Yarbrough and Roy, 1988; Pate, 1986). Atomic hydrogen,
instead o f etching graphite, probably destabilizes the surface o f graphite while stabilizing
the diamond surface and preventing reconstruction of the diamond surface. The graphite
formation is impeded, not necessarily by any preferential etching, but because the graphite
surface is thermodynamically unstable relative to the surface of diamond. Without surface
hydrogenation, reconstruction and graphitization of grown diamond will happen. The
enhanced etching of graphite is just a reflection of the relative instability of graphite
surfaces to gasification.
There is another suggestion also emphasizing surface stabilization by appealing to
the role o f the interface between the substrate and the growing film (Machlin, 1988). A
strong epitaxial relation between the substrate and the deposit enables many solid phases to
be deposited in metastable crystalline forms. The stable diamond growth is a result of the
influence o f the bulk structure for inhibiting reconstruction and graphitization o f the
surface. This explains the pioneering work where growth on diamond seed crystals was
reported, and where little or no atomic hydrogen was present (Eversole, 1962; Deijaguin et
al., 1973; Chauhan et al., 1976).
5.2.3. Experimental approach
Substrates of {100}-oriented single crystal silicon wafer with thickness o f 375 |im
or 75 pm , fused silica and tungsten carbide cutting tool material were used. They were
scratched with 1 p m diamond paste before being submerged into the plasm as for
deposition. Process parameters o f methane concentration, substrate temperature, plasma
etching, film thickness, substrate materials and argon ion bombardment were investigated
for their possible effects on the formation of graphite in diamond films. The air plasma
etching w as accomplished in the same tubular deposition system by introducing
compressed air into the tube at a flow rate o f 100 seem at a pressure o f 10 Torr. The
diamond film was placed in the center o f the air plasma at about 900°C for 5 minutes.
Hydrogen gas or moisture had to be completely removed before initiating the air plasma.
The etching rate of the diamond film was greater than 1.5 pm/h. The hydrogen plasma
etching was carried out in the same system at 90 torr and 1000°C for 2 hours. The gas flow
rate again was at 100 seem. The etching rate was not as fast as that for air plasma etching.
The effect o f argon ion bombardment was examined by depositing a thin diamond film on a
pre-perforated 75 pm thick Si wafer and analyzing the diamond structure around the edge
of the tiny hole by TEM.
Raman spectroscopy, a popular characterization tool in the field o f CVD diamond,
was applied to characterize the diamond phase and to detect non-diamond components in
diamond film s. In addition, a m ore direct and sensitive technique than Raman
spectroscopy, transmission electron microscopy (TEM), was employed to study the
graphite formation in diamond films. As will be seen, the dark field electron microscopic
images and corresponding electron diffraction patterns clearly indicate the existence,
distribution, orientation, size and morphology of graphite in diamond films prepared at
non-optimum conditions.
5.2.4. Results Effect of methane concentrations (CH a %)
D iam ond film s deposited under different m ethane concentrations had
correspondingly different internal structures as indicated in transm ission electron
micrographs, selected area diffraction (SAD) patterns, and Raman spectra (Figure 5.1). At
the low methane concentration o f 0.5%, the film consisted totally o f diamond crystals
without any detectable graphite phase or other non-diamond components, as the Raman
spectrum and the electron diffraction pattern indicated (Figure 5.1a). Diamond crystals
generally contain planar defects o f twins and stacking faults on the {11 1 } planes, as will be
reported later in section 5.4.
W ith an increase in methane concentration to 2%, a broad peak around 1550 cm *1
appeared in the Raman spectrum which indicated some non-diamond phases. The electron
diffraction pattern showed graphite {0 0 .2 } diffraction rings coexisting with diamond
diffraction spots (Figure 5.1b). Graphite {00.4} diffractions were very weak owing to the
scattering nature and the high structural distortion of graphite. The graphite diffraction rings
showed a non-uniform intensity distribution which gave arcs o f stronger intensity to the
graphite ring pattern. This indicated that graphite crystallites were preferentially oriented or
textured about their basal planes in the diamond film. The length o f the arc coiresponded to
the degree o f orientational variation of graphite crystallites. It was noted that the graphite
800 1000 1200 1400
Figure 5.1. Electron diffraction patterns, dark field images and Raman spectra o f diamond
films deposited at different methane concentrations: (a) 0 .5 %CH4 , the dark
field image from the diamond {1 1 1 } diffraction spot, as circled, showing the
twin structure; (b) 2 %CH 4 , the dark field image from the graphite {00.2}
textured diffraction arc, as circled; and (c) 5 %CH 4, the dark field image
from the graphite {0 0 .2 } untextured diffraction ring, as circled.
1200 1400 1600
Figure 5.1. (cont.)
Figure 5.1. (cont.)
{0 0 .2 } diffraction arcs was aligned with the diamond < 111> reciprocal lattice vectors with
a spread o f orientations within a cone angle o f about 40° (Figure 5.2). It was thus
suggested that graphite formed on o r close to the { 1 1 1 } planes of diamond, and the
hexagonal graphite axis was parallel with the trigonal diam ond axis. Such a
crystallographic relationship also was found between diamond and graphite by Evans in a
graphitization study of natural diamond (Evans, 1979).
When the sample was tilted so as to have the diamond <111> zone axis parallel to
Figure 5.2. Electron diffraction patterns from a diamond film deposited at 975°C and
2 %CH4 showing that the graphite {00 .2 } diffraction arcs are aligned with
the diamond {111} diffraction spots with a spread of orientation o f 40°.
(a) B = [Tl2]; and (b) B = [011].
the incident electron beam (Figure 5.3), there was an absence o f graphite {10.0}-type
diffractions, which suggested that graphite grew mainly in the basal plane, or the a
direction, and not in the direction perpendicular to the basal plane, or the c direction. The
dark field image obtained from the graphite {00.2} diffraction arcs (Figure 5.1b) showed
small bright dots which were aligned perpendicular to the graphite <0 0 .2 > reciprocal lattice
vector. Since the Bragg angle for {00.2} diffraction was very small, about 0.3°, these
bright dots corresponded to stacks o f graphite hexagonal layers seen edge-on and
embedded within the diamond matrix. The dimensions of these graphite crystallites were
Figure 5.3. A diamond <111> zone axis electron diffraction pattern showing that no visible
{ 10 .0 }-type diffractions o f graphite are present in a diamond film deposited at
9 7 5 ° C a n d 2 %CH 4.
extremely small, varying between 50-100 A in the a direction o f the basal plane and 20-50
A in the cdirection. Since these graphite crystallites were no more than 100 A in size, the
maximum allowable deviation from the exact Bragg angle was large, i.e., the possible
misorientation o f the film which allowed the dots to remain bright in the dark field image
was large.
The dark field imaging study also revealed an unusual and surely important
phenomenon associated with the textured electron diffraction pattern, that is, a domain
distribution o f graphite crystallites in the diamond matrix, as shown in Figure 5.4. The
dark field images from two different oriented diffraction arcs revealed different areas, or
domains. These domains corresponded to two types of graphite crystallites with different
orientations, that is, they lay on two different {111} planes of diamond. However, no
spatial connection was found between the graphite domains and the underlying diamond
defect structures such as twins and stacking faults.
When the methane concentration was increased further (5%), the resulting Raman
spectrum showed a more intense non-diamond scattering around 1500 cm-1, and the
corresponding electron diffraction pattern showed a strong and complete diffraction ring of
the graphite {0 0 .2 } planes which indicated a loss o f preferred orientation o f graphite
crystallites about their basal planes (Figure 5.1c). Graphite crystallites were no longer
textured and had a random distribution in the diamond matrix. Effect o f substrate temperatures (T J
As indicated in section 3.5, the optimum temperature for diamond film deposition at
the maximum rate was about 1000°C. Increasing the temperature above this optimum
favored graphite formation. As Figure 5.5a shows, the Raman spectrum indicates that the
film basically was composed o f the graphite phase with no diamond 1332 cm ' 1 peak
200 nm
200 nm
Figure 5.4. An electron diffraction pattern, bright and dark field images showing the
domain distribution of graphite crystallites in die diamond matrix of a film
deposited at 975°C and 29 &CH4: (a) the diffraction pattern; (b) the bright
field image showing planar defects present; and (c) and (d) the dark field
images obtained from G1 and G2 graphite diffraction arcs, respectively.
present when deposited at 1100°C. However, the electron diffraction pattern indicated that
the diamond and graphite phases coexisted. Diamond appeared in cluster centers, and
graphite tended to nucleate around boundaries o f the clusters and to grow to larger
dimensions, up to 500 A in the a direction and up to 150 A in the c direction.
A t lower than the optimum temperature, the film contained an increasing amount of
amorphous carbon phases as indicated in the Raman spectrum by the strong peak o f about
1500 cm *1 corresponding to non-diamond components (Figure 5.5b). The electron
diffraction gave a diffuse pattern in the small angle area, and no graphite crystallites could
be detected by dark field imaging.
Similar to a previous report (Badzian et al., 1988a), it is seen here that Raman
spectroscopy was more sensitive for graphite and disordered carbon phases detection due
to the high Raman scattering cross sections (Wada et al., 1980), whereas TEM was better
for diamond phase characterization.
5,2.4.3. Effect of plasma etching
A diamond film containing the graphite phase, as detected by both Raman
spectroscopy and electron diffraction (Figure 5.1b), was exposed to air plasma etching. As
shown in Figure 5.6a, the Raman spectrum indicated no significant difference before and
after the air plasma etch. However, TEM could no longer detect any graphite phase in the
etched film. It appeared that graphite was selectively etched by oxygen and/or nitrogen
atoms from the air plasma due to the difference in the etching rate between graphite and
diamond (Evans and Sauter, 1961; Holland and Ojha, 1975), but the etching was confined
only to the surface region o f the diamond film. With the hydrogen plasma etching of the
diamond film, similar results were obtained because of the preferential etching o f graphite
by atomic hydrogen (Spitsyn et al., 1981; Matsumoto and Katagiri, 1987) at the film
800 1000 1200 1400
Figure 5.5. Electron diffraction patterns, a dark field image and Raman spectra o f diamond
films deposited at different substrate temperatures: (a) a diffraction pattern, a
dark field image from the the graphite {00.2} ring, as circled, and a Raman
spectrum of a diamond film deposited at 5 %CH4 and a high temperature of
1100°C; (b) a diffraction pattern and a Raman spectrum of a diamond film
deposited at 5 %CH 4 and a low temperature of 860°C.
Figure 5.5. (cont.)
1200 1400 1600
Figure 5.6. Electron diffraction patterns and Raman spectra o f a diamond film deposited at
975°C and 2 %CH 4 which was (a) air plasma etched and (b) hydrogen plasma
surface (Figure 5.6b). In this study, Raman spectroscopy characterized the surface and
sub-surface structures, whereas TEM analyzed the surface structure only. Effect o f substrate materials
Different substrates such as silicon, silica and cobalt bonded tungsten carbide were
used to deposit diamond films. Their effects on the graphite formation were examined by
TEM. As shown in Figures 5.7 and 5.8, as well as in previous micrographs, the films
deposited on these different substrates at the high methane concentration (5%) showed
similar electron diffraction patterns. Also the size and the distribution o f graphite were
similar. Effect o f film thickness
It was found that the top surface and the interface regions o f diamond films
deposited on scratched substrates had different structures. The film near the interface with
the substrate (back side) had considerably more graphite and other non-diamond
components than the film close to the top surface (front side) as shown in Figure 5.8. The
front side o f the film had a Raman spectrum with a distinct 1332 cm ' 1 diamond peak and a
relatively small non-diamond carbon background, whereas the back side o f the film was
essentially a graphite layer without any detectable diamond peak. The microstructure of
both sides also was different as revealed by TEM. The front side showed well crystallized
diamond grains with strong electron diffraction spots from the diamond phase, whereas the
back side showed clusters which contained the fine-grain polycrystalline diamond and
graphite phases. The distributions of diamond and graphite at the back side were similar to
those in the high temperature-deposited film as shown in Figure 5.5a, that is, diamond was
in the cluster centers, and graphite crystallites lay along the boundaries. Graphite
1200 1400 1600
Figure 5.7. A n electron diffraction pattern, a dark field image from the graphite {00.2}
ring, as circled, and a Raman spectrum o f a diamond film grown on Si02 at
9750C and 5 %CH 4 -
1200 1400 1600
Figure 5.8. Electron diffraction patterns, dark field images from the graphite {002} rings,
as circled, and Raman spectra o f a diamond film grown on a Co bonded W C
substrate at 950°C and 5%CH4: (a) the front side; and (b) the back side.
Figure 5.8. (cont.)
crystallites were large, comparable with those obtained at 1100°C. Similar results were
found in the films deposited on all three substrates used in this study. Effect of arson ion bombardment
It was possible to induce the graphite formation in diamond films during argon ion
bombardment (ion milling) in the TEM sample preparation due to thermal spike effects or
momentum transfer effects (Weissmantel et al., 1983). If so, samples from all deposition
conditions should show evidence of graphite existence as long as they suffered argon ion
bombardment during the TEM sample thinning. This was found to be not true. In fact, the
film deposited at the low methane concentration showed no detectable graphite phase within
the resolution o f the microscope. In addition, graphite w as detected in the TEM
observations of diamond films deposited on previously thinned silicon substrates which
were not exposed to argon ion bombardment after deposition. Thus, there were no
measurable artifacts induced in diamond films by the argon ion beam thinning during the
TEM sample preparation. If argon ion bombardment played a role in graphite formation, it
would be on a scale smaller than the observations made here, i.e., possibly on a scale of a
few atomic spacings.
5,2,5,. Discussion
5.2.5. LJMechanisms of graphite formation
The principal issue which must be addressed is the mechanism(s) o f graphite
formation during the CVD diamond film growth. From the viewpoint of thermodynamics,
the graphite phase is more stable than the diamond phase and should be the predominant
phase deposited under CVD conditions. However, it is helpful to keep in mind that plasma
assisted CVD is a dynamic process far removed from thermodynamic equilibrium. In this
particular diamond film deposition approach, two specific, important factors, i.e., the
existence of a large amount o f atomic hydrogen and the polycrystalline nature o f the films
which leads to a large internal and top surface area, raise further questions concerning the
application of thermodynamics to the film deposition process and complicate considerations
about the structure o f diamond films. For example, atomic hydrogen could terminate the
dangling carbon bonds on the surface o f diamond to prevent sp 2 bonds and maintain sp 3
bonding in the optimum temperature range of diamond film deposition (Pate, 1986); atomic
hydrogen could preferentially react with graphite, thus removing graphite and allowing
diamond to grow (Spitsyn et al., 1981; Matsumoto and Katagiri, 1987); or the active
surface processes, together with the atomic hydrogen species, could possibly cause
diamond to be the more stable phase than graphite under certain deposition conditions
(Machlin, 1988; Sommer et al., 1989).
From the results presented above, it is suggested that the preferential etching of
atomic hydrogen plays an important role in the elimination and reduction o f the graphite
phase in diamond films. A t the low er methane concentration (0.5%), the hydrogen
concentration is correspondingly higher, and any graphite crystallite which nucleates is
preferentially etched by atomic hydrogen from the plasma before it grows. No graphite
crystallites could actually form and remain in diamond films under this condition.
With an increasing methane concentration (2%), the nucleation density and growth
rate o f both diamond and graphite crystallites increase because of the high supersaturation
of carbon species in the plasma. Graphite crystallites could remain and be embedded in the
diamond matrix before they are completely etched. Once graphite is encompassed within
the diamohd matrix and is no longer in contact with hydrogen species, the embedded
graphite crystallites could grow further. But due to graphite's large specific volume
compared with that o f diamond (Ditchbum, 1965), its growth should cause a compressive
stress field in the surrounding diamond matrix, and the pressure produced would inhibit
extensive growth o f graphite. This possible explanation is consistent with the small sizes
o f graphite crystallites in diamond films as observed. From this point o f view, it is
predicted that the diamond matrix surrounding graphite crystallites should be in an internal
compressive stress state.
The graphite formation on the {111) planes o f diamond can be understood, since a
slight structural distortion can introduce {00.1 )-type o f graphite hexagonal basal planes on
the diamond cubic (111) planes (Badzian, 1987). The graphite crystallites on the {111}
diamond planes could grow easily along its a direction but difficult along its c direction
because o f its strong anisotropic nature of the chemical bondings. That could explain why
graphite is textured on the {1 1 1 } diamond planes, and only the {0 0 .2 }-type o f diffraction
arcs without the {10.0}-type diffraction could be observed. Graphite crystallites are
actually in a turbostratic disordered structure. The fact that the graphite structure is not
perfectly developed also is indicated by the Raman spectra which often show a broad peak
rather than a distinct 1580 cm ' 1 peak o f graphite as pointed out by Nemanich et al. (1988).
As graphite nucleates on the diamond surface, an associated stress field is developed in the
surrounding diamond matrix. An initially nucleated graphite crystallite should have an
influence on the nearby second graphite nucleus, most probably through the stress field. A
likely consequence of this interaction would be the same or close orientation between the
first and the second nucleus, which would minimize the structural energy. The second
nucleated crystallite would certainly affect the third nucleus, and so on, thus forming the
observed graphite crystallite domain distribution within the diamond matrix. Each domain
corresponds to a group of graphite crystallites lying on or close to a specific diamond
{111} plane as the dark field micrographs in Figure 5.4 show. The details o f this collective
interaction between the graphite nuclei are not clear at this stage.
As-the methane concentration is increased further (5%), the nucleation density and
growth rate o f both diamond and graphite further increase, while the hydrogen etching
capability decreases. Thus, more graphite crystallites should survive. Due to the high
supersaturation and consequently the strong driving force, a larger number o f graphite
crystallites nucleate at the same time and form with such a rate that the orientation o f the
later formed nuclei may not be affected by the early formed nuclei, such as through the
stress field interaction. It is conceivable that under this condition, graphite crystallites
would lose their preferred orientation along the diamond {111 } planes and have a high
nucleadon density with a random distribution in the diamond matrix. Another reason for the
observed random orientations o f graphite is that the diamond m atrix is composed of
increasingly disoriented crystals with sm aller dim ensions under the high methane
concentration condition. Therefore, graphite crystallites which form on the (111} diamond
planes would become random, as well.
