LIBRARY i Michigan State University PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE " W. l ll___ll :LJ _;_Jl:L_ 3%: ___] [—— MSU Is An Affirmative Action/Equal Opportunity Institution cMma-DJ DESIGN AND EXPERIMENTAL INVESTIGATION OF A LARGE DIAMETER ELECTRON CYCLOTRON RESONANT PLASMA SOURCE By Fan Cheung 826 A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Electrical Engineering 1993 New for [ cyclc recer. 915 I “hid EXpe plasn ion e Unifo- MPD EValu km a ABSTRACT DESIGN AND EXPERIMENTAL INVESTIGATION OF A LARGE DIAMETER ELECTRON CYCLOTRON RESONANT PLASMA SOURCE By Fan Cheung Sze New plasma sources are needed to meet the stringent processing condition for Ultra Large Scale Integrated circuit (ULSI) fabrication. Two electron cyclotron resonant Microwave Plasma Disk Reactors (MPDRs) have recently been developed. The first one is the MPDR 20, which is operated at 915 MHz and has a 20 cm discharge diameter. The other is the MPDR 325, which is operated at 2.45 GHz and has a 24 cm discharge diameter. Experimental measurements showed that the two plasma sources had high plasma density (about 1011 cm'3), low plasma potential (8-12 volts) and low ion energies (8-15 eV). With downstream magnetic confinement, plasma uniformity of 2.6% (10') over 18 cm has been achieved. Performances of five MPDRs with discharge diameter varying from 5 cm to 24 cm were evaluated. The MPDR 325 had an ion production efficiency of 220 eV per ion and power density of 0.1 Watt/cm3. Etching of 0.1 pm to 0.5 pm features in silicon using an SF6-argon plasma was performed in a similar 24 cm diameter discharge reactor. Anisotropic (900) etch results were achieved with an etch rate of 200-280 Almin. No microloading effect was observed. Etching results indicated that the etching process was limited by gas flow. Etch performances versus other operation parameters were also studied. Copyright by FAN CHEUNG SZE 1993 To my mother, A-Ngo Chan, sisters (Siu-Ying, Siu-Chu, Sin-Hung, Sin-Man and Sin-Fan) and Wai-Ying, for their love and care. 1421 L, Spartan Village East Lansing iv Chm msez appn labor hous achu Pnfib Wouh &flbe nsear from! lflnto ACKNOWLEDEGMENTS I would like to thank my adviser, Dr. J. Asmussen, Professor and Chairperson of the Department of Electrical Engineering, for his guidance in research and support over the last four and a half years. I enjoyed and appreciated all the moments with him, from those weekends in the laboratory talking about experiments to the discussion of my thesis in his house. I would like to thank Professor D.K. Reinhard for his guidance in the etching work and his comment in my thesis. Also I would like to thank Professor T. Grotjohn who reorganized the clean room in the laboratory. I would like to thank Mr. R. Fritz, Mr. L. Mahoney, Dr. J. Hopwood, Dr. G. Salbert and Mr. B. Musson for their assistance during the early stage of my research. Thanks must also be extended to Dr. Alers and Professor Golding from the Department of Physics and Astronomy for the preparation of 0.1 pm to 0.5 pm patterns in the submicron etching experiments. Research funding was provided in part by Wavemat Inc. and Michigan State Research Excellent Fund. 7 1 ~.J , ~44 1 ~46 _C/) [-4 (“J A TABLE OF CONTENTS LIST OF TABLES .............................................................................. x LIST OF FIGURES ............................................................................. xi CHAPTER ONE INTRODUCTION 1 1.1 Introduction .................................................................................. 2 1.2 Research objectives ....................................................................... 4 1.3 Research history ........................................................................... 4 1.4 Outline of thesis ............................................................................. 5 CHAPTER TWO EQNDAMENTQLS 6 2.1 Introduction ................................................................................... 7 2.2 Plasma .......................................................................................... 7 2.2.1 Debye - Hiickel radius .......................................................... 10 2.2.2 Mean free path and collision cross-sections .......................... 13 2.3 Motion of charged particles in time-harmonic E field ................. 15 2.4 Conductivity in plasma .................................................................. 21 2.4.1 AC conductivity .................................................................... 21 2.4.2 Power absorption in plasma .................................................. 23 2.5 Sheath and ion energy .................................................................... 31 2.6 Magnetic confinement of plasma ................................................... 32 vi 0,) J I» I 9‘) ——— {JJ ' J . . 3.4 cm (J) 4.1 4.3 CAL/om:— lr—r er CHAPTER THREE ECR PLASMA SOURCES 34 3.1 Introduction ................................................................................... 35 3.2 Axial divergent magnetic field ECR plasma sources ...................... 37 3.3 Distributed ECR plasma source .................................................... 44 3.4 Hom-type ECR plasma sources .................................................... 46 3.5 Other plasma sources ..................................................................... 53 3.5.1 Helicon source ..................................................................... 53 3.5.2 Helical resonator .................................................................. 55 3.5.3 Inductive coupled plasma sources ........................................ 55 § 3.6 Summary ....................................................................................... 63 CHAPTER FOUR MICROWAVE PLASMA DISK REAQIQR (MPDR! 64 4.1 Introduction ................................................................................... 65 4.2 Structure of MPDR ........................................................................ 65 4.2.] Structure of MPDR 2O ......................................................... 66 4.2.2 Historical development ........................................................ 68 4.3 Operation of MPDR ....................................................................... 78 4.3.1 Electromagnetic resonant modes in cylindrical cavity .......... 78 4.3.2 Ignition and tuned positions in MPDRs ............................... 96 4.3.3 Mode of operation ............................................................... 98 4.3.4 Optimal magnetic configuration ........................................... 110 CHAPTER FIVE DIAGNOSTICS QF MPDR 114 5.1 Introduction ................................................................................... 115 5.2 Experimental techniques ................................................................ 118 5.2.1 Double Langmuir probes ..................................................... 118 5.2.2 Single Langmuir probe ........................................................ 124 5.2.3 Ion energy analyzer ............................................................. 127 vii 6.1 6.3 n‘ .3 6.4 6.5 {/1 r-v-d A — Pn 5.3 6.1 6.2 6.3 6.4 6.5 7.1 7.2 7.3 7.4 7.5 7.6 Experimental results ...................................................................... 133 5.5.1 MPDR 5 .............................................................................. 133 5.5.2 MPDR 20 ............................................................................ 144 5.5.3 MPDR 325 .......................................................................... 160 5.4.4 Summary ............................................................................. 169 CHAPTER SIX PERFORMANCE OF MPDR 177 Introduction ................................................................................... 178 Experimental setup ........................................................................ 179 Definitions of figures of merit ........................................................ 181 6.3.1 Electron temperature versus PA .......................................... 181 6.3.1.1 Free-Fall diffusion .................................................. 182 6.3.1.2 Schottky diffusion ................................................... 184 6.3.2 Microwave coupling efficiency ............................................ 190 6.3.3 Ion production energy cost and power densities ................... 193 Experimental results ...................................................................... 194 Summary ....................................................................................... 205 CHAPTER SEVEN SUBMIORON ETCHING OF SILICON USING MPDR 325 206 Introduction ................................................................................... 207 Chemistry of dry etching for silicon ............................................... 208 Etching methods ............................................................................ 212 Etching system .............................................................................. 216 Preparation of experiment .............................................................. 220 7.5.1 Optical and electron-beam lithography ................................ 220 7.5.2 Etching conditions ............................................................... 222 Etching results ............................................................................... 223 7.6.1 Role of gas flow .................................................................. 223 7.6.2 Effect of rf substrate bias ..................................................... 225 7.6.3 Variations of etch rate with other processing parameters ..... 230 viii 7.6.4 Etch rate uniformity ............................................................. 232 7.6.5 Loading and Microloading effect ......................................... 235 7.7 Summary ....................................................................................... 243 CHAPTER EIGHT FUTURE RESEARCH 245 8.1 Summary of research ..................................................................... 246 8.2 Future research .............................................................................. 248 8.2.1 Plasma source development .................................................. 248 8.2.2 Plasma etching ..................................................................... 249 REFERENCES ........................................................................................ 251 ix Table 3. Table 3. Table 4 Table 4. Table 4. Table 4. Table 4. Table 6, Table 6, Table 7, Table 7. Table 3.1 Table 3.2 Table 4.1 Table 4.2 Table 4.3 Table 4.4 Table 4.5 Table 6.1 Table 6.2 Table 7.1 Table 7.2 LIST OF TABLES Plasma parameters of divergent ECR source ........................ 41 Plasma parameters of distributed ECR source ...................... 47 Operation positions of MPDR 5 without a grid .................... 100 Resonant modes in empty MPDR 20 ................................... 102 Operation positions of MPDR 20 at TE211 mode ................. 106 Operation positions of MPDR 20 at TE111 mode ................. 107 Operation positions of MPDR 325 ....................................... 109 Experimental data for MPDR sources .................................. 180 Characteristic lengths of MPDR .......................................... 188 Reaction set for silicon etching using SF6 ............................ 209 Etching results showing loading effect ................................ 236 Figur Figur. Figure Figure Figure Figure FlEure ; Hflm; Him; 3 Figure 2.1 Figure 2.2 Figure 2.3 Figure 2.4 Figure 2.5 Figure 2.6 Figure 2.7 Figure 2.8 Figure 2.9 LIST OF FIGURES Variations of Debye length with electron temperature at different plasma densities ............................................... Variations of mean free path with pressure at different neutral temperature ............................................................ Snap shot of the motion of a charged particle in time- variant E field and static B field ......................................... Variation of electron energy with time at ECR at different electric field, E 1., strength .................................... Variations of the real parts of Cl with ratio of collision frequency/excitation frequency at different static magnetic field strength at 2.45 GHz excitation frequency .. Variations of the real parts of 0 J. with ratio of collision frequency/excitation frequency at different static magnetic field strength at 915 MHz excitation frequency .. Variations of the real parts of o J. with magnetic field strength at different pressure with 2.45 GHz excitation frequency ........................................................................... Variations of the real parts of o J. with magnetic field strength at different pressure with 915 MHz excitation frequency ........................................................................... Variations of the real parts of (Si with magnetic field strength at different plasma density with 2.45 GHz xi ..12 14 ..16 ..20 24 25 .26 .27 Figure Figure Figure Figure Figure . Figure . Figure ; Figure Q Flgure E Figure 2 Figure 3 Flgure j FlSure j Fl'gure g Fl{lure 3 Fl{lure a FlEllie. 4 F igllre 4 Figure 2.10 Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5 Figure 3.6 Figure 3.7 Figure 3.8 Figure 3.9 Figure 3.10 Figure 3.11 Figure 3.12 Figure 3.13 Figure 3.14 Figure 4.1 Figure 4.2 Figure 4.3 excitation frequency ............................................................. 29 Variations of the real parts of 01 with magnetic field strength at different plasma density with 915 MHz excitation frequency ............................................................. 30 Divergent field ECR source ................................................. 38 Distribution of the electric field (a) strength and (b) orientation in TE12 mode ..................................................... 42 Distribution of the electric field (a) strength and (b) orientation in TE13 mode ..................................................... 43 Distributed ECR plasma source .......................................... 45 Elongated-Mirror-Goemetry (EMG) plasma source ............ 48 Leybold source .................................................................... 49 Distribution of transverse electric field in TMOI mode ........ 51 Longitudinal E field strength in TMOI mode ........................ 52 Schematic drawing of a Boswell source .............................. 54 Side view of a longitudinal section of a Helical resonator .............................................................................. 56 Schematic drawing of ICP-MS instrument constructed by Sciex ............................................................ 57 (a) A schematic diagram of the ICP1 and (b) Variation of induced electric field 2 inches under the coil ................... 59 Schematic illustration of ICPZ .............................................. 60 Illustration of capacitive and inductive coupling .............. 62 Structure of MPDR 20 .......................................................... 67 Photographs showing the components of the MPDR .......... 69 Development of MPDRs for ECR plasma applications .......................................................................... 7 1 xii Figure Figure Figure . Figure . Figure . Figure - Figure - Figure - Figure - Figure - Figure . Figure .‘ Figure .‘ Figure 4 Flgure 4 l:lgure 4 Flgure 4 fifim4 Figure 5 Figure 4.4 Figure 4.5 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 Figure 4.17 Figure 4.18 Figure 4.19 Figure 4.20 Figure 4.21 Figure 5.1 Figure 5.2 Figure 5.3 Figure 5.4 Figure 5.5 Figure 5.6 Figure 5.7 (a) MPDR 325, (b) MPCS .................................................... 75 TE mode chart at 2.45 GHz ................................................. 81 TM mode chart at 2.45 GHz ................................................ 82 TB mode chart at 915 MHz .................................................. 83 TM mode chart at 915 MHz ................................................. 84 Distribution of E-field at TM“ mode .................................. 87 Distribution of E-field at TE01 mode ................................... 88 Distribution of E-field at TE11 mode ................................... 90 Distribution of E-field at TE21 mode ................................... 91 Distribution of E-field at TE31 mode ................................... 92 Distribution of E-field at TE41 mode ................................... 93 Distribution of E-field at TE61 mode ................................... 94 Distribution of E-field at TE71 mode ................................... 95 Schematic setup of the sweeping circuit ............................. 97 Experimental measurement of E-field at TE111 mode .......... 104 Experimental measurement of E-field at TE211 mode .......... 105 Distribution of ECR surface using 4 magnets in MPDR 13 ............................................................................ 111 Tetrapole magnetic configuration using 8 magnets .............. 112 Schematic diagram of the microwave circuit ....................... 116 Experimental setup for plasma diagnostics .......................... 117 Double Langmuir probes technique: (a) schematic circuit and (b) l-V characteristics ........................................ 120 Variation of sheath potential with applied voltage ............... 121 Schematic circuit for single probe measurement ................. 126 EEDF in MPDR 5 ................................................................ 128 EEDF in MPDR 20 .............................................................. 129 xiii Figure Figure Figure . Figure : Figure .‘ Figure 1 Figure 5 Figure 5 Figure 5 “”657 fiRmsg Figure 5.8 Figure 5.9 Figure 5.10 Figure 5.11 Figure 5.12 Figure 5.13 Figure 5.14 Figure 5.15 Figure 5.16 Figure 5.17 Hymim Figure 5.19 Figure 5.20 Figure 5.21 Figure 5.22 EEDF in MPDR 325 ............................................................ 130 Schematic circuit for ion energy measurement ..................... 132 Variations of plasma density with downstream position in MPDR 5 .......................................................................... 134 Radial distributions of plasma density at different downstream positions in MPDR 5 ....................................... 135 Radial distributions of plasma density at different pressures and microwave power in MPDR 5 ........................ 136 Variations of plasma potential in MPDR 5 .......................... 138 Variations of ion energy with downstream position in MPDR 5 .......................................................................... 139 Variation of IEDF with downstream position at 1 mTorr, 160 watts microwave power in MPDR 5 ............... 140 Variation of IEDF with downstream position at 1.5 mTorr, 100 watts microwave power in MPDR 5 ............ 141 Variation of IEDF in radial direction at 1.5 mTorr, 100 watts microwave power in MPDR 5 .............................. 142 Radial variations of electron density at different microwave power in MPDR 20 ............................................ 145 Radial distributions of plasma density at different downstream positions in MPDR 20 ....................... 146 Variations of plasma density with pressure at 0 cm downstream and different microwave power in MPDR 20 ............................................................. 147 Radial distributions of plasma potential at different pressures in MPDR 20 ....................................... 149 Radial distributions of plasma potential at XIV Figur Figur Figuri Fuun Figure Figure Figure Figure : Figure 5 mej Flé‘Ure 5. in Ellrej I (A) Figure 5.23 Figure 5.24 Figure 5.25 Figure 5.26 Figure 5.27 Figure 5.28 Figure 5.29 Figure 5.30 Figure 5.31 Figure 5.32 Figure 5.33 Figure 5.34 different microwave power in MPDR 20 ............................. 150 Variation of plasma potential with downstream positions in MPDR 20 ......................................................... 151 Radial variation of electron energy in MPDR 20 ........................................................................... 152 Variation of electron energy with downstream position in MPDR 20 ........................................................... 153 Variations of IEDF at different downstream positions in MPDR 20 ......................................................... 154 Variation of IEDF with external d.c. bias in MPDR 20 ....................................................................... 156 Radial variations of average ion energy with microwave power in MPDR 20 ............................................ 157 Radial variations of average ion energy with pressure in MPDR 20 .......................................................... 158 Variation of average ion energy with downstream position in MPDR 20 ...................................... 159 Variations of plasma density with pressure at 0 cm downstream and different microwave powers in MPDR 325 ......................................................... 161 Radial distributions of plasma density at different downstream positions in MPDR 325 .................................... 162 Radial distribution of plasma density with and without wafer holder at 14 cm downstream in MPDR 325 ...................................................................... 164 Radial distribution of plasma density with and without wafer holder at 18 cm downstream xv Figure Figure Figure Figure Figure Figure ; Figure! Flgui'e f Flgulre 6 Flgum 6 Flgure 6 F1"éUre 6 FlF'Ure 6 Fig F1 Ure6. $51166 - FIgUl-e 6 5 ‘e Figure 5.35 Figure 5.36 Figure 5.37 Figure 5.38 Figure 5.39 Figure 5.40 Figure 5.41 Figure 5.42 Figure 6.1 Figure 6.2 Figure 6.3 Figure 6.4 Figure 6.5 Figure 6.6 Figure 6.7 Figure 6.8 in MPDR 325 ...................................................................... 165 Radial variations of plasma potential at different downstream positions in MPDR 325 .................................... 166 Radial distribution of plasma potential with and without wafer holder in MPDR 325 .............................. 167 Variations of average electron energy with microwave power at 0 cm downstream and different pressures in MPDR 325 ...................................................................... 168 IEDF at different pressure in MPDR 325 ............................. 170 IEDF at different microwave power and downstream positions in MPDR 325 .................................... 171 Radial distribution of ion energy at different downstream position in MPDR 325 .................................... 172 Radial distribution of ion energy at different pressures in MPDR 325 ....................................................... 173 Radial distribution of ion energy at different microwave power in MPDR 325 .......................................... 174 Variation of electron temperature with CPA at different neutral temperature ................................................ 189 Variation of Te versus CPOt at different neutral temperatures ........................................................................ 191 Distributions of the experimental CPA data ......................... 195 Microwave coupling efficiency versus A ........................... 197 Microwave coupling efficiency versus pressure ................... 198 Microwave coupling efficiency versus microwave power 199 Ion production energy cost versus A .................................... 201 Ion production energy cost versus pressure .......................... 202 xvi Figure 6.9 Figure 6.10 Figure 7.1 Figure 7.2 Figure 7.3 Figure 7.4 Figure 7.5 Figure 7.6 Figure 7.7 Figure 7.8 Figure 7.9 Figure 7.9 Figure 7.10 Figure 7.11 Ion production energy cost versus microwave power ........... 203 Power density versus A ........................................................ 204 Reactions of atomic fluorine with silicon ............................. 21 1 Mass spectra of etch products as a function of substrate temperature ....................................................... 213 Effect of ion bombardment on the mass spectra ................... 214 Etching methods .................................................................. 215 Cross sectional view of the processing system ..................... 217 Variations of potential and current densities for ions and electrons during one rf cycle ........................... 219 Preparation procedure of the etch sample ............................. 221 Variation of etch rate with flow rate at 1 mTorr, 10 cm downstream, 4% SF6 and -10 volts bias (10 watts rf) ........... 224 (3) Variation of self-induced bias with rf power with 5% SF6 at 10 cm downstream ............................................. 226 (b) Variation of self-induced bias with rf power with 5% SF6 at 4.5 cm downstream ........................................... 227 Effect of bias on etch mask: (a) a damaged mask at pressure 9 mTorr, 5% SF6. The bias was -110 volts (50 watts rt) and (b) undamaged mask at 0.6 mTorr, 4% SF6 with -10 volts bias (11 watts rf power). 500 watts microwave power and 10 cm downstream position were used in both cases .......................................... 229 Variation of etch rate with (a) downstream position with -10 volts bias (11 watts rf) and (b) bias at 10 cm downstream position (9-40 watts rf). Etching conditions were 4% SF6, 500 watts microwave at 0.6 mTorr ................ 231 xvii Figure 7. Figure 7. Figure 7. Figure 7. Figure 7.} Figure 7.1 Figure 7.12 Variation of etch rate with microwave power at 0.6 m Torr, 4% SF6, 10 cm downstream and -10 volts bias (11 watts rf) ......................................................................... 233 Figure 7.13 Variation of etch rate with pressure in a 5% SF6 plasma at constant 50 watts rf and 500 watts microwave power ...... 234 Figure 7.14 Etching experiment for uniformity measurement ................. 237 Figure 7.15 Loading effect according to equation (7.3) ........................... 239 Figure 7.16 Anisotropic etching of silicon with the 100 nm thick aluminum mask intact. (a) 500 nm line/500 nm spacing, (b) 200 nm line/300 nm spacing ........................................... 240 Figure 7.17 Anisotropic etching of silicon with the 100 nm thick aluminum mask intact. (a) 500 nm line/200 nrn spacing, (b) 100-150 nm line/200 nm spacing ................................... 241 and (c) 100-150 nm line/500 nm spacing ............................. 242 xviii CHAPTER ONE INTRODUCTION § 1.1 Introduction § 1.2 Research objective § 1.3 Research history § 1.4 Outline of thesis [187]. fire )' eumpi imprm derelo memor memor design um or] both in Undersri fabrieari Hi del'elopr Tabn'carii IDOTS n0 l I0 Prom Flpecja” \ Red ldi‘al [00] j 80 that C0m § 1.1 Introduction In 1947, the world's first transistor was fabricated at Bell Laboratories [187]. The evolution of integrated circuit technology during the last forty- five years has changed people's lives so much that people sometimes trust computers more than themselves. The fabrication technology has been improved dramatically last ten years. One typical example is the development of high speed and high density dynamic random access memory (DRAM). In 1984, 64 k-bit DRAM chips were the most advanced memory chips in the market. Today, 4 Mbit DRAM chips, using 0.8 um design rule, are in mass production. 64 Mbit DRAM prototypes, using 0.25 pm or less design rule have already been produced in research laboratories both in the United States and Japan. This rapid development involves better understanding of the fabrication process as well as an improvement of the fabrication tools. Hundreds of papers have been published in recent years about the development and performance of new plasma sources for single wafer fabrication. As the feature size shrinks, conventional parallel plate plasma tools no longer serve the need. A better dry (plasma) etching tool is needed to provide better anisotropic etching profiles and with low damage, especially for electronic devices using III-V materials. Reduction of the feature size to the submicron range requires better plasma tools for the manufacturing process to eliminate undercutting. The ideal tool is one that can produce a dense, broad and low ion energy plasma so that controllable, uniform and highly anisotropic etching can be achieved. 3 A number of tools, which are intended to meet all these goals, have been proposed and studied in the United States and in other industrial countries. These include electron cyclotron resonant (ECR) plasma sources and non- ECR plasma sources. Examples of the non-ECR sources are the parallel plate plasma source which has been used for many years, the Triode source [74] which is a modification of the parallel plate source, the Helicon source from Australia [29], the Inductively Coupled Plasma (ICP) [49] source and the newest Transmission Couple Plasma (TCP) [153] source from Lam Research Corporation. Examples of the ECR sources are the Axial Divergent Field ECR source [17] which was originated from Japan but has also been studied intensively in the United States, the microwave plasma disk reactor (MPDR) developed at Michigan State University [32], the Leybold source from Germany [26] and the Distributed ECR source from France [24]. All these plasma sources are different in structure and principle of operation. The potential usefulness of these different sources in the IC industries is mainly governed by their performance. However, compatibility of the new source to other commercial processing equipment, the initial capital cost and ease of automation will determine the future role of these sources in semiconductor industry. While the discussion of the usefulness of the above different plasma sources for future ultra-large scale integrated circuit (ULSI) application still continues, this research investigates the scale up of the MPDR and evaluates the experimental performance of large diameter electron cyclotron resonant S OUFCCS . §1.2 Research objectives The objectives of the research described in this thesis were (1) to design and build a large diameter electron cyclotron resonant MPDR, (2) to experimentally measure the plasma parameters of the large diameter MPDR, (3) to evaluate the performance of different MPDR prototypes in terms of electron temperature versus CPA, ion production energy cost, microwave coupling efficiency and power density, and (4) to investigate the performance of the large diameter MPDR in submicron etching application. §l.3 Research history The research described in this dissertation was carried out over the period of 1988 to 1993 at Michigan State University under the direction of Dr. Asmussen, Professor and Chairperson of the Department of Electrical Engineering. The experimental work took place in the Microwave Plasma & Microwave Material Processing Laboratories at Michigan State University. Research funding was provided by Wavemat Inc. and Michigan State Research Excellent Fund. This research builds on previous microwave plasma disk reactor research carried out by Mr. Root [32], Dr. Dahimene [35] and Dr. Hopwood [38]. This past reactor research has lead to MSU owned and patented microwave plasma technologies [320] that have numerous microwave plasma and microwave material processing application. Much of the knowledge and engineering art used in the thesis to design, build and experimentally investigate the property of the large MPDR was based on (TI (L! :31? 