The Physics and Technology of Ion Sources

Ian G. Brown (Ed.) The Physics and Technology of Ion Sources The Physics and Technology of Ion Sources, Second Edition...

Author: Ian G. Brown

48 downloads 1004 Views 16MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Ian G. Brown (Ed.) The Physics and Technology of Ion Sources

The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

The Physics and Technology of Ion Sources Second, Revised and Extended Edition Edited by Ian G. Brown

Editor Ian G. Brown Lawrence Berkeley National Laboratory Berkeley, California [email protected]

&

All books published by Wiley-VCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.: applied for

Cover Picture Photograph of metal ion beam formed by the Mevva II vacuum arc ion source, looking toward the extractor. The visible glow showing the beam is excitation radiation from the background gas (Courtesy of Lawrence Berkeley National Laboratory).

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. Bibliographic information published by Die Deutsche Bibliothek Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data is available in the Internet at . 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – nor transmitted or translated into machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically marked as such, are not to be considered unprotected by law. Printed in the Federal Republic of Germany. Printed on acid-free paper. Composition Khn & Weyh, Satz und Medien, Freiburg Printing Strauss GmbH, Mrlenbach Bookbinding Litges & Dopf Buchbinderei GmbH, Heppenheim ISBN

3-527-40410-4

ad perpetuam rei memoriam

VII

Preface This Second Edition comes 15 years after publication of the First Edition in 1989. Much has evolved in the ion source world in the intervening years. Understanding of ion source behavior has grown, new source variants have been developed, new applications have evolved, and quite generally the beam parameters achievable have improved greatly. This book is designed to serve as a review and reference. The target reader is the researcher actively involved in ion source application and/or development. The objective of the book is to provide a comprehensive, easily understood introduction to and survey of the field. The different chapters have been written by researchers who are expert in the topic discussed. The chapters are more-or-less independent and self-contained. The terminology, notation and units used are as conventional within the subfield of a particular chapter, and vary a little from chapter to chapter. The reader should expect consistent use of symbols and units only within a given chapter and not between chapters. No confusion should follow from this. Ion source research is a very empirical field, and for the most part the emphasis presented in this book is experimental. It has overwhelmingly been the case that theoretical understanding of an ion source has followed after its construction and after its experimental performance has been characterized. This situation is reflected in the presentations here. The academic background assumed of the reader is roughly physics graduate level. A working knowledge of plasma physics, atomic physics, and electromagnetic theory will definitely aid the digestion, and some familiarity with electronics and electrical systems will help too. I am most grateful to my friends and colleagues, the contributors to this book, who patiently persevered a plethora of pernickety proofing. This book is the product of their perseverance, and I thank them. Ian Brown Berkeley, California December 2003


IX

Contents Preface

VII

List of Contributors

XVII

1

Introduction Ian Brown

2

Plasma Physics Ian Brown

2.1 2.2 2.2.1 2.2.2 2.2.3 2.2.4 2.2.5 2.3 2.3.1 2.3.2 2.3.3 2.4 2.4.1 2.4.2 2.4.3 2.4.4 2.5 2.5.1 2.5.2 2.5.3 2.5.4 2.5.5 2.5.6

Introduction 7 Basic Plasma Parameters 8 Particle Density 8 Fractional Ionization 9 Particle Temperature 9 Particle Energy and Velocity 10 Collisions 12 The Plasma Sheath 13 Debye Length 14 Charge Neutrality 15 Plasma Oscillations 16 Magnetic Field Effects 17 Gyro Orbits 17 Gyro Frequencies 18 Magnetic Confinement 19 Magnetic and Plasma Pressure 20 Ionization 21 Electron Impact Ionization 22 Multiple Ionization 23 Photoionization 25 Ion Impact Ionization 26 Negative Ions 27 Field Ionization 27

1

7


X

Contents

3

3.1 3.2 3.3 3.4 3.5 3.6 3.7 4

4.1 4.2 4.2.1 4.2.2 4.2.3 4.2.4 4.2.5 4.2.6 4.3 4.3.1 4.3.2 4.3.3 4.4 5

5.1 5.2 5.3 5.4 5.5 5.6 6

6.1 6.1.1 6.1.2 6.1.3 6.2 6.3 6.3.1

Elementary Ion Sources Ian Brown Introduction 29 Terminology 29

29

The Quintessential Ion Source Ion Beam Formation 34 Ion Beam Parameters 36 An Example 38 Conclusion 40 Computer Simulation of Extraction Peter Spdtke Introduction 41 Positive Ion Sources 42

30

41

Filament Driven Cusp Sources 43 Duoplasmatrons and Duopigatrons 44 Vacuum Arc Ion Sources 44 Laser Ion Sources 45 ECR Ion Sources 46 Penning Ion Sources 50 Negative Ion and Electron Sources 54 Hot Cathode Electron Sources 55 Plasma Electron Sources 55 H– Sources 55 Conclusion 59 Ion Extraction 61 Ralph Hollinger Introduction 61

Fundamentals of Ion Beam Formation in the Extraction System 62 Beam Quality 65 Sophisticated Treatment of Ion Beam Formation in the Extraction System 68 Multi-Aperture Extraction Systems 76 Starting Conditions 83 Beam Transport 87 Peter Spdtke and Ralph Hollinger Introduction 87 Drift 91

Extraction System and Acceleration Gap Low Energy Beam Line 92 Current Effects 93 Space-Charge Compensation 93 Residual Gas Collisions 93

92

Contents

6.3.2 6.3.3 6.3.4 6.4 6.4.1 6.4.2 6.4.3 6.4.4

Sputtering 94 Preserving Space Charge Compensation 95 Influence of Space Charge Compensation 95 A LEBT System for the Future Proton Linac at GSI Compound System 99 Pentode or Two-Gap System 101 Triode System and DC Post-Acceleration 102 Discussion 105

7

High Current Gaseous Ion Sources Nikolai Gavrilov Introduction 107

7.1 7.2 7.2.1 7.2.2 7.2.3 7.3 8

8.1 8.2 8.3 8.4 8.4.1 8.4.2 8.4.3 8.4.4 8.4.5 8.5 8.5.1 8.5.2 8.6 8.6.1 8.6.2 8.6.3 8.6.4 8.6.5 8.6.6 8.6.7

98

107

Basic Types of High Current Ion Sources Filament Driven Ion Sources 110 High-Frequency Ion Sources 116 Cold Cathode Ion Sources 121 Conclusion 130

108

Freeman and Bernas Ion Sources 133 Marvin Farley, Peter Rose, and Geoffrey Ryding Introduction 133 The Freeman Ion Source 134 The Bernas Ion Source 138

Further Discussion of the Source Plasma 141 Plasma and Sheath Potentials 144 Effect of Sputtering on the Plasma 146 Ion Heating of the Cathode and Anticathode in the Bernas Source Current Balance in the Freeman Source 150 Current Balance in the Bernas Source 151 Control Systems 154 Freeman and Bernas Controls 154 Bernas Indirectly Heated Cathode 155 Lifetime and Maintenance Issues 158 Use of BF3 158 Use of PH3, AsH3, P4, and As4 159 Use of Sb, Sb2O3, and SbF3 159 Use of SiF4 and GeF4 159 General Guidelines for the Use of Other Organic and Inorganic Compounds 160 Electrode Cleaning and Maintenance 160 Insulator Cleaning and Maintenance 160

148

XI

XII

Contents

9

9.1 9.2 9.3 9.3.1 9.3.2 9.3.3 9.3.4 9.4 10

10.1 10.2 10.2.1 10.2.2 10.3 10.3.1 10.3.2 10.4 10.4.1 10.4.2 10.4.3 10.4.4 10.5 10.6 10.6.1 10.6.2 10.7 11

11.1 11.2 11.3 11.3.1 11.3.2 11.4 11.4.1 11.4.2 11.4.3

Radio-Frequency Driven Ion Sources Ka-Ngo Leung Introduction 163

163

Capacitively Coupled RF Sources 163 Inductively Coupled RF Sources 165 Source Operation with an External RF Antenna 165 Multicusp Source Operation with Internal RF Antenna 167 Increasing the Ion Beam Brightness of a Multicusp RF Source with Internal Antenna 169 Multicusp Source Operation with External RF Antenna 171 Applications of RF Ion Sources 174 Microwave Ion Sources Noriyuki Sakudo Introduction 177

177

Microwave Plasma in Magnetic Fields 177 Plasma Parameter Changes due to Magnetic Field and Microwave Frequency 177 High Density Plasma at Off-Resonance 178 Some Practical Ion Source Considerations 180 Microwave Impedance Matching to the Plasma 180 High Current Ion Beams Extracted from an Off-Resonance Microwave Ion Source 183 Versatility of Beam Extraction 184 Large Cross Sectional Beam formed by a Multi-Aperture Extractor 184 Slit-Shaped Beam for Ion Implantation 185 Further Improvements in Slit-Shaped Beams 188 Compact Microwave Ion Sources 190 Diversity of Available Ion Species 191 Microwave Ion Sources for Commercial Implanters 194 Semiconductor Device Fabrication 194 SOI Wafer Fabrication 198 Conclusion 199 ECR Ion Sources 203 Daniela Leitner and Claude Lyneis Introduction 203

Brief History of the Development of ECR Ion Sources The LBNL ECR Ion Sources 207 The AECR-U Ion Source 207 The VENUS ECR Ion Source 208 Physics and Operation of ECR Ion Sources 210 Electron Impact Ionization 210 Charge Exchange 211 Plasma Confinement 212

204

Contents

11.4.4 11.4.5 11.5 11.6 11.7 11.7.1 11.7.2 11.7.3 11.7.4 11.7.5 11.8 11.8.1 11.8.2 11.9 12

12.1 12.2 12.3 12.3.1 12.3.2 12.3.3 12.3.4 12.3.5 12.3.6 12.4 12.4.1 12.4.2 12.4.3 12.4.4 12.5 12.5.1 12.5.2 12.5.3 12.6 12.6.1 12.6.2 12.7 13

13.1 13.2

ECR Heating 214 Gas Mixing 215 Design Considerations 216 Microwave and Magnetic Field Technologies 217 Metal Ion Beam Production 218 Direct Insertion 219 Sputtering 220 Gaseous or Volatile Compounds (MIVOC Method) 220 External Furnaces (Ovens) 221 Efficiencies 222 Ion Beam Extraction from ECR Ion Sources 223 Influence of Magnetic Field and Ion Temperature on the Extracted Ion Beam Emittance 223 Influence of Plasma Confinement on Beam Emittance 225 Conclusion 227 Laser Ion Sources 233 Boris Sharkov Introduction 233

Basics of Laser Plasma Physics 234 General Description 235 Laser Characteristics 235 Target Illumination System 237 Target Ensemble 238 Pulse Width and Target-Extractor Separation 239 Extraction System 240 Low Energy Beam Transport Line (LEBT) 242 Beam Parameters 244 Current Profile 244 Charge State Distribution 245 Beam Emittance 246 Pulse Stability and Source Lifetime 247 Sources at Accelerators 248 The LIS at ITEP-TWAC 248 The LIS at CERN 250 The LIS at JINR Dubna 252 Other Operating Options 252 High Current, Low Charge State Mode 252 Influence of Magnetic Field on the Laser Ion Source Plasma Conclusion 254 Vacuum Arc Ion Sources Efim Oks and Ian Brown Introduction 257 Background 258

257

253

XIII

XIV

Contents

13.3 13.4 13.5 13.5.1 13.5.2 13.5.3 13.5.4 13.6 13.6.1 13.6.2 13.6.3 13.6.4 13.7 13.7.1 13.7.2 13.7.3 13.7.4 13.7.5 13.8

Vacuum Arc Plasma Physics 259 Principles of Operation 260 Beam Parameters 262 Beam Current 262 Beam Profile, Divergence and Emittance 264 Beam Composition 265 Beam Noise, Pulse Stability, and Lifetime 267 Recent Improvements in Parameters and Performance 268 Enhancement of Ion Charge States 268 Alternative Triggering of the Vacuum Arc 271 Reduction in Ion Beam Noise and Increased Pulse Stability 273 Generation of Gaseous Ions 274 Source Embodiments 275 LBNL Mevva Sources 276 HCEI Titan Sources 277 NPI Raduga Sources 278 GSI Varis Sources 279 Other Versions and Variants 280 Conclusion 282

14

Negative Ion Sources Junzo Ishikawa Introduction 285

14.1 14.2 14.2.1 14.2.2 14.2.3 14.3 14.3.1 14.3.2 14.3.3 14.4 14.4.1 14.4.2 14.5 15

15.1 15.1.1 15.1.2 15.2 15.2.1 15.2.2 15.3

285

Surface Effect Negative Ion Sources 285 Negative Ion Production by Surface Effect 285 Surface Effect Light Negative Ion Sources 292 Surface Effect Heavy Negative Ion Sources 296 Volume Production Negative Ion Sources 303 Negative Ion Formation by Volume Production 303 History of Source Development 304 Recent Volume Production Negative Ion Sources 305 Charge Transfer Negative Ion Sources 306 Negative Ion Production by Charge Transfer 306 History of Charge Transfer Negative Ion Sources 308 Conclusion 308 Ion Sources for Heavy Ion Fusion Joe Kwan Introduction 311

311

Heavy Ion Beam Driven Inertial Fusion 311 HIF Ion Source Requirements 312 Beam Extraction and Transport 314 Scaling Laws for Beam Extraction and Transport Large Beam vs. Multiple Small Beamlets 317 Surface Ionization Sources 318

314

Contents

15.3.1 15.3.2 15.3.3 15.4 15.5 15.5.1 15.8.1 15.6 15.7 15.7.1 15.7.2 15.8

Contact Ionizers 318 Aluminosilicate Sources 322 Surface Ionization Sources for HIF 326 Gas Discharge Ion Sources for HIF 326 Pulsed Discharge Sources 330 Metal Vapor Vacuum Arc Sources for HIF 330 Laser Ion Sources for HIF 333 Negative Ion Sources for HIF 333 HIF Injector Designs 335 Large Diameter Source Approach 335 Merging Multiple Beamlets Approach 336 Conclusion 338

16

Giant Ion Sources for Neutral Beams Yasuhiko Takeiri Introduction 341

16.1 16.2 16.2.1 16.2.2 16.2.3 16.3 16.3.1 16.3.2 16.4 16.5 16.5.1 16.5.2 16.5.3 16.6

Large Volume Plasma Production 342 Bucket Plasma Sources with Multi-Cusp Magnetic Field Plasma Modeling 344 Atomic Fraction 347 Large Area Beam Extraction and Acceleration 348 Electrode Systems for Large Area Beams 348 Beamlet Steering 350 Giant Positive Ion Sources 353 Giant Negative Ion Sources 359 Operational Principles of Negative Ion Sources 359 Negative Ion Extraction and Acceleration 363 Giant Negative Ion Sources 366 Future Directions of Development 369

Appendices 373 Appendix 1 Physical Constants 373 Appendix 2 Some Plasma Parameters 374 Appendix 3 Table of the Elements 374 Index

341

377

342

XV

XVII

List of Contributors Ian Brown Lawrence Berkeley National Laboratory Berkeley, California

Claude Lyneis Lawrence Berkeley National Laboratory Berkeley, California

Marvin Farley Applied Orion Group Beverly, Massachusetts

Efim Oks High Current Electronics Institute Tomsk, Russia

Nikolai Gavrilov Institute for Electrophysics Ekaterinburg, Russia

Peter Rose Applied Orion Group Beverly, Massachusetts

Ralph Hollinger Gesellschaft fr Schwerionenforschung Darmstadt, Germany

Geoffrey Ryding Applied Orion Group Beverly, Massachusetts

Junzo Ishikawa Kyoto University Kyoto, Japan

Noriyuki Sakudo Kanazawa Institute of Technology Ishikawa, Japan

Joe Kwan Lawrence Berkeley National Laboratory Berkeley, California

Boris Sharkov Institute for Theoretical & Experimental Physics Moscow, Russia

Daniela Leitner Lawrence Berkeley National Laboratory Berkeley, California

Peter Spdtke Gesellschaft fr Schwerionenforschung Darmstadt, Germany

Ka-Ngo Leung Lawrence Berkeley National Laboratory Berkeley, California

Yasuhiko Takeiri National Institute for Fusion Science Toki, Japan


1

1

Introduction Ian Brown

This book is an introduction to the physical principles of ion sources and a detailed review of a number of different kinds of sources. It is intended to serve both as an introductory textbook for newcomers to the field, and as a guide and reference for seasoned ion source users and researchers. The various chapters of the book have been written by researchers who are expert in the topic discussed. The chapters are independent and largely self-contained. While the scope covers much of the ion source field, the book is not encyclopedic. However, the principles described and the kinds of sources considered span a large part of the ion source activity that is taking place around the world today. The ion source research and development community is composed of workers from a rather wide variety of scientific disciplines. These researchers have been drawn from the various application areas of ion sources, and as such they have a variety of different backgrounds and fields of expertise. A major subset of the ion source field has to do with particle accelerator injection, and workers in this area are largely nuclear or accelerator physicists and engineers, with support derived from and ion source goals directed toward the requirements of the various kinds of particle accelerators in operation around the world. A major industrial application of ion sources is for semiconductor ion implantation, a multi-billion dollar world activity, and many workers in this subfield come from backgrounds in solid state or materials science and from electrical engineering backgrounds. There is also a large and active research community worldwide in the field of non-semiconductor ion implantation for material surface modification for purposes as diverse as, for example, wear resistance, biological compatibility, and optoelectronics. An area in which a vast amount of progress has been made is that of the giant neutral beam injectors for heating and fueling experimental fusion reactor devices; the intense, high energy neutral beams are formed by charge-exchange from intense, high energy ion beams, and the ion sources used for this purpose are most impressive indeed. For the most part workers in this ion source subfield come from a plasma physics background. And there are many more application areas, each with its own ion source needs, constraints, and driving forces. Workers in the various application areas naturally tend to communicate mostly with researchers in their own field, to participate in conferences within their own The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

2

1 Introduction

field, and to read journals that address primarily their own field. Thus there arises a problem endemic to the ion source field, that of insularity. This is unfortunate, since there is often much to be learned from the work of researchers in neighboring ion source subfields – cross-field fertilization. The International Conference on Ion Sources, held every two years at changing locations around the world, attempts to bring together workers from a wide range of ion source subfields. This book may also play a helpful role. The first part of the book is introductory, with a presentation of basic plasma physics and the basic ideas behind ion source design and operation. This is followed by a look at some of the computer simulation work that has been used in the field and that is available to the researcher. The remainder and the bulk of the book consists of chapters that address specific kinds of ion sources, written by leading researchers actively engaged in the specific ion source considered. The kinds of sources treated in-depth are: high current gaseous ion sources, Freeman and Bernas ion sources, RF-driven ion sources, microwave driven ion sources, ECR ion sources, laser ion sources, vacuum arc ion sources, negative ion sources, ion sources for heavy ion fusion, and the giant ion sources for neutral beams. Chapter 2 is an outline of the basic plasma physics necessary for an understanding of ion source physics. It is only an outline and summary; references are given to other more complete introductory plasma physics texts. Nevertheless, a familiarity with this basic plasma physics should go a long way toward understanding most of the ion source related plasma physics in the subsequent chapters. Topics addressed include basic parameters, the plasma sheath, magnetic field effects, and the various ionization phenomena. A brief and very basic introduction to the fundamental concepts underlying the physics and operation of plasma-based ion sources follows in Chapter 3. This is the “Ion Sources for Beginners” chapter. The student or scientist who has had no previous experience with ion sources at all should find here the information that he or she needs to understand the principles involved. Following a step-by-step discussion of how the components of an ion source come together to create an ion beam, an example of a simple ion source is presented that demonstrates the fundamentals. Computer simulation of the ion beam formation process has evolved immensely over the years, and it is now possible to simulate the processes to considerable precision. Chapter 4, by Peter Spdtke of GSI Darmstadt, Germany, discusses the computer simulation of beam extraction. The chapter is divided into the cases of positive ion beams and negative ion beams, with various kinds of ion sources discussed in detail. The beam formation process from the perspective of the laboratory physicist is discussed in Chapter 5 by Ralph Hollinger, also of GSI Darmstadt, Germany. Here the extractor geometry is considered in detail, with consideration given to how all of the beam parameters are affected by the extractor design. There are a great many highly interactive considerations that all add up to good overall design of the beam formation and acceleration electrode system. Various different kinds of extractors are considered, for the formation of different kinds of beams. After the beam is formed it must inevitably be transported away from the source and toward the application region. The final application may be an accelerator, an

1 Introduction

implanter, an ion analysis instrument, a fusion reactor, or any of a great many possibilities. The transport distance can be less than a meter or many tens of meters and the beam might be of very small cross-sectional area or it might be a very broad beam, but in virtually all cases the beam must be passed through a vacuum pipe of some sort with minimal loss. A good working knowledge of factors that influence the beam transport is important. The tendency of high current beams to spacecharge blowup and the critical role played by background electrons in space-charge neutralization (or compensation) is one such factor. Beam transport is discussed jointly by Spdtke and Hollinger (GSI) in Chapter 6. High current gaseous ion sources are discussed by Nikolai Gavrilov, from the Institute for Electrophysics at Ekaterinburg, Russia, in Chapter 7. The kinds of sources discussed include filament driven sources, high frequency sources, and cold cathode sources. The performance that can be obtained from modern versions of these sources is impressive. Here the basic plasma physics involved and the considerations related to beam extraction are discussed. Examples are presented of a number of working cold-cathode ion sources. One of the major technological applications of ion sources is in the semiconductor ion implantation industry. The Freeman source and the Bernas source are the usual sources of choice in this field. These sources have been improved over the years to a very fine pitch. Freeman and Bernas sources are considered in depth in Chapter 8 by Marvin Farley, Peter Rose and Geoffrey Ryding of the Applied Orion Group, Beverly, Massachusetts. Following a detailed discussion of the physics of the plasma in these sources, means of ion source control are described, and the key issues of source maintenance and lifetime are discussed. This chapter should provide a valuable reference for workers in the semiconductor implantation community. Radio frequency (RF) driven ion sources have been a staple of the ion source platter available to the experimenter for many years. Recent development has extended the ion beam parameter range and improved the overall performance greatly. RF ion sources are discussed in Chapter 9 by Ka-Ngo Leung of the Lawrence Berkeley National Laboratory. Both capacitively coupled and inductively coupled source types are addressed, as well as the key role played by the multicusp magnetic confinement geometry. Plasma formation by means of RF power and microwave power are related approaches. Thus in a sense the microwave driven ion source is a sibling to the RF driven source. The current status of microwave ion sources is described in excellent detail in Chapter 10 by Noriyuki Sakudo of the Kanazawa Institute of Technology, Ishikawa, Japan. The basic plasma physics involved in the absorption of microwave radiation in a plasma is discussed, including the phenomenon of overdense plasma production, when the plasma density can be greater than the microwave cutoff density by an order of magnitude or more. Microwave power coupling to the source plasma and ion beam extraction techniques are then considered. Finally, the application of microwave ion sources to commercial ion implanters is discussed. Microwave power can be coupled resonantly to a plasma by matching the microwave frequency to the electron cyclotron frequency in the magnetic field by which

3

4

1 Introduction

the plasma is confined – the electron cyclotron resonance (ECR) frequency. If the gas pressure is also sufficiently low and the plasma confinement sufficiently high, then the electrons can be heated to substantial energy by the resonant microwave field and the ions can be stripped to high charge state. An ion source incorporating these principles is called an ECR ion source. Though also of course a kind of microwave ion source, the terminology used has now solidified to distinguish the microwave ion source and the ECR ion source quite clearly. ECR ion sources are described in Chapter 11 by Claude Lyneis and Daniela Leitner of the Lawrence Berkeley National Laboratory. The primary application of ECR ion sources is for generation of high charge state ions for injection into particle accelerators, mostly cyclotrons. The chapter summarizes the history of ECR sources, outlines the plasma physics involved, design considerations, microwave and magnetic field technologies involved, and the production of metal ion beams. The chapter should provide an excellent reference for the laboratory worker. Sources in which the plasma is formed by an intense pulsed laser beam focused onto a solid target are called laser ion sources. These kinds of sources are addressed in Chapter 12 by Boris Sharkov of the Institute for Theoretical and Experimental Physics, Moscow, Russia. Laser ion sources are used primarily for particle accelerator injection into heavy ion synchrotrons. This chapter covers the basics of laser plasma physics, the details involved in putting together a laser ion source, and the beam parameters that can be obtained. Examples of laser ion source facilities at some major accelerator laboratories are described. The vacuum arc ion source is a high current metal ion source. Whereas it is often the case that metal plasma (plasma formed from a metal) is more problematic to form than a gaseous plasma (plasma formed from a gas), the vacuum arc discharge is a relatively simple means of generating large amounts of metal plasma, and the vacuum arc ion source, in turn, is a relatively simple means of generating high current beams of metal ions. These sources are described in Chapter 13 by Efim Oks of the High Current Electronics Institute, Tomsk, Russia, and Ian Brown of the Lawrence Berkeley National Laboratory. Following a brief summary of vacuum arc plasma physics, ion source operating principles are described, then the beam parameters that can be obtained, and some recent new developments in these kinds of sources. The chapter concludes with a summary description of some of the source embodiments that have been made and are in use at various laboratories around the world. Negative ion sources play an important and unique role in the ion source universe. Because the ions carry a negative charge, the plasma physics involved, the means of ion beam extraction, and the kinds of applications for which the sources are used are all significantly different from all the positive ion sources. Negative ion sources are treated in Chapter 14 by Junzo Ishikawa, Kyoto University, Japan. The two different approaches to negative ion formation in plasmas, surface production and volume production, are considered in detail, and the ways in which negative ions sources have been constructed to take advantage of these two approaches are described by reference to many different examples. The ion beam requirements for heavy ion fusion are quite different from beam parameters offered by the mainstream kinds of ion sources. The beam must be very

1 Introduction

high current, short pulse, fast pulse rise and decay times, single charge state, highly reproducible and low noise. At the same time the source lifetime (number of pulses between maintenance downtimes) must be very high. These are difficult requirements. This subfield is discussed in Chapter 15 by Joe Kwan of Lawrence Berkeley National Laboratory. Surface ionization sources, the research approach taken at present, are firstly described, followed by consideration of gas discharge sources, laser and vacuum arc sources, and negative ion sources. By far the largest ion sources are those that have been developed for the neutral beam injection devices for fusion plasma heating. The requirements of these sources and the associated beam requirements are daunting. Nevertheless, impressive progress has been made. Unique features of the sources include the large volume plasma production that must first be done and the extremely precise but very large area beam extraction and acceleration electrodes that must be manufactured. Sources based on positive ions are used in the lower energy regime and negative ion based systems are used for sources with ion energy in the multi-hundreds of keV range. The giant ion sources that have been developed for the neutral beam development program are described in Chapter 16 by Yasuhiko Takeiri of the National Institute for Fusion Science, Toki, Japan. The ion source coverage spanned by the contents of this book is not total. Indeed it would be a nigh impossible task to treat in detail all of the different existing kinds of ion sources in a manageable book. Nevertheless the range of sources considered here does span a diverse spectrum of beam parameters and source techniques. Perhaps, with a little serendipitous fortune, some of the ideas presented here might prove to be fertile material for the generation of new kinds of ion sources and ion beam devices. There are other resources available to the ion source investigator, both novice and expert. These include books, journals, and conference proceedings. Some previous ion source texts are referenced below [1–6]. Many of the topical conferences have considerable ion source content, and the International Conference on Ion Sources (ICIS) is an established conference that is held every two years (odd-numbered years) at various locations around the world. A web search will quickly lead to information about the next planned ICIS. The Proceedings of the ICIS meetings are published in Rev. Sci. Instrum., usually in the first quarter of the (even-numbered) year following the ICIS meeting. The ICIS Proceedings [7] are a very rich source of upto-date information on a wide range of ion source physics and technology.

5

6

1 Introduction

References [1] H. Zhang, Ion Sources (Springer-Verlag, [2] [3] [4] [5] [6]

Berlin, 2000). B.H. Wolf, Editor, Handbook of Ion Sources (CRC, London, 1995). I.G. Brown, Editor, The Physics and Technology of Ion Sources (Wiley, New York, 1989). A.T. Forrester, Large Ion Beams (Wiley, New York, 1988). L. Valyi, Atom and Ion Sources (Wiley, New York, 1977). R.G. Wilson and G.R. Brewer, Ion Beams with Applications to Ion Implantation (Wiley, New York, 1973).

[7] See Proceedings of the International Conference

on Ion Sources, Rev. Sci. Instrum. 61, 221–666 (1990)(January ’90, Part 2); 63, 2351–2912 (1992)(April ’92, Part 2); 65, 1039–1484 (1994)(April ’94, Part 2); 67, 867–1423 (1996)(March ’96, Part 2); 69, 613–1206 (1998)(Feb ’98, Part 2); 71, 603–1242 (2000)(Feb ’00, Part 2); 73, 505–1101 (2002)(Feb ’02, Part 2); 75, 1379–1940 (2004)(May ’04, Part 2).

7

2

Plasma Physics Ian Brown

2.1

Introduction

The characteristics of an ion beam are determined by the plasma and the extractor. Thus, for example, the ion beam current is determined by the plasma density, the plasma electron temperature, the extraction voltage, and the extractor geometry. The beam emittance is determined by the plasma density distribution, the plasma ion temperature, and the extractor geometry; and, clearly, the beam composition is determined by the composition of the plasma. The physics of the ion source is thus largely plasma physics. Here we review those principles of plasma physics required for an understanding of ion source performance and behavior. The review presented here is limited and concise; for more detail than is presented here the reader may consult any of a number of excellent texts on the subject [1–6]. Plasma is sometimes referred to as the fourth state of matter, following the solid, liquid and gaseous states. This description refers to the particle energy, or temperature, of the material – as the temperature of a substance is increased, the material changes firstly from solid to liquid, then liquid to gas, and finally from gas to plasma. In a plasma, one or more of the orbital electrons of the atoms are stripped from the nucleus and participate as individual particles. Whereas in a gas the constituent particles are molecules of the gas species, in a plasma the individual particles that make up the plasma are, in general, of three different kinds – ions, electrons and neutrals. Since now some of the particles are charged, as opposed to the neutral particles of an ordinary gas, the kinds of interactions that take place in the plasma state, between the particles, and between the plasma and external fields, are quite different from the gaseous state. Plasmas exist in nature in those environments where the temperature is adequately high, such as in the sun and stars and in the ionosphere, and on the earth in transient forms such as lightning. Man-made plasmas have become commonplace and are part of the modern world in forms such as fluorescent lamps, neon signs and high voltage sparks. Laboratory plasmas can be created in a wide variety of ways, most commonly as electrical discharges of one kind or another. Industrially, plasmas are used in various forms for semiconductor processing, lighting, materials modification and synthesis, and other purposes. In ion sources, plasmas are the medium from which the ions are extracted. The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

8

2 Plasma Physics

In this chapter we survey and summarize the basic features of the plasma medium. Particle density and temperature are defined and related to some of the different kinds of plasmas encountered in ion sources. Then the plasma sheath, or boundary layer, is discussed; the sheath plays an important role in ion beam formation. Next we describe some of the characteristics of plasma behavior in magnetic fields. Finally, ionization phenomena – the ways in which the ionized plasma may be formed from the neutral medium – are outlined.

2.2

Basic Plasma Parameters

The most basic of all plasma parameters are the plasma density and temperature. These concepts in turn relate to fractional ionization of the plasma and to the concepts of particle distribution functions and their means. Particle collisions play an important role in plasma behavior, and the concepts of collisionless and collisional plasma arise. The plasma frequency is another important fundamental parameter. In this section we define and discuss these concepts and parameters. 2.2.1

Particle Density

The constituents of a plasma are ions, electrons, and usually also un-ionized neutrals. Thus one speaks of plasma electron density ne, plasma ion density ni and neutral particle density nn. It is usual to express the density in units of particles per cm3 or perhaps particles per m3. If the ions are all singly ionized, Qi = 1+, and because the plasma is overall chargeneutral the ion density equals the electron density, ni = ne. In the more general case the plasma may contain multiply charged ions, say Qi = 1+, 2+, ... or negatively charged ions Qi = –1; then the ion and electron particle densities need not be the same. Overall charge neutrality is still preserved, however, as expressed by the general condition RQjnj = 0,

(2.1)

where the sum is taken over all charged species including electrons for which Qe = –1. Note that here we write Q for charge state and q for charge, with q = eQ. The term plasma density is often used to mean the ion density or electron density of the plasma, but note that the term is ill-defined except for the case when all ions are singly charged positive ions. In the case of a plasma containing multiply charged ions, it is more appropriate to refer to the electron density. Most laboratory plasmas have densities in the broad range 108 – 1016 cm–3. For plasmas encountered in ion sources the density is generally in the vicinity of 1012 cm–3, basically because of the conditions imposed on the ion source plasma for good beam formation. As a comparison recall that a room temperature gas at a pressure of 1 10–4 Torr (1.3 10–2 Pa) has a particle density of 3.3 1012 cm–3.

2.2 Basic Plasma Parameters

2.2.2

Fractional Ionization

The fractional ionization or percentage ionization of the plasma is defined as the ratio of ion density to total density of ions and neutral particles, fractional ionization = ni/(ni + nn) .

(2.2)

If there are no neutral particles in the plasma then the plasma is said to be fully ionized. The term highly ionized is loosely used to describe plasmas with percentage ionization greater than about 10% or so. Confusion can occur in the use of the term “highly ionized”, since it can be used to refer either to a plasma with a high percentage ionization or to ions that have several electrons removed. The solution is to be aware of this possibility and to use other terminology. The latter might be referred to, for example, as “highly stripped” or “multiply ionized”. 2.2.3

Particle Temperature

The energy of the plasma particles can be described by a temperature. The plasma temperature is commonly expressed in units of electron Volts (eV), where 1 eV = 11,600 K.

(2.3)

Temperature T can be converted to energy kT by multiplication by the appropriate factor. If T is given in K then k = 1.38 10–23 J K–1 (Boltzmann’s constant), and if T is given in eV then k = 1.6 10–19 J eV–1; (electron volts are, of course, already energy units). The ion temperature Ti and the electron temperature Te are not necessarily equal, and if the plasma is in a magnetic field an anisotropy is introduced and the particle temperatures parallel to and perpendicular to the field may be different; then we can have four different particle temperatures: Ti||, Ti^, Te|| and Te^. The neutral component will likely also be at a yet different temperature, Tn. The term plasma temperature is thus not strictly meaningful unless the ion and electron temperatures are equal. Furthermore, the concept of temperature is strictly valid only for Maxwellian energy distributions. Although this covers most plasmas, the concept of temperature is nevertheless also usually loosely extended to describe plasmas that are not in thermal equilibrium. A plasma map can be drawn with density and temperature as axes, on which we can indicate all the different kinds of plasmas, from ionospheric through “worldly” to white dwarfs, and including ion source plasmas and the plasmas of controlled thermonuclear research. Such a map is shown in Figure 2.1. In this very general representation the term plasma temperature is used loosely, since the ion temperature is not always equal to the electron temperature for all the many different kinds of plasmas shown. The plasma might possess not only energy due to the random thermal motion of its particles, but also energy due to bulk mass flow. In this case the plasma is said to

9

10

2 Plasma Physics

Figure 2.1

n–T plasma map.

have a directed energy, or a drift energy, or a streaming energy, and it is important not to confuse the plasma drift energy with the plasma temperature. For example a plasma gun might produce a puff of plasma with drift energy of perhaps hundreds of electron volts, but the temperature would likely be relatively cold, perhaps of order 1 eV. More usually the plasma has little or no drift energy, and the particle energy is purely thermal. A cold plasma, such as could be produced in a surface ionization plasma source [5, 7] for example, might have a temperature of about 0.2 eV, which, note, is however over 2000 K. Arc-produced plasmas might have ion temperatures ~1 eV and electron temperatures somewhat higher, perhaps several eV, depending on the kind of arc and the discharge parameters. In microwave-produced plasmas the energy input goes directly into the electrons, and at low pressure, for example for electron cyclotron resonance (ECR) discharges, see Ref. [8] and Chapter 11, Te can be over 1 keV while the ions remain cold. It is usually desirable for ion source plasmas to have an ion temperature that is as low as possible. The reason for this is that the ion beam emittance is determined, in large part, by the transverse plasma ion temperature, and a lower temperature leads to lower emittance – a “tighter” beam. 2.2.4

Particle Energy and Velocity

Plasma particle motion can be described by distribution functions. For the case when a plasma species is in thermal equilibrium, its distribution is Maxwellian. Here we summarize the Maxwellian distribution functions of velocity, speed, and energy, and the various averages that can be obtained from them. The velocity distribution function f(v) describes the number of particles in a given velocity interval, dn = f(v)dv = f(vx, vy , vz)dvx dvy dvz. For a plasma in thermal equilibrium the distribution function is Maxwellian, given by

2.2 Basic Plasma Parameters

f ðvx ; vy ; vz Þ

m 3 2 n e 2pkT

¼

2 mðvx

2 vy

þ þ 2kT

2 vz Þ

(2.4)

where m is the particle mass and k is Boltzmann’s constant. The 3-dimensional Maxwellian velocity distribution function is a product of three independent 1-dimensional velocity distribution functions, each of which is of the form 2 m 1 mvx 2 2kT e . (2.5) f ðvx Þ ¼ n 2pkT The distribution functions of speed F(v) and of energy f(E) can be obtained from the velocity distribution function and are given by 2

¼

4pv f ðvÞ

¼

m 3 mv 2 4pv n e 2kT , 2pkT

¼

FðvÞ

¼

nð4=pÞ

FðvÞ

2

2

(2.6)

and f ðEÞ

dv dE 1=2

ðkTÞ

3=2

E

1=2

e

E=kT

.

(2.7)

The mean particle speed and mean energy can be obtained from the distribution functions and are given by rffiffiffiffiffiffiffiffi 8kT ¼ v , (2.8) pm and E

¼

3 kT. 2

(2.9)

In an isotropic plasma the kinetic energy is divided equally between the three degrees of freedom, x E

¼

y E

¼

z E

¼

1 kT. 2

(2.10)

The mean thermal speeds of electrons and ions can be written as e v

¼

pffiffiffiffiffi 67 Te cm ls–1,

(2.11)

¼

rffiffiffiffiffi Ti cm ls–1, 1:57 A

(2.12)

and i v

11

12

2 Plasma Physics

where Te, Ti are in eV and A is the ion mass in amu. Sometimes other slightly different expressions are used for particle speeds. That given above is the mean, Eq. (2.8), as determined from the distribution function Eq. (2.6). The rms (root mean square) velocity is obtained from 1 2 mv 2

3 kT, 2

¼

(2.13)

and is given by ¼

vrms

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3kT=m.

(2.14)

Another speed is the most probable particle speed, that speed vo for which the distribution function F(v) is maximum. It is given by vo

¼

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2kT=m.

(2.15)

, vrms and vo are in the ratio (8/p): These three different possible particle speeds v 3:2 = 1:1.09:0.89, and the uncertainty introduced by the use of one rather than another is unimportant to virtually all laboratory applications. 2.2.5

Collisions

Collisions between the charged particles in a plasma are fundamentally different from collisions between the neutral particles in an ordinary gas. Energy and momentum are exchanged mostly via a large number of distant encounters, rather than by single close encounters. Thus, for example, the velocity vector of a test particle is altered in magnitude and direction through a random walk of many small steps. In this situation the concepts of collision time and mean free path, which have a clear intuitive meaning for the case of neutral particle collisions, need a new definition. This is the relaxation time. The relaxation time is the time required for collisions to make a major change in the distribution function. For example the angular relaxation time sh can be defined as the time for a particle to be deflected through an angle of 90 by the sum effect of many distant encounters; this might also be called the 90 deflection time. Depending on the kind of plasma (constituent particles, anisotropies, different species at different temperatures, etc), there can be many different kinds of relaxation times – between like particles and unlike particles (ions, electrons), change in speed or exchange of energy or momentum, perhaps between species with different properties in different directions (parallel or perpendicular to an applied magnetic field), and perhaps between hot and cold components, etc. Thus one speaks of the energy relaxation time, meaning the time to exchange energy from one component of the plasma to another; for example, if there are hot electrons in a cold ion background then the ions will be heated by electron-ion energy transfer collisions and the ion temperature will rise asymptotically toward a new equilibrium temperature at a rate characterized by the electron-ion

2.3 The Plasma Sheath

energy relaxation time sEei. For simplicity and as a residue from the more familiar case of neutral particle encounters, it is nevertheless common to speak loosely of collision time, with the implicit understanding that what is really meant is relaxation time. The mean-free-path k is related to the cross section r for a given process by k = 1/nr,

(2.16)

where n is the appropriate particle density. The collision time s is related to the mean-free-path and cross section by s = k/v = 1/nrv,

(2.17)

where v is the appropriate mean particle velocity. The collision frequency m is the reciprocal of the collision time, m = 1/s = nrv.

(2.18)

There are a number of different energy exchange times and collision times corresponding to interactions between like and unlike particles and in various parameter ranges. For ion source plasmas these usually lie in the broad range of nanoseconds to milliseconds. Expressions for these parameters can be found in the plasma texts [1–6,9].

2.3

The Plasma Sheath

The plasma particles – the ions and electrons – are charged particles, and therefore interact with one another at a distance via their electric and (if in relative motion) magnetic fields. There is a scale length for the plasma which defines the distance over which the electric field of a test particle extends; field particles closer than this distance know of the test particle, and those beyond it do not. This same phenomenon takes place when an electric field is imposed on a plasma from an external source – the field is shielded from the interior region, quite similarly to the way a metal excludes electric fields from its interior. The origin of the shielding lies in the charged particle redistribution that takes place in response to the applied field. This distance is called the shielding or the screening distance – the distance it takes for the plasma to effectively screen out, or shield itself from, an applied electric field. The electric field is attenuated exponentially over a distance determined by the plasma parameters.

13

14

2 Plasma Physics

2.3.1

Debye Length

A boundary layer is formed at the interface between a plasma and a material wall surrounding the plasma or placed in the plasma. The faster moving (usually) electrons leave the plasma at a greater rate than the more massive ions and the wall acquires a negative charge with respect to the plasma; viewed the other way around, the plasma acquires a positive potential with respect to the wall that confines it. Thus it is usual (but not universal) that the equilibrium potential of a plasma is positive. The boundary layer is called the plasma sheath. Its physical basis is the same as that which determines the shielding distance described above, and the sheath thickness is the same as the shielding distance. These phenomena were first investigated by Peter Debye some 80 years ago [10], and the characterizing scale length is commonly called the Debye length, kD, given by the expression kD

¼

eo kTe , 2 e ne

(2.19)

kD

¼

rffiffiffiffiffi Te 743 , ne

(2.20)

2

or

where in the second expression Te is the electron temperature in eV, ne is the electron density in cm–3, and kD is the Debye length in cm. The magnitude of the sheath drop, i.e., the plasma potential relative to the wall, depends on the precise plasma configuration and parameters but is typically several (3 or 4) times the electron temperature (expressed in volts). A floating probe or electrode inserted into a plasma will normally assume a potential, called the floating potential, that is negative with respect to the plasma by ~(3–4)kTe; one can look on this as being due to an excess electron flux to the probe, with respect to the ion flux, because of the higher electron velocity. Similarly, floating electrodes within an ion source will assume a potential that depends on the surrounding ion and electron flow. Within the plasma sheath the plasma is not charge neutral. This is the region where the plasma particles (mostly the electrons, because of their higher mobility) assume a distribution such as to cancel out the external field or to establish the equilibrium transition layer between the plasma and its boundary. The sheath does not have a sharp boundary, but is an exponential fall-off of uncompensated charge density. The 1/e width of this transition region is the Debye length kD. These concepts of plasma self-shielding, field penetration, and plasma/electrode sheaths are critical to the consideration of ion beam formation at the ion source extractor grids, where one wants a well-defined plasma boundary to form and the electric field applied by the grids to accelerate the ions into a useable beam. The potential distribution within the plasma and sheath region is depicted in Figure 2.2. The Debye length can also be thought of as the distance over which local spacecharge fluctuations cancel out. For dimensions within the plasma greater than kD charge neutrality is preserved to a high degree and local fluctuations in potential are

2.3 The Plasma Sheath

Figure 2.2 The plasma sheath: plasma potential vs. distance from within the plasma volume to the wall.

small, whereas for dimensions small compared to kD there can be spatial and temporal fluctuations in charge neutrality that are not small. The Debye length is the appropriate scale of the boundary or transition region for the case when the wall or electrode assumes its potential only via the plasma, and no additional potential is applied to it. But it is intuitively obvious that if a high voltage is applied to an electrode in a plasma then the sheath must be thicker than for the unbiased case when the plasma-electrode potential (i.e., the sheath drop) is just a few kTe. In this case, the high voltage sheath thickness is greater than the Debye length by (approximately) the root of the ratio of applied voltage to electron temperature, rffiffiffiffiffiffiffi eV . (2.21) dsheath kD kTe This has considerable application to laboratory experiments through such things as probe sizes, dimensions of wire meshes, extractor hole sizes and separation, etc. 2.3.2

Charge Neutrality

Quasi neutrality is a basic plasma property. Over distances large compared to the Debye length, the plasma maintains approximate charge neutrality to a very high degree. Any departure from neutrality gives rise to electric fields which limit the charge build-up. The quasineutrality condition can be written as RQieni – ene = 0,

(2.22)

15

16

2 Plasma Physics

where Qi is the ion charge state (charge qi = eQi) and the sum is taken over all ion species. In the most general case there may be multiply stripped ions and also both positive and negative ions (although the simultaneous presence of both multiplycharged positive ions and also negative ions in the same plasma is improbable). In the simple, but common, case when the ions are positive and singly charged, then ni = ne. The plasma cannot depart from quasineutrality because of the electric fields that would then arise. Random fluctuations of electron density in a plasma can lead to fluctuating electric fields – plasma noise. Potentials of magnitude ~kTe can exist over scale lengths ~kD. 2.3.3

Plasma Oscillations

The discussion immediately above about small departures from charge neutrality causing a restoring force (electric field) which limits the charge accumulation leads naturally to the idea of plasma oscillations. Plasmas have a number of natural modes of oscillation, one of the most fundamental of which is the electron plasma frequency. The (radian) electron plasma frequency is denoted by the symbol xpe and is given by 2

xpe

2

¼

e ne , eo me

(2.23)

where eo is the permittivity of free space. The ions can also oscillate at their own natural frequency, as for example in standing acoustic waves. This frequency is called the ion plasma frequency, xpi, and is given by 2

xpi

2 2

¼

Q e ni eo mi

(2.24)

where Q is the ion charge state. These expressions can be written in the convenient forms fpe

¼

pffiffiffiffiffi 8980 ne (Hz),

¼

210Q

(2.25)

and fpi

rffiffiffiffiffi ni (Hz), A

(2.26)

where Q is the ion charge state, ne, ni are the electron and ion densities in cm–3 and A is the ion mass in amu. Often the electron plasma frequency is referred to simply as the plasma frequency. Usually the electron plasma frequency is in the microwave band, some GHz; the ion plasma frequency is usually in the rf band, low MHz.

2.4 Magnetic Field Effects

2.4

Magnetic Field Effects

Charged particles in motion experience a force in a magnetic field, and so also do the plasma ions and electrons. In a magnetized plasma, the motion of the ions and electrons is circular in the plane perpendicular to the magnetic field (i.e., the transverse direction) and is unmodified in the direction parallel to the field (i.e., the longitudinal direction). Apart from collisions with other plasma particles, the resultant particle motion is helical, a combination of the transverse circular motion and longitudinal constant velocity. The ions and electrons act, to some extent, as though they are tied’ to the field lines. The behavior of a plasma in a magnetic field can be profoundly different from a plasma in the absence of a magnetic field. 2.4.1

Gyro Orbits

A particle with charge q = eQ and velocity v moving in a magnetic field of flux density B experiences a force F given by

F = eQv B.

(2.27)

This is called the Lorentz force. It is directed radially inward on the particle as it moves with perpendicular velocity v^ in a circular motion about the parallel component of magnetic field B||. The velocity component v|| has no interaction with the magnetic field (v|| B|| = 0), and thus the general trajectory of a charged particle in a magnetic field is helical. Equating the Lorenz force to the centripetal force, an expression for the radius of the circular orbit is obtained,

¼

mv? eQB

(2.28)

This is called the cyclotron radius or the gyro-radius of the particle’s orbital motion in the magnetic field. The cyclotron radius can be expressed in terms of the particle temperature via the equality E^ = 12mv^2 = kT^. Recall that temperature is not a single-particle parameter but describes a Maxwellian distribution of velocities, so the new gyroradius is an average over the distribution. To reasonable accuracy the ion gyro-radius can be written as pffiffiffiffiffiffiffiffi ATi (cm), (2.29) i ¼ 0:0014 QB where A is the ion mass in amu, Ti is the perpendicular ion temperature in eV, Q is the ion charge state, and B is the magnetic flux density in Tesla. Similarly the electron gyroradius can be written as pffiffiffiffiffi Te (cm), (2.30) e ¼ 0:00033 B

17

18

2 Plasma Physics

where Te^ is the perpendicular electron temperature in eV and again B is in Tesla. For parameters generally encountered in ion source plasmas i is typically of the order of several millimeters and e of the order of a tenth of a millimeter. 2.4.2

Gyro Frequencies

The frequencies with which the ions and electrons gyrate around the field lines are important parameters. Equating the centripetal force mx2r to the Lorenz force qxrB yields an expression for the (radian) cyclotron frequency, or gyro-frequency, xc

¼

qB , m

(2.31)

where for ions q = Qie and m is the ion mass mi, and for electrons q = e and m is the electron mass me. One can write the ion cyclotron frequency in convenient form as fci = 15.2 QB/A (MHz),

(2.32)

and the electron cyclotron frequency as fce = 28 B (GHz),

(2.33)

where again Q is the ion charge state and B is in Tesla. For magnetic fields of order 1 kG (100 mT), as commonly encountered in ion source plasmas and other small laboratory experimental plasmas, the ion cyclotron frequency is typically in the low to fractional megahertz range and the electron cyclotron frequency in the low microwave band. The charged-particle circular motion carries a sign corresponding to the sign of the charge q, that indicates the sense of the particle rotation about the magnetic field lines. The electron gyromotion is right-handed, and the ion gyromotion is lefthanded. These senses become important for the coupling of some kinds of waves into the plasma – axially propagating waves (of the right frequency) can couple power into the plasma via the electrons if the wave is right-hand polarized or into the ions if the wave is left-hand polarized. The electron component of a plasma can be heated by coupling into it microwave power at the electron cyclotron frequency – the power transfer is resonant, and the scheme is referred to as electron cyclotron resonance heating (ECRH). Plasmas formed and heated in this way are called ECR plasmas. The electrons can be heated to very high energies, and hot electron temperatures in ECR plasmas can readily reach the 10 keV range (the term “temperature” here is used loosely, as the hot electron energy distribution in ECR plasmas is not Maxwellian). For this to be possible it is necessary that the electron collision frequency be small compared to the electron cyclotron frequency, i.e., that the residual gas pressure be not too high; typically, the pressure is in the 10–6–10–5 Torr range. Because of the high electron energy, very highly stripped ions are produced in these kinds of plasma. This approach can

2.4 Magnetic Field Effects

be embodied in an ion source configuration and the ions extracted as a beam, so forming an ECR ion source. Microwave power is often used for plasma formation at higher gas pressures also, say in the broad range 10–3–102 Torr. In this case the coupling is not resonant, since collisions effectively prohibit any ordered electron cyclotron motion. Nevertheless a microwave produced plasma can be formed, with or without magnetic fields present. The terminology for this kind of plasma is still evolving, but one description that has been used and that fits quite well is “high pressure microwave discharge”. Whereas ECR plasmas are typified by high electron temperature and high ion charge states, high pressure microwave plasmas are usually of low electron temperature and low ion charge state. In the simplest case it can be shown that the resonant absorption of microwave power at the electron cyclotron frequency can occur only for plasmas with electron density lower than a certain critical density, ncrit, also called the cutoff density: n < ncrit. Otherwise the microwave power is not transported into the plasma but is reflected. This condition can also be expressed as a condition on the microwave frequency, which must be greater than the electron plasma frequency of the plasma: x > xpe. The wave frequency which is equal to the plasma frequency for a given density is called the cutoff frequency or critical frequency. For a given microwave frequency the critical density is given by ncrit = 1.25 1010 f 2 (cm–3),

(2.34)

where the microwave frequency f is expressed in GHz. In spite of this limit, however, it has been found that under some conditions (having to do with mode conversion, nonlinear effects, and/or finite plasma effects) microwave-produced plasmas can be formed with density greater than the critical density. Such plasmas are called overdense. 2.4.3

Magnetic Confinement

That the ions and electrons in a plasma are tied in their orbital motion to the field lines provides a means of confining the plasma in the direction transverse to the field. Plasma loss along the field can be reduced by increasing the field strength at the ends of the confinement region. In this case the plasma is confined, albeit imperfectly, both longitudinally and transversely – a simple magnetic confinement geometry has been established. The region of increased magnetic field strength is often called a magnetic mirror’, because the plasma particles (some of them) are reflected back into the central plasma region rather than being lost from the plasma. In another magnetic geometry that has become widespread in recent years, a surface array of individual permanent magnets, usually of the high-field rare earth kind (Sm–Co or Nd–Fe), is used to establish a field (essentially a large number, a checkerboard pattern, of magnetic mirrors) that is appreciable (typically of order 1 kG) near the wall and negligible in the interior region; this is called a magnetic multipole confinement geometry, or more vividly a magnetic bucket.

19

20

2 Plasma Physics

The essence of magnetic mirroring of plasma particles lies in the transfer of their kinetic energy between the transverse and longitudinal directions. In the absence of collisions or electric fields, the particle energy is given by E = E^ + E||, where E^ = 12mv 2 and E = 12mv 2. As a particle, either ion or electron, moves in its helical ^ || || trajectory into a region of higher field strength, energy is transferred from E|| to E^; the parallel velocity is decreased until, at the mirror point, the parallel velocity is reduced to zero and the particle reflects. It is as if the particle experiences a force, and indeed the situation can be treated in this way; in a non-uniform magnetic field the particle experiences a force given by lgradB, where l is the magnetic moment of the particle, defined as l = 12mv^2/B. All these ideas are valid only if the particles are free to execute their gyromotion. Often this condition is not met because of particle collisions. If an ion suffers a collision, on the average, in a time short compared to the ion cyclotron period, then it is clear that no gyromotion can be sustained; ion transport is dominated in this case by collisional processes and the magnetic field will have little effect. The ions can experience the magnetic field only if the ion collision frequency is small compared to the ion cyclotron frequency, mi N. The matched case is the most interesting case for accelerator injection application. However, it has been shown that high emission current density and low emittance do not contradict. For the case of maximum divergence angle, the emission current density could be increased in favor of small emittance. The maximum attainable ion beam current is given by the geometry of the extraction system together with the maximum electric field strength. Therefore, every extraction system can be characterized by the extraction current.

Figure 5.17

Rms emittance as a function of the extracted ion beam current.

75

76

5 Ion Extraction

Figure 5.18 Normalized rms emittance as a function of the aperture radius for extraction in the matched case.

It is interesting to consider the influence of different extraction systems, with their maximum specific ion beam current, on the emittance. The maximum field strength is a function of the gap distance d, and for constant aspect ratio, S, also a function of the aperture radius r. Thus the normalized emittance is taken into account in the following considerations. To compare the simulations in a reasonable manner the aspect ratio S, the web width, and the shape of the electrodes will be kept constant. Figure 5.18 shows the normalized rms emittance as a function of the aperture radius r for ion extraction in the matched case. For a given extraction system the maximum ion beam current is fixed. The maximum applied voltage U is proportional to d0.5, and the scale factor b~ is proportional to U0.5 (Eq. 5.15), so it is obvious that the normalized emittance is proportional to r0.5. In order to analyze the influence of ion temperature on beam emittance, the ion temperature is varied from 0 eV to 1 eV. The extraction system is the same as shown in Figure 5.11. The emittance increases by a factor of 4.3 for an increase of ion temperature from 0.05 eV to 1 eV, as shown in Figure 5.6. The dependence of the emittance on the ion temperature scales with T0.5, where T is the ion temperature (Eq. (5.18)) [19]. However, aberrations in the extraction system are not taken into account. As a result, for most extraction systems, the beam emittance is dominated by ion temperature if no external magnetic fields are applied.

5.5

Multi-Aperture Extraction Systems

Some applications call for the use of multi-aperture extraction systems. The reasons for this could be: .

The plasma generator is not capable of increasing the emission current density up to a value required for the application.

5.5 Multi-Aperture Extraction Systems

The mass-to-charge ratio is too high to extract the desired ion beam current with feasible extraction voltages. Extraction of a given ion beam current at lower extraction voltage. Extraction of a given ion beam current with a smaller emission area. Extraction of a higher ion beam current with the same emission area.

.

. . .

Multi aperture extraction systems are often used for high mass-to-charge ratio ions. However, the emittance of the ion beam is much larger. Assume a desired ion beam current of 35 mA Bi+ for the HIDIF/HIF scenario [2] and a plasma generator which can deliver 80% of singly-charged Bi+ ions, and 20% of Bi2+ with emission current densities of hundreds of A/m2 [20, 21]. Further, there is no limitation on extraction voltage. How would a single-aperture extraction syscm 3. 2. 1. .0 -1. -2. -3. .0

1.

2.

3.

4.

5. cm

AXCEL-INP simulation for extraction of 35 mA Bi+ and 8.75 mA Bi2+ with a single aperture extraction system. Figure 5.19

mrad 300 200 100 0 -100 -200 -300 -1.4 -1.

-0.6 -0.2

.2

.6

1.

1.4 cm

Figure 5.20 Emittance pattern of the trajectory plot of Figure 5.19 (e100% = 3000 p mm mrad, erms = 367 p mm mrad).

77

78

5 Ion Extraction

tem be configured? Figures 5.19 and 5.20 show the results of an AXCEL-INP simulation for an extraction system which is favorable for 35 mA Bi+ and 8.75 mA Bi2+. The aperture radius r of the plasma electrode is 21 mm and the aspect ratio S is 0.5. In this case the extraction voltage is 85 kV. The screening electrode is at –8 kV. These values are relatively high for a single gap structure and therefore not very feasible. This large required extraction voltage could provide a reason for designing an alternative multi-aperture system. An important question is: what is the relation between the emission areas F1 and Fn for a single-aperture and for an n-aperture extraction system with the boundary condition of constant ion beam current. Eq. (5.23) gives us the emission current density, and Eq. (5.25) the maximum gap voltage according to Kilpatrick. From the equality of ion beam currents I1 = I2, and since also the aspect ratio S is the same for both cases, it follows that 3=2

F1

3=2

U1 ¼ Fn Un2 , 2 d1 dn 2 d1 2 dn r1 2 ¼ nrn 2 , d1 dn 2

(5.28)

2

S d1 ¼ nS dn , r1 ¼ nrn : The emission current areas are 2

F1 ¼ pr1

(5.29)

and 2

Fn ¼ nprn :

(5.30)

The relationship between them is 2

2

F1 pr1 r12 ¼ n : ¼ 2 ¼ r1 Fn nprn n 2

(5.31)

n

In the case of the same emission areas F1 = Fn it is important to know the relationship between the ion beam currents, F1 ¼ Fn , r1 ¼

pffiffiffi nr n .

(5.32)

For the ion beam currents I1 and In 3=2 2=3 d p1ffiffi 3=2 3=2 n U1 Un F F F n 2 2 1 pffiffiffi In n 1 d21 d1 dn ¼ ¼ n ¼ pffiffiffi ¼ n. 3=2 3=2 U1 U1 I1 n U13=2 F1 2 F1 2 F1 2 d1

d1

d1

(5.33)


This relationship is valid for n holes. We can also calculate the relationship for one of the n holes using Kilpatrick’s law: 3=2

I 1 / F1

2

U1 2 d1

¼

pr1 2 d1 ¼ Sr1 . d1

(5.34)

From Eq. (5.32) In;one hole I1

¼

Srn 1 ¼ pffiffiffi. Sr1 n

(5.35)

For n holes we have to multiply Eq. (5.35) by n: I pffiffiffi In ¼ n n;one hole ¼ n. I1 I1

(5.36)

With the boundary conditions of the same emission current densities j1 = jn and the same emission areas F1 = Fn we can reduce the extraction voltage by a factor of 4=3 3=2 3=2 Un U1 Un rn ¼ , ¼ : (5.37) 2 2 U1 r1 dn d1 This relation is independent of Kilpatrick’s law. All these calculations can be performed considering Coupland’s law that the maximum feasible voltage is proportional to the square root of gap distance. For the same emission areas F1 = Fn and aspect ratio the relationship between currents is given by 1=2 3=2 5 d 5 1=2 3=2 610 p1ffiffi n 610 dn 3=2 3=2 2 F1 F 1 2 n U1 Un r rn 1 F 2 3=8 1 2 In Fn d2n 5=8 S d1 nS ¼ ¼ ¼n ¼n : (5.38) 3=2 ¼ 3=2 3=2 3=2 U1 U1 U1 U1 I1 F1 2 F1 2 F1 2 F1 2 d1

d1

d1

d1

Eq. (5.38) is valid for n holes. For one of the n holes of the extraction system we find 3=4 3=2

In;one hole I1

¼

Fn F

Un

2 dn 3=2 U1 1 2 d1

¼

5=4 3=4 rn 5=4 3=4 S r1

S

¼

r1 pffiffi n

r1

¼ n

3=8

.

(5.39)

For the case of n holes we have to multiply Eq. (5.39) by n, I In 5=8 ¼ n n;one hole ¼ n . I1 I1

(5.40)

Calculation of the relationship between the emission areas Fn and F1 according to Coupland’s law considering the same ion beam currents yields

79

80

5 Ion Extraction

¼ I1

In 3=2 Fn Un2 dn 3=4 2 dn nrn 2 dn 2 3=4 nS dn 4=3 n rn

¼ ¼ ¼

,

3=2 U F1 12 d1 3=4 2 d1 r1 2 d1 2 3=4 S d1

, (5.41)

, ,

¼ r1 :

The ratio of the emission areas is given by 2

2

F1 r r 5=3 ¼ 12 ¼ 1 2 ¼ n . r1 Fn nrn n 4=3

(5.42)

n

Table 5.2 gives an overview of all these results. Relationships between aperture radii, ion beam currents, and emission areas for single aperture and multi-aperture extraction systems.

Table 5.2

F1 = Fn

I1 = In j1 = jn, F1 = Fn

Kilpatrick’s law

Coupland’s law

rn = n–1/2 r1 In = n1/2 I1 (for n holes) In = n–1/2 I1 (for one of n holes) rn = 1/n r1 Fn = 1/n F1 Un = (rn/r1)4/3 U1

rn = n1/2 r1 In = n5/8 I1 (for n holes) In = n–3/8 I1 (for one of n holes) rn = n–4/3 r1 Fn = n–5/3 F1 Un = (rn/r1)4/3 U1

Now let us create, step-by-step, a multi-aperture extraction system for a 35 mA Bi+ ion beam. As a consequence of the relationships in Table 5.2, the emission area for a multi-aperture system is smaller than the area for a single aperture. Thus for a multi-aperture system we need to consider a larger emission current density from the plasma generator Concerning the geometry, we note that the arrangement of apertures strongly affects the beam emittance. The objective might be to increase the brightness of the ion beam by filling up the phase space as much as possible. In this example it is favorable to use small aperture radii. The apertures are arranged as close to each other as possible. In this way the radius of the circle that encloses all the apertures is also minimized. Some different multi-aperture extractor geometries are shown in Figure 5.21. However, for some applications very large multi-aperture extraction systems with thousands of apertures are used [15]. The minimum distance between the holes (web width) is given by the thermal load of the material and by the geometry. Thermal forces are caused by the plasma generator itself, e.g. arc power density, and electrons formed between the plasma and the screening electrode. These are accelerated backwards to the plasma electrode and thus lead to heating. Cole et al. [22] estimated that 1% of the beam energy is deposited on the plasma electrode.


Figure 5.21 Plasma electrode of three multi-aperture extraction systems with web widths of 4 mm.

In the following the geometric parameters of a seven-hole system with a singlehole radius of 3 mm are described. In this case the emission area is 2 cm2. The web width between the inner and outer apertures is 4 mm, hence the radius of the circle that encloses all apertures is 10 mm. The smallest radius which includes all the holes is 13 mm. The transparency is 37%. Based on this seven-hole system one can calculate the parameters of other multi-aperture systems that deliver the same ion beam current of 35 mA Bi+ and 8.75 mA Bi2+. Geometric parameters of single-aperture and multi-aperture systems with constant web width of 4 mm.

Table 5.3

Holes

1

7

13

19

Emission area [cm2] Radius [mm] Transparency [%]

14 21.1 100

2 3 37.7

1.07 1.62 17

0.74 1.11 12.8

The transparency decreases dramatically with increasing number of holes with constant web width. The same calculations with a fixed relationship between web width and radius for the seven–hole system can be performed using Table 5.2. These results are shown in Table 5.3. Table 5.4 summarizes the transparencies and web widths for multi-aperture systems with a constant relationship (web width to radius) of 1.33. Transparency of single- and multi-aperture systems with constant web width to radius ratio.

Table 5.4

Holes

1

7

13

19

Web width [mm] Transparency [%]

– 100

4 37.7

2.16 28.3

1.48 32.5

81

82

5 Ion Extraction

Figure 5.22

Planar (left) and spherical (right) extraction of a multi aperture extraction system.

For the seven-hole system the plasma generator has to deliver 175 A/m2, for a 13hole system 330 A/m2, and for a 19-hole system 473 A/m2. This means that the choice of the right multi-aperture extraction system is not an option in general but depends on the performance of the plasma generator. Now we discuss the beam emittance of a multi-aperture extraction system. For accelerator applications the effective emittance is a most important parameter. In the following we will estimate the effective emittance of a seven-aperture system. Two types of high-brightness seven-hole systems (or in general, multi-aperture systems) are possible: planar and spherical. For planar systems all the beamlets are parallel to the beam axis whereas the beamlets for spherical systems are focused. A spherical system might be useful for direct injection into an accelerator system; for example an RFQ requires a convergent entrance beam. Figure 5.22 shows a cross-sectional view of a planar and a spherical seven-hole system. For the case of planar extraction the outer holes are at a distance from the beam axis; for spherical systems the angle to the beam axis is j. The focal point L of a spherical system is L ¼

R , tanjmax

(5.43)

where R is the radius of the outer hole circle and jmax the maximum angle to the beam axis. For an emittance measurement in the focal plane (only for spherical systems) the emittance pattern is shown schematically in Figure 5.23, together with a similar pattern for a planar extraction system. The single-hole ellipses of the spherical system have the same location in x or y but different angles x¢ and y¢. The single-hole ellipses of the planar system have the same angles in x¢, y¢ but different locations in x and y. Note that some of the single-hole emittance ellipses are formed by two holes. For both systems the emittance is given by: en ¼

P

esingle .

n

For the case of identical beamlets

(5.44)

5.6 Starting Conditions

Figure 5.23 Emittance patterns of multi-aperture extraction systems. Left: planar extraction system. Right: spherical extraction system.

en ¼ n esingle .

(5.45)

The effective emittance is given by the parallelograms in Figure 5.23. For the planar system 0

eeff ;planar ¼

4x þ esingle p

(5.46)

and for the spherical system eeff ;spheric ¼

4xjmax þ esingle . p

(5.47)

5.6

Starting Conditions

The temperature and drift energy of the particles have great impact on ion beam formation by the extraction system. For example, the ion energy in the longitudinal direction influences the extractable emission current density. The temperature and energy of the particles depend on the plasma formation process. The ion energy varies from tenths of eV in filament driven ion sources to tens of eV in vacuum arc ion sources, and up to keV for laser ion sources. Besides this initial ion drift energy due to the plasma generation itself, it is also possible to preaccelerate the ions by biasing the plasma electrode negatively with respect to the plasma chamber. Figure 5.24 shows the results of an AXCEL-INP simulation of the extraction of protons, calculated in Figure 5.14 as a function of the initial ion velocity. The applied voltage is 55 kV. Calculations were performed without any space charge in order to explore the effect on focusing forces with increasing ion velocity. In Figure 5.24 the starting velocity is 43771 m s–1, corresponding to an initial ion energy of 10 eV, generated for instance by a filament driven multi-cusp ion source with a plasma wall

83

84

5 Ion Extraction

v=43771m/s, E=10eV

Figure 5.24 Trajectory plot for protons with initial longitudinal energy 10 eV. In this AXCEL-INP simulation space-charge is not taken into account, so as to analyze the focusing forces.

v=437711m/s, E=1keV

Figure 5.25 Trajectory plot for protons with initial longitudinal energy 1 keV. In this AXCEL-INP simulation space-charge is not taken into account, so as to analyze the focusing forces.

v=43771m/s, E=10eV

Figure 5.26

Trajectory plot for protons with longitudinal initial energy of 10 eV.

5.6 Starting Conditions

v=138416m/s, E=100eV

Figure 5.27

Trajectory plot for protons with longitudinal initial energy of 100 eV.

v=437711m/s, E=1keV

Figure 5.28

Trajectory plot for protons with longitudinal initial energy of 1 keV.

potential hump of 10 V. Figure 5.25 shows the calculation for 1 keV initial ion energy (437711 m s–1) such as produced by a laser ion source. With increasing initial longitudinal velocity the focusing forces decrease. Figures 5.26, 5.27 and 5.28 show trajectory plots taking space-charge into account. The initial energy is again varied from 10 eV to 1 keV. The plasma meniscus becomes more concave with increasing initial energy, resulting in a higher emission current density because of the reduced space-charge in the vicinity of the plasma meniscus. Thus the extractable emission current density is higher for vacuum arc ion sources or laser ion sources compared to the case of filament driven ion sources. In all cases the longitudinal ion energy spread, the transverse energy, and the temperature are neglected.

85

86

5 Ion Extraction

References [1] IFMIF Final Report, IFMIF CDA-Team, edi-

ted by M. Martone, Centro Ricerche Frascati, Roma (1996). [2] The HIDIF Study, Report of the European Study Group on Heavy Ion Driven Internal Fusion for the Period 1995–1098, GSI-Report (1998). [3] S.A. Self, Phys. Fluids 6, 1762 (1963). [4] C.D. Child, Phys. Rev. (Ser. 1) 32, 492 (1911). [5] I. Langmuir and K.T. Compton, Rev. Mod. Phys. 3, 251 (1931). [6] P. Spdtke, INP, Junkernstrasse 99, 65205 Wiesbaden, Germany. [7] J. Liouville, J. Math. 3, 324 (1838). [8] F.J. Sacherer, IEEE Trans. Nucl. Sci. NS-18, 1105 (1971). [9] J. Guyard and M. Weiss, CERN/PS/LIN 76-3, CERN, Switzerland, (1976). [10] P. Lapostolle, Proceedings of the International Conference on Ion Sources, Vienna, Austria, (sterreichische Studiengesellschaft fr Atomenergie, Vienna, 1972). [11] P. Allison, J.D. Sherman and H.V. Smith, Report LA-8808-MS, Los Alamos Nat. Lab. (1981). [12] J.R. Pierce, Theory and Design of Electron Beams (Van Nostrand, Toronto, 1954). [13] R. Becker, Rev. Sci. Instrum. 67, 1132 (1996).

[14] J.R. Coupland, T.S. Green, D.P. Hammond

[15] [16] [17] [18]

[19]

[20]

[21]

[22]

and A.C. Revire, Rev. Sci. Instrum. 44 , 1258 (1973). I.G. Brown, The Physics And Technology of Ion Sources (Wiley, New York, 1989). W.D. Kilpatrick, Rev. Sci. Instrum. 28, 824 (1957). R. Hollinger, K. Volk and H. Klein, Rev. Sci. Instrum. 73, 1027 (2002). R. Hollinger, P. Beller, K. Volk, M. Weber and H. Klein, Rev. Sci. Instrum. 71, 836 (2000). P. Allison, J. Sherman, Comparison of Measured Emittance of an H– Ion Beam with a Simple Theory, CIC-14 Report Collection, LA8808-MS, Los Alamos Scientific Laboratory (1981). M. Weber, K. Volk, P. Beller, A. Lakatos, A. Maaser and H. Klein, Rev. Sci. Instrum. 69, 1066 (1998). M. Weber, K. Volk, P. Beller, R. Hollinger, A. Maaser and H. Klein, Nucl. Instrum. Methods Phys. Res. A 415, 339 (1998). H.C. Cole et al., Electron Heat Dissipation Limits on Multi Aperture Ion Source Performance, Proceedings of the Second International Conference on Ion Sources, Wien (1972).

87

6

Beam Transport Peter Spdtke and Ralph Hollinger

6.1

Introduction

Any assembly of particles can be described by the sum of the coordinates of each particle within the six-dimensional phase space at a given time t. The six dimensions are given by the three space coordinates x, y, z and the three momenta px , py , and pz : n

Lbeam ¼

X ðxi ; yi ; zi ; pxi ; pyi ; pzi Þt :

ð6:1Þ

i¼1

As long as these dimensions are linearly independent, the coordinates of each particle can be projected into sub-spaces and treated separately. For a description of the beam transport the two projections n X Lx ¼ ðxi ; pxi Þ

and

n X Ly ¼ ðyi ; pyi Þ

i¼1

ð6:2Þ

i¼1

are called the transverse phase spaces. The longitudinal phase space is not of interest for cw beams, as long as the momentum deviation Dp=p is zero. To simplify further, the area occupied by all particles can be described by an ellipse whose area divided by p is called the transverse emittance ex , and ey , respectively. The ellipse is chosen for different conditions: rms, 100% ellipse, or a certain fraction of current. The 100% ellipse contains all particles but has the minimum possible area. A numerical algorithm to minimize this area for a given assembly of coordinates has been proposed by Keller [1]. The area of a rms ellipse perms for n particles is a statistical quantity defined by the formula: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u n n n uX 2 X X 02 0 2 ð6:3Þ xi x i ð xi xi Þ : erms ¼ 1=nt i¼1

i¼1

i¼1

This quantity contains a certain fraction of all particles, depending on the distribution of all particles. To compensate for the missing particles which are outside the rms area the erms is sometimes multiplied by a factor of 4. One of these techniques The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

88

6 Beam Transport

is applied to describe a certain fraction of the current. However, sometimes the fraction describes a certain emittance level Z ð6:4Þ e ¼ 1=p ð IdAÞfrac which is typically used in simulation work, and sometimes the differential localized current is used: Z ð6:5Þ e ¼ 1=p Ifrac dA which is typically used in experimental work to eliminate the effect of noise. In both cases integration is made over the area of phase space. According to the law of Liouville [2] these emittances remain constant, if only conservative forces are present. Single particle interactions, such as collisions, are not included in the model. Assuming that the longitudinal momentum is constant, the transverse momenta px and py can be replaced by the angles px =pz and py =pz , respectively. However, this angle will shrink with acceleration, only the normalized transverse emittance stays constant

en ¼ b c e

ð6:6Þ

where b is the velocity, expressed in speed of light units, and c is the relativistic mass. If all present forces are assumed to be linear the transport can be investigated in first order, separately, in both transverse planes. Using the formalism of ellipses the description of beam transport becomes effective because just the three Twiss parameters a, b, and c are sufficient to describe the transport, independent of the emittance e of the beam, which is the area of the ellipse divided by p. The maximum beam radius xmax is given by pffiffiffiffiffiffiffiffiffi ð6:7Þ xmax ¼ b e whereas the maximum angle x_ max is given by pffiffiffiffiffiffiffiffiffi x_ max ¼ c e :

ð6:8Þ

The action of each optical element can be described by a 33 transfer matrix T which is applied to the initial ellipse parameters with index 0, resulting in the new ellipse parameters with index n: 0 1 0 1 b b @aA ¼ T @aA : ð6:9Þ c n c 0 Some optical elements with their transfer matrices are listed here, and many more are available in the literature [3]. A drift section of length Ld modifies the orientation of the ellipse a; b; while keeping the angle c constant:

6.1 Introduction

0

1 T ¼ @0 0

2

2Ld 1 0

1

Ld Ld A: 1

ð6:10Þ

A thin lens with focusing length F does not modify the beam size b, but the divergence parameters a and c: 0 1 1 0 0 1 0 A: T ¼ @ 1=F ð6:11Þ 2 1=F 2=F 1 For bending sections different transfer matrices are available, depending on the type of bending magnet. For the most simple case, i.e. when the magnet ends are perpendicular to the beam, a dipole bends the beam in the deflecting plane according to the momentum and the flux density. 1 0 2 2 2 2=K sin KL cos KL 1=K sin KL cos KL C B 2 2 T ¼ @ K sin KL cos KL 1=K sin KL cos KL A: ð6:12Þ cos KL sin KL 2 2 2 K sin KL cos KL cos KL K sin KL In the non-deflecting plane a dipole can be assumed as a simple drift section. If an angle b exists between the particle path and the dipole magnet exists, an additional transfer matrix at the entrance and the exit of the bending magnet describes the optical action of such an angle in the bending plane with a radius of curvature 0 for the central trajectory 0 1 1 0 0 1 0 A; T ¼ @ tan b=0 ð6:13Þ 2 2 tan b=0 2tan b=0 1 and in the vertical plane 0 1 0 1 T ¼ @ tan b=0 2

2

tan b=0

2tan b=0

1 0 0 A:

ð6:14Þ

1

The transfer matrix for a magnetic quadrupole is different for the focusing plane 0 1 pffiffiffiffiffiffiffi 2 2 2= jKj sin u cos u 1=jKj sin u cos u pffiffiffiffiffiffiffi B pffiffiffiffiffiffiffi C 2 2 T ¼ @ jKjsin u cos u cos u sin u 1= jKjsin u cos u A ð6:15Þ pffiffiffiffiffiffiffi 2 2 2 jKjsin u cos u cos u jKjsin u and for the de-focusing plane 0 pffiffiffiffiffiffiffi 2 2= jKjsinhu coshu cosh u B pffiffiffiffiffiffiffi 2 2 T ¼ @ jKjsinh u cosh u cosh u sinh u p ffiffiffiffiffiffi ffi 2 2 jKjsinh u cosh u jKjsinh u

1 2 1=jKjsinh u pffiffiffiffiffiffiffi C 1= jKjsinh u cosh u A: 2 cosh u ð6:16Þ

89

90

6 Beam Transport

pffiffiffiffiffiffiffi With the abbreviation u ¼ Lq jKj, which describes the effective length Lq of the quadrupole with strength K. Magnetic solenoids are not often used in low energy beam lines, at least for ion beams, because of high power requirements. With an effective length Ls and the ratio of the maximum flux density inside the solenoid B0 and the momentum of the central trajectory B0 , K is defined as K¼

B0 2B0 :

ð6:17Þ

With these abbreviations and h ¼ KLs the transfer matrix reads 0 1 2 2 2 2=Ks sin h cos h 1=K sin h cos h B C 2 2 T ¼ @ K sin h cos h 1=K sin h cos h A: cos h sin h 2 2 2 K sin h 2K sin h cos h cos h 77.5 degree bending magnet

quadrupole doublet

quadrupole triplet

acceleration gap

ð6:18Þ

ion source

quadrupole triplet

12.5 degree switching magnet

quadrupole quadruplet

Figure 6.1

Low energy injection beam line from the ion source to the first rf-accelerator at GSI.

