g {‘21: M
7%

~
,. . .. ~. ~
wflzégfi'k

‘31:.“ ‘r1 ’ ‘-

~02;

1‘. ‘-r“~
min-2x}; 0.
”33"“-

x. " ‘1.

.-

“é

. ‘32
‘t‘

”i
w

”n.9,; 2;:- 2-2‘39- .

2m” ~v
3.“.391.
.711“; av

.,
v :fi‘A 5,53%": ‘ I
_k.~...-«,I. g, :2 «mum ‘ ’   .
L5: ‘ V'NN‘" ‘ _ V ~ . ., u.
"9. . 3'1. , y. , - - v , I“ l ’A
m. n ~ ' wfa » «~39
‘ ' ‘ wow-n47 " . VAN-5f
”a, 2: "t" "MAM.
"he LE'W" ‘3' w v
‘3)? 5'". » 1r! ' I ‘a‘ v: . . V
. «Ag-gs“ .1 - ‘ - u v ' V" " "3‘ >
‘-.“ 5‘ ‘1‘ ~ . . w r U‘ f?"
b» v V.” > ‘
u v. I

. v
, . .
r."~"u

 

 

THEIfii

This is to certify that the

thesis entitled

Characterization of the Burst Strength of
Polycrystalline Diamond Thin Films

presented by

Lyndon Edward-David Flowers

has been accepted towards fulfillment
of the requirements for

Master's degree in Materials Science

 

 

Major fessor

Date 9/13/94;

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

ll’l'lllllilllllllll

 

LIBRARY
Mlchlgan State
University

 

 

 

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

 

DATE DUE . DATE DUE DATE DUE
’. 3:13 9'3
NOV: 928;,3W
l 3 n
a o 3

 

 

V: 3
l- 771

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU Is An Affirmative Action/Equal Opportunity Institution

cmm.W3-D.1

CHARACTERIZATION OF THE BURST STRENGTH OF POLYCRYSTALLINE
DIAMOND THIN FILMS

By

Lyndon Edward David Flowers

A THESIS

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

MASTER OF SCIENCE

Department of Mechanics and Materials Science

1994

CHARACTE

Thebur

i3";'\.fi ‘
«Sleds LII-.33

ABSTRACT

CHARACTERIZATION OF THE BURST STRENGTH OF POLYCRYSTALLINE
DIAMOND THIN FILMS

By

Lyndon Edward David Flowers

The burst strengths of diamond thin film (DTF) diaphragms were determined by burst
testing under uniform loading. Si (100) single crystal wafers were abraded with a 1 micron
diamond powder, and photoresist treated to enhance diamond nucleation sites. DTF’s were
deposited on the abraded Si wafers via a microwave assisted chemical vapor deposition
(CVD) process forming DTF on Si wafer samples.

DTF diaphragms were formed by etching small (~2 to 6 mm diameter) holes through
the Si wafers. The circular DTF diaphragms thus produced were supported around their
edges by the Si wafer. The DTF diaphragms were mounted in a pressure vessel and sub-
jected to an increasing gas pressure until failure.

Bursting of the diaphragms were modeled as circular plates rigidly clamped at the
edges and subjected to a uniform load. The maximum stresses and corresponding strains
are discussed, and correlations are made between normalized diaphragm radius and burst

pressure, and between grain size and burst pressure.

DEDICATION

To my brother and greatest teacher, Michael Adrian Kenworth Flowers.

iii

l u Du}:
kgdt‘fmt tar:
2:131 in. I bei
12.3.: 33mm
undergraduitz
me x. the u or
Him D. Cd\:
is funding c
Bil H. Glime
313' COmmitte

lift-Th.

ACKNOWLE EMNETS

I would like to thank Mrs. Norma Pandy-Flowers for giving me the first boost in my
academic career, by having me recite the multiplication tables at an early age. From that
point on, I believe, my colleagues and I considered myself to be strong in the mathematics.
I also acknowledge the moral and financial support of my immediate family through my
undergraduate and graduate years. My mother was always near, willing, and able to help
me in the worst of times. My father I thank for molding me. I acknowledge my advisor Dr.
Eldon D. Case for his belief in my abilities, even when I thought no other eyes could see it.
His funding of my project is also appreciated. I would also like to thank my colleagues
Bill H. Glime, Kiyoung Lee, and Carol Gamlen for their help in this project. I also thank
my committee members Dr. Don Reinhard and Dr. Stan Flegler for their guidience in my

research.

iv

1. lntroductio
llDumom

‘ fl , A l
1.. Hem.

1.4 Interface

1.} DFPOSIU

Table of Contents

1. Introduction. ....... ..........................................................

1.1 Diamond Thin Films (DTF). ...................................................................

1.2 Methods of CVD. .....................................................................................

1.3 Proposed CVD Growth Mechanisms. ....................................................

1.3.1 A Simple Proposed Mechanism. ..........................................................
1.3.2 Role of Hydrogen in DTF Deposition. .....................................
1.3.3 Role of Oxygen in DTF Deposition. ........................................

1.4 Interface between DTF and Si Substrate. ..............................................

1.5 Deposition Parameters and Their Effects. .............................................
1.5.1 Deposition Precusor Gases. ....................................................
1.5.2 Concentrations of Precusor Gases. .........................................
1.5.3 Deposition Temperature. ..........................................................
1.5.4 Deposition Chamber Pressure. ................................................

1.5.5 Deposition Energy Source. .......................................................

2. Experimental Procedure. ...................................................
2.1 Si Wafer Pre-deposition Treatment. ......................................................

2.1.1 Abrading of the Si Wafers. .......................................................

10

ll

12

12

12

15

15

16

17

19

19

21

5 ‘ F v
-.- Alti

« a -

‘ On“

-.. * L‘
D

., ,
-4 D14:

1‘

'-.5 \IC“;
2.6 Rim

1* .

_ )

I.)

thEXI

I)

2.1.2 Cleaning of the Abraded Si Wafers. ........................................ 24

2.1.3 Fracturing of the Si Wafers. ..................................................... 24

2.2 Film Deposition. ...................................................................................... 28
2.2.1 Microwave Disk Reactor Preparation. .................................... 28

2.2.2 Deposition of DTF’s. ................................................................ 30

2.2.3 Deposition Conditions and Specimen Labeling. ...................... 33

2.3 Optical Microscopy. ................................................................................. 35
2.4 Diaphragm Preparation. ........................................................................... 35
2.4.1 Protective Coating for Si Wafers. ............................................ 35

2.4.2 Etching of Si Wafers. ................................................................ 35

2.4.3 Anisotropic Etching of Si Wafers. ............................................ 40

2.5 Measuring DTF’s Diameters. ................................................................ 42
2.6 Raman Spectroscopy. .............................................................................. 44
2.7 Diaphragm Bursting. ................................................................................ 44
2.7.1 Pressure Vessel and Data Acquisition Apparatus. ............... 44

2.7.1.1 Leak Testing the Pressure Vessel. ......................... 50

2.7.2 Diaphragm Bursting Procedure. ............................................... 52

2.7.2.1 Obtaining the Burst Pressures. ............................... 53

2.8 SEM’s Preparation and Testing Procedures. ......................................... 53
2.8.1 Retrieving and Mounting of the DTF Fragments. .................. 54

2.8.2 Coating the Mounted Fragments. ............................................ 56

2.8.3 Specimen Insertion into SEM’s. .............................................. 56

2.8.3.1 Hitachi S-415A. ......................................................... 59

vi

J

2.8.4 SE

311011111 3128.

 

211.111.;
3.111F11an1t1‘
3.11 Additiini‘:

2.11.1 P

3.11.31

2.11.35

3. Results and I]

1.1 Rexuhx in '

3.11 R
3.1.21
3.: Film Dep.

1 1
.1.) Results in

2.8.3.2 JOEL JSM-35CF. ...................................................... 59

2.8.3.3 Hitachi S-2500C. ....................................................... 63

2.8.4 SEM Procedures for Testing Specimens. ................................ 64

2.9 Grain Size. ................................................................................................ 65
2.9.1 The Average Line Intercept Method .......................................... 66

2.10 Film Thickness. ...................................................................................... 67
2.11 Additional Specimens Obtained for Testing. ........................................ 69
2.11.1 Photoresist Pre-deposition Treatment ................................ 69

2.11.2 Deposition Conditions. ........................................................... 69

2.11.3 Specimen Labeling. ................................................................. 70

3. Results and Discussion. ..................................................... 72
3.1 Results in Wafer Preparation. ................................................................. 72
3.1.1 Results of Abrading. ................................................................. 72

3.1.2 Photoresist Treatment. ............................................................ 72

3.2 Film Deposition. ...................................................................................... 75
3.3 Results in Diaphragm Preparation. ......................................................... 76
I 3.3.1 Results of Etching. .................................................................... 76
3.3.1.1 Anisotropic Etching. .............................................................. 79

3.3.2 Result of Dimpling. ................................................................... 80

3.4 Data Collected. ........................................................................................ 83
3.4.1 Burst Pressure. ........................................................................ 83

3.4.2 Grain Size. ................................................................................. 83

vii

3.5 Rimdn Sp

3.5T11t’01t'11t'.

3.4.3 Film Thickness. ......................................................................... 86

3.4.4 Film Morphology. ...................................................................... 87
3.5 Raman Spectroscopy Results. ................................................................ 89
3.6 Theoretical Modeling of the DTF Diaphragms. ..................................... 92
3.6.1 Timoshenko [Linearized Model]. ............................................ 93
3.6.2 Non-linear Model. .................................................................... 94
3.6.3 Analysis of WF#60 Data Points. ............................................. 96
3.6.4 Empirical Equation. ................................................................... 98
3.6.5 Analysis of the Total Accumulated Data Points. .................... 101
3.7 Other Observed Trends. .............................................................. 102
3.7.1 Thickness Variation. ..................................................... 103
3.7.1.1 Thickness versus Wafer Position. ............................ 104
3.7.2 Grain Size Variation. .................................................... 104
3.7.3 Temperature Variation. ................................................ 107
4. Conclusions. ....................................................................... 109

Appendix A. Micrographs of specimens DF-Nl, DF—N2, DF—N3,

DF-N4, and DF-NS. ................................................................. 111

1. Grain Sizes. .................................................................................... 112

II. Thicknesses. .................................................................................. 1 16

III. Other Observations. .................................................................... 119

Appendix B. Micrograph of specimens WF#25. ............................................. 122

viii

Appendix

Appendix

Appendi}

AWendi

Referem

1. Grain Sizes. .................................................................................... 123
H. Thicknesses. .................................................................................. 126
HI. Diaphragm Diameters. ................................................................ 127
IV. Interfacial Grains. ........................................................................ 129
Appendix C. Micrograph of specimens WF#60. ............................................. 130
1. Grain Size. ...................................................................................... 131
I]. Thicknesses. .................................................................................. 133
v III. Diaphragm Diameters. .................................................................. 135
IV. Interfacial Grains. ........................................................................ 137
V. Other Observations. ..................................................................... 139
Appendix D. Program in basic that runs the burst test. ...................................... 140

Appendix E. Plots of burst pressure versus the fourth power of
the ratio, film thickness divided by diaphragm diameter. ......... 148

Appendix F. Plots of burst pressure versus the square root of
the ratio, film thickness divided by diaphragm diameter. ......... 152

References. ......................................................................................................... 156

Table 1. D
C

Table 2. B
[

Table 3.. E

I

Table {E

W

page

Table l. Deposition conditions of sets DF—Nl, DF-N2, DF-N3,

DF-N4, and DF-NS. ............................................................................... 34
Table 2. Burst pressures, diameters, thicknesses, and grains sizes for

DF-Nl, DF-N2, DF-N3, DF—N4, and DF-NS specimens. ................ 82
Table 3. Burst pressures, diameters, thicknesses, and grain sizes for

WF#25 specimens. .................................................................................. 84
Table 4.Burst pressures, diameters, thicknesses, and grain sizes for

WF#60 specimens. .................................................................................. 85

figure 1 131.
clerical tea
abmcbng a
nest. CII'C1€\
dismal 11111

resents 11111:

figure 1 1b
31.1.1 he:
113331511713 (

FRMKH

“guru 1c
113m 1hr 31

Figul-e l 11
TOI‘mS a C-
ain-16116 1
MEET

“gut-e I (

dilltted :

F1EUre 1

mil£1116.

HRNI

Wei .
kmkm

['1 [E' .

Figure l (a). Proposed growth mechanism showing the sequence of
chemical reactions. A hydrogen atom activates the surface by

abstracting a surface H atom creating a radical site. The large coarse
mesh circles are the top layer carbon atoms and the large circles with
diagonal lines are lower layer carbon atoms. The small circles rep-

resents hydrogen atoms. After [7]. .......................................................................

Figure l (b). An acetylene molecule is added to the radical site. The
radical then propagates to the added molecule 1(b) to 1(c). The car-
bon atoms of the added acetylene molecule are represented by the

large fine mesh circles, and the radical site by a black dot. After [7].

Figure 1 (c). The radical then bonds to the hydrogen atom abstracted

from the surface carbon atom of the lower layer. After [7]. ..............................

Figure l (d). The carbon radical now available on the lower layer
forms a C-C bond with the carbon atom of the originally added
acetylene molecule. This creates a free radical on the adjacent site.

After [7]. ..................................................................................................................

Figure l (e). The sp3 bond is formed by this step. The addition of
another molecule starts a similar reaction sequence which is

depicted in (f) through (i). After [7]. ....................................................................

Figure l (f). The radical site is propagated to the added acetylene

molecule. After [7]. .................................................................................................

Figure l (g). A hydrogen atom is abstracted from a top layer carbon

atom leaving a carbon radical. After [7]. ...............................................................

xi

page

Tuurel cthr‘e
53311317? from 11‘.

mu to '11: 1

figure 1 bribe
tit. "1 leuer 111‘

from be louer 1.

I ’"I. .‘
figure 2. 5m:
trace new:
:3: no. gm 1

rue; fill , ~ . L‘
be 111'“ i. m.
A

1‘11“" * "M1",
“DJ“ f‘S-lklhlll.

“$1193. A st‘

Clix: 13 Float};

ligurel 1 a1. .1
3i '1 11ers um-
1‘31 151 u is p1
3115 sandwich 1
“311184 1b1.

S""731119‘11 sir!

film“ 1c 1.
as indicated,
W 3111311611

“are S 1b 1
“tier 390%

Figures 1c
lob. '

'e
.....
‘a

Figure l (h). The carbon radical forma a CO bond with another car-
bon atom from the newly added acetylene molecule propagating
he radical to the molecule. After [7]. .....................................................................

Figure 1 (i).The radical on the acetylene molecule forms a QC bond
with a lower layer carbon atom while abstracting a hydrogen atom
from the lower layer. After [7]. ..............................................................................

Figure 2. Sketch of DTF/Si substrate interface. The DTF and Si in-

terface showing diamond seeds embedded in Si wafer. Grains nuc-

leate and grow from these sites. Crystal growth is initially lateral

and upwards. Upon contacting their neighbors the crystals grow

upward forming columar structures. .......................................................................

Figure 3. A schematic diagram of the microwave plasma reactor
used to produce the diamond thin films utilized in this study. .............................

Figure 4 (a). Abrading of Si wafers. The shiny polished sides of the

Si wafers were abraded with 1 micron diamond powder. Diamond

Powder was place between the polished surfaces of two wafers, and

the sandwich placed in a dish. ................................................................................

Figure 4 (b). A light load (P) was applied to the Si/diamond powder
sandWich with the fingers. ......................................................................................

Figure 4 (c). The motion applied to the wafers with the fingers is
as indicated. Twisting, side to side, and back and forth motions
were applied to the sandwich. ................................................................................

Figure 5 (a). Fracturing of Si wafers. The edge is first chipped with
the diamond tip pen. ................................................................................................

Figure 5 (b). The wafer is flipped over on tissue paper and the tip

0f the diamond pen used to press down on the opposite side of the

wafer above the previous chip. This causes a crack to propagate

away from the loaded point cross the wafer. .........................................................

Figure 5 (c). Top view of (b) showing the crack propagation direc
tiOn.

00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000

xii

13

20

22
22
23

26

26

, - “I,
T19ure1).AH
. “an.
1"

l T "7"
1w“ .
‘ ‘ h 13”.
1‘11. .'

H .
figure A \
"st: 1:1 put:

o
.

E3 ..........
.

liture 8 1.31.
uttered 55
ire Fifi"???

1
"13 '" “pie 9.
fl‘\ sJku‘“‘ ‘

Flture81b1

be rapture-c

figure 9, E;
fifififlr
twang
’14 heifer.

(ahilflf

F131118 10, ‘A
mflehhmg
Tharp
“3611-5 T
The bar repr

mindihinfi

a“ bolted 1.

Figure 6. Approximately 9 pieces of abraded Si wafers are placed
on the graphite plate. Each piece of Si wafer was approximately

1cm by 1cm. ...........................................................................................................

Figure 7. A schematic of the microwave reactor specimen chamber
used to process the diamond thin film/silicon specimens used in this

study. . ......................................................................................................................

Figure 8 (3). Protective coating for the Si wafers. The Si wafer is
protected by Torr Seal on top and edges and by the diamond coating
on the bottom. The only area unprotected form the etch was through

the circular tube. ......................................................................................................

Figure 8 (b). Top view of the Torr Seal protected Si wafer showing

the unprotected Si wafer that was exposed to the etch. ......................

Figure 9. Etching Apparatus. Each specimen was placed into a 100
ml. beaker ,which had KOH etchant in them. The 100 ml. beakers

were then placed into the deionized water bath in the 2000 ml. jack-
eted beaker. The water bath was kept at 60° C which kept the KOH

etch at ~60° C. ........................................................................................................

Figure 10. Anisotropic Etching Apparatus. The same method as reg-

ular etching was used but the etching occurred in an N2 atmosphere.

Figure 11. Photomicrograph of the diamond diaphragm of specimen
WF#50-6. The diaphragm is the lite gray circular area in the middle.

The bar l"5131‘esents 1.3 mm. ....................................................................................

Figure 12. Burst Testing Apparatus. ....................................................................

Figure 13 (a). Side view of pressure vessel used to burst the dia-
mond thin film specimens used in this study. The pressure vessel

was boned horizontally onto a table top. ..............................................................

Figure 13 (b). Front view of pressure vessel showing the inset rub-

berO-r'mg. .........................................................................

Figure 13 (c). Front view of flat metal disc showing the 2.5 mm hole

in the center and the three holes for the bolts. ......................................................

xiii

00000000000000000

29

37

39
41
43
45
46

47

47

figure1418‘-

use ------------

figure 14 'b'

.1 4“" I.
31:14. c1\.h.““

figure 15 1a!

Rib W311“ Cl

726115 \lt‘ft’ T5
1"."
”wk .‘.IIIID"

figure 15 1b
he and rem;
fer fell.

figure 16. H

* r1 ‘1.»

li'JiUM :1th
m truer. '.
5-251th SE‘
screed sect
seiner. eh;

M2311 there

figure 17. 1f
reflux-1.111 g of
hemenu 0r
P1346. The 31
313mb base

Fleure 18. 1
1liberieal kn
Db1111011. w,
llfilling chm
“911111 “b,"
1M6; y1d1

Figure 14 (a). Diamond thin film diaphragm specimen and flat metal
disk. ......................................................................................................................... 49

Figure 14 (b). The specimen is glued onto the flat metal disk. The
metal disk, with the specimen glued in place, is then ready to be
mounted into the pressure chamber for burst testing. .......................................... 49

Figure 15 (a). Fragment mounting on SEM stubs. The DTF frag-
ments were entangled in cotton fiber after the burst test. The frag-
ments were retrieved by using a toothpick to pull the cotton fibers
apart. ........................................................................................................................ 55

Figure 15 (b). The fragments fell onto the sticky tab on the SEM
stub and remained in the random positions and locations in which
they fell. . .................................................................................................................. 55

Figure 16. Hitachi S-2500C SEM specimen holder. The specimen

holder assembly is attached to the insertion chamber by the temp-

orary cover. The 1.0 cm by 1.5 cm specimen stub was adhered to the

S-ZSOOC SEM stub with a carbon tape. The S-2500C SEM stub was

SCYCWed Securely onto the brass holder which is inserted into the

when chamber by the spherical knob and insertion rod and re-

mained there during testing. ................................................................................... 57

Figure 17. Hitachi S-415A SEM specimen holder. Specimen holder

c(’“SiSting of a metal stub base, and a spring clip by whiCh the DTF

fragments on the 1 cm by 1.5 cm metal specimen stub is held in

place. The Spring clip is adjustable and screws into or out of the met-

a] stub base to adjust the specimen height into focusing range. .......................... 70

“guts 18. JEOL JSM-35CF SEM specimen holder assembly. The
spherical knob was used to screw the steel rod into the brass stage.
Th? .kHOb was also used to move the brass stage between the small
waltlng Chamber and the specimen chamber. The temporary cover

would replace the true cover only while the specimens were being
loaded. . 61

OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

Figure 19. JOEL JSM-35CF SEM specimen chamber and insertion
chamber. .................................................................................................................. 62

xiv

 

‘ 1
Han.
minis:

i . . .
or; 1 JFK
alt Uh. .

mel
ma

“Fair; 1
igmu

I
“a: ‘?o~
TH» Owns:

hwml
iC‘L') 01'] g}

‘1

: ' '9 A.
“Latin

Figure 2.
Ekufi
WEIR .3}

mm. {Esp

“glue 1,
Whine .

c (1
Rafi“ ’_
‘LL‘ “u

Bates.

Figure 20. SEM micrograph of specimen WF#60-25 showing the
thickness of a fractured polycrystalline diamond diaphragm. The

dotted bar represents 5 microns. .....................................................................

Figure 21. The relative positions for the 26 specimens taken from
wafer WF#60 were recorded for later film thickness versus relative

position correlation. ........................................................................................

Figure 22 (a). Micrograph of the interfacial surface of the DTF
specimen WF#25-ll showing distinct grain boundaries define by
the dark lines. A grain size of 0.71 micron was detennined from the

micrograph. The dotted bar represents 5 microns. ..........................................

Figure 22 (b). Micrograph of the interfacial surface of DTF speci-
men

WF#60-l showing distinct grain boundaries define by the dark lines.
A grain size of 1.00 micron was determined from this micrograph.

The dotted bar represents 3 microns. ..............................................................

Figure 23. The flaking off of the DTF from the Si substrate is obv-
lens on specimen WF#25-12. The etch seem to have penetrated the
diamond film and attach the etch beneath. The circular diamond

diaphragm is also observed in the center. The bar represents 1 mm. ...............

Figure 24. Cover slip modification. A glass cover slip was added to
protect the diamond coating from the KOH etch. The inner dia-
meters of the large and the small tubes were 6.8 mm and 2,3, or 4

mm, resP€=Ctively. ...........................................................................................

Figure 25, Typical diamond diaphragm (WF#60-15) produced by
etching a concave and circular dimple that was grounded into the Si

wafer With a TEM sample preparation dirnpler. A1 mm bar is indi-
cated.

0..
ooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo

Flgure 26. Twinning in specimen DF—Nl-4 are indicated by arrows.

A 1 miCTOn bar is represented on the photomicrograph. ................................

Figure 27 (3). Raman Spectroscopy; Raman spectra for specimen
DF‘N 1‘6 shows the characteristic 1333.5 cm'1 diamond peak and

relatively dominant region near graphite 1500 cm'1 peak. ............................

XV

68

71

73

74

77

78

81

88

91

. ‘-
hwm-
19.5 |:-‘:'

I . .l v-
I'h‘i", ,E“
Liiu85‘

Figure 28
rm {:1

i-\Yhafi
.1“ L45 AA

Figure 2‘
5"Vifit

.‘w _. .

midis:

Figure 3
Lakes \\
31:19;
me iii?
5 3 mix

L. lite jf

Figure .1
$382511 3

Hgllre .-‘
its 0586:
(if DF-N

“gum ‘:

M}

V»

Figure 27 (b). Raman curve for specimen DF—N4-4. The character-
istic 1333.5 cm‘1 diamond peak and a dominant region near the
characteristic 1500 cm’1 graphite peak is observed. ............................................ 91

Figure 28. Plot of P versus [i/d14 for 21 WF#60 diamond films burst
in this study, where P is the burst pressure, t is film thickness, and
dis the film diameter. ............................................................................................. 97

Figure 29- Plot of P versus [t/d]1/2 for 21 WF#60 diamond films
burst in this study, where P is the burst pressure, t is film thickness.
and d is the film diameter. ...................................................................................... 99

Figure 30. Thickness versus Position for WF#60. No thickness

values were obtained for specimens #10, #13, #20, #22, and #24. A

slight pattern of thickness decreasing with increasing radial dis-

tance from the bottom center is observed. This pattern might be due

to a higher substrate temperature corresponding to a thicker regions

of the deposited DTF [21]. .................................................................................... 106

Figure Al. The deposition surface of DF-N1-4. A large grain is
seen at the center of the micrograph. A 1 micron length is indicated. ................. 112

Figllre A2. The deposition surface of DF-N3-3. No diamond facets
are observed, therefore this film was consider (as was the entire set

9f PF-NB films) to be mostly graphitic carbon. A 1 micron bar is
l"dim-Med. . ................................................................................................................ 112

Figure. A3. The deposition surface of DF—N2-2. A grain size of
0-53 mlcrons was determined from this micrograph. The film
appears to be covered with minute lumps. A 3 micron dot-bar is

indicated. ................................................................................................................. 113
313“." A4. The deposition surface of DF-N2-4. The surface is
2osmrriated by (111) facets. A grain size of ~0.65 micron is seen. A
' "Heron dot-bar is indicated. .............................................................................. 113
Fliggre A5. The non-diamond film DF-N3-2. The typical (1“),
( Land (110) facets are not present. A 1.5 micron dot-bar is indi-
cated. ....................................................................................................................... 114
Figure A6. The deposition surface of DF-NS-S. Pyramidal (111)
facets dominate the surface. A 2.7 micron dot-bar is indicated. .......................... 114

xvi

figure A'

. ‘ .,.
1am”. 4' J.
7‘ rips-fi*\
..

