Contributions to ECR Plasma Source Dynamics ...

Doctoral Thesis

Contributions to ECR Plasma Source Dynamics: Diagnostics Development and Experimental Results

Author: Ana Mar´ıa Meg´ıa Mac´ıas

Supervisor: Prof. Osvaldo Daniel Cortázar

University of Castilla-La Mancha Doctoral International School Superior Technical School of Industrial Engineering Department of Applied Mechanics and Project Engineering

A thesis submitted in fulfillment of the requirements for the degree of Doctor of Science and Technology Applied to Industrial Engineering

July 2014

Tesis Doctoral

Aportaciones a la Din´ amica de Fuentes de Plasma ECR: Desarrollo de Diagn´ osticas y Resultados Experimentales

Autora: Ana Mar´ıa Meg´ıa Mac´ıas

Director: Prof. Osvaldo Daniel Cortázar

Universidad de Castilla-La Mancha Escuela Internacional de Doctorado ćnica Superior de Ingenieros Industriales Escuela Te ´ nica Aplicada e Ingenier´ıa de Proyectos Departamento de Meca

Tesis presentada como requisito para acceder al t´ıtulo de Doctora en Ciencias y Tecnolog´ıas Aplicadas a la Ingenier´ıa Industrial

Julio 2014

A mis abuelas, Bienve y Pepita

Acknowledgments This thesis is the result of four years of work in which I have been accompanied and encouraged by many people, to all of them, thank you very much. In particular I would like to thank Daniel Cortázar for having helped me so much and above all, for making our work so enjoyable over the years. The time that my PhD work has lasted has been a period full of emotions and good times, it has been an exciting and rewarding challenge that has made me grow professionally. All of this is largely thanks to him, and I must say that the work presented here is as much ours as mine. Thank you very much, Daniel. I also want to thank my family, who have always been part of my success and have provided my main support. Special thanks go to my grandmothers, to whom I dedicate this thesis. They have been present in every important moment of my life ever since I can remember. I am equally grateful to my parents and my brother; without them, this would not have been possible . There have been many people without whose support it would have been impossible to carry out the experiments that form the basis of this thesis, I want to express my gratitude to all of them: to Joan Bordas and Javier Bermejo, for the support they have given us, for believing in us and for giving us the opportunity to undertake this journey; to José Alonso for his support and advice, which was always so useful; to Olli Tarvainen, for first listening to our ideas that hot afternoon in Sicily; to Hannu Koivisto, for giving us the opportunity to work on his team, which was an unforgettable experience; and to Janni Komppula, who came to work with us and shared such good times. I also want to thank everyone who has encouraged me every day: Sira Cordon, for always supporting me; Carmen Abaitua, for that optimism that was so necessary sometimes; Iker Etxebarria, for being so genuine; Maider Camarero, for her sincerity and enthusiasm, Tomaso for his useful suggestions and Roberto Martinez, who was always ready with a good idea. I can not forget Pedro Jimenez, who dedicated so much of his time to us, or Pedro Hungr´ıa, who was always so kind and has been the life and soul of the INEI during this past two years. To both of them, thank you very much. Finally I want to thank Kieron Spackman who has read and reread each page of this thesis with me. Thanks, Kieron. Although these years of work have been fantastic, they have not been without obstacles. Each obstacle has meant a need to go a little further and that, too, has been uplifting. To all who have ever put an obstacle in our way, thank you very much.

Agradecimientos Esta tesis es el fruto de cuatro de a˜ nos de trabajo en los que he estado acompa˜ nada y alentada por mucha gente, a todos ellos, muchas gracias. De un modo particular me gustar´ıa agradecerle a Daniel Cortázar cuánto me ha ayudado y, sobre todo, cu´ anto hemos disfrutado del trabajo durante estos a˜ nos. El tiempo que ha durado mi doctorado ha sido un periodo lleno de emociones y de buenos momentos, ha sido un reto emocionante y enriquecedor que me ha hecho crecer profesionalmente. Todo ello es, en gran parte, mérito suyo y he de decir que el trabajo que aqu´ı se presenta es tan m´ıo como nuestro. Muchas gracias, Daniel. También quiero dar las gracias a mi familia que siempre forma parte de mis éxitos y que es mi principal apoyo. En especial a mis abuelas, a quienes dedico esta tesis, que han estado desde que recuerdo en cada momento importante de mi vida. A mis padres y a mi hermano, sin vosotros nada esto habr´ıa sido posible. Muchas han sido las personas sin cuyo apoyo habr´ıa sido imposible llevar a cabo los experimentos que componen esta tesis, a todos ellos quiero darles las gracias. A Joan Bordas y a Javier Bermejo por el apoyo que nos han prestado, por haber cre´ıdo en nosotros y por habernos dado la oportunidad de emprender este camino. A José Alonso por su apoyo y sus consejos, siempre tan acertados. A Olli Tarvainen, por prestar por primera vez o´ıdos a nuestras ideas aquella tarde calurosa de Sicilia. A Hannu Koivisto, por darnos la oportunidad de trabajar en su equipo, fue una experiencia inolvidable. A Janni Komppula, que vino a trabajar con nosotros y nos hizo pasar tan buenos momentos. También quiero dar las gracias a toda la gente que me ha animado cada d´ıa, a Sira Cord´ on por apoyarme siempre, a Carmen Abaitua por ese optimismo que tan necesario se ha hecho a veces, a Iker Etxebarria por ser tan auténtico, a Maider Camarero por su sinceridad, a Tomaso por sus sugerencias y a Roberto Mart´ınez que siempre ha estado dispuesto a aportar una buena idea. No puedo olvidarme de Pedro Jiménez que tantas horas nos ha dedicado ni de Pedro Hungr´ıa, siempre tan amable, que ha sido el esp´ıritu del INEI durante estos dos u ´ltimos a˜ no. A los dos, muchas gracias. Por u ´ltimo quiero dar las gracias a Kieron Spackman que ha le´ıdo y reléido conmigo cada p´ agina de est´ a tésis. Gracias, Kieron. Estos a˜ nos de trabajo, si bien han sido estupendos, no han estado exentos de obst´ aculos. Cada obst´ aculo ha sido una necesidad de llegar un poco más lejos y eso,

también, ha sido edificante. A todos los que alguna vez han puesto un obstáculo en el camino, muchas gracias.

Abstract Presented herein is a record of experimental research work focused on ECR plasma dynamics with applications in ion source engineering. The results were obtained using a series of novel diagnostics that have revealed the existence of phenomena that have never before been observed and that can be used to acquire a deeper understanding of both ion source physics and its applications for engineering. A wide-ranging and systematic study of breakdown times; original time-resolved measurements of plasma parameters at breakdown/decay, including visible and ultraviolet spectroscopy; and the discovery of eight plasma density distribution modes, all combine to contribute to the state of the art. Moreover, the new tools developed offer the possibility to control plasma parameters in real time, which can lead to major improvements in ECR ion source performance.

Resumen Se presenta un trabajo de investigación experimental enfocado a la dinámica de plasmas ECR con aplicaci´ on en la ingenier´ıa de fuentes de iones. Se muestran los resultados obtenidos en el desarrollo de novedosas técnicas de diagnóstico que han revelado fenómenos no observados hasta el momento y que pueden ser utilizados tanto para un mejor entendimiento de f´ısica de las fuentes de iones ECR como para su ingenier´ıa. Un amplio y sistem´ atico estudio de los tiempos de encendido, mediciones originales con resoluci´ on temporal de los parámetros del plasma incluyendo espectroscopia visible y ultravioleta y el descubrimiento de modos de distribución de densidad no observados anteriormente son parte de los principales aportes realizados al estado del arte. Además, las nuevas herramientas desarrolladas ofrecen posibilidades de control en tiempo real de los par´ ametros del plasma que permitirán mejoras sustanciales en el funcionamiento de fuentes de iones ECR.

Contents Acknowledgments

v

Agradecimientos

vii

Abstract

ix

Resumen

xi

Contents

xiii

List of Figures

xvii

List of Tables

xxi

Abbreviations

xxiii

Physical Constants

xxv

Symbols

xxvii

1 Introduction 1.1 ECR Ion Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.1 ECR Ion Source Architecture . . . . . . . . . . . . . . . . . . . . . 1.1.2 Operating Regimes . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 6 7 9

2 TIPS: Test Bench for Ion Source Plasma Studies 15 2.1 TIPS Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3 Study of Plasma Breakdown Times 3.1 Experimental Set-Up . . . . . . . . . . . . . . . 3.1.1 Diagnostic Port . . . . . . . . . . . . . . 3.1.2 Magnetic Field Profiles . . . . . . . . . 3.2 Measurement Procedure . . . . . . . . . . . . . 3.3 Results . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Breakdown Time Measurements with Bz xiii

. . . . . . . . . . . . . . . . . . . . . . . . . > ECR

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

21 22 22 23 26 29 30

Contents . . . . . . . . . .

30 31 32 32 33 34 34 35 36 41

4 Plasma Density and Temperature Measurements during Breakdown 4.1 Experimental Set-Up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 Diagnostic Port . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.2 Magnetic Field Profiles . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Measurement Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Data Analysis and Calculations . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 Difference between Plasma and Floating Potentials . . . . . . . . . 4.3.2 Slope of a Linear Fitting in a Current Logarithm-Voltage Plot . . 4.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 Temperature and Density Evolution with Asymmetric Bz > ECR . 4.4.1.1 Low Pressure Regime . . . . . . . . . . . . . . . . . . . . 4.4.1.2 High Pressure Regime . . . . . . . . . . . . . . . . . . . . 4.4.2 Temperature and Density Evolution with Symmetric Bz ' ECR . 4.4.2.1 Low Pressure Regime . . . . . . . . . . . . . . . . . . . . 4.4.2.2 High Pressure Regime . . . . . . . . . . . . . . . . . . . . 4.5 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . .

45 46 46 48 49 52 53 54 57 58 58 61 62 62 64 66

5 Plasma Density and Temperature Measurements during Decay 5.1 Experimental Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Data Analysis and Calculations . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Temperature and Density Evolution with Asymmetric Bz > ECR . 5.3.1.1 Low Pressure Regime . . . . . . . . . . . . . . . . . . . . 5.3.1.2 High Pressure Regime . . . . . . . . . . . . . . . . . . . . 5.3.2 Temperature and Density Evolution with Symmetric Bz ' ECR . 5.3.2.1 Low Pressure Regime . . . . . . . . . . . . . . . . . . . . 5.3.2.2 High Pressure Regime . . . . . . . . . . . . . . . . . . . . 5.4 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . .

71 72 73 75 76 76 77 78 79 80 81

6 Plasma Vacuum Ultraviolet Emission during 6.1 Experimental Set-Up . . . . . . . . . . . . . . 6.1.1 Diagnostic Port . . . . . . . . . . . . . 6.1.2 Magnetic Field Profile . . . . . . . . . 6.2 Measurement Procedure . . . . . . . . . . . .

83 84 84 86 86

3.4 3.5

3.3.1.1 Low Pressure Regime . . . . . 3.3.1.2 High Pressure Regime . . . . . 3.3.2 Breakdown Time Measurements with Bz 3.3.2.1 Low Pressure Regime . . . . . 3.3.2.2 High Pressure Regime . . . . . 3.3.3 Breakdown Time Measurements with Bz 3.3.3.1 Low Pressure Regime . . . . . 3.3.3.2 High Pressure Regime . . . . . Simple Model of Breakdown Time . . . . . . . Summary and Conclusions . . . . . . . . . . . .

. . . . . . . . . . ' ECR . . . . . . . . . . < ECR . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

Breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

. . . .

. . . . . . . . . .

. . . .

. . . . . . . . . .

. . . .

. . . . . . . . . .

. . . .

. . . . . . . . . .

. . . .

. . . . . . . . . .

. . . .

. . . . . . . . . .

. . . .

. . . .

Contents 6.3 6.4

Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

7 Influence of Microwave Driver Coupling Design on Plasma Parameters 99 7.1 Preliminary Design Description . . . . . . . . . . . . . . . . . . . . . . . . 100 7.2 Optimized Design Description . . . . . . . . . . . . . . . . . . . . . . . . . 104 7.3 Design Comparison and Experimental Results . . . . . . . . . . . . . . . . 109 7.3.1 Electric Field Distribution . . . . . . . . . . . . . . . . . . . . . . . 109 7.3.2 Magnetic Field Distribution . . . . . . . . . . . . . . . . . . . . . . 113 7.3.3 Beta Coupling Parameters . . . . . . . . . . . . . . . . . . . . . . . 115 7.3.4 Density and Temperature Measurements . . . . . . . . . . . . . . . 116 7.4 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 8 Ultra-Fast Pictures as a Method for Ion 8.1 Experimental Set-Up . . . . . . . . . . . 8.2 Results . . . . . . . . . . . . . . . . . . . 8.2.1 General Behavior . . . . . . . . . 8.2.2 Column Mode . . . . . . . . . . 8.2.3 Hourglass Mode . . . . . . . . . 8.2.4 Slug Mode . . . . . . . . . . . . 8.2.5 Flower Mode . . . . . . . . . . . 8.2.6 Full-Chamber Mode . . . . . . . 8.2.7 Ring Mode . . . . . . . . . . . . 8.2.8 Yin-Yang Mode . . . . . . . . . . 8.2.9 Donut Mode . . . . . . . . . . . 8.3 Summary and Conclusions . . . . . . . .

Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Plasma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

9 Plasma Breakdown Evolution through Ultra-Fast 9.1 Experimental Set-Up . . . . . . . . . . . . . . . . . 9.2 Results . . . . . . . . . . . . . . . . . . . . . . . . . 9.2.1 Column Mode . . . . . . . . . . . . . . . . 9.2.2 Hourglass Mode . . . . . . . . . . . . . . . 9.2.3 Slug Mode . . . . . . . . . . . . . . . . . . 9.2.4 Flower Mode . . . . . . . . . . . . . . . . . 9.2.5 Full-Chamber Mode . . . . . . . . . . . . . 9.2.6 Ring Mode . . . . . . . . . . . . . . . . . . 9.2.7 Yin-Yang Mode . . . . . . . . . . . . . . . . 9.2.8 Donut Mode . . . . . . . . . . . . . . . . . 9.2.9 Rotating Plasma Configurations . . . . . . 9.2.9.1 Rotating Yin-Yang Mode . . . . . 9.2.9.2 Rotating Half-Moon Mode . . . . 9.3 Summary and Conclusions . . . . . . . . . . . . . .

Diagnosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Pictures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . . . .

. . . . . . . . . . . .

123 . 124 . 128 . 128 . 131 . 132 . 133 . 134 . 135 . 136 . 138 . 139 . 140

. . . . . . . . . . . . . .

143 . 144 . 146 . 147 . 149 . 151 . 153 . 154 . 155 . 156 . 156 . 157 . 159 . 160 . 162

10 Conclusions 165 10.1 Work Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165

Contents 10.2 Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 10.3 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 11 Conclusiones 173 11.1 Trabajo realizado . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 11.2 Contribuciones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 11.3 Trabajo futuro . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 Bibliography

183

List of Figures 1.1 1.2 1.3 1.4 1.5 1.6 1.7

Simplified Diagram of an Ion Source . . . . . . . . . . Simplified Diagram of an ECR Ion Source . . . . . . . Afterglow Phenomena After the Microwave Shut-Down Ar13+ Afterglow at 16.6 GHz MINIMAFIOS . . . . . Plasma Electrostatic Potential Distribution . . . . . . Plasma Distribution Ultra-Fast Picture . . . . . . . . . Preglow Current at SMIS 37.5 GHz . . . . . . . . . . .

. . . . . . .

2 8 10 11 12 12 13

2.1 2.2 2.3 2.4 2.5 2.6

Photograph of TIPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Section View of TIPS and Main Subsystems . . . . . . . . . . . . . . . . . Section of TIPS Magnetic Field Generation System . . . . . . . . . . . . . Photograph of TIPS with the Magnetic Field Measurement Set-Up Installed Example of 2D Magnetic Field Simulations . . . . . . . . . . . . . . . . . Experimental vs. Simulated Magnetic Field Values . . . . . . . . . . . . .

16 17 18 18 19 20

3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.14 3.15

Section View of TIPS with the Biased Probe Set-Up Installed . . . . . . . Axial Magnetic Field Profiles Used during Experiments . . . . . . . . . . 2D Representation of Simulated Magnetic Profiles used during Experiments Schematic Representation of Diagnostic Set-Up . . . . . . . . . . . . . . . Typical Oscilloscope Signal for Measuring Breakdown Times . . . . . . . Typical Relative Absorbed Power . . . . . . . . . . . . . . . . . . . . . . . Breakdown Time for Bz > ECR and 3.8 x 10−3 mb of Hydrogen Pressure Breakdown Time for Bz > ECR and 6.2 x 10−3 mb of Hydrogen Pressure Breakdown Time for Bz ' ECR and 3.8 x 10−3 mb of Hydrogen Pressure Breakdown Time for Bz ' ECR and 6.2 x 10−3 mb of Hydrogen Pressure Breakdown Time for Bz < ECR and 3.8 x 10−3 mb of Hydrogen Pressure Breakdown Time for Bz < ECR and 6.2 x 10−3 mb of Hydrogen Pressure Calculation of Temporal and Spatial Density Evolution . . . . . . . . . . Calculation of Seed Electron Temporal Density Evolution at r = 0 . . . . Calculation of Breakdown Time for 3.8 x 10−3 mb and 6.2 x 10−3 mb . .

22 24 25 26 27 28 30 31 32 33 34 35 39 39 41

4.1 4.2 4.3 4.4 4.5

Section View of TIPS with the Lagmuir Probe Installed . . . . Photograph of TIPS with the Langmiur Probe Set-Up Installed Langmuir Probe Diagram . . . . . . . . . . . . . . . . . . . . . Lagmuir Probe Photograph . . . . . . . . . . . . . . . . . . . . Axial Magnetic Field Profiles Used during Experiments . . . .

46 47 47 48 48

xvii

. . . . . . . . . . . . . . . . in an ECRIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . . . .

. . . . .

. . . . . . .

. . . . .

. . . . .

List of Figures 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 4.16 4.17 4.18 4.19 4.20 4.21 4.22 4.23

2D Representation of Simulated Magnetic Fields Used during Experiments I-V Curve Adquisition Conceptual Plan . . . . . . . . . . . . . . . . . . . View of a Synchronism Record During Breakdown Process . . . . . . . . . Schematic Representation of the Diagnostics Set-Up . . . . . . . . . . . . Langmuir Probe I-V Curves Obtained during Experiments . . . . . . . . . Slope of a Linear Fitting Langmuir Probe Curve Analysis method . . . . Electron Temperature Results by Two I-V Curve Analysis Methods . . . Electron Density Results by Two I-V Curve Analysis Methods . . . . . . Plasma Parameters and (Pi − Pr )/Pi Ratio Evoultion during Breakdown . Plasma Parameters for Bz > ECR and 3.8 x 10−3 mb of Hydrogen Pressure MW Coupling Time for Bz > ECR and 3.8 x 10−3 mb of Hydrogen Pressure Plasma Parameters for Bz > ECR and 6.2 x 10−3 mb of Hydrogen Pressure MW Coupling Time for Bz > ECR and 6.2 x 10−3 mb of Hydrogen Pressure Plasma Parameters for Bz ' ECR and 3.8 x 10−3 mb of Hydrogen Pressure MW Coupling Time for Bz ' ECR and 6.2 x 10−3 mb of Hydrogen Pressure Plasma Parameters for Bz ' ECR and 6.2 x 10−3 mb of Hydrogen Pressure MW Coupling Time for Bz ' ECR and 6.2 x 10−3 mb of Hydrogen Pressure Cross Section of the Main Physical Processes in a Hydrogen Ion Source .

49 50 51 51 53 55 56 56 58 59 60 61 62 63 63 65 65 68

5.1 5.2 5.3 5.4 5.5 5.6 5.7 5.8

View of a Synchronism Record during Decay Process . . . . . . . . . . . . Langmuir Probe I-V Curves Obtained During Experiments . . . . . . . . Electron Temperature During Decay by Two I-V Curve Analysis Methods Electron Density During Decay by Two I-V Curve Analysis Methods . . . Decay Parameters for Bz > ECR and 3.8 x 10−3 mb Hydrogen Pressure . Decay parameters for Bz > ECR and 6.2 x 10−3 mb Hydrogen pressure . Decay Parameters for Bz ' ECR and 3.8 x 10−3 mb Hydrogen Pressure . Decay Parameters for Bz ' ECR and 6.2 x 10−3 mb Hydrogen Pressure .

72 73 74 74 76 77 79 80

6.1 6.2 6.3 6.4 6.5 6.6 6.7 6.8

Section View of TIPS with VUV Diagnostics Installed . . . . . . . . . . . Photograph of the Experimental Set-Up . . . . . . . . . . . . . . . . . . . 2D Representation of the Magnetic Field Configuration Used in Experiments Schematic Representation of the Diagnostics Set-Up . . . . . . . . . . . . Time-Resolved Signals and Plasma Parameters during Breakdown . . . . Normalized Rates of Ionization and Molecular Excitation . . . . . . . . . Lyman band Light Signals Recorded for Different Powers . . . . . . . . . Lyman-alpha and Lyman band VUV-signals Saturation Time . . . . . . .

84 85 86 87 89 91 94 95

7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8

Section View of TIPS Preliminary Design and Main Subsystems . . . . . 100 Microwave Coupler Ridged Dimension Diagram . . . . . . . . . . . . . . . 101 Thermal Picture of Waveguide Heating Due to Secondary Plasma Presence102 2D Representation of B-Field Used in Preliminary Design Experiments . . 103 Simulated Electric Field Distribution with the Preliminary Design Geometry104 Simulated Electric Field Distribution without Coupler . . . . . . . . . . . 105 Simulated Electric Field Distribution with the Optimized Design Geometry106 One-Step Ridged Coupler Design. 3D View and Ridged Section Dimensions107

List of Figures 7.9 7.10 7.11 7.12 7.13 7.14 7.15 7.16 7.17 7.18 7.19

Section View of TIPS Optimized Design . . . . . . . . . . . . . . . . . . . 108 2D Representation of B-Field Used in Optimized Design Experiments . . 108 Simulated E-Field along Plasma Chamber Axis on the Studied Designs . . 110 Picture of Plasma Sustained in Optimized Design without B-Field . . . . 111 Preliminary and optimized design Pr /Pi ratio comparision . . . . . . . . . 112 Preliminary and optimized design E-field relative change comparision . . . 113 Simulated Axial B-Field along the Plasma Chamber Axis for both Designs 113 Resonant Electric Field Distributions and Resonant Surface Positions . . 114 Typical Langmuir Probe Curves of Preliminary and Optimized Designs . 117 Electron Density Measurements in Preliminary and Optimized Designs . . 118 Electron Temperature Measurements in Preliminary and Optimized Designs118

8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9 8.10 8.11 8.12 8.13 8.14 8.15

Section View of TIPS with the Ultra-Fast Pictures Diagnostic Installed Photograph of TIPS with the Ultra-Fast Picture Diagnostic Installed . . Close view of TIPS with the Ultra-Fast Picture Diagnostic Installed . . Photograph of the Experimental Set-Up . . . . . . . . . . . . . . . . . . Typical Time Integrated Visible Spectrum . . . . . . . . . . . . . . . . . Different Types of Plasma Distribution Modes . . . . . . . . . . . . . . . Electric field Distribution Inside the Plasma Chamber . . . . . . . . . . Column Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hourglass Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Slug Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flower Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Full-Chamber Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ring Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yin-Yang Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Donut Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

124 125 126 126 127 129 130 131 133 134 135 136 137 138 139

9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9 9.10 9.11 9.12 9.13 9.14 9.15 9.16 9.17

Schematic Representation of Diagnostics Set-Up . . . . . . . . . . . . View of Synchronism Record During the Breakdown . . . . . . . . . . Scope Record Showing a Peak of Visible Light during the Breakdown . Column Mode Breakdown Evolution Study . . . . . . . . . . . . . . . Normalized Photodiode Signals with Column Mode . . . . . . . . . . . Hourglass Mode Breakdown Evolution Study . . . . . . . . . . . . . . Normalized Photodiode Signals with Hourglass Mode . . . . . . . . . . Slug Mode Breakdown Evolution Study . . . . . . . . . . . . . . . . . Normalized Photodiode Signals with Slug Mode . . . . . . . . . . . . . Flower Mode Breakdown Evolution Study . . . . . . . . . . . . . . . . Full-Chamber Mode Breakdown Evolution Study . . . . . . . . . . . . Ring Mode Breakdown Evolution Study . . . . . . . . . . . . . . . . . Yin-Yang Mode Breakdown Evolution Study . . . . . . . . . . . . . . Donut Mode Breakdown Evolution Study . . . . . . . . . . . . . . . . Typical Rotating Plasma Scope Record . . . . . . . . . . . . . . . . . Rotating Yin-Yang Plasma Mode Magnetic Field Profile . . . . . . . . Rotating Yin-Yang Plasma Mode . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . .

144 145 147 148 149 150 151 152 153 154 154 155 156 157 158 159 160

. . . . . . . . . . . . . . . . .

List of Figures 9.18 9.19 9.20 9.21

Rotating Half-Moon Plasma Mode Magnetic Field Profile . . . . Rotating Half-Moon Plasma Mode . . . . . . . . . . . . . . . . . + Temporal evolution for H+ , H+ 2 and H3 . . . . . . . . . . . . . . Typical Normalized Balmer-Alpha and Fulcher Band Photodioide

. . . . . . . . . . . . . . . Signals

161 161 162 163

List of Tables 3.1 3.2 3.3 3.4

Parameters Used during Experiments . . . . . . . . . . . . . Summary of Breakdown Time Measurements for Bz > ECR Summary of Breakdown Time Measurements for Bz ' ECR Summary of Breakdown Time Measurements for Bz < ECR

4.1

Summary of Plasma Parameter Evolution during Breakdown . . . . . . . 66

5.1

Parameters Used during Experiments . . . . . . . . . . . . . . . . . . . . . 75

7.1 7.2

Calculations of Stored Electric Energy Percentages . . . . . . . . . . . . . 112 Coupling Parameters in Vacuum βv and with Plasma βp . . . . . . . . . 116

xxi

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

29 43 43 43

Abbreviations CW

Continuous Wave mode

DAS

Data Acquisition System

DC

Duty Cycle

CCD

Charge Coupled Device

ECR

Electron Cyclotron Resonance

ECRIS

Electron Cyclotron Resonance Ion Source

EEDF

Electron Energy Distribution Function

MCP

Multi Channel Plate

MW

Microwave

OFHC

Oxigen Free High (Termal) Conductivity

TE

Transversal Electric (Resonant Mode)

TIPS

Test Bench (for) Ion-sources Plasma Studies

TM

Transversal Magnetic (Resonant Mode)

VUV

Vacuum Ultra Violet

xxiii

Physical Constants Boltzmann Constant

k

=

8.617 332 24 × 10−5 eV K−1

Electron Charge

e

=

1.602 176 57 × 10−19 C

me

=

9.109 382 91 × 10−31 Kg

mH +

=

1.672 621 77 × 10−27 Kg

c

=

299 792 458 m/s

Vacuum Electrical Permittivity

εo

=

8.854 187 81 × 10−12 F m−1

Vacuum Magnetic Permeability

µo

=

4π × 10−7 N A−2

Electron Mass Hydrogen Ion Mass Speed of Light

xxv

Symbols A

Generic Variable to Designate Area

mm2

Ap

Langmuir Probe Area

mm2

Asheath

Langmuir Probe Sheath Area

mm2

B

Generic Variable to Designate Magnetic Field

T

Bz

Magnetic Field along Plasma Chamber Axis

T

BECR

Electron Cyclotron Resonance Magnetic Field Value

T

De

Ambipolar Diffusion Coefficient for ECR Plasmas

m2 s−1

E

Generic Variable to Designate Electric Field

V/m

Eef f

Effective Electric Field

V/m

Emax

Incident Electric Field

V/m

F

Generic Variable to Designate Force

N

I

Generic Variable to Designate Electric Current

A

Ie

Electron Current

A

Ies

Electron Saturation Current

A

Ii

Ion Current

A

Iis

Ion Saturation Current

A

P

Generic Variable to Designate Power

W

Pa

Absorbed Power

W

Pi

Incident Power

W

Ploss

Power Losses

W

Pmax

Maximum Incoming MW Power Used in our Experiments

W

Pr

Reflected Power

W

Q

Microwave Coupling Factor

a.u xxvii

Symbols R

Plasma Chamber Radius

m

T

Generic Variable to Designate Temperature

eV

Te

Electron Temperature

eV

Ti

Ion Temperature

eV

U

Generic Variable to Designate Electric energy

J

Us

Stored Electric energy

J

V

Generic Variable to Electric Potential

V

Vf

Plasma Floating Potential

V

Vp

Plasma Potential

V

W

Energy Gain of an Individual Electron

eV

f

Generic Variable to Designate Frequency

Hz

n

Generic Variable to Designate Particle Density

m−3

nc

Plasma Electron Critical Density

m−3

ne

Plasma Electron Density

m−3

ne0

Initial Plasma Electron Density

m−3

ne,critical

Minimum Electron Density Necessary to Produce Breakdown

m−3

ne,ss

Plasma Electron Density during Steady State

m−3

ne,t

Plasma Electron Density during Transients

m−3

ne,res

Residual Electron Density after MW Shut-off

m−3

nn

Neutral Gas Density

m−3

p

Generic Variable to Designate Gas Pressure

mb

pmax

Maximum Gas Pressure Used in our Experiments

mb

r

Generic Variable to Designate Radius

m

t

Generic variable to Designate Time

s

tbreak

Breakdown Time

s

v

Generic Variable to Designate Velocity

m s−1

ve

Electron Velocity

m s−1

vd

Drift Velocity

m s−1

vr

Relative Velocity (Electron and Neutral Particles)

m s−1

z

Position along Plasma Chamber axis

m

Symbols β

Generic Variable to Designate Coupling Parameters

a.u

βp

Coupling Parameter with Plasma

a.u

βv

Coupling Parameter in Vacuum

a.u

Generic Variable to Designate Electron Energy

eV

ss

Electron Energy during Steady State

eV

t

Electron Energy during Transients

eV

εr

Relative Electrical Permittivity

eV

µr

Relative Magnetic Permeability

eV

ω

Generic Variable to Designate Angular Frequency

rad−1

σ

Generic Variable to Designate Cross Section

m2

σion

Ionization Cross Section

m2

σrec

Recombination Cross Section

m2

τion

Characteristic Ionization Time

s

τionmin

Minimum Order of Magnitude of Characteristic Ionization Time

s

τof f

Off-Time between MW pulses

s

ξ

Generic Variable to Designate Light Emission Intensity

W/m2

ξss

Light Emission Intensity during Steady State

W/m2

ξt

Light Emission Intensity during Transients

W/m2

Chapter 1

Introduction Ion source technology is an important issue nowadays in both scientific and industrial fields. Over the last few decades, increasing ion source performance and reliability has become one of the main goals for ion source designers. Most ion sources are plasma based, i.e., they contain a plasma from which ions are extracted. The parameters of this plasma are a determining factor for the characteristics of the ion beam extracted from the source and they depend strongly on the design and the technical details of the ion source. For this reason, a deeper understanding of plasma characteristics, its evolution and parameters can be decisive in the design, development and operation of ion sources. The strong relationship between the characteristics of the plasma contained inside an ion source and the performance of the ion source itself have been highlighted by many well know scientists. Thus, Ian Brown starts the plasma physics chapter of his book “The Physics and Technology of Ion Sources” [1] by pointing to the influence of plasma characteristics inside the source on beam current, beam emittance and beam composition. Moreover, he states that the physics of an ion source is largely plasma physics. On the other hand, Richard Geller, known as one of the fathers of Electron Cyclotron Resonance Ion Sources (ECRIS), points out in his book “Electron Cyclotron Resonance Ion Sources and ECR Plasmas” [2] that while ion extraction and transport are aspects that seem to be well understood by engineers, ion generation is a field that is far less understood or developed. He claims that while most studies focus on engineering issues; atomic physics for ion production; surface phenomena on the electrodes and walls or the properties of electrical discharges, the relations between ion source performance and plasma performance are rarely underlined, even though ions are extracted from plasmas. After explaining how the main characteristics of the beam, such as composition, ion charge, emittance or ionization percentage are deeply related to plasma characteristics, 1

2

Geller comes to the conclusion that “specific, high-performance ion beams are extracted from specific, high-performance plasmas”. The object of this thesis is to develop plasma diagnostics that can serve as tools for ion source design and optimization. As will be seen throughout this study, the point of view is always that of the engineer. The work has been focused on the development of plasma diagnostics which permit the exploration of ion source plasma characteristics with the aim of using them as tools for ion source development and optimization. The research presented here is part of the work developed in the Ion Source Engineering Laboratory at the University of Castilla-La Mancha in Ciudad Real (Spain) in collaboration with ESS BILBAO. The team, led by Dr. Daniel Cortázar, started working together in September 2010 with the goal of commissioning an Electron Cyclotron Resonance plasma source that could serve as a test bench for ECR ion sources. All the plasma diagnostics that will be presented in this thesis have been implemented in the plasma source that we called TIPS (Test Bench for Ion source Plasma Studies). It is important to note that the terms “plasma source” and “ion source” are, in some cases, used interchangeably. However, in a plasma source the ions embedded in the plasma usually possess little directed energy. The ion drift energy is zero or, at least, small compared with the mean thermal ion energy, while the ions in an ion source have a large drift energy compared to the mean thermal energy. In most cases, the ion source contains a plasma source as its most essential component. Plasma is formed in the heart of the plasma source and the hardware and electronics needed to produce it are key parts of the overall ion source. The beam is formed from the ions contained in the plasma by means of an electrode system consisting of specially shaped biased metal electrodes. This system is commonly known as an extractor, yet ions are not “pulled out” from the plasma at all; they rather flow from the plasma to the electrode system at a rate quite independent from the extractor voltage and are then accelerated by the extractor to form an energetic beam [1]. Thus an ion source can be considered to be formed mainly by a plasma source and an extractor.

Ion beam

Plasma source

Extractor Figure 1.1: Simplified diagram of the main elements of an ion source: plasma source, extractor and ion beam.

Chapter 1. Introduction

3

Fig. 1.1 shows a simple diagram of an ion source formed by a plasma source and a extraction system placed in a such way that some of the ions contained in the plasma flow toward it. These ions are then accelerated by the extractor to form the ion beam. There are many types of ion sources and usually, they are classified depending on the type of energy used by the plasma source (laser, radiofrequency, electric discharges and so on). Our device, TIPS, is a 2.45 GHz microwave discharge plasma source designed for positive ion production. Many 2.45 GHz plasma sources like TIPS are used as the main part of ion sources around the world, both for industrial and scientific applications. As an example, some of the scientific facilities that use 2.45 GHz ion sources are listed below: • PKUNIFTY, Peking University Neutron Imagining Facility (China). • LEDA, Low-Energy Demonstration Accelerator (USA) • CPHS, Compact Pulsed Hadron Source (China). • TRASCO, High-Current Proton Source (Italy) • FAIR, Facility for Antiproton and Ion Research (Germany). • RCNP, Research Centre for Nuclear Physics (Japan). • VECC, Variable Energy Cyclotron Center (India). • ESS, European Spallation Source (under development in Sweden). In this chapter, ECR ion sources will be presented starting with an overview of their operating principles and a brief summary of its main architectural features. The most important operational regimes; preglow, steady state and afterglow; are detailed in the final section. A study of the transient regimes (preglow and afterglow) requires time-resolved diagnostics, and providing our own diagnostics for time resolution was one of the key points in the design of the plasma diagnostics described in this thesis. Chapter 2 describes the device called TIPS in which all the diagnostics presented here have been implemented. The main parts of the device are described including the magnetic field generation system in more detail. 2D simulations of the magnetic field generated are presented together with the corresponding experimental measurements for validation. Chapter 3 is the first chapter completely dedicated to plasma diagnostics development. The diagnostics described in this chapter are designed to determine plasma breakdown time. In pulsed operation ion sources, the synchronization between pulses

4

and extraction is an important issue, and for this reason determining plasma evolution times could result in a valuable tool for ion source users. Knowing breakdown times and their evolution within the range of working parameters is of special importance for ECR ion sources operating during afterglow and steady state regimes. Time-resolved measurement of the current circulating through a biased probe was combined with signals of incoming and reflected power and visible light emission to determine breakdown times. Measurements were taken under a wide range of working parameters (power, duty cycle, magnetic field profile and pressure). A simple model of the breakdown process was also developed and compared with experimental results. Calculated and experimentally measured breakdown times are demonstrated to be in well agreement. Chapter 4 is focused on the development of a diagnostic to determine the evolution of plasma parameters, electron density and temperature during breakdown and determine how such evolution was related to the breakdown times measured in chapter 3. A Langmuir probe was introduced into the center of the plasma chamber to acquire I-V curves. Taking advantage of the high pulse-to-pulse plasma reproducibility, the system was designed to take each point of the I-V curve at a different pulse. Using this strategy, it was possible to take a set of curves during breakdown evolution with 1 μs time resolution. The experimental set-up and the curve analysis methods are described in the first part of the chapter. Results obtained under a wide range of working parameters are shown and compared in the second part. Unexpected transient temperature peaks that reach 18 eV during 20 μs were observed at the beginning of plasma breakdown. Decays of such peaks reach final stable steady state temperatures of around 5 eV at the flat top of the microwave excitation pulse. These peaks suggest a possible connection with the preglow process observed in ECR ion sources. Under some of the explored working conditions an electron density peak was also observed during breakdown. In chapter 5 the set-up developed in previous chapter is used to study the evolution of plasma parameters during decay. The same working condition range explored in chapter 4 was chosen on this occasion. This study shows the capability of the diagnostic during decay transient process. An interesting structure was found on the reflected MW power signal. It reaches a peak just after the incoming power shut-off and a second smaller rebound approximately 10 μs later. The electron temperature also shows a rebound coinciding with the second reflected power peak. Chapter 6 presents a new diagnostic designed to measure the temporal evolution of microwave-plasma coupling, vacuum ultraviolet light emission and plasma electron temperature in TIPS at the same time. The aim of this diagnostic is to confirm the


5

existence of the electron temperature peak during breakdown reported in chapter 4. A 5-10 μs transient peak of light emission exceeding the steady state by a factor of 3.3, was observed to coincide with an abrupt drop of the MW electric field. Observed light emission intensities combined with cross section data indicate that electron temperature during breakdown should exceed the steady state value of 4-6 eV by a factor ≥ 3 which is in line with the Langmuir probe data shown in chapter 4. Chapter 7 is dedicated to the study of the microwave coupling driver system. The influence on plasma characteristics of stationary electric field distribution throughout the microwave excitation system just before breakdown is analyzed in this chapter. 3D simulations of resonant stationary electric field distributions, 2D simulations of external magnetic field mapping, experimental measurements of incoming and reflected power and of electron temperature and density are all used to compare the performance of both designs. By using these tools, an optimized set formed of plasma chamber and microwave coupler has been designed and built paying special attention to the optimization of the stationary electric field value in the center of the plasma chamber just before breakdown. This optimized system shows a strong stability level for plasma behavior, allowing a wide range of working parameters and even sustaining plasma formation without any external magnetic field. In addition, it is capable of producing electron density values four times higher than the preliminary system. In chapter 8 a new diagnostic is presented. The new optimized set of plasma chamber and MW coupler is used in TIPS from this point on. The plasma reactor was modified to include a transparent double-shielded quartz window allowing the full plasma volume to be viewed. An Image Intensified CCD frame camera, based on a combination of multichannel plate (MCP) light intensifiers and CCD cameras, was used to take plasma pictures. Several different plasma stable modes were discovered, depending on the working parameters. The distribution of these modes also depends on the working parameters: pressure, MW power and magnetic field distribution, with the last of these being the most critical. Visible time-integrated spectra were also taken for each of the plasma modes. The possibility to use these signals as a tool to find information about the plasma ion species fractions was explored. Throughout chapter 9, the breakdown process is studied using the diagnostic previously described in chapter 8. Pictures showing the breakdown process of the plasma distributions described in the previous chapter were obtained for integrated full visible light, Balmer-alpha and Balmer-beta lines and Fulcher band emissions. Photodiode signals of the temporal evolution of such emissions for the high-brightness plasma modes were also recorded. The similarities between these signals and the cluster ion H+ 2 and

6

H+ 3 currents measured by Y. Xu et al. in an ECR plasma source at Pekin University is discussed [3]. Finally chapter 10, is a conclusions chapter. It gives an overview of the work described in this thesis and shows the conclusions that can be extracted from the work. It contains also a contributions section where the publications and conferences generated by this work are listed. The last section is dedicated to the group future plans.