6.1 Introduction

Because of the beam rotation of h in azimuth a solenoid couples both transverse planes to each other, but an existing cylindrical symmetry will be preserved. For an acceleration section of length La , which increases the momentum from p0 to p0 þ Dp, the transfer matrix can be written as 0 1 2 1 2 p0 =Dp La b ðp0 =Dp La bÞ B C T ¼ @0 ð6:19Þ p0 =ðp0 þ DpÞ p0 =ðp0 þ DpÞ p0 =Dp La b A; 2 0 0 ðp0 =ðp0 þ DpÞÞ with b ¼ lnð1 þ Dp=p0 Þ. Note that the determinant of the transfer matrix (6.19) is not equal to 1, because the absolute emittance shrinks during acceleration. With these transfer matrices a first order beam transport simulation can be made, which is demonstrated using the low energy beam line at GSI. The beam line is shown in Figure 6.1, and the result of the first order simulation starting with a measured emittance behind the acceleration gap is shown in Ref. [5]. The ion source is located on a high voltage platform. The ion beam drifts after extraction towards the acceleration gap without optical elements. After acceleration the beam is focused with five quadrupole lenses into a 77.5 analyzing magnet. Behind that magnet the beam is focused by a triplet towards another 12.5 switching magnet. Behind that dipole the beam is focused by a quadrupole quadruplet into the first rf-accelerator. However, one should be aware of the implied simplifications and assumptions in achievable results using this method. 6.1.1

Drift

Because the emittance has a waist in the extraction system (a ¼ 0) and the maximum diameter at that point is known from the geometrical shape of the extraction system (in our case with a 13 3 mm multi-aperture system) the emittance of the beam can be estimated if the diameter after a known drift length is measured. We found a typical value of ˘ 80 mm. This leads to an emittance of 3010 mm mrad. With that estimation b and c can be determined with 3

b0 ¼ ð10 10

2

mÞ =ð300 10

6

m radÞ ¼ 0:333 m

ð6:20Þ

and 3

c0 ¼ ð30 10

2

radÞ =ð300 10

6

m radÞ ¼ 3:

ð6:21Þ

Applying the transfer matrix for a drift of 1 m the beam in front of the acceleration gap is described by 1 0 1 0 1 0 0:33 1 2 1 3:33 @ 3:0 A ¼ @ 0 1 1 A @ 0 A: ð6:22Þ 3:0 0 0 1 3:0 This estimation is only valid for a pure drift, and space-charge forces are neglected. Whether this estimation is valid depends on the experimental conditions:

91

92

6 Beam Transport

either the current is too low to play a role, or the effect of the current is compensated for. In our case the diameter has been measured to be at least 60 mm diameter at that location, therefore the emittance should be of the order of 400 mm mrad, if we assume full space-charge compensation. 6.1.2

Extraction System and Acceleration Gap

Within the extraction section and the acceleration section the first order transfer matrices cannot be applied, because the boundary conditions are not valid. In both cases a strong radial field counteracts the assumption of linear independence. Additionally the space-charge force is not compensated as it is in the drift section. This assumption is confirmed by the following experimental facts: .

.

An emittance of 30 p mm mrad of a single aperture extraction hole (3 mm extraction hole diameter, 2 mm hole diameter of the grounded electrode, 20 mrad angle) seems to be reasonable. However, for an emitting surface with 1.5 mm radius and a longitudinal energy of 100 eV, an angle for the ions within the plasma of about 10 seems to be too high. This number has been calculated assuming that the normalized emittance stays constant and is therefore an indication that the absolute emittance cannot be assumed to be a constant. A current reduction of 56% due to the beam diameter and aperture of the acceleration gap has been proven experimentally, and an emittance reduction from 200 p mm mrad (gap acceptance) to values below 100 p mm mrad according to Liouville’s law could not be measured.

The fact that measured emittances show a strongly aberrated beam indicates again that the simple first order assumptions are not valid. A numerical simulation program has to be used instead; an example is shown in Figure 6.2. The first part of the beam line from Figure 6.1 has been simulated for a 150 mA uranium beam with 32 kV extraction voltage and 89 kV acceleration voltage.

Figure 6.2 Computer simulation [4] of extraction system (on the right-hand side) and acceleration gap (middle of the picture). The beam direction is from right to left.

6.1.3

Low Energy Beam Line

For the rest of the beam line the matrix formalism can be used with high accuracy. A simulation of the above beam line has been presented in Ref. [5].

6.3 Space-Charge Compensation

6.2

Current Effects

Besides the external fields discussed up to now the beam generates internal fields which might influence the beam optics considerably. These are the electric fields caused by the space-charge 2 ð6:23Þ r U¼ e0 and the self magnetic field of the beam Z ~ r ~ : ð6:24Þ B ¼ l0 Ids 3~ 4pr For the parameter range of interest for low energy beams, the magnetic field can be neglected, whereas the space-charge force can be very important. The electrostatic potential caused by an ion beam with current I and velocity v can be estimated for an infinitely long ion beam with radius rb within a beam pipe with radius rw and zero emittance to be U¼

I ð1 þ ln rw Þ: 2pe0 v 2 rb

ð6:25Þ

Inside the beam the potential is parabolic and the force therefore linear. A 10 emA, singly charged argon ion beam with an energy of 30 keV will double its diameter after roughly 20 cm, because of the space-charge, see Figure 6.4 below. The magnetic force due to the current is much too low to compensate this effect. Any deviation of the above assumed boundary conditions, as for example an inhomogeneous beam distribution, will cause non-linear fields and therefore emittance growth. It should be pointed out that in optical elements such as a magnetic dipole or a magnetic quadrupole the magnetic forces acting on the beam and on the spacecharge compensating particles are not compatible with the above demand. However, experimental experience shows that the strong diverging force of the space-charge is not present, indicating a high degree of space-charge compensation. 6.3

Space-Charge Compensation

Two processes are responsible for the passive production of space-charge neutralizing particles: collisions of ions of the primary beam with residual gas atoms, and sputtering. External sources of secondaries, as for example heated filaments, are neglected here. 6.3.1

Residual Gas Collisions

Particles of a charged beam drifting in a vacuum chamber with energy E0 will collide with a certain probability with residual gas atoms or molecules. Depending on

93

94

6 Beam Transport

the specific conditions of the collision, the residual gas atom can be ionized. In the case of a positive primary beam, the secondary ion will be repelled by the spacecharge potential whereas the electron will be trapped. If there is no leakage of this trap, the space-charge potential will be decreased until a balance between spacecharge potential and electron temperature has been reached. ns ¼ np rnrg

ð6:26Þ

where ns is the number of created secondaries, np is the number of primary ions, and r is the cross section. The necessary time s to create enough electrons for the space-charge compensation can be estimated by [6],[7] s¼

1 rvnrg

ð6:27Þ

if the velocity v, and the number of residual gas atoms nrg are known. In this model it is assumed that all generated electrons will be trapped due to the positive polarity of the beam. A more realistic model includes losses of electrons with an energy above the space-charge potential. This will increase the build-up time of compensation [8]. Typical values for the cross section for the production of secondaries by collisions of primary ions with residual gas atoms are of the order of 20 2 10 m . Assuming a typical residual gas pressure within the beam line of 7 10 mbar shows that the rate of secondaries would require a compensation buildup time of the order of ms. Experimental experience [11] however shows a build-up 5 time of the order of 10 s, indicating the existence of another electron source, most probably sputtering. The collisions with residual gas atoms are also responsible for the charge exchange process. High charge state ions tend to capture a free electron: 0

A þA

nqþ

þ

ðn1Þqþ

þ e!A þ A

þe

ð6:28Þ

whereas the process of ionizing has a small cross section at typical energies for beam transport. In both cases the ion is lost from the beam. 6.3.2

Sputtering

If ions of the primary beam hit a metallic surface (e.g. electrodes, beam tube) secondary particles are created. Again, for a positive ion beam, electrons will be trapped in the space-charge potential well, whereas positive secondary particles will be repelled. The cross section for sputtering is huge compared to the cross section for residual gas ionization. It depends on the energy of the primary particle [9] and the angle of incidence [10].

6.3 Space-Charge Compensation

6.3.3

Preserving Space Charge Compensation

The resulting particle distribution is described by the following equation: ! 2 @f ðx; x_ ; tÞ @f qE @f @ f ¼ x_ þ beam F þ D 2 @t @x m @ x_ @ x_

ð6:29Þ

with the ionization rate F, the diffusion term D. Electric fields have to be avoided, so as not to remove electrons from the beam. For the case of a positive ion beam, the ion source is biased with a positive voltage to extract the positive ions. This voltage also attracts negative charges from the beam towards the source and therefore destroys the beam neutralization. To avoid this undesired effect a three-electrode system in accel–decel mode can be used. The negatively biased intermediate electrode screens the positive potential of the ion source with respect to the ion beam. If the beam is further accelerated, decelerated, or injected into an rf structure such screening is necessary also to preserve the space-charge neutralization. Similar to an accel–decel extraction system, all positive potentials have to be screened with respect to the beam. These locations are for example the accelerating gap or the entrance of an rf-accelerator. All elements which are seen by the beam have to be grounded, so as not to destroy the space-charge compensation. Only magnetic lenses should be used so as not to destroy the space-charge compensation. Magnetic fields, for example dipoles, solenoids, and quadrupoles, will influence the mobility of the compensating charges, and, additionally, charges have to be created along the magnetic field lines; this might influence the build-up compensation time, but not the existence of the compensation itself. Experimentally it was found that a beam is neutralized to a very high degree, even in magnetic dipole fields. 6.3.4

Influence of Space Charge Compensation

Within a beam tube with length of 1 m and 80 mm diameter a singly charged argon ion beam with 10 emA current and 30 keV energy is drifting. The absolute starting emittance is 100 mm mrad. In Figure 6.3 the beam envelope is shown for the case of zero current (a) and for the 10 emA case (b). The compensation of an ion beam can be described by Self’s [12] model

ne ¼ ne0 e

UUp kTe

ð6:30Þ

similar to the physics which describes the particle distribution at the plasma boundary of an ion source in the extraction system. Using this model a much more realistic description of the ion beam transport is obtained compared to the assumption of a net current, both shown in Figures 6.3 and 6.4. The resulting potential due to the space-charge for the uncompensated beam is of the order of 800 V (Figure 6.4 a) and the beam becomes strongly divergent (Figure 6.3 b); the potential drops in the longitudinal direction because of this divergence. Using a net current model, the

95

96

6 Beam Transport

Figure 6.3 Beam envelope for a drifting beam: a) no spacecharge, b) full space-charge, c) net current, and d) space-charge compensation with the beam plasma model.

potential is only 80 V, according to the degree of space-charge compensation (Figure 6.4 b) and the longitudinal drop is lower because the divergence is smaller. Using Self’s model the beam plasma potential is constant in the beam direction (Figure 6.4 c) and a gradient is present at the beam edges only; the strength of the gradient depends on the electron temperature. The electrons are oscillating in the beam potential and only the coldest electrons will stay within the beam. The effect of these different assumptions becomes clearer by observing the emittance at the end of the drift section (see Figure 6.5): a pure drift without any spacecharge results in the emittance shown in Figure 6.5 a, strong losses and a divergent a

V 800 600 400 200 0

0.04 0.0 y [m] 0.0 0.2 0.4 0.6 0.8

-0.04 1.0 x [m]

b

V 80 60 40 20 0

0.04 0.0 y [m] 0.0 0.2 0.4 0.6 0.8

-0.04 1.0 x [m]

Figure 6.4 Space-charge potential for a drifting beam: a) full space-charge, b) net current, and c) space-charge compensation with the beam plasma model.

c

V 80 60 40 20 0

0.04 0.0 y [m] 0.0 0.2 0.4 0.6 0.8

-0.04 1.0 x [m]

6.3 Space-Charge Compensation a

mrad 100 60

60

20

20

-20

-20

-60

-60

-100 -4 -3 -2 -1

0

1

2

3

4 cm c

mrad 100

-100 -4 -3 -2 -1

60

20

20

-20

-20

-60

-60

0

1

2

3

4 cm

0

1

2

3

-100 -4 -3 -2 -1

4 cm d

mrad 100

60

-100 -4 -3 -2 -1

b

mrad 100

0

1

2

3

4 cm

Beam emittance for a drifting beam: a) no spacecharge, b) full space-charge, c) net current, and d) space-charge compensation with the beam plasma model. Figure 6.5

beam for the full space-charge of 10 emA in Figure 6.5 b, a divergent beam but still with an undisturbed emittance for the homogeneous 1 emA beam in Figure 6.5 c, and aberrations caused by the nonlinear field of a partially uncompensated beam in Figure 6.5 d. A beam halo develops because of the gradient of the space-charge potential at the beam edge. The degree of space-charge compensation in Figure 6.5 c and in Figure 6.5 d is similar at about 90%. In Figure 6.4 the potential caused by the effective space-charge is shown. The potential is plotted on the vertical axis against the longitudinal and one transverse axis. Because of trapped electrons the potential is constant with beam plasma potential Figure 6.4 c. The gradient in the longitudinal direction for Figure 6.4 b is less strong because the beam does not diverge as strongly as in Figure 6.4 a. The conclusions of this simulation are: .

.

.

Provide a very high degree of space-charge compensation within the beam transport system, to keep nonlinearities small. Remove the space-charge neutralization totally, which might be even more complicated than to keep it [8], and provide lenses with sufficient focusing power. Make the beam transport system as short as possible to reduce emittance growth along the beam line due to the imperfect neutralization.

97

98

6 Beam Transport

The following section will go through a practical example showing how a beam line could be designed and optimized for an intense proton beam.

6.4

A LEBT System for the Future Proton Linac at GSI

As part of a future project, GSI plans to build a 50 MeV proton LINAC for antiproton production [13]. This specialized injector should provide a proton beam of 70 mA current, 0.1 ms pulse duration and 5 Hz repetition rate at the entrance to the existing synchrotron [14]. Several LINACs providing such proton beams for the synchrotron injection are in operation at high energy hadron beam facilities. The layout is still in discussion, however, the present layout consists of an ion source (filament driven volume type or ECR type) on a high voltage platform, a RFQ (radio frequency quadrupole) and a DTL (drift tube LINAC), resulting in a full length of about 20 m. For the injection into the RFQ the ion source and LEBT system has to deliver a 100 mA proton beam with an energy of 100 keV. Three different extraction and acceleration systems will be discussed for the ion source and LEBT section: A compound system with an embedded einzel lens, a pentode system, and a combination of a triode extraction system with a following dc post-acceleration system. The last system has been in operation very successfully at the high current injector at GSI for many years [15,16]. However, the existing high current injector has to deliver ion species with a mass-to-charge ratio from 3 to 65, and therefore has to be more flexible than the proton machine. For the injection into the proton LINAC the normalized rms emittance at the entrance of the RFQ has to be smaller than 0.2 p mm mrad, and the extraction and LEBT system should be manageable for easy routine operation. The normalized rms emittance results in an absolute rms emittance of 13.7 p mm mrad. In the following we will use both, the value of the rms emittance and the effective emittance. For the investigation an intense proton source is assumed with a proton yield of þ þ 90%, 5% of H2 , and 5% of H3 [17,18]. This ion source was built in 1999 as a prototype for the IFMIF (international fusion materials irradiation facility) scenario [19]. Because of the negatively biased plasma electrode relating to the anode the plasmaTable 6.1

Ion source data relevant for the computer simulations.

Ion fraction Emission current density Electron temperature Ion temperature Plasma wall potential Voltage of biased plasma electrode Initial proton velocity Maximum aperture of extraction system Maximum field strength for extraction gap

þ

þ

þ

90% H1 ; 5% H2 ; and 5% H3 2 1000 to 4000 Am 1 eV 0:1 eV 5V 170 V 185,000 m s–1 10 mm for plasma electrode 6:8 kVmm–1

6.4 A LEBT System for the Future Proton Linac at GSI

wall potential drop is increased. As a consequence the initial velocity of the protons before entering the extraction system is 185,000 m s–1, which increases the possible emission current density because of the reduced space-charge forces after passing the plasma boundary. Table 6.1 gives an overview of the most important data necessary for computer simulations with AXCEL-INP [4]. All systems were optimized for a high brightness proton beam with a divergent emittance pattern. Such an emittance pattern is suitable for a LEBT section consisting of a drift section followed by solenoids to focus the ion beam into the RFQ. 6.4.1

Compound System

The compound system consists of a two-electrode extraction gap with 9 mm aperture, resulting in a necessary emission current density of 1750 A m–2. The aspect ratio is 0.56 with a gap of 8 mm. Further on it consists of an einzel lens which decelerates the ion beam to an energy of 6 keV, an acceleration gap, and an embedded screening electrode. The applied voltages of the electrodes and geometric parameter are shown in Table 6.2. Thickness of electrode 1 (0.6 mm) represents the web width of the plasma electrode, electrode 1 has an angle of 70, close to the exact Pierce geometry. Potential and geometric parameter of the compound system. Electrodes 1 to 7 are shown in Figure 6.6 from left to right.

Table 6.2

Electrode

1 2 3 4 5 6 7

Potential [kV]

Aperture [mm]

Thickness [mm]

Gap Field Strength [kV mm–1]

100 65 94 65 0 –5 0

9 9 11 8 8 10 8

0.6/3 5 8 5 5 20 5

4.38 5.8 5.8 5 0.5 0.5 –

The field strength within the extraction gap is 4.38 kV mm–1 and within the acceleration gap 5 kV mm–1. An applied screening voltage of –5 kV is sufficient to ensure a space-charge compensated ion beam. The geometry of the compound system is optimized for a 111 mA at 100 keV full beam with an ion fraction as shown in Table 6.1. Therefore this ion beam includes 100 mA protons. Figure 6.6 presents the trajectory plot of the simulation and Figure 6.7 (left) the emittance pattern at a distance 4 cm from the ground electrode. The maximum angle is 55 mrad, the maximum radius is 4.5 mm. This results in an effective emittance of 110 p mm mrad, and a rms emittance of 10 p mm mrad. It can be seen that 80% of the beam has a much smaller emittance due to lack of aberrations. More than 80% of the beam trajectories have divergence angles less than 20 mrad within a beam radius of 3.5 mm.

99

100

6 Beam Transport [cm] 1

1

2

3

4

5

6

7

0.6

0.2

-0.2

-0.6

-1 0

2

4

6

8

10

12

14

16 [cm]

Figure 6.6 Trajectory plot of 111 mA full beam for a compound system. Geometric parameters and potentials for the electrodes 1 to 7 (from left to right) are described in Table 6.2.

Because of the deceleration to an energy of 6 keV due to the einzel lens the spacecharge forces are dominant at this location. This results in an ion beam bloating. As soon as the ion beam has reached the final energy in the following acceleration gap the ion beam radius is reduced to a minimum. The deceleration effect of the last aperture is negligible. It is assumed that the ion beam is fully space-charge compensated after passing the second ground electrode. As a result the compound system is able to deliver a high brightness 100 mA proton beam with energy of 100 keV in an emittance suitable for injection into the RFQ. However, the compound system reacts very sensitively to fluctuations of emission current density and potential of the einzel lens. Variations of 3% of emission current density or 1 kV of einzel lens potential result in a totally unmatched ion [mrad]

[mrad] 200

140 100

100

60 20

0 -20 -60

-100

-100 -140

-200 -1

-0.6

-0.2

0.2

0.6

1 [cm]

-4

-3

Figure 6.7 Left: Emittance pattern for drifting ion beam. e(rms) = 10 p mm mrad. Right: Emittance pattern in the focus þ plane for H for injection into the RFQ using a solenoid with a flux density of 1.5 T and a half height width of 2 cm.

-2

-1

0

1

2

3

4 [cm]

6.4 A LEBT System for the Future Proton Linac at GSI

beam with too high angle at the exit of the compound system. Due to lens errors the compound system has a large emittance. For injection into a RFQ a solenoid with flux density of 1.5 T and half height width of 2 cm could be an option. After a drift section of 40 cm the ion beam will be focussed by a solenoid. The emittance pattern in the focus plane for protons is shown in Figure 6.7 (right). It can be seen that the ion species will be focussed differently to increase further the proton yield within the acceptance of the RFQ. 6.4.2

Pentode or Two-Gap System

The pentode system extracts and accelerates the ion beam in two gaps to an energy of 100 keV. The embedded screening electrode ensures a space-charge compensated ion beam after extraction. The potentials of electrodes 1 to 5, from left to right in Figure 6. and the geometric parameters are listed in Table 6.3. The aperture of the plasma electrode is 10 mm which results in a necessary emission current density of 1410 A m–2. For the extraction and acceleration gaps maximum field strengths of 4.38 kV mm–1 and 6.5 kV mm–1 are foreseen. The trajectory plot and emittance pattern are shown in Figures 6.8 and 6.9, respectively. The effective emittance 4 cm downstream is 113 p mm mrad, and the rms emittance 8.8 p mm mrad. Compared to the emittance of the compound system, the rms emittance leads to a more laminar ion beam. Also shown in Figure 6.9 (right) is the ion beam after focussing with a 1.5 T solenoid for injection into an RFQ. Here protons are in the focus plane whereas hydrogen molecules are separated. Potentials and geometric parameters of the pentode system. Electrodes 1 to 5 are shown in Figure‚ 8 (from left to right).

Table 6.3

Electrode

1 2 3 4 5

Potential [kV]

Aperture [mm]

Thickness [mm]


100 65 0 –8 0

10 10 10 10 10

0.6/3 5 5 10 5

4.38 6.5 1.6 1.6 –

A nearly laminar ion beam with the postulated energy and intensity is detected for the two gap system. As a disadvantage we note the higher gap field strength for acceleration, but as an advantage the pentode system is insensitive to fluctuations in emission current density. This system would be a very good choice with a solenoid following.

101

102

6 Beam Transport [cm] 1

1

2

3

4

5

0.6

0.2

-0.2

-0.6

-1 0

Figure 6.8

2

4

6

8

10

[cm]

Trajectory plot of 111 mA full beam for a pentode system. [mrad]

[mrad] 200

140 100

100

60 20

0 -20 -60

-100

-100 -140

-200 -1

-0.6

-0.2

0.2

0.6

1 [cm]

-4

-3

-2

-1

0

1

2

3

4 [cm]

Figure 6.9 Left: Emittance pattern, e(rms) = 8.8 p mm mrad. Right: Emittance pattern in focus plane for protons for injection into the RFQ. The flux density of the solenoid is 1.5 T with a half height width of 2 cm.

6.4.3

Triode System and DC Post-Acceleration

The first section consists of a conventional triode extraction system. The designed ion beam current for the extraction is 122 mA full beam because of unavoidable beam losses in the post-acceleration gap. These losses are approximately 9% and necessary to ensure a concave beam–plasma boundary at the entrance to the postacceleration system. The aperture of the plasma electrode is 9 mm in diameter, resulting in a necessary emission current density of 1920 A m–2. The following apertures are 10 mm for the screening electrode and ground electrode, respectively as shown in Table 6.4. With an aspect ratio of 0.5 and voltages of 35 kV for the plasma electrode and –4 kV for the screening electrode the maximum field strength

6.4 A LEBT System for the Future Proton Linac at GSI [cm] 1

1

2

3

0.6

0.2

-0.2

-0.6

-1 0

Figure 6.10

0.4

0.8

1.2

1.6

2

2.4

2.8

[cm]

Trajectory plot of 122 mA at 35 kV full beam for a triode system.

[mrad] 100

60

20

-20

-60

-100 -5

Figure 6.11

-3

-1

1

3

5 [mm]

Emittance pattern. e (rms) = 26 p mm mrad.

is 4.3 kV mm–1. The trajectory plot and the emittance pattern are shown in Figures 6.10 and 6.11. Because of the commonly observed aberrations for a triode system the effective emittance is 166 p mm mrad, whereas the rms emittance is 25.5 p mm mrad. After a drift section of 0.2 m the post-acceleration system starts to accelerate the ion beam to 100 keV. It is made of a high voltage electrode (65 kV) and a screening electrode (–6 kV) embedded by two ground electrodes. The gap distance for the acceleration gap is 30 mm, resulting in a field strength of 2.5 kV mm–1 with an aspect ratio of 0.5 (Table 6.5). The input for computer simulation is derived from output coordinates of the triode system. Figures 6.12 and 6.13 (left) show the trajectory plot and the emittance pattern directly behind the postacceleration. One can see typical aberrations for such a post-acceleration system. As a result the effective emittance is very large: e = 150 p mm mrad. The rms emittance is 26 p mm mrad. In Figure 6.13 (right) a solenoid field with flux density of

103

104

6 Beam Transport

1.5 T and a half height width of 2 cm is shown to separate the ion species and to form a convergent ion beam for input into a RFQ. This solenoid is placed 0.45 m behind the acceleration. Table 6.4

Potentials and geometric parameters of triode extraction system.

Electrode

1 2 3

Table 6.5

Potential [kV]

Aperture [mm]

Thickness [mm]


35 –2 0

9 10 10

0.6/3 4 4

3.89 2 –

Potentials and geometric parameters of post-acceleration system.

Electrode

1 2 3 4

Potential [kV]

Aperture [mm]

Thickness [mm]


65 0 –6 0

30 30 40 30

10 10 30 30

2.5 0.6 0.6 –

The combination of triode system and post-acceleration is manageable very well because of the splitting of the voltages. The increase in the emittance is due to the drift of the space-charge compensated ion beam, resulting in an inhomogeneous distribution at the entrance of the acceleration gap where the space-charge compensation is removed. Any inhomogeneous ion beam distribution tends to become [cm] 4

1 234

3 2 1 0 -1 -2 -3 -4 0

0.2

0.4

0.6

0.8

1

1.2 [m]

Figure 6.12 Trajectory plot of the post accelerated ion beam. Solenoid (1.5 T flux density and 2 cm half height width) is placed 45 cm downstream to focus the ion beam into the RFQ.

6.4 A LEBT System for the Future Proton Linac at GSI [mrad] 200

[mrad] 400 300

100

200 100

0

0 -100

-100

-200 -300

-200 -1.4 -1

-0.6 -0.2

0.2

0.6

1

1.4 [cm]

-400 -1

-0.6

-0.2

0.2

0.6

1 [cm]

Figure 6.13 Left: Emittance pattern of 111 mA full beam at a distance of 0.4 m before separation of the three ion species by solenoid. e(rms) = 27 p mm mrad. Right: Emittance pattern in focus plane for protons after focussing with solenoid (1.5 T, 2 cm half height width).

homogeneous again at the price of an increase in emittance. However, the divergence angles of the post-accelerated ion beam are still adequate. 6.4.4

Discussion

The compound system is able to deliver a high brightness 100 mA proton beam with energy of 100 keV in an emittance suitable for injection into a RFQ. However, because of the einzel lens this system reacts very sensitive to fluctuations of emission current density and the potential of the einzel lens which is a great disadvantage for operation at an accelerator. Variations of 3% in emission current density or 1 kV in einzel lens potential result in a totally unfitted ion beam with maximum divergence angle at the exit of the compound system. The beam quality of the two gap system is very good. A nearly laminar ion beam with the postulated energy and intensity is detected. As a disadvantage we note the higher gap field strength for acceleration, but as an advantage it is insensitive to fluctuations in emission current densities. This system would be the best choice, with a succeeding solenoid for separation of ion species and convergent injection into an RFQ. The combination of triode system and post-acceleration is manageable very well because of the splitting of the voltages. For beginning of operations e.g. after ion source service, it is possible to drive the ion source with constant perveance. Aberrations, however, result in a too large emittance but the divergence angles of the post-accelerated ion beam are very small.

105

106

6 Beam Transport

References [1] R. Keller, J.D. Sherman and P. Allison, IEEE

Trans. Nucl. Sci. 32, 2579, (1985). [2] J. Liouville, J. Math., 3, 324, (1838). [3] Proceedings of the Fifth General Accelerator Physics Course, CERN Accelerator School, CERN 94-01, 1994. [4] INP, P. Spdtke, Junkernstr. 99, 65205 Wiesbaden, Germany, e-mail: [email protected]. [5] L. Dahl and P. Spdtke, The Low Energy Beam Transport System of the New GSI High Current Injector, Proceedings of XX International LINAC Conference, Monterey, 2000. [6] L.R. Evans and D.J. Warner, CERN/MPS/Lin. 71-2 (1971). [7] M.D. Gabovich, Sov. Phys. Tech. Phys., 19(11) (1975). [8] A. Schnlein, Emittanzwachstum und Raumladungskompensation beim Transport intensiver Ionenstrahlen, Johann Wolfgang Goethe-Universitt, Frankfurt, 1987. [9] Yasunori Yamamura and Hiro Tawara, Energy Dependence of Ion-Induced Sputtering Yields from Monoatomic Solids at Normal Incidence, Internal Report, National Institute for Fusion Science, Chigusa-ku, Nagoya 46401, Japan. [10] Yasumishi Yamamura, Yukikazu Itikawa and Noriaki Itoh, Angular Dependence of Sputtering Yields of Monoatomic Solids, Internal

[11]

[12] [13]

[14] [15]

[16] [17]

[18] [19]

Report, Institute of Plasma Physics, Nagoya University, Chikusa-ku, Nagoya 464, Japan. R. Dlling, Raumladungskompensation driftender intensiver Strahlen niederenergetischen Ionen und Techniken zu ihrer Vermessung, Johann Wolfgang Goethe-Universitt, Frankfurt, 1994. S.A. Self, Phys. Fluids 6, 1762 (1963). Conceptual Design Report, An International Accelerator Facility for Beams of Ions and Antiprotons, GSI Darmstadt, Germany, 2001. K. Blasche et al., GSI Scientific Report, ISSN 0174-0814 (2003). H. Reich, F. Heymach and P. Spdtke, Commissioning of the High Current Ion Sources at the New GSI Injector (HSI), Proceedings of XX International LINAC Conference, Monterey, p. 238 (2000). R. Hollinger, F. Heymach and P. Spdtke, Rev. Sci. Instrum. 73, 1024 (2002). R. Hollinger, P. Beller, K. Volk, M. Weber and H. Klein, Rev. Sci. Instrum. 71, 836 (2000). R. Hollinger, K. Volk and H. Klein, Rev. Sci. Instrum. 73, 1027 (2002). IFMIF-International Fusion Materials Irradiation Facility. Conceptual Design Evaluation Report. A Supplement to the CDA by the IFMIF team, edited by A. Mslang, Forschungszentrum Karlsruhe FZKA6199, Germany (1999).

107

7

High Current Gaseous Ion Sources Nikolai Gavrilov

7.1

Introduction

Rapid progress in the physics and technology of high current gaseous positive ion sources was witnessed in the 1960s–1980s, driven by large-scale research projects concerned with heating of magnetically confined thermonuclear plasma [1], development of inert gas electric thrusters for space propulsion [2], and processing of materials [3]. This work resulted in the development of powerful neutral beam injectors employing quasi-steady-state sources of hydrogen and deuterium positive ion beams with current up to 100 A and energy up to 100 keV, as well as large diameter electron bombardment ion thrusters with heavy inert gas ion beams of current up to several amperes and energies of ~1 keV for orientation and correction of the orbit of space vehicles. With the advent of high current ion sources since the 1970s, materials technologies based on high flux ion irradiation, such as ion beam sputter etching and sputter deposition, surface modification of metals and alloys, and ion-beamassisted deposition of coatings have seen widespread growth in both fundamental science and technological applications. The Bohm relationship for the ion current density that can be drawn from a plasma has the form [4] ji = 0.4ne(2kTe/Mi)1/2,

(7.1)

where n is the density of the undisturbed plasma, Te is the electron temperature, and Mi is the ion atomic mass. It follows that the primary means for increasing the current of ions extracted from the plasma, Ii = jiS, is to increase the plasma density and/or the plasma surface area S, noting that for most plasmas Te is only weakly dependent on the discharge parameters. Note also that Eq. (7.1) implies that, all other conditions remaining equal, increase in the atomic mass of the gas leads to a decrease in the extracted ion current. For these reasons the development of high current gaseous ion sources was focused mainly on the search for efficient methods of generation of dense plasma with a large emitting surface and broad ion beam formation. A high ion current density at the plasma boundary equal to ~102 A cm–2 was achieved in Ardenne’s source of the duoplasmatron type with geometric and The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

108

7 High Current Gaseous Ion Sources

magnetic constriction of the discharge column [5]. However, according to the Child– Langmuir law [6, 7] for the space-charge limited current density, ji = (4e0/9)(2e/Mi)1/2(U3/2/l2),

(7.2)

the required electric field exceeds by a large factor the breakdown field strength for such a high current density ji. Here U and l denote the voltage applied across the accelerating gap and the gap length respectively. To increase the ion beam current Lamb and Lofgren [8] proposed reducing the density and increasing the plasma surface area using an expansion cup at the outlet of the plasma generator. However, a single-aperture beam formation system does not provide a large increase in ion beam cross-section, because the angular divergence of the ion beam grows with increasing aperture of the ion optics. To form high current beams in large area ion thrusters multi-aperture ion optics systems were first used by Kaufman [9] and then adapted for neutral beam injectors by Hamilton [10]. Efficient operation of multi-aperture ion optics requires a uniform plasma. The surface area of the plasma in high current sources may be as large as 103 cm2 with a non-uniformity of ~5%. The maximum hydrogen ion beam current density, which is limited by breakdown of the accelerating gap, reaches 0.7–1.2 A cm–2 at energies from 15 to 40 keV [11]. But in quasi-steady-state high current ion sources the maximum ion current density is generally restricted to ji £ 0.25 A cm–2 [12] due to the high power density deposited on the source surfaces and limited cooling rate. As the voltage is increased further, the limiting ji value decreases because the working strength of the accelerating field grows quickly in accordance with Eq. (7.2). The drop in the limiting current density with decreasing voltage is due to the fact that the effective length of the accelerating gap, which depends on the electrode thickness and the hole diameter, decreases more slowly than the distance between the ion optics electrodes. The maximum beam current density for 0.6 keV argon ions is 4.2 mA cm–2 for a two-electrode system and 8.9 mA cm–2 for a three-electrode system [13].

7.2

Basic Types of High Current Ion Sources

High current gaseous ion sources may be divided, according to the type of low pressure gas discharge used, into hot filament cathode sources, cold cathode sources, and high frequency plasma sources [14]. Common features of all include the use of a high current discharge and special means to provide efficient electron impact ionization of the gas, maximum utilization of plasma ions, and minimal flow of neutrals from the gas discharge system. The energy efficiency of the discharge can be defined as the ratio of the ion current at the extractor (screen or plasma) grid to the discharge power input, and is thus specified in units of A kW–1. The energy efficiency of the ion source depends on the transparency of the extraction grid and is determined by the ratio of beam current to discharge power input. Alternatively the reciprocal value – the discharge energy loss per beam ion (eV ion–1) – is frequently

7.2 Basic Types of High Current Ion Sources

used. Thus if the ion energy cost exceeds (as it usually does) 2–3 times the ionization energy of the gas due to additional energy loss for excitation of gas atoms, the energy cost of ions moving toward the extraction grid increases inversely proportional to the ratio of the grid area to the total ion loss area from the discharge, and the energy cost of beam ions increases inversely proportional to the transparency of the extraction grid. The energy efficiency of the ion source is a most important parameter, allowing one to estimate the degree of ion source perfection. The gas efficiency is defined by the ratio of ion current to gas flow in comparable units (ions in the beam/gas atoms into the source). In filament driven discharges, electrons acquire their energy in the cathode sheath of the space charge and the double electric layer that is formed in the region of an abrupt change in the cross-sectional area of the discharge gap [15]. The electrons move towards the anode and dissipate their energy in plasma generation. The ion generation rate per unit volume is determined by the ratio between the density of ionizing electrons and their energy relaxation time sr . To improve the energy efficiency, the condition se > sr should be met, where se is the lifetime of electrons in the plasma before impacting the anode. se can be increased by a transverse magnetic field that impedes the drift of primary electrons to the anode, or an electrostatic potential distribution can be set up so that electrons oscillate within the potential well, perhaps with an axial magnetic field to help (see Chapter 2, Section 2.4.3). Most modern high current ion sources make use of multicusp magnetic field systems in which a strong transverse field is established along the periphery of the discharge system near the anode surface. The field decreases steeply with distance from the magnetic system. Such a magnetic field configuration, usually produced by rows of permanent magnets with alternating polarities, was used first by Moore [16] and then by Limpaecher and MacKenzie [17]. Primary electrons oscillate within the interior region that is virtually field-free and, because of multiple reflections of the electrons within this “magnetic bucket”, the electron distribution becomes isotropic and uniform, while the generated plasma turns out to be quiescent and homogeneous. High frequency discharges make use of the acceleration of electrons in a high frequency electromagnetic field to energies sufficient for ionization of gas atoms. Minimizing electrode erosion is an important design feature of such sources. Whereas ECR ion sources utilize a resonance between the applied electromagnetic field frequency and the electron cyclotron frequency, xappl = xce [18], dense plasmas of singly charged ions can be produced by a microwave discharge in a magnetic field that is not resonant and at higher gas pressures than for the ECR discharge. The working frequency of such microwave sources is typically 1 GHz to several tens of GHz. Absorption of field energy can be increased, however, by choosing a field frequency of the same order as the electron–neutral collision frequency in the plasma, leading to generation of dense plasma [14]. For gas pressures typical of ion sources, this condition is met for radio frequencies (RF) in the broad range 1 to 100 MHz. Cold-cathode ion sources are capable of long term operation with reactive gases. Commonly these sources employ different kinds of glow discharge plasmas with electrons oscillating in a magnetic field or a hollow cathode glow discharge with a

109

110


thin wire anode [19]. However, the energy and gas efficiencies of low pressure glow discharge sources are much lower than for filament driven ion sources. This is explained by the large energy loss for the cold cathode electron emission and the relatively high gas pressure necessary for a self-sustained glow discharge. The high operating voltage of the discharge, several hundred volts, leads to ion sputtering of the cathode material and the appearance of sputtered metal ions in the beam. Such sources have a simple, robust design and are highly reliable, making them attractive for technologies non-critical to gas pressure and metal ion impurities. Gaseous ion sources using a low pressure arc with cathode spots are more efficient in energy terms [20, 21]. An interesting direction in the development of high current ion sources is the use of two-stage discharges. The first stage acts as a plasma cathode and supplies electrons to the second stage, providing conditions for efficient ionization of the gas and formation of uniform plasma. Along with the well-known duoPIGatron [22], whose first stage employs a filament type cathode, these sources include the aforementioned cathode spot arc sources. Sources with RF and microwave discharges in the first stage have been developed [23, 24], and an ion source with injection of electrons to the second stage from a hollow cathode glow discharge [25] has also been used. Most of these sources employ a multicusp magnetic field in the anode chamber of the second stage, and they are capable of long term operation with reactive gases. In the following we consider the principles of operation, design features, and characteristics of the most common types of high current gaseous ion sources. 7.2.1

Filament Driven Ion Sources

In a duoplasmatron ion source [5] the discharge current path to the anode is through a small diameter hole (several mm) in an intermediate electrode, ahead of which a double electrostatic space-charge layer is formed where the electrons are accelerated to an energy of several tens of electron volts. The discharge column is further constricted in the gap between the intermediate electrode and anode by a strong inhomogeneous magnetic field where a second potential jump occurs. Ionization of the gas by a dense flow of energetic electrons confined by the strong magnetic field leads to generation of dense plasma and formation of a negative anode drop in potential (~10 V), which prevents ions from leaving the anode. As a result an intense ion stream arises in the plasma in the direction of the outlet aperture in the anode, which is usually several millimeters in diameter. The feed gas pressure in the duoplasmatron is typically (1–2) 10–2 Torr, the gas efficiency approaches 100%, and the energy efficiency is usually 1 A kW–1. The use of an expansion cup to reduce the plasma density and expand the surface area of the plasma not only increases the beam current but also decreases the beam emittance due to adiabatic expansion of the plasma, and decreases the level of the discharge current oscillations [26]. Duoplasmatron ion sources operating at a DC discharge current of 30 to 50 A and providing 1 A beams of hydrogen ions have been developed by Morgan at al. [27]. The electrodes, including the intermediate


electrode, of such sources are made of copper to facilitate heat removal. A magnetic field coil wound around the anode expansion cup decreases the plasma loss and optimizes conditions for ion extraction and formation of the ion beam. Demirkhanov [28] installed a ferromagnetic anticathode electrode to ensure oscillation of electrons in the anode region of the duoplasmatron. The anticathode was located behind a copper anode and had a potential close to the cathode potential. Electrons oscillating in the magnetic field (1500 G) efficiently ionize the gas at a low pressure of ~5 10–3 Torr. When the anticathode negative bias relative to the anode was increased from 0 to 100 V, the ion current increased by a factor of five. Decreasing the gas pressure from 2 to 0.5 10–2 Torr increased the extracted ion current by 1.5 times. The source, which had an outlet aperture 6 mm in diameter and an expansion cup 50 mm in diameter, provided a pulsed hydrogen ion current of up to 1.5 A at a discharge current of 20 A. This ion source concept, based on the use of a large scale Penning-type electrode system (PIG) to maintain a high current discharge by electrons supplied through a

Figure 7.1 DuoPIGatron. (Reprinted from Ref. [30] with permission from Elsevier). 1 – Filament power supply (65 V, 60 A); 2 – coil power supply (75 V, 1.5 A); 3 – discharge power supply (250 V, 20 A); 4 – accelerating power supply

(50 kV, 0.5 A); 5 – electron suppressor power supply (2.5 kV, 20 mA); GI – gas inlet; MC – magnetic field coil; IE – intermediate electrode; CF – cathode filament; A1 – anode 1; A2 – anode 2; EG – extraction grid.

111

112


double electric layer from a plasma cathode, has been embodied most completely in the DuoPIGatron type ion source [29]. Figure 7.1 shows the design of a source of this kind used in a high-current, high-voltage, oxygen ion implanter [30]. The magnetic field diverges toward the ion optics, ensuring oscillation of primary electrons between the intermediate and screen electrodes and formation of a large, uniform plasma surface. To increase the cathode lifetime, an inert gas (argon) is fed into the cathode stage and oxygen into the anode stage of the source. An ion optics system with 13 holes of diameter 5 mm provided an ion beam current up to 170 mA at an accelerating voltage of up to 50 kV. After mass separation of the beam and acceleration of ions to energy of 200 keV, a beam of O+ ions with current 100 mA was obtained. The filament cathode lifetime is over 25 hours. The periplasmatron-type ion source [31] has a unique electrode system design, which decreases heating of the screen electrode by thermal radiation of the cathode, reduces the effect of secondary electron back flow from the accelerating gap, and ensures a high uniformity of plasma flow to the screen grid. A rectangular variant of this source provided a hydrogen ion current of 96 A from an extractor 40 16 cm in size. Another direction in the development of high-current hot-filament ion sources is the advent of single-stage, large-volume plasma generators. A magnetic-field-free source for use in neutral beam injectors has been developed in which the plasma is generated by a diffuse low-pressure high-current discharge, with a distributed thermionically emitting cathode [32]. Twenty hairpin filaments, 0.5 mm in diameter, are installed around the periphery of the discharge chamber, 14 cm in diameter, near the cylindrical wall, which serves as the anode. The peripheral placement of the cathode provides generation of uniform (–6%), oscillation-free plasma of 12 cm diameter. For a discharge current of 1000 A the deuterium ion beam current density is 0.5 A cm–2 and the beam current 15 A at an extraction voltage of 15 kV. The principle of magnetic confinement of primary electrons and extraction of ions along magnetic lines using a multi-aperture extractor system was first demonstrated in the Kaufman source [13]. Figure 7.2 shows this source with an axial mag-

Figure 7.2

Axial field Kaufman ion source with two-grid extraction optics [13].


netic field, where the cathode is mounted on-axis and the cylindrical anode is located at the periphery of the chamber. The magnetic field strength, which is several tens of gauss on average, decreases in the extractor direction. The screen electrode is at floating or cathode potential. The main drawback of these systems is the radial nonuniformity of the plasma (the average-to-peak current density ratio is 0.4–0.6). For this reason, these sources use a multi-aperture extractor of diameter 10 cm or smaller or a single-aperture. When argon is used as a feed gas, the discharge loss is 300 to 800 eV ion–1; the working gas pressure is about 5 10–4 Torr. A considerable reduction in the ion energy cost and improvement of the mass utilization efficiency, essential for space propulsion ion thrusters, was first accomplished in Moore’s source with a magneto-electrostatic containment discharge chamber [16] where a particular array of permanent magnets and multiple strip anodes was used. The anode electrodes were located in the region of strong transverse magnetic field between the magnetic poles of the multicusp structure. The decrease in the effective surface area of the anode creates a positive anode drop in potential, preventing flow of ions to the anode. An additional anode immersed in the plasma is required for stable discharge operation. Later [17] the use of a multicusp magnetic field was proposed for the formation of dense uniform plasma in large-scale systems. However, at high gas pressure, which is necessary for generation of high-density plasma, negative plasma potential relative to the anode could not be obtained throughout the entire volume of the discharge chamber. Moreover, the discharge efficiency was impaired at high gas pressure [33]. The cathode filaments in electrode systems that use a multicusp magnetic field are located in the magnetic-field-free region, and primary electrons oscillating in the plasma can move to the anode only along magnetic field lines. The electron loss area is small, and is determined by the total length of the poles and the effective thickness of the magnetic slits, which are about twice the primary electron gyroradius [34]. Low energy plasma electrons and ions are confined in the plasma by the combined action of the magnetic field and the self-consistent electric field. A single ring-cusp magnetic field produced by permanent magnets was used in the ion source of a neutron generator (Figure 7.3) [35]. All the components of the cylindrical discharge chamber, except the cathode and the anode, are at floating potential. The shape and position of the discharge system electrodes and the parameters of the magnetic field in this source were optimized carefully. As a result a deuterium ion beam with current up to 0.2 A and ion energy of 200 keV was formed at a discharge current of 10 A and a low gas pressure of 0.25 Pa using a 1.3 cm singleaperture ion extractor. An example of a technological ion source of this type is the CHORDIS (cold or hot reflex discharge ion source) [36], in which the end-plate electrodes of the discharge chamber are at cathode potential and the cylindrical anode is lined with 18 permanent magnets producing a multicusp field. A filament cathode (for the cold version) or a vapor generator (for the hot version) is installed opposite to the screen electrode of the ion optics. The cold version of the source has a single-aperture ion extraction system with hole area of 2 cm2 and provides a 71 mA beam of xenon ions at an extraction voltage of 50 kV and a discharge power of 1.8 kW.

113

114


Figure 7.3 Single ring-cusp ion source [35]. Dashed lines show the shape of the magnetic field. 1 – Vacuum chamber; 2 – heat shield; 3 – magnet ring; 4 – gas inlet; 5 – cathode filaments; 6 – reflector; 7 – anode ring; 8 – aperture plate.

High gas efficiency (up to 90%) and low energy loss (80 W A–1) have been achieved in a laboratory model of an inert gas thruster [37] in which a ring-cusp magnetic field is produced by three rows of samarium–cobalt permanent magnets mounted on the cylindrical wall of the discharge chamber 25 cm in diameter. A thermionic hollow cathode was placed on the axis of the chamber, while all elements of the discharge chamber, except the cathode and the screen electrode of the ion optics, were at anode potential. For a xenon pressure of 1.3–13 10–3 Pa dense highly ionized (‡10%) plasma is generated in the chamber. A beam of xenon ions with current up to 6 A is formed with a multi-aperture ion optics system. About 95% of the beam is contained within a 14 half-angle. High-current ion sources with a large number of magnetic cusps have been termed “bucket-type” ion sources. The discharge chamber of these sources, which is at anode potential, is almost fully lined with permanent magnets. Cathodes may be installed either on the back plate opposite to the screen electrode or the side walls of the chamber. The distance between rows of opposite polarity magnets determines the size of the region in which the field drops by a factor of e ~ 2.7. The design of a source [38] developed for neutral beam injectors is shown in Figure 7.4. The cubic discharge chamber with wall 24 cm long was held at anode potential, while the screen grid could be at floating or cathode potential. Rows of samarium–cobalt per-


Figure 7.4 Cross-section of multicusp ion source [38]. 1 – Cathode filament; 2 – SmCo magnets; 3 – cooling tube; 4 – pulsed gas valve; 5 – insulator; 6 – extraction grid; F – filament power supply (8 V, 1 kA); D – discharge power supply (80 V, 700 A).

manent magnets were placed every 4 cm on the chamber walls. The magnetic field at the poles was up to 4 kG. Eight tungsten filaments maintained a discharge current as high as 600 A at a gas pressure (D2) of 5 10–3 Torr. The ion current density at the extraction grid reached 0.4 A cm–2. The plasma non-uniformity did not exceed 4% over a 10 10 cm2 area. The discharge efficiency was 1.63–1.75 A kW–1. Forrester [39] obtained higher discharge efficiency in the IBIS (intense boundary ion source) having a round chamber 25 cm in diameter and 40 cm long. A specific feature of this source was the use of narrow iron anode strips mounted at magnetic poles, which increased the field to 0.32 T, and a lanthanum hexaboride hollow cathode located on the conical end face of the chamber. The plasma density uniformity is about 4% over an 18 cm diameter. The current density of hydrogen ions was 0.33 A cm–2 at a discharge current of 330 A and a neutral gas pressure of 5 10–3 Torr. The overall power cost, which was estimated assuming a 30% extractor grid transparency, was less than 0.6 kW per extracted ampere of ion current. This high efficiency was probably due to the decreased ion loss area. A source with a 24 57 cm2 rectangular chamber 30 cm long, developed for neutral beam injector application, provided a hydrogen or deuterium ion current density as great as 0.4 A cm–2 at the 10 40 cm2 extractor grid, with a non-uniformity less

115

116


than 5% [40]. The magnetic field was created by SmCo magnets arranged in 36 longitudinal rows. Thirty four tungsten filaments provided a discharge current up to 1 kA in a pulse 30 s long. The energy efficiency of the source was 0.6 A kW–1 at a beam current of 40 A and an accelerating voltage of 80 kV. The gas consumption was 10 to 15 mm Hg l s–1. A similar energy efficiency of 0.63 A kW–1 was obtained in a source [41] with a rectangular chamber 25 40 cm2 in size and 34 cm long at a hydrogen ion current density of 0.25 A cm–2 at the 12 27 cm2 extractor grid with 35% transparency. Continuous linear cusps were normal to the beam axis. Six tungsten cathode filaments were mounted on the side wall of the source. When the magnetic field at the pole surface was reduced from 2.7 to 0.6 kG the ion source energy efficiency decreased to 0.36 A kW–1. To evaluate the suitability of this source for ion implantation and ion beam sputtering technologies, operation of the source [42] with different feed gases was studied. Ion beams of helium, nitrogen (N2+), oxygen (O2+), neon, argon, krypton and xenon were produced at currents of 23.5, 10.7, 6.2, 6.6, 6.2, 5.0 and 4.7 A, respectively, for an accelerating voltage of 35 to 60 kV, with an extractor grid with 1020 apertures each of diameter 4 mm. The discharge efficiency varied between 0.5 and 1 A kW–1 at a gas pressure of 8 mTorr, except for oxygen (0.25 A kW–1 at 9 mTorr). A 38 cm ion source for large-capacity broad-beam etching and deposition applications generated an argon ion beam having a current of 4–5 A at energy of about 1 keV [43]. The cylindrical and end-plate parts of the discharge chamber held permanent magnets producing a multicusp magnetic field. Four hot filaments were installed on the end plate of the chamber. The required gas flow rate was about 14 sccm of argon per ampere of ion beam. To obtain a high beam current at low accelerating voltage it was necessary for the extractor grids to have a large diameterto-gap ratio. The elevated temperature at the center of the thin grids causes them to become dish-shaped. To stop the grids from touching each other, the maximum diameter-to-gap ratio was about 50. The use of “pre-dished” molybdenum grids instead of flat ones allowed a decrease in the dishing depth to about 0.1%. As a result, a diameter-to-gap ratio of nearly 400 was realized in a 38 cm ion source made for uniform (5%–6%) etching of large area substrates. A 50 cm diameter ion source for ion beam etching and deposition is described in Ref. [44]. The source with filament-type cathodes was developed to generate beams of argon ions with energy 300–900 eV and current 1–5 A. Operation at higher current levels was possible with high-emissivity hollow cathodes. 7.2.2

High-Frequency Ion Sources

High-current RF broad beam ion sources were first developed as thrusters for space propulsion in the 1960s [45]. They have been commercially exploited for a number of industrial applications since the 1970s and used in neutral beam injectors since the 1990s. Two types of inductively coupled RF structures are most common. In one type, the induction coil is placed outside a glass or quartz cylindrical tube in which


the plasma is generated. The other type has a metallic discharge chamber with an RF antenna inside the discharge plasma. According to Ref. [46], the drawbacks of ion sources with external antenna include difficulties in producing uniform plasma in large volume and rectangular shaped chambers, thermal limitations for high power RF discharge, and deposition of conducting films on the tube walls. However the beam parameters for these two types of ion sources differ insignificantly and RF ion source performance is close to that of filament driven ion sources. The RIM-350 ion source [47], shown in Figure 7.5, employs a helical, water-cooled RF coil mounted around a quartz discharge vessel 35 cm in diameter. The quartz diffuser distributes injected gas around the periphery of the discharge chamber and thus improves the plasma uniformity. Due to the design of the RF coil and the RF skin-effect, high temperature electrons are distributed near the outer periphery of the discharge. Consequently the usual decrease in plasma density near the walls is compensated for, leading to a uniform distribution of current density. The ion source operates at a frequency of 1.8 MHz, which is in the optimal range for the source size. With an input RF power of about 2 kW and gas pressure of about 1 mTorr, the plasma is of density 5 1011 cm–3, corresponding to an argon ion saturation current density of nearly 9 mA cm–2. The source generates beams of gas ions (Ne+, Ar+, Kr+, Xe+, O2+, etc.) at an ion energy in the range 50–1000 eV and beam current 0.15–1.5 A. The source is intended for various etching and deposition applications. Etch uniformity is better than 3% on targets up to 250 mm in diameter.

Figure 7.5 Schematic of the RIM-350 radio frequency ion source with external coil [47]. RF – RF power supply; MB – match box; 1 – gas inlet; 2 – discharge vessel; 3 – diffuser; 4 – RF coil; 5 – three-grid extraction optics.

A large area RF source with external antenna [48] developed for fusion experiments used a nearly rectangular chamber with quartz walls 32 61 cm2 in size and 19 cm long. The chamber mounted a 6 turn antenna on the outside and a water cooled Faraday shield on the inside. To improve the plasma confinement, the backplate was covered with rows of samarium–cobalt permanent magnets. The working RF frequency was adjustable between 0.75 and 0.95 MHz, and the ion extraction sys-

117

118


tem measured 51 23 cm2. At 120 kW RF power the source produced beams of hydrogen ions at 90 A and 55 kV, and deuterium ions at 75 A and 60 kV. A 25 cm diameter ion source with internal antenna [46], also designed for neutral beam injection (Figure 7.6), had permanent magnets on the cylindrical wall and on the back plate of the aluminum chamber. The magnets produced an azimuthally symmetric multicusp field with magnetic field intensity at the poles of 1.5 kG. The diameter of the three-turn antenna of insulated copper tubing was 10–15 cm. The high frequency generator (~1 MHz) was rated at 30 kW, while the power transfer to the plasma was 90%. The RF plasma was triggered by a 50 mA, 45 V auxiliary discharge between the cathode and the chamber walls. The deuterium ion current density was 0.25 A cm–2 at an RF power of 20 kW, while the plasma non-uniformity was 5% over a diameter of 15 cm. The high density and good homogeneity of the RF-generated plasma led to the development of a large RF ion source for operation with an ion extraction system of size 10 40 cm2. An RF discharge in a chamber measuring 23 52 23 cm3 with a multicusp magnetic system at the side-walls and an internal antenna shaped as a single rectangular turn 13.5 43 cm2, provided an ion current density of 0.15 A cm–2 and 0.18 A cm–2 for deuterium and hydrogen ions, respectively, at a gas pressure of 4 mm Hg and an RF power of 25 kW. The non-uniformity of the ion current density was 7 % over the extraction grid area. The size of the antenna coil exceeded the size of the ion extraction system to avoid damage of the antenna by backstreaming electrons from the accelerator gap. For the same reason, the permanent magnets were removed from the back-plate of the discharge chamber, since the strong magnetic fields can reflect the electron backflow toward the antenna.

Figure 7.6 RF ion source with internal coil [46]. 1 – Pole pieces of soft steel; 2 – igniter; 3 – gas inlet; 4 – antenna feedthrough; 5 – copper end-plate with diagnostic probes; 6 – magnetic softsteel shield.


A rectangular ion source [49] with a 24 cm3 cubic copper chamber, in which Sm– Co permanent magnets produced a magnetic field with axially symmetric line cusps, and a two-turn rectangular antenna measuring 15 15 cm2 was tested for voltage up to 80 kV. The hydrogen ion current density was as high as 0.15 A cm–2. The energy efficiency (0.6 A kW–1) and the angular divergence of the beam (0.7 1.1) generated by a multi-slit ion extractor proved to be no worse than for a hot-filament ion source with the same bucket chamber. Another method for formation of large-area uniform plasmas by means of RF discharges is by using a number of small generators connected to a large volume expansion chamber, as proposed in Ref. [23]. The experiments were carried out using a single plasma generator (Figure 7.7). Dense plasma was produced in a quartz tube 3.2 cm diameter and 30 cm long equipped with an external 15 cm long helical antenna and a magnetic field coil to establish an axial magnetic field of up to 500 G. The RF radiation frequency was 14 MHz and the input RF power was 1 kW. The 26 26 28 cm3 expansion chamber was surrounded by 28 poles of magnetic cusp lines. The hydrogen ion current density from the plasma in the tube was greater than 80 mA cm–2 for a gas pressure of 10–5 Torr and magnetic field strength of 300 G. The ion current density from the expansion chamber plasma increased with increasing magnetic field near the neck of the quartz tube up to values sufficient for formation of a magnetic mirror retaining the plasma in the tube. The max-

Figure 7.7 Schematic of RF plasma generator with an expansion chamber [23]. 1 – Permanent magnets; 2 – end plate; 3 – ringshaped samarium–cobalt magnet; 4,5 – magnetic coils; 6 – gas inlet; 7 – RF antenna.

119

120


imum current density obtained at a pressure of 9 10–6 Torr exhibited a linear dependence on the RF power and was equal to 2.8 mA cm–2 per kW. The nonuniformity of ion current density distribution was –5% over the plasma region of about 14 cm2 in size. The plasma density and uniformity may be improved if one uses more RF generators, but this complicates the design and operation of the device. Moreover, the energy efficiency of this plasma generating system is 3–4 times lower than the energy efficiency of a filament-type source with a similar chamber. Off-resonance microwave sources [50] also provide high-current beams of singly charged ions. The discharge chamber in Sakudo’s source (see Chapter 10) was made as part of a coaxial wave guide separated by a vacuum-tight ceramic window. Microwave radiation (2.45 GHz) was fed into the chamber using a rod antenna, which served as the internal conductor of the coaxial transmission line. An inhomogeneous magnetic field with mirror ratio about 2 was produced in the discharge chamber. When the magnetic field intensity was changed over a range always satisfying the ECR condition, the absorbed power increased with increasing magnetic field, but an ECR-resonance dependence was not observed at an argon pressure of 8 10–3 Torr. Microwave energy transfer efficiency at maximum magnetic field of ~2 kG was over 90% and the plasma density was 10 to 100 times greater than for ECR discharges. A 40 mm diameter multi-aperture ion extractor was used to generate beams of argon (200 mA), hydrogen (400 mA), and oxygen (110 mA) ions with energy 5 keV for a microwave power of 600 W. This corresponds to an energy efficiency of ~0.67 A kW–1 for the case of hydrogen ions. An ion source in which the magnetic field satisfies the ECR condition near the plasma boundary so as to increase the extracted ion current density is described in Ref. [51]. The working frequency was 2.45 GHz. A specific feature of the source was that radiation was coupled into the plasma through a three-layer window. A quartz disk ensured vacuum tightness of the joint, a ceramic disk decreased microwave reflection, and a silicon nitride disk protected the window from the back-flux of energetic electrons from the accelerating gap. The ion beam was formed by a three-electrode ion-optical system with 3.5 mm holes having a total area of 10 mm2. For microwave power of up to 1 kW, beams of oxygen, nitrogen and hydrogen ions with current up to 75 mA were generated at an accelerating voltage of up to 50 kV. After 700 h of source operation, craters up to 0.5 mm deep were formed on the 12 mm thick silicon nitride disk. A bucket-type ion source using microwave plasma produced under ECR conditions as a plasma cathode instead of a filament cathode has been described in Ref. [24]. The ion source (Figure 7.8) employed a microwave plasma generating chamber and an arc plasma generating chamber 180 mm long and 200 mm across, surrounded by permanent magnets forming a multicusp magnetic field near the chamber wall. Electrons were extracted from the microwave plasma through a mesh grid at the sidewall of the electron-extracting electrode. When a potential difference of 30 to 70 V was applied between the mesh grid and the chamber walls, a plasma was formed in the second stage. The plasma discharge current increased with increasing neutral gas pressure (0.13–0.013 Pa) and applied voltage, and reached ~2.5 A. A two-grid 100 mm diameter ion extraction system with 4 mm diameter beamlet


Figure 7.8 Bucket type ion source using a microwave plasma cathode. (Reprinted from Ref. [24] with permission from Elsevier). 1 – Window; 2 – microwave plasma generation chamber; 3 – permanent magnet; 4 – insulator; 5 – arc plasma generation chamber; 6 – gas inlet; 7 – electron extraction electrode; 8 – ion extraction electrodes

holes provided a pulsed (10 ms) 60 mA beam of argon ions. Another type of microwave plasma cathode was describeded in Ref. [52], where electrons were extracted from dense microwave-produced plasma through three holes of diameter several mm that were equally spaced one from another on a circle 60 mm in diameter. The main discharge chamber of diameter 200 mm was surrounded by 12 columns of SmCo magnets forming a line-cusp magnetic field configuration. A 115 mm diameter ion extractor formed high-current low-energy (1 keV) ion beams of 230 mA for argon and 132 mA for oxygen with excellent beam uniformity. 7.2.3

Cold Cathode Ion Sources

The glow discharge can operate in a quasi-steady-state mode at a voltage of several hundred volts and current density at the cathode of up to a few tens of A cm–2 [53], and it is possible to use this discharge in high current ion sources. A high-current low-pressure glow discharge can be formed in hollow-cathode systems and in systems with a magnetic field to provide both radial confinement and axial oscillation (“reflexing”) of ionizing primary electrons. However, large-area uniform plasmas can only be produced in weak magnetic fields. Thus the usual plasma configurations need to be modified. To ensure energy efficiency of the ion source, one needs to increase the ion current extracted from the plasma relative to the ion current to the cathode of the glow discharge. But since the ion current to the cathode determines its electron emission, effective extraction of ions changes the plasma parameters and the conditions for stable plasma operation. Some features of ion extraction

121

122


and broad beam formation with glow discharge plasmas are due to the effect of the cathode sheath, whose thickness is comparable with the size of the ion optics apertures. Basic Physics of Low Pressure Glow Discharges For typical ranges of gas pressure p and magnetic field B used in ion sources, the voltage Ud applied across the discharge gap is localized at the cathode sheath and determines the energy of ions interacting with the cathode surface and of primary electrons supplied to the plasma. For eUd < 1 keV the ratio between the electron and ion currents at the cathode usually does not exceed c ~ 0.1. A primary electron produces a number of ions, N ~ eUd/w, provided that the full path of the electron before it goes to the anode, L, is sufficient for complete energy relaxation of the electron, i.e. the relationship L/ki ‡ eUd/w is fulfilled. Here w is the energy cost of an ion in the plasma and ki is the ionization mean free path of an electron. Therefore, for selfsustained operation of the glow discharge plasma, cN = 1, the gas pressure, which determines the oscillating electron path length L, should be high enough, and the energy required to emit an electron, eUd/c, is an order of magnitude larger than the energy introduced by the electron into the plasma. According to Ref. [54], the minimum neutral gas pressure required for a hollow cathode glow discharge is determined by the ratio between the area of the outlet aperture of the cathode and the area of the cathode surface, g = Sa/Sc. For g ~ 10–2–10–3 the gas pressure may be reduced to ~10–2 Pa. However, a decrease in the anode size impedes the efflux of low energy plasma electrons to the anode in high current discharges, and consequently the discharge operating voltage rises and the gas pressure must be increased. When Sa/Sc < (m/M)1/2 (m and M being the electron and ion masses respectively), a space-charge double layer is formed at the outlet aperture of the hollow cathode. If the anode is placed in the cathode cavity, a positive anode drop in potential occurs. The magnetic field that is used to confine the primary electrons also slows down the diffusion of plasma electrons to the anode. As a result the spatial non-uniformity of the plasma may increase and the plasma structure may even change. In an electrode system of the inverse magnetron type, formed by a cylindrical hollow cathode and a rod anode on the axis, increasing magnetic field leads to formation of a positive anode sheath. For low gas pressure and high discharge current, the voltage drop at the sheath may be up to several hundred volts. In a PIG electrode system with a cylindrical anode and plane end-cathodes, increasing magnetic field causes rotational instability of the plasma. Ion bombardment leads to sputtering of the cathode material, which may consume [55] up to 10% of the energy acquired by ions in the cathode sheath, while most of the ion energy (70–90%) is released as heat. Argon ions with energy between 400 and 800 eV, as is typical for glow discharges, sputter 0.5–1 atoms ion–1 from an Fe cathode [56]. 7.2.3.1


7.2.3.2 Glow-Discharge-Based Plasma Emission Systems The energy efficiency of a glow discharge ion source is defined as the ratio Ib/(IdUd) or as a/Ud, where a = Ib/Id is the efficiency of ion extraction from the discharge plasma (approximately equal to the ratio of extracted ions to the total number of ions in the plasma), Ud is the discharge operating voltage, Ib is the extracted ion beam current, and Id is the discharge current. If the ion current is distributed uniformly over the cathode surface, a may be estimated approximately as the ratio between the surface areas of the plasma emitter and the cathode, a ~ Se/Sc. However, if Sc decreases and Se remains unchanged, the discharge operating voltage and the gas pressure need to be raised. If the primary electrons expend all their energy in the plasma, then the discharge operating voltage Ud ~ 1/(1–a) [57]. Therefore, the a dependence of the energy efficiency varies as a/(1–a), and the maximum of this function is a = 0.5. However the high gas pressure necessary to maintain the discharge at high a values limits the ion extraction and energy efficiency values of high-voltage ion sources. Uniform plasma is generated in the hollow cathode discharge due to multiple reflections of primary electrons by the cathode sheath and collisions in the plasma, leading to an isotropic spatial distribution of electron velocities. Conversely, the use of a magnetic field in the plasma for confinement of primary electrons leads to plasma anisotropy. In an inverse magnetron plasma configuration the anode is in the form of a thin rod on the axis of a cylindrical cathode, parallel to an applied magnetic field. Crossed electric E and magnetic B fields facilitate plasma initiation, and the low primary electron loss rate ensures a high current discharge mode at low pressure. Primary electrons emitted by the cylindrical part of the cathode drift azimuthally. The amplitude of radial oscillations of drifting electrons, which is determined by their Larmor radius = (1/B)(2mUd/e)1/2, influences the radial profiles of the plasma density and electron temperature. Thus it is possible to control the radial distribution of the ion current density ji(r) extracted from the plasma along the magnetic field [58]. As the magnetic field increases, the distribution ji(r) with maximum current density on the axis transforms to a distribution with a minimum on the axis. The ion current density distribution is approximately uniform when is approximately equal to the hollow cathode radius. The Penning electrode system (PIG) is difficult to use in broad ion beam sources because of the substantial radial nonuniformity of the plasma. The plasma nonuniformity may be decreased if the cylindrical electrode system is fitted with a multicomponent cathode. The potential difference between the ring-shaped cathode elements and the anode decreases towards the system axis. As the energy of the primary electrons drops, the ion generation rate near the axis decreases, and consequently the plasma becomes more uniform. The ion extraction efficiency in this modified PIG system is a ~ 30% because of the increased Se/Sc ratio. The power required to maintain a glow discharge plasma may be reduced if conditions are provided for volume multiplication of ionizing electrons. Generation of primary electrons with energy eUd as a result of ion bombardment of the cathode and acceleration of electrons in the cathode sheath requires an energy eUd/c, where c is the ratio between the electron and ion currents at the cathode, while creation of

123

124


an electron in the volume requires an average energy equal to the ion cost w. The ionizing capacity of the latter electron depends on the electric potential where the electron is formed. However, the rate of generation of secondary energetic electrons by ionization of the gas by oscillating primary electrons in the cathode sheath of high-current discharges in large gaps is low because of the low probability of this process, which is determined by the ratio between the sheath thickness and the size of the plasma column [59]. Two-stage electrode systems are more efficient; the low energy plasma electrons acquire additional energy in the space-charge double layer or are injected to the second stage through an artificially supported space charge layer with adjustable potential drop. Electrode systems of each stage are optimized to facilitate triggering and operation of the discharge and increase the ion current and improve plasma uniformity. In the duoPIGatron type ion source with cold cathode (Figure 7.9) [60], ions are extracted from the plasma in the anode region of the discharge, which represents a PIG electrode system in a weak magnetic field (~10–2 T). The cathode stage with gas feed also has a smaller PIG system in a stronger (10–1 T) magnetic field. An auxiliary discharge is excited in the cathode stage at relatively low voltage and produces plasma with density sufficient to disrupt the cathode sheath in the hole between the chambers. The plasma, which penetrates to the anode region, provides conditions for initiating the main discharge. A double layer that accelerates electrons is formed near the hole. Such two-chamber systems with gas-discharge-based triggering systems were developed first for arc plasma electron sources [61] in which contribution

Figure 7.9 Electrode system of DuoPIGatron type ion source with cold cathode [60]. 1 – Hollow cathode; 2 – auxiliary anode; 3 – ring-shaped samarium–cobalt magnet; 4,5 – end plates; 6 – anode; 7 – screen electrode; 8 – magnetic coil; U1 – auxiliary power supply; U2 – main power supply.


of the plasma electrons to the electron beam current was insignificant and, therefore, a magnetic field was not used in the second stage. The main drawback of this kind of ion source is radial nonuniformity of the plasma, which can be eliminated by using a multicusp magnetic field near the anode region. In another type of two-stage system, for example see Ref. [62], ions are extracted from the plasma generated in the cathode region, which communicates through one or several small diameter holes to a small anode chamber where the gas feed is located (Figure 7.10). A double electrostatic layer with potential drop ~50 V that separates the cathode and anode plasmas appears on the cathode side of the holes. The cathode plasma is produced by primary electrons that are accelerated in the cathode sheath and dissipate their energy by interactions with neutrals. Low energy electrons are accelerated and focused by the double layer onto the holes, where dense anode plasma is formed due to high gas density and electron flow. Plasma formation is via a beam–plasma interaction. Ions from the anode plasma are accelerated in the double layer, cross the cathode chamber, and are extracted through the screen grid holes. To facilitate discharge initiation, a PIG structure is located in the cathode region, for which purpose an auxiliary anode is installed and a magnetic field applied. The extracted ion current is distributed more uniformly for configurations with a magnetic field produced by several rows of permanent magnets fitted on the cathode [63].

Figure 7.10 Electrode system of a technological ion source [62]. 1 – Hollow anode; 2 – auxiliary anode; 3,4 – magnetic poles; 5 – perforated cathode disk; 6 – screen grid; 7 – magnetic core; 8 – magnetic coil; 9 – accelerating grid

A very low gas pressure (~3 10–5 Torr) was achieved in a two-stage electrode system with injection of electrons from a hollow-cathode glow-discharge plasma [25]. Electrons are injected through a fine mesh grid separating the cathode and anode regions into a chamber in which a thin-wire anode whose potential relative to the chamber walls can be adjusted independently is located. The fine grid separates the cathode and anode plasmas and ensures formation of a space-charge layer with adjustable potential drop. High transparency of the grid allows extraction of up to 90% of electrons from the first stage plasma. Oscillation of fast electrons in the

125

126


Figure 7.11 Electrode system of the bucket ion source with glow-discharge-based plasma cathode [64]. 1 – Hollow cathode; 2 – igniting electrode; 3 – fine mesh grid; 4 – anode; 5 – screen grid; 6 – permanent magnets; PS1 – glow discharge power supply; PS2 – plasma generator power supply.

chamber housing the thin-wire anode provides efficient ionization of the gas. The efficiency of ion extraction from anode plasma in this system is determined by the ratio between the surface area of the plasma emitter and the second stage chamber walls. To increase the a value a magnetic bucket-type chamber can be used for the second stage of electrode system [64] (Figure 7.11). The power supply PS1 sustains a glow discharge between the hollow cathode and a fine mesh grid placed opposite the output aperture of the hollow cathode and connected electrically to the screen grid. Increasing the output voltage of PS2 up to ~200–300 V results in switching the glow discharge current (1 A) to the anode, while the screen grid current changes its sign and increases to values accounting for ~0.4 of the injected electron current. The ion current density distribution is close to uniform over the magnetic-field-free region of the plasma (8 cm). The rate of ion generation in the second stage depends on the glow discharge current, the energy of injected electrons, and the gas pressure (0.01–0.05 Pa) which determines the ratio of the full path before going to the anode, L, to the ionization mean-free-path ki of injected electrons. A two-grid, 8 cm diameter, ion extraction system produces a 60 mA beam of 3 keV argon ions when the current of the plasma cathode is 0.2 A. The mass composition and charge state of the plasma ions depend on the glow discharge current and operating voltage, and the surface conditions of the cathode. As the discharge current is increased, both the density of ionizing electrons and the flow of sputtered atoms increase. As a result, the rate of metal ion generation rises. The fractional content of metal ions in the ion beam reaches 18% for pulsed mode


operation in argon with a cathode made from Al when the discharge current is 45 A [66]. However if the cathode temperature rises to close to thermal diffusion threshold values for the gas atoms and the feed gas reacts with the cathode material, then the content of metal impurities in the ion beam can be decreased to fractions of a percent due to the low rate of selective sputtering of metal atoms [58]. 7.2.3.3 Extraction of Ions and Broad Beam Formation There is a substantial effect of the space-charge sheath formed between the plasma and the screen grid on the position and shape of the plasma boundary and ion beam parameters; this is due to the significant potential drop and large thickness of the sheath, which is comparable with the hole diameter. In this case the location of the plasma boundary at the center of the holes depends on the field in the accelerating gap, while its position near the hole sides is determined by the sheath thickness. The effect of screen grid potential on the maximum beam current density of low energy (£ 1 keV) argon ions extracted from glow discharge plasma by a multiaperture three-electrode system when the electric field intensity in the beam extraction gap is close to breakdown field intensity, was studied in Ref. [66]. The dependence of ion beam current on the discharge current for the ion optics with floating screen grid was linear up to a value of saturation ion current from plasma ~10 mA cm–2, while when the screen grid was at cathode potential, the rate of ion beam current increase decreased with discharge current. A computer simulation has shown that for conditions when the thickness of the cathode sheath is comparable to the radius of the ion extractor apertures, and the intensity of the accelerating field is insufficient to move dense plasma away, a convex plasma boundary is formed and the ion losses to the screen electrode increase. At lower values of plasma density and field intensity the curvature of the plasma meniscus at the extractor changes with increasing accelerating voltage, the ion current to the extractor grid decreases, and the beam current rises. Consequently the ion extraction efficiency a may differ by 30– 50% from the a value estimated from the ratio of the hole cross-section to the cathode surface area. The parameters of the sheath between the plasma and the grid also influence the angular divergence of the beam. When the discharge voltage in a Kaufman source increases by only 18 V, the ion beam divergence half-angle is decreased from 12 to 9.5 [67]. The large sheath thickness in the glow discharge causes a sizeable decrease in the beam divergence; minimum beam divergence corresponds to an optimal combination of the accelerating field strength and the voltage drop in the sheath [66], if the latter is adjusted independently. In high voltage ion sources [68], (say, extraction voltage of some tens of kV and ion beam current density of 1– 10 mA cm–2), increasing the extraction gap and the hole diameter suppresses the effect of the cathode sheath and ensures both a high ion extraction efficiency and an optimal angular divergence of the ion beam. The latter, along with the ion current density distribution at the plasma boundary, influences the profile of the beam current density at the target.