. if N
7‘1...qu IA '-

Figure A

figure A

WP. 15

uh '. u-
‘

T317333}
5 1

Figure A

Figure A
“bdam
13.11: 6 de

9',
CLUE»

HmmA
T0515 L1 53
531:1:
Figure A

“SW A
$319th
miPOE:
“1 is ind

Fltsure A
9W imm-
and boui-
bar l5 1nd

Figure A
Kim] 3 f3

“RNA
minim
We [hail
*1 A 1.

Figure A7. The deposition surface of DF-N5-6. The surface is
dominated by (111) facets. A grain size of 0.64 micron was deter-
mined from this micrograph. A 3.0 micron dot-bar is indicated. ........................... 115

Figure A8. The deposition (top) surface and edge of DF-N1-4. The
top surface is dominated by pyramidal (111) facets. Columnar stru-
cture is see on the edge of the film. A 1 micron bar is indicated. ......................... 116

Figure A9. The edge of DF-N3—5. No columnar structure is evident
through the thickness. A thickness of 6.4 was determined from this
micrograph. A 1 micron bar is indicated. ................................................................ 116

Figure A10. The edge view of DF-N3-2. No columnar structure is
evident through the edge. The thickness determined from this
micrograph is 3.5 microns. A 6 micron dot-bar is indicated. ................................ 117

Figure All. The edge view of DF-N5-3. A thickness of 3.7 micron

was detern'rined for this DTF. The dominant (111) pyramidal facets

on the deposition surface and the columnar structures through the

thickness are both evident A 1 micron bar is indicated. ...................................... 117

Figure A12. The edge view of DF—N5-4. A thickness of 4.3 mic-
rons was determined from this micrograph. Small lumps on the
mterfacral surface is also obvious. A 4.3 micron dot-bar is indicated. ................ 118

Figure A13. The thickness of DF-N5-6 was determined to be 5.3
nucrons from this micrograph. A 5 micron dot-bar is indicated. ........................... 118

Figure A14. The bottom of the Si wafer supporting DF-N4-7. The

fmrphdogy is the result of etching the (100) single crystal Si wafer

1“ aPmaSSium hydroxide (KOH) bath (section 2.4.2). A 10 micron

bar 18 indicated. ....................................................................................................... 119

Figure A15. The edge and bottom view of the Si wafer supporting
sPecu'nen DF-N4-8. The Si wafer is the thicker material. The edge

and ,b9n0m of the diamond thin film (DTF) is.also see. A 10 micron
b” ‘3 Indicated. ....................................................................................................... 119

“film A16. The interfacial surface of DF-N4-8. The surface is flat
W1 a few scattered lumps. A 1 micron bar is indicated. ..................................... 120

Figure A17. The deposition surface of DF—N5-8. At high (x20k)
magmfication minute black dots are seen on the facets. Small struc-
true that do not resemble square or pyramidal diamond face are also
see- A 1.5 micron dot-bar is indicated. .................................................................. 120

xvii

Figure A18. A high (x70k) magnification of the deposition surface
of DF—N5-8. The small black dots on the surface (see Figure A17)
appears to be holes, perhaps through the gold coating (section

2.8.2). A 430 nm dot-bar is indicated. ...................................................................

Figure Bl.The Deposition surface for WF#25-8. The surface is
dominated by (111) faces. A grain size of 0.75 microns was deter-

mined.A 3.8 micron dot-bar is indicated. ...............................................................

Figure B2. The deposition surface of WF#25-9. The (111) faces
dominate the surface. A few large cracks are observed. A grain size

of 0.6 micron was determined. A 3.8 micron dot-bar is indicated. .......................

Figure B3. The deposition surface of WF#25-10. The grains are
not well defined. Small holes are seen at the grain boundaries. A
grain size of 0.99 was determined. A 5 micron dot-bar is indicated.

Figure B4. The deposition surface of WF#25-11. The (111) faces
dominate the surface. A grain size of 0.67 was determined. A 3.8

micron dot-bar is indicated. ....................................................................................

Figure BS. The deposition surface of WF#25-15. The (111) faces
dominate the surface. A grain size of 0.47 micron was determined.

A 3.8 micron dot-bar is indicated. ..........................................................................

Figllre B6. The edge view of WF#25-7. The edge surface was near-
ly P5931161 and showing the true length thus no tilt correction was
applied, A thickness of 5 microns was determined. A 6 micron dot-

bar iS indicated. .......................................................................................................

Figure 3:7. The edge view of WF#25-9. Columnar structure across
the edge IS evident. No tilt correction was required. A thickness of

2-3 Was determined. A 2.5 micron dot-bar is indicated. .......................................

Figure B8, The diamond diaphragm (light circular center) of speci-
3‘3“ WF#25-7. Many DTF blisters are observed on the Si, possibly
ue to tholes allowing the etch to attack the Si beneath the DTF.

The bar l'epresents 1.3 mm. .....................................................................................

Figure B9, The diamond diaphragm (light circular center) of speci-
men WF#25-10. The good circularity obtained from initial dim-

Figure 310. The diamond diaphragm (light circular center) of
specrmen WF#25-12. The DTF blistering is severe and causes the

DTF t0 flake from the Si substrate. The bar represents 1 mm. ........

xviii

OOOOOOOOOOOOOOOOOOO

ng is Obvious. The bar represents 1 mm. ...........................................................

OOOOOOOOOOOOOOOOOOO

121

123

123

124

124

125

126

126

127

127

128

Figure B]
helix
2‘2: :1: tit

3; ,u‘ w

A .5». u-s
Figure B
“as are
its. deter

LFH'I "if“:
Muted

FigureB

>.-r- '1
L211.) 4.“:

, n1 .‘..
“"115“ It“.

hth

‘1'“; §-.
xuufik 1d

TIGERS ‘

Figure (
56.1.1316 f;
Humans
her is in;

FiEUre (
Square f;
mii'l'ljm
b3} i5 iFJI

Flglue t
5'6 mm
MS of

Fri-£70111 (

Figure Bll.The diaphragm (light circular center) of specimen
WF#25-l4. The Torr Seal is seen on the outer edges surrounding
the protective fused silica tube. Inside the tube is seen the DTF on
Si and the diaphragm. The bar represents 1.3 mm. ............................................... 128
Figure B12. The interfacial surface of WF#25-11. The individual
grains are partially distinguishable. A grain size of 0.71 microns
was determined from this micrograph. A dot-bar of 5 micron is
indicated. ................................................................................................................. 129

Figure B 13. The interfacial surface of WF#25-14. The individual
grains are distinguishable. A grain size of 0.86 microns was deter-
mined from this micrograph. A dot-bar of 3.8 micron is indicated. ....................... 129

Figure C l .The deposition surface of specimen WF#60-2. The

square facets of the (110) or (100) crystals. A grain size of 1.14

microns was determined from this micrograph. A 3.8 microns dot—

bar is indicated. ....................................................................................................... 131

Figure C2. The deposition surface of specimen WF#60-6. The

square facets of the (110) or (100) crystals. A grain size of 1.04

microns was determined from this micrograph. A 3 microns dot-

bar is indicated. ....................................................................................................... 131

Figure C3. The deposition surface of specimen WF#60-18. The

square facets of the (110) or (100) crystals. A grain size of 1.2

microns Was determined from this micrograph. A 3.8 microns dot-

bar ls indicated. ....................................................................................................... 132

Figure C4. The deposition surface of specimen WF#60-19. The

Square facets of the (110) or (100) crystals. A grain size of 1.26

"“Cfons was determined from this micrograph. A 3.8 microns dot-

bar IS indicated. ....................................................................................................... 132

gig”? C5. The edge view of specimen WF#60-6. A thickness of
'6 mlcrons was determined from this micrograph. Columnar struc-
tures 0f the diamond crystals are obvious across the thickness. A 6

micron dOt—bar is indicated. .................................................................................... 133

Figure C6. The edge view of specimen WF#60-14. A thickness of

3‘2 IT“(31'0ns was determined from this micrograph. Columnar struc-

tures of the diamond crystals are obvious across the thickness. A 3

mlCI‘on dOt-bar is indicated. .......... . ........................................................................ 133

xix

 

- -.
Figure C r
3.1, 31:31:
11's; 3f in:

1
. 1‘
m ”1": .

“uh: V‘-

Figure C 8
:3: STJL'I‘J

can. A 5 rr

Figure C 9
hid-1:6.
40;; -;r; .
n-Xy .\£Aa5.
T3 jiffy.
Figure C 1
We‘re! 1-]:
pisietihe

3’3; ._ .
«~21in

Figured
We. 15
mile hi).
3133 . and

Figure C 1
“Fifi 1- 5 a"

he rused 3
and he d1;

i'flre C1.
3161‘» ldugd
6mirlec‘

“iii idem .
l ‘ ‘ ‘
06.5mm“

Cited.

Fié'llre C1:

d: a .

F
' lgure C“
ma‘idual

Figure C7. The edge view of specimen WF#60-18. A thickness of
3.4 microns was determined from this micrograph. Columnar struc-
tures of the diamond crystals are obvious across the thickness. A 2

micron dot-bar is indicated. ...............................................................

Figure C8. The edge and bottom of specimen WF#60-25. Colum-
nar structures of the diamond crystals across the thickness are obvi-
ous. A 5 micron dot-bar is indicated. ................................................

Figure C 9.The diaphragm (light circular center) of specimen
WF#60-6- White Torr Seal epoxy is seen outside of the protective
fused silica tube. Inside the tube, are the DTF on Si (dark area), and

the diaphragm are seen. The bar represents 1.3 mm. ........................

Figure C10. The diaphragm (light circular center) of specimen
WF#60-12. The Torr Sea] is on the outer edges and surrounds the
protective fused silica tube. Inside the tube, are the DTF on Si (dark

3163). and the diaphragm. The bar represents 1mm. .........................

Figure C11. The diaphragm (light circular center) of specimen
WFfGO- 1 5. The Torr Seal is on the outer edges surrounds the pro-
tectlve fused silica tube. Inside the tube are the DTF on Si (dark

area), and the diaphragm. The bar represents 1 mm. .......................

Figure C12. The diaphragm (light circular center) of specimen
WF#60-23. The Torr Seal on the outer edges surrounds the protec-
trve fused silica tube. Inside the tube are the DTF on Si (dark area).

and the diaphragm. The bar represents 1 mm. .................................

Figure C13. The interfacial surface of specimen WF#60-1. The
mleldual grains are well defined. A grain size of 1.00 microns was
determined form this micrograph. A 3 micron dot-bar is indicated.

Figure C14. The interfacial surface of specimen WF#60-6. The
rn “’ldual grains are well defined. A grain size of 0.98 microns was

cated

‘J‘d‘Vidflal grains are well defined. A grain size of 0.82 microns was
6mm form this micrograph. A 3 micron dot-bar is indicated.

Figure C15. The interfacial surface of specimen WF#60-15. The

FigP'fe C16. The interfacial surface of specimen WF#60- 17. The
mdw‘dmll grains are well defined. A grain size of 1.01 microns was

cated

OOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOO

OOOOOOOOOOOOOOOOOOOOO

detemined form this micrograph. A 3.8 micron dot-bar is indi-

detemlined form this micrograph. A 3.8 micron dot-bar is indi-

134

135

135

136

136

137

137

138

138

Figure (
h?:m-

Figure 1
tier.

dlflt Iva}!
' 1

.iLLr...)

Figure

‘n-n '3‘
~.i..,.\_1

Figure

92>; -‘ T‘l
Ans-L. “1

NZ, ..

'0

Figure
"4- a:
14:14. 111

x;

Figure
31:1; ii
55.

FQUre

CA _\4
who. '
“ML. :1
”FHA-
r: x
Q

Fig” re
the con

mt fur

Figure
3125 i0!

Flgllre

5‘25 for

Figllre
sue f0;

Figure
“ZeiOI

Figure C17. A crack on the deposition surface of specimen
WF#60-3. Square (110) or (100) dominate the morphology. Some
intragranular cracks appears to be present. A dot-bar of 3.8 microns
is indicated. ............................................................................................................. 139
Figure C18. A crack on the interfacial surface of specimen
WF#60-26. This crack is clearly intergranular. A dot-bar of 2.7
microns is indicated. ............................................................................................... 139

Figure E1. Insignificantly small data set DF—Nl. Included only to
complete Appendix E. ............................................................................................. 149

Figure E2. Plot of burst pressure versus the fourth power of the
ratio, film thickness divided by diaphragm diameter for data set DF-
N2. ........................................................................................................................... 149

Figure E3. Plot of burst pressure versus the fourth power of the

ratio, film thickness divided by diaphragm diameter for data set DF-
N4. ........................................................................................................................... 150

Figure E4. Plot of burst pressure versus the fourth power of the
11:31:10, film thickness divided by diaphragm diameter for data set DF-
- ........................................................................................................................... 150

Figure E5. Plot of burst pressure versus the fourth power of the

ratio, film thickness divided by diaphragm diameter for data set
WF#25. . . . . ................................................................................................................ 15 1

Figure F1. Insignificantly small data set DF—Nl. Included only for
the Completion of Appendix F. ............................................................................... 153

Figure F2. Plot of the burst pressure versus the square root of grain
srze for data set DF—N2. ........................................................................................ 153

Figure F3. Plot of the burst pressure versus the square root of grain

sue for data set DF—N4. ........................................................................................ 154

Figure I“4. Plot of the burst pressure versus the square root of grain

Slze for data set DF—NS. ........................................................................................ 154

Figure F5. Plot of the burst pressure versus the square root of grain
512° f°r data set WF#25. ........................................................................................ 155

Figure F6. Plot of the burst pressure versus the square root of grain
5‘26 for data set WF#60. ........................................................................................ 155

xxi

1. Iain
0:

M",
ll...lru h-

r _ ‘0‘
Kit Uri:

[3"‘15‘1‘.
Na Mart)
.

.1253. 61

3: 11.121 I

a»; .. . I,
.i'\q~ 34

o~a

. “EL-iii: is”

Tile-'1, i

"me; {1

1. Introduction

One of the earliest documentation of diamond dates back to about 46 AD and centers
around the early city of London [1]. It is fair to say that diamond has been long regarded as
the ultimate in aesthetic value, among materials. Today, with the help of advancing tech-
nology, it is amazing to learn that diamond is indeed a unique material in many respects.
Diamond is unique in its mechanical, optical, electrical, chemical, and thermal properties.
Certainly these superior qualities of diamond compared to other materials were not appre-

ciated even 100 years ago. It is interesting to ponder why people have had such a strong
affinity to diamond so long before it’s many unique properties were even established.

Diamond’s remarkable properties such as hardness [2], broad range of electro-mag-

“etiC transparency [3,4], high electrical resistivity [3,5], excellent chemical resistivity at
room temperature [3,5], and high thermal conductivity [3], makes diamond the ultimate
material candidate for many applications. However, the high cost of natural diamonds and
the limited knowledge of its growth mechanism (which limits the synthetic abilities of dia-
mo“(18) have kept diamonds from wide spread use. It is not likely that diamonds will loose
their aesthetic appeal to people, therefore natural diamonds may never become a cheap
matfitrial. The ability to produce cheap synthetic diamonds seems the only way that dia-
m0“(is can possibly see wide spread use.

Most of the well established properties of diamond are based on its single crystal
fol‘m, as natural single crystal diamonds were always available for testing. The properties
of polycrystalline diamond however, especially its growth mechanism, are not as well
eStablished. Synthetic diamond has only become popular in recent decades, which may be

One of the reasons why many of their properties are not well established. It is

.1
,s‘,‘ 9n. 1"“;
“,2!th a. In

.. . J!
4- a} r
Q‘LJ‘, ilk. \l

“""""t“\\'

Jar-r112 111.1

ifns'tiam

Ybsiusl \
Thu

\ .‘farwn 1
hu‘ulni;
;

53am
5

ir‘At-t'

rim 1

2

conceivable that some of the properties of single crystal diamond might also be found in
diamond’s polycrystalline form. However. it is unlikely that all the properties of the single
diamonds will be realized in polycrystalline diamond. Properties that are grain boundary
dependent might be inherent in polycrystalline diamc‘md.

The optical transparency of polycrystalline diamond may be significantly affected by
scattering due to grain size and concentration. Grain boundary impurities might affect the
transparency of polycrystalline diamond by increasing its absorbancy. Impurity concentra-
tion at grain boundaries might also affect the diffusive, electrical and chemical properties
0f Polycrystalline diamond.

With the ability to synthesize single and polycrystal diamonds has come the desire to
Characterize and to control these properties. This desire is certainly based on the hope of
son'lfbday harnessing the aforementioned exceptional properties of diamond for a multitude
0f Practical uses. To this end much research is being done. In this investigation the burst
Strengths of diamond polycrystalline thin films were tested under uniform loading and

related to film thickness, diaphragm diameter and grain size.

1'1 Diamond Thin Films (DTF)

IMarry situations exist in which the properties of a diamond polycrystalline thin film (DTF)
can be utilized. Diamond’s high electrical resistivity and good heat conductivity suggests it
as a good coating for integrated circuits. The diamond polycrystalline coating, being a good
insulator, would protect the integrated circuit from shorting out due to dust particles etc.
that might fall on the integrated circuit. The diamond polycrystalline coating, being a good

heat conductor, would also enhance heat dissipation. Diamond’s extremely high hardness

ms 1'. '1
IhflRr
rush;
he regal:
15:14:11.: g

DT

90 -‘ . ‘
“Fm:

L'j 9".1;F

\lik‘ij
5

i. .
rials

milieu

imam

”Std C

3

makes it useful as a wear protective coating. Diamond’s optical transparency from the UV
to the IR region, including the visible, makes it a candidate as a window. Diamond’s prop-
erties of high chemical inertness, and optical transparency at high temperatures would fill
the requirements for many unique window applications such as might be found in some
(Noran) energy dispersive x-ray analysis equipments.

DTF’s may be fabricated to be piezoresistive [6], which, combined with the other
Properties mentioned, suggests DTF’s possible use in pressure sensing devices with the
ability to operate in harsh environments.

The results of this experiment may someday be useful information to anyone involved
in erlgineering polycrystalline diamond windows and/or possibly future pressure transduc-

ers Using piezoresistive polycrystalline diamond diaphragms.

1‘2 Methods of CVD
The methods employed to fabricate diamond single crystals and polycrystals can be
gr onped into five categories [3].
(1) Conversion of graphite to diamond by dynamic shock. This method produces
relEltively small diamond crystals.
(2) Conversion of graphite into diamond by a high temperature, high static pressure
r“ethod. This process yields relatively small crystals, as in (1).
(3) Diamond formation from graphite by contact with a molten metal at high tem-
Perature and pressure which is the method used to produce most of the larger single crystals
uSed commercially. ’

(4) A method similar to (3) which uses small diamond crystals as a precursor and

fir n'h'
1 .‘.L

>11; 3'
.1 ‘2‘) ~ .

._ 'V
‘11 ‘k‘
“ ii‘lli‘
7‘.»
“'M‘M

4

can produce diamonds of well-controlled impurity level and produces large single crystals.
(5) Chemical vapor deposition (CVD) of diamond thin films (DTF) onto different
types of substrates.
The chemical vapor deposition (CVD) method is employed in this investigation.
More precisely, a microwave assisted plasma CVD method was used to deposit the DTF’s

(diamond thin films) onto (100) single crystal Si wafers.

1.3 Proposed CVD Growth Mechanisms

An accepted mechanism describing the growth of DTF’s has not yet been established. Dif-
fere nt mechanisms trying to accommodate all the different trends observed in the deposi-
tion processes have been proposed. but many questions are yet unanswered and many
conflicting propositions exists [7,8]. A major disagreement exists on what is the key mono-
met-(S) responsible for the growth of diamond. Researchers generally agree that some
hydrocarbon species is responsible for the diamond growth in CVD processes. Of the dif-
ferent hydrocarbon species suggested as being responsible for diamond growth, acetylene

(C2H2), and methyl (CH3), seems to be the most commonly suggested monomer species
[71.

143.1 A Simple Proposed Mechanism

A Simple mechanism was proposed [7] in which a sequence of chemical reactions utilizing
aCetylene, from the pyrolysis of methane, lead to the formation of sp3 carbon-carbon or dia-
mond bonds. The continued formations of these sp3-bonds leads to epitaxial diamond

growth. This simple mechanism (Figure 1) can be applied to any plane; the dominant { 11 1 }

hgurel .
i313 i-gcr.
‘ . r
”-1“ 1C I:
51% all:

.33“ CilClm

H&WI

 

 

 

 

 

Figure l (a). Proposed growth mechanism showing the sequence of chemical reactions. A
hydrogen atom activates the surface by abstracting a surface H atom creating a radical site,
1(a) to 1(b). The large coarse mesh circles are the top layer carbon atoms and the large cir-
Cles With diagonal lines are lower layer carbon atoms. The small circles represents hydro-
gen atoms. After [7].

/

o/O

 

Figure l (b). An acetylene molecule is added to the radical site. The radical then propagates
to the added molecule, 1(b) to 1(c). The carbon atoms of the added acetylene molecule are
represented by the large fine mesh circles, and the radical site by a black dot. After [7].

Figure 1
Strict '11

Figure ]

i imidt

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

“@112 l (c). The radical then bonds to the hydrogen atom abstracted from the surface
carbon atom of the lower layer, 1(c) t0 1(d)- After [7]-

 

Figure l (d). The carbon radical available on the lower layer then forms a QC bond with
the carbon atom of the originally added acetylene molecule. This creates a free radical on
the adjacent carbon, 1(d) to 1(c). After [7].

Figure l

i .—
wit’ ~11

Figure

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FigIIre 1 (e). The sp3 bond is formed by this step. The addition of another acetylene mole-
cule Starts a similar reaction sequence which is depicted in (f) through (i). After [7]-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure l (f). The radical site is propagated to the added acetylene molecule. After [7].

iihire] lg;

Eur. rlultli

1
t a

Figure] (h)

FM aldded

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 1 (g). A hydrogen atom is abstracted from a top layer carbon atom leaving a car-
bon radical. After [7].

 

 

 

 

 

 

 

 

 

 

 

 

 

F.
néglire l (h)- The carbon radical forma a CC bond with another carbon atom from the
w y added acetylene molecule propagating the radical to the molecule. After [7].

Fi‘gure l

53:31:13 31

 

 

 

 

 

 

 

 

 

 

Figure l (i). The radical on the acetylene molecule forms a C—C bond with a lower layer
carbon atom while abstracting a hydrogen atom from the lower layer. After [7].

.\5

«~.
algk

11)
o
r

I0

and { 100) planes for example. A hydrogen atom activates and removes another H atom
from the surface plane of the diamond to produce a radical site (Figure l (a) to 1 (b)). The
C2112 acetylene molecule then attaches itself to the radical site propagating the radical (Fig-
ure 1 (b) to l (c)) which abstracts a hydrogen atom from the carbon atom of the lower layer
(Figure l (c) to 1 (d)). The carbon radical now available forms a CC bond with the carbon
atom of the originally added acetylene moleculecreatin g a free electron on the adjacent car-
bon (Figure l (d) to 1 (c)). The hybrid sp bond of the acetylene molecule is transformed to
3 sp2 bond upon attaching itself to the solid carbon atom. Another C2112 monomer species
then attaches itself to this radical site in a similar manner. This sequence of C2H2 addition
to radical sites on the diamond surface propagates to form the DTF (Figure 1 (e) to 1 (i)).
It is proposed elsewhere however, that C2H2 is an insignificant species in the deposi-
tion of DTF’s [8]. This conclusion is based on data that shows the mole fraction of C2H2
in the deposition atmosphere to be less than 1.0x10 '7. A mole fraction of 1.0x10'7 was
dretried too low to be responsible for film formation. Besides CH4, whose mole fraction is
0:7 in the deposition atmosphere, only C2116 and CH3 have concentrations above 10 '7, thus

CH4. C2116. and CH3 were suggested to be the only monomers that could possibly be

reSponsible for the diamond film formation [8].

