g 35....
.rpgy-agag
is""¢§"‘§ 35'"
Vet‘éb'; 3?:-

 

3,...

d‘
éfig‘ts 3})

Mi?

‘V‘A::- ... _

mI-ur-

h‘l4t’ 0.... n

m-
. 25:?

i' » -
:-£ 33-5
H' 5;? at}: .

'Fu Wat
1 _L :k

{a 5"

ӣ522.11
. b..-

.........‘ 3.--.
.. Iva-n...—

.. “mas
.

.-u -

:3! '-
.5

fit» r'

.0 p . . 4. “£31
‘- vfiég “.1 “fl";
‘rfeziggem ‘ ~ -- M
H' gl’tfifi: . ‘ P 3;? ,: E5
)iig 1 '4“ ~ ‘ A
51“” fl ‘ Y

i

'u

._.‘,
. 4...“.~

E
i

t'

~ ,
"I'MFU

“‘71:.

..-—n.~A

7.7 'V. ll"; ;.k .1
§ 3 , .2 fig???
.5 ‘ .‘5‘ ~

, g’ 1:“ J3!“
“i u +** «g

,3;

V

-. m.
“a
nuv
w
A...
.—.~.
‘ .,

my
arm

«a... .- “-3-
~zrr‘ '
0...].
11'- M3”? ‘ ‘
~33":
x: .-....,
. “wt.
-4...

u.- g.—

7

I. 9.
"JI'.

3.

3.....-
2.7:-
.._..v
..
.w. .

at-..“ v.4»

Wv-vC-nv -

‘
k
,‘
.s~
.b

m;
_fl\
‘l ‘1

..
..
w

7112‘..-
2

-2:
A - 1:33;. -" :35“ ‘1.—
‘. 1:11.“? - at"
or. n.- “4.3;... vac
Mm.

(‘n

I
:0
I

:l'
4‘:
a

"'E

wu-

..
-
...yu..— u..
M.
Jr‘-

-«

‘1'.
1..

1,-
‘~U . I
u.-

."u.
w

-* ‘Y
... .

u- ‘

‘ 'rz
....-...
am...

an:
“.44.;
‘ T“ '

i

M.

3:1“
-“t

I».

:33... ”3..
:33.
..u
'gafiéar"
H. v m
m
wr-
aaxé

"Q

.w-rm

..-
~Q‘

‘mfi.
.. ""3

, ' . W .I . .
,5  . * ,i _‘ ‘ , I I '
I 'n. ‘j _,u_ ‘.
<‘ 1 < z : ,
IV. . .I ‘I. T "7"»
.~ . 7‘. 7.3. : g :-
'»' _§f . .1
' ,L.3NV?' :
' 7":- I'.. 1:1 1‘
J ‘T §-“'
' “"'I‘- k‘“ V
.g: -. :‘fi, ’._
'. ; .10; V l ‘
E." 4". " '-
A L; t ‘4
V I : ‘jz‘é‘
4%" _. . 'f 3 , ,_
, :4 f" I“ . I
3: {=1 - 7' t
.L ‘:5 5 g‘ .
: > ;, ,gx . ‘ = 5‘, -
' £1. V 5&5”
» . . .k;i
a : if. -. " .
.3.- $50?' 'W" .
w~r”‘ %I,_ 3
. ,- 2‘1-
‘1'“ 2mgéy,
' {Q’Er‘lfi " ‘
a . ' 32:.   .‘
3t “ z- '. <
. ..'V " "L -’ 31--
i ‘35:- ‘1‘ ' " ,
H.,’ .E f
1:: '«‘3
‘!' ,- ' ‘ ' -'
L"... m‘ if".
':_2‘)'.“"".~I:.' . '
3*‘a31.; .‘vx'
»‘ 93%: ‘ 11'“ "- A
'. " ‘
. . n31
. , 3 i'
‘. 1 ' b”
l l
‘ 3;}: m
(E ‘ *
51, -V .15
‘ L’r - "
i: g ‘1

.. m-..- . _
-3~:—..-...~

"no.
M-¢- ..

"E’

 

4';

23m:

~

II‘ICOID

fa llfllllllllllllllIllNIH”!!!HWI(“HillIIHHIHIHHM

310 820 0083

This is to certify that the ‘

thesis entitled
EXPERWAEN‘TAL LuARACTER\LNKmN Ox: A

W “WW“?
EmPLo‘an‘TWREE b\cFEK€N-\ Lquang Carsometmes.

W909. PLAsMA ‘JOURCE

presented by

kmw‘ \—\I\L.\M KHnN

has been accepted towards fulfillment
of the requirements for

MfiL—degree in ththi marlin?“ axing

 

mom

0 Major professor

Date 5LT1?61 bit-‘32)- meswm

 

 

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

 

 

 

 

 

 

LIBRARY

Michigan State
University

 

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1M mumps-p.14

 

 

 

EXPERIMENTAL CHARACTERIZATION OF A COMPACT MPDR PLASMA
SOURCE EMPLOYING THREE DIFFERENT COUPLING GEOMETRIES.
By

Amir Halim Khan

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

MASTER OF SCIENCE

Department of Electrical Engineering

1999

ABSTRACT

EXPERIMENTAL CHARACTERIZATION OF A COMPACT MPDR PLASMA
SOURCE EMPLOYING THREE DIFFERENT COUPLING GEOMETRIES.

By
Amir Halim Khan

In this thesis three compact microwave plasma disc reactor (MPDR) plasma sources, i.e.
MPDR 7 - Side Feed (SF), MPDR 7 - End Feed Probe (EFP) and MPDR 7 - End Feed
Loop GEFL), have been experimentally characterized by varying the input variables space
(U1) and keeping the reactor geometry variables space (U2) fixed. The influence of three
different coupling structures on the plasma source performance have been evaluated and
compared with the global plasma model.

The measured internal variables (X) and output variables (Y) for the three plasma
sources are: 1) plasma density, 2) absorbed power, 3) ion saturation current, 4) electron
temperature, and 5) ion production costs. Additionally, electron energy distribution
function, plasma potential, absolute and spatial electric fields are measured for the side
feed excitation. The experimental technique involves single and double Langmuir probe
measurements and E-field probe measurements.

The performance of the three investigated sources are compared in terms of ion
production costs. The performance predictions from the global plasma model and the
experimental results are compared and the differences are analyzed. Finally, an improved

plasma source design is proposed.

Dedicated to my parents, my wife, and our sons Waleed and Saif

iii

ACKNOWLEDGEMENTS

I would like to thank Dr. Jes Asmussen for his continuous guidance and support which
helped in developing skills that were required to carry out this research. My special
thanks goes to Mark Perrin for his training, useful discussions and the time we made
enjoyable in the laboratory. I extend my thanks to Dr. Timothy Grotjohn who was always
there to assist us during experiments. I would like to thank Dr. Don Reinhard for being
my committee member. Finally, I would thank Jay Kumar for being an excellent team

member.

iv

LIST OF TABLES

x
LIST OF FIGURES xi
Chapter 1
Introduction
1.1 Research motivation I
1.2 Generic microwave plasma reactor 4

1.2.1 Physical description and operation 4
1.2.2 The multivariable plasma reactor 8
1.3 Evolution of MPDR sources at MSU 11
1.4 Brief description of the plasma sources investigated in this thesis -------------- 14
1.5 Research Objectives 18
1.5.1 Specific research objective 18
1.5.2 Experimental plan 19
1.5.3 Experimental evaluation procedure 21
1.6 Thesis outline 26

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

 

 

 

Chapter 2

Experimental systems

 

 

 

 

 

 

 

 

 

 

 

 

 

2. 1 Introduction 27
2.2 Experimental systems 27
2.2.1 Microwave power supply and power monitoring circuit -------------- 30

2.2.2 Vacuum and gas flow system 31

2.3 The Microwave Reactor 34
2.3.1 The plasma source 34

2.3.1.1 The cavity applicator and impedance matching network ----- 37

2.3.1.1.] MPDR 7- Side Feed (SF) 37

2.3.1.1.1.1 Brief History 37

2.3.1.1.1.2 Construction and operation ------------- 38

2.3.1.1.2 MPDR 7 End Feed Loop (EFL) 41

23.1.1.2] Construction 41

2.3.1.1.3 MPDR 7 End Feed Probe (EFP) 42

2.3.1.131 Construction 42

2.3.1.2 Quartz discharge chamber 44

2.3.1.3 Brass ECR magnet keeper assembly 44

2.3.2 Process chamber 48

 

vi

Chapter 3

Experimental techniques

 

3. 1 Introduction

 

3.2 Langmuir probes

3.2.1 Introduction

 

 

3.2.2 Double Langmuir probes

3.2.2.1 Probe design

 

 

3.2.2.2 Principle of operation

 

3.2.3 Single Langmuir probe

 

3.2.3.1 Probe design

 

3.2.3.2 Principle of Operation

50

50

50

50

54

59

59

59

66

 

3.3E1ectric field measurements

3.3.1 Introduction

 

3.3.2 B - field probe design

66

 

3.3.3 Principles of operation — Burkhart’s method

vii

 

69

4.1

4.2

4.3

4.4

4.5

4.6

4.7

4.8

4.9

Chapter 4

Experimental Results

 

Introduction 70

Part I

Electric field measurements and identification Of EM excitation modes in the

 

 

 

 

 

 

 

MPDR 7 — SF plasma source 75
Electron density in the MPDR 7 - SF plasma source 87
Plasma potential and electron temperature 92
Electron energy distribution function (EEDF) 92
Part 11
Global plasma model 100
4.6.1 Ion conservation relation 101
4.6.2 Power balance relation 104
Excitation modes in the three MPDR plasma sources 106

 

Performance comparison between global plasma model predictions and the

experimental measurements of the three different coupling geometries. -------- 108

4.8.1 Comparison of the three plasma sources with the global model predictions
using experimentally measured electron temperatures ------------ 111

Plasma source performance figure of merit — Ion production cost (eV/ion) ---122

 

4.9.1 Experimental measurement of ion production cost -- --122

4.9.2 Derivation of ion production cost from global model equations ---125

viii

4.10

4.11

5.1

5.2

5.3

5.4

5.5

5.6

5.7

4.9.3 Similarities between the five electromagnetic excitations -------

4.9.4 Performance comparison of three MPDR plasma sources -------

Performance comparison of the three plasma sources with other designs

----- 127

----- 127

 

Summary

Chapter 5

Conclusion

 

Work done - Salient features

141

141

 

Characterization of MPDR 7 - SF plasma source

5.2.1 Electric fields

141

 

 

5.2.2 Electron energy distribution function (EEDF)

5.2.3 Electron temperature and plasma potential

142

142

 

 

5.2.4 Electron density

143

143

 

Experimental results of the three plasma sources
5.3.] Comparison of the experimental results with the global model

5.3.2 Influence of the coupling structure on the source performance

Comparison of the three investigated sources with other designs --------

--—143

"-144

«“145

 

Suggestions for an improved plasma source design

Recommendations for the future research

146

146

 

Concluding remarks

148

 

ix

Chapter 1
Table 1.1
Table 1.2
Table 1.3

Table 1.4

Chapter 4
Table 4.1
Table 4.2
Table 4.3
Table 4.4

Table 4.5

Table 4.6

LIST OF TABLES

The trend in IC growth

 

Input variables for MPDR 7 — Side Feed

 

Input variables for MPDR 7 -— End Feed Probe (EFP)

Input variables for MPDR 7 - End Feed Loop (EFL)

Internal and output variables measured for MPDR 7 - SF
Internal and output variables measured for MPDR 7 — EFP

Internal and output variables measured for MPDR 7 — EFL

 

 

 

 

 

 

5,2 / Fab, ratio for different input conditions

23

24

25

72

73

74

86

Excitation modes and their matched tuning positions for the three MPDR

 

plasma sources

107

Experimental and theoretical electron temperatures (T..) for the MPDR 7 —

 

SF plasma source

111

Chapter 1
Figure 1.1
Figure 1.2

Figure 1.3

Figure 1.4
Figure 1.5
Figure 1.6
Figure 1.7
Figure 1.8

Figure 1.9

Figure 1.10

Chapter 2
Figure 2.1
Figure 2.2
Figure 2.3
Figure 2.4
Figure 2.5

Figure 2.6

LIST OF FIGURES

 

 

 

 

Silicon wafer, IC, and transistor 3
The generic ECR reactor 6
ECR plasma reactor applications 7
Figure 1.3a Plasma processing source 7
Figure 1.3b Ion beam source 7

 

Generic plasma processing machine - System variables block diagram -10

 

 

 

 

 

 

 

 

 

 

Evolution of MPDR sources at MSU 13
MPDR 7 - Side Feed (SF) plasma source 15
MPDR 7— End Feed Loop (EFL) plasma source 16
MPDR 7- End Feed Probe (EFP) plasma source 17
Experimental plan 20
Experimental system variables - 22
The generic plasma processing machine 29
Microwave power supply and power monitoring circuit 33
MPDR ion and plasma source — Exploded view 36
Design of bottom end plate adapter 40
End feed probe assembly 43

 

Process and discharge chamber 46

 

xi

Figure 2.7

Chapter 3
Figure 3.1
Figure 3.2
Figure 3.3

Figure 3.4

Figure 3.5
Figure 3.6

Figure 3.7

Chapter 4
Figure 4.1a
Figure 4. lb

Figure 4.2

Figure 4.3a
Figure 4.3b
Figure 4.4a
Figure 4.4b

Figure 4.5

 

 

 

Brass ECR magnet keeper assembly ----- 47
Double Langmuir probe experimental setup 53
Schematic of double Langmuir probe measuring circuit --------------- 57
A typical double Langmuir probe I —V characteristic curve 58

Schematic circuit diagram of single Langmuir probe measurement

 

 

 

 

technique 63
A typical single Langmuir probe I —V curve 64
A typical single Langmuir probe second derivative curve 65
E-field probe design and electric field measurement setup 68

Circumferential variance of electric field strength in MPDR 7 -— SF ------ 79

Longitudinal variance of electric field strength in MPDR 7 — SF ------ 79
Electric field theoretical and measured profile within the MPDR 7 — SF
ion source -- -80

 

Circumferential variance of electric field strength in MPDR 7 - SF ------ 81
Longitudinal variance Of electric field strength in MPDR 7 — SF ------ 81
Circumferential variance of electric field strength in MPDR 7 — SF ------ 82
Longitudinal variance of electric field strength in MPDR 7 — SF ------ 82

Circumferential variance of electric field strength in MPDR 7 — SF versus

 

input power at H = 4.3cm plane --83

xii

Figure 4.6

Figure 4.7

Figure 4.8

Figure 4.9

Figure 4.10

Figure 4.11

Figure 4.12

Figure 4.13

Figure 4.14

Figure 4.15

Figure 4.16

Circumferential variance of electric field strength vs. pressure in MPDR 7

 

- SF at H = 4.3cm plane 84
Circumferential variance of the electric field strength versus input power

in MPDR 7 — SF at H = 4.3cm plane 85

 

Top View projection of coupling antenna probe over the pole faces of

ECR magnets 89

 

Electron density of MPDR 7 — SF plasma source as a function of absorbed

 

power 90

Electron density in MPDR 7 — SF as a function of chamber pressure at a

 

fixed flow rate of 203ccm 91
Electron temperature vs. chamber pressure in MPDR 7 — SF ------ 94
Plasma potential vs. chamber pressure in MPDR 7 -— SF --------------- 95

Electron Energy Distribution Function (EEDF) at z = 3cm downstream
Input Power 100 Watt, Pressure lmtorr, Argon gas flow rate 203ccm ---96
Electron Energy Distribution Function (EEDF) at z = 3cm downstream
Input Power 100 Watt, Pressure 2mtorr, Argon gas flow rate 203ccm ---97
Electron Energy Distribution Function (EEDF) at z = 3cm downstream
Input Power 100 Watt, Pressure 4 mtorr, Argon gas flow rate 208ccm «98
Electron Energy Distribution Function (EEDF) at z = 3cm downstream

Input Power 100 Watt, Pressure 8mtorr, Argon gas flow rate 203ccm ---99

xiii

Figure 4.17

Figure 4.18
Figure 4.19
Figure 4.20
Figure 4.21
Figure 4.22
Figure 4.23
Figure 4.24
Figure 4.25a

Figure 4.25b

Figure 4.25c

Figure 4.26a

Figure 4.26b

Experimental electron temperature (data points) plotted against the

universal curves for uniform & non-uniform Argon gas discharges -

 

 

 

 

MPDR 7 -— SF plasma source 114
Power density load lines for MPDR 7 — SF 115
Power density load lines for MPDR 7 — EFP 116
Power density load lines for MPDR 7 - EFL 117
The Absorbed power load lines for MPDR 7 SF plasma source ----118
The Absorbed power load lines for MPDR 7 SF plasma source ----119
The Absorbed power load lines for MPDR 7 SF plasma source ----l20
The Absorbed power load lines for MPDR 7 SF plasma source ----121

Ion production cost versus pressure — MPDR 7 — SF plasma source ---134
Ion production cost versus pressure -Theoretical and experimental
performance comparison for side feed excitation geometry ------------- 135
Ion production cost versus pressure -Theoretical and experimental
performance comparison for the three MPDR plasma sources ----136
Ion production cost versus pressure -Theoretical and experimental
performance comparison for the TEm in the SF and EFL excitation

geometries 137

 

Ion production cost versus pressure -Theoretical and experimental
performance comparison for the TM01 / TEM mode (low modes) in the

EFP and EFL excitation geometries 138

 

xiv

Figure 4.26c Ion production cost versus pressure -Theoretical and experimental

Figure 4.27

performance comparison for the TE 11 and TM01 mode (high modes) in the

EFP and EFL excitation geometries 139

 

Ion production cost versus pressure — A comparison of the investigated

sources with different reactor concepts 140

 

XV

Chapter 1

Introduction

1.1 Research motivation

Impact of integrated circuit (IC) technology on our lives has been pervasive, and
its domain ranges from consumer goods to information technology, from business
management to manufacturing control. As a result of the exponential growth in IC
fabrication processes, very large scale (VLSI) and Ultra large-scale integrated circuitry
(ULSI) has made a single chip so dense that it bears tens of thousands of transistors.
Figure 1.1 displays the critical dimensions of a single transistor embedded on a silicon
wafer. This dramatic reduction in terms of feature size, and supply voltage, and the
enhancements in processor speed and storage capacity is shown in the Table 1.1. The
quest for an optimum processing speed, minimal dissipated power, least device failure
rate, compact circuit area, blended with a fierce cost competition has revolutionized the
electronic industry. In order to meet the dynamic consumer requirements, manufacturing
groups in the IC industry should be equipped with all the tools that permit them to face
this challenge.