1.1

ECR Ion Sources

Essentially, microwave ion sources consist of a chamber filled with gas or vapor with a superimposed magnetic field and a means of introducing microwaves. In the absence of a magnetic field, microwaves can not propagate in a plasma with an electron density higher than a critical value nc given by [4]:

nc =

εo mω 2 e2

(1.1)

where ω is the microwave frequency, e is the electron charge and me is the electron mass. The critical density for a plasma generated in a 2.45 GHz plasma source like TIPS is 7.45 x 1016 m−3 . The introduction of a magnetic field offers the possibility of acquiring high plasma densities by Electron Cyclotron Resonance Heating. Electrons immersed in a magnetic field travel in circles in a plane perpendicular to the field due to Lorentz force. The angular frequency of this movement is known as the ECR frequency and can be obtained from Eq. 1.2 where B is the magnetic field.

ωECR =

eB me

(1.2)

In ECR ion sources, a magnetic field is generated to match the value, making the ECR frequency equal to the microwave frequency. For a 2.45 GHz microwave frequency plasma reactor like TIPS, the value of the ECR magnetic field (BECR ) is 87.5 mT. Microwaves injected into the plasma chamber should show circular polarization in the direction of the electrons’ rotation, so that the wave can be transferring energy to them continuously. This is a major advantage of ECR ion sources compared to microwave ion sources operated using linearly polarized waves, where the energy transfer depends on collisions since the energy gained by an electron in one semi-period is lost in the next


7

unless it collides against another particle. This energy transfer dependence on collisions makes it necessary to work with high pressure, whereas ECR sources can operate with low pressure, where the low collision ratio allows the electrons to be accelerated by the microwave resulting in an increment in its temperature [2]. On high-frequency ion sources the magnetic field profile is designed in such a way that it provides electron confinement. The ions and electrons are tied in their orbital motion to the field lines and that provides a means of confining the plasma in the direction transverse to the magnetic field lines. In some cases a configuration called magnetic mirror is used, it consist on the increment on the strength of the magnetic at the ends of the confinement region. In this case the plasma is confined, albeit imperfectly, both longitudinally and transversely. Two groups of 2.45 GHz operating frequency ion sources can be found, depending on the shape and value of the magnetic field with respect to the BECR [5]. In some cases, the sources operate at a magnetic field below the ECR: the incoming wave is no longer a pure electromagnetic mode, but interactions with dense plasma trigger non linear phenomena and electromagnetic oscillations. The incident electromagnetic wave absorption can be described by means of a collective approach: the electron gyromotion is disturbed by fluctuating fields and by the plasma “effervescence” with high frequencies [2]. Under this circumstances, Bernstein waves (BWs) can be excited. These waves penetrate the warm plasma core without any cut-off and they can be absorbed at cyclotron harmonic fields BECR /2 or BECR /3. This absorption is useful, but plasma created in such a way is turbulent and non-uniform. The other group is formed by sources operating with magnetic field profiles above ECR. If a first ignition due to single particle ECR heating occurs, resonances between electromagnetic waves and the plasma electrons can occur even for high plasma densities [6, 7]. ECR regions inside the chamber can be understood as triggering areas where the plasma is ignited and then sustained in the whole chamber.

1.1.1

ECR Ion Source Architecture

High current sources working at 2.45 GHz were first used about 30 years ago mostly for industrial applications. They can produce high levels of brightness and high current proton beams and present many advantages in terms of compactness, reliability, reproducibility, low transverse emittance and low maintenance. This makes them widely used for both research and industrial applications [8]. In many industrial applications, and for the low current accelerators, the ion source can be considered as a “black box” whose behavior has almost no consequences on the beam acceleration. However, for high

8

current accelerators, where requirements in terms of reliability and low emittance are more complex, the ion source performance plays an important role.

Magnetic field generator MW generator

Ion beam Plasma chamber

Extractor Plasma source

Figure 1.2: Simplified diagram of an ion source whose plasma source is of type ECR.

Despite the fact that there are many designs for this kind of ion source there are some elements common to all of them. The main parts of an ECR ion source are shown in Fig. 1.2 and briefly described below.

• Plasma chamber: It is the reservoir where plasma is generated and sustained. Gas or vapor from the element chosen to be ionized is introduced into the chamber. Microwaves then ionize it to produce plasma and ions are extracted from it through an extraction hole producing the ion beam. Most of the ions generated inside the plasma diffuse in all directions, collide with the chamber wall and are neutralized. Only ions which happen to diffuse towards the extraction hole can be extracted and form the beam. In TIPS, as will be detailed later, the plasma chamber has cylindrical shape and the gas used to generate plasma was Hydrogen. • Microwave driver system: Microwaves need to be transferred from the microwave generator to the plasma chamber. This can be done by means of an antenna introduced into the plasma chamber [9] or by directing the microwaves to the plasma chamber through a waveguide and a matching transformer to adapt the impedance of the plasma and the waveguide [10]. This second option is generally preferred because it requires less periodic maintenance and makes the equipment more reliable for long-term operations. In most cases, microwaves are injected along the plasma chamber axis. Microwave to plasma coupling design is still an issue under discussion in the ECR ion source community. The microwave driver coupling design for the particular case of TIPS is widely discussed in chapter 7.


9

• Magnetic field generation system: A magnetic field is required for electron cyclotron resonance that corresponds with the microwave frequency. It is also used to prevent the generated plasma, especially hot electrons, from diffusing to the wall of the plasma chamber. This magnetic field can be generated by means of permanent magnets or solenoids surrounding the chamber. In some of the sources using solenoids for magnetic field generation, they can be moved axially to allow changes in the magnetic field profile distribution to try to find the one that results in the best ion source performance. This is the case of TIPS, where we have four solenoids distributed in two movable structures known as pancakes. • Extraction system: The extraction system of an ion source uses voltage differences to accelerate and focus the beam. In general, a positive ion source is polarized to a positive potential while the first electrode is grounded. The power supplies, microwave generator and control subsystems of the ion source can also be polarized at high voltage or, in some cases, a DC break is used to keep only the plasma chamber at high voltage. Different solutions have been found for isolating the microwave generation system device from the high voltage as is shown, for example, in references [10, 11]. In TIPS, no extraction system has been used; instead of extraction electrodes we have designed different diagnostic ports with the aim of giving support to the different diagnostics developed and described in this thesis.

1.1.2

Operating Regimes

ECR ion sources can be operated in either continuous wave (CW) or pulsed mode. The choice of operating mode depends on the application of the source and the requirements of the beam. In the case of TIPS, it can be operated in CW mode or in pulsed mode from 50 Hz to 20 KHz. For sources operating in pulsed mode, the extraction and the microwave pulses can be synchronized. The instant when the extraction takes place strongly determines the source operation regime. Extraction can be done during the flat top of the MW pulse or during its transients, i.e. the rise time and the fall time, taking advantage of two interesting phenomena called preglow and afterglow, respectively. In both cases an unexpected peak of extracted current appears, associated with breakdown and decay transients. The first of these effects to be reported was the afterglow, observed for the first time by the team at DRFMC in Grenoble in 1988 on the MINIMAFIOS 16.6 GHz and described by Melin et al. in 1989 [12]. The same temporal structure has now been

10

observed in many ECR ion sources like MINIMAFIOS 14.5 GHz at CERN, ECR4 14.5 GHz at GANIL, and CAPRICE 14.5 GHz at GSI. When the ion source is tuned to the CW mode and the microwave power is shut off, a sharp peak of extracted ion current appears just after the power switches off; such a peak is called the afterglow current pulse. This peak can be optimized by tuning the ECR ion source magnetic field, the pressure and the MW power for the specific purpose of acquiring the highest and most stable afterglow current. The afterglow peak is more noticeable in ion sources producing very high charge state ions, while in sources for low charge state ion production, this effect is weaker or even non existing. It has also been reported that the afterglow current is generally at its maximum when the source settings are those producing the minimum current in CW mode [13]. As an example, Fig. 1.3 shows the time structure of the ion currents of an ECR ion source: the CW operating mode (a); the afterglow peak after the switch-off of the MW power when the source is tuned to optimize continuous current (b) and the optimized afterglow current (c).

a.u

MW power Extracted current

t

(a)

t

(b)

t

(c)

Figure 1.3: Afterglow phenomena after the microwave shut-down in an ECRS: (a) CW operating mode, (b) afterglow peak when the source is tuned to optimize continuous current and (c) optimized afterglow current [13].

Fig. 1.4 shows the incoming MW power and the Ar13+ extracted current signals measured by Melin et al. on the 16.6 GHz MINIMAFIOS [12]. What is notable is the complex structure that appears in the current signal just after the shut-off of the microwave power. The afterglow phenomena was first used for acceleration purposes in 1992 on the MINIMAFIOS source at CERN and, based on this and the positive results also obtained at GANIL, the new heavy ion injector at CERN was equipped with a source optimized for afterglow operation: the ECR4.


11

In general, the current obtained during the afterglow can reach 2-3 times that which is extracted during the flat top of the pulse. However, K. Langbein reported that if one looks only at a single high ionization state of a heavy element, the relative increase in the ion current can be much higher, reaching, for example, a factor of 100 for P b27+ according to Langbein’s experiments in ECR4 ion source at CERN [14].

Figure 1.4: Ar13+ Afterglow at MINIMAFIOS: MW input power and extracted current signals [12].

Details of the afterglow mechanism are still a topic of discussion in the scientific community. Nowadays, afterglow ion pulses are used with great success, but the explanation of such a process is not unequivocal. Certainly different conditions generate different phenomena that may be superimposed. Non linear effects, as has been experimentally observed, appear to have great influence on both the process itself and the highly charged ion current in the afterglow pulse. In 1991, P. Sortais proposed that in any device containing a hot plasma, the central plasma should be isolated from the wall by an electron cloud due its higher mobility compared to ions. Such a distribution of electrons and ions would produce on the central plasma region a positive potential bias. In an ECR ion source the production of ions in the center of the plasma chamber by means of ECR heating would produce a potential depression in the axis of the plasma chamber as shown in Fig. 1.5. Thus, the sudden shut-off of the MW power would result in an increase to the ion density in the chamber axis area. This model, although it was conceived for higher frequency sources could also be valid in some cases for 2.45 GHz sources like TIPS. However, the discovery of nonhomogeneous plasma modes inside the chamber that is shown in chapter 8 makes this

12

model valid in only a few of the plasma configurations. According to our observations in TIPS, the axis of the plasma chamber is not always the area where maximum electron density is found. In fact, it is not even the region where plasma breakdown necessarily takes place. As an example, Fig. 1.6 shows one of the plasma configurations found in TIPS where it is clear that the plasma is not concentrated in the center of the chamber. The ultra-fast pictures diagnostics presented in chapter 8 opens up an interesting new set of questions related to plasma distribution inside the chamber and its influence on the extracted beam. Wall

central hot plasma

Sheath

Sheath

Figure 1.5: Plasma electrostatic distribution according to P. Sortais [13].

Figure 1.6: Ultra-fast picture of plasma inside the chamber.


13

The relevance that afterglow phenomena has acquired over the last few decades among the accelerator community together with the interest in this process shown by the volume of research done into it, provided the motivation for the plasma decay study described later in chapter 5. This study was undertaken with the aim of trying to find some correlation between the current peak found in many ion sources just after the MW power shut-off and the plasma parameter evolution during decay in our plasma source. The second transient effect commonly used as an ion source operating regime is the so-called preglow, which consists of an ion current peak registered during the plasma breakdown transient. The preglow phenomenon was first reported by P. Sortais et al. in 2004 [15] while the team was looking for ECR ion source operating conditions to produce short operation pulses. Ion sources producing short pulses of multicharged ions are in great demand for researchers in nuclear physics and the physics of elementary particles to be carried out on new generation accelerators. The effect was first observed in both PHOENIX 28 GHz and SMIS 37.5 GHz ion sources. As an example, Fig. 1.7 shows the extracted current from SMIS 37.5 GHz where the preglow current peak at the beginning of the pulse can be observed.

Figure 1.7: 2004 [15].

Total extracted current from SMIS 37.5 GHz reported by Sortais et al. in

Many questions related to the causes of preglow, its characteristics and possible uses are still open to debate in the scientific community. This fact motivated the development of some of the plasma diagnostic tools described in this thesis, and their design was focused on obtaining information about plasma during breakdown. In 2006 V. A. Skalyga et al. proposed a theoretical model of gas breakdown in an ECR ion source. According to such model, plasma breakdown can be understood to

14

occur in two stages. In the first step, the breakdown process is dominated by ionization of the neutral gas produced by collisions with hot electrons: plasma density grows exponentially; the degree of gas ionization is less than unity; small charge ions dominate in the distribution of ions over their charge states; and the power absorbed by the plasma is low. In the second stage, the rate of density growth is slower; the process of ion peeling goes further; the ion charge increases and the absorbed power is greater [16]. After breakdown, the plasma reaches its steady state that can be assumed to be a quasi-gas-dynamic regime. Research carried out on TIPS and described in chapters 3 and 4 gives experimental evidence of the two stages of the plasma breakdown process. We have named these two stages “MW coupling” and “plasma formation time” as detailed in section 3.2. However, while V. A. Skalyga’s theoretical model predicts that plasma breakdown time should rise when power is increased, experimental data measured on TIPS show the opposite behavior. This is probably due to fact that quasi-gas-dynamic conditions can not be assumed in TIPS. A small breakdown time model based on the influence of seed electrons is presented in section 3.4. As stated by V. G. Zorin et al. in Ref. [17] the transition from breakdown to steady state can be attended with an unexpected transient peak of multicharged ions, i. e., the + preglow. Very recently, Y. Xu et al. reported a peak of cluster ions H+ 2 and H3 during breakdown in their 2.45 GHz ECR ion source [3]. The authors suggest that this peak could be strongly related to the existence of the temperature peak found in our device and reported in October 2012. Although preglow and afterglow transients are more noticeable on high-frequency sources, the study of breakdown and decay in TIPS can provide information about plasma dynamic processes that can be useful also for high-frequency sources.

Chapter 2

TIPS: Test Bench for Ion Source Plasma Studies TIPS, as stated in chapter 1, is an Electron Cyclotron Resonance 2.45 GHz Hydrogen plasma source built to be a test bench for ion source engineering. The key aim of the research carried out with TIPS as described in this thesis is to develop diagnostic tools which allow us to acquire a deeper knowledge of the plasma characteristics and plasma physics processes involved in ECRIS performance. In order to access the plasma inside the ion source with different diagnostics, a diagnostic port was placed in the position where the extraction electrodes would be positioned to use it as an ion source [18]. In this chapter, TIPS will be presented and its main subsystems described.

2.1

TIPS Description

In TIPS, pulsed microwaves are generated in a 2.45 GHz, 3 kW adjustable power magnetron and travel through a rectangular waveguide WR340 to the plasma chamber. The chamber is surrounded by a magnetic field generation structure that is described in detail later. Fig. 2.1 shows a photograph of TIPS, the magnetron, the microwave waveguide and the magnetic field generation structure and the pumping system can be seen in the picture.

15

16

Figure 2.1: Photograph of TIPS.

Fig. 2.2 shows a cross section view of the device, including its main subsystems. The plasma chamber (a) is made of OFHC copper and it is 90 mm in diameter and 97 mm long. The chamber wall has four longitudinal channels for water cooling (not visible in the figure). Boron Nitride discs with a thickness of 2 mm (not present in the figure for clarity) are placed at both ends of discharge chamber. Attached to the chamber is a ridged five-step brass microwave coupler (b), designed to adapt impedance between the rectangular waveguide WR284 and the plasma chamber. It also serves for gas injection (c), for chamber pressure measurements (d) and for water cooling input (e) and output (water output channel not visible in the image). Two removable parts made of OFHC copper (f ) form the ridged steps. A tapered waveguide is used to adapt the input of microwave coupler WR284 to the WR340 section of the microwave generator system (g). It is connected to a 30 mm rectangular holder sustaining a 10 mm thick window (h) which separates the volume under vacuum from the atmospheric one. On the atmospheric side, a dual-directional coupler is fitted (i) and a time synchronization signal is obtained from it. A two-stubs tuner (j) is used for fine impedance tuning and, next to it, another dual-directional coupler (k) is used to record incoming and reflected power to and from the plasma respectively. Both dual directional couplers have a 60 dB coupling factor with an approximated coupling loss of 4 x 10−6 dB and a directivity of 25 dB. A closed loop chiller provides cold water to all subsystems.

Chapter 2. TIPS: Test Bench for Ion Source Plasma Studies

17 (j) (k)

(a)

(i)

(g)

(o) (c) (e)

(l)

(b)

(h)

(f ) (d) (m)

Figure 2.2: Section view of TIPS and main subsystems: plasma chamber (a), brass coupler (b), gas inlet (c), pressure gauge flange (d), cooling water inlet and outlet (e), ridged steps (f ), tapered waveguide WR284/WR300 transition (g), vacuum break window (h), dual directional couplers (i) and (k), two-stubs tuner (j), diagnostics port (l), 7 mm pumping hole (m) and magnetic field generation system (o).

On the diagnostics side of the chamber, a diagnostic port (l) fulfills two tasks: it serves as a pumping port through a 7 mm hole in the center (m) and also holds the diagnostic systems. In the course of this thesis, different ports will be described according to the necessities of the diagnostics that have been implemented. Surrounding the plasma chamber, there is a magnetic field generation system composed of four coils (o) arranged in two axially movable pancakes. The magnetic field profile can be adjusted by regulating of the current circulating through each coil and by changing the axial position of the pancakes. The direction of the magnetic field is always toward the MW wave injection side of the chamber. Magnetic field distribution is one of the key points in ECR ion source performance. To characterize these profiles in TIPS, the magnetic field distribution in the plasma chamber volume has been measured by means of a Hall probe capable of measuring in all the three axes with a typical error of ± 1 mT. Fig. 2.3 shows a section view of the set-up used for these measurements. A diagnostics port (a) was specially designed for the purpose. The magnetic probe (b) is fixed inside a plastic tube (c) and this tube is inserted inside in a rotating piece (d). By rotating this piece and displacing the plastic tube axially, the tip of the probe can be

18

moved to map all of the volume where the plasma chamber is usually mounted with a spatial resolution of 2 mm.

(a) (b)

(c) (d)

Figure 2.3: Section of TIPS magnetic field generation system with the diagnostic port (a) for B-field mapping installed: Hall probe (b), plastic tube (c) and rotating holder (d).

Figure 2.4: Photograph of TIPS with the magnetic field measurement set-up installed

Fig. 2.4 shows a photograph of TIPS with the magnetic field measurement set-up installed. Using this set-up to directly measure the magnetic field distribution requires the plasma chamber and MW coupler to be dismounted and the diagnostics port to be changed. Therefore, to provide a quick tool to calculate the magnetic field distribution

Chapter 2. TIPS: Test Bench for Ion Source Plasma Studies

19

MW INJECTION SIDE

r z

DIAGNOSTICS PORT SIDE

obtained by any combination of coil currents and pancake positions, 2D simulations were carried out using FEMM [19].

PLASMA CHAMBER 1 cm

mT 150 146 142 138 134 130 126 122 118 114 110 106 102 98 94 90 86 82 78 74 70

Figure 2.5: Example of 2D magnetic field simulations.

As an example, Fig. 2.5 shows a 2D map of the simulated magnetic field for a particular combination of symmetric coil currents and positions. MW injection takes place on the left side and the diagnostic port is on the right side. The chamber limits are marked with a solid black line and a dotted horizontal line marks the chamber axis. Superimposed to the color map the r and z axes are shown. Simulations were validated with experimental data and the error remains below 2 %. Fig. 2.6 uses red dots to show the experimental B-field measurements along the axis of the plasma chamber (designated Bz throughout this thesis) for the same configurations of coil currents and positions used in the previous figure simulation and a solid red line to show the simulated values of Bz .

20

) ) )

)

Figure 2.6: B-field measurements along the axis (Bz ) of the plasma chamber and the simulated values obtained for the same conditions.

Chapter 3

Study of Plasma Breakdown Times Plasma evolution times are very important parameters in pulsed ion source operation. All over the world, several different ion sources operate in pulse mode to take advantage of the preglow and afterglow transients. In these kinds of sources, the synchronization between the plasma source pulses and the extraction is critical. Knowing the duration of each plasma evolution period, i. e. breakdown, steady state and decay, helps ion source users to achieve better performances. The times associated with the plasma ignition process are of particular interest for ion source designers and users. The diagnostic described in this chapter permits the measurement of the plasma breakdown times in TIPS under a wide range of operating conditions to obtain information about MW coupling and plasma formation stages during the process. The work is useful for designers who need to extract short beam pulses from a 2.45 Hydrogen ECR plasma source for any application because the total breakdown time measured is defined as corresponding to what is required to reach the steady-state plasma parameters. A simple model considering the influence of seed electrons between pulses is proposed in section 3.4 as a first approach to estimating breakdown times. The study has been undertaken using four time-resolved simultaneous diagnostics: electrical biased probe saturation current, visible emitted light, incoming power and reflected power measurements.

21

22

3.1

Experimental Set-Up

3.1.1

Diagnostic Port

Measurements were taken by means of a specifically designed diagnostic port. It allows the biased probe to be placed in the center of the plasma chamber while pumping through the center of the port. This is a very critical point because any change in the vacuum pumping system could result in changes to the plasma breakdown dynamics. As TIPS was designed to be a close reproduction of an ECR ion source, it has been a constant in the design of all the diagnostics to try not to make any changes that could make the plasma reactor operation less comparable to that of an ion source.

(a) (b)

(c) (d)

Figure 3.1: Section view of TIPS where the breakdown time measuring set-up is mounted: diagnostic port (a), probe holder (b), biased probe (c) and observation window (d).

Fig. 3.1 shows a section view of TIPS with the diagnostic port (a) installed. A holder (b) sustains the probe (c) in the plasma chamber axis. An 11 mm-diameter quartz window (d) was also placed at the diagnostic port to allow the observation of the plasma inside the chamber and the connection of a set made up of a collimator, a fiber optics and a photodiode used to register the visible light emitted by the plasma. The 2 mm thickness Boron Nitride disks at both sides of the plasma chamber were kept in the design although they are not shown in the figure for clarity. The disk in the diagnostic port side was machined to fit the new port geometry. It is worth noting that emitted light has become a useful tool to check plasma stability and, above all, the reproducibility of plasma behavior between pulses. Moreover, in a repetitive phenomena like this, where data can be obtained from different pulses (as will be detailed later), keeping jitter very low (under 200 ns in TIPS’ case) is of vital

Chapter 3. Study of Plasma Breakdown Times

23

importance and thus having a measurement of plasma emitted light has turned out in a very valuable tool.

3.1.2

Magnetic Field Profiles

During the reactor commissioning, start up and tuning stages, different magnetic field distributions were measured and tested. Some of them caused a strong plasma tendency to be allocated at the rear part of plasma chamber (the MW injection side and MW coupler piece). Taking into account that plasma quality close to the extraction zone is an important factor in achieving good ECRIS performances, we consider this tendency as a undesirable behavior. Even in the best cases, it clearly produces low-density plasma at the extraction zone. Careful attention was paid to establish a set of parameters where the plasma shows acceptable behaviors. This chapter displays the results of the breakdown study for three of the different magnetic field profiles typically used in TIPS taking as its symmetry reference the center of plasma chamber. They have been named according to their values on the axis as Asymmetric Bz > ECR, Symmetric Bz ' ECR and Symmetric Bz < ECR. Fig. 3.2 shows three Bz magnetic field profiles measured experimentally along the chamber axis with the set-up described in the previous chapter. Plasma chamber limits are indicated by dotted vertical lines, where the left border shows the microwave injection side and the right border, the diagnostics side. The ECR magnetic field level of 87.5 mT is marked by a broken, flat black line. The magnetic field was simulated for each configuration and the corresponding 2D map is represented in Fig. 3.3. Following the pattern of chapter 2, the plasma chamber limits are marked with a solid black line and the left side corresponds with the MW injection side while the right side corresponds with the diagnostic port. Note that the case (a) corresponds with the asymmetric profile where Bz takes higher values that reach 120 mT; case (b) corresponds with a symmetric Bz field with a value coincident to ECR; and case (c) is an symmetric flat Bz magnetic field profile with the values inside the chamber always below the ECR. The surface where the value of the magnetic field is exactly that of the ECR B-field is called the ECR surface. The position of this surface inside the plasma chamber has a strong influence on ECR plasma dynamics [20]. This kind of surface position is marked in Fig. 3.3 with broken black line.

24

Figure 3.2: Axial magnetic field profiles used during experiments. (a) Asymmetric Bz > ECR magnetic profile (b) Symmetric Bz ' ECR magnetic profile and (c) Symmetric Bz < ECR magnetic profile.

These three magnetic field distributions are the B-field experimental conditions that were used during measurements as typical operation modes in order to check the influence of magnetic field on plasma breakdown dynamics. Note that none of the field topologies are magnetic mirrors because this plasma generator was originally designed for the purpose of studying Hydrogen plasma for proton generation, without trapping to enhance confinement time in order to reach high degrees of ionization.


25

(a)

(b)

r z

(c)

mT 150 146 142 138 134 130 126 122 118 114 110 106 102 98 94 90 86 82 78 74 70

Figure 3.3: 2D representation of simulated magnetic profiles used during experiments: (a) Asymmetric Bz > ECR magnetic profile, (b) Symmetric Bz ' ECR profile and (c) Symmetric Bz < ECR magnetic profile. Plasma chamber limits marked with black line. Axial and radial coordinates are represented in image (b).

26

3.2

Measurement Procedure

Four diagnostics were conducted simultaneously to measure characteristic breakdown times during pulsed operation at 50 Hz. Fig. 3.4 shows the experimental set-up. Plasma injected and reflected power were obtained by means of one bidirectional coupler placed in the waveguide (see Fig. 2.2 in chapter 2). An electrical probe made of tungsten wire, 6 mm long and 0.5 mm in diameter, was mounted inside a 1.5 mm-diameter alumina tube placed in the center of the chamber. This probe is polarized by a DC power supply at 100 V. By measuring the voltage across a 10 kΩ resistor connected to the probe, the saturation electron current was obtained above the plasma space potential with a temporal resolution of 100 ns. A floating ground oscilloscope was used because the probe ground needs to be floating at polarization voltage.

Dual directional coupler

Tuner

Dual Quartz directional window coupler Plasma

MAGNETRON 2.45 GHz PULSED MODE Coils

OSCILLOSCOPE CH1

CH2

CH3

CH4

Photo detector

Fiber optics

10 KΩ COMPUTER

Power Supply 0 - 100 V

Figure 3.4: Schematic representation of the diagnostics set-up for breakdown time measurement.

A fiber optics bundle with a diameter of 6.25 mm placed on the observation window was connected to a high-speed photo detector of 14 ns rise time to give a signal proportional to the intensity of the light emitted by the plasma. The range of observed wavelengths was 350 - 600 nm, centered at 540 nm by using an optical collimator. Electron-ion recombination and neutral gas excitation spectral line contributions to light emission are both proportional to the product between electron and ion density


27

and inversely proportional to the square root of electron temperature [21]. Moreover, the visible light, although it is produced by neutral atoms excitation, can be considered as an interesting indicator for evolution of the ionization processes regarding that Hydrogen excitation and ionization cross section ratio is well known. In addition, this signal was very useful in helping to detect anomalous behaviors related to misplaced plasma formation in the MW coupler, or alternating breakdowns between the chamber and the injection side near the quartz window. In chapters 8 and 9 it will be discussed in detail the role of visible light as a diagnostic tool capable to give information about the plasma species fraction. A typical oscilloscope record of 64 averaged signals can be seen in Fig. 3.5 where (a) is the injected power Pi , (b) is the reflected power Pr , (c) the is light intensity signal and (d) is the current signal from the probe.

MW coupling

Plasma formation

Figure 3.5: Typical oscilloscope signal for measuring breakdown times. (a) Incoming power, (b) reflected power, (c) plasma emitted light and (d) probe current.

These signals were obtained under the following experimental conditions: frequency 50 Hz, incoming microwave power 600 W, reflected microwave power 150 W, duty cycle 70 %, Hydrogen pressure 6.2 x 10−3 mb with the asymmetric magnetic profile corresponding to the curve (a) in Fig. 3.2 and the 2D color map (a) shown in Fig. 3.3. The incoming power rise time of about 3 μs allowed us to make clear measurements of plasma breakdown evolution with respect to excitation. Fig. 3.5 shows a significant slope change in light and probe current signals simultaneous with the reflected power drop associated with the MW coupling. It is reasonable

28

to understand these instances as a characteristic time for MW coupling process when an efficient power absorption is taking place. Fig. 3.6 shows the time evolution for the ratio between absorbed power (Pi −Pr ) and the incoming power, Pi , calculated directly from the measurements shown in Fig. 3.5. It is highly significant that this relative absorbed power rises very fast when microwave coupling takes place. This process is closely associated to the drop in electric field strength inside the plasma chamber [22–24]. This issue will be further discussed in chapter 6.

MW coupling

Figure 3.6: Typical relative absorbed power (Pi − Pr )/Pi calculated from direct measurements showing the MW-coupling time definition as proposed.

A closer look at Fig. 3.5 and Fig. 3.6 suggests that plasma processes during breakdown may be understood to be composed of two stages: Microwave coupling: The first stage during which microwave power matching is in progress. This is an early breakdown stage characterized by a low ionization rate and fast coupling dynamics between microwaves with a weak plasma inside the chamber. This process may be closely associated with the drop in electric field strength inside the plasma chamber according to the behavior of the absorbed power [22–24]. We define here the characteristic time for this stage as the time when relative absorbed power (Pi − Pr )/Pi increases rapidly. Plasma formation: The second stage in which light emission increment is associated with an improving ionization rate and where the probe saturation current is rapidly achieved. During this period the microwave coupling is well established and the absorbed energy density is good enough to produce plasma evolution to final steady state


29

parameters. Saturation in both light emission and probe current signals was recorded at practically the same time during the plasma formation stage. We define the characteristic time of this stage as the time when saturation in the probe current is reached, this occurs in most cases in coincidence with the light emission saturation. Fig. 3.5 shows the proposed structure. According to this interpretation, we understand the breakdown time to be the sum of these two partial times. Note that the single measurement of injected and reflected power is not enough to observe the second stage of the plasma evolution, necessitating the implementation of further complementary diagnostics. Our definition therefore involves all the time resolved diagnostics described in this chapter. It is of note that even if plasma is totally established after this breakdown time and we can not appreciate any evolution of the measured signals after that, in some cases the species fractions can be evolving during much longer periods as will be seen in chapters 6 and 9.

3.3

Results

Measurements were conducted to obtain MW Coupling, Plasma Formation and Total Breakdown times. It is important to keep in mind that MW coupling time was measured between the incoming power rising edge and the (Pi − Pr )/Pi ratio strong increase, independently of the finally value reached by this ratio. On the other hand, the Plasma Formation time was measured between the end of the MW coupling time and the instant when probe current saturates which occurs in most cases simultaneously with the light signal saturation. Parameter

Values

Magnetic Field Profile Hydrogen Pressure Peak MW Power (W) Duty Cycle (%)

(a), (b) and (c) of Fig. 3.2 and Fig. 3.3 3.8 and 6.2 x 10−3 mb 300 to 1500 in steps of 300 10 to 90 in steps of 10

Magnetron pulse frequency

Constant at 50 Hz

Probe Polarization (V)

Constant at 100 V

Table 3.1: Parameters used during the experiments.

The range of parameter variation during experiments is shown in Table 3.1. Three main cases were studied, corresponding with the three magnetic field profiles detailed in Fig. 3.2 and Fig. 3.3. Each one was studied for two Hydrogen operation pressures by

30

scanning injected power and duty cycles ranges of 300-1500 W and 10-90 % respectively. Magnetron pulse frequency was 50 Hz and probe polarization voltage of 100 V were both fixed during all of the experiments.

3.3.1

Breakdown Time Measurements with Bz > ECR

This section shows the results of the experiments carried out under the Bz > ECR magnetic field profile shown in Fig. 3.2 (a) and Fig. 3.3 (a).

3.3.1.1

Low Pressure Regime (3.8 x 10−3 mb)

Fig. 3.7 shows measurements corresponding to Hydrogen pressures of 3.8 x 10−3 mb where times are represented as surfaces obtained by linear interpolation between measured points. Such times are plotted as a function of MW incoming power and duty cycles. Fig. 3.7 (a) is the MW coupling time, Fig. 3.7 (b) is the plasma formation time and Fig. 3.7 (c) is the total time as the sum of previous values.

(a)

(b)

(c)

Figure 3.7: Breakdown times for a flat Bz > ECR magnetic field distribution corresponding to Fig. 3.2 (a) and 3.3 (a) for 3.8 x 10−3 mb of Hydrogen pressure. (a) MW coupling time. (b) Plasma formation time. (c) Total breakdown time as the sum of previous values.

Under these experimental conditions, it is notable that MW coupling time shows a tendency to grow while power decreases, as can be seen in Fig. 3.7 (a). Faster coupling times are strictly related to higher microwave incoming power while the dependence with duty cycle is relative smooth. Another interesting issue is the relatively wide range of variation where lower values start at 20 μs, reaching 120 μs in the high duty cycle and low power region. For this case, the maximum value reached for plasma formation time was 90 μs while minimum was 10 μs at the corner of high MW power and low duty cycles. This behavior can be seen in Fig. 3.7 (c), where total breakdown time is represented as the sum of the previous values. It is clear that the main slope follows microwave power, reaching values of 180 μs for high duty cycles and low microwave powers while


31

the opposite corner of low duty cycles and high power is characterized by low breakdown times of 50 μs. Especially interesting is the corner of low power, about 300 W and high duty cycles of 80-90 %, where an unstable area with high jitter is found and it is not possible to record data. This area is shown in Fig. 3.7 as empty. It is important to notice that these empty areas are not representing saturation regions. Such areas are unstable regions where the breakdown process is unpredictable and they represent the cut-off values from where the influence of high jitter does not permit us to take useful data.

3.3.1.2

High Pressure Regime (6.2 x 10−3 mb)

Fig. 3.8 shows times corresponding to Hydrogen pressures of 6.2 x 10−3 mb embedded in a Bz > ECR. The same scheme used in the previous figure is followed, where the plots are a function of MW incoming power and duty cycles. It should be noted that Fig. 3.8 (a) shows that MW coupling time shows a behavior involving a practically flat surface, with lower values for the range of powers and duty cycles studied compared to previous case of low pressure. No remarkable slopes were recorded for this case and values were in the range of 10-30 μs, indicating a fast MW coupling. Fig. 3.8 (b), on the other hand, shows a significant increasing tendency for plasma formation time when the duty cycle is increased. It is notable that the maximum values were reached at 90 % of duty cycle where a smooth curve with a maximum value of 85 μs at 900 W was recorded. This behavior is reflected in Fig. 3.8 (c) where total breakdown time receives the influence of plasma formation time reaching maximum values of 110 μs and minimums of 40 μs.

(a)

(b)

Figure 3.8: Breakdown times for a flat Bz > ECR magnetic field distribution corresponding to Fig. 3.2 (a) and 3.3 (a) for 6.2 x 10−3 mb of Hydrogen pressure. (a) MW coupling time. (b) Plasma formation time. (c) Total breakdown time as the sum of previous values.

In general terms, a comparison between the two pressures at Bz > ECR is interesting because the surfaces of total breakdown time show different behaviors. The lower pressure case presents more aggressively changing behavior depending on microwave power

32

values, and the higher pressure case shows smoother behavior with changes dominated mostly by duty cycle. However, in both cases, the lowest value is reached at the corner of high power and low duty cycles. Another interesting issue is that under this last pressure regime, no unstable high jitter area is observed. The highest value is reached at the low power and high duty cycle corner.

3.3.2

Breakdown Time Measurements with Bz ' ECR

The plasma behavior in the Bz ' ECR magnetic field profile, previously shown in Fig. 3.2 (b) and Fig. 3.3 (b), is stable showing a relatively wide range of powers and duty cycles where measurements can always be conducted with good coupling and high emitted light intensity.

3.3.2.1


Fig. 3.9 shows the measurements corresponding to Hydrogen pressure of 3.8 x 10−3 mb. Following previous section scheme, breakdown times are plotted as function of MW incoming power and duty cycles: Fig. 3.9 (a) is the MW coupling time, Fig. 3.9 (b) is the plasma formation time and Fig. 3.9 (c) is the total time as the sum of previous values.

(a)

(b)

(c)

Figure 3.9: Breakdown times for a flat Bz ' ECR magnetic field distribution corresponding to Fig. 3.2 (b) and 3.3 (b) for 3.8 x 10−3 mb of Hydrogen pressure. (a) MW coupling time. (b) Plasma formation time. (c) Total breakdown time as the sum of previous values.

For this experimental condition, the MW coupling time shown in Fig. 3.9 (a) remains practically constant between values from 30 to 50 μs, showing a slightly upward tendency from low duty cycles and high power to high duty cycles and low power. On the other hand, plasma formation time in Fig. 3.9 (b) behaves differently, and is characterized by a significant drop along the line of middle powers at about typically 900 W. This times are relatively high, starting at values of 40 μs and reaching 95 μs at the corner of


33

low powers and high duty cycles. Total breakdown times are represented in Fig. 3.9 (c) where the influence of plasma formation times on the surface shape is clear. Breakdown times that reach 150 μs for lower power values and high duty cycles just prior to entering in the unreliable area of high jitter behavior are in contrast with the faster values of 50 μs at the opposite corner of high powers and low duty cycles.

3.3.2.2


Fig. 3.10 shows the cases corresponding to 6.2 x 10−3 mb Hydrogen pressure embedded in a Bz ' ECR. The same scheme used in the previous figure is followed with the plots being function of MW incoming power and duty cycles: (a) is the MW coupling time, (b) is the plasma formation time and (c) the total time as the sum of previous values. MW coupling times shown in Fig. 3.10 (a) are faster than those of the previous lower pressure case remaining practically constant at values of 20-30 μs. They show a better coupling behavior practically independent of duty cycle and power variations. A different situation can be seen in Fig. 3.10 (b), where the range of power and duty cycle studied show a strong tendency to increase the plasma formation times to values that reach 130 μs when duty cycles are increased.