1.3.2 Role of Hydrogen in DTF Deposition

Hydrogen is used as the carrier gas in most CVD procesSes. Methane (CH4) and often oxy-
gen (02) gases in small amounts are added to an abundance of H gas to makeup the depo-
Sition precursor gases. The two most important roles of hydrogen atoms in the deposition

Process is (1), hydrogen preferentially etches the graphitic/amorphous phases of carbon

 

(if) FC\

~lf.’ 376

1
l) .- ‘F‘“
A‘ ‘H-nl‘

' I
,‘l ’ 7
sub" d\ n
<' 'i- V
n”; :;\l.ni
s

._‘F ‘
‘. .1 or :hl

111.7336 if
New ii
Ilia-53'
Hg‘d.
‘ERD’ 5F“
nuclear r6-

u'ale p1115

1.3.3 R0
The additi
run that is

added to t

l

11

over diamond phase [7.9.10.1 1.12]. and (2). hydrogen suppresses of the aromatic hydrocar-
bons responsible for diamond [9.13]. The preferential etching of the non-diamond phases
over the diamond phase by hydrogen causes diamond growth rate to dominate. thus pro-
ducing a quality diamond film. The preferential etching effect of hydrogen is the primary
mechanism in CVD diamond formation [2].

Hydrogen is a known impurity within diamond thin films (DTF’s) [4]. Impurities
such as hydrogen. may increase the compressive stresses in a film [14]. It is thought that
increasing the deposition temperature decreases the H content [4,14], which may be the rea-
son for the decrease in the compressive stress associated with increased deposition temper-
ature [14,15]. The overall intrinsic stresses of the film is decreased by increasing the
methane, CH4, concentration of the precusor gas. This intrinsic stress decrease is due to an
increase in the compressive stresses in the film. The compressive stresses are caused by the
increase in‘the hydrogen concentration which is associated with the methane concentration
increase.

Hydrogen content in deposited diamond can be determined by elastic recoil detection
(ERD) spectroscopy [14,18] and by nitrogen nuclear reaction analysis [4,14]. Nitrogen
nuclear reaction analysis has been used to determine that films grown by a particular micro-

wave plasma assisted CVD process contained 0.7 to 8.0 percent hydrogen [4].

1.3.3 Role of Oxygen in DTF Deposition

The addition of oxygen to the precursor gases decreases the quantity of non-diamond car-
bon that is codeposited with the diamond. Oxygen also increases film growth rate [5]. When

added to the methane—hydrogen precursor gas mixture, oxygen increases the effective

--9 cv V

..'v*

F-i‘
l

’l

‘fifve

3“ Li L"
- 4

”rev-o
1‘ "8 l\

1".» »

12

hydrogen concentration in the deposition atmosphere [13]. Hydrogen etches non-diamond
carbon preferentially (section 1.3.2), therefore an increase in its effective concentration aids
in graphite/amorphous carbon suppression. The addition of oxygen thus greatly affects film

growth by increasing the hydrogen concentration and hence the effects of hydrogen etch-

ing.

1-4 Interface between DTF and Si Substrate

A thin layer of SiC is found between the DTF and the Si substrate [12]. The SiC layer is
approximately 0.25 microns thick and may be continuous across the entire diamond/silicon
interface [12]. Residual diamond particles (diamond seeds) left embedded in the Si wafer
after abrasion by diamond powder or diamond paste act as nucleation sites for diamond film
growth [17]. The deposited DTF grows directly on some or all of these embedded diamond
Particles. Initially the crystals grow laterally until they meet, and then they grow upwards
forl‘ning columnar structures (Figure 2). The diamond film deposition rate is initially high

[18] and the SiC layer (perhaps a reaction product) may dominate initially and form much

faSter than the diamond crystal.

1.5 Deposition Parameters and Their Efl'ects
Several deposition parameters affect the films synthesized in a CVD process. The types of
precursor gases used in deposition, the concentration of the precusor gases, the deposition

temperature, the deposition pressure, and the energy source, can each affect the growth rate

and quality of the DTF synthesized in a CVD process.

fix

//
II. / /

11,,1111111/l
.

 

'//,,,,,,,,,,,

W7

1‘. ..
Hall-um

“SUN: 1 <
H red

Scmih i\ l

I3

/////

\
/
WV/

  

W
.C
.3

Silicon Substrate

 

2. Sketch of DTF /Si substrate

mond seeds embedded in Si wafer. Grains nu
grow is initially lateral. Upon co
log columnar structures.

sites. Crystal

ntacting their neighbors the crystals grow upward form—

interface. The DTF and Si interface showing dia-
cleate and grow from these

Fig“ re

-3».
.115.» a.

.
.,._‘._ ‘
N14: ]\
L.
3“ p
..F,\L
I
e
I

'r' .051 s
.1 ...

C

Wu
1..;‘L\‘

14
1.5.1 Deposition Precursor Gases

The type: s and relative concentrations of the precusor gases used in CVD processes have a
significant effect on the nature of the diamonds produced. Varying the types and relative
concentrations of the precursor gases is cheap and simple. This type of variation is often
made in CVD processes to produce desirable diamonds and/or to study diamonds growth
mechanism.

The most common gas combination used in diamond film CVD processes is a meth-
ane—hydrogen (CH4—H2) mixture. Methane is the source of the carbon which forms the car-
bon Sp3 bonds that characterizes diamond. Hydrogen (section 1.3.2) is primarily an etchant
for the other carbon phases codeposited with diamond. Carbon phases such as amorphous
and graphitic phases have a lower Gibb’s free energies than diamond, therefore their
codfitposition with the diamond phase is always expected [11]. The etching of the amor-
Phous and graphitic carbon phases is the most important process in the CVD of diamond
films, therefore the etching effect of hydrogen is quite necessary.

Another precursor gas combination used in diamond film CVD is a methane-oxygen-
hYdrogen (CH4-OZ-H2) mixture [13] where the oxygen increases the effective hydrogen
Concentration in the deposition atmosphere.

Other gases may also be used in the CVD of diamond films, but their uses is usually
as an experimental means of studying the grth mechanism of diamond. For example, the
inert gases helium (He) and argon (Ar) were used in the CVD of diamond films to better
understand the role of hydrogen in the deposition of diamond. In one particular experiment
[8], H2 (the usual carrier) was replaced by helium. CH4 and H2/02 were alternated in the

He canier as the deposition precursor gases. A substantial diamond growth rate was noticed

u\.‘

.3-
‘xu

J .

15

which might indicate that hydrogen as a suppressor to the diamond forming aromatic is a
significant process in the CVD of diamond. In a simulation experiment [9]. Ar was used to
replace H2 to better understand hydrogen’s role in the growth mechanism of diamond. The

results suggested that suppression of aromatic formations by hydrogen was the key factor

in diamond formation.

1.5.2 Concentrations of Precusor Gases

Typical the CH4-H2 mixture used for CVD of DTF ranges from approximately 0.1 percent
to 5—0 percent methane in hydrogen [4,12,14]. However, a higher quality diamond film is
dePOSited at the lower CH4 to H2 ratios, from approximately 0.5 to 1 percent [14,18]. As
the methane to hydrogen ratio increases, the likelihood of codepositing a non-diamond car-
bon phase also increases. The film growth rate also increases with the concentration
increase of methane in hydrogen [12].

In the CH4-02-H2 systems, the concentration range of 02 may be from 0.5 to 5 per-
cent [13]. The film growth rate increases with increasing oxygen concentration in the pre—
Cllrsor gases [13], and as mentioned earlier, oxygen addition effectively increases the
hYdrogen concentration thus improves the quality of diamond film produced. The optimal

oXygen concentration range seem to be 0.5 to 1.0 percent of the precusor gases at 45 -50

Torr [10].

1.5.3 Deposition Temperature

The diamond film growth rate also increases with temperature, but the growth rate of dia-

mond reaches a maximum and further temperature increase leads to the increased formation

1..

Uf‘l

\

I6

of graphite [7,18]. The temperature at which transformation from diamond formation to
graphite formation occurs is not reported. Graphite formation also exhibits a temperature
maximum but at a higher temperature than that of diamond [7].

The usual substrate temperature range for DTF deposition is 700° C to 1,300° C
[10.13, 1 6]. One might assumed that as the temperature increases. so would the surface
mobility for diffusing hydrogen atoms therefore increasing the purity (less hydrogen impu-
rities) of the diamond films produced at the higher growth rate [11]. However, for the tem-
perature range of 600 °C to 887 °C, no correlation between deposition temperature and the
film’s hydrogen content was determined [4].

Intrinsic tensile film stress also varies strongly with deposition temperature [14,15].
Th6 intrinsic stress increases as deposition temperature increases. The intrinsic film stress
versus deposition temperature relationship might be attributed to the non-diamond (sp2)
incorporated during deposition [14]. Higher temperatures corresponds to a lower non-dia-
mond content in the deposited film [14]. Lowering the non-diamond content increases the
film’s tensile stress by lowering the compressive stresses caused by the non-diamond.

The (111) and (100) faces of diamond preferentially grow in the temperature range of
750° C to 1000° C. At the higher temperatures in this range the (111) face dominate,

Whereas the (100) faces dominate at the lower temperatures of this range [19].

1.5.4 Deposition Chamber Pressure

Variation of the deposition chamber pressure in a diamond CVD process affects film mor-
phology and film growth rate [10]. The morphology is the strongest determinant of the

mechanical properties of synthetic diamond [20]. The deposition chamber pressure

"‘4;

”AA

aii"
k'su

ogq‘v

v\~ \
.

'A '

7 \\

.l. 1

m“
nu,” \

.z,‘ ’
:ne

.
rm.
4~u[k

a v
kn“?
'TLOV

35> of]

I7

therefore would seem a reasonable parameter to use in tailoring the synthetic DTF. How-
ever. the deposition chamber pressure is not a parameter that is usually varied to tailor the

deposited diamond thin film (DTF). Film growth rate increases with increasing chamber

pressure [10].

1.5.5 Deposition Energy Source

Several different types of energy (heat) sources are used in the CVD of diamond. Hot fila-
ments[ 19], high temperature furnaces [11], hi gh-intensity infrared lamps [18], and micro-
waves [4] are some of the methods used to excite the precursor gases to a plasma state from
Which diamond is chemically deposited.

As mentioned earlier, the morphology of synthetic diamond is the determining factor
for most of the physical properties for which diamonds are made [20]. The different types
0f energy sources used are not known to have any special effect on the morphology of the
dialmond they produce [20]. The many different methods of plasma generation in use is
1ikaly due to a search for the most economic source for the CVD of diamond. The most eco-
nOl’nic method to generate the heat necessary for the CVD of diamond is still not estab-
1iShed. Perhaps there does not exist a most economic method. The lowest temperature at
Which DTF’s can be synthesized may be the most economical.

Although the heat source is generally accepted as not playing any role in the mechan-
ical properties of the synthetic diamond produced, it seems likely that the heat source
employed may be important to some non-mechanical properties. Non-mechanical proper-
ties of the CVD diamond which are dependent on impurity content and/or possible the

effect of heat gradients on the specimen could make the energy source a non-trivial

i1
19fi'\“
.253‘

(exit Y1
‘ s

Th‘w

l..r‘ ..l
L
‘P
. .l.
“UL Suki
”71";31
Thain”;
~
urv .
H.” :k)\

5mm [

:hsiif\

I8

deposition parameter. 1f the types and concentrations of impurity in a single crystal or a
polycrystalline thin film effect its transparency. electrical. or chemical property. then the
energy source would be important. The impurity of a CVD diamond will be. to some extent,
dependent on the environment in the deposition chamber. The deposition chamber is often
dictated by the energy source used., thus the energy source may influence the types and
amount of impurities incorporated in the synthesized diamond film.

Another non-mechanical property that might be influenced by the heat source is the
uniformity of the thickness of the diamond film produced. The heat gradient in the deposi-
tion chamber is related to the thickness variation of polycrystalline thin films [21]. Mini-
mizin g the heat gradient would produce films of more uniform thickness. Heat gradients
can be strongly dependent on the energy source, as in convection oven the heat gradient is
greater than in a microwave oven. These considerations coupled with controlling cost com-
Plicates the task of establishing any one energy source as the most economical energy
source for the CVD of diamond.

Getting away from the complexity of the CVD process, the burst test experiment dis-
CUSSCd in this thesis is intent on the characterization of the mechanical properties of poly-

CI'yStalline diamond thin films (DTF’s). The DTF’s were deposited by a microwave assisted

CVD Process.

2. E1

, ‘30-

Hit.

’5' ‘

.luh‘

7*: 'M
"Wei“;

2.1 Si

2, Experimental Procedure

Diamond polycrystalline thin films (DTF’s) were deposited on Si single crystal
wafers. Circular holes were etched through the Si wafers to produce circular DTF dia-
phragms. These diaphragm were subsequently loaded to failure with gas pressure.

The deposition of the polycrystalline diamond thin films onto Si wafers was accom-
pli shed with the use of a microwave plasma reactor (Figure 3) via a microwave assisted
CVD process. Circular holes were etched through the Si wafers with a KOH etching solu-
tion to produce circular DTF diaphragms. The circular diaphragms, supported at their edges
by the Si wafer, were then mounted into a pressure vessel and uniformly loaded with nitro-
gen gas pressure to failure. The increasing pressure of the nitrogen gas was closely moni-
tored, and the pressure at failure was recorded for each diaphragm that was burst. The
critical load at failure, film thickness, diaphragm diameter, average grain size, and the qual-
ity 0f diamond polycrystalline film were all considered in the characterization of the

mechanical integrity of the DTF diaphragms.

2.1 Si Wafer Pre-deposition Treatment
The initial Si wafers were abraded with diamond powder to increase their nucleation
Site densities prior to deposition. Microscopic scratches on Si surfaces are known to act as
nucleation sites for diamond crystals during the CVD of diamond on Si wafer [5,12]. Single
Crystal Si wafers were therefore abraded on one side to increase their nucleation site con-
Centration.
A PhOtoresist pre-deposition treatment was also used to like-wise increase the nucle-

ation ° . . . . . .
Sue densrty on later specrmens tested. The photoresrst treatment rs discussed 1n

I9

 

 

thre
311m [if

i

20

Microwave cavity
Flow controllers
“T

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

_ 48%
water flow to cool _
base plate & fl
Roughing
Valve‘ ‘
TC2 (foreline) ‘ "—7 CHAMBER H2 CH4
type 286
'v gas gas
TRAP BARATRON
(type 122-14) Flow controller
PUMP I .____
Alcte12033 E ' ‘
TC] (chamber) N2
type 286
gas
8‘

 

 

 

 

 

Exhaust of H2 and CH4 diluted by N2

5'3“” 3. A schematic diagram of the microwave plasma reactor used to produce the
la"10nd thin films utilized in this study.

 

J Or
\...‘.'

vr .
i
any r
wk. ‘

. fina,
LIMA
V.
«mu.

” .

‘0n

17?. I’!
.

"‘1‘ Di.

\f’fi'l ‘

3 ‘ l-
r “1“”
, .

‘9 ‘h ‘
“ “‘6

‘1
EH.” .
"he”; 1‘

LL.

21
section 2.9.].

2. 1- 1 Abrading of the Si Wafers

Single crystal Si wafers used were received as circular disks, approximately 8 cm in diam-
eter and 0.25 mm in thickness. The wafers each had a dull unpolished side and a shiny pol-
ished side. The shiny polished sides on which the DTF were subsequently deposited were
abraded before deposition.

Abrasion was achieved by placing a wafer into a shallow glass culture dish (Corning)

with its shiny polished side up (Figure 4). The diameter of the Coming glass dish was
approximately 1.5 cm greater than that of the circular wafer. The edges of the dish were
1cm high. Approximately 0.015 gram of a one micron diamond powder was sprinkled over
the top of this polished surface and evenly distributed. A second wafer, the wafer of interest.
was placed with its polished side down on top of the one micron diamond powder that was
SPl’iflkled over the first wafer. With the diamond powder between the two polished surfaces
of the wafers, the fingers were used to gently apply a mild pressure to the wafers forcing
them together (Figure 4 (b)). Under this applied pressure, a circular motion was then gen-
crated With the fingers to rotate the polished side of one wafer on the polished side of the
Other Wafer (Figure 4(c)). This circular motions, with a light pressure from the fingers, was

malllltained for approximately 10 minutes to achieve abrasion.

The abrasion step likely was not very reproducible. Neither the force applied with the
fi . ' .
ngers, nor the exact amount of diamond powder used was measured. However, a possrble

me - . . . . . .
th0d to improve the repeatability of abrasron Is to characterized the micro-scratches pro-

du . . . . -
ced by abrading. Determining the srze, depth, and area densrty of the scratches caused by

“Elite
fii‘gen

22

Si wafer dull side up

..........

  

diamond powderj‘ . ........
‘ '. f Si wafer shiny side up

Corning culture dish

 
 

Figure 4 (a). Abrading of Si wafers. The shiny polished sides of the Si wafers were
abraded with 1 micron diamond powder. Diamond powder was place between the polished
surfaces of two wafers, and the sandwich placed in a dish.

 

 

 

 

P wo Si wafers with diamond powder
between them.
Lm i #— Coming culture dish

 

 

l

P — load applied with fingers

Figure 4 (b)

"gets . A light load (P) was applied to the Si/diamond powder sandwich with the

figure

1.) 31,
‘1‘; w»

23

Coming culture dish

Directions of motions generated

with fingers.

 

Figure 4 (c). The motion applied to the wafers with the fingers is as indicated. Twisting,
side to side, and back and forth motions were applied to the sandwich.

'1.’

is

‘ 9
kn PU

2.1.2 (

i.’ , i
«[9‘ 4'

l'. n;
A.) llys

. L
.9?
“k-six

24

abrading and the size, shape. and disuibution of the residual (seed) diamond powder parti-
cles left embedded in the Si wafer from abradin g may allow reproduction. Such determina-

tions can be made on the SEM using surface replication techniques.

2. 1.2 Cleaning of the Abraded Si Wafers

After abrading the Si wafers on their polished sides with the 1 micron diamond powder it
was necessary to clean residual diamond particles from these surfaces. The abraded Si
wafers were ultrasonically rinsed (using a Fisher Scientific Ultrasonic Cleaner model B-
ZZOOR- 1 , BranSon Cleaning Equipment Company, Shelton, CT. USA) first in methanol for
5 minutes, then in acetone for another 5 minutes. Each rinsing bath was used only once to
optimize the cleaning efficiency. After ultrasonic rinsing, the wafers were blow-dried with
Dust-Off Plus (Faclon Safety Products, Inc., Branchburg, NJ. 08876) directing bursts of
air onto a desired area. Dust-Off Plus is a hand held canister of inert gas with a thin nozzle.
The wafers were held approximately 5 inches from the nozzle of the canister while being
blow-dried. Wafers were blow-dried to speed drying and possibly avoid minimal dust con-

tamination. The blow-drying procedure however, was not considered to be critical and was

not applied to later wafers.

2.1.3 Fracturing of the Si Wafers

The abraded Si wafers were fractured to a final size of approximately 1 cm by 1 cm before
the diamond films were deposited on them. The wafers were fractured prior to film deposi-
tion to avoid possibly disturbing the interfaces between the deposited films and the Si

w . . .
afe’S~ SPeCimens from sets DF—Nl, DF—NZ, DF-N4, and DF-NS were processed m this

r '1”,
:,..i. 's

‘. .0):

asst

N‘"

u llu..

 

0,.
,Lgv

" ~-
‘ I

 

25

manner.
Controlled fractures were obtained by a particular method (Figure 5). The abraded
wafers werelaid on a hard and flat surface; a table top. A diamond tip pen was used to create
a minute -just visible- scratch on one edge of the abraded wafer (Figure 5 (a)). The wafer
was then flipped over on two plies of tissue paper and an increasing load was slowly applied
with the diamond pen tip until fracture occurred (Figure 5 (b)). The tissue paper placed
under the wafers allowed the wafers to bend until it fractured. In this manner a complete
fracture propagated across the wafer. The fracture initiated at the scratch and propagated at
a right angle away from that same edge (Figure 5 (c)). After complete fracture the tissue
papers were relatively undisturbed and many breaks were made on the same pieces of tissue
paper. Fractures were orthogonal and mostly free of any chipping. The final size of each
piece of Si wafer before diamond film deposition was approximately 1 cm by 1 cm.
Later specimens however, were received as large Si wafers (semicircles approxi-
mately 9.7 cm in diameter) with a DTF already deposited on them. The specimens therefore
had to be fractured with the DTF’s already adhered to them. These specimens were also
fractured to 1 cm by 1cm squares.
Although the later specimens were fractured with the DTF’s already attached to the
Si wafers, a severe effect to the adhesion of the DTF on the Si substrate due to fracturing

Was "0t anticipated.

igurt

26

diamond tip pen

pen motion to cause small chip at wafer edge
17 Si wafer

Figure 5 (a). Fracturing of Si wafers. The edge is first chipped with the diamond tip pen.

 

loading motion to propagate crack

diamond tip pen

[1

Si wafer

 

 

 

L two plies of tissue paper to allow buckling of wafer
which propagates the crack

F' . . . . .
tolgu re 5 (b). The wafer is flipped over on tissue paper and the tip of the diamond pen used
cragLCSS dOWn on the opposite side of the wafer above the previous chip. This causes a

to pfOpagate away from the loaded point cross the wafer.

Fig'ur

27

 

 

 

 

~\

Si wafer direction of
\ .
‘rack prc pagation

 

 

 

 

 

 

 

 

‘\ /‘J

_ diamond tip perpressing down

 

 

tissue paper to allow wafer buckling—
and cracking to a complete fracture

Figure 5 (c). Top view of (b) showing the crack propagation direction

 

2.2 FilII
2.2.1 Mi
DTFB it:
LTD T:
:Tfixift’ ¥.
i‘ililih’l pf
r

‘80 \Cn‘.

28
2.2 Film Deposition

2.2. 1 Microwave Disk Reactor Preparation

DTF’S were deposited on (100) single crystal Si wafers by microwave plasma assisted
CVD. The microwave deposition system (Figure 3) was usually kept at or near deposition
pressure as deposition was done at ~ 45 Torrs. If however the system was not at or near dep-
osition pressure, approximately 10 hours were required for the system to be pumped down.
Pump-down was achieved by a roughing pump followed by a diffusion pump. This process
was very time consuming because the pressure transducer (Baratron type 122-A - Figure 3)
was calibrated to zero at a pressure which was less than the transducer‘s resolution.

Approximately nine 1 cm by 1 cm Si wafers were placed on a graphite plate (Figure

6) With their abraded sides facing up. The abraded sides of the wafers were thus. exposed
directly to the microwave stimulated plasma. The graphite plate was square, approximately
3 Cm by 3 cm, with rounded edges and a slight rim. The rim was approximately 0.5 mm by
0.5 mm. An advantage in using a graphite plate was that it reduced the thermal gradients in
the Si wafers [21].

In the microwave reactor chamber (Figure 7), a heavy transparent fused silica dish
approximately 8 cm deep, 8 cm outer diameter, and 1cm thick, was lifted off a rubber O-
ring. The graphite plate with the nine pieces of Si wafers was placed on a fused silica tube
that was under the heavy dish. The fused silica tube. 3.7 cm high and approximately 2 cm
in diameter, sat vertically on one of its ends and the graphite plate carrying the nine pieces
of Si wafers was placed on the top end. The fused silica tube was positioned under the cen-
ter of the heavy cylindrical dish. After the graphite plate with the nine pieces of Si wafers

was placed on top of the vertically standing fused silica tube, the heavy cylindrical dish was

 

Ea. -
ktii ple‘

29

 

~ 1 cm approximately 9 pieces of Si

«4 Q!
Lard
Egg

/ <—-— graphite plate
~ 0.5 mm by 0.5

l'lm

  

wafers with abraded sides up

~3cm

 

 

}‘ ~3cm

Figure 6. Approximately 9 pieces of abraded Si wafers are placed on the graphite plate.
Each piece of Si wafer was approximately 1 cm by 1 cm

30

replaced over the rubber O-ring. The rubber O-ring formed an air tight seal between the
dish ’8 rim and the base plate which allowed the pumping-down of the inside cavity of the
heavy dish. As mentioned earlier, pump-down was necessary for transducer calibration and

to maintain deposition pressure.