The substrate temperature is another important parameter affecting the formation of
graphite. W hen the temperature is low (<900°C), the surface mobility o f carbon atoms is
low. Diamond and graphite crystallites nucleate with more difficulty, and consequently
more amorphous carbon phases are obtained. With an increasing temperature, graphite
formation is favored, as well as diamond crystallization. Carbon atoms would have higher
surface mobility which adjusts their positions to form more perfect crystals and seek
suitable areas on the surface for crystallization. Obviously, grain boundaries, cluster
intersections and other structural defects are preferred places for graphite nucleation and
growth. A t these sites, the structure is relatively loosely bonded, and any stress field due to
lattice mismatch could easily be accommodated and relieved. Thus, graphite crystallites
could grow uninhibited to large sizes as already shown in Figure 5.5a. Spitsyn and
Bouilov (1988) have indicated that the etching rate o f diamond by atomic hydrogen
increases with temperature, while the etching rate o f graphite depends little on temperature.
This etching behavior may contribute to the increasing amount o f graphite in diamond films
at high temperatures.
Different substrates have essentially the same effect on the formation o f graphite in
diamond films as far as the present TEM results show. The probable reason for the back
side o f diamond films having more graphite than the front side o f the films is that the
heteroepitaxy o f diamond films is difficult to achieve, that is, diamond crystals are not
easily nucleated on a foreign substrate surface because o f different lattice spacings and
different thermal expansion coefficients. Once initial diamond crystals are nucleated, the
subsequent diamond nucleation and growth would become easier, since it is equivalent to
an homoepitaxial process, i.e., diamond nucleating and growing on diamond itself. The
above reasoning is supported by recent results (Badzian, 1990), which indicate that both
sides o f a film which was deposited on a diamond powder seeded substrate have the same
structure. It appears that there is a transition layer with different structure and composition
from diamond between the film and the scratched foreign substrate as reported by Williams
and Glass (1989). The various impurity atoms coming from the substrates also may have a
significant effect on the formation of this intermediate layer.
5.2.5,2. Relationship between crystalline defects and graphite inclusions
One might expect to find a close relationship between planar defects o f twins and
stacking faults and graphite crystallites in diamond films, because both planar defects and
graphite form preferentially on the {111} planes o f diamond (section 5.4). However, the
TEM results suggest that this is not the case. N o evidence for any selective graphite
nucleation on such planar defects could be found. This leads us to the conclusion that the
planar defect occurrence and graphite formation may be two independent processes in the
film deposition. Further investigation is required to clarify this issue.
5.2.6. Summary
Graphite can be detected in diamond films deposited at high methane concentrations
in hydrogen (CH 4 %>2 ) by electron diffraction and dark field imaging techniques. In the
film deposited at a methane concentration of 2 %, graphite crystallites form on or close to
the {111 } planes of diamond and are in a highly oriented state with the hexagonal graphite
axis aligned with and close to the trigonal diamond axis. Related to the textured electron
diffraction pattern, graphite crystallites exhibit a domain distribution in the diamond matrix.
Each domain corresponds to a group o f graphite crystallites lying on or close to a particular
{111} plane o f diamond. In the film deposited at a methane concentration of 5%, graphite
crystallites are randomly distributed in the diamond matrix with no preferential orientation
o f the hexagonal basal planes with the diamond {111} planes. The wide range of
orientations o f graphite crystallites is associated with the disturbance produced by the rapid
formation o f new graphite crystallites and the small sizes o f increasingly disoriented
underlying diamond crystals.
Graphite exists in a turbostratic disordered structure with dimensions ranging from
50 A to 100 A in the a direction on the basal planes and 2 0 A to 50 A in the c direction. It is
believed that the graphite growth in the c direction is inhibited by an associated compressive
stress field developed in the diamond matrix which limits its further growth, and this stress
field leads to the texturing and the domain formation of graphite crystallites in the diamond
While low deposition temperatures introduce more amorphous phases in the films,
high temperatures favor the graphite formation with graphite tending to nucleate at
boundaries and to grow uninhibited to larger sizes, up to tens o f nanometers. The two sides
of free-standing diamond films which are deposited on scratched substrates have different
microstructures. The interfacial side has more graphite and other non-diamond components
than the growing surface. Different substrates of silicon, silica and tungsten carbide have
essentially the same effect on graphite formation as far as TEM is able to detect. Both air
plasma and hydrogen plasma can selectively etch the graphite second phase at the surface of
diamond films. The argon ion bombardment has no significant role in the nature o f graphite
formation in the diamond TEM foils at the scale observed in this study.
The graphite formation is believed to be a simultaneous nucleation and growth
process along with the diamond nucleation and growth on the surface, but the graphite
crystallites suffer a severe preferential etching from atomic hydrogen in the plasma.
Although diamond has a strong crystallographic influence on the orientation o f graphite
crystallites, apparently nucleating the graphite structure on its {1 1 1 } planes, no spatial
correlation could be made between the {1 1 1 } planar defects of twins and stacking faults
and the graphite nucleation in diamond films in this study.
5.3. Interfacial structure
The interface between a diamond thin film and a foreign substrate is a critical
structural feature which dictates the usefulness of the film by governing the adhesive
strength. There is a technological urgency to better understand the interfacial structure and
to improve the related film/substrate adhesion, since adhesive failures will likely limit many
applications. The w ork here contributes to such an effort. In this study, structural
characterization was performed on the interfaces o f diamond films grown on both silicon
and silica at two different methane concentrations in hydrogen. The study was aimed at
determining how the process o f initial diamond growth on silicon and silica substrates
differ and how the methane concentrations influence this process. Consequences on the
adhesion strength resulting from the interfacial structures are discussed
5.3.1. Literature review
An interface zone is a transition region between two different materials. It can be
atomic in scale for abrupt interfaces, or it can extend for tens or hundreds o f angstroms for
interfaces where the two materials react chemically. In the former cases where no chemical
reactions occur, the interface generally has been described as a cross-grid array o f misfit
dislocations. The lowest energy state is one in which the film contains a homogeneous
latent strain, so that the spacing difference of the film and substrate is reduced. However,
in the later cases where chemical reactions occur, the interface can be composed o f both a
structurally and compositionally different layer from the film and substrate material with an
appreciable thickness.
The adhesive strength depends critically on the extent o f both chemical and physical
interactions between the film and the substrate materials and the microstructure of the
interface region. Accordingly, the interface structure plays a very important role in
determinirig the adhesive strength o f the film. Better chemical bonding, more interface
contact and less internal stress can markedly improve the adhesion of the film.
The nature and conditions o f substrates strongly influence the nucleation and
growth of diamond films, which in turn, control the interface structure and its properties.
A s far as the internal stress is concerned, it is often observed that the CVD diamond
process leaves compressive stresses at the interface side o f the film, causing the entire film
to curl upward due to a resulting normal tensile stress toward the growth surface.
A carbide intermediate layer has been identified for the interface o f diamond-like
film s (Rabalais and Kasi, 1988; Kasi et al., 1988). Meilunas et al. (1989) studied early
stages o f the nucleation and growth of diamond films on silicon and molybdenum in
microwave plasmas using both X-ray diffraction and transmission infrared spectroscopy.
On both substrates a carbide intermediate layer was formed before the nucleation and
growth o f diamond. For molybdenum, the carbide layer (M 02 C) was estimated to be 1.5
pm thick, whereas on silicon the carbide layer thickness (SiC) was only about 100 A. The
growth of both carbide layers saturated at a point either when the surface was covered with
diamond (for M 02 C) or even before diamond fully covered the substrate surface (for SiC).
Belton et al. (1989) characterized the diamond nucleation and growth after different
intervals o f exposures o f a Si substrate to the growth environment using in-situ X-ray
photoelectron spectroscopy (XPS) without exposing the sample to air. It was found that
before exposure to the growth environment, the silicon surface was covered with a layer of
S i 0 2 and carbonaceous residue. No diamond particles were detected on the surface
pretreated with diamond paste. They identified three distinct phases o f diamond film
growth: removal o f initial substrate contaminants, formation o f a SiC layer and subsequent
nucleation and growth on the SiC layer. The SiC layer was estimated to be 90 A thick.
Williams and Glass (1989) used cross-section TEM and high resolution TEM to
analyze the interface o f diamond films on Si. A SiC layer (~50 A) was found for the film
deposited at a low methane concentration (0.3%), whereas no such intermediate layer was
found for the film deposited at a high methane concentration (2 %), presumably because the
layer was too thin (about 10 monolayers) to be imaged. The SiC layer formed at 0 .3 %CH 4
had an epitaxial relationship to the {111} oriented Si surface, and the diamond crystals
grown at 2 %CH 4 showed a twinned epitaxial relationship with Si. In a further study,
Williams et al. (1989) utilized Auger electron spectroscopy (AES) and XPS together with
cross section TEM to characterize the interface. Their XPS results showed that SiC grew
very rapidly as a layer in the first five minutes at both methane concentrations (0.3% and
2%), then diamond grew slowly as particles on the SiC layer. After two hours of growth,
the SiC phase was non-detectable.
There appears to exist a consensus that a carbide layer is formed before diamond
nucleation starts on a carbide-forming substrate. Diamond deposition on non-carbide
forming substrates generally has not been very successful. How this intermediate layer
affects the adhesive strength o f diamond films and how it correlates with the deposition
parameters are still not clear. So far, little work has been reported on this aspect
5.3.2. Experimental approach
The diamond films used in this study were deposited on two different substrates,
<100> oriented single crystal silicon wafer (Si) and fused silica glass (SiC^) at two
different methane concentrations in hydrogen (0.5-1% and 5%). Other parameters were
kept constant at a pressure of 90 Torr, a total gas flow rate o f 100 seem and a substrate
temperature o f 970-990°C. The substrate was immersed in the plasma during deposition.
To enhance the initial nucleation rate, the silicon substrate surface was pretreated by
abrasion with 1p m diamond paste, and the silica glass substrate was rubbed against a
microscope slide glass sprinkled with 1pm diamond powder. Both substrates were then
cleaned with methanol prior to deposition.
The diamond film/substrate interface can be studied by a number o f approaches.
For example, one can analyze deposits grown for short times to identify phases initially
formed using ex-situ, o r even better, in-situ techniques (Meilunas et al., 1989; Belton et
al., 1989). Also, one can study the interface structure o f thicker films with cross-section
electron microscopy (Williams and Glass, 1989). The approach taken here is to study the
substrate side o f a free-standing diamond film to gain insight into the interfacial structure.
The free-standing diamond films were peeled mechanically from the substrates.
Although in all cases separation apparently occurred between the film and the substrate
surface, the possibility o f disruption or even destruction o f the real interface could not be
ruled out. Some structural details o f the true interfacial region could have been lost.
Therefore, in a strict sense, the structural and compositional features obtained from the
substrate side of a free-standing film were more representative of the characteristics of the
near-interface region than the interface region. Regardless o f this possibility, however, the
substrate side of the free-standing film will hereafter be referred as the film interface, and
the plasma side of the film (also the growing side) will be referred as the film surface,
which was also studied for comparison purposes.
B oth sides o f these free-standing films were analyzed by a number of
characterization techniques. The electron microscopic techniques (SEM and TEM) were the
main characterization tools employed to reveal the interfacial structures. Diamond flakes
were sputtered with gold prior to SEM analysis. For TEM samples, one flake was thinned
from the interface side for analysis o f the surface of the films, and one was thinned from
the top surface to study the interface. Raman spectroscopy also was used.
5.3.3.. Rgsulls Diamond/silicon interface
Figure 5.9 is a series of SEM micrographs and corresponding Raman patterns for
diamond films deposited on Si substrates at two different methane concentrations (1% and
5%). The surface of the diamond films (Figure 5.9a) was rough, composed of protruding,
well defined, faceted crystals, whereas the interface was invariably smooth and had a
greater specular reflection and a shiny appearance as viewed with the naked eye or an
optical microscope. The morphology of the interface of the diamond film deposited at 1%
800 1000 1200 1400 1600 1800
2 pm
t —
- f-
I------------------ 1------------------»
800 1000 1200 1400 1600 1800
Figure 5.9. SEM micrographs and corresponding Raman patterns o f diamond films
deposited on Si substrates under conditions o f (a) 1% CH 4 , surface;
(b) 1% CH 4, interface; (c) 5% CH 4, surface; and (d) 5% CH 4 , interface.
800 1000 1200 1400 1600 1800
2 pm
800 1000 1200 1400 1600 1800
F ig u r e 5 .9 . (c o n t.)
methane concentration consisted o f truncated bases o f individual crystals which initially
nucleated on separate surface sites and then grew across the surface as well as upward
(Figure 5.9b). Under certain conditions, they grew into single crystal needles which
comprised the whole film as delineated by the air plasma etching to remove the loosely
bonded carbon between the columnar needles (Sato and Kamo, 1989). There were voids
around these crystal bases due to the supply shortage of carbon atoms to the interface area
when crystals grew upward and impinged together, cutting off the supply of carbon to the
bases, and hence leaving voids between crystals in the interface of the film. These voids
also have been characterized by real-time and spectroscopic ellipsometry (Collins et al.,
1989). No distinct intermediate layer was found in these SEM micrographs for the diamond
films grown on single crystal Si substrates. Even by using the cross-sectional TEM
technique, the possible intermediate layer could not be detected (Figure 5.10), at least
within the resolution limit of the microscope and for the sample studied. Diamond appeared
to nucleate directly on the Si surface and formed large, flat contact areas with the surface.
These films exhibited a better adhesion strength than films grown on Si0 2 , although voids
were present in the interfaces.
The same general picture appeared for the interface o f the film deposited at 5%
methane concentration (Figure 5.9d). However, the edges of the crystal bases were not so
sharp and angular, rather they were diffused and rounded, indicating a poorly developed
crystal structure. It is apparent that these bases were not single crystal bases but were
polycrystalline aggregates. Again voids were formed between these ball-like bases.
An interesting feature o f these interfaces, which was not found at the interfaces of
the films grown on Si 0 2 substrates, is that there were a lot o f scratch marks at the
interfaces. These are believed to be replicas o f scratch patterns from the Si substrate surface
which were transferred to the interface during the growth process. The scratch marks were
Diamond Him
Diamond film
Silicon substrate
Silicon substrate
Figure 5.10. G oss-sectional TEM micrographs o f diamond film s deposited on Si
substrates under conditions o f (a) 0.5% CH 4, 50 Torr, and (b) 2% CH 4 ,
90 Torr. No intermediate layer was found.
considered to be local hills on the crystals grown in the scratch grooves on the substrate
surface. The loosely bonded intermediate layer on the SiC>2 surface which was composed
of very fine particles obscured such replica patterns.
There was also a crystal size change associated with the methane concentration.
Increasing the methane concentration from 1% to 5% caused a dramatic decrease o f crystal
sizes. The sizes o f most crystals on the surface of the film deposited at 5% methane
concentration (Figure 5.9c) were so small that their crystal facets could no longer be seen.
This phenomenon was related to the high nucleation rates at the high methane concentration
The Raman spectra showed that the surface o f the film deposited at 1% methane
concentration (Figure 5.9a) had a strong 1332 cm ' 1 diamond peak and a small broad hump
at slightly high wavenumbers resulting from non-diamond phases. The top surface o f the
film deposited at 5% methane concentration (Figure 5.9c) had a much higher non-diamond
peak relative to the diamond peak, suggesting that a greater amount of non-diamond
components were introduced in the film. This less perfect structure o f diamond films
obtained at higher methane concentrations has been widely observed and reported in the
literature as well as in this thesis.
It is interesting to compare the Raman spectra from the interfaces formed at different
methane concentrations. The interface of the film deposited at 1% methane concentration
(Figure 5.9b) had an almost identical Raman pattern to that from the surface. The spectrum
from the interface of the film deposited at 5% methane concentration (Figure 5.9d),
however, was different from that obtained from the top surface. It had distinct graphitic
peaks at 1565 cm -1 and 1354 cm-1. The diamond peak was shifted and broadened,
indicating a more poorly developed structure in the interface than on the surface.
To obtain more detailed structural information, electron diffraction was performed
on both films (Figure 5.11). The diamond phase dominated the diffraction pattern from
both the top surface and the interface. The possible (3-SiC phase could not be detected from
the interfaces of either film by this technique. Only a diffused graphite {00.2} diffraction
ring was seen for the interface of the film deposited at 5% methane concentration. The
crystals at this interface had a ball-like morphology (Figure 5.12b). These balls were
polycrystalline, and possibly polyphase, aggregates. They appeared to preferentially
nucleate along the mechanical scratches.The crystals in the interface o f the film deposited at
1% methane concentration exhibited extensive growth features which extended radially
from an apparent initial nucleation site (Figure 5.12a). Diamond/silica interface
The interfaces exhibit considerable differences between the films grown on SiC>2
and Si. For the deposits on vitreous silica, a distinct intermediate layer with a fine
particulate structure formed on the substrate surface before large diamond crystals grew
(Figures 5.13 and 5.14). During peeling, separation occurred between this intermediate
layer and the SiC>2 substrate. This intermediate layer was loosely bonded and had a
thickness of ~2500 A for the film deposited at 0.5% methane concentration. The Raman
spectrum from this intermediate layer was similar to that from the surface (Figures 5.14a
and 5.14b), indicating that it was composed simply of fine sized diamond crystallites. The
intermediate layer of the film deposited at 5% methane concentration had a thickness similar
to that of the film deposited at 0.5% methane concentration and consisted o f more densely
packed, smaller particles (Figure 5.14d). The Raman pattern indicated that this layer had a
non-diamond carbon structure. Since Raman spectroscopy is much more sensitive to non­
diamond carbon than to diamond, the possibility of at least a certain percentage of this
interface consisting o f diamond can not be discounted. The formation of these intermediate
2000 A
1000A ------
2 000 A
Figure 5.11. TEM bright field images and electron diffraction patterns of diamond films
deposited on Si substrates at (a) 1% CH4 , surface, diamond diffraction;
(b) 1% CH 4 , interface, diamond diffraction; (c) 5% CH4 , surface, diamond
diffraction; and (d) 5% CH4, interface, diamond plus diffused graphite
1000 A
1000 A
deposited on Si substrates at (a) 1% CH4 and (b) 5% CH4
2500 A
9 Hm
Figure 5.13. A distinct intermediate layer exists in the interface o f a diamond film grown on
a S i02 substrate at 0.5% CH 4. The thickness of the intermediate layer is about
2500A, compared to the thickness o f the whole films, ~ 9 Jim .