127

128


7.2.3.4 Design and Performance of Cold Cathode Ion Sources The sources described here are used primarily for ion etching of surfaces, surface modification of materials by ion implantation, and ion beam assisted thin-film deposition. Since cold cathode sources do not require power input to maintain cathode temperature, a repetitively-pulsed operational mode in which the pulse repetition rate or pulse length are adjusted can provide an efficient means for controlling the average beam current, while the optimal conditions needed for beam formation at the extractor remain unchanged. Thus materials processing can be carried out over a wide range of sample temperature and implantation dose. Figure 7.12 shows a general view of an ion source with a plasma configuration of the inverse magnetron type [68]. The electrode system of this source consists of a stainless-steel cathode 150 mm in diameter and a tungsten rod anode. A magnetic field coil surrounds the source casing. A broad ion beam with circular cross-section is formed by a two-grid extractor. The total area of the 8–12 mm diameter beamlet holes is 70 cm2, and the pulsed beam current for repetitively pulsed mode is 100– 500 mA for an extraction gap aspect ratio (ratio of hole diameter to grid separation) of 0.4–0.5, and 50–100 mA for DC operation for an aspect ratio of 0.3–0.4. The accelerating voltage was 30 to 40 kV. The discharge operating voltage was 0.5–0.7 kV at a current of 0.5–1 A, the feed gas (Ar) flow rate 30 sccm, and the vacuum chamber

Figure 7.12 Ion source with inverse-magnetron-type electrode system [68]. 1 – Cathode; 2 – anode; 3 – magnetic field coil; 4 – high voltage insulator; 5 – screen grid; 6 – accelerating grid; 7 – water cooling jackets; 8 – cable lead-in; 9 – gas inlet; 10 – insulator.


Figure 7.13 Schematic diagram of an oxygen ion source [63]. 1 – Anode; 2 – cathode; 3 – extraction grid; 4 – permanent magnets; 5 – insulator.

gas pressure 0.03–0.05 Pa. The ion beam current accounts for 5–6% of the discharge current in pulsed mode and up to 7–8% in the DC mode of beam generation. The ion extraction efficiency was increased to 16–20% by replacing the cylindrical hollow cathode by a truncated cone with base plane at the extraction grid. The a value (ratio of the extracted ion current to the discharge current) decreases as the cone base angle increases. The ion current density nonuniformity is less than 10% over an area accounting for ~50% of the surface area of the cathode end face for B ~ 2 10–3 T. This source has been used for ion beam assisted deposition of Ti–TiN multilayer coatings to reduce the abrasive wear of blades in gas turbine engines. Figure 7.13 shows the design of a source of low energy (~1 keV) oxygen ions [63]. The source includes an 18 mm diameter anode that is mounted on the axis of a cold hollow cathode 230 mm in diameter and 60 mm long. The extractor grid is fitted on the end face of the cathode. The magnetic field in the cathode chamber is produced by three circular rows of Sm–Co permanent magnets. Cathode and anode plasmas separated by a double layer with a voltage drop of about 40 V are generated in this two-stage discharge. The voltage drop in the cathode sheath of the discharge is 200– 300 V. Primary electrons that are emitted by the cathode and accelerated in the cathode sheath are confined by a magnetic field and efficiently ionize the gas in the cathode chamber. Electrons that are accelerated in the double layer lose energy through collective interactions in the anode region of the discharge. The discharge current is 0.6–1.2 A, and the gas flow rate 9 to 12 sccm. The current in the extractor circuit accounts for 40% of the discharge current and the ion beam current accounts for 13% of the discharge current. Nonuniformity of the beam current density over a 200 mm diameter is less than 5%. Tests for 400 h showed no change in the electrode geometry by ion sputtering and the source had reproducible parameters throughout the tests. The fraction of impurity metal ions in the beam of sources of this type usually does not exceed 2%. An ion source with external injection of electrons [19] is shown in Figure 7.14. The auxiliary discharge between a hollow cathode 60 mm in diameter and 80 mm long and a fine wire mesh grid fitted in the 6 mm outlet aperture is initiated by dielectric flashover. Electrons are extracted through the grid and maintain the primary non-self-sustained discharge between a 12 cm diameter aluminum cathode

129

130


Figure 7.14 Ion source based on non-self-sustained glow discharge with injection of electrons [19]. 1 – Thin wire anode; 2 – hollow cathode; 3 – fine mesh grid; 4 – hollow cathode for

auxiliary discharge; 5 – screen grid; 6 – accelerating and decelerating grids; 7 – high-voltage insulator; 8 – discharge igniting system; 9 – fan for forced air cooling.

and a 0.7 mm diameter tungsten rod anode. The injected electron current and neutral gas pressure determine the voltage across the primary gap necessary to sustain the required primary discharge current. For a primary discharge with pulsed current of 20 A at an argon pressure of 9 10–5 Torr, the applied voltage may be decreased from 700 to 200 V as the auxiliary discharge current increases from 1 to 10 A. A 100 cm2 ion beam is formed using a three-grid extractor system. The 5 mm diameter holes of the extractor electrode are covered with a grid of 0.5 0.5 mm2 mesh size. The beam current is up to 2 A in pulsed mode and up to 20 mA in DC mode for an extraction voltage of up to 30 kV. Although the fine grid expands the working range of the beam parameters for fixed extractor gap, the ion extraction efficiency decreases nearly proportionally to the grid transparency, and a (ratio of the extracted ion current to the discharge current) equals ~3.5–4%. The fraction of metal ions in the beam is less than 0.15% for a voltage across the main discharge gap of less than 100 V.

7.3

Conclusion

Dense and stable plasmas having a uniform current-density profile over large-area emission surfaces, as necessary for generation of high current beams of positive gaseous ions, can be produced using filament-driven, high-frequency and cold-cathode gas discharges. The excellent performance of available hot-filament gaseous ion

References

sources allows generation of steady-state gaseous ion beams with unprecedented high values of beam current. The ability to generate contaminant-free, intense beams of inert or reactive gas ions has driven the rapid development of high-frequency discharge ion sources, whose beam parameters are now not inferior to those of filament-driven sources. Recent studies of the features of low-pressure glow discharges have shown that it is possible to substantially increase the ion beam current and the energy efficiency of these sources, to decrease the metal ion content in the beam, and reduce the neutral gas pressure. We may look forward to the wide use of these kinds of ion sources in materials processing technologies. They are relatively simple to operate and maintain, and highly reliable.

References [1] W.B. Kunkel, in Fusion, edited by E. Teller

[2] [3]

[4]

[5] [6] [7] [8] [9]

[10] [11]

[12] [13] [14]

(Academic Press, New York, 1981), Chapter 12. E. Stuhlinger, Ion Propulsion for Space Flight (McGraw-Hill, New York, 1964). G. Dearnaley, J.H. Freeman, R S. Nelson and J. Stephen, Ion Implantation (North-Holland, Amsterdam, 1973). D. Bohm, in The Characteristics of Electrical Discharges in Magnetic Fields, edited by A. Guthrie and R.K. Wakerling (McGraw-Hill, New York, 1949), Chapter 3. M. Ardenne, Atomkernenergie, 1, 121 (1956). C.D. Child, Phys. Rev. (Ser. 1) 32, 492 (1911). I. Langmuir and K.T. Compton, Rev. Mod. Phys. 3, 251 (1931). W.A. Lamb and E.J. Lofgren, Rev. Sci. Instrum. 27, 907 (1956) H.R. Kaufman, Advances in Electronics and Electron Physics (Academic Press, New York, 1974), Vol. 36, p. 265. G.W. Hamilton, J.L. Hilton and J.S. Luce, Plasma Physics 10, 687 (1968). N.N. Semashko, A.A. Panasenko, and V.M. Kuligin, in Ion Injectors and Plasma Accelerators, edited by A.I. Morozov and N.N. Semashko (Moscow, Energoatomizdat, 1990), p. 133 (in Russian). W.B. Kunkel, Rev. Sci. Instrum. 61, 354 (1990). H.R. Kaufman, J.J. Cuomo and J.M.E. Harper, J. Vac. Sci. Technol. 21, 725 (1982). M.D. Gabovich, Physics and Technology of Plasma Ion Sources (Moscow, Atomizdat, 1972) (in Russian).

[15] I. Langmuir, Phys. Rev. 33, 954 (1929). th [16] R.D. Moore, in Proceedings of the AIAA 7

[17] [18] [19] [20]

[21]

[22]

[23] [24] [25] [26] [27] [28]

[29]

Electric Propulstion Conference (Williamsburg, Virginia, 1969) (AIAA paper No.69 – 260). R. Limpaecher and K.R. MacKenzie, Rev. Sci. Instrum. 44, 726 (1973). R. Geller, IEEE Trans. Nucl. Sci. 23, 904 (1976). N.V. Gavrilov and E.M. Oks, Nucl. Instrum. Methods A 439, 31 (2000). Yu.I. Belchenko, V.I. Davydenko, G.E. Derevyankin, G.I. Dimov, V.G. Dudnikov, I.I. Morosov, G.V.Roslyakov and A.L. Schabalin, Rev. Sci. Instrum. 61, 378 (1990). N.V. Gavrilov, Yu.E. Kreindel, G.A. Mesyats and F.N. Shvedov, Tech. Phys. Lett. 14, 865 (1988) (in Russian). R.S. Davis, O.B. Morgan, L.D. Stewart and W.L. Stirling, Rev. Sci. Instrum. 43, 278 (1972). Y. Oka, T. Shoji, T. Kuroda, O. Kaneko and A. Ando, Rev. Sci. Instrum. 61, 1256 (1990). Y. Hakamata, T. Iga, K. Natsui and T. Sato, Nucl. Instrum. Methods B 37/38, 143 (1989). E. Oks, A. Vizir and G. Yushkov, Rev. Sci. Instrum. 69, 853 (1998). R. Keller, Radiation Effects 44, 201 (1979). O.B. Morgan, G.G. Kelley and R.C. Davis, Rev. Sci. Instrum. 38, 467 (1967). R.A. Demirkhanov, U.V. Kursanov and V.M. Blagoveschenskiy, Tech. Phys. 34, 30 (1964) (in Russian). M.M. Menon, C.C. Tsai, J.H. Whealton, D.E. Schechter, G.C. Barber, S.K. Combs, W.K. Dagenhart, W.L. Gardner, H.H. Hasel-

131

132

[30]

[31] [32]

[33] [34]

[35] [36] [37] [38] [39] [40]

[41] [42]

[43]

[44] [45] [46] [47]

[48]

7 High Current Gaseous Ion Sources ton, N.S. Ponte, P.M. Ryan, W.L. Stirling and R.E. Wright, Rev. Sci. Instrum. 56, 242 (1985). J.P. Ruffell, D.H. Douglas-Hamilton, R.E. Kaim and K. Izumi, Nucl. Instrum. Methods B 21, 229 (1987). M. Fumelli and F.P.G. Valckx, Nucl. Instrum. Methods 135, 203 (1976). K.W Ehlers, W.R. Baker, K. H. Berkner, W.S. Cooper, W.B. Kunkel, R.V Pyle and J.W Stearns, J. Vac. Sci. Technol. 10, 922 (1973). W.L. Stirling , P.M. Rayan, C.C.Tsai and K.N. Leung, Rev. Sci. Instrum. 50, 102 (1979). K.N. Leung, N. Hershkowitz and K.R. MacKenzie, Phys. Fluids 19, 1045 (1976). J.P Brainard and J.B. O’Hagan, Rev. Sci. Instrum. 54, 1497 (1983). R. Keller, P. Spdtke and H. Emig, Vacuum 36, 833 (1986). J.R. Beattie and J.N. Matossian, Rev. Sci. Instrum. 61, 348 (1990). K.W. Ehlers and K.N. Leung, Rev. Sci. Instrum. 50, 1353 (1979). A.T. Forrester, D.M.Goebel and J.T. Crow, Appl. Phys. Lett. 33, 11 (1978). P.A. Pincosy, K.W. Ehlers, A.E. Lietzke, H.M. Owren, J.A. Paterson, R.V. Pyle and M.C. Vella, Rev. Sci. Instrum. 57, 2705 (1986). Y. Okumura, H. Horiike, and K. Mizuhashi, Rev. Sci. Instrum. 55, 1 (1984). S. Tanaka, M. Akiba, M. Araki, S. Matsuda, Y. Matsuda, Y. Ohara, Y. Okumura, K. Yokoyama and K. Watanabe, Nucl. Instrum. Methods B 37/38, 128 (1989). H.R. Kaufman, W.E. Hughes, R.S. Robinson and G.R. Thompson, Nucl. Instrum. Methods B 37/38, 98 (1989). A.V. Hayes, V. Kanarov and B. Vidinsky, Rev. Sci. Instrum. 67, 1216 (1996). H.W. Loeb, Am. Inst. Aeronautics 69, 285 (1982). W.F. DiVergilio, H. Goede and V.V. Fosnight, Rev. Sci. Instrum. 57, 1254 (1986). V. Kanarov, A.V. Hayes, R. Yevtukhov, B. Vidinsky and A. Navy, Rev. Sci. Instrum. 69, 874 (1997). W. Kraus, E. Speth, J.-H. Feist, P. Frank, B. Heinemann and R. Riedl, Rev. Sci. Instrum. 69, 956 (1997).

[49] H. Goede, W.F. DiVergilio, V.V. Fosnight,

M.C. Vella, K.W. Ehlers, D. Kippenhan, P.A. Pincosy and R.V. Pyle, Rev. Sci. Instrum. 57, 1261 (1986). [50] N. Sakudo, Nucl. Instrum. Methods B 21, 168 (1987). [51] G. Voronin, D. Solnyshkov, M. Svinin and A. Solnyshkov, Nucl. Instrum. Methods B 161–163, 118 (2000). [52] Y. Matsubara, H. Tahara, M. Takahashi and S. Nogawa, Rev. Sci. Instrum. 63, 2595 (1991). [53] L.Yu. Abramovich, B.N. Klyarfeld and Yu.N. Nastich, Tech. Phys. 36, 714 (1966) (in Russian). [54] A.S. Metel, Tech. Phys. 54, 241 (1984) (in Russian). [55] B.S. Danilin and V.Yu. Kireev, Applications of Low-Temperature Plasma for Sputtering (Moscow, Energoatomizdat, 1987) (in Russian). [56] H.H. Andersen and H.L. Bay, in Sputtering by Particle Bombardment I, edited by R. Behrish (Springer-Verlag, Berlin, 1981). [57] S.P. Nikulin, Tech. Phys. 45, 1351 (2000). [58] N.V. Gavrilov, G.A. Mesyats, S.P. Nikulin, G.V. Radkovskii, A. Elkind, A.J. Perry and J.R. Treglio, J. Vac. Sci. Technol. A 14, 1050 (1996). [59] A.S. Metel, Tech. Phys. 55, 1928 (1985) (in Russian). [60] V.Ya. Martens, S.I. Bel’uk and V.N. Posokhov, Instrum. Exp. Techn. No 2, 194 (1992) (in Russian). [61] Yu. E. Kreindel, Plasma Electron Sources (Moscow, Atomizdat, 1977) (in Russian). [62] B.I. Juravlev, V.V. Prilepskiy and V.S. Gorlatov, Instrum. Exp. Techn. No.3, 215 (1993) (in Russian). [63] A.I. Stognij and S.V. Koryakin, Instrum. Exp. Techn. 43, 783 (2000). [64] N.V. Gavrilov and A.S. Kamenetskikh, Rev. Sci. Instrum. to be published 75, 1875 (2004). [65] A.V. Vizir, E.M Oks and G.Yu. Yushkov, Izvestiya Vysshikh Uchebnykh Zavedenii, Fizika. 43, (No.2), 14 (2000) (in Russian). [66] N.V. Gavrilov, D.R. Emlin and S.V. Kuleshov, Rev. Sci. Instrum. 71, 1 (2000). [67] G. Aston, H.R. Kaufman, and P.J. Wilbur, AIAAJ 16, No.5, 516 (1978). [68] N.V. Gavrilov, D.R. Emlin, S.P. Nikulin, Russ. Phys. J. 44, 952 (2001).

133

8

Freeman and Bernas Ion Sources Marvin Farley, Peter Rose, and Geoffrey Ryding

8.1

Introduction

Ion implantation is used for doping wafers during the process of semiconductor manufacture. The production process may require twenty or more implants of different dopant species and doping levels. The ion implanter must be fully automated to obtain an economically viable and repeatable process. For this to be possible the ion source should have smooth and relatively repeatable characteristics. Ion sources used in the research laboratory are not usually known for such characteristics. The two sources described in this chapter have, after much effort, been developed to be adequate for manufacturing processing and are the culmination of source development started 50 years ago. In the 1950s and 1960s the production of highly purified isotopes had a high priority and to meet this need a number of different sources evolved. Most of these source designs were based on experience gained with the calutron source developed by Lawrence and his coworkers for the separation of uranium isotopes for the Manhattan project. A detailed description of this ion source development can be found in Ref. [1] and in an overall review in Ref. [2]. To obtain 100 mA beam current and high mass resolution as required for this application, the extraction slit of the calutron source was 380 mm long in the alpha source and 190 mm long in the beta source. These sources were operated between the poles of massive 180 magnetic mass spectrometers. The source necessarily had to be run in the several kilogauss magnetic field of the mass spectrometer, making it difficult to optimize source conditions even for a single source material. Research on the isotopes of all atomic species required a more flexible solution with higher mass resolution and reduced cross contamination. To achieve this, the ion source was moved to a position outside the spectrometer magnet. The advantages of this geometry are numerous. The source could be optimized independently of the space limitations and magnetic field of the separator magnet, making it easier to generate stable arc plasmas and to provide hash free beams of many isotopes. Of the many sources developed for this purpose, two types have become extensively used for ion implantation. Modifications of the Bernas source developed at the University of Paris by Bernas, Chavet and others is now the most commonly The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

134

8 Freeman and Bernas Ion Sources

Figure 8.1 Schematic diagrams of (a) Calutron, (b) Bernas and (c) Freeman ion sources. The ion beam is extracted radially in all these sources from a slit shaped aperture. The

extraction slit of the original alpha calutron was 380 mm long. Typically a 40 mm length is used in the lower current Bernas and Freeman ion sources.

used ion source for implantation. This source is a descendant of the calutron but unlike the calutron employs a reflex discharge, often also called a PIG (Philips Ionization Gauge) discharge. The Freeman source, previously the most popular ion source used for ion implantation, has an axial filament aligned with an externally applied axial magnetic field. The combination of the externally applied axial magnetic field and the radial electric field near the filament produces an EB drift around the filament that in turn amplifies the ionization to form a dense and stable plasma. Schematics of these different sources are compared in Figure 8.1. The calutron source is not employed in commercial ion implantation although a few early experiments with implantation were conducted with them [3].

8.2

The Freeman Ion Source

To obtain clean separation of the isotopes it was important to operate isotope separator sources in a low hash mode. Even a short breakdown into a hashy mode would be enough to contaminate the separation, and this occurred frequently enough to be a serious problem. As a historical note this problem was one of the factors leading to the abandonment of electromagnetic separation in favor of diffusion for large scale uranium isotope separation. Unstable “hashy” beams are a problem today in ion implanters. The need for a source less susceptible to instabilities and yet giving high quality beams led Freeman [4, 5] to develop a new type of electron bombardment ion source, sometimes referred to as a magnetron source, that did not depend on the basic oscillatory characteristics of the PIG discharge or discharges in strong magnetic fields. In his source Freeman placed a filament running axially along the whole length of the arc chamber and at a negative potential with respect to the

8.2 The Freeman Ion Source

chamber. The source was immersed in a weak magnetic field of from 50 to 200 G parallel to the axis of the filament. The beam was extracted from a slit-shaped aperture in the wall of the arc chamber. Freeman realized that thermal electrons leaving the filament and accelerated by the anode voltage and entering the plasma would move in a complicated pattern under the influence of the filament and axial magnetic fields greatly increasing the ionization efficiency of the electrons and without introducing the intrinsic oscillatory action of the PIG ion source. This new and original source design immediately produced hash free beams over a large range of operating conditions, as is illustrated by the data of Table 8.1. The source gave isotopic enrichments and resolving powers over one hundred times better than normally obtained with the calutron type source previously used. Table 8.1

Isotope Collected

84Sr 53Cr 148Sm 153Eu

Stable high resolution isotope separation results as reported by J.H. Freeman [4]. Source Material

Duration of run (h)

Current at collector (all isotopes) (mA)

Arc Current (A)

Arc Voltage (V)

Source Magnetic field (G)

Sr CrCl3 SmCl3 EuCl2

4 3 3.5 1.5

3.5 1.4 1.8 1.9

2.6 2.0 1.3 2.8

22 37 100 90

60 80 170 90

In addition to the greatly improved mass separation, it was found that the source could operate over the wide range of arc conditions needed to produce an optimum beam of many ion species. The low arc voltage required was a further advantage because it greatly reduced the erosion of the filament by sputtering. The source was integrated into an ion implanter by Freeman in 1968 at Harwell and licensed for manufacture by Lintott Engineering and was incorporated into one of the first commercially engineered ion implanters. Since then other manufacturers have adapted this source, because of its excellent performance, for use first in medium current and later in high current ion implanters [6, 7]. Figure 8.2 shows a cross-section of a typical Freeman source as used in a high current ion implanter. Photographs and drawings of some of these early sources are shown in the previous edition of this book and in the Proceedings of the International Ion Implantation Conferences. The materials of which the source is constructed depend upon the ion species and the application, which at this time is mostly ion implantation. The most commonly required ion species for this purpose have been beams of B+, P+, As+, and now beams of other species such as Ge+, In+, Sb+ and H+ are also becoming important for advanced semiconductor device requirements. To meet the doping requirements the ion source must be capable of delivering these species over a wide range of current from microamperes to tens of milliamperes with the need for high current being the dominating and competitive requirement. Typical operating parameters for the most commonly needed ions are given in Table 8.2.

135

136


Figure 8.2 Drawing of a Freeman ion source, developed for a high current ion implanter, showing some of the details of construction. Table 8.2

Typical performance of the Freeman source for the most common implant species.

Ion Species

11B 31P 75As

Source Material

Lifetime (h)

Current on Wafer (mA)

Arc Current (A)

Arc Voltage (V)

Magnetic Field (G)

BF3 PH3 As

35 50 50

3 10 10

4 2 2

100 60 60

140 100 100

Arc chambers have been made from aluminum, stainless steel, and molybdenum but high density graphite with a particle size less than one micron is probably the most versatile arc chamber material. Graphite is capable of operating well above 1000 C, is resistant to most hot vapors and has the lowest sputtering rate of all the elements. It is also relatively inexpensive and easy to machine. Unfortunately it is not a perfect material and has some properties that must be taken into account. For example BF3, the source gas frequently used to generate an ion beam of boron, reacts with graphite producing an ion beam of C12 that is close to the mass of B11 and so can contaminate the B11 beam if the separation of the peaks at the mass resolving slit is not adequate. Another more subtle problem, present to some extent in all arc chamber materials, is residual contamination after an ion species has been switched. For example, changing the source feed from BF3 to PH3 leaves traces of absorbed fluorine in the chamber walls. The fluorine persists for a long time and reacts with the carbon producing a low current beam of CF+ that has the same mass as phosphorus and so can contaminate the P+ beam. To avoid this problem arc chambers have been made out of molybdenum or fitted with liners of a suitable

8.2 The Freeman Ion Source

refractory metal to reduce the surface area of exposed carbon. Recently the molybdenum liners and arc chambers have been replaced with tungsten due to the Mo++ cross contamination of the BF2+ beam used to produce shallow junction transistors. The first source described by Freeman [4] employed a 4.5 mm diameter tantalum filament machined to 4.5 1 mm2 in the region facing the extraction aperture running the length of the arc chamber. Shaping the filament was expensive and provided no proven benefit in performance. A filament of 2 mm diameter tungsten was found to be the most practical and robust. The filament in this source was located 3 mm away from the extraction aperture but Freeman reported in the same paper that this distance was not found to be critical and could be increased to as much as 10 mm without significant change in performance. This result would appear to indicate that the radial plasma density is essentially constant between 3 and 10 mm, however closer to the filament in the primary ionizing region the density must increase to account for the rate of ion bombardment and sputtering of the filament. Further, the sputtering is not constant along the filament and is most severe at the positive end, an indication of a significant axial gradient of plasma density along the filament. At larger radii, near the anode wall, the uniform density of the extracted beam along the extraction slit indicates that this axial gradient is greatly reduced. Thermal and secondary electrons are accelerated across the cathode sheath into the plasma where they move to the positive end of the filament in complicated cycloidal patterns governed by the Lorentz force by the equations shown below and indicated schematically in Figure 8.3. Fr = eEr – evhBz + evzBh Fh = evrBz Fz = eEz – evrBh

Figure 8.3 (a) Electron trajectories around the filament of the Freeman ion source. Va = 100 V, If = 150 A, and the axial magnetic field is 100 G.

(8.1)

137

138


The cycloidal motion keeps the electrons close to the filament and they migrate to one end of the arc chamber causing some axial asymmetry in the plasma density. Elastic, inelastic and ionizing collisions change the trajectories and increase the volume being ionized. These primary electrons rapidly lose energy by the various processes mentioned, particularly by electron–electron collisions, and become thermalized. The thermalized electrons are collected at the anode, i.e. the arc chamber walls. Bohm [8] has shown that the electrons are driven to the anode predominately by plasma instabilities; diffusion and drift mobility in the plasma fields are not sufficient to account for the anode current. The loss of electrons from the plasma volume causes the plasma potential to increase by a few volts above the anode potential, thus containing the lower energy thermal electrons and stabilizing the plasma with an anode sheath. The higher energy electrons that reach the anode edge of the plasma are of course not stopped by this small voltage and continue to the anode and cause the positive bias of the plasma. It is this sheath that accelerates positive ions to the chamber wall and into the extraction aperture.

8.3

The Bernas Ion Source

The source developed by Bernas in 1958 differs from the construction of the calutron by having the filament inside the arc chamber and an additional electrode to act as a reflector or “anticathode” [9–11]. The source assembly is placed between the poles of a small electromagnet to provide the field that confines the primary electrons to a path between the filament and the anticathode. The purpose of the anticathode is to reflect the primary and thermalized electrons, allowing them to oscillate

Figure 8.4

Drawing of a Bernas source with a directly heated filament.

8.3 The Bernas Ion Source

between the cathode and anticathode, improving the efficiency of ionization in the source. Ions from the original source were extracted radially from a 30 mm by 2 mm slit in the wall of the arc chamber. Many versions of this source have been developed since then with both radial [12, 13 ]and axial extraction through an aperture in the anticathode[14, 15]. A schematic of a typical Bernas configuration is shown in Figure 8.4. The first application of this source to ion implantation was by Gamma Industries for a period in the early 1970s, but it was not fully accepted for commercial implantation until it was modified and used again by Nova Associates in 1982 to replace the Freeman source. Until that time the Freeman source was predominantly used for implantation because of its excellent properties. It gave adequate beams of P+ and As+ but was eventually unable to deliver more than 2 to 4 mA beams of B+, which was insufficient to be competitive. When it was found that a Bernas type source dissociated BF3 more effectively and gave higher currents of B+ with similar lifetime, it rapidly displaced the Freeman source. This source, shown in cross section in Figure 8.4, used a directly heated tungsten filament shaped as a rectangular helix rather than a hairpin filament, improving both the performance and lifetime. The source proved generally advantageous because of the improved electron ionization efficiency and often allowed the source to be operated under conditions that gave extended lifetime and still provided the required beam current. The lifetime of the source, like almost all sources producing heavy ions, is limited by sputtering and chemical erosion, particularly of the filament. To improve this situation the source was modified, the filament being replaced by an indirectly heated cathode of large emitting area and thickness as shown in the schematic of Figure 8.5(a). The emitting surface of an indirectly heated cathode can be made thick enough to last a long time, and the heating filament operates in a good vacuum and, consequently, is not damaged by sputtering. The performance of this type of cathode is also more reproducible compared with a filament which is hard to shape consistently, has a strong magnetic field which affects the plasma, and has erratic erosion and emission characteristic. This change more than any other has made it possible to replace one source with another and achieve almost identical performance. The concept of an indirectly heated cathode was not new. For example, an early introduction of an indirectly heated cathode was made for similar reasons by Morisov et al. [16] before 1960 in a PIG source used for multiply charged ion production with greatly improved performance compared to a source using a filament. The best multi-charged ion yields are obtained with this type of PIG source with high arc voltage of from 500 to several thousand volts and an intense discharge often 10 A or more. At the highest powers the source must be run in a pulsed mode because it becomes impossible to provide adequate cooling. Unfortunately the indirectly heated cathode sputters very rapidly under these conditions and the sputtered tungsten vapor becomes a significant proportion of the gas in the source. It is therefore easy to understand that once the electron confinement sources, which are described in Chapter 12, showed promise for high charge state production, development of PIG sources for this purpose was discontinued. However some of the experimental results and analysis can be applied to the lower voltage PIG sources now used so

139

140

8 Freeman and Bernas Ion Sources ARC PWR SUPPLY

+

HEATER BIAS SUPPLY

-

+

-

DECEL ELECTRODE

HEATER SUPPLY

+

ACCEL ELECTRODE

-

CATHODE

VAPORIZER

WATER JACKET

VAPOR FEED TUBE

GAS FEED

H

HEATER WIRES THERMOCOUPLE

PLASMA COLUMN

ANTICATHODE

(a)

(b)

(c)

(a) Schematic of an indirectly heated cathode Bernas source. (b) Photograph of a Bernas source assembly, and (c) a view of the arc chamber with the front plate removed to show the indirectly heated cathode and the anticathode. Figure 8.5

widely. For example the distribution of tungsten atoms in the arc reported by Kul’kina and Pasyuk [17] is important, even at much lower operating powers as will be discussed later in the chapter. Bernas sources with indirectly heated cathodes with the comparatively minor modifications made by the different implant manufacturers are now provided on more than 90% of the implanters built today. A typical version of the source can give the target currents shown in Table 8.3. These sources all seem to operate with simi-

8.4 Further Discussion of the Source Plasma

lar maximum stable plasma densities and give about the same beam current per square centimeter of extraction area. Performance of a typical Bernas source with a 40 by 3.5 mm2 (1.40 cm2) extraction aperture operating at an extraction voltage above 30 kV. The analyzed target currents depend on the efficiency of the optical system including space charge control but above 30 kV most of the extracted beam selected by the mass analyzer reaches the target chamber. The currents shown in the table should be regarded as average for an extraction slit 40 mm by 3.5 mm in size.

Table 8.3

Ion Species

11B 31P 75As

Source Material

Lifetime (h)

Current on Wafer (mA)

Arc Current (A)

Arc Voltage (V)

Magnetic Field (G)

BF3 PH3 As

35 50 50

15 – 20 20 – 30 20 – 30

5 2 2

100 60 60

140 80 80

Photographs of a recent commercially designed source are shown in Figure 8.5(b) and 8.5(c). One view of this ion source shows the source assembly, the other a top view of the arc chamber with details of the cathode assembly. In this example it can be seen that there are no insulators in the arc chamber to be shorted by sputtered material and the arc chamber is lined with inserts to reduce carbon contamination. When the Bernas source was run for isotope separation, Chavet reported that to achieve the best resolving power and clean separation the anticathode was often connected to the anode, disabling the PIG mode to reduce hash. Unfortunately this technique also reduces the extracted beam by a factor of two or more, making it unsuitable for ion implantation where the extra current that can be obtained with the anticathode at cathode potential is usually essential. Fortunately the acceptable mass separation of an ion implanter is about an order of magnitude less than that required for good isotope separation, however care must always be taken to minimize the hash that is always present in this mode of operation. Generally, for most ion species, low and high beam currents can be achieved with an acceptable range of arc voltage, gas flow and magnetic field. It is therefore possible to design control algorithms which will optimize the performance of the source for a wide variety of ions and even for doubly- or triply-charged ions. A small range of variables is easier to control and generally results in better reliability and longer lifetimes.

8.4

Further Discussion of the Source Plasma

There are a number of reviews of PIG ion sources, mostly concentrating on sources designed to produce highly charged ion beams (Bennett [18], Gavin [19]), and there are many articles describing both axial and radial extraction variations of PIG sources for a multitude of other applications with extremes such as the duopigatron capable of producing ampere beams of protons and small sources suitable for installation in the terminals of high voltage accelerators or for microbeam work. In many

141

142


of these papers the characteristics of these sources are presented together with an explanation of the characteristics of the arc. The different designs and operating conditions make projection of the results from one source to another confusing, if not conflicting, and in all this material very little attention has been paid to the Freeman or Bernas source as developed for implantation. With this in mind it is reasonable to examine here these implant sources running under typical implant conditions and to see to what extent the arc can be understood. For this purpose data taken from an indirectly heated cathode Bernas source run under very routine conditions and also similar data for a Freeman source are listed in Table 8.4, and will be used to examine the operation of the sources and as a guideline for comparison with information from the literature. For clarity this data will be referred to as the reference data. Dimensions and operating parameters of Bernas and Freeman sources from which the comparison data used in the discussion were obtained. Unless stated otherwise the sources were run on argon.

Table 8.4

Source Parameter

Bernas

Freeman

Distance between cathode and anticathode (mm) Cathode and anticathode diameter (mm) Size of extraction aperture (mm2) Gas flow (sccm) Axial magnetic field (G) Arc current (A) Arc voltage (V) Extraction current (mA) Estimated operating temperature (C)

55 17 303.5 2 120 3.0 100 30 800

2 402 2 120 3.0 100 25 800

Basile and Lagrange [20] and Green and Goble[21] have studied the properties of high charge state PIG sources by analyzing the current and energy flow. In these studies the cathodes were heated to emission temperature by ion bombardment. Schulte et al. have extended this analysis by making calorimetric measurements on an indirectly heated cathode source of the type developed at Dubna [22]. One of the most important parameters calculated is the average electron energy w spent per ionization, a number which includes all inelastic processes and not only the ionizing event. w can be interpreted as the efficiency with which a source uses the primary ionizing electrons. The value of w has been measured a number of times in high charge state sources [21] and varies over a considerable range between 30 and 130 eV, depending on the source and the operating conditions. A similar analysis is used here to examine the Bernas source with indirectly heated cathodes operated in the regime useful for implantation, and it is also applied to characterize the current flows in the Freeman source. There are many processes going on in the plasma resulting in complicated particle flows in and out of the plasma, defined as currents in Figure 8.6. Multiply charged ions amount to a few percent of the total positive ion current under the conditions studied and will be ignored in this analysis. However, beams of higher charge states are sometimes required for ion


implantation and a few milliamperes of doubly and triply charged ions can be obtained if the Bernas source is run at higher arc voltage and arc current. See, e.g. Horsky [23].

Figure 8.6 Current flows in (a) the Freeman ion source and (b) the Bernas ion source. The arrows show the direction of motion of the particles.

In Figure 8.6, the following current components are indicated: If = Thermal electron current from the filament (Freeman). + If = Positive ion current striking the filament. se If = Secondary electron current from bombardment of the filament. te Ic = Thermal electron current from the hot cathode (Bernas). se Ic and seIac = Secondary electron currents from ion bombardment of the cathode and anticathode. + Ic = Positive ion current going to the hot cathode (Bernas). + Iac = Positive ion current to anticathode. + Ia = Positive ion current to anode. e Ia = Electron current to the anode. e Iac = Electron current to anticathode. Iex = Radial extraction current from the source extraction aperture. – I = Negative ion current. te

143

144


8.4.1

Plasma and Sheath Potentials

The radial and axial potential distributions in the plasma in a PIG source have often been measured or estimated. The results obtained tend to be specific to the source geometry and the operating conditions. Generally the data support the conclusion that the axial variation of potential between the cathode sheaths is small because of the high mobility of electrons along the magnetic field, and in the radial direction there is a negative potential well of a depth sensitive to the source operating conditions. For example, Kistemaker and Sneider [24] found the depth of the potential well at the axis of a cold cathode PIG discharge run on argon to be 6 to 10 V lower than the potential near the anode, and with electronegative gases the well was much deeper, more than 50 V below anode potential. Similar results were obtained by Gabovich et al. [25] in experiments with a hot cathode discharge run on hydrogen. These authors found that the depth of the potential well could be expressed empirically by the expression DV @ 10–2Hd2,

(8.2)

where DV is the voltage drop (volts), H is the axial field (gauss), and d is the anode diameter (cm). The Bernas source is usually run at higher arc currents and at lower arc voltages and magnetic fields than in the results given above. There are no published data of direct measurements for the plasma potentials for a Bernas source run under these conditions. As a partial remedy for this situation unpublished measurements made on a source running under the conditions of Table 8.4 will be used to help understand the performance of the source. Langmuir probe measurements made by inserting a probe into the extraction aperture gave the results shown in Figure 8.7. The data show that the plasma is 5 V positive with respect to the arc chamber, but of course give no information about the existence or depth of the radial potential well. Eq. (8.2) gives a value of about 4 V for these conditions, but the presence of contaminants, including negative ions, probably makes this estimate somewhat low. In the axial direction the uniform density of the beam along the extraction aperture is a good indication that the plasma density is uniform along most of the distance between the cathode and anticathode. The cathode sheath thickness kD can be estimated to be approximately 0.15 mm from Eq. (2.21), taking the plasma density ne to be 1012 electrons cm–3 and the sheath potential to be 100 V. The voltage drop across the anode sheath as derived from Figure 8.7 is about 5 V or about 3–4 Te, which biases the plasma positively with respect to the anode or arc chamber wall across a 0.03 mm thick anode sheath. This potential is sufficient to accelerate positive ions to the anode and extraction region. If these measurements and assumptions are correct, the axial potential distribution is as shown in Figure 8.8(a). The negative potential well of Figure 8.8(b) is only an illustrative estimate. The negative well in the plasma favors a concentration of positive ions along the axis and this is borne out by the deeper sputter erosion pattern at the center of the cathode and anticathode. The rate of sputtering over the


Figure 8.7 A Langmuir probe measurement of the plasma in the reference source. The cathode is at zero potential. The probe was inserted through the extraction slit and its potential was varied from 50 to 120 V. The ion current

Figure 8.8

reaches an asymptotic value of 44 mA as the probe is driven negative. The thermal electron energy determined from these data is approximately 1 eV.

Illustration of the (a) axial and (b) radial potential distributions in the Bernas ion source.

cathode is the product of the ion current density and the sputtering coefficient. The sputtering coefficient drops rapidly below 100 eV and the ion density at the center of the well is unlikely to increase enough to produce the erosion pattern observed. Therefore it is reasonable to conclude that the well is unlikely to be deep in this low

145

146


voltage discharge given the deep erosion in the center of the cathode, as illustrated by the photograph of Figure 8.10(a) in Section 8.4.2, and is probably less than 10 V. Some further information can be obtained from the probe measurements of Figure 8.7. The probe measurements were made with the source run on argon, and the ratio of electron to ion current in an unmagnetised plasma is theoretically expected to have a value close to the ratio (M/m)1/2 or 271, where M is the ion mass and m the negative particle or electron mass, yet the measured value derived from the probe measurements is approximately 44. The discrepancy can partly be explained by the presence of the axial magnetic field but also by the presence of negative ions. No negative ions should be present in pure argon, but even the presence of 0.01% of negative tungsten ions or other contaminating negative ions per plasma electron would be sufficient to increase the value of m sufficiently to explain the difference from the theoretical prediction. 8.4.2

Effect of Sputtering on the Plasma

Sputtering of the cathodes of the ion source, as well as being the usual reason for failure, has several other negative consequences. The sputtered atoms can be ionized on their way to condensing on the walls of the arc chamber, and in very high power arcs (13 A and 480 V) can even become a majority of the ions extracted from the source, as described by Makov [26]. The deposited atoms capture some of the source gas atoms, acting like a pump and changing the pressure distribution along the arc. The built-up layer of tungsten eventually becomes thick enough to short the anode to the cathode, or more commonly a layer will flake off the arc chamber wall shorting the source and resulting in an unpredictable failure. If neither of these events occur the source will run until the cathode is too severely eroded to sustain an arc, which in the case of the high arc power conditions just described gives only a few hours of operation. Fortunately when the source is used in an ion implanter the arc power is much lower, a few hundred watts, and lifetimes of 100 or 200 h on average are achievable in a production environment which includes both low and high current operation. The sputtering rate of the cathode and anticathode for a Bernas source running under the conditions referenced in Table 8.4 averaged 100 and 200 mg s–1 or 3 to 6 1017 atoms s–1, a rate close to the argon gas flow into the source which at a rate of 2 cm3 is equivalent to 9 1017 atoms s–1. The sputtering rate is found to depend strongly on the condition of the cathodes and increases by a factor of two or more when the cathodes are deeply eroded. When the source is run under the conditions of Table 8.4 it is also observed that the cathode sputters two to three times faster than the anticathode, indicating an unequal division of positive ion current between the cathodes. This result differs from the observations of Schulte et al., who report that the ion current is distributed more evenly between the cathode and anticathode, probably because of the much higher magnetic fields [22]. The energy distribution of the sputtered atoms, most of which are neutral, peaks at 2 eV and has a long high energy tail, and the atoms consequently pass quickly through the plasma to be condensed on the nearest wall. The thickness and distribu-


Figure 8.9 Measurement of the sputtered tungsten deposited on a surface anode 10 mm from the axis of the arc. The thickness of the sputtered layer between the cathode (on the left) and the anticathode shows the asymmetry of the sputtering. The source was run on PH3.

tion of this condensed layer of tungsten is illustrated by the data of Figure 8.9. Since the tungsten atoms make only one pass through the ionizing electrons before condensation, the probability for ionization is greatly reduced compared to the argon atoms, which make multiple passes before escaping through the extraction aperture. This is probably the explanation for the small tungsten currents, amounting to less than 1% of the extraction current, in spite of the large number of atoms passing through the plasma. Self-sputtering, which can be significant in higher power arcs [20], is therefore assumed to be small and is not considered here. According to the data of Stuart and Wehner [27] and Hagstrum [28] the sputtering yield of argon on clean tungsten is 0.08 at 100 eV and drops sharply below that voltage. Experimentally it is observed that the sputtering increases as the cathodes erode. This effect has sometimes been attributed to an increased angle of impact as the cathodes wear, but if this were the case the cathode sheath could not remain conformal to the surface of the cathode. The increase in sputtering rate is most probably due to the increase in cathode area as the cathode erodes. It is possible that the increase in sputtering rate, which occurs mainly on the hotter cathode, can be augmented by the growth of crystalline formations on the cathode, which can have very high sputtering rates [29]. Generally for implantation the source is not run on simple gases like argon but on chemically reactive species like PH3, As vapors or hydrides, and many other gases, and there is no information on the sputter yields. However the same general trends are observed with most gases and the sputter rates are not found to be dramatically different.

147

148


Figure 8.10 (a) Photograph of an eroded tungsten cathode from a source run on PH3. (b) Shows the growth of tungsten on cathode and anticathode with the source running on BF3.

The photograph of Figure 8.10(a) shows the erosion of a cathode after about 10 h running. For electronegative gases the erosion can be significantly offset by the halogen cycle and the cathode and anticathode actually grow from tungsten deposited from the tungsten source liners, as illustrated by the photographs of the cathode and anticathode in Figure 8.10(b). Sputtering of the filament of the Freeman source is also the usual cause of failure. Because the ionizing electrons move towards the positive end of the filament, increasing the plasma density, sputtering at that end of the filament is more severe. Failure occurs when the filament diameter becomes too small to sustain the filament current and the filament over-heats in that region and burns out. Photographs of filaments which have failed in this way are shown in a review of the Freeman source by Aitken [30]. The sputtering rate for the Freeman source running under the conditions given in Table 8.4 can be inferred from the lifetime, which averages about 30 h. The condition of the filament after failure varies significantly, but a quarter to a third of the original mass is a reasonable estimate of the amount sputtered. The rate of mass loss measured in this way is approximately 100 mg s–1. 8.4.3

Ion Heating of the Cathode and Anticathode in the Bernas Source

While sputtering gives an estimate of the ion currents striking the cathodes, it is a very uncertain measurement, given that the sputtering coefficient is rarely known with any accuracy. Fortunately, with an indirectly heated cathode it is relatively easy to measure the ion current. Part of the cathode heating is provided by energetic positive ions from the plasma accelerated across the cathode sheath so that a simple


experiment can determine the fraction of the heating power contributed by the positive ions accelerated from the plasma. Bennett [18] among others has found that the beam current from a hot cathode source is insensitive to arc voltage and to source gas pressure for a fixed arc current. This observation was tested in the reference source by varying the arc voltage VA from 70 to 150 V with the arc current stabilized at 3.0 A. The gas pressure was not changed. Under these conditions it was found that the extracted current for a fixed extraction voltage increased by only 4.5%, a result indicating that the positive ion density and the positive ion current striking the cathode +Ic can be regarded as approximately constant over this range of arc voltage. However as the arc voltage was raised it was necessary to reduce the cathode heater power to hold the arc current constant. This reduction in heater power can be explained by the reduction in the number of primary ionizing electrons required to keep the arc current constant as the arc voltage is raised. The graph in Figure 8.11 shows that the ion heating, that is the product of +Ic and VA, becomes a larger fraction of the total cathode heating power as the arc voltage is raised and can be expressed by the linear relationship of Eq. (8.3), W = Pc – +IcVA,

(8.3)

where Pc is the heater power required in the absence of ion heating and W is the heat generated by the electrons from the external filament. The linear relationship shows that arc current is the dominant variable controlling plasma density. The positive ion current +Ic can be calculated from the data of Figure 8.11. Two corrections are necessary. Although only a small change in the cathode temperature is needed to reduce the thermal emission current, a small correction is necessary to account for the reduced radiation and other thermal losses of the cathode. This cor-

Figure 8.11 Variation of cathode heater power with arc voltage, with the arc current held constant at 3.0 A. W is the power of the electron beam heating the cathode. The extrapolated value Pc is the power required when there is no ion heating.

149

150


rection reduces the ion heating determined from Eq. (8.3) by 2%. A second correction due to the Shottky effect should be considered because the electric field between the cathode and plasma boundary increases as the arc voltage is raised. The increase in electric field modifies the exponent in the Richardson–Duschmann equation, resulting in a higher electron emission as given by Eq. (8.4), J = AoT2exp{–e[u – (eE/4peo)1/2]/kT},

(8.4)

where the constant Ao = 1.2 106 A m–2 K–2, E is the electric field in the sheath in V m–1, u is the work function in V and T is the cathode temperature in K. Although the ion density in the plasma remains nearly constant when the arc voltage is varied, the cathode sheath thickness increases as VA1/2, see Eq. (2.21), so that E varies not as VA but as VA1/2. The Schottky correction for the reference conditions selected lowers the estimated ion heating by 13%. Ion heating of the anticathode can be estimated by comparing the sputtering rate of the cathode and anticathode, and is usually less than one half of the cathode heating, amounting to less than 150 W in this example. The anticathode heating therefore lies between 50 and 200 W for any reasonable operating condition when the source is used for implantation. These power levels are clearly not sufficient to heat the anticathode to emission temperatures. It is worth noting that it was common for PIG sources designed for high charge production to take advantage of the high arc powers and hence ion currents to heat the cathode to emission temperature [18]. The major limitation of this technique is that the arc current can only be controlled by varying the arc voltage. 8.4.4

Current Balance in the Freeman Source

Taking the values of the operating parameters given in Table 8.4 it is possible to make a reasonable determination of the electron and positive ion currents flowing in the Freeman source when operated under these conditions. Figure 8.6(a) defines the current flows that will be considered, the arrows pointing in the direction of particle movement. In the example chosen the source is run on argon and the negative ion current –I is taken to be zero, an assumption likely to be invalidated to some extent by the presence of sputtered tungsten and other contaminating atoms capable of forming negative ions. The currents flowing to the cathode and to the anode must equal the arc current IA, giving the relationships IA = teIf + seIf + +If = eIa – +Ia

(8.5)

The value of the positive ion current striking the filament is not easy to measure in the Freeman source. If the average plasma density around the filament is known, +If can be calculated from the expression +

If = 0.4ne(2kTe/M)1/2Af

(8.6)


where Af is the area (cm2) of the filament including the small incremental thickness of the cathode sheath, ne is the electron density (cm–3) and M is the ion mass (g). Unfortunately the plasma density near the filament has not been measured in a Freeman source and if +If is calculated using the plasma density at the extraction aperture the value obtained is clearly several times too low. With much less accuracy, sputtering measurements can be used to estimate this current and incidentally give a measure of the average plasma density near the filament. Again using the reference conditions of Table 8.4 and a sputtering coefficient of 0.08 and a measured sputtered rate of 100 mg(tungsten) s–1, the positive ion current +If striking the filament is estimated to be 0.66 A. To calculate the positive ion current flowing to the anode, a cylinder as long as the filament and with a radius equal to the distance between the axis of the filament and the extraction aperture will be defined as a virtual anode. Provided the plasma column has a uniform density over the surface of this cylinder the current +Ia of positive ions crossing the surface of this cylinder can be calculated by Eq. (8.7), +

Ia = IexAa/Aex

(8.7)

where Aa = 32 cm2 is the area of the virtual anode and Aex = 0.8 cm2 is the area of the extraction aperture. The extracted beam current taken from the reference data of Table 8.4 and substituted in the equation above gives a positive ion current of 1.0 A. The total positive ion current, which is the same as the rate of production of positive ions +I, is +If + +Ia or 1.66 A. To calculate the thermal emission current Iet the secondary electrons produced by the ions striking the filament must be taken into account. For argon ions on clean tungsten, the sputtering yield s = 0.15 for 100 eV argon ions, (see Krebs [31]). Given seIf = s+Ia the thermal electron current teIf can be calculated from Eq. (8.5) and is 2.24 A. The average energy loss per ionizing collision w is given by w = VA(teIf + seIf )/+I

(8.8)

The value of w for the Freeman source is therefore 141 eV, indicating that it takes more than one electron transit to make an ionizing collision. 8.4.5

Current Balance in the Bernas Source

Current balance in the Bernas is more complex than in the Freeman source, principally because of the PIG or reflex discharge between the electron emitting cathode and the cooler non-emitting anticathode. The currents flowing in the source are defined in Figure 8.6(b). The arc current IA measured at the anode is the sum of three particle flows and is given by IA = eIa – +Ia – Iex ,

(8.9)

151

152


and the sum of the cathode currents, which is of course also equal to IA, ignoring the negative ion current, can be written as IA = teIc + +Ic + +Iac + seIc + seIac .

(8.10)

Other simple relationships exist, for example if the cathode and anticathode are separated electrically and Ihc is the current collected by the hot cathode and Iac is the total anticathode current, then Ihc = IA – Iac ,

(8.11)

and the anticathode current is given by Iac = +Iac + eIac + seIac .

(8.12)

The positive ion current +Ic can be obtained from Eq. (8.3) by measuring the energy deposited by the ions striking the cathode as described in the previous section. The calculated value of +Ic using the data in Figure 8.11 corrected for temperature change and the Schottky effect is 0.99 A. This can be compared with 0.9 A determined from the sputtering rate of the cathode for the same conditions. Experimentally it is found that the anticathode current Iac is generally a small fraction of the anode current as is illustrated by the graph in Figure 8.12 showing the anticathode current as a function of anticathode voltage. The floating potential of the anticathode is in fact within a few volts of the cathode potential. For this to be the case the anticathode must collect a substantial fraction of the primary electrons emitted by the hot cathode. Rewriting Eq. (8.7) for Iac = 0, e

Iac = +Iac + seIac .

(8.13)

This electron current can be suppressed by biasing the anticathode negatively to more than –30 V indicating that some of the electrons in the plasma must have an energy of 130 eV. Similar experimental results have been reported by Schulte et al. [31]. An analysis in this paper of electron plasma interactions shows that they are capable of modifying the primary electron energy distribution and explaining the existence of higher energy electrons. Although the reflex action of the electrons in the discharge increases the ionization efficiency of the source the reflex action also contributes to the instabilities in the plasma and hash in the extracted ion beam. In addition, the accelerated electrons that are captured by the anticathode when it is at cathode potential are immediately lost from the discharge, reducing the efficiency of the source. The current measured at –30 V negative bias is the sum of the positive ion current +Iac arriving at the anticathode and the secondary electron current seIac liberated by the striking ions. The measured current less the secondary emission component is 0.43 A and is in approximate agreement with the value that can be determined from the sputtering data, confirming that the anticathode current is approximately one half of the cathode current. The anode current +Ia can be calcu-


lated by the same technique used for the Freeman source. The extraction current for the Bernas from Table 8.3 is 30 mA, Aa = 34.5 cm2 and Aex = 1.05 cm2. Inserting these values into Eq. (8.7), +Ia = 1.00 A. The total number of ions created per second is therefore the sum of the positive ions going to different electrodes, +I having a value in this case of 2.42 A.

Figure 8.12 Anticathode current versus voltage. The source is run on argon under the standard conditions defined by Table 8.4.

The Bernas source is normally operated with the cathode and anticathode connected, and very little error is introduced into the analysis of the currents in the source if it is assumed that that the anticathode current can be neglected. With this assumption the arc current measured at the cathode is then the sum of only three particle flows, IA = teIc + +Ic + seIc ,

(8.14)

and since the secondary emission is proportional to the impacting ion current this can be rewritten as IA = teIc + (1 + s)+Ic ,

(8.15)

from which the thermal electron current teIc can be calculated to be 1.86 A. The average energy an electron loses to produce an ion is given by a slight modification of Eq. (8.8), i.e. w = VA{teIc + s(+Ic + +Iac)}/+I .

(8.16)

153

154


The value obtained for w, 86 eV, indicates that the electrons are used more effectively in the PIG discharge of the Bernas source than in the Freeman ion source.

8.5

Control Systems

Control of the ion source is generally accomplished through the use of closed loop feedback. The object of the control loop is to maintain a constant ion current, free of short and long term variations. To do this in the Freeman, or for the indirectly heated cathode (IHC) Bernas source, the arc voltage is generally fixed and the arc current is maintained at a constant value through control of the hot cathode temperature. The typical cathode operates at a temperature between about 2500 and 2800 K to produce the required emission current (see Eq. (8.4)). The IHC cathode is large, resulting in a large thermal mass and a long thermal time constant, making it more difficult to control. Cathode sputtering causes the time constant to decrease by up to a factor of 2 during the course of normal operation. A major factor in the design of a control system for the hot cathode ion source is therefore its time-response characteristic. 8.5.1

Freeman and Bernas Controls

Figure 8.13 is a control diagram of the Bernas and Freeman sources with directly heated filaments, showing the major elements of the control system along with the signal flow path. The feed-forward element in the diagram is shown as a PID (proportional, integral, and derivative) controller. The control system could easily be represented through several alternative techniques, see for example [32, 33, 34]. The reference signal Iarc(ref) is algebraically summed with the feedback signal Iarc(fbk) to produce an error signal er . The error signal is processed by the PID controller on a term by term basis, then summed to produce the overall transfer function of the feed-forward element as written in Eq. (8.17), Vf

ðref Þ

¼ Kp ðer Þ þ Ki

R

er dt þ Kd

d ðe Þ dt r

(8.17)

Each term in Eq. (8.17) has a special purpose: The proportional gain Kp is used to adjust the filament voltage instantaneously in response to the error signal. This term alone cannot eliminate the steady state error in the control system, however it does provide a degree of damping. The integral gain Ki is used to accumulate a steady state signal large enough to zero the error term. In this particular application, the integral term represents the stored thermal energy in the cathode retaining a memory of its thermal history. The derivative term Kd provides a degree of damping to compensate for any thermal lag as the cathode reaches an equilibrium temperature. This term is used sparingly due to the high bandwidth of the differentiation process. If the value of Kd is too large, the control tends to be noisy and may trigger

8.5 Control Systems

oscillations. The integral term will also cause the control system to oscillate if it is too large, however the proportional and derivative terms allow for maximum gain in the integral term. The constants Kp, Ki, and Kd may be found through detailed circuit analysis, experimental measurement or computer simulation.

Figure 8.13 Block diagram of the Freeman and Bernas source control. The labeled C, is generally about 2 mm diameter and 20 to 100 mm long. The anode, labeled A, is the arc chamber body including the arc slit. The reflector or anticathode structure is not shown in the schematic

8.5.2

Bernas Indirectly Heated Cathode

Figure 8.14 illustrates a control system used for the Bernas IHC source and is noticeably more complex. The best means of control of this source is through a PID controller using a fast DSP (digital signal processor). The controls for this source are, in general, conditionally stable. For example, the ion heating of the cathode under high arc power settings may reach a level such that the slope of emission current vs. heater power goes to zero, leading to gross instability. When this occurs, the only way to stabilize the loop is to reduce the arc voltage. This can be done automatically through the use of a constant-current constant-voltage crossover control in the arc supply, independent of the external feedback controls. The other potential source of instability is the two feed-forward control blocks in series, however the instability may be avoided through careful selection of control parameters. Furthermore, the time constant of the heater filament is short when compared to that of the main cathode. It should be noted that the control becomes more unstable as the main cathode erodes and its thermal mass is reduced. This must be taken into account in the selection of control constants. The DSP may also be used to produce a schedule of variable gains based on operating conditions and preset arc current. The feedback parameters may also be transformed to produce a linear relationship between the control signal and the feedback. Detailed analysis of this type of control system is complex and difficult to realize in practice. The control system outlined in Figure 8.14 has been analyzed using a computer simulation showing a satisfactory

155

156


Figure 8.14 Block diagram of the Bernas IHC source control. The heater, labeled H, is a small 0.5 mm diameter tungsten wire bent in a serpentine fashion to produce a large emitter area. The cathode, labeled C, is a 1.5 cm dia-

meter disk mounted inside a cylindrical holder and is approximately 3 mm thick. The anode, labeled A, is the arc chamber body including the arc slit. The reflector or anticathode structure is not shown in the schematic.

Figure 8.15 Calculated arc curren response of the Bernas IHC source from cold start to 4.0 A and back to 80 mA.

8.5 Control Systems

Arc Current

Cathode Power

Time (s) Figure 8.16 Plot of the calculated source parameters during the step illustrating the operating range of the cathode power and temperature.

Figure 8.17 Plot of the calculated feed-forward parameters during the step illustrating the control error terms of the PID controller. Note that when the proportional term goes to zero, the residual error is zero and the integral term always contains the final output control.

Arc Current (A)

Temp. (K), Power (W)

Cathode Temp

157

158


response over a wide dynamic range of arc current. A physical model of the source components was generated using the constants of the source, cathode mass, conductive and radiation loss constants, heat flow through the cathode, ion heating of the cathode, and thermionic emission density. In the model, the thermal energy stored in the cathode divided by its thermal mass yields the cathode temperature and consequently the output arc current. Figure 8.15 shows the step response from a cold start to an arc current of 4.0 A followed by a step down to 80 mA. Figure 8.16 provides a more detailed look at the static transfer function of the IHC source; note the long (5 s) delay from the cold start to the onset of emission followed by a more reasonable ~0.5 s response to step changes. Figure 8.17 illustrates the importance of each of the PID control elements and how they are well suited to this type of control application. During the transition period while the arc current is being reset, the proportional and derivative terms dampen oscillations while the integral term zeroes the residual error ultimately satisfying all of the requirements of the control system.

8.6

Lifetime and Maintenance Issues

Perhaps the most difficult technical problems to be solved in these ion sources are those dealing with the destructive nature of the plasma in a very hot arc chamber. In addition to sputtering and evaporation there are the chemical reactions of the hot corrosive gases or ions. In this regard, feed materials are important [8]. Some feed materials such as BF3 rapidly erode the source and extraction electrode components. Other materials such as PH3 and AsH3 deposit dissociated products on the electrodes and in the vacuum chamber, eventually degrading the vacuum system and the voltage holding capability of the system. There is also a group of ion species that are not used frequently but when included in the same ion source system, they create a new set of lifetime and maintenance issues. Included in this grouping are In, Sb, Si, Ge, and Al. The operating mix of ion species and maintenance procedures has a significant impact on the lifetime of the ion source. 8.6.1

Use of BF3

One of the critical factors that affects the lifetime of the source during operation with BF3 is the growth of tungsten crystals on the surface of the cathode and anticathode; this has been described in Section 8.4.2. All that is necessary for robust whisker growth is for the melting point of the scavenged element to be greater than that of the cathode operating temperature. In the case of the anticathode, where the operating temperature is significantly lower than that of the cathode, several different refractory metals, molybdenum, niobium, tantalum, and tungsten, are capable of producing whiskers. The best way to minimize this effect is to operate the source gas feed as lean as possible while still maintaining the beam noise within acceptable

8.6 Lifetime and Maintenance Issues

limits. When tungsten liners are used in the arc chamber this process is beneficial to lifetime because as the cathode is sputtered, it is also replenished with new material. 8.6.2

Use of PH3, AsH3, P4, and As4

Phosphine and arsine of 100% concentration from a sub-atmospheric gas delivery system are becoming the most popular source feeds for 31P+ and 75As+ ions, primarily because it is easier to meet industrial safety standards. When operating a source using vapor feeds, the vapor consists mainly of P4 and As4 molecules. This leads to far greater deposits on the chamber walls and extraction electrodes, thus leading to glitching and high voltage breakdown. The general rule for operating either the vapor feed or the gas feed is to keep the gas flow as low as practical. If the gas or vapor feed is too rich, the source noise is minimized but the tendency for arcing in the extraction region increases dramatically because of deposited material. 8.6.3

Use of Sb, Sb2O3, and SbF3

Elemental antimony is the most difficult material to use for the production of Sb+ ions because the vaporizer oven must be operated near 700 C, above its melting point of 630.5 C, to produce the required vapor pressure. This leads to serious condensation inside the vapor feed tube, if it is not hot enough, as well as on parts of the arc chamber and extraction electrode system. The other negative factor is that the latent heat of melting complicates the thermal cycle. Both Sb2O3 and SbF3 are solids that have significant vapor pressure at temperatures below the melting point of elemental antimony, but both of these feeds are difficult to operate stably and it is a matter of choice which one to use. Sb2O3 has a significant vapor pressure at around 450 C and SbF3 has a significant vapor pressure at around 250 C. SbF3 is probably the best material to use because it is the least prone to the deposits that lead to high voltage arcing, which are especially noticeable when switching the source from antimony to boron ions. 8.6.4

Use of SiF4 and GeF4

Silicon and germanium ions are implanted in silicon to create damage in the surface regions to prevent ion channeling. This is becoming a more important step in the process of forming shallow junctions because the implant energy is being reduced, leading to higher critical channeling angles. In some cases, the wafers must be implanted at normal incidence and this clearly requires that the surface silicon be amorphous prior to implant. The preferred source feed for Si+ and Ge+ ions is SiF4 and GeF4, both of which are available in high pressure bottles or low pressure SDS containers. When using these feed gases, the same precautions for

159

160


the use of BF3 should apply. One additional caution: the presence of N2+ with mass 28 in the source will contaminate the ion 28Si+, the most abundant isotope of silicon. Both SiF4 and GeF4 produce beam line contamination in the form of insulating films, leading to high voltage breakdown in the extraction region. 8.6.5

General Guidelines for the Use of Other Organic and Inorganic Compounds

There are many compounds that can be used in an ion source that contain the atoms needed for implantation, and inorganic compounds produce the best overall results. The organo-metallic compounds can produce the required beam current and operate well for short periods of time. However unacceptable amounts of carbon are deposited in the arc chamber and inside the vacuum system. With any material thought must be given to contamination of the beam by other ions, including care with selection of the construction materials of the source, vaporizer, and gas lines. 8.6.6

Electrode Cleaning and Maintenance

The sources described in this chapter are used in 24 hours-a-day production, and potential problems associated with the maintenance and cleaning of the source should be considered as part of the design. For example the parts of the source system can be separated into two categories. If the part operates at a negative potential with respect to its surroundings it behaves like a cathode and, depending on the magnitude of the electric field, it may be prone to field emission. Its surface must be clean, smooth and free of any insulating material. The use of glass beads in cleaning cathode surfaces should be avoided because the residual beads stuck in the surface become electron emitters. This is very noticeable when the material is copper or aluminum. Furthermore, the roughened surface enhances the local electric field stress, sometimes leading to field emission. The anode surfaces are much more forgiving and may be cleaned using the most effective means available including glass bead blasting. 8.6.7

Insulator Cleaning and Maintenance

The cleaning of insulators is highly dependent on the deposited material and the insulator as well as its use in the source. The insulating bushings used to isolate the ion source from its source chamber are generally made of high purity Al2O3 (alumina) or an epoxy filled with insulating particles. These insulators may be cleaned using a detergent such as Alconox in combination with an abrasive cleaning pad. The most effective means of cleaning an insulator is to use hydrogen peroxide 2H2O2 in a ventilated toxic cleaning station. Hydrogen peroxide is a strong oxidizer that ultimately decomposes into water once its work is done.

References

References [1] R.K. Wakerling and A. Guthrie, Editors,

[2] [3] [4] [5]

[6]

[7]

[8]

[9] [10] [11] [12] [13 [14] [15]

Sources and Collectors for Use in Calutrons, United States Atomic Energy Commission Report TID–5218, (USAEC, Washington, 1949). J.H. Freeman, Radiation Effects 100, 161 (1986). K.E. Manchester, C. B. Sibley, and G. Alton, Nucl. Instrum. Methods 38,169 (1965). J.H. Freeman, Nucl. Instr. Meth. 22, 306 (1963). J.H. Freeman, D.J. Chivers, G.A. Gard, and W. Temple, Nucl. Instrum. Methods 45, 473 (1977). P.H. Rose, in Proceedings of the Second Symposium on Ion Sources and the Formation of Ion Beams, Berkeley (Lawrence Berkeley Laboratory, Berkeley, CA, 1974), paper VII–l–1. M. Guerra, V. Benveniste, G. Ryding, D. H. Douglas-Hamilton, M. Reed, G. Gagne, A. Armstrong and M. Mack, Nucl. Instrum. Methods B 6, 63 (1985). D. Bohm, in The Characteristics of Electrical Discharges in Magnet Fields, edited by A. Guthrie and R.K. Wakerling (McGraw-Hill, New York,1949), Ch. 1 to 5. R. Bernas and A.O. Nier, Rev. Sci. Instrum. 19, 895 (1947). I. Chavet and R. Bernas, Nucl. Instrum. Methods 51, 77 (1967). G. Lempert and Chavet, Nulc. Instrum. Methods 139, 583 (1976). N. White, Nucl. Instrum. Methods, Phys. Res. B 37/38, 78 (1989). S.R. Walther, Rev. Sci. Instrum. 65, 1284 (1994). H. Bauman and K. Bethe, Nucl. Instrum. Methods 122, 517 (1974). B.H. Wolf, Rev. Sci. Instrum. 65, 1284 (1994).

[16] P.M. Morisov, B.N. Makov and M.S. Ioffe,

Atomaya Enerziya 3, 272 (1957); B.N. Makov, Kurchatov Inst. Atom. Energy, Pre-print IAE– 1051 (1966). [17 L.P. Kul’kina and A.S. Pasyuk, Sov. Phys. Tech. Phys. 11, 537 (1966). [18] J.R.J. Bennett, IEEE Trans. Nucl. Sci. 19, 48 (1972). [19] B.F. Gavin, in The Physics and Technology of Ion Sources, edited by I. Brown, (Wiley, NewYork, 1989), Ch. 8. [20] R. Basile and J.M. Lagrange, Nucl. Instrum. Methods 31, 195 (1964). [21] T.S. Green and C. Goble, Nucl. Instrum. Methods 116, 157 (1973). [22] H. Schulte, B.H. Wolf, and H. Winter, IEEE Trans. Nucl. Sci. 23, 1053 (1976). [23] N. Horsky, Rev. Sci. Instrum. 69,1688 (1998). [24] J. Kistemaker and J. Sneider, Physica 19, 950 (1953). [25] M.D. Gabovich, O.A. Bartnovskii, and Z.P. Fedorus, Zh. Tech. Fiz. 30, 345 (1960). [26] B.N. Makov, IEEE Trans. Nucl. Sci. 23, 1035 (1976). [27] R.V. Stuart and G.K. Wehner, J. Appl. Phys. 33, 2345 (1962). [28] D. Hagstrum, Phys. Rev. 96, 325 (1954). [29] L.I. Maissel and R. Glang, Editors, Handbook of Thin Film Technology (McGraw-Hill, New York, 1970), pp. 3–15. [30] D. Aitken, in The Physics and Technology of Ion Sources, edited by I. Brown (Wiley, New York, 1989), Ch. 9. [31] K.H. Krebs, Forts. Phys. 16, 419 (1968). [32] J.J. Distefano, Feedback and Control Systems (McGraw-Hill, New York, 1990). [33] M.E. Van Valkenburg, Network Analysis (Prentice-Hall, New Jersey,1974). [34] B.P. Lathi, Linear Systems and Signals (Oxford University Press, Oxford, 2002), Ch. 4.

161

163

9

Radio-Frequency Driven Ion Sources Ka-Ngo Leung

9.1

Introduction

The use of radio-frequency (RF) voltage to generate a plasma dates back to the 1940s [1, 2]. RF-driven ion sources offer the advantage that they can operate with any type of feed gas, in particular gases such as oxygen that can easily poison tungsten filament cathodes. For this reason, RF ion sources have found important applications in plasma and reactive ion beam etching and ion beam doping. RF ion sources are also useful when long-life operation or clean plasma production is required. Today they are widely employed by the particle accelerator community and by the semiconductor industry. An RF discharge is formed in a chamber filled with a gas at a pressure of about 10–3 to 10–2 Torr. RF power of a few hundred watts is typically needed to establish a stable discharge. The RF frequency can vary from a few megahertz to tens of megahertz. There are two ways in which a low-pressure gas discharge can be excited by RF voltage: (a) a discharge between two parallel plates or electrodes across which is applied an alternating potential (capacitively coupled discharge), and (b) a discharge generated by an induction coil (inductively coupled discharge). Most RF ion sources are operated with the second type of discharge. In this case, an azimuthal electric field is generated by the alternating magnetic field in the discharge region. Electrons present in the gas volume are excited into oscillation by the RF electric field. They quickly acquire enough kinetic energy to form a plasma by ionizing the background gas particles. The ions are then extracted from the source chamber in a manner similar to a dc discharge source.

9.2

Capacitively Coupled RF Sources

Although capacitively coupled RF sources are widely used in plasma etching processes, they are seldom employed for the production of ion beams. Capacitively coupled discharges form a very high positive plasma potential. As a result, material is sputtered from the chamber wall to form impurity ions in the extracted beam. The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

164

9 Radio-Frequency Driven Ion Sources

Figure 9.1 Capacitively coupled RF ion source in cross section: 1 – Mounting flange; 2 – ion source case; 3 – alumina source base; 4 – stainless-steel RF anode chamber walls; 5 – mounting slabs; 6 – ceramic grid holder;

7 – screen grid; 8 – acceleration grid; 9 – electrostatic end confinement made of alumina; 10 – water cooling for RF cathode; 11 – alumina cathode cover made of; 12 – RF cathode; 13 – grid polarization; 14 – gas inlet.

Figure 9.1 shows a capacitively coupled RF ion source developed by Plasma Consult GmbH, Germany [3]. The RF discharge is created between the centrally placed RF cathode and the surrounding cylindrical anode. This radial excitation is especially efficient in the presence of an axial magnetic field generated by the solenoid coils. The high-energy electrons move in the cathode plasma sheath on cycloidal trajectories around the cathode (magnetron effect) and can ionize neutrals or impinge on the cathode surface causing secondary electron emission. The axial magnetic field suppresses the loss of electrons to the anode. To avoid electron loss in the direction parallel to the magnetic field, electrostatic end confinement is used. The floating screen grid and the floating ceramic plate provide the essential confinement geometry. This ion source operates at an excitation frequency of 13.56 MHz and RF power less than 250 W. An integrated extraction system allows one to extract ion beams with current density higher than 0.8 mA cm–2 and beam energy between 100 eV and 2 keV.

9.3 Inductively Coupled RF Sources

9.3

Inductively Coupled RF Sources

In order to produce an inductively coupled plasma, an induction coil or RF antenna is needed. This antenna coil can be placed either outside or inside the ion source chamber. Source operation with these two antenna arrangements is described in the following sections. 9.3.1

Source Operation with an External RF Antenna

The earliest inductively coupled RF ion sources were operated with an external antenna. Figure 9.2 shows a schematic diagram of a Thonemann type RF ion source [4, 5]. It consists of a quartz discharge chamber surrounded on the outside by an RF-induction coil. There are four external variables that affect the characteristics of the discharge and the resulting ion beam: the gas pressure in the chamber, the RF field (its magnitude and coupling to the plasma), the external magnetic field, and the extraction voltage. These parameters are not all independent. For instance, both the pressure and the magnetic field will affect the RF fields by influencing the electrical properties of the plasma, which in turn affect the RF coupling. As a result of these complex relationships, it is not possible to examine the source characteristics as a function of only one of these variables. Typical operating parameters for this source are: gas pressure 10–3 Torr; RF power 350 W; oscillator frequency 13 MHz; extraction voltage 3 kV; magnetic field 4 mT; and estimated ion density 1011 ions cm–3. The energy spread of the ion beam was much larger than that expected from the thermal distribution of ion velocities in the plasma. Ero [6] and Levitskii [7] independently argued that the observed large energy spread anomalies are due to modification and variation of the plasma potential by

Figure 9.2

A Thonemann type RF ion source.

165

166

9 Radio-Frequency Driven Ion Sources

Figure 9.3 Cross-section of RIM-10 with block diagram of the supply units (C: RF-induction coil; D: gas distributor; Q: quartz discharge chamber; S: RF screen; G1, G2, G3: 1st, 2nd and 3rd grids)

energetic electrons that have come under the influence of strong RF electric fields existing in the region between the plasma and the electrode, or between the plasma and the walls. Researchers at the University of Giessen, Germany, later developed large crosssectional area (up to 10 cm in diameter) RF-driven sources for ion thrusters and neutral beam injector applications [8–10]. A schematic diagram of their RIM-10 model RF source is shown in Figure 9.3. A large quartz chamber is surrounded externally by the induction coil of the RF generator (1 to 30 MHz, power 1). That is, at low magnetic field even an RHCP wave is reflected at the overdenseplasma boundary, as for the case of non-magnetized plasma. In order to generate overdense plasma, the first thing to do is to create conditions where microwaves are efficiently introduced even in an overdense plasma. As a result the microwave energy is transformed into kinetic energy of plasma electrons and then consumed in enhancing the plasma density. When the static magnetic field is higher than the ECR field, that is, xce/x > 1, it is clear from Eq. (10.3) that the refractive index n takes a finite real value, even for overdense plasma. Kopecky et al. [5] described how microwaves are efficiently absorbed by a plasma at such high magnetic field and thus the plasma density increases up to several tens of times greater than the critical density which corresponds to the electron plasma angular frequency xpe. Figure 10.1 shows their experimental results. The experiments were carried out in an almost uniform magnetic field. The device operated in a pulse mode with repetition frequency of 50 Hz at a hydrogen pressure of 6.5 10–2 Pa (5.0 10–4 Torr). For microwave absorption, an initial plasma was generated inside the glass discharge tube by electrons from a hot cathode and then microwaves of various power levels (0.4, 1.2 and 2 kW) were supplied. Nstart denotes the minimum initial plasma density that was needed to start microwave absorption, and Nabs denotes the resultant plasma density enhanced by microwave absorption. In the small range of magnetic field intensity around ECR, (xce/x = 0.8–1.4), no initial plasma is needed for both plasma ignition and microwave absorption. It is seen from this figure that for a magnetic field greater than 1.2 times the ECR field, the plasma density increases up to several tens times the critical density, although the density at the ECR field is lower than the critical density. Consequently, in order to form overdense plasma in a uniform magnetic field without any initial plasma, the magnetic field should be established at slightly higher than the ECR field (preferably ~1.3 times the ECR field). Off-resonance microwave discharges can readily form overdense plasma as mentioned above, although the plasma density for the ECR plasmas of Okamoto [15] and Geller [16, 17] did not exceed the critical density, for which the electron plasma frequency equals the microwave frequency. 10.2.2.2

10.3

Some Practical Ion Source Considerations 10.3.1

Microwave Impedance Matching to the Plasma

Microwave ion sources for industrial use are usually driven by 2.45 GHz microwaves. One of the major problems in practical ion sources is how to introduce the microwave power through the plasma boundary into the plasma, even though

10.3 Some Practical Ion Source Considerations

microwaves that have already entered the plasma are absorbed as described above. From a microwave circuitry point of view, the ion source containing plasma is considered a circuit element. However, the impedance of this element depends on the plasma characteristics, which themselves depend on the absorbed microwave power and the superimposed magnetic field. Therefore, if the incident microwave power and the magnetic field are fixed, the plasma parameters and microwave absorption that result from the degree of microwave impedance matching are also fixed by a negative feedback mechanism. Otherwise the plasma could not be maintained. In the discharge chamber, the absorbed microwave power heats the electrons, in turn resulting in excitation and ionization of particles. The absorbed energy is then expended in heating the discharge chamber wall by charged particle bombardment and photon emission from excited particles. One method of realizing a microwave absorption element is to let the resistivity vary gradually along the length of the element. This is accomplished in a practical ion source by using a magnetic mirror field. The field also helps to protect the ceramic window used for vacuum sealing from plasma bombardment heating. An off-resonance microwave ion source described by Sakudo [10] is represented schematically in Figure 10.2. The discharge chamber basically consists of a coaxial waveguide. Microwaves (2.45 GHz) are introduced into the discharge chamber via a coaxial line transition from the rectangular waveguide. The antenna is connected to the inner conductor of the coaxial transmission line, which uses a ceramic window for the vacuum feedthrough. A magnetic field with a mirror ratio of about 2 is superimposed over the entire discharge chamber, using three magnetic coils. The magnetic flux density is higher than 87.5 mT for ECR throughout the entire discharge region. Due to the magnetic field gradient, plasma bombardment of the ceramic is minimized.

Figure 10.2 An off-resonance microwave ion source for high current singly charged ions. The magnetic flux density is higher than that for ECR throughout the entire discharge region. The inner diameter of the discharge chamber is 60 mm [10].