2.2.2 Deposition of DTF’s

The cavity within the dish was pumped down through a hole in the base plate beneath the

fused silica tube (Figure 7). The pressure transducer was calibrated to zero at a pressure less

than the resolution of the transducer. 1f the transducer was previously calibrated, the spec-
ilnen chamber (cavity of the heavy dish) was pumped down to approximately 5 Torrs.
At this low pressure of ~5 Torrs, methane and hydrogen gases were allowed to flow
into the specimen chamber at fixed rates. Methane was allowed to flow in at 1.5 and 2.25
Standard cubic centimeter (sccm) and hydrogen at 300 sccm. These rates allowed for
approximately 0.5 percent and 0.75 percent, respectively, methane in hydrogen as the pre-
cusor gas mixtures.
With ~5 Torr deposition chamber pressure and a constant inflow of the methane/

’ hydrogen precursor gas mixture, the microwave power was turned on and set to about 400
watts. A glowing plasma on the Si wafer specimens was immediately visible through a 4
cm diameter viewing port made of glass and wire gauze. Protective goggles were used to
make these observations in order to avoid possible UV radiation to the eyes. The glow indi—
cated that the plasma necessary for deposition was formed. If a glow was not observed the
microwave power was reduced and another attempt made until a plasma was obtained

Once the glowing plasma was observed, the deposition chamber pressure was

adjustable microwavr—\
guide (tunes mode)

31

/— from microwave

 

{— adjustable shoot (tunes mode)

 

:ylindrical

microwave

chamber \

water cooling
ducts

 

O

 

 

 

heavy transparent fused
silica dish

yrF’

specimen chamber

plasma positioned over

 

 

 

IIIIIIIIIIIIIIIISIIIJI

lilo

   

 

 

 

 

 

 

 

 

  

 

specimen in graphite holder

fused silica tube stand

 

f I

 

 

 

 

O [— heavy base metal

 

 

    

[
Y
0‘ '1'
oooooooo

 

 

 

 

 

 

 

 

 

heavy metal base

J

closed system allows pumping down

heavy transparent fused silica dish

rubber ring allows vacuum

Figure 7. A schematic of the microwave reactor specimen chamber used to process the
diamond thin film/silicon specimens used in this study.

32

increased to approximately 30 Torr. At 30 Torr the microwave power was increased to 600
watts. The deposition chamber pressure was then increased to 45 i 5 Torr. At 45 i 5 Torr
deposition pressure the microwave power was turned to the desired wattage. Wattages of
700 and 900 watts were used in film depositions. At this point the system was left to run
with constant monitoring of the deposition pressure. A leak valve for the methane and
hydrogen gases from the deposition chamber was periodically adjusted to maintain the dep-
osition pressure at 45 i 5 Torr.

The microwave energy was directed from the top of the cylindrical microwave cavity
downward into the cavity by a wave guide (Figure 7). Inside the cylindrical microwave cav-
ity the microwave energy excited the CH4-H2 gas mixture in the fused silica dish into a

Plasma state (Figure 7). The nine pieces of Si wafers on the graphite plate were positioned
by the silica tube to be in the plasma of a mum excitation mode.

The TM012 excitation mode in the cylindrical cavity had a minimum reflected power
of approximately 3.8 percent. The mom excitation mode was tuned by an adjustable slid-
ing short and an adjustable microwave guide. The tuned TMmz excitation mode located the
plasma to coincide with the position of the graphite plate containing the pieces of Si wafers
(Figure 7).

In the excited plasma, carbon atoms are in a supersaturated state. As the carbon atoms
move from the high temperatures of the plasma to lower temperatures outside the plasma
they readily crystallize onto any suitable surface. The Si wafers are positioned in this lower
temperature region of supersaturated carbon atoms and their abraded or photoresist-treated
surfaces are suitable for the crystallization of the supersaturated carbon atoms. Diamond is

nucleated either in the abrasions. on imbedded seed diamonds, or on other suitable sites

33

[5, 1 2]. A continuous 0.25 micron SiC layer forms around these nucleation sites before, dur-
in g, or after nucleation, and the diamond crystals then grew from these sites upwards in col-
umns [12], thus fonning the polycrystalline DTF.

The temperature of the Si wafer surfaces were determined during deposition for DF-
N4 and DF—NS. Four temperature readings were taken for set DF—N4 at approximately one
hour intervals with a pyrometer (IRCON Ultimax infrared thermometer, serial number
11003, Niles, ILL. 60648). For set DF-NS, readings were taken form the four comer spec-
imens and from the center specimen. The emissivity on the pyrometer was set at 0.65 for

the temperature readings.

22.3 Deposition Conditions and Specimen Labeling

Five deposition runs were made under different conditions as listed in Table 1. Approxi-
mately nine specimens were deposited in each run. The first run was done at 0.5 percent
methane in hydrogen and at 900 watts for 6 hours. This set was titled DF-Nl to indicate
diamond film (DF), on n-type Si wafer (N), and first run (1). The second run was done at
0.5 percent methane in hydrogen at 900 watts for 8 hours. This set was titled DF—N2 in a
similar manner, but (2) indicating the second run. The third run was done at 0.75 percent
methane in hydrogen and at 900 watts for 6 hours and was entitled DF-N 3 in a similar man-
ner. The fourth run was done at 0.5 percent methane in hydrogen and at 700 watts for 6
hours and the was entitled DF-N4. The fifth set was deposited at 0.5 percent methane in
hydrogen and 900 watts for 10 hours and was entitled DF-NS. A sixth set, deposited at 0.5
percent methane in hydrogen and 900 watts for 12 hours, was also attempted. However,

deposition was terminated after 2 hours due to the detection of sufficiently strong

 

 

Table l

\i in

34

Table l. Deposition conditions of sets DF-Nl. DF-N2. DF-N3. DF-N4. and DF-NS.

 

 

 

 

 

 

Specimen Percent C4 Deposition Microwave .tefililjlizgiziirfie
Set in H2 time (hr) power (W) (, C)
DF-Nl 0.50 6 900 NA
DF-N2 0.50 8 900 NA
DF—N3 0.75 6 900 NA
DF-N4 0.50 6 700 766
DF-NS 0.50 10 900 774

 

 

 

 

 

 

 

NA -not available

EEC C it

km; E

230

° F
HESS}
LZCTI
jfiaur
“ “£1.
”0”“ 31
2.55101
= grid-

ind}.

(Ir
.7.
._.
(as
-C -
r_

35

microwave radiation leak.
The individual 1 cm by 1 cm specimens from the different runs were labeled as fol-
lows: DF—Nl-l, DF-Nl-2, DF-N1-3, etc. from set DF—N 1, DF—N2-1, DF-N2-2, DF—N2-3,

etc. from set DF-N2, and likewise for sets DF-N3, DF—N4, and DF-NS.

2.3 Optical Microscopy
Immediately after each deposition run a few of the specimens from that set were observed
under an optical microSCOpe to determine film quality. Optical microscopy may roughly
differentiate between a well faceted diamond polycrystalline film and a poor faceted and
Possibly amorphous or graphitic film. A film covered with facets was assumed to indicate
a good polycrystalline diamond film. A smooth film with a few or no facets was assumed
t0 indicate that large regions of amorphous or graphitic carbon were present, thus not of a
good quality diamond film.
Since optical microscopy is a qualitative technique, the films were subsequently ana-
lyzes by Raman spectroscopy, which yields more definitive information on the diamond

and graphitic content of the film (Raman spectroscopy is discussed in sectiOn 2.6).

2.4 Diaphragm Preparation

To obtain circular diamond diaphragms from the ~1 cm square pieces of Si wafers which
were now coated on one side with polycrystalline diamond thin films (DTF), circular holes
were etched through the Si wafers. The etching resulted in circular DTF membranes sup-

ported along the edges by the Si wafers.

 

. 4
r1;

[.[l‘

36
2.4. 1 Protective Coating for Si Wafers

To etch a circular hole through the Si wafer. a circular area of the Si wafer was exposed to
the etchant. All other surfaces of the Si wafer except the circular hole were protected from
the etch. A protective coating was needed to protect the Si areas from the etch.

A thin sputtered gold coating (SEM sample preparation method, section 2.5.2) was
inadequate protection from both a KOH and a NaOH etchant. The thickness of the gold
coating sputtered onto the Si wafer was likely on the typical order of ten to one hundred
angstroms, but the thickness was not ascertained. The gold coating on contacting the KOH
and the NaOH etchants, both at 55 °C, immediately peeled away from the Si wafer surfaces.
The KOH etchant was a 22:28 ratio by weight mixture of dry KOH pellets to deionized
H20 and the NaOH etchant was a 1:3 ratio by weight mixture of dry NaOH pellets to deion-
ized H20.

Torr Seal epoxy, (Varian Vacuum Products, Lexington, MA.), was a successful pro-
teCtive coating against the KOH etchant. Torr Seal epoxy as a protective coating and KOH
as the etchant was therefore used in all subsequent etchings.

To leave only a circular area of the Si wafer unprotected and exposed to the etch,
Small fused silica tubes with outer diameters of 2 mm, 3 mm, and 4 mm, and 2 to 5 mm in
height were centered on the unprotected surfaces of the Si wafers. The silica tubes were
then surrounded with the Torr Seal product (Figure 8 (a)).

The Torr Seal epoxy came in two tubes of different sizes. The larger tube contained
the base resin and the smaller tube contained the hardener resin. Equal lengths of both the
Torr Seal epoxy base and the hardener resins were squeezed from the tubes onto a sheet of

nOtebook paper and thoroughly mixed together. Two pieces of wooden sticks

figure
5? 9

Ni. ii:

his.» 1’
M £2.

37

Torr Seal

  
   

small tube defining circular area

diamond

u..." coating

' I ~IJJ.K‘.2-.‘JAJJJJ.KV.'.‘.I ~.'.‘.‘.'.‘.R'.LWA\W.WMFWW W WUJ.‘

 

 

 

\ diamond coating protecting Si wafer
large tube to protect the diaphragm

Figure 8 (3). Protective coating for the Si wafers. The Si wafer is protected by Torr
Sea] on top and edges and by the diamond coating on the bottom. The only area unpro-
tected form the etch was through the circular tube.

Torr Seal

area of Si wafer exposed to etch

small tube defining circle

 

Figure 8 (b). Top view of the Torr Seal protected Si wafer showing the unprotected Si
Wafer that was exposed to the etch.

 

h‘ii.

’i

“a

‘19:;

38

approximately 5cm long and 1mm in diameter was used to mix the base resin and hardener
resin and to hold the tubes in position on the wafers while the Torr Seal was applied to the
specimens. The mixed epoxy was then spread around the fused silica tubes centered on the
Si wafers to cover the unprotected surfaces and the edges of the Si wafers.

The Torr Seal epoxy cured in 24 hour at room temperature and in 1.5 hours at 70 °C.

2.4.2 Etching of Si Wafers

After coating with the Torr Seal epoxy. the only Si surface area that remained exposed to
the etch was the area inside of the small circular tubes (Figure 8 (b)). The etch was thereby
1irnited to attacking only the exposed circular areas and so only circular holes were etched
through the Si wafers.

Etching began on the exposed Si surface and proceeded through the wafers until it
reached the DTF on the opposite side of the wafers. Etching was stopped when the etchant
reached the DTF to avoid excessive lateral etching and even total etching away of the Si
Wafers.

The KOH etching solution was prepared with a mixture of 22 grams dry KOH pellets
and 28 grams deionized H20. The bath was kept at 60 °C during the entire etching process
by a Constant Temperature Circulator, Model 800 Fisher Scientific, which was connected
t0 a 2,000 ml. jacketed beaker (Figure 9), Fisher Scientific. Pittsburgh, PA. The average
time it took to etch a circular hole through the 0.5 mm thick wafer was approximately 24
hours.

The KOH was mixed in 100 ml glass beakers. The Torr Seal protected specimen was

Placed into the 100 ml beaker. The 100 ml beaker, with the specimen inside, was then

39

f— constant temperature crrculator (with reservoir)

 

 
  

 

 

/ temperature gauge
thermometer (in reservoir)

  

  

 

 

(ani sotropic etc
N2 gas inle

tygon tube

 
   
 

 

Watch glass coversL
etchant (KOH) levels—

bath (deionized H20)—
level

 

 

 

thermometer for bath
(anisotropic etch)

N2 gas outlet

,000 ml. jacketed
beaker

 

100 ml. beakers

 

 

 

 

 

m Illlllllll

 

 

 

 

 

 

 

  

 

:3

Figure 9. Etching Apparatus. Each specimen was placed into a 100 ml. beaker which had
KOH etchant in it. The 100 ml beakers were then placed into the deionized water bath in
the 2.000 ml. jacketed beaker. The KOH etch was kept at ~60° C.

 

-.—-—.. ..-,.._.'p

r

 

M.
.4:

Vii

40

placed into the 2,000 ml jacketed beaker. The water level in the 2,000 ml beaker was
adjusted so that it was approximately the same as the level of the KOH solution in the 100
ml beakers. The walls of the 2,000 ml jacketed beaker was kept at 60 °C by water circulat-
in g through it, the water inside the 2000 ml beaker was kept at 60 °C by the water in the

walls of the beaker and the KOH in the 100 ml beakers were kept at 60 °C by the 60 °C

water surrounding it (Figure 9).

2.4.3 Anisotropic Etching of Si Wafers

In an attempt to produce a more circular DTF diaphragm. an anisotropic etch, 20 percent
dry KOH pellets, 64 percent deionized H20, and 16 percent isopropanol alcohol (IPA), by
Weight, was tried. The anisotropic etch preferentially attacks the (100) surface planes of the
SiIlgle crystal Si wafers [22], and so was expected to produce holes with sides perpendicular
to the surface of the wafers as it proceeded through the wafers. The circular shape of the
area initially exposed to the etch would thus be maintained as the etch proceeded down
through the wafers in the <100> direction.

Nine specimens were etched with the anisotropic etch. Two of the these nine speci-
mens were etched under a nitrogen atmosphere. The nitrogen atmosphere was used in suc-
Cessful anisotropic etching done elsewhere [22], but its exact role was not understood. The
gases produced by the chemical process were not known to be toxic, and any other purpose
for the nitrogen atmosphere, besides that of a diluting solvent gases, was ambiguous.

The apparatus used in the anisotropic etching (Figure 10) consisted of a high pressure

C.‘i’lindrical tank with an Airco regulator, a 2,000 ml jacketed beaker and its lid, an escape

valve, and a 500 ml beaker 3/4 filled with oil. Tygon tubing was used to conduct the N2 gas

i 00 01.x
xllkiwwiooo . \

 

gur
[13.-Lid i

41

Airco regulator: 0 to 30 psi outlet gage and 0 to 4.000 psi tank gage

k

..

 
     

tube from tank to inlet on lid

Tygon tube from outlet on lid
to escape valve

1'
A

_|

 

 

_ V

Cylindrical high pressure
tank containing nitrogen
(tank not drawn to scale)

2
N
lrrrrrrrlil

 

 

 

 

 

 

 

 

 

 

IIIrIl unnurl

 

 

 

 

 

 

N2 [2

tit-'-

  
 

, _ escape valve
<— 500 ml beaker With 011

Figuna 10. Anisotropic Etching Apparatus. The same method as regular etching was
used but the etching occurred in an N2 atmosphere.

 

 

"w

L

 

v.

 

l,‘
-~\

 

 

42
through the etching apparatus. The Airco regulator, screwed onto the top of the high pres-

sure tank, consisted of a tank gage. a output gage. and a manually operated valve that con—
trolled the output pressure. Tygon tubes connected the outlet of the regulator to the inlet
port on the lid of the 2,000 ml jacketed beaker. the outlet port of the beaker lid to a flow
valve, and the flow valve to the 500 ml beaker oil.

With the beaker lid tightly sealed onto the beaker, the pressure regulator on the cylin-
drical pressure vessel was opened to approximately 2 psi. The valve between the 2000 ml
beaker lid and the 500 ml beaker of oil was opened slightly to allow bubbles to escape
through the oil at the end of the tube. Observing the N2 gas escaping through the oil allowed
manipulation of the escape rate of the gas.The bubbles escaping were adjusted to a rate at
Which single bubbles, and not a continuous stream of gas, was coming out of the end of the
TYgon tube in the oil. The flow of nitrogen bubbles seen escaping through the oil confirmed

th at a nitrogen atmosphere existed in the beaker and also assured that no air was allowed

back into the chamber where the anisotropic etching was occurring.

2 - 5 Measuring DTF’s Diameters

Etching circular holes through the Si wafers produced circular DTF diaphragms (Figure

l 1 ) - The diameters of these DTF diaphragms were determined from micrographs of X75
and x10 magnifications. Diameters were determined by multiplying the actual measure-
ment of the inner diameter of the protective tube by a ratio of the diaphragm diameter over
the tube’s inner diameter. The ratio of the diaphragm diameter over the tube’s inner diam-

eter was taken from photomicrographs of the specimens such as Figure 11.

43

 

I‘Nigure ll. Photomicrograph of the diamond diaphragm of specimen WF#60-6. The dia-
Phxagm is the lite gray circular area in the middle. The bar represents 1.3 mm.

44
2.6 Raman Spectroscopy

Raman spectroscopy combined with optical microscOpy was sufficient to verify that a given
specimen was a high quality DTF [8]. A 488 nm (blue) laser light with a 2 microns spot

size, was used on a 1,200 groove grading for the Raman spectroscopy.

2-7 Diaphragm Bursting

2.7 .1 Pressure Vessel and Data Acquisition Apparatus

The apparatus used to burst test the DTF diaphragms (Figure 12) consisted of a generic

Cylindrical high pressure tank containing nitrogen gas, a built-to-order cylindrical pressure

Vessel, an MKS type 122AA-05000BB-SPCAL transducer, and an MKS model PDR-D-l
Power supply/readout (MKS Instruments Inc. Andover MA.). The high pressure tank was

SiI‘nilar to that used in the anisotropic etching. It had an open/close valve on top. On this
Valve were two pressure gages and a manual pressure regulator. The first gage registered
the tank’s internal pressure and the second gage registered the regulator’s outlet pressure.
A manually regulator controlled the output pressure. The 122A pressure transducer had an
Operating pressure range of 0 to 100 psia with an output of 0 to 10 volts of direct current,
ace urate to i 0.5 percent of the reading. The necessary input signal was i 15 volts of direct
Current at 35 milliampere. The operating temperature range of the 122A pressure transducer
Was 0 to 50 °C. The transducer was connected to the pressure vessel by a standard cajon
8 VCR fitting. The PDR-D-l power supply/readout was accurate to :L' 0.01 percent of its dis—
Play and had a resolution of l millivolt. The power supply/readout had a conversion rate of

2-5/ sec. and a power supply output of i- 15 volts at 35 milli-amp and a maximum operating

ternPerature of 50 °C.

W t‘;
‘iiilli; n

\
//'\—

45

MKS transducer

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m

pressure vessel

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

cylindrical high pressure

tank containing nitrogen

(tank not drawn to scale) 000 00

MKS power supply readout AC power inlet
readout later connected to an
A/D board in an IBM PC
r r/ ‘
=

 

 

 

Figure 12. Burst Testing Apparatus

2’ .a
l..i.

46

12 mm thick flange 0.37 cm x 3 cm
bolt

standard female
cajon 8VCR fitting flat metal disk

“—r \

ooooo

I

I

I

I

\

‘ ..... ‘

\ l

\ l

\ l

\ l .
\ I i
\ l '
\ I

\ l

\ I
.......

t

|

‘

I

 

inlet flow control valve

 

 

 

 

 

 

 

leak valve

 

 

 

 

E #7

stand bolted onto ,
a table top _>r // J venting

hole
/

fragment cup filled with cotton

 

 

 

 

 

 

 

 

 

12cm ’ 4—5cm—->

A

 

 

Figure 13 (a). Side view of pressure vessel used to burst the DTF diaphragms used in this
study. The pressure vessel was bolted horizontally onto a table top.

ihun

47

 

 

 

 

O-ring inset on
0 surface of flange

threaded holes for
three 0.37 cm x 3 cm

0 bolts

stand bolted onto
a table top

 

 

I ]

Figure 13 (b). Front view of pressure vessel showing the inset rubber O-ring.

<— 7.5 cm ——->

 

 

Figure 13 (c). Front view of flat metal disk showing the 2.5 mm hole in the center and the
three holes for the bolts.

s] g]

L1.

As
Mr
.M

if

' :-
1M

48

The pressure vessel (Figure 13) was a cylinder 12 cm long by 7.5 cm in diameter. At
one end was the pressure inlet flow control valve (model F02, ARO Inc). The maximum
capacity of the inlet flow control valve was 200 psi. At the other end of the cylinder was a
12 mm thick 7.5 cm diameter flange. The flange had a 4.5 cm diameter 3.5 mm thick rubber
O-ring inset on its outer surface (Figure 13 (b)) so that an air tight seal could be obtained
between the flange and the flat metal disk bolted down onto the flange. A flat metal disk
held the specimens being tested. Three bolts, ~ 0.37 cm diameter by 3 cm long, were used
to fasten the flat metal disk against the flange. The flange was 12 mm thick so that the three
holes threaded into the flange for the bolts would not go through the flange. A thick flange
was used in an effort to form a better seal. Any hole through the flange would be a potential
source of a leak. The flat metal specimen disk, which was fastened over the O-ring, had a
2.5 mm hole bored through its center (Figure 13 (c)) over which the specimen was glued.

The specimen was glued with its larger (5 to 6 mm) protective tube attached to the
metal disk (Figure 14). The glue used was a plastic model cement called Testers (The Tester
Corporation, Rockford IL.). The glue was squeezed onto the flat cross-sectional end of the
large protective tube in an amount just sufficient to wet and form an air-tight seal between
the flat cross-sectional end of the larger tube and the flat metal disk. The glue was then
allowed at least three hours of curing before burst testing.

Connected to the outside of the flat metal disk was a fragment cup, 3.8 cm diameter
by 6.4 cm long, which was filled with cotton and placed behind (relative to the bursting
direction) the flat metal disk (Figure 13). The cotton caught the fragments of the DTF films
as the film burst. The cotton was about 8 mm from the DTF diaphragm when the burst

chamber was assembled. The fragment cup was attached to the flange by the same three

49

small tube defining the area etched

wafer with diamond coating

Torr Seal

 

large protective tube

 

 

flat metal disk with a 2.5 mm aperture in the center

Figure 14 (a). Diamond thin film diaphragm specimen and flat metal disk.

protective tube is glued to the flat metal disk

Figure 14 (b). The specimen is glued onto the flat metal disk. The metal disk, with the

SPeeimen glued in place, is then ready to be mounted into the pressure chamber for burst
testing.

65

[IT

by.»

e
l \

.i,‘

Nu

.n. ‘.
~ 4..

'7'-
.4 1.

Hi
.[k

in

 

50

bolts that held the flat metal disk onto the flange. The fragment cup was aligned concentri-
cally along the cylindrical pressure vessel‘s axis of syrrunetry. A hole drilled into the side
of the fragment cup allowed instantaneous venting of nitrogen gas when the DTF burst.

On one side of the pressure vessel was a leak valve which allowed the internal pres-
sure to be reduced from a high chamber pressure to ambient pressure if necessary. The pres-
sure transducer, mounted on top of the pressure vessel. was connected to a power supplying
LED readout that was visually monitored during testing.

A Computer Board Inc. model CIO-AD08-PGL A/D data acquisition board was later
installed in an IBM PC-8086 for more accurate data collection (Figure 12). The signal from
the 122A pressure transducer was channeled into the ClO—AD08-PGL A/D data acquisition
board. The acquisition board had a resolution of 0.024 psi. The software installed [23] in
the PC, see appendix D, sampled the pressure transducer output approximately ever tenth
0f a second. After five samplings, the highest values was written to a data file. The highest

pressure value attained in the pressure vessel was thus recorded in the data file.

2.7.1.1 Leak 'lbsting the Pressure Vessel

The pressure vessel was tested for air leaks. An air tight chamber was desirable for two rea-
sons. First, so that nitrogen gas was not unreasonably spent during each burst test, and sec-
ond, and more importantly, to obtain more accurate chamber pressure readings. If the
Pressure vessel had any substantial leaks, pressure fluctuation would cause incremental
Pressure increases to be impossible. Small incremental pressure increases were desirable
beCause the LED readout was visually monitored during burst testing, and a more accurate

burSt pressure could be determined if the pressure on the readout was increased in small

SI

increments. After the automatic data acquisition was operational, small pressure leaks did
not interfered with the accuracy of the data collected.

To obtain an air-tight pressure vessel, as many joints as possible were welded (Figure
13). A threaded male joint for the inlet flow control valve was welded to the center of a cir-
cular metal plate and the plate then welded to one end of a cylindrical vessel. Teflon tape
was used on the threaded male joints to seal the flow control and leak valves. The threaded
male joint for a leak valve was also welded on the side of the cylinder. At the other end of
the cylindrical vessel the 12 mm thick flange was also welded into place.