5 these earlier investigations. A brief history of MPDR development at MSU is given in Chapter Four. Research in this dissertation has been published in part in scientific journals [42,43,45,46] and international conferences [180- 190]. §1.3 Outline of thesis Chapter Two outlines some fundamental properties in electron cyclotron resonant plasmas. Chapter Three is a literature review on all the ECR plasma sources and newly developed non-ECR plasma sources which have been developed in recent years. Structure and operation of these sources are discussed. Chapter Four gives the general aspects of the MPDR sources: its structure, its development, operations principle and optimal ECR magnetic configuration. Chapter Five presents all the experimental diagnostic techniques and results of two large diameter MPDRs, the MPDR 20 and MPDR 325, and one small plasma source, the MPDR 5. Chapter Six evaluates the performance of five MPDRs in terms of the electron temperature versus the product of pressure and diffusion length of the discharge volume, ion production energy cost, microwave coupling efficiency and power density. One application of the large diameter MPDR is anisotropic etching of submicron features in silicon which is described in detail in Chapter Seven. Etch performances versus various operation conditions are discussed. Finally, Chapter Eight is a summary of research results. Also included are suggestions for future development and research in MPDR. CHI CHAPTER TWO FUNDAMENTALS § 2.1 introduction § 2.2 Plasma 2.2.1 Debye - Hiickel radius 2.2.2 Mean free path and collision cross-sections § 2.3 Motion of charged particles in time-harmonic electric field § 2.4 Conductivity in plasma 2.4.1 AC conductivity 2.4.2 Power absorption in plasma § 2.5 Sheath and ion energy § 2.6 Magnetic confinement of plasma liar} § 2.1 Introduction This chapter reveals some of the fundamental properties of plasmas, which were encountered frequently throughout the research. Some plasma parameters such as Debye radius, plasma frequency, neutral density, mean free path, collision cross sections and so on are first stated in early sections. These parameters will be useful for discussions in other parts of this dissertation. A general solution for the motion of a charged particle in a time- harmonic electromagnetic field is solved. This general solution describes microscopic charged particle motion in a collisionless plasma and magnetized plasma with a time-harmonic electric field. The energy absorbed by an electron in electron cyclotron resonance is studied. Macroscopic properties such as power absorption in a collisional plasma is also examined in detail. The discussions of sheath potential, ion energy and plasma confinement are also included. §2.2 Plasma Plasma is the most widespread state of matter in the Universe. It is an ionized gas consisting of three components- free electrons, positive ions and neutral atoms or molecules. The Sun and the stars are enormous lumps of hot plasma. The Earth's atmosphere is surrounded by a plasma envelope known as the ionosphere. In the terrestrial environment, plasmas are encountered in the laboratories and technologies. One typical laboratory plasma is the high- temperature plasma, whose electron and ion energies are in the range of 8 MeV, in the application of controlled nuclear fusion and magneto- hydrodynamic conversion of heat energy into electric energy. Another kind of plasma is low energy plasma, whose electron and ion energies vary from a few electron-volts to tens of electron-volts. Most of the low energy plasma is found in the fluorescence tubes. In semiconductor industries, low energy plasmas have also been used in the fabrication of microelectronic devices for many years. The development of a low pressure, low ion energy and high density plasma has become one of the important technologies for the fabrication of ultra large scale integrated circuit (ULSI). Plasma is generated by ionization processes. Two very common processes are photoionization and electric discharge. In the photoionization process, ionization occurs when the energy of an incident photon is equal to or greater than the ionization potential of the absorbing atoms. For example, the ionization potential of the outermost electron in an atomic oxygen is 13.6 electron-volts, which can be supplied by photon with wavelength less than 910 A, i.e., in the far ultra violet region. In nature, the Earth's ionosphere is an example of photoionization phenomenon. Ionization can also be produced, which is found in most laboratory plasma, by applying an electric field across the ionized gas. The electric field accelerates the free electrons to an energy level sufficient to ionize an atom during collisions. One characteristic of this process is that the transfer of energy from the electric field to the electrons is more efficient than to the heavy ions. Such inefficient energy transfer to the ions is because the ions 9 are much heavier than the electrons. The electrons are accelerated by the applied electric field so quickly that the ions, relatively, seem to be not moving at all. As a result, the temperature of the electrons in a gas discharge is usually much higher than the ion temperature. This is why most laboratory plasma can be modeled as a cold plasma in which the ions are stationary but the electrons are free to move. The transfer of energy from electrons to the ions during elastic collisions is very inefficient because of the difference in mass. It has been estimated that the energy transfer from an energetic electron to an ion during a collision is no more than a fraction of 4m6/mi, where me and mi are the mass of electron and ion respectively [69]. Beside elastic collisions, inelastic collisions also take place in which energy is transferred from the energetic electrons to the ions and neutral through excitation [74]. The ions or neutral species are elevated to different excited states momentarily and then drop back to lower energy states during which emissions of particular wavelengths take place. Because different species have different emission spectra or "emission signatures", the composition of the species can be determined through optical techniques. Laser induced fluorescence technique is based on these absorption and emission phenomena to examine different species inside the plasma. If ne, ni and N represent the densities of electrons, ions and neutral molecules respectively, then the degree of ionization is defined as [69]:- (xi =n /(n+N) 10 where n equals to 116 and ni. When “i is larger than 104, the plasma is regarded as a strongly ionized gas in which coulomb interactions become dominant [70]. Plasma encountered in this research, in fact, belongs this category. For example in a 24 cm diameter discharge such as the MPDR 325 plasma source, the argon plasma density at 0.6 mTorr was 1.44 x 1011 cm'3 using 500 watts 2.45 GHz microwave. The neutral density according to equation (2.4) is 1.93 x 1013 em3 if the neutrals are at room temperature 300K. The corresponding value of ori is 0.0075. That means the plasma in this study is a highly ionized plasma. Important plasma parameters, which can be found in most plasma text [54, 55, 60-63, 67-73] are briefly studied below. These parameters help to understand some plasma phenomena and are useful for discussion in later sections. 2.2.1 Debye -Hi‘rckel radius In an ideal cold plasma, when a test charge is introduced inside the plasma, the fast moving electrons will redistribute themselves around the test charge. Such phenomenon is called plasma screening. The potential around the test charge will drop exponential due to this screening effect. In polar coordinates, the potential drop with respect to the radial distance, r, is given as [67]: (2.5) while the collision frequencies between electrons and ions, and electrons and neutrals are: Vei =1.94x10-12 he ln[ l.55x1013(kae/e)3’Z/(ne)%]/(kae/e) (2.6) ven = 6.69x105anen(ka,./e)5’2 (2.7) where an is the collision cross-section for electron and neutral. However, Kaufman and Robinson pointed out that the collision cross-section often has error of i20-30%. Thus the numerical data can only represent the order of magnitude of the collision processes. § 2.3 Motion of charged particles in time-harmonic electric field The motion of a charge particle, such as an electron, under the influence of time-harmonic electric field and static magnetic field is a classic problem in electron cyclotron resonant plasma. A closed form solution is possible though complicated. Figure 2.3 shows a snap shot of the motion of a charge particle in a 3- D space. The orthogonal velocity components of the particle are represented by v”, vx and v_L respectively. u” denotes the direction of a unit vector whose direction is parallel to the static magnetic field, B. ux is a unit vector which is perpendicular to both the static magnetic field and the applied time- l6 ExB u = x lExBl “i=“llxux V = Vx llx + Valli + V//“// Figure 2.3 Snap shot of the motion of a charged particle in a time-variant B field and static B field l7 harmonic electric field, E(t). u_L is in the direction u” x ux. For the sake of clarity, a positive charge is assumed. The solution derived below, however, can also be applied to the electrons. The motion of a particle with charge, q, and mass, In, is governed by the following equation: v=(q/m)(E(t) + v x B) (2.8) As shown in Figure 2.3, the time-harmonic electric field can be decomposed into two components: E(t) = “/IE/l aim! + uiElejm‘ (2.9) After substituting equation (2.9) into (2.8) and rearrangment, we have: v” = (q/m)E// el‘m (2.10a) Vi = (C1/1‘l'l)E_1_€.Tmt - (1)be (2.10b) vx = rubyi (2.100) where o) is the angular frequency of the time-harmonic electric field and (ob is the cyclotron frequency which is equal to qB/m. 18 Assuming that the positively charged particle has non-zero velocity at initial state, equations (2.10a-c) are first converted to s-domain by using Laplace Transform: SVI/(S) - V//(0) = (Cl/HUB” /(S-j(0) (2.1 la) svi(s) - vi(0) = (q/m)E_L /(s-jo)) - (obvx(s) (2.1 lb) svx(s) - vx(0) = (obv_L(s) (2.11c) The solutions of v//(s), V_L(S) and vx(s) can therefore be solved and the final solutions can be obtained by inverse-Laplace Transform. All the intermediate steps are skipped for clarity and the final solutions are: V” = QB” sin(u)t)/(mo)) + V ”(0) (2.12) vi = qE_L[cobsin((obt)-(osin((ot)]/[m(mb2-c02)] + vJ_(0)cos(mbt) - vx(0)sin((obt) (2.13) vx = quwb[cos((ut)-cos((obt)]/[m((ob2-(02)] + vi(0)sin(wbt) + vx(0)cos(wbt) (2.14) When (0 equals to (ob, it can be shown that both vi and vx will grow with time, a condition for cyclotron resonance: vi = qu[(t/2)cos((obt)+(sin((obt))/(20)b)]/m + vi(0)cos(cobt) - vx(0)sin(cobt) (2.15) Vx = qu(t/2)lsin(wbt)l/m + vi(0)sin((obt) + vx(0)cos((obt) (2.16) 19 For an electron, the cyclotron resonance takes place when the magnetic field strength is 875 Gauss using 2.45 GHz microwave excitation energy. At 915 MHz excitation, the required magnetic field strength is 327 Gauss. The energy gained by an electron under the condition of cyclotron resonance can be obtained by me(v//2+V_L2+vX2)/2. If the initial velocity of the electron is zero, the variation of the energy of the electron with time, t, is: Ee=q2{ 4E //Zsin2(cobt)+E i2[t2wb2+sin2((obt)+twbsin(20360] }/ 8mcob2 (2. 17) Figure 2.4 shows the variation of electron energy with time and E .L, assuming E]: E”. In the absence of collision, an electron gains an amount of 8.8 eV in 10 nanoseconds when E] equals to 2 kV/m. It should also be noted that equations (2.12-2.14) are the general solutions of the charged particle motion. By setting the frequency of the electric field to zero, a drift velocity of El/B in the direction of EXB can be obtained. This drift velocity is the same for both positive and negative charged particles and independent of the mass of the particle. The equations can also be applied to situations such as inhomogeneous static magnetic field in which the magnitude and orientation of the magnetic field are functions of position. The solution given in this section gives a microscopic view of the motion of a charged particle under the influences of a static magnetic field and time-varying electric field. The macroscopic property of a magnetized plasma is discussed in next section. Energy of electron in eV 20 ECR energy coupling 9 I I I I 8- ............................................................................ .. 7,... ........................................................................... .1 6... ........................................................................... . 5.. .............. . 4_ .................................................... E-=-21¢:V-/m ......... . 3... ........................................................................... u 2- ....................................................................... 1.. .............. é.......... ....E ............... i ..... .E-LEl-RYIr-fi .......... ..i o i i i 0 2 4 6 8 10 Time(ns) Figure 2.4 Variations of electron energy with time at electron cyclotron resonance at different applied electric field, EL, strength 21 § 2.4 Conductivity in plasma The absorption of power in a plasma or power dissipation in plasma is the theme in this section. The conductivity of a magnetized plasma is first derived and then the variations of conductivity, which determines the power dissipation, at different static magnetic field strengths and pressures are then studied. 2.4.1 AC conductivity In the presence of a static magnetic field and collision processes, the macroscopic motion of the electrons under the influence of a time-harmonic electric field is described by the Langevin equation [68,70]: me(av/at) = -e(E + va) - mevcv (2.18) where v is the average electron velocity, E is the electric field and vc is the collision frequency between the electrons and the neutrals. It is assumed that the electrons are free to move in a stationary uniform background of ions and neutral which provide a viscous damping force to the motion of the electrons. Consider both the electron velocity and the electric field vary harmonically with time, 6.10”, equation (2.18) becomes: mefiwv) = -e(E + va) - mevcv (2.19) Becausi Using [1 fifild Silt field. a expresse Where another nar 0f the eieq; 22 Because J = -enev (2.20) Using the same notations and orientations of the magnetic field and electric field shown in Figure 2.3, the relation of the current to the applied electric field, after substituting (2.20) into (2.19) and rearrangement, can be expressed as: J = o E (2.21) where z T 6.1. —OX 0 T 0': l ox o] 0 I (2.22) I. O O 0'” J cl = (vc+jw)zoo/[(vc+jw)2+(oce2] (2.23) ox = (vc+jw)wceoo/[(vc+jw)2+(occz] (2.24) 0,, 2 0'0 = nee2/[me(vc+j(o)] (2.25) The conductivity, 3, is a tensor, indicating the anisotropic properties of a magnetized plasma. Cl is also called Pederson conductivity [70], ox has another name called the Hall conductivity and woe is the cyclotron frequency of the electrons which is equal to eB/me. 2.4.2 equaui The p0 part of mecon Paris of Sil‘Engtf Should i the Sam magnem the 1W0 ‘ Gauss a mli‘rowa. magnen'c 23 2.4.2 Power absorption in plasma The ohmic loss or power dissipated in the plasma is given by the equation: Pdiss = J ' E = Re(O’_L) IE i|2+ 07/ IE//l2 (2.26) The power absorption by a collisional plasma is thus proportional to the real part of the perpendicular conductivity, oi, and the parallel components of the conductivity. Figure 2.5 and Figure 2.6 show the variations of the real parts of the conductivity, oi, with different ratios of vcju) and magnetic field strength at 2.45 GHz and 915 MHz excitation frequency respectively. It should be noted that when the static magnetic field strength is zero, oi has the same value as 07/. At low ratio of vc/(o, maximum power absorption occurs when the magnetic field strength is close to the value required for ECR condition at the two different excitation frequencies, i.e. 327 Gauss at 915 MHz and 875 Gauss at 2.45 GHz. In a real environment, the electric field of the microwave penetrates into the plasma which has a spatial variation of static magnetic field strength. Assuming that the static magnetic field is always perpendicular to the electric field, the variations of conductivity, oi, with magnetic field strength are then displayed in Figure 2.7 and Figure 2.8 at the two excitation frequencies which were used in the experiments. The plasma density is assumed to be 1011 cm'3 and the electron temperature is assumed to be 5 eV or 58000 K. Since collision frequencies are related to the neutral 24 1.8 W l r r “‘ r p—o O O‘ l f" h r 0.08 T " t" N 9 oo Ne = lEll per cubic cm ‘ Excitation frequency=2.45 GHz Real part of conductivity , cl P as 0.4 . 0.2 . . 0 J l l L O 0.5 1 1.5 2 2.5 3 collision frequency/excitation frequency Figure 2.5 Variations of real part of conductivity, oi, with ratio of collision frequency/excitation frequency at different static magnetic field strength at 2.45 GHz excitation frequency Yeoi Rc-ul part ol‘cnnductivir .CD Ch 1.4- } J :3 OD .0 LA 25 1.6 ‘I I I I I Z“ A I 0.03 T - p— O N I Ne = lEll per cubic cm - Excitation frequency=915 MHz Real part of conductivity , Oi l l 0 0.5 1 1.5 2 2.5 3 collision frequency/excitation frequency Figure 2.6 Variations of real part of conductivity, 0], with ratio of collision frequency/excitation frequency at different static magnetic field strength at 915MHz excitation frequency Real part of conductivity , GI 26 10 : I T I I I I I D I b P d l 10;- l 1: F ‘ Ne=lEll per cubic cm, Te=5 eV . Excitation frequency=2.45 GHz 1 bud O. .— 'I' u—A O. N 'I l “V H or t» V ' I'""" 10 ;- P=5 mTorr 1 1 10" .- P=l mTorr E P=0.6 mTorr ' 10*5 0 500 1000 1500 2000 2500 3000 3500 4000 Magnetic field strength (Gauss) Figure 2.7 Variations of real part of conductivity, oi, with magnetic field strength at different pressures with 2.45 GHz excitation frequency 27 10 E I r I I I I I 100 =- . 10-l :- _ i Ne=1E11 per cubic cm, Te=5 cV : 10’2 T Excitation frequency=915 MHz -. I (a) v Real part of conductivity , 0.1. 8 L 10 - P=5 mTorr -. 10-5 .- P=1 mTorr : I 1 E P=0.6 mTorr ‘ 10*S 0 500 1000 1500 2000 2500 3000 3500 4000 Magnetic field strength (Gauss) Figure 2.8 Variations of real part of conductivity, oi, with magnetic field strength at different pressures with 915 MHz excitation frequency densit are ih pressu absorr strengt. Figure Shows ' Shown sources reaches mai-‘Iieii Pel'pend: the othe energy a Coll-Side“ Elm“TIC j important are Cmcia The l. , . sreddlly( 28 density as stated in equations (2.6) and (2.7), the variations of conductivity are therefore related to the pressure. Figures 2.7 and 2.8 show that as the pressure increases, the conductivity increases which means that power absorption by the plasma also increases. At 0.6 mTorr, the variations of conductivity, oi, with magnetic field strength at different plasma densities are shown in Figures 2.9 and 2.10. Figure 2.9 show the variations at 2.45 GHz excitation while Figure 2.10 shows the variations at 915 MHz excitation. The plasma densities selected shown are the typical experimental values found in most ECR plasma sources. Figures 2.7 to 2.10 indicate that power absorption by the plasma reaches a maximrun when the magnetic field strength is equal to the ECR magnetic strength. This implies that the electric field, which is perpendicular to the static magnetic field, decays faster in ECR region. On the other hand, it also means that a large ECR mung; can have a higher energy absorption. These are two of the aspects which are needed to be considered in the design of an ECR plasma source. The orientation of the electric field with respect to the static magnetic field pattern is also important. Other issues such as ion energy, plasma stability and uniformity are crucial in production environment. The dc conductivity of the electron gas is not used in this analysis but is readily obtained by setting a) to zero in equations (2.23 - 2.25). y.()’_L lite"! purt ut'cunductivil 29 10 E u r r . , , . ‘ F 1 10' - o i 10 g i Excitation frequency=2.45 GHz 1 10-1 P=0.6 mTorr, Te=5 eV T Ne=1e12 per cubic cm Real part of conductivity , 0L 8 10 g .5 ’ Nc=le11 per cubic cm i 10 r 1 10" l T Ne=1e10 per cubic cm 10'7 0 500 1000 1500 2000 2500 3000 3500 4000 Magnetic field strength (Gauss) Figure 2.9 Variations of real part of conductivity, oi, with magnetic field strength at different plasma densities with 2.45 GHz excitation frequency 30 10 E I I I I I I I r 10 r i 100 r 1 6T I Excitation frequency=915 MHz ‘ g. 10'1 r . P=0.6mTorr, Te=5 eV 5 -2 "g 10 1 o o : “a i 10‘3 - - r55. , Ne=le12 per cubic cm 8 F“ 10'4 r ‘3' _5 Ne=1e1 1 per cubic cm i 10 r a Ne=le10 per cubic cm 10 - i ‘ i 10‘7 ' 0 500 1000 1500 2000 2500 3000 3500 4000 Magnetic field strength (Gauss) Figure 2.10 Variations of real part of conductivity, oi, with magnetic field strength at different plasma densities with 915 MHz excitation frequency §2.5: 1 phnna ethor therefor aceelerz electror. electron Phunai are pl‘lrr. Temptirai the dowr Br dlffemnc zone, S: imperfw into the l 31 § 2.5 Sheath and ion energy Because of small mass and at high temperature, electrons in ECR plasma move much faster than the ions inside plasma. When the fast moving electrons reach a surface, they lower the potential of the surface. A sheath is therefore formed in which low energy electrons are reflected and ions are accelerated. In other words, there is a transfer of energy from the energetic electrons to the ions inside the sheath. Because of the loss of the energetic electrons, the temperature of a discharge, electron temperature, drops as the plasma diffuses away from the generation region. These energetic electrons are primarily lost by diffusion towards the vessel wall. The drop of electron temperature is witnessed experimentally by a drop of plasma potential along the downstream position. Bohm's criterion states that a plasma itself cannot support potential differences greater than zkae/2e without breaking down into a space-charge zone. Stangeby [71] indicated that the shielding effect of the sheath was imperfect and a small residual field called the pre-sheath penetrated deep into the plasma. The potential drop of this pre-sheath is determined by the Bohm's criterion and therefore the ions enter the sheath with a velocity component towards the surface, which is called the ion acoustic velocity: via = (kTe/MiW’ (2.27) This is the minimum speed of the ions, which is sometimes neglected in computer simulation of ion energy distribution, entering the sheath. 32 The energy of an ion impinging a surface, such as a substrate, is therefore the summation of its thermal energy in the bulk of the plasma, the pre-sheath potential, the sheath potential between the substrate and the plasma, and the additional negative potential applied to the substrate. When a substrate bias is used, the ions will strike on the substrate surface at nearly 90 degrees. In ion assisted etching, a vertical profile can therefore be achieved. Although the incident angle of the ions impinging the substrate surface can be controlled by the applied bias, a high substrate bias is not desirable because issues such as damages to the etch mask and the substrate surface have to be considered. Therefore an ideal plasma source is one that can create a plasma with a low ion thermal energies and low sheath potential. In experiment, the ion energy distribution can be directly measured by an ion energy analyzer. Details of the measurement techniques and experimental results are discussed in Chapter Five. § 2.6 Magnetic confinement of plasma The concept of magnetic confinement has been implemented by both ECR and non- ECR plasma sources to improve the uniformity and density of the plasma in the processing region. It is an old technique used in plasma research [76-78]. In practice, the plasma pressure always has to be small compared to the pressure of the confinement magnetic field [54]: 132/2p0 >> 2neTekb (2.28) where no is the permeability in vacuum, B is the magnetic strength in Tesla, nc is the plasma density per m3, kb is the Boltzmann constant and TC is the 33 electron temperature in Kelvin. Considering that the plasma density equals to 1011 cm'3 and Te equals to 5 eV, the required confinement magnetic field is: B >> 6.3 Gauss (2.29) Therefore a 100 Gauss magnetic field is sufficient to confine the plasma. The magnetic field for plasma confinement can be generated either by magnetic coils or permanent magnets. Permanent magnets were used in this research to create a multi-cusp magnetic field for downstream plasma confinement. An improvement of the plasma distribution in the downstream location has been observed [43] and is reported in later chapter. CHAPTER THREE ECR PLASMA SOURCES § 3.1 Introduction § 3.2 Axial divergent magnetic field ECR plasma sources § 3.3 Distributed ECR plasma source § 3.4 Hom-type ECR plasma sources § 3.5 Other plasma sources 3.5.1 Helicon source 3.5.2 Helical resonator 3.5.3 Inductive coupled plasma sources § 3.6 Summary 34 35 § 3.1 Introduction In 1977, two apparatus [1,2] were announced by the Central Research Laboratory in Hitachi Ltd., Japan. The first one was a microwave ECR ion source [1] which had an antenna inside the discharge region. A strong axial magnetic mirror, whose field intensity was stronger than 890 Gauss, was generated by three electromagnetic coils to prevent erosion of the antenna due to bombardment of charged particles. This ion source was primarily used for ion beam processing [7] such as ion beam sputtering deposition, ion beam etching and non-mass-analyzed ion implantation. The second apparatus [2] was a modification of the first one. Microwave energy was coupled into a round waveguide and eventually propagated into a 66 mm in inner diameter quartz tube which was placed at the end of the round waveguide. The axial magnetic field near the opening of the quartz tube was adjusted to 875 Gauss. This is the first divergent field microwave electron cyclotron resonance (ECR) plasma source used in semiconductor etching. This source initiated a series of intensive research concerned with the application in semiconductor processing. During the last fifteen years, experiments were carried out first in Japan [3-17] and then more recently in the United States [18-23]. Other ECR sources were also developed worldwide such as the Distributed ECR source (DECR) in France [24,25], the Hom-type ECR source in Germany [26-28] and the multipolar ECR Microwave Plasma Disk Reactor (MPDR) [32—48] in Michigan State University. Even though all the ECR sources use the same principle to generate the plasma, they have different geometries and coupling designs. It was not clear if these different designs resulted in different output plasma 36 properties. Thus many experimental and theoretical investigations have been devoted to the characterization of each source as well as evaluation of the different sources in specific applications. This chapter aims to provide a general review of the divergent field ECR source, DECR and Horn-type ECR source. The MPDR source is discussed, separately, in detail in the next three chapters. This chapter also discusses other types of high density plasma sources such as the helicon source [29-31] from Australia and the recent planar radio frequency inductive coupled plasma source (ICP) from IBM [49,152,182,183]. The Transmission Coupled Plasma (TCP) [153,154] source, which looks very similar to the ICP, is now a commercial product for industrial application. The driving force behind the development of all these new reactors for semiconductor processing, especially etching and deposition, is the demand of high anisotropy, high uniformity, high density, low ion energy and low temperature plasma for fabrication of ULSI devices. Non ECR plasma sources suffer from the fact that the operation pressure has to be above 10 mTorr to maintain a low sheath potential. However, at high pressure, the control of anisotropy is lost. For conventional parallel plate devices, the input power directly determines the energy of the ions impinging the substrate surface. This lack of independent control of ion energy is the main draw back of parallel plate plasma sources. Another important issue which ULSI technology has to deal with is the damage to the substrate. ECR plasma sources, in general, have lower ion energies and can be biased independently so that the energy of the ions impinging on the substrate can 37 be controlled. This is especially important when processing moderately doped III-V semiconductor materials. § 3.2 Axial divergent magnetic field ECR plasma sources The axial divergent magnetic field ECR plasma source is the most extensively studied high density plasma source. It was the earliest ECR source which was applied to plasma processing application such as silicon etching [2]. Many variations of this source have been made both in Japan and in the United States. Research using this divergent field ECR plasma source includes etching [2-4,5,10,15,17,52,53], deposition [5,11,51], ion beam process [l,8,9,13], sputtering deposition [12], reduction [50], hydrogen annealing [147], wafer cleaning [148] and plasma immersion ion implantation [150,151]. The basic structure of a divergent field ECR plasma source consists of a cylinder with radius, d, and length, L, as shown in the Figure 3.1(a). This defines the discharge region. Two magnetic coils are wrapped around the cylinder to create an axial, static magnetic field. Microwave energy is introduced into the discharge region from a waveguide through a vacuum sealed quartz window. The ions and electrons created in the discharge region then diffuse into the downstream processing region where wafer is being processed. Variations of magnetic configuration in these ECR plasma sources include gradient field ECR and mirror field ECR [8]. Figure 3.1 shows (a) a gradient field ECR configuration and (b) a mirror field ECR configuration. Position Position 38 Quartz uw WlndOW 875 G Bz > , , Magnetic coils L L z=0 F—~I" l Downstream (a) Gradient field ECR configuration ECR uw surface 875 G BZ . 