For some time in the semiconductor industry low-pressure gas discharges or
plasmas have been one of the process tools that have enabled the etching of anisotropic
sub-micron features. Besides sub-micron anisotropic etching, many additional plasma
based applications exist in the automotive, aerospace, biomedical and steel manufacturing

industries. Plasmas are also being utilized in material processing such as plasma assisted

crystal growth, i.e. in particular plasma assisted CVD diamond film growth. The last two
decades have seen high density plasma sources, such as electron cyclotron resonance
(ECR) based plasma processing machines edging out parallel plate plasma sources in
terms of high plasma density, high etch rate, low sheath potential etc. These plasma
processing ECR reactors can be configured in a number of ways [1] e.g. Slot-Antenna
(SLAN) reactors, Divergent Field ECR reactor, Cylindrical Distributed ECR reactor
(DECR), Uniform Distributed ECR reactor (UDECR) etc. An additional ECR concept,
the multi-polar ECR reactor or microwave plasma disc reactor (MPDR), which was
developed and patented by Michigan State University researchers [2], is the subject of this
thesis research. These computer-controlled reactors have been demonstrated to fulfill the
highest level of process uniformity and process speed which are the required criteria for
the contemporary 0.35 pm IC fabrication industry [1].

Even though the microwave plasma disc reactor (MPDR) ion/plasma sources have
matured into a commercial product, there still are a number of unanswered questions
related to the performance of these sources. The motivation for manipulating the complex
chemical, physical and electromagnetic reactions within the gas discharge generated by
MPDR ion sources is to obtain selectivity, stability and controllability in the plasma
processing mechanism Thus, in sequel to the on going experimental evaluation of MPDR
ion sources, this study builds upon the research of earlier investigators and aims at design

optimization of MPDR plasma sources developed at MSU.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year
Technology 1992 1994 1996 1998 2000
Channel Length(p.m) 0.5 0.4 0.3 0.2 0.1
DRAM Bits(MB) 16 64 64 256 256
Supply Voltage(V) 3.3/5.0 3.3 2.5 2.5/3.3 2.5
Logic Gates (x 1000) 200 400 600 1000 2000
Supply Voltage(V) 3.3 3.3 2.5 2.5 [.5
Processor Speed(MHz) 50 100 150 200 500
Table 1.1 The trend in IC growth
gate
source drain
100-1000chips —
10-100,000 4 *
transistors 25 um
emitter collector

 

  
   

  

100 -125
(0.2mm

Ola;n'r\

 

base

 

MEX

 

 

{

 

Figure 1.1 Silicon wafer, IC, and transistor

40um

w

1.2 Generic microwave plasma reactor

1.2.1 Physical description and Operation

Although the MPDR sources exist in a number of geometrical configurations, for
ease in understanding the fundamental concepts they can be represented by a generic
reactor model. Figure 1.2 displays a generic low-pressure microwave plasma processing
reactor and Figure 1.3 displays typical applications of ECR plasma reactors, i.e. as an ion
beam source and a plasma processing source. The typical reactor can be broadly divided in
two parts: 1) the plasma source and 2) process chamber. The plasma source consists of
three sub-assemblies: a) microwave cylindrical cavity applicator and matching network, b)
discharge chamber and c) ECR magnets. The cylindrical cavity shell is usually made fi'om
either brass or stainless steel. A sliding short plate forms its top, and the bottom contains
the quartz discharge chamber. Once the high vacuum pump brings the system to a low
base pressure, the process gas is let in. The microwave applicator focuses the microwave
energy into the discharge chamber and creates a microwave discharge. This discharge or
plasma consists of charged particles and chemically active radicals. As shown in Figure
1.3a, these species diffuse out through the discharge chamber (quartz) Opening at the z = 0
plane into the processing chamber. Altemately, as shown in the Figure 1.3b, a single or a
set of double accelerating grids located at the source output, i.e. at the z = 0 plane, can be
utilized to extract an ion beam from the source. The radius and height of the quartz dome,
‘R’ and ‘L’ defines the discharge volume. The radius ‘Rc’ and length ‘Lc’ bounds the
cylindrical process chamber. The process zone is defined by a substrate of radius ‘Rs’

which is positioned downstream at a z > 0 plane. This substrate is mounted on an

adjustable holder. The operating chamber pressure is maintained by continuously pumping
down the working gas.

In low-pressure discharges the mean fiee path of radicals is much larger than the
reactor’s dimensions. Thus in contrast to high-pressure discharges they are dominated by
wall recombination instead of volume recombination. The low-pressure discharges require
the presence of strong external magnetic fields, which are provided either by permanent
magnets or electromagnetic coils. The magnetic fields reduce the surface diffusion losses

to the walls and assist in the discharge stability and improve the reactor efficiency.

Side feed
excitation

 

 

 

Fmd fwd
excitation

EM matching
network

End feed End feed
probe loop

 

 

RA

 

 

Applicator

 

 

 

 

Vacuum
wmpins

Figure 1.2 The generic ECR reactor

   

Process chamber

nnnnnnnnnnnnnnn

Substrate

Figure 1.3a Plasma processing source

 

Microwave
Input Signal yam

Process

/' Accelerating
chambeN

grid

/7 Substrate

 

 

 

 

Figure 1.3b Ion beam source

Figure 1.3 - ECR plasma reactor applications

1.2.2 The multivariable plasma reactor

The plasma reactor is a highly nonlinear electronic device that is dependent on a
number of experimental variables. The generic representation of these variables is
illustrated in the Figure 1.4. It is seen that a host of variables have been grouped together
to form a three variable vector space: 1) input variable vector (U), 2) internal variable
vector (X) and 3) output variable vector (Y).

The input variable (U) is an independent controllable vector space, which is sub-
divided in four vectors U1, U2, U3 and U4. The variables U, are comprised of macroscopic
controllable input variables such as absorbed power (Pabs), operating chamber pressure,
gas type, flow rate etc. The variables U2 relate to the reactor geometry such as the
applicator size, excitation geometry etc. The input variables U3 consist of substrate
processing procedures and include process control cycle, discharge initiation etc. The
variables U, are related to substrate control and are comprised of substrate preparation,
the bias and temperature of substrate holder, etc.

The internal plasma variables (X) are dependent upon the input variables (U). The
internal variables (X) describe, in general, the variables associated with internal plasma
dynamics. The internal variables (X) include plasma potential, charged species energy
distribution functions, absorbed microwave power (Pam), spatial electric field pattern,
uniformity in the density profile etc.

The output variables (Y) can include all the parameters related to the desired
reactor output. They are a function of both input and internal variables. They can be

further sub-divided into reactor performance figures of merit (Y1) and process output

variables (Y2). Reactor performance figures of merit (Y1) include ion production cost,
microwave coupling efficiency, stability, controllability etc. The process output variables
(Y2) consist of processing rate, substrate damage, surface morphology etc.

Researchers are still laboring hard to establish a theory governing the nonlinear
relationships between the plasma source variables. A comprehensive understanding of
these complex variables and their interdependence is essentially required for optimal

machine design and process control.

Processing
procedure
variables (U 3)

Macroscopic

 

 

R to
controllable fizllcre;
input variables of merit (Y 1)
(U1) D
variables (X)
Process Y = h(X) — g(U)
Reactor geometry thut

variables (U2) variables (Y2)

 

 

 

Substrate
control
variables (Ur)

Figure 1.4 Generic plasma processing machine - System variables block diagram

10

1.3 Evolution of MPDR sources at MSU

The genesis of microwave plasma disc reactor (MPDR) took place at MSU in
early 80’s when J. Root and J. Asmussen [4] presented the idea of an electrodeless plasma
source and introduced a first reactor of this kind in 1982. Since then this concept has been
evaluated and optimized and researchers have developed plasma sources that are now
being used in a number of commercial applications.

Figure 1.5 reviews the historical development of MPDR plasma sources at MSU.
The first MPDR prototype was a brass cylindrical cavity employing a side feed microwave
coupling configuration. Besides the stable sustenance of plasma through direct microwave
coupling, this innovative plasma source demonstrated maximum coupling efficiency by
employing an internal tuning mechanism. A few years later (1984), M. Dahirnene [5, 6]
demonstrated second prototype of plasma source, which used permanent magnets to
overcome the difficulties of low-pressure (<1 torr) operation. This reactor was developed
further [7] and is now known as the MPDR 9 plasma source, since the discharge diameter
of the plasma source was approximately 9cm (to be precise 9.4cm). The experiments with
MPDR 9 demonstrated that this source has a low ion beam current. L. Mahoney [8] in
1989 redesigned the base plate of MPDR 9, which accommodated a 5cm discharge. Thus
without carrying out any major changes the MPDR 9 was transformed into MPDR 5
source. This source improved the ion production cost by a factor of two and was able to
produce ion current densities of the order lOmA/cm2 [9]. The MPDR 5 used a cylindrical
cavity shell of 17.8cm inner diameter. Later Mahoney reduced the applicator’s shell

diameter to 8.9cm and this source is now recognized as the MPDR 7 side feed plasma

11

source. Additionally, an even smaller and more compact plasma source was developed in
1990. This source, which is now identified as the MPDR 610, was characterized by A.
Srivastava [10]. A third transformation resulted in the MPDR 20. F. Sze [11, 12] in 1993
investigated this 20cm diameter ECR plasma source. He also characterized a larger 25cm
diameter discharge called the MPDR 325 source [13]. This source has been
commercialized by Wavemat Inc. MPDR 13 sources in both side and end feed
configuration were studied by P. Mak in 1997 [14]. Recently the MPDR 7 End Feed Loop

plasma source was characterized by M. Perrin in 1996 [15].

12

 

 

MPDR 610 [:1-

Donated by IBM
Watson Research
Center

MPDR First Prototype

Q

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O
MPDR Second Prototype
MPDR 9
1:-
@fi
0
MPDR 5 [:21— :—
ISA-LI D x
MPDR l3
SideFeed MPDR 20

MPDR 13
End Feed

 

T

 

Figure 1.5 Evolution of MPDR sources at MSU

13

1.4 Brief description of the plasma sources investigated in this thesis

In order to draw a simple understanding of complex plasma systems, it would be
first convenient to investigate small diameter MPDR sources. The research work
described here spanned approximately one year and focussed on three compact MPDR
sources, namely MPDR 7 — Side feed (SF), MPDR 7 — End Feed Probe (EFP), and
MPDR 7 — End Feed Loop (EFL). Figures 1.6 — 1.8 give a schematic representation of
these plasma sources. The applicator for all the three sources consists of a cylindrical
cavity shell. For the side feed coupling configuration this shell is brass, and in the case of
end excited sources it is stainless steel. The diameter of the end feed excitation geometry
is slightly larger (9.8cm as compared to 8.9 cm) than the side feed source. The family of
MPDR plasma sources is characterized by the user’s control over the propagating mode in
the microwave cavity applicator. Microwave power is coupled through the side feed or
end feed excitation antennas. The sliding short end plates and the excitation antenna
provide a two-dimensional internal tuning mechanism for discharge matching and
electromagnetic mode selection.

The cavity applicator sits on a quartz discharge chamber where the plasma
discharge is ignited using a noble gas. The electron cyclotron resonance (ECR) magnets,
in an alternating pole configuration, surround the discharge chamber in a circular fashion.
They serve to provide the necessary ECR region which is required to sustain a low-
pressure discharge. The operation and physical description of these ion and plasma

sources are discussed in much more detail at Chapter2.

 

1) Finger stocks

2) Sliding short top plate
3) Brass shell

4) Excitation probe

5) Viewing window

IIIIIIIIIIIIIIIIIIIIIIIIIIII
IIIIIIIIIIIIIIIIIIIIIIIIIIIII
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII
eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee

 

7‘ H = 5.6cm Plane

_H=0Plane

z = 0 Plane

6) Quartz bell jar

7) Bottom end plate

8) Horizontal electric field sampling pinholes
9) Vertical electric field sampling pinholes

10) ECR magnets

Figure 1.6 MPDR 7 - Side Feed (SF) plasma source

15

 

 

 

 

Excitation
Loop Antenna

 

 

 

 

 

 

 

 

 

 

Figure 1.7 MPDR 7- End Feed Loop (EFL) plasma source

16

 

 

 

 

Excitation
Probe Antenna

 

 

 

 

 

 

 

 

 

Figure 1.8 MPDR 7- End Feed Probe (EFP) plasma source

17

1.5 Research objectives

The overall aim of this investigation is 1) to establish the performance of small
MPDR plasma sources, 2) to understand the physical phenomenon responsible for the
input/output behavior of MPDR plasma sources and then based on this information 3)
suggest ways and means to improve the design of existing MPDR plasma reactors. An
improved understanding of the physical behavior of MPDR plasma sources will allow

selective optimization of associated system variables and figures of merit.

1.5.1 Specific research objective

Even though a large number of MPDR plasma sources differing in geometrical
configurations have been already extensively characterized at MSU, there still are a
number of poorly understood aspects which directly relate to the operation of these
plasma sources. The objective of this research is to investigate the influence of the method
of cavity excitation on the performance of low—pressure microwave discharges.
Specifically this research evaluates the influence of cavity excitation by experimentally
evaluating three different excitation configurations: 1) side feed probe excitation, 2) end

feed probe excitation and 3) end feed loop excitation.

18

1.5.2 Experimental plan

Plasma generation and processing is a very complex and a non-linear process,
which is interdependent on a large number of what we call as procedural variables
(discussed in Chapters One and Four). A detailed information on these experimental
variables can be traced in the reference [14]. It has been very difficult to find the
correlation of these variables in terms of their combined and individual influence on
reactor’s output variables vector (Y). A complex experimental problem solving technique
isolates the variables and then investigates their independent control over the system
efficiency. This eliminates the influence of other variables on process output. This unique
methodology is employed in this thesis investigation by varying the microwave excitation
configuration, a reactor geometry variable (U2), and studying its influence on the reactor’s
performance. Therefore as described earlier, in this thesis, three MPDR 7 ion and plasma
sources, which only differ in the method of excitation, i.e. side feed, end feed loop and
probe excitation are evaluated (see Figure 1.9). The influence of excitation geometry on
the source performance is investigated by adopting a strategy defined in the following

section.

19

 

 

 

 

 

 

 

 

Plasma Source

 

Process chamber

 

 

 

Figure 1.9 Experimental plan

20

1.5.3 Experimental evaluation procedure

The variables associated with plasma processing machines have earlier been
discussed and defined in the Section 1.2.2. In this thesis the performance of the three
MPDR 7 plasma sources, shown in Figures 1.6 - 1.8, was evaluated by measuring an
exclusive group of internal variables (X) and output variables (Y) against the variations in
macroscopic controllable input variables (U1) and reactor geometry variables (U2). Figure
1.10 displays, in general, the system variables investigated in this work. Tables 1.2, 1.3
and 1.4 describe, in particular, the variance of independently controlled input variables
space (U) for the three MPDR 7 ion and plasma sources. Within the macroscopic
controllable input variables space (U1): 1) absorbed power (Pabs) and chamber pressure
were varied for all the three configurations, 2) gas flow rate was varied for the side feed
excitation only and it was fixed for the end feed sources, 3) Operating frequency was kept
constant at 2.4SGHz, 4) argon gas was used as the input gas for all the experiments, and
5) the tuning parameters Ls and LP were varied to attain the resonant conditions in the
three plasma sources. The applicator and discharge chamber sizes, and ECR magnet
configuration were kept fixed within the reactor geometry variables (U2). Most
importantly, three different methods of microwave excitation were employed in order to

understand the influence of the method of excitation on the plasma reactor performance.

21

 

 

 

Macroscopic controllable C D

 

 

input variables (U1)
0 Absorbed power (Pm) Internal
. Operating frequency Variables (X)
e Discharge gas
. C mum I Reactor Figures
. Flow rate e Plasma density of Merit (Y 1)
0 Tuning W35 L, “‘9 LP 0 Plasma potential

 

. EEDF . 0 Ion production cost

 

 

Reactor geometry 0 Absolute and spatial electric fields

Variables (U2) 0 Electron temperature
0 Absorbed power (Pu)

 

 

 

e Applicator size. type and
method of excitation 0 Ion saturation current (1,)
0 Side feed probe
0 End feed probe
0 End feed loop ’

0 Discharge chamber
dimensions /
o ECR magnets configuration

 

 

 

 

Figure 1.10 Experimental system variables

22

 

 

 

 

 

 

 

 

 

 

 

 

Input Variables Description Experimental Range
(U)
Absorbed power Pab, 50-200 Watt
Operating frequency 2.4SGHz (CW)
Macroscopic controllable Discharge gas 99.99% Argon
input variables Chamber pressure 1-8 mTorr
U1 Flow rate 0.3 — 358ccm
Cavity short length Ls 13.7 — 8.4cm
Depth of excitation probe 1.2 — 1.9cm
antenna Lp
Applicator size, type and 8.9cm diameter cylindrical
_ the method of excitation brass cavity with side feed
Fixed excitation
reactor geometry Discharge chamber Height 56an radius 3.5cm
variables dimensions with a total discharge
U2 volume of 215.5cm3
ECR magnet configuration 8 Pole 8 Magnets
(SP/8M)

 

 

 

Table 1.2 Input variables for MPDR 7 - Side Feed

23

 

 

 

 

 

 

 

 

 

 

 

Input Variables Description Experimental Range
(U)
Absorbed power Pab, 100- 150 Watt
Operating frequency 2.45GHz (CW)
Discharge gas 99.99% Argon
Macroscopic controllable Chamber pressure 1-8 mTorr
input variables Flow rate Fixed at 20 sccm
U1 Cavity short length I.s 3.13 — 10.78cm
Depth of excitation probe 0.97 - 3.82cm
antenna 1..p
Applicator size, type and 9.8cm diameter cylindrical
the method of excitation stainless steel cavity with
Fixed end feed probe excitation

reactor geometry
variables

U2

 

 

 

Discharge chamber Height 5.6cm, radius 3.5cm
dimensions with a total discharge
volume of 215.5cm3
ECR magnet 8 Pole 8 Magnets
configuration (8P/8M)

 

 

Table 1.3 Input variables for MPDR 7 - End Feed Probe (EFP)

24

 

 

 

 

 

 

 

 

 

 

 

Input Variables Description Experimental Range
(U)
Absorbed power Pab, 100- 150 Watt
Macroscopic controllable Operating frequency 2.45GHz (CW)
input variables Discharge gas 99.99% Argon
U1 Chamber pressure 1-8 mTorr
Flow rate Fixed at 20 sccm
Cavity short length L, 4.575 — 8.075cm
Depth of excitation loop 1.95 — 3.2cm
antenna Lp
Applicator size, type and 9.8cm diameter cylindrical
the method of excitation stainless steel cavity with
Fixed end feed loop excitation

reactor geometry
variables

U2

 

Discharge chamber

dimensions

Height 5.6cm, radius 3.50m
with a total discharge

volume of 215.5cm3

 

 

ECR magnet configuration

 

8 Pole 8 Magnets
(8P/8M)

 

Table 1.4 Input variables for MPDR 7 - End Feed Loop (EFL)

25

 

1.6 Thesis outline

The work in this thesis is divided into five chapters. The Chapter two describes the
plasma processing system The Chapter three reviews experimental techniques employed
in this research. Experimental results together with discussions are presented in the
Chapter four. In past investigations the global plasma model has been an effective tool to
understand and predict a plasma source behavior. Global model equations are described in
Chapter four and plasma source performance comparison is drawn for MPDR sources in
three different coupling configurations. Chapter five finally concludes this presentation by

giving the summary of results and conclusion drawn out of them.