A rubber O-iing inset on the outside of the flange would form an air-tight seal with
the flat metal disk as the flat metal disk was fastened down onto the flange. A standard
female cajon 8VRC fitting for the transducer was also welded onto the top of the pressure
vessel. With the male cajon 8VRC fitting of the transducer fastened in place, this joint too
was air-tight.

To test the pressure chamber, the 2.5 mm aperture (figure 14 (a)) in the flat metal disk
was covered with scotch tape. The disk was then fastened down onto the rubber O-rin g on
the flange for an air-tight seal. The leak valve was closed, and the transducer mounted into
its 8VCR cajon fitting. Approximately 5 psi of nitrogen gas from the high pressure tank was
then allowed into the pressure vessel and the inlet flow control valve closed. The tygon tube
to the flow control valve and the electrical wire to the transducer were quickly disconnected
and the pressure vessel then submerged in a sink of tap water. The pressure vessel was
totally submerged in water to cover everything but the attached transducer. If bubbles came
from any joint, the joint was unfastened and re-taped with a thicker layer of teflon tape in

order to improve the seal. This procedure was repeated until leaks were minimal.

52
2.7.2 Diaphragm Bursting Procedure

The burst pressure determined for each specimen was a differential pressure, therefore the
ambient pressure was recorded before each test. The flat metal disk with the DTF dia-
phragm specimen glued onto it was bolted onto the flange. The fragment cup was bolted
behind the small hole on the flat metal disk through which the fragments of the film would
emerge upon bursting (Figure l3(a)). The three 0.37 cm diameter by 3 cm long bolts were
fastened tightly into the flange to insure a good seal between the flat metal disk and the rub-
ber O-ring. The leak valve on the side of the pressure vessel was closed and the N2 gas inlet
flow control valve was opened.

The open/close valve on the high pressure tank was opened and the tank gage imme-
diately showed the pressure inside the tank. The regulator valve was then opened until the
regulator gage read approximately 5 psi. The flow control valve was opened slowly while
the pressure readout was monitored. The manually operated regulator valve was not suffi-
ciently sensitive to allow the pressure in the pressure chamber to be increased
incrementally, thus, the chamber pressure was increased slowly and continuously. The rate
of pressure increase in the pressure vessel was monitored and kept at approximately 0.1 psi
per second. This rate was visually approximated and set only in an attempt to minimize the
human error in recording the burst pressure from the digital readout. If the pressure in the
pressure vessel rose too rapidly the digits on the readout would be changing too rapidly to
visually detect the maximum value displayed at burst. On the other hand, the pressure in
the pressure vessel could be raised too slow for practicality. In adjusting the pressure
increase rate in the pressure vessel, if the rate of pressure increase in the burst chamber was

too fast the N2 inlet flow control valve would be closed slightly to lower the rate. If the rate

H)

9

1.44

53

was too slow the flow control valve would be opened to increase the rate.

Once the pressure-increase rate was set, the pressure was increased at a constant rate
while visually monitoring the LED readout until the DTF diaphragm burst. The DTF dia-
phragm would always be the weakest place in the chamber walls and therefore the first to
fall under the increasing pressure. The DTF diaphragm failed with an audible pop and an
immediate drop in vessel pressure to ambient pressure. The maximum pressure that was
seen on the LED readout was then recorded as the failure pressure of the DTF diaphragm.

After the automated data acquisition system was installed, the testing procedures
were the same except that the LED readout was not monitored for the maximum pressure.
The maximum pressure attained in the vessel was immediately displayed on the PC monitor

after each test. A data file of pressure versus time was also created for each test.

2.7.2.1 Obtaining the Burst Pressures

At the start of each test, the ambient pressure was recorded. The failure pressure, the max-
imum pressure attained in the pressure vessel, was determined at the end of the test.
The burst pressures for each specimen was calculated by subtracting the ambient pressure

from their failure pressure.

2.8 SEM’s Preparation and ’lbsting Procedures

Once the diamond diaphragms were burst and the burst pressures recorded, the fragments
of the burst diamond diaphragms were inspected in the SEM to determine their grain sizes,
film thicknesses, and to study the morphology of the diaphragms. Three different SEM’s

were used in this investigation, a JEOL JSM-35CF, a Hitachi S-415A, and a Hitachi

54
S-2500C.

To prepare the diamond film fragments for SEM testing. it was necessary to retrieve
the film fragments from the cotton-ball fibers (Figure 13 (a)). mount the fragments onto the
SEM stubs, and then gold coat the mounted fragments to make them electron conductive

thus suitable for SEM use.

2.8.1 Retrieving and Mounting of the DTF Fragments

The cotton balls with the entangled fragments were removed from the fragment cup (Figure
13) after each burst. The fragments of the DTF diaphragms were retrieved (untangled) from
the cotton ball fibers with the use of a toothpick.

The fragments from each burst test were untangled over a SEM stub which had a
sticky tab (catalog # AT, ME. Taylor Engineering, Inc.) on its top surface (Figure 1.5 (a)).
Sticky tabs are a standard SEM specimen mounting device. The fragments fell onto the
sticky tabs and stuck in random positions and locations (Figure 15 (b)). They were left in
the positions and locations in which they fell because they were too small to handle, and
positioning them to simplify thickness measurements would be impossible.

A large number of fragments from each burst test was dropped onto a stub in order
to increase the probability of getting one or more fragments to stick at an angle of 30 to 90
degrees with the surface of the stub. This angle of 30 to 90 degrees between the flat frag-
ments and the surface of the stub would allow easier thickness measurements of the edges
of the film fragments to be made.

Each specimen stub was given the label of the DTF fragments mounted on it to avoid

confusion during SEM testing.

55

   
  

<— toothpick

cotton fibers with ____>

diamond fragments
1
k w l
| v . A
‘ S E M stu b

sticky tab

 

 

 

 

Figure 15 (a). Fragment mounting on SEM stubs. The DTF fragments were entangled in
cotton fiber after the burst test. The fragments were retrieved by using a toothpick to pull
the cotton fibers apart.

diamond fragment
sticky tab

' * SEM stub

Figure 15 (b). The fragments fell onto the sticky tab on the SEM stub and remained in the
random positions and locations in which they fell.

56
The specimen stubs used were all made of aluminum. The stubs for the JEOL J SM-

35CF were 0.5 cm height and 0.9 cm in diameter. The stubs for the Hitachi S—415A were
1 cm high and 1.5 cm in diameter. The same stubs of the Hitachi S-415A were also used in
the Hitachi S-2500C. Although not the proper stub for the Hitachi S—2500C, the S-415A

stubs were adhered onto the actual S-2500C stub with a double sided carbon tape (Figure

16). The actual S-2500C stub was 0.75 cm high and 3.5 cm in diameter.

2.8.2 Coating the Mounted Fragments

The DTF fragments mounted on the SEM stubs were then gold coated using an Emscope
Sputter Coater model SCSOOM (Ashford, KT.) The sputtering rate was 7 nm per minute and
sputtering was done for approximately 3 minutes. A gold coating of approximately 21 nm
thick (7 nm per minute by 3 minutes) was thus sputtered over the entire fragment and SEM
stub. The gold coating was necessary to make the specimens conductive to avoid charging
during testing. Charging causes excessive brightness and distortion in the SEM image.

After gold coating the specimens were then ready for SEM observation.

2.8.3 Specimen Insertion into the SEM’s

Specimen insertion and exchange procedures were quite similar for the newer JEOL J SM-
35CF and Hitachi S-2500C SEM’s. However, the specimen insertion and exchange proce-
dures for the Hitachi S-415A SEM was quite different than those of the other two SEM’s.
The main difference was due to the fact that the specimen chamber for the Hitachi S-415A
had to be pumped-down from ambient pressure after each specimen insertion, but for the

JEOL ISM-35CF and Hitachi S-2500C only a small insertion (waiting) chamber had to be

57

1.0 cm by 1.5 cm specimen stub

0.6 cm by 2.4 cm S-2500C SEM stub
\ double sided

carbon tape

/

f <— screw securing
\ brass holder
' cover

attachment to insertion chamber

spherical knob

 

 

   

insertion rod

 

Figure 16. Hitachi S-2500C SEM specimen holder. The specimen holder assembly is
attached to the insertion chamber by the temporary cover. The 1.0 cm by 1.5 cm specimen
stub was adhered to the S-2500C SEM stub with a carbon tape. The S—2500C SEM stub
was screwed securely onto the brass holder which is inserted into the specimen chamber
by the spherical knob and insertion rod and remained there during testing.

58
pumped-down, which required less time. This made the JEOL J SM-35CF and Hitachi S-

2500C SEM’s more convenient to use, because exchanging specimen was less time con-
surning.

However, exchanging specimens on the Hitachi S-2500C SEM was not as quick as
on the JEOL ISM-35CF. On the Hitachi S-2500C SEM, the accelerating voltage was dis-
charged before each specimen removal and recharged after each specimen insertion. This
procedure of discharging and recharging of the LaB6 gun with the accelerating voltage took
approximately 20 minutes and was done as a safety precaution to protect the costly LaB6
gun from possible bum-out during specimen exchange. On the JEOL ISM-35CF this pro-
cedure of discharging and recharging was not performed, making the JEOL ISM-35CF the
most time efficient of the three SEM’s.

Before turning on any of the SEM’s, the water pressure for the cooling pumps were
checked to make sure an adequate pressure was in the lines. The chamber pressures of the
JEOL J SM-35CF and the Hitachi S-2500C were also checked to verify operating chamber
pressure. The specimen chamber pressure for the JEOL J SM-35CF and the Hitachi S-
2500C SEM’s were usually maintained at operating pressure, ~10’6 psi. The specimen
chamber pressure for the Hitachi S-415A SEM however, was not maintained at operational

pressure between use and pumping-down the chamber was necessary before each insertion.

59
2.8.3.1 Hitachi S-415A

Once the water-coolant pressure was checked to be in an operational state, the S-415A
power switch was turned on. The mounted and gold coated film fragments on the 1.0 cm
high by 1.5 cm diameter aluminum stub was then secured onto the specimen holder (Figure
17). The specimen chamber was vented and the cover of the entrance port pulled out to
reveal a stage onto which the specimen holder was place. The stage, built onto the inside of
the door, was designed to hold the specimen holder snugly and allowed x, y, and z motions,
rotation, and tilting of the specimen (the terms x, y, and z signifies three dimensional Carte-
ian coordinates). The specimen on the holder was placed securely on the stage and the
entrance cover to the port was replaced. The specimen chamber was then pumped-down to

operational pressure and the SEM was ready for testing.

2.8.3.2 JEOL J SM-35CF
Once the water coolant pressure and the specimen chamber pressure were checked to be in
operational state, the J S M-35CF power switch was turned on. The mounted and gold coated
film fragments on the 0.5 cm high by 0.9 cm diameter aluminum stub was placed onto the
specimen holder (Figure 18). The height of the stub was adjusted with the height adjust-
ment screw (Figure 18) to make it flush with the top of the brass holder. The stub was then
secured in place with the securing screw (Figure 18). The temporary cover was moved
toward the brass stage until it clipped into place. The clip kept the stage from moving
inwards during pump-down of the insertion chamber.

A red vent/vac button on top of the entrance to the small insertion chamber (Figure

19) was pressed which immediately vented the insertion chamber to ambient pressure. The

60

1 cm by 1.5 cm metal specimen stub

\ spring clip

 

 

 

 

SCTC W’

 

metal stub base

/

 

 

 

Figure 17. The Hitachi S-415A SEM specimen holder consisted of a metal stub base, and
1 spring clip which held the 1 cm by 1.5 cm specimen stub in place. The spring clip is
reight adjustable and screws into or out of the metal stub base to adjust the specimen into
Focusing range.

61

0.5 cm by 0.9 cm

 

 

 

 

 

 

 

temporary cover _
I specrmen stub
insertion rod brass holder securing
\ screw
\ 4' " '——. “““ .

spherical knob height adjustment for stub

Figure 18. JEOL ISM-35CF SEM specimen holder assembly. The spherical knob was
used to screw the steel rod into the brass stage. The knob was also used to move the brass
stage between the insertion chamber and the specimen chamber. The temporary cover
would replace the true cover only while the specimens were being loaded

 

62

heavy high voltage coaxial cable

 

specimen chamber containing electron gun \

view of brass holder in specimen chamber

 

 

locking slide door and knob
red vent/vac buttca— \ . '

spherical knob —>

temporary cover /

   

 

 

view of brass holder in small insertion chamber

 

 

 

Figure 19. JEOL ISM-35CF specimen chamber and insertion chamber.

 

. *cl’I’

63

metal cover on the port of the insertion chamber was removed from the entrance once the
chamber was vented. The metal cover of the entrance was similar in size and shape to the
temporary cover (Figure 18) built onto the specimen holder assembly. The stage with the
specimen was placed in the insertion chamber. The temporary cover on the specimen holder
assembly covered the entrance port. The red button on top of the insertion chamber was
pressed again which pumped-down the insertion chamber. When the pressure in the inser—
tion chamber became equal to that in the specimen chamber, the locking slide door separat-
ing the two chambers (Figure 19) was opened by pulling out the slide door knob. The brass
holder containing the specimen was then moved from the insertion chamber into the spec-
imen chamber with the insertion rod (Figure 19). The rod was unscrewed from the brass
stage and retracted leaving only the stage with the specimen in the specimen chamber. The

locking slide door was returned to lock position and the specimen was ready to be tested.

2.8.3.3 Hitachi S-2500C

The water coolant pressure and specimen chamber pressure were checked to be in opera-
tional state, before the S-2500C SEM was switched on. At the push of a similar vent/vac
button, the insertion chamber was aired and its cover was pulled away from the entrance to
reveal the brass holder (Figure 16). The specimen holder assembly on the S-2500C was
built onto the insertion chamber cover. A 1 cm high by 1.5 cm diameter aluminum stub with
the mounted and gold coated film fragments on the it was then secured onto the S-2500C
stub with a double sided carbon tape (Figure l6).The S-2500C stub was screwed securely
onto the brass specimen holder. The double sided carbon tape which held the two stubs

together was used for several consecutive specimen stubs.

64
Once the specimen stub was secured on the S-2500C stage, the stage was pulled back

until a clip held it in position near the cover. The entire door assembly was then slid back
in place over the port of the insertion chamber. The vent/vac button was depressed again.
which pumped down the insertion chamber.

Mthin 15 seconds the insertion chamber containing the specimen was at a pressure
equivalent to that in the specimen chamber, approximately 106 psi. A locking slide door
separating the insertion chamber and the specimen chamber (similar to Figure 19) was then
opened and the specimen pushed from the insertion chamber into position in the specimen
chamber. The rod was then unscrewed from the brass stage using the spherical knob and
retracted from the specimen chamber. The locking slide door between the two chambers

was then locked. The S-2500C was then ready for testing.

2.8.4 SEM Procedures for 'Ibsting Specimens

Procedures for grain size, film morphology, and film thickness determination were quite
similar for the three different SEM’s once the specimen was in the specimen chamber and
ready for testing. Once in the specimen chamber, each SEM was capable of maneuvering
the specimen in order to take grain size, film morphology, and film thickness micrographs.
The only diflerences were the positioning of the controls and the resolution capability on
each SEM. The Hitachi S-2500C was generally thought to have greater resolution than the
JEOL J SM-35CF, and the JEOL ISM-35CF was thought to have a better resolution than the
Hitachi S-415A. The resolution of the micrographs taken by the Hitachi S-415A were vis-
ibly lower than the resolution of those taken by the JEOL J SM-35CF and the Hitachi S-

2500C. However, the resolution of the micro graphs taken by the JEOL ISM-35CF showed

65
no obvious difference to those taken by the Hitachi S-2500C.

2.9 Grain Size

Once the specimen was in the specimen chamber. the specimen was maneuvered using the
x and y controls to locate an appropriate fragment of which a grain size and/or film mor-
phology micrograph was taken. An appropriate fragment was one with its faceted side up
and also lying flat on the specimen stub.

Not all the fragments on the specimen stub showed their faceted side up. Some frag-
ments had their substrate (interfacial) side up. Nor were all the fragments lying flat against
the specimen stubs. Some were at an angle to the stub, the same position in which they had
fallen onto the sticky tabs (section 2.7.1 Retrieving and Mounting of DTF Fragments). A
fragment showing its faceted side and also lying flat against the surface of the specimen
stub, or only at a small angle with the stub’s surface. would give the best micrograph for
grain size and film morphology because it would minimize foreshortening in each direc-
tion.

An appropriately positioned fragment was located by scanning the specimen stub at low, ~
x60, magnification to find a fragment that appeared to be lying flat on the specimen stub.
This fragment was then observed at a higher, ~ x1800, magnification to see if its faceted
side was up and if the fragment was truly lying flat on the specimen stub. If both Conditions
of faceted side up and flatness to the specimen stub was not met, another fragment was
checked until a fragment satisfying both conditions was found.

If a fragment chosen for grain size micrograph was not flat to the surface of the spec-

imen stub, the fragment was adjusted before the micrograph was taken. The angle between

66

the fragment and the stub’s surface was determined by finding a 90 degrees edge on the
fragment and tilting the fragment until that edge just disappeared. At this position it was
assumed that the fragment was showing a flat surface to the direction of observation and

the grain size micrograph was taken.

2.9.1 The Average Line Intercept Method
The average line intercept (ALI) method required a x8,000 to x12,000 magnification SEM
micrograph of the faceted surface of the DTF so that at least 10 grains spanned the shorter
side of the rectangular photomicrograph. Grain sizes were on the order of a micron.
Points were marked along the perimeter of the micrograph and labeled. A straight
edge was placed across the micrograph from one labeled point to another. The number of
grain (diamond facet) boundaries intercepted by the straight edge were counted. The
straight edges was then moved to another pair of labeled points and the grain boundary
intercepts were counted again. This was done as many times as was necessary to get a total
sum of at least 250 intercepts. The lengths of the line segments, whose intercepts added up
to ~ 250, were also summed. The ALI value was then determined by dividing the sum of
the lengths by the sum of the intercepts, ~ 250. The quotient was then divided by the mag-
nification of the photornicrograph and reported in microns. The ALI value was then multi-
plied by a correction factor of 1 which gave a value characterizing the grain size of the

polycrystalline diamond film. The correction factor could have a value of 1.0 to 1.5.

67
2.10 Film Thickness

Maneuvering the specimen for a film thickness micrograph was a more involved procedure
than maneuvering for a grain size micrograph. It was necessary to get a micrograph of the
edge of the film that showed the true thickness of the edge. Ideally, a fragment positioned
with its edge normal to the direction of observation was desired in order to get a true length
(or thickness). If maneuvering the specimen to such a position was too difficult, the true
length could often be determined by using a tilt correction feature common to most
SEM’s.To get the true thickness of the fragment’s edge, the fragment was usually posi-
tioned with its edge at a known angle to the direction of observation. An angle of 30 or 45
degrees from the direction of observation was most often used. The tilt correction was then
set to correct for the 30 or 45 degrees tilt before a film thickness micrograph was taken (Fig-
ure 20). In this manner the true thickness of the film fragments were imaged. The SEM
micrographs also had bar scale on them which allowed accurate determination of the film
thickness.

The thickness of a CVD diamond film may also be approximated by measuring the
weight gained by the Si wafer during deposition. The weight gained is divided by the mass
density of the polycrystalline diamond thin film (DTF) which gives a volume. This volume
is then divided by the area which is covered by the DTF which gives the film thickness;
assuming a film of uniform thickness. This method of determining the film thickness how-

ever, was not done.

68

 

Figure 20. SEM micrograph of specimen WF#60-25 showing the thickness of a fractured
polycrystalline diamond diaphragm. The dotted bar represents 5 microns.

69
2.11 Additional Specimens Obtained for Testing

Two additional (100) single crystal Si wafers with diamond thin films (DTF’s) chemically
deposited on them were later obtained for testing. These two specimens were deposited by
the same microwave system as the earlier specimens, however, the microwave system was
modified with a larger deposition chamber which allowed the deposition of larger DTF’s.

The two large specimens obtained were received as approximately 4.85 cm radius semicir-
cles. These two semicircular specimens with DTF’s on them were received with the labels
WF#25 and WF#60. These labels were included in the labels given to each individual spec-

imens taken from these wafers (section 2.11.3).

2.11.1 Photoresist Pre-deposition Treatment

The diamond coated semicircular wafers WF#25 and WF#60 were treated by a photoresist
method prior to deposition [6]. The photoresist method involved centrifugally covering a
wafer with a suspension of fine diamond powder in photoresist. During deposition the pho-
toresist was baked off leaving only the diamond powder on the wafer. These diamond par-

ticles acted as nucleation sites for the DTF that was grown on the wafers.

2.11.2 Deposition Conditions

Specimen WF#25 was deposited at a microwave power of 2,098 watts, a deposition pres-
sure of 40 Torr, and precursor gas flows of 400 sccm H2, 4 sccm C02, and 4 seem CH4. The
surface temperature of the wafer during deposition was 850 °C. The weight gain was 7.26
milligram per hour per milligram Si. wafer for a deposition time of 8 hours. The Si wafers

had a uniform thickness of approximately 0.53 mm.

70

Specimen WF#60 was deposited at a microwave power of 2.861 watts, a deposition
pressure of 35 Torr, and precusor gas flows of 400 sccm H2. 2 sccm C02. and 6 sccm CH4.
The surface temperature of the wafer during deposition was 776 °C. The weight gain was
9.42 milligram per hour per milligram Si wafer for a depositing of 8.5 hours. The Si wafer

a uniform thickness of approximately 0.53 mm.

2.11.3 Specimen Labeling
The large semicircular wafers with DTF’s on them were fractured into final specimen size,
approximately 1 cm squares, as mention earlier (section 2.1.3). The individual 1 cm square
specimens from each large semicircular piece were labeled as follows: from specimen
WF#25, WF#25-l, WF#25-2, WF#25-3, etc. and from specimen WF#60, WF#60-l,
WF#60-2, WF#60-3, etc. This was similar to the labeling of the specimens initially tested
and mentioned earlier (section 2.2.3).

During the fracturing of WF#60 the relative positions of each specimen taken from
that wafer was recorded (Figure 21). After all the WF#60 specimens were burst tested and
their thicknesses detemrined on the SEM, the film thicknesses were compared to their rel-

ative position to examine if any thickness-position relationship existed.

71

//i?

 

 

 

 

 

 

11 13 >
22
10 12 18 19
4 2 l6 \ 17 24 2.5
F7 i 3 l \ 14 15 23 26

 

 

 

 

 

 

Figure 21. The relative positions for the 26 specimens taken from wafer WF#60 were
recorded for later. film thickness versus relative position correlation.

3. Results and Discussion
3.1 Results in Wafer Preparation
3.1.1 Results of Abrading

The method of abrading the Si wafers by hand (section 2.1.1) had a low reproducibil-
ity. The amount of diamond powder used, the load applied, the direction of the abrading
motion, the frequency of the motion and time duration of the motion, were all variables that
were either not monitored precisely, or not monitored at all. The average time of abrading
was approximately 10 minutes, but without knowing the applied force and frequency of the
abrading motion. this information would not be very useful in reproduction of the proce-

dure.

3.1.2 Photoresist Treatment

The pre-deposition photoresist treatment to increase nucleation sites, developed by
Aslam et a! [6], seemed an acceptable method for creating nucleation sites. The WF#25 and
WF#60 specimens were prepared by the photoresist pre-treatment. The DTF’s and were
sufficiently adhered to the Si wafer to allow fracturing, dimpling. and etching without the
DTF flaking away from the Si wafer.

A clear difference however was noted between the DTF’s deposited on photoresist
treated Si wafers, WF#25 and WF#60, and those deposited on abraded Si wafers, DF-Nl,
DF—N2, DF—N4, and DF—NS. (Recall the DF-N3 specimens were assumed not to be of a
high diamond content see section 3.2 and were not tested.) Most of the SEM micrographs
taken of the bottom of the WF#25 and WF#60 DTF’s showed distinct grains (Figure 22 (a)

and (b)). No such patterns were noted on any of the other DTFs’ bottoms. The DTF

72

73

 

Figure 22 (a). Micrograph of the interfacial surface of the DTF specimen WF#25-ll
showing distinct grain boundaries define by the dark lines. A grain size of 0.71 micron
was determined from this micrograph. The dotted bar represents 5 microns.

74

 

Figure 22 (b). Micrograph of the interfacial surface of DTF specimen WF#60-l showing
distinct grain boundaries define by the dark lines. A grain size of 1.00 micron was deter-
mined from this micrograph. The dotted bar represents 3 microns.

75
3.2 Film Deposition

Five sets of specimens were deposited; DF-N 1. DF-NZ. DF-N3. DF-N4. and DF-N5 (Table
1). Each set had at least nine specimens. Sets DF-Nl. DF-N2, and DF-NS varied only in
deposition times of 6, 8. and 10 hours respectively. Sets DF—Nland DF-N4 varied only in
their microwave deposition power of 900 and 700 watts respectively. Every set except DF-
N3 was deposited at 0.5 percent methane in hydrogen. Set DF-N3 was deposited at 0.75
percent methane in hydrogen.

