THIN FILM PHENOMENA

TABLE OF VALUES ITEM (Symbol or Abbreviation)

VALUE AND UNITS

1 angstrom (&)

10-8 cm 10-4 cm = 1 0 4 1 10" 4 cm H g - 10"3 Ton

1 micron 0*) s 1 micrometer (pra)

1 micron (pressure) 1 millimicron (ran) 1 Ton Atmospheric pressure

10l

1 electron volt (eV)

Wavelength associated with i ev Wave number associated, with 1 eV Frequency associated with 1 eV Energy associated with 3O0°K AvogadrcA number (N) Boltzmarai's constant (&) Gas constant Planck's constant (h) Speed of light (e) Electron rest mass (m) Electron charge (e}

?(4) \mcV

Classical radius of the electron

Larmor radius of a free electron

1 mm Hg 1.01 x 106:dynes/cm2 (1.6019 K 10-12 ^ . jll605 c K (23.5 kflocal/molc 1.24 x 10"4 enri 8066 cm~l 2.42 x I0 14r scc-I 25.9 x 10-3'cV 6.0225 x 10 23 /mole 1381 x 10-16 8.31 x 10 7 erg/mole/deg 6.626 x 10~27<srgsec 2^98 x 101(5 cm/sec 9.11 x 10-28 gm 4*8 x 10-10 csu = 1.602 x 10"1? ab coulomb

2J& x 1Cr1^ cm n

Occm

THIN FILM

P H E N ^^ E

KASTURI L C H O P R A Staff Scientist/ Ledgemont Laboratory, Kennecott Copper Corporation Lexington, Massachusetts Adjunct Professor of Mechanical Engineering Northeastern University Boston, Massachusetts

McGRAW-HILL BOOK COMPANY Son Francisco

London

Sydney

New York Toronto

St. Mcxico

topjs Panama

THIN FILM PHENOMENA. Copyright © 1969 by McGraw-Hill, Inc. All Rights ReservedPrinted in the United States of America. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the publisher. Library of Congress Catalog Card Number 6S-S5421 10799 34567890 MAMM 754321

TO MY PARENTS

PREFACE

"It seems good for philosophers to move tofreshways and systems; good for them to allow neither the voice of the detractor, nor the weight of ancient culture, nor the fullness of authority, to deter those who would declare their own views: In that way each age produces its own crop of new authors and new arts..FERNEL, 16th Century

Because of their potential technical value and scientific curiosity in the properties of a two-dimensional solid, thin films have been extensively studied for over a century. However, until this decade sufficient technological progress has not been made to give reasonable scientific confidence to thin film research. The developments in this decade have made, directly or indirectly, significant contributions to many areas of basic and applied solid-state research. Physical phenomena peculiar to thin films, and the basis for their study, axe generally the consequence of their planar geometry, size, and unique structure. Epitaxial growth, the occurrence of metastable structures, size-limited electron and phonon transport processes in metals, insulators, and semiconductors, quantum-mechanical tunneling through normal and superconducting metal-insulator junctions, micromagnetics, and plasma resonance absorption are some of the noteworthy contributions of thin film phenomena to solid-state physics. The technical interests which stimulated these studies have also been rewarded in the form of useful inventions such as a variety of active and passive microminiaturized components and devices, solar cells, radiation

sources and detectors, magnetic memory devices, cryotrons, bolometers, interference filters, and reflection and antireflection coatings. While progress in thin film research has been recorded, it is scattered in various scientific journals and thin film conference proceedings. Cleariy, the presentation of a unified picture of the developments in a scientific field in the form of books is just as important as the scientific research itself. There is, at present, no such up-to-date book on the physics of thin films. The popular "know-how" book "Vacuum Deposition of Thin Films" by Holland, published in 1956, and the detailed textbook "Physik diinner Schichten" by Mayer, published in 1950 (vol. 1) and 1956 (voL 2) are quite outdated and need major revisions to represent accurately the present status of our knowledge. That a detailed and authoritative account of the enormous works on thin films in a single volume or by a single author is presently very difficult has led to the present-day trend of publishing edited books comprised of independent review articles on different subjects. The four volumes of "Physics of Thin Films" edited by Hass and Thun, and the NATO conference proceedings, "The Use of Thin Films in Physical Investigations," edited by Anderson, have many excellent articles. An unavoidable lack of coherence and completeness, however, is characteristic of such works. The urgent need for a coherent, unified, and up-to-date account of thin film research inspired me to undertake the laborious task of writing this book. The book provides a comprehensive compilation of the known phenomena associated with the structural, mechanical, electrical, superconducting, magnetic, and optical behavior of films. Because of the necessary limitation on the size of a reasonable book, a detailed treatment and analysis of results is not possible, but sufficient description and literature references are provided to present a critical view of the salient features of the subjects. The experimental aspects of the basic phenomena are emphasized, although technical applications of these phenomena have also been mentioned. Wherever disorder exists in the literature, I have declared my own views and provided a reasonable synthesis. It is natural that the contents of this book draw heavily from my own research interests and scientific contributions. This is reflected in the relatively more detailed treatment of the structure and growth, and transport properties of films. Since the preparation and characterization