26

Chapter 2

Experimental systems

2.1 Introduction

As discussed in Chapter 1, three compact Microwave Plasma Disc Reactor
(MPDR) ion and plasma sources are evaluated in this work. These sources are almost
identical in size but employ three different microwave excitation techniques. Thus these
three plasma sources are very useful in understanding the influence of microwave
excitation on the plasma reactor performance.

This chapter describes the experimental systems used in this research work. In
Section 2.2 microwave power supply, power-monitoring circuit, vacuum and gas flow
systems are illustrated. The microwave reactor is described in detail at Section 2.3. This
section also describes the design and operation of the three MPDR-7 ion and plasma

sources, and the geometry of process and discharge chambers.

2.2 Experimental systems

Even though there are a number of variations in the design of ECR plasma

processing machines, an ECR plasma processing system can still be described by a

generalized system Figure 2.1 displays a generic form of a typical plasma-processing

machine.

27

It can be seen from the figure that the experimental system is divided into three major
subsystems: 1) power supply and power monitoring circuit, 2) microwave reactor, and 3)

vacuum and gas flow system. These subsystems are described in the following sections.

28

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 )
I
1
\
2)
J
r
@1.
J k
\ j
V
3)

1) Power supply and power monitoring circuit
2) The microwave reactor

3) Vacuum and gas flow system

Figure 2.1 The generic plasma processing machine

29

2.2.1 Microwave power supply and power monitoring circuit

Figure 2.2 displays the microwave power supply and the power monitoring circuit
used in the experiments. As per experimental parameter space (see Table1.2, Chapter
one), a variable, 2.45GHz power supply (Micro-Now Inc.’s model 420B 1) was used in the
experiments. This power supply (1) which is specifically designed for research and
industrial applications [16] uses a CW magnetron source and has an output of 50-500
Watt at 2.45 GHz (:10 MHz) with 50 Ohm output impedance. The desired output power
can be obtained by slowly rotating the magnetron current control knob on the front
control panel. The microwave power is fed directly through a barrel connector to a three
port Narda coaxial circulator (4) which has a maximum power rating of 500 Watt. This
unidirectional circulator performs: l) isolation of the power supply fi'om the reflected
power due to sudden variations in the load impedance, and 2) connects power supply to
the microwave cavity applicator. Output power from the microwave supply tends to drift
whenever there is a change in the plasma load impedance. A user must be mindful to
frequently monitor the output power readout and adjust its level to the desired value.
Large reflected power might heat up the circulator and this may eventually burn it out. To
prevent this from happening, a cooling fan is provided at its top. As an additional
precautionary measure, it is advisable to keep the fan running regardless of the amount of
input power. Microwave energy is fed into the port 1 which is then coupled to the port 2.
This port is connected to a Narda 3003-30 (-30db) directional coupler (8), which then

delivers power to the cavity applicator via semi rigid, high power 50 Ohm coaxial cable. A

30

fraction of input power is sampled from the incident wave by directional coupler (8) and is
fed into a HP-8481A power sensor (3) which is connected to HP-435A (2) power meter.
To prevent any excess power into the power sensor, an attenuator (7) with 21.5db rating
is placed in between the directional coupler and the power sensor.

In case of a mismatch, a portion of incident power is reflected back through port 2
to port 3 and is then absorbed via 50 ohm coaxial cable in a Bell 8201, 500 Watt and 50
Ohm oil cooled dummy load (9). Similarly, as described earlier, a fraction of reflected
power is sampled via a directional coupler (6), a 19.5db attenuator (5), a power sensor (3)

and a power meter (2).

2.2.2 Vacuum and gas flow system

The microwave reactor is initially evacuated from atmospheric pressure to
20mTorr (approximately) with the help of a 33m3/hr mechanical pump. Once the chamber
pressure is below 20mTorr, a diffusion pump shares the pumping. This is a NRC - R121
type water- cooled diffusion pump whose fluid charge capacity is 500cc with a rating of
2500 liter/sec and is filled with Krytox oil. Chamber pressure from atmospheric to
lOOmTorr was measured by a thermocouple pressure gauge. A baratron gauge measured
the pressure from lOOmTorr to 0.1mTorr, and an extremely sensitive ion gauge recorded
pressure in the 1.0 - 0.001 mTorr range. Users should not forget to zero the baratron
gauge before carrying out any experiments. Once the reactor is evacuated to a low
pressure (in mTorr range), the baratron gauge is zeroed by shutting off the gas flow and

pressing the zero touch key for a few seconds on MKS 651 pressure controller.

31

Through out this research work 99.99% pure argon gas has been used to ignite the plasma
discharge. This noble gas discharge is far less complex than the other gases due to latter’s
intricate transport and collisional phenomenon. Argon gas is supplied to the processing
chamber from a pressurized cylindrical gas tank through a 0.25” corrugated flexible steel
tubing via MKS 1159 mass flow controller. This is a thermal type flow controller used for
high flow applications with range of 0.1sccm to 3051m and has a full-scale accuracy of _+_
1%. A digital four channel MKS - 247C mass flow controller readout is provided with the
system Flow rate set points can be adjusted either from the mass flow panel controls or
remotely through the rear panel analogue interface. It also features a front panel selectable
gas ratio control mode allowing the user to change (in multi-gas system) the total flow
while holding some gases in a fixed ratio. Argon gas from the flow controller is then fed to

the steel base plate, which has eight equally, spaced 1/64” diameter gas outlets.

32

 

 

 

 

 

1) Microwave power supply
2) PIP—435A power meter
3) HP-8481A power sensor
4) Coaxial circulator

5) 19.5db attenuator

 

 

 

 

 

 

 

To Cavity Applicator

6) 20db directional coupler
7) 21.5db attenuator

8) 30db directional coupler
9) 50 Ohm dummy load

Figure 2.2 Microwave power supply and power monitoring circuit.

33

 

 

 

 

2.3 The Microwave Reactor

The microwave reactor is defined by two assemblies: 1) plasma source and 2)
process chamber. The excitation geometry, design and critical dimensions of these

assemblies are determined by the process application.

2.3.1 The plasma source

The plasma source, as defined earlier in Chapter one, comprises of three sub-
assemblies. Figure 2.3 describes, schematically, the application specific plasma sources
used in this research work. These sources employing various designs in terms of applicator
and discharge chamber sizes, ECR magnet configuration and input gas feed mechanism
have been evaluated at MSU.

The stainless steel base plate (1) which sits on the top plate of process chamber
(discussed later in Section 2.3.2): 1) hosts the input gas feed channels, and 2) provides a
placement platform for the discharge chamber (2). Four 1/16” studs (3) are also welded to
the stainless steel base plate. These studs secure the brass ECR magnet keeper assembly
(5) and the cavity applicator (11) to the stainless steel base plate (1). The brass magnet
keeper assembly sits on the stainless steel base plate via guide post holes (6). This
assembly is comprised of two sub-assemblies: 1) ECR magnets and soft iron magnet
keeper ring (5b), and 2) ECR magnet container (5a). The ECR magnet container, which is
made up of brass, accommodates air and water cooling systems (7 and 8) for the ECR

magnets. The cylindrical ring of ECR magnets and soft iron magnet keeper rests in an

34

annular grove cut in the ECR magnet container. The bottom end plate adapter (9)
provides the electrical contact through the finger stock (10), to both physically and
electrically connect the side and end feed cavity applicators to the base plate assembly (1)
and (5). The plasma source is finally assembled by securing the cavity applicator shell (11)

with the stainless steel base plate studs using the thumbscrews.

35

 

l I
I r
’I

I"

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5b
fi‘ 3
8
;,_......__———.le .
- . - - .5
7 ‘ % /% / J\7
\7/

 

 

 

 

 

 

   
 

 

 

 

E .. gr _ ;
4
1) Stain less steel base plate 6) Guide post holes
2) Quartz discharge chamber 7) Air and water cooling inlets pipes
3) 1/ 16” studs 8) Air cooling inlet holes
4) Operating gas inlet pipe 9) Bottom end plate adapter
5) Brass ECR magnet keeper assembly 10) Finger stock
5a) ECR magnet container 11) The cavity applicator shell

5b) ECR magnets and soft iron magnet keeper ring

Figure 2.3 MPDR ion and plasma source - Exploded view

36

2.3.1.1 The cavity applicator and impedance matching network

The ECR ion and plasma sources in different designs employ either external or
internal microwave impedance matching networks. The external tuning mechanism usually
adopts single or multiple stub tuners, while an internally tuned MPDR has a self-contained
impedance matching mechanism The three MPDR sources investigated in this work are
internally tuned. These MPDR ion and plasma sources are described in the following sub-

sections.

2.3.1.1.1 MPDR 7- Side Feed (SF)

2.3.1.1.l.1 Brief History:

In sequel to the design improvements in ECR sources at MSU, M. Dahimene was
able to extract an 30mA, lOOOV, 3.2cm diameter, argon ion beam from a 9.4 cm diameter
ECR ion source [17]. Low extraction current and high power costs of 8333 Watt/Beam
Ampere were reported for this source. A reason for this was that only 10% of discharge
area was utilized for ion beam extraction. Two approaches were available to L. Mahoney
to reduce the power cost: 1) utilize the entire bottom of the plasma source to extract a
broad ion beam, and 2) reduce the discharge diameter. Since his research was directed
towards the design and characterization of compact ion sources [8], Mahoney adopted the
latter approach to improve the source efficiency. His objectives were: 1) to determine if a

5cm diameter discharge can be maintained with modest amount of input power (around

37

200Watt), and 2) to design a compact, simple, reliable, and long life ion source. His first
prototype MPDR ion and plasma source was a 17.8cm diameter brass cylindrical cavity
with 4.9cm discharge. This source, which is now recognized as the MPDR 5, employed
side feed microwave excitation mechanism Using extraction grids, ion current densities
exceeded lOmA/cm2 with less than 200Watt of input power. In his second prototype,
certain aspects of preliminary design were further improved which made the cavity more
reliable, less expensive and simple to operate. A second prototype of MPDR 5 had a
reduced cavity diameter of 8.9cm which supported the propagation of a single TE”;
electromagnetic mode. A modified MPDR 5 ion and plasma source has been used in this
thesis research. The modification included the design and fabrication of a new brass
bottom end plate that fits the cavity applicator to a 7cm diameter discharge chamber.
Figure 2.4 displays the geometrical design of this adapter. In this thesis this source is now
referred as MPDR 7 Side Feed (SF). Earlier M. Perrin used this 7cm diameter discharge

chamber while evaluating MPDR 7 End Feed Loop (EFL) ion and plasma source [14].

2.3.1.1.1.2 Construction and operation

Referring to the Figure 1.6 (in Chapter one), the microwave applicator consists of
a cylindrical brass shell (3), whose inner diameter is 8.9cm and outer diameter is 9.5cm,
and a sliding short (2). The height of cavity, Ls, varies from 8.5 - 13.7cm as the sliding
short top end plate (2) is adjusted. The lower limit is imposed by the presence of the input
microwave excitation antenna, shown at (I) = 0° in Figure 1.6. The applicator adopts a side
feed excitation configuration by coupling microwave energy from its side-wall through a

probe antenna (4). This side feed mechanism is affixed at H = 5.6cm, where the reference

38

plane H = 0 is the base of the cavity applicator. The excitation probe’s junction with the
cavity is taken as a (D = 0° reference point for other circumferential orientations along the
cylinder. The probe antenna depth, Lp, is made adjustable by using a gearing arrangement
that can extend the probe a maximum of 1.9cm into the cavity. This internally tuned cavity
matches the microwave power source to the unknown plasma load impedance by adjusting
the positions of sliding short top end plate, Ls, and the excitation probe depth, LP. On its
base, the cylindrical cavity applicator sets on the bottom end plate (7) (modified version)
and is secured to the brass magnet keeper plate (see (5) in Figure 2.3). Top and bottom
end plates consist of finger stock (1) which tightly secures the bottom end plate assembly
thereby preventing any microwave leaks. A 2.54cm by 2.36cm viewing window (5) at (I) =
90° is also provided for visual inspection of plasma discharge. The cavity shell is supplied
with both vertical and horizontal arrays of electromagnetic field sampling probe holes.
Three horizontal sampling probe hole arrays (9) run all along the circumference of the
shell and are located at H = 4.6, 8.6 and 10.6cm Each sampling probe hole is 0.95cm
center to center apart and are located in 10° increments all the way upto 360°. Similar set
of vertical probe arrays (8) are located at (I) = 180° and 270°. A total of ten sampling
holes are provided with each vertical array. The first sampling probe hole is drilled at H =
1.12cm and the holes are separated by 0.95cm The diameter of these sampling probe
holes is 2.9mm Since this diameter is very small as compared to the signal wavelength
(12.25cm), microwave energy does not radiate from these holes.

Once the discharge is ignited, sliding short top plate and the excitation probe are

adjusted to minimize the reflected power and tune the cavity to the resonant mode.

39

5%.... as; e5 83.2. Le :38: an 2%...

 

mug... 803.36% 8.”..—

Bum> 03m _

 

:8 e323 .5 a
Low undo: 3:836

Maud:

40

6030:. S 9.3
3222.... a: .z
«qu

 

ESSSaaeeeafiom

2.3.1.1.2 MPDR 7 End Feed Loop (EFL)

2.3.1.1.2.1 Construction

Refer to Figure 1.7 in Chapter 1. This 9.8cm inner diameter cylindrical stainless
steel microwave cavity was designed by L. Mahoney at IBM Watson Research Center and
was later donated to MSU. Prior to this work, M. Perrin had evaluated the MPDR 7 -
EFL for its electromagnetic coupling behavior [15].

M. Perrin demonstrated that this cavity by adjusting the internal matching network
can be tuned to either a TE ,1 or a TM mode. Within the cavity applicator, the TMm mode
is excited for a low sliding short length (L,) while the TEm mode is excited for a large Ls
value. The depth of the excitation loop antenna (LP), which is made up of 0.48cm diameter
copper rod, can be varied from 1.5cm to 5.5cm inside the cavity shell. This loop is secured
at the sliding short top end plate by using two sets of finger stock. Horizontal and vertical
sampling hole arrays are supplied to measure the electric fields in the cavity. This source
can be operated over an input power range of 50—300Watt while maintaining a stable and
uniform plasma discharge. Reference [15] provides additional detailed information on the

cavity dimensions, construction and operation.

41

2.3.1.1.3 MPDR 7 End Feed Probe (EFP)

2.3.1.1.3.1 Construction

Referrring to Figure 2.5 and Figure 1.8 in Chapter One, the MPDR 7 —EFL
applicator can be transformed into end feed probe configuration by replacing the sliding
short top end plate assembly with a similar sliding short arrangement hosting an excitation
antenna probe (Figure 2.5). This sliding short top end plate assembly was designed by M.
Perrin and has been fabricated with technical perfection in Physics Department Workshop
at MSU. Thus a 9.8cm stainless cylindrical shell can be operated with either an end feed
loop or probe excitation configuration. The excitation probe antenna is made up of a

.48cm diameter copper rod which can extend 5.7 cm inside the cavity.

42

 

2

1) Finger stocks 4) Support rods

2) Sliding short top plate 5) Brass probe holder

3) Excitation probe 6) Sliding short top plate adjustment screw

Figure 2.5 End feed probe assembly

43

2.3.1.2 Quartz discharge chamber

Refer to Figures 2.3 and 2.6. The excitation or the discharge region is bounded by
the quartz discharge chamber. A 3.5cm in diameter and 5.6 cm high quartz (2) discharge
chamber (bell jar) contains the plasma discharge. This dome shaped bell jar sits on the
steel base plate (1) and is sealed by a flat, circular, ring shaped vinyl gasket and a Viton
‘O’ ring. The plasma after been ignited in the bell jar diffuses out from its opening toward
the process chamber (16). The ECR magnet ring (12), that surrounds the bell jar, prevents

ion loss to the quartz’s walls.

2.3.1.3 Brass ECR magnet keeper assembly

The brass ECR magnet keeper assembly is shown as 5 in Figure 2.3. A more
detailed sketch is given in Figure 2.7. The brass ECR magnet keeper assembly (5) which
functions: 1) to house eight alternating pole ECR permanent arc magnets (12), and 2) to
prevent heating up of magnets by providing air and water-cooling channels (see 7 and 14
in Figures 2.3 and 2.6 respectively). This 3.17cm high, ring shaped brass ECR magnet
keeper assenme has a 16.5cm outer diameter and 7.6cm inner diameter as shown in
Figure 2.7. This assembly surrounds the 7cm quartz discharge chamber.

The rare-earth magnets employed for electron cyclotron resonance cavity are
sensitive to high temperature and thus require protection from heat exposure. This is

achieved through air and water cooling of the brass magnet keeper assembly. Compressed

air is fed to the magnet keeper through two brass tubes (22) attached to the either sides of
the keeper. The air runs across the annular channels within the keeper and is let out to
cool off the outer surface of quartz bell jar through eight small pinholes (8) (see Figure
2.3).

Due to the presence of magnetic fields from the ECR magnets, charged particles
oscillate around the magnetic field lines at the cyclotron frequency. Once the cyclotron
frequency reaches the applied microwave frequency, the phenomenon of electron
cyclotron resonance takes place. This results in the maximum transfer of microwave
energy to the discharge. For the 2.45 GHz input microwave frequency 875 Gauss of
magnetic field is required to produce electron cyclotron resonance. This is governed by
the relation:

B = me (o /e
where B is the magnitude of magnetic field, me is the mass of electron, e is the magnitude
of charge and (0 represents the angular frequency.
As shown in Figure 2.7, eight are shaped 3.7 x 0.8 x 2cm alternating pole neodyrnium—
iron-boron magnets (12) are contained in the soft iron magnet keeper and they produce an
ECR zone (23) lying approximately 1cm from the magnet face. Magnetic field intensity of
the ring magnets decreases from maximum at the pole face to zero at the center axis (at r
= 0). These magnets are glued together to form a ring around the bell jar. This magnet
ring is enclosed in a soft iron magnet keeper (13) that serves to confine the magnetic
fields. The soft iron magnet keeper prevents downstream leakage of the magnetic fields

into the process chamber.