800 1000 1200 1400 1600 1800
2 Hm
800 1000 1200 1400 1600 1800
Figure 5.14. SEM micrographs and corresponding Raman patterns o f diamond films
deposited on SiC>2 substrates under conditions o f (a) 0.5% CH 4, surface;
(b) 0.5% CH 4, interface; (c) 5% CH 4 , surface; and (d) 5% CH 4 , interface.
800 1000 1200 1400 1600 1800
2 Jim
V '> v»y> >., ‘V<- v i V>
V. > /^ v > V> -;»•' ♦;•*'»:*.
V -$
v£& .
|$ ^ § p M |-
M p ii^ ii^ S S u |:|* P ®
2 Hm
800 1000 1200 1400 1600 1800
Figure 5.14. (cont.)
layers composed of small cluster-like particles was presumably a necessary step to initiate
the growth of polycrystalline diamond films on the fused Si0 2 substrate which are non­
crystalline. The poof adhesion o f the films grown on Si02 substrates as reported by
Pickrell et al. (1990a) could be traced to these intermediate layers.
As described by Pickrell et al. (1990b), SIMS analysis indicated that the
intermediate layer contained various impurities picked up from contamination o f the glass
substrate (Na, Mg, Al, Ca) as well as Si, and that the concentration of these impurities
decreased with the sputtering depth into the sample. The carbon peak in the depth profile
indicated that initially the carbon was sputtered very quickly but then decreased to a
constant level, which was expected when sputtering through a weakly bonded layer o f
carbon to reach the more strongly bonded diamond film.
TEM micrographs and electron diffraction patterns of the intermediate layers formed
on SiC>2 are presented in Figure 5.15. The top surface invariably had larger crystal sizes
than the interface. However, the differences between the surface and the interface were
revealed not only in the apparent crystal sizes but also in the fine structure as shown in the
diffraction patterns. The later difference is more important, since it provides insight into the
nucleation and growth mechanisms. For the film deposited at 0.5% methane concentration,
the diffraction pattern of the interface clearly showed the presence of P-SiC phase in
addition to the diamond phase (Figure 5.15b). The diffraction pattern from the interface o f
the film deposited at 5% methane concentration (Figures 5.15d and 5.16) not only had a
strong P-SiC {111} diffraction ring but also a strong graphite {00.2} ring, indicating the
co-existence of three different phases: diamond, graphite and P-SiC. Table 5.1 gives the
values o f interplanar spacings and lattice constants for these three phases which were
calibrated by a polycrystalline gold film. The lattice constants of the three phases were very
close to their theoretical values. The {200} diffraction from P-SiC was not detected, even
2 0 0 0 A -----
2 0 0 0 A ----
Figure 5.15. TEM bright field images and electron diffraction patterns of diamond films
deposited on Si0
substrates under conditions of: (a) 0.5% CH 4 , surface,
diamond diffraction only; (b) 0.5% CH 4 , interface, diffractions from
diamond, 0-SiC and minor graphite; (c) 5% CH 4 , surface, diffractions from
diamond and minor graphite; and (d) 5% CH 4 , interface, diffractions from
diamond, (3-SiC and graphite.
Figure 5.16. Schematic diagram o f the diffraction pattern in Figure 5.15d obtained from the
interface of the diamond film grown on SiC>2 at 5% CH4 . Diffractions from
three major phases are indexed, vs = very strong; s = strong; w = week;
and vw = very week.
Table 5.1. Interplanar spacings and lattice constants o f diamond, graphite and p-SiC
EX P .2
JCPDS 1 E X P .2
2 .1 0
22 0
1 .0 0
cont(A )
a = 3.57 a = 3.63
a = 2.46
a = 4.36 a = 4.36
c = 6.71 c = 6.75
1 Source: JCPDS International Center for Diffraction Data, 1989
2 From the measurement of the SADpattern in Figure 5.15d calibrated by a polycrystalline Au film
under over-exposed conditions. A possible reason was that the distribution of Si and C
atoms in the p-SiC lattice was so random that the {200} diffraction was prohibited. For the
top surface of both films grown on Si02 (Figures 5.15a and 5.15c), no phases other than
diamond were detected in the electron diffraction patterns.
5 .3 .4.-PigCHSsion
Different types of substrates (Si and SiO ^ and methane concentrations in hydrogen
(CH4 %) are two important parameters governing the interface structures of CVD diamond
films. The chem istry o f the substrate surface influences the phase content and the
morphology o f the initial deposits. Early arriving carbon species can diffuse into the
structure o f many substrates (Joffreau et al., 1988). This diffusion process delays the
nucleation of diamond on the surface. It also can cause the formation o f an intermediate
carbide layer between the film and the substrate, if the two materials can react with each
other. Both Si and Si0 2 substrates can react with carbon to form SiC at elevated
temperatures. A carbide layer has been postulated to be essential for diamond growth and
for good adhesion strength (Badzian and Badzian, 1988; Joffrean et al., 1988). For non­
carbide-forming substrates such as nickel, only graphitic carbon was deposited (Rudder et
al., 1988).
There are three basic parameters o f a substrate influencing the interface structure:
crystal structure, atomic spacing and thermal expansion coefficient. A crystalline Si surface
supplies a good structural match to the diamond. However, the strong tendency o f chemical
reaction between Si and C atoms can lead to the formation o f a SiC phase on a Si surface.
Although a p-SiC intermediate layer was not found in this specific portion of the study due
to the destructive sample preparation manner, there is sound experimental evidence for pSiC formation in diamond deposition onto a Si substrate from various other studies. The
reported layer is thin, in the range up to 100 A. Since P-SiC is a better physical match to
the growing diamond than the Si lattice (Table 5.2), the nucleation of diamond is facilitated
on the carbide covered surface, and an epitaxial relationship could exist between P-SiC and
diamond (Badzian, 1987). By this mechanism, diamond grows across the Si surface on the
p-SiC layer from a nucleation site to form flat bases of crystal columns. The substantial
contact area and strong chemical bonding contribute to the good adhesion strength. As will
be presented later in Chapter 6 , results from the oxidation study indicates that the P-SiC
layer on Si may increase in its thickness with the sustained growth of diamond films via the
interdiffusion of carbon and silicon atoms through the interface. Therefore, it should be
easier to detect the P-SiC intermediate layer in a relatively thick diamond film.
SiC>2 , on the other hand, being an amorphous oxide, does not provide a structurally
compatible surface for either diamond or P-SiC growth. The carbide-forming reaction on
Si0 2 is more complex than on crystalline Si and involves the production of CO 2 and CO
gases. The extent of the reaction will be limited by the diffusion of gaseous species away
from the interface. As a result, only small particulate diamond, graphite and p-SiC were
formed in the intermediate layer. The high impurity concentration in the intermediate layer
also suggests that the chemical etching of the silica substrate surface plays a role in the
formation o f this fine particulate intermediate layer. A fine particulate intermediate layer
leads to an apparent smaller contact area with the substrate surface, which is responsible for
the poor adhesion o f the films on Si0 2 . Under typical conditions, diamond films grow
approximately three times faster on Si than on Si0 2 . This difference is most likely due to
the difference in the rate o f nucleation forming the intermediate layer, since once the
substrate is covered with diamond, the growth rate is expected to be similar.
The methane concentration in the gas phase controls the arrival rate o f carbon
species to the substrate surface. CVD diamond is a co-deposition process of diamond,
Table 5.2. Some physical properties o f diamond, Si, Si0 2 and P-SiC
Structure type
zinc blende
Lattice const
a = 2.46
a = 4.36
-0 .5
c = 6.71
Thermal exp.
coeff. (10-6K-1)
a = 1.3
c = 17
graphite and other non-diamond phases. However, the graphitic and other non-diamond
phases or related adsorbed species are effectively etched by the large amount o f atomic
hydrogen present in the gas phase. If the methane concentration is too high, the deposition
rate o f these non-diamond phases will exceed their gasification rate by atomic hydrogen.
These phases will be formed at the interface and trapped or retained in the bulk of the
diamond films. However, the structural quality improves with the thickness o f diamond
films because of the good physical and chemical compatibility of the existing growing
diamond surface with newly formed diamond nuclei.
This study also illustrates the applicability and usefulness o f the various
characterization techniques used for studying CVD diamond films. W ith Raman
spectroscopy, information on the diamond crystalline quality and phase content of the films
(including amorphous carbon) easily can be obtained on a micrometer scale. However, it is
not surface-specific, since the laser can penetrate substantially into the bulk o f the films.
Furthermore, the Raman scattering is extremely sensitive to sp2 type bonding, so that it can
swamp out any other signal, when this sp2 phase is in an appreciable concentration. TEM
and electron diffraction are essential for analyzing nanometer scale phase mixtures. It is
very sensitive to crystalline phase components. However, for non-crystalline phase, it
gives little information.
5.3.5. Summary
The structural characteristics of the interface or near-interface, including the phase
content and the morphology, strongly depend on both the substrate type and the methane
concentration in the gas phase. Diamond forms large, flat, contact areas with the Si
substrate, whereas on Si0 2 a particulate type of intermediate layer forms before the growth
o f diamond. The films grown on Si substrates exhibit better adhesion strength than films
grown on Si0 2 - At the low methane concentration (0.5-1%), only diamond is detected at
the interface of the film on Si, whereas the film on SiC>2 shows the presence of p-SiC in
addition to diamond. At the high methane concentration (5%), diamond, graphite and
amorphous carbon form at the interface of films on both substrates as well as P-SiC for the
film on silica. The absence of P-SiC layers on Si substrates presumably results from the
very thin nature o f carbide layers which are easily destroyed during the film removal
5.4. Structural imperfections
Structural defects exist extensively in CVD diamond. These defects have an
important effect on the properties o f diamond films. In this section, work on the
characterization o f structural defects, including planar defects (microtwins and stacking
faults), line defects (dislocations) and point defects (impurities and vacancies) by various
techniques will be shown. Particular interest was focused on the TEM study o f planar
defects. Mechanisms of the formation o f the crystalline imperfections are proposed, and
their influences on the properties of CVD diamond films are discussed.
5.4.1. Literature review Defects in natural and HP/HT synthesized diamonds
In 1962, Evans and Phaal (1962a) conducted a TEM study of internal imperfections
in natural diamond. They found three types of defects: straight dislocations and dipoles,
vacancy loops and plate-like precipitates lying on the {1 0 0 } planes which were discs of
substitutional nitrogen atoms. No stacking faults were reported.
Phaal and Zuidema (1966) were the first to report stacking faults as well as
dislocations in synthetic diamond. The concentration of stacking faults was much higher
than that of dislocations, and the dislocation density was at least one order of magnitude
lower than that in natural diamond. They indicated that these faults were formed as a result
of growth around metallic inclusions.
Lawn et al. (1965) observed stacking fault tetrahedra in a natural diamond using a
X-ray topographic technique. Later, Woods (1971) found microtwins at the sub-|im scale
in synthetic diamond, just as in various materials with fee (face centered cubic) or diamond
structure (Au, Al, Cu, and Si). M ade twins (triangular) and spinel twins ({111} twin
planes with <111> twin axis) are commonplace for diamond (Bovenkerk, 1961).
High concentrations of defects in natural and HP/HT-grown diamonds also have
been reported by Evans and Rainey (1975) and Walmsley and Lang (1983). The trigons on
the {111) faces (Varma, 1967) and tetragons on the {100} faces (Moore and Lang, 1972)
were clear indications o f defects on these faces. Frank and Lang (1965) and Evans (1965)
summarized the results o f X-ray topography and TEM studies on dislocations, stacking
faults and nitrogen platelets on the {100} planes. Lang (1979) also reviewed in detail the
internal structure of diamond including dislocations and stacking faults, mixed-habit
growth features, the relation of morphology to internal structure, and the nitrogen impurity
and its {100} platelet form as studied by X-ray topography, X-ray diffraction,
cathodoluminescence and TEM.
As for dislocations, Lang (1979) reported that in natural diamond with low
dislocation densities, most dislocations originated from the crystal nucleus. They appeared
to be very straight but did not necessarily followed low index crystallographic directions.
In diamond with a high dislocation density (~10 8 lines/cm2), the diamond was like a
mosaic crystal consisting of dislocations in irregular polygonal walls enclosing relatively
dislocation-free sub-grains. Dislocations in diamond also could be generated and
propagated by mechanical indentation or abrasion. In such cases, dislocations relieved
internal stresses.
Nearly all natural diamonds are luminescent under irradiation (UV, laser, electron
beam) with emissions ranging from blue to yellow. The activators are point defects, such
as vacancies and impurities, that produce donor or acceptor centers within the energy gap.
Adams and Payne (1974) used laser-stimulated fluorescence to investigate defect centers in
natural diamond. The luminescence spectra were very rich with many features due to
defects and impurities. Many peaks were produced or enhanced by radiation damage. Three
bands were observed at 520 nm, 665 nra, and 780 nm. The 520 nm band was believed to
originate from a defect o f monoclinic symmetry associated with the presence o f a high
concentration of nitrogen. Comprehensive reviews by Clark (1965), Walker (1979) and
Davies (1977) on luminescence of natural diamond are available.
M ost natural diamonds contain nitrogen and boron which can substitute for carbon
in the diamond lattice. Humble et al. (1985) obtained high resolution lattice images o f the
{100} nitrogen plates to study their structure. Other elements (Al, Be and O) also have
been found in natural diamond. Major impurities in synthetic diamond are solvent catalytic
metals, generally group VIII elements (Ni, Fe, Co, etc.) or an alloy thereof which are
trapped in the diamond lattice during growth. Impurities can appear either in the form of
dark inclusions or in a finely dispersed form. Cathodoluminescence (CL) indicated that the
{ 1 1 1 } faces of synthetic diamond absorb impurities more than the { 1 0 0 } faces
(Vishnevsky, 1975). Cathodoluminescence, combined with optical or electron microscopic
observations, also has revealed the distribution of crystal defects and growth sectors of
HP/HT synthetic diamond (Woods and Lang, 1975).
The nature o f vacancies in the diamond lattice appears much more complicated.
Clark et al. (1979) extensively reviewed results from studies of color centers and optical
properties o f diamond. Defects were identified from radiation-stimulated emission or
absorption lines which represent electronic transitions at defects. Important features which
represent the properties and behavior of a defect are the shape, intensity and energy of a
zero-phonon line, and the structure o f the phonon side-bands in the absorption and
emission spectra. However, when an electronic transition is coupled with a simultaneous
excitation o f various vibration modes, the fine structure becomes less well resolved.
Fortunately, the luminescence spectra of many point defects in diamond show sharp, zerophonon emission lines at the same energies as the corresponding zero-phonon absorption
lines. A comprehensive list of diamond defects identified by these methods has been given
by Davies (1977). Some examples are N 3 at 2.985 eV, GR1 at 1.673 eV, H 3 at 2.463 eV,
H 4 at 2.499 eV and ND1 at 3.15 eV.
The GR defect in diamond has received the most experimental and theoretical
attention. It has been observed in absorption, cathodoluminescence and photoluminescence
spectra. GR is either a neutral or a negatively charged vacant atomic site, the precise nature
o f which has not been identified unambiguously. Such a defect has a tetrahedral symmetry.
Recent experimental evidence and theoretical calculations favor the explanation that the
defect is a neutral vacancy. The GR1-8 transition lines correspond to the excitation o f the
neutral vacancy to various excited and charged states. Two zero-phonon lines associated
with GR1 are 1.665 eV and 1.674 eV. There is also a phonon-coupled broad band from
1.4-2.5 eV with a maximum at 2.0 eV. GR1 is reported to be stable up to 930°C. Defects in low pressure CVD diamond
Early in 1983, Matsumoto and Matsui used TEM to investigate the defect structures
of CVD diamond. They reported extensive microtwins formed in diamond crystals. These
defects are closely related to process parameters.
Williams and Glass (1989) found that majority of diamond crystals have a very
high defect density comprised of {111} twins, {111} stacking faults and dislocations. In a
recent paper (Williams et al., 1990), high resolution TEM was used to lattice image
individual defects (twins and stacking faults) and their intersections. It was found that the
five-fold multiply twinned particles accommodated the 7.5° misfit at the twin boundaries
rather than elastically deform. This created a twin boundary coincident with a low angle
grain boundary. The density o f defects in the particles was very high, whereas near the
twin boundaries the density o f defects was dramatically reduced.
Raman spectroscopy has been used widely to examine the structure of CVD
diamond films. The position, shape and width o f the diamond 1332 cm ' 1 peak supply
information about the crystal perfection and stress state of diamond. Strong luminescence
peaks as well as the background in Raman spectra also have been observed and related to
certain defect structures such as vacancies, dislocations and impurities.
Hartnett (1988) studied the luminescence behavior of CVD diamond films with a
Raman spectrometer. He found a broad luminescence background with a maximum at 2.02
eV and sharp luminescence peaks at 1.968 eV, 1.679 eV and 1.637 eV, with the 1.679 eV
peak the strongest. He suggested that the 1.679 eV line is related to the GR1 center under
stress and that the 1.637 eV peak is due to impurities gathered near the 1.679 eV center,
similar to the broadening of the GR1 line to lower energies due to nitrogen. He reported
that Si incorporation increased the 1.679 eV peak, whereas B-doping eliminated this peak
as well as the 2.02 eV luminescence feature. Later, Knight and White (1989) also reported
luminescence observations of CVD diamond films. Three narrow lines with a full width at
half maximum (FWHM) of 0.02-0.04 eV at 1.64, 1.68, and 1.77 eV and three broad
bands (0.3-0.5 eV FWHM) at 1.88,1.94, and 2.06 eV were identified.