of films axe vital parts of thin film research, Chaps. H and ID are devoted to thin film technology. An extensive review of "Ferromagnetism in Films" (Chap. X) was contributed by Dr. Mitchell Cohen, to whom I am very grateful. In the chapters on Mechanical Effects and Optical Properties, respectively, considerable information is derived from the review articles by Hoffman, by Hass and coworkers, and by Heavens in the "Physics of Thin Films" series. My thanks are due to these authors and many others, and to several publishing companies for permission to use their published material freely. These sources are acknowledged in the t e x t The order of presentation is primarily dictated by technical cohesiveness and is not chronological. Historical references are given where some confusion exists in literature. The enormous list of references (over 2,600), kept up to date until the middle of 1968, is only a fraction of the published literature but is adequately representative of the leading and informative articles in all areas discussed. The obviously doubtful and trivial literature has been eliminated. Since commonly used symbols have been employed for each subject, some overlap was unavoidable. This work is intended for use as a reference and research book for graduate students, engineers, and research scientists. It is my earnest hope that it will succeed in inspiring new ideas and help coordinate work on thin films along more fruitful lines. If I have made some glaring errors or omissions, I hope the readers will bring them to my attention. I am grateful to M. R. Randlett and S. K. Bahl for their untiring assistance in preparing and checking the entire manuscript at various stages, and to R. Johnson for editing the draft copy. Thanks are due to D. Barclay, M. R. Randlett, and R. L. Tennis for preparing most of the illustrations. I wish to thank my colleagues W. W. Harvey, 1. L. Gelles, S. H. Gelles, R. H. Duff, J. Chen, J. Pack, and P. C. Clapp for reading parts of the manuscript. The burden of typing parts of the final manuscript was undertaken by Olga O'Brien, Joanne Johnson, Dorothy Smith, and Suzanne Archigian- It is a pleasure to thank E. W. Fletcher, the Director of Ledgemont Laboratory, for his encouragement. Finally, this book could not have been written without the constant help and understanding of my wife, Asha Chopra, who typed the first draft of the manuscript. Lexington, Massachusetts

K. L. Chopra

CONTENTS

Preface vii I

INTRODUCTION

1

n

THIN FILM DEPOSITION TECHNOLOGY

10

1. Introduction 2. Thermal Evaporation 2.1 General Considerations 2.2 Evaporation Methods Resistive Heating; Flash Evaporation; Arc Evaporation; Exploding-wire Technique; Laser Evaporation; RF Heating; Etectron-bombardment Heating 3. Cathodic Sputtering 3.1 Sputtering Process 3.2 Glow-discharge Sputtering Pressure; Deposit Distribution; Current and Voltage Dependence; Cathode; Contamination Problem; Deposition Control 3.3 Sputtering Variants

10 11 11 14

23 23 29

34 XI

4.

5.

6. 7.

3.4 Low-pressure Sputtering Magnetic Field; Assisted (Tripde) Sputtering; RF Sputtering; Ion-beam Sputtering 3.5 Reactive Sputtering 3.6 Sputtering of Multicomponent Materials Chemical Methods 4.1 Introduction 4.2 Electrodeposition Electrolytic Deposition; Electroless Deposition; Anodic Oxidation 4.3 Chemical Vapor Deposition (CVD) Pyrolysis (Thermal Decomposition); Hydrogen Reduction; Halide Disproportionation; Transfer Reactions; Polymerization 4.4 Miscellaneous Methods Vacuum-deposition Apparatus 5.1 Vacuum Systems 5.2 Substrate-deposition Technology Substrate Materials; Substrate Cleaning; Uniform and Nonuniform Deposits; Masks and Connections; Multiple-film Deposition Conclusions Appendix References

THICKNESS MEASUREMENT AND ANALYTICAL TECHNIQUES

. . .