45

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

12) 8 Alternating pole ECR arc magnets
13) Soft iron magnet keeper

14) Air and water cooling channels

15) Argon gas flow channel

16) Process chamber

K‘?

18

 

17) Probe holder assenme

l8) Vacuum feed through connector
19) Optical windows

20) Viewing port

Figure 2.6 Process and discharge chamber

46

21 21

n-. .
---------
........

.....
..................
oooooo
.71!

nnnnn
.........

    
    
 
   

 

   

.-.-I-.-I-.-.-

Q
Q I
ha-n-o-o-acnd

 

nnnnnn
..
---------

12

Magnified view Of a
single magnet piece

 

 

 

 

21) Inlet for water-cooling.
22) Inlet for air cooling
23) 875 Gauss ECR zone

Figure 2.7 Brass ECR magnet keeper assembly

47

2.3.2 Process chamber

As shown in Figure 2.6, the 100 liter capacity processing chamber (16) is a
stainless steel cylinder with 1.6cm thick walls. Height and diameter Of cylinder are 26 and
41.8cm respectively. The top of the chamber is covered by a 1.7cm thick, and 50.5cm
diameter steel plate (top plate). The center of this plate has a 12.32cm diameter annular
aperture. The chamber and the top plate are sealed with a 0.66cm thick Viton ‘O’ ring.
This ring needs to be greased every time the plate is removed from chamber using a Dow
Corning high vacuum silicone grease. At the annular aperture of chamber top plate sits a
stainless steel base plate (1) that houses the input gas inlets. A Viton ‘O’ ring which rests
in a grove cut at the bottom of stainless steel base plate serves to seal the base plate (5)
and the chamber top plate. A steel plate with the dimensions similar to the process
chamber top plate is placed at the bottom of the vacuum chamber (bottom plate). This
plate houses: 1) a vacuum feed through (18), that provides electrical contacts for the
plasma diagnostic equipment, 2) a Langmuir probe holder (17), which can be moved in
radial direction across the Opening face of quartz bell jar (2), and 3) a 6.3cm diameter
annular aperture, in its centre, which is used as a viewing port (20). Two optical windows
(19) both of them 90° apart and 5.9cm in diameter are located on the process chamber

side-walls which can be utilized in the plasma spectroscopic investigations.

48

Chapter 3

Experimental techniques

3.1 Introduction

One of the basic problem in the plasma physics and the plasma based engineering
applications is the determination of plasma parameters. These plasma parameters, for
example include electron temperature and electron density, plasma uniformity, energy
distribution functions of species etc. The goal is to gain quantitative information about the
desired plasma parameters by putting together the available analytical techniques. These
plasma diagnostic techniques, for example comprise microwave and X-ray diagnosis,
photometry, magnetic and electric field measuring probes, mass spectrometry and
spectroscopy etc.

In order to extract useful intelligence, the diagnostic methods should not interfere
with the plasma itself which should remain unperturbed. This is the guiding principle in the
plasma investigations. Before adopting a certain approach, the experimenter should
ascertain that the principle is not violated. The plasma diagnostic techniques used in this
research work are: 1) single and double Langmuir probe based investigations, which are
described in the Section 3.2, and 2) The impressed electric field measurement technique

using E-field probes, which is explained in the Section 3.3.

49

3.2 Langmuir probes

3.2.1 Introduction

Langmuir and his coworkers first used this probe technique in early 1920’s [18]
and since then it has become a one of the most useful plasma diagnostic technique. These
probes being simple in construction and operation are now being widely used to study
laboratory-generated plasmas and space plasma investigations.

In this research, electron density, an estimate of electron temperature and ion
saturation current were measured using double Langmuir probes for the MPDR 7 — Side
Feed (SF), MPDR 7 — End Feed Loop (EFL) and MPDR 7 — End Feed Probe (EFP) ion
and plasma sources. Using the single Langmuir probes, the electron energy distribution
functions (EEDF) including Maxwellian and Druyvesteyn distribution functions, plasma
potential and electron temperature were recorded for the MPDR 7 — Side Feed (SF)

plasma source. These measurements were made under different input conditions.

3.2.2 Double Langmuir probes

3.2.2.1 Probe design

Figure 3.1 displays the design of the double Langmuir probe used in this research
work. These probes were fabricated in the glass working shop of the chemistry department

at MSU.
The probe collecting material should obviate: 1) chemical reactions with the

plasma, 2) sputtering, 3) heating up and 4) secondary emission caused by the impact

50

ionization [19]. Tungsten fulfills the above criterion and is widely used in the probe
construction material. The double Langmuir probe arrangement comprises of two tungsten
rods (1), 0.094cm in diameter, which are sealed inside a glass tube (2). A portion of the
rods, which is 0.56cm in length, is exposed to the plasma from one end of the glass tube.
Within the tube, the other end of the tungsten rod is connected to the copper connecting
wires (4). These copper connectors are then crirnped (5) to the electrical wires (6) which
make an electrical contact with the external measuring circuit via vacuum feed through
(18). The glass tube, in order to electrically isolate the electrodes from the highly reactive
plasma species, is sealed at its base using a shrink—wrap. In addition, this shrink-wrap
mechanically secures the floating structure at the base of the glass tube. Within the glass
tube, the tungsten electrodes are provided additional isolation by enclosing them in the
glass capillary tube (3). Extreme care should be taken once the tungsten rods are being
sealed in the glass tube. Unequal exposure areas of the electrodes give off-centered I-V
characteristic curves causing incorrect interpretation of plasma parameters.

Maintaining a minimum tolerable spacing between the two electrodes is yet
another important consideration in the probe design. This spacing is primarily determined
by calculating the electron Debye length in the discharge. A detailed treatment on the
subject is given in the reference [19]. Electron Debye length is given by the relation:

to = 69 (Tammy
where T. and II. are the electron temperature in degree Kelvin and electron density per m3
respectively. For an argon gas discharge, assuming electron temperature and electron
density to be 6eV and 1 x 10”/ cm3 respectively, the Debye length is 50pm. The sheaths

surrounding the two electrodes will overlap if the electrode spacing is less than three or

51

four Debye lengths. As a rule of thumb, for any plasma diagnostic setup, the adjacent
probes should be kept four Debye lengths apart to avoid any false readings. These
electrodes, in our case, are separated by a distance of 0.22cm and thus fulfill the criterion

of being several Debye lengths away.

52

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1) Tungsten electrode 4) Copper connector To measuring
2) Glass tube 5) Crimps circuit
3) Capillary tube 6) Electrical wires

Figure 3.1 Double Langmuir probe experimental setup

53

3.2.2.2 Principle of operation

The double Langmuir probe technique was first used by E. Johnson and L. Malter
[20] in 1949. This technique employs a very simple method and permits the investigation
of those plasma’s where a reference grounded electrode is missing. The Langmuir probe
electrodes are connected across a variable voltage supply and current through them is
measured as a function of applied voltage. Figure 3.2 shows the measuring circuit
diagram The dc. bias across the two probe electrodes (1) was varied from -50 to +50
Volt using a programmable HP 6634A dc. power supply (4). The output from the
variable voltage source can be obtained either locally from the front panel or remotely
over the HP IB (which is interfaced with a PC). Voltage difference (Vd) across the two
probes is taken as:

Vd = Vpl - sz,

where VP] and sz are the voltages appearing at the probes l and 2. The probe current was
recorded on a HP 3478A digital multi-meter that can also be programmed remotely over
the HP IB. As the probe dc. bias was varied from -50 to +50 Volt (starting at —50 Volt
and incrementing in 1Volt steps through to +50 Volt), the average probe current was
sampled 20 times and the reading was recorded in the computer. This was executed by
using a computer program written in Q-Basic. A typical I-V trace is shown in Figure 3.3,

detailed explanation of this characteristic curve can be found in the reference [14].

54

Using the Kirchoff’s current law, which states that the current flowing into a node point

equals the current flowing out of it, we get:

iil'iel=ie2'i12=1p (3-1)
Where the subscripts “i” and “e” represents ion and electron currents collected by the

probes 1 and 2, and Ip is the total probe current. Assuming the Maxwellian distribution,

the current collected by both the probes is given by:

ie1 = Ieol CXP('eVp1/ kb Te) (32)
and
1.2 =1... exP(-6Vp2/kae) (3.3)

Where 1.01 and 1.02 are the electron currents once voltages across both the probes are zero
1.6. Vprobel and mebez = 0.

By dividing the equation (3.2) with equation (3.3) we get:

ie1 / ie2 = Ieol [[602 CXP‘“ Vp2 ' Vpl) 6/ kb Tel (3-4)
or
iel / ie2 =1eol /Ieo2 exp ('evd/ kb Te) (3-5)

The equation (3.1) :>

tut2=Ip/i.2-1
Using the equation (3.5) we get:-

Ip/ iez - 1 = Ice, / 1,02 exp (-eVd/ kb T.) (3.6)
By taking natural logarithm of equation (3.6) we have:-

111 Up] ie2 " 1) = 1n(Ieol /Ie02) " eVd/ kb Te (37)

55

The electron temperature, T,, can be obtained from the equation (3.7) which is
proportional to the slope of In (Ip/ iez - 1) vs. Vd
From Bohm sheath criterion [21], ion saturation current is given by:

Ii = 0.606 Ape n, (kn/Mo”,
where e is the electron charge in coulombs, n, is the ion density in cm'3, M, is the mass of
the argon ion = 40 x mass of proton = 40 x 1.67x10‘27 kg. Ap is the probe area and is
given by:-

Ap = 2m (probe height + 1t1'2),
where the probe height, in our case, is 0.56cm with a radius of 0.047cm giving a total
probe area of 0.1723cm2. Finally the ion current collected by the two probes was averaged

to give the ion saturation current ( I, ).

56

 

 

 

 

 

 

 

/.

 

_= 52's 0

 

 

BBB 12>

 

 

 

 

 

Analogue To
Digital
Converter

 

 

ii/

 

 

1) Double Langmuir probe

2) HP 3478A Digital voltmeter

 

E6 A 333

 

 

 

 

3) HP 3478A Digital ammeter

4) HP 6634A dc. power supply

Figure 3.2 Schematic of double Langmuir probe measuring circuit

57

 

 

 

 

 

 

 

 

 

 

*5 *~0.2‘
g I
8 l7 f I g l p I 1
§-5o -30 -1o 1.1 10 I 30 50
E . . .......... .Q- l
.4- v » .
' lil
.......... -0.6“'
-o.8—

Probe DC Bias in Volts

 

Figure 3.3 A typical double Langmuir probe I -V characteristic curve

Incident power 150Watt
Chamber pressure 2mTorr

Reflected power 2Watt
Flow rate 20sccm

58

 

3.2.3 Single Langmuir probe

3.2.3.1 Probe design

The probe design is similar to that of a double probe arrangement and the only
difference being the use Of a single electrode. The length of probe exposed to plasma in

this case was 0.56cm with a diameter of 0.051cm
3.2.3.2 Principle of operation

Druyvesteyn [22] in 1930 reported that the electron energy distribution function
(EEDF), f(E), is directly proportional to the second derivative of the Langmuir probe I-V
characteristic curve, i.e.

f(E) = (v, - V)"2 d21./ dv2
where Vs is the plasma potential, V is the voltage applied to the probe and I, is the current
drawn by the probe. It should be noted that the above relation is not restricted to the
Maxwellian distribution of charges.

Normalizing this distribution function to unity we get:-

Tf(E)dE = 1
o

59

Once electron energy distribution function f(E) is known, average electron energies can be

determined by:-

<E>=IE f(E)dE=1
0

The second derivative of the probe current, dzle / de, can be determined by
superimposing a small modulating voltage on the single Langmuir probe and then
measuring the second harmonic of the probe current.

The circuit diagram for the single probe diagnostic set up is shown in Figure 3.4.
The amplitude of 1 kHz applied a.c. signal Obtained from a HP 651B test oscillator is kept
at 0.35Volts rrns which is then superimposed on the dc. voltage. A 128A EG&E amplifier
locks on twice the frequency of the sine wave since the second derivative dZL/ de is
proportional to the amplitude of the current with a frequency 20). The variable HP 6634A
dc. voltage supply was used to vary the bias from —20 to 50Volt.
The probe current is represented by the relation:-

I = I(V + vsincot)
This shows that the current collected by probe is a function of both applied do and ac.
voltages. Where V is the dc. input and v is the amplitude of ac. signal with a frequency

Cl).

60

Using Taylor series expansion we have:-
I = I(V) + v sincot dI / dV + {(v sincot)2/ 2! }d21/ dv2 + {(v sinoot)3/ 3! }d3I / dV3

+ {(v sintnt)"/4! }cl“1/dv4 + {(v sincot)5/ 5! }d51/dV5 + {(v sinoot)6/ 6! }d°I/

Simplifying the above relation

I = I(V) + v sincot dI / dV + v2/2{(1- cos2tnt)/ 2 }d21 / clv2

+ v3/6{(3/4 sintnt - sin 3cot/4)}d31 / dv3 + ------------
after rearranging

I = I(V) + v sintnt dI / dV + (v2/4 - v2/4 cos2tnt) c121 / dv2

+ (v3/8 sintnt - v3/24 sin3cot) d3I / dv3 + ------------
and finally we get:-

I = I(V) + (v2/4 c121 / clv2 + ----- ) + (v (11 / dV + v3/8 d3I / dv3 + ----)sincot

+ (v2/4 c121 / dv2 +—---)(-cosZ(nt) + (v3/24 d3I / dv3 + ------ )(-sin3(nt) + -------
As the amplitude of the applied a.c. signal is kept small we can neglect higher order terms
in the above equation.

Unless otherwise stated, the probe was positioned vertically and its electrode tip
was located at z = 3cm plane, see Figure 3.1. Over the years, the process chamber of the
plasma processing machine used in these experiments became contaminated with a thin
insulating layer. This insulating film might have formed due to the suction of diffusion
pump oil in the process chamber under the low-pressure conditions. Consequently the
chamber walls turned out to be a bad ground. This uncertainty was overcome by using the
processing chamber top plate, which is in the close proximity of the plasma discharge as a

reference ground potential. Figure 3.5 displays a typical single Langmuir probe I-V

61

characteristic curve. It is noted that the saturation current appears to be missing. This may
be attributed to a large current drawn by the probe which disturbs the discharge uniformity
[23]. Figure 3.6 shows a dZIe/ de plot, here the maximum value of dzlel de corresponds

to the plasma potential.

62

c121 / dv2

 

 

 

 

 

 

 

 

 

 

 

 

Process
chm” Lock-in
amplifier
f‘f
1:1 Audio
transformer\
L— Ammeter M
Analogue DC power
To digital supply
converter

 

Reference signal

 

 

Oscillator

 

 

 

 

 

 

 

 

Figure 3.4 Schematic circuit diagram of single Langmuir probe measurement

technique

63

 

 

Probe Current

 

 

 

 

 

 

 

 

 

 

l I I I

-50 -30 -10 1O 30 50
Probe DC Bias in Volts

 

Figure 3.5 A typical single Langmuir probe I -V curve

Incident power 150Watt Reflected power 2Watt
Chamber pressure 2mTorr Flow rate 203ccm

 

 

 

12(1)-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

< 1(XI)
E
.E 800
one
5 1
g 600
o i
3 400
2
“- 200 ,
0 I T V
-30 -20 -10 0 10 20 30
Probe DC Bias in Volts
Figure 3.6 A typical single Langmuir probe second derivative curve
Incident power 150Watt Reflected power 2Watt
Chamber pressure 2mTorr Flow rate 208ccm

65

 

3.3 Electric field measurements

3.3.1 Introduction

Measurement of electric fields inside the microwave cavity applicator is an
important aspect once the ion and plasma sources are characterized. Determination of the
fields inside a plasma-loaded cavity explains certain significant parameters such as
microwave coupling efficiency, and identification and verification of the propagating
electromagnetic modes. These measurements were performed only for the MPDR 7 - SF
ion and plasma source.

The critical dimensions of the side feed cavity applicator makes it possible to
support the propagation of a single electromagnetic mode, which has been identified by
Mahoney [9] as the TEm mode. The impressed electric field measurement technique
employed here is similar to the one used by the earlier researchers at MSU. The normal
component of the electric field, which is incident on the cavity walls, is measured with a
coaxial probe commonly known as the E-field probe. The construction and the principle Of

operation for these probes is described in the following sub-sections.
3.3.2 E - field probe design

Figure 3.7 displays the E-field probe design and the experimental set-up. A semi-
rigid copper coaxial conductor (1) was used in the probe fabrication. The diameters of the
outer (2) and inner conductors (3) were 0.216cm and 0.0254cm respectively. Three

probes were fabricated, each using a 4cm length of the copper coaxial conductor (1). The

66

coaxial conductor was soldered to a SMA connector (5) on one end, and its other end was
kept exposed. The E-field probes made for the experiments performed in this research
work are named as flush, 0.5mm and 1mm probes. These names reflect the depth of the
center conductor, once the field probe is inserted in the cavity applicator (9) through the
E- field sampling holes ( 10 and 11), with respect to the applicator’s inner wall. For
example, the inner conductor in the flush probe has a zero extension in the cavity whereas
0.5mm probe extends 0.5mm In order to obtain the desired length of the inner conductor
during the fabrication process, the exposed end was very carefully sliced to avoid any
burrs and rough edges around the rim The limited extension of the center conductor
within the cavity and its diameter itself evades any chances of perturbing the fields. During
the course of experimentation the sampled power never exceeded 0.006% of the input
power thus satisfying the need of maintaining unperturbed fields. A brass collar (6) is
attached to the probe which: 1) helps inserting the probe in the sampling holes and 2)
maintains the desired extension of center conductor in the cavity. The outer conductor (2),
when inserted inside the sampling hole, provides the electrical contact with the cavity
walls. The sampling pinholes often get oxidized thereby deteriorating the electrical
contact. It is therefore advisable to scrub them, ideally by a wire brush. The E-field probe
is connected to a HP 8481A thermistor (7) whose maximum power rating is 300mWatts.
Finally the power is sampled using a HP 435A power meter (8). The microwave cavity
applicator tends to heat up by absorbing the incident microwave power. This factor would
result in erroneous readings if the power sensor, which is temperature sensitive, is used in
the close proximity of the cavity applicator. The user should therefore keep the thermistor

at a distance to prevent any measurement errors.