Set DF-N3 was believed to be amorphous as determined by optical microscopy. The
thicknesses of fragments from this set were also much larger than those from the other sets
which would also support the claim of the DF-N3 films being mostly amorphous. The
higher concentration of methane used in the deposition precusor gas for set DF—N3 was
attributed to this effect.

Concentration up to 1 percent methane in hydrogen, is typically considered the opti-
mal concentration range for diamond deposition. However, at 0.75 percent methane in
hydrogen, set DF-N3 was considered to be mostly graphitic or amorphous carbon.

The average temperature determined for set DF-N4 was 766 i 5 °C and for set DF-

N5 the average deposition temperature was 774 i 15 °C.

76
3.3 Results in Diaphragm Preparation

3.3.1 Result of Etching

The polycrystalline diamond thin film did not always protect the Si wafer that it covered
from the KOH etch. The DTF deposited onto the single crystal Si wafers was expected to
protect the Si wafer from the strong chemical etch because of diamond ’5 high chemical
inertness [3,5]. However, some of the DTF flaked off from the Si wafers (Figure 23) after
several hours of exposure to the KOH etch. Gas bubbles produced on the surfaces of the
DTF’s during etching indicated that etching occurred on or through those DTF surfaces.
The strong KOH etch was believed to be responsible for the flaking away of the DTF’s from
their Si substrates because gas bubbles were evolving from the surfaces of the DTF’s.

To avoid destroying the DTF’s with the etchant, the DTF’s were protected from the
etchant with the cover slips of microscope slides. The glass cover slip was placed over the
large protective tube on each specimen and then Torr Seal was used to seal the glass cover
slip in place (Figure 24).

After the cover slip alteration to protect the DTF’s, flaking of the DTF’s from their Si
substrates was no longer observed. This result indicated that the strong KOH etch was
affecting the DTF’s by somehow getting through the thin film and attacking the Si substrate
beneath.

Two possible reasons for the adverse reaction of the DTF’s to the KOH etch seems
likely. One reason is that pinholes may have existed through the films, allowing the etch to
attack the Si substrates beneath. Secondly, the DTF’s may not have been of a high diamond
quality (high percent diamond phase), but rather diamond films with an appreciable amount

of graphitic and/or amorphous carbon, allowing the KOH to etch through and attack the Si

77

 

Figure 23. The flaking off of the DTF from the Si substrate is obvious on specimen WF#25-
12. The etch seem to have penetrated the diamond film and attach the etch'beneath. The cir-
cular diamond diaphragm is also observed in the center. The bar represents 1 mm.

‘78

small silica tube

large silica tube 31 substrate

 

“mg:

 

      

diamond coating

I“-‘---“‘
D.--.-.- o. f
-'

 

:-

glass cover slip added

Figure 24. Cover slip modification. A glass cover slip was added to protect the diamond
coating from the KOH etch. The inner diameters of the large and the small tubes were 6.8
mm and 2,3, or 4 mm, respectively.

1'11.“

.jvi

if)

79

beneath. The Raman spectroscopy results showed that there was an appreciable amount of
graphitic carbon in the diamond films (Figure 27 (a) and (b)). supporting the second possi-
bility.However, carbon is reportedly resistant to KOH (from 5 to 90 weight percent) etchant
up to 160 °C [32]. This would rule out the second possibility suggested. leaving pinholes
as the only suspect.

The second possibility, the effect of pinholes. may better explain the unexpected
adverse effect of the KOH etch on the DTF films. It is possible that the pinholes penetrating
through the films allowed the KOH etch through to attack the Si wafer beneath. Pinholes
penetrating through the films would more likely be found in films whose thickness ranged
with the size of the constitutive grains of the polycrystalline film. The thicknesses of the
films that were etched ranged from approximately 2 to 5 microns with grain sizes ranging
from approximately 0.4 to 1.1 microns. However, pinholes were not specifically searched

for, and none were observed during an SEM examination.

3.3.1.1 Anisotropic Etching

The attempts made at anisotropic etching were unsuccessful since the holes resulting from
the anisotropic etch were less circular than those from the isotropic etch. Usually two par-
allel sides resulted instead of the circular shape desired and expected. For the anisotropic
etching done elsewhere [22], the shape desired was rectangular. The area initially exposed
to the KOH etch was therefore also rectangular [22], as opposed to the circular shape used
in this experiment The anisotropic etch therefore seemed more suitable for producing rect-
angular holes than circular ones. 8

Perhaps as the etch proceeds through the wafer, a rectangular shape can easily be

80

maintained since the (010) and (00] ) principal planes are preferentially un-attacked. In con-
trast. an initially circular shape defines no such edges. The preferential etching may explain
the unexpected pattern of two parallel sides noticed with the anisotropic etch. The anisotro-
pic etch was thus aborted.

The isotropic etch, which produced diaphragms of greater circularity, was used in all
subsequent etching. Circularity of the diamond diaphragm was essential to the simplicity
of the theoretical calculations for the biaxial strength (as determined in the diaphragm

bursting experiment).

3.3.2 Result of Dimpling

Diaphragms prepared by dimpling partway through the Si wafers before etching (Figure
25) showed greater circularity than diaphragms produced by etching circular areas of the
Si wafer defined by small tubes. Dimpling to define the area to be etched is a method of
producing diaphragms that was developed by Bill Glime [24]. The greater circularity that
the dimpling procedure produced led to the decision to produce all remaining diaphragms
by the dimpling method.

Although the dimpling method produced diaphragms of higher circularity, it had the
disadvantage that the Si wafer tapered down at the diaphragm edge [24]. The diamond dia-
phragm may be thought of as a clamped circular membrane assuming no bending moment
at the boundary, but the thin edges means that likely (it was not measured), the edge does
rotate, indicating a non-zero moment. Restated, the assumed boundary conditions of no lat-
eral displacement at the diaphragm’s boundary is no longer satisfied if the supporting Si is

so thin that it deflects during the loading of the diamond diaphragm. The assumed boundary

81

 

Figure 25.Typical diamond diaphragm (WF#60-15) produced by etching a concave and
circular dimple that was grounded into the Si wafer with a TEM sample preparation dim-
pler. Al mm bar is indicated.

82

conditions of the diaphragm would also be violated if the Si was so thin at the boundary
that, instead of only the diamond diaphragm breaking, failure also occurred through the thin

Si wafer.

3.4 Data Collected

The data acquired for each specimen included diaphragm diameter. burst pressure, grain
size, and film thickness. Additional data such as Raman spectroscopy and some SEM
micrographs of DTF interfacial surfaces, film morphology, and cracks in the DTF’s were
also taken for some of the films. The numerical data collected are tabulated in Table 2 for
DF-Nl, DF—NZ, DF-N4, and DF-NS, Table 3 for WF#25, and Table 4 for WF#60. A max-

imum error of 10 percent may be present in the SEM measurements as no calibration was

done to the SEM prior to testing.

3.4.1 Burst Pressures

The burst pressures for each diaphragm was determined as mentioned earlier (section
2.7.2.1).The burst pressure for the DF—N 1 diaphragms were 7,172 Pa and 135,586 Pa. The
burst pressure for the DF—N2 diaphragms varied from 60,552 to 265,655 Pa, for DF-N4 dia-
phragms from 14,276 to 252,138 Pa, and for DF-N5 diaphragms from 84,000 to 254,759.
The burst pressure varied from 23,793to 370,758 Pa for the WF#25 diaphragms, and from

10,103 to 390,620 Pa for the WF#60 diaphragms.

83

Table 2. Burst pressures, diameters, thicknesses, and grains sizes for DF-Nl. DF-N2,
DF-N3, DF-N4, and DF—NS specimens. Set DF—N3 was not included in analysis.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Specimen Burst Dlgggggn Thickness Grain Size
Pressure (Pa) (mm) (micron) (rrucron)
DF-Nl-S 7,170 5.6 2.3 1.1
DF-N1-6 136,000 5.0 2.0 0.9
DF-N2-2 144,000 5.2 2.7 0.5
DF-N2-3 60,600 6.8 1.9 0.6
DF—N2-4 267,000 3.4 2.8 0.6
DF—N2-5 e 255,000 5.1 3.1 0.6
DF—N2-7 173,000 4.5 2.1 0.9
DF-N4-2 173,000 4.7 3.0 0.7
DF-N4-3 14,300 4.8 2.0 0.3
DF-NM 94,600 5.6 2.8 0.6
DF-N4-5 252,000 3.6 3.3 0.6
DF-N4—8 5 128,000 6.8 2.5 0.5
DF—N5-2 191,000 3.6 4.5 0.9
DF-N5-3 144,000 6.8 3.7 0.8
DF-N5-4 100,000 6.5 4.3 1.0
DF-NS-S 84,000 6.8 4.7 0.4
DF—N5-6 140,000 6.5 5.3 0.6
DF-N5-8 5 255,000 6.8 3.2 0.9

 

 

 

 

 

N.A. - either not available or not applicable.
- Data points removed from this set for best fit of entire data set.

 

 

84

Table 3. Burst pressures. diameters, thicknesses, and grain sizes of WF#25 specimens.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Grain Size

......... sis... ”3553?? has 3:222:33: 32333?

(mm) (rrucron)
WF#25-7 126,000 2.4 5.0 0.8 NA
WF#25-8 150,000 3.4 2.5 0.8 NA
WF#25-9 155,000 2.1 2.3 0.6 NA
warm-1033 23,800 2.5 2.5 1.0 NA
WF#25-11 98,100 5.2 3.0 0.7 0.71
WF#25-l4 92,600 2.6 2.2 0.6 0.86
WF#25-15 371,000 1.9 2.4 0.5 NA

69

- Data points removed from this set for best fit of entire data set.
NA - not available.

85

Table 4. Burst pressures, diameters, thicknesses, and grain sizes of the WF#60 specimens.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Diaphragm . . . Grain Size
Specimen Burst Diameter Thickness Grain Size of DTF
Pressure (Pa) (mm) (rrucron) (rrucron) (33:33:11)
WF#60-1 342,800 2.3 4.4 1.3 1.0
WF#60-2 94,300 4.6 3.2 1.1 NA
WF#60-3 86,900 4.6 2.5 1.1 1.0
WF#60-4 366,400 1.4 2.8 1.3 NA
WF#60-5 8’ 77,450 4.2 3.5 1.2 NA
WF#60-6 83,520 4.7 3.6 1.0 1.0
WF#60-7 9 87,240 3.8 2.2 0.5 0.8
WF#60-8E 258,300 2.4 2.5 1.1 1.0
WF#60-9 56,550 3.5 2.5 1.1 NA
WF#60-11 151,800 3.3 2.5 1.0 NA
WF#60-12 252,900 3.1 3.5 1.3 NA
WF#60-l4 135,500 5.3 3.2 0.9 NA
WF#60-15 191,200 3.1 3.3 1.3 0.8
WF#60-16 132,300 4.2 3.3 1.3 1.0
WF#60-17 116,200 4.3 3.0 1.3 1.0
WF#60-18 ® 390,600 2.5 3.4 1.2 NA
WF#60-l9 163,000 3.6 3.3 1.3 NA
WF#60-21 ‘8 72,760 3.9 3.0 0.9 NA
WF#60-23 152,500 3.4 2.5 1.3 NA
WF#60-25 119,200 3.0 2.0 1.0 1.3
WF#60-26 10,100 6.5 2.6 1.0 NA

 

® - Data point removed for best fit WF#60 data set.
- Data points removed from this set for best fit of entire data set.
NA - not available.

 

. 86
3.4.2 Grain Size

The grain size of each film was measured by the average line intercept method (section.
2.8.4.1.1) from SEM photomicrographs. SEM photomicrographs of the deposition (fac-
eted) surface were used to determine grain size for the DTF specimens. On DF-N l , DF4N2,
DF-N4, and DF-NS the interfacial surfaces, as photographed by the SEM, showed no gran-
ular‘ structure, therefore no grain size determination were done of the interfacial surfaces
for these specimens. However, some WF#25 and WF#60 films showed distinct granular
structure on their interfacial surfaces. Many specimens from WF#25 and WF#60 were

therefore also tested for the grain size of their interfacial surfaces.

\
t

The grain sizes for the DF-Nl DTF were 0.871 micron and 1.124 micron. The grain
sizes for the DF-N2 DTF varied from 0.529 to 0.860 micron, for DFLN4 from 0.327 to
0.719 micron, and for DF-NS from 0.423 to 1.02 micron. The grain size varied from 0.472
to 0.990 micron for WF#25, and from 0.54 to 1.26 micron for WF#60.

The grain boundaries were seen only on some of the films prepared by the photoresist
pre-deposition treatment. Films WF#25 and WF#60 were prepared by using the photoresist
treatment which may be the reason for their interfacial difference with the films from DF-

Nl, DF—NZ, DF-N4, and DF-NS prepared by abrading with the 1 micron diamond powder.

3.4.3 Film Thickness

Film thickness measurements for each DTF was taken from SEM micrographs (section
2.8.4). The film thicknesses determined for the two DF—N 1 specimens were 2.0 microns and
2.3 microns. The film thicknesses determined for the DF-N2 specimens varied from 1.9 to

3.1 microns, for DF-N4 the thicknesses varied from 2.0 to 3.3 microns, and for DF-NS film

87
thicknesses varied from 3.2 to 5.3 microns. The film thicknesses for WF#25 varied from

2.2 to 5.0 microns, and for WF#60 from 2.0 to 4.4 microns as determined from SEM micro-
graphs. (Additional results on thickness versus relative position for WF#60 specimens is

given in section 3.7.1.1.).

3.4.4 Film Morphology

The grain size and film thickness micrographs also gave information on the DTF’s mor-
phologies. Also while scanning and checking the many DTF fragments on each SEM stub
to find good views for grain size and film thickness micrographs, several interesting aspects
of the different DTF’s morphologies were observed and captured on micrographs.

The (111), (110), and (100) faces dominated the deposition surfaces as is common to
CVD processes [17]. The (111) faces appeared as pyramids while the grains showing the
(110) and (100) faces generally appeared as squares on the film surfaces. Twinning, mostly
observed on the (111) facets, appear as an additional layer on one or more of the (111) pyra-
midal faces (Figure 26).

The pyramidal (111) facets dominated the deposition surfaces of the DF-Nl, DF-N2,
DF-N4, DF-NS, and the WF#25 specimens, while the (110) and/or (100) square facets
dominated the deposition surface of the WF#60 specimens. This observation is verified by
comparing the grain size photomicrographs in appendices A and B to those in appendix C.
These difference observed between deposition surfaces could not be linked to the different
deposition temperature ranges employed for each DTF (section 3.7.3).

Although most of the grain size micrographs showed good (111), (110), and (100) dia-

mond faceting on the DTF’s top surfaces, a few of the surfaces showed facets with small

88

 

Figure 26. Twinning in specimen DF-N1-4 are indicated by arrows. A 1 micron bar is rep-
resented on the photomicrograph.

89

structureless particles dispersed in-between them. The particles were on the order of 5 to
10 microns across and of arbitrary shapes. What the particles were was unclear. The parti-
cles are not likely due to the thin conductive gold coating sputtered onto the films. The con-
ductive coating, ~21 nanometer in thickness (section 2.8.2), would not be distinguishable
at the magnifications at which the photomicrograph were taken.

The DTF’s interfacial surfaces, as mentioned earlier (section 3.4.2), showed a very
distinct difference in morphologies. The interfacial surfaces for the DF-Nl, DF-N2, DF-
N4, and DF-NS films were relatively smoother with small depressions and lumps. The
depressions appeared as darker spots on the surfaces while the lumps appeared as whiter
spots. For the WF#25 and WF#60 films, the interfacial surfaces frequently showed distinct
grain boundaries. Perhaps the different pre-deposition treatments employed in specimen
preparations were responsible for this observed difference.

Films from WF#25 and WF#60 which showed grain boundaries on most of their
interfacial surfaces were quite different than films DF-Nl, DF-NZ, DF-N4, and DF-NS
which showed no grain boundaries on their interfacial surfaces. Film WF#25 and WF#60
were photoresist treated while DF-Nl , DF-N 2, DF-4, and DF—NS were abraded as pre-dep-
osition treatments. The difference in the definition of grain boundaries as observed on the

interfacial surface of the films may be related to the different pre-deposition treatments.

3.5 Raman Spectroscopy Results
Raman spectroscopy (section 2.6) verified that the deposited DTF’s were indeed diamond.

However, Raman spectroscopy was done only on DF-N1-6 and DF-N4-4 due to specimen

size limitation. The large protective tubes on the DF-Nl, DF-N2, DF—N4, and DF-NS spec-

Lit 1

D3171)

0.5131

111161181

 

90

imens (Figure 8 (a)) prevented focusing of the incident beam on the diamond surfaces of
the specimens. Raman spectroscopy was not performed on the specimens from sets WF#25
and WF#60. The diamond quality of the WF#25 and WF#60 films were established by the
graduate student who deposited them [25]. Also, SEM pictures of WF#25 and WF#60 spec-
imens showed distinct (111) pyramidal and (100) and (110) cubo-octrahedral diamond
facets supporting the claim that WF#25 and WF#60 were indeed of good diamond quality.

Raman spectroscopy for the specimens DF-N1-6 and DF-N4—4 showed distinct
1333.5 cm'1 peaks (Figure 27) indicative of the presence of the diamond phase. It was
assumed that all the other specimens in sets DF—Nl and DF- N4 were also of high diamond
concentrations. Sets DF—NZ and DF-NS were also assumed to be high quality diamond
films (no Raman spectroscopy was done on DF—N2 and DF-NS), based only on the faceted
surface textures of the DTF’s of sets DF-N2 and DF—NS as observed under the optical
microscope and SEM. Also, DF-N2 and DF—NS were fabricated in a manner similar to that
of sets DF-Nl and DF-N4.

Carbons’s amorphous and graphitic phases are 50 times more efficient in scattering
Raman infrared radiation than diamond [5], therefore, when the diamond 1332 cm'1 peak
dominates (Figure 27 (a) and (b)), it is probable that the diamond content is much greater
than that of the amorphous and graphitic phases [5,12]. However, the relative intensities of
the diamond peak and the non-diamond carbon (NDC) peak is not an absolute indication of
the quality of the diamond film [26]. The intensity of the NDC peak may be sensitive to the
nature of the NDC and not only the volume fiaction. Variations in crystallite size, morphol-
ogy and possibly the proportions of NDC with film thickness may also affect the relative

intensities of the peaks. A third, and likely most frequent problem that may lead to relative

i
I
l
3
5

WM ‘
‘31“. '4
mm" ’ Hi

”=04.

 

 

‘m ' I ' I v I V I v T ‘ l

 
  
 

a)

 

DF-Nt-l unsupported am fllm

 

 

 

frequency (cm")

Figure 27 (:1). Raman Spectroscopy. Raman spectra for specimen DF-Nl -6 shows the
characterilstic 1333. S cm diamond peak and relatively dominant region near graphite

1500 cm peak.

 

 

 

 

10000 - 1 - . - 1 ~ . . 1 - .
14000 : 1333.5 .
12000 - b) .
f
10000 _ DF-NHuI-uppomddmnun .,
i . .
0000 - .
P 4
0000 _ 1
i 1
4000 i . - - . . . . . . . . .
400 000 000 . 1000 1200 1400 1000
frequency tom")

Figure 27 (b). Raman curve for specimen DF—N4-4. The characteristic 1333. 5 cm dia-
mond peak and a dominant region near the characteristic 1500 cm 1graphite peak 15

observed.

 

92
peak intensities being an inappropriate indicator of a CVD film’s diamond quality is that

the distribution of the NDC will also affect the intensity of the NDC peak.

As mentioned earlier, the unexpected reaction of some of the DTF’s to the KOH etch
may have been due to this substantial amount of non-diamond phase carbon in the film. The
1530 cm'1 peaks seen in both Figures 27 (a) and Figure 27 (b) may indicate the presence of
a substantial graphitic phase. Although diamond is not expected to react to any chemical
below ~300 °C, in some cases the KOH etch, at 60 °C, penetrated the DTF’s and attacked
the underlying Si substrates, often causing the DTF to fracture and curl away from the Si
substrates. This effect could only happened if a non-diamond phase is present in a substan-
tial quantity in the film and is attacked by the KOH etch. The etch would then penetrate the
film by etching through the non-diamond phases and so get to attack the Si wafer beneath.
However, this assumption seems unlikely since carbon is known to resist KOH etch (5 to

90 percent by weight) even at 160 °C [32].

3.6 Theoretical Modeling of the DTF Diaphragms
The DTF diaphragms formed by etching holes through the Si wafer were generally circular
diaphragms supported by a silicon rim. These circular diaphragms were burst by uniformly
loading them with nitrogen gas pressure to failure. The DTF diaphragms were therefore
modeled as circular plates clamped at the edges and uniformly loaded in a transverse direc-
tion.

The DTF specimens’ deformations and stress states were first modeled by a linear
plate theory and then by a nonlinear plate theory. The pressure versus normalized deflection

for the DTF diaphragms were also analyzed by an empirically derived equation. Diaphragm

he

1
b
11:31

1018 c

 

93

deflections were normalized by the thicknesses of the corresponding DTF. The actual
deflections during loading were not measured. The normalized ratio, deflection over thick-
ness, was used as a convenient way to compare the information collected on the different

films.

3.6.1 'l‘imoshenko [Linearized Model]

A classical plate theory model by Timoshenko [27] was used as a model. The circular DTF
diaphragm subjected to gas pressure was modeled as a uniform and transversely loaded cir-
cular plate that is clamped along it’s edge. The burst pressure (transverse load) is related to

material properties and diaphragm geometry by the equation

P: Et4 W0 (1)
3(1 - v2) 04 [T]

where P is burst pressure, E is Young’s modulus, v is Poisson’s ratio, t is the film thickness,
dis the diaphragm (film) radius, and W0 is the maximum deflection of the film. The maxi-

mum stresses developed at the diaphragm’s boundary is given by the equation [27].

o=3 [P02] (2)

The values determined for maximum stresses at the boundary were on the order of 10'6 of
the 1000 GPa Young’s modulus for polycrystalline diamond, or ~l MPa.
The values for maximum deflection however, were unacceptable. Maximum deflec-

tions of nearly 1cm were calculated for the center of diaphragms that were ~0.5 cm in

f; .
U 1

(i117;

lliiig 1

‘L

 

94

diameter. Ceramics are not expected to deform in this manner, making the classical linear
plate model inappropriate.

Trmoshenko’s classical plate model [27] was inadequate since it is limited to cases
in which only small transverse displacements occur. Classical plate theory assumes deflec-
tions on the order of only a fraction (less than 1/3) of the thickness of the plate, so that terms
involving the slope of the deflected plate’s surface can be ignored [27]. Transverse dis-
placements (displacements normal to the plane of the plate) that are on the order of the plate
thickness, or greater, are too large to be treated by classical plate theory.

The DTF diaphragms in this study had diameters ~103 greater than their thicknesses.
The geometry of these specimens thus suggests that the maximum transverse deflection
may be greater than the assumed range in classical plate theory. Indeed, calculations done
based on Chia’s [28] nonlinear model indicated that the central deflections were 10 to 50
times greater than the plate thickness. Additionally, the measured deflections for CVD dia-
mond diaphragms having similar thicknesses and diameters, but under point nanoindenta—
tion loading (rather than uniform loading), show maximum central deflections of more than
six times the plate thickness prior to diaphragm failure [24]. The point loading was done on
a nanoindentor. Mth deflections possibly on the order of several times the thickness, a non-
linear model accommodating larger deflections seemed necessary. A non-linear model by

Chia [28] was thus applied.

3.6.2 Non-linear Model
A geometrically non-linear plate model by Chia of a uniformly loaded circular plate with a

rigidly clamped edge was then used to analyze the deformations and stresses of the DTF

01

M

 

95

diaphragms [28]. Chia’s model accommodates deflections that are greater than the plate
(diaphragm) thickness. This assumption was more appropriate for our diaphragms. Com-
parable deflection analysis of similar diamond diaphragms under point loading showed
deflections greater than six times the plate thickness (section 3.6.1 and [24]).

Chia’s model relates uniformly applied pressure to material constants and film

 

geometry by the equation
P = 1314 [w_0 + (1 + v2)(l73 - 73v) (WC 3 (3)
3(1 - v7)dz t ] 360 L t ]

where P is burst pressure, E is Young’s modulus, v is Poisson’s ratio, tis the film thickness,
d is the diaphragm radius, and W0 is the maximum deflection of the film. The extreme fiber

bending stress is given by the equation

0: 5:2 4 W0 (1-v)(83-43v) w0 3 (4)
<17 [1v]l:['t—] + 360 [1]

 

 

For this study, the pressure P will be taken as the diaphragm burst pressure. The stresses
calculated from the data using equation (3) and (4) were much greater than those deter-
mined by the linear model (section 3.6.1). The stresses calculated are on the order of 10‘2
of the 1000 GPa Young’s modulus of diamond. These large stress values, on the order of
tens of Gigapascals, correspond to a maximum strain of a few percent assuming the simple

Hookean constitutive relationship of

96

where O' is stress, E is Young’s modulus, and 8 is strain.