1. Thickness Measurement 1.1 Electrical Methods Film Resistance; Capacitance Monitors; Ionization Monitors 1.2 Microbalance Monitors Microbalances; Quartz-crystal Monitor 1.3 Mechanical Method (Stylus) 1.4 Radiation-absorption and Radiation-emission Methods . . 1.5 Optical-interference Methods Photometric Method; Spectrophotometric Method; Interference Fringes; X-ray Interference Fringes 1.6 Summary of Methods 2. Analytical Techniques 2.1 Chemical Analysis. 2.2 Structural Analysis 23 Surface Structure Optical Methods; Low-energy Electron Diffraction (LEED); Auger-electron Spectroscopy; Field-emission, Field-ion, and

36

41 42 43 43 44

46

51 55 55 61

67 69 76 83

83 84 90 96 97 99

105 108 108 110 110

\

Sputter-ion Microscopy; Reflection Electron Diffraction; Replica Electron Microscopy 2.4 Volume Structure 121 X-ray Diffraction; X-ray Microscopy (Topographic Methods); Transmission Electron-diffraction and Electron-microscope Methods 3. Conclusions 129 References . . . 130 V

NUCLEATION, GROWTH, AND STRUCTURE OF FILMS . . . . . .

1. Nudeation 1.1 Condensation Process 1.2 Langmuir-Frenlcel Theory of Condensation 1.3 Theories of Nucleation Capillarity Theory; Statistical or Atomistic Theory; Miscellaneous Models; Further Deductions of the Nucleation Theories 1.4 Experimental Results Sticking Coefficient; Observations on Nucleation; Condensation Centers; Condensate Temperature 2. Growth Processes 2.1 General Description 2.2 Liquid-like Coalescence Experimental Observations; Coalescence Model 2.3 Influence of Deposition Parameters General Aspects; Kinetic-energy Effect; Oblique Deposition; Electrostatic Effects 3. Some Aspects of the Physical Structure of Films Crystallite Size; Surface Roughness; Density of Thin Films 4. Crystallographic Structure of Films. 4.1 Lattice Constant of Thin Films Size Effect; Surface Pseudomorphism 4.2 Disordered and Amorphous Structures Impurity Stabilization; Vapor Quenching (VQ): Codeposit Quenching 4.3 Abnormal Mptastable Crystalline Structures Amorphous-Crystalline Transformation; Codeposit Quenching of Metastahle Alloys; Deposition-parametercontrolled and Nucleated Metastable Structures; Pseudomorphs and Superstructures; Conclusions; Metastabilization Mechanisms 4.4 Two-dimensional Superstructures . 4.5 Fiber Texture (Oriented Overgrowth) 4.6 Alloy Superlattices

137

137 138 140 142

149

163 163 166 171

182 189 191 195

199

214 220 223

5. Epitaxial-growth Phenomenon 5.1 Influence of Substrate and Deposition Conditions . . . . Substrate; Substrate Temperature;Deposition Rate; Contamination; Film Thickness; Electrostatic Effects; Deposition Methods; Summary 5JZ Theories of Epitaxy Royer Hypothesis; van der Merwe Theory; Bruck-Engel Theory; Nucleation Theories; Summary 6. Structural Defects in Thin Films 7. Concluding Remarks References

224 225

MECHANICAL EFFECTS IN THIN FILMS

266

1. Introduction 1 2. Internal Stresses 2.1 Experimental Techniques * Ben ding-plate or-beam Methods; X-ray and Electrondif&action Methods; Other Techniques 2.2 Experimental Results Thermal Stress; Intrinsic Stress; Substrate Temperature Dependence; Thickness Dependence; Deposition Rate and Angleof-incidence Dependence; Annealing Effects; Anisotropic Stresses; Stresses in Chemically Prepared Films 2.3 Origin of Intrinsic Stress ' Thermal Effect; Volume Changes; Surface Layer; Surface Tension; Electrostatic Effects; Lattice-misfit Accommodation Model: Structural-defect Hypothesis; Crystalliteboundary Mismatch Model; Anisotropic Growth 3. Mechanical Properties 3.1 Experimental Techniques High-speed Rotor; Bulge Test; Tensile Test; Electronmicroscope Devices; Direct Measurement of Strain 3.2 Experimental Results . Stress-Strain Curves; Tensile Strength; Microhardness 3 3 Origin of the Tensile-strength Effects Structural-defect Hypothesis; Surface Effects; Volume Effects; Phenomenological Approach 3.4 Films vs. Whiskers. 3.5 Stress Relief 4. Adhesion of Films 4.1 Measurement of Adhesion 4.2 Experimental Results 4 3 Origin of Adhesion