Cathodoluminescence spectroscopy can show a w ider range of defects than
radiation excited methods such as laser excitation employed in Raman spectroscopy,
because electron beam excitation can not only provide the required powers for facilitating
luminescence detection in wide band gap materials such as diamond but also allow for
imaging the spatial distribution of the emitting centers. For CVD diamond films, Kawarada
et al. (1988) obtained visible cathodoluminescence spectra. The peaks occurred at different
energies between 2.4-2.8 eV (green to purple-blue) and are attributed to donor-acceptor
pair recombinations. The {111} and {100} growth sectors can be differentiated by the
intensity of luminescence.
The most common impurities contained in CVD diamond films are hydrogen and
silicon. The concentration of hydrogen has been reported to be about 500 ppm and 2500
ppm for MPECVD diamond films grown at methane concentrations of 1% and 5%,
respectively, as measured by the resonant nuclear reaction technique (Hartnett, 1988).
Other elements such as Na, Al, Ca, W, etc., also have been detected by secondary ion
mass spectrometry or neutron activation analysis, and presumably they result from the
contamination o f the growth environment or the etching of the substrate materials. Laser
Microprobe Mass Analysis spectroscopy (LAMMA) also has been used successfully to
detect impurities in diamond films (Matsumoto et al., 1985).
5.4.2. Experimental approach
TEM was the main technique used to characterize the various lattice defects. The
cross-section TEM method, as well as the plan-view TEM method, have proven to be
powerful techniques for the study of such types of defects as stacking faults, twins and
dislocations. To obtain information on point defects, IR spectroscopy, SIMS and laser
excited luminescence spectroscopy were used.
5.4.3. Planar defects TEM results
Figure 5.17 shows the main features of diamond films under TEM observation.
They consist o f a large number o f stacking faults and microtwins. With an increase of the
methane concentration during the film deposition, the density of defects increased, and their
sizes became smaller, about lOOOA, 500A and
100A for methane concentrations of 0.5%,
2% and 5%, respectively, as shown in Figures 5.18. These defect dimensions were related
100 A
Figure 5.17. TEM micrographs o f stacking faults (a, b) and twins (c, d) in CVD diamond
films. The film in (a) and (d) was deposited at 2% CH4 , and the film in (b)
and (c) was deposited at 0.5% CH 4 .
Figure 5.18. TEM micrographs of planar defects in CVD diamond films deposited at
different methane concentrations: (a) 0.5% CH 4 ; (b) 2% CH 4; and
(c) 5% CH4.
to the interspacings between boundaries of the planar defects. This observed behavior was
apparently related to the high nucleation rate and consequently the small crystal size at high
carbon supersaturations during growth. The planar defects occurred on the {111} planes as
indicated in Figure 5.19a, in which the additional spots cutting the main diffraction spots
by one-third along the < 111> directions represent the presence of twinning on the {111 }
planes (Hirsch et al., 1965), and streaked lines joining the main spots in the <111>
directions indicate stacking faults o f random repetition of the {111} planes. Also shown in
the figure are a bright field image (Figure 5.19b) and dark field images (Figures 5.19c and
5.19d) which reveal the relationship between the matrix and the twins. Figure 5.19c is a
dark field image obtained from the (111) matrix spot (M). This shows the matrix (111)
orientation as bright areas and the twins as dark areas. Figure 5.19d was obtained from a
diffraction spot (T) which was due to twinning on the (111) plane and showed those twins
as bright areas and the matrix as dark areas. The stacking faults on different {111} planes
could interact with each other as illustrated in Figure 5.17b and indicated by arrows. Such
kind o f interactions between stacking faults has been studied earlier by Matthews (1959). Formation mechanisms o f planar defects
It is apparent that stacking faults and twins are important structural imperfections in
diamond films. In the diamond structure (Figure 5.20a), the stacking sequence in the
< 111> direction is AocB pCyAaB pCy, or simply ABCABC, as is shown in Figure 5.20b.
If there are some mistakes in the stacking sequence, intrinsic and extrinsic stacking faults
will occur on the {111} planes as shown in Figure 5.20c. These kinds o f stacking faults of
the {1 1 1 } planes can form easily in a deposition process due to a number o f possible
reasons: misfit between the deposit and substrate lattice (Matthews, 1962); separation of
unit dislocations into partials (Phillips, 1960); collapsed vacancy disks (Jaccodine, 1963);
1000 A
1000 A
Figure 5.19. An electron diffraction pattern and TEM micrographs of a microtwin:
(a) the diffraction pattern within a selected area about 0.8 mm in diameter,
(b) a bright field image; (c) a dark field image from the matrix spot
M (1 1 1 ); and (d) a dark field image from the spot T which is due to
twinning on the (1 1 1 ) plane. The imaged diamond crystal was
deposited at 0.5% CH 4 .
Figure 5.20. The diamond structure and its {111} stacking sequences: (a) The cubic unit
cell of diamond structure; (b) its {111 } planes; and (c) a projection o f
stacking sequences of the { 1 1 1 } planes in diamond on the ( 110 ) plane,
where (I) is an intrinsic stacking fault, (II) is an extrinsic stacking fault,
and (HI) is a monolayer twin.
Figure 5.20. (cont.)
and nucleation from slip lines, microscratches, regions o f impurity segregation and local
stress developed during the growth process (Mendelson, 1964). In epitaxial silicon films,
the { 11 1 ) stacking faults typically are tetrahedra which form when the faulted crystals
grow and meet with the correctly deposited crystals (Booker and Stickler, 1962). Although
they also were found in diamond, stacking faults in the tetrahedral form are not common.
Stacking faults in diamond films usually appear as ribbons bounded by Shockley partial
dislocations. The substrate surface usually is not perfect, because it is mechanically
scratched; thus, the formation of small areas of stacking faults on the {111 ) planes is most
likely to start at the interface. It is expected that the degree o f perfection of the substrate
surface will influence the occurrence of the defects in the films. However, mechanical
polishing of the substrate surface appears to enhance the nucleation density of diamond
films. Therefore, a compromise between the high nucleation rate and the structural
perfection of diamond films is likely. The fact that diamond films deposited at higher
methane concentrations, which cause higher nucleation densities and higher growth rates,
have higher defect densities supports the above consideration. Badzian and Badzian (1988)
proposed that the mechanically damaged substrate surface could introduce a buffer layer of
(3-SiC between the silicon substrate and the diamond film. This layer was found by
Williams and Glass (1989). On the other hand, the defects also can be initiated within the
films by impurity atoms which are likely to be deposited in incorrect sequence, creating a
nucleation center for a faulted crystal. The intrinsic and extrinsic {111} stacking faults can
introduce monolayer stacking sequences of hexagonal diamond (Lonsdaleite) and twins,
possibly leading to the formation of Lonsdaleite stacking and twin structures in cubic
Twinning on the {111} planes of diamond corresponds to a very small amount of
twin energy, since little structural disturbance is created (Figure 5.21). The twin structure is
Figure 5.21. A projection o f a twinned diamond structure on the (110) plane. K i= (l 11),
and T| j= [1 1 2]. The large filled circles are atoms in the (110) plane passing
through the center of the unit cell; the small filled ones are atoms in a ( 110 )
plane separated from the first by a quarter o f a diagonal; the open circles are
initial positions; and the circles with lines represent atoms after the twinning
displacement. (Source: Klassen-Neklyudova, 1964)
closely related to the stacking faults as can be seen in the faulted stacking sequence where a
monolayer twin is formed (Figure 5.20c). In the SAD pattern (Figure 5.19a), streaks
around the main diffraction spots and twin spots indicate that either the stacking faults are
present, or the twins are very thin. Therefore, twins and stacking faults often have common
origins, and it should not be surprising that twinning occurs so extensively in diamond
films. External crystal shapes examined by SEM also often display this twinning behavior
(Zhu et al., 1990). The strong tendency to form such kinds o f planar defects is a major
obstacle to the growth of single crystals of diamond.
5.4.4. Dislocations
Dislocation lines and end-on dislocations (perpendicular to the surface) are shown
in Figure 5.22. Diamond usually has its dislocations aligned along the <110> directions on
the {11 1 } close-packed planes to minimize its energy because of the strong directional
nature o f valence bonds. The Burgers vectors of unit dislocations in the diamond structure
are also in <110> directions. It is obvious that in most cases they are screw dislocations or
60° dislocations. The atomic models of a 60° dislocation and a screw dislocation in the
diamond structure are shown in Figure 5.23. It can be seen that there is an incompleted
atom plane associated with the 60° dislocation (Figure 5.23a). The atoms on the edge of
this half plane have unsaturated dangling bonds which could be donors or acceptors of
electrons, thus possibly establishing space charge areas around them. The structure along
the dislocation line is relatively open and appears like a channel. This kind of dislocation
line could increase phonon and electron scattering and, for instance, act as recombination
centers for minority carriers in semiconductors. Furthermore, they could absorb impurity
atoms and be the paths for diffusion o f these impurities. Such possible effects should be of
concern in making diamond semiconductor devices.
Figure 5.22. TEM micrographs of (a) dislocation lines and (b) aligned end on dislocations,
as indicated by arrows, in a diamond Him deposited at 0.5% CH 4 .
Figure 5.23. Atomic configurations of (a) a 60° dislocation and (b) a screw dislocation in
the diamond structure (Source: Hu and Qian, 1980)
5.4.5. Point defects
The nature o f point defects is very important if diamond is to be used as a
semiconductor or an optoelectronic material. In this section, results from three
characterization techniques, namely, IR spectroscopy for impurity chemical bondings,
SIMS for impurities, and laser excited luminescence for impurities, vacancies or their
combinations, will be presented to gain insight into the point defect centers in CVD
diamond and their correlations with deposition parameters. IR spectroscopy
IR spectroscopy mainly was used to detect the possible IR absorptions by C-H
bonds and C-Si bonds, thus to examine the presence o f hydrogen and silicon atoms in
diamond films. Both transmittance and reflectance spectra in the range from 800 to 4000
cm -1 were measured (Figure 5.24) from two polished diamond films, of which both the top
growth surface and the surface against the substrate were smooth (Wang et al., 1990). The
intrinsic absorption o f diamond (two-phonon peak) around 2000 cm ' 1 can not be resolved
because o f the insufficient absorption due to the small thickness of the films (~20 |im ). An
absorption feature at about 2900 cm -1 can be observed in the transmission spectra which is
due to carbon-hydrogen (C-H) stretching involving carbon with sp3 hybridized bonding.
There are no absorptions from sp 1 and sp2 bonded carbon structures. An absorption band
near 800 cm -1 corresponds to the stretching vibration of silicon carbide, apparently formed
at the interface between the diamond film and the silicon substrate. Other impurities such as
nitrogen o r oxygen, the main cause of absorption in the 900-1600 cm -1 region for HP/HT
synthesized diamond (Badzian et al., 1986), were not detected in CVD diamond films.
Since the control o f nitrogen and oxygen impurities is possible in the CVD diamond
process, at least the absorption in this range can be eliminated.
j2 55
40 0 0
30 0 0
Wavenumber (cm-1)
= 30
<r 20
3 0 00
Wavtnumber (cm’ 1)
Figure 5.24. Infrared spectra o f polished diamond films: (a) transmission spectra and
(b) reflection spectra. Sample X I was 21 |im thick deposited at 0.5% CH 4 ,
and sample X2 was 16.5 |im thick deposited at 5% CH 4.
202 SIMS results
Both the surface composition and the depth profiling of diamond films were
measured by SIMS as shown in Figure 5.25. Apart from C, H and Si, as well as SiC and
SiO, which were somewhat expected, there were strong signals from Na, Mg, Al, K and
Ca which are presumably resulted from the contaminations of the Si substrate, the graphite
susceptor, or the CVD reactor tube wall. It should be noted that the two diamond films
measured in Figures 5.25a and 5.25b were deposited under the same conditions, but the
growth rate o f the film in (a) was three times higher than that o f the film in (b). No
additional information can be deduced from the SIMS measurements, except that the
concentrations o f Na and K were much higher in the film (a) than in the film (b). The
reason for the growth rate disparity is unknown.
It is seen from the depth profile (Figure 5.25c) that the concentrations o f C and H
are essentially constant in the profiled thickness (0-60 nm). However, the concentrations o f
Si and SiC increase with the depth from the surface, suggesting that Si plays a more
important role in the deposition of diamond films at the interface than at the top surface. Laser induced luminescence spectroscopy
The luminescence spectra were excited at room temperature with an argon ion laser
(514.532 nm) and scanned through the range from 400-8000 c m '1. Alm ost all CVD
diamond films exhibit strong luminescence features as shown in Figure 5.26. A peak at
about 1.678 eV (738.4 nm) (0.02 eV FWHM) is veiy intense (Figure 5.26a); this line is
very close to the GR1 line 1.674 eV common in natural diamonds and associated with
neutral vacancies. Many deposition conditions were changed to examine the behavior o f
this peak. It was found that the peak is relatively independent o f deposition methods
(MPECVD, HFCVD, or oxy-acetylene flame); gas flow patterns (the normal introduction
L°9 C intensity (c/s)]
20 25 30 35 40 45
O -beam 5^.
Log [Intensity (c/s)]
20 25 30 35 40 45
Figure S.2S. SIMS spectra o f diamond films: (a) and (b) surface mass spectra; and (c) a depth profile.
The films in (a) and (b) were deposited under the same typical MPECVD conditions of
1% CH4 ,1000°C, 90 Torr and 100 seem, and the film in (c) was deposited at 5% CH4
with a thickness o f 2 .2 |im.
of gases at the top o f the reactor vs. the direct introduction o f gases close to the substrate
surface through a nozzle configuration); types o f substrate (Si, Mo, Cu, or graphite);
substrate orientations ( S i< lll> vs. Si<100>); diamond crystallographic faces (triangular
{111 } vs. cubic {100 }); surface morphologies (continuous film-like vs. individual grain­
like); noble gas additions (He, Ne, or Ar); the partial surface oxidation (500-1000°C in a
gas of 0 .0 1 % 0 2 +Ar); and the graphitic phase content in the films related to the methane
concentrations in the gas phase. However, the 1.678 peak could be eliminated when
hexagonal BN substrate was used (Figure 5.26c). Thus, it is suspected that B atoms
incorporated in the diamond lattice effectively passivate the defects responsible for the
1.678 eV peak.
There exists also a small satellite peak on the lower energy side of this strong peak
at about 1.637 eV (757.4 nm), but the relative intensities o f the 1.637 eV and 1.678 eV
lines can vary with the deposition conditions o f the films and the structural quality.
Therefore, they are considered as two different defect centers. In addition, the 1.637 peak
is always related to another small peak at 1.968 eV (630.3 nm). The 1.678 eV peak can
become weak, and both the 1.637 and 1.968 peaks can disappear when the films contain a
relatively large amount of graphitic phases (Figure 5.26b). In such cases, additional broad
luminescence features around 2.059 eV (602.0 nm) and a small peak at 1.767 eV (701.0
nm) can appear.
Badzian et al. (1988a) have ascribed the 1.678 eV luminescence peak to the
incorporation of Si in octahedral holes in the diamond lattice. It seems that Si is not the only
factor determining the appearance of the strong 1.678 eV luminescence peak. Because these
luminescence features consistently appear over a wide range o f deposition conditions and
deposition methods as outlined above, they are most probably due to commonly occurring
impurities and defects which are inherent to the low pressure CVD processes. Also it is
Wavenumber (cm-1)
Figure 5.26. Luminescence spectra of diamond films deposited at: (a) 1% CH4 on Si;
(b) 4.5% CH 4 on Si; and (c) 1% CH 4 on hexagonal BN by HFCVD.
likely that the defects responsible for the strong luminescence in CVD diamond Alms are
different-from GR1 neutral vacancies in natural diamond in size, symmetry and
composition. A precise assignment o f these peaks to specific defects is not possible at this
time. The polycrystalline nature and graphitic phase inclusions complicate the study due to
internal reflections, extra absorptions and strain. Low temperature luminescence spectra
and decay times o f these luminescent defects need to be measured.
5.4.6. Relationship between crystalline defects and surface morphologies
The internal crystal defects are closely related to the external surface morphologies
by the growth history. For example, by examining the internal structures of the natural,
rounded dodecahedral diamond (growth bands and dislocations), it was determined to be a
dissolution form of octahedral diamond rather than a growth form (Moore and Lang,
1974). In the case of natural diamond of cubic habit, two distinct internal structures have
been discovered (Moore and Lang, 1972). One was a fibrous (or columnar) pattern in the
< 111> directions with branching and equal growth velocities in symmetrically equivalent
directions. The cubic habit resulted from the branching mode of growth. The other was a
truncated cuboid pattern. The growth horizons and dislocation trajectories showed that the
cube was developed with a combination of forms: normal faceted growth of the {1 1 1 }
faces plus non-faceted, hummocky growth o f non-crystallographic faces of a mean
orientation {100}. Both structures provide experimental bases for proposing growth
models of cubic diamond. For synthetic diamond, the history of relative development of
forms during growth also can be read from the internal structural features such as the
growth facet traces, dislocation trajectories and growth sector boundaries. Therefore, an
adequate understanding o f this relation would be useful in the analysis o f the growth
behavior o f CVD diamond films and in predicting growth morphology and crystal quality
under a wide range of experimental conditions. However, this relation between the internal
defective structure and the external morphology generally have been neglected. Recently,
Clausing et al. (1989) studied the relationship between the internal and external growth
features for HFCVD-grown diamond films.
In this study, diamond films with typical surface morphologies of the {100} square
faces and the {111 } triangular faces and a film with a cauliflower morphology composed of
small crystal aggregates were investigated by TEM. The { lll} -faceted films showed
extensive twinning and stacking fault features which are mostly visible on the {111 } planes
with rare twins on the {100} planes (Figure 5.27a). The crystal shown here was cyclic
twinned with pseudo five-fold symmetry. Since there is only a relatively slight energy
difference between the normal and twinned diamond structures, diamond is predisposed
toward twinning. The diamond deposition is believed to proceed via frequent twinning
growth. If these twinning bands extend and intersect with the surface, striking twinning
features will be recorded in the final morphology, and this has been frequently observed.