1 wafer ’2 % z=0 Downstream >3 (b) Mirror field ECR configuration Figure 3.1 Divergent field ECR plasma sources 39 In the mirror field ECR configuration, a third coil is placed at the same horizontal level of the wafer. As a result, a magnetic mirror is formed as shown in Figure 3.1(b). The mirror field ECR source has a lower ion energy than the gradient field ECR source. As the third coil current increases from 0 A to 8 A at 0.45 mTorr, the ion energy decreases from about 13 eV to 5 eV and the FWHM (width of ion energy spectrum at 50% maximum) decreases from 8 eV to 5 eV [8]. In the gradient field source, the magnetic field lines penetrate the substrate at different angles. Ions and electrons move along these magnetic flux lines by ambipolar diffusion and impinge on the surface, resulting in different etched profiles in the center and the edge of the wafer [52]. The problem is solved by the third coil in the mirror field ECR configuration, resulting in collirnated magnetic flux lines which are perpendicular to the substrate surface. Thus, beside achieving lower ion energy, the mirror field ECR also shows a better etching profile. Sarnujkawa et al. reported that when a 87 5 G magnetic strength was maintained right above the substrate, a uniform etched profile could be obtained [52]. The size of the divergent field ECR plasma source has been increased from the original 66 mm in inner diameter and approximately 150 mm high [1,2] to a recent version 200 mm in inner diameter and 197 mm high [5,52]. It is believed that the length of the discharge is critical to create a stable plasma [16]. Several different operation modes, such as TE112 [8], TE113 [5,12,18] and TM01 [23] modes have been reported. 40 Table 3.1 shows the typical discharge properties of the divergent ECR plasma sources reported in the collected literature. This data represents measurements taken in different sources at different locations. However, all measurements were taken in downstream locations as shown in Figure 3.1. Etch uniformity is one of the concerns in this divergent field ECR source. Because the motion of the ions and electrons are restricted in the radial direction, any nonuniforrnity of the plasma densities in the plasma generation region will be projected onto the wafer if it is positioned close to the ECR surface. This was reported by a study made by Iizuka et a1. [16]. He suggested that higher TE modes such as TE12 and TE13 modes might have been excited in his source. Figure 3.2 and Figure 3.3 display the distribution of the electric field strength and its orientation for the TE12 and TE13 modes respectively. Derivations of the mode equations and their interpretations are discussed in detail in Chapter Four. Therefore they are not repeated here. In each plot, the central axis of the divergent ECR source is represented by the coordinate (41,41). Iizuka reported that a non-uniform etched result was obtained whose pattern matched with the electric field strength patterns shown in the Figure 3.2 and Figure 3.3. Within the discharge region, strong electric field shown in the figures creates more ions. Since lateral diffusion of the ions was restricted because of the axial magnetic field, a non-uniform etch rate was observed when the wafer was placed close to the source [16]. Substrate bias was usually used during etching. However, no self- induced bias can be obtained if the chuck is too close to the discharge generation region. This is because the axial magnetic field prevents the flow of electrons to the grounded side wall. Due to the lack of grounded surface, 41 TABLE 3.1 PLASMA PARAMETERS FOR DIVERGENT ECR SOURCE Pressure Gas Plasma] Power Ne Te Ion energy Ref # (mTorr) Potential(volts) (W) (cm'3) (eV) (eV) 3 Ar - 400-900 2-7 E12 - - 1 1 H2 - 400-900 1-2 E12 - - 1 0.45 Ar - 100 - - 12-48 8 0.2- 5 Ar - 120-600 0.2-1 E11 3-6 - 11 0.6-3 H2 - 120-600 1.2E10 5-8 - 1 1 0.3-3 N2 - 1000 - - 12-17 15 0.5 C12 - 1000 - - 14 15 0.4 Ar 10-40 100 3-4 E10 4-4.3 - 16 0.4-30 02 12-20 200-300 1-2 E10 - - 19 0.3 Ar 20-30 500-600 6-8 E10 6-12 - 20 42 "5'2 MODE: Electric field strength Tiar- ,;1;'.’\\\‘l‘l‘i( “I" . “\Tllllm " \lllllllli‘i l were \\“ \l ..\ u‘ uuuuu. \““\\\\\\\\\i‘“ 0.; o o o'.‘ '/ - 0 . co 9 .— o ' ’- —— ' ' .— '- ‘f’ .I ~ " ii i i , I ' «6.1/Ill”) Uri/I, "l." ‘.r ‘90:, ”’1’, Ill!” 4" O -9 -o - -- ’_. ..- .. 1 ill 1 “(TEN l - 1 ll 1 TT 1 1 ll TI‘t‘llTill .ll“ - - ’6‘. '.. - Figure 3.2 Distribution of the electric field (a) strength and (b) orientation in TE12 mode 43 T513 MODE: Electric field strength \ iii ll‘l‘ il‘l'i‘il‘l‘ ii“ “iii \lllll‘ 1.111,...ui‘i‘ii‘l“ . a] 7, '51:) Iii," Till," I“ (l 700'." [Ill/I’ll!- "MT i‘l‘i Milliii T513 MODE: Orientation of electric field Figure 3.3 Distribution of the electric field (a) strength and (b) orientation in TE13 mode 44 no induced d.c. bias can be estabilished in the chuck which carries the wafer. This phenomenon was reported by Dane et al. [114]. Another concern of this type of source is switching of modes during a process. Separate reports show that there are two different operation modes. One is called the high (density) mode and the other is called low (density) mode [23,146]. The difference of the two modes is witnessed by a change of plasma density and plasma potential with little or no change in external input power [23]. Aydil et al. [146] reported that switching of the modes, during the process, could not be predicted. Such phenomenon of switching mode was also observed in a similar source with TM01 excitation by both Carl et a1. [23] and Gorbatkin et al. [184]. More work is needed to be done in this source to provide a controllable and repeatable process at specific operation conditions. § 3.3 Distributed ECR plasma source This type of plasma source, originated from a group of French researchers, is one of the two ECR sources developed in Europe. The concept was evolved from another idea called Microwave Multipolar Plasma (MMP) [156]. The MMP, in fact, is a plasma magnetically confined in a chamber. One or more separate plasma sources, such as the divergent field ECR plasma sources, are used to generate the plasma. The idea of the Distributed ECR source is that those magnets used to confine plasma have also an ECR magnetic field strength of 875 Gauss near the pole in each magnet. As shown in Figure 3.4, an antenna is then placed at a nearby position to generate the plasma. Figure 3.4 Distributed ECR plasma source 46 As shown in Figure 3.4, eight rows of permanent ferrite magnets with 1500 Gauss pole strength are arranged outside a grounded chamber of 28 cm in diameter [25]. Microwave energy is fed into the chamber through eight 8 mm O.D. tubular stainless steel antennae which are placed inside the plasma. Each antenna has a length of 10 to 40 cm long and is cooled by water. A single microwave generator and a matching device are used to couple the microwave energy to all antennae. Table 3.2 shows some typical plasma density and electron energies that are achieved with this source. The method of plasma excitation in DECR source suffers from the fact that the antennae embedded inside the plasma will be a constant source of contamination due to the bombardment by the high energy ions or reactive species which reach the surface of the antennae during an etching process. When the source is used in deposition application, the deposited material on the surface of the antennae may degrade the performance of the antennae. Because of these two reasons, the tubular antennae may have to be replaced periodically. Finally, since the wafer is placed inside the chamber and is very close to the antenna array, it is necessary to investigate any damage to the wafer because of the microwave radiation during an etching or deposition process. § 3.4 Horn-type ECR plasma sources These types of hom-type ECR plasma sources were developed in Germany. Two versions have been produced. One is the Elongated-Mirror- Geometry (EMG) plasma source [26] as shown in Figure 3.5. The other, a 47 TABLE 3.2 PLASMA PARAMETERS OF DISTRIBUTED ECR SOURCE Pressure Gas Plasma Potential Power Ne Te Ion energy Ref # (mTorr) (volts) (W) (cm'3) (eV) (eV) 0.4-1.8 Ar 24-30 1600 l-l.5 E11 3-3.8 - 25 48 (l) Horn antenna, (2) microwave transparent window, (3) magnetic race-track system, (4) substrate hold: r. Figlre 3.5 Elongated-mirror- Geometry (EMG) plasma source RF- water 49 Dc-Blaa Mi may Rectangular f.“ ' Short circuitlng waveguide / slide I 0 Horn shaped I radiator Quartz chamber Ring with , \; magnet! In] J l 'm Isl -I . . . u ‘ ‘- - . Precaaa and . Analysis Lanter- alupport | chamber ........... 0 Figure 3.6 Leybold source 50 cross section of which is shown in Figure 3.6, is here referred as the Leybold source [27,28]. The EMG plasma source uses a rectangular horn of dimension 60 cm x 9 cm. It has been used primarily for deposition such as silicon dioxide [26]. The Leybold source uses a circular horn with 20 cm in diameter and is used for etching [27]. It has a dome-shaped quartz disk which is similar to the MPDR described in the next chapter. Both EMG and Leybold sources use permanent magnets to form the ECR magnetic surface. A TMOI microwave excitation mode is used to generate the plasma. Two "plasma modes", ring mode and bulk mode, have been reported from the Leybold source, which are dependent on pressure and input power [28]. The transition from ring mode to bulk mode is accompanied by a change of plasma density and reflected microwave power. A study of mass spectroscopy of the ring mode and the bulk mode indicates a change of distribution of charged species inside the plasma. For example, in an oxygen plasma at 0.75 mTorr, 02+ ions are the majority ions in the ring mode When the microwave power is between 120 watts to 360 watts. When microwave power is higher or equal to 480 watts, the source is operating in bulk mode, whose concentrations of the 0+ and 02+ ions are basically the same. Moreover, during the transition from ring mode to bulk mode, the plasma potential changes from 10 volts to 50 volts. To understand the cause of the ring mode and the bulk mode, the mode diagram of TMOI is examined. Figure 3.7 shows the distribution of the t_ra_r_1_§1erse electric field in TMOI mode. It has a maximum electric field strength off the center, but a zero strength at the center. The 1913101011111 electric field, however, has a maximum strength at the center as shown in 51 TMOI MODE: Transverse E field strength 1 x . n“: I ‘ [1,, ., . “T“\|\‘“\ 5“ r , .‘m‘mu ,‘I/ 7,’ ‘ Him“ “\‘ET‘ “H" W0'1’I’II7 " 0.8 ‘ ii(m“\\“\‘ |“Hon/,7’ 7,117 . \\ t on 7"][17/ \\\\\\ \‘\\“‘ “.“""7,” I 7,! ‘\l\\\\“ \“‘.‘0Wooit,7, 7, I7 ‘ 77,”, “I, 7 (1.6 a . {"71” t .» r u u I u‘ \‘ ul‘ 1, , ‘7 , ‘ u‘ (v i i ll l x mm MODE: -Oriaitau'on of transverse E field 1le 7’ ill llllllllll Figure 3.7 52 mm MODE: Longimainu E field sucngvh I I I. I I ,I 7/7’7’7""‘ ,7,/I,70¢ ‘\\‘ 7 I I I In \ \ ‘III/’7IIII’Ill":‘7'.',"‘.|“l“\“|‘\‘\‘: \\ ‘7/I"I,’I '7" 'I I o o I.‘ s\\‘ \‘ \ 7,, I,II,III . ‘7 ’7 I 7’ no . l I I I ’I I I’ 7,7,7! I I I’ ’II I ’I I ’77 \\\\l\\\\ ' u l. \ ‘ \ \\\\ 'I u H nu \ \\ “mun ’II’H'IH' ‘n.|un\‘\\\\ .7 v’/'l'nuu 0l|\\‘\\ \uu 7,7577. '37” 5.”. u. I“‘|“\“\‘\‘\“\\‘_‘\\\\\ ' ‘ . I n 7 - Figure 3.8 Longitudinal E field strength in TMm mode 53 Figure 3.8. It is believed that in the ring mode, only the transverse electric field is responsible to generate the plasma. In the bulk mode, both the longitudinal and transverse electric fields generate the plasma. § 3.5 Other plasma sources Three types of new sources which are non-ECR plasma sources are also studied both in the industry and universities. They are the Helicon source [29-31,154,157-161], the Helical resonator and the planar rf Inductive Coupled Plasma (ICP) [49,162-164] or Transmission Coupled Plasma (TCP) [153,154]. All these plasma sources use 13.56 MHz microwave energy to generate the plasma. 3.5.1 Helicon source A second name for the Helicon source is the Boswell source. It was developed in Australia [29-31]. Figure 3.9 shows a schematic drawing of a typical Boswell source sitting on a processing chamber A rf antenna is wrapped around a cylindrical glass chamber in which the plasma is formed. The outer side of the rf antenna is surrounded by magnetic coil that creates a static axial magnetic field inside the glass cylinder. The magnetic coil helps to confine the charged species created inside the glass cylinder, thus reducing the diffusion loss of the ions to the glass wall. Typical dimensions of the glass cylinder are 15 cm in diameter and 22 cm long. 54 I Match Network 8‘ 030? '." fifiéo .‘.. ‘ ' ' ozozz O ‘ O A b .0 A Glass chamber f??? 000 ”$39 1 fi A .. o §§ 0 av . Magnetic coil ".' d’flflP ”9“ O A 0.0.0... 0 020203 .' 9 O E E3. 29 ' V V °3°2°1 . . . , O O .0 '...'....'.' 0900000000000 {’3§¢’3§’3§3¢ 'I O. o O .0 OO- .'."...'. 00000000000 Ia§£§§vdfi§o o 0 O O 0.0? .0 ) Wafer Match Network Figure 3.9 Schematic drawing of a Boswell Source 55 3.5.2 Helical resonator Another high density rf source is from Prototech in the United States, called Helical Resonator. Figure 3.10 shows the side view of the longitudinal section of a Helicon resonator. Two differences can be found in this source. First, a helical coil is used to surround a quartz tube. Second, there is an electrostatic shield between the quartz tube and the helical coil. Another major difference between this source and the Boswell source is that no magnetic coil is used. Instead, the whole structure is hosted inside a metallic cylinder as shown in Figure 3.10. The function of the electrostatic shield is believed to be the same as the Faraday shield in ICP2 source which is described in next section. 3.5.3 Inductive coupled plasma (ICP) sources Inductive coupled plasma sources (ICP) have been used as sources of charged species for mass spectrometers for many years. Figure 3.11 is a schematic diagram of an ICP-Mass Spectrometer constructed by Sciex to detect neutral species. The neutral species are forced to flow through a small diameter glass tube into a second but larger glass tube. A high voltage rf is applied to a coil wrapped around at the end of the second glass tube to form a discharge. The ionized neutral species then enter the mass spectrometer and are analyzed. The use of planar inductive coupled plasma in semiconductor processing, however, is relatively new. Yet, it gets considerable attention in the industry since its development is supported by a major computer 56 metallic cylinder RF helix Electrostatic shield with multiple vertical slots on the Q‘mz wall Figure 3.10 Side view of longitudinal section of a Helical Resonator 57 ‘f—he / 5 ICP MS Sample \ (neutrals) @@3 Figure 3.11 Schematic drawing of ICP-MS instrument constructed by Sciex 58 company and a major semiconductor equipment supplier. More than one version of this type of source have been developed, even though they look very similar to each other. The first one is the planar ICP from IBM [49,164,182,183], hereafter referred as ICP]. The second version is from University of Wisconsin in Madison [162,163], hereafter referred as ICP2. The third one is the Transmission Coupled Plasma (TCP) from Lam Research, California [ 153,154]. Figure 3.12 (a) displays the cross-sectional view of the ICP] and the bottom view of the plasma showing the cusp-shaped confined plasma. One characteristic of this source is the use of a rectangular coil on a flat half inch thick quartz plate. The argon plasma density is reported to be increased linearly with power from 1011 cm'3 at 300 W to 6 x 1011 cm'3 at 1.2 kW at l mTorr [182]. The variation of the induced electric field strength at 2 inches under the coil is shown in Figure 3.12 (b). Typical argon ion production efficiencies are about 175 eV/ion at l mTorr [182]. The source has been used for ashing application [49]. The ICP; is basically the same as the ICP1 except two major differences. As shown in Figure 3.13, the first difference is that a circular coil is used instead of a rectangular coil. The second difference is a major improvement of the ICP1 that a Faraday shield is placed in between the coil and a one inch thick quartz plate as shown in Figure 3.13(b). This Faraday shield serves the same purpose as the shield used in the Helical Resonator that the plasma is created by rf inductive coupling rather than capacitive coupling. matchning '“‘ network T substrate holder g 1 confinement magnet coil Figure 3.12 (a) A shematic diagram of the ICP] w position k ) Figure 3.12 (b) Variation of induced electric field 2 inches under the coil 60 Spoke Faraday shield (a) RF coil / E. Vacuum 1“ chamber J \ Brf (b) Figure 3.13 Schematic illustration of ICP2 61 It is believed that during start up or at low plasma densities, without a Faraday shield as shown in Figure 3.14, the high voltage in the coil and any ground surface in the system including the substrate holder behave like the two electrodes of a capacitor. Plasma is generated in part by the electric field, EC, between the two "electrodes" and by the electric field, Bin, induced by the time varying magnetic field, Brf. The presence of the capacitive electric field, EC, caused fluctuation of plasma and caused sputtering on those ground surface near to the coil. With the present of a metallic Faraday shield as shown in Figure 3.13(b). the capacitive electric field, EC, is isolated from the plasma. In this case, only the induced electric field, Bin, is responsible for the generation of the plasma. The vertical drilled through slots in Helical resonator and the radial spokes in the Faraday shield probably serve the purpose of removing power loss due to eddy current in the shield. The ICPz has been used for polymer coating using HMDSO (hexamethyldisiloxane) [163] at 13.56 MHz with 1.2-1.6 kW power. Other applications include optical film coating, fluorination of thermal plastic and chemical treatment of synthetic and natural fiber surface [186]. The Transmission Coupled Plasma (TCP) is a planar ICP source marketed by Lam Research. No detailed structure of the source has been published. The source is claimed to be a variation of the previous two ICP sources. It can be operated from 1-10 mTorr while most of the ICPz applications are at 20 mTorr or above. The TCP has demonstrated anisotropic etching of 0.3 pm features [153] in polysilicon and has an etching uniformity of 5%, 3 a over 200 mm wafer probably with 62 High voltage RF coil Quartz Bi Vacuum “ chamber I Chuck 1 Grounded chamber \. J: Figure 3.14 Illustration of capacitive and inductive coupling 63 downstream confinement magnets. The etch rate varies from 0.3 to 1 pm per minute, depending on the gas chemistry and the type of substrate film. § 3.6 Summary All the ECR sources and the non-ECR sources described in this chapter are developed for future ULSI applications. Three, but not the least, important issues which all these plasma sources have to fulfill are low damage, high uniformity and scalability of the source for large area processing. The Microwave Plasma Disk Reactor is one type of ECR source which fulfills all these three requirements. In next three chapters, the structures, operation and performances of the MPDRs will be examined. The capability of the MPDR in etching of silicon features down to 0.1 pm is then presented in Chapter Seven, which demonstrates that the MPDR, especially the large diameter MPDR, may play an important role in semiconductor processing. CHAPTER FOUR W § 4.1 Introduction § 4.2 Structure of the MPDR 4.2.1 Structure of the MPDR 20 4.2.2 Historical development § 4.3 Operation of the MPDR 4.3.1 Electromagnetic resonant modes in cylindrical cavity 4.3.2 Ignition and tuned positions in MPDRs 4.3.3 Mode of operation 4.3.4 Optimal magnetic configuration 64 65 § 4.1 Introduction The Microwave Plasma Disk Reactor (MPDR) is one type of the ECR plasma sources developed for future Ultra Large Scale integrated circuits (ULSI) application. Five different sizes of ECR Microwave Plasma Disk Reactors have been fabricated and investigated in the MSU laboratories. Investigation of two large diameter reactors, the MPDR 20 and MPDR 325, is one of the major subjects in this dissertation. Properties of other smaller MPDR plasma sources, which have not been investigated, are also included in this dissertation. Experimental investigations of the MPDR sources are discussed in Chapter Five and Chapter Six. This chapter details the structural and historical development of the MPDR plasma sources. The operation of the MPDRs is also described in detail. Electromagnetic resonant modes in a perfect cylindrical empty cavity are also presented for the illustration of the excitation modes used in most of the MPDRs. They are also used in the discussion of optimal magnetic configuration for the MPDRs. § 4.2 Structure of the MPDR The structure of the large diameter MPDRs, the MPDR 20 and MPDR 325 are described in detail in this section. MPDR 20 has a discharge diameter of 20 cm while the MPDR 325 has a discharge diameter of 24 cm. All the features found in these two reactors can be found in other small prototype reactors developed in the laboratory. All MPDRs share two basic 66 features: a cylindrical microwave resonant cavity enclosing a quartz discharge chamber and a multicusp ECR magnetic configuration. § 4.2.1 Structure of the MPDR 20 MPDR 20 was the first large diameter ECR MPDR developed in the MSU laboratories. It was also the first 915 MHz ECR plasma source developed for plasma processing. Besides operating at 915 MHz microwave frequency, it can also be operated at 2.45 GHz. Figure 4.1(a) displays the longitudinal cross section of the reactor. Microwave energy is coupled into the 45.7 cm diameter x 45.7 cm high aluminum cavity by a microwave input probe <3>. By adjusting the position, Lp, of the microwave input probe and the height, Ls, between the sliding short <2> and the baseplate <4>, the reactor can be tuned to a specific electromagnetic resonant mode with zero reflected power. Illustration of the excitation modes is covered later in this chapter. The baseplate <4> is 51 cm in diameter. It consists of two pieces: a top brass piece <4a> and a bottom stainless steel piece <4b>. The baseplate hosts 24 of 5.4 cm x 5.4 cm x 2.7 cm rare earth permanent magnets made of neodymium iron boron <9>. The magnets are arranged in alternate poles inside the baseplate by stacking two magnets together. They are kept in position by a magnet keeper <6> made with low carbon steel. A 12-pole multicusp ECR magnetic field is formed as shown in Figure 4.1(b). The ECR surface, which has a field strength of 875 Gauss for 2.45 GHz excitation frequency, is located 2.33 cm from the surface of the magnet. If the excitation frequency is reduced to 915 MHz, then the ECR surface has a A t 7 . ! 1 = l 8 , E 2 I ’ 'A'A'AV-VA'A'k-‘g ; 3 4a 4b 5 (a) Longitudinal section 6 ECR surface E CR at 2.45 GHz sunace are» 9;" "I" 7"; l 4,2 9 4b (b) Top view of plane A-A Figure 4.1 Structure of MPDR 2O 68 field strength of 327 Gauss which is located 3.7 cm from the magnet's surface. To protect the magnets from thermal demagnetization, a water channel <5> is used to cool the baseplate. The discharge volume is defined by the 20 cm diameter opening in the baseplate and the dome-shaped quartz disk <7>. A discharge volume of about 3000 cm3 is formed. Gas is introduced into the discharge region through twelve gas pin holes <8> drilled on the inner surface of the baseplate. The quartz disk sits on the baseplate <4b> with a L-shaped Viton vacuum gasket <10> in between as shown in Figure 4.1(a). Figure 4.2(a) is a picture of the components in MPDR 20, showing the aluminum cavity, the probe and the quartz disk. There are two circular rings mounting on the outside of the aluminum cavity. 2 mm in diameter holes are drilled on these rings. They are used for electric field measurement to verify the excitation mode in the reactor. Figure 4.2(b) shows the stainless steel baseplate <4b> with the ECR magnets. § 4.2.2 Historical development of MPDR The historic development of the MPDR can be divided into two stages since 1981. The first seven years was the first stage of development. It was the evolution stage in which major components reached their final form. The next stage of development took about five years. More than five MPDRs were constructed for different applications. Each of these newly developed MPDRs had its own specific features and improvement, but components such as the cylindrical cavity and the baseplate remained the same even though the sizes of these components varied. 69 (b) Magnetic ring in the baseplate Figure 4.2 Photographs showing the components of the MPDR 20 70 The first stage of development The first prototype of the microwave plasma disk reactor (MPDR) was presented in 1982 by Root and Asmussen in a MSU/NASA Workshop in Michigan State University [32a]. This first reactor was designed as an electrodeless ion source [32a] using 2.45 GHz microwave excitation. The work was subsequently published [32b] and an US. Patent was granted [32c] later in 1985. This prototype is referred as MPDR I in Figure 4.3. It has a 7-inch inside diameter cylindrical cavity made from a brass cylinder. Microwave energy was introduced into the cavity by a side—feed monopole antenna or probe. The plasma was generated and confined in a flat 9-cm diameter cylindrical quartz disk and an ion extraction grid attached to the bottom of the baseplate. This prototype reactor was not an ECR plasma source because there was no static magnetic field inside the quartz disk for the electron cyclotron resonance to take place. The reactor was originally designed to be a high efficiency ion source. Ions were extracted through the grid for either downstream processing such as oxidation of silicon wafer [34] or as an electrodeless microwave electrotherrnal thruster for spacecraft propulsion [185]. Another paper was published in 1985 by Root and Asmussen [33]. This paper was important because it pointed out the concept of using a multipolar magnetic cusp to reduce plasma diffusion to the wall inside the plasma disk, and the idea of using electron cyclotron resonance to generate plasma for low pressure application. This was the first time when the concept of using rare earth permanent magnet to form a multipolar cusp- 71 ”my N. MPDR I microwave probe (Root et al., 1982) quartz disk MPDR II Magnet (Dahimene et al., 1984) MPDR 9 (Dahimene et al., 1987) [Qt-L'- \--.~,\\ MPDR for / a“ Diamond I Thin Film (Norton) MPDR 610 _ l I l (Asmussen et al., 1988) (Wavemat) MPDR 20 (Sze et al., 1990) MPDR 5 (Mahoney et al.,1990) \\ \r... DTP (Norton) MPDR 13 ak etal., 1992 (M ) MPDR 325 (Sze etal., 1990) (Wavemat) MPDR 325i . (Sze et al., 1993) wavegulde ' (Wavemat) Figure 4.3 Development of MPDRs for ECR plasma Applications 72 shaped ECR magnetic field to generate a plasma was ever published. About the same year in the MSU laboratories, Dahimene tested the first electron cyclotron resonant MPDR by placing a ring of permanent magnets around the quartz disk. This is referred as MPDR II in Figure 4.3. The worked was first represented in the 19th EPIB conference [35a] and later published in 1986 [35b]. An improved version of the reactor was published in 1987 [36]. The ECR magnets, instead of being exposed to the microwave field, were hosted inside the baseplate. This version of MPDR is referred hereafter as the MPDR 9, as shown in Figure 4.3, because it has a 9 .cm diameter of plasma. Early versions of the MPDR 9 had metal grids attached to the bottom of the baseplate [32-36]. The reactor was operated as a broad ion beam source. The advantage of using a grid is to establish a pressure gradient between the plasma generation region and the downstream processing chamber. Such pressure gradient allows several mTorrs in the region enclosed by the quartz disk and the grid so that a high density plasma can be generated. The processing region below the grid can be maintained at pressure less than 0.1 mTorr so that a mean free path between collision of 52 cm can be obtained. Ions can be accelerated by the grid through proper biasing, forming a broad directed beam to bombard the substrate surface. The disadvantage of using a metal grid or grids is that the processing rate is reduced due to the recombination loss of ions on the grid surface. The grid itself, in addition, may become the source of contaminant due to physical sputtering by energetic ions, or chemical reaction between reactive species in plasma and the grid. 73 The MPDR 9 can also be operated without a grid for different applications. Examples of these applications are plasma oxidation by G. Salbert [182] and etching by J. Hopwood [60]. In fact all current ECR MPDRs are now operated without any grid attached to baseplate [43, 45-48, 183,184]. A gridless reactor increases the processing rate due to elimination of recombination loss of the ions on the grid. Potential contamination from the grid is also removed. The lowest processing pressure in a gridless MPDR 9 is about 0.3 mTorr depending on how much microwave power is used. However, the amount of microwave power that could be used in the MPDR 9 source was limited to 350 watts at pressure less than 10 mTorr. This is because a Viton O-ring is used in the MPDR 9 as the vacuum seal between the quartz disk and the baseplate. The O-ring will burn if too much microwave power is used. The second stage of development Since 1988, a family of MPDR plasma sources was developed with some structural modifications. The original MPDR 9 has a 17.8 cm inside diameter brass cylinder with 9 cm diameter discharge. Eight 2.54 x 2.54 x 2.54 cm Ne-Fe-B magnets are arranged in a multicusp configuration to create the ECR surface. On the basis of the blueprint of the MPDR 9, three different sizes of MPDR prototypes were designed and fabricated. One reactor was a 3.75 cm diameter discharge reactor designed by Asmussen and Fritz in 1988 and published in 1990 [40]. It was later modified to a commercial product known as MPDR 610 [48,165]. MPDR 610 is an ion, plasma or free radical source designed for molecular beam 74 epitaxy applications such as superconducting thin film deposition. Both the method of microwave excitation and orientation of the ECR magnetic field in this source are different from the rest of the plasma sources in the MPDR family. Either 915 MHz or 2.45 GHz microwave excitation can be used to create the plasma. It is now being studied by Srivastava [48]. The second reactor has a 5 cm discharge diameter designed by L. Mahoney (hereafter referred as MPDR 5) and was published in 1991 [41]. It has a 8.9 cm inner diameter cavity. Because of the small diameter of the cavity, this reactor can only be operated at TE111 mode. The discussion of mode excitation is covered in later section and therefore is not discussed here. Eight 1 cm x 1 cm x 1 cm permanent magnets are used to form an octapolar static magnetic field. This source was previously configured as an ion beam source in Mahoney's work [41,61] using a multi-grid to accelerate the ions. The source can also be operated without a grid as a plasma source, which is studied in this thesis. The third reactor [42,43] is the MPDR 20. The design of which is one of the major subjects in this dissertation. This reactor is a scale up of the MPDR 9. A twelve-pole multicusp static magnetic field is used. However, the reactor can be run with either 915 MHz or 2.45 GHz excitation. One significant step was achieved in this reactor. It was the first MPDR which was operated with an opening bigger than the cutoff radii of the cylindrical cavity resonant modes. Because the reactor runs very well at 2.45 GHz without a grid as the smaller reactors, a commercial reactor was released later by WAVEMAT with some modifications. This is the MPDR 325 [45] as shown in Figure 4.4(a). This reactor is another important subject in this 75 cylinder microwave input probe quartz magnet water channel O-ring gas channel Alinco bar magnets cooling aluminum tube channel Longitudinal view View of section A-A (b) Figure 4.4 (a) MPDR 325 and (b) MPCS 76 dissertation, especially its performance and application in etching of submicron features in silicon. Although MPDR 325 looks similar to the MPDR 20, it has several different features which are improvement over the MPDR 20. The inner diameter of the cavity is 35.56 cm which is smaller than the MPDR 20. As indicated in later section, the resonant modes in MPDR 325 are further apart than those in the MPDR 20 if both are operated at 2.45 GHz. As shown in Figure 4.4(a), the long quartz vessel in MPDR 325, which covers most of the baseplate area, reduces contaminant from the metal wall. The whole reactor is made of aluminum. Therefore it is much lighter than the MPDR 20. Unlike the MPDR 20, a waveguide port is used for microwave input. The use of waveguide is better than using a coaxial cable, which is used in MPDR 20, in that power ratings of coaxial cable limit the input power to the cavity. The baseplate of the MPDR 325 is also water cooled. It has the same magnetic configuration as the MPDR 20. The input gas holes are equally spaced in a metal gas ring near the opening of the quartz vessel as shown in Figure 4.4(a). The 24 cm inner diameter quartz vessel is clamped vacuum tight between the baseplate and the gas ring piece. This forms the plasma generation region with a volume of 3600 cm3. The experience of running a large diameter MPDR 325 lead to the modification of the original MPDR 9. A 13 cm inner diameter quartz disk is used in a modified baseplate in MPDR 9 which is hereafter referred as 77 MPDR 13 [47]. MPDR l3 shares the same 17.8 cm diameter cavity as in MPDR 9, but it uses the same type of quartz disk in MPDR 325. Therefore the differences between MPDR 9 and MPDR 13 are just the size and shape of the quartz disk as shown in Figure 4.3. The MPDR 13 is currently investigated by Mak [47] and King [46]. Three improvements in MPDR l3 and MPDR 325 can be found. One is the use of a longer quartz vessel to reduce contamination by reducing the exposed area of the metal wall to the plasma. The second improvement is moving the O-ring seal away from the plasma so that it would not burn at high microwave power. Microwave input powers up to 1000 watts have been used in these two reactors. It is believed that both reactors can be operated at power over 1000 watts. The third improvement is the use of a multipolar plasma confinement structure (MPCS) [43] as shown in Figure 4.4(b). The MPCS can be attached underneath the baseplate of the MPDR 20 and MPDR 325. It is made from an aluminum tube which has an outside diameter of 28 cm and length of 22 cm. The inside diameter is 24 cm, which is large enough for 6- inch wafer processing. Twelve slots are located on the outside of the MPCS. A 0.5 x 0.5 x 6 inches Alinco bar magnet with pole strength of l kGauss is placed in each of the twelve slots. When attaching to the baseplate, all magnets, both inside the applicator baseplate and those in the MPCS, are oriented in a manner that one of the poles of each magnet is facing toward the central axis of the system and poles of neighboring magnets, either above, below, or side by side, are opposite to each other. Such a configuration creates magnetic cusps oriented horizontally and vertically. 