266 267 267

238

244 252 254

271

287

295 296

298 306

310 311 313 314 316 321

/I

5. Concluding Remarks References

322 323

E L EOT RON-TRANSPORT PHENOMENA IN METAL FILMS

328

1. Introduction 328 2. Electrical Conduction in Discontinuous Films 329 2.1 Conduction Mechanisms 329 Thermionic (Schottky) Emission; Quantum-mechanical Tunneling; Activated Tunneling; Tunneling between Allowed States; Tunneling via Substrate and Traps 2.2 Experimental Results 335 2 3 Temperature Coefficient of Resistivity (TCR) 340 2.4 Network (Porous) Films 342 2.5 Elastoresistance 344 3. Electrical Conduction in Continuous Films 344 3.1 Theories of Size Effect 345 3.2 Experimental Results SB Thickness Dependence; Low-temperature Results; Size-effect Anisotropy; Magnetic Boundary Scattering; Superimposed Films 3 3 Temperature Coefficient of Resistivity (TCR) 365 Theory; Experimental Results 3.4 Specular Scattering 368 3 .5 Field Effect 375 3.6 Influence of Absorption and Adsorption on Conductivity 378 3.7 Conductivity Changes Due to Annealing 381 3.8 High-resistivity Films 387 4. Galvanomagnetic Size Effects in Thin Films 390 4.1 Introduction 390 4.2 Longitudinal Magnetoresistance 392 4 3 Transverse Magnetoresistance (Field Perpendicular to the Film Surface—HI) 396 4.4 Transverse Magnetoresistance (Field Parallel to the Film Surface—HU) 398 4.5 Hall Effect in Thin Films 401 4.6 The Anomalous Skin Effect 405 4.7 Eddy-current Size Effects 409 5. Transport of Hot Electrons 411 6. Thermal Transport 417 6.1 Thermal Conductivity 417 6.2 Thermoelectric Power 419 6 3 Heat Transport across Film-Insulator Interface 423

vn

7. Concluding Remarks References

425 425

TRANSPORT PHENOMENA IN SEMICONDUCTING FILMS

434

1. Introduction 2. Theoretical Considerations Mobility; Galvanomagnetic Surface Effects; Anisotropy Effects; Quantum Size Effects 3. Experimental Results 3.1 Size Effects 3.2 Transport Properties of Thick Films 4. Photoconduction in Semiconductor Films 4.1 Activation Process 4.2 Photoconductivity Mechanisms 4 3 High-voltage Photovoltaic Effect 5. Field Effect - Thin-film Transistor (TFT) 6. Concluding Remarks References

434 435

V H I TRANSPORT PHENOMENA !N INSULATOR FILMS

1. Introduction 2. Dielectric Properties 2.1 Thin Films 2.2 Thick Films 2 3 Dielectric Losses 3. Piezodectiic Films 4. Electrical Conduction in Insulator Kims 4.1 Conduction Mechanisms 4.2 Thermionic (Schottky) Emission 4 3 Quantum-mechanical Tunneling Theories; Image-force Correction; Temperature-field (TF) Emission; Temperature Dependence; Experimental Results; Conclusions 4 A Bulk-limited Conduction Space-charge-limited Current (SCLC) Flow; Trap and Impurity Effects 4.5 Voltage-controlled Negative Resistance (VCNR) . . . . 4.6 Current-controlled Negative Resistance (CCNR) 4.7 Tunnel Emission (Hot-electron Transport) 4.8 Tunnel Spectroscopy 5. Photoeffects in Tunnel Structures 5.1 Electroluminescence

440 440 444 452 452 454 455 456 4S9 460 465

465 468 466 470 471 477 479 479 481 483

499

503 506 508 514 516 516

5.2 Photoconduction and Photoemission 6. Concluding Remarks References SUPERCONDUCTIVITY IN THIN FILMS

518 522 522 529

1. Introduction 529 2. Basic Concepts 530 3. Transition Temperature of Thin-film Superconductors . • . . . 5 3 5 3.1 Introduction 535 32 Thickness Dependence * . - 537 3 3 Superconductivity-enhancement Phenomenon 539 3.4 Mechanisms of Enhanced Superconductivity 544 3.5 Influence of Stress 547 3.6 Influence of Impurities

4.