Film deflections of ~10 to 50 times the plate thickness were calculated from Chia’s
geometrically non-linear model [28] which seemed intuitively more accurate than those
determined by the linear model [27]. Central deflections of 10 to 50 times the diaphragm
thickness seemed more credible for a diaphragm of this geometry than deflections of
~10,000 times the thickness as calculated from the linear model [27].

Both Timoshenko [27] and Chia [28] related burst pressure, radius, Young’s modulus
and thickness to the deflection normalized by film thickness. Equations (1) and (3) are iden-
tical except for the second term in (3), which indicates that as t/d becomes very small, the
equations become identical. Banerjee [29] also shows that equation (3) characterizes the

large deflections for uniformly loaded thin plate very well.

3.6.3 Analysis of WF#60 Data Points
In both equation (1) and (3) the burst pressure P is a function of the film thickness divided
by the diaphragm diameter taken to the fourth power, (t/d)4. Therefore the data set WF#60

was fitted to a curve of the form

P=E_L]4 c1 +c2. (5)
(1

Data set WF#60 was used because it was large compared to the other data sets, containing
21 data points. The next largest data set was for WF#25 which had 7 data points. A least-
squares fit to equation (5) determined by a least—squares method gave a correlation coeffi-

cient of 0.85 with Cl equaling 1.67 x 1016 and c2 equaling 1.13 x 105. A correlation coeffi-

97

 

 

 

 

 

4E+05 i ‘ l 1 J 1 l 1 1
VVFhKMJe4» 4‘
’11? 3905-— +—
9;
(D
5 2E+05~ -
(D
3 _
15‘. 1E+05 —-
05+00.

' l ' l ' I i 1
0E+00 4E-12 8E-12 113-11 2E-11
[t/d]**4 - Dimensionless '

Figure 28. Plot of P versus [1/014 for 21 WF#60 diamond films burst in this study, where P
is the burst pressure , t is film thickness, and d is film diameter.

98

cient of 0.85 indicates no correlation, and no correlation can be seen from the fit of the
data to the curve (Figure 28).

Plots of burst pressure versus the fourth power of the thickness-diameter ratio also
were also made for sets DF-Nl, DF—N2, DF-N4, DF—NS, and WF#25 (Appendix E). The
data points were very scattered, even more so than the data for set WF#60, therefore a

least—squares fit to equation (5) was not attempted for these data sets.

3.6.4 Empirical Equation
Burst pressures were also plotted (Figure 29)as a function of the non-dimensional parame-

ter R, defined as

R = [t/d] 1’2 (6)

where t is the thickness of the film and dis the diameter of the DTF diaphragm.

The burst pressure data was fit to the equation
P = clR + c2 (7)

where C] and c2 are fitted constants and P is the burst pressure of the diaphragms. A
least-square fit of the 21 points of data set WF#60 to equation (7) gave a correlation coef-
ficient of 0.9 with Cl equaling 69.9 and c2 equaling -41.9. The removal of WF#60-5,
WF#60-8, WF#60—18, and WF#60-21, the four most scattered points, gave a correlation of

0.953 with el equaling 8.65 x 105 and C2 equaling -6.73 x 105.

 

99

 

Pressure (Pa)
N
m
+
O
01
l

 

OE+OO ' I ' l l ' l '
0.7 0.8 0.9 1.0 1.1 1.2
[t/d]**0.5 - Dimensionless '

 

l

 

 

Figure 29. Plot of P versus [t/d]l/Z for 21 WF#60 diamond films burst in this study, where
P is the burst pressure , t is film thickness, and d is film diameter.

100
The four data points WF#60-5, WF#60-8, WF#60-18, and WF#60-21, removed

from set WF#60 for the best fit exhibited no peculiarities to which their deviation from the
curve fitting might be attributed. However, it is likely that the diameter measurements for
specimens WF#60-5 and WF#60-21 may have both been inaccurate. The boundaries of
diaphragms for WF#60-5 and WF#60-21 were difficult to determine possibly due to pin-
hole effects (section 3.5). The pinhole effect may occurs because the etch permeates the
DTF through small holes and attacks the Si beneath. The diaphragm diameters were not
clearly defined because of the pinhole effect which may have cause these data points to be
scattered.

The burst pressure data also was fit to an equation of the form

P = Clxc2 + C3 (8)

where X is the ratio (t/d). Equation (7) and equation (8) fit the data equally well as deter-
mined from the nearly identical correlation coefficients obtained from regression analysis
for the two equations. The correlation was 0.954 and the three free parameters c1, c2, and
C3 were 501,000, 0.430, and 313,000, respectively. Since equation (7) involves only two
free parameters and equation (8) involves three free parameters, the curve fit was dis—
played only in terms of equation (7) (Figure 29).

In both equation (1) and (3), P is a function of (t/d)4,while in the empirical equation
(7) P is a function of (t/d)1/2. The empirical equation (7) better represents the behavior of
the limited data range collected for WF#60. Further comparison between theory and

experiment would be facilitated by measurements of the central plate deflections w0 dur-

101
ing loading.

3.6.5 Analysis of the Total Accumulated Data Points

The total data set including all diamond diaphragms burst in this study (46 data points rep-
resenting specimens from six different films) was fit to equation (7) relating burst pressure
to DTF diaphragm geometry. The 7 most scattered points of the 46 points data set were
taken out in order to increase the correlation coefficient. The 7 data points removed from
the set were DF-N2-5, DF-N2-7, DF-N4-8, DF-N5-8, WF#25-7, WF#25-10, and WF#60-
7. The constants c1 and c2 were 60.94 and -3l.67, respectively, with a correlation coeffi-
cient of 0.674. The correlation of 0.674 is insignificant, but the values for CI and c2 are
interesting.

The values for CI and c2 obtained from the entire data set are quite similar to those
found for the best fit of data set WF#60 to equation (7). The values of Cl and oz for the
best fit of set WF#60 compared to Cl and 02 for the best fit of the entire set is for CI, 62.17
compared to 60.94, and for c2, -32.38 compared to -35. 14, indicating that the entire data
set follows a curve similar to that of the WF#60 data set.

The correlation coefficient of 0.674 for the entire data set compared to 0.975 for the
WF#60 data set may be due to several factors. First, the entire set is taken from 6 different
DTF’s. The different DTF’s were deposited under different conditions which produces
films with difl’erent physical properties, e.g. impurity content, film thickness, grain size,
etc., and thus film’s with different physical properties. The scattering may be partially due
to these physical differences in the deposited DTF’s. The data set WF#60 however, was

taken from one DTF which could possibly contribute to the correlation coefficient of

102
0.975. Secondly, specimens DF-Nl, DF—N2, DF-N4, and DF-NS were not as circular as

the diaphragms for the specimens of WF#60. The difference in circularity was due mainly
to the different methods used in defining the area of the Si wafer exposed to the KOH etch
(section 3.3.2 of this thesis). Thirdly, a possible cause for the scatter in the data is the
shape of the Si wafer supporting the DTF diaphragms. The Si wafer tapers down to a very
thin edge [24] where it supports (modeled as rigid clamped edge) the DTF diaphragm.
Failure may occur only in the DTF at the edge of the diaphragm, or through the DTF and a
thin part of the tapered Si wafer. If failure occurs through both the DTF and part of the
tapered Si wafer, the model would not represent the boundary conditions. This misrepre-

sented boundary conditions would cause scattering.

3.7 Other Observed 'h'ends

The different sets of DTF’s were deposited under different conditions, such as deposition
time, deposition temperature, and precusor gas concentrations. Thus the physical proper-
ties of each DTF set and, most likely their mechanical behaviors, differed from set to set.
Trends among the different sets might be established based on deposition conditions and
the effects the conditions have on their respective films.

Sets DF-Nl, DF—N2, DF-N4, and DF-NS were deposited under differing microwave
powers, differing methane concentrations, and for different deposition times. Increasing
the power of the microwave generator, with all other parameters held constant, is expected
to yield a high film growth rate[10]. Increasing the methane concentration, with all other
parameters held constant, is expected to increase the likelihood of producing a graphitic

carbon film [14], and also to increase the film growth rate [12,30]. Increasing the deposi-

I 03

tion time, with all other parameters held constant, is expected to yield thicker diamond
films simply, because after an initially high rate the film thickness growth rate slows to
being pretty much constant [18]. Longer deposition times is also expected to yield films
with larger facets (grain sizes), because grain size increases with film thickness [12].
These trends can be investigated in the data collected from sets DF-Nl , DF-N2, DF-N4,
and DF-N5.

The specimens from sets DF-N l, DF—NZ, DF-N4, and DF-NS were investigated for
their burst pressures, diaphragm diameters, film thicknesses, and grain sizes. From two to
six DTF specimens were tested from each of these four sets. The size of these sets was
deemed too small for any substantial inferences to be made, therefore only the trends

observed were addressed.

3.7.1 Thickness Variation

The film thicknesses for sets DF-N 1, DF—N2, DF-N4, and DF-NS varied from 1.9 to 4.7
microns (Table 2). The film thickness of set WF#25 varied from 2.2 to 5.0 microns (Table
3). The film thickness for DTF WF#60 varied from 2.0 to 4.4 microns (Table 4).

Sets DF-Nl, DF-N2, and DF—NS differed only in their deposition times. The deposi-
tion times for DF—Nl, DF—NZ, and DF—NS were 6, 8, and 10 hours respectively. The aver-
age film thicknesses for these three sets were 2.2 microns for DF—Nl, 2.5 microns for DF-
N2, and 4.3 microns for DF—N5. Increasing deposition time results in an increasing film
thickness. The time increases were in 2 hour increments, but a corresponding incremental
increase in film thickness was not observed. If thickness increases proportionally with

time [18], an incremental increase in film thickness would be expected for the incremental

104

increase in deposition time. However, the set sizes were 2 data points for set DF-Nl, 5
data points for set DF—N2, and 6 data points for set DF-NS, and the DTF’s thickness distri-
bution over the substrate is known to vary considerably from thicker at the center of the
substrate to thinner at the edges [21] (also section 3.7.1.1). Perhaps the small data sets,
especially DF—Nl, made them inadequate in exhibiting incremental increases between

films with expected thickness variations.

3.7.1.1 Thickness versus Wafer Position

The relative positions of the 26 specimens taken from WF#60 were recorded and com-
pared to corresponding film thicknesses. The pattern noticed is one of decreasing thick-
ness with increasing of the radial distance from the center of the deposition plasma (Figure
30) and [24]. The thickness variation might be attributed to the temperature variation that

also decreases from the center to the edges [21].

3.7.2 Grain Size Variation

The burst pressure versus square root of grain size was plotted for the six different sets of
specimens (Appendix F). No pattern was evident from the plots. The data seemed very
scattered and without any correlation between burst pressure and grain size for this range
of grain sizes. It is possible that the lack of correlation between burst pressure and grain
size is due to the narrow range of grain sizes tested. The grain sizes from sets DF—N 1, DF-
N2, DF-N4, and DF-NS ranged from 0.33 to 1.12 microns (Table 2). The grain sizes of the
DTF’s taken from WF#25 varied from 0.47 to 0.99 microns (Table 3), and for WF#60 the

grain size varied from 0.54 to 1.26 microns (Table 4).

105

#20
‘< >3}

 

 

 

 

 

 

 

 

(#13)
2.5
2-5 (#10) 3.5 3.4 3.3 (#22)
2.5 3.6 2.8 3.2 3.3 \ 3.0 (#24) 2.0
[2.2 3.5 2.5 4.4 i 3.2 3,3 2.5 2.6

 

 

 

 

Figure 30. Thickness versus Position for WF#60. No thickness values were obtained for
specimens #10, #13, #20, #22, and #24. A slight pattern of thickness decreasing with
increasing radial distance from the bottom center is observed. This pattern might be due to
a higher substrate temperature corresponding to a thicker regions of the deposited DTF

[21].

106

In cases where the initial flaw is limited by the grain size, and scales with grain size,
and in which brittle behavior is observed. strength (considered to be burst pressure) should

vary with grain size as [31],

of = K-d‘l/z.

Due to the relatively narrow grain size range of ~0.9 micron to ~12 microns, the
corresponding square root of grain size range is 0.949 um”2 to 1.09 umm. The square
root range varies by only a ~15 percent of its lowest value, therefore the expected effect on
burst pressure would only be a 15 percent change.

The range of the grain size data collected in this study may have been too narrow to
make a grain size versus burst pressure investigation. However, in studies involving a
wider range of grain sizes, microstructural effects may be very significant and should be
investigated.

Grain size tends to increase as DTF thickness increases. For sets of DTF’s having an
increasing average thicknesses (DF-Nl , DF—N2, and DF—NS) the associated average grain
sizes are 1.00, 0.65, and 0.79 microns, respectively. The average grain size of 1.00
microns for the DF-Nl set does not conform to the expected trend. However, only two
specimens from set DF—Nl were tested, and not the average grain size over the entire DTF.

When it was possible, the grain size on the interfacial surface of the DTF’s were deter-
mined. Grain boundaries on the interfacial surface of the DTF’s were evident only for some
WF#25 and WF#60 specimens. Two’interfacial grain size determinations were make from

the seven WF#25 films tested and nine from the twenty one WF#60 films tested. Interfacial

I 07

grain size determinations were limited to SEM photomicrographs which showed well
defined interfacial grains which is why only a total of 11 interfacial grain size
determinations were made although several other photomicrographs from WF#25 and
WF#60 showed a similar structure on their surfaces. Interfacial grain sizes are presented in
Table 3 for the WF#25 films and Table 4 for the WF#60 films '

The grain sizes of the interfacial and deposition surfaces of the same film were com-
pared to determine if indeed grain size in polycrystalline diamond films increased with film
thickness. If so, the grain size of the deposition surface should be larger than the grain size
of the interfacial surface for the same film. Two of the nine interfacial grain sizes from set
WF#60 were larger than their corresponding deposition surface grain sizes. This would
support the what is expected. Two of the two interfacial grain sizes from set WF#25 were
larger than their corresponding deposition surface grain sizes. This is not as would be

expected. Perhaps again the sample size of two is just too small.

3.7.3 'Ibmperature Variation

Sets DF-Nl, and DF—N4 had the same deposition conditions except for the microwave
power used in their deposition. The microwave power used for depositing DF-N l was 900
watts, and for DF-N4 it was 700 watts. The associated specimen surface temperatures
were ~800 “C at 900 watts, and ~760 °C at 700 watts (section 2.2.2). The Raman spectros-
copy for specimens DF-Nl and DF—N4 (Figure 26) did not show differences in the relative
intensities of the diamond peaks that might indicate the degree of the diamond content.
Higher temperatures typically produce diamonds of higher purity [8,11,18].

As mentioned earlier (section 3.7.1.1) the film thickness versus film position noted

108

across the Si wafer is likely attributable to the temperature variation across the wafer during
deposition.

Another observation made from the photomicrographs was the difference that pyra-
midal (111) faces dominated the DF-Nl, DF-N2, DF-N4, DF—N5 and WF#25 specimens’
deposition surfaces, but square (110) and/or (100) faces dominated the WF#60 specimens’
deposition surfaces. The different faces are known to dominate at different temperatures.
The (100) faces usually dominates at the lower temperatures than the (111) faces which
usually dominate between 750 °C and 1,000 °C [19].

Sets DF-N] , DF-N2, and DF-NS were deposited at approximately 774 °C, set DF-N4
at approximately 766 °C, WF#25 at approximately 850 °C, and WF#60 at approximately
776 °C. All sets were deposited between 750 °C and 1,000 °C and are expected to show dep-
osition surfaces dominated by pyramidal (111) faces. However, the WF#60 set shows a

square (110) and/or (100) dominated surface.

4. Conclusions

A more reproducible method for abrasion would aid in standardizing the deposition proce-
dure. Abrasion is done to increase the nucleation site density on the substrate. Nucleation
site density strongly affects the adhesion of the DTF to its substrate [21] and thus is an
important parameter to control. Although SEM surface replication technique may be used
to determine nucleation site density . a method in which the site density can be prescribed
may still be very desirable necessary. SEM testing of Si wafers that have been submerged
for different periods of times in an ultasonic bath of diamond powder suspension might be
tried to see if alteration in an ultrasonic bath of diamond suspension would ofter a method
of producing a prescribed nucleation site density.

The different methods of nucleation site generation, abrasion and photoresist treat-
ments, show a distinguishable difference in the morphology of the interfacial surfaces of
the DTF’s. No adhesion tests were done on these different DTF coatings but it would be
interesting to see if differences in adhesion were associated with this morphological differ—
ence.

The modelling of the diaphragms as uniformly loaded plates with clamped edges is
acceptable as a rough estimate for the particular DTF specimen included in this study. The
Si rims supporting the diamond diaphragms for the specimens fabricated by using small
tubes to define the holes etched through the Si wafers were not observed under the SEM.
Therefore nothing is know of the shape and thickness of the Si rims where they supported
the diamond diaphragms. The edges of the diamond diaphragms fabricated by dimplin g
were supported by a very thin Si rim [24]. The support of the edges of the diaphragms might

not satisfy the assumed boundary conditions of the model. Changes must be made in either

109

110

the fabrication of the diaphragms or the modelling of the diaphragms' boundary conditions.

APPENDICES

III

APPENDIX A

Micrographs of some specimens from sets DF-Nl, DF-N2, DF-N4, and DF-NS. The

micrographs are presented in the order of grain sizes, thicknesses, and other observations.

.112

 

 

 
 

.3329: 2 b3 c828 fl

< doecmu ozfimfim 330E on on GEE m Z n5 .8 new BEE o .3806
33 as 5928 .83 EE .3: 2882: 33080 2m Boom mm Emce— aobub _ < .cafiwobfi: 05 mo .6250 05 3 :08 m
SEE 0 8m ha 528% u 5% owes . 0 3a m cosaoao o . 0.5”.

 

 

.83 580 ._ .< 5:53:

 

 

68865. mm Hm
68865 8 Hanson cocoa. Wm < duo 8c 88:: m < .383 82:8 HE? @838 on 8 88%: a:
2 55:: Me. oz we 85 58w < 888 2 H C .3. B3580 8E. .nmfimohfi: 85 Eot. 8.558% mm? 885:: mm o w

0 00m 5m $0550an 0 Qua—w 0N5 gum

 

.35 :80 .H .< 5.05.344.

 

    

.U8m
-65. 2 :58vo .588 5m < .885; 8: 8e.EEou 88$ ASA
398mg; 6-975 no 8&8; .5983 o . 2:».

 

 

«32 309mm

114

 

 

    

68865 2 a
bow 58:: m; < .8808 8: us 883 AA: 3 was Abe: A A 2
.825 2: {N 2- .5 E

 

V .
EC 3335;:

I...

 

.85 gene .8 .< 55%?

115

 

  

68m
65. 3 Hanson cocoa: o.m < .nmfimobg £5 an 35880
33 5.8%: $6 we aim 53w < .838 AAZV \3 335:5

. . 85w.

 

.65 580 .8 .< 55...?

 

 

    

116

8838888 8 885 88888 8 .888 888880888 85 888 88888
-8888 33 88.0 80 8888886 < 88888888 05 888588 8528»
8 880888 :88: 8 oz 618178-15 80 “Who of. .588 988%.

 

8828888888 88 s

808688 8 < .83 888 mo omen 888 880 new 8 088888 8888888880
.883 AS: 88888883 3 88888.88 8 88.8888. 88 8F.

‘ sovlmlilx :4 :m

 

80888880882. .88 .< 88.8.8593.

 

 

68868 8 8
:83: H < .2825 Son 03 32285 8: @508: 8:325

3:638 05 ER Satan :oEmomon 2: so 888 REES»
E“ 88 3888888 £3 :08.

 

mmomo :9; momm

117

 

 

.8382

8 53.88 5888 o < .8588: Wm 8 :mfiwobg 8.: EB
3588888 32885 2:. .owvo 05 swag: E885 8 8
.02: 3:5: 8 oz NAZLD «o 38> ow o o

    

 

E 3 S . m

$885885. d .< 86:22?

 

 

 

6886.: m

85.83 2888 me < 85830 08m 8 8&8; 8828888 2: c
85:: 385 .naflwobub 86 Set BEEBBU 8;» 858.
SN mo wmocwom: < .vfizim mo 32> @o 2.? .N~< 85m.

.33
BE 8 53-828 5888 m < .nmflwobfic 85 Sod 88me m.
u 8 058880 82> - - 0 m8 8 8 u . v.5».

   

  

 

:2 323m :2 $993

118

.Gum 75:05:50 Hip o:

8.19.

 

83356:: d .< 83.0%:

 

 

.8389: 8 883 88088888 08 < .03. 08 . 8888888888 8 8.3 880888: 08

8 EH9 SE 2888 9588888888 2: 80 onon 888888 owuu 2:. .888. Améd 888888 58888 AEOMV 88888883 888888888088 88 888 808825. 8
-8288 8888888 288 8 88883 8m 8:. m- v2. n5 58888.0on wEfio 8.88?» o8m888m 803 E88 M88856 80 88:88 08.88 8 $2.292: o:

as. momv sow; 38.0.8

 

3088883890 .8050 .E .< 86:258..

120

 

88888888 8 82-8088 858828 m8 < .08 0888 08m 83 E80
.8888 8888888888888 80 288283. 0388888 8088 on 888888 8835 88.885
.888 88 :0 888m Ba 8088 883 8888886 888888888me Qomx

 

m n. I .8 m 3 m m _H_ H

 

Dom

mo

mmomo so .8

 

888888888. 8 883 8588888 8 < 88888888 88888888 38. a 88883 888: m

 

momv

880888380880 82880 .5 .< 8:85.883

121

 

38888888 88 883.58V E: on. < .83.“ 858883 8888888 .

Eow 05 88958888 388888088 .8808— on 9 9888288888 $.38 o8=w8

83 8888885 0888 880 8088 880288 888888; 088.8. .w-mz-n8.n8 Mo 88888
88088883 0

menu 8.

 

308888833880 88880 .5 .< 8888:8288:

122

APPENDIX B

Micrographs of some specimens from WF#25. The micrographs are presented in the

order of grains sizes, thicknesses, diaphragm diameters, and interfacial grains.

 

 

      

885-8088 880808888 w.m < .UoEEBBU mm? 880808888 0.0 80 on8m 8888
< .33va0 vs $808880 owbx 33 < .oowtsm 05 8888888886 moon
:83 28% 569$? 80 8888.83 8808888880 v . «.8889

.3888988 m8 8.3-8088 880588.88 w.m 8886288888808
33 £80888 mud 80 B88 88888.8w < .388 88 88V 83 BEEEoo m
888 888m 2: $6.983? 83 8588; 880888885 u . 9888M.

 
 

 

 

 

123

 

8% 888885 .8 .88 8888:8288?

 

 

      

68880888888 m
883-888 880808888 m < 8588888888088 33 mod 80 o~8m :8m8w < .88
88888588 888m8w 0888 :8 888m 288 88088 8888885 .coccuw 8803 8088 M
£88 888% 08: 3.8 £888.83 80 8888885 888888880 0 P888?

   

688888888 .88 88-8088 880588: w.m < .8828.
-8828 83 no.0 8o o~8m :8m8w < $98883. 0888 888888888088 38
888: 2: 88.38883 80 83885 8808888880 u . 9888M.

 

124

 

.8888”. 88.820 .8 .88 8888:8283.

125

 

 

     

8888888888 8 885-888 880808888 w.m < .8888888880
33 880808888 53 80 BE 8888a < .888me 05 8888888088 3088
88888 2:. 68-3th 80 888L885 8808888880 v . 988a.

 

.35 8.88:0 .8 .88 88.88.828.848

 

.8888
888888 m8 883-8088 880808888 Wm < 6088888888088 .8883 md 80 305808

< 8888888838 W883 880888888 8888 0 Z 8882885 8 owwu 0888 $0888
088880888; 8888888888885 5.2888885 80 32> um? 08; .8: 0.88%.

 

126

.3”; 8.388898%“;

 

 

      

     

8888888888 88 883-8088 8888888
«18828888888088 3? £80588: m 80 835808888 < 8808888888. 33 8808
8888 8888 on 888888 888M858 08888 0888 w888>>o88m 888888 8088888888 8888880
33 com m uwwo Bit. 8 388883 80 32> am e o «.888?