However, unlike the internal structure of natural diamond crystals of cubic habit which
possess fibrous or cuboid features (Moore and Lang, 1972), the crystals in the {100}oriented films are planar-defect free (Figure 5.27b). Nanometer scale roughness could be
discerned on the {100} square faces. These features are interpreted as the sites where
atoms could attach and result in growth in the normal direction. Therefore, the roughness is
an indication of the relative ease with which the crystal growth could proceed. Along the
boundaries o f the {100 }-oriented crystals where the {111 } planes lie, planar defect features
again could be detected extensively. For the film with cauliflower morphology deposited at
a high methane concentration, no obvious growth features other than small defective
crystals were present (Figure 5.27c). The defective internal structure of the {111}-oriented
films could seriously degrade the optical properties o f the diamond films as Badzian has
2 00 0 A
Figure 5.27. SEM and TEM micrographs of the surface morphologies and corresponding
internal structures o f (a) a diamond crystal with the {1 1 1 } surface
morphology which is highly defective; (b) a diamond crystal with the {100 }
surface morphology which is planar defect-free; and (c) a film with a
cauliflower morphology containing numerous defective small crystals.
2000 A
1000 A
Figure 5.27. (cont.)
recently found in his homoepitaxial experiments (Badzian, 1990).
5.4.7. Discussion
Crystalline defects like stacking faults, microtwins and dislocations, as well as
second phase, non-diamond components such as graphitic and amorphous carbon, are
common features in CVD diamond films because of the non-equilibrium nature o f the
process. They can degrade film performances and, therefore, are critical to many
applications. In electronics and optoelectronics, impurities, grain boundaries, lattice defects
and spurious phases will determine device performance, because they will affect carrier
generation, mobility, lifetimes, recombination, scattering, trapping, and radiative quantum
efficiency. Indeed, the thermal and electrical properties o f CVD diamond films have been
found to be inferior to those o f natural diamonds (Ono et al., 1986; Gildenblat et al.,
1988). This is due largely to the extensive existence o f various structural defects.
Therefore, they should be o f great concern when considering practical applications of
diamond films.
There exists a close relationship between the internal structure and the external
morphology. It provides valuable clues in elucidating the growth mechanisms o f CVD
diamond. It also helps in designing experimental conditions to develop relatively defect-free
structures o f diamond films by choosing the { 100 } faces as the preferential developing
faces. This has been verified experimentally in the homoepitaxial diamond deposition
process (Geis, 1989; Fujimori et al., 1989; Badzian, 1990). Perfect diamond single
crystals can be epitaxially grown on the {1 0 0 } diamond surfaces, whereas highly defective
structures are developed on the {111} oriented diamond substrates. However, the growth
rate is much faster on the {111} surfaces than on the {100} surfaces. The most obvious
reason is that the {111 } surfaces have a much higher defect density, and these defects
subsequently act as nucleation and growth sites.
5.4.8. Summary
CVD diamond films consist o f a large number of microtwins and stacking faults
which occur on the {111 } planes o f diamond and correspond to a very low distortion
energy structure. With an increase of methane concentration during the film deposition, the
density o f these defects increases, and their sizes become smaller. The {lll} -faceted
crystals were usually highly defective, containing many planar defects. However, the
{100}-oriented crystals were generally perfect with no detectable planar defects. CVD
diamond also contains extensive dislocations. Various impurities from etching of the
reactor tube and the substrate also can be incorporated easily in diamond films. These
structural imperfections can degrade many physical properties o f diamond films and are
obstacles in growing single crystals o f diamond.
Chapter 6
6.1. Introduction
This chapter will address the oxidation properties of diamond films at elevated
temperatures. The oxidation experiments were used as a tool to reveal the structural defects
o f CVD diamond. The behavior o f diamond films in a series o f oxygen-containing
atmospheres (0 .01 - 1 %0 2 +Ar and air) was investigated by thermogravimetric analysis
(TGA), electron microscopies (SEM and TEM) and Raman spectroscopy. Experimental
data on oxidation kinetics, oxidation mechanisms, and structural and morphological
consequences were obtained. Annealing of diamond films in hydrogen, argon and nitrogen
atmospheres was also investigated.
6.2. Literature review
The carbon-oxygen reaction system has been extensively studied with the carbon
ranging in structures from essentially amorphous to single crystals o f graphite as reviewed
by Walker et al. (1959). The porosity o f the carbon specimen was always a factor in
complicating the interpretation o f experimental results. By studying the diamond-oxygen
reaction, this difficulty was expected to be overcome. However, since a layer o f free
carbon often forms on the diamond surface as a result of oxidation, diffusion effects again
become very important
It has long been known that if a diamond is heated to a temperature o f about 600700°C or higher (depending on the initial quality of the diamond) in pure oxygen or even in
air, it will be wholly or partially oxidized to CO and CO 2 . If diamond is heated in the
absence o f oxygen, preferably in vacuum, to a temperature o f about 1400-1500°C or
higher (again depending on the quality of the diamond), it will be wholly or partially
converted to graphite (Lonsdale and Milledge, 1965).
Lambert (1936) has reported an activation energy o f 43 kcal/mol for the oxidation
o f diamond powders. Later, Evans and Phaal (1962b) measured oxidation rates of low
index faces o f diamond single crystals in a fast flowing stream of oxygen. The reaction was
found to be independent o f the flow rate (500-1500 cm/s) in a temperature range of 6501350°C but sensitive to the crystal orientation and the oxygen partial pressure. The
oxidation rate increased linearly with oxygen partial pressures from 0.05 to 0.5 Torr.
Below 1000°C, the {111} surfaces had the highest oxidation rate, followed by the {110}
surfaces, with the {100} surfaces having the lowest rate. The {111} and {110} faces had
thin coatings of graphitic carbon on their surfaces at all the temperatures in an oxygen
atmosphere o f 0.4 Torr. Activation energies of 44 kcal/mol between 650-750°C and 23
kcal/mol between 750-1000°C were measured. The {100} faces were not always covered
with a carbon layer. It was completely free of visible surface carbon layer between 650850°C, and above 850°C the carbon layer appeared. Activation energies o f 55 kcal/mol
were measured between 650°C and 750°C and 37 kcal/mol between 850°C and 1000°C.
At a pressure of 1 atm, an activation energy o f 55 kcal/mol was measured for the {111}
and {110} faces reacting with oxygen between 600-700°C without the surface carbon layer
formed. Therefore, the true activation energy was considered to be 55 kcal/mol, because
other values were strongly affected by the diffusion of oxygen through the carbon layer to
the underlying diamond surface.
The diamond surface of {111} or {110} was attacked uniformly, resulting in a matt
type of surface. The oxidation was more concentrated on steps (kinetically favored) than on
flat surfaces. Therefore, the reaction rate for the {100} faces will be more structure
sensitive than the {111 } or {110 } faces as it would depend on the density of the source of
steps such as microcracks, dislocations and possible impurities in the lattice. All surface
atoms were reactive on the { 111 } or {1 1 0 } faces, but reactive sites on the {100 } faces
were located primarily on steps.
They further proposed that the diamond-oxygen reaction is not simply oxidation of
diamond to gaseous CO 2 or CO, but involves an intermediate process in which a carbon
layer forms on the diamond surface. Whether or not a visible surface carbon layer was
formed depended on the relative rates of formation at the diamond surface o f the carbon
layer and its rate of removal by oxidation. Both these processes had their individual
dependence on the oxygen presssure. They also found that water vapor and chlorine
inhibited the diamond-oxygen reactions.
For CVD diamond films, oxygen is known to be an effective etchant o f non­
diamond phases. This function has been demonstrated both during the deposition process
and in post-deposition oxidation. Sato and Kamo (1989) used air plasmas to etch diamond
films, resulting in a selective oxidation o f grain boundaries and defect etching in those
grains with high defect densities. Needle-like columnar grains o f single crystal diamond
were revealed.
Plano et al. (1989) reported oxidation of diamond films (together with the silicon
substrate) in a 22%C>2+78%N2 atmosphere at 0.5 atm in a temperature range o f 600750°C. They found that diamond films oxidized in a similar manner to, but at a slower rate,
than natural diamond due to the selective etching behavior (sp2 bonding vs. sp 3 bonding).
Diamond transformed to graphite after sufficient oxidation. The activation energy was
found to be about 41 kcal/mol between 600-750°C.
Johnson et al. (1989) conducted a TGA study o f diamond powders (<0.5 |im ),
graphite powders (2-20 |im), and CVD diamond films in air at 760 Torr and 50 Torr. They
found that diamond powders oxidized at lower temperatures due to the high surface area,
and graphite powders oxidized at higher temperatures, while diamond films oxidized at
intermediate temperatures. The oxidation o f diamond powders was limited by the oxygen
concentrations, and the oxidation of CVD films was limited by the surface area exposure to
the oxidizing atmosphere. The oxidation rate o f diamond films was approximately 0.1
pm/h at 600°C and 4 pm/h at 700°C, corresponding to an activation energy of 62 kcal/mol.
The oxidation rate also was a decreasing function of time.
6.3. Experimental approach
Diamond films were deposited under typical conditions in a methane concentration
range o f 1-5%. F or the TGA study, free-standing diamond films were prepared by
dissolving the substrate in an H F solution. The thicknesses of the two films used were 36
p m and 40 pm , respectively. The TGA experiments were performed at 1 atm in a flow
system o f 1%0 2 +Ar or dry air (~ 21 % 0 2 +N 2 ) on a TGA instrument which was described
in section 2.4.6. F or comparison purposes, diamond powders o f both 0.5-1 p m and 3-6
p m sizes as well as graphite powders o f 10-15 p m size also were used. To keep
approximately constant surface area, the initial weight of these samples was maintained at
9.5-10 mg. The gas flow speed in the furnace was about 3 cm/s, and the heating rate was
programed at 5°C/min.
For the study o f structural and morphological consequences caused by oxidation,
CVD diamonds were heated in flow systems (either in a furnace or in the deposition tube
using a tungsten heater) of 0 .0 1 %O 2+Ar and 0.25%O2+Ar at latm in a temperature range
of 500-1100°C. A microwave air plasma operated at 15 Torr also was employed to etch
films. This method o f air plasma etching also is a practical and effective means o f cleaning
the reaction tube by gasifying the deposits on the wall. The oxidized diamond crystals and
films were subsequently examined by SEM and Raman spectroscopy. Diamond films also
were annealed in an oxygen-free gas such as H 2 , N 2 or Ar in a temperature range o f 8001000°C. A microwave hydrogen plasma (90 Torr) also was used to heat a film to 900°C.
An argon plasma could not effectively heat the sample under similar conditions.
6,4. Results
The results of the oxidation study will be presented below in four parts: oxidation
kinetics obtained from TGA; structural and morphological consequences analyzed by SEM
and Raman spectroscopy; characterization of the oxidation residues by TEM; and finally,
annealing in oxygen-free gases.
6.4.1. Oxidation kinetics
Figure 6.1 presents curves o f weight loss vs. temperature o f diamond powders,
graphite powders and diamond films heated in flowing gases of 1 % 0 2 +Ar and air
(~ 2 1 %C>2+N 2 ) as recorded by the TGA instrument. It is seen that the diamond powders
started to oxidize at about 600°C, while diamond films and graphite powders began to
oxidize at about 800°C. The large surface area of powder materials was believed to cause
diamond powers to oxidize at the lowest temperature. Graphite oxidized at higher
temperatures than diamond which is consistent with the literature (Fedoseev and
Uspenskaya, 1974). Although the starting temperature for oxidation of diamond films was
independent o f the oxygen partial pressure (comparing the curves of diamond films in air
and in 1 % 0 2 +Ar), the oxidation rate did change significantly. The weight loss was much
faster in air than that in 1 %0 2 +Ar.
Figure 6.2 shows the oxidation rate as a function o f temperature (900-1250°C) in
the two gases. The oxidation rates were similar below 975°C in both 1 % 0 2 +Ar and air.
However, the rate in air above 1000°C was about 10 times higher than that in 1% 0 2 +Ar
Diamond Film
Diamond Powder
Diamond Film
Diamond Powder
(3 -6 /z m )
Graphite Powder (IO-15/x.m)
Weight (mg)
(0 .5 - l/im)
Tem perature (#C)
Figure 6.1. Weight loss vs. temperature o f diamond powders, graphite powders and
diamond films heated in flowing gases o f 1% 0 2 +Ar and air
Reaction Rate (m g /h )
Temperature (°C)
Figure 6.2. The oxidation rate of the diamond film as a function of temperatures
gas. This behavior indicates that the rate-controlling mechanisms were changing in this
temperature range. It should be noted that the oxidation rate at a specific temperature shown
in Figure 6.2 was obtained by measuring the weight loss over a certain period of time at
that given temperature. The weight loss expressed in Figure 6.1 occurred in a dynamic
system with temperature continuously rising; the slope of the weight loss, when the heating
rate is considered, might not represent the actual oxidation rate at a certain temperature.
The oxidation rate also was found to decrease with time as indicated in Figure 6.3.
The decrease in surface area with the progress o f oxidation probably was responsible for
the decrease of the rate, although in a much shorter time scale, the surface area might
increase due to the surface etching through oxidation. Therefore, the rate given in Figure
6.2 also was an average value over a certain period of time.
6.4.2. Structural and morphological consequences
The first indication of oxidation was the formation o f a black surface layer at
temperatures above 700°C, as shown in Figure 6.4. The surface turned black even in a gas
containing only a trace amount of oxygen (0.0 1 % 0 2 +Ar). This layer was believed to be
graphite or amorphous carbon as indicated by the Raman spectra (Figure 6.5). The Raman
peak around 1580-1600 cm '1, which represents well-developed graphitic structure (sp 2
bonding), became more and more well-defined as the temperature o f oxidation increased.
The diamond peak width at 1332 cm ' 1 also increased with the extent o f oxidation,
indicating the degradation of the diamond structure. Evans and Sauter (1961) found that
this black carbon layer varied from effectively amorphous carbon to an oriented layer o f
graphite with a crystallite size of greater than 200 A and was formed under the presence of
oxygen. The process o f its formation possibly involved the chemisorbed oxygen, a
rearrangement of the surface complexes, and the transference of carbon from the diamond
1% O2 + Ar (I230°C)
2 10
Time (min)
Air (I020°C )
E 140
Time (min)
Figure 6.3. The oxidation rate of the diamond film as a function of time
Figure 6.4. Optical images o f a diamond film (a) before being oxidized and
(b) after being partially oxidized at 700°C in an atmosphere
o f 0.01% O2 + A r for 4 hours
1000 1200 1400 1600 1800
Figure 6.5. Raman spectra o f diamond films oxidized at different temperatures: (a) a
diamond film deposited at 0.5% CH 4 oxidized in 0.25% O 2 + Ar, and (b) a
diamond film deposited at 5% CH 4 oxidized in 0.01% O2 + Ar.
lattice to the layer. Therefore, it was probably a product of both chemical reaction and
phase transition. This layer also has been speculated as being deposited rather than being
the product of a physical phase change (Evans and Sauter, 1961).
SEM observations showed that oxidation etch pits became visible at about 600°C
with significant, distinctive pits appearing at 700°C (Figure 6 .6 a). Below 700°C, the
oxidation occurred preferentially on some faces such as {111} (Figure 6 .6 b), while at
temperatures above 800°C, the surface was attacked uniformly, resulting in a matt-like
topography (Figures 6 .6 a and 6 .6 b). The films oxidized in the microwave air plasma
exhibited similar morphologies. The morphological details caused by oxidation were more
clearly revealed on separate CVD crystals than on continuous films. The etch pits
concentrated on the {111} faces, as indicated in Figure 6.7, in which the morphology of
oxidized individual crystals shows that the {1 1 1 } faces were subjected to extensive
etching, while the {100} faces suffered little or no attack. The hole-shaped etch pits were
apparently defect sites, suggesting that the oxidation might not be limited to the surface
only, and it could proceed into the crystals along these defects.
Oxidation also could occur, of course, on the {100} faces (Figure 6 .8 ) in a bit
higher oxygen partial pressure. Again oxidation started from discrete weaker or defective
sites and then spreaded over all the surface. It also was found that there were ball-like or
ribbon-like materials attached to the oxidized diamond surfaces (Figure 6.9). The black
surface layer was believed to be composed of such substances. These were most probably
some kinds of intermediate reaction products between diamond carbon and oxygen. For
example, diamond might first transform to some intermediate complexes involving
chemisorbed oxygen or transitional forms of carbon before converting completely to
gaseous CO or CO 2 .
2 *,m
2 Hm ------
2 Hm
Figure 6 .6 . SEM micrographs o f the surface morphologies o f the diamond films oxidized
at different temperatures: (a) in 0.01% O 2 + A r for 4 hours; and (b) in 0.25%
O 2 + A r for 1.75 hours.
(•juoo) ' 9 ’9 sinSijj
2 p m ---------
] pm
Figure 6.7. Oxidized diamond crystals at 700°C in 0.01% C>2 + A r for 4 hours showing etch
pits concentrated on the {111 } faces
Figure 6 .8 . Oxidation occurs on the {100} surfaces in 0.25% O 2 + Ar: (a) at 800°C for
1.75 hours; and (b) at 900°C for 1.75 hours.
in 0.25% O 2 + A r for 1.75 hours
Another interesting phenomenon was that material transport possibly occurred
during the oxidation as shown in Figure 6.10, where initially separate individual particles,
which were phase mixtures o f diamond and graphite or amorphous carbon deposited at a
high methane concentration, connected together by joints formed through material transport
during oxidation. The driving force for this phenomenon was the minimization o f surface
free energy, because the particles were all sphere-shaped, and the joints were smoothly
curved. Such material transport driven by surface free energy was a strong indication of a
recrystallization process in many metallic systems. Whether it was true for the carbon
system investigated here remains to be established.