5. 6. 7.

8. 9.

10.

548

3.7 Electrostatic-charge (Field) Effect 549 3.8 Proximity Effects in Superimposed Films . . . . . . 550 Critical Magnetic Field 555 4.1 Type I Films 556 4.2 Type n Films 560 4.3 Mean-free-path Dependence of H ^ 562 4.4 Transverse Critical Field . 563 Critical Current '.566 Propagation of Normal and Superconducting Phases . . . . 572 Superconductive Tunneling 573 7.1 Experimental Results 576 General; Tunneling in Superimposed Films; Gapless Superconductivity 7.2 Super current (Josephson) Tunneling 582 Infrared Transmission through Thin Films -' . 586 Superconductive Thin-film Devices 588 9.1 Cryotrons 588 Wirewound Cryotron; Crossed-filra Cryotron (CFC); The Shielded CFC; In-line Cryotron; Multicrossover Cryotron (MCC); Ferromagnetic Cryotron Configuration 9.2 Computer Memory Devices 593 Crowe Cell; Persist or, Persistatron, and Continuous-film Memory (CFM) 9.3 Superconducting Magnets 595 9.4 Low-frequency Devices 596 9.5 Bolometers — Radiation Detectors 598 9.6 Tunnel Devices 598 Concluding Remarks ^ 599 References 600

FE RROMAGNET1SMJN FILMS (fay M. H. Cohen)

1. 2. 3. 4.

5.

6.

7.

8.

9.

10.

608

Introduction 608 Magnetization vs. Thickness 610 Stoner-Wohlfarth Model (A First Approximation) 614 Magnetic Anisotropy 619 4.1 Shape Anisotropy 619 4.2 Magnetocrystalline Anisotropy 620 4.3 Strain-Magnetostriction Anisotropy 621 4.4 M-induced Uniaxial Anisotropy 622 4.5 Oblique-incidence Uniaxial Anisotropy 629 4.6 Unidirectional Anisotropy 633 4.7 Perpendicular Anisotropy 633 Magnetization Ripple 635 5.1 Anisotropy-dispersion Model 638 5.2 Micromagnetic Model 639 Hoffmann's Theory; Harte's Theory Quasi-static Manifestations of Ripple 649 6.1 Rotational Hysteresis 649 6.2 HA Domain Splitting (HA Fallback) 652 6.3 Initial Susceptibility 654 6.4 Effects of W$lKD!Ku 658 Domain Walls 659 7.1 Structure of Walls 659 Bloch Walls; Ne'el Walls; Crosstie Walls; Other Wall Calculations; Experimental Results 7.2 Wall Motion 666 Domain Nucleation and Growth; Coercive Force; Wall Velocity; Magnetic Viscosity; Creep; Wall Streaming Resonance 677 8.1 Ferromagnetic Resonance 677 8.2 Spin-wave Resonance 681 8.3 Energy-loss Mechanisms 685 Switching 685 9.1 Coherent Rotation Theory 685 9^2 Experimental Results 687 9.3 Interpretation 691 Complex Magnetization Configurations 692 10.1 M Normal to Kim Plane 693 MnBi Films < 0); Stripe Domains > 0) 10.2 Multilayer Films 696 Interaction Mechanisms; Domain Wall Interactions 10.3 Cylindrical Films 698'

11. Measurement Techniques 11.1 Microscopic Observations Bitter Technique; Kerr and Faraday Magneto-optical Effects; Lorentz Electron Microscopy 11.2 Macroscopic Measurements Mechanical Detection (Torque Magnetometer); Fluxpickup Detection; Magneto-optic Detection; Magnetoresistive or Hafl-effect Detection; Ripple-independent H g Determinations 12. Applications 12.1 Flat-film Memory 12.2 Cylindrical-film Memory 12.3 Future Developments References OPTICAL PROPERTIES OF THIN FILMS