67

 

\:e°

 

 

 

 

 

 

 

 

 

 

 

 

II-.
7
9
o o . . 10
O 0
0 O

11

 

Figure 3.7 E-field probe design and electric field measurement setup

1) Rigid copper coaxial cable 7) HP 8481A thermistor

2) Outer conductor 8) HP 435A power meter

3) Inner conductor 9) Cavity applicator shell

4) Dielectric 10) Vertical sampling holes
5) SMA connector 11) Horizontal sampling holes

6) Brass collar

68

3.3.3 Principles of operation - Burkhart’s method

S. Burkhart in 1985 developed a technique by using an open ended coaxial probe
(E-field probe) for the measurements of high power microwaves [24]. The probe insertion
no more than flush with the cavity inner walls has an advantage of presenting no
discontinuity to the incident microwaves. The field measurements were carried out once
the cavity was operating under fully matched conditions i.e. with minimum reflected
power (always less than 5% of the Pine). The normal electric field component on the E -
field probe inner conductor and is given by [24]:-

E, = [2P/ 201‘” x [KatazeoclII‘ (V/m),
where P is the power sampled with the E—field probe, 20 is the characteristic impedance of
the cable, which in our case is 500, “a” is the center conductor radius (0.0254cm), so is
the permitivity of free space, and (n is the frequency of the oscillating fields (2.45GHz).
The area multiplier Kr for a flush probe is equal to 3.846. By substituting the values of all
the constants the above relation reduces to:

E, = (1.88 x 106) P"2 (V/m)

This relationship was employed in this thesis research to measure the electric field strength

in the microwave cavity shell.

69

Chapter 4

Experimental Results

4.1 Introduction

As mentioned earlier three MPDR plasma sources are characterized in this thesis.
In this chapter the MPDR 7 - Side Feed (SF) is extensively evaluated against a number of
input variables and then MPDR 7 — End Feed Probe (EFP) and MPDR 7 - End Feed Loop
(EFL) plasma source output performances are compared to the performance of the MPDR
7 - Side Feed (SF) plasma source. Finally, the experimental results are compared with the
global plasma model. Thus in a broad sense this chapter is divided into two parts. The
first part describes the experimental results of the measured internal variables (X) for the
SF plasma source. Then part two compares: 1) the output performance of the three
MPDR plasma sources and 2) the experimental results with the predictions from global
plasma model.

The input variables parameter space (U) for the three plasma sources has already
been given in Tablesl.2 — 1.4 in Section 1.5.3 of Chapter One. A selected group of
internal and the output variables were measured versus changes in the input variables U1.
Tables 4.1 - 4.3 display the internal variables (X) and the output variables (Y) measured
for the MPDR plasma sources investigated in this study.

As seen from Table 4.1 the internal variables measured for the MPDR 7 — SF
plasma source were the power absorbed (Pabs), absolute and spatial electric fields, plasma

density, plasma potential, electron energy distribution functions (EEDF), electron

70

temperature and ion saturation current. The output variable (Y) that was measured for this
source was ion production cost.

For the end feed sources, both loop and probe excitation, the measured internal
variables were the power absorbed (Pam), plasma density and the ion saturation current.
Again the measured output variable was ion production cost.

From Tables 4.1 — 4.3 it can be seen that the macroscopic controllable input
variables space (U1) is almost identical for all of the three plasma sources. The difference
is in the input regimes of argon gas flow rate and absorbed power. Except for the method
of excitation the reactor geometry variables space (U2) for the three plasma sources are
the same. The side and end feed sources differ slightly in applicator diameter and the
method of coupling. The cavity applicator shell diameter of the end feed IVIPDR plasma
sources is 0.9cm larger than that of the side feed source. This larger diameter allows the
excitation of an additional propagating mode in the end feed sources. The cavity
applicator shell of the side and the end feed plasma sources is made of brass and stainless
steel respectively. For the three MPDR plasma sources it can be noted from the Tables
4.1 — 4.3 that the entire input variable space (U) is approximately constant except the
method of microwave excitation. Thus for a given input condition, keeping the reactor
geometry variables space (U2) fixed, the influence of the method of excitation on the
internal and output variables can be evaluated from the experiments presented in this

thesis.

71

 

 

 

 

 

 

 

 

 

 

 

Input Variables Description Experimental Range
(U)
Absorbed power Pabs 50-200 Watt
Operating frequency 2.45GHz (CW)
Macroscopic controllable Discharge gas 99.99% Argon
input variables Chamber pressure 1-8 mTorr
U1 Flow rate 0.3 - 358ccm
Cavity short length Ls 13.7 — 8.4cm
Depth of excitation probe 1.2 — 1.9cm
antenna Lp
Applicator size, type and 8.9cm diameter cylindrical
the method of excitation brass cavity with side feed
Fixed excitation
reactor geometry Discharge chamber Height 5.6cm, radius 3.5cm
variables dimensions with a total discharge
U2 volume of 215.5cm3

 

 

ECR magnet configuration

 

8 Pole 8 Magnets
(8P/8M)

 

ii

 

Power absorbed (Pm)

Plasma density
Plasma potential
EEDF

Electron temperature
Ion saturation current

 

Internal variables (X)

Absolute and spatial electric fields

11

 

 

Output variables (Y)

o Reactor performance figures of merit

Ion production cost

 

 

 

 

Table 4.1 Internal and output variables measured for MPDR 7 — SF

72

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Input Variables Description Experimental Range
(U)
Absorbed power Fab, 100- 150 Watt
Operating frequency 2.45GHz (CW)
Discharge gas 99.99% Argon
Macroscopic controllable Chamber pressure 1-8 mTorr
input variables Flow rate Fixed at 20 sccm
U1 Cavity short length Ls 3.13 - 10.78cm
Depth of excitation probe 0.97 — 3.82cm
antenna Lp
Applicator size, type and 9.8cm diameter cylindrical
the method of excitation stainless steel cavity with
end feed excitation
Fixed Discharge chamber Height 5.6cm, radius 3.5cm
reactor geometry dimensions with a total discharge
variables volume of 215.5cm3
U2 ECR magnet configuration 8 Pole 8 Magnets
(8P/8M)
Internal variables (X) Outputgiables (Y)

0 Plasma density

 

0 Power absorbed (P2,)

0 Ion saturation current

 

 

 

Reactor performance figures of merit

Ion production cost

 

 

Table 4.2 Internal and output variables measured for MPDR 7 - EFP

73

 

 

 

 

Input Variables Description Experimental Range
(U)
Absorbed power Fab, 100-150 Watt
Macroscopic controllable Operating fi'equency 2.45GHz (CW)

input variables

 

Discharge gas

99.99% Argon

 

 

 

 

 

 

 

 

Ul Chamber pressure 1-8 mTorr
Flow rate Fixed at 20 sccm
Cavity short length L, 4.575 — 8.075cm
Depth of excitation loop 1.95 — 3.2cm
antenna Lp
Applicator size, type and 9.8cm diameter cylindrical
the method of excitation stainless steel cavity with
end feed loop excitation
Fixed Discharge chamber Height 5.6cm, radius 3.5cm
reactor geometry dimensions with a total discharge
variables volume of 215 .5cm3
U2 ECR magnet configuration 8 Pole 8 Magnets

 

 

(8P/8M)

 

fl

1)

 

 

0 Plasma density

 

Internal variables (X)

0 Power absorbed (Pin)

0 Ion saturation current

 

 

 

Output variables (Y)

Reactor performance figures of merit

Ion production cost

 

 

Table 4.3 Internal and output variables measured for MPDR 7 - EFL

74

 

Part I

4.2 Electric field measurements and identification of EM excitation modes in the
MPDR 7 - SF plasma source

Measurement of the electric field radial component (13,) impressed on the cavity
walls is presented in this section. These measurements were made by sampling power
using an E-field probe. The sampled power was then used to calculate the electric field
strength by employing Burkart’s method [24]. This technique was discussed in Section
3.3 of Chapter 3. For all input conditions the measurements were made once the
discharge loaded cavity was tuned to the best possible match and the sampled data
obtained from the E- field probes was then plotted to identify the excitation modes.

Figures 4.1a and 4.1b display the measured strength of the radial electric field at
the cavity inner-conducting walls in the SF plasma source. This plot represents the field
strength for an absorbed power of 200Watt, with a chamber pressure of 4mTorr, and
20sccm of argon gas flow rate. The electric field strength which was measured at two
circumferential sampling probe hole arrays, at H = 4.3cm (low ring) and at H = 7.3cm
(high ring), is shown in the Figure 4.1a. It is seen that the mode pattern verifies the
presence of a TEm mode. The field strength is found out to be maximum at d) = 0°, 180°
and 350°. The sampling holes at d) = 0° and 350° are next to the location where the
excitation probe antenna is located (at H=5.6cm plane), and the sampling hole at (b = 180°
is located directly across from the antenna probe.

Figure 4.1b displays the variation of electric field strength versus the z - direction.

The data presented here is obtained from the vertical sampling holes located at (b = 180°

75

(see Figure 4.2). The maximum electric field strength was recorded at the sampling hole
located at H = 5.3cm This sampling hole is located directly across from the excitation
antenna probe. The radial component of the electric field is reduced as one moves axially
from H = 5.6cm plane (this plane defines the junction of the excitation antenna probe and
the cavity applicator shell) towards the sliding short top plate or towards the discharge.
This demonstrates the existence of a standing electromagnetic field inside the cavity
applicator. The electric field strength was very low and thus unrecordable along the entire
vertical sampling hole array located at d) = 270° (see Figure 4.1a). This was found out to
be true for all the input conditions.

Figure 4.2 displays the measured and the theoretical profiles of the electric field in
the plasma source. Figures 4.3 and 4.4 display the electric field strength recorded at a
chamber pressure of 8mTorr with two levels of absorbed power i.e. 200 and 150Watt. It
is noted from the Figures 4.1, 4.3 and 4.4 that the circumferential variations in the electric
field strength are the same for both low and high sampling rings. However in all the
cases, at d) = 180°, the field strength at the low sampling ring slightly exceeds that of the
high ring.

The increase in the absorbed power accounts for an increase in the electric field
strength in the cavity. This can be seen from the Figures 4.1, 4.3 and 4.4. Figure 4.5
summarizes this observation for the field strength at a fixed pressure of 8mTorr and three
different power levels. Figure 4.6 shows the pressure dependence of the impressed

electric field strength. Higher field intensities are recorded as the operating pressure is

reduced.

76

E‘: ‘

The observations made from the electric field measurements are:-

1)

2)

3)

A single mode is excited in the cavity and the spatial distribution of the
electric field indicates that the TE“ 1 mode is impressed on the discharge.

The excitation antenna probe does influence the measured electric field
patterns i.e. as seen from the Figures 4.1, 4.3 and 4.4 the impressed E-field is
higher at the probe plane, H = 5.6cm, and is also higher at the “ probe side” of
the cavity. This is probably caused by the evanescent probe “near fields”.

The loaded quality factor “Q” of the cavity is mathematically expressed as:-

Q. a EZIP...

where Fab, is the power absorbed in the cavity and E is the impressed electric
field. From the Figure 4.6 it is seen that for a constant input power, Pabs, the
impressed electric field strength in the cavity decreases as the chamber
pressure is increased. Thus the loaded quality factor “Q,” of the cavity
increases with decreasing pressure. On the other hand Figures 4.5 and 4.7
show that for a constant pressure, the electric field strength increases with the
increasing absorbed power. The quality factor in this case increases slightly
with increase in the input power. These results are summarized in the Table
4.4 by calculating the impressed electric field and finding out the ratio of
EZ/Pabs at a fixed location in the cavity applicator, d) = 170°, under various
input conditions. It should be noted here that the Burkart’s method is valid
only for the flush probe. The 1mm probe was thus calibrated with the flush

probe impressed electric fields at d) = 170°.

77

Based on this limited data following conclusions are drawn:
(a) Holding the chamber pressure constant and increasing the absorbed
power, increases the Q slightly.
(b) Holding absorbed power constant and decreasing the chamber
pressure, increases the Q.
These results require further investigations by measuring the quality factor

under a broader range of input conditions.

78

 

 

 

 

 

 

 

22 -, . . ........
20.F 3 0°. o. :F
16 Q ......... IIOO ...... .2...
E14. . .0 , y I .
c 1 I OI I0 I oH=4.3cm
"0‘ ' ‘ ’ Io ' o IH=7.3cm
u_J 3, I I . , .5 .o .
6* I I ,
4 . ..... . I. I .........
0 I I v I I Tm—Y——I
0 45 90 135 180 225 270 315 360
Circumferential angle w.r.t. input probe

 

 

 

Figure 4.1a Circumferential variance of electric field strength in
MPDR 7 - SF
Input power 200Watt, Pressure 4mTorr, and flow rate 20sccm
Data is obtained using flush probe

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Top ofthedischarge
chamber
1.86cm within the
cavity
:5
. e
3) . ....... 9. . .
25 1e .1
E 20 1 i ’ At180°
3 O . . e
.5 '
- Probeantennaat
1° ' ‘ ‘ ' ¢=0°,depth=l.15cm
0 . fl . . ' ; . . .' .
0 1.5 3 4.5 m6 7.5 9 10.5 12
Vert’calSanphgholelocaiortH'ncm

 

 

 

Figure 4.1b Longitudinal variance of electric field strength in
MPDR 7 - SF
For vertical sampling hole array located at (I) = 180°
Input power 200Watt, Pressure 4mTorr, and flow rate 20sccm
Data is obtained using flush probe

79

 

 

 

 

Excitation
Probe at
H=5.6cm
& ¢=0°

 

 

 

 

 

O
O
C
e g . g e 9
Dated llines Solid and Dashed
. =H—fie ds i lines.=E-fields
1
9 e e 30 0 '
¢=90°

 

 

Vertical Sampling
Holes at ¢=270°

 

 

 

 

Vertical Sampling
Holes at ¢=180°

 

Circumferential
Sampling Holes
. at Hfl3cm

 

 

 

Measured radial

electric field

patternonthe

cavity wall

 

Figure 4.2 Electric field theoretical and measured profile
within the MPDR 7 — SF ion source

80

 

 

 

 

 

 

Excitation
Probe at
H=5.6cm
& ¢=0°

 

 

 

 

_--—-- ' '------~--
- --
~~

”if: '11:” Solid and Dashed
- ' e s g lines=E-fields

. s .
. O o i o O .
i

¢=90°

 

 

Vertical Sampling
Holes at ¢=270°

 

 

 

 

Vertical Sampling
Holes at ¢=180°

 

 

H=7 .3cm

eeeeeeeeeeeeeeeee

Circumferential
Sampling Holes
. at H=4.3cm

 

 

 

Measured radial

electric field

pattern on the

cavity wall

 

 

Figure 4.2 Electric field theoretical and measured profile
within the MPDR 7 — SF ion source

80

———n‘ ‘-

 

 

 

22 -
20 ill.
18 - 0
16 ‘ 89 o o'-
E o 0'
§ 14 .0 I
I
’c‘ 1: I :0 .20 I o H=4.3cm
“—1 8 _ I o .0 I l H=7.3cm
6 - I' B
4 - I
2 .
0 . a . . . e .= .
o 45 90 135 180 225 270 315 360
Circumferential angle w.r.t. input probe

 

 

 

Figure 4.33 Circumferential variance of electric field strength
in MPDR 7 — SF
Input power 200Watt, Pressure 8mTorr, and flow rate 20sccm
Data is obtained using flush probe

 

 

Top of the discharge
chamber

1.86cm within the
cavity

 

 

 

IE.I in kV/m

 

Vertical Sampling hole location. H in cm

 

 

 

Figure 4.3b Longitudinal variance of electric field strength in
MPDR 7 — SF
For vertical sampling hole array located at <1) = 180°
Input power 200Watt, Pressure 8mTorr, and flow rate 205ccm
Data is obtained using flush probe

81

 

18
1:1 .5 00 3° .8
i” o

E
x
.9. el 0 in oI lH=7.3cm
1 6‘ I I 0I
y l I I

4 I °I

2i

o

 

04590135180225270315360
Circumferential angle w.r.t input probe

 

 

 

Figure 4.4a Circumferential variance of electric field strength in
MPDR 7 — SF
Input power 150Watt, Pressure 8mTorr, and flow rate 205ccm
Data is obtained using flush probe

 

 

Top of the discharge
chamber
1.86cm within the
cavity

 

 

L

 

m
25 ° . O

O

‘0 I Probe antenna at i
s O ¢=0°, depth=l.15cm

0
0 L5 3 4,5 7.5 9 ‘05
Vertical Sampling hole on, H in em

 

[-

 

 

 

Figure 4.4b Longitudinal variance of electric field strength in
MPDR 7 - SF
For vertical sampling hole array located at d) = 180°
Input power 150Watt, Pressure 8mTorr, and flow rate 205ccm
Data is obtained using flush probe

82

 

IEI in WM
76

l
a i to
6 'yh Izmwaes
. ;;¢ ilk-WM
4‘“ 2 5 ’ " " ' iH‘Iwowm
2 ' i

o

 

 

 

l l

 

°eeeeeeeeeeuaeeaa$
Cicurfererfnl anpb w .r.t. 'lw probe

 

 

 

Figure 4.5 Circumferential variance of electric field strength in MPDR 7 - SF versus
input power at H = 4.3cm plane
Input power 200,150&100Watt, Pressure 8mTorr, and flow rate 20mm
Data is obtained using flush probe

83

 

 

I4 ntorr
E 8 ntorr

 

o‘b“©°%°{l9~6°®°¢°%b9$°,§°éb°

Circumferential angle w.r.t. input probe

 

Figure 4.6 Circumferential variance of electric field strength vs. pressure in
MPDR 7 — SF at H = 4.3cm plane

Input power 200Watt, Pressure 8 & 4mTorr, and flow rate 205ccm
Data is obtained using flush probe

 

 

 

 

 

 

 

 

 

15 A
12 ~
E
>
x
.s 9 ‘
E
6 a
3 a
0 . .
9 ‘19 9 6° 65° @419 .99 \b°\%°¢19°q39$b9q§°qg>°¢b§rg9¢fo
Circumferential angle w.r.t. input probe Egggwgtt
I100Watt

 

Figure 4.7 Circumferential variance of the electric field strength versus input
power in MPDR 7 - SF at H = 4.3cm plane
The data is obtained using 1.0mm E-field sampling probe.
The probe is calibrated by flush probe.

85

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0:2 .I. o E:— 233 n... u m «a 2.33.58 3...: 303.51 .3.— otau a..— \N. v... nigh

mod 8.8 S .N am .ou v com
3; _5 m3; «.2 w

m; 3.: no.“ «:2 o2
e.— 52 w: v.2 w on—
92 m: :2 «.2 com

3.2mm €35 .m 3.3mm EEK €35 A55

Guyana—“3 8% 38m 8:: 8% 38¢ 53E 2885 gem

 

 

4.3 Electron density in the MPDR 7 - SF plasma source

In the plasma based engineering applications, the need for process time estimation
is very vital during the design optimization of a plasma-processing machine.
Determination of this output parameter directly relates to the efficiency and the process
economy. Within the plasma discharge, the higher the reactive species concentrations
are, the faster the process rate and correspondingly the shorter procedural time. This
section gives the results of the electron density measurements carried out for the MPDR 7
- SF ion and plasma source. The objectives of this section are: 1) to ascertain if the
MPDR 7 —SF plasma source can be characterized as a high density plasma (HDP) source
(the electron densities in excess of 1x1011/cm3 is considered to be high density), and 2) to
verify the sustainability of high density discharge at low pressure using ECR magnets.