78 This configuration minimizes the diffusion of ionic species towards the wall, therefore reducing loss of plasma due to surface recombination. The MPCS has a magnetic field strength of 100 Gauss 2 cm from the inner surface and the field strength decays rapidly toward the center of the confinement chamber. The MPCS can be cooled either by gas or water using the cooling channel as shown in Figure 4.4(b), which run parallel to the bar magnets. Experimental observation showed that the MPCS did not heat up at all. The MPCS improves the uniformity of the plasma for large wafer application. Experimental results showed that a uniformity of 2.6% standard deviation of plasma density over 18 cm diameter was achieved at 12.5.cm downstream in MPDR 20. A similar but shorter MPCS was used by M. Dahimene in MPDR 325 and a 3.0% standard deviation of plasma density at 6 cm downstream was obtained over 15 cm diameter [43]. At 15 cm downstream in MPDR 325, a uniformity of 8.1% (3 0') was achieved [45]. The downstream plasma confinement structure has already become an integral component in all ECR MPDR applications. § 4.3 Operation of the MPDR 4.3.1 Electromagnetic resonant modes in cylindrical cavity The resonant modes of electromagnetic wave in a cylindrical cavity are represented by two groups [66, 75]. One is called the transverse electric field (TE) mode in which the electric field of the electromagnetic wave lies in the transverse plane of the cylinder. The magnetic field of the 79 electromagnetic wave in TE mode has both transverse and axial components. The other is called transverse magnetic field (TM) mode which means that the electric field has both transverse and axial components, but the magnetic field only lies in the transverse plane of the cylinder. Each cylindrical cavity resonant mode is governed by a mode equation. For TE mode, the mode equation is given as [66]: ‘11an = Jn (x'npr/a)[b cos(n¢) +c sin(n¢)] sin(q1tz/d) (4.1) where n=0,1,2,...; p=1,2,3,...; q=l,2,3,...; a is the radius of the cavity; d is the length of the cavity and x'np is the pth zero of the first derivative of Bessel function, J n. For TM mode, the mode equation is: ‘1’an = Jn (xnpr/a) [b cos(nq>) +c sin(n¢)] cos(q1tz/d) (4.2) where n=0,1,2,...; p=1,2,3,...; q=0,1,2,...; a is the radius of the cavity; (1 is the length of the cavity and x“p is the pth zero of Bessel function, J". The relation between the resonant frequency and the geometry of a perfect cylindrical cavity is given by Harrington {66]. For TE mode: d = qua/[41t2a2f26 u - x'np2]%- (4.3) 80 For TM mode: d = qua/[4n2a2f26 u - xnp2]l/2 (44} where f is the frequency, 6 is the permittivity and u is the permeability in vacuum. At fixed microwave frequency, the variation of cavity radius versus the resonant length in a cylindrical cavity can be represented by a mode chart. Figure 4.5 and 4.6 show the mode chart for the TE and TM modes at 2.45 GHz respectively. In these two mode charts, there are repeated patterns. These repeated patterns represent higher order modes. For example, in a 10 cm diameter cylindrical cavity, the TE11 mode can be excited at cavity lengths about 9 cm, 18 cm and 27 cm which are represented by TE111, TE112 and TE 1 13 in Figure 4.5 respectively. These mode charts give the theoretical relation between cavity radius and the resonant length for an empty perfect cylindrical cavity. In a non-perfect cavity such as the one in MPDR, the resonant lengths of these cavity modes shift. The presence of the quartz disk and the hole in the baseplate shortens the resonant length. With the presence of a discharge, the protruded plasma lengthens the resonant length. These resonant lengths also change as the plasma densities change with input microwave power and pressure. The number of resonant modes which exist in a cylindrical cavity is determined by its radius and length. A large diameter cavity such as the MPDR 20 has resonant modes close to each other as shown in Figure 4.5. A small diameter cavity such as MPDR 9 has its resonant modes widely separated. As a result, single mode excitation is possible in small diameter Cavity Radius (cm) 30.00 N p. O O —h .0 o o 0.00 O 81 : mono + I I. J‘e "‘ 3 MPDR 325 *4, 1 :4, -l v- ‘1 I '3,” J r , ..mun .. 7:5", : MPDR 5 T5”;\ ’5’“ 7’5”: J MPDR610 ; TE+mode at 2.45 Hz IIIIIIIIUIIUIIlllll'llIIlIle .00 1 0.00 20.00 30.00 Cavity Length (cm) Figure 4.5 TB mode at 2.45 GHz Cavity Radius (cm) 82 30.00 \‘iil‘l‘iw‘i‘li‘w \ '\ e ‘\\ \\‘i||! \\\ 1., [\r\‘~\\\\wu\|\“§§\ ., \\ \\ l I \‘\\\“““§‘\ MIR \l‘t -l MPDR 20 '8 8 llrrLl If/ / / a MPDR325 ‘ ‘ . \\ ‘ : '~ ‘\ 4 \\ 1 ‘\ — 10'00 A MPDR9&13 OI q MPDRS :m o 00 4 TM mode at 2.45 GHz 0 lllllIIIIIIIIITIITjIIITlleII 0.00 10.00 20.00 30.00 Cavity Length (cm) Figure 4.6 TM mode chart at 2.45 GHz 83 60 - r5421 50- A : TE321 o 40.1 15021 V - TE221 1n 2 ~ 03 " ._ TE121 '8 30_ r5411 0: . 4?: :m TE311 .; 20_ r5011 0 " 11 0 _ TE2 ‘ TE111 10- l‘ 915 MHz 0 IIlllllllllllllllll'lllillllI'llllllIlIIIITIIIIIIIIIIIIlllllllilllllllllllilli o 10 20 30 4o 50 60 7o 80 Cavity Length (cm) Figure 4.7 TB mode at 915 MHz 60 01 O 4.». O Cavity Radius (cm) '63 ‘6‘ 84 1 K '7 W321 —i ‘— 4 ”221 1—1 W411 _‘ W121 - W311 ‘ W021 ‘ TMZ11 2m m1“ .. W011 - 915 MHZ IIIIIIIIIIIIIITIIIIIIIIIIII'IIIIIITTI[IIIIIII'III O 20 40 60 80 100 Cavity Length (cm) Figure 4.8 TM mode at 915 MHz 85 cavity. When the cavity diameter is large such as the MPDR 325, the resonant modes are close to each other. As a result, more than one resonant mode can be excited. In general, orientation of the microwave excitation probe has to match with the electric field pattern of the mode which is excited. This can be explained by the electric field distribution in each mode diagram described below. W Consider a cylindrical cavity with its axis on the z-axis. The electric field for the TEnp mode can then be obtained by following relations [66]: 8‘1’ 13r = - (4.5a) r 81]) B‘P 13¢ _ (4.5b) 8r E2 = () (4.50) The magnetic field is perpendicular to the electric field with a non-zero 2- component: Hr = (4.621) 86 82‘1’ H¢ = (4.6b) 8(1) 82 Hz = (x'np/a)2 w (4.6c) For TM mode: 8‘? Hr = (4.7a) r 84) 8‘1’ ['14, - - (4.7b) 8r Hz = 0 (4.70) 832‘? [3r = (4.8a) Br 82 82‘? 13¢ — (4.8b) 8(1) 82 Ez = (xnp/a)2 ‘1’ (4.80) Figure 4.9 to Figure 4.10 are the graphic displays of TM” and TEOI resonant modes in a cylindrical cavity. Each figure consists of a two- dimensional plot of the electric field strength and the orientation of the 87 TMll MODE: Longitudinal E field strength \‘ \ 11 11111“ 1\‘$1“::§1 “\\ \\\‘\\\\‘ 0 . 1 1111\1\\\\\‘ l‘11111111\\\‘\\\\ ‘111\\\\\\‘ 1111\\\\\ \\\\\\\\1 M “‘1“ 1“,“ -0.5 mil MODE: Transverse B field mph \\‘-‘\\ \\ 13:“ \l 1 1 . .1111 .1 9:111 0-4 , IOO“|\\\\\\1‘|I ""-]111111 ill 11 111111 111 10 20 30 .111 50 m 10 1111 88 T50] MODEEIecuicficldsuumh 1 ‘1 1 ‘unu 1; 1‘1‘1\\\“\‘|.:‘.1’o.o"l"ll’;,l;’; 1 O I A "\\\1‘11‘1‘0.43.0!!!” . n 0 I l I I 1 I, I l ' u. “‘1“1“1““ | “\ \ 1“ ‘3‘ -,- .. - ,. 1 \\\1 ‘1! 1‘ 1\\ .§“‘\\\\‘\ . o o o. \ ~“~‘ - - - ‘ .;.~ ~ - - '. - - - - - - - or: O O. -- .. .0 - - .’ .. .‘ .. o o o - o ' " " o "' ’ ’ ’ ' _. ’a v- ’ .—-” ’ ’ .— ’ ’d 9 - o - o- - - - - o .. .. - .. o o - ‘o - ‘.- o o'. o - o .o \ ¢\‘ \\ ‘¢~ “ ~¢~ ~ .s"- ...~ \‘ ‘ ~ ‘ ‘ ‘ - . ..- f - - 9.. .. o o o o- - - . ~' - -' / I” g ..r’ ‘ - - o o \ - - .. -0 o . o o ...p .p’ ’ .. .- f.— ’ 5 ,1,— ffigfi’ (,4 I (,4 A o. - If . o - o o o - " ' ‘\ \\' - o y ‘ 1 - - o - - ’ 1 111 1 11 1 1 131313 - O O o o .— _A A -‘e‘ \ .— I :- 11 1111591111. \ 1"; . \1““ o I I 1 1 50101."? 9,991”! o o - o o .o f. - . o .. .. o o - ’4 ,— g \, ‘o - ’. u m1 MODE: Onenmon' ' ofelecmc' field Figure 4.10 Distribution of E-field at TED] mode 89 transverse electric field. A polar coordinate system is used even thought a square matrix is displayed. The numerical model uses 80 x 80 elements that gives sufficient resolution for the vector presentation. The radius of the cylindrical cavity is represented by 40 elements. The axis of the cylindrical cavity or the center of each polar plot is represented by the coordinates (41 ,41). It should be noted that in TM modes, the total electric field strength is not a direct vector sum of the longitudinal electric field and the transverse electric field as shown in Figure 4.9. From equations (4.2) and (4.8a-c), it is indicated that there is a 90 degrees phase difference between the longitudinal and transverse electric field. All the electric field strength in TE and TM modes are normalized. The quivers in the contour plots in Figures 4.9 to Figure 4.16 represent the direction of the transverse electric field. Because of using a side-feed antenna, TE12, T1313 and TE01 modes are not likely to be excited in the MPDR. In the case of TE]; and TE13 modes, a side-feed probe antenna cannot excite an azimuthal electric field near to the wall (see element (40,15) in Figure 3.2 and element (40,10) in Figure 3.2). In the case of T1301 mode as shown in Figure 3.10, there is no radial component of the electric field, thus a side-feed loop antenna is likely to excite this mode, but not a side-feed monopole antenna. Figure 4.11 to Figure 4.16 display the mode diagrams of those resonant modes which are excited or can be excited in MPDR plasma sources except the MPDR 610 which will be discussed in later section. These mode diagrams are different from the conventional plots in most text books which only indicate the orientation of the transverse electric field, but no information about the field strength. 90 TEll MODE: Electric field strength 0.6 0.4 \111 111111 1111111111 11‘ \1 “11111111111111 \\ 0 l 00 11 ”1"] 1‘1 1‘ 111111 1 \ \ 1 1 1 1 111 \ 11.1111111‘ 1111111 1 “15‘0““ \ \ ‘ 1 1 1 ‘ “ “11‘ \“ \\1 11 1 1 ‘1 \ ‘ \\1 “11““ \‘ \\ \ 1 1‘ 11 1 11“ 11111“ \\“ \\ \\1 1 “1111111111“ 11111111 ‘“‘\‘\S‘\“““\““"11111‘11‘ ‘11 \\ 1‘ \ \“: 1‘ ‘1 1‘... 1““ 1 1 “1111131111 “‘ \\111“ 111 1511 MODE: Orienmion of electric field 111 Figure 4.1 1 Distribution of E-field at TE“ mode 91 T31 MODE: Electric field strength 0000 p “0000 p 11.0 1 111 1,1 111 00.1.1‘“ 1. O .00" o o - N. H. I '1 11 111‘1“| ‘ 1 0‘.-- -'-'. - o - - - - - .‘.§' - —. o - ..‘ ~ 0 o o . - é ’ é 4...; o - . T321 MOD&Orienu’onofdecuicfield Figure 4.12 Distribution of E-field at TE21 mode 92 11331 MODE: Electric field strength ,,I 1 ,II’, /,’, MINE“ /’Im11\11 H’I’I'...‘ 1\\ I I "’ '1’1’11 M .111 I", 1111/1 "I, ”1, 1 I] ‘1‘ ”\ “\\\1 1 l“ ‘11“ 1 I 117/I 111:1 1 ' 3,171 1111111111 '1 Fais'onI I' 11‘. 11 N 1 ‘ 1 “‘11‘1 ‘1‘1'1‘11'1‘1‘11' ‘ ‘\\1\\\\““:.’// 11““ ‘M \\\\\“9// 1111111“ Figure 4.13 Distribution of E-field at TE3] mode 93 m1 mmwmmm Figure 4. [4 Distribution of E-field at TE41 mode 94 T561 MODE: Electric field my}! 11.1111 /l01"1'\ .1311“: .IH11‘1 1 \y ‘ I '1" 111' 1"11 ' 1'11'1'1l11‘1 ' '1 1'1" 11"" 1:11“ 1M ”a” Gui "‘“1 ‘11'11'111'1‘11‘1'1 11 11,1"111 1'11 1 1111 11E11‘1 ’1 1%,, 11111111 1 1' 1111 Figure 4.15 Distribution of E-field at TEN mode 95 TE]! MODE: Elearicficldflmpglh Mlmmdelecuicfield Figure 4.16 Distribution of E-field at TE71 mode 96 4.3.2 Ignition and tuned positions in MPDRs Every MPDR has two degrees of freedom in tuning. The input micro- wave probe and the sliding short allow tuning and tracking of an operation mode. The starting position of a MPDR is defined as the positions of the sliding short and the microwave input probe such that a minimum microwave reflected power is observed without a plasma. The ignition positions are those starting positions that can ignite a discharge. It should be noted that no grid was used in all the MPDRs investigated. More than one ignition position may be found in each MPDR and they are repeatable. The tuned position of a MPDR is defined as the positions of the sliding short and the microwave input probe such that a minimum microwave reflected power is observed with a discharge. Each starting position, ignition position or tuned position represents either a single resonant mode or several resonant modes in the MPDR depending on the diameter of the cavity. Starting positions of the MPDR can be found by using a frequency sweeping circuit as shown in Figure 4.17. Operation of the sweeping circuit has been described in detail by Manring [166], therefore not repeated here. In brief, the MPDR is swept through a wide range of microwave frequencies. When the resonant frequency of the cavity, such as fo=2.45GHz, matches with the input microwave frequency, all the microwave energy at that frequency will be absorbed. This is indicated by an "inverted" power absorption curve shown on the oscilloscope in Figure 4.17. By adjusting the positions of the sliding short and the microwave input probe, the sweeping circuit can determine all the starting positions in each MPDR. 97 Transmission _— attenuator line Figure 4.17 Schematic setup of sweeping circuit In ignition; found at ' TE mode plasma. 1 used to tu mTorr. th the TE ni MPDR 6l Afté found by nmmmef Usually, 1 microwave be tuned t Variations pCflected p dfinsity, H. “11.1 bad 4‘33 Mod The e"'plalned 1. mOdeS Of 0] 98 In MPDR 5, there is only one starting position which is also the only ignition position to start the plasma. In MPDR 20, six starting positions were found at 915 MHz, forty at 2.45 GHz. Some are TM modes. The others are TE modes. However, not all of these positions can be used to ignite the plasma. The my way to tell whether a particular starting position can be used to turn on the plasma is to try it. If no plasma can be turned on at 100 mTorr, the starting position is useless in the reactor. It was found that only the TB modes can create a stable plasma in all the MPDRs, except the MPDR 610, at pressure below 100 mTorr. After ignition, the resonant position shifts. The tuned position can be found by adjusting the sliding short and the microwave input probe in alternate fashion until a minimum microwave reflected power is observed. Usually, the sliding short has to be moved upward slightly while the microwave excitation probe is moved towards the cavity. Most MPDRs can be tuned to less than 1% reflected power while in single mode excitation. Variations of the pressure and input microwave power will increase the reflected power to a few percentages because of the change of the plasma density. Howver, a slight adjustment of the sliding short alone can bring the cavity back to a tuned condition. 4.3.3 Mode of operation The mode of operation of each MPDR plasma source can best be explained by using mode charts as shown in Figure 4.5 to Figure 4.8. The modes of operation in different diameters of MPDR are discussed separately. MPDR ' Figure -' Figure that allo radius f( radius f( cavity di cavity re and mag has been reaCIQr Wd Table 4. 1, from the fi 99 MPDR 610 MPDR 610 is different from the rest of the plasma source in the MPDR family in two aspects. First, it uses an end-feed antenna as shown in Figure 4.3. Second, its radius is less than the cutoff radius as shown in Figure 4.5. The cutoff radius is the smallest radius of a cylindrical cavity that allows a resonant mode to be excited. As shown in Figure 4.5, the cutoff radius for TB mode at 2.45 GHz is about 3 cm. In Figure 4.6, the cutoff radius for TM mode at 2.45 GHz is about 4.5 cm. The MPDR 610 has a cavity diameter of 5 cm. Therefore it is not possible to excite a cylindrical cavity resonant mode. Instead a TEM mode, whose time-varying electric and magnetic fields are perpendicular to the cylindrical axis, is excited and has been reported earlier by Asmussen et a1. [40]. MPDR é All MPDR plasma sources, except the MPDR 610, use a side-feed probe antenna similar to that shown in Figure 4.1. TB modes are more likely to be excited with this kind of configuration although TM modes can also be excited. The diameter of the cylindrical cavity of the MPDR 5 is 8.9 cm. This means that this reactor can only be operated at TE111 mode or its high order modes as shown in Figure 4.5. This was the original idea when the reactor was designed [61]. The ignition position of the source is shown in Table 4.1. The initial position of the sliding short is about two centimeters from the final tuned position. Once the plasma is turned on, the sliding short has to be moved up to keep the cavity in a resonant mode so that all 100 TABLE 4.1 OPERATION POSITIONS OF MPDR 5 WITHOUT A GRID Experimental conditions: Gas: Argon Flow: 50.2 sccm Pressure: 1.5 mTorr Power: 100 watts Cavity Height (cm): Ls Probe Length (cm): Lp Theoretical position1 10.5 - Ignition 8.6 1.1 Tuned position2 10.6 1.8 1 For a perfect empty cylindrical cavity 2 In MPDR with plasma . NH 0 Quit-’oillrsu. “not w - 101 microwave energy is coupled into the plasma. Compared with previous report by Leonard Mahoney who attached a grid to the baseplate for his ion beam experiments [61], the main differences are the initial and final positions of the microwave input probe. MPDR 9 and MPDR 13 The diameter of the cylindrical cavity in MPDR 9 and MPDR 13 is 17.8 cm. Figure 4.5 and Figure 4.6 indicate that four single TE modes, TEM], T1321], TED” and TE311, and two TM modes, TM011 and TMlll, can be excited in these cavities with 2.45 GHz microwave energy. The T1321] mode is used in most of the ECR applications and diagnostics in these two plasma sources. The change of the size of the quartz disk did not have major influence on the excitation modes. The choice of which mode to be used was based on the uniformity, stability and density of the plasma. The difference in performance in the MPDR versus mode excitation is under investigation by Mak [47]. MPDR 20 MPDR 20 was designed to be excited with 915 MHz microwave energy. The cavity has a diameter of 45.7 cm. At 915 MHz, the resonant modes are wider apart as shown in Figure 4.7 and Figure 4.8. Six possible resonant modes were detected in this reactor by using the frequency sweeping circuit. They are listed in Table 4.2. Two modes have been used to create a discharge in this reactor. The first one is the TE111. The other is the TE211 mode which was used in most of the diagnostics experiments. TE112, 102 TABLE 4.2 RESONANT MODES IN EMPTY MPDR 20 Cavity Height2 (Ls) Probe Length (Lp) Theoretical Height1 Mode 18.78 cm 3.90 cm 18.06 cm TE111 20.34 cm 4.35 cm 19.61 cm TMOll 23.54 cm 3.44 cm 22.86 cm TE211 34.24 cm 5.67 cm 33.82 cm TM111 36.79 cm 3.66 cm 36.13 cm TE112 39.89 cm 4.09 cm 39.22 cm TMmz 1 For a perfect empty cylindrical cavity 2 In MPDR without plasma 103 which has similar field distribution as the TE111, was not used. The other modes, TM011, TM111 and TM012, could not ignite the plasma. Verification of the mode was done by direct measurement of the electric field pattern along the cavity wall with a 2 mm diameter microcoaxial probe [33] shown in Figure 5.2. The theoretical distribution of electric field of TEnpq mode on the cavity wall is given by the equation: E,(r,¢,z) = (n/r)E0Jn(x'npr/a)[b cos(n¢) + c sin(n¢)] Sin(q1tz/d) (4.9) where Jn is the Bessel function of order n, x'np is the pth zero of J 'n, a is the radius of the cavity and d is the cavity length. Figure 4.18 and Figure 4.19 show the experimental data and the theoretical distributions of the electric field strength in TE111 and TE211 respectively. As shown in the figures, the experimental results match very well with the theoretical distribution although there are some deviations of the electric field strength near the peak, which are caused by the variations of the inserted length of the micro- coaxial probe inside the cavity. Table 4.3 and 4.4 shows the operation positions at these two modes. The tuned position of the sliding short with the presence of a discharge is very close to the theoretical value for a perfect empty cylindrical cavity at TE211 mode. A slightly bigger deviation between the theoretical cavity length and the height of the sliding short at tuned condition was found in the TE111 mode. This is because the distortion caused by the dome shaped plasma becomes significant. 104 300.0 Fm : 915 rm: 0 Experimental Data 250") Theoretic value 0 O 0 200.0 0 0° 0 150.0 0 100.0 . 50.0 Field strength ( microwgtt) 0.0 IIIIIIrIIITITIT|IITIAIIIYIIIIII 0.0 45.0 9010 135.0 180.0 225.0 270.0 315.0 Angle w.r.t. input probe (degree) Figure 4.18 Experimental measurement of E-field at TE, 11 mode 360.0 Field strength ( microwott) 105 500-0 j PM a 915 In: 3 0 Experimental Data 3 Theoretic value or o .0 o N o P D 100.0 JJIIIIIIILIrriiirulmiir 0.0 [III ITIIWII !!!!! ITIIIIIIIIIII.IIIIIIIIIIII IIIIIIIIITIFI‘T IIII 0.0 45.0 90.0 135.0 180.0 225.0 270.0 315.0 350.0 Angle w.r.t. mput probe (degree) Figure 4.19 Experimental measurement of E-field at TE211 mode TABLl EXperimental Gas: How: Pressure: POWCr; \ \ 11160131ij p0 Ituition “W posing. \ I For a Pfirfeq 3b MPDR Wll 106 TABLE 4.3 OPERATION POSITIONS OF MPDR 20 AT TEzu MODE Experimental conditions: Gas: Argon Flow: 30.4 sccrn Pressure: 0.8 mTorr Power: 200 watts Cavity Height (cm): Ls Probe Length (cm): Lp Theoretical positionl 22.86 - Ignition 22.6 5.3 Tuned position2 22.8 9.5 1 For a perfect empty cylindrical cavity 2 In MPDR with plasma 107 TABLE 4.4 OPERATION POSITIONS OF MPDR 20 AT TEln MODE Experimental conditions: Gas: Argon Flow: 25.7 sccm Pressure: 0.6 mTorr Power: 200 watts Cavity Height (cm): Ls Probe Length (cm): Lp Theoretical position1 18.06 - Ignition 18.8 5.8 Tuned position2 18.95 9.9 1 For a perfect empty cylindrical cavity 2 In MPDR with plasma E: ..U U 70 \ I I from th 17.78 c is well Figure 1 about 6 and ign strength electric is discu pressure The igni as show M. re~‘0nant From Fig Cavity l€r TE313 an etpenmel ”We thar I“fled 10 ‘d pOWer, In lnpm of S( 108 MPDR 325 A large diameter reactor at 2.45 GHz has some difference properties from those reactors with small radii. The MPDR 325 has a cavity radius of 17.78 cm. The inner diameter of the plasma disk is 24 cm in diameter which is well above the cutoff radii of many modes as shown in Figure 4.5 and Figure 4.6. For example, at 2.45 GHz, the cutoff radius of TE211 mode is about 6 cm. Microwave energy thus propagates through the quartz vessel and ignites the plasma. Once a plasma is formed, the microwave field strength will decay as it moves through the discharge. The decay of the electric field in the plasma depends on the conductivity of the plasma, which is discussed in Chapter T wo. In MPDR 325, plasma can be started at pressure as low as 1.5 mTorr. This makes the reactor very easy to operate. The ignition position and the final tuned position are very close to each other as shown in Table 4.5. Therefore only a small adjustment is needed. MPDR 325 is operated in a controlled multimode [42] in which resonant modes are close to each other for a cavity radius of 17.78 cm. From Figure 4.5, four TE resonant modes, which are close to the tuned cavity length shown in Table 4.5, can be identified. They are TE131, TE711, TE313 and TE612. For TM modes, TM031 and TM221 are close to the experimental positions (see Figure 4.6). In controlled multimode mode, more than one resonant mode are excited. However, the MPDR can still be tuned to a position such that the plasma is stable and has minimum reflected power. In MPDR 325, typical reflected power is less than 7 watts for an input of 500 watts microwave energy at 0.6 mTorr argon plasma. Expenmental Gas: Flow: Pressure: Power; \ \ Theoretical p0 Ituition 11119110311101 BID MPDR wit 109 TABLE 4.5 OPERATION POSITIONS OF MPDR 325 Experimental conditions: Gas: Argon Flow: 20 sccm Pressure: 0.6 mTorr Power: 500 watts Cavity Height (cm): Ls Probe Length (cm): Lp Theoretical position1 «- Ignition 21.5 Tuned position2 21.1 1 For a perfect empty cylindrical cavity 2 In MPDR with plasma 1.5 1.4 4&4 I l? tnagnnt static m static m of the t This is IFdIlSVEl' Ft configur center at creates t] We mag C0Uflgu rt Th EXCltatjm maxim”n ECRsunl MPDR 9 magmas ; any 1W0 SOIDeu’ht1 broken E 110 4.3.4 Optimal magnetic configuration If the energy coupling efficiency is directly proportional to the magnitude of the cross product of the microwave electric field, E(t), and the static magnetic field, B, created by the permanent magnets, then the optimal static magnetic configuration will be the one which has the same distribution of the transverse H-field of the microwave in the excited resonant mode. This is because the transverse H-field is always perpendicular to the transverse E-field in each resonant mode. For example, if TE111 mode is being excited, the optimal magnetic configuration will be a dipole magnetic field which has a maximum at the center and runs perpendicular to the E-field as shown in Figure 4.11. This creates the maximum value of E(t) xB. If TE211 mode is excited, then a 4- pole magnetic configuration should be used. Similarly, the optimal magnetic configuration for a TEM mode excitation would be a 2n-cusp magnetic configuration. The MPDR plasma source is intrinsically ideal for T En] mode excitation where n can be any non-zero integer. This assumption for maximum efficient energy coupling holds as long as there is a continuous ECR surface within the plasma disk. For illustration, TE211 mode is used in MPDR 9. A 4-pole magnetic configuration should be used. If only four magnets are used to form the four magnetic cusps, then the distance between any two neighboring magnets is too far apart. The ECR surface is broken somewhere in between the two magnets as shown in Figure 4.20. This broken ECR surface, if used, creates an unstable plasma. Jeff Hopwood 111 ECR 1 surface 2m Figure 4.20 Distribution of ECR surface using 4 magnets in MPDR 13 112 ECR surface S N N S Figure 4.21 Tetrapole magnetic configuration using 8 magnets 113 indicated in his thesis that an octapole magnetic configuration should be used to keep a stable plasma in MPDR 9 using TE211 mode. However, a better solution has been proposed by Mak [47] in his 4-pole/8-magnet configuration. Figure 4.21 illustrates the formation of a tetra-pole magnetic configuration using eight magnets. By using two different sets of magnets, a continuous ECR surface is created. This technique is useful to maintain an unbroken ECR surface inside a large diameter plasma disk such that the number of static magnetic pole matches with the electric field pattern of the resonant mode. CHAPTER FIVE W § 5.1 Introduction § 5.2 Experimental techniques 5.2.1 Double Langmuir probes 5.2.2 Single Langmuir probe 5.2.3 Ion energy analyzer § 5.3 Experimental Results 5.5.1 MPDR 5 5.5.2 MPDR 20 5.5.3 MPDR 325 5.5.4 Summary 114 1 15 § 5.1 INTRODUCTION Characterization of three MPDRs is the main topic in this chapter. They are the MPDR 5, MPDR 20 and MPDR 325. The MPDR 9 has been studied by Jeff Hopwood in his graduate research and MPDR 13 is currently under the investigation by Mak and King [46,47] while the MPDR 610 is the subject of Aseem Srivastava's thesis research [48,165]. Although MPDR 5 was studied by L. Mahoney, yet the source has not been characterized without using a grid. It is included in this dissertation because it has some properties that have not been observed in other MPDRs. The MPDR 5 and MPDR 325 were characterized at 2.45 GHz microwave energy while the MPDR 20 was characterized at 915 MHz microwave energy. A schematic diagram of the microwave circuit for the characterization is shown in Figure 5.1. The diagram shows the essential elements for the operation of a MPDR. Microwave input power to the MPDR was monitored by the incident and reflected power meters. In the cases of characterizing MPDR 5 and MPDR 20, 50 Q coaxial cables were used to connect the components together. However, experiments carried out with the MPDR 325 used rectangular waveguides instead of coaxial cables. Figure 5.2 shows the experimental setup of the MPDR 325 in a vacuum system for plasma diagnostic experiments. A similar experimental setup was used for the MPDR 5 and MPDR 20. A downstream plasma confinement structure, which has been described in Chapter Four, was mounted below each reactor as shown in Figure 5.2. Because MPDR 5 has a smaller diameter than the confinement structure, downstream confinement 116 Transmission line _— attenuator attenuator Figure 5 .1 Schematic diagram of microwave circuit 00 Ch; Micr Prot ECR 81. Gas 1011 Cooling Water . '0” Ener VacUUm . Che""ber 117 Quartz Discharge Microwave Input Chamber Pr°be Sliding Short Micro-Coax Resonant Cavity Probe L ECR Magnet ECR surface N Gas Inlet L Z=0 cm \ r / ...... . P. I Downstream (1)/83:39 / Confinement / 1' K Magnets .4 \ Ion Energy 4 \~. Analyzer Langmuir / X Probe Vacuum Y 3-D Translation Chamber 2 Stages ‘To Pumps Figure 5.2 Experimental setup for plasma diagnostics magnets hWDR: defined 2 Pl: analyzer the meas function plasma dc Wath Langmuir help of u. probe Wag measuremt meaSUremE Chapter. 