181

182

10 Microwave Ion Sources

Since the antenna is always in contact with the plasma, it is heated by charged particle bombardment. To avoid destruction of the ceramic, a water-cooled structure is employed. Usually, the coaxial waveguide is connected at right angles at a point kg/4 distant from the end of the rectangular waveguide, where kg is the wavelength in the rectangular waveguide. To cool the antenna, the inner conductor is made of a double tube with one end penetrating the end plate of the rectangular waveguide. The characteristic impedance of the coaxial waveguide is about 50 X. Therefore, the impedance of the plasma-filled discharge chamber should approximate this value for accurate impedance matching. Since the plasma is considered to be a resistive dielectric, the ratio of antenna diameter to chamber diameter should be smaller than that of the coaxial waveguide. The practical dimensions were determined empirically. The impedance of a plasma-filled ion source depends on the electron density and electron temperature, as well as on the magnetic field intensity and its distribution. In addition, the electron density and electron temperature change in a highly complicated manner, depending on the absorbed microwave power, gas pressure, and magnetic field. Therefore, it is practically impossible to examine the relation between any two of these parameters simply by holding the others constant. In practice, when the geometric structure of the ion source, gas species, and magnetic field are determined, the plasma parameters are fixed depending on the absorbed microwave power. This in turn fixes the extent of impedance matching between the microwave circuit and ion source, thus determining the absorbed microwave power. The relation between microwave absorption and magnetic field intensity is shown in Figure 10.3. In this case, the magnetic field intensity is varied while the field shape is maintained. The argon gas pressure is 1.0 Pa and the incident microwave power is 1 kW. Magnetic field intensity is normalized to that shown in the inset in Figure 10.2, which was determined empirically so as to obtain ion beams with high currents. Microwave absorption is expressed as a percentage of incident microwave

Figure 10.3 Microwave absorption as a function of magnetic field intensity. The magnetic field intensity is varied while the field shape is maintained. The argon gas pressure is 1.0 Pa [10].

10.3 Some Practical Ion Source Considerations

power. This absorption increases gradually with magnetic field intensity; its value exceeds 90% when the magnetic field intensity is above 95%. Below the 85% magnetic field level, there should be one or two points that satisfy the ECR condition. However, no resonant property can be seen in Figure 10.3. This means that the microwave power is mainly absorbed by off-resonance processes, in this case. This also suggests that even in a so-called ECR ion source, microwave absorption may be partially due to off-resonance effects, especially when the source is operated at a relatively high gas pressure. 10.3.2

High Current Ion Beams Extracted from an Off-Resonance Microwave Ion Source

When microwave absorption is plotted as a function of argon gas pressure, the change in absorption is gradual from 6.5 10–2 to 1.0 Pa (5.0 10–4 to 8.0 10–3 Torr). Over this entire pressure region the absorption is above 75%. However, by employing an EH tuner between the ion source and the directional coupler, more than 90% of the microwave power can be absorbed. Plasma parameters are measured with a movable Langmuir probe as shown in Figure 10.2. The magnetic field intensity at the position where the probe is placed is about 130 mT (1300 G). Since the probe axis is perpendicular to the magnetic field, the effect of magnetic field on probe measurements is negligible at the expected plasma parameters (ne > 1010 cm–3, Te > 1 eV) as shown by Dote [18]. The electron temperature of argon plasma and hydrogen plasma gradually increases from 3.7 to 6.8 eV and from 4.5 to 7.3 eV, respectively, as the absorbed microwave power varies from 200 to 1000 W. The electron densities are shown in Figure 10.4. These are 10–100 times higher than the critical density, which is often the uppermost value obtained by ECR. Any kind of waveguide can be used in place of the coaxial waveguide in the structure of the discharge chamber. A round waveguide, as in the microwave plasma etching apparatus described by Suzuki [19], is suitable for a large volume discharge chamber.

Figure 10.4 Electron density as a function of absorbed microwave power. Argon and hydrogen pressures are 0.49 and 0.13 Pa, respectively [10].

183

184


10.4

Versatility of Beam Extraction

Microwave plasmas can be generated over a wide range of volume and in various forms. This permits a variety of beam extraction geometries. 10.4.1

Large Cross Sectional Beam formed by a Multi-Aperture Extractor

The current density extracted from the plasma Ji depends upon the plasma parameters as follows: rffiffiffiffiffiffiffi kTe 1 exp ð Þ (10.4) Ji ¼ eni M 2 where k is Boltzmann’s constant, e electron charge, ni ion density, Te electron temperature, and M ion mass. A high degree of microwave energy absorption must be realized since high electron density and/or high electron temperature are necessary for a high current ion source. This equation is applied to all the ions that leave the source plasma through the plasma boundary. Since most of the boundary faces the chamber wall, only a small portion of the generated ions diffuses toward the extraction slit and is extracted as a beam. The remaining ions bombard the chamber wall with a plasma-potential energy that corresponds to roughly five times the electron temperature in eV. Therefore, the power dissipated as heat on the discharge-chamber walls is considerably greater than that used for ionization of the extracted ions. A three-element multi-aperture extractor grid system of 50 mm diameter is placed at the bottom of the discharge chamber, as shown in Figure 10.2. The potentials of the upper element and discharge chamber are at the same level as the ion acceleration voltage. The middle element is supplied with negative voltage to prevent low-energy electrons in the beam from entering the discharge chamber. Argon and hydrogen ion currents are measured with a cone-shaped Faraday cup having an effective area of 40 mm diameter and equipped with a secondary electron suppressor. The extracted argon ion current is shown in Figure 10.5. Current increases with ion acceleration voltage. Maximum argon ion current is 200 mA at 5 kV, and that of hydrogen is around 400 mA. The critical voltage at which it is possible to extract argon ions as a beam is around 2.0 kV, as shown in the figure, and that for hydrogen ions around 800 V. Ion sheath widths have been obtained from measurements made by the probe. In argon and hydrogen plasmas, these widths are 0.38 and 0.62 mm, respectively, when the absorbed microwave power is 950 W and the probe potential is –70 V. These values also suggest a difference between the critical voltages for argon and hydrogen plasmas. Oxygen ions, which are rather difficult to obtain with conventional hot-filamenttype ion sources, can be stably produced with this ion source [20]. At an ion extraction voltage of 5 kV, the oxygen ion current is 110 mA with a microwave power of 600 W. If the aperture radius of the extractor and the distance between the upper

10.4 Versatility of Beam Extraction

Figure 10.5

Argon ion current as a function of ion acceleration voltage [10].

and middle electrode elements are increased, ion extraction at the same current level can be optimized at a higher voltage, that is, several tens of kilovolts, as shown by Shimada [21]. 10.4.2

Slit-Shaped Beam for Ion Implantation

Sakudo [11] constructed a high current ion source using off-resonance microwave plasma, which provides a slit-shaped ion beam suitable for the electromagnetic mass separator of an ion implanter. This kind of separator requires a narrow ion beam to obtain high mass resolution. This is because the mass resolution Re in a symmetric-sector mass separator is (Jayaram [22]) Re ¼

r ; 2S

(10.5)

where S is the beam width at the ion exit, r is the beam trajectory radius in the magnetic field, and ion energy spread is neglected. Therefore, ion beams for an electromagnetic mass separator should be extracted through a narrow slit. The details of the first microwave ion source for use with an implanter [11, 23] are shown in Figure 10.6. Microwaves are introduced into the discharge chamber through a vacuum-sealing dielectric plate placed between the rectangular waveguide and the tapered, ridged waveguide. Two iron blocks are placed just on the outer periphery of the latter waveguide. The cross section of the discharge region has a rectangular shape, similar to the exit slit, to effectively utilize the generated ions. The remaining space for the discharge region is filled with dielectric (boron nitride). Since the electric field between the ridges, that is, the discharge electrodes, is fairly uniform, this structure helps to produce a uniform plasma.

185

186


Figure 10.6 The first microwave ion source used by an ion implanter for semiconductor device fabrication [11, 23].

Figure 10.7 The first high current ion implanter equipped with a microwave ion source. This machine was used on a semiconductor device production line of Hitachi Co. Ltd. for more than 10 years from 1977 [24].


An ion implanter equipped with this microwave ion source was constructed [24]. The microwave ion source with its slit-shaped beam is combined with a 40 cm radius, 90 deflection magnetic mass separator and a rotating disk target chamber, as shown in Figure 10.7. The ion energy can be varied from 20 to 80 keV. The implant current is continuously measured with a Faraday cup, which is composed of a rotating disk, an electron shield, and an electron suppressor, as shown in the inset of Figure 10.7. A high current ion beam of 40 mA is extracted through the 2 40 mm2 slit. Since the mass separator design is based on double-focusing optics with oblique entry and exit of ion beams, a high transmission level of up to 60% is attained. This permits 10 mA (maximum 15 mA) P+ implantation and 4 mA (maximum 6 mA) B+ implantation. These current values were the highest among those of ion implanters for semiconductor device fabrication until the early part of the 1980s. This implanter was used on a semiconductor device production line of Hitachi Co. Ltd. for more than 10 years from 1977. As described later in Section 10.6, the ion currents of modern implanters are much higher due to improvements in both the source design and the mass separator optics. The dependence of the mass peak ratio on the microwave power with PH3 gas introduced into the ion source is shown in Figure 10.8. The P+, H+, and P2+ intensities increase, and those of H2+, P2+, and PH+ decrease as the microwave power is increased. In the plasma, each PH3 molecule is bombarded many times by electrons. The larger the number of electron collisions, the more the molecule decomposes. As microwave power increases, both the electron density and the electron temperature of the plasma increase, resulting in an enhanced atomic-ion percen-

Figure 10.8

Dependence of mass peak heights for PH3 on incident microwave power.

187

188


tage. On the other hand, as the gas pressure increases, the electron temperature decreases and recombination of decomposed particles increases, resulting in an enhanced molecular-ion percentage. Thus, the molecule decomposition rate is widely controlled by changing both or either of the microwave power and the gas pressure. The lifetime of this ion source is 1–2 A h (implant current duration) for 10 mA implantations of P+ and As+. This is several times longer than that of most conventional hot cathode type ion sources. The PH3 gas consumption for these high current operating conditions is 0.1–0.5 atm cm3 min–1, which is several times lower than that of conventional hot cathode type ion sources. This long lifetime and low gas consumption rate leads to reduced maintenance frequency for the ion source and vacuum system, and to an increased operating rate for the implanter. 10.4.3

Further Improvements in Slit-Shaped Beams 10.4.3.1 Increasing Extraction Voltage When a source is operated at a higher voltage simply by scaling up the standard microwave ion source with a slit-shaped beam (e.g., the source as shown in Figure 10.6), a number of problems occur:

1.

It is difficult to achieve sufficient voltage isolation between the high potential discharge chamber and the magnetic field coil at ground potential, because their relative distance cannot be altered while the magnetic field intensity and distribution are maintained throughout the discharge region.

Figure 10.9 Microwave ion source having a closed magnetic circuit. Since the magnetic field between the accel–decel electrodes is completely eliminated, high-voltage beam extraction is stably operated. This structure is also used with a double-stage extractor to obtain higher ion energy [25].


2.

3.

Sparks between electrodes in the vacuum are liable to occur in the ion acceleration space, since a magnetic field is superimposed on the accelerating electric field. It is difficult to use a multistage extraction lens because of the magnetic coil at ground potential.

To avoid these problems, the ion source has been further modified [25]. The magnetic coil is positioned surrounding the discharge chamber at the high voltage terminal, where it is also surrounded by a thick wall and an acceleration electrode, which are made of iron or high-permeability metal as shown in Figure 10.9. Thus a magnetic circuit is formed to focus the magnetic field into the discharge chamber. Consequently no magnetic field exists in the ion accelerating space in the new ion source. A double-stage extraction lens, which is not shown in this figure, is also adopted. This ion source provides stable, mass-separated ion beams at extraction voltages of 20 to 120 kV with current of up to 15 mA. A further increase in ion acceleration voltage can be realized by increasing the number of lens stages. Increasing Extraction Current In general, to increase the current from an ion source in which the plasma density has reached a maximum, the ion-emitting area must be enlarged. It is very easy to lengthen the ion exit slit of the microwave ion source, although this may be somewhat difficult in conventional Freeman and Bernas ion sources because the filament for Freeman sources and the cathode to anode distance for Bernas sources must 10.4.3.2

Mass-separated currents are increased by combining a microwave ion source with lengthened slit with a conventional mass separator. a) is the normal slit and b) is the lengthened and curved slit [26].

Figure 10.10

189

190


also be lengthened. Extracted current from a microwave ion source can easily be increased by lengthening the slit without changing the ion optics in the narrower direction of the slit, as shown in Figure 10.10. If the extraction lens is curved along the longer direction to converge the ion beam into a magnet pole gap of a conventional mass separator, mass-separated currents are also increased without degrading the mass resolution. In this figure, a) is the normal slit and b) is the lengthened and curved slit. The cross section of the discharge chamber made of boron nitride is shown as the inset of the figure. In the case where the ion exit slit is 80 mm high, a mass-separated P+ ion of 30 mA was obtained at the collector, although the effective height of the magnet-pole gap is 45 mm [26]. The curved extraction lens is not adaptable to Freeman and Bernas sources because their source plasmas are basically straight and parallel to the extraction slits of their standard ion sources. 10.4.4

Compact Microwave Ion Sources

Ishikawa [27] constructed a compact microwave ion source that extracts an ion beam of several milliamperes through a hole of 2 mm in diameter as shown in Figure 10.11. To make the source small, a permanent magnet was used to replace the solenoids. The source is 50 mm in diameter and 65 mm in height. The microwave fre-

A compact microwave ion source using a permanent magnet: 1, sealed coaxial connector; 2, top flange; 3, antenna; 4, boron nitride; 5, sheath heater; 6, middle flange;

Figure 10.11

7, gas inlet; 8, copper gasket; 9, permanent magnet; 10, base flange; 11, plate with an extraction aperture; 12, insulator; 13, extraction electrode; 14, coolant; and 15, spacers [27].

10.5 Diversity of Available Ion Species

quency is 2.45 GHz and the power consumption is only 7 to 30 W. The ion beam has low emittance of 10–8 m rad and high brightness of 1011 A m–2 rad–2. A microwave cavity plasma disk ion source was developed by Asmussen [28]. It uses a cylindrical microwave cavity operating in a hybrid mode associated with the TE211 empty cavity mode. Without the aid of a static magnetic field, microwave fields produce plasma in a disk-shaped quartz tube placed in the cavity. Ion beams are extracted with a multi-aperture lens adjacent to the plasma. Since discharge losses per extracted ion are roughly proportional to the total surface area divided by the extraction area, the dish shape, when the height is much less than the radius, is an efficient discharge configuration. The performance of the ion source was improved by redesigning to surround the discharge zone with many closely spaced rare earth magnets producing a multi-cusp static magnetic field (Dahimene [29]). Leung [30] constructed a small microwave ion source from a quartz tube with one end enclosed by a two-grid extraction lens. The source is also enclosed by a microwave cavity. An ion beam is extracted through a 0.8 mm diameter extraction aperture. The current density exceeds 200 mA cm–2 when the microwave power is 400 W.

10.5

Diversity of Available Ion Species

General ion species for implantation into silicon semiconductors have traditionally been P+, As+ and B+ ions. However, new silicon devices using silicon-on-insulator (SOI) technology require oxygen ion implantation at very high doses. Microwave ion sources can stably use even reactive materials as the source materials. Some of these sources, with slit-shaped beams, can provide a mass-separated O+ ion beam of more than 10 mA for such implantation. In this case, the ratio of O+ ions to the total ions in the mass spectrum is about 0.85, as shown in Figure 10.12. This high atomic fraction is obtained by feeding high power microwaves up to 1 kW. Decomposition of oxygen molecules is enhanced due to the high electron density and/or the high electron temperature of the microwave plasma. A research group at NTT Japan was the first to fabricate a 4-kb CMOS/SIMOX SRAM by O+ ion implantation with a commercial implanter (Hitachi IP-815), as reported by Omura [31]. For mass production of SOI wafers, higher current oxygen ion implanters are needed. Commercial implanters dedicated to SIMOX wafer fabrication are described later in Section 10.6. For implantation into compound semiconductors and for materials modification in metals and insulators, various other kinds of metal ions are required. Source materials for most metal ions are readily obtained in the form of halides, the evaporation temperatures of which are usually lower than those of the original metals. In traditional hot-filament-type ion sources halides have sometimes caused problems due to rapid erosion of the filament by the highly reactive halide plasma. However, the microwave ion source does not have any quickly consumed parts. Furthermore, halogen ions and radicals in the plasma clean the chamber walls by both ion sputter-

191

192


Figure 10.12 Mass spectrum of oxygen ions generated by an off-resonance microwave discharge. High atomic fraction is obtained by feeding high-power microwaves.

ing and plasma-chemical etching. This helps prevent metallic elements from depositing on the inner walls of the discharge chamber, through which microwaves enter the plasma. To introduce liquid or solid materials into the chamber, the microwave ion source has two types of oven, an external one and an internal one, in addition to the usual gas inlet system as shown in Figure 10.6. Low vapor pressure materials can be introduced with either oven, depending on the required evaporation temperature. The mass spectrum for TiCl4 extracted at 30 kV is shown in Figure 10.13. A Ti+ isotope ion with a 3.5 mA peak is obtained. Since TiCl4 is a high vapor pressure liquid (about 103 Pa at 20 C), it is fed through the needle valve like gaseous material. A Ti+ ion implant current of 3.5 mA is obtained at 30 kV. Source materials that have to be heated up to around 200 C to obtain a vapor pressure greater than 100 Pa are used with the external oven. Hf+ isotope ions of 2 mA are obtained with the source material HfCl4 being heated to 170 C in the external oven. Solid materials that have very low vapor pressure at room temperature are heated in the internal oven up to several hundred C. This ion source can provide most metal ion beams at a current level of several milliamperes by vaporizing the metal halides with either of the ovens. Some elements, whose compounds are hard to evaporate as well as the pure elements, can be ionized by plasma chlorination. For example, in order to get an Sc+ ion beam, Sc2O3 powder is placed in the discharge chamber and CCl4 is fed through the needle valve. In the chamber, Sc2O3 is chlorinated and vaporized by a plasma-chemical reaction with chlorine ions or radicals in the plasma. A Sc+ ion implant current of 0.5 mA is obtained. An ion implanter that is dedicated to materials modification in metals and insulators was designed by Iwaki [32], and is shown in Figure 10.14. The ion beam line is basically that of a commercial ion implanter, Lintott series of Applied Materials Corp., except for the ion source and the implant chamber. Ions are extracted at 30 kV from the microwave ion source, the structure of which is the same as that shown in Figure 10.6. As the ions are post-accelerated, the maximum acceleration voltage can be enhanced up to 200 kV. The target chamber has a secondary mass

10.5 Diversity of Available Ion Species

Figure 10.13 Mass spectrum of TiCl4. By feeding TiCl4 through the needle valve, 3.5 mA of Ti+ ion current was obtained at 30 keV [23].

Figure 10.14

An ion implanter for materials modification in metals and insulators [32].

193

194


analyzer for observing the modified surface composition. Analysis can be performed after ion implantation without breaking vacuum. Most metal ions at several milliamperes beam current are obtained with the halides containing the elements fed into the source. Many experiments on high dose implantation into non-semiconductor materials were carried out with this machine.

10.6

Microwave Ion Sources for Commercial Implanters 10.6.1

Semiconductor Device Fabrication Brief History of Ion Sources for Commercial Implanters Industrial applications of ion beams began in the early 1970s with their application to semiconductor device fabrication. Ion implantation was firstly utilized for adjusting the VTH, threshold gate voltage, of MOS (metal oxide semiconductor) devices. B+ or P+ ions were implanted into the channel region through a gate-oxide layer of about 0.1 lm in thickness. Since the required doses were fairly low (4 1011 to 6 1012 cm–2), low current ion implanters that could provide an implant current of up to several tens of lA were used in the production lines of semiconductor devices. On the other hand, towards the middle of the 1970s, most doping processes for LSI fabrication besides the above-mentioned channel doping shifted from thermal diffusion to ion implantation as device integration proceeded. Higher dose implants of up to 1016 cm–2 become necessary to form source and drain regions of MOS as well as emitter regions of bipolar transistors. The Freeman source, which had been invented for high-resolution isotope separations in the early 1960s [33], was adopted as the ion source for the high current ion implanters of Extrion Corp. and Nova Associates Corp. These implanters could provide several mA of ion current for every ion species needed for LSI fabrication. For the next ten years towards the mid-1980s, the Freeman source was almost the only source to be employed by commercial manufacturers [34]. Then the Bernas source [35] largely replaced the Freeman source due to its more stable operation. 10.6.1.1

Microwave Ion Sources In order to resolve the lifetime problem of conventional sources, a microwave ion source for implantation was developed by Sakudo [11, 34]. Microwave ion sources for industrial use are usually driven by 2.45 GHz microwaves. They can provide higher ion currents than conventional Freeman and Bernas sources for most elements. In addition, microwave ion sources are free from metallic contamination that is often a serious problem in modern silicon semiconductor device production. In practice, some of the conventional DC discharge ion sources have discharge chambers made of molybdenum that may be sputtered and ionized in the source plasma. Since Mo2+ ion has the same mass to charge ratio as the BF2+ ion that is used for ultra-shallow doping of boron into modern ULSI (Ultra Large Scale Inte10.6.1.2

10.6 Microwave Ion Sources for Commercial Implanters

grated Circuit) devices, Mo2+ ions cannot be mass-separated from BF2+ ions. Furthermore, the filament material, tungsten has a similar problem. The mass to charge ratio of the W2+ ion is so close to that of the BF2+ ion that mass separation of these two ion species is rather difficult without a high-resolution mass separator [36]. The first commercial high current implanter equipped with a microwave ion source was the IP-815 of Hitachi Ltd. Since the source structure was similar to that shown in Figure 10.6, the maximum implant energy and current were 80 keV and 10 mA, respectively. Then, the new implanter, IP-825, adopted a closed magnetic circuit for the microwave ion source, as shown in Figure 10.9, as well as a double-stage extraction system. It could provide a maximum As+ ion current of 15 mA over an energy range from 20 to 120 keV. About 40 implanters of this series have been sold and are working on ULSI production lines in Japan. As the integration of semiconductor devices proceeds, each device becomes smaller, resulting in the need for smaller p–n junction depth. Thus the implant energy of low-mass ions such as B+ must be lowered. Since the transmission of a low energy ion beam through a mass-separator beam line becomes poor due to high beam emittance, much higher currents must be extracted to maintain the implant currents, even at very low extraction voltages. Applied Materials Corp. has constructed a new microwave ion source that is basically the same as the Hitachi commercial implanter IP-825 but has been improved to accommodate the new processes for advanced semiconductor devices, as shown in Figure 10.15. The magnetic circuit modeling of the microwave ion source using a 2D FEM (Finite Element Method) program, Quick Field [37] is shown. The source enclosure forms a closed magnetic circuit to confine and direct a magnetic flux into the discharge chamber.

Figure 10.15 Magnetic field analysis of a microwave ion source for optimizing the discharge chamber volume. The magnetic circuit is drawn in black. The discharge chamber in

this figure is that used before optimization. The field is simulated by the 2D FEM code Quick Field. Magnetic flux lines are shown by dotted lines [39].

195

196


Microwaves are introduced horizontally from the right side of the figure through a dielectric conduit into the discharge chamber, which is surrounded by a solid waveguide, omitted in the figure. The discharge chamber in this figure is the old one before optimization. The highest magnetic flux points are outside the discharge

Figure 10.16

Magnetic field in the volume-optimized discharge chamber [40].

10.6 Microwave Ion Sources for Commercial Implanters

chamber and the flux lines are short because most of them are terminated on the sidewalls of the chamber [38]. The discharge chamber volume of the new microwave ion source has been optimized by comparing with the magnetic field analysis [39, 40]. Figure 10.16 shows (a) the magnetic flux lines and (b) the magnetic flux density distribution in the volume-optimized discharge chamber. The high magnetic flux density regions of over 20 mT are inside the discharge chamber. The ions generated in the high magnetic flux regions will flow into and remain in the spacious region between the pole pieces of the magnetic circuit and the extraction electrode, resulting in enhancement of the plasma density. This ion source has been mounted on the beam line of a commercial high current implanter, xR80 of Applied Materials Corp., which was originally equipped with a Bernas source, providing maximum currents of 15 mA of As+ ions and 10 mA of B+ ions. The microwave ion source has significantly enhanced the currents. The maximum currents of As+ and B+ ions are 25 mA and 15 mA, respectively. In addition, the ion current dynamic ranges have been extended as shown in Figure 10.17 [39, 40]. This source provides a very wide range of As+ ion current from 1 lA to 26 mA. Solid arsenic is vaporized with an internal oven. Both the microwave power and the oven temperature are varied in order to control the plasma density. Although the ion current is proportional to the plasma density, the dynamic range of plasma density that is controlled by these two parameters is up to 2 or 3 orders of magnitude. However, a new technique has extended it to over 3 orders of magnitude. For very low ion current, argon gas is added at a flow rate of 0.7 sccm as the discharge supporting gas. The dynamic range of B+ ion current is also from 1 lA to 15 mA by using the same method. These wide dynamic ranges cover the performance of both mediumcurrent and high-current implanters.

Dynamic range of As+ ion beam at 20 keV. Solid arsenic is vaporized with an oven. Both the microwave power and the oven temperature are varied to control the plasma density as widely as possible. For a very low ion current, argon is added as the discharge supporting gas at a flow rate of 0.7 sccm [40]. Figure 10.17

197

198


As for the BF2+ ion current that is used for ultra-shallow doping of boron in modern ULSI, the microwave ion source provides 18 mA, twice that obtained from a Bernas source on the same xR80 beam line. In order to produce high current molecular-ion beams, the electron energy of the high-density plasma should be lowered. Although the microwave ion source can realize low temperature and high density plasma by changing both the microwave power and the gas pressure, the DC discharge type source originally has high energy electrons due to the arc voltage (several tens of volts) besides the plasma electrons. The high energy electrons promote too much dissociation of BF3 molecules, resulting in a low current of BF2+ ions with the Bernas source. 10.6.2

SOI Wafer Fabrication

Most SOI wafers for advanced semiconductor devices are competitively made by two major processes: SIMOX [42], and wafer bonding with hydrogen implantation [42]. The common key technology for both processes is very high dose ion implantation. The SIMOX process employs oxygen implantation at a dose of about 1 1018 O+ cm–2. The wafer bonding employs hydrogen ion implantation at a dose of about 2 1017 H+ cm–2, a fifth of the oxygen dose for SIMOX. 10.6.2.1 Oxygen Ion Implanters for SIMOX Process The required O+ dose for SIMOX is about 100 times the highest dose for ordinary implantation into semiconductor devices used in large scale production. Another specific feature of the SIMOX process is wafer heating. The wafer temperature should be kept at 500–650 C during O+ implantation. Eaton Corp. developed a 100 mA class O+ ion implanter dedicated to SIMOX, the NV-200, in 1987 [41]. This machine employed a microwave ion source with a round-waveguide type discharge

Microwave ion source for a commercial SIMOX ion implanter [44]. (Courtesy of K. Tokiguchi.)

Figure 10.18

10.7 Conclusion

chamber [21]. In order to realize a high current O+ ion beam, the mass resolution of the separator was reduced to 8 by increasing the ion emitting area of the ion source; this is the lowest value for separating O+ ions from the neighboring N+ ions. The implantation time for the required dose was 6 min per 100 mm wafer, i.e. 2.5 h per batch of 25 wafers. In the mid-1990s, Ibis Technology Corp. and Hitachi Co. Ltd. started producing commercial SIMOX implanters, the Ibis1000 [43] and the UI-6000 [44], respectively. Figure 10.18 shows the microwave ion source for the Hitachi UI-6000. The ion source operates at the off-resonance condition where the magnetic flux density is higher over the entire plasma region than for ECR (87.5 mT for 2.45 GHz). The source incorporates a unique transform waveguide that consists of 3 parts: a tapered rectangular waveguide, a ridged circular waveguide and a conventional waveguide. Ions are extracted with a multi-aperture extractor having 13 holes of 5.5 mm in diameter. 10.6.2.2 Hydrogen Ion Implanters for Wafer Bonding This process is based on the delamination of a thin single-crystal film from a silicon wafer by H+ ion implantation. The ion implantation induces formation of an indepth weakened layer at a depth corresponding to the mean projected range of the ion in silicon. The SOI wafer is made by bonding the thin film to the oxidized surface of another wafer. The advantages of this process over SIMOX are the lower dose (1/5 of the O+ dose for SIMOX) and the removal of the need for wafer heating. Thus, ordinary high current implanters for semiconductor device fabrication can readily be used as H+ ion implanters for SOI wafer fabrication without any special modification. A normal high current implanter that is equipped with a Bernas source provides an H+ ion beam of 20 mA. This current level allows the same implantation time as that of a 100 mA class O+ ion implantation for SIMOX. Use of a microwave ion source in place of the Bernas source extends the performance remarkably. A high current beam of molecular hydrogen ions H2+ is easily obtained by controlling both the microwave power and the gas pressure. The implanter equipped with a microwave ion source provides an H2+ ion current of 50 mA that corresponds to 100 mA H+ ion current from the viewpoint of implantation time. Thus the implantation time for the bonding process with H2+ ions from a microwave ion source is 1/5 of that for SIMOX with 100 mA O+ ion implantation.

10.7

Conclusion

Microwave ion sources that are operated in the off-resonance condition can provide long-life stable ion beams for a variety of ion species by using a wide range of source materials including reactive chemical compounds. In practice, an ion source with large cross-sectional beams of several hundred milliamperes has been developed, whose ion current can be further increased by increasing the discharge chamber volume and the area of the multi-aperture extractor. Another ion source, producing slit-shaped beams, has been constructed for application to ion implantation. An

199

200


implanter equipped with this ion source can provide implantation currents of several mA to several tens of mA for most ion species, including metallic elements. To increase the ion acceleration voltage, this ion source can be easily combined with multistage extraction systems. To increase the ion current, the ion exit slit can be easily lengthened. With versatility of beam extraction, many other kinds of modification can be made to suit particular industrial applications. Microwave ion sources have many advantages in applications to industrial use besides the long lifetime. For semiconductor device fabrication they are free from metallic contamination that is often a serious problem for producing silicon semiconductor devices with traditional DC-discharge ion sources. They can provide beams of either atomic ions or molecular ions at a higher current level than other types of ion sources, by controlling the operational conditions. For SOI wafer fabrication, microwave ion sources will be exclusively used. In this field, two different processes, SIMOX and wafer bonding with hydrogen implantation, compete. However, both processes use microwave ion sources; O+ ions are used for SIMOX and H2+ ions for wafer bonding.

References [1] J. Musil and F. Zacek, Czeck. J. Phys. B 22, [2] [3]

[4] [5] [6]

[7] [8] [9] [10]

133 (1972). J. Musil and F. Zacek, Czeck. J. Phys. B 23, 736 (1973). J. Musil and F. Zacek, Experimental study of the absorption of intense electromagnetic waves in a magnetoactive plasma, Research Report of the Institute of Plasma Physics, Czechoslovak Academy of Science, Rpt. No. IPPCZ-180, (1973) p16 J. Musil. F. Zacek, and P. Schmiedbergen, Plasma Phys. 16, 971 (1974). V. Kopecky, J. Musil, and F. Zacek, Plasma Phys. 17, 1147 (1975). S. Nanobashvili, G. Rostomashvili, and N. Tsintsadze, Sov. Phys. Tech. Phys. 20, 280 (1975). C. Cook, O. Heinz. D. Lorenz, and J. Peterson, Rev. Sci. Instrum. 33, 649 (1962). E. Szuszczewicz, Phys. Fluids 15, 2240 (1972). I.G. Brown, Plasma Phys. 18, 205 (1976). N. Sakudo, K. Tokiguchi. H. Koike, and I. Kanomata, Rev. Sci. Instrum. 48, 762 (1977).

[11] N. Sakudo, K. Tokiguchi. H. Koike, and

I. Kanomata, Rev. Sci. Instrum. 49, 940 (1978). [12] K. Tokiguchi, N. Sakudo, and H. Koike, J. Vac. Sci. Technol. A 2, 29 (1984). [13] T.H. Stix, The Theory of Plasma Waves (McGraw-Hill, New York, 1962), p. 32. [14] M. Heald and C. Wharton, Plasma Diagnostics with Microwaves (Wiley, New York, 1965) [15] Y. Okamoto and H. Tamagawa, Rev. Sci. Instrum. 43, 1193 (1972). [16] R. Geller, IEEE Trans. Nucl. Sci. NS-23, 904 (1976). [17] R. Geller, B. Jacquot, and C. Jacquot, in Proceedings of the International Ion Engineering Congress, Kyoto (1983), p. 187. [18] T. Dote, H. Amemiya, and T. Ichimiya. Jpn. J. Appl. Phys. 3, 789 (1964). [19] K. Suzuki. S. Okudaira, N. Sakudo, and I. Kanomata, Jpn. J. Appl. Phys. 16, 1979 (1977). [20] K. Tokiguchi, H. Itoh, N. Sakudo, H. Koike, and T. Saitoh. Vacuum 36, 11 (1986). [21] M. Shimada, I. Watanabe, and Y. Torii, in Proceedings of the 10th Symposium on Ion Source

References and Ion-Assisted Technology, Tokyo (1986), p. 131. [22] R. Jayaram, Mass Spectrometry, (Plenum, New York, 1966), p. 25. [23] N. Sakudo, K. Tokiguchi, and H. Koike, Vacuum 34, 245 (1984). [24] N. Sakudo, K. Tokiguchi, and H. Koike, Rev. Sci. Instrum. 54, 681 (1983). [25] N. Sakudo, Nucl. Instrum. Methods Phys. Res. B 21, 168 (1987). [26] N. Sakudo, Rev. Sci. Instrum. 69, 825 (1998) [27] J. Ishikawa, Y. Takeiri, and T. Takagi, Rev. Sci. Instrum. 55, 449 (1984). [28] J. Asmussen and J. Root, Appl. Phys. Lett. 44, 396 (1984). [29] M. Dahimene and J. Asmussen, J. Vac. Sci. Technol. B 4, 126 (1986). [30] K.N. Leung, S. Walther, and H.W. Owren, IEEE Trans. Nucl. Sci. NS-32 (5), 1803 (1984). [31] Y. Omura, S. Nakashima, and K. Izumi, in Proceedings of the Symposium on VLSI Technology, Kobe (1985), p. 24. [32] M. Iwaki, K. Yoshida, N. Sakudo, and S. Satou, Nucl. Instrum. Methods Phys. Res. B 6, 51 (1985). [33] J.H. Freeman, Electromagnetic Separation of Radio Isotopes, edited by M.J. Higatsberger and F.P. Viehbock (Springer, Vienna, 1961), p. 204. [34] P.H. Rose, Rev. Sci. Instrum. 61, 342 (1990).

[35] R. Bernas and O. Nier, Rev. Sci. Instrum. 19,

895 (1948). [36] R.B. Liebert, G.C. Angel and M. Kase, in Pro-

ceedings on 11th Ion Implantation Technology, Austin, Texas (1996), p. 135. [37] Tera Analysis, Quick Field, CA (1995). [38] H. Ito, S. Ito, M. Takahashi, N. Sakudo and A. Kawasaki, in Proceedings on 12th Ion Implantation Technology, Kyoto (1998), p. 558. [39] N. Sakudo, Rev. Sci. Instrum., 71, 1016 (2000). [40] N. Sakudo, H. Ito, Y. Matsunaga, K. Hayashi, A. Miyamoto and Y. Tazaki, Rev. Sci. Instrum. 73, 812 (2002). [41] K. Izumi, Nucl. Instrum. Methods B 21, 121 (1987). [42] B. Aspar, C. Lagahe, H. Moriceau, A. Soubie, E. Jalaguir, B. Biasse, A. Papon, A. Chabli, A. Clavier, J. Grisolia, G. Benassayag, T. Barge, F. Letertre and B. Ghyselen, in Proceeding on 12th International Conference on Ion Implantation Technology, Kyoto (1998), p. 255. [43] G. Ryding, T.H. Smick, M. Farley, B.F. Cordts, R.P. Dolan, L.P. Allen, B. Mathews, W. Wray, B. Amundsen, and M.J. Anc, in Proceeding on 11th International Conference on Ion Implantation Technology, Austin (1996), p. 436. [44] K. Tokiguchi, T. Seki, K. Amemiya and Y. Yamashita, in Proceeding of 11th International Conference on Ion Implantation Technology, Austin (1996), p. 287.

201

203

11

ECR Ion Sources Daniela Leitner and Claude Lyneis

11.1

Introduction

ECR (electron cyclotron resonance) ion sources are widely used for the production of high quality multiply charged ion beams for accelerators, atomic physics research and industrial applications. A wide variety of designs have evolved, tailored to these applications. The development and refinement of ECR ion sources over the last three decades has provided remarkable improvements in their performance. For example in 1980, Micromafios produced 15 elA of O6+ [1] and 16 years later in 1996 the CAPRICE source produced 1100 elA [2], an increase by a factor of 73! Recently the RIKEN 18 GHz source produced 2000 elA of Ar8+ [3]. Development efforts for even more powerful ECR ion sources are underway, made possible by technical progress on magnets and microwave generators, which allows the application of stronger magnetic fields and higher frequencies. Research in nuclear physics with heavy ions continues to drive and fund the development of these sources to achieve higher charge state ions, and more intense beams with better emittance. The growing field of radioactive beam nuclear physics has added new impetus for the development of a new generation of high frequency, high magnetic field ECR ion sources and for sources with high ionization efficiency for the production of beams from rare and or radioactive isotopes [4–6]. Recently, ECR ion sources have been developed to produce very high currents (>100 mA) of singly charged ions for accelerators [7]. Several general characteristics of ECR sources explain their widespread application in the accelerator community. Most important is the ability to produce CW beams from any element at useful intensities for nuclear and atomic physics research. Another characteristic of ECR sources is that the discharge is produced without cathodes. Therefore, only the source material injected into an ECR source is consumed. As a result, ECR sources can be operated continuously for long periods without interruption. Maintenance required on ECR sources is also minimal, consisting mainly of occasional repair of vacuum equipment, external ovens and electrical support equipment.


204

11 ECR Ion Sources

11.2

Brief History of the Development of ECR Ion Sources

The field of ECR ion sources has its roots in the plasma fusion developments in the late 1960s. The use of electron cyclotron resonance heating (ECRH) in plasma devices to produce high charge state ions was suggested in 1969 [8]. The first sources actually using ECRH to produce multiply charged ions were reported in

Figure 11.1 Schematic drawing of the SUPERMAFIOS ECR ion source. (Courtesy of the Lawrence Berkeley National Laboratory.)

11.2 Brief History of the Development of ECR Ion Sources

1972 in France by Geller et al. [9] and in Germany by Wiesemann et al. [10]. These devices, which used solenoid magnetic mirror configurations and operated at pressures of 10–4 to 10–5 Torr, were capable of producing plasma densities of the order of 1 1012 cm–3 and keV electrons. However, the ion confinement times were 10–4 s or less and this resulted in low charge state distributions (CSD) which, for example, peaked at N2+ for nitrogen and Ar2+ for argon. A breakthrough occurred in 1974 when Geller and coworkers transformed a large mirror device used for plasma research (CIRCE) [11] into an extremely successful ion source, SUPERMAFIOS [12], which is shown in Figure 11.1. Unlike the earlier ion sources using ECRH, the magnetic field of CIRCE, renamed SUPERMAFIOS, used a hexapole field in addition to the solenoidal mirror field. This produced a minimum-B magnetic field configuration that stabilized the plasma against MHD instabilities. It also provided a closed ellipsoidal surface (the resonant zone) inside the plasma chamber where the electrons were heated by ECRH. With this configuration, the loss rate of hot electrons and of ions from the plasma was significantly reduced. In addition, CIRCE was a two-stage device. A cold plasma was generated in the first stage, operating at a higher pressure of about 10–3 Torr. This cold plasma flowed along the magnetic field lines, feeding the main confinement stage, which operated at much lower pressure, about 10–6 Torr. These new features, plasma stabilization in a minimum-B field configuration, a closed ECR surface, and a main stage operating at low neutral pressure, resulted in an enormous improvement in the lifetime of the ions in the source (~10–2 s) and shifted the CSD to much higher charge states. Several variations of this ECR source were tested between 1974 and 1977, however the basic features of the main stage design remained unchanged. The magnetic topology and axial extraction system associated with the main stage of SUPERMAFIOS are found in all modern high charge state ECR ion sources. The main drawback of SUPERMAFIOS was the 3 MW power consumed by the large room temperature copper coils needed to produce the solenoid and sextupole fields in the main stage. Different solutions were considered to reduce this power. The use of samarium–cobalt (Sm–Co) permanent magnets was initially considered impractical because it was assumed that ECR sources should be as large as SUPERMAFIOS (diameter 30 cm, length 100 cm). Therefore, projects to build large superconducting ECR sources were started in Louvain-la-Neuve [13], Karlsruhe [14] and Jlich [15]. In 1979 Geller transformed a reduced scale permanent magnet model hexapole built in Louvain-la-Neuve into a successful source of much smaller size (diameter 7 cm, length 30 cm) called MICROMAFIOS [1]. After development and modification it was renamed MINIMAFIOS [16]. MINIMAFIOS sources have since been built by Geller’s group for laboratories at KVI-Groningen, SARA-Grenoble, GANILCaen, GSI/CERN-Geneva, and NAC-South Africa. In the period from 1980 to 1985, the development of ECR ion sources spread to other laboratories [17–22]. A second generation of ECR ion sources was launched in 1984 when the Grenoble group constructed a new source, MINIMAFIOS-16 GHz, using higher microwave frequencies and stronger magnetic fields [23]. These improvements significantly increased both the total extracted current and the charge state distribution.

205

206

11 ECR Ion Sources Table 11.1

f(GHz) O

16

40

Ar

Xe

Performance of various 14 and 18 GHz ECR ion sources.

6+ 7+ 8+ 9+ 11+ 12+ 13+ 14+ 16+ 17+ 18+ 20+ 25+ 26+ 27+ 28+ 29+ 30+ 31+ 32+ 33+ 34+ 35+ 36+ 37+ 38+

SNanogan AECR-U Caprice

GTS

RIKEN

SERSE

SERSE

14 150 20 330 130 70 24 8 1

18 1500

18 1000

14+18 540 225

28

405

1900 1000 300

6.5

0.3

10+14 840* 360*

270 192 120 82 25.6 3.1 0.25 46 46 30 26.5 21 12 7 4.6 2.9 * 1.6 1 0.6 0.25

14 1000 190 800 500 200 100 40 15 1

1.8 80 40 40 20 10 4

216

295

260 200 130 84 21 2.6 0.4 135

92 65 28 18

4

38.5 23.5

300 216

9.1 5.2

164

84

100

*

7

18 1100 324

290 180

78

11

VENUS

28 15 9 3

35

2 8

*

Performance date for three frequency (8.6+10+14 GHz) operation have been taken from Refs. [2, 37–39].

Many second-generation ECR ion sources followed that also used higher frequencies and stronger magnetic fields to increase performance (see references in Table 11.1). While most of the sources used permanent magnets for the sextupole and copper coils for the solenoids, a number of fully superconducting ECR ion sources were built [24–26]. Recently, with advances in superconducting magnet technology, a new generation of high field superconducting sources is beginning to emerge, the first being the superconducting source SERSE. The most advanced superconducting source now in operation is the VENUS ECR ion source, which is briefly described later in this chapter. However, several projects and design have been started around the world to build third generation superconducting ECR ion sources [27–31].

11.3 The LBNL ECR Ion Sources

11.3

The LBNL ECR Ion Sources

In this section we describe two sources, the advanced Electron Cyclotron Resonance-Upgrade (AECR-U) [32] and the Versatile ECR ion source for NUclear Science (VENUS) [29, 31], as examples of second and third generation ECR ion sources. To give the reader some perspective the performances of the two LBNL sources are compared with other high performance sources for selected beams in Table 11.1. In addition, SuperNanogan, a high performance, compact, fully permanent magnet ECR ion source, is listed for comparison with the conventional 14 GHz ECR ion sources. Furthermore, the SERSE results at 28 GHz for Xe are also listed in Table 11.1 as a demonstration of the frequency scaling [33]. 11.3.1

The AECR-U Ion Source

The AECR-U ion source is a conventional ECR ion source that has been optimized for the production of high charge state ions. The design combines all of the current ECR ion source improvement techniques. Shown in Figure 11.2 is a cross-sectional view of the AECR-U ion source. It was built in 1990 and was upgraded in 1996 to further enhance its performance. Slightly modified versions of this source are in operation at three other laboratories [34–36].

Figure 11.2 An elevation view of the AECR-U. The ECR zones for 10 and 14 GHz are indicated by ellipses at the center of the plasma chamber, since the source can be operated with twofrequency heating.

207

208

11 ECR Ion Sources

Its maximum peak fields on axis are 1.7 and 1.1 T at the injection and extraction regions, respectively. The axial fields are produced by hollow core copper coils with two iron yokes that enhance the field at injection and extraction while reducing it at the center. The maximum radial field at the inner surface of the aluminum plasma chamber is 0.85 T, which is produced by a 12 piece NdFe-B sextupole. The aluminum plasma chamber is 30 cm in length with an inner diameter of 7.6 cm. Six radial slots and an external turbo pump give about 150 l s–1 pumping speed. The radial slots also provide easy oven access to the plasma chamber. The AECR-U plasma is driven by microwaves at two frequencies (14 and 10 GHz) launched through two of the three off-axis waveguides terminated at a bias plate in the injection region. Gas is bled into the source through one of the off-axis waveguides. Although its magnetic field strength is only about 60% of the highest field strength ECR ion sources in operation, the AECR-U is one of the highest performing 14 GHz ECR ion sources, especially for very high charge states. For example, a low intensity U64+ beam was produced with the AECR-U ion source and accelerated by the 88 in Cyclotron to 2.06 GeV [32]. 11.3.2

The VENUS ECR Ion Source

VENUS is a third generation superconducting ECR ion source designed to produce high current, high charge state ions for the 88 in Cyclotron at the Lawrence Berkeley National Laboratory [29, 31, 40]. VENUS also serves as the prototype ion source for the RIA (rare isotope accelerator) front end. The magnetic confinement configuration consists of three superconducting axial coils and six superconducting radial coils in a sextupole configuration. The nominal design fields of the axial magnets are 4 T at injection and 3 T at extraction; the nominal radial design field strength at the plasma chamber wall is 2 T, making VENUS the first ECR ion source to have optimum magnetic fields for 28 GHz operation. This frequency choice has several design consequences. To achieve the required magnetic confinement, superconducting magnets have to be used. The size of the superconducting magnet structure implies a relatively large plasma volume. Consequently, high power 28 GHz microwave coupling becomes necessary to achieve sufficient plasma heating power densities. Finally, the extraction of the high current, multi-species ion beam from the ion source plasma in the presence of a high magnetic field is a challenging task, and VENUS will provide an essential database for the design of future ECR high current accelerator injector systems. Figure 11.3 shows the mechanical layout of the ECR ion source. The mechanical design is described in detail elsewhere [29]. The water-cooling of the ion source is sufficient to allow continuous operation at 15 kW microwave input power. The design and development of the superconducting magnets are described in Refs. [41, 42]. The sextupole coils are wound around a core with iron in the center, which enhances the peak field by about 10%. The superconducting sextupole coils experience strong forces in the axial field of the solenoids. Therefore, a new clamping scheme utilizing liquid metal filled bladders was developed to prevent any move-

11.3 The LBNL ECR Ion Sources

Figure 11.3

Mechanical layout of the VENUS ion source and cryogenic system.

ment of the energized coils [42]. During commissioning of the superconducting magnets, the sextupole reached 110% of its design field after a few training quenches (2.4 T) with the solenoids operating at their design fields, 4 T at injection and 3 T at extraction respectively. The cryogenic system for VENUS has been designed to operate at 4.2 K with two cryocoolers each providing up to 45 W of cooling power at 50 K and 1.5 W at 4 K in a closed loop mode without further helium transfer. The two cryocoolers provide sufficient cooling power for 18 GHz operation [40]. For 28 GHz operation, the expected X-ray flux will add to the heat load of the cryostat. Therefore, we are currently constructing a cryostat extension with an additional (third) cryocooler. During 2003 VENUS was commissioned at 18 GHz and preparations for 28 GHz operation were started. Tests with various gases and metals were performed with up to 2000 W of 18 GHz RF power. Promising performance was measured in these preliminary beam tests. For example, 180 plA of O6+, 15 plA of Ar12+, 7.5 plA of Xe20+ and 6.8 plA of Bi24+ were produced in the early commissioning phase. In FY04 a 10 kW, 28 GHz gyrotron system will be added, which will enable VENUS to reach full performance. A similar 28 GHz gyrotron system was used successfully for the first time in Catania in 2000 to power the SERSE ECR ion source [33]. The waveguide system planned for VENUS takes advantage of this development. The components are expected to be delivered in December of 2003 and commissioning of VENUS at 28 GHz will begin in the Spring of 2004.

209

210

11 ECR Ion Sources

11.4

Physics and Operation of ECR Ion Sources

ECR ion sources use magnetic confinement and electron cyclotron resonance heating to produce a plasma made up of energetic electrons and relatively cold ions. The plasma electrons have two components, a cold population (~20 eV) and a hot population that has a high energy tail reaching up to 100 keV or more. Typical ion energies are a few eV. The electrons produce the high charge state ions primarily by sequential impact ionization. The ions and the electrons must be confined for sufficient time for the sequential ionization to take place. In a typical ECR ion source, the ion confinement times need to be about 10–2 s to produce high charge state ions. The ionization rate depends on the plasma density, which typically ranges from about 1011 cm–3 for low frequency sources to more than 1012 cm–3 for the highest frequency sources. Charge exchange with neutral atoms must be minimized, so operating pressures are typically 10–6 Torr or less. The plasma chamber is biased positively so that at extraction the ions can be accelerated out of the plasma and into the beam transport system. While the basic principles of operation of an ECR ion source are straightforward, the details of the plasma physics, atomic physics and electron cyclotron heating on the one hand, and the technologies needed to produce the high magnetic fields and high frequency microwave power on the other, are not simple and will be described briefly in the following sections. 11.4.1

Electron Impact Ionization

In ECR sources multiply charged ions are created mainly by step-by-step ionization, caused by the successive impact of energetic electrons. Therefore the electron impact ionization cross sections are significant parameters. Semi-empirical formulae, fitted to existing experimental results, have been proposed by Mller and Salzborn [43] and Lotz [44]. Figure 11.4 shows electron impact ionization cross sections for argon ions calculated using the Mller–Salzborn formula. Note that the cross section for ionizing argon from 15+ to 16+ is more than 3 orders of magnitude lower than the cross section for ionizing argon to 1+. The rate at which ions of charge state i are produced by electron impact ionization of ions with charge state i–1 is given by Rprod,i = ne hri–1,i vei ni

(11.1)

where ne is the electron density, hri–1,i vei, the rate coefficient, is the product of the electron impact ionization cross section from charge i–1 to i and the electron velocity averaged over the electron energy distribution, and ni–1 is the density of ions of charge state i–1.

11.4 Physics and Operation of ECR Ion Sources

Figure 11.4 Computer calculation of some argon electron impact ionization cross sections using the Mller–Salzborn formula.

11.4.2

Charge Exchange

Two processes dominate the loss of high charge state heavy ions from the plasma. These are charge exchange with neutral atoms in the plasma and loss of confinement. The cross sections for charge exchange between highly charged ions and neutrals are extremely large. An empirical formula, fitting some existing experimental data, has been proposed by Mller and Salzborn [45]. The cross section for charge exchange from initial charge state i to final charge state i–1 is given by r i;i1 ¼ 1:43 10

12 1:17

i

2:76

V0;1

2

ðcm Þ

(11.2)

where V0,1 is the first ionization potential (eV) of the neutral atom. Typical charge exchange cross sections are three to four orders of magnitude larger than corresponding electron impact ionization cross sections. Fortunately, the reaction rates are proportional to the projectile velocities, and neutral atoms are much slower than hot electrons. However, to keep the rate of production by electron impact equal to the rate of loss by charge exchange for high charge state ions, it is necessary for the neutral atom density in the plasma to be two orders of magnitude lower than the electron density.

211

212

11 ECR Ion Sources

11.4.3

Plasma Confinement

The processes that govern particle confinement in an ECR ion source are complex. The temperatures of electrons and ions are different and their confinement times vary considerably. Furthermore, the electron velocity distribution function consists of more than one population, usually described as cold, warm, and hot populations in the literature [46]. Since the confinement times are proportional to the electron temperatures, they are different for each electron population. The ion confinement time is naturally a very critical parameter in ECR sources. In ECR sources, unlike in other sources like EBIS, it is not possible to arbitrarily interrupt the ion confinement to extract the confined ions. All extracted ions have undergone a loss of confinement – the ion beam is formed from the mirror loss flux. However, if this confinement time is too short, ions do not have time to reach high charge states and if the confinement time is too long, the high charge state ions decay by charge exchange instead of being extracted as a useable beam. The following short discussion about confinement models in the ECR plasma oversimplifies the processes involved. For a more thorough discussion, we refer to some excellent papers and books [46, 47]. However, most models are based on the assumption of a quiescent plasma and one important conclusion of the available literature is that to improve the confinement of the ECR plasma, turbulence should be avoided. As a result, there is a maximum power density for each ECR source that can be coupled in and a maximum plasma density (typically lower than the cut-off frequency) that can be achieved before plasma instabilities begin to decrease the high charge state performance. These limits increase with frequency and magnetic field strength. In a simple magnetic mirror structure, particles are confined due to conservation of the magnetic moment l, l ¼

mv2? 2B

(11.3)

where v? is the velocity component transverse to the magnetic field, m is the particle mass and B the magnetic flux density. As the particle moves in the direction of increasing magnetic field strength, its transverse velocity increases so as to satisfy the conservation of magnetic moment. Thus its parallel velocity component, v||, decreases according to the conservation of energy. When v|| reaches zero the particle is reflected. However, this magnetic confinement is not perfect. If the ratio of the transverse velocity to the parallel velocity is too small the particle will be lost from the plasma. A loss cone can be defined by ! rffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffi Bmin v? 1 ¼ (11.4) ¼ sinh ¼ vk Bmax Rm crit

where Bmax is the peak mirror field, Bmin is the minimum field, Rm is the mirror ratio, and h is the loss cone angle (in velocity space) defined by this relationship.


Particles, which oscillate back and forth in the magnetic mirror, can be scattered into the loss cone by collisions. Thus the confinement time for ions is related to the scattering rate for ion–ion collisions and the confinement time for electrons is related to the scattering rate for electron–ion collisions. For particles to be magnetically confined, the rate for scattering through a large angle must be small compared to their gyro frequency in the magnetic field. High energy electrons, which have low collision rates, are well confined. The warm electrons suffer more frequent scattering collisions and are less well confined. On the other hand, the ions are highly collisional and are therefore not magnetically confined [48]. For high charge state ECR ion sources, simple axial mirrors do not provide sufficient magnetic confinement. In addition to the axial magnetic field produced by solenoids, the typical high charge state ECR source uses a sextupole (also called hexapole) or other multipole magnet to produce a radially-increasing field. The combination of the axial mirror field and the radial multipole field produces a “minimum-B” magnetic field configuration, where the magnetic field is a minimum at the center of the device and increases in every direction away from the center. Such a field provides a plasma confinement geometry that is stable against MHD instabilities. The ratio of the maximum field strength at the peak of the magnetic mirrors to the minimum field strength at the center of the device is defined as the axial mirror ratio, Rm = Bmax/Bmin, as given above. The ratio of the minimum field at the center of the plasma chamber to the maximum field at the plasma chamber wall (moving radially at midplane) defines the radial mirror ratio. While the early ECR ion sources operated with axial and radial mirror ratios with values less than two, the newer sources use mirror ratios as high as 4 at injection, 2 at extraction, and slightly greater than 2 in the radial direction. These higher mirror ratios improve the plasma confinement and this shifts the charge state distribution to a higher average charge. The minimum-B configuration also makes the ECR heating of the electrons more efficient. Since the electron impact ionization cross sections for high charge state ions is very small, a large flux of high energy electrons is needed to produce a significant density of highly stripped ions. If the electrons travel only once through the system, as in an EBIS (electron beam ion source), the power flux needed is enormous. For instance, 1 to 10 MW cm–2 would be required to produce 100 lA of Ar8+. ECR ion sources with a minimum-B structure have excellent plasma confinement properties, thus allowing the electrons to make many passes through the system before they escape from the confinement region, significantly reducing the energy dissipation. Low energy electrons have poor magnetic confinement and much higher velocities than the ions, which means that their loss rate is higher than the ions. As a result, a positive plasma potential builds up to retard the loss of thermal electrons and enhance the ion loss rate. The positive plasma potential, which has been measured to be 10 to 40 V [49], reduces the ion confinement time and equalizes the global electron and ion losses. There are several experimental techniques for adding cold electrons to the plasma and thereby lowering the plasma potential, which has several advantages. It reduces the ion loss rate and decreases the voltage gradients in the plasma, which can add to

213

214

11 ECR Ion Sources

the energy spread and longitudinal emittance of the extracted ion beam. The primary source of cold electrons in these devices is from the stepwise ionization of atoms and ions. However, cold electrons can also be supplied to the plasma from a first stage as used in SUPERMAFIOS and many first generation ECR ion sources, or by enhanced secondary electron emission from the walls [50, 51], or by an electron gun [52], or by a biased disk [53]. A variety of wall coatings have been successfully used to enhance secondary emission including ThO2 [51], SiO2 [50], and Al2O3 [54], and even special metal–dielectric layers have been tested [55]. Building the plasma chamber from aluminum, which always forms an oxide layer on the surface, is a common method for taking advantage of this effect. Measurements on the AECR showed that the plasma potential could be reduced to about half of its previous value by increasing the supply of cold electrons from wall coatings or an electron gun [49]. In the most recent ECR ion sources, a negatively biased disk located on the axis at the injection end of the plasma chamber is used in place of a first stage or an electron gun. The biased disk reduces the loss of cold electrons at the injection mirror by reflecting those electrons with energies less than the bias voltage. In addition, secondary electrons are generated by ions accelerated into the bias disk. Typical bias disk voltages are 10 to 100 V and the disk currents are from about 1 to 10 mA. For the production of high charge state ions the biased disk voltage is rather critical and should be tuned precisely. At the plasma core the hot electron density is high. The hot electrons have a low collision rate and their confinement time is much longer than the lower energy electrons outside the core. The high density of hot electrons can produce a local negative potential in the center of the trap, retaining high charge state ions and increasing their confinement time [47]. Radially, transverse to the magnetic field, particle losses occur in discrete steps roughly equal to the Lamor radii i or e. Since the electrons have a much smaller Lamor radius, this diffusion process is mainly important for the ions. The diffusion time is proportional to !2 r (11.5) sD ¼ s90 i where i is the ion Lamor radius, s90 is the 90 ion scattering time, and r is the plasma radius. 11.4.4

ECR Heating

To optimize the rate of ionization by electron impact, electron temperatures between 1 keV and 20 keV are typically needed. The ion temperature on the other hand should be as low as possible because the ion temperature is one source of emittance and energy spread of the extracted beam. Therefore, a method to selectively heat the electrons in the plasma is desirable. The use of ECRH meets this requirement. If we introduce into the plasma an electromagnetic wave, whose frequency is equal to the cyclotron frequency of the electrons in the magnetic field, an extremely efficient en-


ergy transfer occurs between the wave and the electron population. The exact nature of this energy transfer is complex, and has been discussed in numerous papers. A comprehensive discussion has given by Eldridge [56] and Girard et al. [57]. In a minimum-B configuration, the magnetic field is not uniform but increases from the center to the outside. Therefore the ECR condition is normally met only on a closed surface around the center, called the “ECR surface”. It has been found experimentally that it is essential to have the ECR surface closed inside the plasma chamber. For high performance ECR ion sources, the ECR surface sits well inside the plasma chamber wall. The propagation of electromagnetic waves in a plasma is quite complex. In a simplified picture, the plasma can be described as a high-pass filter. Microwaves with frequencies higher than a critical frequency, called the plasma frequency, can propagate while microwaves with frequencies lower than the plasma frequency are reflected. The plasma frequency, fp, is a function of the plasma density, sffiffiffiffiffiffiffiffiffiffi 2 1 ne e fp ¼ (11.6) 2p e0 me where ne is the electron density, e is the electron charge, me is the electron mass, and e0 is the permittivity of free space. For a given microwave frequency, frf, the critical density, ncrit, is defined as the density for which fp equals frf. Rewriting Eq. (11.6) in practical units gives the critical density in terms of the microwave frequency as 8 2

ncrit ¼ 1:26 10 frf

(11.7)

where now ncrit is in units of electrons cm–3 and frf is the microwave frequency in Hz. The critical density is an upper bound on the useful plasma density that can be achieved in high charge state ECR ion sources. The production of overdense plasma (ne > ncrit) has been demonstrated in some plasma physics experiments using ECRH but the electron temperature remains very low, the ratio of neutral atom density to plasma density is high, and production of high charge state ions is strongly suppressed [58]. Measurements of plasma densities in ECR ion sources have shown that the hot electron population makes up roughly 10% of the total electron density, which can be close to the critical density [59]. Therefore, the use of higher frequencies seems the only practical way to reach higher plasma densities in ECR sources. Low frequency ECR ion sources operate at 2.45 GHz where the critical density is 7.6 1010 cm–3, many sources now operate at 14.5 GHz or 18 GHz where the critical densities are 2.7 1012 and 4.1 1012 cm–3, and sources operating at 28 GHz are being developed where the critical density is 9.9 1012 cm–3. 11.4.5

Gas Mixing

It was discovered during operation of ECR sources that the production of high charge state ions can be substantially enhanced by adding a light support or mixing

215

216

11 ECR Ion Sources

gas (typically oxygen) to the ECR plasma [60]. Normally about 80% mixing gas is used and it can be up to 95% or higher for very heavy elements. However, a too high ratio of mixing gas increases the neutral pressure inside the ECR plasma and may also limit the production of higher charge states and the maximum intensity of heavier ions. Gas mixing has been extensively studied both experimentally and theoretically [48]. A widely accepted explanation of this effect is that energy is transferred in collisions between the lighter mixing gas ions and the heavier ions, which cools the heavier ions and increases their confinement time. In addition, the lighter ion lowers the average charge of the plasma, which again increases the plasma confinement time [47].

11.5

Design Considerations

In this section we will discuss some guidelines for the choice of frequency and magnetic field for high charge state ECR ion sources. Over the last three decades, a wide variety of ECR ion sources have been built and tested. These sources have operated at frequencies that range from 2.45 GHz to 28 GHz. While the axial and radial magnetic mirror ratios in the early sources were modest, in recent years it has become clear that strong confinement fields are important for optimum performance at a given frequency. Early evidence for this came from a series of ECR ion sources called CAPRICE in which the sextupole fields were increased to provide a mirror ratio of at least 2 in both the axial and radial direction [61]. Tests on superconducting ECR ion sources, where the axial and radial fields can be independently adjusted, have provided further experimental guidelines for optimum magnetic field values [28, 62]. The electron cyclotron resonance magnetic field is given by Becr ¼

f 28

(11.8)

where f is the microwave frequency in GHz and Becr is in T. The optimum fields can be written as ratios as shown in Table 11.2. In addition, to ensure suitable extraction of the beam, the field on the axis at extraction should be slightly less than the radial field at the wall. The “optimum” Table 11.2

Binj/Becr Bext/Becr Bmin/Becr Brad/Becr Bext/ Brad

Optimum magnetic field ratios for high performance ECR ion sources. ~4 ~2 ~ 0.8 2 0:9

Binj is the maximum axial field strength at the injection end of the plasma chamber, Bmin is the minimum axial field strength between the mirrors, Brad is the radial field produced by the sextupole at the plasma wall, and Bext is the axial field at extraction.

11.6 Microwave and Magnetic Field Technologies

field geometry given in Table 11.2 can be understood in a simple picture where stronger magnetic confinement is provided for the trapped electrons moving either toward injection or radially outward than for those moving toward the extractor. Since the plasma remains neutral up to the plasma meniscus at the extractor, this flow of electrons toward the extraction hole takes the ions with it and enhances the ion density at extraction. While the dependence of ECR ion source operation on frequency and magnetic field has been well demonstrated, the dependence on the size of the plasma chamber is less clear. At a minimum, the plasma chamber diameter should be at least twice the free space wavelength of the microwave operating frequency so that the plasma chamber acts as a multimode cavity [63]. Recent experiments indicate that, in high charge state ECR plasmas, the ion confinement times are due to ion diffusion and therefore should depend on the axial length and radial size of the plasma [48].

11.6

Microwave and Magnetic Field Technologies

As described above, the plasma density scales with the square of the microwave frequency and the magnetic field for confinement scale linearly with frequency. The only constraints on using very high frequencies in ECR ion sources are technical and economic. These are related to the generation of high magnetic fields and microwave power at high frequency. Table 11.3 gives the optimum magnetic field values for a range of frequencies used in ECR ion sources. The first generation ECR sources operated at between 6 and 10 GHz and the magnetic fields for these frequencies could be supplied with room temperature copper coils and permanent magnet sextupoles using samarium–cobalt. The second generation sources operated at 14 GHz and typically used copper coils with carefully shaped iron yokes to provide the axial fields and the strongest permanent magnets made with NdFe-B. Another approach is to use permanent magnets to generate both the axial and radial fields. These all permanent magnet ECR ion sources are very compact and have low power consumption [64]. The strongest permanent magnet sextupoles have about 1.4 T at the pole face and due to the thickness of the plasma chamber, the fields at Table 11.3

Critical density and optimum fields as function of microwave frequency.

Frequency (GHz)

ncrit (cm–3)

Becr (T)

Binj (T)

Brad (T)

2.45 6.45 10.0 14.5 18.0 28.0 37.0

7.56 E10 5.24 E11 1.26 E12 2.65 E12 4.08 E12 9.88 E12 1.72 E13

0.088 0.230 0.357 0.518 0.643 1.000 1.321

0.350 0.921 1.429 2.071 2.571 4.000 5.286

0.175 0.461 0.714 1.036 1.286 2.000 2.643

217

218

11 ECR Ion Sources

the inner plasma wall are typically reduced to 1.2 T or less. Applying the above guidelines, the optimum fields for 18 and 28 GHz are shown in Table 11.3. Third generation ECR ion sources with frequencies of 18 GHz or greater typically require superconducting magnets to reach the optimum fields although the 18 GHz RIKEN source uses conventional coils and permanent magnets as does PHOENIX which has been tested at 28 GHz. There are also several hybrid ECR ion sources that use superconducting solenoids to produce the axial magnetic field and permanent magnets to produce the radial field [65]. The cost and the complexity of the microwave power also increase with frequency. ECR ion sources can be powered by magnetrons, traveling wave tubes, klystrons, extended wave oscillators, or gyrotrons. For frequencies between 5 and 18 GHz and power levels above 500 W, klystrons are most commonly used. They are commercially available at 6.4, 10, 14 and 18 GHz with CW power capabilities of 2 to 3 kW. In the last few years, laboratory scale gyrotrons have become available that can produce more than 2 kW CW at frequencies above 20 GHz. The SERSE ECR in Catania was tested with a 10 kW 28 GHz gyrotron [62] and the same gyrotron was then used on the PHOENIX ECR ion source [66].

11.7

Metal Ion Beam Production

The diversity of different atomic and nuclear physics experiments performed at facilities around the world requires great variety and flexibility in the production of ion beams. High performance ECR ion sources can provide this flexibility, since they can produce beams of ions as light as hydrogen and as heavy as uranium. Several methods to produce ions from solid materials have been developed over the last 20 years. The most important techniques are: (1) evaporation from external furnaces (ovens), (2) the use of gaseous or volatile chemical compounds MIVOC, (3) sputtering, (4) direct insertion of solids into the plasma, and (5) evaporation by a laser beam [67]. In some respects all of these techniques are complementary. Figure 11.5 shows charge state distributions for metal ion beam production from various ECR ion sources using all the mentioned techniques [35, 68–72]. The selection of the best method to feed solids into ion sources depends on the temperature required to evaporate the metal, the availability of an isotope in a certain chemical form, the intensity, flexibility required, etc. For example, the LBNL ECR and the LBNL AECR-U ion sources are built with radial access. Therefore, several techniques are used simultaneously to provide the maximum flexibility for the production of metal ion beams (see Figure 11.6). The production methods at LBNL include the use of gaseous compounds, the oven technique, the direct insertion method, and the MIVOC method. The following section gives a short overview of the most important techniques.

11.7 Metal Ion Beam Production

Figure 11.5 Performance of various sources for the production of metal ion beams. The production methods used are also indicated.

Figure 11.6 Cross section through the AECR-U ion source showing all the different methods used for injection of solid feed.

11.7.1

Direct Insertion

The direct insertion technique was one of the first to be utilized for a wide variety of metals [73] and is still used by most ECR groups. A support gas such as oxygen sustains the plasma while a solid rod is positioned close to the plasma, where it is vaporized and the vapor is subsequently ionized. Though the insertion technique is simple and effective, it has the operational disadvantage of strong coupling between

219

220

11 ECR Ion Sources

plasma and sample heating. However, for some refractory metals (e.g., tantalum) it may be the only choice available. Stable operation can be achieved by carefully controlling the position of the rod. 11.7.2

Sputtering

For the sputtering technique [69, 74], a sample is mounted radially at the periphery of the plasma and is negatively biased with respect to the plasma. The sample is sputtered by the plasma ions, and neutral metal atoms diffuse into the ECR plasma where they are ionized. This technique partly decouples the ECR plasma from the evaporation process. However, adequate plasma density is required to ensure sufficient sputtering rate, and therefore the ion source tuning is not completely decoupled from the evaporation process [70]. Furthermore, the achievable intensities are dependent on the sputtering yield. Thus the sputtering method is very convenient for low intensities, but has limitations for the production of high intensity metal beams. 11.7.3

Gaseous or Volatile Compounds (MIVOC Method)

In some cases, gaseous compounds can be used, especially if the components of the compound can serve as a good mixing gas (see Section 11.4.5). Generally, one has to look for compounds with light elements. Sulfur for example can easily be produced from SO2, CS2 or H2S. Oxygen is an excellent mixing gas, therefore a compound containing oxygen is the best choice if the desired metal is heavier. If the desired metal is lighter than oxygen, hydrogen is the better choice. Carbon contaminates the plasma chamber walls and should be avoided whenever possible.

Figure 11.7 Schematic drawing of the MIVOC chamber connected to the plasma stage of an ECR ion source. (Courtesy of Hannu Koivisto, JYFL, Finland.)

11.7 Metal Ion Beam Production

The use of volatile compounds was pioneered by Koivisto and Arje [75]. Typically, metallocene materials are used for this method. The metallic compound is loaded into an external chamber and connected to the source by a leak valve. After the initial pump down of the residual gas in the MIVOC chamber, the compound vapor can be introduced into the source as an ordinary gas (see Figure 11.7, [75]). Very successfully applied MIVOC compounds for high intensity metal ion beams are ferrocene C10H10Fe or m-carborane [35, 68]. The main advantage is the fast set-up time and ease of use, but rare isotopes are often not readily available in the appropriate chemical form. Another major drawback is the impurities (in particular carbon, e.g. ten carbon atoms for one Fe atom), which contaminate the plasma chamber walls. The ion source performance and stability can be compromised, particularly for long duration, high intensity applications. 11.7.4

External Furnaces (Ovens)

Of all the methods for the production of metal ion beams, the oven technique is the least intrusive, especially if pure metals can be used. The oven technique was originally developed by Clark and Lyneis at Berkeley [76]. Later the micro-oven concept was developed by the Grenoble group [77] for Caprice type ECR ion sources. Generally, a metal vapor pressure of about 10–3 to 10–2 Torr is required inside the oven (for an oven aperture of about 3 mm diameter) to supply the right amount of atomic flux to the ECR plasma. The temperature needed to produce a particular metal ion beam can be estimated from the vapor pressure curve of the metal. Besides the temperature required to evaporate the metal, chemical compatibility of the hot liquid metal and the crucible must be considered [78]. For metals with temperature requirements less than 1600 C ceramic inserts such as zirconia, alumina or yttria can be used to prevent alloying of the heating crucible with the molten metal. For higher temperatures, ceramics begin to sublime and can even react with the hot metals. Therefore, the material has to be either loaded directly into the W or Ta furnace or special crucibles must be used. Chemical compatibility of crucibles for most elements is discussed in Ref. [78]. If the pure metal is hazardous to handle such as some alkaloids, or the isotope is not available as a pure metal, online chemical reactions can be used. For example, alkali metals can be loaded in as alkali chlorides and mixed with calcium. When heated in a low temperature oven CaCl2 is formed and the pure alkali metal is released into the plasma [79]. If the temperature required is too high, chemical compounds can be utilized to produce the desired metal vapor flux. However, careful consideration should be given when choosing the compound. Oxides, which sublime at lower temperatures, are the best choice for refractory metals since oxygen is an excellent mixing gas. For example, rhenium oxide (Re2O7) sublimes sufficiently at temperatures above 140 C, whereas for the pure metal 2790 C is needed to reach a vapor pressure of 10–3 Torr. Fluorides and chlorides typically also have relatively low working points. However, fluorine and chlorine contaminate the plasma chamber walls and should be avoided.

221

222

11 ECR Ion Sources

Figure 11.8 High temperature oven designed to be inserted axially into the VENUS ECR ion source. The Ta/W crucible is heated by electric current carried by water-cooled leads. The maximum operating temperature is 2100 C.

LBNL [71, 76] and GSI [80] have developed high temperature furnaces that are capable of temperatures up to 2100 C. As an example, Figure 11.8 shows the LBNL high temperature oven developed for VENUS, consisting of a resistively heated Ta or W furnace. 11.7.5

Efficiencies

High efficiency production of high intensity heavy ion beams is desired for various reasons. Firstly, rare isotopes like 48Ca, 36S, and 85Rb are very expensive. Secondly, reloading the oven during long experiments requires several hours of accelerator downtime. Third, it is desirable to minimize the contamination of the ion source over consecutive runs. At LBNL typical consumption rates for the radial ovens are from 100 to 200 lg h–1 lA–1, and for the axial oven typical consumption rates are 10 to 40 lg h–1 lA–1. As shown by the Dubna group [81], a hot tantalum liner covering the plasma chamber walls can be very effective in increasing the overall ion source efficiency. This technique was also tested at Berkeley [71]. The liner, which is heated by the plasma loss flux and by the microwave power, minimizes the condensation of metal vapor on the chamber wall. At a microwave power of 300 W the liner reached a temperature of 400 C, corresponding to a vapor pressure of 10–4 Torr for Ca. Typical consumption rates of 5 to 10 lg h–1 lA–1 for this set-up were measured for 48 Ca. The main disadvantage of this method is that the peak charge state for a given metal will shift to lower charge state (for example from 11+ to 9+ for 48Ca) and there is no independent control of the liner temperature. To extend this method for high temperature metals an external heated liner would be necessary.

11.8 Ion Beam Extraction from ECR Ion Sources

11.8

Ion Beam Extraction from ECR Ion Sources

Traditionally, ECR ion source extraction and analysis systems have been designed for low current, low space-charge ion beams. However, with the dramatic increase in performance over the last decade, modern sources can produce tens of mA of heavy ions, and the extracted ion beams are highly space-charge dominated. Therefore, the beam transport has to be designed as a high current injector system [82]. However, the design is yet more challenging since the ECR ion source produces a spectrum of ions. Consequently, the analyzing system has to have a mass resolution of m/Dm of 100 or better to separate the high charge state ions. A typical set-up for the analyzing section consists of a combination of a double focusing magnet and one or two solenoid lenses. Simulation of the extraction and beam transport from the ECR ion source plasmas is complicated for several reasons. First, there are multiple ion charge states present in the plasma (charge state distributions of several ion species), which are all extracted together. Typically, 8 to 30 different ions need to be taken into account to simulate the extraction system for a heavy ion beam. Secondly, the ions are extracted from a strong magnetic field, which as we will show is the dominant cause of beam emittance for ECR ion sources. Third, plasma physics effects also play an important role in the ion source extraction; these are not included in the traditional ECR ion source optics codes. Finally, the ion beam density distribution across the plasma extraction aperture may not be uniform and probably varies for different charge states, which further complicates simulation efforts. In the following section we briefly describe these phenomena. 11.8.1

Influence of Magnetic Field and Ion Temperature on the Extracted Ion Beam Emittance

Ion beam formation from a plasma is described in detail elsewhere in this book. Here we describe phenomena specific to ECR ion sources. For an ECR extraction system two main contributions to the ion beam emittance have to be considered: (1) the ion beam temperature, and (2) the induced beam rotation due to the decreasing axial magnetic field [82]. The emittance due to ion temperature can be estimated by assuming a Maxwellian temperature distribution inside the plasma [83]: sffiffiffiffiffiffiffiffiffiffiffi 0 kTi xx rmsnorm (11.9) ¼ 0:016 r eTEMP M=Q where e is the normalized x-x¢ rms emittance in p mm mrad, r is the plasma outlet hole radius in mm, kTi is the ion temperature in eV, and M/Q is the ratio of ion mass in amu to ion charge state and is dimensionless. Assuming a uniform plasma density distribution across the plasma outlet hole, the emittance due to beam rotation induced by the decreasing magnetic field in the vicinity of the extractor can be described by Busch’s theorem ([84], assuming e 100% = 5e rms, a waterbag distribution)

223

224

11 ECR Ion Sources 0

xx rmsnorm

eMAG

2

¼ 0:032 r B0

1 M=Q

(11.10)

where e is the normalized x-x¢ rms emittance in p mm mrad, r is the plasma outlet hole radius in mm, B0 is the axial magnetic field strength at the extractor in T, and M/Q is the ratio of ion mass in amu to ion charge state and is dimensionless. Beam rotation due to the decreasing magnetic field becomes the dominating contribution to the ion beam emittance when the following condition, derived by combining Eqs. (11.9) and (11.10), is satisfied: B0 r 0:5

pffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffi kTi M=Q

(11.11)

where B0 is in T, r is in mm, kTi in eV, and M/Q is dimensionless. Table 11.4 summarizes the minimum magnetic field values for which the emittance starts to be dominated by the ion beam rotation. A plasma outlet hole radius of 4 mm and an ion temperature of 0.3 eV has been assumed. It is evident that the magnetic field is the main contribution to the ion beam emittance in most ECR ion sources [82]. Magnetic field for which emittance starts to be dominated by ion beam rotation.