Double Langmuir probes were used to extract the information about the electron
densities in the argon gas plasma discharge which was ignited under different operating
conditions. The Langmuir probe techniques employed are discussed in detail in Section
3.2 of Chapter 3. The diagnostic probe was positioned in the processing chamber at z =
3cm and r = 0cm as shown in the Figure 3.1 of Chapter 3. The equations of the global
plasma model, which is discussed later in the Section 4.6, are based on the charge density
at the source opening i.e. at z = 0cm plane. Since the Langmuir probe was positioned at z
= 3cm plane, all the charge density measurements are multiplied by a factor of two. This
factor is determined from the density measurements of P. Mak [14].

The cavity applicator was aligned so that its probe excitation antenna projection

was located directly in between the two ECR magnets. See Figure 4.8. Throughout the

87

entire experimentation process, the alignment of the coupling probe antenna with respect
to the ECR magnets was kept identical.

Figure 4.9 displays the electron density data obtained as the source was operated
with three absorbed power levels of 100, 150 and 200 Watt. The chamber pressure ranged
from 1 - 8 mTorr while the gas flow rate was kept constant at 20 sccm. The measured
data is plotted as symbols. The solid lines are the results from global model and are
plotted here for reference. The experimental results demonstrate that the electron density
increases with increase in the absorbed power and the operating pressure. The lowest
electron density was 1.74x10”/ cm3 which as recorded at a pressure of 1 mTorr for 100
Watt of absorbed power. The presence of ECR zones aided in the low—pressure operation
of the MPDR source. The diffusion losses to the quartz discharge chamber walls may
also be reduced due to the presence of ECR confining zones and thus high densities were
recorded for this source. The ion source produced a maximum density of 4.08x10‘1/cm3
for 8 mTorr at 200 Watt.

It would be worth mentioning here that once the Langmuir probe was placed at z
= 1cm (1cm below the ECR magnets), it produced erroneous results. This was due to the
static magnetic fields from ECR magnets which interacted with the Langmuir probe
electrodes. It is therefore advisable to keep the Langmuir probe electrodes away from the
influence of the magnetic fields. Figure 4.10 displays the electron density vs. pressure
curves for 100, 150 and 200 Watt at a fixed gas flow rate of 20 sccm. It may again be
noted that at 1 mTorr, for all absorbed power values, the rate of increase in electron

density is quite low as compared to the 8-mTorr data point.

88

 

 

Plasma source

 

 

 

 

 

 

 

 

 

 

Brass magnet keeper assembly

 

 

 

 

 

 

Top View

  
    
 

'-o-.-.-.-o‘
Q
.-.-.-.-.-a

Figure 4.8 Top View projection of coupling antenna probe over the
pole faces of ECR magnets

89

 

 

 

 

 

 

12 l
8mtorr 4mtorr
10 -
('0
E
:3 8 A 2mtorr
.5
g 6 — lmtorr
@
o
E 4 -— I
E a 0
III I a
2 ‘ . 3 O
IBmMH
0 i . T. j . 04mmn
o 50 100 150 200 250 ‘ 2 mtorr
Absorbed Power(W) O 1 mtorr

 

 

 

Figure 4.9 Electron density of MPDR 7 - SF plasma source as a function of
absorbed power
Solid lines indicate Global model predictions

Pressure range 1 — 8mTorr at a fixed flow rate of 205ccm

9O

 

“T (’s

 

Q.
.1. o
.
.u p...
.
.J
. .
.
I
1
i
,
r
“N iv
6
.
. .r
. v
.t

P—m—_—.—- . -m-—-

 

Electron Densnty( x 10 "/cm3)

 

 

+ hput Pow er 100w
—I— hput Pow er 150w
4.5 + hput POW er 200w

 

 

 

3.5 1
3 .i
2.5 -
2 .
1.5 .
1 .4
0.5 1
0 a T I T I T i r

O l 2 3 4 5 6 7 8 9
Pressure in mTorr

 

 

 

 

 

 

Figure 4.10 Electron density in MPDR 7 — SF as a function of
chamber pressure at a fixed flow rate of 20sccm

‘91

 

4.4 Plasma potential and electron temperature

Plasma potential (VP) and the electron temperature (Te) were calculated from the
data obtained from single Langmuir probe. The single Langmuir probe technique is
discussed at Section 3.2.3 of Chapter 3. The probe was located at r = Gem and the z =
3cm plane. The measurements were made for 100 and 150 Watt of absorbed power at a
fixed flow rate of 203ccm and the chamber pressure range was 1 - 8mTorr.

Figures 4.11 and 4.12 show the experimental pressure dependency of both
electron temperature and plasma potential. It is seen that both plasma potential and
electron temperature decrease as the pressure increases. However the downward slope
levels off as the pressure exceeds 4mTorr. This trend was also observed during the
experimental evaluation of MPDR 9 and MPDR 13 plasma sources (both of them
adopting the side feed coupling) where Vp and Te exhibited a negligible pressure
dependency for higher pressure (>3mTorr) [14, 25]. It is also observed from the Figures
4.11 and 4.12 that the plasma potential and electron temperature are relatively insensitive

to the absorbed power.

4.5 Electron energy distribution function (EEDF)

Process control in plasma based applications requires an understanding of the
transport properties of plasma constituents. The transport mechanism in a plasma
discharge depends upon the kinetics and energy distribution of the reactive species.
Plasma based processes are susceptible to the damage inflicted by the high-energy

electrons impinging on the processed surface. Thus electron energy distribution function

92

(EEDF) becomes an important internal variable. This is the motivation for measuring the
electron energy distribution (EEDF).

In a discharge, the electrons after gaining energy from the input microwave power
may either collide or diffuse out of the source opening. Thus they are not mono-energetic
rather they maintain an energy distribution within a discharge. Figures 4.13 - 4.16 show
the typical normalized EEDF for the MPDR 7 - SF plasma source. The measurements
were made using single Langmuir probe technique that is discussed in Section 3.2.3 of
Chapter3. These plots represent the experimental results for the chamber pressure range
of l - 8mTorr with a flow rate of 20sccm and 100 Watt of absorbed power. It is found out
that the measured EEDFs more closely follow the Druyvesteyn distribution rather then
the Maxwellian distribution. This is in agreement with the results presented by the earlier
researchers [12, 14]. It is noted from the Figures 4.13 — 4.16 that the measured electron
energies are even slightly lower than the Druyvesteyn distribution. Thus absence of high-
energy electrons in the processing region is attributed to elastic and inelastic collisions as

they diffuse downstream into the measurement region.

93

 

 

12

10 5‘fi
1

 

 

 

 

 

 

—o— 100Watts
+ 150Watts

 

 

 

 

 

Electron Temperature (eV)
0)

 

 

 

0 2 4 6 8 10
Pressure in mTorr

 

 

Figure 4.11 Electron temperature vs. chamber pressure in MPDR 7 — SF
for 100 and 150 Watt of input power at 20sccm

94

 

 

Plasma Potential in Volts

30

25

20

15

1O

 

 

4\><._\

 

 

 

 

 

 

2 4 6
Pressure in mTorr

10

 

 

—o— 100Watts
+ 150Watts

 

 

Figure 4.12 Plasma potential vs. chamber pressure in MPDR 7 - SF
for 100 and 150 Watt of input power at 20sccm

95

 

 

 

“U” Represents Maxwellian Distribution

“0” Represents the Measured Distribution
“Solid line” Represents Druyvesteyn Distribution

 

 

 

 

 

 

 

 

 

Figure 4.13 Electron Energy Distribution Function (EEDF) at z = 3cm downstream

Input Power 100 Watt, Pressure lmtorr, Argon gas flow rate 20sccm

96

 

 

“o” Represents Maxwellian Distribution

“0” Represents the Measured Distribution
“Solid line” Represents Druyvesteyn Distribution

 

 

 

 

 

 

 

Electron Energy (eV)

 

 

 

Figure 4.14 Electron Energy Distribution Function (EEDF) at z = 3cm downstream
Input Power 100 Watt, Pressure 2mtorr, Argon gas flow rate 20sccm

97

 

“o” Represents Maxwellian Distribution

“0” Represents The Measured Distribution
“Solid line” Represents Druyvesteyn Distribution

 

 

 

 

 

 

 

 

 

 

 

 

Electron Energy (eV)

 

 

 

Figure 4.15 Electron Energy Distribution Function (EEDF) at z = 3cm downstream

Input Power 100 Watt, Pressure 4 mtorr, Argon gas flow rate 205ccm

98

 

“o” Represents Maxwellian Distribution

“0” Represents The Measured Distribution
“Solid line” Represents Druyvesteyn Distribution

 

 

 

 

EEDF

 

 

 

0 5 10 15 20 25 30 35 4O 45
Electron Energy (eV)

 

 

 

Figure 4.16 Electron Energy Distribution Function (EEDF) at z = 3cm downstream

Input Power 100 Watt, Pressure 8mtorr, Argon gas flow rate ZOsccm

99

Part II

4.6 Global plasma model

The global plasma model, which is based on simple particle balance and power
balance equations, has proved to be a useful model that can lend understanding to the
behavior of the low-pressure argon gas discharges. A brief description of this model is
presented here. A detailed description of global plasma model can be found in reference
[26].

This model is based on the assumptions that:

1) electron energy follows a Maxwellian distribution.

2) electron-ion recombination occurs only on the chamber walls and is absent

within the discharge itself.

3) at high pressures, electron and ion diffusion is governed by Schottky ambipolar

diffusion and at low-pressures free fall ambipolar diffusion is dominant.
Based on the above assumptions, two equations were developed to describe the low-
pressure discharge behavior. These equations are: 1) ion conservation, i.e., ion creation
equals ion lost in every differential volume throughout the discharge, and 2) power
balance, i.e., power absorbed (Pabs) in the plasma discharge equals the power lost by the
electron gas by electron-neutral elastic and inelastic collision processes, and by the

transfer of electron and ion kinetic energy to the walls.

100

4.6.1 Ion conservation relation

The low-pressure discharge is characterized by both collisional and collisionless
diffusion of electrons and ions. The first is explained by the Schottky diffusion model
where diffusion of charged species are governed by the ion and electron collision
frequencies for momentum transfer. The later is based on an assumption that the ion
mean free path, 7g, is greater than the discharge dimensions. The ion mean free path for
an argon gas is given by the following relation [27].

A. = l/ngoi cm E l/(33O P) cm, (4.1)
where oi is the collisional cross section of argon ion (10"4cm2), and P is the pressure in
Torr. The neutral gas density, ng, is given by

ng = P/KTg m‘3, (4.2)
where P is the chamber pressure in Pascels, K is the Boltzmann constant and T3 is the gas
temperature. Equation (4.1) has been derived assuming that Tg is 293°K. In this thesis the
gas temperature Tg is assumed to be 600°K for the entire experimental pressure range of
1 — 8mTorr. Thus when using T3 = 600°K, equation (4. 1) becomes it, E 1/(161 P) cm. The
calculated A,- for the operating pressures of 1, 2, 4, and 8mTorr are 62an 3.1cm, 1.550m
and 0.77cm respectively. Except near the walls (sheath region), the free fall diffusion
model in the discharge assumes a quasi-neutral state i.e. ion and electron densities are
equal, n, = it... With a larger mean free path (1,), the charge species after being ionized
diffuse from the discharge chamber onto the walls without experiencing any collisions.
Thus the discharge is fairly uniform. A low-pressure discharge does not have any volume

recombination processes, and thus every ion or electron created is lost at the walls. The

101

plasma densities near the sheath in the axial and radial directions are represented by hL
and h; respectively and are given by:

hL = 0.86 / (3.0 + L my” (4.3)

hR = 0.8 / (4.0 + R 0..)”2 (4.4)
The discharge Chamber’s (quartz dome) height, L, is 5.6cm and R the discharge radius is
3.5 cm. The experimental pressure regime in this thesis is 1 - 8mTorr. Within this range,
the pressures of 1 and 2mTorr fall in the low-pressure regime thus hL and h; are
approximated as 0.4 [26]. For 4 and 8mTorr, hL and h; are calculated using equations 4.3
and 4.4.
The ion conservation in the discharge is governed by the particle balance equation [26]:

Kill Us = 1/ (ng den), (45)
where Ki; is the electron ionization rate constant. Here uB, the Bohm sheath criterion or
the Bohm velocity is given by:

Lu; = (eTJM)”2 , (4.6a)
where M is the mass of argon ion.
Thus

uB = 1547.5(T.)"2 . (4.6b)
In the equation (4.5), deg, is the characteristic discharge length which is represented by:

deff = 0.5RL / (RhL + LhR) (4.7)

102

Since both K, and uB are functions of electron temperature, Te, equation (4.5)
demonstrates that the discharge electron temperature is a function of both neutral gas
density 118 and characteristic diffusion length deg. For the experiments described in this
thesis the calculated deffl and den: 2 are 269an deff4 and dm 3 are 3.1cm and 3.62cm

respectively, where the subscripts 1 — 8 correspond to the chamber pressure.

103

4.6.2 Power balance relation

The second global plasma model equation is the power balance equation which is
given by [26]:

Pa = noeuBAerrET, (4.8)
where 110 represents the electron or ion density at the center of the discharge. Aeff is the
effective discharge area, which is equal to 27tR (RhL + LhR). The calculated values for
each experimental pressure are Aer“ and A.“ = 80.04ka, A.“ and Aeffg are 69.6cm2
and 59.535 cm2 respectively. In the equation (4.8) ET is the total energy lost per ion—
electron pair lost from the discharge. Mathematically it can be expressed as:

ET = BC + E.3 + E, (4.9a)
The right hand side of the above equation represents collisional energy loss in creation of
an electron-ion pair EC, loss of kinetic energy as an electron collides with the walls 15...,
and finally the loss of ion’s kinetic energy across the sheath E. For Maxwellian electrons
Be is 2T. and for argon ions the ion-bombarding energy, E], is 5.2T... Thus the equation
4.9a can be rewritten as,

ET = BC + 7.2Tc_ (4%)
In the above equation T3 is the electron temperature which can be obtained either from a
direct experimental measurement or by solving the conservation equation (4.5) for an
uniform argon gas discharge. Ec is read from the Ec versus T.3 curve plotted for argon, i.e.
Figure 3.17, in the reference [26]. The values for ET are En = 77.4eV, En = 70.326V,
E74 = 67.68eV and ETg = 66.64eV, where the subscripts 1, 2, 4 and 8 denote the pressure

in mTorr.

104

Equation 4.8 is also known as plasma load line equation and this relation can also be
expressed in terms of average discharge power density <Pa> as:

<P,> = (noe uB ET) / deff = P, / (n R2 L) (4. 10)

105

4.7 Excitation modes in the three MPDR plasma sources.

Out of the three plasma sources compared in this section, only the MPDR 7 - SF
plasma source excites a single TE”. mode. The MPDR 7 - EFL and MPDR 7 - EFP
plasma sources both excite two electromagnetic modes. These two modes by virtue of
their location in the cavity applicator are identified in the discussion to follow as either a
low or a high mode. For clarity, compare the matched tuning positions in the Table 4.5
for both low and high modes.

M. Perrin [15] demonstrated that the MPDR 7 - EFL plasma source can be tuned
to either a TEm mode (high mode) or a TMm/I‘EM mode (low mode). In the preliminary
investigations of the MPDR 7 - EFP plasma source, M. Perrin found out the existence of
the TEM/TMm (low mode) and TMm modes (high mode). In case of the end feed MPDR
plasma sources, the excitation antenna probe! or loop in the low mode is close to the
discharge chamber and thus the electromagnetic energy is coupled to the plasma
discharge through electromagnetic fields that are the sum of the mode and the probe near
fields. The five excitation modes and their matched tuning positions are tabulated in the

Table 4.5.

106

 

89:3. «Eu...— mnm—z 09:: 2.. .5.
2833: 9:55 92.8.3. :2: a... 8.3:. aezgonfl m... 93:.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Em :2 9.2 a: 8.2 :3 $2 2 8.: 3:: oeeeime Ea
_g Bea
33 En New a; N3 «on an :5 Sn 8 30:5

23 oeeeime

.3: 2% as.» mm #3 E a: an 2: EB 285
SE E8
33 82 ES 32 33 2 m8.“ 2." R3. 3 83
3:: creed—am 9 Bow Em
88 99
286 m2 2: S 2: 2 .3 m2 and SE 285 203.285

J J J J J J J J

8038 22m
333% 538 a: a eoaawéo: 88.358. 5257.502 525-302 82: Reese 25
.8332 once 2620M 2m 05 .«o 350m

comwatomon

:5 EJE J

 

auction mafia 3:832

 

 

 

 

 

107

4.8 Performance comparison between global plasma model predictions and the

experimental measurements of the three different coupling geometries.

The global plasma model discussed previously in the Section 4.6 provides an
estimate of plasma source behavior by relating the variables (charge density and electron
energy) with the input variables (chamber pressure, operating power and the discharge
geometry). The objectives of this section are two fold:- 1) to compare the measured
performance of the three MPDR ion and plasma sources with the predictions from the
global plasma model, and 2) to ascertain the validity of the global plasma model for low
pressure, high density, compact MPDR ion and plasma sources.

The solution to the particle balance equation 4.5 yields the curves, which have
been reproduced here in the Figure 4.17. The dashed curve represents solutions from
equation (4.5), i.e., the free fall diffusion solutions, and the solid curve symbolizes
solutions from the Schottky diffusion model. The data points in the figure were recorded
for the MPDR 7 - SF plasma source. These measurements were made for 100 and 150
Watt of absorbed power, chamber pressure of 1, 2, 4 and 8mTorr and the gas flow rate
was fixed at 205ccm. It is seen that the experimental data points deviate considerably
from the predicted free-fall diffusion model for the argon gas. They lie between the
Schottky and free fall diffusion curves. The measured values continue to follow a
downward trend, but as compared to the theory the slope is off by a large factor.

Using the power balance relation (discussed in the Section 4.6.2) a family of
power loss load lines are shown in the Figure 4.9 of Section 4.3. The solid lines in the

Figure 4.9 have been plotted using the equation 4.8. The data points in Figure 4.9

108

representing lower plasma density also correspond to lower input power. It is seen that
the measured discharge density vs. Fab, follows the trend predicted by the global model
but the experimental data points are offset by a certain factor for the entire pressure
range. It can be observed from the figure that once the source is operated at higher input
power the divergence between the measured and the predicted value increases. For
example at 100 Watt the experimental data points are off by a factor of 1.4 - 1.8 while for
200 Watt it increases to 2.6 - 2.8.