(a)

(b)

(c)

Figure 3.10: Breakdown times for a flat Bz ' ECR magnetic field distribution corresponding to Fig. 3.2 (b) and 3.3 (b) for 6.2 x 10−3 mb of Hydrogen pressure. (a) MW coupling time. (b) Plasma formation time. (c) Total breakdown time as the sum of previous values.

It is notable that coupling and plasma formation times show opposite behavior when Hydrogen pressure is increased. While the first one decays, showing a faster coupling dynamics, the second one increases, showing the necessity of more time for plasma parameter evolution. These two opposing dynamics are practically compensated in the calculation of total breakdown time, as shown in Fig. 3.10 (c), there is a big difference in comparison with previous lower pressure case. This surface starts at higher values that show a gentle upward tendency of between 70 and 150 μs from low to high duty

34

cycles respectively. Another issue is the absence of the depression, corresponding to the middle value MW power at 900 W, that is present at lower pressure.

3.3.3

Breakdown Time Measurements with Bz < ECR

The plasma breakdown times set out in this section correspond with the magnetic field profile representations in Fig. 3.2 (c) and Fig. 3.3 (c).

3.3.3.1


Fig. 3.11 shows measurements corresponding to Hydrogen pressures of 3.8 x 10−3 mb where Fig. 3.11 (a) is the MW coupling time, Fig. 3.11 (b) is the plasma formation time and Fig. 3.11 (c) is total breakdown time as the sum of previous values. In general terms, plasma behavior for this relatively low Bz < ECR profile and low pressure regime is unstable, showing a narrow range of powers where measurements could be conducted, and always with poor coupling, low emitted light intensity and a remarkable level of jitter.

(a)

(b)

(c)

Figure 3.11: Breakdown times for a Bz < ECR magnetic field distribution corresponding to Fig. 3.2 (c) and 3.3 (c) for 3.8 x 10−3 mb of Hydrogen pressure: (a) MW coupling time. (b) Plasma formation time. (c) Total breakdown time as the sum of previous values.

Fig. 3.11 (a) shows MW coupling times where a narrow power range between 1200 and 1500 W was the only one where any measurements could be taken. For these cases, the system presents a poor MW coupling with high reflected microwave power values. The rest of powers and duty cycle values tested presented a behavior characterized by high jitter or even plasma allocation outside of the plasma chamber (inside the MW coupler or the transition piece). Such high jitter behavior practically disqualifies these data for any application requiring a reasonable level of reproducibility. Fig. 3.11 does not show data in such experimental conditions and the corresponding parameter area looks empty.


35

Fig 3.11 (a) shows MW coupling times for those experimental conditions where measurements could be made. Fig. 3.11 (b) shows plasma formation times with practically the same behavior, and always with values between 40 μs and 80 μs. Finally, Fig. 3.11 (c) shows the total breakdown time as the sum of previous values, reaching maximum values of 180 μs at 40 % of duty cycles.

3.3.3.2


Fig. 3.12 shows the measurements corresponding to Hydrogen higher pressure of 6.2 x 10−3 mb. Fig. 3.12 (a) is the MW coupling time, Fig. 3.12 (b) is the plasma formation time and Fig. 3.12 (c) is the total time as the sum of previous values. This case shows a more reproducible behavior and data could be taken under a wider range of incoming power and duty cycle conditions.

Figure 3.12: Breakdown times for a flat Bz < ECR magnetic field distribution corresponding to Fig. 3.2 (c) and 3.3 (c) for 6.2 x 10−3 mb of Hydrogen pressure. (a) MW coupling time. (b) Plasma formation time. (c) Total breakdown time as the sum of previous values.

Fig. 3.12 (a) shows that MW coupling times remained practically constant and gave values ranging from 20 μs to 50 μs. An exception was recorded at the corner of low power and high duty cycles where an unstable area was found. Plasma formation time is represented in Fig. 3.12 (b), where it can be seen that smaller values of around 70 μs were obtained for high power and low duty cycles, while the rest of the surface is characterized by a saddle-like shape that reaches higher values of 150 μs at low powers and high duty cycles, just before the flat top unstable high jitter area. Total breakdown time is represented in Fig. 3.12 (c) as the sum of previous surfaces. The plasma formation time surface is mainly reflected in the breakdown time surface, making it the same shape as at Fig. 3.12 (b), with an unstable area at the top. However, the faster breakdown times keep values of 70 μs in the corner of high power and low duty cycles, where good coupling conditions are evident. It is significant that in most of the cases the breakdown time evolves toward an unstable area where jitter makes the phenomena completely unpredictable. From the

36

point of view of reproducibility and applicability to the design of an ECRIS, the existence of such unstable areas represents a serious limitation.

3.4

Simple Model of Breakdown Time

The neutral gas pressure dependence of breakdown time can be understood by means of the model proposed by O. Tarvainen et al. [25] where:

tbreak = τion ln (ne,critical /ne0 )

(3.1)

where tbreak is the breakdown time, τion = [nn hσion vr i]−1 is the characteristic ionization time, σion the ionization cross section, vr the relative velocity (electron and neutral particle) with h i denoting the average over the velocity distribution, nn is the neutral gas density, ne,critical is the critical value that the electron density has to reach to produce breakdown and ne0 is the electron density in the plasma chamber at the beginning of the microwave pulse. Assuming that during breakdown hσion vr i and nn are independent of time, Eq. 3.1 predicts that the time required for plasma breakdown is inversely proportional to neutral particle density. This was successfully checked with different gas and the calculations gave the order of magnitude correctly. The strong impact of the initial electron density ne0 at the instant that the microwave pulse starts is also predicted by Eq. 3.1. Considering that this initial value is the final value of the seed electron density evolution during plasma off-time that comes from the decay of the previous pulse, the dependence of breakdown time with incoming power and duty cycle may be studied. It needs to be highlighted the fact that the dynamic associated to the seed electrons is still unknown. Without confinement the plasma off-time may be long enough for the electron population to diffuse towards the walls. However, experimental results show that even for long plasma off-times they can survive enough to influence the next pulse breakdown. We will now consider a simple model in order to explain our experimental results. Accepting that after the microwave excitation pulse is off, an electron density remains embedded in the neutral gas during a time that is sufficient to produce some influence on the breakdown of the following pulse, this density should evolve between pulses (plasma off-time), decreasing according to the rate of recombination with ions and radial diffusion under the influence of magnetic field applied inside the chamber. Based on these assumptions the electron density evolution should obey the following:


37

dne = De ∇2 ne − ne ni hσrec vr i dt

(3.2)

where De is the ambipolar diffusion coefficient for ECR plasmas, ne is the electron density, ni the ion density, σrec the recombination cross-section under the influence of the magnetic field, vr the relative velocity (electrons and ions) and h i denotes the average over the velocity distribution. Assuming that hσrec vr i remains practically constant, ne = ni = n during the recombination process and infinite cylindrical symmetry, then Eq. 3.2 takes the following shape in cylindrical coordinates: De ∂ dn = dt r ∂r

∂n r ∂r

− n2 hσrec vr i

(3.3)

This equation was solved by Gray and Kerr [26] with the following dimensionless variables: N = n/ne,res ; τ = tDe /Λ; Λ = R/λ1 ; ρ = r/Λ where, for our case, we use ne,res as the residual value of ne soaked in neutral gas immediately after plasma is off in the z axis (r = 0), R is the plasma chamber radius and λ1 = 2.405 for cylindrical symmetry, producing: dN 1 ∂ = dτ ρ ∂ρ

∂N ρ − γN 2 ∂ρ

(3.4)

with boundary conditions;

N (λ1 , τ ) = 0 and;

N (ρ, 0) = J0 (ρ)

where the first is the zero density condition at the plasma chamber wall and the second is the initial radial density distribution assumed as Bessel’s function by solving eq. 3.2 at t = 0. Note that Eq. 3.4 has only one parameter,

γ=

hσrec vr i n2e,res hσrec vr i ne,res Λ2 = De ne,res De /Λ2

(3.5)

that can be understood as the ratio between the initial axial electron loss rate in the absence of diffusion, and the corresponding loss rate resulting from diffusion

38

alone. Thus γ is a measure of the degree to which the plasma is initially recombination controlled (γ 1) or diffusion controlled (γ 1). In our case, to calculate γ we need to estimate hσrec vr i and De . Assuming that during plasma-off time between pulses, seed electrons are at 1 eV of temperature (order of magnitude estimation), we can use the approximation from Ref. [27] for a neutral Hydrogen plasma:

−19

hσrec vr i = 0.7 × 10

13.6 Te (eV )

1/2 (3.6)

and the Bohm semi-empirical approximation from Ref. [28] to estimate the diffusion coefficient for an ECR plasma:

DBohm =

kTe 16eB

(3.7)

Considering ECR magnetic field and taking a deliberately high initial electron density, like the typical measured value of 1016 m−3 during flat top pulse, as a means to check the best case for collisions with ions and the worst case for a diffusive regime:

γ = 3.8 × 10−6 1

(3.8)

This shows that for our ECR plasma, the predominant initial condition is completely diffusive, even when considering the best parameter set for collisions. This calculation confirms the behavior proposed by Tarvainen et al. in [28] on the basis of experimental data. The solution of Eq. 3.2 is shown in Fig. 3.13 where the diffusive behavior can be observed by looking at how the Bessel’s function profile radial density distribution decays progressively in the time. Especially interesting for our experimental data comparison is the level curve at r = 0 because this is the position where our electrical probe was placed during measurements. Fig. 3.14 shows this curve where time electron density evolution is displayed. Furthermore, and taking into account that characteristic ionization time in Eq. 3.1 is τion = [nn hσion vr i]−1 , if we assume the approximation hσion vr i ≈ σion hvr i, we can relate hvi with the incoming power by using the Fokker-Planck-based model proposed by Guest [20]. This model proposes that the electron dynamic of plasma cyclotron heating is driven by an energy transfer mechanism between high energetic electrons (that are directly


39

1.0

1.0 0.9

0.8

0.8

) N (n/n res

0.7

0.6

0.6

0.5

0.4 0.4

0.3

0.2 0. 0

0.2

0. 5

1. 0

1 .0

(r

/ D (t

1 .

2. 0

0.1

)

1. 5

e

0 2.

5

.5 0 02.

0.0

/

0 .0

0 .5

)

2. 5

3. 0

Figure 3.13: Calculation of density temporal and spatial evolution.

1.0

0.9

0.8

N (n/n

res

)

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0 0.0

0.5

1.0

1.5

(t D / e

2.0

2.5

3.0

)

Figure 3.14: Calculation of seed electron density temporal evolution at r = 0.

accelerated by RF) and a population of low temperature electrons. The dynamic friction force that produces this energy transfer is calculated as being proportional to hvr i, and the rate of transferred energy is proportional to hvr i2 . Under these assumptions, and considering that neutral gas density nn is proportional to neutral gas pressure, we can estimate the characteristic ionization time by:

τion

pmax ≈ τionmin p

r

Pmax P

(3.9)

where τionmin is the minimum order of magnitude of the characteristic ionization

40

time estimated from our data corresponding to maximum power and neutral gas pressure used in our experiments, pmax is the maximum neutral gas pressure, Pmax is the maximum value of power, p is pressure and P is power. Finally, Eq. 3.1 takes the shape: tbreakdown ≈ τion ln

ne,critical 1 ne,res N

(3.10)

with τion from Eq. 3.9, N coming from calculations shown in Fig. 3.14 by converting dimensionless time to real and using the off-time between pulses τof f = (1−DC/100)/f , where DC is the duty cycle and f is the operation pulse frequency. By using Pmax = 1500 W; pmax = 6.2 x 10−3 mb and f = 50Hz; we can use Eq. 3.10 to estimate the order of magnitude of breakdown time as a function of duty cycle and power under the following assumptions:

i) Microwave coupling is the most significant factor related to the breakdown time calculated by Eq. 3.1 because the process of ionization by collision is mainly driven by the electric field enhancement produced during this stage. ii) The minimum characteristic ionization time is estimated from the ECR microwave coupling times shown in Fig. 3.9 (a), where the order of magnitude is about 1 μs. This is coincident with calculations of optimal breakdown conditions in an ECR 2.45 GHz plasma by Guest [20]. iii) The ratio ne,critical /ne,res = 106 is assumed in order to maintain the assumption of ne,critical /ne0 = 107 used in Ref. [25] considering the variation range of N. Fig. 3.15 shows the two surfaces calculated for our working pressures of 3.8 x 10−3 mb (a) and 6.2 x 10−3 mb (b). By comparing theses surfaces with Fig. 3.9 (a) and Fig. 3.10 (a) respectively, it can be seen that microwave coupling time data match the order of magnitude with calculations. It is significant that constant minimum breakdown times of about 20 μs are obtained over the whole range of duty cycles at 1500 W for both pressure cases. High incoming powers may produce high residual electrons resulting in a sufficiently high seed electron density at the beginning of pulses. This can guarantee fast coupling in a wide range of duty cycles. Under these circumstances, the saturation of faster coupling times with respect to duty cycle may be produced as can be seen in Fig. 3.9 (a) and Fig. 3.15. However, a completely different situation was observed in our case for over ECR magnetic fields. Looking at data shown in Fig. 3.7 (a) and Fig. 3.8 (a), there are significant differences between calculations and experimental data. Especially interesting is the case of lower pressure at 3.8 x 10−3 mb represented in Fig. 3.7 (a) where


41

160 140

s)

120 Time (

100 (a) (b)

80 60

(W cy cl e

ow er

y

(%

20

P

D ut

)

40

)

0

Figure 3.15: Calculation of breakdown times for 3.8 x 10−3 mb (a), and 6.2 x 10−3 mb (b).

the extended range of measured values and the increment of breakdown times are not described by our simple model.

3.5

Summary and Conclusions

The plasma diagnostic tool presented in this chapter is based on an analysis of the temporal evolution of four different measurements: incoming and reflected power, emitted light and electron saturation current. The capability of this tool to determine plasma breakdown time and its different stages was clearly demonstrated. An experimental study of breakdown time in TIPS was carried out for three different magnetic field profiles Bz > ECR, Bz ' ECR and Bz < ECR as shown in Fig. 3.2 and 3.3. Measurements revealed a two-stage structure for the breakdown process. There is an early breakdown stage that we have called MW Coupling; where the microwave power matching is in progress under a low ionization rate, and fast coupling changes between microwaves and the weak plasma inside the chamber are taking place. Then there is a second stage that we have called Plasma Formation; during which light emission increment is associated with an ionization rate increment and where saturation probe current is rapidly reached. During this last stage, the MW coupling is well established and the absorbed energy density is good enough to produce plasma evolution to final steady state parameters. Fig. 3.5 shows the structure proposed above.

42

According to this interpretation, the study described in section 3.3 was conducted on the basis of measurements of MW coupling and plasma formation times for the three magnetic field configurations, for two different Hydrogen operation pressures scanning power and duty cycle to establish typical breakdown times. For Bz > ECR and a relatively low Hydrogen pressure of 3.8 x 10−3 mb, breakdown shows a strong dependency on injected microwave power, where the best coupling times are obtained for low duty cycles and high microwave powers. This effect is moderated significantly when a higher Hydrogen pressure of 6.2 x 10−3 mb is used. For Bz ' ECR, the general behavior is characterized by good microwave coupling and light emission over a wide range of powers and duty cycles for the two Hydrogen pressures under study. While MW coupling times show small variations, plasma formation times reach an instability area for low powers and high duty cycles that affects the total breakdown time behavior. For Bz < ECR and a relatively low Hydrogen pressure of 3.8 x 10−3 mb, breakdown is dominated by instabilities that determine a narrow operating range in which measurements can be taken. Low light emission and also low probe electrical current are symptoms of low density and low temperature plasma. However, if the pressure is increased to 6.2 x 10−3 mb, behavior is significantly improved, showing just a small working area of powers and duty cycles where instabilities and jitter are decisive at low powers and high duty cycles. Tables 3.2, 3.3 and 3.4 show a brief data summary with maximum and minimum measured times to keep in mind the range of values recorded. Note that breakdown times are not necessarily the sum of previous values in these tables because maximum and minimum values are not normally coincident in surfaces, as can be seen in the respective figures 3.7 to 3.12. In general terms, the process with the deepest impact on total breakdown is plasma formation, which dominates the general behavior. Disregarding the case of Bz < ECR where range operation is too small, coupling times display smooth behavior with a relatively small variation. The measurements of injected and reflected power suggest a process associated with the drop in the electric field strength inside the the plasma chamber [22–24]. The effect of this E-field drop will be further discussed in chapter 6. At the beginning, the power absorption is poor, allowing the electric field strength in the plasma chamber to reach characteristic values for the coupling system and cavity, but once the ionization process starts, the absorption becomes significant and the electric field drops.


Times

3.8 x 10−3 mb Min. Max.

43

6.2 x 10−3 mb Min. Max.

Remarks

MW Coupling

20

120

10

35

Stable behavior.

Plasma Formation

20

90

30

85

Breakdown

50

170

50

110

High light emission Unstable area for high duty cycle and low power.

Table 3.2: Breakdown summary for Bz > ECR: profile of Figs. 3.2 (a) & 3.3 (a)

Times

3.8 x 10−3 mb Min. Max.

6.2 x 10−3 mb Min. Max.

Remarks

MW Coupling

40

55

10

30

Good coupling.

Plasma Formation

40

90

40

130

Breakdown

60

150

60

150

High light emission. Unstable area for high duty cycle and low power.

Table 3.3: Breakdown summary for Bz ' ECR profile of Figs. 3.2 (b) & 3.3 (b)

Times

3.8 x 10−3 mb Min. Max.

6.2 x 10−3 mb Min. Max.

Remarks

MW Coupling

40

105

20

50

Unstable behavior.

Plasma Formation

40

80

60

150

Narrow operating range.

Breakdown

60

170

60

190

Low light emission.

Table 3.4: Breakdown summary for Bz < ECR: profile of Figs. 3.2 (c) & 3.3 (c)

In all cases, an increment in breakdown time for high duty cycles and low powers was reported, while the best situations for good MW couplings were observed for high microwave powers and low duty cycles. On the other hand, it should be pointed out that measurements of injected and reflected power are not sufficient to describe the full breakdown process because they are not sensitive to changes associated with the plasma formation stage. This fact makes the plasma diagnostics developed in this chapter a valuable tool to describe the plasma breakdown process. The influence of plasma formation time in breakdown process, and the impossibility of determining plasma formation time just by measuring incoming and reflected power should be a consideration for developers of automatic plasma optimization control systems based only on measurements of injected and reflected power. A simple model based on the influence of the seed electrons remaining in the neutral gas between pulses was developed with the aim of describing our experimental data as

44

a function of duty cycle and power. The diffusive nature of the seed electron dynamic during the switch-off time between pulses was shown to be in coincidence with the experimental data generated by other experiments [28]. The calculations showed a good correspondence with our experimental data for the case in which the ECR magnetic field profile was used but also showed a mismatch for magnetic field profile over ECR.

Chapter 4

Plasma Density and Temperature Measurements during Breakdown From the study of breakdown times carried out in the previous chapter arose the idea of developing a diagnostic to further explore the breakdown phenomena. The evolution of plasma parameters, especially plasma electron density and temperature, during breakdown can give valuable information about the physical processes involved. In this chapter, a new plasma diagnostics is presented, designed to determine plasma electron density and temperature evolution during breakdown. Understanding plasma physics processes during transients (breakdown and decay) in pulsed plasma sources is of special interest for many application fields such as particle accelerator science, nuclear fusion reactors and the plasma processing industry [5, 29]. Extensive research has been conducted on this subject by different researchers with electrical probes, spectroscopy and radiation diagnostics under a wide range of parameters for different plasmas. The processes involved during breakdown should be definitive for monocharged beam current optimization as well as for the improvement of multiple charged ion production efficiency. Both cases are of great interest and under close study in the ECRIS community [22, 25, 30]. Together with the diagnostics design and set-up, this chapter presents a study of the plasma breakdown process in TIPS by means of time-resolved Langmuir probe diagnostics and incoming and reflected microwave power measurements. The experimental conditions were set to be the same as those in breakdown time study in the previous chapter to make results comparable and to be able to extract conclusions.

45

46

4.1

Experimental Set-Up

4.1.1

Diagnostic Port

As is well-known in the plasma community, Langmuir probes are used immersed in plasmas to acquire characteristic I-V curves which permit us to estimate plasma electron temperature and density. In the experiments described in this chapter, the same diagnostic port described in the previous chapter was used to access the plasma with the probe. A Langmuir probe was used in place of the polarized probe employed previously to measure breakdown evolution times. Fig. 4.1 shows a section view of TIPS with the Langmuir probe installed.

(a) (b)

(c) (d)

Figure 4.1: Section view of TIPS where the Langmuir probe set-up is mounted: diagnostic port (a), probe holder (b), Langmuir probe (c) and observation window (d).

Fig. 4.2 shows a photograph of TIPS with the Langmuir probe system installed. Note that a millimetric probe displacement system (not visible in previous figure for clarity) was installed on the device. This system, based on axial and radial bearings, allows the displacement of the Langmuir probe inside the chamber with a spatial resolution of 1 mm. Fig. 4.3 shows a diagram of the probe construction with its dimensions. The probe was made of a 0.5 mm diameter tungsten wire placed inside a 1/16” outer diameter and 1/32” inner diameter alumina tube. To keep the tungsten wire well centered inside the tube, a hypodermic needle was placed inside it to help the wire pass through. This is very important, especially when the tube becomes metalized due to plasma. Any

Chapter 4. Plasma Density and Temperature Measurements during Breakdown

47

contact between the probe tip and the metalized part of the tube would result in a bigger collecting area and thus in false readings for plasma electron density and temperature.

Figure 4.2: Photograph of TIPS with the Langmuir probe set-up installed.

Figure 4.3: Diagram with details of the Langmuir probe construction and dimensions.

Fig. 4.4 shows a photograph of the probe holding piece with the Langmuir probe mounted in it. The tube is sustained by an Ultratorr 1/16” bored fitting that allows the axial displacement of the probe without breaking the vacuum. As is shown in the picture the alumina tube is metalized after several hours embedded in the plasma.

48

Figure 4.4: Photograph of the Langmuir probe mounted in the probe holder.

4.1.2

Magnetic Field Profiles

On the basis of experience gained during the experiments described in the previous chapter, we decided to use asymmetric Bz > ECR and symmetric Bz ' ECR as our main experimental conditions in terms of magnetic field, and abandon below-ECR configuration due to the instability of the plasma under these conditions and the limited range of working parameters already shown in previous chapter.

Figure 4.5: Axial magnetic field profiles used during experiments. (a) Asymmetric Bz > ECR and (b) Symmetric Bz ' ECR magnetic profile.


49

For easy reference, Fig. 4.5 reproduces once again the Bz magnetic field profiles along the z-axis experimentally measured. Fig. 4.6 shows the 2D simulated magnetic maps already shown in chapter 2. Magnetic map (a) represents the asymmetric case where Bz > ECR and (b) the symmetric case where Bz ' ECR in most of chamber volume.

Figure 4.6: 2D view of magnetic field profiles used during the experiments obtained by simulation. (a) Asymmetric Bz > ECR and (b) Symmetric Bz ' ECR magnetic profile.

4.2


Langmuir probe time-resolved measurements made on several kinds of plasma generators distinct from TIPS have been conducted recently for repetitive pulsed operation during plasma transients [18, 30–33]. In general terms, these measurements have

50

been obtained taking the I-V curve points one by one from different consecutive pulses, completing the voltage sweep in a predetermined time. When synchronization is done carefully, keeping the jitter sufficiently low throughout the process, it is possible to obtain an estimation of electron density and temperature at a precise, predefined instant. Fig. 4.7 shows a conceptual plan where this technique is represented using a typical I-V curve and an oscilloscope record, both obtained from the experiment. (a)

(b)

(c)

Figure 4.7: Conceptual plan showing that each point of the I-V curve (left) is obtained by averaging 100 measurements obtained in different consecutive pulses (right). Note that signal (a) is the incoming power, (b) the reflected power and (c) are the synchronism pulses used to check proper timing.

The Langmuir probe driver circuit (ESPION) is made by Hiden Analytical Ltd. and can acquire a single I-V point in 62.5 ns before rearming itself in another 14.6 μs which is required for data handling. Each point of the I-V curve is an average of the probe current acquired over 100 consecutive pulses at fixed probe voltage. Altogether, it takes several minutes to acquire a complete I-V curve corresponding to a single temporal data point, which inevitably exposes the measurement and data analysis to long-term shifts in plasma conditions and pulse-to-pulse oscillations. These effects can be minimized by carefully tuning the microwave coupling and averaging out the data. Applying this method, time resolved I-V curves can be obtained by synchronizing the Langmuir probe trigger driver and the magnetron via a delay generator. Several pulses can be seen in the oscilloscope record in Fig. 4.7, where signal (a) is the incoming power, (b) is the reflected power, and (c), the synchronism pulses from the probe driver system. Note that amplification for both channels (a) and (b) is the same. In order to see more precisely the instant where data are collected, it is necessary to zoom the breakdown stage time interval. An example of a typical timing record during the study of the breakdown transient is shown in Fig. 4.8 where signal (a) is the incoming power, (b) is the reflected power,


51

and (c) is the synchronism pulse from the probe driver system. The front of this pulse represents the instant where measuring is taking place.

Figure 4.8: View of a synchronism record during the breakdown process: (a) is the incoming power, (b) is the reflected power and (c) is the synchronism pulse where its rise indicates the instant of measuring.


Tuner



OSCILLOSCOPE CH1

CH2

Langmiur Probe Controller

CH3 Synchrony pulses Monitor

COMPUTER

Pulse Delay Generator

Figure 4.9: Schematic representation of the diagnostics set-up for time resolved density and temperature measurements during breakdown. Resolution is 200 ns.

52

Fig. 4.9 represents the experimental set-up where the timing strategy is shown. Note that the delay pulse generator produces a predetermined synchronization by firing the probe driver system with a TTL pulse. A sample of such signal is recorded in the oscilloscope in order to check the instant of time where the I-V curve is obtained (Fig. 4.8 curve(c)). Initially, 6 mm tungsten wire probes with diameters of 0.05 and 0.1 mm diameter by long were used following Chen’s design criteria for RF Argon low density plasmas [34]. However, noisy curves of low intensity were obtained. After increasing the probe diameter to 0.5 mm, reliable results with a remarkable reproducibility were measured in our Hydrogen plasma by placing the probe tip in the middle of plasma chamber. Jitter was carefully checked, obtaining a value lower than 200 ns in order to ensure measurement quality. By modifying Langmuir probe trigger delay, a set of I-V curves produced during the plasma breakdown were obtained.

4.3

Data Analysis and Calculations

Langmuir probe I-V curve analysis is an issue still under discussion in literature. For low plasma densities and small probes, the sheath expansion effect produces an increase in the collected current because the effective area for particle collection is the sheath area and not the probe’s geometric area [35, 36]. This situation is reflected by ion currents that do not show clear saturation values, increasing gradually with increasing negative voltage. The slope in the I-V curve ion current branch is an unequivocal sign of low density plasma with values constantly below the critical density of 7.45 x 1016 m−3 for 2.45 GHz microwaves. Fig. 4.10 shows typical I-V curves obtained during the experiments. Curve (a) has been measured at 60 μs i.e. during the steady-state and curve (b) at 15 μs i.e. during the plasma breakdown transient discussed in the following section. The electron temperature Te is generally estimated from I-V curves, assuming a Maxwellian Electron Energy Distribution Function (EEDF) even when it is well known that such an assumption can not be supported during transient stages like the breakdown or decay of pulsed plasmas. In this way, the development of methods to calculate the EEDF and plasma parameters from experimental probe data during plasma transients with non-maxwellian EEDFs is still a matter open to research and discussion [37, 38]. In this work, temperature and density calculations are made by two different methods that are explained in the following subsections.


53

(a)

(b)

Figure 4.10: Langmuir probe I-V curves obtained during the experiments. Curve (a): typical case during the steady-state. Curve (b): Typical case corresponding to the breakdown plasma condition (60 μs and 15 μs after incoming microwave pulse).

4.3.1

Difference between Plasma and Floating Potentials

This method estimates plasma temperature by using the difference between plasma and floating potentials. Two expressions are well known for these calculations, the one proposed by Lieberman and Lichtemberg in their book “Principles of Plasma Discharges and Material Processing” [39].

p Vp − Vf = kTe ln (mi /2πme )

(4.1)

and the practically equivalent equation proposed by Saudit and Chen [40],

Vp − Vf = (kTe /2) ln (2mi /πme )

(4.2)

where Vp is the plasma potential, Vf is the floating potential, kTe is in units of eV , mi is the unit mass of the ion and me is the unit mass of the electron. For Hydrogen plasmas with a varying species fraction, Eq. 4.1 and Eq. 4.2 take the form Te = 0.26...0.35(Vp − Vf ). We have used a coefficient of 0.35 because the external magnetic field may increase the coefficient to values > 0.3 [41]. Obviously, to use this method it is necessary to determine Vf and Vp from the experimental I-V curve. Vf is defined as the I-V curve zero crossing and Vp as the

54

maximum value of the I-V curve first derivative. With this two assumptions we obtain an estimation of the temperature value Te,1 . Electron density is then calculated by:

ne =

4I p es eAp 8kTe /πme

(4.3)

where Ies is the electron saturation current, e is the electron unit charge and Ap is the surface of the cylindrical probe. It is clear that to use Eq. 4.3 it is necessary to estimate Ies . If we accept that at Vp the measured current is the sum of Ies and the ion current value at this point, Ies can be estimated. This estimation has to be made assuming some approximations about the ion current. In other words, we have to fit the behavior of the ion current on the left branch of the I-V curve to estimate Ies . If the fitting employed is linear, then Ies is calculated as the difference between the current measured at the plasma potential I(Vp ) and the value obtained with the linear fitting at the same voltage Vp [35]. However, if the fitting employed is parabolic in accordance with Laframboise’s calculations [36, 42], the value of the ion current at the plasma potential is null, Ii (Vp ) = 0, and Ies is calculated as the value of the current at the plasma voltage, ergo Ies = I(Vp ) can be assumed, obtaining Ies directly from the I-V curve. These two calculation branches produce two possible values for electron densities, ne1 and ne2 , from Eq. 4.3 with Te,1 .

4.3.2

Slope of a Linear Fitting in a Current Logarithm-Voltage Plot

This method estimates the electron temperature (Te,2 ) on the basis of the exponential dependence of the electron current on the probe polarization voltage for V ≤ Vp i.e.:

Ie (V ) = Ies

−e(Vp − V ) exp kTe

(4.4)

where Te is determined by means of the linear fitting slope on a current logarithm vs. voltage representation of the probe curve. In order to determine the electron density it is necessary to obtain Vp . At this point, it is important to take into account that Vp can be calculated by using two techniques. The first is to calculate it as the maximum value of the first I-V derivative, as in previous method, and the second is to use the crossing point of two linear fittings in a current logarithm vs. voltage representation: the first used to find Te and the other fitting the V > V p branch of the curve. Such values of Vp are usually close. Fig. 4.11 shows a typical I-V curve obtained during experiments and


55

Slo

pe

Te

,2

it corresponding log(I)-V curve. The image on the left shows the I-V curve where the floating potential has been marked for clarity although it is not used in this method. The image on the right shows the current logarithm plot where the two linear fits and the plasma potential are marked.

Vf

Vp

Figure 4.11: Langmuir probe curve and its corresponding log(I)-V curve with linear fittings to determine electron temperature and plasma potential.

Electron density is calculated by Eq. 4.3 and the approximations relating to the ion current have to be made again with two alternatives: linear or parabolic, as described in the previous method. These calculation branches give two density values for each plasma potential estimation to end up with four more values for electron density ne3 , ne4 , ne5 and ne6 . Comparisons between temperature and electron density evolution during breakdown obtained by the two methods described are shown in Fig. 4.12 and Fig. 4.13 respectively under the following experimental conditions: incoming peak power of 900 W, pulsed frequency of 100 Hz, duty cycle of 10 %, Hydrogen pressure 3.8 x 10−3 mb, and ECR symmetric magnetic profile corresponding to curve (b) and magnetic map (b) of Fig. 4.5 and Fig. 4.6 respectively. These results are an example of temperature and density evolution obtained during the breakdown process to see how the results depend on the method employed. Note that results keep the shape of the evolution within a relatively narrow band of values. This characteristic and its reproducibility suggest that estimations of plasma parameters can be taken with a reasonable reliability. The error bar in temperature measurement is estimated to be below 10 %, reaching 1 eV at high values during transient and 0.25 eV during the steady state. This error is associated to the assumed Maxwell-Boltzmann EEDF, however, if the EEDF is different the error would be bigger. This fact would not be in any case associated to analysis method. Accuracy in electron density is estimated to be about 1 x 1016 m−3 .

Te (eV)

56

30 28 26 24 22 20 18 16 14 12 10 8 6 4 2 0

T T

0

20

40

60

80

e1

e2

100 120 140 160 180 200

Time (

s)

Figure 4.12: A comparison between electron temperatures obtained during the breakdown process calculated by the two methods described. Te1 , is estimated using the difference between plasma and floating potential. Te2 , is estimated by measuring the slope in a semilog I-V curve. The error bar in temperature measurement is estimated to be below 10 %, reaching 1 eV at high values during the breakdown transient and 0.25 eV during steady state.

6x1016 5x10

n n

16

n

4x10

16

n

-3

ne (m )

n

3x1016

n

e1

e2

e3

e4

e5

e6

2x1016 1x1016 0

0

20

40

60

80

100 120 140 160 180 200

Time (

s)

Figure 4.13: A comparison between density values calculated for the breakdown process. Densities ne1 and ne2 were obtained from the difference between floating potential and plasma potential with linear and parabolic fittings on the ion current branch. Densities ne3 , ne4 , ne5 and ne6 were obtained from the I-V curve using two plasma potential approximations with both linear and parabolic fittings in each case. Accuracy in electron density is estimated to be about 1 x 1016 m−3 .


57

In the following section, we present our results on the basis of calculations made using the method of section 4.3.2 with a Laframboise ion current fitting and the plasma potential calculated as the crossing point of the two linear fittings in the semilog I-V curve.

4.4

Results

A study of the breakdown process was conducted, taking measurements every 1 μs during breakdown followed by points every 10 and 20 μs on the microwave flat top, in order to obtain the temporal evolution of electron temperature and density. Fig. 4.14 shows a typical evolution of measured parameters for Hydrogen pressure of 3.8 x 10−3 mb, 1500 W peak microwave incoming power and 90 % duty cycle. The upper window shows the evolution of electron temperature and density during the breakdown process until flat top microwave excitation pulse is reached. The lower window shows the evolution of the absorbed/incoming power ratio (Pi − Pr )/Pi on the same time-base. It shows that at the very beginning of the pulse, the electric field inside the chamber is much higher than later on, during steady-state, when the plasma damps the microwave and the electric field drops. This behavior reflects the interplay between the electric field strength and the plasma load of the cavity during the ignition transient [22], showing microwave coupling time as directly the width of the (Pi −Pr )/Pi pulse. This E-field drop and its influence on plasma parameter evolution will be further discussed on chapter 6. It is notable that in the upper window, a temperature peak reaching almost 20 eV can be seen in coincidence with a drastic increment in absorbed power during the microwave coupling process. Such peak is followed by decreasing behavior that reaches about 5 eV as a final steady state temperature, remaining practically constant during the flat top microwave pulse. Electron density reaches stable values of about 1.5 x 1016 m−3 at the time when the temperature peak is produced, a fact that suggests that this process can be closely associated to plasma evolution during breakdown. It should be pointed out that the microwave coupling time shown in the Fig. 4.14 lower window was previously defined in chapter 3 (see Fig. 3.6). It is very interesting to notice how temperature peak and density increasing behavior occur during rapid increase of the absorbed power over incoming power ratio, i.e, in coincidence with the end of the MW coupling period. Systematic studies were carried out to measure plasma parameter evolution during breakdown under different working parameters. Conditions were set to be the same as in the experiments described in chapter 3 with the aim allowing conclusions to be drawn

58

MW coupling time

Figure 4.14: Upper window: Plasma temperature and density evolution during breakdown. Lower window: ratio of absorbed power and incoming power (Pi − Pr )/Pi showing the coupling time definition as the width of the pulse drop.

about the relationship between breakdown times and the evolution of plasma electron density and temperature evolution during this working period. Therefore, asymmetric Bz > ECR and symmetric Bz ' ECR magnetic field profiles were studied, both with the two different pressure level regimes previously used in chapter 3: a low pressure regime of 3.8 x 10−3 mb and a high pressure regime of 6.2 x 10−3 mb. As stated in section 4.1.2, the magnetic field profile below ECR has been excluded from this study due to its limited range of working parameters and high jitter level.

4.4.1 4.4.1.1

Temperature and Density Evolution with Asymmetric Bz > ECR Low Pressure Regime (3.8 x 10−3 mb)

Fig. 4.15 shows the temperature and density temporal evolution for three cases with Hydrogen pressure maintained at 3.8 x 10 −3 mb and the asymmetric over ECR magnetic field profile as a function of duty cycle and incoming power. Fig. 4.15 (a) is


59

a duty cycle scan at 1500 W of power, Fig. 4.15 (b) is a duty cycle scan at 900 W of power and Fig. 4.15 (c) is a power scan at 30 % of constant duty cycle.

(a) 1500W

(b) 900W

(c) 30% DC

Figure 4.15: Temperature and density evolution during the breakdown process for a Hydrogen pressure of 3.8 x 10−3 mb and Bz > ECR magnetic field. (a) Constant power at 1500 W. (b) Constant power at 900 W. (c) Constant duty cycle at 30 %.

In all of the cases, the beginning of the temperature and density pulses are coincident with the (Pi − Pr )/Pi pulse fast increase as shown in Fig. 4.14. As stated in the previous chapter, the microwave coupling time can be estimated by measuring the width of this (Pi − Pr )/Pi pulse. To make it easier to analyze the data and to establish connections between breakdown times, plasma density and plasma temperature evolution during breakdown, Fig. 4.16 plots the MW coupling time surface already shown in chapter 3 (see Fig. 3.7), where the level curves of 1500 W (a), 900 W (b) and 30 % of duty cycle (c) are marked as dashed lines. They correspond to the measurements represented in Fig. 4.15 (a), Fig. 4.15 (b) and Fig. 4.15 (c) respectively. It is interesting to compare Fig. 4.15 with Fig. 4.16 because a connection can be found between microwave coupling time and plasma parameter evolution. For example,

60

when incoming power is kept constant at 1500 W as indicated in level curve (a) in Fig. 4.16, practically constant microwave coupling times are observed with values of around of 35 μs for the whole range of duty cycles studied. The corresponding plasma parameter evolution in Fig. 4.15 (a) also shows non-relevant differences throughout the studied range. The temperature and density evolution suggests that this case is characterized by sharp temperature peaks of 16-18 eV followed by decays to the final temperature steady state at around 5 eV.