.: 8memou

8885808888. .88 .88 8888888288888

 

 

.8888: m8 8888808888 883 28.8 .n8-8-Q 0888 8888880

.8888 8 888882888 88 8888-8- 388830 88 9888888888 8m. 8m 0888 8888888 08 8880 0888 w8883o88a 8808888888 08 2888 8888888888 .8

-8888 888088 8828888880 8888888886880 85% 2F 3- at; 8808888888 288 880 8808:8880 2m 880888888 Eb 88888882 8-888.883 882888888
0 88888 888 858808 M8 Ema 88888 . P888».

 

.m8o80888m8n8 Emma-8888889 .888 .m 888888828838

 

 

.EE m8 888883888888 :3 BF .EmeE:@ 0888 888a 8m :0 n88- .
288 58 .08 28888 88888 0883888 088888 88885 888888 3880080888 288 w:. 8 388828808 88888 088.8. 6888385 8m “8888 Soc 888888 8 ”Sb 0
8888880883 335 $8880 0888 :0 :08 8 883m 880.8. 0888-. 1888 mat 888888 888888 @8988 .8 8888880383 88.0 088.8. 1N8 3th 88888888388

 

8808088889 8888388888885 .888 .88 x_u:u8888<

 

 

880888 w.m 80 88888-888 < .8888m8wo8o8E £888 88.8088 88088888888088 3
88808888888 owd 80 388 888888» < .0888888888888888888888 08a madam 88888888
-8888: 08: 68.38883 80 888888;. 888898888888 0 . 0.88%.

 

129

 

38888.88 8.

 

 

       
 

88088888888 88 88080.
n 80 883-8088 < 88888888808088: 8888 888088 8588888882088 33 £898.
88.0 80 B88 888888w < .o8nmnm8swc88m888 8888888888888 0888 £88m 88888888

-6888 RF .88-nm#-8>> 80 888 w 8888 8888 o . P888».

 

.2820 8883882888 .2 .88 888.588.8888

130

APPENDIX C

Micrographs of some specimens from WF#60. The micrographs are presented in the
order grain sizes, thicknesses, diaphragm diameters, interfacial grains, and other observa-

tions.

 

 

      

880883888888 88 88888.8

280898888 m < 888888880588: $888 E8888 8888888808088 33 £580.
38 80 B88 888888-88. < 888888.880 So: 80 S8 88 288 80 88888 8888888
08:- .o-ac#n8>PcoE8oo88m8o 88888888 88088888888 0 9888».

86888888888 m8 885-80
280808888 w.m < .8888m8wo8o8888 A88888 E088 8828888888088 283 282,88

v8.8 8o 38 88888888 < 8888888880 8088 80 88 88 D888 80 88888-8 2888888
088.8. .m-88c#n8>> 8828-880on 80 888888;. 8808820888888 Etna-u 08889

   

 

 

 

 

131

 

.mon8m 888888880 .8 .U 8888888288848.

 

      

880883888888 .88 883-88888 £8080.
Wm < .8888m8wo8o8888 88888 E088 8588888880828 33 $808088: 38

8o 8.8m 888% < 88888380 3088 80 88 88 @888 80 888888 0888888888 28
«88-398883 88288888? 80 8888.83 8808888888888 2F .85 P888?

 

 

 

132

 

 

     

88883888888 8 885-8088 £8089
888m < .8888a8wo8u8888 .8888 888088 88888888885828 £3 20808888 N8
80 88.0. 888888 < 288888888 8088 80 88 88 @888 80 $8888. 88888888 088

 

1888 3088885 888888888888. 80 888888;- 88o888m088o o 0.888?

 

 

 

.mon8m 888880 .8 .0 88888888888888..

 

8888888888 88 88888-888 880808888 0 < 880580888
2.88 9.68088 $58850 0888 888883.885 8888088888888 @888 80 388880885 8

88888800 888818808088 88888 E8888 8808888888828 33 8880808888 0. m 8
858888on 0 32> ow o o . c.8889

 

880888888888 8 8888-8088 880808888 m < 885808888 288 $05
885830 0888 288889880 8888088888888 0888 80 8888888888 8988888880
.8888888wo8o8888 8888 888088 8588888888088 33 8880808888 N. m 8o 30

8808888888.“. 88 32> am e o . 88.88%. mm... 0

:888888888

8880888808888- .88 .U 888888828884...

 

134

 

88088888888 88 88888-88888 880808888 m < 8880888880 0888 .30

208.888 288 30828 288888888 8888088888888 288 80 8888808886 8888888880
8888:880on o E88880

 

:_H_88m.mmmn_m_ 8.888

 

 

      

88088888888 .88 8.3.8088 8808888888 m < 8388882888 288 8880
8.8808280 0888 8888888880 8888088888888 @888 80 3888888888 888888888880
.8888m8wo.8o8888 88888 888088 88288888808088 £3 8880808888 v m 80 mm»
88088888888 80 32> owno “888-8- .8-U c.8889

 

8335808888. .88 .U 888888808885.

I35

 

 

 

      

.858: 38808838 885 2:. .EwaEmmE 288 888.888 Amos 888883 8
co ”Ch 288 888 60888 288 28828 .008: 8885 0380 388880888 on
$858888; 088m wuwco 82880 288 no m8 Row :8. 28.8. .3603
5:880on 80 888888 REESEME EwaEmmfi o .

SE m8 £883,888“: .8880 08F 85% 888 898888888880 @888 888888 Ame
V8588 8m 85 ”ED 288 8.8 .288: 288 288%: .288 888mm 03888 e?
080E 288 80 0083880 88qu .88 @8880 88m tor—L 888;? .0- 00¢

 

 

8880:8885 88883888888889 .5 .U x8=:~an<

 

      

.EE 8 $883888“: 30 08:. 8889888898888 288 0:88 .8888.“ 888880

8m :0 ”Eb 0:8 288 28888 288 20838 .8888 88888.8 883888 9888888
288 28850888; 8%» 88:0 or: so 88m 8.808. 28.8. .mm-©0#n8
8808885on 80 $850 88888586 888M888 88888888888880 2: .N8U c.8889

 

 

I36

 

.EE 8 38588882 8880 08:. .EmemmE 288 088 83888 8888888
8m 88o nab 288 288 28888 288 005888 .3888 88888.8 888888 388088
0888 $855888; 3%» 8885 @888 :o w8 83m 880.8. 28% 08-008888
cuE8ooam 8o 888888 88888888 858888 EmSFaEU 2: .883 P8889

 

 

52088889 8883:8985 .888 .0 88888882883.

 

       

8888888888 .88 888.80

85588: w.m < 098838088: 2888 E88 0288888888888 33 £858
mod 80 38¢. 88888888 < 88280088 8895 0888 288885 888880380888 28
0-308383 8888;85qu E98883. «588 88888 e . 288w.

 

 

 

E, 8.8 m” . 8..

I37

 

 

 

       

88888888888 .88 885.80
888088 m < .8888m8wo8o8888 £888 E808 8888888880588 £83 £85.
088.8 .80 ou8m 888.8% < 88280088 8803 @888 £82m 88886888888888 28
8.30%? 882888898 8o 8888.83. 8888 8888 u . 288w.

 

£88.80 8888088888888 .>8 .0 8888888283..

I38

 

 

       

08880888888 m8 80.80

880808888 w.m < 88888883588: m8£ E808 8588888888088 £83 28080.
80.8 80 38m :8888w < 888888088 8895 288 madam 8888803888888 28

. 818-308.88n8 3 8.088888% 80 8888.83. 88828888888 o P888».

.

 
 

 

 

 

68880888888 m8 88.8
880588: m < .8888888wo8u8E E888 E808 882888888828 .8883 £588
mw. o 80 38m 888858 < 880880088 8803 888 £88888.“ 8888.803888888 28

 

$888880 8888088888888 .>8 .0 88888882883.

 

88888080888 m8 8:05.

Wm 80 88.800 < 88888858898888 8888888,) v.8 80885 288.8. .0N-o0#n8
8:888on 80 388883. 8885888828: 288 so 88088.80 < .8889 9:89

 

 

I39

 

    

8888888888 88 282088 88.
8o 8880-588 < 853:8 on O8 $882898 $808880 88888888888m885888 2880
882288858: 288 28888858 so: 5 88 88 228% 8-88.8

8808:80on 80 888888: 8585088088 @888 85 88080 < 8.80 Paw.

 

 

.mco8888>8800 808880 .> .U 88888882888<

I40

APPENDIX D

Data Acquisition Program. The program was written in quick basic. The data points
are read continuously during testing, and 2.33 points are saved every second. Five points
are read and the highest one is written to the created file. After the test is stopped, the high-

est value written to the data file is also written on the monitor

I41

APPENDIX D
' ' Wilt?
' this progm seam to work OK
DIM 11(8)

DD! ”(1000). GIANNIOOO)
newsman. GEN“). mu
m SEE CIOB an. 32mm, Fir)

'WC
DmeOOO)
'MC
Delay-4
Macadam-1

' ---------- Initializeaectim --------------------------------------
ant-moo 'Settheaasehddress
I‘m-0 'Setmdennberwo
11(0) -m . '
CALL cream. li0)). Ft) ' Send Initialization
IPR<>Om ’Iftherehasaprtblen...

MW '

as:mmr-mnmmmnt-'

muss:s'mp '
EDI? '
wr(ms+3).e 'Setthegainregiater

 

75

100

175

200

I42
APPENDIX D

CLS

PRJNI‘ : PRINI‘ WHO, '-");

m 10, 10: WMW9, " ")

man 12, 10: rams-121x359. " “l: 10mm 10, 10
W'mmemmy (inpei) ", Targets
auger. -VAL('Ihxgac$)

“121812.10

mmmingavalmotm.fipai. isthiaoorrect (WM 2’"; Target
ops-mysznops-"mmmmo

RUBEN”) <>'Y'mm75

Getuatflmtoaavethedutain

mm, 10

m'mnerthenauaofuteflletoaavedatain (incltxiepaths) "
1005315, 15: ERINI‘S'RJICMS‘I. " ')

11111815, 15: W", Film

mmoomzoo

mmmnmmn

$10300

mmumfflembecmaced

zoom 17, 10

ammunuapztblanqeningmetile';me1hm$
manna, 10

I43
APPENDIX D

PRJNI‘ "Press any key to enter a different filenane "
momma (mars - "'): LmP

mam; 17, 10: WWMQ, " ")

1m 19, 10: WWW9. " "l

mm 175

 

300 mmoomo

(11021

M 8, 31: m 'erxmt Prmsure'

mm

tom: 10, 33: mm -W-

m 11, 33: PRINI‘ '0 0"

1m 12, 33: PRINT “WY“

(11014

m 16, 20

mm Whiting for the pressure to reach m.“ psi": Target:
1m 22, 23: m '<< PM 'X' to halt: the program >>";
(110112: W22, 33: WW“: (1)1324

' Mttmmvoltagetogathighmagh

 

400 oos-mvs:nmlovs)-'x-moomomez
mama:
max 11, 36: mm 'm.##'; Pressure;
nm8nzgeemoomsoomsaooro4oo

 

I44
APPENDIX D

500 mam: 16, 20: PRJNI‘S'IRMGO, ~ ~)
mm, 20
PRJNI" Reoomdingdata...
seam-m
mm, mat-same), Preamre

sso mixelmoalay
room 16, 50: RM '/"3
sesame
HW»WmW-W
room 11, 36:PR1N1‘IBDG "##3##“: Pressure;
Ops-ms: IPIDBEMOPS) - 'x-moomomel
Mac

mm. (I'M-8mm). Pregame

mm 16, 50: mm "W
W-W-rl

ops-ms: nWlws) - 'x'mcmoomel

oomsso

(Invert:
MM+2LO 'Setchaxmelmnber
mm-1m1o:m1u 'Slightpame
MM+1LO "I‘riggeromvemim

mono:

I45
APPENDIX D

mc-mtam+2)ma1za

REE-12831316710311.0313 'O‘Iecksincmveraimisdcne
ink-INNER) 'Readinlavbyte
nu-1mmmzxrk1 'migrmpame
m-M(am+1) 'Readinhighbyte
ydaea-m1r16+xu/16-2048 'Crnbinedata

Volts - ydaca * (20 _/ 4096)
Pregame - Volts * 10

(1183

W10, 10

PRM'medanausaavadinatilecalled '; memes

woman, 10

ERMtBIm "mummtorW3 aeoarls.'; (mm-seam)
mm, 10

ammonium; 'mmmmfl

111121816, 10

mmrtsm: 'Avuage points/second MM '; mmmta / (m - Start'fim
1.001818, 10

mmmmmnm

PRJNI':PRINI‘:PR1NI'

I46
APPENDIX D

cross

MIC, 10

mmmmpmgranuasatqpedwiththepressum at ##4## psi."; Pres
mm, 10

anr ”namgetpreammuaa'; We; "psi."

PRINI':PRJNI':PRM

SHRIMP?

mas-mom"
mes-'Drivermtinitializedusirgmdem
ms-mmwtotrange'
mas-'Baseadlmaaisoutofrange'
ms-Wmlindtamtotrange'
bus-“cmlnnberiamtofnrye'

Mn'wmth
C3337

 

I47
APPENDIX D

M-‘Wlevelmtofrange”

mas-machtatramferpameteramtof range"

mas-MSdncauamterpametersmtofrange'

bus-'mnnbarwtotnnge'

mas-'Oamteromtigmticnnnber>smmdem"

IDs-'OPfldntagmt-lanm'

M-Mmdmlmtotmige"

mas-"Triggerlevelmtofrange'

ms-flflguflopenotlorm
am (16611701180119)
M-‘Nocined'
(18340
ms-mssmmmm'
08341
Bus-mamotnrgetormdeasomsm
um
lees-'Nonmmrthiamnnber'
mm
m

I48

APPENDIX E

Plots of burst pressure versus the fourth power of the ratio. film thickness divided by
diaphragm diameter. Plots are presented for sets DF-Nl. DF-NZ. DF-N4. DF-NS, and

WF#25.

1.5E+05

1

1.0E+05 -

5.0E+04-

Pressure (Pa)

 

I49
APPENDIX E

DF-N1le4 i

 

 

0.0E+00—$ v I

' i
2.8000E-14 2.8400E-14 2.8800E-14

[t/d]**4 Dimensionless

Figure El. Insignificantly small data set DF—N 1. Included only to complete appendix.

 

 

 

 

3E+05 ' ‘ ‘ '
A ‘ ‘ __
m -1
9” 2E+05e _
a) A
5 A
g 1E+05- —-
a“. i‘ DF-N2e4

0E+00 I l ' I

0.0E-I-00 2.0E-13 4.0E-13

[t/d]**4 - Dimensionless

Figure E2. Plot of burst pressure versus the fourth power of the ratio, film thickness divided
by diaphragm diameter for data set DF-NZ.

150

 

 

 

 

APPENDIX E

3E+05 ‘ 1 1
g . DF-N4e4 A »
V 2E+05 ~ r-
2 . -
3 q A
(D
3 1E+05 — A -

. CL .
OE+OO ‘ l l
4.0E—1 3 6.0E-13 8.0E-13

[t/d]**4 - Dimensionless

Figure E3. Plot of burst pressure versus the fourth power of the ratio. film thickness divided
by diaphragm diameter for data set DF-N4.

 

 

 

 

3.0E+05 4 1 . l .
a .A
g 2.0E+05“ A b
9
D “A A
$3 105+05— -
e ' 1 .
0- * DF-N5e4 “
0.0E+00 r , T 8
0E+00 1E-12 2E-12 3E-12

[t/d]**4 - Dimensionless

Figure E4. Plot of burst pressure versus the fourth power of the ratio, film thickness divided
by diaphragm diameter for data set DF-NS.

4E+05

15]

1 r i

 

3E+05e

8+

2E+05a

 

1E+05

Pressure (Pa)

A
0E+00|

A

WF#25e4

r T—I

l

 

 

0E+OO

r ' i '
1E-11 2E-11 3E-11
[t/d]**4 - Dimensionless

Figure E5. Plot of burst pressure versus the fourth power of the ratio, film thickness divided
by diaphragm diameter for data set WF#25.

152

APPENDIX F

Plots of burst pressure versus the square root of the ratio, film thickness divided by
diaphragm diameter. Plots are presented for sets DF—Nl, DF-N2, DF-N4, DF-NS, WF#25,

and WF#60.

Pressure (Pa)

I 53

 

 

 

 

APPENDIXF

2E+05 i 1 l 1 l L i L

2E+05- GS'M ~
‘ A

1E+05~ ~

5E+O4— -
J

05+00‘1 I , I , :W
O£9 10 ‘L0 1t) 10 ‘lO 1.1 11

[gs]**-1/2 (micron)**-1/2

Figure F1. Insignificantly small data set DF-Nl. Included only for the completion of

 

 

 

 

Appendix F.
3E+05 ‘ ‘ 1 ' l
A ‘ ‘ "
cu ‘ GS-N2
Q; 2E+05— ‘ "
(D 8
a: 1E+05—
9. A
0. _
0E+00 I I I I
1.0 1.1 1.2 1.3 1.4

[gs]**—1/2 (micron)**-1/2

Figure F2. Plot of the burst pressure versus the square root of grain size for data set DF—NZ.

154

 

APPENDIX F

3E+05 1 J 1 l 1 l 1 l 1 l L l 1
33? J GS-N4‘ .
V 2E+05- *-
0) A
5
(D A
g 1E+05- A -
D. ‘ L

OE+OO . , . , .

 

 

 

l ‘ I ' I ' I ‘
1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8
[gs]**-1/2 (micron)**-1/2

Figure F3. Plot of the burst pressure versus the square root of grain size for data set DF—N4.

 

1 l l 1 l l l J
2.5E+05- ‘ ~—
33? - GS-N5 .
j; 2.0E+05- ‘ L-
h -1 .
3
am) 1.5E+05“ A ‘ b
g . -
°- 1.oe+05a A —
I 1*

 

 

 

' .' l ' l ' l l l l
0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6
[gs]**-1/2 (micron)**-1/2

Figure F4. Plot of the burst pressure versus the square root of grain size for data set DF—NS.

 

 

155

 

 

 

 

APPENDIXF

4.0E+05 1 l 1 l 1 l 1 l 1‘
’11? ‘ GS#25 *
g; - .
w -1 l-
‘5 2.0E+05— —
(D - A A _
(D A
9 ~ 1 1
D. .. .

0.0300! 1 I 1 1 1

1.0 1.1 1.2 1.3 1.4 1.5

[gs]**-1/2 (micron)**-1/2

Figure F5. Plot of the burst pressure versus the square root of grain size for data set WF#25.

Pressure (Pa)

 

 

 

 

1 l l
‘ 4E+05-— GS#69

. g r
I a A .
2E+05- i ' ‘ -
. ‘ ‘ _
Z “A ‘ ‘ A

0E+oo f I

8.0E-01 1.0E+00 1.2E+00 1.4E+00

[gs]**-1/2 (micron)**-1/2

Figure F6. Plot of the burst pressure versus the square root of grain size for data set WF#60.

10.

11.

12.

13.

14.

156
References

Life Application Bible, Tyndale House Publishers, Inc., Wheaton 111., xii, ( 1988).
MB. Hursthouse, J. Kane, and AG. Massey, “Plastic Deformation and Anisotropy
in the Hardness of Diamond”, Nature, 228: 660-661, (1970).

R.M. Chenko and HM. Strong, “Physical Property of Diamond”, Report No.
75CRDO89, General Electric Company Schenectady, N.Y.,October (1975).

Henry Windischmann and Glen F. Epps, “Properties of diamond membranes for x-
ray lithography”, J. Appl. Phys. 68 [11]: 5665-5673, (1990).

Karl E. Spear, “Diamond-Ceramics of the Future”, J. Ceram. Soc., 72 [2], 171-191,
(1989).

A. Masood, M. Aslam, M.A. Tamor, and T.J. Potter, “Techniques for Patterning of
CVD Diamond Films on Non-Diamond Substrates”, J. Electrochemic. Soc., 138,
L67, (1991).

Michael Frenklach and Karl E. Spear, “Growth Mechanisms of Vapor-Deposited
Diamond”, J. Mater. Res., 3: 133-140, (1988).

Stephen J. Harris, “Gas-phase kinetics during diamond growth: CH4 as-growth spe-
cies”, J. Appl. Phys. 65 [8]: 3044-3048, (1989).

Michael Frenklach, “The role of hydrogen in vapor deposition of diamond”, J.
Appl. Phys. 65 [12]: 5142-5149, (1989).

C.-P Chang, D.L. Flamm, D.E. Ibbotson, and J .A. Mucha. “Diamond crystal growth
by plasma chemical vapor deposition”, J. Appl. Phys. 63 [5]: 1744-1748, (1988).
John C. Angus, Herbert A. Will, and Wayne 8. Stanko, “Growth of Diamond Seed
Crystals by Vapor Deposition”, J. Appl. Phys. 39 [6]: 2915-2922, (1967).

DJ. Piclcrell, W. Zhu, A.R. Badzian, R.E. Newnham, and R. Messier, “Near-inter-
face characterization of diamond films on silica”, J. Mater. Res. 6 [6]: 1264-1277,
(1991).

J .A. Mucha, D.L. Flamm, and DE. Ibbotson, “0n the role of oxygen and hydrogen
in diamond-forming discharges”, J. Appl. Phys. 65 [9]: 3448-3452, (1989).

H. Windischmann and Glenn F. Epps, “Intrinsic stress in diamond films prepared by
microwave plasma CVD”, J. of Appl. Phys. 69 [4]: 2231-2237, (1991).

 

 

15.

16.

17.

18.

19

20.

21.

22.

23.

24.

25.

26.

27.

28.

157
ES. Berry, W.C. Pritchet, J .J . Cuomo. C.R. Guarnieri. “Internal stress and elasticity
of synthetic diamond thin films”, Appl. Phys. Lett. 57 [3]: 302-303, (1990).
Xiang Xin Bi, P.C. Eklund, J .G. Zhang, and A.M. Rao, “Optical properties of chem-
ical-vapor-deposited diamond films”, J. Mater. Res. 5 [4]: 811-817, (1990).
Sumio Iijima, Yumi Aikawa, and Kazuhiro Baba, “Growth of diamond particles in
chemical vapor deposition”, J. Mater. Res. 6 [7]: 1491-1497, (1991).
Satya P. Chauhan, John C. Angus, and Nelson C. Gardener, “Kinetics of carbon on
diamond powder”, J. Appl. Phys. 47 [11]: 4746-4754, (1976).
T. DebRoy, K. Tankala, W.A. Yarbrough, and R. Messier, “Role of heat transfer and
fluid flow in the chemical vapor deposition of diamond”, J. Appl. Phys. 68 [5]:
2424-2432, (1990).
M.A. Tamor, “Practical control of CVD diamond film morphology: chemistry,
geometry, and reality.” seminar, Physics an Astronomy Bldg. Michigan State Uni-
versity, Oct. 29‘", 1993.
Carol Gamlen, “Adherence of diamond thin films on silicon wafers and FTIR spec-
troscopy damaged ceramic materials”, Masters Thesis, Department of Materials
Science and Mechanics, Michigan State University, (1990).
H. Keller, “Si-ExraKtionsoptik fur filamentlose Ionenquellen”, Diplomar beit zum
Thema, University of Wuppertal, Germany, June 1987.
Henry Webe, Graduate Student, Department of Material Science and Mechanics,
Michigan State University, created and installed software, (1993).
Bill Ghe, “ Mechanical behavior of CVD diamond thin film diaphragms”, Master
Thesis, Department of Materials Science and Mechanics, Michigan State Univer-
sity, (1994).
J. Zhang, Ph.D. candidate Electrical Engineering Department, personal communi-
cation, MSU, November 1993.
Walter A. Yardbrough and Russell Messier, “Current issues and problems in the
chemical vapor deposition of diamond”, Science 247: 688-695, (1990).
Stephen Timoshenko, mm, 2“d ed., New York: McGraw-
Hill, (1959). '
Chuen-Yaun Chia, WW 1St ed., New York: McGraw-Hill
International Book Co., (1980).

 

29.

30.

31.

32.

158
B. Banerjee, “Large deflections of circular plates of variable thickness”, Int. J. Sol-
ids Structures, 19 [2]:l79—182, (1983).
RF. Davis, Z. Sitar, B.E. “Williams, H.S. Kong, H.J. Kim, J .W. Palour, J .A. Edmond.
J. Ryu, J .T. Glass and CH Carter, Jr., “Critical Evaluation of the Areas for Future
Research Regarding the Wide Band Gap Semiconductors Diamond, Gallium
Nitride and Silicon Carbide”, Material Science and Engineering, BI, 77 -104.
(1988).
W.D. Kingery, H. K. Bowen, and DR. Uhlmann,1_nt_rod_mm§§mmrg§, 2nd Edi-
tion, \Vrley-Interscience publication, 794, (1976).
Phillip A. Schmeitzer, “Corrosion Resistance Table: Metals, Nonmetals, and Rub-
bers”, 2nd Edition, 90-97, (1986).

 

‘__..—-—-‘.-.A— w 9* .H———

I...