13 near to ill delennined. 118 magnets were mounted underneath a stainless steel plate on which the MPDR 5 was placed. The bottom of the baseplate in each MPDR was defined as the zero downstream position, Z=0, as shown in the figure. Plasma diagnostic tools used included Langmuir probes, an ion energy analyzer and a micro-coaxial probe. A single Langmuir probe was used for the measurements of plasma potential and electron energy distribution function (EEDF) while double Langmuir probes were used to measure plasma density. The ion energy distribution function (IEDF) was measured by a multi-grid ion energy analyzer. Both the ion energy analyzer and the Langmuir probe could be moved at different downstream locations with the help of three remotely controlled translation stages. The micro-coaxial probe was a 50 Q 2-mm diameter coaxial cable. It was used both for the measurement of electric field strength inside the resonant cavity and the measurement of microwave coupling efficiency, which is described in next chapter. By measuring the circumferential and axial electric field strength near to the wall inside the cavity, the mode of excitation could be determined. § 5.2 Experimental techniques 5.2.1 Double Langmuir probes Double Langmuir probes technique was first proposed by Johnson and Malter [56]. This technique has widely been used to measure the plasma density in a discharge because of its simple structure. The two probes used were made with a 0.9 mm diameter tungsten wire. Part of the wire, 8 mm in kngdt‘ ghsstu dmus characte memot [fighaD cmmm: Dwm1v dulnott uhenthe 119 length, was exposed to the plasma. The rest of the wire was sealed inside a glass tube. The distance between the two probes was 6 mm. Figure 5.3(a) shows the circuit diagram for the measurement of current-voltage characteristics. Both the current through the probes and the voltage across the probes were sampled and taken by a computer through A/D (Analog to Digital) converters as shown in the figure. As shown in the figure, the current sampled by the computer was the total current flow in the circuit. There was current loss in the parallel resistor and the voltmeter. This current did not flow through the probes. It was deducted by the computer at the time when the l-V data was taken. Figure 5.3 (b) shows an actual current-voltage plot of a double-probe measurement taken in one of the MPDRs. The curve can be distinguished by two regions. A nearly linear region, which is indicated by segment OA in the figure, and a saturated region, which is indicated by segment AB. If the surface areas of the two probes are made the same and if there is no spatial variation of the plasma potential around the two probes, a symmetric l-V curve should be obtained as the one shown in Figure 5.3(b). What it means is that when Vd is zero, there is no net current flowing between the two probes. The I-V curve should pass through the origin. If the probes are unevenly coated with an insulated layer, the I-V curve will shift either to the left or to the right. If the probes are shorted due to the deposition of conductive film, a nearly straight I-V curve will be observed. Figure 5.4 is an illustration of the I-V characteristics in double probes measurements. When Vd is zero, both electrons and ions reached the probe surface at the same rate. When Vd is equal to a small negative voltage, the 120 ND ——1 l V“ A/D 1d 1500.00 3 iel : a? I l a 500.00 — ipi E 1 O E - 11 v o _ d -500.00 : 1p2 . 2 1e2 2 B A —1500000-1'IIIIIIIIIUII ll'lljililllli -60.00 -20 oo 20 00 60.00 .Voltage (V) . (b) Figure 5.3 Double Langmuir probes technique: (a) schematic circuit and (b) I-V characteristics 121 sheath Probe 1 “‘33: Probe 2 plasma Vd = O I Vol = small negative Vd = large negative Figure 5 .4 Variation of sheath potential with applied voltage 8.2 \1' CC Fl 1P 911 tel 122 potential of probe 1 is raised. More electrons are collected in probe 1 because of the decrease in sheath potential between probe 1 and the plasma. A negative current flows through the external circuit to neutralize some the ions collected by probe 2. When Vd equals to the space potential of the plasma, the electrons collected by probe 1 are nearly the sum of the ions collected by probe 1 and probe 2. Probe 2 is at a state of saturation and is indicated by point A in Figure 5.3 (b). Further decrease of the applied voltage causes an expansion of the sheath between probe 2 and the plasma, which explains why segment AB is not flat. The knee point, such as point A, is the most critical point in double probe analysis to obtain the correct saturation current. The ion density, ni, is related to the ion saturation current, ip, according to following equation [55]: n, = (5/3eAp) ip (M, / kae)'/2 (5.1) where Ap is the probe area, Mi is the mass of the ions, kb is the Boltzmann constant, e is the electron charge and T6 is the electron temperature. In order to find the electron temperature, two assumptions are made. First, the total ion current collected by the two probes is independent of the applied external circuit. Second, a Maxwellian distribution of electron energy is assumed. Let i6 and ip be the electron current and ion current collected by each probe. According to Kirchhoffs law of current, following relation is established: let V potent The el potenti Where current have; I! ll Therefor VOIIage ‘ Al Which 00 good for 123 1p] - 161 = 162 - ip2 (5.2) => II) = 1p] + 1p2 = i6] + 11:2 = constant (5.3) Let mebe be the potential of each probe with respect to the plasma potential. Then the applied voltage, Vd would be: vd = vprobe] ‘vprobe2 (5-4) The electron current reaching the probe is governed by the retardation potential of the sheath [55]: 161 = Icolfixlx'evprobel/kae) 65) where 1e] is the electron current collected by probe 1, 1.30 is the electron current when mebe] is zero. Therefore using equations (5.4) and (5.5), we have: (1e1/1e2) = (Al/A2) exp (-eVd/kae) (5.6) => (lp/iez -1) = (Al/A2) exp (-eVd/kae) (5.7) =.~ map/1,2 -1)=1n(A1/A2)— eVd/kac (5.8) Therefore the slope of a semi-logarithmic plot of (Ip/iez -1) versus applied voltage Vd can be used to find the electron temperature. All plasma density measurements were carried out in argon plasma in which both the charge and mass of the ions were known. The analysis is not good for gas with large molecular structure in which different charged species 1‘ quite di microwa concentt density 1 and the optical 1 inside tl 5.2.2 S measun functior taken 0 [0 VP.\ dEfined d”11110 CUrrenI 124 species may exist. This is because the composition of the discharge can be quite different at different operation conditions such as pressure and microwave power. For example, in an oxygen plasma, the 0*“ and 02+ concentrations vary with operation conditions [28]. In this case, the plasma density should fall between two values, one assuming that all the ions are 0*“ and the other assuming that all the ions are 02+. A better way is to use optical methods to measure the concentrations of different charged species inside the plasma [167,168]. 5.2.2 Single Langmuir probe Plasma potential and electron energy distribution function can be measured by a single Langmuir probe. The electron energy distribution function, f(E), is related to the second derivative of the I-V characteristics taken by a single Langmuir probe [57-60]: dzl f(E) oc (Vp-V)l/2——— (5.9) de where V1) is the plasma potential, I is the current through the probe, E equals to Vp-V, and V is the applied d.c. voltage. Once f(E) is found, the average electron energy can be determined numerically. The plasma potential is defined as the voltage which gives the maximum value of the second derivative of the current with respect to the voltage [59]. There are two possible methods to find the second derivative of the current with respect to the voltage. One method is to take the I-V data using a single twice to accuraC) attempt the smal techniqu II with res] HCIOSS a 3 Th applied v Where V expanding 125 a single Langmuir probe and then a numerical differentiation is carried out twice to find the second derivative. The problem of this method is that the accuracy is very poor unless the original data is free from noise. Any attempt to remove the noise by numerical technique will also screen away the small population of the high energy electrons. As a result, numerical technique is not recommended. The other method is to measure the second derivative of the current with respect to the voltage directly. A small sinusoidal signal is applied across a single Langmuir probe with a d.c. bias as shown in Figure 5.5. The current flowing through the probe can be treated as a function of applied voltages: I: I(V+vsin(ot) (5.10) where V is the d.c. voltage and v is the small a.c. signal voltage. By expanding equation (5.10) into a Taylor series, it can be shown that: v2 d21(V) v4 d4l(V) = 1(V) + * + * + 4 dV2 64 dV4 dl(V) v3 d3I(V) .. + [ v* + * + ] sinwt dV 8 dV3 V2 d21(V) V4 (141(V) -[ :1: + * + ] cosZwt + (5-11) 4 dV2 48 (1V4 126 Vacuum chamber wall dc voltage A/D Figure 5.5 Schematic circuit of single probe measurement When tl higher ( second frequent T) except I used. Tl chosen 1 long ‘dIlC 127 When the small signal, v, is smaller than one volt, the coefficient of the higher orders terms can be neglected. As a result the magnitude of the second derivative can be picked up by setting the lock-in-amplifier to a frequency which is twice the frequency of the small a.c. signal. The experimental setup was similar to the one used by Hopwood [60] except that an audio transformer with frequency response up to 3kHz was used. The schematic circuit is shown in Figure 5.5. The small signal was chosen to be 0.35 volt peak to peak at 1 kHz. The single probe was a 10 mm long and 0.9 mm in diameter tungsten wire. Figure 5.6 to 5.8 show the typical normalized electron energy distribution function in MPDR 5, MPDR 20 and MPDR 325. In each figure, the average electron energy of the experiment data was calculated from the normalized EEDF and then the Maxwellian and Druyvesteyn distributions with the same average energy were plotted for comparison. As shown in the figures, the EEDF in MPDR plasma sources was neither a Maxwellian nor Druyvesteyn type distribution. In all these cases, the population of the high energy electrons in each EEDF is less than that in Maxwellian distribution. 5.2.3 Ion energy analyzer Ion energy distribution function (IEDF) is probably the most important plasma parameter in ECR plasma sources. A multi-grid energy analyzer is used to find the ion energy distribution in a discharge. 0.10 0.08 ....... ---.--41lr.¢ 6 Au. 0 . A00N205L02v «13.. 0 04 - 0.02 0 128 0.10 0.08 .0 o m f(E) (Normalized) O b a 0.02 0.00 0.00 1.5 mTorr. Ar on. 5 cm downstream ioowattet .1-14562 r-Ocrn Average energy = 8. 44 eV 0 Experimental data \ o —— Maxwellian distribution — - Druyveeteyn distribution 10.00 20.00 30. 00 40.00 Electron Engergy (eV) Figure 5.6 EEDF in MPDR 5 f(E) 0.12 0.08 0.04 0.00 129 lllllllllelljllLIJIJ 0.0 5.0 a Experimental data (E) = 7.05 eV Maxwellian : (E) = 7.05 eV Druyvesteyn : (E) = 7.05 eV 0 cm downstream, 0.8 mTorr Argon, 110 watts at 915 MHz \ \'—_n . 3"-‘—-H 10.0 15.0 20.0 25.0 30.0 35'. Electron engergy (eV) Figure 5.7 EEDF in MPDR 20 TIII 0 40.0 1(E) 130 0.16 5 cl 1 .1 j o Maxwellian distribution: < E>-6.01 eV 0.12 E 0 Experimental data: < E) -6.01 eV 2 O Druyveeteyn distribution: (DI-6.01 eV 1 I a 1.0 mTorr. 550 Watts, 2.45 GHZ. Argon 0.08 1 \t 14 cm downstream. center I I -1 a \ -1 I, ‘ ' \ .' . °s 0.04 to o .. \ 1:: ~‘ 0 y a . D - 0.00 IIrIII [I111] IIIII I llrtrII I I III1II III ;."‘.j~r’:“.--:~ ’;-' '. ’. ' 0.00 10.00 15.00 20.00 25.00 30.00 Energy (eV) Figure 5.8 EEDF in MPDR 325 F Th ion 6 which h has to b collectet analyzer electrons retardatit retardatic collecting during thl Th1 131 Figure 5.9 shows a schematic circuit for the ion energy measurement. Th ion energy analyzer consists of a 50 pm thick stainless steel membrane which has a center hole with 100 pm in diameter. The diameter of the hole has to be smaller than the thickness of the sheath so that only the ions are collected. There are two grids and a collecting plate inside the energy analyzer. The first grid is an electron repulsion grid, which repels away any electrons that diffuse through the center hole. The second grid is an ion retardation grid. Only those ions that have energies greater than the retardation potential can pass through the second grid and reach the collecting plate. A picoarnmeter was used to monitor the ion current, Iic: during the experiment. The ion energy distribution function is given as [60]: (1 lie f1(5) °< * (Vs-Via'Vz (5.12) d v.1r where Iic is the ion current, V11r is the potential of ion retardation grid, Vic is the potential of the collecting plate and E is the energy of the ion which is equal to qV-u. Similar to the single probe measurement, a small a.c. signal was added to the retardation potential, Vir- The lock-in-amplifier was set to the same frequency of the a.c. signal, which was chosen between 40 to 45 Hz. The magnitude of the small signal was maintained at 0.35 volt peak to peak throughout the experiment. A typical ion current was in the range of a few houndreds nanoammperes. The ion energy analyzer can also be biased relative to the grounded chamber wall. During the experiments, Ver was set to 60 volts and Vic was set to 13.8 volts. Vacuum ch \ 8f 132 Vacuum chamber wall . . Electron repulsron grid Ion retardation grid 100 um diameter hole Collecting plate 1 __ l I Vbias <9 Vi, + a.c.signal Figure 5.9 Schematic circuit for ion energy measurement cm3 a This pl 2.45 G: which 1 the plas downm Constant Alconst Plasma . drastical generate. 101110 sp‘ confmEm diameter distribute Shows the and mic“) 133 § 5.3 Experimental Results 5.5.1 MPDR 5 The plasma densities obtained in the MPDR 5 were as high as 7x1011 cm'3 at zero downstream, which was the highest of all the MPDR sources. This plasma density is ten times the theoretical value of the critical density at 2.45 GHz. The critical density at 2.45 GHz is the density of a cold plasma in which microwave frequency less than 2.45 GHz cannot propagate through the plasma. At 915 MHz, the critical density is 1010 cm'3. Figure 5.10 displays the variation of the plasma density versus the downstream position for different pressures and input microwave power. At constant 1.5 mTorr pressure, a higher input power produces a denser plasma. At constant 160 watts microwave power, there is no significant difference in plasma densities at l mTorr and 1.5 mTorr. The plasma densities drop drastically as the plasma diffuses downstream. This is because plasma is generated in a 5 cm diameter region. Once coming out of the MPDR, the ionic species diffuse freely inside the downstream chamber. A plasma confinement structure is situated right under the MPDR 5 source. It has a diameter of 20 cm. As a result, plasma distribution is more uniformly distributed in the downstream locations as shown in Figure 5.11. Figure 5.12 shows the variations of radial distribution of plasma at different pressures and microwave input power. 134 A E 5.00 D 1.0 mTorr. 160 watt 0 1.5 mTorr, 160 watt A 1.5 mTorr, 100 watt :1‘ o o 5" o o 2.00 1.00 center, Argon IllllllllllllllLllJlJJLllllllllllllllllllllllllll Plasma density (x iEii per cubic c .0 o o IIIIIIIIIIIIIIIIIII[IIIIIIIIIIITIIIIIIIIIIIIIIIIIII[ITIIIII 0.00 1.00 2.00 3.00 4.00 5.00 6.00 Downstream posrtlon (cm) Figure 5.10 Variations of plasma densi with downstream positrons in MPD 0 5. 0. 5. 0. 5 C. 3 2 2 11 1| 0 0 ACCO 0530 1.00 rem: XV \AthDD DEmDEQ /'\ E 3.00 CC 9 .0 7‘ 7" I" I“ 01 0 U1 0 01 o o o o 0 Plasma density (x 1Eii per cu .0 o o 135 O 1.5 cm downstream D 3.5 cm downstream A 5.5 cm downstream 160 watt, 1.0 mTorr, Argon 1.00 2.00 Radial distance (cm) 3.00 4.00 |IIIIIIIITITIllllIII'lllIIIIII'IIIIIIIIIIIIIIIIIIIIIIIIITTII 0.00 5.00 Figure 5.11 Radial distributions of plasma density at different downstream positions in MPDR 5 6.00 O O 8. 5 0 0. AEO 2030 1.00 0 A 0 PPMI. O 2 0. XV \AthDD OEmD‘ ——r«-H--—HJ—#4~————fi—~———---T O 0. 0 Q 01 Flgurj O 0 Plasma density (x 1E11 per cubic cm) 136 .0 00 o - -I q -l q q —1 0.60 .0 .3; O .0 to o O 1.5 mTorr. 160 watt 0 1.0 mTorr, 160 watt A 1.5 mTorr, 100 watt 5.5 cm downstream, Argon JLIIlllllIllllllllll .0 o o O .0 1.00 2.00 3.00 4.00 5.00 Radial distance (cm) Figure 5.12 Radial distributions of plasma density at different pressures and microwave power in MPDR 5 IIlIITIII[IrTIIIIIIITTTITTTIITPIIIIIIIIIIIIII[IllIIIIIIIIll Variz shown in direction at cm downstt downstream dropped frc downstream the ion ener One sources is i plot in Figu the radial d is the dowr Maximum , distribution AS in dOWHStFEan mlCI‘OWaVe downstreart 137 Variations of plasma potential in radial and downstream directions are shown in Figure 5.13. Plasma potential decreases in the downstream direction and radial distribution of the plasma potential is very uniform at 5 cm downstream. Figure 5.14 shows the average ion energies as a function of downstream position at different power and pressure. The energy of the ions dropped from nearly 30 eV at 2 cm downstream to half of its value at 5 cm downstream. Such a big drop in ion energy can be explained by examining the ion energy distribution functions. One important difference in MPDR 5 from other MPDR plasma sources is its ion energy distribution function shown in Figure 5.15. Each plot in Figure 5.15 represents a normalized IEDF at different locations. r is the radial distance in centimeter from the central axis of the system while (I is the downstream position, also measured in centimeter. Full Width Half Maximum (FWHM) of the ion energy distribution is given only when the distribution is Gaussian type, i.e., single peak distribution. As indicated in Figure 5 .15, there was a gradual change of IEDF from Gaussian type distribution to a bimodal distribution as the plasma diffused downstream. This phenomenon was also observed at a different pressure and microwave power as shown in Figure 5.16. As the plasma diffused downstream, the population of the high energy ions dropped and a low energy peak emerged. However, there was no significant change in the radial distributions of the IEDF at 4 cm downstream as shown in Figure 5.17. 25.00 20.00 .—t 5" o 0 Plasma potential (V) 3 8 138 1:1 radial direction, z=5 cm, 100 watt 1 .5 mTorr, Argon IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 0 downstream direction, r=O cm, 110 watt 'o 0 1.00 2.00 3.00 4.00 5.00 POSItlon (cm) Figure 5.13 Variations of plasma potential in MPDR 5 IIIIIIITIIIIIIIIIITIIIIIIIllIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 6.00 a>®v >DL®C® COL Ion energy (eV) 139 40.00 _ 30.00 {1 20.00 E 10.00 E D 160 watt. 1.0 mTorr : 0 100 watt. 1.5 mTorr 3 center, Argon 0.00 - lllllllllllllllIIIIlllllllllITTTllllTIIIIITlllIll 1 .00 2.00 3.00 4.00 5.00 6.00 Downstream position (cm) Figure 5.14 Variations of ion energy with downstream positions in MPDR 5 Cl m cl ‘qaifl‘i‘1“31.il‘§flfljl-i' o m m m m m m m m m 933.9502 0? 305.0502 0v; 0.1 F1] 140 0.10, E/ 0.10. : momma-chm : : 1.0m1’ - OI : IQmI-O - m-isfn’w : 1.0mwr.- MW 5..-. 1 "3 ‘"° mjjr‘UI? I I IIIIIIIII I IIIIIIIIIIIIIIIIII I ''''''''' Ion energy eV) 2 0.10. 0.005 10011001. 3 1.0 mTorr.\ :t." (I) C Downstream (D " 915 MHz. 200 watts U 0.6 mTorr, Argon O 4 cm With confinement Cl 10 cm 8 Flow rate: 26 sccm alt 13 cm L. +1 0 (D 10 E10 Irlrjrlrrliir1lfrlrffiurT1111TFI11Iltrf —-1o.o —7.5 — .o —2.5 0.0 2.5 7.5 10.0 0 Radial distance (cm) Figure 5.19 Radial distributions of plasma density at different downstream positons in MPDR 20 147 A E O o : °- - 0 .0 _ A A 3 6.0 _ O .. L I 0 _ O. - o I 1— 1 La 5.0 '1 .— + x _ 1 0 ° v _ >~ I .t -l 2 4.0 _. o I o 190watts '0 - n 253 watts a q A 317watts n g jFrequencyz915MHz 0 380 watts u d argon, 0 cm downstream €300 IllllllllllllljlllllllllIIIIITTITI 0.4 0.8 1.2 1.5 Pressure (mTorr) Figure 5 .20 Variations of plasma density with pressure at 0 cm downstream and different microwave power in MPDR 20 148 distributions of the plasma potential at different pressures and microwave power are shown in Figure 5.21 and 5.22 respectively. Plasma potential is related to the population of high energy electrons in the EEDF. When pressure increases, the collision frequency increases. Therefore the energetic electrons are more likely to loss part of their energies as a result of the collision processes. At constant microwave input power, plasma potential will drop as the pressure increases. An increase in microwave input power also increased the plasma potential as shown in Figure 5.22. The distributions were very uniform at 12 cm downstream over 9 cm in radius. Variation of plasma potential with downstream location is shown in Figure 5.23. The plasma potential decreased from about 10 volts at zero downstream to about 7 volts at 14 cm downstream. The radial distribution of the electron energy is shown in Figure 5.24. As expected, the distribution was similar to the radial distribution of the plasma potential. Figure 5.24 shows the variation of the electron energy with downstream positions. The electron energy dropped very slightly because of the downstream magnetic confinement. Plasma potential in the MPDR 20 is much lower than the plasma potential in the MPDR 5. The low plasma potential in MPDR 20 has a direct impact on the ion energy distribution. A high plasma potential means a thick sheath between the plasma and the substrate surface, in which ions are accelerated. Figure 5.26 shows the IEDF in MPDR 20 without any d.c. bias to the ion energy analyzer. The figure shows a decrease of average ion energy from 12.8 eV at 4 cm downstream to 8.8 eV at 14 cm downstream without any broadening. Such decrease in ion energy was the result of a 12.00 10.00 ' O A>0v .O~+CO*OQ Orcnorl 149 12.00 A > o V1000 :é " a 4— . c d 0 . ..— O -1 a. 1 a 0.6 mTorr D j o 0.8 mTorr E 8.00 m - 0 1.0 mTorr .2 . 0- I 1 915 MHz. 100 Watt. 12 cm downstream. Ar 1 6.00 ITTIIUIIIIYIIIUIITTIIIIIITTYIIIIIITIIIIIIVIIlllTI 0.00 2.00 4.00 6.00 8.00 1 0.1 Radlal dlstance (cm) Figure 5.21 Radial distributions of plasma potential at different pressures in MPDR 20 12.00 11.00 0 0 O. O. 9 8 A>0v _0_+CO+OQ 02.520 10.00 12.00 11.00 1 0.00 Plasma potentlal (eV) .m (D 8 8 7.00 150 0 80 Watt 0 100 Watt 0 120 Watt 915 MHz, 0.8 mTorr, 53.7 scorn Ar 12 cm downstream Radlal dlstance (cm) Figure 5 .22 Radial distributions of plasma potential at different microwave power in MPDR 20 I'UIIFUUI 6000 I'II'II" I‘TTY‘TI' IIIIIIIII TI'I'IUII 0.00 2.00 4 00 e 00 8.00 1 0.! A \/ V _ 0V . .u FL Luv d AWV Aim nu. etulu m" BL ~ 1% Plasma potential (V) 12.00 10.00 8.00 6.00 4.00 2.00 0.00 151 915 MHz. 110 watts 0.78 mTorr. Argon fi C1 '1 .1 d -i '1 u—d .1 q d d .. .. .. .. d .1 .1 .1 .1 .1 111117111111IIITIIIITIIYYIIUUITIIII11111111111ITTITTTTTTITIIIIIIIITIIIIIIIIIIII 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 Downstream pasution (cm) Figure 5 .23 Variation of plasma potential with downstream positions in MPDR 20 Average electron energy (eV) 152 8.00 6.00 z“ o o O 915 MHz. 0.76 mTorr. 43.8 scam Ar. 80 Watts 12 cm downstream I" o o llllllllllLlLLllllllllllllllll1111 0000 rITIYITITTIIIII‘IUIITIIIIIIIIlIIIWIrIIIIIIUUFTrII 0.00 2.00 4.00 6.00 8.00 10.00 Radial distance (cm) Figure 5.24 Radial variation of electron energy at 0.76 mTorr and 80 watts microwave power in MPDR 20 Average electron energy (eV) 7" 1° 9" 5‘ .U‘ 9’ .\‘ 9° 0 O O C) O O O O O O O O O O O O .0 o o 153 I\O\OGO\0 1 o -1 1 -1 '1 I .1 .1 915 MHz. 110 watt: 3 0.78 mTorr. Argon I '1 I dllIIIIYIIIITTITIIIYIII11111111‘ITTITIII{IIIIIIITIIWTI11111111111YTT1]IIIIITIIT 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Downstream position (cm) Figure 5.25 Variation of electron energy with downstream position at 0.78 mTorr in MPDR 20 16.0 0.15 "r— 111 .dil «1 «1.4i41.1114..j11]1111l--1W.11111111111NJ11I]1111114\1 5 0 ll 0 A0vu M 1414‘0. 00 0. 0 154 0.15 G——-O 4 cm downstream G——6 8 cm downstream A-———0 12 cm downstream 6—0 14 cm downstream 1 0.10 -* ' 915 MHZ. 110 watts A Z 0.78 mTorr. Argon LL] .- v -I W- .1 0.05 ‘ fl 0.00 l 1 i r I i "— 20.00 300 ‘ 40.00 50.00 60.00 Ion energy (eV) Figure 5 .26 Variations of IEDF at different downstream positions in MPDR 20 155 decrease in plasma potential. The full width at half maximum (FWHM) of each IEDFs in Figure 5.25 was about 8 eV, which was smaller than those results obtained in MPDR 9 by Jeff Hopwood and the FWHM shown in the MPDR 5. When a d.c. bias voltage was applied to the ion energy analyzer, the ion energy distribution shifted by about the same amount without broadening as shown in Figure 5.27. As a result, the energy of the ions impinging on a substrate surface can be controlled by the use of an external bias. One application is to increase the selectivity in the etching of silicon using silicon dioxide etch mask in chlorine plasma. Silicon dioxide can be sputtered off by an ion with energy of 20 0V or above. Therefore, by controlling the bias to the substrate, selective etching can be achieved. Radial variations of the average ion energy with power and pressure are shown in Figure 5.28 and Figure 5.29. The ion energies were higher at 0.6 mTorr than at 1.0 mTorr. This was primarily caused by the variation of the sheath potential at different pressure. Radial uniformity of the ion energy was very good at 12 cm downstream as shown in Figure 5.28. Variation of the average ion energy is shown in Figure 5.30. The ion energy decreased from about 13 eV at 4 cm downstream to 8 eV at 14 cm downstream. The most significant features in MPDR 20 are its low ion energy and high uniformity in the downstream position at about 12-14 cm below the baseplate. A low ion energy plasma is essential for anisotropic etching of semiconductor with minimum substrate bias. This is critically important because the etching process is controllable by an external bias and the damage to the substrate material is minimal. A uniform plasma over a large f(E) 156 0.15 0.10 0.05 1111141111111111411111111114 0.00 0.00 I ' H 0 V bias G—a -10 V bias A-—-0 -21 V bias 915 MHz, 110 watts 0.78 mTorr. Argon 14 cm downstream 10.00 20.00 30.00 40.00 50.00 60.00 lon energy (eV) \V\. - e ~_. Figure 5.27 Variations of IEDF with external d.c. bias in MPDR 20 .-...r-fiuuquq141qlu1uju1-qq~1-d~40 0 0 0 0 0 5 he. 5 0 A>0v samLOCO C0. 000.5.)6‘ Average ion energy (eV) 157 15.00 10.00 i 3 a 100 Watts 1 o 120 Watts 5.00 - - 12 cm downstream. 0.8 mTorr. 915 NH: 0000 UTIIITIIIIUWUIIIIIFIFITIIIUI‘ljllIIIIIIITjIIIT1UI 0.00 2.00 4.00 6.00 8.00 10.00 Radial distance (cm) Figure 5.28 Radial variations of average ion energy with microwave power in MPDR 20 O —114J.J:- -1qu Mil-11F»? «1.1 11431111131110. 0 0 0 0 0M 0 O 0 0 Kw Rd. A>0V \nmLoco com 00°L0>< 200 100 158 20.00 “4.14—1.1.1.1 0| 0 O 10.00 a 0.6 mTorr a 1.0 mTorr Average ion energy (eV) 9' O O 5.4 cm downstream, 100 Watts,915 MHz 1104111111L1111111L11111 0.00 IllIITIIY[TlIIITIIIIITTijiITITIIIITIIIIIIIIIITUIT 0.00 2.00 4.00 6.00 8.00 10.00 Radial distance (cm) Figure 5.29 Radial variations of average ion energy with pressure in MPDR 20 16.0 12.01 8 0C A>0v \AOmeme CO_ 0 0 4n 000L0>< 0.00 159 16.00 4 w, r 1 A : i > 4 (D '1 . V1200 3 o >\ I Ch . L. .1 Q) a 9 m 6.0051 C d o ; 91_5 MHz. 110 watts ._ 4 0./' mTorr. Argon g .1 a 4.00 1 l. I Q) - > 1 < . I 0.00 IIIIIIIIIIITTTYTWIIIIIIIIIII]IIIIIIIIIIiTITYIlTlfIIllIIIII]IIIIIIIIIIIIIIIII 0.00 2.00 4.00 6.00 8.00 10.00 12.00 Downstream posntton (cm) Figure 5.30 Variation of average ion energy with down- stream position in MPDR 20 14.00 16.00 160 area is important for the processing of large diameter wafer. Currently, up to eight inches in diameter wafer are being used for production. Thus a uniform plasma is necessary to achieve uniform processing and high production yield. The MPDR 20 was also tested at 2.45 GHz. It was the first MPDR source that had a discharge diameter larger than the cutoff radii of the cavity resonant modes. However, most of the characterization experiments at 2.45 GHz were performed in another large diameter MPDR, the MPDR 325. It is reported in next section. 5.5.3 MPDR 325 The discharge diameter of the MPDR 325 was 10 inches, the largest among all the MPDRs. This reactor is a commercial plasma source which was used in the etching application in this dissertation [43,45]. Figure 5.31 shows the variation of plasma density with pressure and microwave power in this source. The plasma densities increased with both power and pressure. Although the plasma density is not as high as in the MPDR 5, the ion production efficiency in the MPDR 325 is the highest among all the MPDRs. More discussions about the production efficiency are given in next chapter. Uniformity was again one of the major concerns in this reactor. Most of the characterization experiments were conducted at downstream positions where etching experiments were conducted [43,44]. The radial distribution of the plasma density is shown in Figure 5.32. At 14 cm downstream, the variation of the plasma density was 8.1 % over 18 cm. At 18 cm 161 A E 2.0 O A .2 g 1.8 0 L 0 04.6 A Z LIJ " 1.4 X 0 V ..2‘ 'fi 1-2 O 290 watts g a 380 watts '0 A 480 watts o 1.0 5 Frequency: 2.45 GHz 0 argon, 0 cm downstream Q 0.8 ' 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Pressure (mTorr) Figure 5.31 Variations of plasma density with pressure at 0 cm downstream and different microwave powers in MPDR 325 8.00 9’ o o 4.00 2.00 Plasma density (x1E10 /cublc cm) 0000 000000000 000000000 000000000 000000000 60 2.60 4.6 6 s. 162 0) II 1| 0 1| 1 0 U U D 14 cm downstream without wafer holder 0 18 cm downstream 550 watts at 2.45 GHz, 0.8 mTorr, Argon o 6. 