(b)

(c)

(a) Figure 4.16: Microwave coupling time as a function of incoming power and duty cycle for Bz > ECR and low pressure regime (3.8 x 10−3 mb).

However, the case of constant power at 900 W that is shown as level curve (b) in Fig. 4.16 shows a remarkable tendency for the coupling time to increase when duty cycle is also increased, starting at values of 40 μs at 10 % of duty cycle and reaching 120 μs at 90 %. The corresponding plasma parameter evolution in Fig. 4.15 (b) shows that plasma temperature evolves by decreasing temperature peak values and increasing final temperatures at plasma steady state. In this case, the relationship between microwave coupling times and temperature evolution indicates that for longer coupling times, longer and lower temperature pulses are observed with slightly higher final temperatures. The last case, where duty cycle is kept constant at 30 %, is shown as the level curve (c) in Fig. 4.16. For power ranging between 300 W and 1500 W, microwave coupling times show a tendency to rise when power decreases and the corresponding plasma parameter evolution in Fig. 4.15 (c) shows how temperature pulse decreases with power and practically vanishes at low powers. On the other hand, electron density shows a general behavior characterized by almost no dependence of duty cycle and a slight tendency to grow with incoming power. Higher values reach 2 x 1016 m−3 at 1500 W and lower values of 0.8 x 1016 m−3 are registered at 300 W.


4.4.1.2

61


The same process as in the previous section was conducted for the higher Hydrogen working pressure. Fig. 4.17 shows three cases of temperature and density evolution for different values of incoming power and duty cycles, maintaining Hydrogen pressure at 6.2 x 10−3 mb and using the Bz > ECR magnetic field profile.

(a) 1500W

(b) 900W

(c) 50% DC

Figure 4.17: Temperature and density evolution during the breakdown process for a Hydrogen pressure of 6.2 x 10−3 mb and Bz > ECR magnetic field. (a) Constant power at 1500 W. (b) Constant power at 900 W. (c) Constant duty cycle at 50 %.

Fig. 4.18 shows the microwave coupling times obtained for this experimental condition keeping the same representation scale as in Fig. 4.16 to facilitate comparison. The cases shown in Fig. 4.17 are represented on the surface as level curves (a), (b) and (c) for 1500 W, 900 W and 50 % of duty cycle respectively. These data show that temperature peaks reach lower values compared to the previous low pressure regime and the value of this peaks is more sensitive the incoming power level than in previous case, almost disappearing for low incoming power cases. Especially interesting is the the case of Fig. 4.17 (a), where the influence of duty cycle is remarkable for our maximum power input value under study. In addition,

62

electron densities show higher values reaching 2 x 10−16 m−3 , following the relative increment for this pressure regime.

(c)

(b)

(a) Figure 4.18: Microwave coupling time as a function of incoming power and duty cycle for the high pressure regime (6.2 x 10−3 mb).

4.4.2

Temperature and Density Evolution with Symmetric Bz ' ECR

The study described in the previous section was repeated under Bz ' ECR magnetic field profile conditions. Both low and high pressure regimes were studied and are detailed in this section. It needs to be point out that under this working conditions the plasma presented non negligible jitter during the early breakdown. Measurements acquired under jitter conditions were disregarded and, for this reason, some of the data sets of Fig. 4.16 are incomplete in the first breakdown instants. However, despite the lack of this points the curves still suggest the presence of the temperature peak and they were plot for completeness.

4.4.2.1


Fig. 4.19 shows plasma parameter evolution for different power and duty cycle working conditions. It is notable that both density and temperature show a significant increase in their steady state values, reaching, during the flat top of the pulse, values between 8 and 10 eV for temperature and between 2 x 1016 and 3 x 1016 m−3 for density.


63

Following the format used in previous cases, Fig. 4.20 represents breakdown times corresponding to these experimental conditions and superimposed broken lines (a), (b) and (c) correspond to Figs. 4.19 (a), 4.19 (b) and 4.19 (c) respectively.

(a) 1500W

(c) 50% DC

(b) 900W

Figure 4.19: Temperature and density evolution during the breakdown process for a Hydrogen pressure of 3.8 x 10−3 mb and Bz ' ECR magnetic field. (a) Constant power at 1500 W. (b) Constant power at 900 W. (c) Constant duty cycle at 30 %.

(c)

(b)

(a)

Figure 4.20: Microwave Coupling time as a function of incoming power and duty cycle for Bz ' ECR magnetic field profile and low pressure regime (3.8 x 10−3 mb).

64

Measurements corresponding to 4.19 (a) show fast, narrow temperature peaks reaching up to 28 eV, while density shows a peak that has not been registered in any of the previous cases. Neither of them seems to be dependent on duty cycle, following the tendency of MW coupling times, which do not present any significant changes along Fig. 4.20 curve (a). Equally notable are the high density values in the flat top of the pulse, reaching 3 x 1016 m−3 in the lower duty cycle cases. For the case of 4.19 (b), the power decrease results in a decrease in the temperature peak value and in the steady state density that in every case remains below 2 x 1016 m−3 . Finally, Fig. 4.19 (c) shows the evolution of plasma parameters along line (c) in Fig. 4.20. Note that temperature peaks are wider for low powers following the increment in MW coupling time shown in Fig. 4.20 along curve (c). The density peak is also noticeable in all cases, reducing its height as power is reduced, and nearly disappearing for the 300 W case.

4.4.2.2


Fig. 4.21 shows the plasma parameter evolution under Bz ' ECR and high pressure working conditions for (a) constant power at 1500 W, (b) constant power at 900 W and (c) constant duty cycle at 30 %. Fig. 4.22 shows, with the same color scale used in previous cases, the MW coupling times measured under the same working conditions. Broken lines (a), (b) and (c) mark the paths corresponding with 4.21 (a), 4.21 (b) and 4.21 (c) respectively. With respect to the previous case, it is important to point out that both temperature and density have suffered an significant drop in steady state values: temperature during flat top is around 4 eV while density is between 1.5 x 1016 and 2 x 1016 m−3 . Temperature peaks have also reduced their values to reach maximums of 12 eV. Fig. 4.21 (a) shows a decrease in the temperature peak maximum, going from 12 eV in the case of the 90 % duty cycle to 8 eV in the 10 % duty cycle case. This corresponds to the decrease in MW coupling time that can be seen in Fig. 4.22 following curve (a) in the decreasing duty cycle direction. Peak width and density evolution do not evolve at all with duty cycle, while steady state values for temperature are lower for low duty cycles. The case of the 900 W constant power study, corresponding to 4.21 (b) and curve (b) in Fig. 4.22, shows smaller temperature peaks and presents similar behavior: when duty cycle is decreased, peaks lower their level and nearly disappear for low duty cycles, while density reflects no changes and steady state values for temperature reduce their levels.


(a) 1500W

65

(c) 50% DC

(b) 900W

Figure 4.21: Temperature and density evolution during breakdown process for a Hydrogen pressure of 6.2 x 10−3 mb and Bz ' ECR magnetic field. (a) Constant power at 1500 W. (b) Constant power at 900 W. (c) Constant duty cycle at 30 %.

(c)

(b)

(a) Figure 4.22: Microwave Coupling time as function of incoming power and duty cycle for Bz ' ECR magnetic field profile and low pressure regime (6.8 x 10−3 mb).

66

Finally, the constant 50 % duty cycle study shown in 4.21 (c) shows a decrease on temperature peaks when power is reduced. This occurs, once again, in coincidence with an increment in MW coupling times (curve (c) of Fig. 4.22). Steady state temperature values present almost no changes while density reflects a reduction in its steady state value when power is decreased. It is noticeable that a density peak can also be observed under this working conditions in those cases of high incoming power and high duty cycle, However, this peak does not present any remarkable evolution when the incoming power level or the duty cycle are changed.

4.5


To summarize, table 4.1, shows the main values for plasma electron density and temperature as a function of the magnetic field profile and the gas pressure in the chamber. Pressure x 10−3 (mb)

Te peak (eV)

Te flat top

Ne x 1016

(eV)

(m−3 )

B>ECR(' 120 mT in chamber axis) 3.8

16 − 18

6−7

1−2

6.2

8 − 12

4−5

1−2

B'ECR(≈ 87.5 mT in chamber axis) 3.8

24 − 16

8 − 10

2−3

6.2

18 − 12

5−7

1−2

Table 4.1: Summary of plasma density and temperature evolution during breakdown.

In this chapter, a plasma diagnostic based on Langmuir probe measurements has been presented and it has been shown to be a useful tool for the study of the temporal evolution of plasma parameters (electron density and temperature). Based on the results of the breakdown study described in previous chapter, systematic measurements of plasma parameters were carried out under a wide range of working parameters: Hydrogen pressure, MW power and magnetic field distribution. Different kinds of plasma parameter evolution were observed depending on the incoming power and duty cycle in connection with microwave coupling times. Under


67

relatively high power conditions, where short microwave coupling times were recorded, high temperature peaks were observed during the microwave coupling process at very beginning of plasma breakdown. This behavior is in line with the idea of a breakdown process divided into the two stages described in chapter 3: microwave coupling, characterized by a fast heating peak, followed by a plasma formation time where plasma parameters reach their final, stable temperature and density. However, for low incoming powers, temperature evolution shows a tendency to reduce its peak values, evolving gradually towards a higher final temperature without peaking. An explanation of this behavior may be found in the influence of seed electrons already mentioned in chapter 3 section 3.4 [25, 43]. Immediately after each microwave pulse is extinguished, a residual electron density remains embedded in the neutral gas during sufficient time to produce an effect on the breakdown of the pulse that follows. If incoming power is high enough, residual electrons may produce an electron density at the beginning of pulses that is also high enough to guarantee a fast coupling. Under these circumstances, saturation of fast coupling times with respect to duty cycle may occur as can be seen in the curve (a) of Fig. 4.16. This effect could be related to the electron temperature peaks described in this chapter and it may be a determining factor for the microwave coupling process at the early breakdown stage [44]. However the increase in breakdown times for high duty cycles at lower incoming powers remains to contradict this explanation. Further research on this point is included in the scope of our future work. Moreover, the early instant, when temperature peaks were observed in our experiment, also suggests a relationship between these observations and preglow processes during breakdown in ECRIS plasmas [12, 45, 46]. Equally notable from the preglow point of view are the breakdown density peaks registered in cases where the magnetic field is the symmetric Bz ' ECR. Finally, it is interesting to consider the impact that this phenomenon can have on plasma ionization dynamics. A recent publication by Y. Xu et al. [3] records a + peak of cluster ions H+ 2 and H3 during plasma breakdown. The authors point to the temperature peak reported here as a possible cause of this cluster ion peak. In Hydrogen plasma several reactions take place which produce or destroy the different ions. H+ 2 ions + are created by a direct ionization process (4.5) while H3 is produced by the dissociative attachment reaction 4.6. Meanwhile, protons are produced mainly by two multiple collision processes in 4.7 and 4.8.

e + H2 −→ H2+ + 2e

(4.5)

68

H2+ + H2 −→ H3+ + H

H2 + e −→ 2H + e H + e −→ H + + 2e

(4.6)

(4.7)

H2 + e −→ H2+ + 2e H2+ + e −→ H + + H + e

(4.8)

or H2+ + e −→ 2H + + 2e Besides these production processes, various ion destruction mechanisms take place in the ion source at the same time. Fig. 4.23 shows together the cross sections of the main processes for ion generation and destruction together.

Figure 4.23: Cross section for main physical processes in a Hydrogen ion source.

The cluster ion peak reported by Y. Xu et al. occurs during plasma breakdown and could be related the electron temperature peak measured and described in this chapter. The cross section diagram shows a narrow area of particle energy marked between broken lines. This narrow gap represents the energy conditions when H+ 2 is generated but not destroyed, between 15 and 20 eV approximately. It is possible that in some cases the


69

temperature peak coincides with this energy range (temperatures between 10 and 13 eV) enhancing the production H+ 2 . However, the peak of cluster ions could also be due to the fact that during breakdown the protons, in which formation process cluster ions are destroyed, have not been produced yet.

Chapter 5

Plasma Density and Temperature Measurements during Decay From the breakdown study described in chapter 4 arose the idea of exploring the decay process as a means to provide a view of plasma evolution along the whole pulse. We used the same set-up, measuring strategy and Langmuir probe analysis method as in chapter 4 to study the evolution during decay. The range of working parameters were also set to be the same to make the results fully comparable. Measurements of Langmuir probe curves were taken with 2 μs starting during the plasma steady state and following its decay. Special attention was paid to the analysis of the low intensity Langmuir probe curves corresponding to the late instants of the decay process. The results of the decay study are presented in this chapter following the same structure already used in chapters 3 and 4. This study, in conjunction with the breakdown study described in chapter 4, provides a characterization of plasma parameter evolution along the whole microwave pulse.

71

72

5.1

Experimental Conditions

The experimental conditions used previously in chapters 3 and 4 for Hydrogen pressure and magnetic field intensity in the breakdown study were also set for these new experiments. Magnetic field configurations asymmetric Bz > ECR and symmetric Bz ' ECR were explored (see Figs. 4.5 and 4.6 in chapter 4) and plasma behavior during decay was studied for different magnetron powers. The influence of duty cycle was carefully checked and no changes were detected when it was either increased or decreased. Duty cycle was thus fixed at 10 % and was not changed throughout the study. During the first tests it was detected that the magnetron produced jitter at MW pulse shut-off when working at 50 Hz, this was solved by increasing the frequency to 100 Hz.

Figure 5.1: View of a synchronism record during the decay process: (a) is the incoming power, (b) is the reflected power (amplified ten times with respect to incoming power) and (c) is the synchronism pulse, where the rising edge indicates the instant of measuring.

As in the study of the plasma parameter evolution during breakdown, incoming and reflected power measurements were recorded together with the trigger synchronization signal in order to know the precise moment when the Langmuir probe I-V curve was recorded. An example of a typical timing record during the study of the decay transient is shown in 5.1, where signal (a) is the incoming power, signal (b) is the reflected power and signal (c) is the synchronism pulse from the probe driver system. The front of this pulse represents the exact instant when the measurement of the I-V curve takes place. Note that in this case, signal (b) is amplified ten times with respect to signal (a). This is because the reflected power is low during the pulse body and it is necessary to increase the amplification in order to see the structure in the reflected power.

Chapter 5. Plasma Density and Temperature Measurements during Decay

5.2

73

Data Analysis and Calculations

Typical I-V curves obtained during experiments are shown in Fig. 5.2. Case (a) is a curve obtained 100 μs before the incoming microwave pulse is off, where the plasma can be considered to be in steady state, and case (b) is a curve obtained during the decay transient. Note that the current intensities recorded in case (b) are noisier and ten times lower than in case (a) in line with the density drop during decay.

(a)

(b)

Figure 5.2: Langmuir probe I-V curves obtained during the experiments. Curve (a): typical case during the steady-state plasma condition (100 μs before incoming microwave pulse is off). Curve (b): typical case corresponding to the decay stage interval of time.

Due to these lower and noisier curves obtained during decay, special attention was paid to the analysis of low intensity curves measured in the final instants of the decay process. The same Langmuir probe analysis methods described in section 4.3 of chapter 4 were used with excellent results to analyze a full set of curves. Following the same notation and representation scheme of chapter 4, the results for both plasma electron temperature and density are shown in Figs. 5.3 and 5.4, respectively. In the following section we present the decay study results based on calculations made using the method of section 4.3.2, where the electron temperature is obtained from the slope of the linear fitting on the current logarithm-voltage, the ion saturation current is assumed to follow Laframboise’s hypothesis and the plasma potential is considered the crossing point of the two linear fittings on the semilog I-V curve.

74

Figure 5.3: A comparison of the electron temperatures obtained during the decay process calculated by two methods. Te1 , is estimated from the difference between plasma and floating potential. Te2 , is estimated by measuring the slope in a semilog I-V curve.

4x1016 n n

3x1016

n

-3

ne (m )

n n

2x1016

n

e1

e2

e3

e4

e5

e6

1x1016

0

0

5

10

15

20 Time (

25

30

35

40

45

s)

Figure 5.4: Densities ne1 and ne2 were obtained from the difference between floating potential and plasma potential either by linear or parabolic fittings on ion current branch, respectively. Densities ne3 , ne4 , ne5 and ne6 were obtained from the slope of log(I) − V curve for the two plasma potential values with both linear and parabolic fittings.


5.3

75

Results

As was done for the breakdown transient study in chapter 4, time resolved electron density and temperature evolution were measured in our plasma source. The incoming and reflected microwave power signals were recorded simultaneously. Table 5.1 shows the range of parameters within which experiments were conducted.

Parameter Magnetic Field Profile Hydrogen Pressure (mb) Peak MW Power (W)

Values Bz > ECR and Bz ' ECR 3.8 and 6.2 x 10−3 300 to 1500

Duty Cycle (%)

10 %

Magnetron Pulse Frequency (Hz)

100

Table 5.1: Parameters Used during Experiments

In the following subsections, results are shown using the same structure already used in previous chapters 3 and 4. As before, the B-field denominated Bz > ECR corresponds with the asymmetric magnetic field distribution of Fig. 4.5 (a) and 4.6 (a) while the Bz ' ECR corresponds with the symmetric magnetic field distribution of 4.5 (b) and 4.6 (b). In each set of data, the incoming power varies from 300 W to 1500 W. Every graph shows five data sets following the same representation as in previous chapters: (a) is the incoming power represented by a solid yellow line, (b) is the reflected power represented by a solid blue line, (c) is the electron temperature marked with triangular points and (d) is the electron density represented by empty dots. Curves (a) and (b) are in arbitrary units but (b) is amplified ten times with respect to (a) in order to see the structure of the reflected power.

76

5.3.1

Temperature and Density Evolution with Asymmetric Bz > ECR Low Pressure Regime (3.8 x 10−3 mb)

5.3.1.1

Fig. 5.5 shows a sequence with constant Hydrogen pressure of 3.8 x 10−3 mb. 5

14

4

2

4

5

10

15

20

25

30

Time (

35

40

45

0

0 50

5

0

5

10

15

20

25

900 W

30

35

40

45

5

12

4

0 50

s)

14 600 W

4

6

1

2 0

5

10

15

20

25

30

Time (

35

40

45

2

4 1

2 0

0 50

Ne x 10

2

4

16

Te (eV)

16

Ne x 10

6

3

8

-3

-3

3

8

(m )

10 (m )

10 Te (eV)

1

2

Time (

12

0

2

4

s)

14

-3 16

6

1

2

3

8

Ne x 10

16

6

0

4 (m )

-3

3

8

0

1200 W

10

Ne x 10

Te (eV)

1500 W

(c) (d)

12

Te (eV)

(a) (b)

10

5

14

(m )

12

0

5

10

15

20

s)

25

Time (

30

35

40

45

0 50

s)

5

14 12

300 W

4

16

-3

(m )

3

8 6

2

4 1

2 0

Ne x 10

Te (eV)

10

0

5

10

15

20

25

Time (

30

35

40

45

0 50

s)

Figure 5.5: Evolution of plasma parameters and the incoming and reflected power signals for low pressure (3.8 x 10−3 mb) and for the Bz > ECR magnetic field profiles used in the experiments: (a) incoming power from magnetron in a.u., (b) reflected power from plasma in a.u. amplified ten times with respect to incoming power, (c) electron temperature and (d) electron density. The error bar in temperature measurement is estimated at below 10 % and the accuracy in electron density is estimated at about 1 x 1016 m−3 .


77


5.3.1.2


14

12

4

2

4

5

10

15

20

25

30

Time (

35

40

45

0

0 50

5

0

5

10

15

20

25

900 W

30

35

40

45

5

12

4

0 50

s)

14 600 W

4

6

1

2 0

5

10

15

20

25

30

Time (

35

40

45

2

4 1

2 0

0 50

Ne x 10

2

4

16

Te (eV)

16

Ne x 10

6

3

8

-3

-3

3

8

(m )

10 (m )

10 Te (eV)

1

2

Time (

12

0

2

4

s)

14

-3 16

6

1

2

3

8

Ne x 10

16

6

0

4 (m )

-3

3

8

0

1200 W

10

Ne x 10

Te (eV)

10

1500 W

(c) (d)

Te (eV)

(a) (b)

(m )

12

5

14

0

5

10

15

20

s)

25

Time (

30

35

40

45

0 50

s)

5

14 12

300 W

4

16

-3

(m )

3

8 6

2

4 1

2 0

Ne x 10

Te (eV)

10

0

5

10

15

20

25

Time (

30

35

40

45

0 50

s)

Figure 5.6: Evolution of plasma parameters and the incoming and reflected power signals for high pressure (6.2 x 10−3 mb) and for the Bz > ECR magnetic field profiles used in the experiments: (a) incoming power from magnetron in a.u., (b) reflected power from plasma in a.u. amplified ten times with respect to incoming power, (c) electron temperature and (d) electron density. The error bar in temperature measurement is estimated at below 10 % and the accuracy in electron density is estimated at about 1 x 1016 m−3 .

78

It should be noted that, for both cases, a pulse of reflected power appears immediately after incoming power starts to decay showing a peak during the final part. After this, an unexpected second pulse or shoulder is observed while the incoming power continues its decay without any alteration. Temperature evolution starts at typical values of 8-12 eV corresponding to the steady state of the plasma during the flat top for the power pulses. When the incoming power comes down as it starts its decay process, temperature initially follows the decay, but it shows a rebound or peak in coincidence with the second pulse of the reflected power signal just described. Note that the temperature decay process closely follows the reflected power signal evolution. In contrast, the plasma density evolution behaves more in line with expectations, closely following the incoming power signal decay process without extending decay time. The temperature rebound is more visible in the higher pressure case, generally in coincidence with the second shoulder in the reflected power signal.

5.3.2

Temperature and Density Evolution with Symmetric Bz ' ECR

The measurements described in the previous section were repeated for the symmetric Bz ' ECR magnetic field. Both low and high pressure regimes were explored and their results are shown in this section. The arrangement of graphics follows the same pattern as that of the previous Figs. 5.5 and 5.6 with the same symbols and line styles formats to facilitate comparison. In general terms, plasma coupling shows a better behavior for the asymmetric case of Bz > ECR of the previous section than for Bz ' ECR in this section. The level of reflected power is initially higher in this case, but the coupling improves for higher incoming power. It is noticeable that reflected power drops during the beginning of the incoming power decay but then rebounds, showing the first narrow peak mentioned earlier. A second pulse or shoulder is again recorded while the incoming power continues its decay. Density behavior stays closely in line with incoming power decay as in previous cases, showing practically no influence from the structure found in the microwave reflected power.


79


5.3.2.1


14

12

4

2

4

5

10

15

20

25

30

Time (

35

40

45

0

0 50

5

0

5

10

15

20

25

900 W

30

35

40

45

5

12

4

0 50

s)

14 600 W

4

6

1

2 0

5

10

15

20

25

30

Time (

35

40

45

2

4 1

2 0

0 50

Ne x 10

2

4

16

Te (eV)

16

Ne x 10

6

3

8

-3

-3

3

8

(m )

10 (m )

10 Te (eV)

1

2

Time (

12

0

2

4

s)

14

-3 16

6

1

2

3

8

Ne x 10

16

6

0

4 (m )

-3

3

8

0

1200 W

10

Ne x 10

Te (eV)

10

1500 W

(c) (d)

Te (eV)

(a) (b)

(m )

12

5

14

0

5

10

15

20

s)

25

Time (

30

35

40

45

0 50

s)

5

14 12

300 W

4

16

-3

(m )

3

8 6

2

4 1

2 0

Ne x 10

Te (eV)

10

0

5

10

15

20

25

Time (

30

35

40

45

0 50

s)

Figure 5.7: Evolution of plasma parameters and the incoming and reflected power signals for low pressure (3.8 x 10−3 mb) and for the Bz ' ECR magnetic field profiles used in the experiments: (a) incoming power from magnetron in a.u., (b) reflected power from plasma in a.u. amplified ten times with respect to incoming power, (c) electron temperature and (d) electron density. The error bar in temperature measurement is estimated at below 10 % and the accuracy in electron density is estimated at about 1 x 1016 m−3 .

80


5.3.2.2


14 1500 W

12

4

2

4

5

10

15

20

25

30

Time (

35

40

45

0

0 50

5

0

5

10

15

20

25

900 W

30

35

40

45

5

12

4

0 50

s)

14 600 W

4

6

1

2 0

5

10

15

20

25

30

Time (

35

40

45

2

4 1

2 0

0 50

Ne x 10

2

4

16

Te (eV)

16

Ne x 10

6

3

8

-3

-3

3

8

(m )

10 (m )

10 Te (eV)

1

2

Time (

12

0

2

4

s)

14

-3 16

6

1

2

3

8

Ne x 10

16

6

0

4 (m )

-3

3

8

0

1200 W

10

Ne x 10

Te (eV)

10

(c) (d)

Te (eV)

(a) (b)

(m )

12

5

14

0

5

10

15

20

s)

25

Time (

30

35

40

45

0 50

s)

5

14 12

300 W

4

16

-3

(m )

3

8 6

2

4 1

2 0

Ne x 10

Te (eV)

10

0

5

10

15

20

25

Time (

30

35

40

45

0 50

s)

Figure 5.8: Evolution of plasma parameters and the incoming and reflected power signals for high pressure (6.2 x 10−3 mb) and for the Bz ' ECR magnetic field profiles used in the experiments: (a) incoming power from magnetron in a.u., (b) reflected power from plasma in a.u. amplified ten times with respect to incoming power, (c) electron temperature and (d) electron density. The error bar in temperature measurement is estimated at below 10 % and the accuracy in electron density is estimated at about 1 x 1016 m−3 .


5.4

81


The study of plasma parameters evolution during decay set out in this chapter completes a panoramic view of the whole plasma pulse. Following the structure of previous chapters 3 and 4, two Hydrogen working pressures at two different magnetic field profiles were studied for a range between 300 W and 1500 W of peak incoming microwave power. Figs. 5.5, 5.6, 5.7 and 5.8 show typical records where the evolution of plasma density and temperature with the incoming and reflected microwave power as a temporal reference can be seen. A remarkable structure is observed in the reflected power signal, while incoming power decay shows exponential behavior without any remarkable alteration. The reflected power signal presents a peak just after the MW power shut-off and a second smaller peak approximately 10 μs later. This structure was observed within the working parameter range of our experiments with a very high degree of reproducibility. Experiments with several different duty cycles from 10 % to 90 % did not show any significant differences. The first pulse typically has a duration of 5 μs and it is usually higher than the second one, which has a typical duration of 15 μs. It is interesting to note that the first pulse takes place during the last part of the incoming microwave power decay, when its value has decreased by approximately 70 %. Meanwhile, the second one is produced while the incoming power is almost negligible, being in most cases lower than 10 % of its initial value. Simultaneously, electron temperature evolution shows an interesting rebound in coincidence with the second pulse of the plasma reflected microwave emission. In contrast, density evolution always stays in close line with the incoming power decay, with no structures and without the extended decay time shown by temperature. Temperature peaks show a relative increment at the Bz ' ECR magnetic field distribution, while the Bz > ECR profile structure shows shorter decay times, reaching only half the values of the ECR case. This behavior was recorder for both studied pressures. The initial level of the incoming power signal does not seem to have a strong effect on the evolution. However, it was observed that in those cases where the coupling ratio between incoming and reflected power before decay is better, the structure in the reflected power emission is more noticeable and temperature peaks show an slightly incremental tendency. The data showed that the main factor influencing on the general behavior is the incoming and reflected power ratio from the plasma. This fact strongly suggests that final values for the plasma parameters before shut-down and their interplay

82

with the incoming microwave power during this kind of decoupling process may be decisive for the ulterior dynamics of the plasma parameter evolution. Temperature rebounds could be explained in terms of energy distribution within a collective of particles (electrons) whose population is descending. Once the incoming power starts to fall, it is reasonable to suppose a decrease in plasma density from the steady-state value due to the ionization rate drop. This gradual decrease in density is shown by our measurements in Figs. 5.5, 5.6, 5.7 and 5.8. Such behavior could produce a relative increment in the reflected power because coupling became worse and worse until it reached a situation where the rapid loss of electrons during this stage produced a profound change in the EEDF. It also seems that the influence of plasma parameter evolution may well have a determining influence on the power absorbed during plasma decay. Further work on calculating EEDFs based on experimental probe data to explain these observations is one of the aims of our group for the near future.

Chapter 6

Plasma Vacuum Ultraviolet Emission during Breakdown It is well known that Langmuir probe results are somewhat sensitive to the chosen analysis method, as shown in chapter 4 section 4.3 with regard to the different curve analysis methods. The Langmuir probes used for measurements can also be influenced by the external magnetic field and by the exact nature of the electron energy distribution as discussed thoroughly by Chen et al. in Ref. [41]. For this reason, we used a complementary diagnostics set-up described in this chapter to confirm the existence of the electron temperature peak previously reported in chapter 4. This complimentary diagnostics involves the measurement of Vacuum Ultraviolet (VUV) emission, in a range of 100-250 nm, from the plasma during transients. VUV-light emission is a consequence of the interaction between energetic (> 10 eV) electrons and neutrals in excitation reactions. The VUV-emission signal can also be considered as a measure of the ionization rate because the ratio of the excitation and ionization cross sections for Hydrogen is well-defined [47]. In the narrative that follows we focus a single experimental condition to concentrate on the interpretation of the data. However, additional data obtained through the VUV-diagnostics is also presented to demonstrate that the conclusions are valid over a wide range of experimental conditions. To put this research into a wider context, it is significant that the transient effects observed took place on a μs-scale and are thus interpreted as originating from the temporal evolution of the electron energy distribution, rather than the neutral gas heating and subsequent depletion that is often observed in various plasma generators as discussed first by Boswell [48].

83

84

The experiments described in this chapter were carried out in collaboration with the Ion Source Group at Jyv¨ askyl¨ a University (Finland). The Finnish group brought the VUV spectrometer to our installations and a member of their team participated in the five-weeks measuring campaign during which all the measurements recorded in this chapter were carried out.

6.1 6.1.1

Experimental Set-Up Diagnostic Port

(a)

(c)

(b)

(d)

Figure 6.1: Section view of the combined VUV and Langmuir probe plasma diagnostics installed in the plasma reactor: diagnostics port (a), pumping connection and line-of-sight for optical diagnostics (b), tilted Langmuir probe (c) and observation window (d).

A section view of TIPS is shown in Fig. 6.1 to present the VUV diagnostics set-up. The diagnostic port (a) was designed to leave a line-of-sight path through the pumping line (b). To be able to take Langmuir probe measurements in the center of the plasma chamber, a tilted Langmuir probe (c) was installed. The probe is sustained by an Ultratorr fitting placed in a tilted flange and it is possible to retrofit the probe, without breaking the vacuum, to remove the tip from the line-of-sight. Observation window (d) was kept in the design. 2 mm thick Boron Nitride disks on both sides of the plasma chamber were also kept in the design although they are not shown in the figure for the sake of clarity. The disk on the diagnostic port side was machined to fit the new port geometry.

Chapter 6. Plasma Vacuum Ultraviolet Emission during Breakdown

85

A photograph of the experimental set-up is shown in Fig. 6.2, where the VUV spectrometer can be seen on the left side and the plasma source on the right.

Figure 6.2: Photograph of the experimental set-up. The vacuum ultraviolet spectrometer on the left is connected to the plasma generator on the right.

The main advantage of VUV-diagnostics over the less complicated visible ones is that the excitation cross sections to the first excited states of atoms and molecules yielded in VUV emissions are an order of magnitude higher than the excitation cross sections to higher states e.g. Balmer series and Fulcher band [47]. Moreover, the excitation energy is lower, which means that the excitation rate is less sensitive to changes in the tail of the electron energy distribution function. This is especially important when studying transient effects. Measuring molecular emission within the Lyman band (∼ 161 nm) is a good diagnostic tool for excitation rate because these transitions do not suffer significantly from cascade effects from higher excited states [49]. The Langmuir probe was used to acquire I-V curves using the same procedure described in previous chapters. The probe tip was again made of 6 mm long, 0.5 mm diameter tungsten wire. Transient effects were studied by synchronizing the probe with the VUV-emission measurement via a delay generator.

86

6.1.2

Magnetic Field Profile

The magnetic field profile used in the experiments described in this chapter is the asymmetric over ECR one previously used in chapters 3, 4 and 5, which is very representative of the rest of the working conditions. For convenience, Fig. 6.3 shows again the 2D cross-sectional view of the magnetic field profile. Keeping the representation patterns of previous chapters, the plasma chamber boundaries are represented by a rectangular shape with a solid black line where the left side corresponds to the microwave injection side, and the right side to the diagnostic port. A dashed black line indicates the position of the 87.5 mT ECR surface. The magnetic field at the Langmuir probe position (in the center of the chamber), is about 120 mT .

r z

mT 150 146 142 138 134 130 126 122 118 114 110 106 102 98 94 90 86 82 78 74 70

Figure 6.3: 2D representation of the magnetic field configuration used in the experiments. The plasma chamber is represented by a solid black line. The dashed black line indicates the position of the ECR surface (87.5 mT).

6.2


Fig. 6.4 shows a diagram of the diagnostics and timing system integrated within the plasma reactor. The timing signal for data acquisition, provided as in previous cases by the dual directional coupler, was used to trigger the VUV-diagnostics as well as the Langmuir probe via a delay generator. The positions of the tuner stubs were adjusted in order to obtain the minimum reflected power readout while keeping stable plasma and low jitter between the timing signals for consecutive microwave pulses (jitter was always kept below 0.1 μs).


87

The VUV spectrometer consists of a monochromator (McPherson, model 302) with a holographic grating, a photomultiplier tube (ET Enterprises, model 9406B) and a PC controlling the wavelength in the range of 100-250 nm. A transimpedance amplifier connected to an NI USB-6255 Data Acquisition System (DAS) card was used to record the photomultiplier current. The spectral resolution of the VUV spectrometer was set to 9 nm, with slits to increase light intensity at the PMT, while its temporal resolution was fixed at 2 μs. In order to reduce noise, 100 consecutive pulses were recorded and averaged, which means that data acquisition time was a few seconds. Dual directional coupler

Tuner

Vacuum UV Spectrometer



PMT

Transimpedance amplifier

OSCILLOSCOPE CH1 CH3

CH2

DAS Card

Langmiur Probe Controller

Spectrometer Controller

Synchrony pulses Monitor

COMPUTER


Figure 6.4: Schematic representation of the diagnostics set-up for measuring the temporal evolution of microwave-plasma coupling, vacuum ultraviolet light emission and plasma electron temperature and density.

The electron temperature Te was estimated by two of the methods developed in section 4.3 of chapter 4. The first method uses the slope of the logarithm I-V curve assuming a Maxwellian electron energy distribution function (EEDF) [35, 42] with the Laframboise hypothesis on the ion current fitting and the plasma potential assumed to be the crossing point of the two linear fittings in the semilog I-V curve. The second method takes the difference between plasma potential (assumed to be the maximum of the I-V curve first derivative) and floating potential i.e. Te = 0.26...0.35(Vp − Vf ), where Te is in units of eV [36, 39, 41]. The tilt of the probe made it impossible to estimate the coefficient precisely, so we have used a coefficient of 0.26 which corresponds to the

88

minimum (theoretical) value for Te and assumes perfectly Maxwellian EEDF instead of 0.35, as was assumed in section 4.3.1 of chapter 4. Deriving the error directly from the measured quantities is almost impossible since commonly accepted probe analysis techniques are not always supported by theory and include assumptions [35, 41, 42]. The shape of the curve is defined by the EEDF as well as the sheath expansion at high probe voltages [35, 36]. The orientation (tilt) of the probe with respect to the external magnetic field results in a slight underestimation of the saturation currents since the electron Larmor radius is smaller than the diameter of the probe tip. This causes the field lines intercepting the probe to become partially depleted due to the limited diffusion rate across the field [36]. In the experiments described here, the plasma breakdown transient was studied keeping the repetition rate (50 Hz) and duty cycle (50 %) fixed while varying the microwave power from 300 to 3000 W and the neutral Hydrogen pressure from 3.8 to 9.3 x 10−3 mbar. Probe data were taken at 1 μs intervals during the ignition transient, followed by a gradual increment to these intervals up to 20 μs after reaching the steady state.

6.3

Results

Fig. 6.5 shows a typical result for the temporal evolution of the measured parameters. The corresponding settings are 900 W incident power, 9.3 x 10−3 mbar Hydrogen pressure and the magnetic field profile is the Bz > ECR depicted in Fig. 6.3. Fig. 6.5 shows the evolution of the Pr /Pi ratio. The relative change of the averaged electric field in the plasma chamber can be estimated form the incoming and reflected power signals based on Ref. [50], where electron heating in the ECR zone is studied theoretically. Due to the long wavelength it can be assumed that the incident microwave power is either reflected or absorbed i.e. Pi = Pr + Pa . The absorbed power Pa consists of the fraction absorbed by the plasma and transmission losses, Ploss . Following Ref. [50] the ratio of the “effective electric field” to the incident electric field can be expressed as

v Eef f (t) u u (Pr +Ploss )/Pi− 1 =t Emax ln Pr +Ploss Pi

(6.1)


89

Figure 6.5: Time-resolved signals and plasma parameters during the breakdown of Hydrogen discharge in TIPS. (a) Pr /Pi and average electric field, (b) Lyman band and Lymanalpha VUV light emission, (c) floating and plasma potential, (d) electron temperature calculated by two different methods, (e) saturation electron current and the ratio between Lyman band emission signal and saturation electron current.

90

The ratio of reflected to incident power measured at the beginning of the microwave pulse was about 90 %, indicating that transmission losses account for 10 % of the incident power. The temporal evolution of the electric field calculated from Eq. 6.1 is also shown in Fig. 6.5 (a). The result suggests that the electric field drops to ∼ 65 % of the initial value due to plasma damping. The temporal evolution of VUV light emission at 160 nm, i.e. at the maximum of the molecular Lyman band extending from 125 to 161 nm, and at 121.6 nm, corresponding to the Lyman-alpha transition, is presented in Fig. 6.5 (b). Lyman band transitions 1 + B 1 Σ+ u → X Σg are caused by electron impact on excited neutral Hydrogen molecules (e + H2 → Lyman band). Since electron losses are dominated by diffusion, the rate of recombination reactions resulting in Lyman band emission is insignificant in comparison to electron impact excitation [51]. Lyman-alpha corresponds to the electronic transition of a Hydrogen atom from the first excited state to the ground state (n = 2 → n = 1). The transient peak for light emission, lasting some μs, is observed in coincidence with the drop in the electric field. After the ignition transient, the Lyman band and Lymanalpha emission intensities exhibit opposite trends i.e. molecular emission decreases while atomic emission increases slowly and saturate at ∼ 500 μs. Fig. 6.5 (c) shows the temporal evolution of the floating potential and the plasma potential calculated as the maximum of the I-V curve’s first derivative. It is evident from the data that both potentials fluctuate during the breakdown process which increases the uncertainty of the probe analysis in comparison to the steady state. Fig. 6.5 (d) shows the temporal evolution of electron temperature obtained with the two analysis methods already mentioned i.e. from the slope of the logarithm IV curve assuming Maxwellian electron energy distribution function (EEDF) with the Laframboise’s hypothesis on the ion current fitting and assuming the plasma potential to be the crossing point of the two linear fittings on the semilog I-V curve; and from the difference of plasma potential and floating potential assuming the plasma potential to be the maximum of the I-V curve’s first derivative. Taken together, the probe data show, as it was reported in chapter 4, the existence of a breakdown transient before the temperature saturates to about 4-6 eV at about 20 - 50 μs and remains at this level until the end of the microwave pulse. The temporal evolution of the electron saturation current, defined at Vprobe = Vp , following Laframboise’s hypothesis, is presented in Fig. 6.5 (d). The electron saturation current is proportional to the plasma electron density (according to Eq. 4.3) and is therefore interpreted to increase monotonically up to 20 μs, which is equal to the time it takes for the average electric field to collapse. Note that studying the evolution of the


91

electron saturation current instead of the plasma electron density calculated through Eq. 4.3 guarantees that the result is independent of the electron temperature evolution. The volumetric rate of ionizing collisions is proportional to the electron and neutral gas densities and the ionization rate. Thus, it is expected that the electron density first increases exponentially and then saturates to a level where the loss rate of electrons due to diffusion and recombination equals the ionization rate of the neutral gas, which agrees well with the data in Fig. 6.5 (e). It is concluded that the decrease in the Pr /Pi ratio and average electric field is caused by the presence of electrons absorbing the microwave power and damping the microwave electromagnetic field. A similar effect on the microwave-plasma coupling has been observed with high frequency ECR ion sources [22, 52] but has never been directly correlated with the temporal evolution of the plasma density.