Using the global model equations 4.5 and 4.10, the power density load lines are
plotted in the Figures 4.18, 4.19, and 4.20. These figures compare the theoretical power
density vs. plasma density with the measured experimental data. The solid lines in these
figures show the discharge behavior as predicted by the global model for the chamber
pressures of 1, 2, 4, and 8 mTorr. The data points display the experimental results for the
pressure regime of 1 - 8mTorr and the input power of 100, 150 and 200 Watt for the side
feed excitation configuration, 100 and 150 Watt for the end excited sources. As explained
earlier, those data points in the Figures 4.18 — 4.20 which represent lower plasma density
also correspond to lower input power levels.

The following general conclusions are drawn from the Figures 4.18 - 4.20:-

( 1) It is seen that the end feed coupling geometry, in general, follows the global
curves more closelythan the side feed excitation. When the three excitations
are compared, the MPDR 7 — EFP plasma source follows the theory more
closely than the other two plasma sources.

(2) The power density in the MPDR 7 — SF plasma source is higher by a factor of

1.7 - 2.6 from the predicted value once the source operates at high powers

109

(150 — 200Watt). For the low power operation (lOOWatt) it deviates by a
factor of 1.4 — 2.0.

(3) When the three MPDR plasma sources are operated at higher power (150 -
200 Watt for the side excited and 150 Watt for the end excited sources) the
measured power density is higher from the predicted value by a factor of 1.2 -
2.6. While for the low power this factor varies between 1.5 below and 2.0

above the predicted value.

110

4.8.1 Comparison of the three plasma sources with the global model predictions

using experimentally measured electron temperatures

The Figures 4.18 — 4.20 were plotted using the theoretical electron temperature,
Te, given by the global model equations. The Table 4.6 compares the theoretical and the

experimentally measured Te for the MPDR 7 — SF plasma source.

 

 

 

 

 

 

Experimentally measured Te
Pressure Theoretical Te eV
mTorr CV lOOWatt 150Watt
l 7.0 9.36 10.1
2 5.6 9.3 1 9.205
4 4.4 8.3 8.96
8 3.7 8.312 8.529

 

 

 

 

 

 

Table 4.6 Experimental and theoretical electron temperatures (Te)

for the MPDR 7 — SF plasma source

It is seen from the Table 4.6 that the theory predicts low electron temperatures as
compared to the temperatures measured experimentally. Thus it is worthwhile to use
experimental T3 in the global model equations and compare the results with the measured
data. The Figures 4.21 — 4.24 compare the theoretical and the experimental power load
line curves when the measured T6 is inserted into the global model power balance
equation (4.8) for two input power levels of 100 and 150Watt. Figures 4.21 and 4.22

show that at low input power (100 Watt), the experimental results agree fairly well with

111

the theory, whereas the data in the Figure 4.9 displayed the differences of a factor of 1.4 —
1.8. It is also noted that as the operating power increases to 200Watt, the experimental
data deviates from the theory by a factor of 1.2 — 1.7. Using the same power levels, the
experimental results and the theory deviated by a factor of 2.6 — 2.8, as noted from the
Figure 4.9. These new load lines curves suggest that it is possible that the discharge
electron temperatures are higher than the temperature predicted by the global model
equation. The agreement between the experimental and the theoretical load lines
improves when the higher electron temperatures are used. This suggests that the higher
experimentally measured electron temperatures may be valid and the theoretical electron
temperatures predicted by the equation (4.5) are low. The power load lines shown in the
Figures 4.23 and 4.24 use the experimental electron temperatures. Again the agreement
between the theory and experiments is improved when experimental temperatures are
used.

The difference between the measured electron temperatures and the electron
temperatures predicted by the global model should be investigated further, first by
checking the measured values and then by re-evaluating equation (4.5). If the measured
electron temperatures are higher than the predictions by the global model then the ion
conservation equation must be modified to include the correct physical phenomenon that

results in higher electron temperatures.

112

The possible sources of error between the measured data and the theory are:-
l) The plasma density is measured at the edge of the discharge volume. The
actual discharge density might still be higher than the measured density, even
though the charge densities have been compensated by incorporating a
correction factor of two.
2) The applicator conducting wall losses may possibly be larger (> 10% of the

absorbed power) than has been assumed.

113

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

10 - .100 Watt
0 150 Watt
9 ‘ From M.A.Lieberman and AJ.Lichtenberg,
8 + “Principles of Plasma Discharge and Materials
Processine New York: Wilev. 1994.
7. x * .....
6 .
9
3 5
o
'—
4 .
Free Fall Diffusion
3 « - -‘ ‘ - - (uniform discharge)
‘1' SchottkyDiffusion 'f _ . i ' \ ‘7“.fs s x _
(non-uniform discharge) . - » fl
0 F Y 1' V I V i ‘1
l.OE+17 1.0E+18 l.0E+l9 LOB-+20 l.0E+21 1.0E+22 l.0E+23 1.0E+24
ng'lanbda(m‘-2)

 

 

 

Figure 4.17 Experimental electron temperature (data points) plotted against the
universal curves for free-fall and Schottky diffusion Argon gas discharges.
MPDR 7 — SF plasma source
Input power 100 &150 Watt, chamber pressures of 1, 2, 4, & 8mTorr
and gas flow rate fixed at 205ccm

114

 

 

10

Power Density (W/cma)

0.01

 

 

 

r .

1 .

 

 

 

 

 

 

1.5

2 2.5 3 3.5 4

Plasma Density ( x10“cm3)

 

 

A 8 mTorr
e 4 mTorr
I 2 mTorr
o 1 mTorr

 

 

Figure 4.18 Power density load lines for MPDR 7 - SF
Pressure range 1 - 8mTorr
Solid curves represents global model predictions, and the
experimental data is plotted as symbols.

115

 

 

 

 

10

 

 

 

 

 

 

 

 

”E
Q 1
3
g +.
C
Q)
o
E
g e
O
a
0 01 I I I I I +8 mTorr (TBVl-L)
. +8 mTorr(TM-H)
1.5 2 2.5 3 3.5 4 4.5 1 mTorr (NH)
Plasma Density ( x 10“cm3) ' 1 ”WNW”

 

 

 

Figure 4.19 Power density load lines for MPDR 7 - EFP
Pressure range 1 - 8mTorr
Solid curves represents global model predictions, and the
experimental data is plotted as symbols.

116

 

 

 

10

Power Density (W/cma)

0.01

 

 

 

 

 

+1 mTorr(TM-L)
+1 mTorr(TE-H)
+ 8 mTorr(TM-L)

o 8 mTorr(TE~H)

 

 

 

1.5

 

2 2.5 3 3.5 4 4.5

Plasma Density ( x 1011cm3 )

 

Figure 4.20 Power density load lines for MPDR 7 - EFL
Pressure range 1 - 8mTorr
Solid curves represents global model predictions, and the
experimental data is plotted as symbols.

117

 

 

 

 

 

 

12 -
1O —
CO
22, 8 _
E 8mTorr
L“
'5 6 ‘ 4mTorr
8 1 d
an
E 4 ‘ ' 2mTorr
g
LU A
2 a 6 o
I 8 mtorr
0 l 1 I I I 0 4 mtorr
o 50 100 150 200 250 ‘ 2 mtorr
Absorbed Power (W) O 1 mtorr

 

 

 

 

 

Figure 4.21 The Absorbed power load lines for MPDR 7 SF plasma source.
The solid lines are plotted using the measured electron temperature in the global model

equations and the experimental data is displayed in symbols.
For input power of lOOWatt.

118

 

 

12*

 

 

 

 

10 ~
on
E a —
El] 8mTorr
LE; 6 — 4mTorr
‘g 2mTorr
§ 4 A lmTorr
1.1-J A
2 a 6 o
I8mmn
O t . . . . 04mmn
o 50 100 150 200 250 ‘ 2 mtorr
Absorbed Power (W) O ‘I mtorr

 

 

 

 

Figure 4.22 The Absorbed power load lines for MPDR 7 SF plasma source.
The solid lines are plotted using the measured electron temperature in the global model

equations and the experimental data is displayed in symbols.
For input power of 150Watt.

119

 

 

 

 

 

 

 

 

10
”i=2
8 ‘ ‘
.é‘
a)
c
a:
0
$01- .....................
O
a.
0.01 l l T W T

1.5 2 2.5 3 3.5 4 4.5

Plasma Density ( x 1011cm3)

 

 

A8 mTorr
e 4 mTorr
l2 mTorr
o1 mTorr

 

 

 

Figure 4.23 The Absorbed power load lines for MPDR 7 SF plasma source.

The solid lines are plotted using the measured electron temperature in the global model

equations and the experimental data is displayed in symbols.
For input power of 150Watt.

120

 

 

 

 

 

 

 

 

 

 

 

 

10 .....
”E
s 1 ‘
.E‘
(D
C
d)
a
“£01“ :3: .....
no ..............................
A8mTorr
.................................. 04mm"
. . . l2mTorr
0.01 I T T T ‘ o1mTorr
15 2 2.5 3 3.5 4 45
Plasma Density ( x1011cm3)

 

 

Figure 4.24 The Absorbed power load lines for MPDR 7 SF plasma source.
The solid lines are plotted using the measured electron temperature in the global model

equations and the experimental data is displayed in symbols.
For input power of 150Watt.

121

4.9 Plasma source performance figure of merit - Ion production cost (eV/ion)

The ion production cost is a useful figure of merit that is helpful for comparing
different plasma source designs. This section evaluates the performance of the three
plasma sources in terms of ion production cost. The plasma source ion production costs
are discussed in Sections 4.9.1 and 4.9.2. Similarities between the five excitation modes
are discussed in the Section 4.9.3. The comparison of the ion production cost for the three
MPDR sources is made in the Section 4.9.4 and then in the Section 4.9.5 these ion
production costs are collectively compared with the other ECR reactor designs. In the

Section 4.9.6 a summary of some interesting questions is given.
4.9.1 Experimental measurement of ion production cost

The ion production cost is defined as a ratio of the total input microwave power
absorbed (Pam) to the total ion current that can be extracted from the source opening at
the z = 0 plane. This is also defined as the energy required to create a useful ion. It is
assumed that despite the influence of the confining magnets, ions are still lost to the
discharge chamber walls. The ion production cost takes into account only those ions,
which diffuse downstream in the process chamber. The source efficiency which is

determined by the ion production cost, is given by the relation [28]:

11:12.4 lids (411)
As

122

where Pabs is the total power absorbed by the plasma source, J, is the ion saturation
current density at the z = 0 plane. The experimentally measured ion saturation current
density, J s, is given by:

Js = 15/ AP, (4.12)
where Is is the saturation current and Ap is the probe area = 2 7: r x probe height + 1t r2.
The probe height and radius ‘r’ are 0.56 cm and 0.047 cm respectively and thus the probe
area is 0.1723 cmz. Assuming that the density is constant over the z = 0cm plane and then
integrating the above equation yields:

n = P,.,,/ [(1,/ Ap) x 1: R2], (4. 13)

where “R” is the 3.5cm discharge chamber radius.

In the experimental data presented here the saturation current (1's) was measured
using double Langmuir probes which were located at the z = 3cm plane, and the absorbed
power (Pabs) was taken as the difference between the incident power (Pm) and the
reflected power (Pref) i.e. Pat,s = Pine - Pref. The Pine and Pref were read directly from the HP
power meters. See Figure 2.2 in the Chapter Two. The discharge density at the source
opening (2 = 0 cm) was determined by multiplying 1's, measured at the z = 3cm plane, by
a factor of two to estimate the ion saturation current 15 at the z = 0cm plane.

It should be noted here that during the course of experimentation the electron
temperature (Te), which was measured with double Langmuir probes, varied under
different input conditions. For example in case of end feed probe excitation Te varied

between 1.9 —- 3.00 eV and for the end feed loop excitation it

123

 

varied between 2.1 — 3.1 eV. This in return caused variations in the ion saturation current
(1,) thereby influencing the ion production costs. Thus the data displayed in the Figures

4.18 — 4.20 may not correspond to the ion production cost data presented later in this

section.

124

4.9.2 Derivation of ion production cost from global model equations

The source efficiency, defined earlier by the equation 4.11, is the ratio of total
power absorbed by the plasma source (input) to the flux diffusing downstream in the
process chamber. The power balance equation (4. 10) can be rewritten in terms of

characteristic discharge length, deg, as
P... = (enouBE-r nRZL) / am. (4.14)
where TERZL defines the discharge volume. The diffusing flux, JS, in the equation 4.11 is

the flux of ions or electrons passing through the discharge opening at z = 0cm plane and
is expressed as

IS = enouBhL (4. 15)
Using this definition of flux and integrating it over the source opening, the denominator

in the equation 4.11 can now be written as
I] st = enouBertR2 (4. 16)
A.

Using the equation 4.14 and 4.16, the plasma source efficiency can now be expressed as
1'1 '-' ET L / deft hL (4.17)
By substituting deff from equation 4.7,

T1 = 213T { 1+ (UR)(hR/hL)} (4.18)

125

This equation indicates that the ion production cost is a function of L, R, ET, ha, and hL. L
and R are the reactor geometry variables (U2), and ET, which is also a function of
pressure, is an internal variable (X). The terms h, and h, can be best described by the
input variables space (U). However the equation (4. 18) predicts that ion production costs

are independent of input power.

126

4.9.3 Similarities between the five electromagnetic excitations

As seen from the Table 4.5 in Section 4.8, the three investigated MPDR plasma
sources can excite five different electromagnetic excitations. The SF and the EFL (high
mode) both excite the same TB in mode. Thus these two excitations under the same input
pressure and power conditions may yield similar ion production costs. The two end feed
low mode excitations, i.e. the low mode in EFP and EFL, both couple to the discharge
through near probe fields and TM (or TEM modes). Thus these two excitations may also

yield very similar ion production costs under the same input conditions.

4.9.4 Performance comparison of three MPDR plasma sources

Figures 4.25 (a), (b) and (c) compare the performance, i.e. the ion production cost
versus the SF, EFL and the EFP excitations as the pressure and the absorbed power are
changed. The experiments clearly demonstrate that the ion production costs are a strong
function of pressure and Pabs. A careful analysis of experimental data yields interesting
observations and conclusions. Some of these are summarized below:—

(1) Ion production costs increase as the absorbed power increases, i.e.
while the ion/electron density increases with an increase in the
absorbed power the incremental power required to add an additional
ion/electron pair also increases. This is clearly shown in the Figure
4.25a where the ion production costs vs. Fab, are presented for the SF

excited plasma source. Earlier researchers, F.Sze and P.Mak [12, 14],

127

(2)

(3)

have also reported the degradation in the source efficiency with the
increase in absorbed power for the larger diameter plasma sources
(MPDR 325 and MPDR 13).

From the equation (4.18), in Section 4.9.2, we see that the global
model predicts that the plasma source efficiency should be
independent of the absorbed power. But the experimental results in
this thesis and also from the experimental work of the earlier
researchers show that the source efficiency is a function of absorbed
power. These experimental results, collectively, indicate that there are
important experimental phenomenon which are not included in the
global model. For example, the experiments point out that ET is a
function of absorbed power or is indirectly a function of the electric
fields impressed on the discharge.

The Figure 4.25b compares the theory with the experimental results
for the side feed excitation. The theoretical data presented in the
figure is obtained using the global model power balance equation,
which employs both theoretical and experimental electron
temperature (T e). Following conclusions are drawn from the Figure
4.25b:-

(a) At higher pressures of 4 and 8 mTorr, the theory and

experimental results agree within approximately 30%.

128

(4)

(b)

(C)

((1)

At the low pressures of l and 2 mTorr the measured ion
production costs are almost a factor of two higher than those
predicted by the global model.

The theory predicts an increase in the ion production cost from
308eV/ion at 8mTorr to 402eV/ion at lmTorr, which is a 30%
increase. Whereas the experimental data show a much larger
increase in the ion production cost as the chamber pressure is
decreased. For example as the pressure changes from 8mTorr to
lmTorr the ion production costs increase more than 50% for
lOOWatt and more than 80% for 150 and 200Watt.

The agreement between the theory and the experimental results
improves once experimental electron temperature (TC) is used

in the global model power balance equation.

The Figure 4.25c compares the theory with the experimental results

for the three plasma sources. For all the five excitations at a high

pressure of 8mTorr and lOOWatt of absorbed power, the ion

production cost vary between 320 - 350 eV/ion and for 150Watt it

vary between 355 - 440 eV/ion. That is the source efficiencies differ

by less than 10% at lOOWatt and ~ 20% at 150Watt. Thus it could be

easily said here that the three investigated sources become more

efficient when they were operated at higher pressure and low power.

At these higher pressures their efficiencies are similar. However the

source efficiency can degrade by a factor of two as we decrease the

129

(5)

(6)

operating pressure to lmTorr. This is consistent with the earlier
investigations [14].

It is noted fi'om the Figure 4.25c that the TEm mode becomes least
efficient when the MPDR 7 — SF and EFL plasma sources are
operated at a high power (150Watt). Also note that these TE”,
excitations become dramatically less efficient as pressure is reduced
thereby giving much higher ion production costs than that predicted
from the global model. The ion production cost increases from 425
eV/ion at 8mTorr to 800 eV/ion at lmTorr. This difference may be
due in part to the higher impressed EM fields in the cavity applicator
at lower pressures which produces higher conducting wall coupling
losses. The electric field measurements presented in the Figure 4.6 of
Section 4.2 show an increase in the impressed electric fields as the
pressure is reduced. The difference in the ion production costs may
also be related to different heating and the diffusion mechanisms
when the directions of the impressed electric field are varied on the
low-pressure plasmas. This must be investigated further.

It is seen from the Figure 4.26a that the TEm mode produces similar
ion production cost in both the SF and EFL excitations. This supports
the observation discussed earlier that these two excitations are similar
thus may produce similar ion production costs. Under the same input
conditions of pressure and power the ion production costs are

approximately the same for a similar excitation mode. The side feed

130

(7)

(8)

(9)

probe and the end feed coupling methods must have similar coupling
efficiencies.

From the Figure 4.26b it is seen that where the ion production cost
data is available it indicates very similar ion production costs for both
low EFL and EFP modes thereby supporting the idea that these two
excitations are similar.

Figure 4.26c displays a comparison of the high modes in the EFL and
EFP excitations. It is seen that the TM01 mode, which is the high
mode in the EFP excitation, is more efficient than the high mode
(TEm) excited by the EFL excitation. This implies that the high TMo]
modes couple more efficiently to the discharge than the high TEm
mode excitations.

Again referring back to the Figure 4.26b, it is noted that the ion
production cost vary less than 30% for the low mode excitation in the
end feed sources over the entire range of experimental input
conditions. In fact the most efficient ion production cost occurs at
lmTorr with the EFL low mode exciting the discharge. The fact that
the ion production cost do not dramatically increase at low pressures
suggests that some unknown heating and/or diffusion processes are
sustaining the discharge at low pressure when these low modes excite

the discharge.