Table 11.4

M/Q

B0 (T)

1 5 10 30

0.07 0.15 0.22 0.38

Figure 11.9 Emittance pattern as calculated with AXCEL [85, 86] for a CSD distribution optimized on Ar16 with oxygen used as mixing gas.

11.8 Ion Beam Extraction from ECR Ion Sources

As an example the beam rotation in the strong axial magnetic field of the superconducting ECR ion source VENUS is shown in Figure 11.9 for an argon ion beam. Different charge states have different focal lengths and emittance orientations in phase space [85]. 11.8.2

Influence of Plasma Confinement on Beam Emittance

In addition to the magnetic field at the extractor, the ion confinement in the plasma and plasma stability also play important roles in determining the ion beam emittance. Experimentally it has been found that, within the charge state distribution for a particular element, the measured emittance decreases for higher charge states [87, 88]. As an example, the dependence of the normalized xx’ and yy’ emittance values on charge state is shown in detail in Figure 11.10 for a bismuth ion beam extracted from the AECR-U ion source, which was optimized for medium charge state ion production. The beam currents measured for each charge are plotted in Figure 11.11 for reference. Clearly the emittance value is predominantly dependent on the charge state and not on the current at these ion beam intensities. For instance, the ion beam emittance for an 18.8 elA beam of Bi21+ was measured to be 0.07 p mm mrad, while the emittance for an 18 elA beam of Bi32+ was 0.03 p mm mrad [88]. Figure 11.12 shows emittance values for beams of several ion species having different masses and charge states as measured after the analyzing magnet of the AECR-U injector ion source at the 88 in Cyclotron. The total extracted current was approximately equal for all the different masses (1.5 mA), the peak axial magnetic field at the extractor was ~0.9 T, and the plasma outlet hole radius was 4 mm. Also plotted in this figure is the theoretical emittance due to magnetic field in the extraction region.

Figure 11.10 Dependence of the normalized xx¢ and yy¢ emittances on the charge state for a bismuth ion beam, measured on the AECR-U.

225

226

11 ECR Ion Sources

Figure 11.11

Bismuth charge state distribution for the emittance measurements of Figure 11.10.

The emittance values show a M/Q dependence, which is contrary to the behavior expected from Eq. (11.10). The measurements indicate a strong mass dependence of the normalized emittance. For example the normalized rms emittance for O3+ (0.146 p mm mrad) is about 6 times higher than for Bi38+ (0.024 p mm mrad). This behavior cannot be explained by ion optics considerations but must be due to plasma physics and ion production processes, e.g. radial transport phenomena for different ions and electrons. The highly charged heavier ions appear to be concentrated more on the axis of the ion source, whereas the low charge state ions could also be produced in the outer shell of the ECR plasma. Therefore, an effective outlet hole radius may be defined for the higher charge states and heavier masses by comparing the measured emittance values with the theoretical emittance value computed from Eq. (11.10)

Emittance values for various ion masses and charge states as measured after the analyzing magnet of the AECR-U injector ion source at the 88 in cyclotron.

Figure 11.12

11.9 Conclusion

Figure 11.13

reff

Effective extraction radii as defined by Eq. (11.12) for different ions.

¼ ract

rffiffiffiffiffiffiffiffi eact eth

(11.12)

where reff is the effective extraction aperture (plasma outlet hole) radius; ract is the actual radius; eact is the measured emittance, and eth is the theoretical emittance value according to Eq. (11.10). Such effective radii for various ions are plotted in Figure 11.13 for the AECR-U ion beams taken from Figure 11.12. The concentration of high charge state heavy ions on the source axis is clearly visible. The larger extraction radii compared to the actual outlet hole radius for the low charge state ions are most probably caused by beam transport imperfections, e.g. aberrations or unmatched extraction conditions. The other major factor determining the ion beam emittance is the stability of the plasma. For unstable plasma conditions, the ion beam emittance can easily vary by a factor of 2 or 3 for comparable ion beam intensity [88]. This makes an emittance scanner an extremely useful ion source tuning aid and an important step in understanding the ion beam transport from the source.

11.9

Conclusion

In the 15 years that have elapsed since the first edition of this book, there has been remarkable progress in the development of high charge state ECR ion sources. This progress encompasses the major gains in the capability of ECR ion sources to produce intense, high charge state ion beams, and in the development of specialized ECR ion sources for specific applications. In addition, there have been significant improvements in our understanding of the plasma physics of ECR ion sources. We found that fitting this burgeoning field into a single chapter was a major challenge and we were forced to choose between many interesting developments and focus on

227

228

11 ECR Ion Sources

those that would, in our view, be most useful to the reader. As a result, some excellent work has gone unmentioned and some areas are only lightly covered. A more detailed description of the plasma physics of ECR sources can be found in Geller’s comprehensive book [47] and modeling of ECR plasma has been recently reviewed by Girard et al. [89]. In order to produce even more intense, higher charge state beams, higher frequency, higher magnetic field ECR ion sources are being developed. This direction presents major technical and financial challenges. Therefore, the efforts are focused at a few laboratories, which are working to develop a new generation of ECR ion sources for heavy-ion accelerators such as the LHC at CERN [90], RIA in the United States [29, 31] the new heavy ion accelerator storage ring in Lanzhou, HIRFL-CSR [91] and at RIKEN in the future [92]. One of the challenges in the coming years will be to find ways to simplify the technology so that very high frequency sources can be applied more widely. With the development of increasingly powerful ECR ion sources, research has shifted to the better understanding of the transport of the intense, multi-species and multi-charge ion beam extracted from ECR ion sources. With a better understanding of ion beam formation at the ECR extraction and the role of plasma physics in the ion beam emittance, it will be possible to further improve the beam quality and intensity from ECR ion sources. While the most direct path to ECR ion source development is to increase the frequency and magnetic field strengths, it is still possible that other approaches and refinements will lead to significant progress, just as they have over the last 15 years. This is especially true in the development of ECR ion sources for special applications such as radioactive ion beams and charge breeders. ECR ion sources remain a fascinating subject for ion source scientists and there is still room for innovation. An especially fruitful effort comes from atomic physics groups at universities, which are developing ECR to conduct basic research and continue to contribute significantly to the field [93–97]. One of the great strengths in the field is that young physicists and engineers continue to build new sources, explore new ideas and take on the challenge of developing a new generation of ECR ion sources.

References

References [1] V. Bechtold, N. Chan-Tung, S. Dousson,

R. Geller, B. Jaquot, and Y. Jongen, Nucl. Instrum. Methods 178, 305 (1980). [2] D. Hitz, F. Bourg, M. Delaunay, P. Ludwig, G. Melin, M. Pontonnier, and T. M. NGuyen, Rev. Sci. Instrum. 67, 883 (1996). [3] Y. Higurashi, T. Nakagawa, M. Kidera, T. Kageyama, T. Aihara, M. Kase and Y. Yano, Nucl. Instrum. Methods A 510, 206 (2003). [4] R. Geller, C. Tamburella, and J. L. Belmont, Rev. Sci. Instrum. 67, 1282 (1996). [5] P. Sortais, J.F. Bruandet, J.L. Bouly, N. Chauvin, J.C. Curdy, R. Geller, T. Lamy, P. Sole, and J.L. Vieux-Rochaz, Rev. Sci. Instrum. 71, 618 (2000). [6] T. Lamy, J-F. Bruandet, N. Chauvin, J.-C. Curdy, M. Fruneau, R. Geller, G. Gimond, P. Sole, J-L. Vieux-Rochas, G. Gaubert, L. Maunoury, P. Sortais, and A. C. C. Villari, Rev. Sci. Instrum. 69, 1323 (1998). [7] J. Sherman, A. Arvin, L. Hansborough, D. Hodgkins, E. Meyer, J.D. Scheider, H.V. Smith Jr., M. Stettler, J.R.R Stevens, M. Thuot, T. Zaugg, and R. Ferdinand, Rev. Sci. Instrum. 69, 1003 (1998). [8] H. Postma, Phys. Lett. A 31, 196 (1970). [9] S. Bliman, R. Geller, W. Hess, and B. Jaquot, IEEE Trans. Nucl. Sci. NS-19, 200 (1972). [10] K. Bernhardi and K. Wieseman, Plasma Phys. 14, 1073 (1972). [11] R. Bardet, P. Briand, L. Dupas, C. Gormezano, and G. Melin, Nucl. Fusion 15, 865 (1975). [12] P. Briand, R. Geller, and B. Jaquot, Nucl. Instrum. Methods 131, 407 (1975). [13] Y. Jongen, C. Pirart, G. Ryckewaert, and J. Steyart, IEEE Trans. Nucl. Sci. NS-26, 3677 (1979). [14] V. Bechtold, H.P. Ehret, L. Friedrich, J. Mllenbeck, and H. Schweikert, IEEE Trans. Nucl. Sci. NS-26, 3680 (1979). [15] H. Beuscher, H.G. Mathews, C. MayerBrcke, and J. Reich, Proceedings of the 9th International Conference on Cyclotrons and Their Applications (Les ditions de Physique, Caen, France, 1981). [16] F. Bourg, R. Geller, B. Jaquot, and M. Pontonnier, Proceedings of the 4th International Workshop on ECR Ion Sources and Related Topics

(Centre dtudes Nuclaires-Grenoble Press, Grenoble, France, 1982). [17] V. Bechtold, L. Friedrich, and H. Schweickert, Proceedings of the 9th International Conference on Cyclotrons and their Applications (Les ditions de Physique, Caen, France, 1981). [18] Y. Jongen, C. Pirate, and G. Ryckewaert, Proceedings of the 4th International Workshop on ECR Ion Source and Related Topics (Centre d’Etudes Nuclaires-Grenoble, Grenoble, 1982). [19] D.J. Clark, J.G. Kalnins, and C.M. Lyneis, IEEE Trans. Nucl. Sci. NS-30, 2719 (1983). [20] F.W. Meyer, Proceedings of the 6th International Workshop on ECR Ion Sources, Berkeley (LBNL, Berkeley, California, 1985). [21] T.A. Antaya, H.G. Blosser, L.H. Harwood, and F. Marti, Proceedings of the 6th International Workshop on ECR Ion Sources (LBNL, Berkeley, California, 1985). [22] R. Pardo, E. Minehara, F. Lynch, B. Billquist, W. Evans, B. E. Clift, and M. Waterson, Proceedings of the 7th International Workshop on ECR Ion Sources (KFA-Jlich, Jlich, Germany, 1986). [23] R. Geller, Proceedings of the 11th International Conference on Cyclotrons and Their Applications (Ionics Publishing Co., Tokyo, Japan, 1986). [24] Y. Jongen and G. Ryckewaert, IEEE Trans. Nucl. Sci. NS-30, 2685 (1983). [25] H. Beuscher, W. Krauss-Vogt, W. Brutigam, J. Reich, and P. Wucher, Proceedings of the 11th International Conference on Cyclotrons and Their Applications (Ionics Publishing, Tokyo, 1986). [26] T.A. Antaya and S. Gammino, Rev. Sci. Instrum. 65, 1723 (1994). [27] D. Hitz, G. Melin, and A. Girad, Rev. Sci. Instrum. 71, 839 (2000). [28] S. Gammino and C. Ciavola, Rev. Sci. Instrum. 71, 631 (2000). [29] M.A. Leitner, C.M. Lyneis, D.C. Wutte, C.E. Taylor, and S. R. Abbott, Phys. Scr. T 92, 171 (2001). [30] H.W. Zhao, X.Z. Zhang, Z.M. Zhang, X.H. Guo, P. Yuan, Y. Cao, L.T. Sun, B.W. Wei, Y.F. Wang, W.L. Zhan, and D.Z. Xie, Rev. Sci. Instrum. 73, 525 (2002). [31] C.M. Lyneis, D. Leitner, S.R. Abbott, R.D. Dwinell, M. Leitner, C. Silver, and

229

230

11 ECR Ion Sources

C. E. Taylor, Rev. Sci. Instrum. 75, 1389 (2004). [32] Z.Q. Xie, Rev. Sci. Instrum. 69, 625 (1998). [33] S. Gammino, G. Ciavola, L. Celona, D. Hitz, A. Girad, and G. Melin, Rev. Sci. Instrum. 72, 4090 (2001). [34] M. Schlapp, R.C. Pardo, R.C. Vondrasek, J. Szczech, P.J. Billquist, J. Vieregg, Z.Q. Xie, C.M. Lyneis, and R. Harkewicz, Rev. Sci. Instrum. 69, 631 (1998). [35] H. Koivisto, P. Heikkinen, V. Hnninen, A. Lassila, H. Leinonen, V. Nieminen, J. Pakarinen, K. Ranttila, J. Arje, and E. Liukkonen, Nucl. Instrum. Methods B 174, 379 (2001). [36] H. Koivisto, J. DeKamp, and A. Zelle, Nucl. Instrum. Methods B 174, 373 (2001). [37] S. Gammino, G. Ciavola, L. Celona, M. Castro, F. Chines, and S. Marletta, Rev. Sci. Instrum. 70, 3577 (1999). [38] D. Hitz, A. Girard, K. Serebrennikov, G. Melin, D. Cormier, J.M. Mathonnet, J. Chartier, L. Sun, J.P. Briand and M. Benhachoum, Rev. Sci. Instrum. 75, 1403 (2004). [39] T. Nakagawa, T. Aihara, Y. Higurashi, M. Kidera, M. Kase and Y. Yano, Rev. Sci. Instrum. 75, 1394 (2004). [40] D. Leitner, S. R. Abbott, R. D. Dwinell, M. Leitner, C. Taylor, and C. M. Lyneis, Proceedings of the Particle Accelerator Conference (PAC’03) (IEEE, Portland, OR, 2003). [41] C.E. Taylor, S.R. Abbott, D. Leitner, M. Leitner, and C.M. Lyneis, Adv. Cryogenic Eng. 49 and 50, (2003). [42] C.E. Taylor, S. Caspi, M. Leitner, S. Lundgren, C. Lyneis, D. Wutte, S.T. Wang, and J. Y. Chen, IEEE Trans. Appl. Superconductivity 10, 224 (2000). [43] A. Mller, E. Salzborn, R. Frodi, R. Becker, H. Klein, and H. Winter, J. Phys. B 13, 1877 (1980). [44] W. Lotz, Z. Phys. 216, 241 (1968). [45] A. Mller and E. Salzborn, Phys. Lett. A 62, 391 (1977). [46] D. Hitz, G. Melin, and A. Girad, Rev. Sci. Instrum. 71, 839 (2000). [47] R. Geller, Electron Cyclotron Resonance Ion Source and ECR Plasmas (Institute of Physics Publishing, Bristol, 1996). [48] A.G. Drentje, A. Girard, D. Hitz, and G. Melin, Rev. Sci. Instrum. 71, 623 (2000). [49] Z.Q. Xie and C. M. Lyneis, Rev. Sci. Instrum. 65, 2947 (1994).

[50] C. M. Lyneis, Proceedings of the International

Conference of ECR ion sources and Their Applications (Michigan State University, East Lansing, Michigan, 1987). [51] R. Geller, F. Bourg, P. Briand, J. Debernardi, M. Delaunay, B. Jaquot, P. Ludwig, R. Pauthenet, M. Pontonnier, and P. Sortais, Proceedings of the International Conference on ECR Ion Sources and Their Applications (MSU, East Lansing, Michigan, 1987). [52] Z.Q. Xie, C.M. Lyneis, R.S. Lam, and S. A. Lundgren, Rev. Sci. Instrum. 62, 775 (1990). [53] G. Melin, C. Baru, F. Bourg, P. Briand, J. Debernardi, M. Delaunay, R. Geller, A. Girard, K.S. Golanovanisky, D. Hitz, B. Jaquot, P. Ludwig, J.M. Mathonnet, T.K. Nguyen, L. Pin, M. Pontonnier, J.C. Rocco, and F. Zadworny, Proceedings of the 10th International Workshop on ECR Ion Sources, (ORNL, Knoxville, Tennessee, 1990). [54] T. Nakagawa, Jpn. J. Appl. Phys. 30, 930 (1991). [55] L. Schachter, S. Dobrescu, K.E. Stiebing, and J.D. Meyer, Rev. Sci. Instrum. 75, 1511 (2004). [56] O. Eldridge, Phys. Fluids 15, 676 (1972). [57] A. Girard, D. Hitz, G. Melin, and K. Serebrennikov, Rev. Sci. Instrum. 75, 1381 (2004) [58] K. Golovanivsky, Proceedings of the 12th International Workshop on ECR Ion Sources, (Institute for Nuclear Study, University of Tokyo, RIKEN, Japan, 1995). [59] A. Girard, P. Briand, G. Gaudart, J.P. Klein, F. Bourg, J. Debernardi, J.M. Mathonnet, G. Melin, and Y. Su, Rev. Sci. Instrum. 65, 1714 (1994). [60] H. Beuscher, W. Krauss-Vogt, and H.G. Mathews, Proceedings of the 5th International Workshop on ECR Ion Sources (LBNL, Berkeley, California, 1985). [61] B. Jaquot and M. Pontonnier, Proceedings of the 10th International Workshop on ECR Ion Sources (ORNL, Knoxville, Tennessee, 1990). [62] D. Hitz, A. Girard, G. Melin, S. Gammino, G. Ciavola, and L. Celona, Rev. Sci. Instrum. 73, 509 (2002). [63] C.M. Lyneis, Proceedings of the 1987 IEEE Particle Accelerator Conference (IEEE, Washington DC, 1987). [64] P. Sortais, J. Debernardi, R. Geller, P. Ludwig, and R. Pauthenet, Proceedings of the International Conference on ECR Ion Sources and

References Their Applications (MSU, East Lansing Michigan, 1987). [65] T. Nakagawa, T. Kurita, M. Imanaka, Y. Higurashi, M. Tsukada, S. M. Lee, M. Kase, and Y. Yano, Rev. Sci. Instrum. 73, 513 (2002). [66] T. Thuillier, J.L Bouly, J.C. Curdy, T. Lamy, C. Peaucelle, P. Sole, P. Sortais, J.L. VieuxRochaz, and D. Voulot, Proceedings of the 15th International Workshop on ECR Ion Sources, ECRIS’02 (JYFL, Jvskyl, Finland, 2002). [67] R. Harkewicz, Rev. Sci. Instrum. 66, 2883 (1995). [68] T. Nakagawa, J. rje, Y. Miyazawa, M. Hemmi, T. Chiba, N. Inabe, M. Kase, T. Kageyama, O. Kamigaito, M. Kidera, A. Goto, and Y. Yano, Rev. Sci. Instrum. 69, 637 (1998). [69] R.C. Vondrasek, R. Scott, and R. C. Pardo, Rev. Sci. Instrum. 75, 1532 (2004). [70] R.C. Vondrasek, personal communication, (Argonne National Laboratory, 2003). [71] D. Wutte, S. Abbott, M. Leitner, and C.M. Lyneis, Rev. Sci. Instrum. 73, 521 (2002). [72] H. Koivisto, J. rje, R. Seppl, and M. Nurmia, Nucl. Instrum. Methods B 187, 111 (2002). [73] F. Bourg, R. Geller, and B. Jacquot, Nucl. Instrum. Methods A 254, 13 (1987). [74] D.P. May, Rev. Sci. Instrum. 69, 688 (1998). [75] H. Koivisto, J. Arje, and M. Nurmia, Rev. Sci. Instrum. 69, 785 (1998). [76] D.J. Clark and C.M. Lyneis, J. Phys. 50, C1-759 (1989). [77] G. Melin, F. Bourg, P. Briand, M. Delaunay, G. Gaudart, A. Girard, D. Hitz, J.P. Klein, P. Ludwig, T.K. Nguyen, M. Pontonnier, and Y. Su, Rev. Sci. Instrum. 65, 1051 (1994). [78] K.J. Ross and B. Sonntag, Rev. Sci. Instrum. 66, 4409 (1995). [79] B. DeMarco, H. Rohner and D.S. Jin, Rev. Sci. Instrum. 70, 1967 (1999). [80] R. Lang, J. Bossler, R. Iannucci, and K. Tinschert, Proceedings of the ECR Workshop 2002, (University of Jyvaskyla, Jyvaskyla, Finland, 2002). [81] V. Kutner, S.L. Bogomolov, A.A. Efremov, A.N. Lebedev, V.N. Loginov, A.B. Yakushev, and N.Y. Yazivitsky, Rev. Sci. Instrum. 70, 860 (2000). [82] M.A. Leitner, D.C. Wutte, and C.M. Lyneis, Proceedings of the Particle Accelerator Conference (PAC’01), (American Physical Society, IEEE, Chicago, 2001).

[83] I.G. Brown, The Physics and Technology of Ion

Sources (Wiley, New York, 1989). [84] A. Septier, Focusing of Charged Particles, Vol. 2

(Academic Press, New York, 1967). [85] D. Wutte, M. Leitner, C.M. Lyneis, C.E. Tay-

lor, and Z.Q. Xie, Proceedings of the Heavy Ion Accelerator Technology Conference (HIAT’98), AIP Conference Proceedings, (American Institute of Physics, Argonne National Laboratory, Illinois, 1998). [86] P. Spaetke, AXCEL, ion optics simulation program, (INP, Junkernstr. 99, 65205 Wiesbaden, Germany; http://www.inp-dme.com/) [87] P. Sortais, L. Maunoury, A.C. Villari, R. Leroy, J. Mandin, J.Y. Pacquet, and E. Robert, Proceedings of the Proceedings of the 13th International Workshop on ECR Ion Sources, (Texas A&M University, College Station, TX, 1997). [88] D. Wutte, M. Leitner, and C.M. Lyneis, Phys. Scr. T 92, 247 (2001). [89] A. Girard, K. Serebrennikov, G. Melin, R. Vallcorba, and C. Lcot, Rev. Sci. Instrum. 73, 1146 (2002). [90] S. Gammino, G. Ciavola, L. Celona, L. And , M. Menna, L. Torrisi, D. Hitz, A. Girard, P. Seyfert, and D. Guillame, Proceedings of the 15th International Workshop on ECR Ion Sources, (JYFL, Jyvskyl, Finland, 2002). [91] H.W. Zhao, Z.M. Ahang, W. He, X.Z. Zhang, X.H. Guo, Y. Cao, P. Yuan, L.T. Sun, L. Ma, M.T. Song, W.L. Zhan, B.W. Wei, and D. Z. Xie, Rev. Sci. Instrum. 75, 1410 (2004). [92] T. Nakagawa, T. Kurita, M. Imanaka, H. Arai, M. Kidera, Y. Higurashi, S.M. Lee, M. Kase, and Y. Yano, Proceedings of the 15th International Workshop on ECR Ion Sources (JYFL, Jyvskyl, Finland, 2002). [93] V. Mironov, K. E. Stiebing, O. Hohn, L. Schmidt, H. Schmidt-Bcking, S. Runkel, A. Schemp, G. Shirkov, S. Biri, and L. Kenz, Rev. Sci. Instrum. 73, 623 (2002). [94] P. Grbling, J. Hollandt, and G. Ulm, Rev. Sci. Instrum. 73, 614 (2002). [95] M. Leitner, D. Wutte, J. Brandsttter, F. Aumayr, and H. Winter, Rev. Sci. Instrum. 65, 1091 (1994). [96] L. Muller, A. Heinen, H. W. Ortjohann, and H. J. Andra, Rev. Sci. Instrum. 73, 1140 (2002). [97] A. Heinen, M. Rther, J. Ducre, J. Leuker, J. Mrogenda, H.W. Ortjohann, E. Reckels, Ch. Vitt, and H.J. Andr, Rev. Sci. Instrum. 69, 729 (1998).

231

233

12

Laser Ion Sources Boris Sharkov

12.1

Introduction

The laser ion source is based on plasma generation by a high power laser beam focused by a mirror or lens system onto a solid, movable target. The focused laser light is used to form plasma from the target which is made of the material to be ionized. The first two detailed proposals to use laser-produced plasma as a source of ions for particle accelerator injection were made independently by Peacock and Pease [1] and by Byckovsky et al. [2], both in 1969. Neodymium glass lasers can produce highly charged ions of various materials with high kinetic energy (several keV per ion) [3, 4]. The high charge state ions were found to expand as a jet emitted perpendicular to the target surface with a narrow conical angle of 20 to 30 [5]. The first operation of a laser ion source, using a Nd-glass laser, for particle injection into a cyclotron was reported by Anan’in [6]. The application of a laser-produced plasma as a source of ions for injection into a high energy synchrotron was done at the 10 GeV synchrotron at JINR (Joint Institute for Nuclear Research) in Dubna, Russia, in 1977 [7]. Most of this development used light or medium mass ions up to chromium (Cr13+) [8]. Laser ion sources for Van-de-Graaf accelerators have been employed at the Technical University of Munich [9] and at ITEP (Institute for Theoretical and Experimental Physics) in Moscow, Russia, [10] since 1988, and a laser ion source has been constructed and investigated in great depth at the University of Arkansas [11]. The first successful attempt to match a laser ion source to the high current MAXILAC RFQ linac was done at GSI (Darmstadt) in 1994 [12]. Direct injection of carbon ions into an 80 MHz RFQ was done using a 4 J, 40 ns, CO2 laser at TIT (Tokyo) in 2002 [13]. At the present time, laser ion sources based on the use of repetitively-pulsed CO2 lasers are used routinely at the Laboratory of High Energies, JINR, Dubna, and at ITEP, Moscow. A new generation of laser ion sources has been developed and tested for the LHC (Large Hadron Collider) CERN [14], and for the ITEP-TWAC (TeraWatt ACcumulator) accelerator facility at ITEP-Moscow [15].


234

12 Laser Ion Sources

12.2

Basics of Laser Plasma Physics

Plasma electrons are heated by the intense laser radiation to temperatures up to several hundred eV. Energy absorption is by the inverse Bremsstrahlung mechanism, a classical absorption process due to the scattering of plasma electrons, accelerated in the light wave, by plasma ions [16]. The absorption coefficient (ratio of absorbed laser energy to the energy of the incident laser radiation) is determined by the electron–ion collision frequency in the underdense plasma corona – the region of the plasma where the plasma frequency, xp, is less than or equal to the frequency of the laser light, xL. The surface where xp = xL is referred to as the critical surface and the density at that surface, ncr , as the critical density. The inverse Bremsstrahlung absorption coefficient is given by Kab = mei(ncr)Lh/c ,

(12.1)

where mei(ncr) = 4(2p)1/2Ze4Keincr/3me1/2(kTe)3/2

(12.2)

is the electron–ion collision frequency, Te is the temperature of the plasma electrons, Z is the ion charge state, e and me are the charge and mass of the electron, respectively, Kei is the Coulomb logarithm (Kei » 8 – 10), ncr = xL2me/4pe2 is the critical density, c is the speed of light, Lh » vsL is the scale length of the underdense plasma region, v is the plasma velocity, and sL is the laser pulse duration. A detailed analysis of the inverse Bremsstrahlung absorption process shows that the absorption efficiency decreases as the laser intensity and wavelength increase, and increases with the duration of the laser pulse. For laser ion source designs, the laser intensity (or power density) P, wavelength kL, and laser pulse duration (sL) of interest are P ‡ 1010– 5 1013 W cm–2, kL ‡ 1000 nm, and sL ‡ 1–100 ns. Experimental results and theoretical predictions of absorption under these conditions have shown that absorption of the incident laser energy is typically between 70% and 95%. Plasma ions are stepwise ionized due to electron–ion collisions in the dense, high temperature plasma. The temperature of the plasma Te and the final ion charge state distribution depend strongly on the laser power density P [W cm–2] at the target. The ion pulse expands longitudinally due to the energy spread of ions in the plasma, resulting in a plasma pulse duration that is considerably longer than the laser pulse duration. The length of the drift space between the target and the extraction system determines the ion pulse duration; it is in this drift space that the plasma expands.

12.3 General Description

12.3

General Description

A typical or representative laser ion source (LIS) configuration is shown in Figure 12.1. It consists of a number of individual components or subsystems: 1. 2. 3. 4. 5.

A repetitively-pulsed CO2 laser of ~5–100 J output energy, capable of producing more than ~106 pulses without maintenance or adjustment. A target illumination system. A target holder and manipulator capable of delivering 105–106 pulses without maintenance or adjustment. An extraction system for forming the ions from the expanding laser-produced plasma into an energetic ion beam of typical energy 10–30 keV u–1. A low energy beam transport line (LEBT) for matching the ion beam from the LIS to the subsequent pre-accelerator.

In the following we consider these five subsystems in more detail.

CERN Laser Ion Source - 2003

POWER AMPLIFIER

MASTER OSCILLATOR TARGET AREA

GRIDDED LENSES

EXTRACTION SYSTEM

R Scrivens CERN AB/ABP

Figure 12.1

Layout of the Laser Ion Source at CERN. (Courtesy CERN).

12.3.1

Laser Characteristics

The laser-plasma scaling laws for charge state distribution, plasma density, and plasma velocity impose requirements on the minimum laser energy necessary to produce the required ion charge state and other beam parameters [17, 18]. The repetition rate of the accelerator (into which the ions are to be injected) and reliability

235

236


requirements set strong constraints on the maximum laser energy available from each active medium, due to the limits of laser technology and the basic physics involved. For these reasons, Nd:glass and ruby lasers cannot be used, nor can electron-beam-pumped CO2 or excimer lasers. A TEA (transverse excitation atmospheric) type CO2 laser has the best compromise of parameters for present requirements – delivering energy to a solid target fabricated from the material to be ionized. The pulse energy of the laser can vary from 0.1 J to 100 J, in a pulse length of ~0.02–1 ls, and with a repetition rate up to about 50 Hz. The beam divergence is typically about 2 mrad. The laser gas mixture is pumped by an electrical discharge in a direction perpendicular to the light direction. The laser gas mixture consists primarily of CO2, with N2 to keep the CO2 molecules in an excited state for a relatively long time and He to create a homogeneous discharge in the laser. The laser radiation is infrared at a wavelength of 10.6 lm. The laser pulse shape, which influences the charge state distribution of the generated plasma, depends critically on the composition ratio of the three gases [18, 19]. The resulting ion current (amount of plasma produced) is mainly a function of laser energy. Detailed measurements show that the ion current density is directly proportional to the square of the focal spot diameter df and to the laser power density P, and hence is directly proportional to laser energy: J ~ df2 P ~ EL. In contrast to lasers used for more conventional experimental investigations, the final set of parameters for the CERN and ITEP sources required considerable upgrading of the laser. Experimental results and numerical simulation indicate that only a fraction of the total laser pulse reaching the target zone and focal spot within the first 50 ns contributes to the ion yield. The laser energy within these constraints is termed the useful energy of the laser, E*. In a free-running laser, E* can be as low as 25% of the total energy. The non-useful energy contributes, however, to damage of the target, contamination of the optical elements, and degrades the target chamber vacuum. The final performance requires at least 90% of the total 100 J to be useful. Only the master oscillator-power amplifier (MOPA) operating scheme is capable of fulfilling this condition [19–21].

Figure 12.2 Typical MOPA laser optical scheme. 1 – master oscillator; 2, 11 – absorption cells; 3 – diffraction grating; 4, 5 – confocal spherical mirrors; 6 –spatial filter diaphragm; 7 – 10 plane mirrors; 12 – convex spherical mirror; 13 – focusing spherical mirror; 14 – amplifier module. (Courtesy TRINITI [20]).

12.3 General Description 160

Intensity, a.u.

140 120 100 80 60 40 20 0 0

10

20

30

40

50

60

70

80

90 100

Time, ns Figure 12.3 Typical MOPA laser pulse shape (laser energy ~100 J, 27 ns pulse FWHM). (Courtesy TRINITI [18]).

A typical MOPA laser scheme is shown in Figure 12.2, and a typical laser output pulse shape in Figure 12.3. 12.3.2

Target Illumination System

The target illumination system provides guiding and high quality focusing of the laser beam on the target surface. The target illumination optical scheme employed at ITEP and CERN [21, 22] is shown in Figure 12.4. This scheme has the following features: . .

Small angle of incidence of the laser beam to the target surface (near normal). Geometrical protection of the final focusing mirror and input window (KCl, NaCl or ZnSe for a CO2 laser) against coating by deposition of ions and neutrals produced by the laser beam-target interaction.

Figure 12.4

Target chamber with target illumination system.

Experiments with the illumination angle of incidence varied have indicated the importance of minimizing the angle between the laser beam and the target normal. Even a small 6 off-normal angle of incidence the ion generation efficiency was reduced by approximately 50–70%.

237

238


The diffraction limited focal spot size df can be estimated from df 1:22

kF ; D

(12.3)

where k is the wavelength of laser radiation (10.6 lm for CO2), D is the diameter of the laser beam, and F is the focal length of the focusing mirror. The maximum possible distance between focusing and plane mirrors (xmax) is restricted by energy density limitations for the plane mirror. For a copper surface in vacuum we can take 5 J cm–2 as the limit, and one can write h i 4Ecos45 2 (12.4) ¼ 5 J=cm 2 pdmin ðxmax Þ and dmin ðxmax Þ ¼

DðFxmax Þ F

(12.5)

where in Eq. (12.4) we have taken an angle of incidence at the plane mirror (Figure 12.4) of about 45. From these two equations one finds that dmin = 4.3 cm and xmax = 180 cm. The distance between the mirrors, x = F/2 =125 cm, can be chosen taking into account the necessity of geometrical protection of the focusing and plane mirrors against coating by ions and atoms from the laser-produced plasma. Then the diameter of the laser beam at the plane mirror will be equal to 80 mm. The energy density at the surface of the input window can be estimated as E 4E 2 2 < 0:5 J=cm S pD

(12.6)

where S is the area of the beam at the mirror. For NaCl and KCl windows the energy density limits are more than 5 J cm–2; in this case the windows can withstand more then 106 laser shots. The entrance window should be inclined to the laser beam so as to avoid reflection back into the laser system. The angle depends on the geometry and is normally 3–5. 12.3.3

Target Ensemble

The CO2 laser beam enters the target chamber through a NaCl (or KC1) window that is transparent to 10.6 lm laser radiation. The stainless steel chamber is insulated from ground to a potential of up to ~100 kV. The chamber is pumped down to »10–6 Torr after each laser shot with repetition rate 1 Hz. This requires a pumping speed of about 1000 l s–1, in turn necessitating installation of a pump system on the high voltage platform with power consumption about 5 kW. Besides the optical focusing system, the chamber houses the target mounted on a target manipulator mechanism. The target is a cylindrical tube, externally rotatable


without breaking vacuum. The dimensions of the target are determined by the number of laser pulses required before target replacement is necessary. For example, a cylindrical target of diameter 150 mm and height 200 mm allows more than 2 106 shots before replacement (assuming 10 laser pulses at each target position). Any solid material can be used for the target in the source. However, some target materials with low melting temperature can produce a coating on the surface of the focusing optical elements. The target position changing mechanism has to provide ~106 shots without replacement of the target block. The precision that is needed for target positioning is given by the Rayleigh length of the focused laser beam and the dimensions of the focal spot. In the case of diffraction limited beam divergence, the positioning precision is given by the relative aperture of the focusing objective D @ 2(F/D), where D is the largest acceptable deviation of the focal plane from the target surface. For F/D = 3 and a corresponding focal spot diameter df » 50 lm, the maximum acceptable displacement of the target is D = €100 lm. Apart from the stepping motor associated with the target position changing mechanism to expose a fresh part of the target surface, no power is required at the high voltage platform for the sources at JINR and at ITEP. 12.3.4

Pulse Width and Target-Extractor Separation

The pulse duration of the ion beam extracted from a laser ion source is determined by the drift space between the target and the extraction plane (typically 70 to 300 cm), because of the energy spread of the plasma during the expansion (of the order of 1 keV per charge-state). The ion beam pulse duration (FWHM) can be estimated from [19]: sI = 2LDv/(v2 – Dv2)

(12.7)

where L is the drift length, Dv is the characteristic half-width of the ion velocity spectrum f(v). The ratio v/2Dv remains approximately constant and equals 1.6 € 0.5 for the power density region of interest. The experimental scalings obtained in this way are v [cm s–1] = 87P0.42 [W cm–2],

(12.8)

sI [ls] = 2 106 P–0.43 L [W cm–2 m–1].

(12.9)

Since the target irradiation conditions are fixed and there are no additional external fields in the space between the target and extraction electrodes, the ion pulse length is proportional to the distance L between the target and the first extraction electrode: sI ~ L. On the other hand the number of ions N and the current I extracted through a given aperture both decrease steeply with increasing L: N ~ 1/L2 and I ~ 1/L3. Hence the ion pulse duration can be adjusted by varying the distance

239

240


between the target and the extraction system, although limitations on this distance exist: The minimum distance is limited by the maximum plasma density that one can tolerate in the extraction gap without gap breakdown. The maximum distance is limited by the current level required, since the extracted current density decreases with the square of the distance (following the solid angle from the extraction hole). The combination of the required ion beam current and pulse width at the extraction aperture of diameter D determines the drift distance L. Usually L is 100–300 cm long. This distance is large enough to locate optics with focal length F < L. The updated design of the target chamber (shown schematically in Figure 12.4) employs a self-consistent target illumination scheme [21] developed for an interrelated set of parameters: . . . . . . .

Laser beam diameter D = 160 mm. Focal length of the focusing mirror F = 230–260 mm. Diameter of the focusing mirror O = 220 mm. Distance to the plane mirror l = F/2. Diffraction spot diameter d = 1.22kF/D > 100 lm. Laser energy density limitation for plane mirror < 5 J cm–2. Maximum illumination angle < 5.

The target is a rotatable cylinder with a surface area over 400 cm2, provided with fine adjustment (€100 lm) in two axes, with a lifetime of at least 106 shots before replacement is needed. This will allow several days of uninterrupted operation at 1 Hz repetition rate. Since the whole ensemble is designed to be modular, the length of the plasma expansion drift tube (which determines the ion beam pulse duration [19, 21]) can be changed from about 1.5 m to about 3 m. 12.3.5

Extraction System

The plasma expands through the drift region, free of external fields, before the ions are extracted. This is to allow the plasma density to fall to a value at which ions can be extracted with voltages of about 100 kV. The extraction system used consists of a three-electrode acceleration–deceleration arrangement often used in conventional ion sources [23]. The plasma is formed at the positive source potential and the ions are accelerated by a negative extraction electrode. The beam is then decelerated by the final grounded electrode of the extraction system. The negative potential of the extraction electrode is used to prevent the acceleration of secondary electrons back into the source, as these electrons provide some compensation for the space charge of the accelerated beam; i.e., this electrode is a “suppressor electrode”. The negative voltage also allows variation of the extraction voltage at fixed source potential. Moreover, the experience gained at CERN shows that without a negative electrode it is not possible to avoid extractor breakdown during the ion pulse. To ensure extraction of all of the ions in the high charge state peak, it is important to choose a sufficiently high extraction voltage to match the ion current to the extrac-


tion system (optimized extraction). Before extraction the plasma consists of multiply charged ions with a current density ~10 mA cm–2 [14]. The required extraction potential can be estimated from the Child–Langmuir law, 3=2

J ¼

4 pffiffiffiffiffiffiffiffiffiffiffi Ud e 2q=m 2 , 9 0 d

(12.10)

where J is the current density, Ud is the extractor gap voltage, d is the gap spacing, and q and m are the ion charge and mass, respectively. For example, for Ta20+ ions and a gap spacing of 30 mm, an extraction voltage Ud ~30 kV is required for optimal extraction. The current density can be increased by a factor of 1.25–2 when the electrodes have two circular apertures for beam extraction [24], and the initial ion velocity of the plasma stream (approximately 105 m s–1) leads to a further enhancement by a factor of 1.4–1.6 for a streaming energy of 2 keV per charge at an extraction potential of 60 kV. However, the ion extraction optics is affected by the electric field distribution and the shape of the plasma-beam boundary (the plasma meniscus). These effects have been studied for quiescent plasma sources [25] but not for a source with a high ion streaming velocity. These considerations lead to a higher electric field requirement in the extractor gap. With so many factors affecting the extraction of ions from a LIS, it is necessary to optimize the process by simulation and experiment. The potential of the high voltage side of the source is fixed (60–100 kV) by the injection energy requirements of the RFQ pre-accelerator. Since the emittance of the extracted ion beam is strongly dependent on the geometry of the extraction system, simulation becomes very important for minimizing the emittance. At present the best code for this purpose is the 3D KOBRA3-INP code, which shows good agreement with experimental results [26]. The results obtained show that an rms emittance (87%) of about 200–250 p mm mrad can be achieved under optimal conditions. Optimization of the extraction system geometry has been done for typical average currents of ~100 mA that have been extracted from the LIS test-stand at CERN [14, 27, 28] In order to meet the goal of 9 mA of Pb25+ beams (single charge state only), an aperture of 24.4 mm is needed for the plasma electrode, resulting in 92 mA total current. For higher current, a solution is found with a 30 mm aperture plasma electrode resulting in 13.6 mA of Pb25+. The simulation code IGUN [29] was used to optimize the geometry of the extraction system in 2D. Of particular interest is the gap between the plasma electrode and the suppressor electrode, and the thickness and angle of the plasma electrode. The plasma input conditions were fixed in the simulations as follows: . . .

Ion current (over 24 mm): 92 mA. Electron temperature: 10 eV. Ion streaming energy: 2.0 keV per charge.

The geometry optimized by IGUN was then input into KOBRA3 for a 3D simulation of the extraction. Using a uniformly distributed ion current streaming in the

241

242


Figure 12.5 (a) Projection of ion trajectories as modeled by KOBRA; (b) emittance diagram. (Courtesy CERN).

plasma as a starting condition, the beam profile and the emittance diagram after extraction were calculated as shown in Figure 12.5. A non-uniform distribution of ion current density at the input of the simulation was also simulated. The values for non-uniform input distributions are the best estimates of the source output emittance. The resulting emittance values are summarized in Table 12.1. The data confirm an appreciable gain in emittance by working with smaller apertures, and hence lower currents, which are still within the design specification of the LIS. Table 12.1

Simulated emittances for different electrode apertures and ion density distribution.

Aperture (mm)

Current Distribution

Emittance (mm mrad)

24

Uniform Non-uniform Uniform Non-uniform

57 70 69 310

30

12.3.6

Low Energy Beam Transport Line (LEBT)

The LEBT plays an important role as a matching section to the subsequent accelerator, normally a radio frequency quadrupole (RFQ) accelerator. It is a key element of the LIS in that it must work properly with a space-charge dominated ion beam; the total current of the ion beam extracted from the laser-produced plasma can easily reach 50–100 mA. The RFQ requires a small diameter beam at the input (normally < 10 mm), and hence the ion beam emerging from the 30 mm aperture extraction electrodes must be tightly focused and matched. At the output of the laser ion source, a high current of highly charged ions with a large range of charge states is available. Focusing of this beam by magnetic elements


4

3

2

BEAM

1

400mm

5

60mm Figure 12.6 Drawing of the gridded electrostatic line (GEL). 1– Ceramic isolator; 2 – high voltage electrode; 3 – grid; 4 – ground electrode; 5 – HV feedthrough mounting port.

can cause a non-linear space-charge field to develop which can induce large aberrations and emittance growth in the beam. The CERN LIS has used both two solenoids and a gridded electrostatic line (GEL) to focus the high current, multiply charged beam into the RFQ accelerator. The GEL uses a set of high voltage electrodes (up to 40 kV) separated by grids as shown in Figure 12.6 [14, 30]. A single solenoid line was also investigated. Each LEBT is characterized by the maximum transmitted current into a 6.5 mm aperture Faraday cup. A comparison of the three different configurations is shown in Table 12.2, where it can be seen that the highest measured transmission is available for the GEL, which is well predicted by simulations. Comparison of the transmitted current from the LIS to a small aperture Faraday cup, for three different focusing line configurations. Source output was 60–70 mA with 7–8 mA of Ta20+.

Table 12.2

Total current (mA) Ta20+ current (mA) Transmission of Ta20+

Two Solenoids

One Solenoid

GEL

10 2 26%

17 3.3 42%

33 4 51%

For full simulation of the ion optics in 3D, the commercial code KOBRA3 [26] was used; the code solves the ray trajectories in a self-consistent manner with the Poisson and Lorenz equations in a 3D geometry which can include magnetic fields. It was supposed that the beam from the laser ion source had an energy spread DW/W = 2.5%. It has been demonstrated by simulation, and experimentally observed, that a significant aberration of the beam quality can occur due to the non-linear space-charge

243

244


electric field that develops in the beam because of the different focusing strengths of a magnetic lens for different ion charge states in the beam. This aberration does not exist when electrostatic focusing is used.

12.4

Beam Parameters 12.4.1

Current Profile

The laser produced plasma delivers ions to the extraction region with a charge state distribution that changes with time. Normally the higher charge states have higher velocities. As a result the charge states of interest are present for only a few ls during the first spike of the ion pulse. Typical Pb ion pulses generated by a 95 J CO2 laser are shown in Figure 12.7. A typical early-time ion beam current pulse shape is shown in Figure 12.8. The 3.5 ls window shown corresponds to the time slot with major abundance of Pb27+

Figure 12.7 Average Pb ion beam current density, showing also the pulse-to-pulse statistical spread. P ~ 3 1010 W cm–2. (Courtesy TRINITI).

Figure 12.8 Example of early-time, high-charge-state phase, pulse shape of a Pb ion beam current pulse. P ~ 3 1013 W cm–2. (Courtesy CERN).

12.4 Beam Parameters

Figure 12.9 Typical current outputs from the LIS-RFQ, using the GEL at input and accelerating the Ta20+ beam component from the laser ion source [31].

ions. Finally, the ion pulse shape for Ta20+ ions separated in an RFQ accelerating Ta ions to 100 keV u–1 is shown in Figure 12.9. Two pulses are shown. The first is measured directly after the RFQ and the second after a bending magnet. Both measurements used a Faraday cup [31]. 12.4.2

Charge State Distribution

The charge state distribution of the extracted ion beam is determined by the laser power density P. The power density should be chosen according to the ionization potential of the principal charge state required from the LIS. For a given P the plasma consists of a mixture of charge states in the vicinity of the principal charge state, as shown in Figure 12.10 for the case P » 3 1010 W cm–2. An example of the charge state distribution for a lead beam formed at a laser power density at

Figure 12.10 Measured ion charge state distribution of a Pb beam for ion analyzer registration energy setting E = 300Z eV. The laser power density on the target was P » 3 1010 W cm–2.

245

246


Charge state distribution of a lead ion beam generated by a CO2 laser at a power density P » 3 1013 W cm–2 [26].

Figure 12.11

P » 3 1013 W cm–2 is shown in Figure 12.11 for the time window during which Pb27+ ions are dominant. Typically the principal ion charge state does not exceed 20% of the total beam current. 12.4.3

Beam Emittance

The charge states of interest are only present for a few ls during the first spike of the extracted ion pulse. Thus, to measure the emittance, it is necessary to use a device which can be activated for that specific part of the beam pulse only. Due to variation from pulse to pulse and also because of the low repetition rate, it is desirable to perform the measurement during a single pulse. Two types of experimental diagnostic devices have been used for time-resolved emittance measurements of LIS ion beams in different laboratories – a pepper-pot system and a system based on an array of slits. At CERN, beamlets which pass through 80 lm holes of a pepper-pot are detected after a drift of 145 mm by means of a gated microchannel plate (MCP) and phosphor screen assembly which turns the spatial pattern of the arriving beamlets into a light pattern on the screen. This pattern is then imaged onto the input plane of an image intensified CCD camera and recorded by computer [27]. The phase space plot is obtained in an 800 ns time window at the peak of the 60 mA high charge state pulse. The measured emittance is 100 mm mrad for 90% of the beam current at an extraction voltage of 50 kV. Measurements made at two different times and thus at different currents during the first pulse show phase space ellipses; the emittance remains constant but the shape of the ellipse changes. At ITEP, an array of 100 lm slits with 3 mm spacing and a CsI crystal as scintillator and a gated PCO camera-based read-out system are used for measurement of the angular distribution of a Pb ion beam [32] at each of the slit points. Respective emittance measurements for a lead ion beam of about 10 mA total current and ion pulse length of 75 ls result in about 400 p mm mrad (rms) as shown in Figure 12.12.


Figure 12.12 Phase-space ellipse for a Pb ion beam. Extraction voltage 50 kV, middle electrode voltage –10 kV, PCO camera gate width 10 ls, gate opening time: 40 ls (circles), 60 ls (squares) and 80 ls (triangles).

12.4.4

Pulse Stability and Source Lifetime

Here, by pulse stability we mean the pulse-to-pulse repeatability in pulse shape and energy. The required pulse stability of the beam current is mainly determined by the stability of the laser impinging the target. The stability typically demonstrated by routine LIS operation at JINR (Dubna) and at ITEP (Moscow) is about –20% (referred to pulse amplitude) over long term runs of about 105 shots. Stability test runs have been undertaken at CERN for a 1 Hz repetition rate, 95 J laser in MOPA configuration. The best value achieved was 1 h 15 min, non-stop 1 Hz operation [22]. Shot-to-shot statistical variation in the output laser pulse energy was less than –15%. Upgrading of the prototype, using more radiation resistant optical materials and lowering the energy density at optical surfaces, will lead to substantial improvement in lifetime. The number of pulses, N, required from the source without any maintenance or any adjustment depends on the synchrotron ring filling scenario. For the scenario described in [33], 624 pulses would be needed from the source to fill both LHC rings once. On the basis of four fillings per day, and with N = 105 shots, 40 days LHC operation could be covered. However, in non-stop operation as needed for machine development, 105 shots would allow just 1 day, 9 h of operation. This explains the necessity for high reliability. For an experimental campaign lasting two weeks on the TWAC facility (ITEP), about 5 105 LIS non-stop operation cycles are required. The overall lifetime of the source is mainly limited by the optical elements of the laser system.

247

248


12.5

Sources at Accelerators 12.5.1

The LIS at ITEP-TWAC

Modification and upgrade of the existing heavy ion accelerator chain at ITEP, for the production of TeraWatt power level (100 kJ, 100 ns) of intense ion beams, is in progress at the ITEP TWAC project [15, 21, 34]. The ITEP laser ion source aims to produce high current pulses (5 mA, 5 ls) of highly charged (He-like preferably) ions of medium atomic mass elements (Co, Cu, Zn) for single turn injection into a booster synchrotron ring. The basic idea of the project is the accumulation of nuclei of elements as heavy as possible using non-Liouvillean injection into the accumulator ring. A foil stripper will be used to convert highly charged ions to fully-stripped nuclei. Ions with very high charge states, in the range C-like to He-like ions for various elements, have to be used to minimize loss of ions during the non–Liouvillean injection into the accumulator ring. As the accumulation time is limited, the ion beam current at the exit of the source needs to be about 10 mA. According to the ITEP-TWAC acceleration– accumulation scenario, the required parameters of the LIS are the following: . . . . . . .

Element: as heavy as possible. Ion charge state: in the range C-like to He-like ions. Ion pulse length (for 95% of ions with desirable charge state): 10–15 ls. Number of ions with the required charge state: ~ 5 1010 ions pulse–1. Emittance of extracted beam: < 500 p mm mrad. Repetition rate: 1 Hz. Number of source operation cycles between maintenance periods: >105.

The LIS will be used for the ITEP accelerator–accumulator complex in two stages. In the first stage of the project the existing 5 J, 0.5 Hz repetition rate CO2 laser will generate an intense beam of C4+ ions to prove the project principles. In the second stage, a 100 J, 1 Hz CO2 laser will be built and used as a driver for the LIS, generating intense beams of highly charged ions (Z/A ~ 0.3–0.4) with atomic masses of up to 60 [34, 35]. Carbon C4+ ions (charge-to-mass ratio 1/3) are being used in a “run-in” phase of the TWAC facility for testing complicated beam gymnastics in the accelerator chain [12, 32, 35]. The ion beam current density was measured by a Faraday cup placed just behind the extraction system. The drift region distance from the target to the extraction system was 130 cm. The peak value of ion current density reaches 8–9 mA cm–2. The charge state distribution of the ion beam was measured at the outlet of the bending magnet. The most abundant charge state is the He-like ion C4+, with 35% of the total number of ions in the beam. Time-resolved ion beam emittance measurements were made for different geometries of the three-electrode accel–decel extraction system. Phase space ellipses of

12.5 Sources at Accelerators

the charge integrated (containing all charge states) ion beam for an extraction voltage of 50 kV and an extraction aperture of 20 mm were measured. An emittance of » 300 p mm mrad was obtained at 4 rms. The emittance values do not change significantly with time during the ion pulse. The yield of highly charged ions of light and medium masses was investigated using a CO2 laser system with output energy ~75 J in a 16 ns laser pulse FWHM. Beams of highly charged ions such as F7+...9+, Mg8+...10+, Al9+...11+, Ca12+...15+, and Ti14+...16+ have been extracted with the total number of particles in a given charge state ~1010–1011. Ion pulse widths for various charge states containing 95% of each charge state are in the range 3.5–4.5 ls for highly charged Ti ions at a distance 308 cm from the target. These results are necessary for the selection of the optimal species and charge state for maximum power of the accumulated beam for the ITEP-TWAC project. A LEBT containing three Einzel lenses and a buncher (2.5 MHz, 10 kV) has been used to match the LIS C4+ ions to the heavy ion injector I–3 (2.5 MHz, 2 MV/gap) shown in Figure 12.13 [36]. A C4+ ion beam was chosen for the first step as the most intense and stable beam generated by a LIS with a 5 J CO2 laser. A three-aperture accel–decel system is used to extract the ion beam at a voltage of 50 kV. Grids with a transparency of about 95% are installed inside the middle electrode of all Einzel lenses to reduce aberration. The buncher consists of two grounded apertures and a middle tube operating at 10 kV and 2.5 MHz. The total length of the LEBT is 225 cm.

Figure 12.13

The ITEP-TWAC LIS injection line. (Courtesy ITEP).

Different extraction system geometries have been tested to obtain as high as possible yield of C4+ ions at the exit of the I-3 injector [37]. First electrode apertures of diameter 20 mm, 40 mm and 80 mm were used. A 90% transparent grid was placed in the first electrode to stabilize plasma boundary variations in time. The highest yield of C4+ ions at the injector exit was found for the largest extraction aperture, 80 mm, and using the first electrode grid. The carbon ion beam peak total current, measured by a current transformer at the inlet of the I–3 Injector, reaches about 90 mA. The LIS was routinely operated with a rep-rate of 0.25 Hz to inject C4+ ions

249

250


into the ITEP–TWAC ion accumulator facility for more than a year. The focal length of the focusing mirror used for the 5 J, CO2 laser was short, F = 135 mm; replacement of the focusing mirror was required after 3–4 months of operation. Significant extension of the maintenance-free LIS operational period was achieved by using a CO2 laser gas mixture regeneration circuit. A palladium catalyst was used for gas mixture regeneration. Refilling of gas mixture was required once in 3 months. 12.5.2

The LIS at CERN

For the highest luminosity in hadron colliders, the injector chain is best served by a source providing a high brightness beam with a pulse length similar to that of the revolution time in the first synchrotron. In this way, multi-turn injection and the associated emittance increase is avoided. At CERN there is a demand for high currents of highly charged ions, presently, for fixed target experiments at the SPS and, in the future, for the LHC project. The LHC, however, requires an intensity that is significantly greater than the intensity presently achieved by the ECR ion source. Laser ion sources have the potential to fulfill the LHC requirements. The objective of the CERN laser ion source [14, 27] is to produce high current pulses (5 mA, 5 ls) of highly charged heavy elements (Pb25+, In, Nb) for single-turn injection into a synchrotron (the PS Booster). The CERN accelerator complex runs with a repetition

CERN - High Current Laser Ion Source CO2 Oscillator

CO2 Amplifier

100 J, 1Hz, 15ns

150 mJ, 3Hz Bldg 363

Emittance Current Charge State Distribution Measurements

Target

Expanding plasma

Current Emittance

Current

Extraction system 120 kV

LEBT : Grided Electrostatic Lenses

Double aperture simulating the RFQ acceptance

Bldg 236 Pierrick Fournier-2001

Figure 12.14

Layout of the laser ion source experiment at CERN. (Courtesy CERN).

12.5 Sources at Accelerators

rate of 0.8 Hz and the LIS should run with at least the same repetition rate. Lower ion current with higher repetition rate would require an intermediate accelerator for storage and electron cooling (LEIR project [38]). Figure 12.14 shows the general layout of the laser ion source experiment at CERN. The new 1 Hz rep-rate ~100 J, ~25 ns CO2 laser is a key element of the CERN LIS. The chamber containing the target is connected to a variable high voltage power supply (maximum voltage 120 kV) and is isolated from the main vacuum chamber by a ceramic insulator. The laser beam enters the target chamber through a NaCl window and is focused onto the target by a parabolic copper mirror of focal length 130 cm. The plasma is formed on the target surface, and drifts into the expansion region which consists of a system of telescopic tubes directly connected to the target chamber. This creates a region free of external fields for the expansion of the plasma. At the end of the expansion region the ions are separated from the electrons and accelerated by the electric field in the extraction assembly. The ion beam, which contains a variety of charge states, then enters a LEBT for transverse matching to an RFQ. A project to investigate the feasibility of the application of laser ion sources was started in 1990 and a source capable of producing high currents of highly charged ions was constructed. A joint CERN-ITEP-TRINITI project was carried out for the design and construction of a new, unique CO2 laser. Construction of the LIS has been completed and first experimental investigations of the extracted beam were performed in March 2003. A master oscillator and power amplifier laser chain was chosen as the basic scheme for generation of the laser pulse with parameters required for the laser ion source. The optical scheme was built on the basis of an existing hybrid CO2 laser with a TEA module (master oscillator) of active volume 1.7 1.7 45 cm3 and power amplifier of 15 15 100 cm3 active volume. The four-pass amplifier optical scheme has been assembled and tested. The gas mixture is CO2:N2:He = (1.5:1:7.5). The following parameters have been achieved. The output laser pulse energy is 90–100 J, and the pulse duration is 25 –30 ns. Shotto-shot statistical fluctuation in the laser pulse amplitude and the laser pulse duration are = –14% and –9%, respectively. The master oscillator output pulse was of energy 150 mJ and pulse duration ~75 ns. The results obtained for a laser energy 90–100 J, pulse-length FWHM 26–27 ns, and repetition rate 0.25 Hz, are scaled to a distance of 1.765 m and summarized in Table 12.3 for a total Pb ion current including charge states 18 < Z < 33, with 16% content of Pb27+, with an extraction aperture d = 24.4 mm and extraction voltage U = 100 kV. A normalized emittance e(rms) = 0.2 mm mrad has been measured at a current density of 4.3 mA cm–2. These results provide information about the laser ion source capabilities that are required for LIS application in accelerator laboratories. Further technical improvements will significantly improve the laser reliability and will encourage wide usage of LIS for different applications.

251

252

12 Laser Ion Sources Table 12.3

Number of ions scaled for extraction distance 1.765 m.

Distance Current Total current Pb27+ current Pulse length Number of ions

1.765 10.71 50.1 7.87 3.50 0.64 1010

m MA cm–2 mA mA ls

12.5.3

The LIS at JINR Dubna

Since the middle of the 1970s a LIS has been in operation at JINR-Dubna [39], producing a number of light ion species, from fully stripped Li3+ ions to He-like Mg10+ and Si12+ ions, for injection into the Dubna synchrophasotron and more recently into the Nuclotron [40]. After pre-acceleration up to 5 MeV u–1 in the LU-20 linac, He-like ions are fully stripped. A two-module TEA CO2 laser with ~1 Hz rep-rate and output energy of several Joules has been successfully used for more than 20 years, but a growing demand for increased ion beam current has led to the decision to upgrade the laser to an output energy of ~20 J, but still in free-running laser generator configuration and with refracting focusing optics.

12.6

Other Operating Options 12.6.1

High Current, Low Charge State Mode

There is a need for an ion source of heavy elements of relatively low charge state, e.g., U4+, Pb4+, Ta4+ etc., but with rather long pulse duration, 60–80 ls. A long pulse corresponds to the multi-turn injection mode of a synchrotron ring. This mode of source operation can be achieved using a nitrogen-rich laser gas mixture for a freerunning laser (for instance CO2:N2:He = 1:1:3) and/or by defocusing the laser beam at the target surface, i.e., by decreasing the laser power density P. The power density level required for efficient production of low charge state ions is typically ~1010 W cm–2. Experiments to optimize the Pb4+ ion yield from laser-produced plasma by variation of focal spot size and laser pulse duration have been carried out using a ~100 J CO2 laser with laser power density on the target surface in the range (1–9) 1010 W cm–2. The emittance of the lead ion beam extracted from the laser produced plasma was measured as a function of time with plasma parameters (electron temperature, ion velocity, and ion charge states) optimized for Pb4+ production. Simulations of an existing GSI matching channel between source and RFQ have

12.6 Other Operating Options

been carried out to provide information about transport of the LIS ion beam. The main results are: .

.

.

.

.

.

.

.

.

The highest yield of Pb4+ ions was obtained for a laser power density on the target surface in the range (3–4) 1010 W cm–2 for a laser pulse duration 15 ns. The yield of Pb4+ ions for a laser pulse duration of 15 ns (optimized conditions) is greater than for a laser pulse duration of 40 ns, over the range of target power densities explored, (3–4) 1010 W cm–2. A Pb4+ ion current of 10 mA with pulse length 80 ls was obtained for optimized conditions. This corresponds to a total number of Pb4+ ions in the pulse of about 1012. Numerical simulations show that further decrease in the laser pulse duration should lead to increased yield of Pb4+ ions. (Note, however, that the hydrogen-like atomic model for low charge state heavy elements is not reliable). The yield of lead ions was measured as a function of angle under optimized laser irradiation conditions. It was found that the yield is practically constant (within 15%) over a range of angles –2 from normal to the target surface. The yield of lead ions decreases with increasing charge state over the range of laser power densities investigated. The yield of Pb7+–Pb10+ ions is about an order less than the yield of Pb4+ ions and is in the range 1010–1011 ions per pulse, calculated for a distance of 3 m from the target and 3.4 cm diameter extraction aperture. The momentum spread Dp/p of the different charge states of lead ions was measured in the expanding plasma. Using the data obtained, the momentum spread of Pb4+ ions in the beam extracted by DC potential 50 kV can be estimated as (Dp/p)b @ 0.2. This value can be decreased to a few percent by ramping the extraction voltage. The emittance of the lead ion beam extracted under conditions optimized for the Pb4+ yield was measured as a function of time for 30 kV and 50 kV extraction voltage. The emittance of a beam with total current about 10 mA and ion pulse length 75 ls is about 400 p mm mrad (for about 75% of ions in the beam). The geometry and main dimensions of the LIS for generation of a 10 mA, 80 ls beam of Pb4+ ions can be specified.

12.6.2

Influence of Magnetic Field on the Laser Ion Source Plasma

Investigations at the U-200 cyclotron at JINR Dubna and at the Moscow Engineering Physics Institute have shown some specific aspects of the influence of longitudinal and transverse magnetic fields on the laser ion source plasma [8, 41]. A longitudinal magnetic field mainly confines the plasma, narrowing the expansion angle of the plasma plume, and forming a cylindrical plasma along the field axis with increasing magnetic field. Increased pulse length is a result of particle

253

254


rotation and collective effects in this plasma cylinder. The ion charge states remained the same, for power densities P £ 1010 W cm–2 and a field length of about 10 cm. A transverse magnetic field has less influence on the plasma expansion angle but can lead to instability of the expanding plasma for higher power densities and longer plasma plume drift space. The plasma expansion rate is decreased by the transverse magnetic field and in this way changes the ions energy spectrum. The transverse magnetic field does not lead to a decrease in the maximum ion charge state, e.g., Si8+, V11+, Zr13+, In11+, Ta9+, and Bi8+ [8]. A more detailed description of the influence of longitudinal and transverse magnetic fields on the laser ion source plasma can be found in Ref. [41].

12.7

Conclusion

The laser ion source is well suited for the generation of intense, high brightness, pulsed ion beams from any solid material, with high, medium and low ion charge states, at moderate cost, and with low power consumption of the high voltage platform. The high intensity, short pulse length (~10 ls), and pulse repetition rate (~1–10 Hz) are very well matched to the requirements needed for filling of synchrotron rings in single turn injection mode. The operational parameters of the source are summarized in Table 12.4. Table 12.4

Summary of LIS operational and beam parameters.

Duty cycle Lifetime Repetition rate Laser power density Laser radiation wavelength (typical) Laser pulse length Power consumption Gas consumption Ion current Ions from solid material Charge state distribution Ion extractor aperture diameter Extraction voltage Emittance (4 rms)(current dependant) Pulse shape

10–3 % > 106 pulses £ 10 Hz 1010 – 1013 W cm–2 10.6 lm; 1.06 lm 20 ns–1 ls 0.5– 5 kW 10 l min–1 0.1–20 mA any solid material 1+ to 30+ 5–30 mm 50–100 kV 50–300 p mm mrad quasi-gaussian

References

References [1] N.J. Peacock and R.S. Pease, Br. J. Appl. Phys. [17] I. Roudskoy, Laser Particle Beams 14, 369

(J. Phys. D), Ser. 2, 2, 1705 (1969). [2] Yu.A. Byckovsky, V.F. Eliseev, Y.P. Kozyrev

and S.M. Silnov, Sov. Patent No. 324938, 1969. [3] Yu.A. Byckovsky, N.N. Degtiarenko, V.F. Eliseev, Y.P. Kozyrev and S.M. Silnov, Sov. Phys. JETP 33, 706 (1971). [4] G.F. Tonon, IEEE Trans. Nucl. Sci. 19, 172 (1972). [5] O.B. Anan’in, Yu.A. Byckovsky, V.F. Eliseev, Y.P. Kozyrev and S.M. Silnov, Sov. Phys. JETP Lett. 16(10), 543 (1972). [6] O.B. Anan’in, N.N. Degtiarenko, V.F. Eliseev, Y.P. Kozyrev and S.M. Silnov, Sov. Phys. JETP Lett. 17(69), 460 (1973). [7] O.B. Anan’in, Yu.A. Byckovsky Y.P. Kozyrev, B.Yu. Sharkov and S.M. Silnov, Sov. J. Quantum Electron. 7, 873 (1977). [8] V.B. Kutner, Rev. Sci. Instrum. 63, 2835 (1992). [9] G. Korshinek and J. Sellmair, Nucl. Instrum. Methods A 268, 473 (1988). [10] L.Z. Barabash, A.A. Golubev, B.Y. Sharkov and A.V. Shumshurov, Sov. J. Atomic Energy 6(4), 395 (1988). [11] R.H. Hughes and R.J. Anderson, in The Physics and Technology of Ion Sources, edited by I.G. Brown (Wiley, New York, 1989), p. 299. [12] V. Dubenkov, B. Sharkov. A. Golubev, A. Shumshurov, O. Shamaev, I. Roudskoy, A. Streltsov. Y. Satov, K. Makarov. Y. Smakovsky, D. Hoffmann, W. Laux, R.W. Muller, P. Spaedtke, C. Stoekl, B. Wolf and J. Jakoby, Laser Particle Beams 14, 385 (1996). [13] M. Okamura, T. Katayama, R.A. Jameson, T. Takeuchi and H. Kashiwagi, Laser Particle Beams 20, 455 (2002). [14] P. Fournier, G. Gregoire, H. Kugler, H. Haseroth, N. Lisi, C. Meyer, P. Ostroumov, J.-C. Schnuriger, R. Scrivens, F. Varela Rodriguez, B.H. Wolf, S. Homenko, K. Makarov, Y. Satov, A. Stepanov, S. Kondrashev, B. Sharkov and A. Shumshurov, Rev. Sci. Instrum. 71, 924 (2000). [15] B.Yu. Sharkov, S. Kondrashev, S.M. Savin and A.V. Shumshurov, Nucl. Instrum Methods Phys. Res. A 415, 20 (1998). [16] T.W. Johnston and J.M. Dawson, Phys. Fluids 16, 722 (1973).

(1996). [18] B.Yu. Sharkov, A.A. Golubev, R.T. Khaidarov,

S.A. Kondrashev and A.V. Shumshurov, Rev. Sci. Instrum. 63, 2841 (1992). [19] V.Yu. Baranov, K.N. Makarov, V.C. Roerich, Yu.A. Satov, A.N. Starostin, A.E. Stepanov, B.Yu. Sharkov, K. Langbein and T.R. Sherwood, Laser Particle Beams 14, 347 (1996). [20] S. Homenko, K. Makarov, V. Roerich, A. Stepanov and Yu. Satov, Master-Oscillator-PowerAmplifier laser system for a laser ion source, Preprint TRINITI 0045-A, (1998). [21] B.Yu. Sharkov, S. Kondrashev, I. Roudskoy, S. Savin, A. Shumshurov, H. Haseroth, H. Kugler, K. Langbein, N. Lisi, H. Magnusson, R. Scrivens, S. Homenko, K. Makarov, V. Roerich, A. Stepanov and Yu. Satov, Rev. Sci. Instrum. 69, 1035 (1998). [22] A. Balabaev, S. Kondrashev, K. Konukov, A. Lozhkin, B. Sharkov, A. Shumshurov, A. Charushin, K. Makarov, Yu. Satov, Yu. Smakovskii, O. Camut, J. Chamings, H. Kugler, R. Scrivens, Rev. Sci. Instrum. 75, 1572, (2004). [23] Handbook of Ion Sources, edited by B. Wolf (CRC, Boca Raton, 1995). [24] R. Scrivens, Ph.D. Thesis, University of Wales, Swansea, 1999. [25] J.R. Coupland, T.S. Green, D.P. Hammond, A.C. Riviere, Rev. Sci. Instrum. 44, 1258 (1973). [26] P. Spadke and C. Muhle, Rev. Sci. Instrum. 71, 820 (2000). [27] J. Collier, G. Hall, H. Haseroth, H. Kugler, A. Kuttenberg, K. Langbein, R. Scrivens, T.R. Sherwood, J. Tambini, O.B. Shamaev, B.Yu. Sharkov, A. Shumshurov, S. M. Kozochkin, K.N. Makarov and Yu.A. Satov, Rev. Sci. Instrum. 67, 1337 (1996). [28] http://scrivens.home.cern.ch/scrivens/lis/ params/rfq/currentfromrfq.html [29] R. Becker and W.B. Herrmannsfeldt, Rev. Sci. Instrum. 63, 2756 (1992). [30] P. Ostroumov, CERN Technical Note 9.3.98 (1998). [31] H. Haseroth, H. Kugler, K. Langbein, N. Lisi, H. Magnusson, R. Scrivens, S. Homenko, K. Makarov, V. Roerich, A. Stepanov, B.Sharkov and Yu. Satov, Rev. Sci. Instrum. 69, 1051 (1998).

255

256


[32] S. Kondrashev, A.V. Balabaev, N.D. Mescher-

yakov, S.M. Savin, B.Y. Sharkov and A.V. Shumshurov, Rev. Sci. Instrum. 71, 1409 (2000). [33] The Large Hadron Collider Conceptual Design, CERN/AC/95-05 (LHC), http://www.cern.ch [34] N.N. Alexeev, V.N. Balanutsa, D.G. Koshkarev, V.I. Nikolaev, B.Yu. Sharkov, A.V. Shumshurov, P.R. Zenkevich and G.L. Mamaev, Laser Particle Beams 20, 385 (2002). [35] N. Alexeev, D. Koshkarev and B. Sharkov, JETP Lett. 77(3), 149 (2003). [36] N.N. Alexeev V.N. Balanutsa, V.I. Nikolaev and B.Yu. Sharkov, in Proceedings of EPAC 2000, p. 1283.

[37] N.D. Mescheryakov, N.N. Alexeev, A.N. Bala-

[38]

[39]

[40] [41]

nutsa, S.A. Kondrashev, V.I. Nikolaev, I.V. Roudskoi, B.Y. Sharkov and A.V. Shumshurov, Laser Particle Beams 20, 455 (2002). P. Lefvre and D. Mhl, in The Large Hadron Collider Conceptual Design, CERN/AC/95-05 (LHC), http://www.cern.ch, Lead Ion Accumulation Scheme for LHC, LHC Note 257. V.A. Monchinsky, L.V. Kalagin and A.I. Govorov, Laser Particle Beams 14, 439 (1996). A.D. Kovalenko, in Proceedings of EPAC-94, London, 1994, Vol.1, p.161. V.B. Kutner, Z. Phys. (Suppl). D 21, S 165 (1991).

257

13

Vacuum Arc Ion Sources Efim Oks and Ian Brown

13.1

Introduction

The distinguishing feature of a vacuum arc ion source is its capability of forming high current beams of metal ions. The metal plasma from which the ion beam is formed is produced by a vacuum arc discharge, and hence the name. Vacuum arc ion sources have been used primarily for ion implantation in material surface modification research, and for particle accelerator injection for fundamental nuclear physics research, and also for other fundamental and applied purposes. The source has provided a valuable addition to the spectrum of ion sources available to the experimenter. Beams have been produced from over 50 of the solid metals of the Periodic Table, with ion energy up to several hundred keV and with beam current up to several amperes. Typically the source is repetitively pulsed with pulse length of order a millisecond and duty cycle of order 1%, and operation of a dc embodiment has been demonstrated. The vacuum arc discharge provides a relatively direct method for generating metal plasma. No additional gas is needed, as for example in sputtering, and there is no necessity for low melting point, as for example with atomic ovens; the vacuum arc works in high vacuum and for virtually all metals. High density pure metal plasmas can be produced and consequently high current pure metal ion beams can be formed. The maximum ion current that can be produced is in fact determined by the beam formation system and not by the plasma; beam current can be much greater than for other types of metal ion sources. The ions are in general multiply ionized but of low charge state. The mean charge state of the ion beam lies mostly between 1+ and 3+, depending on the particular metal species, but the charge states can be increased in a number of different ways. On the other hand, there are also some disadvantages of the vacuum arc ion source. The metal plasma is generated at “cathode spots”, which by their fundamental nature are in chaotic motion throughout the arc pulse. This leads to an inherently noisy plasma and so also a relatively noisy ion beam. Another feature of the vacuum arc plasma discharge that can often be a disadvantage is the formation not only of copious amounts of metal plasma but also of “macroparticles” (solid cathode debris) that can be a contaminant to the ion beam. Thus the vacuum arc ion The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

258

13 Vacuum Arc Ion Sources

source is best suited to applications in which these negative features can be tolerated.

13.2

Background

Attempts to use the vacuum arc for a high current metal ion source were first made in the U.S. as part of the Manhattan Project during World War II [1]; this effort was abandoned, however, for reasons of arc instability. In 1979 Prewett and Holmes developed a carbon vacuum arc ion source at the University of Liverpool, U.K., to produce a low energy C+ ion beam current of up to 0.5 A [2]. A gridless vacuum arc ion generator was used by Adler and Picraux [3] at Sandia National Laboratories in the early 1980s for a kind of plasma immersion ion implantation application. At about the same time Humphries and co-workers at the University of New Mexico embarked on an extensive investigation of the vacuum arc for production of ion beams for heavy ion fusion research application [4–6]. Research in the general area of vacuum arcs has a long and rich history in the former Soviet Union (FSU) [7]. The development of vacuum arc based ion sources began in the late 1950s and early 1960s, led primarily by Plutto and co-workers at the Sukhumy Physical Technical Institute in Georgia [8]. This early work was largely a demonstration of the basic concept of metal ion extraction from vacuum arc plasma. In the mid 1960s a kind of vacuum arc ion source was developed by workers in Ukraine for the production of beryllium ion beams with current up to 170 mA [9]. The vacuum arc ion source program at the Lawrence Berkeley National Laboratory (LBNL) was initiated in 1982 for the production of high current uranium ion beams for injection into the LBNL heavy ion synchrotron (the Bevalac) for fundamental heavy ion nuclear physics research, and later for ion implantation application. The LBNL sources, called “Mevva” (metal vapor vacuum arc) ion sources, were developed in a number of different directions, including embodiments with multiple-cathode assemblies, very large extractors, miniature sources, and a test dc version. This work has been described in detail [10, 11]. At virtually the same time, development of vacuum arc ion sources was initiated at the High Current Electronics Institute (HCEI) of the Russian Academy of Sciences, Tomsk [12, 13]; these sources (called “Diana” [12] and “Titan”[13]) played a significant role in the rapid growth of activity in the field throughout the FSU. A series of vacuum arc ion sources called “Raduga” were developed at the Nuclear Physics Institute (NPI) of the Tomsk Polytechnic University [14, 15], and the “Tamek” sources were developed at the Tomsk Institute of Automatic Control Systems and then at the Applied Physics Institute, Sumy, Ukraine [16]. Programs were subsequently established in Australia [17], China [18], Germany [19], Japan [20], and Moscow [21, 22], among other world universities, institutions and laboratories. An historical review of the early development of vacuum arc ion sources has been given elsewhere [23].