0 Radial distance (cm) 6 000000000 0 1 OJ Figure 5.32 Radial distributions of plasma density at different down- stream positions in MPDR325 163 downstream, the variation was 7.5% over 18 cm. Uniformity of the plasma within the downstream confinement structure was further improved by the presence of a 18 cm diameter wafer holder as shown in Figure 5.33 and 5.34. At 14 cm downstream the uniformity of the plasma right above the wafer holder was 3.8 % over 16 cm in diameter. Without the wafer holder, it was 5.4% over 16 cm in diameter. At 18 cm downstream, the uniformity was 3.1% over 16 cm in diameter above the wafer holder. Without the wafer holder, it was 4.7% over the same diameter. This shows that current configuration of the MPDR 325 is capable to process substrate up to seven inches in diameter. Further increase of the substrate diameter require a larger downstream confinement structure. The one used has a 24 cm inner diameter which is too small for the processing of wafers with diameter of eight-inch or above. Radial distributions of the plasma potential at two different downstream positions and pressures are shown in Figure 5.35. When a wafer holder was placed in the downstream chamber, the plasma potential right above the wafer holder was more uniform than the distribution of plasma potential without the wafer holder. This is shown in Figure 5.36. Without the wafer holder, the uniformity was 2.9% over 16 cm in diameter. With the wafer holder it was 1.9% over the same diameter. The results agreed with the results of the plasma density distribution with wafer holder shown earlier. The variations of average electron energy with pressure and microwave power are shown in Figure 5.37. Without exception, the electron energy went up when the pressure was decreased. At constant pressure, the electron energy increased with the input microwave power. 8.00 9’ o o 2.00 Plasma density (x1E10 /cubic cm) .5 b O 0.00 0. 164 1L1 L111llLlLlllLLlllL4lJlllL o 14 cm downstream without wafer holder 014 cm downstream with water holder 550 watts at 2.45 GHz. 0.8 mTorr. Argon 0 0000j0000100T0000I01000000700100000000IIIII000000 o 2.00 4.00 6.00 8.00 10.00 Radial distance (cm) Figure 5.33 Radial distribution of plasma density with and with- out wafer holder at 14 cm downstream in MPDR 325 1° 5» 9 .w .m >1 O O O O O 8 O O O O 0 Plasma density (xii-210 /cubic cm) '0 O 0.00 _£\ev—r o 18 cm downstream without wafer holder 0 18 cm downstream with water holder 550 watts at 2.45 GHz. 0.8 mTorr. Argon 2.00 4.00 6.00 Radial distance (cm) I000000000T000000000]000000r00'0000000001000000000 0.00 8.00 Figure 5.34 Radial distribution of plasma density with and with- out wafer holder at 18 cm downstream in MPDR 325 10.00 166 11.00 .. _ h ;-—-."7 0 “§=\ - _. 10.00 ‘-——~\~a o 9 9 a V 9.00 E ..— c 43 8.00 01.0 mTorr, 14 cm downstream a. D 0.8 mTorr. 14 cm downstream O E 7.00 0 0.6 mTorr, 14 cm downstream t0 0% it 0.8 mTorr. 18 cm downstream 6.00 550 Watt. Argon. 2.45 GHz 5.00 0.00 2.00 O 6.00 8.00 10.00 4.0 Radlal dlstance (cm) Figure 5.35 Radial distributions of plasma potential at different pressures and positions in MPDR325 12.00 10.00 a x A \>/ M _ 8.00 e .2 ..- C ‘2 6 00 o e O. D g 4.00 a 18 cm downstream without wafer holder 5 o 18 cm downstream with water holder 2.00 550 watts at 2.45 GHz. 0.8 mTorr. Argon 0000 [000000000l0000000001000000000ll000000001000j00000 0.00 2.00 4.00 6.00 8.00 167 Radial distance (cm) Figure 5 .36 Radial distribution of plasma potential with and without wafer holder at 18 cm downstream in MPDR 325 10.00 168 8.0 d I 4s) A 7 > I a) - v _, >10 -: a, u L d 2 4 o I S 6.0 3 L . 4.. -t o I 2 4 m ... a) I a: .2 25'0 - o 0.6mTorr a): : D 0.8mTorr < j A 1.0mTorr : 2.45 GHz, argon - Ocmdownstram 4.0 DTTTIIIIITITTI0000001000100000]lIIIIIIIIIITTIIIITI 260 300 350 400 450 500 Power (watts) Figure 5.37 Variations of average electron energy with microwave power at 0 cm downstream and different pressrues in MPDR 325 169 The ion energy distribution functions (IEDFs) of the MPDR 325 are shown in Figure 5.38 and 5.39. Figure 5.38 shows the IEDFs at different pressure 14 cm down below the reactor. As the pressure was decreased, the average ion energy increased from 11.5 eV at l mTorr to about 12.8 eV at 0.6 mTorr. FWHM of the IEDFs varied from 7 eV to 9 eV, which were about the same observed in the MPDR 20. Figure 5.39 shows the variations of the IEDFs with microwave input power. Only a slight increase in the average ion energy when the input power was increased from 550 watts to 750 watts. This is indicated by a shift of 2 eV to the right in the IEDF in Figure 5.39. At constant power, the IEDFs at 14 cm downstream and 18 cm downstream were essentially the same. This is also observed in the radial distributions of the average ion energy in Figure 5.40. An average ion energy of about 12 eV is distributed over 16 cm in diameter at both downstream locations. Figure 5.41 shows the radial distribution of the average ion energy at different pressure at 14 cm downstream. Similar plots of the average ion energy are shown in Figure 5.42 at different microwave input power. At 550 watts microwave input power, the ion energy distribution has a uniformity of 4% (16) over 16 cm in diameter in a location 14 cm below the reactor at 0.8 mTorr. 5.5.4 Summary In summary, the MPDR 5 plasma source had a bimodal ion energy distribution that was not observed in other MPDR sources. The bimodal distribution was repeatable and consistent both in downstream direction and radial direction indicating a charge exchange process takes place within 5 cm downstream. Downstream ionization may exist but play a minor role for f(E) (Normalized) 170 0.16 E 01.0 mTorr. =11.53 eV 0.12 1 Z a 0.8 mTorr, (E>=11.92 eV : 3 0 0.6 mTorr. < E>= 12.82 eV 0.08 f 1 14 cm downstream. center 5 550 watts at 2.45 GHz. Argon 1 0.04 f 3 0.00 .m...n.n...... . . . : . 0.00 10.0 20.00 30.00 40.00 50.00 Ion energy (eV) Figure 5.38 IEDF at different pressures in MPDR 325 f(E) (Normalized) 171 0.16 q i O 14 cm downstream. 550 watts 0.12 1 1 u 14 cm downstream. 750 watts E O 18 cm downstream. 550 watts 0.08 f 1 0.8 mTorr. 2.45 GHz. Argon 0.04 f d [Jr C "\ .2. 0.00 ‘l: IIIII1IIIiTI' ‘ "41 '4 '-‘ “‘ 0.00 10.0 20.00 30.00 40.00 Ion energy (eV) Figure 5.39 IEDF at different microwave power and downstream positions in MPDR 325 50.00 Average ion energy (eV) 172 15.00 111111 10.00 lllJlllllLLllllJlllllll 0.00 a 14 cm downstream o 18 cm downstream 0.8 mTorr. 550 WattS, 2.45 GHz 0.00 TTIYTIIITIIIIIITTjIII[IIITII’IIUTIT]IIIIITIIIIIIT 2.00 4.00 6.00 8.00 10.00 Radial distance (cm) Figure 5.40 Radial distribution of ion energy at different downstream positions in MPDR 325 Average ion energy (eV) 173 15.00 M *5 10.00 a 0.6 mTorr D 0.8 mTorr 5.00 0 1.0 mTorr 14 cm downstream. 550 Watts, 2.45 GHz 0.00 [ITI1TIIII]ITTITITIIIIfIIIIIIIIIITIIIIIIITIIIIVIII 0.00 2.00 4.00 6.00 8.00 10.00 Radial distance (cm) Figure 5.41 Radial distribution of ion energy at different pressures in MPDR 325 Average ion energy (eV) 174 15.00 + + 4 M 10.00 - 3 u 550 Watts 5.00 - o 700 Watts ~ 0.8 mTorr. 14 cm downstream. 2.45 GHz 0.00 TUITIII11111ITIIIIIIIIIIIIIYIIVYIIIIjITIIIIIIrIT—Il 0.00 2.00 4.00 6.00 8.00 10.00 Radial distance (cm) Figure 5.42 Radial distribution of ion energy at different microwave power in MPDR 325 175 the emergence of the low energy peak in the IEDFS. The source had a high plasma density (5x1011 cm‘3) at zero downstream position. With an 8-inch in diameter downstream confinement structure, the uniformity of plasma density improved from 92.4% (10) at 1.5 cm to 18% (10) at 5.5 cm downstream. The average ion energies varied from 12 to 30 eV depending on downstream location. Because of the small size of the discharge volume, the lowest operation pressure is limited to 1.1 mTorr. The microwave power used in this source was limited to 100-190 watts with a power density varied from 2 to 4 watts/cm3. An increase in microwave power is possible by redesigning the quartz disk. However, a much higher power density is not desirable since the sheath potential will be too high for some applications. The MPDR 20 plasma source is the only MPDR which was operated with 915 MHz microwave energy. It produced a uniform plasma, 2.6% (10) over 18 cm in diameter at 13 cm downstream. It had a low plasma potential, about 10 volts at zero downstream and 7 volts at 14 cm downstream. It has low ion energies, 8 to 10 eV, and a small FWHM (about 8 eV), making the source ideal for etching wafer up to 6 inches in diameter. The reactor was operated at less than 500 watts which was limited by the 915 MHz microwave source. The MPDR 325 shares all the advantages in MPDR 20. It had a low ion energy plasma (from 10 to 14 eV) and small FWHM (about 8 eV). The plasma was uniform (3.1% over 18 cm at 14 cm downstream) for large diameter wafer processing at downstream location. The MPDR 325 had a high plasma density in the downstream chamber, about 6x1010 cm'3 at 14 cm downstream. The source was operated at about 600 watts microwave. 176 However, it can be operated at power up to 1500 watts. In addition to those advantages discussed earlier in Chapter Four, the MPDR 325 is the best plasma source in the MPDR family for semiconductor processing. CHAPTER SIX 7 r. r ‘ .> - r K o - r. r. - C A 0 § 6.1 Introduction § 6.2 Experimental setup § 6.3 Definitions of figures of merit 6.3.1 Electron temperature versus PA 6.3.1 . l Free-Fall diffusion 6.3.1.2 Schottky diffusion 6.3.2 Microwave coupling efficiency 6.3.3 Ion production energy cost and power densities § 6.4 Experimental results § 6.5 Summary 177 I. 178 §6.1 INTRODUCTION In Chapter Five, the experimental results of three MPDRs are reported. The operation conditions in each MPDR are different from each other. For example, the small MPDR 5 cannot be operated at microwave power greater than 200 watts and the lowest operation pressure is about 1.1 mTorr while microwave power up to 1000 watts can be used in the MPDR 325 and the lowest operation pressure is about 0.4 mTorr. In the past, many questions concerning specific designs have been solved by intuition or by a "cut and paste" approach. However, important questions remain unanswered. For example, "What is the optimal size for the MPDR ECR plasma source?", "How do the properties of the discharge vary as the input power, discharge pressure and discharge size change?", "How does the microwave coupling efficiency change as the excitation mode or discharge size change?" and "How does the overall ion production cost vary as the design changes?" Answers to these questions will evolve from experimental evaluations and the development of theoretical models of the ECR plasma sources. This chapter investigates the experimental performance of five different diameter MPDRs using argon plasma. Source performance versus pressure, input power and discharge size are experimentally evaluated using four figures of merit. They are the electron temperature versus pressure- diffusion length product, ion production energy cost, microwave coupling efficiency and discharge power density. The MPDR 5, MPDR 9, MPDR l3 and MPDR 325 are evaluated with 2.45 GHz microwave excitation 179 frequency while the MPDR 20 is evaluated at 915 MHz. These figures of merit may also be applied to other ECR sources. The definition of each figure of merit is given in section 6.3. §6.2 Experimental setup A number of experiments were conducted in each of the MPDRs to evaluate the four figures of merit. Figure 5.2 shows the experimental setup of the experiments. The experimental setup for each MPDR was the same, except that different MPDR was mounted on the top of the vacuum systems. The zero downstream position, z=Ocm, was defined as the bottom of the baseplate in each MPDR. Probe measurements of the average electron energy and plasma density were performed at zero downstream position. A 2 mm diameter micro-coaxial probe, as shown in Figure 5.2, was used to measure the electric field strength inside the cavity for the calculation of microwave coupling efficiency. All experiments used downstream plasma confinement structure. As shown in Figure 4.4(b) the MPDR 5, MPDR 9 and MPDR 13 had a 20 cm diameter, eight pole, downstream confinement structure mounted below the stainless steel plate. A 28 cm diameter, twelve pole, downstream confinement structure was used in the evaluation of the MPDR 20 and MPDR 325. Table 6.1 is a list of information and a collection of plasma data of all the MPDRs. In the table, the cavity inner diameter is listed for each MPDR. The plasma diameter is the inner diameter of the quartz disk. The magnetic configuration represents the number of ECR magnetic cusps arranged in a manner shown in Figure 4.1(b). The discharge volume is defined as the 180 TABLE 6.1 EXPERIMENTAL DATA FOR MPDRs Parameters MPDR 5 MPDR 9 MPDR 13 MPDR 325 MPDR 20 A D 0 if >K Cavity dia. (cm) 8.9 17.8 17.8 35.6 45.7 Plasma dia. (cm) 4.8 9.1 13.1 24.0 19.6 Mag. config. 8 poles 8 poles 8 poles 12 poles 12 poles A (cm) 0.72 1.29 1.81 2.51 2.46 Plasma vol. (cm3) 54 325 876 3596 2967 Excitation 2.4SGHz 2.4SGHz 2.4SGHz 2.4SGHz 915MHz , frequency Mode TE111 TE211 TE211 (multimode) TE211 Input power (watt) 99-184 168-260 247-616 BOO-500 190-380 Pressure (mTorr) 1.1 - 1.6 0.7-1.3 0.8-1.5 0.6-1.0 0.6-1.5 Vp (V) a 18.3-22.6 255-36 0 17.5-18.5 9 94-108 6.8 -11.8 Ion density (cm3) b 3-5 xlEll 1.3-2 xlEll 1.2-2.5 xlEll 1-2 xlEll 4.6-6 xlEIO ion (eV) a 12.5 -28.0 19.7 -36.1 C 19.5 -21.3 d 10.5-14.5 8.4 - 15.0 FWHM (eV) 3 10.7 -13.6 8.2 -12.5 C 10.8 -15.8 d 7.0 - 8.7 7.0-8.0 a Downstream data (2 2 0 cm) b At 2 = 0 cm C From I. Hopwood [60] d From Mak et a1. [46] Plasma volume = V: 11121.. 181 volume of discharge enclosed by the quartz disk and the boundary at zero downstream position, i.e., z=0 cm, and is equal to 1trzL as shown in Figure 5.2. The parameter, A, is the characteristic diffusion length of the discharge volume. The range of input power and pressure listed were the experimental conditions in which the figures of merit were evaluated. The plasma potential (VP), ion density, ion energy and FWHM were the data either obtained from the experiments or collected from earlier publications [48,60,6l,l65]. §6.3 Definitions of figures of merit 6.3.1 Electron temperature versus PA One important figure of merit is the discharge electron temperature versus the pressure-characteristic diffusion length product. The electron temperature of the plasma was obtained by single probe measurements taken in the center at z=0 cm downstream position. The characteristic diffusion length, A, is defined later by equation (6.7). Two theoretical curves were also generated for reference. One of the two curves was based on Free-Fall Diffusion. The other was based on Schottky or Ambipolar Diffusion. The mathematical approach used here is similar to Mahoney's work [61]. Details of the derivation in both Free-Fall Diffusion model and Schottky Diffusion model are given. The influence of the neutral temperature on the pressure-characteristic diffusion length curve is also examined. 182 The starting point of the theoretical analysis is to balance the ionization rate of the neutral molecules with the diffusion loss to the boundary. The ionization rate is represented by NoVchff’ where N0 is the plasma density, vi is the ionization frequency and Veff is the plasma volume. 6.3.1.1 Free-Fall Diffusion In Free-Fall Diffusion, the Bohm sheath criterion [62] is used to describe the diffusion of the ions towards a wall through a sheath. This model of diffusion through a sheath is only good when the mean free path is much larger than the thickness of the sheath and therefore no collision takes place inside the sheath. A quasi-neutral plasma with Maxwellian distribution outside the sheath is assumed. Because the electrons move much faster than the ions, they reach a surface and start to accumulate. An electric field is created which repels the incoming low energy electrons but meanwhile attracts the ions towards the surface. At steady state, the flux of the electrons towards the wall is equal to the flux of ions through the sheath. If a planar sheath is assumed, the flux of the ions reaching the surface is [55]: I“, = 1/2N0(kae/Mi)% (6.1) where No is the plasma density, kb is the Boltzmann constant, T3 is the electron temperature and Mi is the mass of the ions. Assuming a cylindrical 183 discharge of radius r and length L as shown in Figure 5.2, the volume of the discharge is: veff = ma. { (6.2) If ions and electrons cannot diffuse across an azimuthal magnetic field and M is the number of poles in the magnetic configuration, then the effegtjye area for diffusion loss in the plasma volume becomes [171]: A6“ = 2m(r + L-qu) + MA,“ (6.3) where Am is the pole area of each magnet. At steady state, the rate of ionization is equal to the rate of diffusion to the wall. Thus: I‘i Aeff = NoViVerr (6-4) The ionization frequency is given as [61]: 2 \lz .. v, = Nn l 6 0(2) exp (—e/kae) de (6.5) (Tune)56 (kae) 3,2 0 where e is the electron energy, 6(8) is the ionization cross-section by the electrons and Nn is the neutral density. The neutral density can be expressed in terms of pressure and temperature: Nn = P/I‘nkb (6.6) 184 The characteristic diffusion length is defined as: A = Veff/ Aeff (67) After substituting equation (6.6) into (6.5) and equations (6.1), (6.2), (6.3) and (6.5) into equation (6.4), following relation can be obtained after rearrangement: (kae)2 1% (mneyfrn PA = - (6.8) I 8 0(8) exp (—e/kb'1‘e) de 4(2Mi)% where P is the pressure in Pascal and A in m. The total cross section, 6(8), for ionization in argon gas by electron can be found in an earlier publication by Rapp and Englander-Golden [65]. 6.3.1.2 Schottky Diffusion This diffusion model assumes that the mean free path between collision is much smaller than the size of the discharge volume. Since electrons move much faster than the ions, a space electric field is established which accelerates the ions but slows down the electrons. As a result, both ions and electrons diffuse with the same diffusion coefficient, D3, which is known as Schottky or ambipolar diffusion coefficient. Considering an infinitesimal volume of discharge with a non-zero space charge field, E, the flow of electrons and ions are: BecaL where The de and ele Where f0r the j The con 185 rc = — NepeE - DCVNC (6.9) I} = Niui E - DiVNi (6.10) Because Nc=Ni=N and I} = I} = F, it can be shown that: 1“ = -DaVN (6.11) where uiDe + “eDi Da= (6.12) He + Hi The definitions of the mobility, u, and diffusion coefficient, D, for the ions and electrons are given as below: l~lr=6/MiVim , 11e=C/meVie Di = kai / Mivim , De = kae / meVie (6.13) where vim and vie are the collision frequencies of the momentum exchange for the ions and electrons respectively. The continuity equation is given below: aN -WN+VT=O (an) at 186 Using equation (6.11), it can be shown that, at steady state, the following relation can be obtained: VZN + (vi/Da)N = 0 (6.15) By the method of separation of variables and using a cylindrical coordinate system, an approximate solution for equation (6.15) is given below: N(r,z)=Nolo(2.405r/R)cos(1tz/L) (6.16) where R and L is the radius and length of the discharge respectively. The plasma density is zero on the boundary when 2 = :l:L/2 or r =R. No is the plasma density at the center of the cylindrical discharge volume. The relation of the vi and Da can be obtained using equation (6.15): V, = Da [ (2.405/R)2 + (n/L)2 ] (6.17) Since the temperature of electron is much higher than the temperature of ion inside the plasma, Da can be approximated by mkaJe and the mobility is: 111 = re Ni (6.18) I = A/ (6.19) where t is the relaxation time. For a Maxwellian distribution, the mean free path between collision is: x = l/vZNnan (6.20) 187 an is the collision cross section of the argon molecules [63] and equals to 4.2x10‘19 m2. and for Maxwellian distribution are: .—. (awn/mire (6.21) = 3kan/Mi (6.22) where Tn is the neutral temperature. Using equation (6.5) and substituting all the variables into equation (6.17), following relation can be obtained: (kae)2 kszn(meTeTn)1’6 (P002 = * (6.23) 1 8 0(8) exp (—e/kae) de 3‘12 M95an where P is the pressure and at is the characteristic diffusion length in Ambipolar Diffusion model and is defined as below: a = [(2.405/R)2 + (1t/L)2] V2 (6.24) The values of A and at for each reactor are listed in Table 6.2. Figure 6.1 shows the variation of electron temperature with PA at different neutral temperatures. The unit for the pressure is converted into Torr and A into centimeter because they are commonly used in experimental environment. As shown in the figure, at constant pressure, electron temperature has to be high to sustain a plasma in a small discharge volume, which has a small A value. On the other hand, at constant A, the electron temperature goes up when the pressure goes down. A higher ionization rate, which is related to the electron temperature, is therefore needed to compensate the diffusion loss at low pressure. Figure 6.1 also indicates that an increase of the neutral 188 TABLE 6.2 CHARACTERISTIC LENGTH FOR MPDR MPDR A (cm) (1 (cm) 5 0.72 .69 9 1.29 1.21 13 1.81 1.64 20 2.46 2.26 325 2.51 2.48 Electron Temperature (eV) 189 10.0 _ i 8.0 5 5,0 : 373 K 4.0 E 300 K 2.0 5 3 Free Fall Diffusion .1 0.0 T 1111111] I 1111111] I I 111111] I IIIIIII] 1 1111111 10“ 10" 10'2 10'1 1 10 PA (Torr cm) Figure 6.1 Electron Temperature versus CPA at different neutral temperatures 190 temperature from 27 CC to 100 DC shifts the curve to the right. Such shift represents a small increase in the electron temperature when PA is large, but a big increase electron temperature at small PA. A similar result is also obtained for ambipolar diffusion as shown in Figure 6.2. It is necessary to point out that the values of at and A are very close to each other as shown in Table 6.2. Thus the values of Pat and PA are nearly the same in a semilogarithmic plot as shown in Figure 6.1 and 6.2. 6.3.2 Microwave coupling efficiency The microwave coupling efficiency is another figure of merit to examine the performance of the MPDRs. The technique was developed by Rogers [192] and Manring [166] and is defined as: Coupling efficiency = 100 % x (1- QulQuo) = 100% x (Pa/Pt) (6.25) where Quo is the empty cavity quality factor, Qu is the (plasma) loaded cavity quality factor, P21 is the power absorbed by the plasma and Pt is the total power input to the cavity. Without repeating Manring's work, the relation of the loaded and empty cavity quality factor is listed below: P10 Ital2 Qu = Qua " (6.26) |13,.,|2 Pt where Pm is the net power to the cavity without the plasma, Er0 is the radial electric field at a specific spot of the resonant mode, Pt is the net power to 191 10.0 .. .J A E > 8.0 - (D I v 1 8 : 373K 2 Z O 6.0 : L .— 0 2 Q- i E : 19 40 J 300K 5 i o I L —1 _._ -l 0 I 2 2.0 '3 '41 : E Schottky Diffusion 000- I IOTTIIIF l IIIIIIII I 1111111 10"3 10'2 10" 1 P01 (Torr cm) Figure 6.2 Electron Temperature versus CPCX at different neutral temperatures 192 the cavity with plasma and E, is the radial electric field at the same spot of the resonant mode with plasma. By substituting equation (6.26) into equation (6.25), following relation is obtained: P10 IEr | 2 Coupling efficiency = 100 % x (1 - . ) (6.27) IE,., I 2 P. The radial electric field strengths, IE1 and IEr0 I , can be measured by a micro-coaxial probe as shown in Figure 5.2. Two millimeter diameter holes were drilled both vertically on the wall and along the circumference of the cavity in the MPDR 5, MPDR 9, MPDR l3, MPDR 20 and MPDR 325 sources. The mode of excitation was identified before the measurement of the coupling efficiency. This measurement technique is valid only for single mode excitation. The MPDR 5 was operated at TE111 mode. The MPDR 9, MPDR 13 and MPDR 20 were operated at TE211 mode. A TE711 mode, which was the only single mode identified in the reactor, was used in the MPDR 325. The problem of this mode was that it was not a good resonant mode. Without a discharge, more than 20% of the incident power was reflected. All other reactors had less than 2% reflected power. The plasma created by this mode was also weak and microwave propagated down the vacuum chamber. Moreover, the reflected power went up to about 25%. Therefore, this mode is not recommended for normal operation. A spot that had the strongest electric field strength was chosen to insure the best accuracy of measurement. Calibration of the ratio Pto/Ppo was performed by using a sweeping oscillator (HP8350B) operating in the 193 cw single frequency mode. The frequency of the signal was adjusted to the same frequency of the microwave source which was used to excite the plasma. Pm and Pt, were the net microwave power absorbed by the cavity. They should not include any power loss in the transmission cables. 6.3.3 Ion production energy cost and power densities The ion production energy cost is defined as [191]: Energy cost = Pt li‘ JsdA (eV/ion) (6.28) AS where Pt is the net microwave input power, .18 is the ion saturation current density which can be extracted from the plasma generation region and As is the area of the opening at z=0cm for each MPDR through which the ions diffuse downstream for processing. The ion saturation current density, J5, which flows through a planar sheath is: Is = 0.5eN0(kae/Mi) V2 (6.29) where N0 is the plasma density, Tc is the electron temperature, Mi is the mass of argon ion and e is the electron charge. If the plasma density is uniform inside the plasma volume, then the ion production energy cost is: Energy cost = P[/ (JSAs ) (6.30) 194 where AS is the total cross sectional area of the discharge in each MPDR which opens to the downstream chamber. Since current can be expressed as rate of flow of charged particle, therefore the energy cost can be expressed in units of eV per singly charged ion. Both the plasma density and the electron temperatures were taken at the center of the discharge opening at zero downstream. The power density is defined as: Total microwave input power Pt Power density = = (6-31) plasma volume Veff When determing the ion production energy cost and the absorbed power density, all MPDRs were operated at their normal operation mode as shown in Table 6.1. In other words, the MPDR 325 was operated at multimode instead of TE711 mode, which was used in the evaluation of microwave coupling efficiency. §6.4 Experimental results Figure 6.3 shows the distributions of the experimental 'I‘e versus PA data as well as the two theoretical diffusion curves at room temperature. Data points from a divergent ECR source [1 1] are also displayed. Figure 6.3 indicates that electron temperatures in the MPDRs are lower than those predicted by the two simple diffusion models. In general, the electron Electron Temperature (eV) 44*ODD 195 10.0 9° 0 O) O O 111111111111111111L111111111111111111111111111111 .2" o E " ‘ Ambipolar Diffusion Free-Fall Diffusion i" o O O I I IIIIIII I I I IIIII] I I IIIIIII I I ITIIIII I I IIITII '10“ 10* 10'2 10'1 1 PA (Torr cm) MPDR 5 MPDR 9 MPDR 13 MPDR 20 MPDR 325 Axial Divergent Field ECR source Figure 6.3 Distributions of the experimental CPA data 1O 196 temperature of the discharge in the MPDRs increases when the PA value decreases. Figure 6.4 shows the microwave coupling efficiency versus the characteristic length of the MPDR. As shown Figure 6.4, within the operation conditions listed in Table 6.1, the microwave coupling efficiencies of all the MPDR sources, except the MPDR 325, were higher than 75%. The coupling efficiencies were above 90% in the MPDR 9 and MPDR 13 while the MPDR 20 had an efficiency between 80-90%. The MPDR 325 has lower coupling efficiencies than the other sources. Figure 6.5 shows the relation between the coupling efficiency and the discharge pressure. For MPDR 5, the coupling efficiencies were basically the same over the range of pressure in which it was operated. The same was true to the large MPDR 20 source. However, for MPDR 9 and MPDR 13, there was a slight decrease in coupling efficiency as the pressure went up from 0.6 to 1.2 mTorr. For MPDR 325, the coupling efficiency went down as the discharge pressure went up. Figure 6.6 is another view of the coupling efficiency in each MPDR versus the microwave input power. At constant pressure, the coupling efficiencies were about the same as the power increased in most MPDRs. The only exception was the MPDR 325, which showed an increase in efficiency with the microwave power. Figure 6.5 and Figure 6.6 indicate that the microwave coupling efficiency in each MPDR is not strongly dependent on the pressure and microwave input power within the range of operation conditions investigated in these experiments. 100.00 90.00 80.00 70.00 60.00 50.00 40.00 Microwave coupling efficiencty 0/0 30.00 197 1444* m 1310} 111111111111111111111111111111111111111111111111111111111111111111111 0.0 0.5 1.0 1.5 2.0 , 2.5 A (cm) Figure 6.4 Microwave coupling efficiency versus A IIIIIIITIIIIIIIIITTIIIIIIIIII[IIIfiIIIIIIIIIIIIIIIIIIIIIIII 3.0 Microwave coupling efficiencty °/o 1 00.00 90.00 80.00 70.00 60.00 50.00 40.00 30.00 198 3 Ba 5 B a 9 g E o 5 * .1 9 A 2 1: 1: at g 11 g r a A a : 1‘: ‘k 3 «e; * s g 5 1: w _ i * 1. 5 1k -IIIIIIIIIIIIITIIIIIIIIIIIIIII[IIIITIIIIIIIIIIIIIIITIIIIIIII 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Pressure (mTorr) Figure 6.5 Microwave coupling efficiency versus pressure 199 1003 1 - 5 D B o : :1 @ (B 5* 90'":E %*D% 8 O C : .2 3 :1: Q: o 5 AAA H- : A O : A U) E .E 70E * fit E 5 * o 60% * * *1» 0 E #11: g i as 1k a 50: fir 3 i at: 1k 2 a w .2 40-: 2 3 30.4IIIIIIIIIIIIIIIIIIIITIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII[IIIIIIIII 0 100 200 300 400 500 600 700 Power (watt) Figure 6.6 Microwave coupling efficiency versus microwave power 200 The ion production energy cost versus A is shown in Figure 6.7. The solid line represents those MPDRs which were operated at approximately the same pressure and power (1.0-1.2 mTorr and ISO-260 watts). It shows that the ion production energy cost reaches a maximum in MPDR 9, about 620 eV per ion, and a minimum in the MPDR 325, about 220 eV per ion. This means that the MPDR 325 is the most efficient plasma source among the MPDRs. It should be noted that in this case the MPDR 325 was operated in multimode. Variation of the ion production energy cost with pressure is shown in Figure 6.8. At constant microwave power, the ion production efficiency goes down as the pressure is decreased. This is because the plasma density decreases when the pressure is decreased as shown in Figure 5.20. Figure 6.9 shows the variation of the ion production energy cost versus microwave input power. Results show that at constant pressure ion production cost increases with input microwave power. In other word, the efficiency goes down as the microwave input power goes up. Figure 6.10 displays the experimentally measured power densities versus A. The MPDR 5 was operated at power densities an order of magnitude higher than the MPDR 325. This is because the surface to volume ratio in a smaller discharge is large. The diffusion loss is high resulting a high power density and a high ionization rate to maintain the discharge. The lowest power density needed to create a discharge in the MPDR 5 was about 2 watts/cm3. On the other hand, the lowest power density for the large diameter MPDR to sustain a discharge is about 0.06 watt/cm3 in the MPDR 201 1 200 O O O 800 600 400 Ion production energy cost (eV/ion) 200 11111r11111111111111111111111111111111111111111111111111111 CD IIIIIIII[IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII .0 0.5 1.0 1.5 2.0 2.5 3.0 A (cm) 0 Figure 6.7 lon production energy cost versus A Ion production energy cost (eV/ion) 1 200 O O O 800 600 .p O O N O O 00 202 IIlllllllllllljlllllllijlllllllllllllllllllllllllllllllllL 1:1 8 a B D B a * . * EA 2 . * * * A 3 *A *09 * =1: :1: :1: 0 Gr: :1: o * 3 t .1. 3 IIIIIIII[IIIIIIIIIIIIITIIIIIIIIIIIIIII[IIIllIIIIIIIITTIIII .4: 0.6 0.8 1 0 1.2 1.4 1.6 Pressure. (mTorr) Figure 6.8 Ion production energy cost versus pressure Ion production energy cost (eV/ion) 203 1200_ 3 1:1 5 :1 1000: 1 5 D o 8003 1:1 '3 o O E 1&1 *0 0 o E A A3! 