131

 

4.10 Performance comparison of the three plasma sources with other designs

Figure 4.27 compares the three MPDR ion and plasma sources investigated in this
thesis with larger diameter MPDR 325 plasma source and three other ECR reactor
designs. The three different reactor concepts are: 1) divergent field ECR plasma source
(div), 2) distributed ECR plasma source (decr), and 3) slotted antenna type ECR plasma
sources (slan) in two discharge diameters i.e. 4cm diameter (uslan) and 16cm diameter
(slanl). These plots represent the data already published in the reference [1,14] and is
reproduced here for comparison. The efficiencies for the different ECR configurations
vary over a large range. It may be noted that the MPDR concept for both compact and
larger diameter sources stands out in terms of efficiency. While the reasons for the
differences in between the efficiencies of the different reactor concepts are not
understood, the large dissimilarities may be due to inefficient microwave coupling in the
div, decr, and slan type reactors. Within the MPDR clan of plasma sources the larger

diameter sources are found out to be more efficient than the compact plasma sources.

132

4.11 Summary

On the basis of the experimental results and their analysis presented in this section the

following important questions have been left unanswered:-

The influence of the conducting wall on the source efficiency needs to be determined.
Since the most efficient discharges have some or substantial near field coupling to the
discharge, does this near field plasma coupling play an important role in increasing

plasma source efficiency?

Why is TMm mode coupling is more efficient than the TB] 11 mode coupling?

133

8.58 25...: hm I 5 E52 I 9:38.:— 323 :8 3:269:— E: «mud 2:3...—

 

tOwE 5 0.50005

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

o m w v v v N N N F F F
_ _ _ b _ _ F o
-8N m.
l .08 m
1 as. _l L . . e e - 8v m
.l w -- gar ;. - w.
SON >69. L .gr [I 1 com
, >62 1 . . ; 252 l m
as. 80 n
[ la . 02 m
260.4. 269 U
, w a. .8. m
r. . - 8o
260m

r 000..

 

 

134

.859. «Ema—q mm I N mag 08 280 5002020 :2 .2565qu u D
.00. 3:25.596 @32qu - 000260 02803 838 E02: 0320 03...: 02028 330 8003005 :3 u I
.00. 0300820 wage—05m - 000250 8:23 .838 _o0oE 0320 0003 0803.3 380 0002.080 :3 u I

552.300 5:535 008 «0.. .30
Eaten—=8 85.53.22. 3:25.305 05. 1902809- 238.:— 323 :8 newt—.095 :8 an"... 9:3:—

 

tBE c. 2332.”.
o v

 

 

.302 . . 392 268 .

 

 

 

 

 

 

 

135

 

 

   
  

I
EIIIHHIIHHHHNIHHHHHHI

NMIIlliI!IHINIHHIHIHIHIHIII ..
_

llfillllllllllllllllllllIllllllllllllllllllllIlll

  
 

I||M|||I|||||||||||l||||||||||||

'2‘;

   

 

(uoyAe) isoo uouonpmd u0|

L=L ow MODE (mm-EFL, TEM-EFP)
H=High mode (TEm-EFL, TM-EFP)

Pressure in Man

 

 

136

Figure 4.25c Ion production cost versus pressure -Theoretical and experimental performance comparison for the three

MPDR plasma sources.

I = Ion production costs obtained using global model power balance equation - Employing theoretical Te.

I = Ion production costs obtained using global model power balance equation - Employing experimental Te.

D = Expenmental Ion production costs for MPDR 7 — SF plasma source.

Experimental ion production costs for MPDR 7 -— EFP plasma source.

S = Experimental ion production costs for MPDR 7 — EFL plasma source.

.858 25.2: Hmm I 5 mafia :8 330 50308: :0. _Scmeucqu u m
.0958 2:3: um I 5 ~55 Sm 330 802.08: :2 32.3086 u D
.3. 03:08:35 wig—cam - 80260 3:23 .530: 302: 032m 9%: 00:330 280 50309: :3 u I

.3. 0:28:85 mfizoEEm . 33260 3:23 830: 0.00:: 032m 9:2. 02:83 230 2:28:08: :3 n I

55058» 5:335 H: 0:: mm 2.. 5 EH8 2: .8.—
Eztanfiou 3583.22. 3530.25. 0:: 3328...“... 9:38.:— 259» :8 53309:. uwué :3 9:35

 

:9. E :_ oSwmmt
a _

”WWI
1mm

mumumum

“[["‘1Jl|“‘yw..,

 

(uog/Aa) 1300 uoncnpmd uol

 

 

 

 

137

 

.838 :63: .Em I 0 ”5:2 3: 88 2:28:08: :0_ 3580030 I I
.838 282: Hmm I h 05:04 .3: 2800 302.00.: :0. 3:86:86 n m

.0 H 3:058:06 maze—9:5 :00230 8:23 .630: #0005 052m :5»: 005800 8800 30260:: :0: u I
0:. 0802020 mfixoafim 300300 8:22. .630: _0008 052w m5? 00:00:00 800 302.08: :0: .I. I

8050300: 0033.30 Ham 0:: mun.— 2: :_ A802: 30: 002: SE \ SE 2: .3:
32.35.30 02333.03: 3:25.330 0:: _3303008. 0.388: 2.20.» .30 3:00:02: :0— awn... 0.33m

 

 

 

 

 

:BE 5 0580.:

I O w
, 02 m
W W... D.
W W V -8: m.
W I - 8: m
l - 8... .0.
III. a.

262 . . r
rJ\|I\ 262 com m
>>om_. I 000 w

 

 

138

.0058 0:82: mum: I 0 0:99: 00.: 3000 :0005000: :0. 3:05.003 n I
.0058 0:82: Amm: I 0 M99: 00.: 800 5:00:08: :0. _200E..0:xm n m

.0 .:. 08:25:09.0 9.3295: . 5:05:00 00:22: :030: _0008 052w wfima 002500 0800 50500:: :0: u I
.0 :. 30:00:02: 332:8: - 5:03:00 00:23 .830: :002: 0303 ms»: 002030 0.800 30500:: :0: n I

5.00580 8.3.05 :5 E: 5: 2: 5 032.. :0:: 002.. as: 0... BE. 2.. 3.
:02..a::.00 00:05:90.0: 3:22.330 0:: 30:00:00.5- 0539.: 2.20.» .000 5:00:02: :0: 00m... 052,.—

 

 

..20. c. 05000.:
n a : :

W a o w

M .0

WM. 0

W I oow W

W 0.

W: w.

W - 09. u

W. m

- 80 m

w

I com M

 

 

 

 

139

5:00:00 080:0: 0:05.50 :23 .0058 0052.005: 05 :0 :025::.00 < I 0508:: 2.0.—0.» 0000 5:00:02: :0: 5.: 050:..—

 

 

 

      

 

 

O —. 0.5%.: 0580...: :0:—:30 —. .O
I-----I---Im...:.?;

.I II. >6 I 0 - com 1
. m

- 82.
W
3:00-23... - com: m.
I I I 3
I I I II . ooow m
£0.55“... 2% W I i m
1.4.04.0 E m . I I I.
:w-t0::. / I.- I 00mm W

T ooom

II I IIIII E38 IIIII +

 

 

 

 

140

Chapter 5

Conclusion

5.1 Work done - Salient features

(1)

(2)

(3)

(4)

A new bottom end plate was designed and fabricated that fit the MPDR 7 - SF
cavity applicator to a 7cm discharge. The design of this bottom end plate adopter
is given in Figure 2.4 in Chapter 2.

Performance characterization of MPDR 7 — SF plasma source is carried out by
measuring a selected group of internal variables (X) and output variables (Y).

The influence of coupling structure on the plasma source performance is
experimentally evaluated for the three MPDR plasma sources employing three
different coupling geometries.

The experimental results are compared with the predictions from the global plasma

model and the differences are analyzed.

5. 2 Characterization of MPDR 7 - SF plasma source

5.2.1 Electric fields

(1) A single mode was excited in the cavity which and was identified as TE 11
mode.
(2) The location of excitation antenna probe does influence the spatial electric

field strength in the cavity applicator i.e. the impressed electric fields

141

 

measured near the probe are higher than the predictions from single mode
excitation. These higher fields are associated with the evanescent probe
fields.

(3) The plasma-loaded cavity Q is a function of absorbed power (Pabs) and
chamber pressure. The plasma loaded cavity Q increases as the pressure

decreases. Whereas with an increase in Pabs, Q increases slightly.

5.2.2 Electron energy distribution function (EEDF)

(1) The measured EEDFs more closely follow the Druyvesteyn distribution

rather then the Maxwellian distribution.

(2) The absence of high-energy electrons supports the utilization of MPDR 7 —

SF plasma source in the plasma based processing applications.

5.2.3 Electron temperature and plasma potential

(1) Both electron temperature (Te) and plasma potential (Vp) are weak

functions of Pan and chamber pressure.

142

5.2.4 Electron density

5.3

5.3.1

(1)

(2)

(3)

The lowest and the highest electron densities recorded for the MPDR 7 -
SF plasma source are 1.74 x 10”/cm3 (at 100 Watt and 1 mTorr) and 4.08
x loll/cm3 (at 200 Watt and 8 mTorr) respectively.

The MPDR 7 — SF plasma source sustained high-density plasmas (greater
than or around 1 x lOll/cm3) specifically at low pressures and generally
under all input conditions.

An increase in electron density has been recorded with a corresponding

increase in the operating power and the chamber pressure.

Experimental results of the three plasma sources

Comparison of the experimental results with the global model

(1)

(2)

(3)

The experimentally measured electron temperatures were higher than the
electron temperatures predicted by the global mode].

When the experimentally measured electron temperatures are inserted into
the global model power balance equation the agreement between the
experimental and the global model load lines improves.

The experimentally measured plasma load lines do not match the load lines
obtained from the global model equations. The theoretical load lines

produce higher electron densities for a given input power Pabs. The slopes

143

(4)

(5)

of the load lines, i.e. Anc vs. AP...” are lower for the experimental data, i.e.
as Pa... increases the differences the theory an experiment increase.
According to the global model, ion production costs, 11, are independent of
Pabs~ However the experimental measurements indicate that ion production
costs are functions of Fab, and chamber pressure. Ion production costs
increase as pressure decreases and increase as power increases.

The experimental measurements demonstrate that the discharge properties
vary as Fab. varies. The difference between global model predictions and
experimental data increases as input power increases and chamber pressure

decreases.

5.3.2 Influence of the coupling structure on the source performance

(1)

(2)

Overall efficiencies do not have a major dependence upon loop or probe
applicator excitations but appear to be dependent upon the type of mode
excited, i.e. efficiencies depend on TE vs. TM mode excitation and/or on
near field coupling.

The three investigated MPDR plasma sources maintained a stable and
controllable discharge for the entire input parameter space. However it was
observed that it is easier to maintain a discharge under low-pressure

conditions by the side feed excitation.

144

5.4

(3)

(4)

(5)

(6)

The ion production costs vary considerably with mode excitation i.e. the
five different excitation modes yield different efficiencies. These differences
can be greater than a factor of two.

The most efficient operation (i.e. the lowest ion production costs) occurs
with low mode excitation in the presence of near field coupling, i.e., either
low mode loop or probe near field coupling.

“High mode” excitations are the least efficient. Using the results from other
investigators, it is unlikely that the efficiency differences can be attributed
to applicator conducting wall losses.

The most inefficient operation is with TB”, mode excitation. At low
pressures of lmTorr and with TE“ 1 mode excitation, the plasma source is

more than two times as lossy as excitation with a low mode.

Comparison of the three investigated sources with other designs

(1)

(2)

MPDR ion and plasma sources are the most efficient as compared to the

other ECR reactor designs.

The larger diameter MPDR 325 plasma source is most efficient when

compared to the compact MPDR plasma sources.

145

5.5

5.6

Suggestions for an improved plasma source design

The experiments demonstrate that overall source efficiency is improved by
near field coupling. Thus an improved plasma source design may incorporate probe
or loop coupling placed very close to the discharge. This type of near field
coupling may be optimized by varying the geometry of the coupling structure
adjacent to the quartz dome/discharge interface. Experiments also have
demonstrated that the TM mode excitation yields better efficiencies than TE mode
excitation. Combining these two major experimental observations would result in a
plasma source design with end feed probe coupling that excites TEM, or TM
modes in the coaxial section, and has near field coupling at the end of the probe.
This coupling may be improved by placing a flat disk on the end of the excitation

probe.

Recommendations for the future research

(1) The measured electron temperatures are higher than the predictions from
the global model. The measured plasma densities are lower than the
densities predicted from the global model and ion production costs are a
function of input power. Thus it is thus imperative to verify the accuracy of
these experimental measurements by repeating selected experiments. If the

differences between global model and experiments are proven to be valid,

146

(2)

(3)

(4)

(5)

then the global plasma model should be modified to accommodate the
phenomenon responsible for these differences.

The measured plasma source efficiency is different for the five-excitation
modes. The conducting wall losses might be different for each excited
mode i.e. they may be relatively lower or higher. Earlier researchers (P.
Mak and M. Perrin) have shown that the wall losses are very small i.e. less
than 3-5 % of Pm. It is therefore unlikely that the efficiency differences can
be attributed to the applicator conducting wall losses. It is still required to
validate the earlier results by measuring the wall losses for the five different
excitations evaluated in this thesis.

Characterize, in greater detail, the end excited source performance in terms
of : 1) Pan, vs. ne and Te, and 2) ne and T. vs. chamber pressure.

It is seen that the influence of Pabs on the plasma source efficiencies is
significant. This might be contributed by the neutral and ion gas heating in
the discharge. The influence of gas heating should be evaluated further.
Experimentally verify the concept of an improved plasma source design,

which is discussed in the Section 5.5.

147

5.7 Concluding remarks

The experimental evaluation of the three MPDR ion and plasma sources which is
presented in this research work gives a better understanding of the characterization of low
pressure and high density argon gas discharge and most importantly the influence of

microwave excitation geometry on the plasma discharge parameters.

148

REFERENCES

[1] J. Asmussen, T. Grotjohn, P. Mak, M. Perrin, “The design and application of electron
cyclotron resonance discharges,” IEEE Trans. Plasma Sci., vol. 25, pp1196-1221,
1997.

[2] J. Asmussen and J. Root, “Ion generation apparatus and methods for the use thereof,”
US. Patent #4,507,588, (Mar.26, 1985).

[3] J.M. Cook and K Donohoe, “Etching issues at 0.35 um and below,” Solid State
Technol. 34(4), 119-124 (1991).

[4] J. Root and J. Asmussen, “MSU Microwave Ion Source,” presented at the
MSU/NASA workshop on Advance Propulsion Concepts using Time Varying
Electromagnetic Fields, East Lansing, Michigan, Feb. 24-25 (1982).

[5] M. Dahimene and J. Asmussen, 19th EPIB conference, (1984).

[6] M. Dahimene and J. Asmussen, “The performance of a microwave ion source
immersed in a multicusp static magnetic field,” J. Vac. Sci. Technol., B4(l), 126-130
(1986).

[7] J. Asmussen and M. Dahimene, “ The experimental test of a microwave ion beam
source in oxygen,” J. Vac. Sci. Technol., B5(1), 328-331 (1987).

[8] L. Mahoney and J. Asmussen, “ A compact, resonant cavity, five centimeter,
multicusp, ECR broad-beam ion source,” Rev. Sci.Instrum., 61(1), 285-287 (1990).

[9] L. Mahoney, “ The design and testing of a compact electron cyclotron resonant
microwave cavity ion source, MS. Thesis, Dept. of Electrical Engineering, Michigan

State University (1989).

149

 

[10] A. Srivastava, M. Dahimene, T. Grotjohn and J. Asmussen, “ Experimental
characterization of a compact ECR ion source,” Rev. Sci.Instrum., 63(4), 2556-2558
(1992).

[11] F. Sze, D. Reinhard, B. Mussen and J. Asmussen, “ Experimental performance of a
large-diameter multipolar microwave plasma disk reactor,” J. Vac. Sci. Technol.
B8(6), 1759-1762 (1990).

[12] F. Sze, “ Design and experimental investigation of a large diameter electron
cyclotron resonant plasma source,” Ph.D. Thesis, Dept. of Electrical Engineering
Michigan State University (1990).

[13] B. Mussen, F. Sze, D. Reinhard, and J. Asmussen, J. Vac. Sci. Technol. B9(6), 3521
(1991).

[14] P. Mak “An experimental evaluation of a 12.5-cm diamter multipolar microwave
electron cyclotron resonance plasma source,” Ph.D. Thesis, Dept. of Electrical
Engineering Michigan State University (1997).

[15] M. Perrin “The electromagnetic evaluation of a compact ECR microwave plasma
source,” MS. Thesis, Dept. of Electrical Engineering, Michigan State University
(1996).

[16] Micro-Now Instruments, “Owner’s manual Model 429B1 microwave power
oscillator,” (1986).

[17] M. Dahimene, Ph.D. Thesis. Michigan State University, (1987).

[18] I. Langmuir and H. Mott-Smith, Gen.Elec. Rev. 26 (1923).

[19] W. Lochte-Holtgreven, “ Plasma diagnostics,” North-Holland Publishing Company,

Amsterdam (1968).

150

[20] E. Johnson and L. Malter, “ Double-probe method for determination of electron
temperature in steady and time-varying gas discharges,” Phys. Rev. 76, 1411-1412
(1949)

[21] B. Chapman, “Glow discharge processes,” New York, NY: Wiley (1980).

[22] M. J. Druyvesteyn, “ Der Niedervoltbogen,” Z. Phys. 64, 781-798 (1930).

[23] H. Rundle, D. Clark and J. Deckers, “ Electron Energy Distribution Functions in an
02 discharge,” Can. J. Phys. Vol. 51, 144-148 (1973).

[24] S. Burkart, “ Coaxial E—field probe for high-power microwave measurement,” IEEE
Trans. Microwave Theory Tech. MTT-33 (3), 262-265 (1985).

[25] J .A. Hopwood, “ Macroscopic properties of a multipolar ECR microwave cavity
plasma source for anisotropic silicon etching”, Ph.D. Dissertation, Michigan State

University, (1990).
[26] M.ALieberman and AJLichtenberg, “ Principles of Plasma Discharge and
Materials Processing”, New York: Wiley, (1994).

[27] EA McDaniel, “ Collisional phenomena in ionized gases, New York ”, NY: Wiley,
(1964).

[28] J .R. Brophy and RI. Wilbur, “ Simple performance model for ring and line cusp ion
thrusters ”, AIAA Journal, 23(11), 1731 — 1736, (1985).

151

"11111111111001“