It is interesting to note that TGA-recorded weight loss started at about 800°C, but
Raman spectroscopy started to reflect the structural changes caused by oxidation at about
900°C, and the morphology examined by SEM showed etch pits at temperatures as low as
600°C. This was due to the sensitivity of the different characterization techniques.
Therefore, a statement o f the temperature at which oxidation starts should be taken with
caution. At least the characterization technique ought to be specified.
6.4.3. Characterization of oxidation residues
There were materials left after the diamond film was completely oxidized in the
1 % 0 2 +Ar gas at temperatures up to 1230°C (Figure 6.11). These residue substances were
obviously resistant to oxidation at such a high temperature. They were very thin (thousands
of angstroms), continuous layers with a whitish translucent color. The surface of these
residues retained the morphology of diamond crystals as shown in Figure 6.12. However,
their Raman spectra provided no information on the structure o f these residues.
Since these layers were very thin, they were ideal for TEM analysis. The electron
diffraction pattern (Figure 6.13a) indicated that the residue substance was a pure J3-SiC
5 pm
Figure 6.10. Material transport during oxidation at 1000°C in 0.25% O 2 + Ar
for 1.75 hours
0.5 mm
Figure 6.11. A transmission optical image o f the oxidation residues left from a complete
oxidation o f the diamond film at 1230°C in 1% 0 2 +Ar
800 1000 1200 1400 1600 1800
800 1000 1200 1400 1600 1800
Figure 6.12. SEM micrographs and Raman spectra o f (a) the diamond film and (b) its
oxidation residues
1000 A
Figure 6.13. An electron diffraction pattern and a bright field image of the oxidadon
residues: (a) the diffraction pattern and (b) the bright field image.
phase with a lattice constant of 4.54±0.01
A, compared to the theoretical value o f 4.36 A
(see Table 5.1). The diffraction spots were elongated, and the {111} rings were widened,
indicating that the SiC phase was defective, and the crystallite size was very small, in the
range of 100-300 A, as revealed in a bright field image (Figure 6.13b). This was direct and
physical evidence of the formation of SiC at the interface between the Si substrate and the
diamond film during deposition.
6.4.4. Annealing in oxygen-free gases
Diamond films were heated in gases of hydrogen, argon and nitrogen in the
temperature range of 800-1000°C. No structural changes could be observed by Raman
spectroscopy. The surfaces of these films did not show any visible changes, either. This
confirms that the formation of a black carbon layer requires the aid of oxygen. Since natural
diamond can sustain temperatures up to 1400°C in an oxygen free atmosphere or vacuum
without showing structural changes, it is not surprising that no visible effects were
observed in the present low annealing temperature range of 800-1000°C. However,
annealing in a hydrogen plasma did result in surface etching around the crystal boundaries
at about 1000°C. This suggests that atomic hydrogen or the higher energy level species
obtained in plasmas were responsible for the surface etching phenomenon by non-oxidizing
6.5. Discussion
The oxidation resistance of CVD diamond films is vital for many high temperature
applications such as cutting tools and infrared windows in ambient atmosphere. The
starting temperature of oxidation is one o f the most important parameters forjudging the
oxidation property. It is seen from Figure 6.1 that the temperature required for the onset of
oxidation depends on the specific surface area o f the samples. Another factor influencing
the starting temperature which might be easily ignored is the heating rate o f the sample. The
slower the heating rate, the lower and more accurate the starting temperature can be
determined. Therefore, special attention has to be paid to the heating rate when
comparisons of starting temperatures from different experiments are to be made.
A t a constant surface area, possible rate-controlling mechanisms for oxidation of
diamond films are the oxygen concentration in the gas phase, the gas phase diffusion of
oxygen to the surface, and the surface reaction between carbon and oxygen. The surface
reaction is believed to be a rapid process in the relatively high temperature range (9001250°C) of this study, because most of the reaction-controlled oxidation experiments were
performed in the temperature range of 600-750°C associated with an activation energy of
55 kcal/mol. According to the oxidation rates as functions of temperature and oxygen
partial pressure obtained in this study (Figure 6.2), the diffusion process is the most
probable rate-controlling mechanism below 975°C for both 1% 0 2 +Ar and air, with an
activation energy of about 10 kcal/mol. This value differs from the activation energy of 2337 kcal/mol which is characteristic of the diffusion controlling step through the surface
carbon layer (Evans and Phaal, 1962) and is, therefore, attributed to the gas phase
diffusion process. The black carbon layer can be gasified very rapidly, and hence it does
not limit the overall oxidation rate. Above 975°C, the diffusion rate accelerates, and the
oxidation rate apparently is proportional to the oxygen partial pressure. The oxygen partial
pressure becomes the param eter governing the rate o f oxidation reactions. The
determination of the activation energy in this temperature range becomes meaningless, since
no obvious relationship could be expected between the oxygen partial pressure and the
temperature, although an apparent activation energy of 10 kcal/mol easily but falsely could
be assigned to the process in 1% 0 2 +Ar gas.
The oxidation etching is an effective means for studying diamond defects and their
distribution, because etching normally occurs selectively around defect sites. Previous
etching experiments on diamond surfaces mostly were done with fused potassium nitrate or
sodium carbonate as well as air in which characteristic shapes of etch pits were produced
(Williams, 1932; Omar et al., 1954; Omar and Kenawi, 1957; Patel and Tolansky, 1957).
Evans and Sauer (1961) reported in detail their etching experiments on diamond with
oxidizing etchants. They indicated that for well defined etch pits, the etchant had at least
two functions, namely, as the substance which attacks the surface and as an inhibitor which
modifies the attack. In air, the water vapor accelerated the oxygen-carbon reaction, but one
o r more o f the products o f the reaction (associated with H) acted as an inhibitor. In a flow
system, however, it is difficult to get well-defined pits, because the reaction products are
easily swept away from the surface.
In the experiments described here, the etch holes concentrated on the {111} faces
are believed to be a clear indication of the high defect densities on the triangular faces. Of
the six <110> axes along which Burgers vectors may lie in diamond, three are parallel to a
given triangular face, and three make an angle o f 54°44' with it. Dislocations running out
roughly normal to the {1 1 1 } faces consequently will either be largely screw or nearly pure
edge dislocations. The former type might be expected to etch more rapidly than the latter
and hence produce deeper etch pits. An X-ray topography study performed by Frank and
Lang (1965) confirmed that dislocations terminated on the {111} faces. They indicated that
trigons, which were observed often on the {111 } faces during etching experiments, were
actually dislocation outcrops. Therefore, oxidation reactions could proceed selectively
along these dislocation lines, and a dislocation-facilitated mechanism of oxidation was very
likely as schematically shown in Figure 6.14. However, many other kinds o f defects, such
as stacking faults and microtwins, also form on the {111} surfaces. In addition, the {111}
Figure 6.14. Schematic diagrams of the selective oxidation around dislocation lines:
(a) a edge dislocation; and (b) a screw dislocation.
faces could be in a mosaic structure with many differently oriented subgrains interwoven
(Badzian, 1990), and also they could provide selective oxidation sites.
It is worth noting that the morphological and structural consequences were
examined for diamond films oxidized at different temperatures for a constant period of
time. It is reasonable to suppose that the different morphological and structural changes
caused by different oxidizing temperatures as shown in Figure 6.6 also can be produced by
different oxygen partial pressures o r different periods o f time at a same oxidation
The finding of the p-SiC phase in the oxidation residues is of significance, since
enormous interest and efforts have been focused on the interfacial phenomena. Issues such
as whether a SiC phase forms and what role it plays for diamond nucleation on a Si
substrate have been debated widely in the literature. The oxidation residues are clear
evidence of the formation o f P-SiC on the Si surface exposed to diamond-forming plasmas.
This provides further support that the formation o f the P-SiC material is a necessary step
for diamond nucleation and growth on Si substrate. Since the p-SiC formation is diffusion
limited on a Si surface (Mogab and Leamy, 1973), the P-SiC layer probably thickens with
the diamond film, considering the relatively large thickness of the diamond film which was
analyzed here. This may account for the fact that the p-SiC layer was difficult to detect for
many films (section 5.3), even using sophisticated, high resolution TEM imaging
techniques (Wiliams and Glass). Therefore, the thicker the diamond film, the thicker the
SiC interfacial layer, and the easier it is to find.
6.6. Summary
The starting temperature of oxidation for MPECVD diamond films is about 800°C,
as evidenced by weight loss, while the surface morphology shows visible oxidation etching
pits at temperatures as low as 600°C, and Raman spectroscopy starts to reflect the
structural changes at temperature only as high as 900°C. The oxidation rate depends on the
surface area and the partial pressure of oxygen. The possible rate-controlling mechanisms
in the temperature range o f 900-1230°C are the oxygen concentration and the gas phase
diffusion o f oxygen. The oxidation o f diamond probably proceeds through a mechanism of
diamond transforming to graphitic or amorphous carbon before converting completely to
gaseous CO or CO2 . The diamond oxidation preferentially occurs around defect sites on the
{111} faces, whereas the {100} faces are more resistant to oxidation. The oxidation o f
diamond provides an effective means to study the defects of diamond crystals and the SiC
interfacial layers formed between CVD diamond films and silicon substrates.
Chapter 7
7.1. Evaluation o f the present work
A comprehensive study of the structural properties of MPECVD diamond thin films
has been accomplished with a thread relating the film structure to the deposition and to the
properties throughout the context. A particular emphasis was placed on the investigation o f
the overall internal structural features of diamond films which included the graphite
inclusions, the interfaces between the films and the substrates, and the crystalline defects.
The work contributes to the general understanding of the low pressure CVD o f diamond in
several areas.
1) A parametric understanding of the MPECVD diamond process has been
achieved. The effects of major deposition parameters, such as methane concentration,
substrate temperature, gas pressure, gas flow rate, substrate position relative to the plasma,
and noble gas additions on the film growth rate and the structural quality (diamond phase
purity) have been understood. It is recognized that there is not a single unique set o f
optimum conditions for diamond deposition, rather there are several sets o f them depending
on which parameter is being optimized (growth rate, film uniformity, phase purity, etc.).
The results provide an experimental basis on which we can begin to establish predictive
CVD models to understand the more fundamental aspects of the chemical, thermodynamic,
kinetic and structural properties of CVD diamond processes.
2) New and meaningful results are added to the data base of gas phase chemistry by
exploring the plasma systems o f methane-hydrogen-noble gases. With noble gases as
agents and emission spectroscopy as a characterization tool, knowledge o f the plasma
species, their relative concentrations and their relations with the film growth rate and
structure is obtained. The results fully demonstrate that a high degree o f molecular
excitation and dissociation in the plasmas is critical for achieving a high film growth rate,
and the region for diamond formation can be expanded by changing or controlling the gas
composition. The study provides valuable clues about the plasma chemistry-involved
growth mechanisms of CVD diamond.
3) The complex internal structures o f CVD diamond have been extensively
characterized. Particularly, the following areas have been studied in detail: the unique
diphasic structure of CVD diamond deposited under non-ideal conditions, i.e., the graphite
second phase inclusions in the diamond matrix; the interfaces o f diamond films grown on
Si and SiC>2 substrates; and crystalline defects. The deposition parameters-intemal structure
relations and the structural imperfections-surface morphologies relations have been
explored. The surface morphology and its evolution related to the experimental parameters
and process variables also have been investigated in the course of this thesis; the results are
not included here but will be reported elsewhere. The results achieved represent a
significant step toward a full understanding of the CVD diamond structure which is a key
link between the preparation and properties of diamond films.
4) Finally, the oxidation behavior was investigated as a primary step in exploring
various physical properties of diamond films with relation to the film deposition and
structure. Besides the results on oxidation kinetics, the structural and morphological
consequences of oxidation are particularly interesting. The oxidation experiments are an
excellent means to reveal crystalline defects and carbide interfaces through selective
oxidation reactions or etching and consequently aid in the understanding o f the growth
mechanisms of CVD diamond.
7.2. General concluding remarks
Since detailed discussions and topical conclusions have been given in each chapter,
the following remarks serve as broader conclusions related to the major experimental
results obtained in this thesis research.
7.2.1. MPECVD diamond growth behavior
A systematic investigation of the diamond growth behavior and its correlations with
controllable experimental parameters indicates that the film growth rate is sensitive to the
methane concentration in hydrogen, the substrate temperature, the gas pressure, the gas
flow rate, and the substrate position relative to the plasma. These experimental parameters
are dependent upon each other and are closely interconnected. The conditions for optimal
growth of diamond films, i.e., maximum growth rates, minimum structural defects and
large area uniform coverages, can only be obtained by carefully considering and combining
each of these parameters. Compromises in choosing the experimental parameters to meet
the above requirements for an optimal diamond film deposition are necessaiy in most cases.
For example, large area uniform films can be obtained at a low pressure, but their growth
rates decrease. The growth rate can be increased by depositing films at high methane
concentrations, but the structure o f these films becomes more defective, and a graphite
second phase is introduced.
7.2.2. Noble gas-involved methane-hvdropen plasma chemistry
Optical emission spectroscopic characterization of the noble gas-involved methanehydrogen plasmas supplies ample information on the plasma chemistry-involved growth
mechanisms o f diamond films. The noble gases, which influence the degree of excitation of
reactant molecules by energy transfer or charge transfer from their excited and ionic states,
are active in the deposition process by inducing additional ion-molecule and excited atommolecule reactions. As a result, the parametric range for diamond deposition can be
extended, and enhanced deposition rates are achieved. The non-diamond carbon phases can
be suppressed by small oxygen additions along with the noble gases, leading to an effective
way to rapidly deposit diamond films at high methane concentrations while still retaining
minimal non-diamond carbon components in the films.
7.2.3. Strucmre and defects of CVD diamond
The internal structure o f CVD diamond films is very complex. They are
polycrystalline in nature with varying grain sizes and orientations depending on the
deposition conditions. Typically, the {111} or {100} faces dominate the surface
morphologies of diamond films. Extensive non-diamond carbon phases can be introduced
when the deposition conditions are not optimized. Interfaces with various structure and
compositions between diamond films and silicon or silica substrates are always present.
Furthermore, CVD diamond contains extensive crystalline defects.
The graphite inclusions as a second phase in CVD diamond is inevitably present
when deposition conditions deviate from ideal. High substrate temperatures (>1100°C) and
high methane concentrations (2-5%) are favorable conditions for the graphite formation.
The graphite crystallites exhibit a turbostratic, disordered structure which preferentially
forms in the {111} planes of diamond. Graphite is textured in a domain-like distribution at
moderate methane concentrations, while it becomes randomly distributed at high methane
concentrations. No spatial connections between diamond planar defects and the graphite
second phase are found, although both appear preferentially in the diamond {111} planes.
For the interface structure, the substrate chemistry as well as the methane
concentration greatly influence the nature of the carbon deposited initially onto the substrate
surface. Diamond forms large flat contact areas on silicon, whereas on silica a particulate
type o f intermediate layer forms first due to the chemical reactions occurring on the surface.
A silicon carbide phase exists between the diamond film and the silicon or silica substrate,
although it may not always be detected easily on silicon.
Overwhelming planar defects are present in CVD diamond structure. They include
microtwins and stacking faults, both o f which occur predominantly on the {111} planes.
Their densities increase with the methane concentration during the deposition process, and
their dimensions become correspondingly smaller. Planar defects on other planes are rare.
Dislocations and various point defects also are detected. All these structural defests have
detrimental effects on the electrical, thermal and optical properties of CVD diamond films.
The tendency to form these structural defects is one of the main obstacles in growing single
crystals or developing tailored structure o f CVD diamond.
7.2.4. Oxidation properties of CVD diamond
The temperature at which considerable oxidation for MPECVD diamond films
begins is about 700°C. The oxidation rate depends on the surface area o f the sample and
the oxygen partial pressure in the gas phase. Also it strongly depends on the
crystallographic orientation o f diamond. The oxidation preferentially occurs around
defective sites on the {111} faces, whereas the {100} faces are more resistant to oxidation.
The oxidation of diamond is believed to proceed through a mechanism o f diamond
transforming to graphitic or amorphous carbon before converting completely to gaseous
CO or CO 2 . The oxidation experiments are also very helpful in studying the defect
structures and the possible carbide intermediate layers formed between diamond films and
silicon substrates.
7.3. Directions of future work
As indicated by Yarbrough and Messier in a recent Science article (1990), and
shown in specific details in this thesis, there are many issues and problems which remain to
be solved in the current field of low pressure CVD of diamond before a realistic assessment
of the potential impact on the materials science and technology can be made. There is still
much work to be done, such as along the lines of this thesis research, in order to reach that
elusive goal o f a detailed basic and applied understanding. In particular, three major
research areas, which were addressed in part in this thesis, are the following:
1) Further work on the parametric deposition process is needed to clarify the effects
o f various experimental parameters and process variables. Techniques of effectively
decoupling the deposition parameters are required. Based on the experimental data from the
parametric investigation, thermochemical modeling of the diamond growth process is
necessary. The kinetics and thermodynamics of the CVD system have to be developed and
understood. This will enable a more profound understanding of the fundamental growth
mechanisms and, in turn, a better control of the deposition process.
2) Some structural features need to be analyzed in more detail. Examples are: the
fine details o f the turbostratic structure of graphite; the cause of the domain distribution of
graphite crystallites in the diamond matrix; the type, nature and distribution of point defects;
and noble gas-induced defects and their activities. Based on a good understanding of the
relations between the film deposition and structure, experimental methods o f reducing or
eliminating various defects need to be explored. This will eventually lead to the growth of
heteroepitaxial single crystal diamond. Relations between the film structure and various
physical properties also are of great importance.
3) The plasma chemistry and surface chemistry in diamond-forming plasmas need
to be characterized more fully. A combined use of laser/optical and mass spectroscopic
techniques is desired. A broad and reliable experimental data base on the plasma properties
and surface phenomena will be particularly useful in testing various theoretical works.