13.3 Vacuum Arc Plasma Physics

13.3

Vacuum Arc Plasma Physics

Reviews of vacuum arc plasma discharges have been given by a number of authors [24–26]. The vacuum arc is a discharge between two electrodes in vacuum. At the cathode the current is concentrated to a small number of tiny, discrete sites, called cathode spots. The formation of cathode spots is a fundamental characteristic of the vacuum arc discharge. The spots are where the metal plasma is produced, and it is this plasma that provides the current path between cathode and anode that keeps the arc alive. Thus some of the plasma that is generated at the cathode must necessarily deposit on the anode so as to form the current path; and some of the metal plasma can be taken away, using suitable geometry, and used for another purpose – for example to form the ion beam of an ion source. At a cathode spot the cathode surface is heated, vaporized, and ionized into the plasma state. The spot is of diameter of the order of 10 lm, and the current density is extremely high, of the order of 106–108 A cm–2. The arc current is constricted to a small number of such spots. Most of the parameters of the vacuum arc plasma are determined by the plasma physics within the cathode spot. Individual spots move around on the cathode surface, and the lifetime of a particular spot may be only microseconds or less; on the other hand, small surface irregularities like edges or protuberances tend to anchor the spots to these sites. The plasma pressure within a cathode spot is high, and the strong pressure gradient causes the plasma generated there to plume away from the surface in a manner reminiscent of the plasma plume generated by an intense focused laser beam at a solid surface. The current carried by a cathode spot is typically of the order of a few to a few tens of amperes, depending on the metal, and if the arc is caused to conduct a higher total current, then more cathode spots are formed. Thus a typical metal vapor arc discharge of several hundred amperes current might involve the participation of several tens of cathode spots. The assemblage of cathode spots gives rise to a dense plasma of cathode material that streams away from the cathode as a jet. Ambient gas is not essential to the discharge. Much vacuum arc research has been carried out in the 10–6 Torr range, and ~10–4 Torr might be considered a rough upper limit. Residual gas can have a significant influence on vacuum arc plasma parameters, especially on the metal ion charge state distribution [27]. The arc current can be anywhere in the range from several tens of amperes up to many kiloamperes, and for the mode of discharge employed in most vacuum arc ion sources the current is typically one hundred to several hundred amperes. The arc voltage (when the arc is alight, i.e., the burning voltage’) lies in the range 10 – 30 V, say typically about 20 V, and varies according to the cathode metal used. Unlike the more common high pressure gaseous arc discharge, the vacuum arc displays a positive resistance characteristic, that is, the arc voltage increases with arc current, albeit only slowly. The mechanism by which the arc is established in the first place is poorly understood. Certainly the concept of arc growth by an electron avalanche in a gaseous medium between the electrodes – the usual picture of gaseous breakdown – is not

259

260


applicable. Here conduction of the arc current is supported by metal plasma that is evolved from the solid electrode (cathode) material itself. Thus the problem is taken back to the initial step of establishing the first cathode spot and the evolution of plasma from the cathode. In most vacuum arc ion sources, triggering of the vacuum arc has been accomplished by using surface flashover to the cathode across a thin insulating surface, and other methods have also been used [28]. A well-established basic feature of the vacuum arc discharge is the relationship between the metal ion flux that is generated and the current that drives the arc. The arc current is composed of both an electron component and an ion component. The plasma ion current is the plasma flux generated at the cathode that is in the form of metal ions. Over a wide range of conditions the plasma ion current is a constant fraction of the arc current , Iion = eIarc,

(13.1)

where e is in the range 0.06–0.12, say typically about 10%. Thus the electrical efficiency of the vacuum arc – the ratio of available metal ion plasma current to arc current – is high, about 10%. The electrical efficiency of the vacuum arc ion source can thus also be high, although the steps (total metal plasma generated by the arc) to (plasma transported to the beam formation electrodes) to (formation of ion beam from plasma presented to the extractor) to (ion beam delivered to a downstream target or beam-line) conspire together to yield an overall electrical efficiency that is of course considerably less than the fundamental vacuum arc efficiency of 10%. The relationship between the mass evolved from the cathode and the arc parameters has also been investigated. We can generalize to say that the mass of the plasma generated by the vacuum arc is of the order of several tens of micrograms per Coulomb of arc current, where the precise value depends primarily on the metal used. At the cathode spots, cathode material is converted into metal plasma and solid “macroparticles” (so called because they are macroscopic compared to plasma particles, ions). The macroparticles are metallic globules that are ejected from the cathode in the molten state and then rapidly solidify; they are typically of diameter in the range 0.1– 0 lm. For many cases (e.g., for high melting point cathode materials) the macroparticle content of the ion beam is small and is not a concern.

13.4

Principles of Operation

In an enclosed geometry all of the plasma that is generated at the cathode strikes the surrounding anode, and the current needed to keep the arc alight is conducted through the metal plasma in the space between the cathode and the anode. In an ion source embodiment, one wants to use some of the metal plasma for other than arc current conduction; this is the plasma that is presented to the extractor for ion beam formation. Bearing in mind that the plasma is formed as a jet that plumes away from the cathode in a direction normal to the cathode surface, one way of

13.4 Principles of Operation

accomplishing this is to employ a coaxial geometry as shown in Figure 13.1, a schematic of the LBNL Mevva II vacuum arc ion source [10, 11]. A part of the plasma created at the cathode flows through the anode hole, of diameter about 1 cm, and through a drift space of several centimeters to the extractor grids. The cathode is a simple cylindrical rod (typically 5–10 mm diameter) of the material of interest, and the trigger electrode surrounds the cathode, separated from it by a thin (about 1 mm) alumina insulator. Some of the plasma strikes the anode and serves to carry the current that keeps the plasma alight, while a central part of the plasma plume streams through the central hole in the anode to the extractor grids where the ion beam is formed from the plasma. The kind of beam formation electrodes (extractor grids) used can be chosen to suit the particular application, and the conventional multi-aperture, accel–decel configuration is simple and effective. A set of three circular grids is located appropriately with respect to the expanding plasma plume, with the first grid (nearest the plasma) at or near plasma potential, the middle grid typically several kilovolts negative with respect to ground so as to suppress the backflow of low energy electrons, and the third grid at ground potential. The first grid and the entire plasma generation part of the source are biased to high positive potential. The magnetic field coil shown in Figure 13.1 can be considered as optional. A modest longitudinal magnetic field in the arc region can be used to control the amount of plasma that is transported from the cathode to the extractor. In the absence of magnetic field, much of the plasma is lost to the walls; with an applied field of just a few hundred gauss, the radial plasma loss can be much reduced and a greater fraction of the plasma that is formed can be presented to the extractor and converted into beam. Thus for highest ion beam current and maximum ion source efficiency, a variable magnetic field is advantageous.

Figure 13.1 Simplified schematic of the LBNL Mevva II ion source. (Courtesy of Lawrence Berkeley National Laboratory.)

261

262


A generalized schematic of the electrical system to drive the source is shown in Figure 13.2. The arc current can be supplied by a floating transistor-switched current supply, or more simply by a low-impedance LC pulse line if the required pulse length is no greater than about a millisecond. Lines of length 100–500 ls and impedance 0.5–1.5 X, charged to several hundred volts, are compact and convenient. This is a simple way of providing the 50–500 A necessary for the arc, and the problem of floating the supply up to high extractor voltage is minimized. The trigger pulse is about 10 kV (O/C) in amplitude and 10 ls in duration, supplying a peak current of a few tens of amperes, conveniently applied trigger-to-cathode by a stepup transformer which also serves for high voltage isolation.

Figure 13.2

Simplified schematic of the Mevva overall electrical system.

13.5

Beam Parameters

While the precise source performance is of course determined by the specific source embodiment used, one can nevertheless generalize, referring to specific source embodiments for specific experimental results. We summarize here the beam parameters that are typical of vacuum arc ion sources. 13.5.1

Beam Current

The ion beam current can easily be as high as several amperes, and for very broad beam sources up to several tens of amperes [11]. The fraction of this beam that is useable, however, for example for injection into an accelerator low energy beam line or for implantation of a downstream target, depends on the particular set-up. The


beam transported downstream depends on the beam emittance and the target (or beam line) acceptance as well as losses in the beam transport system. For example, a 10 cm extractor source version (Mevva V) has produced a titanium ion beam current into a nearby large area Faraday cup of up to 3.5 A at 90 kV extraction voltage [11]. In another series of measurements a source embodiment with a 2 cm diameter extractor was used to determine the beam extraction characteristics as shown in Figures 13.3 and 13.4. Figure 13.3 shows the Ti beam current as a function of arc current for a range of extraction voltages, and Figure 13.4 shows the Ta beam current as a function of extraction voltage for a range of arc currents. Measurements of this kind have been made for a number of different cathode materials (ion beam species) and, in summary, good agreement was found with the well known Child–Langmuir prediction for extracted ion current under spacecharge-limited conditions, 1 3 4 2eQi V2 2 eo S I ¼ Mi d2 9 ¼

1 3 Q V2 1:72 S i 2 2 , A d

(13.2)

where S is the extractor open area, Qi is the ion charge state (ion charge q = eQ), Mi = Amu is the ion mass where A is the atomic weight in amu and mu the mass of 1 amu, V is the extractor voltage, d is the extractor gap, and in the second expression I is in mA, V in kV, S in cm2, and d in cm. The implication is that the ion beam

Figure 13.3

Ion beam current as a function of arc current. Titanium beam.

263

264


Figure 13.4

Ion beam current as a function of extractor voltage. Tantalum beam.

current that can be expected from any embodiment of vacuum arc ion source can be predicted with reasonable accuracy from the Child–Langmuir equation above. 13.5.2

Beam Profile, Divergence and Emittance

The beam divergence is determined primarily by the extraction optics, and assuming that the extractor grids have been designed and fabricated reasonably, the extraction optics can be empirically matched by variation of the plasma density via the arc current. At optimum (the “perveance match” condition), the beam divergence is typically about 3 half-angle. The normalized emittance, eN, is typically about 0.2 p mm mrad (normalized) for uranium at perveance match. The initial beam shape, i.e., the beam radial profile immediately after extraction, is determined by the radial plasma density distribution at the extractor location, together with the extraction optics. If the plasma density is not flat across the extractor, the extraction optics cannot be matched everywhere. A multipole magnetic bucket configured of rare-earth magnets and located in the ion source drift region can be used to flatten the plasma profile at the extractor [29], but even so, the beam profile loses memory of its shape at the extractor fairly rapidly, and after some tens of cm downstream propagation the beam profile reverts to the usual Gaussian [30]. Experimentally, control of the arc current provides a convenient means of optimizing the extraction via the plasma density. Alternatively, the extraction optics and so also the extracted beam profile can be tailored by the extractor voltage for a fixed arc current (plasma density).


13.5.3

Beam Composition

The vacuum arc ion source produces ions that are multiply ionized, and the charge state spectrum of the ion beam is important for most vacuum arc ion source applications. Ion charge state distributions (CSD) have been studied in detail experimentally [31] using time-of-flight (TOF) charge state diagnostics [32], and theoretically [33]. Almost all of the solid metals of the Periodic Table have been successfully used in vacuum arc ion sources, as well as cathodes made from metallic alloys, compounds, and pressed mixtures. Compound cathodes produce ions of the cathode constituents, and it is interesting to note that beams containing non-metallic elements, such as B and S, can be made by using conducting compound electrodes of which the non-metal is a constituent such as LaB6 or PbS. An example of the charge state spectra obtained is shown in Figure 13.5, where an oscillogram of an iridium time-of-flight spectrum is shown. Note that for multiply-charged ions, electrical current ielec is not the same as particle current ipart, since each particle can carry multiples, Q, of the electronic charge, e: ielec = Qipart, This can be important, as for example in the case of the beam current measured by a Faraday cup being electrical current, while the current needed for estimating ion implantation dose is particle current. The ions generated by the vacuum arc are in general multiply-stripped with a mean charge state of from 1 to 3, depending on the particular metal species, and the charge state distribution can have components from Q = 1+ to 6+; thus the ion energy is greater than the extraction voltage by this same factor. However, the ion charge states can be increased substantially in a number of different ways; this is discussed below in Section 13.6.1. The measured charge state distributions and mean charge states for a wide range of elemental species, under conventional vacuum arc operational parameters and in the absence of techniques to enhance the highly charged ion composition, are given in Table 13.1. More detail has been given in Ref. [31].

Figure 13.5 Time-of-flight charge state distribution for an iridium ion beam. The amplitudes are electrical current measured by a Faraday cup.

265

266

13 Vacuum Arc Ion Sources Table 13.1

Ion charge state fractions and mean charge states, expressed in terms of particle cur-

rent. Element

Li C Mg Al Si Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn Ge Sr Y Zr Nb Mo Rh Pd Ag Cd In Sn Sb Ba La Ce Pr Nd Sm Gd Dy Ho Er Tm Yb Hf Ta

Z

3 6 12 13 14 20 21 22 23 24 25 26 27 28 29 30 32 38 39 40 41 42 45 46 47 48 49 50 51 56 57 58 59 60 62 64 66 67 68 69 70 72 73

Charge State 1+ 2+ 100 100 46 38 63 8 27 11 8 10 49 25 34 30 16 80 60 2 5 1 1 2 46 23 13 68 66 47 1 1 3 3 2 2 2 2 1 13 3 3 2

54 51 35 91 67 75 71 68 50 68 59 64 63 20 40 98 62 47 24 21 43 67 61 32 34 53 * 100 76 83 69 83 83 76 66 66 63 78 88 24 33

hQi 3+

11 2 1 6 14 20 21 1 7 7 6 20

4+

5+

1 1

1

* 33 45 51 49 10 9 25

7 22 25 1 1 1

2 3

*

23 14 28 17 15 22 32 32 35 9 8 51 38

* 1

21 24

1 3

6+ 1.0 1.0 1.5 1.7 1.4 1.9 1.8 2.1 2.1 2.1 1.5 1.8 1.7 1.8 2.0 1.2 1.4 2.0 2.3 2.6 3.0 3.1 1.7 1.9 2.1 1.3 1.4 1.5 1.0 2.0 2.2 2.1 2.2 2.2 2.1 2.2 2.3 2.3 2.4 2.0 2.1 2.9 2.9

13.5 Beam Parameters Element

W Ir Pt Au Pb Bi Th U

Z

74 77 78 79 82 83 90 92

Charge State 1+ 2+ 2 5 12 14 36 83 20

23 37 69 75 64 17 24 40

hQi 3+

4+

5+

43 46 18 11

26 11 1

5 1

64 32

12 8

6+ 1

3.1 2.7 2.1 2.0 1.6 1.2 2.9 2.3

* Trace (under 1%).

13.5.4

Beam Noise, Pulse Stability, and Lifetime

By beam noise we mean the fractional fluctuation level of ion beam current about the mean beam current: di/ib. Beam noise is unimportant for ion implantation but very important for accelerator injection application. The noise level varies according to the arc current. There is an optimum operating point at which the rms noise is a minimum of about 7%; this corresponds approximately to the perveance match condition, when the plasma density is optimally matched to the extraction optics. Characteristics of the beam noise have been studied and reported [6, 11, 34, 35]. Work carried out collaboratively between GSI at Darmstadt, Germany, HCEI at Tomsk, Russia, and LBNL at Berkeley, USA, has resulted in techniques for greatly reducing the beam noise level; see Section 13.6.3. Reproducibility of the beam current pulse shape from pulse to pulse – pulse stability – is unimportant for most implantation applications but important for accelerator injection application. Pulse stability is improved by operating the source at the same optimum arc current as for minimizing beam noise (perveance match), and by ensuring that the arc cathode and trigger are not severely eroded. Pulse-topulse reproducibility becomes poor as the cathode end-of-lifetime is approached. It is important that the source triggers reliably, with only a small triggering failure rate. When the trigger pulse is electrically adequate and the annular alumina trigger insulator is in good condition, the triggering reliability can be excellent, with a failure rate of just several misfires per 1000 pulses, or ~0.1%. The peak voltage of the trigger pulse should be not much less than about 10 kV and the current drawn upon triggering breakdown should be a few tens of amperes, as needed to form the first cathode spot. There should be no vacuum gap in the cathode–insulator–trigger surface, so that the trigger discharge is indeed a surface breakdown; and the plane of the trigger ring should be slightly below the plane of the cathode so that recondensed metal cannot form a short from cathode to trigger. Alternative triggering methods are discussed below in Section 13.6.2; using the “triggerless” method of arc initiation, a lifetime of over 1 million pulses has been obtained from a single cathode.

267

268


The time for which the source can run between necessary maintenance periods is determined by the need for cathode replacement and/or by triggering failure. Note however, that if meshes are used, they can become covered with metal; transparency suffers and the mesh may need to be replaced. For typical operating conditions one can obtain up to several hundred thousand pulses before the cathode is eroded away enough to require change, depending on the arc current and pulse length. With a multiple cathode source (see Section 13.7.1) one can simply switch to another cathode and in this way the total number of shots can be extended up to several million before needing to replace all the cathodes in the cathode assembly. The source can be operated steadily at a pulse rate that might suit synchrotron application, say 1 pulse per second or lower, for 24 hours per day for a whole week between changes; or at a higher pulse rate of 20 to 40 pulses per second as might be suitable for ion implantation applications for several 8-hour days. Soft materials such as Li, Sn and Pb, and C, have a shorter lifetime because much of the cathode mass is eroded in the form of macroparticles rather than as plasma from which an ion beam can be formed.

13.6

Recent Improvements in Parameters and Performance 13.6.1

Enhancement of Ion Charge States

The generation of high ion charge states is important for both fundamental research and industrial applications. Increasing the high charge state ion fraction provides a means of increasing the ion energy without increasing the ion accelerating voltage. For example, ions with charge state Q = 10+ acquire an energy of 1 MeV for an acceleration voltage of just 100 kV. Another perspective follows from the high ion density that can be obtained in the vacuum arc plasma and thus the high current beams that can be produced: even if a particular high charge state species forms only a small fraction of the total ion beam current, nevertheless that fraction may still constitute an impressive beam current. For example 1% of a total ion beam current of 1 A is still 10 mA, which is “high” for many applications. Methods for increasing the ion charge states in vacuum arc plasmas include the use of strong magnetic fields in the arc discharge region, modification of the arc current pulse shape by adding high current “spikes”, and passage of an intense electron beam through the vacuum arc plasma. Each method has its advantages and disadvantages, and each method does result in an increase in the ion charge states. The maximum increase in mean ion charge state is a factor of 2.5. In the following we summarize the work that has been done in this area and the results achieved; these methods have been reviewed in detail [36]. Electron cyclotron resonance (ECR) heating of the vacuum arc plasma falls into a somewhat different category, and one can alternatively view the hybrid device as an ECR ion source with vacuum arc plasma injection. This approach has also been explored and is described below.

13.6 Recent Improvements in Parameters and Performance

Finally, we note that a sibling kind of discharge – the vacuum spark – forms metal plasmas with much more highly ionized species. The spark discharge is typified by its short pulse (microseconds or less) and high current (many kA). The ion charge state spectrum formed in vacuum sparks has been explored, but conventional ion sources of this kind appear not to have been developed. A vacuum arc ion source has been operated in a spark-like mode and species such as Ti6+ and W8+ have been seen [37]; and a test version of a vacuum spark source revealed copper species up to Cu7+ [38]. 13.6.1.1 Magnetic Field Effects The effect of magnetic field on vacuum arc ion charge state distribution has been explored [39–43]. Typically a pulsed magnetic field is established by a small coil surrounding the arc. The coil may be either internal or external to the vacuum chamber; there is advantage to it being internal and small, thus allowing high fields of over 1 T to be produced relatively easily. The coil may be driven by a separate power supply or alternatively connected in series with the arc current. The influence of the magnetic field on the ion CSD first becomes apparent for field strengths B > 0.1 T, and fields up to 2 T have been investigated. The general result is that the stronger the applied field, the greater the charge state increase. As an example, the influence of magnetic field on a silver plasma is shown in Figure 13.6 [43]. The enhancement factor – the ratio of mean charge state with magnetic field to mean charge state without magnetic field – depends on the magnetic field strength and the cathode material, and varies from a low of 1.3 (for Sc and Sr) up to a high of 2.3 (for Ni and Bi). New high charge states, not visible in the CSD spectra under B = 0 conditions, appear; for example ions such as C3+, Ti5+, Cr5+, Ni5+, U6+ and others can be produced in this way. These results have been described in detail [36, 43]. Ion charge state enhancement in a strong magnetic field occurs because of the increased temperature of the plasma electrons that are responsible for the ionization. Numerical simulations carried out by Anders [44] show that an increase in electron temperature of just 1 eV can explain the observed charge state increase.

Figure 13.6 Influence of magnetic field on the ion charge state distribution for a silver plasma.

269

270


13.6.1.2 The “Current Spike” Method When a magnetic field is applied to the arc, as described in the preceding, an increase in the arc voltage (the “burning voltage” of the arc) is observed. That is, there is an apparent correlation between the production of high charge states and increased arc voltage. This observation led to another approach to the production of increased high charge state fractions, the formation of a transient “current spike” added to the main arc discharge current. In experiments investigating this effect [45, 46] a current spike of up to 1 kA with pulse length about 4 ls was added to the main arc current pulse of magnitude 100–250 A and 400 ls duration at a delay time of 150–200 ls after arc initiation. At the time of the spike the arc voltage jumped to the charging voltage of the spike capacitor, as high as 1 kV, and then decreased to a few tens of volts in several microseconds. An increase in the high charge state ion fraction was observed over the same timescale for which the arc voltage was high. Charge state enhancement was observed for all cathode materials explored, and the magnitude of the effect was comparable to that seen when a magnetic field is applied as described above. The duration of enhanced high charge state production is limited to the duration of the current spike, which for good effect needs to be less than about 20 ls. For many ion beam applications the pulse length should be greater than this. A train of pulsed current spikes occurs as a possible solution to this problem, but experiments with double spike and subsequently also with multiple spikes have shown that the enhancement effect diminishes from pulse to pulse if the time between pulses is less then the pulse duration.

Addition of a High Current Electron Beam – the E-Mevva Injection of an intense electron beam into a vacuum arc metal plasma and the subsequent extraction of ions has been the subject of some ongoing research [47]. The name E-Mevva has been given to this kind of device. In the experimental work, an 13.6.1.3

Figure 13.7 TOF spectrum for the HCEI E-Mevva operating with a Pb cathode. Upper trace: no electron beam. Lower trace: with electron beam.


electron beam of energy about 20 keV and current about 40 A was formed and injected into a vacuum arc plasma. An example of the influence of the electron beam on the ion charge states is illustrated in the results with a Pb cathode, as shown in Figure 13.7. A significant influence of the electron beam on the high charge state ions is clearly visible. In the conventional mode of vacuum arc operation one obtains only singly- and doubly-charged ions in the Bi TOF spectra, for a mean ion charge state of 1.4. Application of the electron beam moves the CSD toward much higher charge state species. The maximum charge state that can be seen is Pb7+, and the mean charge state is around 4+. The E-Mevva approach could have considerable potential as a high charge state ion source, and work in this direction is continuing. Electron Cyclotron Resonance Another possible approach to enhanced high charge state production in vacuum arc plasmas is the application of electron heating in an ECR (electron cyclotron resonance) configuration. Early attempts to form a hybrid Mevva-ECR device were reported in Ref. [48]. More recently the challenge has been taken up again at Nizny Novgorod [49]. The experimental set-up used at Nizny Novgorod made use of a high power pulsed gyrotron (37.5 GHz, 70 kW, 1.5 ms) for microwave generation. A miniature vacuum arc plasma gun (100 A, 0.5 ms) with an anode aperture of just 2 mm was located within the vacuum chamber, on the axis and at the magnetic field maximum so as to inject metal plasma into the ECR mirror zone. The main ECR region was 27 cm in length and configured as a simple magnetic mirror system (Bmax ~2.5 T, mirror ratio 4). The charge state distribution of the ion efflux from the ECR trap was measured by a multi-channel magnetic analyzer. Preliminary results have been encouraging, indicating that the normal vacuum arc Pb spectrum containing only Pb+ and Pb2+ species is elevated up to Pb6+. This work is continuing. 13.6.1.4

13.6.2

Alternative Triggering of the Vacuum Arc

The triggering system most often used for vacuum arc ion sources is based on a surface flashover discharge across a thin insulator coaxially located between the cathode and a surrounding trigger ring, as described in Section 13.4. The first cathode spot is formed at the metal–insulator boundary. However, there is significant erosion of the ceramic insulator material, and (elsewhere) metal film deposition on the insulator surface. These factors limit the lifetime of this kind of triggering system to ~105 pulses, or less for high arc current and/or long pulse operation. The reliability of the surface flashover mode of triggering is improved if the surface of the insulator is slightly conductive; this occurs naturally after some arc operation since metal plasma and macroparticles (droplets) are deposited onto the insulator surface. In fact, operation with a new, clean, ceramic insulator can fail to trigger because the surface resistance may be too high for breakdown, and contaminating the insulator with a graphite pencil helps to get started. Extension to the limit of this thinking can lead to the kind of triggering shown schematically in Figure 13.8 [28].

271

272


Figure 13.8 Schematic of “triggerless” vacuum arc initiation. The insert shows an enlarged view on the cathode–insulator– vacuum triple-junction. (From Ref. [28].)

In this system no trigger electrode is present; instead, the surface of the insulator separating cathode and anode has a conducting surface layer. A completely adequate surface coating can be applied simply by heavily marking the entire insulator surface with a graphite pencil. When the main arc circuit is switched on (by an additional high current switch in the arc circuit), current across the surface of the insulator causes the first cathode spot to be initiated on the edge of the cathode near the insulator. The high current switch can be a thyristor (SCR), high-current transistor, or other; and the arc supply, shown as a simple capacitor in the figure, can be a pulse forming network, or a high-current supply such as an arc welder or other. This system has been called “triggerless” arc initiation, and has been used in vacuum arc plasma sources and ion sources with lifetimes in excess of 106 pulses [28]. A gaseous pre-discharge can provide another approach to cathode spot ignition in vacuum arc ion sources. In order to avoid the negative influence of background gas on the ion charge states distribution [27], a low pressure gas discharge should be used. Two closely related kinds of gaseous discharges that are suitable and that have been used for vacuum arc triggering are the Penning and the cylindrical magnetron discharges; both are E B discharges. In both cases the electrons are confined by a strong magnetic field and so have a long lifetime within the discharge gap. For both systems the main arc is ignited between the cathode and the hollow anode of the gaseous discharge; the hollow cylindrical anode thus serves a dual purpose. A small solenoid coil is used in both cases to establish a magnetic field in the plasma region. With the same trigger pulse parameters as for the surface flashover method (10 kV, 10 ls), stable arc ignition can be obtained even for background pressure as low as 10–6 Torr. Operation for long lifetimes has been obtained by this approach, greater than 106 pulses at a repetition rate of 50 pps. This approach has been described fully elsewhere [50].


13.6.3

Reduction in Ion Beam Noise and Increased Pulse Stability

Low beam current fluctuation level (low beam noise) and good shot-to-shot reproducibility of beam pulse shape are essential requirements for application of vacuum arc ion sources to accelerator injection [19] and heavy ion fusion [51]. The primary origin of beam noise and pulse shape instability has to do with the random, explosive nature of cathode spots in the vacuum arc. Humphries and co-workers developed a technique for reduction in ion beam noise level by the use of two fine meshes to reflect plasma electrons and to limit pure ion flow by its own space charge [5, 6]. In this case of space charge limited flow the extracted ion beam current is independent of the level of plasma fluctuations and random plasma generation at the cathode spots. The method has been further developed specifically for accelerator injection application [52, 53]. As mentioned in Section 13.5.4, minimum noise level is found for optimal matching of extraction voltage and plasma density – perveance match. For lower density or higher extraction voltage, the extraction gap can transmit more current than the plasma can provide, and all plasma noise is effectively transformed into beam noise. For an overdense plasma at the extraction system, the ion current is too high to be extracted, resulting in an unstable ion beam. Note that an increase in gas pressure in the discharge gap and in the extraction region improves the beam quality, but this method shifts the CSD to lower charge states [27]. There is also a significant influence of magnetic field on beam noise and pulse stability. Work carried out at GSI Darmstadt, Germany, over a period of several years has resulted in a vast improvement in both the beam noise and the beam pulse shape reproducibility [54, 55]. By combining a number of techniques, including the use of meshes, addition of optimized magnetic field, and geometrical optimization, the performance of the GSI vacuum arc sources has been brought to the point where they are fully acceptable for routine accelerator injection application. The most recent such source has been called the Varis; this source is described below in Section 13.7.4. The good performance of this source with respect to beam noise is evident from Figures 13.9 and 13.10. Figure 13.9 shows the pulse shape of a uranium ion beam, pre-analysis with all charge states, as measured by a Faraday cup 30 cm downstream; current is 156 mA and beam noise is €4%. Figure 13.10 shows the

Figure 13.9 Uranium ion beam current pulse from the GSI Varis source. 40 mA cm–1 vertical, 100 ls cm–1 sweep speed.

273

274


Post-analysis U4+ beam at the RFQ entrance; 5 mA div–1 vertical, 100 ls cm–1 sweep speed. Figure 13.10

post-analysis U4+ beam as measured by a beam transformer at the entrance to the pre-LINAC RFQ accelerator, a point about 12 m downstream from the ion source; U4+ beam current is about 25 mA and beam noise is €5%. The pulse shape, beam noise, and reproducibility are adequate for the Varis source to be used routinely at GSI for high current injection of metal ions into the accelerators. See Section 13.7.4 for more details. 13.6.4

Generation of Gaseous Ions

The vacuum arc discharge requires no ambient gas. The current carrying medium is the metal plasma formed from the conducting cathode material. However, the presence of added gas in the arc region can influence the arc processes and can affect the metal plasma parameters [22]. Importantly, gaseous ions can be added to the metal ion beam at experimenter option. Gas is fed into the source near the arc region where the plasma density is high and when a modest magnetic field (B ~ 200–400 G) is applied. The gas is ionized and gaseous ions are introduced into the metal plasma and extracted ion beam, in addition to a down-shifting of the metal ion charge states. The fraction of gaseous ions can readily be comparable to the metal ion fraction; the gas-to-metal ion ratio can be controlled from zero to as much as about ~99% gaseous. An example of this effect is shown in Figure 13.11(a) for a titanium ion beam [57]. In this case a magnetic field has been added to the arc region as described above in Section 13.6.1.1; thus the low pressure Ti charge states are elevated over the “usual” charge state distribution for the case of zero-B of Table 13.1. As the pressure is increased, there is a slow change of metal ion charge states from high to low, and a simultaneous increase in the fraction of gaseous ions in the beam. The overall metal-to-gas ion fraction changes with pressure as shown in Figure 13.11(b). Two conclusions follow from these kinds of observations: (i) high fractions of highly charged ions call for the lowest possible background gas pressure, and (ii) ambient gas pressure provides a means for controlling the ion charge state distribution toward lower states. These effects can be used to advantage. For example, a hybrid Ti–N ion beam has application for implantation of sub-surface TiN layers, and similar buried ceramic layers can be formed from other species such as Al–O, Zr–O, etc.

13.7 Source Embodiments

Figure 13.11 (a) Change in charge state composition for a Ti ion beam as a function of nitrogen background gas pressure. Also showing the nitrogen ion fractions. (b) Metal-to-gas ion beam composition fraction as a function of nitrogen gas pressure, for the case of a hybrid Ti–N ion beam.

[57]. In work related to accelerator injection, a vacuum arc ion source was used to provide Mg+ ions for injection into the GSI heavy ion accelerator, having increased the ion source gas pressure so as to maximize the fraction of singly-charged ions with respect to the doubly-charged Mg2+ fraction [58, 59]. A way of operating a vacuum arc source in a 100% gaseous mode has been reported [60]. In this case the vacuum arc feature is not used at all, but instead a hollow cathode glow discharge is formed in the same ion source geometry. There is a small gaseous contamination of the metal ion beam even at the lowest pressures at which the vacuum arc ion source is normally operated (~10–6 Torr or less). The primary source of this is condensation of gas (oxygen, nitrogen, water vapor) on the fresh front surface of the cathode from the residual ambient between vacuum arc pulses. This also affects, to a small extent, the metal ion charge state distribution. The magnitude of the effect depends on the time between pulses, and can be minimized by operating at high repetition rate, say greater than about 10 pps [61].

13.7

Source Embodiments

We describe here a number of vacuum arc ion sources that have been developed at several different laboratories around the world. These sources demonstrate the features outlined in the preceding and exemplify the versatility possible.

275

276


13.7.1

LBNL Mevva Sources

A number of vacuum arc sources have been developed at the Lawrence Berkeley National Laboratory (LBNL) [11], of which the broad-beam Mevva-V has evolved to be the most used. This source has been operated steadily for many years with infrequent need for maintenance or repair. It remains a reliable workhorse and has served as a basic model for vacuum arc ion sources elsewhere. A photograph of the source is shown in Figure 13.12 and a schematic in Figure 13.13. This embodiment uses an extractor grid 10 cm in diameter, incorporates an 18cathode multiplex assembly, and can operate at up to about 65 kV extraction voltage (absolute maximum is 100 kV). For optimum beam optics the plasma profile across the extractor aperture can be kept somewhat uniform by means of a samarium-

The LBNL Mevva-V source, partly disassembled, showing the 10 cm extractor grids on the left and the multiple cathode assembly on the right. (Courtesy of Lawrence Berkeley National Laboratory.)

Figure 13.12

Schematic of the Mevva-V source, fitted with pulsed solenoid surrounding the arc region and with gas inlet feed. (Courtesy of Lawrence Berkeley National Laboratory.)

Figure 13.13


cobalt permanent magnet multipole structure located within the plasma expansion chamber, but this is not essential and has not normally been used. The cooling of the source was increased and made more efficient than in the earlier versions, and so also the mean power dissipation capability. Upgrades to the Mevva-V that have been incorporated over the years include the addition of a small high-field pulsed solenoid around the arc region for charge state control, and the addition of a gas feed inlet to the arc region for the formation of hybrid metal-gas ion beams [59]. Finally we note that the Mevva-V ion source has been operated as a high current electron source by simply changing the source bias from positive to negative [62], without any other change to the set-up. The Mevva-V has been fully described [11, 63, 64]. 13.7.2

HCEI Titan Sources

The vacuum arc ion sources made and developed at the High Current Electronics Institute, (HCEI), Tomsk, Russia, were given the name “Titan”. These sources were a parallel development to the Berkeley Mevva sources. The Titan sources have the very nice feature of being able to produce either metal or gas ion beams by using either of two different types of arc discharge – a vacuum arc (metallic) or a constricted arc (gaseous) – in the same discharge system. This allows the formation of surface layers in the implantation target material of metal-gas compounds with high strength characteristics. Metal ions are generated by operating in the usual vacuum arc mode. To generate gaseous ions, a constricted arc discharge with cold cathodes is used. In this kind of discharge cathode material ions are produced in the cathode region, while plasma is generated in the anode cavity by ionization of the working gas. Several versions of the source have been developed, each with its own special features. A schematic of the Titan-3 ion source is shown in Figure 13.14. This source uses gas for the high voltage insulation, as opposed to earlier versions that used oil insulation; use of oil leads to carbon in the ion beam due to the diffusion of oil through various vacuum seals. Water cooling is used, and provision is made for electrical isolation, 3, of the high voltage discharge chamber from the grounded cooling water supply unit. If distilled water is used, this decoupling ensures adequate high voltage hold-off and low leakage current for voltages up to 80 kV. Distilled water circulates in a closed loop. The arc current is 40–150 A and the pulse duration is 400 ls. The repetition rate is variable over the range 10–50 Hz. The extractor diameter is 15 cm. Since the discharges have independent power supplies and the plasma generation processes in the discharges are different and quite separate from each other, it is possible to produce a two-component metal-gas ion beam with the ion fractions in each component being readily controllable by variation of discharge currents. The total current extracted from the source ranges between 0.1 and 0.3 A for gas ions and from 0.2 to 0.5 A for metal ions. The Titan sources have been described in detail [13, 65, 66].

277

278


Figure 13.14 Schematic of the Titan-3 ion source: 1 – gas feed inlet pipe; 2 – high-voltage cable; 3 – water cooling pipe; 4 – outer shield; 5 – constricted arc channel; 6 – vacuum arc cathode; 7 – high-voltage discharge chamber screen; 8 – tapered anode; 9 – ion beam formation electrodes.

13.7.3

NPI Raduga Sources

A series of vacuum arc sources was developed at the Nuclear Physics Institute (NPI) of the Tomsk Polytechnic University, given the name “Raduga” [15]. The sources incorporated a number of novel features, including multiple cathodes between which the arc could be rapidly switched, thus generating beams alternating on a short timescale between various metal ions. In the same source the extraction voltage could also be controlled from pulse to pulse, thus offering the additional versatility of ion energy control of each of the different metal ion beams formed from each sub-cathode. By means of this rapid control of ion species and ion energy, it was possible to carry out ion implantation of multiple species with experimenter-

Figure 13.15 Schematic of the arc and filter region of the NPI broad beam source, which incorporates a large area dc cathode and a linear “venetian blind” macroparticle filter.


determined control of the implantation depths and so to achieve excellent overlap of the implantation profiles for each species. In another source version [67], the arc cathode was a large rectangular block over which the cathode spots were guided magnetically; the arc source was operated dc. Cathodes with operating areas of 300 and 1500 cm2 were investigated. A macroparticle filter of novel design [68] was also included in the plasma drift region. A schematic of the source is shown in Figure 13.15. The extractor voltage was repetitively pulsed, with pulse duration 400 ls and repetition rate up to 50 pps. Because of the motion of the cathode spots on the large area cathode, the ion beam emission region on the extractor surface also moves, and on a long-term integrated basis a very broad beam ion source is effected. 13.7.4

GSI Varis Sources

Vacuum arc ion sources have been used successfully at the GSI heavy ion accelerator research center at Darmstadt, Germany, for injecting several kinds of metal ions into a linear accelerator and synchrotron, including uranium. For accelerator application the quality of each individual ion beam pulse is important, as opposed to the case for ion implantation application. Beam noise and/or pulse shape variation from pulse to pulse can lead to poor accelerator performance. Thus this source has incorporated a number of the developments referred to in the preceding to enhance the high ion charge state fraction, to reduce the beam noise, and to improve the pulse-to-pulse beam shape reproducibility [54–56].

Figure 13.16 GSI Varis ion source.1 – Cathode flange, 2 – Cathode, 3 – Anode (stainless steel), 4 – 10 SmCo cusp magnets, 5 – Coil I, 6 – Coil II, 7 – Plasma electrode and grid, 8 – Screening electrode, 9 – Ground electrode, 10 – Insulators.

279

280


A schematic of a recent GSI source, the Varis, is shown in Figure 13.16. The cathode assembly holds 17 individual cathodes, each 17 mm long and 5.7 mm in diameter. The stainless steel anode, located at a distance of 15 mm from the cathode, has a central aperture 15 mm in diameter. Magnetic field coils are located external to the vacuum chamber. Two stainless steel grids are installed to reduce the beam fluctuation level and to reduce plasma density. A multi-aperture accel–decel extraction system (13 holes, each 3 mm in diameter, aspect ratio 0.5) is used to form the ion beam. As an example of source performance, 25 mA of U4+ ion current with an energy of 2.2 keV u–1 (156 mA full beam, or 170 mA cm–2, electrical current) was measured at the entrance of the RFQ, for a typical extraction voltage of 32 kV, with a high fraction (67% electrical current fraction) of U4+ ions, a post-analysis U4+ beam noise level of less than €10%, and good beam pulse shape reproducibility. This ion source has proven its capability in extended tests at the GSI high current injector and has been put into regular operation for the generation of high current metal ions for the GSI accelerator facility. This vacuum arc ion source has provided metal ion beams of current an order of magnitude greater than previously possible. The GSI sources have been described in detail elsewhere [57, 69, 70]. 13.7.5

Other Versions and Variants

A series of vacuum arc ion sources have been developed at the Institute of Low Energy Nuclear Physics at Beijing Normal University (BNU) and incorporated into ion implantation facilities for metallurgical surface modification [71, 72]. BNU implantation machines have been installed at a number of industrial and government sites throughout China. A 50 mA average beam current implanter is shown in Figure 13.17. Beam pulses are 2.4 ms long at a repetition rate of up to 25 pps, with a cathode lifetime of over 8 h. The beam diameter at the implantation target is 50 cm. A 100 mA average current source is under development. Industrialization of vacuum

The Beijing Normal University 50 mA average current vacuum arc ion source based implanter. The ion source can be seen in the top center of the photo.

Figure 13.17


Figure 13.18 Vacuum arc ion source utilizing a 50 cm diameter extractor. (Courtesy of Lawrence Berkeley National Laboratory.)

arc ion source-based implanters for non-semiconductor surface modification has been accomplished impressively by the Beijing group. A vacuum arc source embodiment making use of a 50 cm diameter extractor has been demonstrated at LBNL [11]. The large extractor was mated with several different plasma gun assemblies; a photograph of one such setup is shown in Figure 13.18. At 50 kV extraction voltage and an arc current of 300 A, a beam current of about 7 A was obtained during the beam pulse flat-top and a peak current at the early-time current overshoot part of the pulse of about 20 A. This extracted beam current is 2.3% during the current flat-top and 4.6% during the current overshoot. Since the total metal plasma production is about 10% of the arc current (see Section 13.3), and given the 40% extractor grid optical transmission, this implies that an impressively high fraction of the metal plasma is converted into ion beam – the plasma expansion chamber and large area extractor configuration are very efficient in terms of plasma utilization. The suitability of very large area and high power ion sources for high throughput industrial application has been discussed [73].

Figure 13.19

MicroMevva ion source. (Courtesy of Lawrence Berkeley National Laboratory.)

281

282


At the other end of the size spectrum are the “MicroMevvas” that have been made by the Berkeley group [11, 74]. Several versions were made and tested, and a photograph of the smallest of these is shown in Figure 13.19; it is about 1.5 cm in diameter and 6 cm in overall length. Beam currents of 15 mA at 15 kV were produced. Power dissipation is an obvious concern, and these sources are suited only for low duty cycle operation.

13.8

Conclusion

The vacuum arc ion source has proven itself to be an important addition to the spectrum of ion sources available to the experimenter. A widespread community of source developers and users has evolved, and a rich literature in the field is growing. The primary application has turned out to be for non-semiconductor ion implantation – surface modification of metals, polymers, and other materials; the source and beam parameters are well matched to the requirements. Large sources have been developed and incorporated into implanters, and inroads have been made into commercialization of the technology. A secondary application is for particle accelerator injection. Considerable research effort has been invested in the development of techniques for reduction of beam noise and improvement in beam pulse shape reproducibility, and this work has seen a high degree of success. A vacuum arc ion source has been used for injection of several different kinds of metal ion beams, including uranium, into the GSI heavy ion synchrotron at record high current levels. Further development in several areas would lead to better source performance and characteristics. These areas include, for example, improved triggering methods, extended source lifetime (between necessary down-time for cathode maintenance), high charge state elevation and control, and further reduction of beam noise, among others. All of these needs are receiving experimental attention, and it is to be expected that the future will witness some exciting advances, as has the past.

References

References [1] R.K. Wakerling and A. Guthrie, Editors, Elec-

[2] [3] [4] [5]

[6]

[7] [8] [9]

[10] [11] [12]

[13]

[14]

[15] [16] [17]

tromagnetic Separation of Isotopes in Commercial Quantities, National Nuclear Energy Series, (USAEC, Washington D.C., 1951), p. 324. P.D. Prewett and R. Holmes, J.Phys. E: Sci. Instrum. 12, 179 (1979). R.J. Adler and S.T. Picraux, Nucl. Instrum. Methods Phys. Res. B 6, 123 (1985). S. Humphries, Jr., M. Savage and D.M. Woodall, Appl. Phys. Lett. 47, 468 (1985). S. Humphries, Jr., C. Burkhard, S. Coffey, G. Cooper, L.K. Len, M. Savage, D.M. Woodall, H. Rutkowski, H. Oona and R. Shurter, J. Appl. Phys. 59, 1790 (1986). S. Humphries, Jr., C. Burkhart and L.K. Len, in The Physics and Technology of Ion Sources, edited by I.G. Brown (Wiley, New York, 1989), First Edition, p. 397. I.G. Kesaev, Cathode Processes in Electric Arcs, (Nauka, Moscow, 1968) (in Russian). A.A. Plutto, Zh. Eksp. Teor. Fiz. 39, 1589 (1960) (in Russian). E.I. Revutskii, G.M. Skoromnyi, Yu.F. Kulygin amd I.I. Goncharenko, in Proc. Soviet Conf. Charged Particle Accelerators, Moscow, Oct. 9-16, 1968, edited by A.A. Vasiliev (USAEC, Washington D.C., 1969), p. 447. I.G. Brown, J.E. Galvin and R.A. MacGill, Appl. Phys. Lett. 47, 358 (1985). I.G. Brown, Rev. Sci. Instrum. 65, 3061 (1994). G.P. Bazhenov, S.P. Bugaev, G.P. Erokhin, V.N. Kiselev, A.E. Ligatchev, S.M. Chesnokov and A.V. Ianchiuck, in Proceedings of the 5th All-Union Symposium on High Current Electronics, Tomsk, Part II, p. 93 (1984) (in Russian). S.P. Bugaev, A.G. Nikolaev, E.M. Oks, P.M. Schanin and G.Yu.Yushkov, Rev. Sci. Instrum. 65, 3119 (1994). N.M. Arzubov, G.P. Isaev and A.I. Ryabchikov, in Proceedings of the 6th All-Union Symposium on High Current Electronics, Tomsk, Part III, p. 184 (1986) (in Russian). A.I. Ryabchikov, S.V. Dektjarev and I.B. Stepanov, Rev. Sci. Instrum. 65, 3126 (1994). A.M. Tolopa, Rev. Sci. Instrum. 65, 3134 (1994). P.J. Evans, G.C. Watt and J.T. Noorman, Rev. Sci. Instrum. 65, 3082 (1994).

[18] H. Zhang, X. Zhang, F. Zhou, S. Zhang, Q. Li

and Z. Han, Rev. Sci. Instrum. 65, 3088 (1994). [19] B.H. Wolf, H. Emig, D. Rueck and P. Spdtke, Rev. Sci. Instrum. 65, 3091 (1994). [20] J. Sasaki, K. Hayashi, K. Sugiyama, O. Ichiko and Y. Hashiguchi, Surf. Coat. Technol. 51, 166 (1992). [21] V.A. Batalin, J.N. Volkov, T.V. Kulevoy and S.V. Petrenko, Rev. Sci. Instrum. 65, 3104 (1994). [22] M.M. Katz, L.N. Kondratiev, N.N. Pomelov and A.D. Rogal, Rev. Sci. Instrum. 65, 3101 (1994). [23] I.G. Brown and E.M. Oks, IEEE Trans. Plasma Sci. 25, 1222 (1997). [24] G.A. Mesyats, Cathode Phenomena in a Vacuum Discharge, (Nauka, Moscow, 2000). [25] R.L Boxman, P.J Martin and D.M Sanders, Editors, Handbook of Vacuum Arc Science and Technology (Noyes, New York, 1995). [26] J.M. Lafferty, Editor, Vacuum Arcs – Theory and Application (Wiley, New York, 1980). [27] P. Spdtke, H. Emig B.H. Wolf and E.M. Oks, Rev. Sci. Instrum. 65, 3113 (1994). [28] A. Anders, I.G. Brown, R.A. MacGill and M.R. Dickinson, J.Phys.D: Appl. Phys. 31, 584 (1998). [29] S. Anders, S. Raoux, K. Krishnan, R.A. MacGill and I.G. Brown, J. Appl. Phys. 79, 6785 (1996). [30] R.A. MacGill, A. Vizir and I.G. Brown, Rev. Sci. Instrum. 71,672 (2000). [31] I.G. Brown and X. Godechot, IEEE Trans. Plasma. Sci. 19, 713 (1991). [32] I.G. Brown, J.E. Galvin, R.A. MacGill and R.T. Wright, Rev. Sci. Instrum. 58, 1589 (1987). [33] A. Anders, Phys. Rev. E 55, 969 (1997). [34] I.G. Brown, P. Spdtke, H. Emig, D. Rueck and B. Wolf, Nucl. Instrum. Methods Phys. Res. A 295, 12 (1990). [35] A. Anders and R. Hollinger, Rev. Sci. Instrum. 73, 732 (2002) [36] E.M. Oks, IEEE Trans. Plasma Sci. 30, 202 (2002). [37] A. Anders, I. Brown, R. MacGill and M. Dickinson, Rev. Sci. Instrum. 67, 1202 (1996). [38] A. Anders, I.G. Brown, R.A. MacGill and M.R. Dickinson, IEEE Trans. Plasma Sci. 25, 718 (1997).

283

284


[39] S. Humphries Jr. and G. Rutkowski, J. Appl. [40] [41] [42]

[43]

[44]

[45]

[46] [47]

[48] [49]

[50]

[51] [52] [53] [54]

[55] [56]

Phys. 67, 3223 (1989). I.G. Brown, Nucl. Instrum. Methods Phys. Res. B 37/38, 68 (1989). F.J Paoloni and I.G. Brown, Rev. Sci. Instrum. 66, 3855 (1995). E.M. Oks, I.G. Brown, M.R. Dickinson, R.A. MacGill, H. Emig, P. Spdtke and B.H. Wolf, Appl. Phys. Lett. 67, 200 (1995). E.M. Oks, A. Anders, I G. Brown, M.R. Dickinson and R.A. MacGill, IEEE Trans. Plasma Sci. 24, 1174 (1996). A. Anders, G. Yushkov, E. Oks, A. Nikolaev and I.G. Brown, Rev. Sci. Instrum 69, 1332 (1998). A.S. Bugaev, E.M. Oks, G.Yu. Yushkov, A. Anders and I.G. Brown, Rev. Sci. Instrum. 71, 701(2000). G.Yu. Yushkov, E.M. Oks, A. Anders and I.G. Brown, J. Appl. Phys. 87, 8345 (2000). V.A. Batalin, A.S. Bugaev, V.I. Gushenets, A. Hershcovitch, B.M. Johnson, A.A. Kolomiets, R.P. Kuibeda, T.V. Kulevoy, E.M. Oks, V.I. Pershin, S.V. Petrenko, D.N. Seleznev and G. Yu. Yushkov, J. Appl. Phys. 92, 2884 (2002). M. Cavenago and A. Vassiliev, Rev. Sci. Instrum., 67, 1207 (1996). A.V. Vodopyanov, S.V. Golubev V.G. Zorin, S. Razin, A. Vizir, A. Nikolaev, E. Oks and G. Yu. Yushkov, Rev. Sci. Instrum. 75, 1888 (2004). A.G. Nikolaev, G.Yu. Yushkov, E.M. Oks, R.A. MacGill, M.R. Dickinson and I.G.Brown, Rev. Sci. Instrum. 67, 3095 (1996). A. Anders and J. Kwan, Nucl. Instrum. Methods Phys. Res. A 464, 569 (2001). E. Oks, P. Spdtke, H. Emig and B.H. Wolf, Rev. Sci. Instrum. 65, 3109 (1994). E. Oks, G. Yushkov, I. Litovko, A. Anders and I. Brown, Rev. Sci. Instrum. 73,735 (2002). R. Hollinger, M. Galonska, F. Heymach and P. Spadtke, in Proceedings of the XXth International Symposium on Discharge and Electrical Insulation in Vacuum (ISDEIV), Tours, France, July 2002 (IEEE Catalog number 02CH37331, ISBN 0-7803-7394-4) p. 447. R. Hollinger, M. Galonska, F. Heymach, and P. Spdtke, GSI Report 2002-03 April, 2002. M. Galonska, F. Heymach, R. Hollinger, and P. Spdtke,in Emerging Applications of Vacuum Arc Produced Plasma, Ion and Electron Beams, NATO Science Series, Vol. 88, edited by

E.M. Oks and I.G. Brown (Kluwer, Netherlands, 2002), p. 123. [57] E.M. Oks, G.Yu. Yushkov, P.J. Evans, A. Oztarhan, I.G. Brown, M.R. Dickinson, F. Liu, R.A. MacGill, O.R. Monteiro and Z. Wang, Nucl. Instrum. Methods Phys. Res. B 127/128, 782 (1997). [58] E.M. Oks, Rev. Sci. Instrum. 69, 776 (1998). [59] H. Reich, P. Spdtke and E.M. Oks, Rev. Sci. Instrum. 71, 707 (2000). [60] G. Yushkov and I.G. Brown, Rev. Sci. Instrum. 75, 1582 (2004). [61] G. Yushkov and A. Anders, IEEE Trans. Plasma Sci. 26, 220 (1998). [62] E.M. Oks and I.G. Brown, IEEE Trans. Plasma Sci. 26, 1562 (1998). [63] I.G. Brown, IEEE Trans. Plasma Sci. 21, 537 (1993). [64] I.G. Brown, A. Anders, M.R. Dickinson, R.A. MacGill and O.R. Monteiro, Surf. Coat. Technol. 112, 271 (1999). [65] A.S. Bugaev, V.I. Gushenets, A.G. Nikolaev, E.M. Oks, K.P. Savkin, P.M. Schanin, G.Yu. Yushkov and I.G. Brown , in Emerging Applications of Vacuum Arc Produced Plasma, Ion and Electron Beams, NATO Science Series, Vol. 88, edited by E.M. Oks and I.G. Brown (Kluwer, Netherlands, 2002), p. 79. [66] A.S Bugaev, V.I. Gushenets, A.G.Nikolaev, E.M Oks and G.Yu.Yushkov, Izvestiya Vysshikh Uchebnykh Zavedenii, Fizika, 43, (no.2), Consultants Bureau, 21 (2000). [67] A.I. Ryabchikov, I.B. Stepanov, S.V. Dektjarev, E.I. Lukonin and I.A. Shulepov, Rev. Sci. Instrum. 71, 704 (2000). [68] A.I. Ryabchikov and I.B. Stepanov, Rev. Sci. Instrum. 69, 810 (1998). [69] R. Hollinger et al, Rev. Sci. Instrum. 75, 1595 (2004). [70] P. Spdtke, J. Bossler, M. Galonska, F. Heymach, R. Hollinger, R. Iannucci, R. Lang, K.D. Leible, K. Tinschert, Ion Source Development and Operation, GSI Scientific Report, ISSN 0174-0814, May 2003) [71] H. Zhang, X. Zhang, F. Zhou, S. Zhang, Q. Li and Z. Han, Rev. Sci. Instrum. 65, 3088 (1994). [72] T. Zhang, H. Zhang, C. Ji, X. Zhang, Y. Wu, F. Ma, H. Liang, H. Shou and J. Shi, Surf. Coat. Technol. 128-129, 1 (2000). [73] I.G. Brown, J. Vac. Sci. Technol. A 11,1480 (1993).

285

14

Negative Ion Sources Junzo Ishikawa

14.1

Introduction

A negative ion is formed by physical mechanisms in which an atom or molecule captures an electron into an electron affinity level from an electron source. Negative ion production mechanisms can be classified in terms of the source of electrons, i.e., electrons in the conduction band of a metal, plasma electrons, and free particles. The corresponding mechanisms are then referred to as “surface effect”, “volume production” and “charge transfer”, respectively. In the surface effect, an electron at the Fermi level in the conduction band of a metal shifts by tunneling to the electron affinity level of an atom or molecule which is approaching the metal surface. The probability of the electron shift is greatly enhanced as the effective work function of the metal surface is lowered. This mechanism is used for both light and heavy negative ion production, and also takes place in particle reflection and sputtering phenomena. On the other hand, in the volume production process, a highly vibrationally-excited hydrogen molecule effectively captures a low energy plasma electron to form a negative hydrogen ion through the dissociative electron attachment process. By use of a magnetic filter the plasma region in which dissociative electron-attachment takes place should be separated from the plasma region for hydrogen molecular excitation where relatively high energy electrons are needed for the excitation. In the charge transfer mechanism, alkali metals or alkaline earth metals are used as electron donors for negative ion production because of their large charge transfer cross sections. 14.2

Surface Effect Negative Ion Sources 14.2.1

Negative Ion Production by Surface Effect

An atom emitted energetically from a low work function metal surface may leave, with reasonably high probability, in the form of a negative ion. In this case an elecThe Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

286

14 Negative Ion Sources

Figure 14.1 Illustration of the mechanism of secondary negative ion emission by light particle reflection and heavy particle sputtering

tron shifts from the Fermi level of the metal surface to the electron affinity level of the atom by tunneling. Two processes are involved in atom ejection from metal surfaces: light particle reflection and heavy particle sputtering by positive ions. This negative ion production mechanism is called secondary negative ion emission. Figure 14.1 illustrates the mechanism of secondary negative ion emission by light particle reflection and heavy particle sputtering [1]. The negative ion current generated at the metal surface is expressed as the product of the positive ion current for reflection or sputtering, I+, the reflection yield or the sputtering yield, A, and the negative ion production efficiency, g–. Before negative ions are extracted from the ion production region, they suffer electron detachment collisions with gas particles within the ion production region. The negative ion current is decreased by a factor exp(–n0Lrd) during the transport from the region, where n0, L and rd are the neutral gas particle density, the transport length, and the electron detachment cross section for the negative ion, respectively. Thus the extractable negative ion current produced through this process, I –, is given by þ I ¼ I Ag exp n0 Lr d

(14.1)

Reflected or sputtered particles have a velocity distribution function f(v). Thus the negative ion production efficiency g– is given by the integral of the product of the negative ion production probability P–(v) and the velocity distribution function, g

¼

R

P ðvÞf ðvÞdv

where v is the emitted particle velocity.

(14.2)

14.2 Surface Effect Negative Ion Sources

14.2.1.1 Negative Ion Production Probability The negative ion production probability has been analyzed theoretically by Rasser et al. [2]. They introduced a fairly simple equation as follows, 2 pðuEa Þ (14.3) P ðvÞ ¼ exp p 2avcosh

where u is the work function of the metal surface, Ea is the electron affinity, h is the ejection angle of the emitted particle, and a is the decay factor. The decay factor for hydrogen is evaluated to be 2 10–5 eV m–1 from the experimental data of van Wunnik et al. [3]. The above equation is valid when the normal velocity of the particle emitted from the metal surface (vcos h) is greater than 104 m s–1. Thus this equation can be used for the case where a light particle such as hydrogen is emitted with a relatively high velocity due to reflection. In the case of particle sputtering, the normal velocity of the emitted particle is sometimes lower than 104 m s–1, especially for heavy particles. Then Eq. (14.3) is no longer valid and needs to be modified. A modified semi-empirical equation for the negative ion production probability has been given by Ishikawa et al. [4], in which an effective local temperature Teff proportional to the mass number of the emitted particle (M) was introduced. " # 2 ðuEa Þ P ðvÞ ¼ exp (14.4) p ð2avcoshÞ=pþkTeff where kTeff = 0.073 + 2.0 10–3 M and k is Boltzmann’s constant. The negative ion production probability depends strongly on the work function of the metal surface and decreases sharply with increasing work function. A low work function metal surface is therefore essential for high efficiency negative ion production by secondary negative ion emission. Lowered Work Function by Cesium Coverage In general, the work function of a metal surface is lowered when the metal surface is covered by low work function material such as an alkali metal or alkaline earth metal, and its minimum value is lower than the work function of the adsorbed material. Cesium has the lowest work function (1.81 eV) of all elements, and thus is frequently used to obtain a low work function metal surface for secondary negative ion emission. Alton has reported on the work function of cesium covered surfaces for various metals and proposed a semi-empirical equation for the minimum work function of a cesium covered metal surface [5]: 14.2.1.2

umin ¼ 0:62ðVi þ Ea Þ 0:24u0 [eV]

(14.5)

where u0 is the inherent work function of the metal surface, Vi and Ea are the first ionization potential and electron affinity of the adsorbed material on the metal, respectively; for the case of cesium, the sum of Vi and Ea is 4.35 eV, which is the smallest of all elements. Graham has measured the work function of cesium covered metal surfaces as a function of cesium coverage [6]. Figure 14.2 shows the gen-

287

288


Figure 14.2

General behavior of the surface work function for cesium coverage on the metal surface.

eral behavior of the surface work function as a function of cesium coverage. The inherent work function of metal surfaces, uo, is usually between 4.5 and 6 eV. In the low cesium coverage region, the work function decreases with increasing cesium coverage and reaches a minimum value umin (1.37 to 1.76 eV) at a coverage of about hmin = 0.6 (0.6 atomic layers). Then it increases with further increase in cesium coverage up to a value of l.56 to 2.15 eV (hmono = 1). Above room temperature cesium atoms no longer adhere to the surface because the Cs–Cs bonding is very weak, and the work function does not change. However if the metal surface is cooled too much by a coolant such as liquid nitrogen, a cesium multilayer is formed. The work function then approaches the work function of bulk cesium, uCs. Metal atom–Cs bonding is usually much stronger than Cs–Cs bonding, and is broken at a temperature of several hundred C. Therefore a cesium monolayer on a metal surface (cesiated surface) can be formed over quite a wide temperature range of between room temperature and several hundred C. This wide temperature margin for highly efficient negative ion production shows that secondary negative ion emission is a very effective method for the production of negative ions. In secondary negative ion emission via reflection of light ions on the cesiated surface, the cesium atom on the metal surface is not sputtered by the light ion because the mass difference is quite large. In this case optimal surface conditions can be maintained during operation without a steady cesium atom flux supply, once the minimum work function is attained on the cesiated surface. However, in secondary negative ion emission by the sputtering method, the cesium atoms on the metal surface are sputtered together with metal atoms. In this case, a continuous cesium atom supply to the metal surface is needed in order to maintain the optimal surface condition. Electron Detachment Cross Section Negative ions are converted to neutral particles or positive ions due to single-electron-detachment collisions or double-electron-detachment collisions with gas particles, respectively. The electron detachment cross section rd indicated in Eq. (14.1) 14.2.1.3


Figure 14.3

Electron detachment cross sections for various negative ions and target gases.

refers to the sum of the single- and double-electron-detachment cross sections. Figure 14.3 shows the electron detachment cross sections for various negative ions (B, C, O and Si) and target gases (He, Xe, H2 and N2) as a function of negative ion energy (5 to 50 keV) [7, 8]. The single-electron-detachment cross section r-10 is almost independent of negative ion energy, but weakly dependent on the physical parameters of the negative ion and the target gas. It ranges from 0.6 10–15 to 2.5 10–15 cm2. On the other hand, the double-electron-detachment cross section r–11 is smaller by more than an order of magnitude than the single-electron-detachment cross section. Therefore, the value of rd can be considered to be that of the single-electron-detachment cross section. Since the value of the electron detachment cross section is relatively large, a negative ion beam can be seriously affected by electron detachment collisions. If a negative ion beam is transported through a region with a gas pressure of 10–2 Torr (a pressure typical of plasma production chambers used for conventional positive ion sources) and a length of 10 cm (typical source size), the survival rate for negative ions is only about 1/1000. This means that almost all of the negative ions are lost during passage through the plasma production chamber under typical discharge conditions. Therefore, in negative ion sources the gas pressure in the plasma production chamber should be much lower than 10–2 Torr. If the gas pressure is 10–3 Torr or 10–4 Torr, the survival percentages are 50% or 93%, respectively. A low pressure discharge is essential for negative ion production. The gas pressure during the transport of a negative ion beam after extraction from an ion source is also very important, and should be lower by an order of magnitude in comparison with positive ion beam transport.

289

290


14.2.1.4

Experimental Data for Negative Ion Production Efficiency

(a) Reflection of light particles

Granneman et al. have measured the negative ion production efficiency for secondary negative ion emission for the case of lithium as the light particle, [9]. Figure 14.4 shows the lithium negative ion production efficiency as a function of ion velocity normal to the tungsten surface when lithium ions of various energies are irradiated at an incident angle of 75 onto a cesiated W(110) surface with a minimum work function. The efficiency depends on the ion velocity normal to the tungsten surface but not on the incident ion energy, and this dependence is in good agreement with Eq. (14.3).

Figure 14.4 Negative lithium ion production efficiency as a function of ion velocity normal to the tungsten surface [9].

Wada et al. measured the energy distribution of backscattered hydrogen ions from a cesiated metal surface (converter) which is placed in the generated hydrogen plasma and is negatively biased at a voltage of –Vpc, as indicated in Figure 14.5 [10]. In the hydrogen plasma there exist H+, H2+, and H3+ ions. These ions bombard the converter surface and are then backscattered. Neutral hydrogen atoms H0 are also absorbed on the converter surface. These four kinds of particles are changed to negative ions on the converter surface due to the secondary negative ion emission process. These particles have energy equal to the sum of the incident positive ion energy per particle and the acceleration energy of the negative ion. Therefore backscattered particles H+, H2+, H3+ and H0 have different negative ion energies of 2, 1.5, 1.33 and l eVpc, respectively. Figure 14.5 shows the hydrogen negative ion energy distribution as a function of the backscattered polar angle where an angle normal to the surface is taken as 0. In the normal direction (polar angle 0), the main particle for negative ion production is H0 because the distribution has a peak at about l eVpc. However in the high polar angle direction, H3+ or H2+ particles mainly contribute to the negative ion production because the distribution peak shifts to 1.3 or 1.5 eVpc.


Thus, the particle with high incident energy has a relatively high transverse energy by collision on the surface, and this is the reason why this type of ion source has a relatively high beam emittance.

Figure 14.5

Energy distribution of backscattered hydrogen ions from a cesiated metal surface [10].

(b) Sputtering of heavy particles

Negative ion production efficiencies in secondary negative ion emission by sputtering for various heavy particles have been measured by Ishikawa et al. [4]. Particles sputtered from a cesiated target by 10 to 15 keV xenon ion bombardment contain negative ions together with neutral particles, and the fluxes of negative ions and of

Figure 14.6 Measurement system for negative ion production efficiency by heavy-particle sputtering on a cesiated surface.

291

292


the total sputtered particles were separately measured using the system shown in Figure 14.6. The negative ion flux density was measured by a EB mass separator and the total sputtered particle flux density was measured by RBS analysis for sputtered particles deposited on a silicon substrate. The ratio of negative ions to total sputtered particles, i.e., the negative ion production efficiency, was measured for various targets under the condition of various cesium supply and target temperatures. Table 14.1 shows the maximum negative ion production efficiencies for typical heavy particles. For most of the metallic elements, the efficiencies are 10% or more. This shows that heavy negative ions can be effectively produced through the secondary negative ion emission process by sputtering. Table 14.1

Element

Maximum negative ion production efficiencies. C

Efficiency (%) 18.3

Si

Cu

Ge

W

15.6

12.1

13.6

9.4

14.2.2

Surface Effect Light Negative Ion Sources 14.2.2.1 History of Source Development The off-axis duoplasmatron [11] was an early source in which effective negative hydrogen ion production by the surface effect was found. In this source the central axis of the anode and ion extraction electrodes was intentionally shifted by about l mm from the central axis of the intermediate electrode. Several tens of microamperes of negative hydrogen ion current were extracted from the generated plasma, however at the same time a few milliamperes of electron current was also extracted. From this fact it was concluded that dense negative ions could be produced near or on the electrode surface. In order to enhance negative ion production by the surface effect, the hol1ow discharge duoplasmatron (HDD) was developed [12, 13]. In this source, an isolated metal pole was positioned within the plasma on the central axis so as to increase the metal surface area for negative hydrogen ion production. Milliamperes of negative hydrogen ion currents were extracted from this source, e.g., 9 mA (0.25 A/cm2): no cesium, of pulsed negative hydrogen ion current, and 18 mA (0.57 A/cm2): cesium supply, were extracted. This source development clarified that negative hydrogen ions can be effectively produced by the surface effect, and also that production is especially enhanced on a cesiated electrode surface. Next, high current negative hydrogen ion sources were developed in which the reflection of positive hydrogen ions from a large cesiated metal surface was used, such as the magnetron type [13–18], PIG type [19–21], multicusp type [22–24] and duopigatron [25] type. Amperes of beam current were produced. In the development of these sources, the mechanism of reflection type light negative ion production was investigated in depth.


14.2.2.2 Recent Reflection Type Negative Ion Sources The reflection type negative ion source has a metal electrode with a low work function, i.e., a converter electrode, which is placed in the generated plasma and is negatively biased with respect to the plasma. The bombarding energy for positive ion reflection is determined by this bias voltage. The sources can be divided into two groups. In the first group, the converter electrode is one of the electrodes for plasma generation. The magnetron type and PIG type without a converter are classified in this category. In the other group, the converter electrode is separated from the plasma generation electrodes, and then the converter voltage can be controlled independently of discharge conditions. The PIG type, multicusp type and duopigatron type with converter are classified in this latter group. (a) Magnetron type negative hydrogen ion source

The magnetron type negative hydrogen ion source [13–18] is also called a surface plasma negative ion source by Bel’chenko et al. [14–16]. In this ion source, firstly the negative ion current was increased by a large margin in a cesiated mode of operation. Figure 14.7 shows the basic structure of the source [13]. The source has a spool-shaped cathode which is surrounded by an anode. Between these two electrodes a cold cathode discharge is maintained by the assistance of a perpendicular magnetic field of l to 2 kG at a relatively high hydrogen gas pressure (about 1 Torr). Negative ions are produced on the cathode surface by the surface effect, and then accelerated by the cathode fall potential difference to enter the grooved region. In the groove region, dense hydrogen gas particles together with the discharge plasma particles co-exist, and thus the high velocity accelerated negative ions suffer resonant charge exchange collisions with hydrogen gas particles to produce slow negative ions. Both slow and fast negative ions are extracted from the slit aperture. In the cesiated mode of operation, 880 mA (1 ms, 1 Hz) of negative hydrogen ion current was extracted at a discharge current of 450 A. Many versions of magnetron type negative ion sources have been developed, including the multigrooved type [26].

Figure 14.7

Basic structure of magnetron type negative hydrogen ion source [13].

293

294


(b) PIG type negative hydrogen ion source

(i) PIG source without converter The structure of the PIG source without a converter is shown in Figure 14.8 [20]. In this source a PIG discharge at a relatively high hydrogen gas pressure is used and thus the negative ion production mechanism is very similar to the magnetron type. In the PIG electrode structure the cathode surface does not face the ion extraction aperture, and thus fast negative ions are not extracted from it. Only slow negative ions produced through resonant charge exchange are extracted, and several hundred milliamperes of pulsed negative hydrogen ion current can be obtained. (ii) PIG source with converter In order to directly extract negative ions produced on the metal surface in the PIG discharge plasma, a PIG type source with a converter was developed. The structure of the source is illustrated in Figure 14.9 [21]. In this source a PIG discharge at a

Figure 14.8

PIG type negative hydrogen ion source without converter [20].

Figure 14.9

PIG type negative hydrogen ion source with converter [21].


relatively low hydrogen gas pressure (4 10–3 Torr for hot PIG discharge and 1 10–1 Torr for cold PIG discharge) is used to prevent charge exchange collisions of negative ions. A 500 to 600 mA pulsed negative hydrogen ion current was extracted from a hot PIG discharge source. (c) Multicusp type negative hydrogen ion source

In the multicusp type negative hydrogen ion source with converter, the negative ions produced on the converter surface are directly extracted without collisions. The ion source developed by Ehlers et al. [22–24] is shown in Figure 14.10. A dense plasma (~1012 cm–3) was generated by an arc discharge in a multicusp type plasma chamber at a relatively low gas pressure (less than 10–3 Torr) to realize both high current irradiation of positive ions to the converter and low electron detachment collision probability for negative ions. The shape of the plasma chamber is an elliptic cylinder, and the chamber surface is surrounded with Sm–Co permanent magnets to form cusp magnetic fields. Eight tungsten filaments were placed in the zero magnetic field region for the arc discharge (90 V, 100 A). A converter made of molybdenum (8 25 cm2), which was negatively biased, was placed near the center of the plasma chamber and neutral cesium particles were supplied to its surface. More than 1 A of negative hydrogen ion current was extracted using a cesiated converter in dc mode.

Figure 14.10

Multicusp type negative hydrogen ion source [24].

295

296


14.2.3

Surface Effect Heavy Negative Ion Sources History of Source Development Heavy negative ions were originally required for tandem accelerator injection. The first sputter type heavy negative ion source was developed by Mueller and Hortig as shown in Figure 14.11 [27]. The source had a ring target which was rotated at 180 to 400 turns per minute. On one side, neutral cesium particles were sprayed onto the target surface. On the other side, a relatively high energy positive ion beam was irradiated onto the cesiated target. When a Kr+ ion beam (10 to 30 keV) was used for sputtering, several microamperes of heavy ions (C–, Cr–, Cu–, Ag–, Au–, etc.) were extracted as described in Table 14.2. 14.2.3.1

Figure 14.11

Sputter type heavy negative ion source developed by Mueller and Hortig [27].

In a second stage of development (a few years after the Muller and Hortig source), two important concepts for the sputter type heavy negative ion source were established: beam sputter type and plasma sputter type. The former was the universal negative ion source (UNIS) developed by Middleton and the latter was the Aarhus negative ion source (ANIS) developed by Andersen. The structure of the UNIS [28–30] is shown in Figure 14.12. In the UNIS, a cesium ion beam with an energy of 20 to 30 keV was irradiated in a high vacuum onto a cone shaped sputtering target for both sputtering and as a supply of cesium. High energy ions give a high sputtering yield, and then the quantity of the cesium remaining on the sputtering target is relatively low. Therefore, this source is suited for negative ion production of low sputtering yield materials such as carbon rather


than high sputtering yield materials such as copper. A wide variety of negative ions (46 elements) at a current of sub-microampere to tens of microamperes could be produced. The negative ion currents are described in Table 14.2. After the UNIS, many kinds of modified UNIS such as an inverted type [31] and a reflected beam mode type [32] were also developed. Table 14.2

Negative ion currents obtained by typical sputter type heavy negative ion sources.

Negative ion Source

Mueller & Hortig

UNIS

C– C 2– Cr– Cu– Ag– Au–

B– C– Al– Si– P– Ge– As– In– Sb– Te– Ti– Ni– Cu– Nb–

2.2 3.4 0.1 6.0 14 12

Negative ion current (microamperes)

Positive ion Energy Current

Figure 14.12

Kr+ or Ar+ 10–30 keV

ANIS

2.3 50 0.2 27 0.6 1.1 1.0 0.03 0.32 1.8 0.005 6.6 7 0.027

Cs+ 20–30 keV 1–1.5 mA

Li– B– BO– C– C2– Al– Ti– Ni– Cu– Nb– Mo– Ag– Ta– Au–

VNIS

1.0 0.0 2.5 20 15 2.3 0.9 55 51 1.0 0.14 36 3 80

Be– B– C– F– Al– Si– P– S– Fe– Cu– Ni– Pt– Au–

NIABNIS

0.1 35 300 250 6 250 25 200 4 150 150 150 200

Cs++(Xe+ orAr+)Cs+ 1–2 keV 5–10 keV ~ 10 mA 5–15 mA

C– C2– B– B2– Cu– Al– Al2– Sb– Sb2– Si– As– Ni– Nb– Mo– Ta–

735 ~200 15 20 320 6.0 7.7 0.6 2.1 170 2.4 59 1.6 1.0 0.4

Cs+ 10–25 keV 1–10 mA

Universal negative ion source (UNIS) [28].

The ANIS is illustrated in Figure 14.13 [33, 34]. In this source a sputter target with a spherical structure was placed in a PIG discharge xenon plasma, and was negatively biased to about 1 kV. Neutral cesium particles were supplied to the

297

298


plasma chamber. Then Xe positive ions were bombarded onto the cesiated sputter target surface. Negative ions produced on the target were focused to a small extraction aperture. Thus this source has the typical structure of a plasma sputter type heavy negative ion source. However the extracted negative ion current from this source was relatively low, ranging from several microamperes to several tens of microamperes, as described in Table 14.2. This is because, in this source, the PIG discharge gas pressure was relatively high (2.3 10–1 Torr), and thus most of the negative ions produced on the sputter target could not survive during transport to the extraction aperture due to electron detachment collisions with gas particles. Since the cesium particles are supplied sufficiently to the target surface, this source is suitable for negative ion production of high sputtering yield materials. After the ANIS, several kinds of modified ANIS such as a simple negative ion sputter source (SNICS) [35, 36] and others [37] were developed.

Figure 14.13

Aarhus negative ion source (ANIS) [33].

The UNIS and ANIS had disadvantages in efficient negative ion production, i.e., insufficient cesium particle supply to the sputter target surface and high gas pressure of the negative ion path in the plasma region, respectively. In order to overcome these disadvantages, several kinds of heavy negative ion sources of the beam sputter type and the plasma sputter type were developed in a third stage. 14.2.3.2

Recent Sputter Type Heavy Negative Ion Sources.

(a) Beam sputter type

(i) Versatile negative ion source The versatile negative ion source (VNIS), developed by Middleton et al. as shown in Figure 14.14 [38, 39], is a source which overcame the disadvantage of the UNIS. In this source a cooled sputter target is placed in front of a surface cesium ionizer


made as a cylindrical tantalum filament, and sufficient cesium ions together with sufficient neutral cesium particles are supplied to the sputter target surface. Thus a relatively high negative ion current of a few hundred microamperes could be obtained, as described in Table 14.2. In this source it is difficult to bias the sputter target to voltages of more than several kilovolts because of the disposition of the electrodes.

Figure 14.14

Versatile negative ion source (VNIS) [38].

(ii) NIABNIS The NIABNIS (neutral- and ionized-alkaline metal bombardment type heavy negative ion source) developed by Ishikawa et al. as shown in Figure 14.15 [40, 41] is another source which overcame the disadvantage of the UNIS. In this ion source a plasma type cesium ion source such as a compact microwave ion source is used for cesium supply. The cesium ion source can deliver sufficient neutral cesium particles, as well as a cesium ion beam with energy of 10 to 25 keV, to a cooled sputter target which also serves as cesium ion extractor. Various kinds of heavy negative ion currents of a few hundred microamperes could be obtained, as described in Table 14.2. (b) Plasma sputter type

(i) Multicusp plasma sputter type The multicusp plasma sputter type heavy negative ion source (BLAKE-II) developed by Mori and Alton as shown in Figure 14.16 [42] is the first source which overcame the disadvantage of the ANIS. In this source, a large sputter target is placed in a plasma generated at a low xenon gas pressure of order 10–3 to 10–4 Torr. The plasma

299

300


Neutral- and ionized-metal bombardment type heavy negative ion source (NIABNIS) [40].