Є

Є

Fig. 6.6 shows rates for ionization and excitation of H2 molecules as a function of the average electron energy for Maxwell-Boltzmann and Druyvesteyn EEDFs, which is considered more to be representative for plasmas where Te > Ti [53]. The curves are normalized to the value at average electron energy of 7.5 eV, which corresponds to an electron temperature of 5 eV, i.e. the estimated steady-state value, for the Maxwellian distribution. Cross-section σ(ε) data are taken from Ref. [47].

Figure 6.6: Normalized rates of ionization and molecular excitation resulting in Lyman band emission for Hydrogen as a function of average electron energy.

The comparison of the Lyman band VUV-signal and the electron saturation current shown in Fig. 6.5 (e) can be considered as evidence of the average electron energy peak. The electron saturation current collected by the probe can be expressed as Ies = ene hve iAsheath , where Asheath is the area of the probe sheath, which is considered as

92

constant for Vprobe < Vp . The average electron velocity ve is proportional to the squareroot of the average electron energy i.e. hi0.5 . Thus, Ies ∼ ne hi0.5 . On the other hand, the Lyman band emission signal is proportional to the electron density and the molecular excitation rate, the latter being proportional to hix , where x > 1 as shown in Fig. 6.6. Dividing the Lyman band VUV-signal by the electron saturation current yields a quantity that is proportional to hiy , where y > 0.5 (the plasma density dependences cancel each out). The peak of the Lyman band signal/Ies ratio in Fig. 6.5 (e) is therefore interpreted as evidence of electron average energy (or temperature) peaking during the plasma breakdown transient. The intensity of Lyman band light emission, ξB 1 Σ+ 1 + , is directly proportional u →X Σg to neutral gas density nn , plasma density ne and the molecular excitation rate hve σ()i, which allows the measured Lyman band light signal and Fig. 6.6 to be used for estimating the ratio of the transient to steady-state average electron energy. The ratio of the transient to steady-state light emission intensity can be expressed as

R R [ σ(v)f (v)dv]t nn,t ne,t [ σ(v)f (v)dv]t ξt R ≤ R = ξss nn,ss ne,ss [ σ(v)f (v)dv]ss [ σ(v)f (v)dv]ss

(6.2)

where f (v)dv is the electron velocity distribution function and subscripts t and ss refer to transient and steady state respectively. Similar proportionality can also be derived for Lyman-alpha intensity ξn=2→n=1 by considering the appropriate reactions and cross sections [47]. The inequality in Eq. 6.2 arises from two assumptions: (1) the degree of ionization shown by the plasma is small, implying constant neutral gas density and (2) the electron density does not exceed the steady-state value during the breakdown transient, i.e. ne,t ≤ ne,ss (see Fig. 6.5). According to Fig. 6.5 (b) the transient to steady-state intensity ratio for Lyman band emission is 3.3. The normalization of the data in Fig. 6.6 allows us to read the minimum value of the average electron energy during the breakdown transient at the intersection of the rate curve and the horizontal (dashed) line at 3.3 on the vertical scale. It follows for Maxwell-Boltzmann and Druyvesteyn EEDFs that the average minimum electron energies corresponding to the ignition transient, i.e. the peak of the light emission, are hie,t ≥ 13.5 eV and hie,t ≥ 10.5 eV, respectively. These values are somewhat sensitive to the steady-state electron temperature, as Fig. 6.6 indicates. Overall, it is concluded that the observed peaks of VUV-light emission are clear evidence of the average electron energy peaking during the plasma breakdown process, which is in qualitative agreement with the Langmuir probe data. As shown in Fig. 6.6, the rate of


93

ionization grows with the average electron energy more rapidly than the light emission, again in good agreement with the experimental data. In stochastic heating process the energy gain of an individual electron is proportional to the square of the average electric field [50] i.e. ∝ E 2 , which holds for both, ECR and collisional off-resonance heating. Thus, the temporal evolution of the electric field, displayed in Fig. 6.5 (a), suggests that the average electron energy, which is proportional to the temperature of the distribution - e.g. for Maxwell-Boltzmann distribution hi = (3/2)kTe - should be expected to decrease to 35 - 45 % of the transient value during the build-up of the plasma density. The steady-state electron temperature of 5 eV yields an estimate of 11-15 eV for the transient value. Both results are in good agreement with the values shown in Fig. 6.5. Transient electron energies (and temperatures) obtained through these independent analysis methods are very similar and they are in reasonable agreement with the Langmuir probe data that exhibits transient behavior. The microwave coupling and VUV-emission data are considerably more reliable during transients than the plasma properties obtained by analyzing the Langmuir probe I-V curves. They provide direct information about the plasma characteristics while the probe curves and their analyses are somewhat sensitive to the temporal fluctuations of plasma properties. Averaging the I-V data over consecutive pulses artificially inflates the electron temperature when the plasma potential fluctuates from pulse to pulse, even if the slope of the I-V curve and saturation current remain constant. Similar effects are induced if the tail of the EEDF oscillates over time from pulse to pulse. Thus the reliability of the probe data in steady-state plasma conditions can be deemed to be much better than it is during the ionization cascade, when the plasma potential is in a transient phase as shown in Fig. 6.5. VUV-emission spectroscopy and microwavecoupling diagnostics are non-invasive techniques providing “global” information on the plasma without their sampling power being restricted to a small localized volume. For the reasons given, it is important to estimate the electron temperature during the transient peak by cross-checking the probe data with independent analysis methods as discussed above. This highlights the versatility and temporal resolution of the diagnostics set-up which, to our knowledge, is one of the most sophisticated ever used for studying the transient effects of Hydrogen plasmas in 2.45 GHz microwave discharges. In order to demonstrate that the interpretation of the data is valid for a wide range of experimental conditions, Lyman band light signals recorded with different microwave powers are presented in Fig. 6.7 (a) and different magnetic field configurations in Fig. 6.7 (b). The magnetic field configurations have been labeled according to the distance measured from the injection side of the plasma chamber to the (simulated) resonant

94

resonance 44 mm from injection side





no resonance (B < ECR)

(a)

(b)

Figure 6.7: Lyman band light signals recorded for different powers at 9.3 x 10−3 mb of gas pressure (a) and magnetic field configurations labeled according to the distances measured from the microwave window to the simulated ECR surface at 3000 W of incoming microwave power (b).


95

surface. In all cases, the VUV-signal exhibits a ∼ 10 μs transient peak, indicating higher electron energy at the beginning of the discharge pulse. These data have been averaged over 128 consecutive pulses. Interestingly, the breakdown transient peak is also observed with the magnetic field configuration for which there is no resonance in the plasma chamber, i.e. B < 87.5 mT. This highlights the fact that the nature of the transient is independent of the exact mechanism by which energy is transferred from the microwave to the electrons. Finally, Fig. 6.8 shows the saturation time for the plasma parameters. The saturation of the Lyman-alpha and Lyman band emission signals is reached in about 400 500 μs. It is of note that the plasma parameters obtained with the Langmuir probe, i.e. plasma potential and electron temperature and density, reach their steady-state values much faster (typically in 20 - 50 μs). The saturation behavior of the VUV-signals is therefore attributed to the evolving atomic-to-molecular H/H2 species fraction. This conclusion is supported by the fact that the VUV-signal saturation time is similar to the typical saturation time of the H + /H2+ /H3+ species fraction of the ion beam extracted from 2.45 GHz ion sources measured, for example, by R. Xu et al. [54].

Figure 6.8: Lyman-alpha and Lyman band VUV-signals showing a typical saturation time of ∼ 400 μs to reach the steady-state plasma condition.

96

6.4


It is concluded that this combined diagnostics supports the time resolved Langmuir probe diagnostics described in chapter 4 to provide a valuable tool for the study of plasma parameter evolution during transients. The existence of the temperature peak associated with the breakdown process reported in chapter 4 is corroborated by the results shown in this chapter. The VUV-light emission depends strongly on the average electron energy and, therefore, exhibits a strong transient peak associated with the plasma breakdown. This peak coincides with the electron temperature peak shown by analysis of the Langmuir probe curves. The results obtained can be summarized using the following qualitative model for the plasma breakdown: initially the plasma density is low, allowing the average microwave electric field in the plasma chamber to build-up. Under such conditions electrons are heated efficiently and the average energy of their population reaches values exceeding the saturation value. The exponential increase in the plasma density causes the average electric field and electron energy to decrease until the diffusive loss rate of electrons matches the ionization rate of the neutral H2 -gas. The ionization rate (dne /dt) is proportional to neutral gas density and average electron energy, i.e. the plasma density increases almost exponentially until the steady state is reached. The 10 μs ignition transient corresponds to the progress of the ionization cascade. The interpretation given for this time-scale is supported by the calculated average collision time of 100 - 500 ns for 15 eV electrons in the neutral gas pressure range corresponding to our experimental conditions (cross-section data from [47]). The saturation of the probe data is achieved in 20 - 50 μs, which corresponds to the temporal scale of the ion motion (Ti < 1 eV) at the relevant length scale, i.e. chamber dimensions of ∼ 0.1 m. Finally, the 400 - 500 μs saturation time of the atomic and molecular VUV-signals corresponds to the temporal evolution of the species fraction. The model is similar to the one suggested for high frequency ECR ion sources [55]. However, the present study provides significantly more data, both direct and indirect, which supports the interpretation and extends the validity of the model from highfrequency minimum-B ECR ion sources to 2.45 GHz microwave plasma discharges, where the temporal evolution of plasma parameters is significantly faster due to higher neutral gas pressure. The experiments indicate that maintaining the electron temperature at the transient level could enhance ionization rate and improve the efficiency of 2.45 GHz microwave ion sources.


97

From the breakdown study presented in this chapter, arose the idea of using the same combined diagnostics to study the decay process too. However, the limited time that the VUV spectrometer remained in our laboratory made it impossible to carry out a decay study. All the experiments were carried out during a five-weeks working period.

Chapter 7

Influence of Microwave Driver Coupling Design on Plasma Parameters Microwave coupling within a resonant cavity has been a widely studied engineering problem ever since microwaves were first used [56], and it is a critical process for ECR ion source performance. The way in which energy is transferred from the microwave generator to the plasma, along with the physical processes involved, have been discussed many times in literature [24, 57, 58], and the effect of plasma ignition on resonant frequency, electric field distribution and coupling parameters are issues still under discussion in the ECRIS community [5, 22, 23]. One goal of the work carried out in TIPS was to acquire a better understanding of the microwave coupling process. The coupling process, the way in which energy is transferred to the plasma chamber and the electric field intensity inside it, all influence the plasma density that is acquired. Chapter 6 showed the strong influence of the electric field drop produced by the plasma breakdown on the evolution of the plasma parameters. In this chapter the influence of the electric field distribution just before breakdown on plasma parameters will be explored. Plasma density is one of the most important factors in ion source performance, having a major impact on the ECR ion source current, with the consequent natural interest on the part of ion source designers in its optimization. In this chapter, we present the results of a comparative study made between two different plasma chamber and coupler systems, the preliminary system (used in all previous chapters) and an optimized version. The aim of this study was to improve our understanding of the relationships between electric field distributions inside the full

99

100

microwave driver system before plasma disruption and final plasma density that is obtained. The study is composed of E-field simulations and experimental measurements of incoming/reflected power ratios and plasma parameters along the discharge chamber axis. The plasma parameters measured showed that the optimized system has the capability to produce values of plasma density four times higher than the preliminary one.

7.1

Preliminary Design Description

Figure 7.1: Section view of TIPS and main subsystems with preliminary design: plasma chamber (a), microwave coupler (b), gas inlet (c), pressure gauge flange (d), cooling water inlet and outlet (e) and (f ), tapered waveguide WR284/WR300 transition (g), vacuum break window (h), dual directional couplers (i) and (k), two-stub tuner (j), diagnostics port (l), observation window (m), Langmuir probe (n) and magnetic field generation system (o).

Although TIPS preliminary design was described in detail in chapter 2, a section view of the device and its MW coupling system is again shown in Fig. 7.1 to facilitate comparisons between the preliminary and the optimized design. In the figure, the Langmuir probe diagnostic port is installed. The Langmuir probe set-up described in chapter 4 was used to determine the evolution of plasma electron density and temperature along the plasma chamber axis in both the preliminary and the optimized designs. The preliminary chamber design was produced by Elytt Energy, as one of the engineering supply companies for ESS Bilbao. The microwave system was considered as an ensemble composed of coupled independent parts and the resonant dimensions of the plasma chamber were calculated by considering it as a cylindrical cavity. The microwave coupler design was conceived to adapt the impedance between the plasma chamber and a standard WR284 microwave guide, following the engineering criteria of impedance

Chapter 7. Influence of Microwave Driver Coupling Design on Plasma Parameters

101

matching [59]. This coupler corresponds to letter (b) in Fig. 7.1. It is made of brass and the inner geometry consists of five ridged steps. Fig. 7.2 shows a diagram of the dimensions for the preliminary coupler ridged steps dimensions. 15

12 15.7

26.1 20

34

(a)

72.1

39.8

39.9

43.3

(b)

48.2

58.6

230

Plasma chamber side

Waveguide side

Figure 7.2: Diagram of microwave coupler dimensions in the preliminary design. Dimensions in millimeters.

It should be noted that the design methodology and calculations followed by the preliminary designers were completely different from the criteria used in this chapter to optimize the design. Although we started from the same point by calculating the resonant frequencies of the plasma chamber as a perfectly conductive cylindrical cavity, for all the other calculations, the entire microwave system was considered as a resonant system. The results shown in this chapter support this second option as a better design strategy. During the first testing stage, immediately after commissioning of the system, a strong plasma tendency to ignite outside the plasma chamber was detected. This behavior is characterized by the formation of plasma inside the microwave coupler or even farther afield, and was first noticed due to anomalous heating in the tapered waveguide area. Fig. 7.3 shows a thermal image of the vacuum break window holder and the tapered waveguide ((g) and (h) in Fig. 7.1), showing how this region heated up. Image temperatures are expressed in degrees Celsius and show good agreement on the side of the plasma chamber (right) with water cooling temperature supplied by the chiller. The mirror image below the waveguide corresponds to the infrared image reflection in the aluminum plate where the device is resting. A quartz observation window was placed in the vacuum window holding part. By means of a photodiode and fiber optics, we checked the plasma presence in this region

102

to ensure that heating is due to plasma and not to any other reason like microwave mismatching between parts, etc.

Figure 7.3: Thermal picture of the tapered WR284/WR300 waveguide transition piece and vacuum break window holder part.

The area near the vacuum window is the most distant point from the pumping system. Relatively speaking, therefore, it is always the zone with the highest pressure in the system. A gauge was installed in the vacuum window holder to check whether this higher relative pressure could be the cause of plasma ignition. As expected, pressure was inevitably higher in this area than in the rest of the system. However, the values measured do not justify the plasma formation at this point. This tendency could be corrected by carefully setting the parameters to allow plasma stabilization inside the discharge chamber. This point was especially critical with regard to the magnetic field profile. Different B-field configurations were tested and only a few of them could be used to confine plasma ignition to inside the plasma chamber. The configuration that produced stable plasma over the widest range of pressure and MW power conditions was the B>ECR configuration already used as a working condition in all of the previous chapters. This configuration was also chosen as a working condition in the study described in this chapter. Fig 7.4 reproduces again the 2D simulated map of the magnetic field configuration to make comparisons easier. The relatively narrow parameter range that is capable of sustaining the plasma in its proper position, without formation of a secondary discharge inside the microwave coupler and/or waveguide, is what motivated the electromagnetic study of the system in the first place. After a parametric study, we found that this tendency toward plasma ignition outside the discharge chamber can be counteracted by working under certain experimental


103

conditions: low gas pressure, low magnetron power and a longitudinally asymmetric magnetic field profile, with this last factor having the greatest impact.

r z

mT 150 146 142 138 134 130 126 122 118 114 110 106 102 98 94 90 86 82 78 74 70

Figure 7.4: Asymmetric magnetic field distribution required with the preliminary design to avoid plasma formation outside the discharge chamber while keeping a sufficiently wide pressure and MW power working ranges.

A critical behavior type was observed for relatively small changes in magnetic field configuration producing alternating plasma ignition in the discharge chamber and microwave coupler or farther afield. The tendency of the plasma to be located in the microwave injection area becomes unavoidable. At this point, is important to highlight that in all the electric field simulations shown in this chapter we are considering the situation before the plasma disruption and, thus, the plasma presence was not taken into account. As it was stated in chapter 6, the drop of the averaged electric field produced by the plasma breakdown can be estimated from Pr /Pi data by Eq. 6.1. However, we can not estimate the electric field distribution when the plasma is ignited because, as it will be seen in chapter 8, the plasma distribution is not homogeneous. Fig. 7.5 shows the simulated electric field distribution for the geometry of the preliminary design. The microwave input power port, where TE10 mode is excited assuming 1500 W of microwave input power, is marked in the figure. The section denominated Waveguide contains a microwave two-stub tuner with bidirectional couplers attached at both sides corresponding to Fig. 7.1 (i), (j) and (k). The Transition piece corresponds to the tapered waveguide WR284/WR300 transition piece of Fig. 7.1 (g), followed by the Microwave Coupler and Plasma Chamber, (b) and (a) in the same figure respectively. Note that the whole microwave system, from the plasma chamber to the end

104

of the second bidirectional coupler (where experimental data for incoming and reflected power are measured) has been considered as a resonant cavity.

ve Wa

tion nsi Tra

MW a sm P la b e r m cha

p le cou

p ie

gui

de

ce i MW

r

t npu

t por

ak bre uum w Vac indo w

Figure 7.5: Simulated electric field distribution with the preliminary design geometry and 1500 W continuous input power through the 2.45 GHz TE10 port.

Fig. 7.5 clearly shows that the highest electric field maximum is inside the microwave coupler where plasma tends to ignite in most cases. By using this electric field distribution as a clue to such behavior, we modified the geometry until we reached a different electric field distribution with the maximum located in the center of the plasma chamber. This new design was denominated ‘optimized ’ and is described in the following section. Additionally, several simulations were done for different tuner stub positions to check the influence of this subsystem without registering significant differences in the electric field distributions obtained.

7.2

Optimized Design Description

When a microwave resonant cavity is tightly coupled to some other circuit, the impact of any physical modification to the cavity and the change in the internal field distributions are so great that it can no longer be called the same cavity. In this case, the concept of an intrinsically resonant frequency for the cavity becomes meaningless, and we can only talk about the resonant frequency of the system as a whole [56]. Bearing in mind this concept, we started from the design of a separated ideal cylindrical plasma chamber as a first step. The resonant frequency of a particular TE mode in a cylindrical cavity is given by the following equation [60]:


TE fnml =

c √ 2π µr εr

s

p0nm r

2

+

lπ d

105

2 (7.1)

where: c is the speed of light, εr and µr are respectively the relative electrical permittivity and relative magnetic permeability of the medium filling the cavity (in the case of vacuum, εr = µr = 1), r is the chamber radius and d, its length, and p0nm is the zero of order m for the first derivative of the Bessel function of order n. Indices n, m and l identify the electromagnetic field pattern of the TEnml mode. Based on Eq. 7.1 and taking into account the available room due to solenoid dimensions, the new chamber was designed to have its TE111 resonant mode at our working frequency, 2.45 GHz, and its dimensions were set to be of 85 mm diameter and 113 mm in length. This new chamber was made with two coaxial aluminum parts sealed with O’rings to leave a water cooling bath surrounding the resonant cavity. Boron Nitride discs were again used at both sides of the plasma chamber. The second step was to inject microwaves into this new chamber after attaching the microwave guide. The introduction of any coupling into a cavity involves a certain amount of perturbation of the internal field distribution, which has two significant effects: the resonant frequency must be considered as belonging to the entire system, since it is generally different from the original resonant frequency of the cavity without coupling, and the losses in the cavity are increased by the amount of power fed out through the coupling system. In fact, if we simulate this new chamber directly attached to the microwave guide, resonant frequency shifts to 2.42 GHz.

MW t port u inp

ve Wa

a sm P la b e r m cha

tio nsi Tra

np

de gui

ie c e


Figure 7.6: Simulated electric field distribution with the φ85 x 113 mm chamber directly attached to the waveguide and 1500 W continuous input power through the 2.45 GHz TE10 port.

106

Fig. 7.6 shows the stationary electric field distribution obtained for this case with a 2.45 GHz excitation of 1500 W input power at the port. Note how this distribution gains intensity in practically the entire system without any effective concentration of field inside the chamber. Unfortunately, it was impossible for us to experiment with this configuration due to flange incompatibility between the plasma chamber and the waveguide. The plasma chamber has a design that does not permit us to attach a standard flange directly onto it. This is the reason for the lack of experimental data in Table 7.2. However, the electric field simulation for this particular case is interesting because it highlights the importance of the part denominated coupler between the plasma chamber and the rest of the microwave guide system. This part has the critical function of returning the resonant frequency of the entire system (including the coupler itself) to the working microwave frequency of 2.45 GHz at the same time that it maximizes the E-field value inside the chamber and minimizes it in the rest of the system. MW t port u inp

ve Wa

a sm P la b e r m cha

tio nsi Tra

np

gui

de

ie c e


MW r p le cou

Figure 7.7: Simulated electric field distribution with the optimized design geometry and 1500 W continuous input power through the 2.45 GHz TE10 port.

Simulations showed that a one-step ridged coupler of length λ/4 was the best option among those that were studied. Fig. 7.7 shows the results of the electric field simulation once this one-step coupler is included in the system. It is remarkable how, with this design, the maximum electric field is higher than the one obtained with the preliminary design and most importantly, that the maximum is located in the center of the plasma chamber. Note that the color scale is not the same in Figs. 7.5, 7.6 and 7.7 due to the big difference in the electric field values obtained in each case. Fig. 7.8 shows a 3D view of the new MW coupler design that was machined in aluminum. The ridged shape is formed by means of two exchangeable inserts screwed to the main body of the coupler. This enabled us to change the shape of the ridged section easily without machining a complete new coupler. It is 30 mm long and is provided


107

with channels for gas feeding and pressure measurements with the same diameters and positions that the original coupler had. In addition to the improvement in the electric field distribution previously mentioned, reducing the length of the coupler and simplifying its geometry resulted in a decrease in the vacuum volume and minimized the number of air traps, thereby improving the pumping. Gas Inlet

Exchangeable inserts

Vacuum gauge channel 21.6 mm

34 mm

8 mm

72 mm

Figure 7.8: One-step ridged coupler design. 3D view and ridged section dimensions.

Fig. 7.9 shows a section view of TIPS with the optimized chamber (a) and MW coupler (b) installed. The new chamber’s water cooling bath (c) is also visible. In the MW coupler ridged inserts (d) are marked in blue. The coupler has also for gas intake (e) and vacuum gauge connection (f ). Applying the magnetic field distribution previously used with the preliminary design produced an unstable plasma behavior in which high jitter in the incoming and reflected power signals was observed. Although we wanted to compare both designs while keeping the working parameters constant, this unstable plasma behavior forced us to change the B-field profile in the optimized design case. A 2D plot of this new B-field distribution is shown in Fig. 7.10 with the same color scale used previously in the preliminary design magnetic field distribution shown in Fig. 7.4 to make comparisons easier. With the optimized design and this magnetic field configuration, the plasma showed robust behavior for relatively small changes in coil position as well as its currents. The tendency to ignite outside the discharge chamber also disappeared.

108 (a) (c) (e) (b)

(f )

(d)

Figure 7.9: Section view of TIPS with optimized design: plasma chamber (a), one-step ridged coupler (b), water cooling chamber (c), ridged exchangeable steps (d), gas inlet (e) and vacuum gauge connection channel (f ).

r z

mT 150 146 142 138 134 130 126 122 118 114 110 106 102 98 94 90 86 82 78 74 70

Figure 7.10: 2D color map of the magnetic field distribution used with the optimized design. The resonant surface is marked with a dashed black line.

The optimized plasma chamber and MW coupler set is fully compatible with the diagnostic port, allowing us to take Langmuir probe measurements of electron density and temperature to compare the characteristics of the plasma generated in both designs.


7.3

109

Design Comparison and Experimental Results

The first qualitative difference observed during the optimized design testing is that the plasma tendency to be located inside the MW coupler and the tapered waveguide transition disappeared. Another remarkable difference is that the range of working parameters is much wider with the optimized design. In the following subsections we present a comparison of electric field simulations, magnetic field distributions, coupling parameters and measured plasma density and temperature for both cases of study.

7.3.1

Electric Field Distribution

Fig. 7.11 shows the simulated electric field norm along the axis for the three cases studied. Superimposed on each graph are schematic cross section views of each system to show the electric field distributions. Fig. 7.11 (a) shows the preliminary design distribution, Fig. 7.11 (b) shows the new chamber directly attached to the microwave guide and Fig. 7.11 (c) shows the case of the optimized design including the new λ/4 one-ridged step MW coupler. In the preliminary design, the electric field maximum is situated inside the coupler, more precisely in the second ridged step, while similar relative maximum values are periodically distributed throughout the system. The relative local maximum inside the chamber is lower than any of the others inside the coupler and tapered transition piece. The case where the new chamber is attached directly to the microwave guide presents practically the same distribution, but for the case of the optimized design, simulations show a completely different E-field distribution where the absolute maximum is centered inside the plasma chamber. The electric field maximum inside the chamber is five times higher in the case of the optimized design. It was observed that the optimized system can sustain a plasma ignited inside the chamber even without the presence of magnetic field. Fig. 7.12 shows a photograph of the plasma generated in the optimized design chamber when the B-field is completely removed. The photo was taken from the diagnostics side of the chamber using the set-up developed for the ultra-fast intensified frame image diagnostics that will be described in the next chapter. This fact suggests that the E-field has been effectively improved. It was tried without success to measure plasma density and temperature in this nonmagnetic embedded plasma. The plasma was extinguished every time that polarization voltage was applied to the Langmuir probe without any way to sustain the plasma during probe measurements. We also noted a low visible light emission intensity in relation to

110

the plasma embedded in the magnetic field, and a very low microwave absorbed power of less than 2 %, both of which also suggest a low plasma density. (a)

(b)

(c)

Figure 7.11: Simulated electric field norm along the plasma chamber axis for the cases studied: (a) preliminary design (chamber with 90 mm diameter by 97 mm length and fivestep ridged coupler), (b) plasma chamber with 85 mm diameter by 113 mm length directly attached to the waveguide and (c) optimized design with plasma chamber 85 mm in diameter by 113 mm long, one-step ridged coupler and waveguide.

The stored time-averaged electric energy in the system formed by the chamber, the coupler and the waveguide (Us ) can all be calculated from the electric field distribution using the equation:


111

Figure 7.12: Digital photograph of the plasma sustained in the optimized design plasma chamber without magnetic field.

1 Us = εo 2

Z

~ |2 dV |E

(7.2)

V

where εo is the vacuum electric permittivity and E is the electric field. Using electric field distributions obtained by simulations to calculate Us by Eq. 7.2, we found that the energy stored in the optimized system is 30 times higher than in the preliminary one. The distribution of this energy in each part of the system, i.e the chamber, the coupler and the waveguide, is of special relevance in understanding the tendency of the plasma to ignite inside or outside the discharge chamber. It reasonable to state that plasma will show a tendency to ignite in those areas where the available energy to produce plasma breakdown is higher. Table 7.1 shows the percentage of energy stored in each part of the system for the preliminary design, the new chamber with the MW guide case and the optimized one. Note that in Fig. 7.11 (c) a narrow E-field peak is located at approximately at the point of union between the coupler and the transition piece reaching a value of approximately 80 % of the maximum inside the chamber. However, this fact does not disturb the stability of the plasma inside the discharge chamber, with no plasma detected outside. The difference between stored energy in the plasma chamber and the coupler already shown in table 7.1 may offer a good reason for such stability.

112

Stored energy (%) Chamber

Coupler

W aveguide

Preliminary design

32 %

18 %

50 %

New chamber

33 %

—

67 %

Optimized design

93.1%

1.2 %

5.7 %

Table 7.1: Calculations of Stored Electric Energy Percentages

Fig. 7.13 shows experimental the Pr /Pi ratio for both designs. It is noticeable the different behavior of each designs: while the preliminary design shows Pr /Pi ratio during the steady state of approximately 0.05, the optimized one shows a higher value of around 0.2. Knowing that the performance of the optimized design is better, as will be shown in section 7.3.4, it is concluded that optimizing the Pr /Pi ratio might not be the best strategy for reaching the best performance in a plasma source.

(a.u)

It needs to be pointed out that Fig. 7.13 shows a shorter coupling time for preliminary design according to the criteria defined the section 3.2 of chapter 3. However, the large range of breakdown times measured in chapter 3 does not allow us to extract conclusion from Fig. 7.13 about the existence of differences between both designs regarding the breakdown times.

Figure 7.13: Preliminary and optimized design Pr /Pi ratio comparison

As it was stated before, estimating the electric field value or its distribution once the plasma is ignited is subject to a large number of uncertainties, especially regarding the plasma density distribution. However, it is possible to estimate the relative change on the average electric field in the optimized by Eq. 6.1 as it was described in chapter 6. According to Fig. 7.13, for both designs the initial Pr /Pi ratio is approximately 0.9, thus, the transmission losses Ploss can be estimated to be 10 % of the incoming power. Then, following Eq. 6.1 we obtain the results shown in Fig. 7.14.


113

Figure 7.14: E-field relative change comparision for the preliminary and the optimized designs.

The electric field drop associated to the plasma filling the cavity is smaller (20 %) in the optimized design than in the preliminary one (40 %). From Fig. 7.11 it can be seen that the maximum electric field inside the chamber just before breakdown is five times higher in the optimized design case than in the preliminary one. Then, we can estimate the electric field when the plasma is filling the cavity to be approximately six times higher in the optimized design.

7.3.2

Magnetic Field Distribution

(a) (b)

MW injection

Preliminary design chamber

Optimized design chamber

Figure 7.15: Simulated axial B-field along the plasma chamber axis for both designs: curve (a) corresponding to preliminary design and curve (b) to optimized.

As mentioned before, the B-field distribution used is different for each design. Fig. 7.15 shows a plot of the B-field along the axis of the plasma chamber for both cases: curve (a) corresponds with the preliminary design magnetic field and curve (b) with the optimized one. Marked between broken vertical lines is the length of each plasma

114

chamber from the MW injection end (left side of the plot), set at the z=0 mm position for both cases. MW injection

Diagnostics port (a)

[V/m]

B > B ECR

B < BECR

1 cm B > BECR

(b)

[V/m]

B < BECR

Figure 7.16: Resonant electric field distributions obtained in simulations with the resonant surface superimposed to see the combination of the E and B fields inside both discharge chambers: (a) corresponding to preliminary design and (b) to optimized design.

It is also interesting to compare the resonant surface position with respect to the chamber geometry in each case [20]. Fig. 7.16 shows the simulated electric field distribution inside the plasma chamber for both designs. The position of the resonant surface is marked with a broken black line showing the limit between two volumes, one with B-field values over the ECR and the other one with values under the ECR. Case (a) represents the preliminary design while (b) represents the optimized one. MW injection is located on the left side of each chamber and the diagnostics port on the right. Note that the electric field in the chamber center is nearly ten times higher in the optimized design case than in the preliminary one. It is also noticeable that for the optimized design, the resonant surface is positioned deeper inside the chamber while maintaining plasma


115

stability conditions and allowing for a higher amount of resonant surface embedded in the plasma under the relative high E-field.

7.3.3

Beta Coupling Parameters

The design strategy proposed in this chapter is characterized by considering the full microwave system as a resonant cavity and optimizing the electric field in just one part (the plasma chamber) instead of the criteria where an ensemble of parts is impedance matched. Under this framework, it may be interesting to estimate coupling parameters β because their relationship with quality factor matching between parts can help to understand the coupling behaviors of both systems. We can express coupling parameters β by using the following expression [22]: p (Pr /Pi ) p β= 1 ± (Pr /Pi ) 1∓

(7.3)

where Pi is the incoming power and Pr is the reflected power. The upper signs correspond to the undercoupled cavity case and the lower ones to the overcoupled. By studying the behavior of scattering parameters S11 as a function of frequency [61, 62] for both preliminary and optimized design simulations, we can conclude that in both cases the system is undercoupled and we would thus use upper signs in β calculations. Pi and Pr were recorded with a scope connected to the bidirectional coupler shown as (k) in Fig. 7.1, as been already shown in previous chapters and with special detail in chapter 3. When the system is considered under vacuum conditions, i.e before plasma breakdown, coupling parameters in vacuum βv can be calculated both theoretically and experimentally. The first can be achieved using the simulated values for incoming and reflected power and the second, from signals of incoming and reflected power signals measured on the bidirectional coupler placed in the MW waveguide ((k) in Fig. 7.1). The calculation of coupling parameters when plasma is filling the cavity βp can not be done theoretically because this simulations do not take plasma into account. They were instead obtained using experimental data. Table 7.2 summarizes such coupling factor values for the systems under study when they are calculated from simulations and from experimental data. The experimental case of the new chamber directly attached to the microwave guide is not present due to the practical impossibility of attaching both parts, as has already been mentioned.

116

Simu. βv

Exp. βv

Exp. βp

Preliminary design

0.003

0.005

0.63

New chamber

0.003

N/A

N/A

Optimized design

0.038

0.041

0.3

Table 7.2: Coupling Parameters in Vacuum βv and with Plasma βp

There is a remarkable difference in the behavior between vacuum and plasma β parameters for the preliminary and optimized designs. While βv shows a value one order of magnitude higher for the optimized system, if plasma is ignited the βp shows the opposite behavior: it is smaller for the case of the optimized design. This is very interesting because the fact that βp reaches a higher value in the case of the preliminary design does not necessarily mean that the performance of this design is better than the optimized one. It is important that bear in mind that plasma is an electrical conductor and its relative permittivity and permeability are different from 1. Due to this fact, plasma presence inside the chamber produces a shift in resonant frequency that is bigger when plasma density increases. This implies that having higher plasma electron density would result in a worse coupling once the plasma starts filling the chamber. The shift in chamber resonant frequency can be estimated by considering the case of an homogeneously distributed plasma filling the chamber with a constant electron density [23, 63] and it shows major variations from 2.45 to 3.3 GHz when electron density changes between 1 and 6 x 1016 m−3 respectively. This effect of electron density on the plasma chamber resonant frequency suggests that β optimization may not be a good enough criterion to optimize plasma parameters. Experimental results shown in next the section may support this point.

7.3.4

Density and Temperature Measurements

The Langmuir probe diagnostic port described in chapter 4 and also used in chapters 5 and 6 was used to obtain Langmuir probe curves to determine plasma parameters during the steady state state in pulsed operation mode with both plasma chamber and MW coupler sets along its z-axis. The calculation of the plasma parameters from the I-V curves was made using the method described in section 4.3.2, where the electron temperature is obtained from the slope of the linear fitting on the semilog I-V curve, the ion saturation current is assumed to follow Laframboise’s hypothesis and the plasma potential is estimated as the crossing point of the two linear fittings on the semilog


117

I-V plot. The aim of these experiments was obviously to determine the impact of the changes made by our system on the plasma parameters. Fig. 7.17 shows two typical Langmuir probe curves, one of them obtained with the preliminary plasma chamber and MW coupler set (a) and the other with the optimized one (b). Both curves were measured under the same operating conditions: 1500 W of input power and 3.8 x 10−3 mb for Hydrogen pressure. The magnetic field distribution for each case is the one described in the previous section and shown in Fig. 7.15 and 7.16. In both cases the Langmuir probe was placed at the axis of the plasma chamber and longitudinally in the center. The current measured by the probe on the right-hand flat region (electron saturation current) is proportional to the electron density [36]. Note how this value is higher for the optimized design case.

(b)

(a)

Figure 7.17: Typical Langmuir probe curves obtained with the original plasma chamber and coupler design (a) and with the optimized one (b).

Fig. 7.18 shows the results of electron plasma densities obtained under the experimental conditions previously mentioned along the z-axis of the plasma chamber for both of the cases under study. Data were measured at intervals of the 10 mm taking the zero reference on the left side of chamber which corresponds to the MW injection port. Note that the differences between chamber lengths are marked with two vertical broken lines. Electron density values are indicated by blue dots for the preliminary design and red for the optimized one. The behavior of density along the z-axis for the preliminary design shows values of around of 1 x 1016 m−3 without noticeable variations. In contrast, density values obtained for optimized designs show values that reach 4 x 1016 m−3 in the proximity of the diagnostic port (corresponding to the extraction zone in an ECRIS).

118


MW injection


Figure 7.18: Densities obtained by Langmuir probe measurements along the z-axis in both plasma chambers under study. Values were measured during the steady-state stage of plasma in pulse mode at intervals of 10 mm.


MW injection


Figure 7.19: Electron temperatures obtained by Langmuir probe measurements along the z-axis in both plasma chambers under study. Values were measured during the steady-state stage of plasma in pulse mode at intervals of 10 mm.

On the other hand, Fig. 7.19 shows the electron temperature values corresponding to the same points as in Fig. 7.18. The same kind of representation has been kept with the only exception being the blue and red triangles that are used for the temperature points of the preliminary and optimized design respectively. Note that, in both cases, temperatures reach maximum values between 13-15 eV on the left side that corresponds to the MW injection port. For the preliminary design, these maximum values are reached in the proximity


119

to the resonant surface that lies at 10 mm. However, for the optimized design, where the resonant surface is at 30-40 mm, temperatures are relatively lower and show no important variations associated with the combination of high electric field values and the resonant surface as might be expected. It is also noticeable that temperatures reach close values of 3 and 5 eV for both cases in proximity to the diagnostics port (extraction port for an ECRIS), showing no practical differences at this point. Also notable is the coincidence of temperature values at both sides of the plasma chamber, but with differences in the middle that may be related with the improvement to the electric field distribution. The product Te x ne can be considered as a relative measurement of the energy contained in the plasma. From Figs. 7.18 and 7.19 we can estimate that this product was approximately six times higher in the case of the optimized design which in good agreement with the estimation made on section 7.3.1 where the electric field inside the plasma chamber after plasma ignition was expected to be six times higher for the optimized design.