Adams, D. M. and S. J. Payne (1974), J. Chem. Soc. Lond. Faraday Trans,
Aisenberg, S. and R. Chabot (1971), J. Appl. Phys., 4
n, ZQ, 1959
2 , 2953
Akatsuka, F., Y. Hirose and K. Komaki (1988), Jpn. J. Appl. Phys.,
2Z ,
Alder, B. J. and R. H. Christian (1961), Phys. Rev. Lett., Z. 367
Anderson, L. P. (1981), Thin Solid Films, M , 193
Angus, J. C. (1989), in Proceedings of the First International Symposium on Diamond and
Diamond-like Carbon, edited by J. P. Dismukes, A. J. Purdes, B. S. Meyerson, T. D.
Moustakas, K.E. Spear, K. V. Ravi and M. Yoder, The Electrochemical Soc. Inc.,
Pennington, NJ, Vol. 89-12, p. 1
Angus, J.
C .,
H. A. Will and W. S. Stanko (1968), J. Appl. Phys.,
Angus, J.
C .,
P. Koidl and S. Domitz (1986), in Plasma Deposited Thin Films, edited by
J. Mort and F. Jansen,
Angus, J.
C. C.
Press, Boca Raton, FL, p. 89
Hayman (1988), Science, 2 4 1 .913
Anthony, T. R. (1987), in Final Program and Abstracts of the MRS Fall Meeting. Boston,
Massachusetts, Paper No. N3.6
Anthony, T. R. (1988), in MRS Short Course on Chemical Vapor Deposition of Diamond
Films. Boston, Massachusetts
Bachmann, P. K., W. Drawl, D. Knight, R. Weimer and R. Messier (1988), in MRS
Extended Abstracts on Diamond and Diamond-like Materials Synthesis, edited by
G. H. Johnson, A. R. Badzian and M. W. Geis, Vol. EA-15, p. 99
Badzian, A. R. (1987), in Advances in X-Rav Analysis, edited by C. B. Barrett, J. V.
Gilfrich, R. Jenkins, J. C; Russ, J. W. Richardson Jr. and P. K. Predecki, Plenum,
New York, Vol. 31, p. 113
Badzian, A. R. (1990), not published
Badzian, A. R., B. Simonton, T. Badzian, R. Messier, K. E. Spear and R. Roy (1986),
Proceedings SPIE, 6 8 3 .127
Badzian, A. R., P. K. Bachmann, T. Hartnett, T. B. Badzian and R. Messier (1987), Les
Editions de Physique (Paris), IS, 63
Badzian, A. R. and R. C. DeVries (1988), Mater. Res. Bull., 2
2 , 385
Badzian, A. R. and T. Badzian (1988), Thin Solid Films, 26.283
Badzian, A. R., T. Badzian, R. Roy, R. Messier and K. E. Spear (1988a), Mater. Res.
Badzian, A. R., T. Badzian and D. Pickrell (1988b), Proceedings SPIE, 262,14
Badzian, A. R., T. Badzian and L. Pilione (1990), Int. J. Refract. Metals and Hard Mater.,
(in press)
Balamuta, J. B., M. F. Golde and Y. S. Ho (1983), J. Chem. Phys., 22, 2822
Barin, I. and O. Knacke (1973), Thermochemical Properties of Inorganic Substances.
Springer-Verlag, Berlin and New York, p. 116
Baum, R. M. (1985), in Chemical and Engineering News. December 23, p. 20
Belton, D. N., S. J. Harris, S. J. Schweis, A. M. Weimer and T. A. Perry (1989), Appl.
Phys. Lett., 24, 416
Boenig, H. V. (1982), Plasma Science and Technology. Cornell University Press, Ithaca
and London, p. 47
Booker, G. R. and R. Stickler (1962), J. Appl. Phys.,
Bovenkerk, H. P. (1961), Amer. Mineral., 146. 952
Bridgman, P. W. (1947), J. Chem. Phys., 12, 92
Bridgman, P. W. (1955), Sci. Amer., 1 9 3 .42
Bundy, F. P. (1963), J. Chem. Phys.,
Bundy, F. P. (1980), J. Geophys. Res., £5, 6930
Bundy, F. P., H. T. Hall, H. M. Strong and R. H. Wentorf, Jr. (1955), Nature Lond.,
176. 51
Bundy, F. P., H. P. Bovenkork, H. M. Strong and R. H. Wentorf, Jr. (1961), J. Chem.
Bundy, F. P. and J. S. Kasper (1967), J. Chem. Phys., 4£, 3437
Bundy, F. P., H. M. Strong and R. H. Wentorf, Jr. (1973), in Chemistry and Physics of
Carbon, edited by P. L. Walker, Jr. and P. A. Thrower, Marcel Dekker Inc., NY,
Vol. 10, p. 213
Butler, J. E. and F. G. Celii (1989), in Proceedings of the First International Symposium
on Diamond and Diamond-like Films, edited by J. P. Dismukes, A. J. Purdes, B. S.
Meyerson, T. D. Moustakas, K.E. Spear, K. V. Ravi and M. Yoder, The
Electrochemical Soc. Inc., Pennington, NJ, Vol. 89-12, p. 317
Capitell, M. and M. Molinari (1980), in Topics in Current Chemistry, edited by S. Veprek
and M. Venugopalan, Springer, Berlin-Heidelberg-New York, Vol. 90, p. 70
Carter, N. L. and G. C. Kennedy (1964), J. Geophys. Res., £2, 2403
Celii, F. G„ P. E. Pehrsson, H. T. Wang and J. E. Butler (1988), Appl. Phys. Lett., 52.
Chang, C. P., D. L. Flamm, D. E. Ibbotson and J. A. M ucha (1988), J. Appl. Phys., 63,
Chauhan, S. P., J. C. Angus and N. C. Gardner (1976), J. Appl. Phys., 42., 4746
Clark, C. D. (1965), in Physical Properties of Diamond, edited by R. Berman, Clarendon
Press, Oxford, p. 295
Clark, C. D„ E. W. J. Mitchell and B. J. Parsons (1979), in The Properties o f Diamond.
edited by J. E. Field, Academic Press, London, p. 23
Clausing, R. E., L. Heatherly and K. L. More (1989). Surf. Coat. Technol., 39/40.199
Collins, R. W., Y. Cong, Y. T. Kim, K. Vedam, Y. Liou, A. Inspektor and R. Messier
(1989), Thin Solid Films, I £ L 565
Davies, G. (1977), in Chemistry and Phvsics of Carbon, edited by P. L. Walker, Jr. and
P. A. Thrower, Marcel Dekker, Inc., NY, Vol. 13, p. 1
De Carli, P. S. (1966), in Science and Technology o f Industrial Diamonds, edited by
J. Burls, Industrial Diamond Information Bureau, London, Vol. 1, p. 49
De Carli, P. S. and J. C. Jamieson (1961), Science, 1 3 3 .1821
De Larue, W. and H. W. Muller (1978), Phil. Trans. Roy. Soc. Lond., 1 6 9 .155
Deijaguin, B. V., D. V. Fedoseev, V. M. Lukyanovich, B. V. Spitsyn, V. A. Ryabov and
A. V. Lavrentyev (1968), J. Cryst. Growth, 2, 380
Deijaguin, B. V., V. G. Lyutsau, D. V. Fedoseev and V. A. Ryabov (1970), Dokl. Akad.
Nauk. SSSR, m
Deijaguin, B. V. and D. V. Fedoseev (1973), Carbon, H , 299
Deijaguin, B. V., D. V. Fedoseev and K. S. Uspenskaya (1973), Russ. J. Phys. Chem.,
42 ,1 3
Deijaguin, B. V. and D. V. Fedoseev (1975), Sci. Amer., 2 3 3 .102
Deijaguin, B. V. and D. V. Fedoseev (1977), in.Growth of Diamond and Graphite from
the Gas Phase. Jzd. Nauka, Moscow (in Russian)
Deijaguin, B. V., L. L. Bouilov and B. V. Spitsyn (1986), Archiwum Nauki o
M ateriatach,2, 111
DeVries, R. C. (1987), Annu. Rev. Mater. Sci., 12,161
Ditchbum, R. W. (1965), in Physical Properties o f Diamond, edited by R. Berman,
Clarendon Press, Oxford, p. 5
Du Pont, E. J. (1965), Netherland Patent 6506395
Evans, T. (1965), in Physical Properties of Diamond, edited by R. Berman, Clarendon
Press, Oxford, p. 116
Evans, T. (1979), in The Properties of Diamond, edited by J. E. Field, Academic Press,
London, p. 403
Evans, T. and D. H. Sauter (1961), Phil. Mag., £> 1129
Evans, T. and C. Phaal (1962a), Proc. Roy Soc. Lond., A 270.538
Evans, T. and C. Phaal (1962b), in Proceedings o f the Fifth Biennial Conference on
Carbon. The Pennsylvania State University, University Park, PA, p. 147
Evans, T. and P. Rainey (1975), Proc. Roy. Soc. Lond., A344. I l l
Eversole, W. G. (1962), U.S. Patents 3030187 and 3030188
Fedoseev, D. V. and K. S. Uspenskaya (1974), Russ. J. Phys. Chem.,
Fedoseev, D. V., V. P. Varnin and B. V. Deryagin (1984), Russ. Chem. Rev., 52, 435
Field, J. E. (1979), in The Properties of Diamond, edited by J. E. Field, Academic Press,
London, p. 641
Firestone, R. F. and L. M. Dorfman (1971), in Actions Chimiques et Biologiaues des
Radiations, edited by M, Haissinsky, Masson & Cie Editeurs, Paris, p. 7
Fischbach, D. B. (1971), in Chemistry and Phvsics of Carbon, edited by P. L. Walker,
Jr., Marcel Dekker, Inc., NY, Vol. 7, p. 1
Frank, F. C. and A. R. Lang (1965), in Physical Properties of Diamond, edited by R.
Berman, Clarendon Press, Oxford, p. 69
Frenklach, M. and K. E. Spear (1988), J. Mater. Res. 5 ,1 3 3
Frondel, C. and V. B. Marvin (1967), Nature Lond., 214. 587
Fujimori, N., T. Imai and A. Doi (1986), Vacuum, 26 .9 9
Fujimori, N., A. Ikegaya, T. Imai, K. Fukushima and N. Ota (1989), in Proceedings of
the First International Symposium on Diamond and Diamond-like Films, edited by
J. P. Dismukes, A. J. Purdes, B. S. Meyerson, T. D. Moustakas, K.E. Spear, K. V.
Ravi and M. Yoder, The Electrochemical Soc. Inc., Pennington, NJ, Vol. 89-12,
p. 465
Geis, M. W. (1989), in Abstracts of Symposium on Preparation and Characterization o l
Diamond and Diamond-like Materials, sponsored by the American Vacuum Society and
the Materials Research Society, The Pennsyvania State University, University Park,
PA, Sept. 20
Gildenblat, G. Sh., S. A. Grot, C. R. Wronski, A. R. Badzian, T. Badzian and R.
Messier (1988), Appl. Phys. Lett., 52.586
Gildenblat, G. Sh., S. A. Grot, C. W. Hatfield, C. R. Wronski, A. R. Badzian, T.
Badzian and R. Messier (1990), Mater. Res. Bull., 22.129
Goma, J. and M. Oberlin (1980), Thin Solid Films, 55.221
Gonzalez-Hemandez, J., R. Asomoza, A. Reyes-Mena, J. Rickard C., S. S. Chao and D.
Pawlik (1986), J. Vac. Sci. Technol., A£> 83
Grundy, P. J. and G. A. Jones (1976), Electron Microscopy in the Study of Materials.
Edward Arnold Publications, London
Hammond, M. L. (1988), in Handbook of Thin Film Deposition Processes and
Techniques, edited by K. K. Schuegraf, Noyes Publications, Park Ridge, NJ, p. 26
Hannay, J. B. (1880), Proc. Roy. Soc. Lond., 2Q, 450
Hannemann, R. E., H. M. Strong, and F. P. Bundy (1967), Science, 155. 995
Hanssen, L. M., W. A. Carrington, J. E. Butler and K. A. Snail (1989), Mater. Lett., Z,
Harris, S. J., A. M. Weiner and T. A. Perry (1988), Appl. Phys. Lett., 52.1605
Hartnett, T. M. (1988), Characterization of Diamond Deposition in a Microwave Plasma.
M. S. Thesis, The Pennsylvania State University, University Park, PA
Haymet, A. D. J. (1985), Chem. Phys. Lett., 1 2 2 .421
Heidenreich, R. D. (1964), Fundamentals of Transmission Electron Microscopy.
Interscience Publishers, New York
Heimann, R. B., J. Kleiman and N. M. Salansky (1984), Carbon, 22.147
Herzberg, G. (1971), The Spectra and Structure o f Simple Free Radicals. An Introduction.
to Molecular Spectroscopy. Cornell University Press, Ithaca, NY
Hirose, Y. and N. Kondo (1988), in Proceedings of the First International Confemece on
New Diamond Science and Technology, editd by Japan New Diamond Forum, Tokyo,
Japan, p. 39
Hirose, Y. and Y. Terasawa (1986), Jap. J. Appl. Phys., 21, L519
Hirsch, P. B., A. Howie, R. B. Nicholson, D. W. Pashley and M. J. Whelan (1965),
Electron Microscopy of Thin Crystals. Butterworths, London,
Holland, L. and S. M. Ojha (1975), Vacuum, 21, 53
Holland, L. and S. M. Ojha (1979), Thin Solid Films, 5 3 . 107
Huang, D., M. Frenklach and M. Maroncelli (1988), J. Phys. Chem., 22. 6379
Humble, P. J. K. Mackenzie and A. Olsen (1985), Phil. Mag. 52,605
Hu, G. Q. and M. G. Qian (1980), Metallography. Shanghai Science and Technology
Publications, Shanghai, China (in Chinese)
Inspektor, A. (1987), Surf. Coat. Technol., 21, 31
Inspektor, A., U. Carmi, A. Raveh, Y. Khait and R. Avni (1986), J. Vac. Sci. Technol.,
M , 375
Inspektor, A., Y. Liou, T. McKenna and R. Messier (1989a), Surf. Coat. Technol.,
Inspektor, A., T; McKenna, Y. Liou, L. Bourget, K. E. Spear and R. Messier (1989b), in
Proceedings of the First International Symposium on Diamond and Diamond-like-
Films, edited by J. P. Dismukes, A. J. Purdes, B. S. Meyerson, T. D. Moustakas,
K.E. Spear, K. V. Ravi and M. Yoder, The Electrochemical Soc. Inc., Pennington,
NJ, Vol. 89-12, p. 342
International Resource Development Inc. (1987), Diamond Thin Film: The Threshold of
Commercialization. Report No. 733, New Canaan, CT
Jaccodine, J. (1963), Appl. Phys. Lett., 2,201
JCPDS International Center for Diffraction Data (1989), Powder diffraction File.
Swarthmore, PA
Joffreau, P. O., R. Haubner and B. Lux (1988), in MRS Extended Abstracts on Diamond
and Diamond-like Materials Synthesis, edited by G. H. Johnson, A. R. Badzian and
M. W. Geis, Vol. EA-15, p. 15
Johnson, C. E., W. A. Weimer and D. C. Harris (1989), Mater. Res. Bull., 2 4 , 1127
Kaiser, W. and W. C. Bond (1959), Phys. Rev., U S , 857
Kamo, M„ Y. Sato, S. Matsumoto and N. Setaka (1983), J. Cryst. Growth, £2, 642
Kasi, S. R., H. Kang and T. W. Rabalais (1988), J. Chem. Phys., M , 5914
Katz, M., D. Itzhak, A. Grill and R. Avni (1980), Thin Solid Film, 22,497
Kawarada, M., K. Mar and A. Kiraki (1987), Jpn. J. Appl. Phys., 2£, L1032
Kawarada, H., K. Nishimura, T. Ito, J. Suzuki, K. S. Mar, Y. Yokota and A. Hiraki
(1988), Jpn. J. Appl. Phys.,
Kawato, T. and K. Kondo (1987), Jap. J. Appl. Phys., 2d, 1429
Kennedy, G. C. (1966), in Science and Technology o f Industrial Diamonds, edited
by J. Burls, Industrial Diamond Information Bureau, London, Vol. 1, p. 41
Klassen-Neklyudova, M. V. (1964), Mechanical Twinning of Crystals, translated by J. E.
S. Bradley, Consultants Bureau, New York, p. 32
Knight, D. S. and W. B. White (1989), J. Mater. Res., 4,3 8 5
Knights, J. C., R. A. Lujan, M. P. Rosenblum, R. A. Street, D. K. Bieglesen and J. A.
Reimer (1981), Appl. Phys. Lett., 2&» 331
Knox, B. E. and K. Vedam (1976), Coating Science and Technology. Final Technical
Report, Rome Air Development Center Contract No. RADX-TR-76-96 (ARPA order
No. 2415)
Kobashi, K., K. Nishimura, Y. Kawate and T. Horiuchi (1988), Phys. Rev., 2&, 4067
Komatsu, T., H. Yamashiola, Y. Tamou and N. Kikuchi (1987), in Proceedings of the
Eighth International Symposium on Plasma Chemistry, edited by K. Akashi and A.
Kinbara, International Union o f Pure and Applied Chemistry, Oxford, U. K.,
Vol. 1, p. 2487
Krogh, O., T. W icker and B. Chapman (1986), J. Vac. Sci. Technol., A4, 1796
Kurihara, K., K. Sasaki, M. Kawarada and N. Koshino (1988), Appl. Phys. Lett., 52.
Lambert, J. D. (1936), Trans. Faraday Soc., 2 2 ,4 5 2
Lander, J. J. and J. Morrison (1963), J. Chem. Phys., M> 1403
Lander, J. J. and J. Morrison (1964), Surf. Sci., 2. 553
Lang, A. R. (1979), in The Properties of Diamond, edited by J. E. Field, Academic Press,
London, p. 425
Latham, J. L. and A. E. Burgess (1977), Elementary Reaction Kinetics. Butterworths,
London, p. 33
Lawn, B., Y. Kamiya and A. R. Lang (1965), Phil. Mag., 12. 177
Lewis, R. S., T. Ming, J. F. Wacker, E. Andrers and E. Steel (1987), Nature, 3 2 6 .160
Libby, W. F. (1963), Phys. Rev., m
Lind, S. C., C. J. Hochanadel and J. A. Ghormley (1961), Radiation Chemistry of Gases.