00 600: 8 I 0 E G E 4 00 400: O 5 a *3 E * *1? 11' 200-5 '1" 3 o-IIIIIIIII]IIIIIIIIIIIIIIIIIIIIIIIIIIIII]IIIIIIIIIIIIIIITIIIIIIIIIIIII 0 100 200 300 400 500 600 700 Power (watt) Figure 6.9 Ion production energy cost versus microwave power Power density (watt per cubic cm) 204 10 _ 1: 1 El _ D - a a ‘ 6 at 10"“: 3* : ’fi 2 =1: 10-2 IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIITIIWIIIIrIIIIIIIIIIrIIII 0.0 0.5 1.0 .5 2.0 2.5 3.0 1 A (cm) Figure 6.10 Power density versus A 205 20 and 0.1 watt/cm3 in the MPDR 325. It should be noted that even the MPDR 20 has a slightly smaller power density than the MPDR 325. The discharge density in MPDR 20 is about 30% less than the MPDR 325 at the same microwave input power. Thus the MPDR 325 is more efficient than the MPDR 20. §6.5 Summary Discharge temperature in MPDR goes up as PA decreases. High microwave coupling efficiencies (80% to 92%) were observed in most of ' the MPDRs. The coupling efficiency is not strongly dependent on the discharge pressure and microwave power. It is believe that the microwave coupling efficiency is pimarily determined by the design of the MPDR and the ECR magnetic configuration. A variation of the ECR magnetic cusp in MPDR 13 has shown a coupling efficiency of 97% [47] which is a few percentages higher than that observed in this study. Ion production is more efficient in a large diameter discharge such as the MPDR 325 than in small diameter discharge. As the discharge diameter increases, the minimum power density needed to keep the plasma decreases. The MPDR 325 has the best ion production efficiency among all the MPDRs. As indicated in Chapter Five, it also has a uniform discharge over 7 inches in diameter. The ion energies in the MPDR 325 are low and the IEDF has a small FWHM. It is therefore concluded that the MPDR 325 is the best candidate for large diameter wafer processing. CHAPTER SEVEN § 7.1 Introduction § 7.2 Chemistry of dry etching for silicon § 7.3 Etching methods §7.4 Etching system § 7.5 Preparation of experiments 7.5.1 Optical and electron-beam lithography 7.5.2 Etching conditions § 7.6 Etching Results 7.6.1 Role of gas flow 7.6.2 Effect of rf substrate bias 7.6.3 Variations of etch rate with other processing parameters 7.6.4 Etch rate uniformity 7.6.5 Loading and microloading effect § 7.7 Summary 206 207 § 7.1 Introduction Plasma etching is one of the important applications of an ECR plasma source [81-121, 123-125, 127-149, 173-175]. The primary objective is to achieve anisotropic submicron etching so that more circuits can be integrated on a small piece of silicon substrate. Since there is a trend of processing single large diameter wafers up to 200 mm [126,147], uniformity of the etching process is an important factor in determining the practical value of a new plasma source in a production line. Other issues such as etch rate and the damage to the substrate are also important. All these issues have to be addressed in all the newly developed ECR and non-ECR plasma sources, including the MPDR. The damage issue in the MPDR has been investigated by researchers in AT&T [102,108,115-117,176]. III-V compound substrates, which are very sensitive to the etching process, such as GaAs, AlGaAs, InP, InGaAs and AlInAs were etched by the MPDR 9. Results showed that the MPDR source had little damage to these materials. Although silicon substrates are less sensitive compared to the III-V compounds, damage to this substrate still needs to be investigated. Other issues such as oxide charging, hydrogen passivation and so on are also important. This chapter is concerned with the etching of silicon with feature sizes down to 0.1 micron. A new MPDR 3251 was used in the etching experiments. The only difference between MPDR 325 and MPDR 325i is the cavity diameter, which is reduced from 14 inches in MPDR 325 to 12 inches in MPDR 3251. Both MPDRs use the same 24 cm diameter quartz 208 vessel and the same ECR magnetic configuration. Previous results show that the properties of a discharge are related to the discharge geometry rather than the cavity diameter. Thus properties of the plasma in MPDR 325i should be very similar to the MPDR 325. § 7.2 Chemistry of dry etching for silicon Silicon can be etched by fluorine atom or chlorine atom based plasmas. The most widely used fluorine containing gases are CF4, SF6 and NF3. These gases generate free fluorine radicals which react with silicon to form volatile products. Since SF6 gas was used in the etching experiments in this dissertation, only its chemistry is described in this section. The chemistry of the SF6 with silicon in a plasma is a complicated subject. Table 7.1 shows the reaction set which includes all possible reactions taking place in the discharge [144]. The symbol, M, appeared in some equations represents a third body, such as a surface, required for the reactions to take place. Oxygen appears in some of reactions because it can be released from the quartz vessel or added deliberately. There are at least two different models to describe the dissociation of SF6. The first one is d'Agostino and Flamm model [142] which is summarized as below: SFx+e—>SFX,1+F+e (7.1) SF; + sr=6 —> 15+ SF“, + 31:6,“, (7.2) 209 TABLE 7.1 REACTION SET FOR SILICON ETCHING USING SF5 leade- heede- In alter m at 17m '1 sr.-'- sane 20 z 37,—!» sit.» 2: 1 SF.-4-o SF,+F 211 4 8&4. saw 2: s SF,—'-o SF+F 28 6 sr—'. 3+1: 2: '1 sr,+1=-£-‘-s1=. m 10'" '0 SF.+P-5o sri, 1.1x10‘" '9 saw—'1. SF. 11:: 10'" '10 sr,+1=-‘-'~ SF, 26x 10'“ 11 SH» i=-"-‘-o 5F, 2.91:10'“ 12 sir-=1. 39 2.0: 10'“ 13 sr,+sr,-o sane. 1.5:: 10'" 14 mar, .. 315+“. 2.5:: 10'“ is snsr-o use. 1.5:! 10'" '10 0,—Lo0+o 23 '11 saw-o sor.+1= 20x 10‘“ 10 saw-t sap.”I 20x 10'" ‘1! sumo-o 80H P 1.1111?" 20 SING-o 50+ F 1.7x10‘” '21 sauna-o so,+i= 7.9x10"' :2 sor,+o-so,r.+r son 10'" 13 son-L» sort,“I 21 a son-Lo sor=,+1I a as son—Loser" 20 a ear-Lesa” 21 u so—Los+o 23 a so.-'-oso+o 13 '19 earns—'1» sari. 5.01110’" '10 sor.+ 9.1. sari, :2: 10°“ '31 sour-‘1» 3015 1.0: 10’" n eon-Lear Half“ '31 so.+1=—“-so,r unit” 0» sonar-lo 30,3, 1.0:: 10'" J: sow-Leo, 1.41: 10'" 3 3+03‘ ”+0 11".-“ 11 ao+o,-oso,+o um'” 10 ' so+so-. so,+s 1.1x 1r” 19 no.4. so, ”1110'“ a Iii-r0.» 15w, 1.0: 10‘" 41 o+Po,-o Fo+o. 3.0: 10‘" a O+FO-o 9+0, ”1: If" 43 F.—'- F-o-F a «a his! to '40 3113+ r-o err, in 10‘" '06 IF,+F-o 81', 1x 10" ‘41 3115+ F-o SIP. in 10" 210 where x is a number between I to 6 inclusive and n equals to l or 2. These reactions dissociate SF6, producing lighter SFx radicals and fluorine atoms. Another model is suggested by Ryan and Plumb [144] from IBM. They have argued that the dissociation of SF6 through electron impact mainly gives SF2 and four fluorine atoms by the reaction listed as reaction (1) in Table 7.1. Sadeghi et al. [142] also has indicated that with the presence of a fluorine atom-consuming material such as silicon or tungsten, the concentrations of SF and SF2 molecules drastically increase. Two different reaction mechanisms between fluorine atoms and silicon have been presented. Figure 7.1 shows the reaction model suggested by Flamm. A layer of fluorinated silicon is formed on the surface of the silicon substrate. Fluorine atoms are adsorbed to form bonding with the silicon atoms on the surface. As described by Flamm, two reaction paths take place in phase I as shown in Figure 7.1. The reaction may either proceed in path a to form 8in molecule, which leaves the surface as the desorption product [137]. This accounts for about 5% to 30% of the material leaving the silicon surface. The Sin molecules then react in the gas phase with atomic fluorine or fluorine molecule to form SiF3 with the emission of spectrum peaked around 500 nm [74]. SiF3 then reacts with atomic fluorine to form the end product SiF4. According to Flamm, most of the reactions go through path b in phase I and then phase II and 111. As a result, most of the material leaving the silicon surface is SiF4. Another study by Winters and Plumb [135] indicates that the spontaneous etching at room temperature produces stable gases SiF4, Si2F6 211 fluorinatod skin F F by i F r F y '“\ ------- / si 81 / \ / \ k 11) F\ ./F _F__..F___.§!__F_ \ / i / Si si \ / \ F F l:\ \ t ., / SI F-Si _. --E. __. __L_E. n / l / ' si /SI\ / \ F F sir \ F-SI/ \3/ -— -——E—— —> -—————F—— m / si< ' / /SI\ Figure 7.1 Reactions of atomic fluorine with silicon [74] 212 and Si3Fg with SiF4 being the dominant product. Figure 7.2 shows the mass spectra of the products at various substrate temperatures. Winters and Plumb indicated that the SiF+ and SiF2+ peaks, shown in the mass spectrum analysis, should be assigned to Si2F6 and Si3F3 rather than Sin. However, as the temperature of the substrate increases from room temperature to 1000 0C, the relative contributions of the SiF4, Si2F6 and Si3F3 etch products decrease while concentration of 8in increases and eventually becomes the major product. Figure 7.3 shows the effect of ion bombardment in an ion assisted etch process. With the help of the argon ions, the etch process is significantly increased. § 7.3 Etching methods According to Flamm [74], there are four basic methods of plasma etching. As shown in Figure 7 .4, the first method is sputtering. Etching is achieved by energetic bombardment of ions on the substrate, kicking off the substrate material. The second method is chemical method. Reactive neutral species react spontaneously with the substrate. The product must be volatile. This process rarer takes place alone in plasma environment because of the formation of sheath. The third method is ion-enhanced energetic etching. In this case, ions are attracted towards the substrate surface through a sheath, striking the surface where reaction between reactive neutrals and the substrate take place. The ions enhance the etch rate through two possible mechanisms. Mass Spectrometer Intensity 213 SiF+ 3in" F-atoms -» Sil111) 7 n-type fl fl 2x 1 0'2 thm Figure 7.2 Mass spectra of etch products as a function of substrate temperature [135] Mass Spectrometer Intensity 214 SiF° Sin‘ 311:3: Si(111), p-type. IO'ZQ-crn Electron Energy ,= 70 eV 2x10‘6 F-atoms/cm2~s Current Density=13 pA/cm2 Axial Configuration i‘Ar=2.5pA . . $12F5 1 Sir; / 1 l \ i’A,=O Figure 7.3 Effect of ion bombardment on the mass spectra [135] 1. Sputtering 2. Chemical 3. Ion-enhanced energetic 4. Ion-enhanced protective 215 ION I NEUTRAL VOLATILE / PRODUCT o o v o NEUTRAL ION VOLATILE / PRODUCT lW NEUTRAL ION VOLATILE PRODUCT _ «n— INHIBITOR Figure 7.4 Etch methods [74] 216 They break the surface lattice, opening more reaction sites for the reaction to take place. Furthermore the ions kick off the byproducts or end-products of the etching reaction, thus enhancing the desorption rate. An example is SF6- argon plasma. Because the paths of the ions are perpendicular to the substrate surface, anisotropic etching can be achieved. The fourth method is ion-enhanced protective etching. In this process, a protective coating is formed on the surface of the substrate during the reaction. An example is CF4-argon plasma in silicon etching. After the dissociation of one or more fluorine atoms from CF4 molecules, the carbon skeletons link themselves together to form a polymer on the silicon surface. The polymer will prohibit the etching reaction to go on. In this case, the ions sputter the polymer away and allow the etch process to proceed. This type of reaction gives a very good vertical etch profile since the side wall is coated with the polymer, prohibiting lateral etching. However, the coating has to be removed. This adds additional step in the fabrication process and therefore this method is not desirable in manufacturing process. § 7.4 Etching system The etching system used for the submicron etching experiments is a customized Plasma Quest model 357-W etching and deposition system. Figure 7.5 shows the cross sectional view of the processing system. It consists of a WAVEMAT model MPDR 325i ECR plasma source, which has a twelve-inch inside diameter cavity and a discharge diameter of 24 cm. The MPDR 325i is similar to MPDR 325. It is equipped with 12 permanent magnets which form the twelve ECR magnetic cusps. 217 Quartz Discharge Chamber Microwave Input Probe ECR surface Sliding Short 00011091 Resonant Cavity Water Ma net Gas 9 Inlet ,. Wafer —> ‘— Load Look To Pumps 6 Vacuum Chamber Movable Chuck To Chiller <— Figure 7.5 Cross sectional view of the processing system 218 The vacuum system consists of a Leybold TMP 1000C turbomolecular pump, which has a pumping capacity of 1150 liters per second. The system is provided with a rotary vane pump, Stokes V-023-2, placed in line with the turbo pump. The turbo pump can be isolated from the main processing chamber by a VAT series 64 200-mm throttling gate valve. Chamber pressure is monitored by a MKS 390HA capacitance manometer and is controlled by a VAT PMS adaptive pressure controller. The base pressure of the water jacketed l6-inch diameter main chamber is 6x10-8 Torr. A single wafer load lock system is equipped in the system. Wafer size up to 6 inches in diameter can be processed. The circular chuck is made of aluminum and has an anodized outer layer of thickness between 2 to 5 mils. It can be moved to different downstream locations. The chuck temperature was maintained between 20 °C to 22 °C by a recirculating unit during all experiments. As indicated in Figure 7.5, all experiments described in this investigation were performed without downstream confinement magnets. Thus, it is expected that process uniformity can be further improved. A separate 13.56 MHz rf power supply with an automatic match network is connected to the chuck. Since electrons move much faster than the ions, a self-induced d.c. bias is established on the chuck. Figure 7.6 shows a schematic variation of the potential and current densities for ions and electrons during an rf cycle. The rf voltage, Vrf, is shifted downward by a self-induced voltage, VS. As a result, ion current, Ii, continuously reaches the substrate surface. The ion current is neutralized by a huge electron EOE: U OMS-Ox» Current Voltage 219 90 180 270 360 Phase Angle (deg) Figure 7.6 Variation of potential and current densities for rons and electrons during an rf cycle current, positive §7.5 P 7.5.1 I prepare alumini PFEpar resoluti The dc Similar Similar SPinnin baking) Seconds (IEIQmZI depoxit: prOCedU 220 current, je, which reaches the substrate surface when the rf voltage is positive. This happens at 90 degree phase angle in the Figure 7.6. § 7.5 Preparation of the experiments 7.5.1 Optical and electron-beam lithography Both optical and electron-beam lithographic techniques were used to prepare the submicron features before the inversion of the pattern using an aluminum liftoff procedure. Preparation of sample using optical lithography A Suss MJB 3 UV3OO mask aligner system, which has an optimal resolution of 0.4 micron in the vacuum contact mode, was used to pattern features sizes down to 0.5 micron using AZ5209 positive photoresist. The preparation procedures for the samples are shown in Figure 7.7. The cleaning procedures for the three-inch p-type silicon wafers are similar to those given by Reinhard [178]. Preparation of the pattern was similar to those in previous etching experiments [45]. The photoresist spinning for AZ5209 was 4000 rpm for 40 seconds. Soft bake (pre-exposure baking) temperature was 87 0C for 30 minutes and the UV exposure was 50 seconds. The wafer was developed in a solution consisting of one part deionized water and one part developer. About 400 nm thick aluminum was deposited by thermal evaporation at about 10'6 Torr. Aluminum liftoff procedure was performed by putting the sample wafer in a beaker containing 221 p-type <100> I I PR spining (A25209, PMMA) I I I I PMMA 80ft bake I I A25209 soft bake I E-beam exposure I I UV exposure I I wafer cleaning I l * ll— P.R. development I P.R. development I I Negative PR mask I Positive PR mask I l 1 Aluminum Liftoff I LAluminum Liftoff l _1 Positive Al mask Negative Al mask 0.1-0.5 11m 0.5 um l I AZ5209 hard bake 1 Second UV exposure I P.R. development I l I Negative PR mask 1 I Aluminum Liftoff 1 Positive Al mask 0.5 pm Figure 7.7 Preparation procedure of the etch sample 222 acetone which can dissolve both AZS209 and PMMA. An ultrasonic bath was used to assist the liftoff process. All samples were cut to a size of about 1 cm x 0.5 cm in order to be consistent with the samples prepared by the electron-beam lithography which were restricted to this size. The sample was then mounted with graphite paint on a 3-inch silicon carrier wafer for etching, except during the gas flow experiments and uniformity experiments in which a 6-inch silicon wafer was used. Preparation of sample using electron-beam lithography An electron-beam lithographic technique was used for preparation of feature sizes from 0.1 pm to 0.5 pm. The negative resist used was PMMA with a thickness of 250 nm after a spin of 40 seconds at 4000 rpm. The soft bake was 160 0C for 60 minutes. A JEOL ISM-840A scanning electron microscope with external beam control was used to write the pattern into the PMMA using a beam dosage of 250 nC/cm2 and accelerating voltage of 35 kV. The PMMA was developed in a solution consisting of one part of methyl isobutyl ketone and three parts of isopropyl alcohol for 60 seconds. A three hundred nanometer thick aluminum film was then deposited by thermal evaporation to define the final etching mask after aluminum liftoff. 7.5.2 Etching conditions Mixtures of SF6 and argon were used as the etching gas, which has been used in earlier ion assisted etching of silicon [24,25]. One advantage of 223 this etch gas composition is no polymer film will form on the substrate surface. The etching performance was investigated under different etching conditions. Gas flow varied from 16 to 32 sccm. Microwave input power was varied from 400 to 800 Watts. Processing pressure was varied from 0.6 mTorr to 9 mTorr, with no change in gas flow, by adjusting the throttle valve. The gas composition was 4% SF6 which gave highly anisotropic etch profiles as indicated in a previously etching experiment which used d.c. bias [45]. In this study, bias was via rf, with rf power varied from 5 to 50 Watts. An investigation of the etch rate at different downstream locations was also conducted. § 7.6 Etching results Etch depths were measured by Dektak II profilometer after removing the aluminum mask. Microloading and etch profiles of line width less than 500 nm were determined by scanning electron microscopy. Both JEOL T- 330 and JEOL JS M-84OA SEM machines have been used. 7.6.1 Role of gas flow The variations of the etch rate versus the gas flow are shown in Figure 7.8 for microwave input power of 500 and 800 Watts. The sample was mounted on a 6-inch wafer. Gas composition was fixed at 4% SF6 and 96% argon. The rf power was 10 Watts which produced an negative induced d.c. bias of 10 volts. By adjusting the throttle valve, the pressure was kept constant at l mTorr while the flow was increased from 16 sccm to 32 sccm. The curves in Figure 7.8 showed that as the flow rate was doubled (from 16 260 ) N .p. O N N O 1111111111111111r11l11111114111111111111111111111 Etch rate (A per min N O O 180 160 224 800 watts 500 watts 1 IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIITIIIIIII 0 15 20 25 30 35 Total Flow Rate (sccm) Figure 7.8 Variation of etch rate with flow rate at l mTorr 10 cm downstrem, 4%SF6 and -10 volts bias (10 rf watts) Eli-ii 225 to 32 seem), the etch rate increased by about 38%. Increasing the microwave power from 500 Watts to 800 Watts only increases the etch rate by less than 1%. The etching process was operating under a condition in which fluorine atoms, produced from dissociation of SF6, are limiting the etch rate. For example, a 16 sccm total flow with 4% SF6, will produce 1.72x1019 fluorine atoms per minute assuming each SF6 molecule coming into the system is perfectly dissociated and liberates six fluorine atoms. If all fluorine atoms participate in etching and producing SiF4, these will result in an ideal etch rate of 291 Almin. The actual etch rate was 182 Almin. Taking into account that dissociation of each molecule is less than 100%, the etch rate seems to be limited by the supply of fluorine atoms. As the flow increases, more SF6 molecules are introduced into the plasma. Therefore more fluorine atoms are available for etching as a result of SF6 dissociation. The etch rate thus increases. The slight increase in etch rate as the microwave power increased can be attributed to a slight increase in dissociation of the remaining SF6 molecules or associated radicals. Another possible reason is an increase in substrate temperature which increases the reaction rate. 7.6.2 Effect of rf substrate bias The rf source which is connected to the chuck is autotuned by a match network. The source can be operated in two different modes: a constant rf power mode or a constant bias (self-induced bias) mode. The relation of the rf power and the induced bias is shown in Figure 7.9. Figure 7.9(a) shows the variations of the induced bias versus pressure and power at 10 cm downstream. The discharge consisted of 5% SF6 and 95% argon. As the rf I. bias (volts) d.c. 226 10.0 i 3 O 0.6 mTorr -l 1 0.0 1 a 1.0 mTorr 1 - A 3.0 mTorr : \ 100: 500watts ' : ............. 700watts -20.03 : R -30.0-1 \ 2 \ i \ : \a -4000 _ IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIITTIIIIIIIIIIIIIIIIIIIII 0.0 10.0 20.0 30.0 40.0 50.0 60.0 RF power (watts) Figure 7.9 (a) Variation of self-induced bias with rf power with 5% SF6 at 10 cm downstream ur- 227 20.0 . 5 500watts 1-4 E ............ 700 watts 10.0 E 97 ‘ ~ 9 ‘ .1 \ 7 A E m 0.0 : z _ O 2 > 2 v .i — J g 10.0 2 '5 i d —20.05 '6 E O 0.6 mTorr 5 E] 1.0 mTorr -300 g A 2.0 mTorr 3 0 3.0 mTorr -4000 ‘ TITImIIITrIrIIIITIIIfiTITIIIITIIIIWI]IIIIIIIIIIIIIIITIII 0.0 10.0 20.0 30.0 40.0 50.0 60.0 RF power (watts) Figure 7.9 (b) Variation of self-induced bias with rf power with 5% SF6 at 4.5cm downstream 228 power increases, the induced bias becomes more negative. For a fixed rf power, an increase in pressure will also increase the bias. At 0.6 mTorr an increase in microwave power from 500 Watts to 700 Watts reduced the induced bias. This is because due to an increase in plasma density. The relation of rf power and induced bias also depends on downstream position as shown in Figure 7.9(b). At 4.5 cm downstream the induced bias voltage is no longer negative at low rf power. A positive induced bias was observed. The appearance of the positive induced bias is due to a change of the ratio of "effective" conductive area between the chuck and the plasma source/chamber. The addition of the rf changes the plasma such that the induced voltage drops as the rf power increases. Another observation is that at constant 500 Watts microwave power, an increase in pressure from 0.6 mTorr to 1 mTorr causes the induced bias to become less negative. Compared with Figure 7.9(a), at 10 cm downstream, the induced bias becomes more negative as pressure increases from 0.6 to 1.0 mTorr. This shows that the plasma density at 1.0 mTorr is higher than that at 0.6 mTorr in the source region. However, the plasma density is lower at 1.0 mTorr than at 0.6 mTorr in lower downstream position. Thus it indicates that the decrease in mean free path between collision due to high pressure results in a larger gradient of plasma density along the z-axis. This suggested that at high pressure, the wafer needs to be placed closer to the source in order to obtain a higher processing rate. For a given pressure, the higher the bias, the more perpendicular the ions hit the surface. This translates to a better etch profile. However, a high bias can also be deleterious because damage to the etch mask will become 229 ISKU x1a,aaa (a) a damaged mask at 9 mTorr, 5% SF6, -110 volts (50 watts rf) 15KU X35,666 B-Sl-‘Lm 268843 (b) undamage mask at 0.6 mTorr, 4% SF6, -10 volts bias (11 watts rf) Figure 710 Effect of bias an etch mask. (500 watts microwave power, 10 cm downstream) 230 an issue. Figure 7.10 shows two SEM pictures illustrating the effect on the etch mask with increasing bias. At 9 mTorr, a bias of -110 volts was needed to maintain anisotropic etch profile. As shown in Figure 7.10(a), the aluminum mask on the top of the 2 11m features was damaged by the high bias. On the other hand, when the operating pressure was 0.6 mTorr, a bias of ~10 volts was sufficient to achieve a vertical etch profile without any damage to the aluminum mask as shown in Figure 7.10(b). The feature shown was 0.5 pm line with 1 pm spacing. Compared with the 20 volts d.c. bias used in earlier work [45], a smaller rf bias voltage was sufficient to achieve a vertical etch profile. 7.6.3 Variations of etch rate with other processing parameters Variations of etch rate with respect to the bias and the downstream positions are shown in Figure 7.11. To examine the effect of downstream position an etch rate, the pressure, microwave power, substrate bias and gas composition were fixed at 0.6 mTorr, 500 Watts, -10.3 volts and 4% SF6, respectively. As shown in curve (a) in Figure 7.11, there was a steady decrease in etch rate as the substrate was moved further away from the source. It is expected that if downstream confinement magnets were used, the etch rate would be higher and the rate of decrease would be smaller. The lowest etch rate was about 238 Almin when the substrate was at 17 cm downstream. As shown in the curve (b) in Figure 7.11, the etch rate went up as the bias potential was decreased from -5 to -45 volts. In this case, the substrate was placed at 10 cm downstream while the rf bias voltage was varied. Etch rate (A per min) 231 Downstream Position (cm) O 5 10 15 20 320- llllJllllIlLlllllllIllJllLLLlIllllllllJ E \ E 30053“ \ 280-5 3 260-: 2405 2205 200-IIIIIIIIIIIIIIIIIIIIIIIIlIIIIIITIIIIIIIIIIIIIIIII 0 10 20 :50 40 50 Bias (negative volts) Figure 7.11 Variations of etch rate with (a) downstream position with -10 volts bias (11 watts rf) and (b) bias at 10 cm downstream position (940 watts rf). Etching conditions were 4% SF6, 500 watts microwave power at 0.6 mTorr 232 Variations of etch rate with the microwave power are shown in Figure 7.12. In this case, the etch sample was mounted on a 3-inch wafer carrier. The pressure was 0.6 mTorr and the rf bias was -10 volts. The increase of the etch rate almost saturated as the power was increased from 400 Watts to 800 Watts. This rate of increase in etch rate was slower than the associated increase in ion density [47]. As discussed earlier, the etch rate was limited by the amount of fluorine atoms, not the ions. Thus the slight increase in etch rate was primarily the result of higher dissociation rate of the SF6 molecules or associate radicals. As the pressure increased, the etch rate increased dramatically as shown in Figure 7.13. The gas flow rate was kept constant at 20 sccm (5% SF6 and 95% argon). A constant 50 Watts rf power was used to maintain a highly anisotropic etched profile. Although Figure 7.3 indicates that ion bombardment significantly changes the chemistry on the substrate surface, the high etch rate at 9 mTorr can not only be the result of a high ion flux. In addition to a possible higher dissociation rate of the SF6 molecules, it is suspected that the increase in residence time of the fluorine atoms in the chamber from 24 ms at 0.6 mTorr to 355 ms at 9 mTorr may also play a role in the enhancement of the etching process. 7.6.4 Etch uniformity Previous results using a similar 24 cm discharge reactor and d.c. bias have shown a 2% (lo) variation over a 6-inch diameter with downstream confinement magnets at 18 cm downstream [45]. This study was performed, without downstream plasma confinement, on a 6-inch silicon wafer on which 233 320 Etch rate (A 240 1111111111111111111l1111111111111111111111111111 220 IIIIIIIIIIITIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII 300 400 500 600 700 800 900 Power (watts) Figure 7.12 Variation of etch rate with microwave power at 0.6 mTorr, 4% SF6, 10 cm downstream and -10 volts bias (11 watts rf) Etch rate (A per min) 1 200 1100 1 000 (O O O 800 700 600 234 11]]lllllllllllllllllllllllllllllllllllllllllllllljljlllljl IIITIPTIIIITIIIIIIIIrIlIIIIIIIITIIIIIIIIIIIIIIIII o 2 4 6 8 10 Pressure (mTorr) Figure 7.13 Variation of etch rate with pressure in a 5%SF6 plasma at constant 50 Watts rf and 500 watts microwave power Yam-.m-ffih...‘ _" Y 1 235 16 samples were mounted in a cross manner using graphite paint as shown in Figure 7.l4(b). A uniformity of 4% (10) over 150 mm diameter was obtained at 10 cm downstream position using 500 Watts microwave power. Four percentages of SF6 and 96% argon gas mixture was used and the pressure was 0.6 mTorr. A single, patterned 3-inch wafer was also etched as shown in Figure 7 .l4(a), under the same condition and a uniformity of 1.4% (10) was obtained. 7.6.5 Loading and Microloading effect The depletion of the etching species from the gas passing through the wafers is known as the loading effect. Table 7.2 displays the loading effect observed in the etching experiments. When a 6-inch wafer was used to hold a small sample instead of using a 3-inch wafer, the etching rate dropped from 273 Almin to 200 A/min. A separate experiment was also conducted. A patterned 3-inch silicon wafer, with about 30% coverage of aluminum mask, was etched and the etch rate was found to be essentially the same as the etch rate of a single small patterned sample which was placed on a blank 3-inch wafer. This is because the probability of fluorine atom loss in silicon is approximately the same in aluminum. Thus the loading effect, in this case, is related to the size of the wafer rather the percentage of mask coverage on the wafer. A mathematical model was given[74] as: Rm/RO = 1/( 1+ (mAwKw)/(ASKS)) (7.3) where Rm is the etch rate, RO is a constant under a fixed operating condition, in is the number of wafers, AW is the wafer area, KW is the loss factor of 236 TABLE 7.2 ETCHING RESULTS SHOWING LOADING EFFECT Experimental conditions: 500 watts microwave power, -10 volts, 4% SF6, 10 cm downstream at 0.6 mTorr Diameter Area Etch rate Number of sample 3 inches 45.6 cm2 273 A/min 1 6 inches 182.4 cm2 200 A/min 16 3 inches 45.6 cm2 285 A/min (patterned wafer) 237 1.4%(1 o) 4% (1 a) (b) Figure 7.14 Etching experiment for uniformity measurement 238 fluorine from etching silicon, AS is the area of the vessel wall which confines the plasma, and KS is a combined loss factor accounting for the volume recombination and wall loss of fluorine atoms. Figure 7.15 shows the variations of Rm/R0 versus substrate area, Aw. The geometry of the plasma is assumed to be cylindrical with a radius, R, and height, L. The coefficients, KS and Kw, are given as 0.0017 and 1.5 x 10'4 respectively [74]. The experimental results as shown in Table 7.2, in general, follow the trend predicted by equation (7.3). As the wafer size increases, the etch rate decreases. However, it is necessary to understand that this model may only apply to a flow limited conditions in which the amount of etch material exposed to the plasma will affect the dynamics of the chemical reaction. Microloading is here defined as the variation of etch rate due to packing density of the features. It is an important issue when fabricating features of different sizes on a single wafer. Figure 7.16 shows the etch profile of two different features on the sample. As shown, the etch depth in the 500 nm line/500 nm spacing was the same as the 200 nm line/300 nm spacing. The result of another experiment which was run under exactly the same condition as in Figure 7.16, is shown in Figure 7.17. No microloading was observed in the 500 nm line/200 nm spacing, 100-150 nm line/200 nm spacing and the 100-150 nm line/500 nm spacing patterns. All SEM pictures show the submicron features with the intact 100 nm thick aluminum mask. Highly anisotropic (900) etching was demonstrated. The etch condition was 500 Watts microwave power, -10 volts bias and 4% SF6 at 0.6 mTorr located at 10 cm downstream. The total gas flow was 20 sccm. 239 R=5 inches, L=l to 10 inches 0.9.... \ ......' ........ ........ ........ ....... . M- ...... ....‘.....=, ........ ........ g ......... ....... . 0.7_ ....... ....... .. .. .. .. ......... §L=11ilini:hes---‘ oooooooooooooooooooooooooooooooooo Rm/RU O 0‘ I 0.5... ....... . ....... ...... .'. ....... ... 0.4- ....... g ......... g ........ ........ g ........ g ........ 0.3 _ ....... ......... ........ ........ ......... -=1-‘tncii ........ ..... g ....... . 0, ; 1 i i 1 ; g i () 20 40 60 80 101 l 120 140 1 60 l 80 Substrate area (square cm) Figure 7.15 Loading effect according to equation (7.3) 240' flamers .‘ .131 15KU $458,888 8.514": 188882 (a) 500 nm line/500 nm spacing ....Emm atmwam. 1 5 K U X S 8 , 8 B 8 (b) 200 nm line/300 nm spacing Figure 7.16 Anisotropic etching of silicon 111013 (a) 500 nm line/200 nm spacing 1531 . t. r- .ahh 1595i HDIE (b) 100-150 nm line/200 nm spacing Figure 7.17 Anisotropic etching of silicon 242 11013 (c) 100-150 nm line/500 nm spacing Figure 7.17 Anisotropic etching of silicon 243 § 7.7 Summary Highly anisotropic etching (z 90°) of submicron features from 500 nm to 100 nm on p-type <100> silicon has been achieved. The etch rate varied from 200 Almin to 1120 A/min depending on etching pressure. Etching was done under flow limiting conditions, and was limited by the supply of fluorine atoms. Higher flow rate would allow higher etch rate. At high pressure, a high induced bias is needed to achieve anisotropic etch profile. In that case, a stronger mask such as nickel mask or other hardened resist has to be used other than aluminum mask. The induced substrate bias voltage is also a function of the chuck position. A positive bias voltage was obtained when the chuck was 4.5 cm downstream. A negative bias voltage can be obtained either by increasing the rf power or increasing the ground surface near the baseplate. Etch rate increased dramatically with processing pressure, but increased slightly with rf bias and microwave power. The increases of etch rate with the flow rate indicates a higher pumping speed is needed to increase the etch rate when the operation pressure is decreased. As the substrate moved from 5 cm downstream to 17 cm downstream, the etch rate dropped from about 290 A/min to 230 A/min at 0.6 mTorr. Etch rate can be increased by placing the substrate closer to the source or change the gas composition. Uniformity of 4% (16) over 150 mm diameter and 1.4% (16) over 75 mm diameter without downstream confinement magnets were obtained. Loading effect was observed when the wafer size changed from 3- 244 inch to 6-inch in diameter. However, no microloading was observed in the experiments. CHAPTER EIGHT W § 8.1 Summary of research § 8.2 Future research 8.2.1 Plasma source development 8.2.2 Plasma etching 245 246 § 8.1 Summary of research A large diameter electron cyclotron resonant microwave plasma disk reactor, the MPDR 20, was designed, built and characterized as part of this Ph.D. thesis research. The source has a 45.7 cm diameter cavity and a 20 cm diameter discharge. It was the first MPDR operated at 915 MHz. At 500 watts, the source produced a plasma density of 1011 cm'3 at zero downstream, which was ten times of the critical density at 915 MHz. At 13 cm downstream, a 2.6% (10') uniformity over 18 cm in diameter was obtained. This highly uniform plasma is ideal for large diameter wafer processing. The plasma potential was low, between 7 to 10 volts. Measurements of ion energy distribution function revealed a small FWHM of 8 eV with average ion energies in the range of 8-10 eV. The plasma source was operated with TE211 excitation mode at 915 MHz. Initial experiments of operating the MPDR 20 at 2.45 GHz was also successful which lead to the construction of the MPDR 325 which is now sold as a commercial plasma processing source. Characterization of the MPDR 325 was also investigated in this Ph.D. thesis. The MPDR 325 was the first large diameter commercial plasma source operated at 2.45 GHz microwave frequency. Multimode excitation was used to create the plasma. It has a 35.56 cm diameter cavity and a 24 cm diameter discharge. Unlike the other smaller MPDRs, the MPDR 325 has an opening which is larger than the cutoff radii of some cavity resonant modes. Plasma can be ignited at pressure at several mTorr while the other MPDRs have to ignited at pressure about 100 mTorr. The Experimental result showed that the source has all the advantages which the MPDR 20 247 processes. The plasma density was high at downstream position, about 6x1010 cm'3 at 14 cm downstream. A 3.1% (10) variation of plasma distribution over 18 cm at 14 cm downstream was obtained. Plasma potential was low about 10 volts at 14 cm downstream. The average ion energies were about 10 to 14 eV. The ion energy distribution functions had small FWHM of about 7 to 9 eV. Unlike the MPDR 20, this large diameter source is capable of operating at microwave up to 1500 watts. The experimental performance of several different diameter MPDRs was also investigated to understand the difference in operation versus the size of the plasma source. Results showed that the MPDR 5, which has a 5 cm discharge, produced the highest density at zero downstream. The plasma potentials were high, about 23 volts at zero downstream and 20 volts at 5 cm downstream. The most interesting result in the MPDR 5 was its bimodal distribution in the ion energy distribution function, which was believed to be caused by resonant charge exchange. Four figures of merit were used to compare the performance of the MPDRs. They were the electron temperature versus the product of pressure-diffusion length, microwave coupling efficiency, ion production energy cost and power density. The coupling efficiencies of all MPDRs, which were excited by single cavity resonant modes, varied from 75% to 95%. Within a pressure between 1.0- 1.2 mTorr and power between 200 to 300 watts, the MPDR 9 had the highest ion production energy cost, 640 eV per ion, while the MPDR 325 was the most efficient in the ion production with 220 eV per ion. In terms of power density, the MPDR 5 required a minimum of 2 watts/cm3 to sustain the plasma while the MPDR 20 and MPDR 325 required only 0.06 watts/cm3 and 0.1 watts/cm3 respectively. 248 Submicron application of the large diameter MDPR was performed in a new etching and deposition system. Feature sizes varied from 0.5 pm down to 0.1 pm were patterned on a p-type <100> silicon wafer both by optical and electron-beam lithography. An aluminum mask was used as the etch mask. Optimal gas composition for anisotropic etching was 4% SF6 in argon plasma. The pressure was 0.6 mTorr with a total flow of 20 sccm. Five hundred watts of microwave power at 2.45 GHz was used and the rf induced bias voltage was -10 volts. The etch rate varied from 200 Almin to 1120 A/min depending on discharge pressure. The etch process was limited by the supply of fluorine atoms. Higher pumping speeds would result higher etch rate. Etch unifonnities were 4% (16) over 150 mm diameter and 1.4% (10') over 75 mm diameter without downstream confinement magnets. § 8.2 Future research 8.2.1 Plasma source development Development of microwave plasma disk reactor can be divided into following categories: (A) Structure of the MPDR Magnetic configuration is the subject of interest in this category. As shown in Chapter Four, the idea magnetic configuration is one which matches with the transverse electric field of the cavity resonant mode. Study of variation of plasma properties versus magnetic configuration, in fact, has started. Initial results showed that there were improvements both in plasma 249 density and ion energy distribution. The work can be applied to the large MPDR 20 which was operated at TB 211 made with 915 MHz excitation. The objective is thus to determine the optimal excitation mode and its magnetic configuration. (B) Automation of the MPDR All MPDRs can be operated by computer since all operation positions are repeatable. The task is to build a robust real-time program which can start the discharge, tune the cavity and monitor the process. (C) Scale up of the MPDR 20 The performance of the MPDR 325 is better than the smaller MPDRs in terms of operation and ion production efficiency. If a higher power 915 MHz microwave source is available, the MPDR 20 can be scaled up to a larger reactor with discharge diameter larger than 34 cm. The immediate concern in this scale-up to large diameter is the cost of the components such as the quartz discharge chamber. A high power 915 MHz microwave source, which has a maximum of at least one thousand watts is needed for the operation of the scale-up source. Current commerical unit has a maximum of 500 watts. 8.2.2 Plasma etching Etch rate is currently limited by the pumping speed. Without changing the hardware, the etch rate can be increased by using other etch 250 gas such as Hydrogen Bromide. Another method is to move the chuck upward, i.e. closer to the plasma source, where the plasma density is higher. Anisotropic etching of silicon is only one of the fabrication processes [178]. The etching of silicon nitride is also an important process. Recently Tungsten has become an important material for metal contact in ULSI. It is because voids are less likely formed in via holes [179] than the aluminum. Both silicon nitride and tungsten can be etched by the SF6-argon gas. All the end products, SiF4, NF3 and WF6 have low boiling point [180] and therefore are volatile in vacuum environment. Finally, process control is equally important in plasma processing. The end point detection is a crucial technique and yet has not been studied in the MPDR sources. LIST OF REFERENCES o ...“ q T_ ‘ I. ll- [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] 251 LIST OF REFERENCES N. Sakudo, K. Tokiguchi, H. Koike and I. Kanomata, Rev. Sci. Instrum., Vol. 48, No. 7, 762 (1977). K. Suzuki, S. Okudaira, N. Sakudo and I. Kanomata, Japan. J. Appl. Phys., Vol. 16, No. 11, 1979 (1977). K. Suzuki, S. Okudaira and I. Kanomata, J. Electrochem. Soc., Vol. 126, No. 6, 126 (1979). K. Ninomiya, K. Suzuki and S. Nishimatsu, Japan. J. Appl. Phys., Vol. 22, No. 4, 139 (1983). S. Matsuo and M. Kiuchi, Japan. J. Appl. Phys., Vol. 22, No. 4, L210 (1983). K. Suzuki, K. Ninomiya, S. Nishimatsu and S. Okudaira, J. Vac. Sci. Technol., B3 (4), 1025 (1985). K. Tokiguchi, N. Sakudo and H. Koike, Rev. Sci. Instrum., 57 (8), 1526 (1986). M. Matsuoka and K. Ono, Appl. Phys. Lett. 50 (26), 1864 (1987). K. Asakawa, N. Takadoh, M. Uchida and T. Yuasa, NEC Res. & Develop., 87, l (1987). S. Sugito, S. Ishida and Y. Iida, NEC Res. & Develop., 92, 18 (1989). K. Shirai, T. Iizuka and S. Gonda, Japan. J. Appl. Phys., Vol. 28, No. 5, 897 (1989). M. Matsuoka and K. Ono, Japan. J. Appl. Phys., Vol. 28, No. 3, L503 (1989). [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] 252 K. T omotsune, Y. Yamamoto, M. Morita, K. Suzuki, Y. Shirnizu, K. Fuji, N. Takadoh, K. Asakawa and K. Okada, NEC Res. & Develop., 95, 15 (1989). H. Nihei, J. Morikawa and N. Inoue, Japan. J. Appl. Phys., Vol. 29, No.5, L822 (1990). K. Kamata, T Inoue, K Maruyama and I. Tanabe, Japan. J. Appl. Phys., Vol. 29, No.7, L1203 (1990). ...... S. Iizuka and N. Sato, J. Appl. Phys., 70(8), 4165 (1991). H. Abe, NEC Res. & Develop., Vol. 32, No. 2, 179 (1991). E. Ghanbari, I. Trigor and T. Nguyen, J. Vac. Sci. Technol. A7(3), 918 (1989). Y.H. Lee, J.E. Heidenreich III and G. Fortuno, J. Vac. Sci. Technol. A7(3), 903 (1989). O.A. Popov and H. Waldron, J. Vac. Sci. Technol. A7(3), 914 (1989). O.A. Popov, J. Vac. Sci. Technol. A7(3), 894 (1989). J. Forster and W. Holber, J. Vac. Sci. Technol. A7(3), 899 (1989). DA. Carl, M.C. Wiliamson, M.A. Lieberman and AJ. Lichtenberg, J. Vac. Sci. Technol. B9(2), 339 (1991) RR. Burke and C. Pomot, Solid State Tehnol., 67, Feb., 1988. J. Pelletier, Y. Amal and L. Vallier, Rev. Sci. Instrum., 59(7), 1072 (1988). M. Geisler, J. Kieser, E. R-uchle and R. Wilhelm, J. Vac. Sci. Technol. A8(2), 908 (1990). K.H. Kretschmer, K. Matl, F. Lorenz, I. Kessler and B. Dumbacher, Solid State Tehnol., 53, Feb., 1990. G. Neumann, J. Vac. Sci. T echnol., B9(2), 334 (1991). R.W. Boswell and D. Henry, Appl. Phys. Lett., 47 (10), 1095 (1985). 253 [30] AJ. Perry and R.W. Boswell, Appl. Phys. Lett. 55 (2), 148 (1989). [31] AJ. Perry, D. Vender and R.W. Boswell, J. Vac. Sci. Technol., B9(2), 310 (1991). [32a] J. Root and J. Asmussen, "MSU microwave ion source", presented at the MSU/NASA Workshop on Advanced Propulsion Concepts using Time Varying Electromagnetic Fields, East Lansing, Michigan, Feb (1982). [32b] J. Asmussen and J. Root, Appl. Phys. Lett., 44(4), 396 (1984). [320] J. Asmussen, J. Root, "Ion generation apparatus and method for the use thereof", US patent No. 4507588, March (1985). [33] J. Root and J. Asmussen, Rev. Sci. Instrum., 56(8), 1511 (1985). [34] T. Roppel, D.K. Reinhard and J. Asmussen, J. Vac. Sci. Technol., B4(1), 295 (1986). [35a] M. Dahimene and J. Asmussen, 19th EPIB conference, (1984). [35b] M. Dahimene and J. Asmussen, J. Vac. Sci. Technol., B4(1), 126 (1986). [36] J. Asmussen and M. Dahimene, J. Vac. Sci. Technol., 85(1), 328 (1987). [37] L. Mahoney, M. Dahimene and J. Asmussen, Rev. Sci. Instrum., 59(3), 448 (1988). [38] J. Hopwood, D.K. Reinhard and J. Asmussen, J. Vac. Sci. Technol., 36(6), 1896 (1988). [39] J. Asmussen, J. Vac. Sci Technol., A7, 883 ( 1989). [40] J. Asmussen, R. Fritz, L. Mahoney, G. Foumier and G. Demaggio, Rev. Sci. Instrum., 282 (1990). [41] L. Mahoney and J. Asmussen, Rev. Sci. Instrum., 61(1), 285 (1990). [42] [43] [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] 254 J. Asmussen, J. Hopwood and RC. Sze, Rev. Sci. Instrum., 61(1), 250 (1990). EC. Sze, D.K. Reinhard, B. Musson, J. Asmussen and M. Dahimene J. Vac. Sci. Technol., B8(6), 1759 (1990). J. Hopwood and J. Asmussen, Appl. Phys. Lett., 58(22), 2473 (1991). B. Musson, F.C. Sze, D.K. Reinhard and J. Asmussen, J. Vac. Sci. Technol., B9(6), 3521 (1991). G. King, F.C. Sze, P. Mak, T.A. Grotjohn and J. Asmussen, J. Vac. Sci. Technol. A 10(4), 1265 (1992). P. Mak, G. King, T.A. Grotjohn and J. Asmussen, J. Vac. Sci. Technol. A 10(4), 1281 (1992). AK. Srivastava, M. Dahimene, T.A. Grotjohn and J. Asmussen (to be published). J. Hopwood et al., IBM, (seminar presented in Michigan State Univeristy). Y. Sakamoto and Y. Ishibe, Japan. J. Appl. Phys., Vol. 19, No. 3, 839 (1980). K. Kamata, T. Inoue, K. Maruyama and I. T anabe, Japan. J. Appl. Phys., Vol. 29, No. 7, L1203 (1990). S. Sarnujkawa, S. Mori and M. Sasaki, Japan. J. Appl. Phys., Vol. 29, No. 4, 792 (1990). S. Tachi, K. Tsujimoto and S. Okudaira, Appl. Phys. Lett., 52(8), 616 (1988). DE. Post and R. Behrisch, W anmlledfiusion, NATO ASI series, Plenum Press, New York, p17, 1988. [551 [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [661 [67] 255 W. Lochte - Holtgreven, Plasma Diagnostigs, North - Holland, Amsterdam, 705-725. E.O. Johnson and L. Malter, Phys. Rev., Vol. 80(1), 58 (1950). MJ. Druyvesteyn, Z. Phys., 64, 781 (1930). H.W. Rundle, D.R. Clark and J.M. Deckers, Can. J. Phys., Vol. 51, 144 (1973). LE. Heidenreich III and J .R. Paraszczak, J. Vac. Sci. Technol., B6(l), 288 (1988). J. Hopwood, Macroscopic pmpeflias Qf a malfipalar alagtran o 011‘ or..- ‘11ro .v- .1Vi oltm. 0. ‘ o 11.0.” siliggn atahing, Ph. D. Thesis, Dept. of Electrical Engineering, Michigan State University (1990). L. Mahoney, 11- var. i. :11 .. om... rn l o W MS. Thesis, Dept of Electrical Engineering, Michigan State University (1989). RF. Chen, W, R.H. Huddlestone and S.L. Leonard, editors, Academic Press, New York, 113 (1965). HR. Kaufman and RS. Robinson, WW, Common-wealth Scientific Corporation, Virginia, (1987). RR McKelvey, W, Robert E. Krieger Publishing Co. Inc., Florida (1986). D. Rapp and P. Englander-Golden, J. Chem. Phys., 43, 1464 (1965). R. F. Harrington, Wigs, McGraw- Hill (1961). A. Theodore Forrester, W Wan, John Wiley & Sons, 10 (1988). [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] [81] 256 BE. Cherrington, Gaseous electronics and gas lasers, Pergamon Press, New York (1979). J. L. Delcroix, WW, translated by M. Clark, D. J. BenDaniel and J. M. BenDaniel, Interscience publishers, Inc., New York (1964). J.A. Bittencourt, WW, Pergamon Press, (1988). P. C. Stangeby, ’1 ' o ’1... - 1 ..- ‘ .._ 'n _-1 11 run. Frisian, D. E. Post and R. Behrisch, editors, Plenum Press, New York, 41 (1988). B, M. Smimov, MW, translated by Oleg Glebov, Mir Publishers, Moscow (1981). D. R. Nicholson, W. John Wiley and Sons, 1983. DM. Manos and D. L. Flamm, WW Academic Press (1989). LC. Shen and J.A. Kong, MW, Brooks/Cole Engineering Division (1983). M. Sadowski, Rev. Sci. Instrrun., Dec., 1545 (1969). L. Pomathiod, R. Debrie, Y. Amal and J. Pelletier, Phys. Lett., Dec., 301 (1984). R. Lirnpaecher and KR. MacKenzie, Rev. Sci. Instrum., vol. 44, 6, June, 726 (1973). Seitaro Matsuo, Yumiko Takehara and Akira Ozawa, Japn. J. Appl. Phys., Vol. 17, No. 11, 2071 (1978). Kiyokatsu Jinno, Japan. J. Appl. Phys., Vol. 17, No. 7, 1281 (1978). Seitaro Matsuo and Yoshio Adachi, Japan. J. Appl. Phys., Vol. 21, No. 1, L4 (1982). [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] 257 D. L. Flamm, V. M. Donnelly and DE. Ibbotson, J. Vac. Sci. Technol. B 1(1), 23 (1983). HF. Winters, J.W. Coburn and T.J. Chuang, J. Vac. Sci. Technol. B 1(2), 469 (1983). Y. Amal, J. Pelletier, C. Pomot, B. Petit, and A. Durandet, Appl. Phys. Lett., 45 (2), 132 (1984). C. Pomot, B. Mahi, B. Petit, Y. Amal and J. Pelletier, J. Vac. Sci. I1 Technol. B 4(1), 1 (1986). I N.N. Efremow, M.W. Geis, R.W. Mountain, GA. Lincoln, J.N. Randall and NP. Economou, J. Vac. Sci. Technol. B 4(1), 337 (1986). Susumu Asada and Katsumi Mari, J. Vac. Sci. Technol. B 4(1), 318 (1986). J. Sugiura, W.J. Lu, K.C. Cadien and A.J. Steckl, J. Vac. Sci. Technol. B 4(1), 349 (1986). Yukinori Ochiai, Kazuhiko Shihoyama, Takao Shiokawa, Koichi Toyoda, Akio Masuyama, Kenji Gama and Susumu Namba, J. Vac. Sci. Technol. B 4(1), 333 (1986). Gregory W. Grynkewich, T .H. Fedynyshyn and RH. Dumas, J. Vac. Sci. Technol. B 8(1), 5 (1990). Wei-Xi Chen, L.M. Walpita, C.C. Sun and W.S.C. Ghang, J. Vac. Sci. Technol. B 4(3), 701 (1986). Toshiro Ono, Masatoshi Oda, Chiharu Takahashi and Seitaro Matsuo, J. Vac. Sci. Technol. B 4(3), 696 (1986). B. Mahi, Y. Amal, C. Pomot, J. Vac. Sci. Technol. B 5(3), 657 (1987). N. Vodjdan and P. Parrens. J. Vac. Sci. Technol. B 5(6), 1591 (1987). [95] [96] [97] [98] [99] [100] [101] [102] [103] [104] [105] 258 J. Hopwood, M. Dahimene, D.K. Reinhard and J. Asmussen, J. Vac. Sci. Technol. B 6(1), 268 (1988). Y. Tobinaga, N. Hayashi, H. Araki, S. Nakayama and H. Kudoh, J. Vac. Sci. Technol. B 6(1), 272 (1988). SW. Pang, M.W. Geis, W.D. Goodhue, N.N. Efremow, D.J. Ehrlich, R.B. Goodman and J.N. Randall, J. Vac. Sci. Technol. B 6(1), 249 (1988). Ahmed M. El-Masry, F-O. Fang, J .C. Wolfe and J.N. Randall, J. Vac. Sci. Technol. B 6(1), 257 (1988). SW. Pang, W.D. Goodhue, T.M. Lyszczarz, ' DJ. Ehrilich, R.B. Goodman and GD. Johnson, J. Vac. Sci. Technol. B 6(6), 1918 (1988). R. Cheung, Y.H. Lee, K.Y. Lee, T.P. Smith, HI, D.P. Kern, S.P. Beaumont and C.D.W. Wikinson, J. Vac. Sci. Technol. B 7(6), 1462 (1989). T.R. Hayes, M.A. Dreisbach, P.M. Thomas, W.C. Dautremont-Smith and LA. Heirnbrook, J. Vac. Sci. Technol. B 7(5), 1130 (1989). S. J. Pearton, W.S. Hobson and K.S. Jones, J. Appl. Phys., 66 (10), 5009 (1989). S. J. Pearton and W.S. Hobson, J. Appl. Phys., 66 (10), 5018 (1989). A. Durandet, O. Joubert, J. Pelletier and M. Pichot, J. Appl. Phys., 67(8), 3862 (1990). S. Salimian, C.B. Cooper, [[1 and A. Ellingboe, Appl. Phys. Lett., 56 (14), 1311 (1990). [106] A. Scherer, B.P. Van der Gaag, E.D. Beebe and P.S.D. Lin, J. Vac. Sci. Technol. B 8(1), 28 (1990). R1 259 [107] J.M. Cook, D.E. Ibbotson and D.L. Flamm, J. Vac. Sci. Technol. B 8(1), 1 (1990). [108] C. Constantine, D. Johnson, S.J. Pearton, U.K. Chakrabarti, A.B. Emerson, W.S. Hobson and AP. Kinsella, J. Vac. Sci. Technol. B 8(4), 596 (1990). [109] S. Salimian, C. Yuen, C. Shih and CB Cooper, 111, J. Vac. Sci. Technol. B 9(1), 114 (1991). [110] Daniel L. Flamm, Solid State Technology, March, 47 (1991). [111] SW. Pang, K.T. Sung and K.K. Ko, J. Vac. Sci. Technol. B 10(3), 1118 (1992). [112] Yoju Saito, Masahiro Hirabaru and Akira Yoshida, J. Vac. Sci. Technol. B 10(1), 175 (1992). [113] Seji Samukawa, Tomohiko Toyosato and Etsuo Wani, J. Vac. Sci. Technol. B 9(3), 1471 (1991). [114] D. Dane, P. Gadgil, T.D. Mantei, M.A. Carlson and ME. Weber, J. Vac. Sci. Technol. B 10(4), 1312 (1992). [115] 8.]. Pearton, F. Ren, J.R. Lothian, T.R. Fullowan, R.F. Kopf, U.K. Chakrabarti, S.P. Hui, A.B. Emerson, R. L. Kostelak and SS. Pei, J. Vac. Sci. Technol. B 9(5), 2487 (1991). [116] L.R. Lothian, J.M. Kuo, W.S. Hobson, E. Lane, F. Ren and SJ. Pearton, J. Vac. Sci. Technol. B 10(3), 1061 (1992). [117] S.J. Pearton, U.K. Chakrabarti, A. Katz, A.P. Perley, W.S. Hobson and C. Constantine, J. Vac. Sci. Technol. B 9(3), 1421 (1991). [118] K. Suzuki, S. Okudaira, S. Nishimatsu, K. Usami, and I. Kanomata, J. Electrochem. Soc., Solid-State Sci. and Technol., 129(12), 2764 (1982). 260 [119] Takashi Matsuura, Hiroaki Uetake, Tadahiro Ohmi, Junichi Murota, Koichi Fukuda, Nobuo Mikoshiba, T adashi Kawashima and Yoshihiro Yamashita, Appl. Phys. Lett., 56 (14), 1339 (1990). [120] Tohru Saitoh, Toshiya Yokogawa and Tadashi Narusawa, Appl. Phys. Lett., 56 (9), 839 (1990). [121] Konstantinos P. Giapis, G.R. Scheller, R.A. Gottscho, W. S. Hobson and Y.H. Lee, Appl. Phys. Lett., 57(10), 983 (1990). [122] Yasuhiro Yokota, H. Hashirnoto, M. Song and M. Awaji, Japan. J. Appl. Phys., 29(4), 739 (1990). [123] Shin'ichi Tachi and S. Okudaira, J. Vac. Sci. Technol. B 4(2), 459 (1986). [124] Jer—Shen Maa, H. Gossenberger and L. Hammer, J. Vac. Sci. Technol. B 8(4), 581 (1990). [125] Akira Shigetomi, S. Matsuda, K. Moriizumi, HJ. Kusunose, H. Onoda, T. Imai and Y. Watakabe, J. Vac. Sci. Technol. B 8(2), 117 (1990). [126] Masaharu Watanabe, Solid State Technol., March, 39 (1991). [127] Masaharu Watanabe, Solid State Technol., April, 133 (1991). [128] Philippe Laporte, Y. Melaku and M. Goethals, Solid State Technol., April, 69 (1991). [129] Mr. Hanhu Pu, private conversation. [130] Seiji Samukawa, M. Sasaki and Y. Suzuki, J. Vac. Sci. Technol., B 8(5), 1062 (1990). [131] 0.]. Glembocki and EA. Dobisz, J. Vac. Sci. Technol., B 9(2), 1403 (1991). [132] R. Cheung, S. Thomas, I. McIntyre, C.D.W. Wilkinson and SP. Beaumont, J. Vac. Sci. Technol., B 6(6), 1911 (1988). 261 [133] M. Surendra, D.B. Graves and U. Morey, Appl. Phys. Lett., 56 (11), 1022 (1990). [134] J.L. Cecchi, J.E. Stevens, R.L. Jarechi, Jr., and Y.C. Huang, J. Vac. Sci. Technol., B 9(2), 318 (1991). [135] Harold F. Winters and LC. Plumb, J. Vac. Sci. Technol., B 9(2), 197 (1991). [136] M. Haverlag, A. Kono, D. Passchier, G.M. W. Kroesen, W.J. Goedheer and FJ. de Hoog, J. Appl. Phys., 70(7), 3472 (1991). [137] S. Vanhaelemeersch, J. Van Hoeymissen, D.Vermeylen and J. Peeters, J. Appl. Phys., 70(7), 3892 (1991). [138] R. K.F. Teng and HG. Yang, J. Vac. Sci. Technol., B 9(2), 380 (1991). [139] R. Venkatasubramanian, M.L. Timmons and TS Colpitts, J. Appl. Lett., 59(17), 2153 (1991). [140] R. J. Davis, Appl. Phys. Lett., 59(14), 1717 (1991). [141] LC. Arnold and H.H. Sawin, J. Appl. Phys., 70 (10), 5314 (1991). [142] N. Sadeghi, H. Debontride, G. Turban and MC. Peignon, Plasma Chemistry and Plasma Processing, 10(4), 553 (1990). [143] Jéréme Perrin, J. Méot, J -M. Siéfert and J. Schmitt, Plasma Chemistry and Plasma Processing, 10(4), 571 ( 1990). [144] K.R. Ryan and LC. Plumb, Plasma Chemistry and Plasma Processing, 10(2), 207 (1990). [145] N. Mutsukura and G. Turban, Plasma Chemistry and Plasma Processing, 10(1), 27 (1990) [146] ES. Aydil, J.A. Gregus and R.A. Gottscho, PS-ThAS, 39th AVS National Symposium, Chicago, Nov. (1992) 262 [147] Joel M. Cook and K.G. Donohoe, Solid State Technol., April, 119 (1991). [148] Tatsuya Takeshita, T. Unagami and O. Kogure, Japan. J. Appl. Phys., 28(3), L358 (1989). [149] T. Shibata, Y. Nanishi and M. Fujirnoto, Japan. J. Appl. Phys., 29(7), L118] (1990). [150] Semiconductor International, June, 46 (1992) [151] Carey Pico, Solid State Technology, May, 81 (1992) [152] Solid State Technology, Dec., 22 (1991) [153] Semiconductor International, May, 48 (1992) [154] Peter Singer, Semiconductor International, July, 52 (1992) [155] A. J. van Roosmalen, J. A. G. Baggennan and S. J. H. Brader, Day M, Plenum Press, New York (1991). [156] B. Petti and J. Pelletier, Japan. J. Appl. Phys., 26(6), 825 (1987). [157] A.J. Perry, H.Persing and R.W. Boswell, PS2-ThM10, 39th AVS National Symposium, Chicago, Nov. (1992). [158] G. Giroult-Matlakowsk, S. Armand, R. W. Boswell, PS-MoP8, 39th AVS National Symposium, Chicago, Nov. (1992). [159] R.W. Boswell and R.K. Porteous, PS-MoP9, 39th AVS National Symposium, Chicago, Nov. (1992). [160] M.A. Jamyk and R.W. Boswell, PS-WeP24, 39th AVS National Symposium, Chicago, Nov. (1992). [161] C. Charles, A. Goullet, C. Cardinaud, G. Turbanand R.W. Boswell, EM-ThP36, 39th AVS National Symposium, Chicago, Nov. (1992). [162] Jeffrey A. Tobin, Guifan Li, L. T. Mahoney, D.D. Denton and J. L. Shohet, J1 l-ThMl 1, 39th AVS National Symposium, Chicago, Nov. (1992). 263 [163] L.T. Mahoney, G.Li, A.E. Wendt, J.L. Shohet and H. Yu, PS-MoP7, 39th AVS National Symposium, Chicago, Nov. (1992). [164] J. Hopwood, C.R. Guarnieri and I]. Cuomo, PS-ThA8, 39th AVS National Symposium, Chicago, Nov. (1992). [165] A.K. Srivastava and J. Asmussen, PS-ThA3, 39th AVS National Symposium, Chicago, Nov. (1992). [166] E.B. Manring, x ' n l ' ' ' of i r w v heatin_ f 001i 0 - w. 'v matril ° 1 ' lr linc.ricl resonagt Qavity, M.S. thesis, Department of Electrical Engineering, Michigan State University (1988). [167] John S. Mckilop, J.C. Forster, and W. M. Holber, J. Vac. Sci. Technol. A 7(3), 908 (1989). [168] T.I. Cox, V.G. Deshmukh, D.A. Hoe, A. J. Hydes, N. St. J. Braithwaite and N. M. P. Benjamin, J. Phys. D: Appl. Phys. 20, 820 (1987). [169] Edbertho Leal-Quiros and M.A. Prelas, Rev. Sci. Instrum. 61(6), 1708 (1990). [170] J. A. Meyer, G-H. Kim and N. Hershkowitz, PS-ThA6, 39th AVS National Symposium, Chicago, Nov. (1992). [171] R.A. Stewart, B. Troyanovsky and M.A. Lieberman, "Modeling of magnetic bucket confinement in an ECR processing reactor," IEEE Conference on Plasma Science, Williamsburg (1991). [172] Masayuki Fukao and Yasuo Harada, J. Appl. Phys. 72(9), 3943 (1992). [173] D. Henry, J.M. Francou and A. Inard, J. Vac. Sci. Technol. A10(6), 3426 (1992). 264 [174] ME. Day, M. Delfino, and S. Salimian, J. Appl. Phys. 72 (11), 5467 (1992). [175] R. A. Gottscho, C.W. Jurgensen and DJ. Vitkavage, J. Vac. Sci. Techol. B 10(5), 2133 (1992). [176] F. Ren, T. R. Fullowan, S. J. Peraton, J. R. Lothian, R. Esagui, C. R. Abernathy and W.S. Hobson, J5-ThA6, 39th AVS National Symposium, Chicago, Nov. (1992). [177] K.T. Kung, J.P. Fourier and SW. Pang, PS-WeM7, 39th AVS National Symposium, Chicago, Nov. (1992). [178] D.K. Reinhard, mmdaatign ta Intagrataa Qimait Enginaasm' g, Houghton Mifflin Company, Boston, 181 (1987). [179] Peter Singer, Semiconductor International, August, 36 (1992) [180] Robert C. Weast, CRQ Handbgak af Qhamistsy Ed Physics, 1St student edition, B (1990). [181] SM. Gorbatkin and LA. Berry, J. Vac. Sci. Technol. A 10(5), 3104, Sept. (1992). [182] J. Hopwood, C.R. Guarnieri, S.J. Whitehair and 1.]. Cuomo, J. Vac. Sci. Technol. A 11(1), 147 (1993). [183] J. Hopwood, C.R. Guarnieri, S.J. Whitehair and 1.]. Cuomo, J. Vac. Sci. Technol. A 11(1), 152 (1993). [184] SM. Gorbatkin, L.A. Berry, and J .B. Roberto, J. Vac. Sci Technol. A 8 (3), 2893 (1992). [185] S. Whitehair and J. Asmussen, J. propulsion and power, 3 (2), 136 (1987). [186] L. Mahoney, seminar given in Electrical Engineering Department, Michigan State University, Feb (1993). 265 [187] D. Peterson, The OSHA Compliance Manual. McGraw-Hill Book Company, New York, New York. [188] EC. Sze, G. Alers, D. K. Reinhard, B. Golding and J. Asmussen, "Etching of silicon with features to 0.2 micron using a multipolar ECR plasma reactor," 39th AVS National Symposium, Chicago, Nov., 1992 (to be published in J. Vac. Sci. T echnol.) [189] F .C. Sze and J. Asmussen, "Experimental scaling laws for ECR reactors," 39th AVS National Symposium, Chicago, Nov., 1992 (to be published in J. Vac. Sci. Technol.) [190] RC. Sze and J. Asmussen, "Downstream characterization of a 20 cm diameter 915MHz/2.4SGHz multipolar electron cyclotron resonant plasma source," (presented at 1991 IEEE international conference on plasma science, Williamsburg, Virginia). [191] IR. Brophy and R]. Wilbur, AIAA Journal, 23, 1731 (1985). [192] J. Rogers, "Properties of steady-state high pressure argon microwave discharge", Ph.D. thesis, Michigan State University, East Lansing (1982). ell; .- .Hbli. (6.101;...Itltttltlf.d .. .1 . rattle... 111...! (1.1.3:: t...‘ 11114 .2. t 7 1 4.1 n¢\r3to xv. 4. VII-.7011‘ll33‘00 . Iv ...l .0.Iditai .... .«I 4 ..I t .0 .I..rolr.|t?l.flrt.¥lk! t ... .1171. r ..vI1« ....l. .3531? «1411.11... ...?i I: 1 s so 1.6-I'M) i. .3 v I .9 u : . fr 131! i .r. {0. «$11!: 1.... ......htteuvcir..fl.u... «D. In]. . . .ut 4.11.511 1123363. a at 1!..mnrlrt. 1 1(- ‘v '\ . \ (”8:931:11- lfiskrot. as. I] <71 11; or.) (111’ [Fr 9113.- .3I1l. I. 921:; — : 25.336; . 43Q"$.v I4 Loft DIP») III.- olxrrai v? u 1.3, ..- 4toflu .../I713. .1111” oft—rec): v... . . ' ‘Iu’.’ 9 614;: — 1.»- Mn. ‘9...- OK 5...“. n34. I I t p.) . I I. L. .1... . at ....11...: . . . ... i . . ..11111...iillillini.tJIi ..1 t . 1 . t . ID 111611 111.111.11-111”... . .. l .-.l 1.11.11 . it..- 1 ttqtr 1111......1111 \(t It ittd.......l......t.l. ii ..16. us 1. ..11...1 £1.11 ll..ti..li ....ultiii...lluiii. . . .101. . . .110 VIVISIuboo. ,. . 1.1.3.112... 111:" v . . 11.071“ 1:3 I. a... .1II| 14 I! “A i. . at.“‘. 6 let g I .ohtuh v '5‘ on. .96. 1.1 Q . 11‘. avian o "H.131“. 711.4 ... If 3.3.. e. m3! . 0 too- . rd 1 x. 0 v - swans? s. 11 kg. \ ham—m I 15*.” N Mt move... a 0 Id 0!... I Iv an. ‘ Al on x - bl: i1 '4 1t ‘4]? . 0 lu‘ ' .tt .. . 13 . .111. t . at»... a. it 1. _. I. . . .-i ..tzhnntthi. ~ .31.“ . 11:11.. 1H1. o 0.1..th burr.“ I not. ... i A», . 11.. 411