7.4


A comparative study of two microwave driver systems (preliminary and optimized ) to produce a 50 Hz pulsed 2.45 GHz Electron Cyclotron Resonance Hydrogen plasma has been presented in this chapter. 3D simulations of resonant electric field distribution in vacuum, 2D simulations of magnetic field distribution produced by external coils, experimental measurements of incoming/reflected power and Langmuir probe measurements of plasma parameters along the chamber z-axis have been all covered. Special attention was paid to the stationary electric field pattern in the entire microwave excitation system as an engineering tool for obtaining an optimized design. The E-field is maximized in the discharge chamber and minimized in the rest of the MW driver system. Although such distribution may be considered only as a reasonable estimation until the breakdown instant because of the shift in the resonant frequency produced by the plasma [23, 63], its effects on plasma behavior and its parameters were experimentally studied. As shown in Fig. 7.16, the B-field profile had to be modified from its preliminary settings for the optimized system, suggesting a strong connection between the E-field and the B-field combination. It is also worth noticing how the resonant surface gains penetration for the optimized design to reach higher electric field values in the proximity of the plasma chamber center. If the design criteria conform to the maximization of electric field values in the center of the plasma chamber, keeping it as low as possible in

120

the rest of the system, a new, stable operating range may be reached where the resonant surface could play a decisive role. From an engineering point of view, the design of two of the parts in both systems have been demonstrated to be critical: the plasma chamber and the microwave coupler. The initial symptom, detected in the preliminary design of plasma tendency to be located outside the plasma chamber, has been solved by using the optimized system design. The abnormal heating in the proximity of the vacuum window deriving from plasma formation has also disappeared. The new design has enabled us the possibility to study plasma behavior in TIPS under a much wider range of stable working parameters, and this has proved extremely important for other studies and diagnostics like the ultra-fast picture diagnostics that will be described in next the chapter. In addition to the remarkable stability mentioned above, the optimized design can sustain a plasma inside the chamber even without a magnetic field, as shown in Fig. 7.12 supporting the hypothesis that the E-field has been effectively improved. This plasma absorbs just 2 % of power and shows low intensity visible light emission that also suggests low density. Measurements of these plasma parameters by Langmuir probe were unsuccessful, producing plasma extinction when voltage was applied. Calculations of beta factors for both systems in vacuum conditions and with the presence of plasma suggest that the optimization of this factor may be not a good enough criteria for an optimized plasma generator. This process of de-coupling in favor of a higher plasma density may be connected with an effect reported by other researchers, where output current in some ECR ion sources increases when coupling efficiency is decreased, proving that the best output is not necessarily that which gives the minimum reflected power [64]. In order to determine the impact of the changes in both designs on their respective plasma parameters, a study of density and temperature along the z-axis of both discharge chambers was undertaken by measuring with a Langmuir probe. A significant improvement in electron density was recorded for the optimized system, reaching values of more than four times the values of the preliminary one. However, temperatures show similar behavior for both designs with only a slight increment for the optimized design detected. For the preliminary design, high temperature values are reached in the proximity of the resonant surface that is at 10 mm (see Fig. 7.16). However, nothing really remarkable is observed for the optimized design at 30-40 mm where the resonant surface is located.


121

Finally, 3D stationary vacuum resonant electric field simulations across the entire MW system show that it is a valuable tool for approaches to ECR plasma source design. The result is a maximization of the E-field in the discharge chamber while minimizing it in the rest of the MW system and a deeper penetration of the resonant surface while simultaneously maintaining plasma stability. This increment in the amount of resonant surface in contact with the plasma at relatively higher E-field values seems to play an important role in producing higher densities and more stable plasma behavior.

Chapter 8

Ultra-Fast Pictures as a Method for Ion Source Plasma Diagnosis Ultra high speed photography by image intensified gated cameras has been an important diagnostic tool for decades, especially for hot and dense plasmas with high light emission like those found in Z-pinches, plasma foci and exploding wires, among others. These kinds of pictures yield valuable information about plasma density distributions and their time evolution as well as their interaction with magnetic fields. However, for the case of Electron Cyclotron Resonance (ECR) plasmas, where low densities and temperatures place them on the borderline of the camera systems’ capabilities, few studies of the spatial and/or temporal light emission profiles have so far been conducted [65–68]. In this chapter, a new plasma diagnostic is presented. The aim of this design was to use ultra-fast pictures as a new tool for plasma diagnostics. A “transparent plasma electrode” developed using a 10 mm thick quartz window placed between two grounded tungsten meshes, was used to maintain the microwave enclosure and the resonant frequency of the cavity. This system enables us to view the volume of the entire plasma chamber, allowing us to obtain images of plasma distribution in a wide range of experimental conditions. An image Intensified CCD frame camera system, which is based on a combination of multichannel plate (MCP) light intensifiers and CCD detectors, was used to obtain sets of pictures with sufficient quality to recognize different emission patterns with a temporal resolution of 1 μs. As in the experiments described in previous chapters, incoming and reflected microwave power were recorded to check the coupling characteristics for each case under study. Visible integrated spectra showing Balmer α, β and γ lines and the Fulcher band were also measured using a fiber optic visible spectrometer. These spectra are obtained by adjusting the spectrometer integration time according to intensity, but 123

124

always averaging the sum of several entire plasma pulses. This kind of measurement is representative of the steady-state plasma stage. Such spectra show interesting information about atomic and molecular ionization/recombination processes that may help to understand the ratios between ion species present in the distributions observed.

8.1

Experimental Set-Up (a)

(d)

(e)

(b)

(c)

pump

Figure 8.1: A cross sectional view of the plasma source: (a) optimized plasma chamber, (b) optimized microwave coupler, (c) optical window with shielding grids, (d) vacuum tee, (e) standard quartz window.

A completely new diagnostic port was designed to create a wide observation window to obtain frame pictures of full plasma volume. Fig. 8.1 shows a cross section view of the system with the new diagnostics installed so that its components can be seen. In the work described in this chapter, we used the optimized aluminum plasma chamber (a) and ridged wave guide coupler (b) described in the previous chapter. The optical window that plays the role of a transparent plasma electrode can also be seen in Fig. 8.1, where the double electrical shielded window design is shown (c). Two meshes, each one a 2.5 mm square structure made of 0.25 mm tungsten wires, were placed on both sides of a 10 mm thick quartz window that has a cylindrical hole with a 7 mm diameter at its center for pumping. Meshes were connected to the plasma chamber body to keep them grounded. This design met three important criteria: it allowed us to obtain an image of the volume of the whole plasma chamber, to pump gas through a centered hole and to maintain the electromagnetic chamber resonance while avoiding microwave losses. This last aspect was checked by direct measurements

Chapter 8. Ultra-Fast Pictures as a Method for Ion Source Plasma Diagnosis

125

of non-ionizing radiation emissions with a specific radiometric device mod. NARDA NBM-550. The Boron Nitride disk on the incoming power side of the plasma chamber was kept (not visible in the figure), while the one corresponding to the diagnostic port side was removed. Once the system had been tested with the plasma discharge established, no differences were observed in general plasma behavior in relation to previous experiences using a solid aluminum plasma electrode. This supports the assumption that this double shielded transparent window is basically equivalent to a solid metal plasma electrode with an extraction hole in the center as is usual in an ECRIS. The pumping system was connected by using a standard ISO DN 100 tee (d) that allowed us to place a standard quartz window (e) on the optical discharge axis to maintain the vacuum enclosure. Fig. 8.2 shows a photograph of TIPS with the ultra-fast picture set-up installed. The fiber optics (in blue) can be seen on the bottom left side of the standard quartz window.

Figure 8.2: Photograph of TIPS with the Ultra-Fast Picture Diagnostic Installed. Blue fiber optics can be seen on the bottom left side of the standard quartz window.

A photograph showing a closer view of the device can be seen in Fig. 8.3. The standard quartz window, the DN 100 tee and part of the tungsten mesh on the plasma electrode can also be seen in the picture. Black and white images were acquired with an Intensified Frame CCD Camera System made by Cordin Corp. (model 220A). A Pentax 67 x 150 mm telephoto lens with a focus ring to reduce the minimum focus distance was attached to the camera. A simple synchronizing arrangement by means of a delay generator was used to guarantee

126

that images were taken during the steady-state plasma stage, typically at the middle of the microwave pulse. The typical exposure time used in this study was from 1 to 3 μs depending on the emission intensity of the plasma. The MCP and CCD gains can be adjusted depending on the light intensity to optimize the images.

Figure 8.3: Close view of TIPS with the Ultra-Fast Picture Diagnostic Installed.

Fig. 8.4 shows a photograph of the set-up. TIPS is on the left side of the picture and on the right side of the picture it can be seen the ultra-fast camera.

Figure 8.4: Photograph of the experimental set-up.


127

The auxiliary diagnostics include measurements of incident and reflected power signals recorded by bidirectional couplers and a fiber optics collimator is placed in the border of the optical window to acquire time integrated visible spectra via a fiber optics spectrometer.

Light intensity (a.u)

A wide range of experimental conditions were studied using visible spectra. The main advantage of the visible light emission diagnostic over the VUV study carried out in chapter 6 is its simplicity. Visible light can be carried to the spectrometer by a fiber optics. Moreover, the visible spectrometer is smaller and cheaper than the VUV one and it does not require vacuum. This makes possible to use it as a complementary diagnostic as it was done in the experiments described in this chapter, where the spectra were taken at the same time that the ultra-fast pictures by collimator and a fiber optics placed in the quartz window.

(a) Balmer-alpha 656.3 nm Hα

(b) Balmer-beta 486 nm Hβ

(c) Balmer- gamma 434 nm Hγ

(d) Fulcher Band 600nm

Wavelength (nm) Figure 8.5: Typical integrated time spectrum obtained by a fiber optics spectrometer where Balmer Series and Fulcher band are shown.

Fig. 8.5 shows a typical time-integrated visible spectrum obtained from a fiber optic spectrometer placed at the external quartz window during experiments. The Balmer series lines alpha (a) at 656.3 nm, beta (b) at 486.1 nm, gamma (c) at 434.1 nm and the Fulcher band (d) are all clearly indicated in the figure. These series are important to obtain valuable information about the relative ion species composition of plasma. The visible light emission of Hydrogen is due to the excitation of neutral Hydrogen atoms and molecules by electron impact from level q to level p, and the decay into level k by spontaneous emission resulting in line emission pk. The Hydrogen Balmer Series corresponds to the transitions from the excited states of atomic Hydrogen, with

128

the primary quantum number np > 2 to the excited state where nk = 2. Balmeralpha radiation at 656.3 nm is emitted when np = 3, i.e. it is the primary transition within the Balmer series. For Balmer-beta, the corresponding numbers are 486.1 nm and np = 4. The n = 3 and n = 4 states are populated by atomic excitation and dissociative excitation of H2 - molecules by electron impact. The relative importance of these processes depends on the density of neutral atoms and molecules as well as the distribution of electron energy. In low temperature plasmas (T e in the order of 10 eV), the rate for atomic excitation is typically greater by more than one order of magnitude [69]. However, the exact ratio of the two excitation channels depends on the vibrational distribution of the molecules. The Fulcher band emission around 600 nm corresponds to the molecular triplet transition d3 Πu → a3 Σ+ g , with 3 + 3 + a further transition (emission in UV) a Σg → b Σu to a repulsive state (or molecular continuum, see for example ref. [70]). Thus, Fulcher band emission indicates dissociation of the emitting molecule. The radiative lifetimes of the excited states resulting in light emission within the Balmer series and Fulcher band are in the order of 10−7 s and 10−8 s, respectively [71], [70].

8.2 8.2.1

Results General Behavior

Different distributions of visible emitted light have been recorded with this diagnostic under a wide range of experimental conditions. Hydrogen was used with pressures ranging from 3.8 x 10−3 mb to 8.6 x 10−3 mb so as to working conditions the same as in previous chapters. For the same reason, incoming microwave powers were used in the range of 300 W and 1500 W, showing reflected power from the plasma ranging from 1 % to 70 % of the incoming power. Fig. 8.6 shows a pictorial survey of configurations observed during the experiments. All photos were obtained with a single shot at the final steady state reached for the plasma once the microwave power pulse is well stabilized. A simple synchronization scheme was used with a delay generator triggered by the incoming power signal. Note that the images were originally obtained in B&W but have been transformed by software into a representation where red corresponds to the highest intensity (saturated white) and blue to the lowest (deep black). Eight modes were recognized, all of them showing a remarkable reproducibility within the experimental parameters: gas pressure, magnetic field profile and microwave power ratio (incoming/reflected). Note that plasma chamber diameter is marked by a


129

Figure 8.6: Different types of plasma distribution modes observed during the experiments. Images correspond to the final steady state stage reached for the plasma. False color calibration is used where red corresponds to highest intensity and blue the lowest.

thin, dark blue circular line, especially visible in cases (a), (b), (d), (e) and (g). The shadow produced by the fiber optic collimator of the spectrometer can also be observed in the lower left corner of almost all of the pictures. We chose some names to facilitate reference for each pattern as suggested by their shapes: Fig. 8.6 (a) Column, Fig. 8.6 (b) Hour Glass, Fig. 8.6 (c) Slug, Fig. 8.6 (d) Flower, Fig. 8.6 (e) Full Chamber, Fig. 8.6 (f) Ring, Fig. 8.6 (g) Yin-Yang and Fig. 8.6 (h) Donut. The experimental conditions were defined by the magnetic field profile B, neutral gas pressure p and microwave power ratio Pi /Pr (incoming/reflected). Magnetic field profiles were obtained by 2D simulations using FEMM as in previous chapters. Note that the color scale of the B-field 2D maps shown in this chapter has been slightly modified with respect to the previous ones to adapt it to the wide range of different B-field conditions explored in this chapter. In general terms, we have observed that the magnetic field profile is a determining factor for transitions between different patterns while neutral gas pressure and microwave power show a weaker influence. The set of pictures in Fig. 8.6 can be separated into two groups depending on the bulk emission intensity: one high intensity emission set composed of Column, Hour Glass and Slug and another lower emission set with Flower, Full Chamber, Ring, Yin-Yang and Donut. Moreover, the ratios between lines of Balmer series and the ratios between

130

the Fulcher band and the Balmer lines show remarkable differences for some of recorded plasma distributions. In addition, some patterns can produce several kinds of spectra depending on their magnetic field profile while others show the same kind of spectrum for different magnetic profiles. It is remarkable that three of the configurations, Hour Glass, Slug and the stationary Yin-Yang are not symmetric with respect to the plasma chamber axis. For these cases the alignment of the plasma distribution is vertical, as is the direction of the electric field inside the plasma chamber. This suggests a possible relationship between the electric field orientation and the plasma distribution. Fig. 8.7 shows a simulation of the electric field distribution inside the plasma chamber. A lateral and a frontal view are represented in the figure where the limits are marked in solid black line and the axis of the chamber is marked with a broken line in the lateral view. In the frontal view, black arrows are used to indicate direction. The simulation was carried out considering a magnetron power of 1500 W.

Figure 8.7: Electric field distribution inside the plasma chamber.

In the subsections that follow, we describe the experimental conditions under which each plasma pattern distribution was observed, with special emphasis on the magnetic field profile and spectra. The B-field configurations shown in the following cases correspond to experimental situations where plasma behavior is stable and reproducible. Magnetic field distributions falling between such cases keep their shapes but show unstable behavior, mainly characterized by microwave coupling oscillations that are reflected as unstable light emission intensity over the whole time and high jitter during the breakdown process. These intermediate cases were disregarded. Another important aspect to point out is that the change from one determined magnetic field configuration to another can take place through a number of different


131

paths. This is because each profile is obtained by the independent selection of currents in the four coils and the longitudinal position of its pancake holders. So, the transition between the different plasma distributions strongly depends on the way these parameters are modified.

8.2.2

Column Mode Hα

(b)

(a) Hγ

Hβ

}

B-Field

Fulcher

Hβ

1 cm

Hα

(c)

Hγ

}

B-Field

Fulcher

Hα

} Fulcher

(d)

mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

B-Field

Figure 8.8: Column shape recorded with its corresponding visible spectra and magnetic field profiles where this shape is observed.

Fig. 8.8 shows a typical picture, where plasma distribution is concentrated on the axis of the plasma chamber (a), that we denominated Column. A 3D map of light intensity is shown in the lower picture of image (a) and the spectra, and their corresponding magnetic field distributions, are also presented. The magnetic field distribution 2D color maps follow the same representation used in previous chapters. The left side of the plasma chamber is the area from which the microwaves come through the microwave coupler, and the right side is where the transparent electrode (a quartz window between tungsten meshes) is located. Broken, curved black lines represent the position of the ECR surfaces. It is significant that three different spectra were observed in the Column distribution for three different magnetic profiles. The first case corresponds to Fig. 8.8 (b) where

132

the intensity of the magnetic field is relatively high, reaching values of over 120 mT, and one ECR surface can be seen on the left side (the microwave entrance). Note that for this case, the magnetic field is over the value of the ECR in practically the entire chamber volume, showing a strong gradient along the axis. Under these conditions, the light intensity is relatively high showing a bright spot where the Hα line is absolutely dominant as can be seen in the associated spectrum. The case corresponding to Fig. 8.8 (c) shows another experimental situation where the Column is observed with relatively lower magnetic field values in the chamber. Note that values around the ECR dominate the volume of the plasma chamber with two ECR surfaces inside the chamber. The Hα line is still dominant but the ratio between Hα and Hβ has decreased, while Hγ and Fulcher band emissions do not show any changes in the associated spectrum. The last case for this kind of plasma distribution is shown in Fig. 8.8 (d), where the magnetic field profile was characterized by a values of around BECR /2. The total light intensity is notably lower for this experimental situation and the spectrum is radically different to that of the previous cases. The ratio between Fulcher band emission and Hα is practically 1, showing a remarkable amount of molecular ionization activity probably related to a relative increment of H2+ molecular ion density in the plasma.

8.2.3

Hourglass Mode

Fig. 8.9 shows a typical picture with the plasma distribution denominated Hourglass (a), in which a waist centered on the plasma chamber axis with two vertical symmetrical lobes can be seen. A 3D map of light intensity is also presented in the lower picture. The spectrum and its corresponding magnetic field distributions are also visible. Just one kind of spectrum was observed in four different magnetic profile distributions inside the discharge chamber. This spectrum is different to the previous case because although Hα is still dominant, the relative increment of Hβ is remarkable, while Fulcher band does not show any significant change, keeping a low intensity profile. It is also interesting to see how this Hourglass plasma distribution and its characteristic spectrum are maintained in four different magnetic profiles. The first B-field of Fig. 8.9 (b) shows a field map where B is over the ECR value in practically the whole discharge chamber, with the exception of relatively small volumes at the microwave injection and diagnostic side. The case corresponding to Fig. 8.9 (c) shows another experimental situation where B values are under the ECR in almost all of the chamber volume with the exception of two lobes on the left side that form a toroidal shape. The magnetic field distributions of cases (d) and (e) show even lower values for B field with


133

no ERC surfaces and where a relatively high volume with B ' ECR/2 can be observed for (d) at the microwave injection side of the chamber.

(b) (a) mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

(c) Hα Hβ 1 cm

Hγ

(d)

} Fulcher

(e)

B-Field

Figure 8.9: Hourglass shape recorded with its corresponding visible spectrum and magnetic field profiles where this shape is observed.

8.2.4

Slug Mode

Fig. 8.10 (a) shows a typical picture with the plasma distribution that was denominated Slug in which a plasma column with an inclined ‘S ’ shape was recorded. The corresponding 3D map of light intensity is shown in the lower picture of image (a) and the spectrum with its corresponding magnetic field distributions are also presented. Just one kind of spectrum was observed under these two different magnetic profile distributions. The spectrum shows a predominant Hα line but with a remarkable increment in Hβ line emission. This fact also suggests a preeminent H + ion population for this plasma configuration. This interesting distribution was obtained for just two magnetic field profiles, both with relatively intense values for B-field. Fig. 8.10 (b) shows B values over the ECR in nearly all of the chamber volume, with the resonant surface close to the MW injection side. This magnetic field configuration has a characteristic convex shape volume and two

134

symmetric lobes on the diagnostic side to produce a toroidal volume of high intensity (120 mT) close to the chamber wall. The case corresponding to Fig. 8.10 (c) shows a situation with more concentrated values of about 105 mT in the center of the plasma chamber.

(a) mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

(b)

Hβ

1 cm

Hγ

Hα

(c)

} Fulcher

B-Field

Figure 8.10: Slug shape recorded with its corresponding visible spectrum and the magnetic field profile where this shape is observed.

It is notable how this plasma configuration is reproducible in two different magnetic field profiles, showing no differences in the spectra recorded for both cases. Another interesting issue is that this distribution could be interpreted as a stable intermediate stage between the Column and Hourglass modes. Depending on the path chosen to change the magnetic field configuration, it is possible to pass from the Column to the Hourglass via the Slug with a remarkable reproducibility in both directions.

8.2.5

Flower Mode

Fig. 8.11 (a) shows a typical picture with the plasma distribution that was denominated Flower, in which four lobes around chamber axis were observed. The corresponding 3D map of light intensity is shown in the lower picture of image (a) and also the spectrum and the corresponding magnetic field distribution are presented. Just one


135

magnetic field profile was to produce this plasma configuration. Note that the symmetry is characterized by the existence of two kinds of lobes: a pair of horizontally-aligned big ones and two vertically-aligned smaller ones that coincide with the direction of the electric field inside the chamber.

(a)

Hα

1 cm

Hγ

(b)

Hβ

} Fulcher

B-Field

mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

Figure 8.11: Flower shape recorded with its corresponding visible spectrum and magnetic field profiles where this shape is observed.

The spectrum also shows a predominant Hα line with nothing remarkable in the molecular Fulcher band. This fact again suggests a preeminent H + ion population for this plasma configuration. Just one magnetic field distribution was able to produce this plasma distribution. Fig. 8.11 (b) shows B values around the ECR value in the full discharge chamber with two symmetric lobes formed by the resonant surfaces close to the chamber wall.

8.2.6

Full-Chamber Mode

Fig. 8.12 shows a typical picture with the plasma distribution denominated FullChamber (a), in which a practically homogeneous distribution of the plasma emission is observed. The corresponding 3D map of light intensity is shown in the in the lower picture of image (a) along with the spectrum and the corresponding magnetic field distributions. Just one kind of spectrum was recorded in three different magnetic profile

136

distributions. The spectrum shows some increment to Fulcher band emission in spite of the predominant Hα line.

(b) (a)

Hα

1 cm

Hγ

mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

(c)

Hβ

} Fulcher

(d)

B-Field

Figure 8.12: Full-Chamber shape recorded with corresponding visible spectrum and magnetic field profiles where this shape is observed.

Fig. 8.12 (b) shows a magnetic field map where B is around the ECR in practically the whole discharge chamber with a deep bi-convex volume where it reaches values of 95 mT. The case corresponding to Fig. 8.12 (c) shows another experimental situation where the ECR surface is a practically flat disc near to the center of the chamber separating it into two large volumes where the left side corresponds to B ≤ ECR and the right side to B ≥ ECR. The magnetic field distribution of case (d) shows lower B-field values where practically the entire volume of the discharge chamber is filled by B ≤ ECR, with the exception of just two small lobes on the diagnostic side defining a small toroidal volume of ECR field near to the diagnostic side. Case (d) is also interesting because a relatively high volume at the left side of the chamber is embedded in ECR/2.

8.2.7

Ring Mode

Fig. 8.13 (a) shows a typical picture with the plasma distribution denominated Ring in which a plasma layer close to the chamber wall can be seen. The corresponding 3D


137

map of light intensity is shown in the lower picture of image (a) picture along with the spectrum and the magnetic field distributions where this shape was observed. Only one kind of spectrum was observed in these two different magnetic profile distributions. The spectrum shows a predominant Hα line emission without anything remarkable in the molecular Fulcher band.

(a) mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

(b)

Hα

1 cm

Hγ

Hβ (c)

} Fulcher

B-Field

Figure 8.13: Ring shape recorded with its corresponding visible spectrum and magnetic field profiles where this shape is observed.

Fig. 8.13 (b) shows a magnetic field map where B is well over the value of the ECR in the entire discharge chamber with a characteristic convex volume delimited by the resonant surface on the diagnostic side. Values of 120 mT were achieved in the center of the chamber with a symmetric distribution. The case corresponding to Fig. 8.13 (c) shows another experimental situation where the ECR surfaces are outside the plasma chamber and its full volume is established well under the ECR value. It is interesting to see how this plasma configuration is reproducible in two such different magnetic field profiles, showing no differences in the spectra recorded for these extreme cases. Its important to highlight that the Ring distribution shows behavior that is extremely sensitive to any change in the magnetic configuration. The two cases presented here can be considered as stability islands. For this case, the ratio between incoming and reflected power was always very low, with coupling values never higher than 5 %.

138

8.2.8

Yin-Yang Mode

(a)

Hα

1 cm

Hγ

(b)

Hβ

} Fulcher

B-Field

mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

Figure 8.14: Yin-Yang shape recorded with its corresponding visible spectrum and magnetic field profiles where this shape is observed.

Fig. 8.14 (a) shows a typical picture of the plasma distribution denominated YinYang with an asymmetric large lobe followed by a tail. This shape was recorded for only one experimental condition, characterized by the magnetic field profile and the spectrum shown in Fig. 8.14 (b). The corresponding 3D map of light intensity is shown in the lower picture of image (a). Only one magnetic field profile has been able to produce this plasma configuration. The spectrum shows a predominant Hα line with a remarkable increment in the molecular Fulcher band that may also reflect an increment in the H2+ population. Note that the magnetic field distribution is well under the ECR value throughout the chamber volume showing a smooth positive gradient in the direction of the diagnostic side with maximum values of around 75 mT at the center and 80 mT on the external border.


8.2.9

139

Donut Mode

Fig. 8.15 (a) shows a typical picture of the plasma distribution denominated Donut where a thick plasma layer separated from the chamber wall were recorded. As in the previous case, only one kind of spectra and one magnetic profile embedding the chamber were observed as the experimental conditions to produce this kind of structure, they are shown in Fig. 8.15 (b). The corresponding 3D map of light intensity is shown in the lower picture of image (a).

(a)

Hα

1 cm

Hγ

(b)

Hβ

} Fulcher

B-Field

mT 120 114 108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

Figure 8.15: Donut shape recorded with its corresponding visible spectrum and magnetic field profiles where this shape is observed.

The spectrum shows a predominant Hα line with an increment in the molecular Fulcher band that may also reflect an increment in the H2+ population. Note that the magnetic field distribution is characterized by a strong gradient in the z axis, from low values at the microwave injection side of typically 50 mT to values reaching 120 mT at the diagnostic side. Equally noticeable is the the strong asymmetry respect to plasma chamber center.

140

8.3


The plasma diagnostics described in this chapter have been shown to be a valuable tool in determining the plasma configurations in ECR ion sources like TIPS. Eight new stable plasma density distribution modes were found. The combination of a high speed photography study under a wide range of experimental conditions with measurements of time-spatial integrated visible spectra helps us to understand the relationship between plasma distribution patterns and ionization/recombination processes. According to Refs. [70] and [71], the radiative lifetimes of the excited states resulting to light emission within the Balmer series and Fulcher band are on the order of 10−7 s and 10−8 s, respectively. The typical speed of the Hydrogen atoms and/or molecules can be estimated from their temperature (kinetic energy). It is assumed that the maximum temperature of the neutral population is on the order of 1 eV. This value, corresponding to typical ion temperature in the plasma, can be used as a superior limit for the neutrals. This translates to rms-speed of 104 m/s (order of magnitude). Therefore, the maximum distance governed by the neutral particles between the collisional excitation and subsequent light emission is on the order of 10−3 m. Actually, the energy distribution of the neutral particles is much narrower but estimating the maximum rms-speed/distance serves the purpose in this case. The evident conclusion is that the spatial light emission profile corresponds to the spatial distribution of excitation collisions i.e. distribution of energetic electrons and ionization. Furthermore, the distribution of Fulcher band emission indicates the spatial distribution of molecular dissociation. Magnetic field configuration appears to be the most important parameter to produce changes between the different shapes, whereas neutral gas pressure shows a weak influence. The ratio between incoming/reflected microwave power is self-adjusted during experimental conditions where plasma shows stable behavior, showing two typical cases: high coupling of 75 % and low coupling of less than 5 % both of which were observed in connection with the bulk light intensity emitted by the plasma. Looking at the pictorial survey shown in Fig. 8.6, in general terms, plasma distributions can be separated into two groups: a high-brightness group characterized by a relatively high intensity of bulk light emission composed of the Column, Hourglass and Slug modes and a low-brightness group composed of the Flower, Full-Chamber, Ring, Yin-Yang and Donut modes. High microwave coupling factors of 75-90 % were recorded for the high-brightness cases, showing at the same time a high ratio between Hβ and Fulcher emission as can be seen on the spectra. Conversely, low microwave couplings of 5 % or less were always recorded for the low-brightness cases with a inversion in the


141

ratio between Hβ and Fulcher. This interesting behavior may be closely related to major changes in the ion population ratios of the plasma. Surely the most interesting mode, due to its obvious applications in ECRIS, is the Column, where plasma is fully concentrated on the chamber axis. This case shows interesting possibilities because of the strong dependence of its plasma characteristics on the applied magnetic field configuration. As can be seen in Fig. 8.8, we found three stable Column mode cases, with different characteristic spectral emissions. The first two cases, represented in Fig. 8.8 (a) and (b), correspond to a high-brightness emission, as previously described, with a high corresponding microwave coupling between 80-90 % and spectra showing a significant component of Balmer lines, as is usual in Hydrogen plasmas with a relatively high population of H+ . However, Fig. 8.8 (c), where a low intensity magnetic field configuration is used, shows a remarkable spectrum where Fulcher band reaches a very high relative intensity, suggesting a drastic change in the ion species composition. Low microwave coupling and also low bulk light emission were recorded for this case, which falls into the category of low-brightness. Finally, the existence of several stable modes of plasma density distribution in a 2.45 GHz Hydrogen plasma source was demonstrated to be mainly driven by the magnetic field configuration where the plasma is embedded. Some connections between modes and visible spectra were reported as fingerprints related with the ratios of the different ions present in the plasma. These spectra show that some degree of tuning on ion species output could be possible in ECRIS devices by modifying the experimental conditions with special emphasis on the magnetic field. Future plans include a more-in-depth research into density distributions in the chamber volume by Langmuir probe measurements and also the implementation of a Wien Filter diagnostic now under testing stage. This device is able to extract an ion beam sample from the plasma and measure the currents of H+ , + H+ 2 and H3 and its temporal evolution.

Chapter 9

Plasma Breakdown Evolution through Ultra-Fast Pictures This chapter presents further results obtained with the ultra-fast picture plasma diagnostics described in the previous chapter. The high temporal resolution of the Cordin 220A Image-Intensified CCD frame camera allows the study of the plasma evolution during transients for all the different plasma configurations found in the previous chapter. Only a few studies probing the spatial and/or temporal light emission profiles of ECR-heated plasmas have so far been conducted [67, 68, 72]. Here we present an overview of the breakdown process for the different plasma configurations. The CCD camera is capable of taking up to four different pictures in a single shot with a minimum delay between frames of 100 ns. The evolution of the 2D plasma spatial distribution during breakdown was studied for the eight different configurations described in the previous chapter. Single shot sets of ultra-fast pictures for full visible light, Balmer-alpha, Balmer-beta and Fulcher band are presented for the high-brightness cases: Column, Hourglass and Slug. The low emission levels for the cases we have called low-brightness, i.e., Full-Chamber, Ring, Flower, YinYang and Donut, especially during first stage of plasma breakdown, makes it impossible to take pictures with the filters using short enough exposure times. Two rotating plasma configurations were discovered. This phenomenon, which has never been reported before in these kinds of plasma source, is very stable and reproducible.

143

144

9.1

Experimental Set-Up

Although the same set-up was used in previous chapter, in the experiments presented in this chapter the timing was of special importance. Careful synchronization between the plasma breakdown process and the high speed camera trigger required a timing arrangement similar to the one used in chapters 3, 4 and 5. Fig. 9.1 shows the schematics of the experimental set-up and timing system.


Tuner

Dual Quartz directional window coupler MCPs Intensified Gated Multiframe CCD Camera

Plasma MAGNETRON 2.45 GHz PULSED MODE Pump Coils

OSCILLOSCOPE CH1

CH2

CH3

CH4

Photo detector

Fiber optics

Synchrony pulses Monitor

COMPUTER


Single shot trigger pulse

Figure 9.1: Schematic representation of the diagnostics set-up for the ultra-fast pictures synchronization during breakdown.

The set-up enabled us to take up to four images with independently adjusted exposure times and delays between each shot in synchronization with the breakdown process. The typical exposure time used in this study was 1 μs for the high-brightness cases (Column, Hourglass and Slug) and 2 μs for the low-brightness ones (Full-Chamber, Ring, Flower, Yin-Yang and Donut). The delay between frames depends on the plasma dynamics and is always related to the rising edge of the plasma generator incoming power signal. It is written below each picture shown in the following sections. The MCP and CCD gains were adjusted depending on the light intensity to optimize the images taken through different filters at various stages of the microwave pulse.

Chapter 9. Plasma Breakdown Evolution through Ultra-Fast Pictures

145

It should be noted that the independent settings for the MCP and CCD of each picture made it necessary to perform a calibration in order to compare the images. The images were transformed through the calibration into the same representation previously used in chapter 8, where red corresponds to the highest intensity (saturated white on black and white gamma) and blue, the lowest (deep black). The auxiliary diagnostics included the measurement of incident and reflected power signals recorded by bidirectional couplers. The time-resolved light emission was also obtained through a fiber optics placed at the periphery of the optical window and connected to a fast photodiode. The set-up allowed bandpass filters to be mounted in front of the CCD camera and the photodiode in order to study (nearly) monochromatic light emission in the visible light range.

Figure 9.2: View of a synchronism record during the breakdown process: (a) is the incoming power, (b) is the reflected power and (c) is the plasma emitted visible light signal and (d) shows the high speed camera frame synchronization pulses.

The camera trigger pulse signal from its output monitor was also recorded in the scope, which permitted us to know the exact instant when each picture was taken with respect to the experiment’s characteristic signals (incoming power, reflected power, and emitted light). Fig 9.2 shows a typical synchronism record taken during the experiments. Following the same representation and color code used in previous chapters, (a) is the incoming power, (b) is the reflected power, (c) is the plasma emitted visible light signal and (d) shows the high speed camera frame synchronization pulses from its monitor output. Each pulse corresponds with the instant when its picture was taken.

146

9.2

Results

The main purpose of this chapter is to demonstrate the temporal capabilities of the diagnostics set-up from the point of view of temporal resolution and to discuss the interpretation of the images. For this reason, only a representative sample of the measurement results for each plasma configuration is presented in this section. To fully characterize the breakdown evolution of each case, parametric studies will have to be conducted under different working conditions for each of them. This task is included in the group’s working plans for the near future. Visible light (390-700 nm), Balmer-alpha (656.3 nm), Balmer-beta (486.1 nm) and Fulcher band (around 600 nm) bandpass filters (Thorlabs models FB660-10, FL488-10 and FB600-40) were placed inside the camera photo-lens to take filtered pictures of the breakdown process. These sets of pictures show the evolution of the atomic and molecular ionization processes. Unfortunately, filtered pictures could only be taken in the high-brightness modes. The long exposure time necessary to take filtered pictures of the low-brightness cases makes it impossible to obtain well-defined images to study the evolution of plasma distributions during breakdown. Moreover, the photodiode signals for high-brightness cases corresponding to Visible light, Balmer-alpha, Balmer-beta and Fulcher band were also recorded. It is notable that the evolution of the atomic emissions (Balmer-alpha and Balmer-beta) and the molecular emission (Fulcher band) presented opposite trends. While Balmer-alpha and Balmer-beta showed a tendency to increase during the breakdown process, Fulcher band presented an early peak and after that, a decay toward the steady state. This will be further developed in sections 9.2.1, 9.2.2 and 9.2.3 through Figs. 9.5, 9.7 and 9.9. The opposite evolution in the Balmer lines and Fulcher band is completely in line with the behavior of the VUV Lyman-alpha and Lyman-band emissions described in chapter 6 and shown in Fig. 6.8. This correspondence between ultra-violet and visible light behaviors supports the idea that measurement of the latter is a valuable tool to study the atomic and molecular evolution of the plasma. However, it should be mentioned that the breakdown light peak shown in the Lyman-alpha and Lyman-band lines (Fig. 6.8) can not be seen in the visible line record, this is probably due to the high emission level reached by those lines during the steady state. This steady state value exceeds the level of the peak and, then, it remains integrated within the overall pulse shape. This is confirmed by the fact that the breakdown light peak has only been observed under the experimental conditions yielding in very low light steady state emission profiles. As an example, Fig. 9.3 shows the scope record of one of those low emission cases where the visible light peak during


147

the breakdown is clearly seen. Curve (a) is the incoming power, curve (b) the reflected power and curve (c) the total visible light measured by a photodiode.

Figure 9.3: Scope record showing a peak of visible light during the breakdown. Incoming power (a), reflected power (b) and total visible light emission.

It was also found that at least two of the plasma modes can rotate inside the plasma chamber with frequencies that vary between 5 and 50 KHz depending on the working conditions. Pressure and absorbed power seem to be the parameters with the greatest impact on the frequency of the rotating phenomena.

9.2.1

Column Mode

Fig. 9.4 shows a set of pictures taken at different stages of the breakdown process of a Column shape Hydrogen discharge for visible emission range, Balmer-alpha, Balmerbeta and Fulcher band by using bandpass filters. The corresponding operational parameters were 2100 W incident power at 50 Hz / 10 % duty factor, 3.8 x 10−3 mb neutral gas pressure and a B-field corresponding to case (d) in Fig. 8.8 in the previous chapter. The set of pictures is divided into two groups separated by a dashed vertical line. The group on the left is obtained at the very beginning of the plasma breakdown, when the corresponding intensity of light emission is relatively low. The second set of pictures on the right is obtained at a later stage of the discharge when the light intensity is significantly higher. Thus, the two groups of pictures were measured on different pulses using different MCP gains to avoid overexposure or excessively dark pictures. This fact means that it is not possible to compare the intensities of the false color images between

148

the groups. Exposure times were set to 1 μs for the full visible light set, to 3 μs for the Balmer-alpha and Balmer-beta series and 2 μs for the Fulcher band series.

Figure 9.4: Normalized CCD-MCP images of Column breakdown evolution: visible light (a), Balmer-alpha (b), Balmer-beta (c) and Fulcher band (d) taken at different stages of the plasma breakdown. The times given are measured from the leading edge of the microwave pulse.

The light emission profile is a circular shape in the plasma chamber axis. The evolution of the morphology is similar for the Balmer-alpha, Balmer-beta and Fulcher band sets, starting from a small circle in the center and then growing in diameter. The final dimensions are reached after 270 μs when the seventh picture was taken. Although in each set of four pictures the MCP gains have been set to reach the best possible image quality, it is very interesting to analyze the breakdown process dynamics from the point of view of the way that the different visible lines evolve. Fig. 9.5 shows the evolution of the normalized light signals through filters corresponding to this high-brightness Column case. Curve (a) corresponds to visible emission range, curve (b) to Balmer-alpha, curve (c) to Balmer-beta and curve (d) to Fulcher band. The temporal scale was been shifted by 250 μs for visual reasons. Both Fig. 9.4 and Fig. 9.5 clearly show how the Balmer-alpha and Balmer-beta lines present a very similar evolution. In Fig. 9.5 it can be seen that both curves show a slowly increasing behavior towards the steady state. It takes the full length of the pulse


149

(2 ms) for these two Balmer lines to reach the steady state. This behavior matches the high-speed frames evolution of Fig. 9.4.