Reinhold Publishing Corporation, New York
Liou, Y., A. Inspektor, R. Weimer and R. Messier (1989), Appl. Phys. Lett., 51,631
Lochte-Holtgreven, W. (1968), in Plasma Diagnostics, edited by W. Lochte-Holtgreven,
North-Holland Publishing Company, Amsterdam, p. 181
Lonsdale, K. (1947), Phil. Trans. Roy. Soc., A 2 4 0 .219
Lonsdale, K. and H. J. Milledge (1965), in Physical Properties of Diamond, edited by
R. Berman, Clarendon Press, Oxford, p. 48
Lonsdale, K., H. J. Milledge and E. Nave (1959), Mineral. Mag., 22,185
Loretto, M. H. and R. E. Smallman (1975), Defect Analysis in Electron Microscopy.
Chapman and Hall, London
Machlin, E. S. (1988), J. Mater. Res., 2, 958
Mania, R., L. Strobierski and R. Pampuch (1981), Cryst. Res. Technol., 15, 785
Manory, R. R., U. Carmi, R. Avni and A. Grill (1988), Thin Solid Films, 155,79
Matsumoto, S. (1985), J. Mater. Sci. Lett., 4, 600
Matsumoto, O. and T. Katagiri (1987), Thin Soild Films, 146.283
Matsumoto, S., Y. Sato, M. Kamo and N. Setaka (1982a), Jpn. J. Appl.Phys., 21, LI 83
Matsumoto, S., Y. Sato, M. Tsutsumi and N. Setaka (1982b), J. Mater. Sci.,12, 3106
Matsumoto, S and Y. Matsui (1983), J. Mater. Sci., 12,1785
Matsumoto, O., H. Toshima and Y. Kanzaki (1985), Thin Solid Films, 122,341
Matsumoto, S., M. Hino and T. Lobayashi (1987), Appl. Phys. Lett, 51,7 3 7
Matthews, J. W. (1959), Phil. Mag., 4, 1017
Matthews, J. W. (1962), Phil. Mag., 2, 915
McKenna, T. A. (1990), Diamond Film Deposition and Microwave Plasma Processes in
the C-H-O System. M. S. Thesis, The Pennsylvania State University, University Park,
McLintock, I. S. and J. C. Orr (1973), in Chemistry and Phvsics of Carbon, edited by
P. L. Walker, Jr. and P. A. Thrower, Marcel Dekker, Inc., NY, Vol. 11, p. 243
Meilunas, R., M. S. Wong, K. C. Sheng, R. P. H. Chang and R. P. Van Duyne (1989),
Appl. Phys. Lett., 54,2204
Mendelson, S. (1964), J. Appl. Phys., 2 5 ,1570
Messier, R., A. R. Badzian, T. Badzian, K. E. Spear, P. Bachmann and R. Roy (1987),
Thin Solid Films, 1 5 3 .1
Mitsuda, Y., K. Kojima, T. Yoshida and K. Akashi (1987), J. Mater,' Sci., 22,1557
Mitsuda, Y., T. Yoshida and K. Akashi (1989), Rev. Sci. Instrum., 6Q, 249
Mitura, S. (1987), J. Cryst. Growth, &Q, 417
Mogab, J. and H. J. Leamy (1973), in Silicon Carbide. Proceedings of the Third
International Conference on Silicon Carbide, edited by R. C. Marshall, J. W. Faust, Jr.
and C. E. Ryan, University of South Carolina Press, Columbia, SC, p. 58
Moisson, H. (1894), Compt. Rend., Circ. Acad. Sci. Paris (France), 118. 320
Moore, C. E. (1949-1958), Atomic Energy Levels. Natl. Bur. Std. (U. S.) Circ. No. 467,
Vol. 1-3
Moore, M and A. R. Lang (1972), Phil. Mag., 26,1313
Moore, M and A. R. Lang (1974), J. Cryst. Growth, 26.133
Mori, T. and Y. Namba (1983), J. Vac. Sci. Technol., AL 23
Mrozowski, S. (1980), Thin Solid Films, 65. 221
Mucha, J. A., D. L. Flamm and D. E. Ibbotson (1989), J. Appl. Phys., 65. 3448
Muncke, G. (1979), in The Properties of Diamond, edited by J. E. Field, Academic Press,
London, p. 473
Musil, J. (1986), Vacuum, 25.161
Namba, Y. and T. Mori (1985), J. Vac. Sci. Technol., A2> 319
Nemanich, R. J., J. T. Glass, G. Lucovsky and R. E. Shroder (1988), J. Vac. Sci.
Technol., M
Ohtake, N., H. Tokura, Y. Kuriyama, Y. Mashimo and M. Yoshikawa (1989),
Proceedings o f the First International Symposium on Diamond and Diamond-like.
Films, edited by J. P. Dismukes, A. J. Purdes, B. S. Meyerson, T. D. Moustakas,
K.E. Spear, K. V. Ravi and M. Yoder, The Electrochemical Soc. Inc., Pennington,
NJ, Vol. 89-12, p. 93
Omar, M., N. S. Pandya and S. Tolansky (1954), Proc. Roy. Soc., A225. 33
Omar, M. and M. Kenawi (1957), Phil. Mag., 2,859
Ono, A., T. Bada, H. Funamoto and A. Nishikawa (1986), Jpn. J. Appl. Phys., 25. 1808
Park, S. C., J. R. Bodart, H. X. Han and B. J. Feldman (1987), in MRS Proceedings on
Plasma Processing, edited by J. W. Cobum, R. A. Gottscho and D. W. Hess, The
Materials Research Society, Pittsburgh, PA, Vol. 68, p. 199
Parsons, C. A. (1907), Proc. Roy. Soc. Lond., 22. 532
Parsons, C. A. (1920), Phil. Trans. Roy. Soc. Lond., A220. 67
Pashley, D. W. and M. J. Stowell (1963), Phil. Mag. & 1605
Pate, B. B. (1986), Surf. Sci., 165. 83
Patel, A. R. and S. Tolansky (1957), Proc. Roy. Soc. Lond., A 243.41
Pearse, R. W. B. and A. G. Gaydon (1963), The Identification of Molecular Spectra.
Chapman and Hall Ltd., London
Phaal, C. and G. Zuidema (1966), Phil. Mag., M> 79
Phillips, V. A. (1960), Phil. Mag., 5, 571
Pickrell. D., W. Zhu, A. R. Badzian, R. Messier and Newnham (1990a), Appl. Phys.
Lett, (in press)
Pickrell. D., W. Zhu, A. R. Badzian, R. Messier and Newnham (1990b), J. Mater. Res.
Piekarczyk, W., R. Roy and R. Messier (1989), J. Cryst. Growth,
Plano, L. S., S. Yokota and K. V. Ravi (1989), in Proceedings of the First International
Symposium on Diamond and Diamond-like Films, edited by J. P. Dismukes, A. J.
Purdes, B. S. Meyerson, T. D. Moustakas, K.E. Spear, K. V. Ravi and M. Yoder,
The Electrochemical Soc. Inc., Pennington, NJ, Vol. 89-12, p. 380
Poferl, D. J., N. C. Gardner and J. C. Angus (1973), J. Appl. Phys., M . 1428
Prince, J. F., C. B. Collins and W. W. Robertson (1964), J. Chem. Phys., 40,2619
Rabalais, J. W. and S. R. Kasi (1988), Science, 2 3 9 .623
Raman, C. V. and K. S. Krishnan (1928), Nature, 121.501
Rapaport, E. (1967), J. Chem. Phys., 4£, 2691
Robertson, J. (1986), Advances in Physics, 25.317
Roman, W. C., M. B. Colket, S. O. Hay and A. C. Eckbreth (1989), in Proceedings of
the First International Symposium on Diamond and Diamond-like Films, edited by
J. P. Dismukes, A. J. Purdes, B. S. Meyerson, T. D. Moustakas, K.E. Spear, K. V.
Ravi and M. Yoder, The Electrochemical Soc. Inc., Pennington, NJ, Vol. 89-12,
p. 330
Rouzaud, J. N., A. Oberlin and C. Beny-Bassez (1983), Thin Solid Films, 105.75
Rudder, R. A., G. G. Fountain and R. J. Markunas (1986), J. Appl. Phys., 60, 3519
Rudder, R. A., D. J. Vitkavage and R. J. Markunas (1988), in MRS Extended Abstracts
on Diamond and Diamond-like Materials Synthesis, edited by G. H. Johnson, A. R.
Badzian and M. W. Geis, Vol. EA-15, p. 19
Saito, Y., S. Matsuda and S. Nogita (1986), J. Mater. Sci. Lett., 5,565
Saito, Y., K. Sato, H. Tanaka, K. Fujita and S. Matsuda (1988), J. Mater. Sci., 21,8 4 2
Sato, Y., M. Kamo and N. Setaka (1987), in Proceedings of the Eighth International
Symposium on Plasma Chemistry, edited by K. Akashi and A. Kinbara, International
Union of Pure and Applied Chemistry, Oxford, U. K„ Vol. 1, p. 2446
Sato, Y. and M. Kamo (1989), Surf. Coat. Technol., 3 9 /4 0 .183
Sawabe, A. and T. Inuzuka (1985), Appl. Phys. Lett., 4£, 146
Sawabe, A. and T. Inuzuka (1986), Thin Solid Films, 137. 89
Setaka, N. (1984), Hyomen (Surface), 22,1 1 0 (in Japanese)
Setaka, N. (1989), J. Mater. Res., 4, 664
Shiomi, H., Y. Nishibayashi and N. Fujimori (1989), Jpn. J. Appl. Phys., 22, L2153
Sibiende, F. and G. Benezech (1988), J. Mater. Sci., 22.1632
Singh, B., O. R. Mesker, A. W. Levine and O. R. Mesker (1988), Appl. Phys. Lett., 52.
Skinner, B. J. (1957), Amer. Mineral., 42, 39
Smithson-Tennant (1797), Phil. Trans. Roy. Soc. Lond., £2,123
Sokolowski, M., A. Sokolowski, B. Gokieli, A. Michaelski, A. Rusek and Z.
Ramanowski (1979), J. Cryst. Growth, 47,421
Sommer, M., K. Mui and F. W. Smith (1989), Solid State Comm., 52,775
Spear, K. E. (1984), in Proceedings of the Ninth International Conference on Chemical
Vapor Deposition, edited by McD. Robinson, C. H. J. van der Brekel, G. W. Cullen,
J. M. Blocher, Jr. and P. Rai-Choudhury, Vol. 6, p. 81
Spear, K. E. (1989), J. Amer. Ceram. Soc., 22, 171
Spencer, E. G., P. H. Schmidt, D. C. Joy and F. J. Sansalone (1976), Appl. Phys. Lett.,
22, 118
Spinks, J. W. T. and R. J. Woods (1976), An Introduction to Radiation Chemistry. John
W iley and Sons, New York, p. 141
Spitsyn, B. V. and L. L. Bouilov (1988), in MRS Extended Abstracts on Diamond and
Diamond-like Materials Synthesis, edited by G. H. Johnson, A. R. Badzian and M. W.
Geis, Vol. EA-15, p. 3
Spitsyn, B. V., L. L. Bouilov and B. V. Deijaguin (1981), J. Cryst. Growth, 5 2 ,2 1 9
Suzuki, K, A. Sawabe and T. Inuzuka (1988), Appl. Phys. Lett., 52 .1818
Suzuki, K., A. Sawabe, H. Yasuda and T. Inuzuka (1987), Appl. Phys. Lett., 5Q, 728
Suzuki, M. and H. Tanji (1987), in Proceedings of the Tenth International Conference on.
Chemical Vapor Deposition, edited by G. W. Cullen and J. M. Blocher, Jr., Vol. 8,
p. 1089
Technical Insights Inc. (1987), Diamond Films: Evaluating the Technology and
Opportunities. Emerging Technologies No. 25, Eaglewood/Fort Lee, NJ,
Thomas, G. and M. J. Goringe (1979), Transmission Electron Microscopy of Materials.
Wiley, New York
Thornton, J. A. (1982), in Deposition Technologies for Films and Coatings, edited by
R. F. Bunshah, Noyes Publications, Park Ridge, NJ, p. 40
Tsai, H. C. and D. B. Bogy (1987), J. Vac. Sci. Technol., A2, 3287
Tsuda, M., M. Nakajima and S. Oikawa (1986), J. Am. Chem. Soc., 1 0 8 .5780
Vakil, H., W. Banholzer, R. Kehl and C. L. Spiro (1989), Mater. Res. Bull., 2 4 ,7 3 3
Varma, C. K. R. (1967), Phil. Mag., 16, 959
Vamin, V. P., I. G. Teremetskaya, D. V. Fedoseev and B. V. Deryagin (1984), Sov.
Phys. Dokl., 22, 419
Vastola, F. J. and J. P. Wightman (1964), J. Appl. Chem., 14, 69
Venugopalan, M. and R. Avni (1985), in Thin Film From Free Atoms and Particles, edited
by K. J. Klabunde, Academic Press, Orlando, p. 68
Venugopalan, M. and S. Veprek (1983), in Topics in Current Chemistry, edited by
F. L. Boschke, Springer, Berlin-Heidelberg-New York, Vol. 107, p. 1
Vishnevsky, A. S. (1975), J. Cryst. Growth, 22.2 9 6
Von Bolton, W. (1911), Z. Elektrochemie, 12,971
Von Heimendahl, M. (1980), Electron Microscopy of Materials, translated by U. E. Wolff,
Academic Press, New York
Wada, N., P. J. Gaczi and S. A. Solin (1980), J. Non-Cryst. Solids, 25.5 4 3
Walker, J. (1979), Rep. Prog. Phys., 42,1605
Walker, P. L„ Jr., F. Rusinko Jr. and L. G. Austin (1959), in Advances in Catalysis.
edited by D. D. Eley, P. W. Selwood and P. B. Weisz, Academic Press, New York
and London, Vol. XI, p. 133
Walmsley, J. C. and A. R. Lang (1983), J. Mater. Sci. Lett., 2. 785
Wang, X. H., L. Pilione, W. Zhu, W. A. Yarbrough, W. Drawl and R. Messier (1990),
J. Mater. Res. (in press)
Wedlake, R. J. (1974), Diamond Res., pp. 2
Wedlake, R. J. (1979), in The Properties of Diamond, edited by J. E. Field, Academic
Press, London, p. 502
Wei, J., H. Kawarada, J. Suzuki, K. Yanagihara, K. Numata and A. Hiraki (1989), in
Proceedings of the First International Symposium on Diamond and Diamond-like
Films, edited by J. P. Dismukes, A. J. Purdes, B. S. Meyerson, T. D. Moustakas,
K.E. Spear, K. V. Ravi and M. Yoder, The Electrochemical Soc. Inc., Pennington,
NJ, Vol. 89-12, p. 393
Weissmantel, C., K. Bewilogua, C. Schurer, K. Breuer and H. Zscheile (1979), Thin
Solid Films, 6L, L I
Weissmantel, C., K. Bewilogua, D. Dietrich, H. J. Erler, H. J. Hinneberg, S. Klose, W.
Nowick and G. Reisse (1980), Thin Solid Films, 22 ,1 9
Weissmantel, C., K. Breuer and B. Winde (1983), Thin Solid Films, 100.383
Wentorf, R. H., Jr. (1965), J. Phys. Chem., £2, 3063
Wentorf, R. H., Jr. (1974), Advances in High Pressure Research, edited by
R. H. Wentorf, Jr., Academic Press, London and New York, Vol. 4, p. 249
Wild, Ch., N. Herres, J. Wagner, P. Koidl and T. R. Anthony (1989), in Proceedings of
the First International Symposium on Diamond and Diamond-like Films, edited by
J. P. Dismukes, A. J. Purdes, B. S. Meyerson, T. D. Moustakas, K.E. Spear,
K. V. Ravi and M. Yoder, The Electrochemical Soc. Inc., Pennington, NJ,
Vol. 89-12, p. 283
Williams, A. F. (1932), Genesis of the Diamond. Ernest Benn, London
Williams, B. E., D. A. Asbury and J. T. Glass (1989), in Proceedings of the NATO
Advanced Research Workshop on the Physics and Chemistry of Carbides. Nitrides and
Borides. Manchester, U. K.
Williams, B. E. and J. T. Glass (1989), J. Mater. Res., 4, 373
Williams, B. E., H. S. Kong and J. T. Glass (1990), J. Mater. Res. (in press)
Woods, G. S. (1971), Phil. Mag., 473
Woods, G. S. and A. R. Lang (1975), J. Cryst. Growth, 2&, 215
Yarbrough, W. A., A. Inspektor and R. Messier (1990), in Properties and
Characterization of Amorphous Carbon Films. Materials Science Forum, edited by J. J.
Pouch and S. A. Alterovitz, Trans Tech Publications, Brookfield, VT, Vols. 52 and
53, p. 151
Yarbrough, W. A. and R. Messier (1990), Science, 2 4 7 .688
Yarbrough, W. A. and R. Roy (1988), in MRS Extended Abstracts of Diamond and
Diamond-like Materials Synthesis, edited by G. H. Johnson, A. R. Badzian and M.
W. Geis, Vol. EA-15, p.33
Yarbrough, W. A., M. A. Stewart, J. A. Cooper, Jr. (1989), Surf. Coat. Technol.,
Zhu, W., A. R. Badzian and R. Messier (1990), not published
W ei Zhu was born on November 9, 1963, in Zhejiang Province, China. H e
attended The Zhejiang University from 1979 to 1986, graduating with an M.S. degree in
Metallurgy. In July 1986, he was awarded a financial support from The K. C. Wong
Education Foundation Ltd. in Hong Kong and enrolled in the graduate school o f The
Pennsylvania State University. Since then, he has been pursuing a Ph.D. degree in the field
of Solid State Science.
Mr. Zhu is a member of the American Vacuum Society and the Materials Research
Society. He has several papers in the open literature.
Без категории
Размер файла
10 663 Кб
Пожаловаться на содержимое документа