Figure 14.15

Figure 14.16

Multicusp plasma sputter type heavy negative ion source (BLAKE-II) [42].


is generated efficiently due to both the cusp magnetic fields on the plasma chamber wall and the lanthanum hexaboride (LaB6) filaments. The lanthanum hexaboride thermionic filament can deliver a high electron current at a relatively low filament temperature of 1400 to 1500 C. Since electrons from the two sets of filaments are effectively confined in the cylindrical chamber, a dense plasma is generated in it, even for very low gas pressure. Neutral cesium particles are supplied to the spherical sputter target from a cesium oven. Thus, intense negative ions produced on the sputter target surface are effectively extracted without suffering electron detachment collisions. The source is operated in a pulsed mode. Typical pulsed arc voltage and current were 30 to 40 V (500 ls, 20 Hz) and 10 to 20 A, respectively. When a sputter target voltage of –500 to –1000 V was applied, a target current of about 300 mA was produced. The extracted beam currents were 10 mA for Au–, 10 mA for Cu–, 6.4 mA for Pt–, 6 mA for Ni–, 5.4 mA for Ag–, and 8 mA for C– beams. The normalized beam emittance was about 0.3 p mm mrad at the Ni– current of 6 mA (ii) Microwave discharge plasma sputter type The microwave discharge plasma sputter type heavy negative ion source (BLAKE-IV) as shown in Figure 14.17 [43] was developed by Mori et al. after developing the multicusp plasma sputter type. In this source a microwave discharge plasma is generated in a cylindrical plasma chamber, on the wall of which a magnetic cusp field is formed by Sm–Co permanent magnets for confining the plasma as well as for generating ECR conditions. Microwave power at 2.45 GHz is introduced into the plasma chamber through a sputter target that also serves as an antenna for the

Figure 14.17

Microwave discharge plasma sputter type heavy negative ion source (BLAKE-IV) [43].

301

302


microwaves. Pulsed microwave power (at about 3 kW, 0.5 ms, 20 to 50 Hz) is used to form a plasma in xenon gas with flow rate around 0.1 sccm. Bias voltage for the water-cooled sputter target is between –0.2 and –1.5 kV. In order to eliminate electrons from the extracted ion beam, a small Sm–Co dipole magnet which generates a magnetic field perpendicular to the beam axis is placed at the ion extraction aperture. A 6 mA Cu– beam current was extracted at the optimum cesiated surface condition. (iii)Rf plasma sputter type An rf plasma sputter type heavy negative ion source as shown in Figure 14.18 [44– 48] was developed by Ishikawa et al. just after developing the BLAKE-II. This source can deliver milliamperes of negative ion current such as B–, C –, Si–, P– and Cu– in dc mode. In this source, a dense xenon plasma is generated inside an rf coil with several turns located in the plasma chamber. The rf (13.56 MHz) discharge takes place even at a relatively low xenon gas pressure (~10–4 Torr) by introducing rf power of few hundred watts. A sputter target is placed perpendicular to the rf coil axis, and neutral cesium particles are supplied to its surface. The bias voltage of the sputter target is about several hundred volts. Heavy negative ions are effectively produced on the cesiated sputter target surface by xenon ion bombardment and are transported to the extraction aperture with minimal electron detachment collisions. In the negative ion extraction region a weak magnetic field perpendicular to the beam axis is applied to eliminate electrons from the beam. Table 14.3 shows the maximum extracted negative ion currents from the source. More than 10 mA of copper negative ion current and milliamperes of boron, carbon, phosphorus, and

Figure 14.18

Rf plasma sputter type heavy negative ion source [44].

14.3 Volume Production Negative Ion Sources

silicon negative ion currents, which would be useful for implantation into semiconductors, were obtained. Table 14.3

Maximum extracted negative ion currents from rf plasma sputter type heavy negative

ion source. Negative ion

B–

B2–

C–

C2 –

P–

Si–

Si2–

Cu–

Cu2–

Current (mA)

0.03

1.0

1.6

2.3

0.8

3.8

0.27

12.1

0.15

14.3

Volume Production Negative Ion Sources 14.3.1

Negative Ion Formation by Volume Production

A highly vibrationally excited H2 or D2 molecule (excitation levels from 5 to 10 for H2 and from 10 to 15 for D2) has a large cross section, i.e., rate coefficient of dissociative attachment with an electron having an energy of around 1 eV [49, 50], as shown in Figure 14.19. Through this process hydrogen negative ions are effectively produced, even in a plasma. This hydrogen negative ion production method is called volume production. However, an electron having an energy of around 40 eV is suitable to generate the highly vibrationally excited hydrogen molecules as shown in Figure 14.20. Since the optimal electron energies for these two processes are quite different, the plasma within the ion source is usually divided into two regions by a magnetic filter (several tens of Gauss of transverse magnetic filed) in volume production type negative ion sources. One part is a high electron temperature region (or electron bombardment region) for the generation of highly vibrationally excited hydrogen molecules and the other part is a low electron temperature region for the production of negative ions through dissociative electron attachment. In the region for hydrogen excitation, there exist many electrons with energy several tens of eV,

Rate coefficient of dissociative attachment with an electron having an energy of 1 eV.

Figure 14.19

303

304


Figure 14.20 Cross section of vibrational excitation for hydrogen molecule as a function of electron energy.

extracted from an electron source in a plasma region, as is usual for positive ion sources. These relatively high energy electrons cannot cross the magnetic filter. The excited hydrogen molecules and low energy electrons selectively diffuse to the second region for the dissociative electron attachment, and negative ions are effectively produced in this region. Negative hydrogen ions are extracted from a hole in the wall of the dissociative electron attachment region. Since many electrons are extracted together with the negative ions, a transverse magnetic field of about several hundred Gauss is applied in the ion extraction region to eliminate these electrons. 14.3.2

History of Source Development

The concept of volume production was originally proposed by Bacal et a1. [51–53]. The effect was first found in a plasma formed at relatively high hydrogen gas pressure. However, the negative ion density in the single region plasma was not greatly increased, and they divided the ion source plasma into two regions with different electron temperatures so as to realize the optimum conditions for these regions: one for the generation of highly excited molecular hydrogen with high temperature electrons, and the other for the production of negative hydrogen ions by dissociative electron attachment with low temperature electrons [54, 55]. The extracted negative ion current density was several tens of milliamperes per square centimeter. Recently an increase in the current density by an order of magnitude was found when a small quantity of cesium is supplied to the plasma chamber [56, 57].

14.3 Volume Production Negative Ion Sources

14.3.3

Recent Volume Production Negative Ion Sources

Since multicusp magnetic fields can effectively confine the plasma particles, many kinds of volume production negative ion sources have been developed using a multicusp plasma chamber with a magnetic filter. Several tens of milliamperes of H– current was extracted from a single aperture [55, 57], and more than 20 A of H– current was extracted from a multiaperture extraction electrode for neutral beam injection in fusion research [58]. Single-Aperture Volume Production H– Source Figure 14.21 shows a volume production dc H– source with a single extraction aperture, as developed by Holmes et al. [55, 59]. The plasma chamber (191410 cm3) is a multicusp bucket source with a magnetic filter (dipole filter or tent filter) to divide the two different electron temperature regions. The source has a relatively large extraction aperture of 16 mm diameter so as to investigate the characteristics of negative ion extraction. An electron suppressor near the extraction aperture is used to reduce the electron flux in the extracted beam. The negative ion extractor is a triode (G1, G2, and G3) accelerator type to collimate the extracted beam while dumping the extracted electrons. In the first step the beam is extracted by an extraction voltage Vext (up to 20 kV) and then it is accelerated to a high voltage Vbeam (up to 100 kV). In the high electron temperature plasma region, a dense plasma is generated by an arc discharge current (210 A) at a gas pressure of 8–9 10–3 Torr. Without and with cesium, 44 mA and 86 mA of negative hydrogen ion currents were obtained, respectively. Figure 14.22 shows a single aperture volume production H– source with an rfdriven multicusp plasma source, as developed by Leung et al. [57]. This source was 14.3.3.1

Figure 14.21

Single aperture volume production H– source developed by Holmes et al. [55, 59].

305

306


Figure 14.22 Single aperture volume production H– source with an rf-driven multicusp plasma source [57].

developed for particle accelerator injection application. The cylindrical source chamber (10 cm in diameter and 10 cm long) is surrounded by Sm–Co permanent magnets, forming longitudinal line-cusp magnetic fields. A magnetic filter is formed near the extraction region by locating a pair of permanent magnet rods so as to divide the high and low electron temperature regions. The high electron temperature plasma region was generated by pulsed rf power (about 2 MHz, 10 to 50 kW) at a gas pressure of about 1.2 10–2 Torr. The rf power is supplied to the plasma by a two turn antenna with inductive coupling. About 40 mA and 90 mA of H– currents were extracted from a 5.6 mm diameter aperture without and with cesium, respectively. Multi-aperture Volume Production H– Source These kinds of ion sources are described in detail in Chapter 16. 14.3.3.2

14.4

Charge Transfer Negative Ion Sources 14.4.1

Negative Ion Production by Charge Transfer

A negative ion can be produced from a positive ion via a charge transfer process in which the positive ion acquires two electrons from electron donor particles by collision. The electron donor must readily release electrons, and thus the kind of particles usually used as electron donors are alkaline metals or alkaline earth metals. In a negative ion source using this process, a positive ion beam is needed as a raw particle beam for charge transfer. The positive ion beam is passed through a charge transfer cell in which vapor of the electron donor material is contained. Maximum

14.4 Charge Transfer Negative Ion Sources

Figure 14.23 Maximum negative ion production efficiency by charge transfer as a function of proton energy [60].

negative ion production efficiency by charge transfer can be obtained at a certain ion energy when the length of the charge transfer cell is tuned to the optimum condition. For protons, the maximum negative ion production efficiency is 10 to 50% at a relatively low ion energy (a few hundred electron volts) as shown in Figure 14.23 [60]. On the other hand, for heavy positive ions the efficiencies have maximum values at relatively high ion energy (a few tens of kiloelectron volts) as shown in Figure 14.24 [61]. The maximum value of the efficiency for Cl, I, Te and Au, whose electron affinities are relatively high, shows a quite high value of over 40%.

Figure 14.24 Maximum negative ion production efficiency by charge transfer as a function of ion energy for various heavy ions [61].

307

308


14.4.2

History of Charge Transfer Negative Ion Sources

Conventionally, low current negative ion beams were produced by charge transfer processes because positive ion beams are easily obtained and the construction of the source involves only an electron donor gas cell. When the electron donor gas is the same element as the positive ions, the charge transfer efficiency is as low as about l%. On the other hand, when the electron donor gas is an alkali metal or alkali earth metal, the efficiency is increased to 10% or more. Since lithium is less active and of lowest mass among the alkaline metals, it is frequently used as an electron donor. Freeman et al. obtained 200 mA of Te– current using a lithium donor [62].

14.5

Conclusion

Negative ion sources have been developed together with their applications: heavy negative ion sources were first required for tandem accelerator injection and then for materials science applications; negative hydrogen ion sources were required for heating fusion reactor plasma and also for accelerator injection applications. From the motivation to obtain high current negative ion beams in these application fields, negative ion production mechanisms such as surface effect, volume production etc. have been deeply investigated. Negative ion currents comparable to positive ones can now be obtained. Negative ions have many interesting characteristics in comparison with positive ions, especially with respect to their interaction with materials. For example, when a negative ion beam is irradiated onto insulated materials or insulators, the surface charging voltage is extremely low, and then a new field of negative ion implantation has started. Thus further new applications are expected from negative ion physics and technology.

References

References [1] J. Ishikawa, Nucl. Instrum. Methods B 37/38,

38 (1989). [2] B. Rasser, J.N.M. van Wunnik, and J. Los, Surf. Sci. 118, 697 (1982). [3] J.N.M. van Wunnik, J.C. Geerings, E.H.A. Granneman, and J. Los, Surf. Sci. 131, 17 (1983). [4] J. Ishikawa, H. Tsuji, Y. Gotoh, and S. Azegami, Particles and Fields Series 53: Production and Neutralization of Negative Ions and Beams, 6th Int. Symp.(AIP, New York, 1994) AIP Conference Proceedings No.287, p.66. [5] G.D. Alton, Surf. Sci., 175, 226 (1986). [6] W.G. Graham, Proc. 2nd International Symposium on Production and Neutralization of Negative Hydrogen Ions and Beams, (BNL, Berkeley, 1980) BNL Report-51304, p. 126. [7] J. Ishikawa, H. Tsuji, and T. Maekawa, Vacuum 39, 1127 (1989). [8] H. Tsuji, J. Ishikawa, T. Maekawa, and T. Takagi, Nucl. Instrum. Methods B 37/38, 231 (1989). [9] E.H.A. Granneman, J.J.C. Geerlings, J.N.M. van Sunnik, O.J. van Bommel, H.J. Hopman, and J. Los, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p. 206. [10] M. Wada, R.V. Pyle, and J.W. Stearns, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p.247. [11] G.P. Lawrence, R.K. Beauchamp, and J.L. McKibben, Nucl. Instrum. Methods 32, 357 (1965). [12] Th. Sluyters and K. Prelec, Nucl. Instrum. Methods 113, 299 (1973). [13] Th. Sluyters, Proceedings of the 2nd Symposium on Ion Sources and Formation of Ion Beams (LBL, Berkeley, 1974) LBL Report-3399, VIII2-1. [14] Yu.I. Bel’chenko, G.I. Dimov, and V.G. Dudnikov, Nucl. Fusion 14, 113 (1974). [15] Yu.I. Bel’chenko, G.I. Dimov, and V.G. Dudnikov, Sov. Phys. Tech. Phys. 18, 1083 (1974). [16] Yu.I. Bel’chenko, G.I. Dimov, and V.G. Dudnikov, Proceedings of the 2nd Symposium on Ion Sources and Formation of Ion Beams (LBL, Berkeley, 1974) LBL Report-3399, VIII-1-1.

[17] T.S. Green, Nucl. Instrum. Methods 125, 345

(1975). [18] K. Prelec, Proc. 2nd International Sympsympo-

[19] [20]

[21]

[22]

[23]

[24]

[25]

[26]

[27] [28] [29] [30]

sium on the Production and Neutralization of Negative Hydrogen Ion and Beams (BNL, Brookhaven, 1980) BNL Report-51304, p. 145. K.W. Ehlers, Nucl. Instrum. Methods 32, 309 (1965). P. Allison, H.V. Smith, Jr., and J.D. Sherman, Proceedings of the 2nd International Symposium on the Production and Neutralization of Negative Hydrogen Ion and Beams (BNL, Brookhaven, 1980) BNL Report-51304, p. 171. W.K. Dagenhart, W.L. Stirling, G.M Banic, G.C. Barber, N.S. Ponte, and J.H. Whealton, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p. 353. K.W. Ehlers and K.N. Leung, Proceedings of the 2nd International Symposium on the Production and Neutralizaition of Negative Hydrogen Ion and Beams (BNL, Brookhaven, 1980) BNL Report-51304, p. 198. K.W. Ehlers, Proceedings of the International Ion Engineering Congress, (Institute of Electrical Engineers of Japan, Tokyo, 1983) p. 59. A.F. Lietzke, E.W. Ehlers, and K.N. Leung, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p. 344. C.C. Tsai, R.R. Feezell, H.H. Haselton, P.M. Ryan, D.E Schechter, W.L. Stirling, and I.H. Whealton, Proceedings of the 2nd International Symposium on the Production and Neutralization of Negative Hydrogen Ion and Beams (BNL, Brookhaven, 1980) BNL Report-51304, p. 225. K. Prelec, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p. 333. M. Mueller and G. Hortig, IEEE Nucl. Sci. 16, 38 (1969). R. Middleton and C.T. Adams, Nucl. Instrum. Methods 118, 329 (1974). R. Middleton, Nucl. Insturm. Methods 122, 35 (1974). R Middleton, Nucl. Instrum. Methods 144, 373 (1977).

309

310


[31] K.R. Chapman, Nucl. Instrum. Methods 124,

299 (1975). [32] K. Brand, Nucl. Instrum. Methods 154, 595 (1978). [33] H.H. Andersen and P. Tykesson, IEEE Nucl. Sci. 22, 1632 (1975). [34] P. Tykesson, H.H. Andersen, and J. Heinemeier, IEEE Nucl. Sci. 23, 1104 (1976). [35] G.T. Caskey, R.A. Douglas, H.T. Richards, and H.V. Smith, Jr., Nucl. Instrum. Methods 157, 1 (1978). [36] H.V. Smith, Jr., Nucl. Instrum. Methods 164, 1 (1979). [37] G.D. Alton and G.C. Blazey, Nucl. Instrum. Methods 166, 105 (1979). [38] R. Middleton, Nucl. Instrum. Methods 214, 139 (1983). [39] R. Middleton, Nucl. Instrum. Methods 220, 105 (1984). [40] J. Ishikawa, Y, Takeiri, H. Tsuji, T. Taya, and T. Takagi, Nucl. Instrum. Methods B 232, 186 (1984). [41] J. Ishikawa, Y. Takeiri, and T. Takagi, Rev. Sci. Instrum. 57, 1512 (1986). [42] G.D. Alton, Y. Mori, A. Takagi, A. Ueno, and S. Fukumoto, Rev. Sci. Instrum. 61, 372 (1990). [43] Y. Mori, Rev. Sci. Instrum. 63, 2357 (1992). [44] J. Ishikawa, Rev. Sci. Instrum. 63, 2368 (1992). [45] J. Ishikawa, H. Tsuji, Y. Okada, M. Shinoda, and Y. Gotoh, Vacuum 44, 203 (1993). [46] H. Tsuji, J. Ishikawa, Y. Gotoh, and Y. Okada, Particles and Fields Series 53: Production and Neutralization of Negative Ions and Beams, 6th International Symposium (AIP, New York, 1994) AIP Conference Proceedings No.287, p. 530. [47] J. Ishikawa, Rev. Sci. Instrum. 65, 1290 (1994). [48] J. Ishikawa, Surf. Coat. Technol. 65, 64 (1994). [49] C. Muendel, M. Berman, and W. Domeke, Phys. Rev. A 2, 181 (1985). [50] J.M. Wadehra and J.N. Bardsley, Phys. Rev. Lett. 41, 1795 (1978). [51] M. Bacal, A.M. Bruneteau, W.G. Graham, C.W. Hamilton, and M. Nachman, Low Energy Ion Beams (Institute of Physics, London, 1980) Inst. Phys. Conf. Ser. No. 54, 139. [52] M. Bacal, A.M. Bruneteau, H.J. Doucet, W.G. Graham, and G.W. Hamilton, Proceed-

ings of the 2nd International Symposium on the Production and Neutralization of Negative Hydrogen Ion and Beams (BNL, Brookhaven, 1980) BNL Report-51304, p.95. [53] M. Bacal and A.M. Bruneteau, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p.31. [54] K.N. Leung and K.W. Ehlers, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p.67. [55] A.J.T. Holmes and T.S. Green, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p. 42. [56] Y. Okumura, M. Hanada, T. Inoue, H. Kojima, Y. Matsuda, Y. Ohara, M. Seki, and K. Watanabe, Production and Neutralization of Negative Ions and Beams, 5th International Symposium (AIP, New York, 1990) AIP Conference Proceedings No.210, p. 169. [57] K.N. Leung, D.A. Bachman, and D.S. McDonald, Particles and Fields Series 53: Production and Neutralization of Negative Ions and Beams, 6th International Symposium, (AIP, New York, 1994) AIP Conference Proceedings No.287, p. 368. [58] Y. Takeiri, K. Ikeda, M. Hamabe, M. Osakabe, O. Kaneko, Y. Oka, K. Tsumori, E. Asano, T. Kawamoto, and M. Sato, Rev. Sci. Instrum. 73, 1087 (2002). [59] R. McAdams, R.F. King, and A.F. Newman, Production and Neutralization of Negative Ions and Beams, 5th International Symposium (AIP, New York, 1990) AIP Conference Proceedings No.210, p. 255. [60] A.S. Schlachter and T.I. Morgan, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p. 149. [61] A.S. Schlachter, Production and Neutralization of Negative Ions and Beams, 3rd International Symposium (AIP, New York, 1984) AIP Conference Proceedings No.111, p. 300. [62] H. Freeman, W. Temple, and D. Chivers, Nucl. Instrum. Methods 94, 581 (1971).

311

15

Ion Sources for Heavy Ion Fusion Joe W. Kwan

15.1

Introduction

The two main approaches to achieving fusion energy are magnetic confinement and inertial confinement. High power ion beams are important components in both cases. In magnetic fusion, neutral beam injectors are used for heating, driving current and diagnostics (see Chapter 17). In inertial fusion, ion beams can be used as “drivers” to deliver pulsed power to ignite a D-T fusion target. For a typical 5 mm diameter target, the required beam energy is about 3–7 MJ with a pulse length of 10 ns, and the peak power is about 1015 W cm–2 [1]. The present mainline research for inertial confinement fusion is to use high power laser beams as drivers. For example the National Ignition Facility (NIF) that is presently under construction will have 192 beams of neodymium glass laser [2]. Although we expect to see demonstration of target ignition from experiments at NIF, there are still many practical issues concerning the use of lasers in a fusion energy power plant. Today’s glass lasers simply do not have the required power efficiency, duty rate and reliability. In comparison, ion beams can be potentially more advantageous than laser beams because of their higher power efficiency (~30%), their higher repetition rate (~10 Hz), and their greater resistance to radiation damage at the target chamber windows. Our goal is to develop high power ion beam accelerators that can be used as drivers for commercial inertial fusion energy plants. 15.1.1

Heavy Ion Beam Driven Inertial Fusion

When applying pulsed power to a fusion target, the energy must be deposited within a short penetration range, e.g. 0.02 to 0.2 g cm–2 of the target material, in order to achieve implosion. The implosion will compress the D-T fuel before ignition. This target range depends on the ion mass and kinetic energy. For heavy ions with atomic mass near 200, the allowable kinetic energy is 100 are ultimately needed for fusion drivers, lighter ions such as K+ and Ar+ are useful for the near future because they allow experiments at high ion velocities on medium length accelerator facilities during the early development phases. Within the accelerator, high charge state heavy ions

15.1 Introduction

are similar to light ions of the same charge-to-mass ratio, but have the advantage that accelerating heavy ions (to the same energy per nucleon) with less beam voltage requires a shorter accelerator. However, the lower beam voltage must be compensated by higher beam current in order to deliver the same total power. Also, it is difficult to make ion sources that can produce pure high charge state ions (a single, but high, charge state) with large currents. A pure charge state is desirable because ions with different charge states will not stay together as a bunch in an induction linac. HIF beams require fast rise-time to minimize the waste of volt-seconds in the induction cores and to reduce the current and energy variation at the beam head. Energy spread results in chromatic aberration at the final focussing lens system. Large energy spread also limits the longitudinal beam compression for achieving high current. One common origin of energy spread is resonance charge exchange occurring in the region just outside the ion source aperture where the gas pressure is high. Thus HIF sources require low pressure. In order to focus ion beams onto a mm-size fusion target, the beam emittance must be small. This means that HIF requires beams with both large current and high brightness. Since the beam brightness is proportional to J/T, where J is the current density and T is the effective ion temperature, high brightness demands either high current density and/or low ion temperature. Furthermore, a heavy ion injector must have a low energy beam transport (LEBT) system that can adequately handle the severe space-charge forces. This LEBT structure can limit the maximum current density in the injector and often imposes restrictions on the ion source. Table 15.1 shows a summary of typical ion source and injector specifications for HIF using induction linac drivers. These are approximate numbers and they may vary depending on the actual ion mass, charge state and other system design parameters. Some numbers such as the charge state purity, current profile uniformity and energy spread are simply today’s best guess and have not been fully analyzed. Table 15.1

HIF driver ion source and injector specifications.

Beam energy (MeV)

1.6–2.0

Beam current per beam channel (A) Beam pulse width (ls) Beam rise-time (fraction of pulse width) Repetition rate (Hz) Ion mass (amu) Charge state purity Emittance (p mm mrad, normalized, 4rms) Current fluctuation and variation Current profile uniformity Beam energy spread (kV) Lifetime (pulses)

0.5 20 £ 5% 10 84–238 > 90% < 1.0 £ 1% > 95% £2 108

313

314

15 Ion Sources for Heavy Ion Fusion

Up to now ion sources traditionally used in HIF experiments have been surface ionization sources. The reason for this choice is mainly due to limitations in the LEBT. In fact, the problems encountered in HIF are fundamental to high current injectors and therefore we will dedicate the next section to discussing the relevant physics issues. Recently we have started to investigate a different way of building the HIF injector using high current density small beamlets. This approach opens up the possibility of considering many other types of ion sources such as RF discharge sources, metal vapor arc sources, laser ion sources, and negative ion sources. Since these other kinds of ion sources are discussed individually elsewhere in this book, we will elaborate more on the surface ionization sources in this chapter.

15.2

Beam Extraction and Transport

The prime characteristic of an HIF injector system is the delivery of a large amount of beam current with high average current density. High average current density results in small beamline cross-section and therefore minimizes the size of induction cores. As we have found in most cases, the real limiting factor is not due to the ion sources but is related to the beam transport system. In this section we briefly discuss the beam transport scaling laws. 15.2.1

Scaling Laws for Beam Extraction and Transport

The extractor, aperture lens and electrostatic quadrupoles (ESQ) are all electrostatic devices that are governed by the “three-halves scaling law”: J /

V 3=2 L2

(15.1)

where J is the current density, V is the applied voltage, and L is the characteristic length of the system. This scaling law is a direct result of the equations of motion for charged particles in an electrostatic field [4]. The most common place to apply this scaling law is in space charge limited flow regions where the current is selfadjusting. We can consider varying the voltage and the dimensions separately or in combination. If the dimensions are fixed and all the voltages applied to the electrodes (including the initial kinetic energy of the charged particles) increase by the same factor x, then the particle trajectories remain identical as long as the current density is increased by the factor x3/2. Since the dimensions are fixed in this case, the current is also increased by the same factor as the current density. The time of flight is decreased by the factor x1/2. In another situation, if all the voltages in the system are fixed but the dimension is increased by a factor y, then the particle trajectories remain similar as long as the current density is decreased by the factor y2. The current, which is the product of

15.2 Beam Extraction and Transport

current density and area, remains unchanged. The time of flight is increased by the factor y. The two factors can be combined to vary both the dimension and the voltage together (in order to avoid breakdown). The scaling can be extended as long as the system remains true to the physics involved. Thus the charged particle emission remains space-charge limited, the kinetic energy scales with beam voltage, there are no gas ionization or other atomic physics effects, and there are no secondary electrons or other surface physics effects. One intrinsic source of error that is usually ignored in many experiments is the ion temperature and contact potential at the emitting surface. Also, secondary electrons are frequently a major problem for high current beams with large beam potential. 15.2.1.1 High Voltage Breakdown Scaling The maximum value of V that can be applied to a device is limited by voltage breakdown, so an important rule to consider here is how the breakdown threshold scales with dimension. For gaps shorter than about 1.0 cm, V scales linearly with distance, whereas for larger gaps typical of HIF applications, V is approximately proportional to the square root of distance [5, 6]. In either case, and according to Eq. (15.1), the current density is found to decrease with increasing size. The nonlinear relationship is especially unfavorable to high current systems that require very high voltages and therefore need very large gaps. 15.2.1.2 Beam Extraction Diode The extractor forms an ion beam by accelerating ions out of the ion source aperture. To the first order of approximation, the current density is governed by the Child– Langmuir relationship for space-charge-limited flow in a diode

J ¼ v

V 3=2 d2

(15.2)

where v = (4eo/9)(2q/M)1/2 is called the Child–Langmuir constant, with q and M being the charge and mass of the ions respectively, d is the effective extractor gap length, and V is the extraction voltage. Clearly, this equation is a special form of Eq. (15.1). For a circular aperture with radius a, and ignoring any curvature of the emitting surface, the extracted beam current is given by 2 a 3=2 V (15.3) ICL ¼ pv d The aspect ratio (a/d) is an important geometric parameter affecting the beam optics. Typically, the aspect ratio is kept to less than 0.5 in order to prevent spherical aberration at the beam edge, so the maximum beam current of a diode is determined only by the extraction voltage and does not directly depend on the actual diode size. In general the beam current increases with diode size because a larger diode can hold off higher extraction voltage, but, as mentioned above, the current density decreases with size.

315

316


Since the typical pulse length of HIF driver beams at the injector is about 20 ls, the beam pulse rise-time is expected to be about 1 ls or less in order to conserve volt-seconds in the induction cores. Some near-term HIF experiments may require shorter rise-time because the main pulse length is designed to be only long enough (e.g., about 1 ls) for studying beam dynamic physics. Rise-time control is therefore an important issue for developing HIF ion sources. For plasma ion sources, including the plasma plume types such as metal vapor vacuum arcs and laser ion sources, rise-time control is complicated by the time required to establish both a steady state plasma and a “soft” boundary (the meniscus) at the extraction aperture. The situation is simpler in the case of a surface ionization source because the emitting surface is a rigid boundary. Nevertheless, even for a simple planar diode, a minimum beam pulse rise-time can only be produced with an extraction voltage pulse that has a special waveform to compensate the discontinuity in space charge at the propagating beam front. The ideal extraction voltage waveform for a 1-dimensional space-charge-limited flow is determined by adding the space-charge potential increase from the anode to the moving beam front and the vacuum potential difference from the beam front to the cathode. The result is: !3=2 # pffiffiffi rffiffiffiffiffiffiffiffiffiffiffi " pffiffiffi qe 2 2 J qe 2 J 4 4dt t (15.4) VðtÞ ¼ 27 v A mp 9 v A mp where J is the current density, d is the gap distance, t is time, q is the charge state, A is the ion mass in amu, and mp is the proton mass [7]. Of course, the actual extraction diode is not 1-D because of the Pierce electrode and the exit-aperture on the downstream electrode. In computer simulation, the 3-D problem is solved by empirically constructing a voltage waveform such that the emission current can be kept constant [8]. 15.2.1.3 Beam Transport in an Ion Gun Typically the ion gun is an extension of the extraction diode; it provides aperture lens focussing while accelerating the ion beam to sufficient energy before entering the LEBT section. In the case of high current density application, an Einzel lenses system using accel–decel electrodes can enhance the ion gun focussing capability. The scaling relationship for an Einzel lens can be understood through a simple model of balancing the beam’s space charge force with the Einzel lens’ focussing force. This can be written as 2 DV 1 pffiffiffiffiffi (15.5) J 2v L U

where U is the ion kinetic energy in eV, and DV and L are the voltage difference and gap distance between the electrodes respectively. In order to avoid optical aberration, the electrode should be thin and the radius of the aperture should be much smaller than L. This equation still follows the above-mentioned V3/2 scaling law as long as U scales together with DV. Similar to the problem of the extractor diode, the current density decreases for large size ion gun because DV scales as the square root of L.

15.2 Beam Extraction and Transport

For an Einzel lens system with fixed DV/L, the focusing capability diminishes as the beam gains velocity, thus limiting the beam energy range of the ion gun. A different kind of beam transport such as a quadrupole system is needed at moderately high beam energy, say greater than about 1 MeV. Beam Transport in an Electrostatic Quadrupole (ESQ) Channel Although the RF quadrupole (RFQ) is a common LEBT device used in modern RF accelerators, it is not suitable for HIF induction linacs because the beam is compressed into short bunches in an RFQ. For long beam pulses the electrostatic quadrupole (ESQ) is more effective. The maximum transportable beam current in an ESQ channel can be written as [9]: pffiffiffiffiffi DVq U (15.6) J v b b 15.2.1.4

where DVq is the voltage across the quadrupole electrodes and b is the bore radius. Here, the ESQ focussing capability increases with beam velocity (with corresponding increase in the lattice period), thus making it an effective LEBT device at medium energy. Average Current Density in an Array of Beam Channels In an array of multiple beams, the average current density (Jav) is the total beam current divided by the array cross-sectional area, including the space occupied by the beams, the electrodes, and the clearance in between. Thus a system study is required in order to determine the optimum Jav and the corresponding emission current density for the ion source. 15.2.1.5

15.2.2

Large Beam vs. Multiple Small Beamlets

As mentioned in the Introduction, the HIF driver system contains a large array of parallel ion beam channels (N » 100) with each channel carrying about 0.5 A of beam current. Assuming that the heavy ion has a mass of 100, the space-charge potential of this beam is equivalent to that of a 5.0 A proton beam. In today’s ion beam technology, the systems that have similar demand in beam current are the deuterium neutral beam injectors used in tokamaks (experimental magnetic fusion devices) and the xenon ion propulsion systems used in spacecraft. In both cases, a plasma ion source is fitted with an extraction grid containing multiple small apertures to form a large number of beamlets. The beamlets are allowed to merge into a single beam upon leaving the accelerator. As explained in Section 15.8, the process of merging small beamlets leads to emittance growth. Therefore the beamlets must start with very high brightness in order to compensate for this loss. So far, the traditional approach for single-beam HIF experiments is to use a large diameter ion source that is capable of producing current >0.5 A (without requiring beam merging). Since it is nearly impossible to create a well-behaved large diameter emitter surface (the meniscus) from a plasma

317

318


ion source without using a mesh at the aperture, the surface ionization source was adopted for HIF. In a way similar to electron guns, the surface ionization source provides a solid emitting surface for designing good optics. It is interesting to note that the surface ionization source was also the type of ion source used historically for spacecraft ion propulsion before it was replaced by the xenon plasma source.

15.3

Surface Ionization Sources

The surface ionization source emits ions from a hot plate. This configuration is ideal for designing beam optics with minimum aberration because the emitting surface is a solid boundary. In fact, the surface ionization source can be designed to produce a beam with very large aperture (and therefore high current per beam) because of the rigid emitting surface. The ion temperature from a surface ionization source is typically low, of the order of a fraction of an eV; therefore the beam emittance is kept low, even when the source diameter is large. The two main types of surface ionization sources are contact ionizers and aluminosilicate sources. 15.3.1

Contact Ionizers 15.3.1.1 Basic Principles of Contact Ionization The definitive work on contact ionization was published by Taylor and Langmuir in 1933 [10]. When an atom evaporates from a metallic surface, it can become an ion by leaving an electron behind. The ionization is due to quantum tunneling of an electron between the metal’s Fermi level and the atom’s electron cloud. An atom is more likely to become an ion if the atom’s ionization potential is lower than the metal’s work function. For this reason, surface ionization sources are typically used to generate singly ionized alkali ions such as cesium and potassium. The equilibrium ratio of ion flux (mi) to neutral flux (ma) departing from the surface is given by the Saha–Langmuir equation: mi g eðuIÞ (15.7) ¼ i exp kT ma ga

where gi and ga are the weighting factors, u is the work function, and I is the ionization potential. For example for the case of cesium ions emitted from a tungsten surface, u = 4.62 eV, I =3.87 eV, gi =1 and ga =2, therefore we find (mi/ma) = 465 at T = 1273 K. Iridium is better than tungsten because of its higher work function but the material is considered to be too expensive for general use. The amount of alkali metal coverage on the emitting surface is a result of the dynamic balance between incoming and outgoing fluxes of ions and atoms. The optimum coverage, typically a fraction of a monolayer, will lower the tungsten work function, thus enhancing ion emission. For high current, the emitter must be at high temperature to elevate the desorption rate. Unfortunately, according to the

15.3 Surface Ionization Sources

above equation, the ratio (mi/ma) is low when the temperature is high. This means that high current sources will inherently have significant evaporation of neutral atoms. For example at T near 1500 K, the cesium current density can approach 50 mA/cm2 while the neutral fraction is more than 1%. For most applications the neutral atom emission is negligible, but the issue is an important one for HIF because of the typically low duty rate. An ionizer will continuously evaporate neutrals (at a steady temperature) even though the ion beam duty factor for HIF is < 2 10–4. Contact Ionizer Construction A simple way to apply cesium to the emitting surface is to use a nozzle to spray cesium vapor from the front side. The challenge is to uniformly deliver the correct amount of cesium vapor onto the emitting surface while simultaneously avoiding depositing the vapor onto other neighboring surfaces. This is especially difficult if the source is large in diameter and has a wide extraction gap because the vapor jets typically do not have a high degree of aiming accuracy and sharp boundaries. The minimum nozzle temperature must be above 600 K in order to keep the cesium in the vapor phase. However even at such temperatures an adsorbed layer of cesium can diffuse along the nozzle wall and around to the outside part, thus allowing cesium vapor to come off in all directions. Higher nozzle temperature will reduce the diffusion length, allowing the nozzle to work properly. A great deal of cesium ionizer technology was developed for the ion propulsion thruster [11]. It was found that ionizers made of porous tungsten have two major advantages. First, it was more cost effective to produce large emitters using porous material than by using solid tungsten. Second, it provided a convenient way to continuously feed alkali metal vapor from the bulk volume to the front side through diffusion. Although the porous tungsten emits at a lower current density than solid tungsten (at a given temperature) and has a high neutral fraction, the above two advantages are sufficiently strong reasons to make porous tungsten the preferred choice. Porous tungsten is produced by pressing tungsten powder (spherical particles) to 80% of solid tungsten density and then sintering at about 1800 K. The particle size of the tungsten powder should be chosen to maximize the ion emission by allowing cesium adatoms to diffuse evenly on the exposed surface but minimize the neutral efflux from deep inside the pores. This means that the hole size and the distance between the holes should be of the order of the diffusion length. According to Forrester [12], this diffusion phenomenon is governed by the effects of trapping (at low temperature) and migration (at high temperature). The diffusion length decreases with increasing temperature; the diffusion length is about 1 micron at 1300 K. Thus the tungsten spheres used in fabricating surface ionization sources are usually 2–10 microns in diameter. The porous material can be easily machined to a specific geometry while it is in the “green state” before sintering. For low current density the emitting surface can be flat, but for high current density the emitting surface should be curved to produce beam focusing. As shown in Figure 15.2, the emitter is typically embedded in 15.3.1.2

319

320


a beam-forming electrode. The “Pierce angle” at the tip of the beam-forming electrode is most significant in controlling the beam optics and will determine the level of aberration at the beam edge.

emitter

Pierce electrode

extraction electrode

67.5 deg Ion Beam

heater wire heat shields

Figure 15.2

Schematic diagram of a surface ionization source with a Pierce electrode.

The emitter can be heated to operating temperature (up to 1400 K) either by using heater wire, or by electron bombardment. The most common heater wires are made of tungsten or molybdenum. Heat shields are necessary to minimize radiation heat loss and to keep some components from overheating. In particular, the temperature of the Pierce electrode should be kept low in order to prevent ion emission from its surface. Tantalum should be avoided in the heater package because the material is too reactive in case of air or water leaks. Typical failure mode of the heater wire is when some part of the wire makes electrical contact to the porous tungsten block, thus producing an electrical short. One way to solve this problem is to use potted heater wire but this can introduce contamination issues at high temperature. Another way is to minimize the shorting probability by using an isolation transformer to power the heater wire. A floated heater wire must make at least two contacts to cause any damage. Heating and cooling of the surface ionization source should be done slowly (up to an hour for a 17 cm diameter source) to avoid thermal shocks. A rear-feed system has a manifold behind the emitter and is connected to a cesium reservoir via a heated tube. Special brazing or e-beam welding is required to join the manifold to the porous tungsten emitter without closing up the pores. At steady state, the emitter is hot and cesium vapor continues to diffuse to the front side. A valve can be used to shut off the cesium vapor flow at the tube but there is no easy way to stop cesium that is already in the manifold from diffusing to the front except by cooling down the assembly. A common mode of failure is blockage in the feed tube or pores. This can be a result of contamination turning pure cesium into oxides with higher melting temperatures. One convenient way of loading a surface ionization source with alkali metal is to use a “doped” source. In this case, an aqueous solution of potassium or cesium carbonate is absorbed into the porous tungsten (done in air). The carbonate decomposes at about 1000 K releasing carbon dioxide. The alkali metal oxide left on the surface will slowly break down at high temperature to provide alkali metal atoms.


Typically there is enough alkali metal in the solution to produce several atomic layers of coverage throughout the pores in the block. Initially the emitting surface will be over-saturated with alkali metal atoms, but since the rate of desorption of neutral atoms from a saturated surface is very high, it takes only a few hours for the doped source to reach steady state operation (with a desirable alkali metal coverage). Without further replenishment, the coverage will gradually be depleted, so a doped source is only good for a few days of experiments. Experimentally we have found that if the doped source is opened up to air before the potassium is completely depleted, the source resumes working again after pump-down and a brief activation period. Typical Ionizer Performance in HIF Experiments Figure 15.3 shows beam current density as a function of extraction voltage for a diode using a 2 cm diameter cesium vapor ionizer operated at 4 different temperatures. Extraction voltage pulse length was about 3–4 ls, with a repetition rate of 0.1 Hz. At the highest temperature, the beam current density is space charge limited. Therefore the slope of the J–V curve in log–log plot is 3/2 (see scaling law in the previous section). At the maximum extraction voltage used, the beam current was 50 mA, corresponding to 15.9 mA/cm2. 15.3.1.3

Cs ion current density (mA/cm^2)

100 1145C 1077C 1039C 1012C

10

1 10

100 Extraction Voltage (KV)

Figure 15.3 J–V characteristics for a 2 cm diameter cesium vapor ionizer. Typical pulse length is a few ls, firing at about 0.1 Hz.

At the lower temperatures, the beam current became emission limited at high extraction voltage. The transition from space-charge limit to emission limit was not sharp. This is possible if some parts of the surface were more capable of emitting ions than other parts so that overall the entire surface gradually became emission limited. In order to ensure predictable beam current and therefore fixed beam

321

322


optics, all HIF injectors that use surface ionization sources are designed to operate in space charge limited mode. Figure 15.4 is a photograph of a 30 cm diameter cesium contact ionizer developed for HIF application that had produced up to 1 A of cesium ion current [13].

Figure 15.4 A 1 A high brightness cesium surface ionization source mounted on top of a 400 kV insulator. (Courtesy Lawrence Berkeley National Laboratory.)

15.3.2

Aluminosilicate Sources Basic Principles of Aluminosilicate Sources Emission of lithium ions from aluminosilicate material was first reported by Hundley in 1927 [14]. In 1936 Blewett and Jones showed promising results from filament sources made of aluminosilicate [15]. According to Feeney et al. [16] aluminosilicate sources are thermionic ion emitters emitting ions in the same way that electrons are emitted from a hot surface. In the absence of electric field, the emission is governed by the Richardson–Dushman equation: eu 2 (15.8) J ¼ A T exp kT 15.3.2.1

where J is the current density and u is the work function. With strong electric field, the ion emission can be enhanced, mainly due to a reduction in the work function. This phenomenon is known as the Schottky effect in thermionic electron emission. The concept of treating an aluminosilicate source as a thermionic emitter is adequate for most applications, but it is not exactly correct for describing the underlying physics of ion emission from the aluminosilicate material. Since the material is not a conductor, the work function is not as well defined as it is for electron emission. In fact the work function may vary throughout the emitting surface. Typically when the ion current density is found to scale according to the Richardson–Dushman equation, an empirical work function can be defined to fit the data. Since it is not a conductor, the emitting surface is not an equipotential surface. As ions are emitted


from the surface, the electrical potential will change unless there are ions flowing from underneath the surface to fill in the potential void. Thus the empirical work function is an average quantity, and the variation in work function may lead to a high effective ion temperature (e.g. 0.5 eV) in comparison to the actual thermal temperature (0.09–0.12 eV) of the emitter. In contrast to the contact ionization of adsorbed cesium atoms on a tungsten surface, as described in the previous section, thermionic ion emission is based on a model in which ions are already available on the surface of the aluminosilicates. Ions are liberated from a hot surface due to thermal agitation and the ion flow increases if there is an electric field to pull the ions along. Experimental data have shown that the neutral fraction in ion beams generated from aluminosilicate sources is at least an order of magnitude lower than in beams from contact ionizers [17]. It is perhaps a coincidence that both contact ionizers and aluminosilicates sources have similar operating temperatures of around 1000–1400 K and both produce alkali metal ions. The chemical formula for aluminosilicate is X2O · Al2O3 · n(SiO2), where X is the desired alkali metal element such as Li, Cs or K, and n is an integer determining the type of aluminosilicate crystals. For n = 2, 4 and 10, the corresponding analog natural minerals are b-eucryptite, spodumene and mordenite. The basic structure of this material is silica with some portion of the Si4+ ions in the crystal replaced by Al3+ ions. Adding a loosely bound alkali metal ion compensates the electric charge deficiency produced by this replacement. Since the silica crystalline structure is very open with large tunnels through which the alkali metal ions can move freely, aluminosilicate is a good ionic conductor but not an electronic conductor. This special property of conducting alkali metal ions is the fundamental reason why aluminosilicate can be used as an ion exchanger as well as a thermionic ion emitter. Aluminosilicate Source Construction Like contact ionizers, an aluminosilicate source has all the good features of a surface ionization source such as a rigid emitting surface, a low ion temperature and low gas flow. In fact, the aluminosilicate source is in some ways superior to the contact ionizer because it does not require a metal vapor feed system, does not critically depend on the porosity of the tungsten substrate, and does not evolve a lot of alkali metal vapor. Aside from the possibility of slightly higher ion temperature (due to non-uniform work function), the only major disadvantage of an aluminosilicate source is the total ion charge delivered during the source’s lifetime due to the limited amount of alkali metal ions that can be chemically stored in the aluminosilicate material. In applications that do not require a large amount of beam charge, averaged over the source’s lifetime, the aluminosilicate source is almost ideal. Examples that are frequently found in the literature are in the area of ultra-high vacuum beam collision experiments, ion sputtering, secondary ion mass spectrometer, and plasma diagnostics. Aluminosilicate sources are also suitable for near-term HIF experiments because of the low duty rate requirement of a few ls per pulse and 0.1 Hz repetition rate. 15.3.2.2

323

324


It is not yet clear if aluminosilicate sources can be developed to meet the requirements of a future HIF driver at a duty rate of 20 ls per pulse, 10 Hz repetition rate, and 108 pulses lifetime. One interesting idea is to use the “solid-state cesium ionizer”, invented by Seidl [18], which has a layer of porous tungsten thin film deposited on top of an aluminosilicate block. The rear-side of the block is also coated with a metal layer, so that when voltage is applied between the two layers (across the aluminosilicate block), alkali metal ions in the block are driven forward. During operation, cesium ions are emitted from the porous tungsten layer by contact ionization while the aluminosilicate functions as the cesium reservoir. The fact that one can drive ions out of an aluminosilicate block with an applied voltage implies that a high fraction of the ions stored in the aluminosilicate can be deployed even if the block is thick [19, 20]. The basic configuration of an aluminosilicate source is nearly identical to that of a doped ionizer, using similar designs for the heater and the tungsten substrate. The only difference is in the layer of aluminosilicate coating on the emitting surface. Commercially available aluminosilicate sources are typically small in diameter (e.g., 14 in) and their coatings are not very uniform. Since HIF experiments require beams of high brightness and high current, the production of large diameter, uniformly coated aluminosilicate sources was a major challenge. Previously the aluminosilicate coatings that were produced often had physical defects such as cracks and fractures, or had difficulty enduring many heat cycles. Recently we have successfully advanced the technique of uniformly coating large sources with diameter up to 10 cm and a spherical emitting surface. Nowadays we can reliably produce 0.25– 0.5 mm thick aluminosilicate layers that have no defects [21]. The first step in fabrication is to prepare the aluminosilicate powder. Aluminum oxide, silicon dioxide and the carbonate of the alkali metal are mixed stoichiometrically according to the type of aluminosilicate required. The mixture is then heated in an air furnace to chemically react according to the following equation: X2 CO3 þ Al2 O3 þ nðSiO2 Þ ! X2 O Al2 O3 nðSiO2 Þ þ CO2 "

(15.9)

After reaction, the aluminosilicate must be ground to a fine powder of 100– 400 mesh size. By mixing the aluminosilicate powder with a small amount of water (deionized to remove impurities) to form a “mud”, and then using a specially shaped tool to do the spreading, one can build up a layer of aluminosilicate with fairly uniform thickness on the substrate. The next step after drying in air is to sinter the powder in a vacuum furnace at about 1800 K. There are a few important points that can significantly affect the outcome. First, the tungsten substrate must be etched to open up the pores and fine aluminosilicate powder (400 mesh) should be used in this region in order for the aluminosilicate to melt into the pores. This method produces a good interface between the aluminosilicate and the metal substrate, thus preventing future separation after many heat cycles. Similar to the process of firing ceramics, the aluminosilicate mud must be air dried slowly to prevent cracking. Furthermore, we found that it was better to use a mixture of various powder sizes than just using the finest powder, because coatings made from the finest powder shrink too much.


An innovation called the aluminosilicate-composite source was invented to circumvent the difficulty in sintering pure aluminosilicate coating. The idea was first reported by Chow et al. in 1967 [22], and further developed by Hughes et al. in 1980 [23]. The basic concept here is to embed aluminosilicate powder in a tungsten (or molybdenum) matrix. A mixture of aluminosilicate and tungsten powders can be pressed and then sintered together at a temperature lower than that required for sintering pure tungsten. The sintered material can be further machined to a desirable emitter shape. In effect the embedded aluminosilicate is a reservoir releasing alkali metal ions to the porous tungsten matrix when heated. Based on the published data and our own experience, ion sources made by this method tend to have lower current density than conventional aluminosilicate sources. The poor thermal conductivity of the composite material also makes it difficult to reach high temperature at the emitting surface without overheating the rear side. Typical Aluminosilicate Source Performance Aluminosilicate sources have been widely used in HIF experiments with source diameters ranging from 0.5 cm to 17 cm. Figure 15.5 shows typical I–V curves indicating how the maximum beam current density varies with emitter temperature. At T = 1373 K, a current density of up to 80 mA/cm2 of K+ was achieved using an extraction voltage that has a pulse length of a few ls and repetition rate of about 0.1 Hz. It is expected that the beam current would droop over the pulse, for pulse lengths of a few tens of ls or greater, or would be at a lower level if the repetition rate is much higher [24]. 15.3.2.3

K ion current (mA / cm2)

100 1065C 1000C 985C 945C

10

1 10

100 Extraction Voltage (KV)

Figure 15.5

J–V characteristics of a fully activated potassium aluminosilicate K+ source.

A 17 cm diameter aluminosilicate source was developed for a 2 MeV injector built for the ILSE/Elise project [25]. The injector performance was satisfactory in meeting the project specifications of 0.8 A K+ ion beam with a normalized edge emittance

325

326


< 1 p mm mrad. Recently the injector was modified for the HCX project by replacing the 17 cm diameter source with a 10 cm diameter one that operated at a higher current density and has improved the beam optics [26, 27]. A schematic diagram of the injector is shown in Figure 15.6.

Figure 15.6

Schematic diagram of the HCX injector with a large diameter aluminosilicate ion source.

15.3.3

Surface Ionization Sources for HIF

The surface ionization sources, especially the aluminosilicate type, are the preferred type of ion sources for many recent HIF experiments because they meet the requirements of high current, high brightness, low gas pressure, high reliability, and are generally easy to use. However, this type of ion source has some inherent disadvantages such as the emission of alkali metal vapor into the accelerator, low power efficiency (because the heater is on continuously), limited lifetime, and low average current density for high current beam transport. These disadvantages provide an incentive to look for better ways to build ion sources for future HIF drivers while the nearterm single-beam HIF experiments (which have less stringent requirements than for a multiple-beam HIF fusion driver) can still use the surface ionization sources.

15.4

Gas Discharge Ion Sources for HIF

Gas discharge ion sources have the ability to generate large beams and high currents because they contain large but well-controlled plasma volumes. One example is the neutral beam injectors developed for experimental fusion tokamaks. Another example is the ion propulsion thruster used in spacecraft. The gas discharge ion sources discussed below are mainly the multicusp type. Strictly speaking, ECR sources are also gas discharge ion sources and can be considered for HIF application. ECR

15.4 Gas Discharge Ion Sources for HIF

sources typically produce a mixture of multiple charge state ions. While high charge state can be beneficial, a mixture of charge states is problematic for a HIF induction linac (see Section 15.1.2). Furthermore the configuration of a large solenoid magnet surrounding the ion source is not compatible with the geometry of HIF drivers with multiple beams. So far, little ECR ion source development has been directed toward HIF application. Gas discharge ion sources have been discussed in detail in previous chapters in this book; thus here we concentrate on issues related to HIF application. In simulating the merging of beamlets into a large beam [28], it was found that the typical 1–2 eV ion temperature for gas source is acceptable. For an accelerator grid transparency of less than 20%, current density of each beamlet must be near 100 mA/cm2 in order for the average current density to be high enough for HIF considerations. This level of heavy ion current density has been demonstrated in a recent experiment using a 10 cm diameter RF-driven multicusp gas source with a 3 mm aperture at the center [29]. In the experiment, the pressure was set to give the maximum beam current. The results shown in Figure 15.7 indicate that the maximum current density increases with RF power and is nearly independent of the gas species.

Ion current density (mA / cm2)

500

Neon 64 mT Argon 20 mT Krypton 20 mT Xenon 20 mT

400

300

200

100

0 0

10

20

30

RF power (KW)

Figure 15.7

Maximum current density from an RF multicusp source using a 3 mm aperture.

Another important criterion for HIF ion sources is the level of gas flow. It is important to minimize the gas flow because significant charge exchange can occur in the extraction region below a few hundred keV where the charge exchange crosssection is highest. Charge exchange produces ions at less than full energy and therefore increases the beam’s energy dispersion (equivalent to high longitudinal emittance). Clearly, the gas pressure used to produce the high current density data in Figure 15.7 is too high for HIF beams. Fortunately, the gas efficiency of a multicusp source typically improves with the size of the source volume (because higher volume-to-surface ratio results in better plasma confinement). So when larger ion sources are built for high current using multiple beamlet extraction, the required source pressure will be reduced.

327


In a recent experiment, a 26 cm inner diameter multicusp source was constructed. The data shown in Figure 15.8 was taken from a single 2.5 mm diameter beamlet on axis. These results indicate that the beam current density reaches a maximum at an optimum source pressure of 2 mTorr even though the plasma density inside the ion source increases monotonically with source pressure [30]. Figure 15.9 is an emittance diagram of the beamlet showing an equivalent ion temperature of about 2 eV. 400

Ion current density (mA cm^2)

328

Faraday cup Lang. probe

300

200

100

0 0

1

2

3

4

5

6

Gas Pressure (mTorr) Figure 15.8 Beam current density as a function of pressure for a 26 cm ID multicusp source. The optimum argon pressure is 2 mT.

Figure 15.9 Ar+ emittance diagram for the 26 cm multicusp source. The emittance number in the figure is un-normalized.

15.4 Gas Discharge Ion Sources for HIF

Figure 15.10 A 26 cm ID multicusp source for HIF multiple beamlets study. (Courtesy Lawrence Berkeley National Laboratory.)

Figure 15.10 shows a cut-out view of the 26 cm ion source and Figure 15.11 a photograph of the extraction grid containing 61 apertures for producing an array of beamlets. The ultimate goal is to produce uniform plasma in the source, such that the beamlet currents are uniform over the array. Highly non-uniform beamlets will result in large emittance and frequent breakdowns in the injector. In Section 15.2.1.2 we discussed the rise-time of a beam pulse in an ideal extraction diode where the emitting surface is a rigid boundary. The corresponding situation for a plasma source is more complicated because both the plasma and the beam-forming meniscus require time to reach equilibrium. Since both the plasma and the extraction voltage must be present in order to form a beam, either one can be used as an on-off control. The plasma build-up time is typically of the order of a few tens of ls. So switching the extraction voltage should be utilized for producing rise-times near 1 ls. Gating at the beam-forming electrode (aperture) using a positive low voltage with respect to the plasma shows a rise-time of the order of a few ls [31]. Argon beams with rise-times about 1 ls and current density near 100 mA/cm2 were produced in subsequent experiments [32]. In summary, the gas discharge source is probably the most promising way of producing a large and uniform array of beamlets with adequately high current density and adequately low emittance for HIF applications. Beamlets will be merged to form larger beams at about 0.5 A each. This method, in comparison to the large

Figure 15.11 A 61 aperture extraction grid for HIF multiple beamlets study. (Courtesy Lawrence Berkeley National Laboratory.)

329

330


diameter surface ionization sources, is more power efficient, has a higher average current density, and is scalable to very high beam current. The remaining issues that require further study are reducing the charge exchange fraction, demonstrating a fast rise-time, and a low emittance of merging beamlets.

15.5

Pulsed Discharge Sources

In contrast to a steady state gas discharge, another way of generating plasma is to use a high power density pulsed discharge. Typical examples are the laser ion sources and the arc discharge sources that include the metal vapor vacuum arc (MEVVA) ion sources. Due to the nature of high power density, pulsed discharge sources usually produce a spectrum of ion charge states. In comparison to the steady-state gas discharge, the pulsed discharge either does not require gas or needs only a very low pressure. Details of these types of ion sources are discussed in their own chapters in this book. In a typical set-up, a plasma plume is created by the discharge and the flow is directed towards an extraction grid for beam formation. The velocity distribution of ions in the plume results in ion separation at the end of a long drift distance. Usually ions with higher charge state acquire more kinetic energy and so they arrive early at the extraction grid. Since the plume also expands transversely, the plasma ions are cooled to a lower transverse ion temperature—an effect that is good for making low emittance beams. On the other hand, the unsteadiness of the plasma plume results in fluctuation of the beam-forming meniscus—an effect that can cause emittance growth. One way to control the beam formation based on space-charge limited flow is to use grid(s) [33]. However, early experimental results showed only limited success because the passage of the plasma through a screen would always reduce the average current density and increase the emittance. Fine mesh screens are fragile and non-rigid, so it is difficult to apply this method to very large apertures. In addition, a large aperture requires a large extraction gap (according to the aspect ratio) and therefore high extraction voltage. During a voltage breakdown, large breakdown energy can easily destroy a fine mesh screen. Thus even when a plasma plume covers a large extraction area, beam extraction should be done using an array of small apertures (maybe with additional fine mesh) similar to those of the gas source. 15.5.1

Metal Vapor Vacuum Arc Sources for HIF

In a metal vapor vacuum arc source (MEVVA), a plasma plume is produced when an arc creates a cathode spot to vaporize the cathode material. Thus the Mevva source can produce ions of conductive solids and it does not require gas to operate. For typical arc voltage and current of 1 kV and 500 A, ion current density in the beam extraction area can be in the tens of mA/cm2 range. Early attempts to use

15.5 Pulsed Discharge Sources

Mevva source for HIF [33, 34] showed that while the source could produce significant ion current, there were problems related to noise, pulse-to-pulse reproducibility, and mixed charge states. Recently improvements have been made in many of these areas, thus bringing back the interest in using Mevva ion sources for HIF. One major achievement was in improving the beam uniformities both spatially and temporally. In the MEVVAIV developed at GSI, by directing the plasma plume to flow through a pair of grids under the guidance of a magnetic field, the beam noise was significantly reduced [35]. Voltage bias on the double-grid was supposed to space-charge limit the plasma flow thus reducing the current density fluctuation. Furthermore, it was found that a strong magnetic field applied near the cathode could enhance the average charge state of ions in the beam. On the other hand introducing gas pressure in the chamber and in the beam line improved the beam quality but lowered the average charge state at the same time [36]. More recently, experimental results were reported showing that a gadolinium ion beam with current density of 21 mA/cm2 was uniform to within 2% over a diameter of 60 mm, and the fluctuations were within 3% rms over a pulse length of 20 ls [37, 38]. These results are shown in Figures 15.12 and 15.13. Further improvements in the uniformities were observed when a magnetic field was presented in the drift region between the arc and the extraction grid. Among the various metals (Pb, Y, Gd, Ba) being tested, barium was found to have the best charge state distribution for HIF because it contained more than 95% of 2+ 700

0.6

500

0.5

400

0.4

300

0.3

200

0.2

100

0.1

0 -100

0 0

700

20

30 40 time(µs)

50

-0.1 60

0cm 1.3cm Arc 2.5cm 3.5cm 4.0cm

500

18 15

400

12

300

9

200

6

100

3

0

0 0

10

20

30 40 time(µs)

50

60

Figure 15.12 Total ion beam current for 60 A arc current and 40 kV extraction voltage.

21

-3

Ion current density ( mA/cm 2 )

Arc_current (A)

10

030499.dat

600

-100

Total ion current (A)

Arc_current (A)

0.7

030499.dat

600

Figure 15.13 Current density pulses at various positions with respect to the beam center.

331

332


charge state ions. This is probably due to the fact that the ionization potential from 2+ to 3+ is 34.45 eV, which is very high in comparison to the 5.54 eV and 10.00 eV for the first 2 levels of ionization respectively. This charge state distribution was obtained when the arc reached steady state at 200 ls after the arc initiation. In general, there were more high charge state ions and impurity ions at the instant when the plume just arrived at the extraction grid and so the charge state distribution was not as pure as that during the later time. On a much shorter time scale, strontium (Sr) was found to have a charge state distribution dominated by 2+ at 10 ls after the arrival time [39]. Bismuth is another material that can be a good candidate for HIF because it is the heaviest non-radioactive metal and it can have a charge state distribution dominated by 1+ (at low arc current). Furthermore it is isotopically pure and low cost. In a recent experiment [40], the emittance of a Bi+ beamlet was measured using the slit-scanner method. At a current density of 18 mA/cm2, and aperture diameter of 2 mm, the normalized 4 rms emittance was 0.006 p mm mrad. Assuming a semiGaussian velocity distribution at the source aperture, the equivalent ion temperature is 1.76 eV. It was found that both the beam current and the emittance increased almost linearly with the arc current. Unfortunately this means that the brightness, which is proportional to the beam current divided by the square of the emittance, decreases with increasing beam current. The Mevva configuration can be used to indirectly ionize background gas molecules in the source, but the process was effective only when the source was equipped with a strong magnetic field. Figure 15.14 shows the spectra from an experiment using argon gas at 5 10–5 Torr [39]. The results showed that a significant amount of Ar+ and Ar2+ was produced while the aluminum ions were suppressed. So far the issue of lifetime using Mevva ion sources for HIF has not been seriously addressed. At a repetition rate of 10 Hz, the number of shots per 24 h day is

Figure 15.14 Time of flight spectra from a Mevva source using aluminum cathode. Top: vacuum operation (10–6 Torr), bottom: with argon gas flow (5 10–5 Torr).

15.6 Negative Ion Sources for HIF

8.64 105. Therefore ideally a reliable ion source for an HIF driver should have a life time of >108 shots which is about 2 orders of magnitude larger than typical Mevva sources. 15.8.1

Laser Ion Sources for HIF

To a large extent, the laser ion source is similar to the Mevva source. They both produce a plasma plume that expands over a drift region before beam extraction. Even the repetition rates and lifetimes are similar. At up to 100 J per pulse, the laser ion source can produce very high power density (> 1013 W cm–2) on a target spot, so the average charge state can be higher than that produced by the Mevva. With advances in laser technology, the laser ion source is continuing to improve its performance for heavy ion high energy physics accelerators in CERN and TWAC [41, 42]. While the laser ion source can be useful for the storage ring-based HIF approach, the same is not necessarily true for induction linac-based HIF. Although the laser ion source can produce a large current, the very wide spread of the ion charge state distribution is a problem for the induction linac. One way to address this problem is to select a lower wavelength, lower power (but longer pulse length) laser. According to Yoshida et al. [43], the electron temperature resulting from inverse bremsstrahlung absorption is proportional to (Ik2)2/3, where I and k are laser intensity and wavelength, respectively. Their experiment using a frequency-doubled Nd:YAG laser (k = 532 nm) on a copper target showed a charge state distribution dominated by only Cu+ and Cu2+ with no higher charge state components. Another obstacle for applying laser ion sources to a HIF driver system is due to the requirement of multiple-beam geometry. Since the laser light illuminates the target from the front, the most typical single-beam set-up is to locate the laser on the side and use mirrors to avoid the target debris. While contamination of the mirrors is a serious lifetime issue, a more subtle but yet difficult problem to solve is how to arrange multiple lasers for a HIF driver system that may require up to 100 beams. Using 100 lasers, one for each beam, can also be a cost issue. So far, it is not clear if a single laser can be used to drive several targets by doing fast switching.

15.6

Negative Ion Sources for HIF

There are two possible advantages to using negative ion beams for HIF. First, in a way similar to neutral beam injectors, the negative ions can be converted into neutral atoms via photo-detachment after the final focusing stage, thus producing an intense particle beam that can be focussed down to a small spot without the adverse space charge repulsion. Second, a negative ion beam will not attract the secondary electrons that are produced by the beam halos scraping on beam pipes. Trapping the secondary electrons is an undesirable effect found in high-current, positive ion beams with large beam potentials.

333

334


Negative ions are generally more difficult to produce in large quantity and they have a large cross-section for beam loss via electron detachment. This means that a HIF driver based on negative ion beams must have extremely good vacuum. Fortunately, unlike charge exchange of positive ions, negative ion beam loss does not produce new negative ions to cause beam energy dispersion. Nevertheless, if the pressure in the target chamber is too high, then the neutral atoms that are converted from negative ions will be ionized to become high charge-state positive ions. In 2001 Grisham calculated the laser photo-detachment efficiency and the required vacuum conditions for HIF applications, and later proposed a driver system based on halogen negative ions [44, 45]. An experiment was performed to demonstrate the feasibility of producing high current density heavy negative ion beams. It was believed that chlorine, bromine and iodine would behave similarly because they all have high electron affinities, at 3.62 eV, 3.36 eV and 3.06 eV, respectively. Chlorine was chosen in the experiment because the substance is in the gas phase at room temperature. The experiment used a typical negative ion source set-up consisting of a 10 cm diameter, 15 cm deep multicusp chamber, with a magnetic filter in front of a 2 mm diameter aperture (see Chapter 15 on negative ion sources). The dipole filter had a peak field of 135 G. At 2.2 kW of RF (12.56 MHz) power, and 28 mTorr of source pressure, a negative chlorine ion current density of 45 mA/cm2 was achieved. The negative ions were 99.5% pure atomic Cl–, and the e/Cl– ratio was as low as 7:1. Figure 15.15 shows the negative ion current and co-extracted electron current as a function of gas pressure [46]. The negative ion current density scaled linearly with RF power and was expected to reach 100 mA/cm2 at 5 kW. The ion temperature was found to be 0.3– 0.5 eV using a pepper-pot diagnostic.

Figure 15.15 Cl– and co-extracted electron current versus source pressure (aperture diameter = 2 mm).

In comparison with the 0.75 eV electron affinity for hydrogen, the Cl– production was highly efficient and without the need for adding cesium. In fact, the small ratio of e/Cl– suggested that the negative ion to positive ion density ratio in the plasma was more than 69%. This model is also supported by the fact that both the Cl– current and the electron current were not sensitive to the bias potential on the plasma electrode.

15.7 HIF Injector Designs

In summary, the experimental results are promising for using negative ion beams as HIF drivers. The negative ions are almost 100% singly charged, and have low energy dispersion. If the background pressure in the target chamber is low enough, negative ion beams can be converted to neutral atomic beams to produce ultra-high beam intensity. Furthermore, if the secondary electron problem for positive ion beams is detrimental, then negative ion beams could be a solution that saves the HIF concept. One area of improvement needed is reduction of the gas pressure in the source. In fact, the beam line vacuum requirement may turn out to be the most stringent condition for considering heavy negative ion beams for HIF.

15.7

HIF Injector Designs

In Section 15.2 we discussed beam transport scaling laws that are applicable to HIF injectors. The general rule is that a high current beam must be transported with low current density. So for current of the order of a few hundred mA and considering only a single beam, a large surface ionization source is the preferred choice. However for much higher current, or when considering a large number of beam channels, the method of merging high current density beamlets is preferred because it has a higher average current density resulting in a more compact injector design. 15.7.1

Large Diameter Source Approach

A preliminary design of a multiple beam injector based on the large diameter source approach was reported by Kwan et al. [47]. Computer simulation of the 1.6 MV ion gun with 0.5 A K+ current is shown in Figure 15.16. Since the beam size at the ion gun exit is much larger than that required in the ESQ channel in the induction linac, an ESQ matching section is needed. Consequently a multiple-beam injector system will have a funnel shape. The array is arranged in such a way that neighboring beams will share common ESQ electrodes. Figure 15.17 is a schematic diagram of the outermost beam line in an 84 beam matching section. Beams in the matching section are gradually compressed to smaller radii and also must be steered towards the axis using long ESQ channels.

Figure 15.16

Simulation of a 1.6 MV, 0.5 A ion gun with a 14 cm diameter K+ surface source.

335

336


Center line

Schematic diagram of the outermost beam line in an 84 beam array injector using large diameter surface ionization sources.

Figure 15.17

The main issues with this design are the complexity caused by beam steering and the large size. Aside from being costly, the very large size raises concerns in high voltage breakdown damage. Since the stored energy between electrodes is proportional to the capacitance and the square of the voltage, for a given electric field the stored energy increases linearly with electrode size. Typically injector components require in situ high voltage conditioning, but breakdowns inevitably occur during the conditioning process. Therefore it is necessary to limit the energy involved in a breakdown in order to prevent damage that can lead to more breakdowns. For most electrode material, this energy threshold is about a few joules. Electrodes that are too large may exceed this threshold and therefore be limited to lower voltage gradient. 15.7.2

Merging Multiple Beamlets Approach

In the merging multiple beamlets approach, the high current density beamlets (each beamlet has only a few mA current) are kept separated from each other by the acceleration grids in order to overcome the space charge expansion. After they have gained sufficient kinetic energy and reduced charge density, the beamlets are allowed to merge together to form a high current beam (e.g., > 0.5 A). The final beam brightness depends on the emittance of the merged beam. Figure 15.18 is a

Figure 15.18

Conceptual injector design based on merging an array of high current density beamlets.

15.7 HIF Injector Designs

Figure 15.19

An Einzel lens system for accelerating a high current density beamlet to 1.2 MeV.

schematic diagram of an injector beamline based on this concept. The total length from the ion source to the end of the merging/matching section is of torder 1.0 m. With a high enough average current density, the injector’s transverse dimension can match that of the ESQ channel, so the system is not funnel-shaped and minimum beam steering is required. Comparing the two approaches (Figures 15.17 and 15.18), we found that the merging beamlet method can be up to a factor of six smaller in size [48]. A proof-of-principle experiment that merged 19 H– beamlets into an ESQ channel was done for developing neutral beam injectors for the next-generation tokamaks [49, 50]. HIF injectors will require much higher beam current, voltage, and brightness. In order to obtain high average beam brightness, it is necessary to develop high transparency accelerator grids operating at high voltage gradient to focus the beamlets. Figure 15.19 shows an Einzel lens system designed to accelerate a 2 mm diameter K+ beamlet to 1.2 MeV kinetic energy. The accel–decel scheme provides enhanced beam focussing for current density up to 100 mA/cm2. The main physics issues for merging beamlets are envelope matching and emittance growth in the merging process. An innovative way to match beams into an ESQ channel is to aim the beamlets differently in the two transverse planes so that when merged the final beam spot is an ellipse with matching envelope divergence.

Figure 15.20 An elliptical arrangement of beamlets allows the merged beam to be exactly matched into an ESQ channel. Figure on the right contains beam envelopes in x–z and y–z planes.

337

338


In other words, the system is purposely designed to be astigmatic with different x and y focal planes. The result as shown in Figure 15.20 is a very short matching section with minimal beam envelope excursion. The space between beamlets creates free energy which is converted to emittance when the beamlets are merged together. This emittance growth phenomenon was investigated by Anderson [51]. Basically, the emittance growth (normalized to a constant beam current) is minimized when the beamlet energy is high, the number of beamlets is large, and the beamlets are close to each others. Recent computer simulations by Grote [28] further showed that the final emittance also depends on the initial beamlet convergence angle (best at around –2 to – 4 mrads) and the ion temperature. Nevertheless the ion temperature dependence was weak, and thus ion sources with high ion temperature can be used as long as the average current density remains high. Figure 15.21 shows the evolution of 91 beamlets in configuration space and phase space. The x and y rms emittance was found to initially rise to different values because of the elliptical shape but later came to an equilibrium value (average between x and y emittance) after a few undepressed betatron periods (about 10 m distance). The beam halo produced was about 1%.

Figure 15.21 Merging 91 beamlets in configuration space and phase space.

15.8

Conclusion

Very high current beams with high brightness are required for heavy ion beam driven inertial fusion. Brightness can be achieved by using either high current density or low emittance, or a combination of both. The ideal case is to make high current

References

beams with high current density, but fundamental beam transport physics prevents the two conditions from occurring simultaneously. In the traditional approach, a large area ionization source is used to produce a high current beam (per channel) at relatively low current density. Even with a large diameter the emittance can still be low because the ion temperature is low for surface ionization sources. A new approach is to merge a large number of small but high current density beamlets to form a large beam. With sufficiently high beamlet current density and grid transparency, the average current density can be adequate. Furthermore, the emittance of the final merged beam depends mostly on the emittance growth in the merging process and only weakly on the intrinsic ion temperature at the ion source. Therefore this approach allows many ion source options, and is not limited to surface ionization sources. We have considered in this chapter various ion source options for HIF applications including contact ionization sources, aluminosilicate sources, gas plasma sources, metal vapor vacuum arc sources, and negative ion sources. Experiments are continuing to improve these ion sources in order to meet the challenging and sometimes evolving HIF requirements. The ultimate goal for HIF is to produce fusion energy, and our effort to develop ion sources for HIF is part of this endeavor.

References [1] D.A. Callahan-Miller and M. Tabak, Nuclear [2] [3]

[4] [5]

[6] [7] [8]

[ 9]

Fusion 39, 1547 (1999). E.I. Moses and C.R. Wuest, Fusion Science and Technology 43, 420 (2003). R.O. Bangerter, in Proceedings of the International Symposium on Heavy Ion Inertial Fusion, Frascati, Italy, 25–28 May, 1993, Nuovo Cimento 106A, N.11, 1445, (1993). J.R. Pierce, in Theory and Design of Electron Beams, (Van Nestrand, New York, 1954), p.16. L.L. Alston, in High Voltage Technology (Oxford University Press, New York, 1968), p.65. L. Cranberg, J. Appl. Phys. 23, 518 (1952). M. Lampel and M. Tiefenback, Appl. Phys. Lett. 43, 57 (1983). J.-L. Vay, A. Friedman, D.P. Grote, Proceedings of the International Computational Accelerator Physics Conference, East Lansing, MI, Oct. 15– 18, (2002). E.P. Lee, R.O. Bangerter, C.F. Chan, A. Faltens, J. Kwan, E. Henestroza, K. Hahn, P. Seidl, J.J. Barnard, A. Friedman, D.P. Grote, W.M. Sharp, in Proceedings of the International

[10] [11] [12] [13]

[14] [15] [16] [17] [18] [19]

Symposium on Heavy Ion Inertial Fusion, Princeton, New Jersey, Sept. 6–9, 1995, Fusion Eng. Design, 32-33, 323, (1996). J.B. Taylor and I. Langmuir, Phys. Rev. 44, 432 (1933). G.R. Brewer, in Ion Propulsion, (Gordon and Breach, New York, 1970), p. 112. A.T. Forrester, in Large Ion Beams, (Wiley, New York, 1988), p. 236. S. Abbott, W. Chupp, A. Faltens, W. Herrmannsfeldt, E. Hoyer, D. Keefe, C.H. Kim, IEEE Trans. Nucl. Sci. 26, 3095 (1979). J.L. Hundley, Phys. Rev. 30, 864 (1927). J.P. Blewett and E.J. Jones, Phys. Rev. 50, 464 (1936). R.K. Feeney, W.E. Sayle, II, and J.W. Hooper, Rev. Sci. Instrum. 47, 964 (1976). E. Chacon-Golcher, D. Baca, J.W. Kwan, Rev. Sci. Instrum. 73, 1036 (2002). M. Seidl, U.S. Patent No. 4783595 (8 November 1988). S.I. Kim and M. Seidl, J. Appl. Phys. 67, 2704 (1990).

339

340


[20] A.E. Souzis, W.E. Carr, S.I. Kim and M. Seidl, [21]

[22] [23] [24] [25]

[26] [27]

[28]

[29]

[30]

[31]

[32]

[33]

[34] [35] [36] [37]

Rev. Sci. Instrum. 61, 788 (1990). D. Baca, E. Chacon-Golcher, J. W. Kwan, J. K. Wu, in Proceedings of the Particle Accelerator Conference, Portland, Oregon, May 12–16, (2003). K.K. Chow, H.S. Maddix, and P. Chorney, Appl. Phys. Lett. 10, 256 (1967). D.W. Hughes, R.K. Feeney, and D.N. Hill, Rev. Sci. Instrum. 51, 1471 (1980). A.N. Pargellis and M. Seidl, J. Appl. Phys. 49, 4933 (1978). S.S. Yu, S. Eylon, E. Henestroza, C. Peters, L. Reginato, A. Tauschwitz, D. Grote, F. Deadrick, in Proceedings of the International Symposium on Heavy Ion Inertial Fusion, Princeton, New Jersey, Sept. 6–9, 1995, Fusion Eng. Design 32-33, 309 (1996). F.M. Bieniosek, E. Henestroza and J.W. Kwan, Rev. Sci. Instrum. 73, 1042 (2002). J. W. Kwan, F. M. Bieniosek, E. Henestroza, L. Prost and P. Seidl, Laser Particle Beams 20, 441 (2002). D.P. Grote, E. Henestroza, J.W. Kwan, Phy. Rev. Special Topics-Accel. Beams 6, 014202 (2003). http://prst-ab.aps.org/ J. Reijonen, M. Eardley, R. Keller, J. Kwan, K.N. Leung, D. Pickard, R. Thomae, and M.D. Williams, in Proceedings of the Particle Accelerator Conference, New York, (1999), p. 1943. G. Westenskow, R.P. Hall, E. Halaxa, and J.W. Kwan, in Proceedings of the Particle Accelerator Conference, Portland, Oregon, May 12–16, (2003). L.T. Perkins, J. W. Kwan, K. N. Leung, M. Rickard and M. D. Williams, Rev. Sci. Instrum. 69, 1060 (1998). L. Ahle, R.P. Hall, and A.W. Molvik, J.W. Kwan and K.N. Leung, Rev. Sci. Instrum. 73, 1039 (2002) S. Humphries, Jr., C. Burkhart and L.K. Len, in The Physics and Technology of Ion Sources, edited by I.G. Brown, (Wiley, New York, 1989), pp. 397–419. S. Humphries, Jr., H. Rutkowski, J. Appl. Phys. 67, 3223 (1990). E.M. Oks, P. Spaedtke, H. Emig, and B.H. Wolf, Rev. Sci. Instrum. 65, 3109 (1994). H. Reich, P. Spaedtke, and E.M. Oks, Rev. Sci. Instrum. 71, 707 (2000). F. Liu, N. Qi, S. Gensler, R.R. Prasad, M. Krishnan, and I.G. Brown, Rev. Sci. Instrum. 69, 819 (1998).