(

)

(a) (b) (c) (d)

( ) Figure 9.5: Normalized Column mode photodiode signals corresponding to visible light (a), Balmer-alpha (b), Balmer-beta (c) and Fulcher band (d). The temporal scale was shifted by 250 μs for visual reasons.

On the other hand, the Fulcher band signal shows a completely opposite behavior. It rises to the maximum level in approximately 60 μs and starts a slow decay. In the Fulcher set of pictures shown in Fig. 9.4, the maximum brightness is situated between the fourth and the fifth pictures. Unfortunately, this occurs just in the middle of the two sets and, consequently can not be determined precisely. Regarding the Fulcher curve in 9.5, the same evolution is observed: fast growth up to the maximum in less than 60 μs and then a softly decreasing behavior during the whole pulse length. Both fast pictures and photodiode signals show that molecular ionization associated to Fulcher band emission presents a fast breakdown. It reaches its maximum in a few microseconds and then starts a slow decay toward the steady state. In contrast, the atomic ionization, related to Balmer-alpha and Balmer-beta emission, shows a different behavior, evolving gradually towards its maximum during the steady state of the pulse.

9.2.2

Hourglass Mode

Fig. 9.6 shows the Hourglass morphology evolution following the same representation as in the previous case. The working conditions for this study were 2700 W input power at 50 Hz / 10 % duty factor, 8.6 x 10−3 mb, and the magnetic field corresponding with Fig. 8.9 (d) in the previous chapter. Exposure times were set to 1 μs for the full visible light and Balmer-alpha sets, to 2 μs for the Balmer-beta series and 3 μs for the

150

Fulcher band one. Again the dashed vertical line separates sets of pictures that were taken with the CCD and MCP gains adapted to the light intensity of the process. As stated in the previous section, this makes it impossible to establish quantitative comparisons regarding light intensity between different sets. However, it can clearly be seen how the atomic emission (Balmer-alpha and Balmer-beta) starts off weak and then the grows slowly. On the other hand, the molecular emission (Fulcher) reaches its maximum intensity in approximately 30 μs and then decays.

Figure 9.6: Normalized CCD-MCP images of Hourglass breakdown evolution: visible light (a), Balmer-alpha (b), Balmer-beta (c) and Fulcher band (d) taken at different stages of the plasma breakdown. The times given are measured from the leading edge of the microwave pulse.

It is remarkable how the first picture of each series shows that ionization starts off weak in two areas, one in the center of the chamber and another toward the left side. The second picture, 20 μs from the incoming power pulse rising edge, shows a more homogeneous, very weak circular region of ionization filling almost the whole chamber with some brighter areas in the center. The third picture (30 μs) in every series already shows two bright areas from which the Hourglass will grow. In picture number four (40 μs), the lobes are already formed in alignment with the electric field (vertical direction). The second set of pictures in each series shows the evolution of these two lobes, growing and gaining in intensity towards the steady state.


151

Three observations can be made: (a) it takes only a few tens of μs to reach the light emission profile, (b) the light emission profile is similar for atomic (Balmer lines) and molecular emission (Fulcher band) and (c) the light emission profile is not homogeneous, but is instead extended along the direction of the microwave electric field polarization. Following the plasma breakdown, the atomic and molecular emission intensities follow opposing trends. The atomic emission intensity increases while molecular emission intensity decreases towards the end of the pulse.

(

)

(a) (b) (c) (d)

( ) Figure 9.7: Normalized Hourglass mode photodiode signals corresponding to visible light (a), Balmer-alpha (b), Balmer-beta (c) and Fulcher band (d). The temporal scale has been shifted by 250 μs for visual reasons.

Fig. 9.7 shows the evolution of the normalized light signals through filters corresponding to this high-brightness Hourglass stable plasma mode. The evolution of the normalized signals Balmer-alpha (b), Balmer-beta (c) and Fulcher (d) is very similar to the previous Column case. However, the shape of the total visible light (a) matches almost perfectly with the Balmer-alpha curve shape. This indicates that the influence of Fulcher emission on the total visible light is low and we can thus conclude that the molecular ionization is low in these cases and, overall, lower than in the Column mode, where the influence of Fulcher on the full visible light signal was noticeable.

9.2.3

Slug Mode

Fig. 9.8 shows the breakdown process evolution of the Slug plasma mode. The high brightness of this case allowed for a full study with filters. The working conditions for these experiments were set to 750 W input power at 50 Hz / 10 % duty factor and 3.8 x 10 −3 mb Hydrogen pressure with the magnetic field profile of Fig. 8.10 (b). Exposure

152

times were set to 1 μs for the full visible light and Balmer-alpha sets and to 2 μs for those of the Balmer-beta series and the Fulcher band.

Figure 9.8: Normalized CCD-MCP images of Slug breakdown evolution: visible light (a), Balmer-alpha (b), Balmer-beta (c) and Fulcher band (d) taken at different stages of the plasma breakdown. The times given are measured from the leading edge of the microwave pulse.

Regarding the light intensity time evolution, from the full visible light series we can highlight the long time needed for this mode to reach the steady state morphology (170 μs). Focusing on the filter series, it is clearly seen how slow is the evolution of atomic ionization (Balmer-alpha and Balmer-beta series). In contrast, molecular ionization associated with the Fulcher series shows a fast evolution towards the maximum, reached in the second frame (30 μs), and, then, a slow decay. This behavior is similar to the Column mode previously described in this chapter. The morphology of the Slug mode evolves in a similar way in the four series shown although intensity evolution is different. Therefore, it can be understood that atomic and molecular ionization occur in the same areas but with different characteristic times. The first picture (30 μs) shows a weak circular ionization area in the center of the chamber in the full visible light and Fulcher band series. However, nothing can be seen in the Balmer-alpha and Balmer-beta series. The second picture of each series (40 μs) shows a small circular area in the center of the chamber. The evolution of this circular region becomes bigger in the vertical direction in the third and fourth images, 50 and


153

60 μs respectively. The fifth picture (70 μs) is the first one to show clearly in all the series the ‘S’ -like shape that characterizes the Slug mode. In picture number six (170 μs) the final Slug shape is already formed and in the next two pictures (270 and 370 μs respectively) it gains intensity without any morphological modification.

(

)

(a) (b) (c) (d)

( ) Figure 9.9: Normalized Slug mode photodiode signals corresponding to visible light (a), Balmer-alpha (b), Balmer-beta (c) and Fulcher band (d). The temporal scale has been shifted by 200 μs for visual reasons.

Fig. 9.9 shows the evolution of the normalized light signals through filters corresponding to this high-brightness Slug case. The filter line evolution is similar to that of previous cases: Balmer-alpha (a) and Balmer (b) show a slowly growing behavior while Fulcher band (d) peaks at the early instants of breakdown and then decays. The shape of the full visible light (a) can be used to give an idea of the influence of atomic and molecular ionization.

9.2.4

Flower Mode

The Flower mode is a stable plasma configuration with four lobes, two vertical small ones and two horizontal large ones. All of them are curved counterclockwise. The working conditions for the study shown in this section are 1500 W input MW power at 50 Hz / 10 % duty factor and 3.8 x 1016 mb Hydrogen pressure. The magnetic field profile necessary to produce this mode corresponds with the one shown in Fig. 8.11. Exposure times were set to 2 μs. Fig. 9.10 shows the breakdown evolution of this mode in two sets of ultra-fast images taken during the breakdown process. The first noticeable characteristic of this breakdown process is that it is very long. It takes more than one 1 ms of evolution for

154

the flower shape to be fully defined. In fact, this is the longest breakdown process of all of the cases described in this chapter.

Figure 9.10: Normalized CCD-MCP images of the Flower breakdown evolution. The times given are measured from the leading edge of the microwave pulse.

The first picture (8 μs) shows a nearly homogeneous ionization in the whole chamber volume. In pictures two (10 μs) to five (50 μs), no well defined morphology can be appreciated but some dark areas do appear. The sixth picture (550 μs) is the first one where the flower shape can be seen albeit very weakly. In picture number seven (1050 μs) the Flower mode is fully developed and in the last picture (1550 μs) an increase in the light intensity is observed.

9.2.5

Full-Chamber Mode

The Full-Chamber mode is one of the low-brightness modes. For this reason, it was impossible to record quality pictures with filters during breakdown. Following the representation used in previous cases, Fig. 9.11 shows the total visible light evolution in two sets of four images divided into two groups and separated by a vertical dashed line. For this study, working conditions were set to 1950 W input power at 50 Hz / 10 % duty factor, 3.8 x 10−3 mb Hydrogen pressure and the magnetic field distribution shown in Fig. 8.12 (c). Pictures were taken with 2 μs exposure time.

Figure 9.11: Normalized CCD-MCP images of Full-Chamber mode breakdown evolution. The times given are measured from the leading edge of the microwave pulse.

It is highly remarkable how fast the breakdown process of the Full-Chamber plasma mode is under this set of working conditions. The final morphology is reached in approximately 6 μs. The evolution starts from a small circular shape in the center of the


155

chamber that can barely be seen in the first picture. In the second picture, this shape is brighter and better defined. 2 μs later, in the third picture, the ionization has extended to the full chamber, although it is very weak. Finally, from the fourth picture on, the shape of the Full-Chamber has completely evolved and no significant changes can be appreciated between pictures. In some of the images two dark circles can be detected on both sides of the chamber center. They correspond with the holes made for pumping reasons in the Boron Nitride disc placed on the MW injection side of the plasma chamber. This fast breakdown process may have some relationship with high molecular ionization because Fulcher emission always has a faster evolution than Balmer-alpha and Balmer-beta lines. However, it is important to take into account that the exact mechanism producing protons in this kind of plasma source is still unknown.

9.2.6

Ring Mode

The Ring plasma mode breakdown process is shown in Fig. 9.12 following representation scheme of previous cases. Only full visible light images are shown due to the low brightness of this case. The working conditions on this study were set to 1950 W input power at 50 Hz / 10 % duty factor and 3.8 x 10−3 mb with the magnetic field profile shown in 8.13 (a) of the previous chapter. Pictures were taken with 2 μs exposure time.

Figure 9.12: Normalized CCD-MCP images of Ring breakdown mode evolution. The times given are measured from the leading edge of the microwave pulse.

This breakdown evolution is probably the most unexpected one of all those studied in the laboratory. It is significant that both the reproducibility and the pulse-to-pulse stability were quite high. As can be seen in the first image (6 μs), the ionization starts in the center of the plasma chamber in a vertical low brightness area. Then, in the second frame (8 μs), this region grows vertically ant becomes thinner in the middle. The third picture (10 μs) shows a high brightness morphology, the vertical part acquires an ‘S’ shape and the vertical ends grow following the edge of the chamber to form four lobes. In the fourth picture (12 μs), the vertical structure starts to fade while the lobes keep growing. The light emission drop between this image and the previous one is notable. 4

156

μs later, the fifth picture shows that the vertical structure has almost disappeared and the radial lobes have intersected to produce the ring shape. The final ring configuration is reached in the seventh frame (20 μs).

9.2.7

Yin-Yang Mode

The Yin-Yang plasma mode is a low-brightness case. The working conditions for the study presented in this section were 1200 W input power at 50 Hz / 10 % duty factor and 4.3 x 10−3 mb. The magnetic field profile corresponds with the one shown in Fig. 8.14 of the previous chapter. Exposure times were set to 2 μs. Fig. 9.13 shows two sets of four pictures taken during the breakdown process. Long exposure times were required to take good quality pictures in terms of intensity. On the other hand, excessively long exposure times were not useful for this time resolved study. A 2 μs exposure time was finally chosen as a compromise between image quality and time resolution.

Figure 9.13: Normalized CCD-MCP images of Yin-Yang breakdown evolution. The times given are measured from the leading edge of the microwave pulse.

As can be seen in the first picture (5 μs), the ionization starts on the left side of the chamber axis. This region grows across the next four pictures and in the sixth image (130 μs), the tail of the Yin-Yang shape can be appreciated for the first time. In image number seven (230 μs) the Yin-Yang shape is nearly formed and in the picture number eight (330 μs) the shape is completely formed.

9.2.8

Donut Mode

The low-brightness of the Donut mode, especially during breakdown, makes it impossible to take a set of pictures early enough to appreciate the morphology evolution. However, a series of images is shown in this chapter for the sake of completeness. The exposure time used was 3 μs. The working conditions in this experiment were set to 2100 W MW input power at 50 Hz / 10 % duty factor and 8.6 x 10−3 mb Hydrogen


157

VISIBLE

pressure. The magnetic field profile corresponding to this study was the one shown in Fig. 8.15.

Figure 9.14: Normalized CCD-MCP images of Donut breakdown evolution. The times given are measured from the leading edge of the microwave pulse.

9.2.9

Rotating Plasma Configurations

Two plasma configurations were observed to present a rotating behavior with respect to the plasma chamber axis. One of them, the Yin-Yang, can also be static, as described in the previously, while the other, which we have called Half-moon, has only been observed rotating. Both configurations show high levels of reproducibility and stability. Equally significant were the high pulse-to-pulse reproducibility and the low jitter observed. These phenomena were first noticed due to a periodic oscillation present in the visible emitted light signal pulse of the plasma. A typical scope corresponding to a rotating plasma condition can be seen in Fig. 9.15 where curve (a) is the incoming power measured in the second bidirectional coupler, curve (b) is the reflected power, also measured in the second bidirectional coupler and curve (c) is the photodiode signal of the plasma’s visible emitted light signal. In particular, the scope record shown in Fig. 9.15 corresponds to the rotating Half-Moon configuration that will be detailed later in this section. It is important to remember that the fiber optics collecting the plasma emitted light are placed on one side of the standard quartz window (corresponding to letter (e) on Fig. 8.1). This eccentric positioning ensures that any rotating plasma configuration causes periodic oscillations in the light signal. On occasions, the light signals can present oscillations in cases of non-rotating plasmas, but if these oscillations are not also present in incoming and reflected power signals, they indicate the presence of a rotating plasma condition. Rotating conditions are extremely sensitive to changes in working parameters, and especially to changes in the magnetic field profile. Changes of 100 mA on one of the coil currents can completely change the plasma configurations. The Hydrogen pressure

158

and the magnetron power, however, can be changed within a reasonable range while still sustaining the rotating behavior. Pressure and power strongly changes affect the rotating frequency that can vary within a range that goes from approximately 5 Hz to 50 Hz. Both cases show a poor coupling behavior with very low absorbed power between 2 and 10 %. Light emission is also low and, in consequence no filtered light study was carried out.

Figure 9.15: Typical rotating plasma scope record: incoming power (a), reflected power (b) and visible emitted light (c).

It is worth noting that, even if the magnetic field is always pointing in the same direction (parallel to the chamber axis and pointing toward the MW injection side), the plasma configurations rotate in opposite directions. While the Half-Moon rotates clockwise, the Yin-Yang rotates counterclockwise. It was observed that if the B-field direction is inverted the rotation direction is inverted too. In Eq. 9.1 vd is the drift velocity for a particle of charge q immersed in a magnetic field B and subjected to an external force F . From the equation we know that if the magnetic field is inverted the drift velocity would also invert its direction. The fact that in the observed plasma rotating phenomena the rotation changes its direction when the magnetic field is inverted suggests that this rotation can be produced by a drift phenomenon although the force producing it is still unknown.

v~d =

~ F~ xB ~ 2 q|B|

(9.1)


9.2.9.1

159

Rotating Yin-Yang Mode

The rotating Yin-Yang was the first rotating configuration observed. It is produced by a relatively low asymmetric magnetic field profile where no resonant surface is present inside the plasma chamber. Fig. 9.16 shows a simulated 2D map of this magnetic field configuration. Following the usual representation scheme, the left side corresponds with the MW injection side while the right corresponds with the diagnostics port side. The magnetic field presents a strong gradient towards the diagnostic port side, going from less than 30 mT on the MW wave injection side to 90 mT on the opposite one. The intensity of the magnetic field is similar to that which produces the standing Yin-Yang shape represented in Fig. 8.14. However, the gradient is stronger. mT 120 114

r z

108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

Figure 9.16: Rotating plasma mode magnetic field profile.

The upper image in Fig. 9.13 shows a record of rotating Yin-Yang incoming power (a), reflected power (b), visible light emission (c) and high speed camera trigger pulses (d). The lower image shows a set of four 3 μs exposure time pictures corresponding with the trigger pulses of curve (d), where the rotating behavior can be seen. The edge of the fiber optic holder position has been marked in the pictures with dashed white line. The working conditions corresponding with these measurements are: 8.6 x 10−3 mb Hydrogen pressure and 2700 W magnetron power at 50 Hz / 10 % duty factor.

160

Figure 9.17: Yin-Yang Mode rotating behavior. Upper image: Yin-Yang mode rotating plasma typical scope signals (incoming power (a), reflected power (b), visible emitted light (c) and high speed camera trigger signal (d). Lower images: Set of four 3 μs exposure time pictures taken in the instants marked by upper picture signal (d) pulses.

9.2.9.2

Rotating Half-Moon Mode

The rotating Half-Moon shape is produced by a completely symmetric magnetic field profile. A simulated 2D map is shown in Fig. 9.18 following the usual representation scheme. Two resonant surfaces can be seen inside the plasma chamber. They form in the center of the chamber a bi-convex volume where the magnetic field is over the ECR value while at both ends of the chamber the values are below it. The range of magnetic field values present inside the chamber is very narrow compared with the previous Yin-Yang case, going from 90 mT to 110 mT. The upper image of Fig. 9.19 shows a record of rotating Half-Moon incoming power (a), reflected power (b), visible light emission (c) and high speed camera trigger pulses (d). The lower image shows a set of four 3 μs exposure time pictures corresponding with the trigger pulses of curve (d), where the rotating behavior of the plasma can be seen. The edge of the fiber optic holder position has been marked in the pictures with dashed white line. The working conditions corresponding with these measurements are: 3.1 x 10−3 mb Hydrogen pressure and 1000 W magnetron power at 50 Hz / 10 % duty factor.


161 mT 120 114

r z

108 102 96 90 84 78 72 66 60 54 48 42 36 30 24 18 12 6 0

Figure 9.18: Rotating Half-Moon plasma mode magnetic field profile.

Figure 9.19: Rotating Half-Moon mode behavior. Upper image: Typical Half-moon mode rotating plasma scope signals (incoming power (a), reflected power (b), visible emitted light (c) and high speed camera trigger signal (d). Lower images: Set of four 3 μs exposure time pictures taken in the instants marked the by upper picture signal pulses (d).

It is interesting how different the working conditions are to produce the rotating Half-Moon with respect to those of the rotating Yin-Yang. While the Yin-Yang magnetic

162

field profile is strongly asymmetric and relatively low, the Half-Moon one is higher but perfectly symmetric with respect to the chamber center. At the same time, the HalfMoon working pressure is three times smaller than that of the Yin-Yang. As for the absorbed power, it is higher in the case of the Half-Moon (≈10 %) than in the Yin-Yang case, where it is well below 2 %.

9.3


The initial results presented in this chapter demonstrate the capabilities of the diagnostics set-up, which is ideal for studying the microwave-plasma coupling and physics of the discharge breakdown (or decay). The photodiode coupled with Balmer series and Fulcher band filters is an extremely useful diagnostic for ion sources intended for proton beam production as the temporal signals of the atomic and molecular light emission could be used to determine the species fraction of the extracted beam without separation of the charged particles downstream in the beam line. Such diagnostics could significantly reduce the time required for ion source tuning and troubleshooting. These arguments are especially well supported by the recently published results of Y. Xu et al. from Peking University. In these experiments, time resolved beam ion + current for H+ , H+ 2 and H3 species was measured in a 2.45 GHz Ion Source of similar characteristics to that of TIPS. The results published by the authors in Ref. [3] are reproduced in Fig. 9.20. The temporal evolution of the ion currents shows a noticeable similarity to the photodiode signals of Figs. 9.5, 9.7 and 9.9.

H

+ +

H2 + H3

+ Figure 9.20: Temporal evolution for H+ , H+ 2 and H3 measured by Y. Xu et al. in a 2.45 GHz ECR Ion Source at Peking University [3]


163

Fig. 9.21 shows a typical record of Balmer-alpha and Fulcher-band photodiode signals recorded during the experiments. The similarities between atomic ion H+ and + molecular ions H+ 2 / H3 with Balmer-alpha and Fulcher band respectively are very significant. It offers perhaps the strongest proof in favor of the representativity of Balmeralpha as an indicator of atomic ionization and Fulcher band for the molecular equivalent.

Figure 9.21: Typical Balmer-alpha (a) and Fulcher band (b) photodioide signals.

The Wien Filter under development in our laboratory has been designed to leave a line-of-sight that allows us to measure photodiode signals of the Balmer series and Fulcher band plasma emissions while the beam is extracted. These experiments will be critical to determine how Balmer-alpha, Balmer-beta and Fulcher band measurements can be used to determine the species fraction in the plasma, and, in consequence in the ion beam.

Chapter 10

Conclusions 10.1

Work Overview

As Ian G. Brawn states in his book “The Physics and Technology of Ion Sources” [1], the physics of an ion source is largely plasma physics. The aim of the work described throughout this thesis is to develop of plasma diagnostic tools that can provide a deeper knowledge of ECR ion source plasmas. Different diagnostics were developed and implemented in the ECR plasma source that we called TIPS, including electrical biased probes, Langmuir probe density and temperature measurements, VUV spectroscopy, optical spectroscopy, monochromatic time-resolved light measurements and ultra-fast pictures. The design, the set-up and the results obtained for each of them have all been presented in this thesis. Other secondary measurements as signals of incoming and reflected power and plasma integrated emitted visible light were used throughout the studies as fingerprints to check plasma reproducibility. The knowledge acquired on plasma source performance and its plasma characteristics led us to study the microwave driver system. This study was the starting point for an optimized design for the set composed of the plasma chamber and the microwave coupler, taking a different engineering approach that was shown to produce better plasma stability and up to four times higher electron density. ECR ion sources are commonly used in both scientific and industrial areas, which makes its optimization of great interest for ion source designers and users. Nowadays many of these kinds of sources are used in pulsed mode, and a lot of them take advantage of two phenomena associated with breakdown and decay transients: the preglow [13, 16, 17, 44, 45, 73] and the afterglow [12–14, 74–76]. As a consequence, a critical issue in the 165

166

design of all the plasma diagnostics developed was to provide them with time resolution. This gave us the chance to study the plasma evolution during transients. Moreover, the study of the plasma evolution during breakdown can be useful also in ion sources working during the steady state because knowing the plasma breakdown “‘history” can help in the understanding of the characteristics once the steady state is reached. The second key point in the diagnostics design was to make them as non-invasive as possible. Plasma characteristics depend to a large extent on the device in which the plasma is generated and sustained. For this reason, all the diagnostics were designed with the aim of minimizing their impact on the plasma source characteristics: geometry, pumping, gas flow, microwave resonance, etc. The first of the diagnostics that we developed is presented in chapter 3. It consisted of an electrical biased probe inserted into the plasma. The evolution of the probe saturation current, together with incoming and reflected power, measured in a bidirectional coupler, and plasma emitted visible light were used to study the breakdown process. Three different magnetic field profiles and two Hydrogen pressure values were used as working conditions and measurements were taken under a wide range of powers and duty cycles. A breakdown structure composed of two stages is proposed. The first of these is an early breakdown stage that we called MW Coupling, where the microwave power matching is in progress under a low ionization rate, and fast coupling changes between the microwaves and the weak plasma inside the chamber are taking place. Then, during the second stage, which we called Plasma Formation, light emission increment is associated with an ionization rate increment and probe saturation current is rapidly reached. In this last stage, the MW coupling is well established and the absorbed energy density is good enough to produce plasma evolution to final steady state parameters. A simple model based on the influence of the seed electrons remaining in the neutral gas between pulses was developed with the goal of describing our experimental data as a function of duty cycle and power. Calculations showed a good agreement with our experimental data for the case in which the ECR magnetic field profile was used, but also show a mismatch for the magnetic field profile over ECR. Further research on this topic is among our future plans. In chapter 4, a different diagnostics is presented: a time-resolved Langmuir probe plasma electron density and temperature set-up. In the light of the results obtained in chapter 3, this diagnostics was developed with the aim of studying the evolution of plasma parameters during the breakdown transient and its relationship with the breakdown time evolution.

Chapter 10. Conclusions

167

The working conditions were set to be the same as those used in the breakdown time study described in the previous chapter. However, the lower magnetic field profile was disregarded due to the narrow range of working parameters (power, pressure and duty cycle) capable of producing stable plasma condition under this B-field configuration. Graphs of measured MW coupling times taken from the previous chapter were used as a guideline to interpret the electron density and temperature evolution study. In cases of high incoming power, generally characterized by fast MW coupling times, a temperature peak was found to appear just after MW coupling. These results support the model of a breakdown process composed of two stages: MW coupling and plasma formation. However, in low power cases, the peak tends to disappear and temperature evolves smoothly upward towards the steady state. On the other hand, the plasma electron density shows a different behavior, depending on the magnetic field profiles chosen. While for the asymmetric over ECR B-field profile it grows smoothly, and does not present any remarkable changes in the breakdown process, in the symmetric B-field case, a density peak is found in coincidence with the maximum temperature. In chapter 5, a study into plasma electron density and temperature evolution during decay is presented. Working conditions were chosen to be the same as in the previous breakdown study. In consequence, this study, together with the breakdown one described in the previous chapter, completes a panoramic view of plasma parameter evolution along the whole pulse under the selected working conditions. Incoming and reflected power signals were shown as a temporal reference, together with plasma density and temperature measurements. A remarkable structure was observed in the reflected power signal, showing a peak just after incoming power shut-off and another smaller rebound approximately 10 μs later. Temperature measurements also showed a rebound in coincidence with the second reflected power peak. Electron density, in contrast, shows an independent, smoothly decreasing behavior. No remarkable changes were observed with the magnetic field profile, the incoming power or the duty cycle. However, MW coupling seems to play an important role: in those cases where the coupling during the flat top of the pulse was better, the structure in the reflected power and electron temperature evolution was more noticeable. A combination of diagnostics is presented in chapter 6. A diagnostic port was designed to allow simultaneous measurements of plasma VUV emission and Langmuir probe plasma electron density and temperature. As is widely known, Langmuir probe results are somewhat sensitive to the analysis method chosen, the existence of a magnetic field profile, the plasma electron distribution function, and other aspects. For this reason,

168

this complimentary diagnostic was designed to measure the average energy of electrons during breakdown. Calculations of the electric field drop associated with plasma breakdown and of the ratio between Lyman band light signal and electron saturation current predicted an electron energy transient during the breakdown in agreement with the existence of the electron temperature peak previously measured. The temporal evolution of Lymanalpha and Lyman band emissions was recorded as they have been demonstrated to be good tools for the study of atomic and molecular excitation evolution since they are not sensitive to changes in the tail of the electron energy distribution function. This is especially important when studying transient effects. This combined diagnostics confirmed the existence of the temperature peak measured in chapter 4. It also highlighted the impact of the electric field drop calculated from experimental measurements of incoming and reflected power on the plasma parameter evolution. Chapter 7 is probably the section that is most strictly related with ion source engineering. The narrow range of working parameters capable of sustaining stable plasma inside the chamber motivated a deep study into the resonant characteristics of the device. The relationship between the electric field evolution during breakdown and the plasma parameter evolution that could be deduced from the results in chapter 6 also led us to study the possible effects that E-field distribution just before breakdown could have on plasma parameters. The engineering company that initially designed the plasma chamber and the MW coupler considered the system as an assembly of independent parts, and the MW coupler part was designed to adapt the impedance between the plasma chamber and the WR284 standard waveguide. Chapter 7, proposes a different approach. The system was considered as a whole resonant cavity and the geometry was changed to try to maximize the electric field inside the plasma chamber while minimizing it in the rest of the system. Under these premises, a new plasma chamber set with an MW coupler was designed, built and installed at the source. The optimized system, where values reached by the E-field before plasma disruption were calculated to be five times higher, was proven to produce a better performance. The tendency of the plasma to be located outside the plasma chamber completely disappeared and the range of working parameters became wider, allowing for deeper studies into plasma behavior under different conditions. Measurements of incoming and reflected power were recorded and used to calculate the E-field drop produced by plasma ignition in the optimized system. By comparing these results with those of the preliminary design shown in chapter 6, we estimated that


169

when plasma is ignited, the average value of the electric field inside the plasma chamber can be up to six times higher in the optimized design case. The Langmuir probe diagnostics was installed to check the influence of this optimized design on plasma electron density and temperature. Measurements were made along the plasma chamber axis in both designs. No differences were observed in electron temperature values or in its evolution along the axis. A growing tendency towards the MW injection side was observed in both cases. Measured density, in contrast, was up to four times higher in the case of the optimized design. It is important to note that the plasma electron density and temperature measurements were made using a different magnetic field profile for each design. This was due to the impossibility of producing stable plasma in the optimized design using the asymmetric Bz > ECR magnetic field profile that produces the most stable plasma in the preliminary one. This suggests that the improvement in plasma density could be due to a combination of the changes in both the electric and the magnetic field profiles. Trying to find a B-field configuration that works in both designs and comparing electron density evolution under each of these conditions is one of our future objectives. In chapter 8, the least invasive plasma diagnostics of this work is presented, with the development of a high speed photography diagnostic. Although high speed photography is a well known tool for plasmas in experiments like exploding wires, Z-pinches or plasma foci, among others, the low light intensity emitted by ECR plasmas place them on the borderline for applying this technology. On the other hand, the extended duration of phenomena in ECR plasma makes it possible to use relatively long exposure times. A “transparent plasma electrode” was developed in order to provide a view of the full plasma chamber volume. Two tungsten meshes separated by a 10 mm thick quartz window with a 7 mm hole in the middle for pumping prevented the microwaves from leaking out of the chamber. An image intensified CCD camera was used to take photos of plasma distribution with an exposure time of 1 μs. Direct experimental evidence of eight different stable plasma density distribution modes in an ECR 2.45 GHz plasma generator operated with Hydrogen at 50 Hz are also recorded here for the first time. A high speed photography study was conducted under a wide range of experimental conditions, including measurements of time-spatial integrated visible spectra. A number of connections between modes and visible spectra are reported as fingerprints related with the ratios of the different ions present in the plasma. These spectra show that some degree of tuning to ion species output could be possible on ECRIS devices by modifying the experimental conditions with special emphasis on the magnetic field.

170

Chapter 9 makes a first approach to the study of the breakdown process of the different plasma configurations through ultra-fast pictures. The breakdown process under the conditions producing each plasma distribution mode was studied. Each breakdown study is composed of two sets of four pictures, the first one corresponding to an early stage, where the light emission is low, and the second to a later stage of higher brightness. Sets of pictures corresponding to full visible light, Balmer-alpha, Balmer-beta and Fulcher-band were presented for the high-brightness cases: Column, Hourglass and Slug. The low emission of the cases that we called low-brightness, i.e Flower, Full-Chamber, Ring, Yin-Yang and Donut, specially during first stage of plasma breakdown, made it impossible to take pictures with filters using short enough exposure times, especially during the first stage of plasma breakdown. The development of the time resolved Balmer and Fulcher diagnostics using fiber optics, band filters and a photodiode is a promising tool for the tuning of ion sources. It offers an effective alternative to other much more complicated and expensive systems that need to interrupt the ion beam, to be under vacuum conditions or to be built as part of the source itself. Moreover, this simple and comparatively inexpensive device could allow for installation of an “in flight” plasma monitoring system in the ion source. The fact that it does not need to be under vacuum conditions makes the system very flexible and versatile. The experimental evidence of the connection between atomic and molecular ionization processes and the Balmer and Fulcher emissions is the keystone upon which this diagnostic rests. The evolution of VUV lines (Lyman-alpha and Lymanband) studied in chapter 6 and the visible lines recorded in this chapter are compared. The similar behavior of both the VUV and the visible lines studied confirms the second as a good tool for ion source monitoring and optimization. + The similarities between the H+ , H+ 2 and H3 current evolution measured by Y. Xu et al. and the photodiode signals shown in this chapter were highlighted. The Wien filter currently under development in our laboratory should allow us to measure the visible line emission at the same time as the different ion currents. Testing the correlations between both is one of our plans for the near future.

Finally, two rotating plasma configurations were discovered. These phenomena,which have never before been reported for these kinds of plasma source, are very stable and reproducible. Although magnetic field configuration appears to be critical for this rotational behavior, some dependence of the rotational frequency on power and neutral gas pressure was also found. An inversion in the direction of the rotation was observed when the magnetic field is also inverted. This fact suggests a connection with plasma drift velocity although the nature of the actuating force is still unknown.


10.2

171

Contributions

The international publications generated by the work described in this thesis are listed in this section.

• O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Experimental Study of Breakdown Time in a Pulsed 2.45 GHz ECR Hydrogen Plasma Reactor. IEEE Transactions on Plasma Science, Vol. 40, no 12 (Dec. 2012). Chapter 3. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Experimental Study of Temperature and Density Evolution during Breakdown in a 2.45 GHz ECR Plasma. TUYO01, Proceedings of The 20th International Workshop on Electron Cyclotron Resonance Ion Sources (ECRIS 2012), 25 - 28 September 2012, Sydney. Chapter 4. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Temperature Peaking at Beginning of Breakdown in 2.45 GHz Pulsed Off-Resonance Electron Cyclotron Resonance Ion Source Hydrogen Plasma. Review of Scientific Instruments, Vol. 83, p. 103302 (Oct 2012). Chapter 4. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Temperature and Density Evolution during Decay in a 2.45 GHz ECR Hydrogen Plasma Reactor: Off Resonance and Resonance Cases. Review of Scientific Instruments, Vol. 84, 093301 (Sept 2013). Chapter 5. • O. D. Cort´ azar, J. Komppula, O. Tarvainen, A. Meg´ıa-Mac´ıas, A. Vizca´ıno de Juli´ an and H. Koivisto. Experimental Study of Hydrogen Plasma Breakdown in a 2.45 GHz Microwave Discharge. Plasma Sources Science and Technology, Vol. 22. p. 015026. (Feb. 2013). Chapter 6. • A. Meg´ıa-Mac´ıas, O.D Cortázar and A. Vizca´ıno de Julián. Influence of Microwave Driver Coupling Design on Plasma Density at Testbench for Ion sources Plasma Studies, a 2.45 GHz Electron Cyclotron Resonance Plasma Reactor. Review of Scientific Instruments, Vol. 85, 033301 (March 2014). Chapter 7. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas, O. Tarvainen, A. Vizca´ıno de Julián, J. Komppula and H. Koivisto. High Speed Photographic Study in a 2.45 GHz Microwave Hydrogen Plasma Source: Direct Evidence of Several Stable Plasma Density Distribution Modes. Under review. Chapter 8.

172

• O.D Cort´ azar, A. Meg´ıa-Mac´ıas, A. Vizca´ıno de Julián, O. Tarvainen, J. Komppula and H. Koivisto. Ultra-fast Intensified Frame Images from an ECR Hydrogen Plasma at 2.45 GHz: Some Space Distributions of Visible and Monochromatic Emissions. Review of Scientific Instruments, Vol. 85, 02A902 (Feb 2014). Chapter 9.

10.3

Future Work

The first question that may occur to the reader is how these results affect ion source performance, and in particular, how our diagnostics development and experimental observations can be transferred to ion sources. For this reason, our first priority for the near future is to transform TIPS into an ion source, i.e., to add an extraction system to the plasma source and extract a sample of the ions composing the plasma: the ion beam. In this way, an extraction electrode system has been developed. It permits to obtain a beam sample of 1 mA to be analyzed by a Wien Filter under commissioning. Both systems have been developed in the laboratory together with a Faraday’s Cup to measure the value of the current of the different ions coming from the filter. The Wien Filter is + capable to separate the H+ , H+ 2 and H3 species from the beam, allowing only one of them to reach the Faraday’s Cup. Moreover, the Wien Filter has been designed with a free optical path that permits to undertake optical measurements from the plasma chamber axis. This will make possible to conduct simultaneous measurements of the visible light evolution and the ion currents. The main goal is to demonstrate the value of the Balmer series and the Fulcher band emissions to determine the ion composition for these plasmas. Moreover, it is planed to carry out a systematic study of the species fraction for the different plasma density distribution modes determined by the magnetic field configurations shown in chapter 8. Finally, we plan to used the acquired knowledge to find optimization ways for the different ion species. In this way, a collaboration with “DAEδALUS” project has been started, where the development of high current H+ 2 ion sources is a key factor in the context of the new high current Cyclotron generation.