[38] N. Qi, J. Shein, R.R. Prasad, M. Krishnan,

A. Anders, J.W. Kwan, and I.G. Brown, Nucl. Instrum. Meth. Phys. Res. A 464, 576 (2001) [39] A. Anders and J.W. Kwan, Nucl. Instrum. Meth. Phys. Research A 464, 569 (2001). [40] A. Anders and E. Chacon-Golcher, J. Appl. Phys. 93, 2298 (2003). [41] B. Yu. Sharkov, A.V. Shumshurov, V.P. Dubenkow, O.B. Shamaev, and A.A. Golubev, Rev. Sci. Instrum. 63, 2841 (1992). [42] B. Yu. Sharkov, S. Kondrashev, I. Roudskoy, S. Savin, A. Shumshurov, H. Haseroth, H. Kugler, K. Langbein, N. Lisi, H. Magnusson, R. Scrivens, J.C. Schnuringer, J. Tambini, S. Homenko, K. Makarov, V. Roerich, A. Stepanov, and Yu. Satov, Rev. Sci. Instrum. 69, 1035 (1998). [43] M. Yoshida, J. Hasegawa, S. Fukata, Y. Oguri, M. Ogawa, M. Nakajima, K. Horioka, and M. Shiho, Rev. Sci. Instrum. 71, 1216 (2000). [44] L.R. Grisham, Nucl. Instrum. Meth. A 464, 315 (2001). [45] L.R. Grisham, Fusion Sci. & Tech. 43, 191, (2003). [46] S.K. Hahto, S.T. Hahto, J.W. Kwan, K.N. Leung, and L.R. Grisham, Rev. Sci. Instrum. 74, 2987 (2003). [47] J.W. Kwan, O.A. Anderson, D.N. Beck, F.M. Bieniosek, C.F. Chan, A. Faltens, E. Henestroza, S.A. MacLaren, P.A. Seidl, L. Ahle, D.P. Grote, E. Halaxa, C.T. Sangster, and W.B. Herrmannsfeldt, in Proceedings of the Particle Accelerator Conference, New York City, New York, (Mar 28, 1999), p. 1937. [48] J.W. Kwan, L. Ahle, D.N. Beck, F.M. Bieniosek, A. Faltens, D.P. Grote, E. Halaxa, E. Henestroza, W.B. Herrmannsfeldt, V. Karpenko, and T.C. Sangster, Nucl. Instrum. Meth. Phys. Res. A 464, 379 (2001). [49] T. Inoue, K. Miyamoto, M. Mizuno, Y. Okumura, Y. Ohara, , G.D. Ackerman, C.F. Chan, W.S. Cooper, J.W. Kwan, and M.C. Vella, Rev. Sci. Instrum. 66, 3859 (1995). [50] J.W. Kwan, G.D. Ackerman, C.F. Chan, W.S. Cooper, G.J. DeVries, W.F. Steele, M.E. Stuart, M.C. Vella, R.P. Wells, T. Inoue, Y. Okumura, and M. Mizuno, Rev. Sci. Instrum. 66, 3864 (1995). [51] O.A. Anderson, Proc. of Inter. Symp. on Heavy Ion Inertial Fusion; Fusion Engineering and Design, 32–33, 209 (1996).

341

16

Giant Ion Sources for Neutral Beams Yasuhiko Takeiri

16.1

Introduction

Mega-power neutral beams for heating high temperature plasma are required for realization of controlled thermonuclear fusion as an energy source. Ion sources used for such neutral beam injection (NBI) systems are characterized as high energy (several hundred keV at present and above 1 MeV in future) and high current (several tens of A), and the source dimensions are large so as to produce several to several tens of MW of beam power. In this chapter, the giant ion sources used in neutral beam injectors for fusion plasma heating are described with respect to both conventional positive ion sources and the presently-developed negative ion sources. The injected beam energy is determined by the expected penetration depth in the target plasma. Beam energy has increased from about 20 keV for the first experiments on injection into magnetically confined plasmas in 1974 [1] up to about 400 keV at present [2], as the target plasma size has become larger. In a future fusion reactor the beam energy will need to be up to 2 MeV. The neutral beam is created by charge-exchange conversion from an ion beam, and at higher energies, above 100 keV nucleon–1, the neutralization efficiency for positive ions decreases drastically while maintaining at around 60% for negative ions. Thus large scale, positive ion sources were developed in the 70s and 80s when the required beam energy was below 80 keV nucleon–1, followed by negative ion source development for the high energy NBI system aimed at reactor-sized fusion machines. Ion sources for the NBI systems have the distinct features of large volume plasma production and large area beam acceleration at high power density. The beam species are hydrogen and hydrogen isotopes (deuterium and tritium). In the following sections, we firstly survey the characteristics of the large volume plasmas produced for large area beam extraction, and the properties of multi-beamlet extraction and acceleration including beamlet steering. Next, the giant positive ion sources, as used in the present NBI systems, are illustrated from the viewpoint of operational performance, including technological aspects. Finally, the giant negative ion sources, which are just now operational and still being developed for future high energy NBI systems, are described, including the operational principles and the source performance. The Physics and Technology of Ion Sources, Second Edition. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

342

16 Giant Ion Sources for Neutral Beams

16.2

Large Volume Plasma Production

In NBI ion sources, efficient source plasma production at high density is required, with high uniformity over a large volume. Although high density plasmas can be efficiently produced using magnetic fields penetrating the plasma for confinement of the ionizing electrons, such as the PIG discharge, it is difficult to maintain the uniformity while expanding the plasma production area; also, the produced plasma is not necessarily stable, i.e., it can be noisy. On the other hand, although stable and quiescent plasmas can be produced without a magnetic field, the plasma production efficiency in a field-free source is not high, due to the absence of plasma confinement. The multi-cusp bucket source satisfies the requirements, in that the plasma is produced in a field-free region and is surrounded by a multipole cusp magnetic field for plasma confinement [3–7]. A uniform and stable plasma can be efficiently produced in a large scale multi-cusp bucket source, and this is commonly used in NBI ion sources. Here, we survey the properties of large volume plasma produced in multi-cusp bucket sources. 16.2.1

Bucket Plasma Sources with Multi-Cusp Magnetic Field

The bucket plasma source is characterized by a multipole cusp magnetic field generated with permanent magnets that are attached to the plasma chamber wall. Since the plasma produced in a large volume field-free region is surrounded by the multicusp magnetic field near the chamber wall, plasma diffusion to the wall is reduced, resulting in highly efficient production of a uniform plasma. Figure 16.1 illustrates

Figure 16.1 Schematic illustration of the improvement in plasma uniformity with increase in the pole multiplicity of the multicusp magnetic field; (a) without the cusp field, (b) adding a low-multiplicity multipole field; and (c) increasing the multipole multiplicity. The illustration is only qualitative and does not indicate exact density profile shapes.

16.2 Large Volume Plasma Production

Figure 16.2 Cross-sectional view of a plasma formed in a multicusp source with permanent magnets arranged parallel to the source axis. The plasma is produced by an RF discharge. The six horizontal rods seen in the plasma are RF antennas. Courtesy of Dr. M. Matsuoka, Mie University, Japan.

schematically the improvement in plasma uniformity by a multicusp magnetic field. Without the cusp field, the plasma diffuses to the wall and the radial density profile is parabolic. With a cusp field the central part of the plasma is uniform, and the uniform region extends closer to the wall as the field pole multiplicity increases. A cross-sectional view of a plasma produced in a multicusp source with permanent magnets arranged parallel to the source axis is shown in Figure 16.2. The plasma boundary is shown along the cusp field. The maximum distance between the plasma boundary and the wall corresponds roughly to the spacing between the permanent magnet rows, with opposing poles near to each other. Thus the uniform plasma region approaches the wall as the spacing decreases. The magnetic field rapidly attenuates with distance from the wall, and the field strength at the plasma boundary is several tens of G. The cusp field strength, which is related to the plasma confinement, influences the plasma production efficiency and the electron density of the uniform plasma produced in the field-free region. In NBI ion sources, the magnetic field strength at the wall surface is as high as 1–2 kG. Several different kinds of multi-cusp sources can be configured according to the arrangement of permanent magnets, as shown in Figure 16.3. Note that the ion extraction surface has no magnetic field, so that ions diffusing to the plasma sheath are extracted. The axial line-cusp configuration, in which the magnets are aligned parallel to the beam axis, is used mainly in circular cross-section ion sources. The azimuthal line-cusp configuration, in which the magnets are aligned perpendicular to the beam axis, is used mainly in rectangular cross-section ion sources. As well as the line-cusp configurations, a checkerboard multi-cusp configuration is also utilized, in which the magnets are arranged so that the poles facing the plasma are alternatively north and south in two dimensions. The checkerboard mulit-cusp has the feature of shorter magnetic field connection length. For negative ion

343

344


sources, the azimuthal line-cusp is utilized so as to match the magnetic filter field to the cusp field, as will be described in Section 16.5.1.

Figure 16.3 Arrangement of permanent magnets of multicusp sources in (a) axial line-cusp configuration, and (b) azimuthal line-cusp configuration. Note that there are several variations for the arrangement of permanent magnets on the back-plate.

16.2.2

Plasma Modeling

For most multi-cusp bucket sources, the source plasma is produced by a dc arc discharge between the filament-cathode and the wall-anode. Here we model the large volume plasma produced by a filament-driven dc arc discharge, based on the analyses of Goede and Green [8] and Goebel [9]. First we derive the particle balance equations of the source plasma to find conditions for efficient plasma production. To simplify the argument, the plasma is assumed to be mono-atomic for both ions and neutrals, i.e., we neglect dissociation of molecular species. In the filament-driven dc arc discharge, the plasma contains primary fast electrons emitted from the filament-cathode, plasma electrons, and ions. The primary electrons, which initially have an energy corresponding to the cathode sheath potential, are lost either at the anode-wall after a containment time sf or by thermalization through inelastic collisions with neutral gas and interactions with plasma electrons. Since the primary electron flux is balanced by these loss rates, using the electron emission current from the cathode, Ie, the balance equation becomes h i 1 Ie=e ¼ n0 nf < rv >col þ nf sf V, (16.1) where n0 is the neutral gas density, nf is the primary electron density, col is the reaction rate coefficient for inelastic collisions including ionization and excitation averaged over the primary electrons with a velocity degraded to thermal energy by single inelastic collisions, V is the plasma volume, and e is the electron charge. The ions are produced by collisions of primary electrons with neutral gas, and the ionization by plasma electrons can be neglected because the observed electron tem-


perature is much lower than the ionization energy. Then the ion production rate I+ is given as Iþ=e ¼ n0 nf < rv >ion V,

(16.2)

where ion is the ionization reaction rate averaged over the velocity distribution of the primary electrons. The ions produced are lost at the anode-wall, at the ion extraction electrode (usually at floating potential), and at the cathode after a containment time si. The ion loss rate can also be expressed as the ion flux to the total loss area, A = Aai + Ae + Ac, and therefore the balance equation becomes n0 nf < rv >ion V ¼

ni ni vi Aai þ Ae þ Ac , si

(16.3)

where ni and vi are the ion density and ion velocity at the sheath edge, respectively, and Aai, Ae, and Ac are the effective anode area for ion loss, the electrode area, and the cathode area, respectively. The ion temperature is low enough to allow vi to be a nonthermal, monoenergetic velocity. For the low gas pressure satisfying collisionless conditions, vi is the collisionless ion drift velocity determined by the Bohm sheath criterion, and then Eq. (16.3) is rewritten as 1=2 kTe Aai þ Ae þ Ac , (16.4) n0 nf < rv >ion V ¼ ni M where Te is the plasma electron temperature, M is the ion mass, and k is the Boltzmann constant. The plasma electrons derive either from ionization or from thermalization of primary electrons, and are lost at the anode-wall. Assuming positive plasma potential, the random electron flux which overcomes the anode sheath is described by the Boltzmann equation. Therefore the balance equation becomes h i eu n0 nf < rv >col þ n0 nf < rv >ion V ¼ 1 ne ve Aae exp k Te 4 1=2 eu kTe ¼ ne Aae exp k Te 2pm

(16.5)

where ne and ve are the electron density and the mean electron velocity at the sheath edge, respectively, m is the electron mass, u is the plasma potential at the sheath edge with respect to the anode potential, and Aae is the effective anode area for plasma electron loss. The anode area corresponds to the effective particle collecting area of the anode, and it has been shown that its width is of the order of the hybrid Larmor radius, h = 2(ei)1/2 [10]. Here e and i are the Larmor radii of the plasma electrons and ions, respectively. Therefore Aai and Aae can be taken to be the same, Aa ” Aai = Aae. Using this set of balance equations, we estimate the ion production efficiency. From Eqs. (16.1) and (16.2) we obtain

345

346


col Ie 1 ¼ þ . Iþ ion n0 ion sf

(16.6)

The first term is related to the electron energy distribution, which depends on the source geometry and the operational conditions. At the limit of infinite gas pressure, this term represents the ion production efficiency, and is a constant depending on the initial primary electron energy, i.e., the discharge voltage, and is independent of the source geometry. The inverse ion production efficiency Ie/I+ has a linear correlation with 1/n0sf, indicating that the containment time of the primary electrons should be long for efficient ion production. Although higher gas pressure is also preferable for efficient ion production, low gas pressure operation is required for reduction of the heat load at the electrodes, which is especially important for a negative ion source, as well as for a compact vacuum pumping system. sf is dependent on the effective anode area Aa as sf

4V vf Aa

(16.7)

where vf is the primary electron velocity. Therefore a larger plasma volume with a smaller anode area leads to efficient ion production. In the filament-driven arc discharge, the ion flux to the filament-cathode should satisfy the Langmuir sheath criterion for space-charge flow as follows, 1=2 m Je (16.8) Jþ M where J+ = I+/A is the ion current density and Je = Ie/Ac is the current density of the emitted electrons. From Eqs. (16.6)–(16.8) we obtain the minimum gas pressure condition for a stable arc discharge as n0

A 4V

vf Aa . M1=2 col ion Ac A m

(16.9)

ion

The minimum gas pressure depends roughly on the ratio of the total loss area to the volume. The gas pressure should be lowered in a larger volume source with smaller anode area. On the other hand, the plasma discharge occasionally switches into an inefficient mode in a source with reduced anode area for confinement improvement. This discharge mode switch accompanies a large change of the plasma potential from positive to negative with respect to the anode. As a result, the primary electron energy is reduced due to a voltage reduction across the cathode sheath, and consequently the ionization efficiency is degraded. Using Eqs. (16.4) and (16.5), the plasma potential can be expressed by !1 col eu M 1=2 Aa ¼ 1 þ . (16.10) exp kTe 2pm Aa þAe þAc ion


To keep the plasma potential positive, the anode area should satisfy the following condition for the anode area limitation: ! col Aa 2pm 1=2 . (16.11) 1 þ Aa þAe þAc ion M As for the cathode, we obtain the cathode area limitation from Eq. (16.8) as follows: 1=2 Ac I m þ . (16.12) Aa þAe þAc Ie M Although reduction of the anode area is important for high ionization efficiency, the ratio of the anode area to the total loss area should be large enough to satisfy Eq. (16.11). On the other hand, although a large number of thick cathode-filaments satisfy Eq. (16.12) for high ionization efficiency, they equivalently reduce the anode area as indicated by Eq. (16.11). Sources that are designed so that the ion extraction electrode has a relatively small area possibly mitigate the anode area limitation. 16.2.3

Atomic Fraction

In positive ion NBI systems, the energies of the neutral atom beams that are formed by neutralization by molecular H2+ and H3+ ions are a half and a third of the ion beam energy, respectively. In order to optimize the beam deposition profile in the target plasma, the fraction of the full energy neutral atom beam must be as high as possible. Since the optimum conditions for ion beam extraction and acceleration depend on the ion mass, a high atomic fraction is also required for efficient ion source operation. Therefore, the plasma source should be designed to produce plasmas with a high atomic fraction. On the other hand, since no molecular hydrogen negative ions can be produced, all neutral atom beams have the full energy in negative ion NBI systems. However, a high atomic fraction is also required in the source plasma for cesium-seeded operation of negative ion sources, as will be described in Section 16.5.1. The atomic fraction in a discharge increases with plasma density and ion confinement time [11]. Therefore the atomic fraction increases with discharge power. The ion containment time depends on the ratio of plasma volume to the effective ion loss area [12], and consequently a larger arc chamber with a stronger cusp magnetic field is desirable for the production of plasma with a higher atomic fraction. The presence of primary ionizing electrons with a high energy in the vicinity of the plasma grid (ion extraction region) leads to an increase in H2+ ion extraction, because H2+ ions are generated in the extraction region from H2 molecules by impact ionization with primary electrons and are directly extracted. On the other hand, for thermal electrons with a temperature of several eV, the cross section for the ionization of molecules is sufficiently low while the cross sections for the dissociation of neutral molecules and molecular ion species are large. Thus another approach to increased atomic fraction is to divide the plasma volume into two regions using a sheet of transverse magnetic field, called a magnetic filter, so that

347

348


primary high energy electrons are prevented from entering the extraction region [13]. With this magnetic filter the floating potential of the plasma grid is decreased with respect to the anode wall, indicating elimination of primary ionizing electrons from the vicinity of the plasma grid. The concept of this magnetic filter is control of the electron energy distribution, and is applied successfully to negative ion production in the plasma volume as will be described in Section 16.5.1.

16.3

Large Area Beam Extraction and Acceleration

In order to obtain a high current ion beam, a large area beam extractor and accelerator is required. Since the extractable ion current from a single aperture is not determined by the aperture diameter but by the aspect ratio, defined as the ratio of the aperture diameter to the extraction gap width, multi-apterture (multi-hole or multislot) extraction and acceleration electrode systems are applied to NBI ion sources. A bundle of beamlets has a large cross section and multiple beamlets should be focused so as to pass through the injection port of the fusion device. Since an electrostatic or magnetic lens makes the accelerator complicated and does not work effectively with a large area, high current beam, individual beamlet steering is an important technique for large area beam focusing. In this section we survey the properties of large area beams consisting of multiple beamlets, and beamlet steering in multi-hole electrode systems. 16.3.1

Electrode Systems for Large Area Beams

The extractable ion current density, Ji, is limited by the ion space charge. In the case of the parallel plane diode it is determined by the Child–Langmuir law, and expressed as follows: 3=2 4 2Ze 1=2 Vext (16.13) e0 Ji ¼ 2 9 M ds where ds is the ion sheath thickness, Vext the extraction voltage, Z the ion charge number, and e0 the permittivity of free space. Considering an extraction system with a single round aperture of diameter 2a and an extraction gap separation of d (including the electrode thickness at the plasma boundary), the extractable ion current Ii can be written as pe 2Ze 1=2 2a 2 d 2 3=2 pe0 2Ze 1=2 2a 2 3=2 2 Vext Vext : (16.14) Ii ¼ pa Ji ¼ 0 9 9 M d ds M d Here, ds » d holds whenever the ion-emitting surface is kept flat or concave, leading to good beam optics. Eq. (16.14) indicates that the extractable ion current is determined only by the aspect ratio of the extraction system, 2a/d, when the extraction voltage is fixed. The space-charge limited current in a cylindrically symmetric extrac-

16.3 Large Area Beam Extraction and Acceleration

tion system has been derived by Langmuir and Blodgett [14], and its perveance P is expressed as P ¼ P0 ð1 þ 0:8kÞ

2

(16.15)

where P0 = Ii/Vext3/2 is the plane diode perveance of Eq. (16.14) [15]. k is a constant in the range –0.5 < k < 0.5, representing a measure of the curvature of the ion-emitting plasma boundary, and for a concave boundary (converging trajectories) k is negative. Since the aspect ratio is usually less than unity so as to keep the ion-emitting surface stable, the extractable current is as low as 0.5 A at an extraction voltage of 50 kV for H+ ions and independent of the aperture diameter. To enhance the perveance equivalently, multi-hole or multi-slot extraction systems are utilized for high current NBI ion sources. The total extractable ion current is then simply determined by the total aperture area multiplied by the space charge limited current density. Since the current density is about 0.6 times Ji in Eq. (16.13) for a beamlet extracted from a round aperture with minimum divergence angle, for the case of 50 kV H+ extraction a multi-hole extraction system delivering a total current of 70 A can be designed with a circular grid of 20 cm diameter and a grid transparency of 50%. Although the current density increases linearly with Vext3/2, the maximum holdoff voltage against breakdown is roughly dependent on d1/2. Therefore the maximum current depends on d3/4 if the aspect ratio is kept constant. Since a gap length of 1–2 cm is required for 100 kV extraction, a multi-stage extraction and acceleration electrode system is utilized for high energy ion sources. In this kind of system the electric field in each acceleration gap stage is gradually strengthened as the beam is accelerated downstream, so as to suppress beam divergence during the acceleration. After acceleration the ion beam drifts in the neutralization gas cell, and a thin plasma is generated along the beam path. Consequently the electrons flow backward in the accelerator and strike the back-plate of the arc chamber, in positive ion sources. This electron backstreaming decreases the overall acceleration efficiency and also leads to damage to the arc chamber. To suppress electron backstreaming, a suppressor electrode, biased negatively with respect to ground potential, is inserted just upstream of the ground electrode. An example of an ion extraction and acceleration electrode system is shown in Figure 16.4 for a single beamlet [16]. This is a four-electrode system (tetrode) with two-stage acceleration. The ions are extracted and accelerated in the first and second gaps, respectively. The third electrode is the suppressor electrode against backstreaming electrons. In the figure, various phenomena related to beam ion collisions with different types of particles are illustrated schematically. These collisions lead to deterioration of the beam acceleration efficiency and enhancement of the heat load on the electrodes. The beam acceleration efficiency is quite important especially in high power ion sources such as NBI sources. As shown in Figure 16.4, neutral particles in the accelerator lead to beam loss and backstreaming electrons. Therefore the operational gas pressure should be lowered in order to achieve a high acceleration efficiency.

349

350


Example of a four-electrode system (tetrode) with two-stage acceleration. The various collision processes that accompany the acceleration of positive ions are also illustrated schematically, with possible trajectories of different types of particles. (1) Primary transmitted ions (Hf+). (2)–(4) Direct interception by electrodes of ions (Hf+) on highly aberrated trajectories, leading to secondary emission of electrons. Many of these secondary electrons go back to the upstream electrodes or the plasma source. (5) and (6) Secondary (slow) Figure 16.4

ions (H+) produced by charge exchange or ionization of the background gas, some of which hit electrodes giving a high secondary emission of electrons. Electrons produced by the ionization also go back to the plasma source. (7) Slow ions (H+) extracted from the neutralizer plasma by the negative potential of the suppressor electrode, are incident on the suppressor electrode and create secondary electrons. This figure is modified from the original one in Ref. [16]. Courtesy of Dr. Y. Okumura, Japan Atomic Energy Research Institute.

16.3.2

Beamlet Steering

The large-area beam consisting of multiple beamlets should be focused to pass through the injection port. A simple way is to shape the grid electrodes geometrically with a slight spherical curvature so that each beamlet aims at a common focal point. This is utilized for short-pulse ion sources with three-grid accelerator (triode),

16.3 Large Area Beam Extraction and Acceleration

where the grid electrodes have no cooling channel and therefore can be pressed spherically. However, the geometrical shaping of the grid electrodes is inapplicable to long-pulse ion sources, the grid electrodes of which have cooling channels. In this case multiple beamlets have to be focused by steering individual beamlets programmably. Beamlet steering by the aperture displacement technique, where the aperture axis deviates from the beam axis, is useful for multiple beamlet focusing. In general, an aperture with a different electric field strength on each side works as an electrostatic lens. Figure 16.5 shows a general schematic of a two-stage acceleration system with three electrodes, where the aperture axes deviate from the beam axis. Here, the suppressor, with a small potential difference from the corresponding electrode, is omitted for simplicity because it has a negligibly small influence on the beamlet steering. Divergent and convergent lenses are formed at the entrance and the exit apertures of the electrode, respectively, when the electric fields on both sides are applied for beamlet acceleration, as shown in Figure 16.5.

Figure 16.5 General schematic view of a two-stage acceleration system with three electrodes. The aperture axes of electrodes 1 and 2 deviate from the beam axis. The suppressor grid is omitted for simplicity.

Usually an electrode is regarded as a single thin lens with a combination of both lens effects. According to the thin lens approximation, the focal length f of this type of lens is given from the paraxial ray equation as [17, 18], f ¼ and

4V (for a round aperture), Ed Eu

(16.16)

351

352


f ¼

2V (for a slot aperture), Ed Eu

(16.17)

where V is the ion energy at the lens position, and Eu and Ed are the electric fields at the upstream and downstream sides of the electrode, respectively. Here we consider a positive ion accelerator with a round aperture as shown in Figure 16.5. From Eq. (16.16), it is found that the sign of the focal length, i.e., the relative strength of the electric field on each side of the electrode, determines whether a lens acts as a convergent lens or a divergent lens. When the lens axis (i.e., the aperture axis) is displaced by a distance d from the beamlet, the beamlet deflects both for convergent and divergent lenses; see Figure 16.6. The beamlet deflection angle h is then given as h ¼ d=f ,

(16.18)

and the deflection direction is the same as the aperture displacement for a convergent lens and the opposite for a divergent lens.

Figure 16.6 Equivalent beam optics showing (a) convergent lens, and (b) divergent lens. The beamlet deflection is illustrated for the case when the lens axis is a distance d off the beamlet.

The final steering angle is determined by the sum of the steering angle at each electrode [19] in consideration of the axial beam acceleration in each electrode gap. For the accelerator shown in Figure 16.5, the steering angle h1 at electrode 1 is written as 1=2 d V1 h1 ¼ 1 , (16.19) f1 V1 þV2

16.4 Giant Positive Ion Sources

which contains the effect of the axial beam acceleration in the second gap. The steering angle h2 at the electrode 2 is written, taking the radial beam deviation by the beam steering at the electrode 1, db, into account, as 1=2 d2 R d1 d d V1 db ¼ dx. (16.20) h2 ¼ 2 b , f2 V1 þE2 x 0 f1 Since the focal lengths of electrode 1 and 2 are f1 = 4V1/(E2 – E1) and f2 = 4(V1 +V2)/ (–E2), respectively, for round apertures, the final steering angle hf is, then,

hf

¼ h1 þ h2

" # ( ) rE d1 þd2 rE 1=2 d1 rE 1 ¼ 3 1 d2 , (16.21) 4 d1 þd2 rE 2 rE d1

where rE is the electric field ratio E2/E1. When rE = 1, electrode 1 does not work as a lens and the beam steering is determined only by the displacement of electrode 2. The field ratio is usually larger than unity, rE ‡ 1, for the intermediate electrode (electrode 1), because the beam divergence is suppressed by the convergent lens. From Eq. (16.21) we find that the final steering angle is determined by a linear combination of the displacements of electrodes 1 and 2, to a good approximation. For simplicity and effectiveness of the steering, the ground electrode (including the suppressor) is usually displaced. The action of a thin lens on the beam trajectories can be analyzed systematically using beam transfer matrices [15]. Non-linear effects such as space charge, potential sag at the aperture center, grid thickness, and finite beam radius, also influence the beam steering [20], but not strongly. A 3-D beam trajectory simulation is useful for analysis of beamlet steering by aperture displacement.

16.4

Giant Positive Ion Sources

Here we consider the large and powerful positive ion sources that are presently used for high power NBI system in the large tokamaks. These sources are designed with physics emphasis on the uniformity, atomic fraction, beam acceleration and steering, and with technical emphasis on dissipation of beam heat load at the grids and backstreaming electron power at the arc backplate. In general, high density plasmas of (1–5) 1012 cm–3 are produced in large volume multicusp sources corresponding to an ionization rate of 1–5%, leading to high current ion beams, and the gas efficiency, the ratio of the beam current to the gas flow rate, is as high as 20–40%. The grid electrodes are actively water cooled with cooling channels embedded between each row of apertures. Since the cooling channels are narrow so as not to reduce the grid transparency, the channel length is restricted, not exceeding about 25 cm although depending on the structure. Therefore, the large area grid is usually rectangular, elongated in the direction perpendicular to the cooling channels. In the U.S. magnetic fusion energy program, high-power, long-pulse, large positive-ion sources were developed with common structure for use in the NBI systems

353

354


(a)

(b)

Figure 16.7 (a) Schematic diagram of the CLPS (common long pulse source) positive-ion source developed in the U.S. fusion program, and (b) a photograph of the accelerator of the CLPS. Courtesy of Lawrence Berkeley National Laboratory.


of the TFTR (Tokamak Fusion Test Reactor) tokamak and the DIII-D tokamak. This was called the common long pulse source (CLPS) [21]. The CLPSs were designed to have the same plasma source, accelerator grid modules, and insulator stack, except for the grid area and the gap length for beam acceleration, which depend on the injection port size and the injection energy specified in the individual tokamaks. Figures 16.7 (a) and (b) show a schematic diagram of the CLPS and a photograph of the accelerator, respectively. The bucket plasma source has dimensions 24 cm wide, 57 cm long, and 30 cm deep, surrounded by axial line-cusp which is generated with 40 rows of samarium–cobalt permanent magnets aligned parallel to the beam axis. For long-pulse operation the back-plate is actively cooled to dissipate power loading by the backstreaming electrons from the accelerator. The arc efficiency, defined as the ratio of the extractable ion current to the arc power, is 0.6–0.7 A kW–1, showing good confinement of the produced plasma. The atomic fraction is as high as 80– 85% in deuterium operation, because the high energy primary electrons seem to hardly approach the vicinity of the plasma grid due to the deep arc chamber and confinement of the primary electrons by the magnetic field around the filaments. The accelerator is an accel–decel tetrode consisting of four grids – plasma, gradient, suppression, and ground grids. The TFTR source with a specified energy of 120 keV has a grid area of 12 43 cm2 with four identical modules per electrode, and the grid module is characterized by a multi-slot structure, as shown in Figure 16.7 (b). The actively water-cooled grid tubes of shaped molybdenum form 45 beamlet slots. The multi-slot grid has a high transparency of 60% despite the active water cooling although extension of the grid width is restricted by the mechanical strength of the water cooled tubes. The TFTR source produced a 120 kV, 73 A deuterium beam for a few seconds with a current density of 230 mA/cm2. The beam divergence (1/e half angle) was as small as 7 mrad parallel to the slots and 12 mrad perpendicular to the slots. It is a special feature of the slot grid accelerator that the parallel and perpendicular divergence angles are different. Figure 16.8 shows a cut-away drawing of the positive ion source developed for the NBI system for the JET (Joint European Torus) tokamak, which is called PINI (PlugIn Neutral Injector) [22]. The PINI plasma source has a racetrack cross section of 50 22 cm2, and is surrounded by checkerboard multi-cusp field where the north and south poles alternate in two dimensions. In this high-order multipole field, compared with the line-cusp field, the magnetic field rapidly attenuates with increasing distance from the anode wall, resulting in a larger magnetic field free volume. However, the primary electrons emitted from the filaments can approach the plasma grid, and consequently the atomic fraction is not high, 68%. In order to increase the proton yield a magnetic filter is superimposed on the checkerboard multi-cusp field, as shown in Figure 16.9, which separates the electron-emitting filaments from the plasma grid [23]. Although, in general, the plasma uniformity is degraded by the presence of the magnetic field in the plasma due to E B and B B drifts, it can be compared to the checkerboard source without magnetic filter as the superimposed filter field is symmetric relative to the source geometry. The proton fraction increased to 87% with the magnetic filter. The source operates efficiently at low pressures (0.34 A kW–1 at 3 mTorr). The accelerator of the PINI

355

356


source is a tetrode with 262 circular beamlet apertures over a grid area of 45 18 cm2. The water cooled grid consists of two sets of half grids inclined to each other, facing a common point, and the beamlets inside each grid half are steered towards a focal point by the aperture displacement technique. The hydrogen beam thus produced is 80 keV, 60 A with a current density of 200 mA/cm2 for 15 s with a divergence of around 10 mrad. For deuterium ion beam production the accelerator was modified to a triode (three-electrode system with single-stage acceleration) by the removal of the second (gradient) grid and by increasing the extraction gap. As a result, a deuterium ion beam of 160 keV, 37 A was obtained, keeping almost the same beam power as the hydrogen beam.

Figure 16.8 Cut-away drawing of the PINI (plug-in-neutralinjector) positive ion source used in the JET NBI system. From Ref. [22], with permission from IAEA.

The ion source in the NBI system for the JT-60 tokamak is shown schematically in Figure 16.10 [24, 25]. The arc chamber is rectangular, 25 40 cm2 in cross section and 34 cm in depth, and equipped with a backstreaming electron beam dump at the back-plate. The confinement magnetic field is an azimuthal line-cusp configuration where the cusp lines are perpendicular to the beam axis. To obtain a high proton yield, in addition to the larger chamber the cusp magnetic field strength is as strong as 2.7 kG on the inner surface of the chamber. As a result, a high proton fraction of 90% was achieved. In this source there exists an axial field in the central region of the chamber, which would influence the plasma uniformity. The small anode area due to the strong cusp field would also lead to an inefficient discharge mode with a negative plasma potential, as discussed in Section 16.2.2. Therefore, particular attention is paid to the filament position and the filament current direction for a uniform distribution of the primary electrons [25]. Consequently, a uniform plasma with a high arc efficiency of 0.7 A kW–1 is produced. The accelerator is a tetrode, each grid of which has 1020 circular apertures 4 mm in diameter within a rectangular area


Figure 16.9 Illustration of the magnetic field pattern and the filament positions of the PINI source with checkerboard cusps and superimposed linear cusps for the magnetic filter. From Ref. [23], with permission from AIP.

12 27 cm2 (transparency of 40%). The beamlets are focused to a common point by the programmed aperture displacement in the suppressor grid. A hydrogen ion beam of 75 keV, 35 A (current density of 250 mA/cm2) was produced for 10 s with a divergence angle of 17 mrad. In the above ion sources the source plasma is generated by a dc arc discharge with multiple filaments for the cathode. The filament lifetime, several to several tens of hours, determines the maintenance period of the ion source. On the other hand, for RF-driven plasma generation the source lifetime is extremely long because there are no electrodes such as filaments that suffer from fatigue and erosion. Figure 16.11 shows a schematic diagram of the RF-driven large positive ion source which is used in the second NBI system of AUG (ASDEX-Upgrade) tokamak [26]. The source is compatible with the PINI accelerator. The RF power, of the order of 100 kW with frequency around 1 MHz, is inductively coupled to the plasma, which is produced in a quartz vessel surrounded by a copper coil antenna of a few turns, both embedded in a stainless steel vacuum vessel to withstand atmospheric pressure. A Faraday shield is installed between the plasma and the quartz vessel to avoid sputtering of the quartz. Since an RF transformer for insulating the high voltage

357

358


Schematic diagram of the positive ion source in the JT-60 NBI system. From Ref. [25] with, permission from AIP.

Figure 16.10

potential (up to 100 kV) is mounted on the backplate of the source together with the matching network, the RF generator can be operated at ground potential. The source delivers about 95 A of H+ at 55 keV and 80 A of D+ at 60 keV, comparable to the filament-arc dc-PINIs. Since the RF causes fluctuations in the plasma density and potential, the beam divergence is a little larger than for dc sources. The atomic fraction is not so high, 70–75%, presumably because of high energy electrons accelerated inductively in the vicinity of the plasma grid.

16.5 Giant Negative Ion Sources

Figure 16.11 Schematic diagram of the RF-driven positive ion source used in the AUG NBI system. Courtesy of Dr. E. Speth, Max-Planck-Institut fur Plasmaphysik, Germany.

16.5

Giant Negative Ion Sources

In a positive ion based NBI system, the injection energy is limited to around 80 keV nucleon–1, above which the neutralization efficiency is drastically reduced. For the next step in fusion research, where the required injection energy is higher than several hundred keV, a negative ion based NBI system is inevitable due to its high neutralization efficiency for such high energy beams. Although various kinds of negative ion sources have been developed for NBI systems, such as the double charge exchange source [27], the magnetron type source [28], and the multicusp type surface production sources [29], the volume production source [30] is now utilized for the present NBI system. The negative ion current density available to the injector is around 30 mA/cm2, which is 5–10 times lower than for a positive ion system. As a result, the dimensions of the ion source need to be quite large in order to realize high power injection. In this section the operational principles of the volume production negative ion source and the negative ion accelerator are overviewed, followed by a description of the presently used and developed negative ion sources. 16.5.1

Operational Principles of Negative Ion Sources

Hydrogen negative ions are produced in the plasma through dissociative electron attachment to vibrationally excited molecules. Since this process does not take place on the surface but in the plasma, it is called “volume production” [30]. Hydrogen molecules are vibrationally excited by collision with high energy electrons (> 20 eV),

359

360


and negative hydrogen ions are produced by dissociative attachment of low-energy electrons (»1 eV) to vibrationally excited molecules, expressed as follows: H2 + efast(>20 eV) ! H2*(excited state) + efast,

(16.22)

H2*(excited state) + eslow(»1 eV) ! H– + H.

(16.23)

The cross section in reaction (16.23) is larger for more highly excited states, n ‡ 7. On the other hand, the negative ions produced are easily destroyed by collisions with electrons with energies greater than 2 eV, which is incompatible with negative ion production by reaction (16.22) because of the electron energy. To resolve this incompatibility, a transverse magnetic field, called the magnetic filter, is applied in the plasma so that the plasma is divided into two regions. In the upstream region, called the driver region, in which the arc discharge is driven, the high energy electrons exist to produce vibrationally excited molecules, and in the downstream region, called the extraction region, in which the negative ions are produced and extracted, the electron temperature is low enough to produce negative ions and to prevent negative ion destruction. This “tandem” method or two-chamber method is illustrated in Figure 16.12. Vibrationally excited molecules diffuse across the magnetic filter while the electron energy distributions appropriate to the driver region and to the extraction region are controlled by the magnetic filter. For volume production of negative ions a multicusp bucket plasma source can be used, and therefore the basic technologies of positive ion sources can be applied. Three representative methods for generating the magnetic filter field are shown in Figure 16.13. In the rod filter method [31], the filter field is generated by inserting several rods containing permanent magnets into the plasma. A pair of permanent magnet rows outside the plasma generate the filter field in the external filter method [32]. In the PG filter method, the filter field is generated in front of the plasma grid by a large current flowing on the plasma grid [33]. In the volume production source, the negative ion current density obtained is not enough in low pressure operation below 1 Pa, which is required for reduction of the

Driver Region

Extraction Region

B H2 + efast → H2* + efast’

H2* + eslow → H– + H

Magnetic Filter

Plasma Grid

H– Ion

Extraction Grid

Illustration of the “tandem” or two-chamber method and the production mechanism of negative ions in volume production with a magnetic filter.

Figure 16.12


Figure 16.13 Methods for generating the magnetic filter: (a) rod filter, (b) external filter, and (c) PG (plasma grid) filter.

neutralization loss of negative ions during the acceleration. For low pressure operation, a small amount of cesium vapor is supplied to the source plasma. This cesium injection leads to a large enhancement, 3–10 times, in negative ion production and a reduction in the extracted electron current for low operational gas pressure [34]. An example of the cesium effect is shown in Figure 16.14 [35]. It is found that the negative ion current increases by a factor of 6–7 at lower operational gas pressure. This cesium catalysis is thought to be a surface effect in which hydrogen atoms and

Negative ion current as a function of arc discharge current, with and without cesium seeding, showing the cesium effect in negative ion production. From Ref. [35], courtesy of Dr. Y. Okumura, Japan Atomic Energy Research Institute.

Figure 16.14

361

362


Ho Atom Cs-Seeded Source Plasma

H– Ion

H+, H2+, H3+ Positive Ion Plasma Grid Cesium Coverage Illustration of the production mechanism of negative ions through the surface effect in a cesium-seeded negative ion source.

Figure 16.15

ions are converted to negative ions on the cesium covered surface with low work function, as illustrated in Figure 16.15. Since the conversion efficiency to negative ions is low for molecular neutrals and ions, a source plasma with a high fraction of atomic neutrals and ions is required for efficient negative ion production. This is realized with a high confinement plasma source, where the cusp magnetic field is strengthened and the ratio of plasma volume to plasma loss area is large, as described in Section 16.2. Note that the operational gas pressure should be reduced in such a high confinement source. A characteristic feature of the Cs-seeded volume production source is that the negative ion yield depends on the temperature of the plasma grid (PG) [36], as shown in Figure 16.16 [37]. The work function of the Cs-covered surface shows a minimum at a half to one monolayer of Cs, which is achieved at a PG temperature of 200–400 C depending on the amount of injected Cs. As shown in the figure, with an increase in the PG temperature, the negative ion current increases and the extracted current, which contains the electron current extracted together with the negative ions, decreases, and both saturate at around 250 C. They would start to

Negative ion current and the extraction current as a function of the temperature of the plasma grid in a cesiumseeded volume production source. From Ref. [37].

Figure 16.16


decrease and to increase, respectively, at a higher PG temperature due to evaporation of Cs from the surface. 16.5.2

Negative Ion Extraction and Acceleration

Although the negative ion extraction and acceleration system is basically the same as a positive ion system, negative ion extraction is distinguished from positive ion extraction in that electrons are extracted together with negative ions. Figure 16.17 shows an example of a grid system for negative ion extraction and acceleration. Permanent magnets are embedded in the second (extraction) grid (EG) to separate the electrons extracted with the negative ions. Since the extracted electrons are deflected by this magnetic field and are incident on the EG, the extraction voltage is as low as 5–15 kV so as to reduce the heat load to the EG. The electron suppression grid (ESG) is placed at the same potential as the EG to prevent secondary electrons generated inside the EG apertures from entering the acceleration gap. Extracted negative ions are then accelerated toward the ground grid (GG). For multi-stage accelera-

Figure 16.17 Example of a negative ion extraction and acceleration grid system; (a) a cross sectional view perpendicular to the rows of permanent magnet embedded in the extraction grid, and (b) a cross sectional view parallel to the magnet rows.

363

364


tion, an acceleration grid (AG) is simply added, corresponding to each acceleration stage. In the negative ion source, a grid electrode for suppression of backstreaming positive ions is not used because the flux is low compared with backstreaming electrons in a positive ion source. As shown in Figure 16.17, the magnetization direction of the permanent magnets in the EG is usually parallel to the beam axis, and the polarity is changed after every two lines of permanent magnets. As a result, the direction of the magnetic field on each side of the EG is opposite to the other, and the magnetic field on the plasma grid side extends into the ion emitting surface, which is effective in suppressing extracted electrons. On the other hand, the magnetic field causes a small deflection of the extracted negative ions, and the deflection direction of negative ion beamlets alternates line by line, leading to expansion of the whole beam size. Since a deflected beamlet deviates from the aperture axis of the grid, the beamlet is also deflected due to the electrostatic lens effect as described in Section 16.3.2, and this beamlet deflection is more significant. Beamlet deflection is compensated by beamlet steering using the aperture displacement technique [38]. To the extent that the near-axis thinlens approximation is valid, aperture displacement steering angles can be linearly superimposed. Therefore the GG apertures are displaced both for compensation of beamlet deflection due to the EG magnetic field and for multiple beamlet focusing. Figure 16.18 shows the effectiveness of the aperture displacement technique for compensation of negative ion beamlet deflection [38]. In the case of GG aperture displacement only for multiple beamlet focusing, the whole beam splits due to the alternate beamlet deflection by the EG magnet, as shown in Figure 16.18(a). By linearly superimposing the GG aperture displacement for compensation of alternate beamlet deflection on that for multiple beamlet focusing, all the beamlets are gathered at a common focal point, as shown in Figure 16.18(b). It is also possible to dis-

Beam profiles 11.2 m downstream in the direction parallel to the permanent magnet rows embedded in the extraction grid, for the cases (a) aperture displacement only for the multiple beamlet focusing, and

Figure 16.18

(b) aperture displacement for both multiple beamlet focusing and compensation of alternate beamlet deflection by the magnetic field at the extraction grid. 270 beamlets are produced from a grid area of 25 cm 26 cm. From Ref. [38].


place the ESG apertures for beamlet steering, although the steering angle is more sensitive and more complicated by the amount of displacement, as indicated by Eq. (16.21). Combination of GG and ESG aperture displacement is available for multiple beamlet focusing and compensation of beamlet deflection, respectively. Suppression of accelerated electrons within the negative ion accelerator is very important for improvement of the acceleration efficiency and reduction of the heat load on the grid electrodes. There are two main electron sources. One is secondary electrons generated on the EG, and the other is electrons generated by stripping in the destruction of negative ions by collisions with background neutrals during the acceleration. Although secondary electrons are generated on the surface inside the EG aperture, they are shielded from the acceleration electric field with the ESG. The EG aperture is shaped so as to be gradually widening toward the downstream side, and the diameter of the end aperture is larger than that of the ESG aperture to improve the shielding effect for secondary electrons. It has been reported that EG aperture shaping in which the end opening is narrowed again so as to have the same function as the ESG, similar to attachment of a thin ESG to the EG, is also effective for suppression of secondary electron acceleration [39]. This configuration simplifies the accelerator. Reduction of the operational gas pressure is essential for reducing stripped electrons. Figure 16.19 shows the acceleration efficiency, defined as the ratio of H– ion current to the drain current of the acceleration power supply, as a function of gas pressure [39]. The efficiency increases linearly as the gas pressure decreases, indicating reduction of stripped electrons at lower gas pressures. The acceleration efficiency is extrapolated to about 97% for zero gas pressure in Figure 16.19. The remaining 3% corresponds to secondary electrons and direct interception of negative ions by the GG. Thus the operational gas pressure should be as low as possible, when consequently negative ions will be efficiently produced.

Figure 16.19 Ratio of H– ion current to the acceleration drain current, IH–/Iacc, and the ratio of the equivalent heat load current to the ground grid to the H– ion current, Igg/IH–, as a function of gas pressure in the arc chamber. This figure is modified from the original one in Ref. [39].

365

366


16.5.3

Giant Negative Ion Sources

Negative ion sources utilized in NBI systems are quite large. Figures 16.20 and 16.21 show photographs and a schematic, respectively, of the giant negative ion source [40] for the Large Helical Device (LHD) NBI system [41]. Beam energy and current specifications are 180 keV and 30 A, respectively, and the beam species is hydrogen. The source is equipped with an external magnetic filter, and the arc chamber is 35 cm 145 cm in cross section and 21 cm in depth, surrounded by a multicusp magnetic field with permanent magnets aligned perpendicular to the beam axis. The external magnetic filter is generated as a transverse magnetic field in front of the PG electrode by a pair of permanent magnet rows facing each other with separation 35 cm. The filter field strength at the center is about 50 G. Twenty four tungsten filaments are used for the high power arc discharge of more than 200 kW. Three cesium ovens are attached to the backplate of the arc chamber, and the Cs injection rate is controlled by the oven temperature. The accelerator is a three-grid, single-stage system with 770 beamlet holes. The grid area is 25 cm 125 cm and is divided into five sections, each of which is mounted at a

Photographs of the giant negative ion source for the NBI system of the LHD (Large Helical Device) fusion machine. The left photograph is taken from the arc chamber side, and the right is taken from the downstream side, showing the ground grid.

Figure 16.20


Figure 16.21 Schematic diagram of the giant negative ion source in the LHD-NBI system. This diagram is modified from the original one in Ref. [40].

small angle so as to face a common focal point 13 m downstream. The PG is made of molybdenum and thermally insulated to maintain the PG temperature above 200 C. The EG and the GG are made of copper and water cooling pipes are embedded for active cooling of the grid electrodes. Permanent magnets are embedded in the EG along the cooling pipes for electron suppression. The inside of the EG aperture is specially shaped so as to prevent secondary electrons from entering the acceleration gap, and therefore the ESG is not used. The GG apertures are displaced so that all beamlets arrive at a common focal point and so that alternate beamlet deflection by the EG magnetic field is compensated. The H– current increases linearly with arc power and reaches 25 A. A higher energy giant negative ion source is shown in Figure 16.22, which is used in the JT60U NBI system [42]. The nominal deuterium beam energy and current are 500 keV and 22 A, respectively. The filter magnetic field is formed by a 3–5 kA current passing directly through the PG (PG filter). The overall size is 2 m in diameter and 1.7 m in height. The beam area is 45 cm 110 cm, in 5 segments with 216 apertures. The three-stage accelerator has six grid electrodes and operates at low gas pressure, < 0.3 Pa, in a high-confinement arc chamber where the surfaceto-volume ratio is minimized by the semi-circular shape. A MeV acceleration negative ion source has been conceptually designed for the NBI system for ITER (international thermonuclear experimental reactor), as shown

367

368


Illustration of the high energy giant negative ion source used in the JT60U-NBI system. From Ref. [42], courtesy of the author.

Figure 16.22

in Figure 16.23 [43]. The designed beam energy and current are 1 MeV and 40 A, respectively, and the beam species is deuterium. The extracted negative ion beamlets are accelerated to 1 MeV with a five-stage acceleration system (multi-aperture and multi-grid acceleration) in the reference design. The size of the grid electrode is 1.64 m 0.6 m containing 1300 apertures, and the accelerator is huge due to the insulation and shield structure of each grid support. As a result, the size of ion source is about 3.3 m in diameter and about 2.6 m in height. To simplify the accelerator structure and to reduce the cost of high voltage components, single gap acceleration of merged beamlets is also proposed and is being developed as an alternative design [44]. In this case the acceleration gap is around 1 m.

Illustration of the MeV negative ion source designed for the NBI system for ITER (international thermonuclear experimental reactor). From Ref. [43], with permission from AIP.

Figure 16.23

16.6 Future Directions of Development

16.6

Future Directions of Development

We have described the giant ion sources that have been developed for neutral beam injection systems, laying emphasis on large volume source plasma production and large area beam extraction and acceleration. There are almost no problems remaining to be solved in the large positive ion sources as presently used in NBI systems. The RF-driven ion source, which will be utilized in future because of its simplicity and long lifetime, needs further improvement in the control of electron energy and plasma fluctuations that derive from the RF. Advanced plasma modeling for RFplasma generation should be helpful in the improvement process. On the other hand, a number of problems still remain to be solved for the giant negative ion sources, although these sources are presently used in NBI systems. The magnetic filter field generated in front of the plasma grid degrades the plasma uniformity, and this problem must be solved for improvement of beam uniformity. The suppression of electron acceleration is quite important for reducing the heat load on the downstream grid. Reduction of the operational gas pressure is required in addition to sophisticated grid design for suppressing secondary electrons. The ITER NBI system calls for a 1 MeV negative ion beam, and the structure of the accelerator support including high voltage insulation and power feeding should also be developed for the electrostatic acceleration of high current ion beams. In the future fusion reactor the NBI must operate continuously over one year, and for long lifetime a continuous source utilizing the RF or microwave should be developed. Recent development of three-dimensional (3D) computer simulation techniques for ion sources should be useful for these challenges. Simulation of plasma production by following primary electron trajectories would be helpful for improvement of plasma uniformity, and following negative ion beam trajectories including secondary electrons generated on the grid surface would also provide a powerful tool for optimization of the accelerator. The principles and technology of giant ion sources contribute to the general development of ion sources through their sophisticated physics and their highly advanced techniques.

369

370


References [1] J.G. Cordey, J. Hugill, J.W.M. Paul, J. Shef-

[2]

[3] [4] [5]

[6]

[7] [8] [9] [10] [11] [12] [13] [14] [15] [16]

[17] [18] [19] [20]

field, E. Speth, P.E. Stott, V.I. Tereshin, Nucl. Fusion 14, 441 (1974). K. Ushigusa and JT-60 Team, Proceedings of the 16th International Conference on Fusion Energy, Montreal, 1996 (International Atomic Energy Agency, Vienna, 1997), Vol. 1, p. 37. R. Limpaecher and K.R. MacKenzie, Rev. Sci. Instrum. 44, 726 (1973). W.L. Stirling, P.M. Ryan, C.C. Tsai and K.N. Leung, Rev. Sci. Instrum. 48, 533 (1977). R.S. Hemsworth, D.A. Aldcroft, T.K. Allen, D.V. Bayes, J.N. Burcham, H.C. Cole, M.C. Cowlin, J.C. Coultas, J.H. Hay and W.J. McKay, Proceedings of the Joint VarennaGrenoble International Symposium on Heating in Toroidal Plasmas, Grenoble, 1978 (Pergamon Press, 1978), Vol. 1, p. 83. A.P.H. Geode, T.S. Green, A.R. Martin, G.A.G. Mosson and B. Singh, Proceedings of the Joint Varenna-Grenoble International Symposium on Heating in Toroidal Plasmas, Grenoble, 1978 (Pergamon Press, 1978), Vol. 1, p. 77. K.W. Ehlers and K.N. Leung, Rev. Sci. Instrum. 50, 1353 (1979). A.P.H. Geode and T.S. Green, Phys. Fluids 25, 1797 (1982). D. M. Goebel, Phys. Fluids 25, 1093 (1982). K.N. Leung, N. Hershkowitz and K.R. MacKenzie, Phys. Fluids 19, 1045 (1976). C.F. Chan, C.F. Burrell and W.S. Cooper, J. Appl. Phys. 54, 6119 (1983). Y. Okumura, H. Horiike and K. Mizuhashi, Rev. Sci. Instrum. 55, 1 (1984). K.W. Ehlers and K.N. Leung, Rev. Sci. Instrum. 52, 1452 (1981). I. Langmuir and K.G. Blodgett, Phys. Rev. 24, 49 (1924). A.J.T. Holmes and E. Thompson, Rev. Sci. Instrum. 52, 172 (1981). Y. Okumura, Y. Ohara, and T. Ohga, JAERI-M 7696 (Japan Atomic Energy Research Institute, Japan, 1978). G.J. Davisson and C.J. Calbick, Phys. Rev. 38, 585 (1931). J.R. Coupland and T.S. Green, Nucl. Instr. Methods 125, 197 (1975). J.H. Whealton, Rev. Sci. Instrum. 48, 1428 (1977). J.R. Conrad, Rev. Sci. Instrum. 51, 418 (1980).

[21] M.C. Vella, W.S. Cooper, P.A. Pincosy,

R.V. Pyle, P.D. Weber and R.P. Wells, Rev. Sci. Instrum. 59, 2357 (1988). [22] T.S. Green, J.R. Coupland, D.P. Hammond, A.J.T. Holmes, A.R. Martin, R.S. Hemsworth, E. Thompson, Proceedings of the 10th International Conference on Plasma Physics and Controlled Nuclear Fusion Research, London, 1984 (IAEA, Vienna, 1985), Vol. 3, p. 319. [23] A.J.T. Holmes, T. S. Green and A. F. Newman, Rev. Sci. Instrum. 58, 1369 (1987). [24] M. Akiba, M. Araki, H. Horiike, T. Ito, M. Kawai, M. Kuriyama, S. Kitamura, S. Matsuda, M. Matsuoka, H. Mukaida, Y. Oguchi, Y. Ohara, T. Ohga, H. Ohtsuki, Y. Okumura, K. Shibanuma, T. Shibata, H. Shirakata and S. Tanaka, Rev. Sci. Instrum. 53, 1864 (1982). [25] S. Tanaka, M. Akiba, H. Horiike, M. Matsuoka, Y. Ohara and Y. Okumura, Rev. Sci. Instrum. 57, 145 (1986). [26] E. Speth, M. Ciric, J.H. Feist, P. Frank, B. Heinemann, W. Kraus, F. Probs, R. Riedl, R. Trainham, O. Vollmer and R. Wilhelm, Fusion Eng. Design 46, 383 (1999). [27] J.E. Osher, F.G. Gordon, G.W. Hamilton, Proceedings of the 2nd International Conference on Ion Sources, Vienna, 1972, p. 876. [28] Yu.I. Belchenko, G.I. Dimov and V.G. Dudnikov, Nucl. Fusion 14, 113 (1973). [29] K.N. Leung and K.W. Ehlers, Rev. Sci. Instrum. 53, 803 (1980). [30] M. Bacal and G. W. Hamilton, Phys. Rev. Lett. 42, 1538 (1979). [31] K.N. Leung, K.W. Ehlers and M. Bacal, Rev. Sci. Instrum. 54, 56 (1983). [32] G. Dammertz and B. Piosczyk, Proceedings of the 4th International Symposium on Heating in Toroidal Plasmas, Rome, 1984 (Italian Commission for Nuclear and Alternative Energy Sources, 1984), Vol. 2, p. 1087. [33] M. Hanada, T. Inoue, H. Kojima, Y. Matsuda, Y. Ohara, Y. Okumura, K. Watanabe and M. Seki, Rev. Sci. Instrum. 61, 499 (1990). [34] S.R. Walther, K.N. Leung and W.B. Kunkel, J. Appl. Phys. 64, 3424 (1988). [35] Y. Okumura, M. Hanada, T. Inoue, H. Kojima, Y. Matsuda, Y. Ohara, Y. Oohara, M. Seki, Y. Suzuki and K. Watanabe, Proceedings of the 16th Symposium on Fusion Technology, London, 1990, (North-Holland, Amsterdam, 1991), Vol. 2, p. 1026.

References [36] Y. Okumura, M. Hanada, T. Inoue, H. Kojima, [41] O. Kaneko, Y. Takeiri, K. Tsumori, Y. Oka,

[37]

[38]

[39]

[40]

Y. Matsuda, Y. Ohara, M. Seki and K. WataM.Osakabe, R. Akiyama, T. Kawamoto, nabe, Proceedings of the 5th International SymE. Asano and T. Kuroda, Proceedings of the posium on Production and Neutralization of 16th International Conference on Fusion Energy, Negative Ions and Beams, Brookhaven, NY, Montreal, 1996 (International Atomic Energy 1989, AIP Conference Proceedings No. 210, Agency, Vienna, 1997), Vol. 3, p. 539. [42] M. Kuriyama, N. Akino, T. Aoyagi, M. Araki, p. 169. N. Ebisawa, Y. Fujiwara, A. Honda, T. Inoue, A. Ando, K Tsumori, Y. Takeiri, O. Kaneko, M. Kawai, M. Kazawa, J. Koizumi, K. MiyaY. Oka, T. Okuyama, H. Kojima, Y. Yamashita, moto, N. Miyamoto, K. Mogaki, Y. Ohara, R. Akiyama, T. Kawamoto, K. Mineo, T. KurT. Ohga, Y. Okumura, H. Oohara, K. ata, and T. Kuroda, Proceedings of the 6th InterOhshima, F. Satoh, K. Shimizu, S. Takahashi, national Symposium on Production and NeutraH. Usami, K. Usui, K. Watanabe, M. Yamalization of Negative Ions and Beams, Upton, moto, T. Yamazaki, Y. Ono and S. Kawashima, NY, 1992, AIP Conference Proceedings No. Proceedings of the 16th IEEE/NPSS Symposium 287, p. 339. Y. Takeiri, O. Kaneko, Y. Oka, K. Tsumori, on Fusion Engineering, Illinois, 1995 (IEEE, E. Asano, R. Akiyama, T. Kawamoto, T. Kur1995), Vol. 1, p. 491. oda and A. Ando, Rev. Sci. Instrum. 66, 5236 [43] R.S. Hemsworth, J.-H. Feist, M. Hanada, (1995). B. Heinemann, T. Inoue, E. Kussel, A. Krylov, P. Lotte, K. Miyamoto, N. Miyamoto, D. MurY. Takeiri, Y. Oka, M. Osakabe, K. Tsumori, doch, A. Nagase, Y. Ohara, Y. Okumura, O. Kaneko, T. Takanashi, E. Asano, T. KawaJ. Pamela, A. Panasenkov, K. Shibata, M. Tanii moto, R. Akiyama and T. Kuroda, Rev. Sci. and M. Watson, Rev. Sci. Instrum. 67, 1120 Instrum. 68, 2003 (1997). (1996). Y. Takeiri, M. Osakabe, K. Tsumori, Y. Oka, O. Kaneko, E. Asano, T. Kawamoto, R. [44] A. Simonin, J. Bucalossi, C. Desgranges, Akiyama and M. Tanaka, J. Plasma Fusion M. Fumelli, C. Jacquot, P. Massmann, Res. 74, 1434 (1998). J. Pamela, D. Riz, R. Trainham and Y. Belchenko, Rev. Sci. Instrum. 67, 1102 (1996).

371

373

Appendices Appendix 1

Physical Constants c e0 l0 e me mp u h r k NA n0 a0 e/k

Speed of light in free space 2.9979 108 m s–1 –12 Permittivity of free space 8.8542 10 F m–1 –7 Permeability of free space 4p 10 H m–1 Electron charge 1.6022 10–19 C Electron mass 9.1094 10–31 kg Proton mass 1.6726 10–27 kg Atomic mass unit, amu 1.6605 10–27 kg Planck’s constant 6.6261 10–34 Js J m–2 s–1K–4 Stefan–Boltzmann constant 5.6705 10–8 –23 Boltzmann’s constant 1.3806 10 J K–1 23 Avogadro’s number 6.0221 10 Molecules mole–1 Loschmidt’s number 2.6868 1019 Molecules cm–3 –8 Bohr radius (radius of hydrogen atom) 0.5292 10 cm Temperature associated with 1 eV 11,605 K Joules per eV 1.6022 10–19 J eV–1 Wavelength of a 1 eV photon 12,399 1 amu = 931.49 MeV me = 0.5110 MeV mp/me = 1836.15 p = 3.14159265 e = 2.71828183 Molecular density at 20 C and 1 mTorr pressure = 3.3 1013 molecules cm–3

The Physics and Technology of Ion Sources. Ian G. Brown (Ed.) Copyright 2004 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 3-527-40410-4

374

Appendices

Appendix 2

Some Plasma Parameters fpe = 8.98 103 ne1/2 fpi = 2.10 102 QA–1/2ni1/2 fce = 2.80 106 B fci = 1.52 103 BQ/A vThe = 6.69 107 Te1/2 vThi = 1.57 106 (Ti/A)1/2 cs = 9.79 105 (cQTe/A)1/2 e = 3.81 102 Te1/2/B i = 1.64 102 (ATi)1/2/(QB) kD = 7.43 102 (Te/ne)1/2

Electron plasma frequency Ion plasma frequency Electron cyclotron frequency Ion cyclotron frequency Electron mean thermal velocity Ion mean thermal velocity Ion sound speed Electron cyclotron radius Ion cyclotron radius Debye length

Hz Hz Hz Hz cm s–1 cm s–1 cm s–1 cm cm cm

In these expressions ni and ne are the ion and electron densities in cm–3, Ti and Te are the ion and electron temperatures in eV, B is the magnetic field strength (flux density) in G, Q is the ion charge state, A is the ion mass in amu, and c is the specific heat ratio. The cyclotron radii are given in terms of mean thermal velocities, r = vTh/xc.

Appendix 3

Table of the Elements

Element

Symbol Z

A

MP (C)

BP (C)

Hydrogen Helium Lithium Beryllium Boron Carbon Nitrogen Oxygen Fluorine Neon Sodium Magnesium Aluminum Silicon Phosphorus Sulfur Chlorine Argon Potassium Calcium Scandium Titanium

H He Li Be B C N 0 F Ne Na Ng Al Si P S Cl Ar K Ca Sc Ti

1 4 6.9 9 10.8 12 14 16 19 20.2 23 24.3 27 28.1 31 32.1 35.5 39.9 39.1 40.1 45 47.9

–259 –272 180 1290 2300 3550 –210 –218 –220 –249 98 649 660 1410 44 115 –100 –189 64 840 1540 1660

–253 –269 1350 2970 2550(s) 4800 –196 –183 –188 –246 890 1090 2470 2400 280 445 –35 –186 760 1480 2830 3290

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Electrical Ionization Density Resistivity Potential (g cm–3) (lX cm) (eV)

0.53 1.85 2.34 2.26

8.5 4.0 4E12 1375

0.97 1.74 2.70 2.33 1.8–2.7 2.07

4.2 4.45 2.65 1E17 2E23

0.86 1.55 3.00 4.54

6.15 4.0 51 42

13.60 24.59 5.39 9.32 8.30 11.26 14.55 13.62 17.42 21.56 5.14 7.65 5.98 8.15 10.48 10.36 13.02 15.76 4.34 6.11 6.56 6.84

Appendix 3 Table of the Elements

Element

Symbol Z

A

MP (C)

BP (C)

Vanadium Chromium Manganese Iron Cobalt Nickel Copper Zinc Gallium Germanium Arsenic Selenium Bromine Krypton Rubidium Strontium Yttrium Zirconium Niobium Molybdenum Technetium Ruthenium Rhodium Palladium Silver Cadmium Indium Tin Antimony Tellurium Iodine Xenon Cesium Barium Lanthanum Cerium Praseodymium Neodymium Promethium Samarium Europium Gadolinium Terbium Dysprosium Holmium

V Cr Mn Fe Co Ni Cu Zn Ga Ge As Se Br Kr Rb Sr Y Zr Nb Mo Tc Ru Rh Pd Ag Cd In Sn Sb Te I Xe Cs Ba La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho

50.9 52 54.9 55.8 58.9 58.7 63.5 65.4 69.7 72.6 74.9 79 79.9 83.8 85.5 87.6 88.9 91.2 92.9 95.9 98.9 101.1 102.9 106.4 107.9 112.4 114.8 118.7 121.7 127.6 126.9 131.3 132.9 137.3 138.9 140.1 140.9 144.2 145 150.4 152 157.2 158.9 162.5 164.9

1890 1860 1245 1535 1495 1453 1083 420 30 937 817(a) 217 –7 –157 39 770 1520 1852 2470 2620 2172 2310 1966 1554 962 321 157 232 631 450 113 –112 28 725 920 798 931 1020 1080 1075 822 1311 1360 1410 1470

3380 2670 1960 2750 2870 2800 2570 907 2400 2850 613(s) 685 59 –152 687 1380 3340 4380 4740 4610 4880 4200 3730 2950 2210 765 2080 2300 1700 990 184 –107 680 1700 3450 3460 3510 3070 2460 1800 1600 3250 3200 2600 2720

23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67

Electrical Ionization Density Resistivity Potential (g cm–3) (lX cm) (eV) 6.11 7.18 7.4 7.87 8.92 8.90 8.95 7.13 5.91 5.32 5.73 4.79 3.12

25 13 185 9.71 6.25 6.84 1.68 5.9 17.4

1.53 2.54 4.46 6.51 8.57 10.22 11.5 12.45 12.41 12.02 10.50 8.65 7.31 7.31 6.62 6.24 4.93

12.5 23 60 40 12.5 5.2

1.89 3.50 6.15 6.77 6.77 7.00

20 50 80 75 68 64

7.54 5.25 7.90 8.23 8.55 8.78

105 90 140 115 90 82

33 12

7.2 4.51 10.5 1.59 7.0 8.37 11 40 4.4E5 1E15

6.74 6.76 7.43 7.90 7.86 7.63 7.73 9.39 6.00 7.88 9.81 9.75 11.85 14.00 4.18 5.69 6.53 6.95 6.88 7.10 7.28 7.36 7.45 8.33 7.58 8.99 5.78 7.34 7.84 9.01 10.45 12.13 3.89 5.21 5.61 5.57 5.42 5.45 5.55 5.60 5.64 6.16 5.98 5.93 6.02

375

376

Appendices Electrical Ionization Density Resistivity Potential (g cm–3) (lX cm) (eV)

Element

Symbol Z

A

MP (C)

BP (C)

Erbium Thulium Ytterbium Lutetium Hafnium Tantalum Tungsten Rhenium Osmium Iridium Platinum Gold Mercury Thallium Lead Bismuth Polonium Astatine Radon Francium Radium Actinium Thorium Protactinium Uranium

Er Tm Yb Lu Hf Ta W Re Os Ir Pt Au Hg Tl Pb Bi Po At Rn Fr Ra Ac Th Pa U

167.3 168.9 173 175 178.5 180.9 183.8 186.2 190.2 192.2 195.1 197 200.6 204.4 207.2 209 210 210 222 223 226 227 232 231 238

1530 1545 820 1660 2225 3000 3410 3180 3040 2410 1772 1064 –39 303 327 271 254 302 –71 27 700 1050 1750

2900 1950 1200 3400 4600 5425 5660 5650 5030 4430 3900 2810 357 1460 1750 1560 965 337 –62 677 1200 3200 4790

9.05 9.32 6.96 9.84 13.20 16.65 19.30 21.02 22.57 22.65 21.45 19.30 13.55 11.85 11.35 9.75

11.72

13

5.28 6.90 6.20

1132

3900

18.95

30

6.20

68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92

86 70 25 58 35 13 5.6 19.3 9.5 5.3 10.6 2.21 98 18 20.6 107

6.10 6.18 6.22 6.15 6.80 7.88 7.98 7.87 8.75 9.05 8.96 9.23 10.44 6.11 7.41 7.29 8.43 10.75

MP, melting temperature; BP, boiling temperature; (s), sublimes; (a), at a pressure of 28 atm. Electrical resistivity is at or near 20 C.

377

Index a Accel-decel extraction system 35, 42, 95, 99 Aluminosilicate sources 322ff Analyzing magnet 91 ANIS ion source 297 Atomic fraqction 347 Auxiliary discharge 118 AXCEL code 41ff, 64ff, 72

b Beam halo 97 Beam transport 61, 87ff Beam quality 65 Beam rotation 91, 223 Beamlet steering 336, 351, 364 Bernas ion sources 133ff, 138, 151 Broad beam formation 34, 127 Bucket type ion sources 114, 120, 167, 295, 329, 342, 355, 357

c Calutron ion sources 133 Cathode sputtering 146, 154 Charge exchange 27, 37, 94, 211, 306, 327 Charge neutrality 15 Chasman Lapostolle 66 Child-Langmuir law 61, 184, 263, 314, 348 CLPS (Common Long Pulse Source) 355 CO2 laser 235ff, Cold cathode ion sources 121, 128 Collisions 12, 93, 213 Compound extraction 99 Computer simulation of extraction 41ff Conservative forces 88 Contamination 136 Contact ionizer 318 Control systems 154, 155 Coupland’s law 70

Critical density 19, 178, 180, 183, 215, 217 Cyclotron radius 17

d Debye length 14 Discharge energy efficiency 108 Dissociative electron attachment 303, 359 Divergence angle 68, 119, 127, 264 Drift 91 Duopigatron 44, 112 Duoplasmatron 44, 110

e ECR heating 18, 214 ECR ion sources 46, 177, 203ff AECR-U 206, 207, 216 CAPRICE 203, 206 CIRCE 205 GTS 206 MicroMAFIOS 203 MiniMAFIOS 205 PHOENIX 218 RIKEN 18GHz 203, 206 SERSE 206, 209, 218 SNanogan 206 SUPERMAFIOS 204, 214 VENUS 206, 208, 225 Einzel lens 99, 316 Electrode cleaning 160 Electron affinity 287 Electron cyclotron frequency 18, 216 Electron detachment 286, 288 Electron impact ionization 22, 210 Electron sources 54 Electron temperature 9 Electrostatic quadrupole 317 Elements (properties) 374 E-Mevva ion source 270 Emittance 65, 82, 87, 223, 226, 246, 264


Index

378

Ion cyclotron frequency 18 Ion extraction 42, 61ff, 107 Ion impact ionization 26 Ion implantation 133, 185, 195 Ionization 21 Ion temperature 9

4-rms 67 100% ellipse 87 Fraction of current 87 Growth 47, 338 Normalized 88 Rms 66, 87 Extraction 31, 61, 92, 234, 240, 262 Accel-decel 35, 42, 95, 99 Beam current 62 Diode 62 Multi-aperture 34, 61, 76, 108, 184, 348 Multi-stage 349 Pentode 98, 101 Tetrode 349 Triode 64, 102, 356

j

f

l

Feed gases 159 Filament driven ion sources 43, 110ff, 133ff Filaments 137, 148 Filament sputtering 135 Fractional ionization 9 Freeman ion sources 133ff

Laminar ion beam 101 Lamor radius 17, 123, 214 Langmuir probe 183 Laser ion sources 45, 233ff, 333 Laser produced plasmas 234, 244, Lifetime 158, 188, 247, 267 Liouville’s theorem 65, 88 Lotz formula 24, 210 Low energy beam line (LEBT) 92, 98, 242

g Gas efficiency 109 Gas mixing 215 Giant ion sources 341ff, 366 Glow discharge 122ff Gyro frequency 18 Gyro orbit 17, 123, 214

h H– sources 55 Hash 134, 273 Hexapole cusp field 47, 208 HIF (Heavy ion fusion) ion sources 311ff High current ion sources 107 High frequency ion sources 116 Hollow discharge duoplasmatron 292 Hot cathode electron sources 55

i ICIS (International Conference on Ion Sources) 5 IFMIF 72 Impedance matching 179, 182 Indirectly heated cathode 139 Induction linac 312 Inductively coupled sources 165 Inertial fusion ion sources 311ff Insulator cleaning 160 Ion beam formation 29, 62

JET (Joint European Torus) 355

k Kapchinskij-Vladimirskij distribution 67 Kaufman ion source 112 Kilpatrick’s law 70 KOBRA code 41ff

m Magnetic confinement 19, 212 Magneto-electrostatic confinement 113 Magnetic filter 168, 303, 347, 355, 357, 360 Magnetic quadrupole 89 Magnetic mirror 19, 181, 212 Magnetic moment 20, 212 Magnetic pressure 20 Magnetic solenoid 90 Magnetron source 134, 293 Maxwellian distribution 10 Mean free path 13 Metal ion beams 218, 257ff Direct insertion 219 Efficiency 222 MIVOC method 220 Ovens 221 Sputtering 220 Vacuum arc 257, 271 Mevva ion source 276ff, 330 Microwave discharges 177 Absorption 182 Off-resonance 180, 183 Overdense plasmas 19, 178, 180 Microwave ion sources 177ff Microwave plasma cathode 121

Index Minimum-B magnetic field 205, 213, 215 MOS 194 Mller-Salzborn formula 210ff Multicusp magnetic field 19, 109, 295, 329, 342, 343 Multicusp bucket sources 114, 120, 167, 295, 329, 342, 355, 357 Multiple ionization 23, 126, 206, 245, 266 Multiply charged ions 8, 23, 204, 210, 242, 245, 249, 265

n Negative ions 27, 285 Charge transfer 306 Extraction 363 Secondary emission 290 Surface production 27, 285, 292, 296 Volume production 27, 303, 305, 359, 362 Negative ion sources 54, 285ff, 333 Neutral beams 341 NIABNIS ion source 299

o Off-resonance plasma 180 Overdense plasma 19, 178, 180, 183, 215, 217

p Particle in cell (PIC) codes 41 Penning ion source 50 Periplasmatron ion source 112 Perveance match 68 Phase space 87 Physical constants 373 Photoionization 25 Pierce geometry 46, 68, 320 PIG discharge 22, 50, 111, 134 Negative ion source 294 PINI (Plug-in neutral injector) 355 Plasma boundary 42, 95 Location 42 Meniscus 69 Thickness 42 Shape 42 Plasma cathode 121, 126 Plasma density 8 Plasma electron sources 55 Plasma frequency 16, 178, 215 Plasma generator 29, 61 Plasma gun 30 Plasma oscillations 16 Plasma parameters 374 Plasma potential 15, 33, 138, 213

Plasma pressure 20 Plasma sheath 13, 144 Post acceleration 98, 102 Proton LINAC 98

r Radioactive ion beams 228 Radioactive isotopes 203 Raduga ion source 278 Refractive index 179 Resolving power 141 Richardson-Duschmann equation 150, 322 RIM ion sources 117, 166 RF (Radio frequency) ion sources 117, 163ff, 302, 357

s Saha-Langmuir equation 318 Shottky effect 150 SIMOX 198 198 SOI 191 Space charge force 92 Space charge compensation/neutralization 93, 95 Maintenance of 95 Residual gas collisions 93 Sputtering 94 Sputtering 291 Suppressor 35, 64, 99 Surface ionization 285, 318, 326

t Target illumination 237 Thermionic emission 322 Titan ion source 277 Trajectory codes 41 Transfer matrix 88ff Transverse particle energy 9, 50 Twiss parameters 88

u ULSI 194 UNIS ion source 296

v Vacuum arc ion sources 38, 44, 257ff, 330 Vacuum arc plasmas 259 Varis ion source 279 VNIS ion source 298

w Work function 287

379

The Physics and Technology of Ion Sources

The Physics and Technology of Ion Sources

Ion Exchange Technology II: Applications

Ion Exchange Technology I: Theory and Materials

The physics of information technology

The Physics of Information Technology

Ion Plating Technology: Developments and Applications

Renewable Energy: Sources and Methods (Green Technology)

Ion

ion

Ion

Ion

Ion

Ion

Ion

Ion

Fiber Optics: Physics and Technology

Fiber Optics: Physics and Technology

Advanced Technology and Particle Physics

Paper Products Physics and Technology

Advanced Technology and Particle Physics

Photonic crystals: physics and technology

Fiber Optics: Physics and Technology

Vacuum Physics and Technology (Methods of Experimental Physics)

Ion

Ion

Ion

Ion

Ion

Ion

The Sources of Normativity

The Physics and Technology of Ion Sources