Chapter 11

Conclusiones 11.1

Trabajo realizado

Tal y como Ian G. Brown avanza en su libro “The Physics and Technology of Ion Sources” [1], la f´ısica que rige una fuente de iones es, en su mayor´ıa, f´ısica del plasma. El objetivo primordial del trabajo a lo largo de esta tesis ha sido el desarrollo de herramientas novedosas de diagn´ ostico que puedan servir para ampliar el conocimiento existente sobre plasmas de fuentes de iones ECR. Una serie de diferentes diagnósticas han sido dise˜ nadas, desarrolladas e implementadas en el reactor de plasma denominado TIPS: medición de la corriente circulante a través de una sonda polarizada, mediciones de densidad y temperatura de electrones en el plasma mediante sondas de Langmuir, espectroscopia ultravioleta, espectroscopia visible, mediciones monocrom´ aticas de la luz, fotograf´ıas ultra-rápidas del volumen completo de la c´ amara de plasma. Además, las se˜ nales de la potencia de microondas incidente y reflejada y la medici´ on de la luz visible emitida han sido utilizadas en todas las diagnósticas como elemento identificador de las condiciones del plasma, garantizando as´ı la reproducibilidad. El dise˜ no de cada una de las diagnósticas, as´ı como los resultados de los estudios llevados a cabo con las mismas han sido presentados a lo largo de esta tesis, prestando especial atenci´ on al dise˜ no ingenieril de cada una de ellas as´ı como a los detalles técnicos de su implementaci´ on. El conocimiento de las caracter´ısticas de funcionamiento del equipo desarrollado durante la implementaci´ on de las diferentes diagnósticas motivó un estudio del sistema de microondas. Este estudio fue el punto de partida para el dise˜ no de un nuevo conjunto optimizado compuesto por la cámara de plasma y el acoplador de microondas. Este 173

174

dise˜ no se llev´ o a cabo desde un punto de vista ingenieril diferente al original que demostró ser capaz de mejorar ampliamente la estabilidad del plasma y, a la vez, incrementar su densidad electr´ onica en un factor cuatro. Las fuentes de iones ECR son frecuentemente usadas tanto en el ámbito cient´ıfico como en el industrial. Es por ello que su optimización resulta de gran interés tanto para los dise˜ nadores de fuentes de iones como para los usuarios. En la actualidad, muchas fuentes de iones de este tipo operan en modo pulsado aprovechando, en su mayor´ıa, dos fenómenos asociados a los transitorios de encendido y apagado del plasma: preglow [13, 16, 17, 44, 45, 73] y afterglow [12–14, 74–76] respectivamente. Como consecuencia, uno de los puntos cr´ıticos a la hora de afrontar el dise˜ no de nuestras diagnósticas fue dotarlas de resoluci´ on temporal de modo que se dispusiese de la posibilidad de llevar a cabo estudios de la evoluci´ on del plasma durante estos periodos transitorios. Además, el estudio de la evoluci´ on del plasma durante el encendido puede resultar de interés en el caso de fuentes que operen durante el estado estacionario, ya que es un modo de conocer la “historia” del plasma que puede arrojar información sobre las caracter´ısticas del mismo una vez alcanzado el estado estacionario. Por otro lado, un segundo punto importante fue la realización de dise˜ nos lo menos invasivos posible dado que existe una fuerte relación entre las caracter´ısticas del plasma y las del equipo en el que es generado. Es por ello que todas las diagnósticas fueron dise˜ nadas tratando de minimizar el impacto en las caracter´ısticas de la fuente de plasma: geometr´ıa, sistema de vac´ıo, flujo de gas, resonancia de las microondas, etc. La primera diagn´ ostica desarrollada se presenta en el cap´ıtulo 3. Se insertó en la cámara de plasma una sonda eléctrica polarizada y se utilizó la evolución de la corriente que circulaba por la misma, junto con mediciones de las potencias incidente y reflejada y de la luz visible emitida por el plasma, para estudiar el proceso de encendido del plasma. Se realizaron estudios bajo tres perfiles de campo magnético diferentes y dos valores de la presión de Hidr´ ogeno en un amplio rango de potencias y ciclos de trabajo. A la vista de los resultados obtenidos, se propuso una estructura del proceso de encendido del plasma compuesta por dos etapas. En una primera etapa de encendido, que llamamos MW coupling (etapa de acoplamiento de microondas), tiene lugar una transferencia de la potencia de las microondas con un bajo grado de ionización del plasma a la vez que se produce un r´ apido acoplamiento de las mismas con el plasma poco denso que se est´ a formando dentro de la cámara. En la segunda etapa, que hemos llamado Plasma Formation (formaci´ on del plasma), la emisión luminosa del plasma aumenta asociada a un incremento de la ionización y la corriente en la sonda alcanza su nivel de saturaci´ on. Durante esta u ´ltima etapa el acoplamiento de las microondas con el

Chapter 11. Conclusiones

175

plasma es mucho m´ as alto y la densidad de energ´ıa absorbida es suficiente para alcanzar la condici´ on estacionaria. Un modelo sencillo de este proceso, basado en la influencia de los electrones “semilla”, ´ capaces de sobrevivir durante el periodo entre pulsos, fue desarrollado. Este trata de describir los resultados experimentales en función de la potencia y el ciclo de trabajo. Los c´ alculos realizados mostraron estar de acuerdo con las mediciones experimentales para el caso en el que el campo magnético utilizado fue simétrico y próximo al ECR. Sin embargo, se alejaron de los datos experimentales obtenidos con el campo magnético asimétrico cuyos valores se encontraban por encima del nivel del ECR. Un estudio m´ as profundo de este caso se encuentra entre nuestros planes futuros. El cap´ıtulo 4 presenta una nueva diagnóstica de mediciones de la temperatura y la densidad del plasma con resolución temporal mediante sondas de Langmuir. A la vista de los resultados obtenidos en el cap´ıtulo 3, la diagnóstica fue desarrollada para el estudio de la evoluci´ on de los parámetros del plasma durante el proceso de encendido. Las condiciones de trabajo elegidas fueron las mismas que en el estudio de los tiempos de encendido desarrollado en el cap´ıtulo anterior. Sin embargo, el caso del campo magnético por debajo del ECR fue descartado debido al estrecho rango de valores de potencia y ciclo de trabajo capaz de producir plasma estable. Los estudios de tiempos de acoplamiento de microondas del cap´ıtulo 3 fueron utilizados como hilo conductor para las mediciones de la evolución de la densidad y la temperatura de electrones. En los casos en los que la potencia incidente es alta, caracterizados en general por tiempos de acoplamiento breves, se descubrió un pico de temperatura en los primeros instantes del encendido. Esto apoya la hipótesis de que el proceso de encendido podr´ıa dividirse en dos etapas: acoplamiento de microondas y formaci´ on del plasma. Por otro lado, en los casos en los que la potencia incidente es baja dicho pico de temperatura tiende a desaparecer y la temperatura evoluciona suavemente hasta su valor final. La densidad del plasma muestra un comportamiento completamente diferente dependiendo del perfil de campo magnético elegido. Bajo el perfil de campo magnético por encima del ECR la densidad no presenta cambios significativos durante el proceso de encendido. Mientras que si el perfil elegido es el campo simétrico en el ECR, la densidad muestra un pico en coincidencia con el pico de temperatura. El cap´ıtulo 5 desarrolla un estudio de la evolución de la densidad y la temperatura de electrones durante el proceso de apagado. Las condiciones de trabajo elegidas fueron las mismas que en el estudio del proceso de encendido llevado a cabo en el cap´ıtulo 4. En consecuencia, este estudio, junto con el realizado en el cap´ıtulo anterior, ofrecen una

176

visión de la evoluci´ on de los par´ ametros del plasma a lo largo de todo el pulso bajo las condiciones de trabajo elegidas. Las se˜ nales de potencia incidente y reflejada fueron utilizadas como referencia temporal junto con los datos de temperatura y densidad medidos. Se encontró una estructura muy llamativa en la se˜ nal de potencia reflejada, mostrando un pico justo después del inicio del descenso de la se˜ nal de potencia incidente y un segundo pico más peque˜ no unos 10 μs después. Coincidiendo con este segundo incremento en la se˜ nal de potencia reflejada, las mediciones de temperatura muestran un pico también. La densidad de electrones, sin embargo, muestra un decaimiento suave e independiente de los picos observados. No se observaron cambios significativos en este comportamiento asociados a los diferentes perfiles de campo magnético, las variaciones en la potencia incidente o en el ciclo de trabajo. Sin embargo, la calidad del acoplamiento de microondas durante el régimen estacionario parece jugar un papel importante: en aquellos casos en los que el acoplamiento fue mejor, la estructura de la se˜ nal de potencia reflejada y de la evolución de la temperatura de electrones result´ o más notable. En el cap´ıtulo 6 se presenta una combinación de dos diagnósticas. Se dise˜ nó un nuevo puerto que permitiese realizar al mismo tiempo medidas de la emisión ultravioleta del plasma y de su densidad y temperatura mediante una sonda de Langmuir. Como es sabido, los resultados de las sondas de Langmuir son sensibles al método de análisis utilizado, a la existencia de campo magnético, a la función de distribución de energ´ıa de los electrones, etc. Por esta raz´ on, fue dise˜ nada esta diagnóstica complementaria para medir la energ´ıa electr´ onica media durante el proceso de encendido. Se realizaron estimaciones del descenso del campo eléctrico asociado al encendido del plasma y del ratio entre la se˜ nal luminosa de la banda de Lyman y la corriente de electrones de saturaci´ on que corroboraron la existencia del pico en la temperatura de electrones previamente detectado. El cap´ıtulo 7 es, probablemente, el que está más estrictamente relacionado con la ingenier´ıa de fuentes de iones. El estrecho rango de parámetros de trabajo capaces de producir comportamiento estable, motivaron un estudio más profundo de las caracter´ısticas de resonancia y acoplamiento electromagnético del equipo. La empresa de ingenier´ıa que dise˜ nó la cámara y el acoplador de microondas en primer lugar, consider´ o el sistema como un ensamblaje de partes independientes y el acoplador fue dise˜ nado para adaptar la impedancia entre la cámara y la gu´ıa de onda estándar WR284. En el cap´ıtulo 7 se propone un punto de vista diferente. El sistema completo fue considerado como una cavidad resonante y la geometr´ıa fue modificada


177

para maximizar el campo eléctrico en la cámara a la vez que se minimizaba en el resto del sistema. Bajo estas premisas se dise˜ nó un nuevo conjunto optimizado de cámara y acoplador de microondas que fue construido e instalado en el reactor. El sistema optimizado produjo mejores resultados. La tendencia del plasma a encenderse fuera de la cámara desapareci´ o completamente y el rango de parámetros de trabajo se hizo m´ as amplio permitiendo estudios m´ as profundos bajo diferentes condiciones. Se utilizaron las mediciones de potencia incidente y reflejada para calcular la ca´ıda del campo eléctrico producida por el encendido del plasma en el dise˜ no optimizado. Comparando estos resultados con los obtenidos para el dise˜ no preliminar en el cap´ıtulo 6 fue posible estimar que el campo cuando el plasma está encendido, podr´ıa ser hasta seis veces m´ as intenso en el caso del dise˜ no optimizado. La diagn´ ostica de mediciones con sondas de Langmuir fue nuevamente instalada para analizar la influencia del dise˜ no optimizado en la densidad y la temperatura. Se llevaron a cabo mediciones a lo largo del eje de la cámara con ambos dise˜ nos. No se registraron diferencias en los valores de temperatura de electrones obtenidos ni en su evoluci´ on, aunque s´ı un incremento de la temperatura hacia la zona de inyección de microondas en ambos casos. La densidad de electrones, sin embargo, resultó ser hasta cuatro veces superior en el caso del dise˜ no optimizado. Es importante resaltar que la densidad y la temperatura de electrones fueron medidas utilizando perfiles de campo magnético diferentes para cada caso. Esto fue debido a la imposibilidad de producir un plasma estable con el dise˜ no optimizado utilizando el campo asimétrico por encima del ECR que produc´ıa las condiciones de plasma m´ as estables con el dise˜ no preliminar. Esto sugiere que la mejora en la densidad del plasma podr´ıa deberse a una combinación de los cambios en los perfiles de campo magnético y campo eléctrico. Entre nuestros planes de trabajo futuro se encuentra buscar una configuraci´ on de campo magnético que produzca plasma estable con ambos dise˜ nos y comparar la evoluci´ on de la densidad electrónica en ambos casos. La diagn´ ostica menos invasiva de todas las presentadas a lo largo de esta tesis es la que se muestra en el cap´ıtulo 8. Se trata de una diagnóstica de fotograf´ıas ultra-rápidas. Aunque la fotograf´ıa de alta velocidad es bien conocida como técnica de diagnóstico de plasmas en experimentos como “exploding wires”, “z-pinches” o “plasma foci” entre otros, la baja emisividad luminosa de los plasmas ECR hace que se encuentren en el l´ımite de utilizaci´ on de esta tecnolog´ıa. Por otro lado, la larga duración de los fenómenos asociados a los plasmas ECR hace posible el uso de tiempos de exposición relativamente largos. Para poder llevar a cabo esta diagnóstica en TIPS, se desarrolló un original “electrodo de plasma transparente” que permitiese observar el volumen completo de la cámara. Con el fin de evitar la salida de las microondas fuera de la misma, se utilizaron

178

dos mallas de tungsteno separadas por una ventana de cuarzo de 10 mm de espesor con un orificio central de 7 mm de di´ ametro para el bombeo. Un convertidor de imágenes CCD fue utilizado para realizar fotograf´ıas de la distribución del plasma dentro de la cámara con 1 μs de tiempo de exposición. Por primera vez se encontraron evidencias experimentales directas de la existencia de ocho diferentes modos estables de distribución de la densidad del plasma en un reactor ECR de 2.45 GHz. Se llev´ o a cabo un estudio fotográfico de alta velocidad acompa˜ nado de mediciones de los espectros visibles integrados en el tiempo para cada uno de los modos encontrados. También se establecieron algunas conexiones entre los diferentes modos y los ratios entre las distintas especies de iones presentes en el plasma a través de los espectros de luz visible. Estos espectros muestran que ser´ıa posible realizar un cierto ajuste de las especies i´ onicas presentes en una fuente de iones ECR modificando las condiciones de trabajo, especialmente el perfil de campo magnético. En el cap´ıtulo 9 se realiza una primera aproximación al estudio de la evolución de los diferentes modos de distribuci´ on del plasma durante el encendido a través de las fotograf´ıas ultra-r´ apidas. Se estudi´ o el proceso de encendido para una de las condiciones de trabajo que producen cada uno de los modos. Cada estudio está compuesto de dos conjuntos de cuatro fotos, el primero tomado durante una etapa inicial en la que la emisión luminosa es baja y el segundo durante una etapa más tard´ıa en la que la emisión luminosa es mayor. En los casos de alta emisión luminosa (Column, Hourglass y Slug) se tomaron conjuntos de fotos en todo el espectro de luz visible y, a través de filtros, de las emisiones de Balmer-alpha, Balmer-beta y de la banda de Fulcher. Sin embargo, para los casos de baja emisividad (Flower, Full-Chamber, Ring, Yin-Yang y Donut) fue imposible tomar fotograf´ıas a través de los filtros con tiempos de exposición lo suficientemente breves para obtener imágenes n´ıtidas. Por otra parte, el desarrollo de diagnósticas de las emisiones de Balmer y Fulcher con resoluci´ on temporal mediante el uso de fibras ópticas, filtros y fotodiodos es una prometedora herramienta para la optimización las fuentes de iones. Esta diagnóstica ofrece una alternativa a otros sistemas mucho mas complejos y costosos que requieren la interrupci´ on del haz de part´ıculas. Además, esta diagnóstica simple y que a penas requiere inversi´ on, permitiria la instalación de un sistema de monitorización del plasma de una fuente de iones en tiempo real. La evidencia experimental de la conexión de los procesos de ionizaci´ on at´ omica y molecular con las emisiones de Balmer y Fulcher es la clave sobre la que se asienta esta diagnóstica. Merece ser destacada la similitud entre las evoluciones de las emisiones ultravioletas (Lyman-alfa y banda de Lyman del cap´ıtulo 6) y las de las lineas visibles estudiadas en este cap´ıtulo. Esto sugiere que la medición de la emisi´ on visible puede ser una diagnóstica valiosa para la monitorización y la optimizaci´ on de fuentes de iones.


179

+ Las similitudes entre las corrientes de iones H+ , H+ 2 y H3 medidas por Y. Xu et al. y las se˜ nales de fotodiodo presentadas en este cap´ıtulo merecen ser mencionadas. El filtro de Wien que se encuentra actualmente en construción en nuestro laboratorio nos dará la posibilidad de medir las corrientes de los diferentes iones al mismo tiempo que se toman las se˜ nales de fotodiodo. De este modo será posible establecer correlaciones entre ellas.

Por u ´ltimo, dos configuraciones de plasma rotante han sido descubiertas. Este fenómeno, que no hab´ıa sido visto antes en este tipo de fuentes de plasma es muy estable y reproducible. Aunque la configuración de campo magnético es cr´ıtica para la aparici´ on de este fen´ omeno, se ha observado la dependencia de la frecuencia de rotaci´ on con la potencia incidente y la presión de gas neutro. Se observó que la rotación invierte su sentido de giro si se invierte la dirección del campo magnético. Esto u ´ltimo sugiere que este fen´ omeno puede tener origen en la velocidad de deriva del plasma aunque se desconozca la naturaleza de la fuerza implicada.

11.2

Contribuciones

A continuaci´ on se presenta un listado de las publicaciones en congresos y revistas internaciones a las que ha dado lugar el trabajo presentado en esta tesis. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Experimental Study of Breakdown Time in a Pulsed 2.45 GHz ECR Hydrogen Plasma Reactor. IEEE Transactions on Plasma Science, Vol. 40, no 12 (Dec. 2012). Cap´ıtulo 3. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Experimental study of temperature and density evolution during breakdown in a 2.45 GHz ECR plasma. Proceedings of The 20th International Workshop on Electron Cyclotron Resonance Ion Sources (ECRIS 2012), 25 - 28 September 2012, Sydney. Cap´ıtulo 4. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Temperature peaking at beginning of breakdown in 2.45 GHz pulsed Off-Resonance Electron Cyclotron Resonance Ion Source Hydrogen Plasma. Review of Scientific Instruments, Vol. 83, p. 103302 (Oct 2012). Cap´ıtulo 4. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas and A. Vizca´ıno de Julián. Temperature and Density Evolution During Decay in a 2.45 GHz ECR Hydrogen Plasma Reactor: Off Resonance and Resonance Cases. Review of Scientific Instruments, Vol. 84, 093301 (Sept 2013). Cap´ıtulo 5.

180

• O. D. Cort´ azar, J. Komppula, O. Tarvainen, A. Meg´ıa-Mac´ıas, A. Vizca´ıno de Julián and H. Koivisto. Experimental Study of Hydrogen Plasma Breakdown in a 2.45 GHz Microwave Discharge. Plasma Sources Science and Technology, Vol. 22. p. 015026. (Feb. 2013). Cap´ıtulo 6. • A. Meg´ıa-Mac´ıas, O.D Cort´ azar and A. Vizca´ıno de Julián. Influence of Microwave Driver Coupling Design on Plasma Density at Testbench for Ion sources Plasma Studies, a 2.45 GHz Electron Cyclotron Resonance Plasma Reactor. Review of Scientific Instruments, Vol. 85, 033301 (Marzo 2014). Cap´ıtulo 7. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas, O. Tarvainen, A. Vizca´ıno de Julián, J. Komppula and H. Koivisto. High Speed Photographic Study in an 2.45 GHz Microwave Hydrogen Plasma Source: Direct Evidence of Several Stable Plasma Density Distribution Modes. Actualmente en revisión. Cap´ıtulo 8. • O.D Cort´ azar, A. Meg´ıa-Mac´ıas, A. Vizca´ıno de Julián, O. Tarvainen, J. Komppula and H. Koivisto. Ultra-Fast Intensified Frame Images from an ECR Hydrogen Plasma at 2.45 GHz: Some Space Distributions of Visible and Monochromatic Emissions. Review of Scientific Instruments, Vol. 85, 02A902 (Feb 2014). Cap´ıtulo 9.

11.3

Trabajo futuro

La primera pregunta que puede surgir al lector al finalizar esta tesis es qué influencia pueden tener estos resultados en las de fuentes de iones y, en particular cómo pueden extrapolarse a la optimizaci´ on de las mismas. Por esta razón, el objetivo de nuestro trabajo futuro es transformar TIPS en una fuente de iones, es decir, a˜ nadir un sistema de extracción y extraer un haz. En este sentido, se ha construido un sistema de electrodos que permite extraer un haz de muestra de 1 mA para ser analizado por medio de un filtro de Wien que se está poniendo a punto. Ambos sistemas han sido desarrollados en el laboratorio junto con una copa de Faraday que permitir´ a medir la corriente de iones que atraviesa el filtro. + ´ Este es capaz de separar las especies de H+ , H+ 2 y H3 presentes en el haz haciendo que llegue a la copa de Faraday solamente una de ellas. Además, el filtro ha sido dise˜ nado de tal forma que ser´ a posible realizar mediciones ópticas en el eje de la cámara de plasma. Esto dar´ a la posibilidad de obtener datos tanto de la evolución temporal como espectroscópicos de la luz emitida y, simultáneamente, de la corriente de cada una de las especies de iones.


181

El primer objetivo es demostrar el valor de las emisiones luminosas de la serie de Balmer y de la banda de Fulcher para determinar la composición iónica del plasma. Adem´ as, se planea llevar a cabo un estudio sistemático de los diferentes modos de distribuci´ on de la densidad del plasma descritos en el cap´ıtulo 8, con el campo magnético como factor principal, que permita determinar la fracción de especies iónicas de cada uno de ellos. Por u ´ltimo, se pretende utilizar el conocimiento obtenido para encontrar medios de optimizaci´ on de la producción de haces de los diferentes iones presentes. En este sentido se ha comenzado a participar en la colaboración “DAEδALUS”, donde es necesario desarrollar fuentes de iones de alta corriente de H+ 2 en el contexto de una nueva generaci´ on de Ciclotrones.

Bibliography [1] I. G. Brown. The Physics and Technology of Ion Sources. WILEY-VCH, 2nd edition, 2004. [2] R. Geller . Electron Cyclotron Resonance Ion Sources and ECR Plasmas. Institute of Physics Publishing, Bristol, UK, 1996. [3] Y. Xu, S. Peng, H. Ren, J. Zhao, J. Chen, A. Zhang, T. Zhang, Z. Guo, and J. Chen. + High current H+ 2 and H3 beam generation by pulsed 2.45 GHz Electron Cyclotron Resonance Ion sources. Review of Scientific Instruments, 85(2):02A943, 2014. doi: http://dx.doi.org/10.1063/1.4850717. [4] T. Taylor and J. S. C Wills. A high-current low-emittance dc ECR proton source. Nuclear Instruments and Methods in Physics Research A, 309:37–42, June 1991. [5] S. Gammino, L. Celona, G. Ciavola, F. Maimone, and D. Mascali. Review on High Current 2.45 GHz Electron Cyclotron Resonance Source. Rev. Sc. Instrum., 81: 02B313, 2010. [6] G. Lisitano, R.A. Ellis Ellis, W.M. Hooke, and T.H. Stix. Production of Quiescent Discharge with High Electron Temperatures. Rev. Sc. Instrum., 39(3):295–297, 1967. [7] G. Lisitano, M. Fontanesi, and E. Sindoni. Nonresonant absorption of electromagnetic waves in a high-density plasma. Appl. Phys. Lett., 16(13):122–124, 1969. [8] Z.Q. Xie. State of the art of ecr ion sources. In Proceedings of Particle Accelerator Conference 1997, Vancouver, Canada., May 1997. IEEE. [9] J. Ishikawa, Y. Takeiri, and T. Takagi. Axial magnetic field extraction type microwave ion source with a permanent magnet. Rev. Sc. Instrum., 55(4):449–456, 1984. [10] N. Sakudo, K. Tokiguchi, H. Koike, and I. Kanomata. Microwave ion source for high-current implanter. Rev. Sc. Instrum., 849(7):940–943, 1978. 183

Bibliography [11] T. Taylor and J. F. Mouris. An advanced high-current low-emittance dc microwave proton source. Nuclear Instruments and Methods in Physics Research A, 336:1–5, November 1993. doi: 10.1016/0168-9002(93)91074-W. [12] G. Melin, F. Bourg, P. Briand, J. Debernardi, M. Delaunay, R. Geller, B. Jacquot, P. Ludwig, T.K. N’Guyen, L. Pin, M. Pontonnier, J.C. Rocco, and F. Zadworny. Some particular aspects of the physics on the ECR sources for multicharged ions. Rev. Sc. Instrum., 61:236–238, 1990. [13] P. Sortais. Pulsed ECR ion source using the afterglow mode(invited). Rev. Sc. Instrum., 63(4):2801–2805, 1992. [14] K. Langbein. Experimental investigation of the afterglow of the pulsed Electron Cyclotron Resonance Discharge. Rev. Sc. Instrum., 37(3):1334–1336, 1996. [15] P. Sortais, J.-L. Bouly, J.-C. Curdy, T. Lamy, T. Sole, P. Thuiller, J.-L. VieuxRochaz, and D. Voulot. ECRIS development for stable and radioactive pulsed beams. Rev. Sc. Instrum., 75(5):1610–1612, 2004. [16] V. A. Skalyga. Gas breakdown in Electron Cyclotron Resonance ion sources. Rev. Sc. Instrum., 77(3):03A325–1 – 03A325–1, 2006. [17] V.G. Zorin. Status and development of ECR ion sources. In Proceedings of ECRIS 2010, pages 90–89, Grenoble, France, 2010. [18] O. D. Cort´ azar, A. Meg´ıa-Mac´ıas, and A Vizca´ıno-de Julián. Experimental Study of Breakdown Time in a Pulsed 2.45-GHz ECR Hydrogen Plasma Reactor. IEEE Trans. Plasma Sci., 40(12):3409–3419, 2012. Digital Object Identifier 10.1109/TPS.2012.2214064. [19] David Meeker. Finite element method magnetics v4.2. www.femm.info, 2010. [20] G. Guest. Electron Cyclotron Heating of Plasmas. KGaA, Weinheim, Germany, 2009. ISBN ISBN 978-3-527-40916-7. Cap. 6. p. 103-114. [21] H. R. Griem. Principles of Plasma Spectroscopy. Cambridge Univ. Press, New York, 1997. ISBN ISBN 0 521 45504 9. Cap. 9. p. 248-250. [22] C. Lyneis, J. Benitez, D. Leitner, J. Noland, M. Strohmeier, H. Koivisto, and O. Tarvainen. Charaterization of the microwave coupling to the plasma chamber of the LBL ECR ion source. In Proceedings of ECRIS 2010, pages 162–164, Grenoble, France, Sept 2010. [23] L. Celona, S. Gammino, F. Maimone, D. Mascali, N. Gambino, R. Miracoli, and G. Ciavola. Observations of resonant modes formation in microwave generated magnetized plasmas. Eur. Phys. J. D, 61(1):107, 2010.

Bibliography

185

[24] L. Celona, S. Gammino, G. Ciavola, F. Maimone, and D. Mascali. Microwave to plasma coupling in electron cyclotron resonance and microwave ion sources (invited). Rev. Sc. Instrum., 81:02A333, 2011. ¨ [25] O. Tarvainen, T. Ropponen, V. Toivanen, J. Arje, and H. Koivisto. Plasma breakdown diagnostics with the biased disc of electron cyclotron resonance ion source. Plasma Sources Sci. Technol., 18:035018, 2009. [26] E. P. Gray and D. E. Kerr. The Diffusion Equation with a Quadratic Loss Term Applied to Electron Volumen Recombination in a Plasma. Ann. of Phys., 17:276– 300, 1959. [27] R. J. Goldston and P. H. Rutherford. Introduction to Plasma Physics. Taylor and Francis Group, 270 Madison Ave. New York, 1995. ISBN ISBN -13: 0 7503 0183 1. Cap. 10. p. 152. ¨ [28] O. Tarvainen, T. Ropponen, V. Toivanen, T. Kalvas, J. Arje, and H. Koivisto. Diagnostics of plasma decay and afterglow transient of an electron cyclotron resonance ion source. Plasma Sources Sci. Technol., 19:1–13, 2010. [29] J. Pelletier and A. Anders. Plasma-Based Ion Implantation and Deposition: A Review of Physics; Technology, and Applications. IEEE Trans. Plasma Sci., 33(6): 1944–1959, 2005. [30] H. B¨ acker, J. W. Bradley, P. J. Kelly, and R. D. Arnell. Using Langmuir probes to measure the plasma decay rates in pulsed RF magnetron discharges. J. Phys. D: Appl. Phys., 34:2709–2714, 2001. [31] A. Rousseau, E. Teboul, N. Lang, M. Hannemann, and J. Ropcke. Langmuir probe diagnostic studies of a pulsed hydrogen plasmas in planar microwave reactors. J. Appl. Phys., 92(27):3463–3471, 2012. [32] N. Britun, T. Godfroid, S. Konstantinidis, and R. Snyders. Time-resolved gas temperature evolution in pulsed Ar-N2 microwave discharge. Appl. Phys. Lett., 98 (14):1502–1504, 2011. [33] O. D. Cort´ azar, A. Meg´ıa-Mac´ıas, and A. Vizca´ıno-de Julián. Temperature peaking at beginning of breakdown in 2.45 GHz pulsed offresonance electron cyclotron resonance ion source hydrogen plasma. Rev. Sc. Instrum., 83(10):3302, 2012. [34] F. F. Chen. Langmuir probes in RF plasma: surprising validity of OML theory. Plasma Sources Sci. Technol., 18:035012, 2009. [35] R. L. Merlino. Understanding Langmuir probe current-voltage characteristics. Am. J. Phys., 75(12):1078–1085, 2007.

Bibliography [36] F. F. Chen. Plasma Diagnostics Techniques. Number edited by R. H. Huddlestone and S.L Leonard. Accademic Press, New York, 1965. [37] J. L. Jauberteau and I. Jauberteau. Electron energy distribution function in plasma determined using numerical simulations of multiple harmonic components on Langmuir probe characteristic-Efficiency of the method. Rev. Sc. Instrum., 78(4):3501, 2007. [38] V. A. Godyak and V. I. Demidov. Probe measurements of electron-energy distributions in plasmas: what can we measure and how can we achieve reliable results? J. Phys. D: Appl. Phys., 44:233001, 2011. [39] M. Lieberman and A. Lichtemberg. Principles of Plasma Discharges and Material Processing. Wiley, 1994. p. 189. [40] I. D. Saudit and F. F. Chen. RF compensated probes for high-density discharges. Plasma Sources Sci. Technol., 3:162–168, 1994. [41] F. F. Chen, J.D. Evans, and D. Arnush. A floating potential method for measuring ion density. Phys. Plasmas, 9(4):1449–1455, 2002. [42] M. Tuszewski and J. A. Tobin. The accuracy of Langmuir probe ion density measurements in low-frequency RF discharges. Plasma Sources Sci. Technol., 5:640–647, 1996. [43] E. V. Suvorov and M. D. Tokman. Understanding Langmuir probe current-voltage characteristics. Sov. J. Plasma Phys., 15(8):540–545, 1989. [44] O. Tarvainen, T. Ropponen, T. Thuillier, J. Noland, V. Toivanen, T. Kalvas, and H. Koivisto. The role of seed electrons on the plasma breakdown and preglow of electron cyclotron resonance ion source. Rev. Sc. Instrum., 81:02A303, 2010. [45] T. Thuillier, T. Lamy, L. Latrasse, I.V. Izotov, A.V. Sidorov, Skalyga V.A., V.G. Zorin, and M. Marie-Jean. Study of Pulse Electron Cyclotron Ion Source Plasma Near Breakdown: The preglow. Rev. Sc. Instrum., 79:02A314, 2008. [46] V. Skalyga, V. Zorin, S. Golubev, S. Razin, A. Sidorov, and A. Vodopyanov. Micropulses generation in ecr breakdown stimulated by gyrotron radiation at 37.5 ghz. In Proceedings of ECRIS 2010, Grenoble, France, 2010. [47] J. S. Yoon, M. Y. Song, J. M. Han, S. H. Hwang, W. S. Chang, and B. J. Lee. Cross Sections for Electrons Collisions with Hydrogen Molecules. J. Phys. Chem. Ref. Data, 37(2):913–931, 2008. [48] R. W. Boswell. Very efficient plasma generation by whistler waves near the lower hybrid frequency. Plasma Phys. Cont. Fusion, 26(10):1147–1162, 1988.

Bibliography

187

[49] J. M. Ajello, S. K. Srivastava, and Y. L. Yung. Laboratory studies of uv emissions of H2 by electron impact. The Werner- and Lyman-band systems. Phys. Rev. A, 25(5):2485–2498, 1982. [50] M. C. Williamson, A. J. Lichtenberg, and M. A. Lieberman. Self-consistent electron cyclotron resonance absortion in a plasma with varying paramenters. J. Appl. Phys., 72(9):3924–3933, 1992. [51] M. Goto, K. Sawada, and T. Fujimoto. Relations between the ionization or recombination flux and the emission radiation for hydrogen and helium in plasma. Phys. Plasmas, 9(10):4316–4324, 2002. [52] V. Toivanen, O. Tarvainen, C. Lyneis, J. Kauppinen, J. Komppula, and H. Koivisto. Electron cyclotron resonsance ion source plasma chamber studies using a network analyser as a loaded cavity probe. Rev. Sc. Instrum., 83:02A306, 2012. [53] M. Druyvesteyn and F. Penning. The mechanism of electrical discharges in gases of low pressure. Rev. Mod. Phys., 12(2):87–174, 1940. [54] R. Xu, J. Zhao, S.X. Peng, Z.X. Yuan, Z.Z. Song, J.X. Yu, and Z.Y. Guo. Study on proton fraction of beams extracted from Electron Cyclotron Resonance Ion Source. Rev. Sc. Instrum., 79(2):B713, 2008. [55] T. Ropponen, O. Tarvainen, I. Izotov, J. Noland, V. Toivanen, D. Machicoane, G. Leitner, H. Koivisto, T. Kalvas, O. Peura, P. Jones, V. Skalyga, and V. Zorin. Studies of plasma breakdown and electron heating on 14GHz ECR ion source through measurement of plasma bremsstrahlung. Plasma Sources Sci. Technol., 20:055007, 2011. [56] E. T. Jaynes. Theory of microwave coupling system. Technical report, Naval Research Laboratory, 8 1945. CRG Report 84. [57] F. Maimone, L. Celona, F. Ciavola, G. Consoli, S. Gammino, S. Barbarino, and L. Timino. Electromagnetic study and optimization of the pm-trips ion source and the related microwave line. Technical Report INFN/TC-07/04, Instituto Nazionale de Fisica Nucleare - Laboratori Nazionali del Sud, April 2007. [58] L. Celona, G. Ciavola, S. Gammino, N. Gambino, F. Maimone, D. Mascali, and R. Miracoli. On the observation of standing waves in a cylindrical cavities filled by microwave discharge plasmas. In Proceedings of ECRIS 2008, Chicago, U.S.A., Sept 15-18 2008. [59] J. Helszajn. Ridge Wavequides and Passive Microwave Components, volume 49 of IET Electromagnetic Waves. MPG Books Ltd., Bodmin, Cornwall, UK, 2000. ISBN ISBN 978-0-85296-794-2.

Bibliography [60] D. M. Pozar. Microwave Engineering. Jhon Wiley and Sons Inc., University of Massachusetts at Amherst, fourth edition, 2012. ISBN 978-0-470-63155-3. [61] A. Mostacci. S11 fitting routines for rf cavity resonance, March 2009. Lecture at La Sapienza Universita di Roma. [62] D. Alesini. Power coupling. http://cas.web.cern.ch/cas/Denmark-2010/Lectures/, June 2010. Lecture at CERN Accelerator School, Ebelfolt, Denmark. [63] F. Consoli, S. Barbarino, L. Celona, G. Ciavola, S. Gammino, and D. Mascali. Investigation about the modes in the cylindrical cavity of an ecr ion source. In Proceedings of PIBHI 2005 and PPLA 2005, volume 160 of Radiation effects and defects in solids, page 467. Taylor and Francis, London, 2006. Taormina, June 8-11. Italy. [64] C.E. Hill and K. Langbein. Experiments on a 14.5 GHz electron cyclotron resonance source. Rev. Sc. Instrum., 69(2):643–645, 1998. [65] A. Okamoto, K. Hara, K. Nagaoka, S. Yoshimura, J. Vranges, M. Kono, and M.Y. Tanaka. Experimental observation of a tripolar vortex in a plasma. Phys. Plasmas, 10(6):2211–2216, 2003. [66] S. Biri, R. Racz, and J. Palinakas. Studies of the ECR Plasma in the Visible Ligth Range. In Proceedings of ECRIS 2010, Grenoble, France, 2010. [67] R. Racz, S. Biri, and J. Palanikas. ECR plasma photographs a a plasma diagnostic. Plasma Sources Sci. Technol., 20:025002, 2011. [68] N. Nishino, Y. Nakashima, Y. Higashizono, S. Kobashashi, K.Md. ISLAM, Y. Kubota, M. Yoshikawa, Y. Mishima, and T. Cho. First Observation of Plasma Motion in GAMMA 10 Using a Fast Camera. Plasma Fusion Res., 1:035, 2006. [69] K. Behringer and U. Fantz. The influence of opacity on hydrogen excited-state population and applications to low-temperature plasmas. New Journal of Physics, 2, 2000. [70] U. Fantz and D. W˝ underlich. Frank-Condon factors, transition probabilities, and radiative lifetimes for hydrogen molecules ans thei isotopomeres. Atomic Data and Nuclear Data Tables, 92:853–973, 2006. [71] Ya. F. Verola˘ınen and Ya. Nikolaich. Radiative lifetimes of excited states of atoms. Sov. Phys. Usp, 25(6):431–447, 1982. [72] G. Castro, D. Mascali, F.P. Romano, L. Celona, S. Gammino, D. Lanaia, R. DiGiugno, R. Miracoli, Serafino T., F. Di-Bartolo, N. Gambino, and G. Ciavola.

Bibliography

189

Comparation between off-resonance and electron Bernstein waves heating regime in a microwave discharge ion source. Rev. Sc. Instrum., 83:02B501, 2012. [73] I. V. Izotov, A. V. Sidorov, V. A. Skaliga, V. G. Zorin, T. Lamy, L. Latrasse, and T. Thuiller. Experimental and Theoretical Invertigation of the Preglow in ECRIS. IEEE Trans. Plasma Sci., 36(4):1494–1501, 2008. [74] G. Melin, F. Bourg, P. Briand, P. Ludwig, G. Gaudart, G. Girard, Hitz D., J.P. Klein, P. Ludwig, T.K. NGuyen, M. Pontonnier, and Y. Su. Status of development of ECR ion sources at Grenoble (invited). Rev. Sc. Instrum., 65(4):1051–1056, 1994. [75] C. Lyneis. ECR Ion Sources for Accelerators. In Proceedings of the 13th International Conference on Cyclotrons and their Applications, pages 301–307, Vancouver, BC, Canada, 6-10 July 1992. [76] P. Briand, R. Geller, and G. Melin. A newly designed ECR source for the lead injector of CERN. Nucl. Instrum. Meth. Phys. Res A, 294:673, 1990.

Presented herein is a record of experimental research work focused on ECR plasma dynamics with applications on ion source engineering. The results were obtained using a series of novel diagnostics that revealed the existence of phenomena that have never before been observed and that can be used to acquire a deeper understanding of both ion source physics and its applications for engineering. A wide ranging and systematic study of breakdown times; original time-resolved measurements of plasma parameters at breakdown and decay, including visible and ultraviolet spectroscopy; and the discovery of eight plasma density distribution modes, all combine to contribute to the state of the art. Moreover, the new tools developed offer the possibility to control plasma parameters in real time, which can lead to major improvements in ECR ion source performance.

July 2014 Edited by Boziztem

Contributions to ECR Plasma Source Dynamics ...

Contributions to ECR Plasma Source Dynamics ...

Suggest Documents

Ion Beam Cocktail Development and ECR Ion Source Plasma Physics

An ECR Table Plasma Generator

A Radiation Hard ECR Source for ISAC

A Radiation Hard ECR Source for ISAC

Installation of the Legnaro ECR Ion Source

Mass Spectrometry with an ECR Ion Source

Operation of the LNL ECR Ion Source

Operation of the LNL ECR Ion Source

A Vapour Source for ECR Ion Sources

Jan 31_Business Contributions to Canadian ... - Sector Source

S1 Table. Mean source contributions to invertebrate

ECR Systems Modelling Workshop - System dynamics theory

Source contributions to seasonal vegetative exposure to ozone

An advanced ECR ion source with a large uniformly distributed ECR ...

In-Situ Electron Cyclotron Resonance (ECR) Plasma Potential

ECR Based Chemically Assisted Plasma Etching ... - Semantic Scholar

ECR Plasma assisted deposition of zinc nanowires. - arXiv

Study of plasma parameters in a continuous ECR discharge sustained ...

Electric Probe Measurements in an ECR Plasma CVD ...

Discharge Characterization of 40 cm-Microwave ECR Ion Source and ...

Progress Report of the 3rd Generation ECR Ion Source Fabrication

Charge state breeding applications with the ECR PHOENIX source ...

Measurement of the emittance of an ECR ion source ...

Project of the Superconducting ECR Ion Source DECRIS-SC2