699 699

702

706 707 709 710 710 721

1. Introduction 721 2. Thin Film Optics 722 2.1 Reflection and Transmission at an Interface 723 2.2 Reflection and Transmission by a Single Film . . . . 724 2.3 Anisotropic and Inhomogeneous Films 728 2.4 Multilayer Films 728 2.5 Optical Absorption 729 3. Optical Constants of Thin Films 732 3.1 Experimental Methods 732 Reflection Methods; Reflectance and Transmittance Methods; Interferometric Methods; Spectrophotometric Methods; Critical-angle Method; Polarimetric Methods (EHrpsometry); Summary 3.2 Results on Optical Constants 741 General Remarks; Metal Films; Abnormal Absorption Phenomenon; MaxweD-Garnett Theory of Abnormal Absorption; Dielectric and Semiconducting Films 3.3 Size Effects in Optical Properties. - • 757 Semiconductor and Dielectric Films; Metal Films 4. Thin Film Absorption and Photoemission Phenomena. . . . 761 4.1 Infrared Absorption 761 Optical Modes; Surface Modes 4.2 Magneto-optical Absorption 763 4 3 Plasma-resonance Absorption 764 4.4 Ultraviolet Absorption 767 4.5 Photoemission from Metal Films 767

5. Multilayer Optical Systems 5.1 Antireflection Coatings Inhomogeneous Films; Homogeneous Single Films; Multilayer Films; Infrared Antireflection Coatings 5.2 Reflection Coatings Metal Mirrors;. All-dielectric System 53 Interference Filters -. Reflection Filters; Transmission Interference Filters; Frustrated Total-reflection Filter 5.4 Absorptance and Thermal Emittance of Coatings . . . 5.5 Thin-film Polarizers 6. Concluding Remarks References

Author Index

795

Subject Index

827

769 770

777 781

785 786 786 787

INTRODUCTION

No, 'tis not so deep as a well, nor so wide as a church-door, 'tis enough, 'twill serve.

SHAKESPEARE, Romeo and Juliet

While nonsolid films and the associated phenomenon of interference colors have been studied for over three centuries, thin solid films were probably first obtained by electrolysis in 1838. In the recorded literature, however, Bunsen and Grove obtained metal films in 1852 by means of a chemical reaction and by glow-discharge sputtering, respec* tively. Faraday obtained metal films in 1857 by the thermal evaporation on explosion of a current-carrying metal wire. The usefulness of the optical properties of metal films, and scientific curiosity about the behavior of two-dimensional solids have been responsible for the immense interest in the study of the science and technology of thin films. The varied and irTeproducible results often obtained on films led most workers to conclude that vapor-deposited films represent a high state of disorder and that no two films are alike. This chaotic state of the knowledge has actually been a blessing in disguise since it sustained and energized continued interest in thin film research.

The technology and understanding of films less than 1 micron thick have made tremendous advances in the last decade, primarily because of the industrial demand for reliable thin-film microelectronic devices to fulfill the urgent needs of the Sputnik era. This progress has brought maturity and much scientific confidence in the use of thin films for basic and applied research. In addition to major contributions to a variety of new and future scientifically based technologies, thin film studies have directly or indirectly advanced many new areas of research in solid-state physics and chemistry which are based on phenomena uniquely characteristic of the thickness, geometry, and structure of films. A discussion of these phenomena, of course, forms the central theme of this book. This chapter will introduce the uninitiated reader to some of the outstanding scientific and technological achievements based on thin film research. The details and the appropriate references are provided in the main text. Thin Film Technology (Chaps. II and J J I). The demand for clean ultrahigh-vacuum conditions during vacuum evaporation has been largely responsible for the growth of a highly specialized and scientifically based vacuum industry. In addition to the conventional oil pumps, a variety of high-speed, oil-free getter-ion, sublimation, and cryogenic pumps has emerged. Consequently, large deposition chambers maintained at ~1(T 7 to 10' 1 0 Torr are now commonplace in any laboratory. A multitude of techniques have been developed to prepare polycrystalline and nearly single-crystal films of all types of materials. Deposition rates may range from a fraction of an angstrom to thousands of angstroms per second. Among these deposition techniques, those of thermal evaporation by resistive and electron-bombardment heating, sputtering by means of glow discharge, rf and ion beams, and vapor deposition "by a variety of chemical reactions have been perfected. With the introduction of electronic monitoring and control, thermal evaporation may be carried out at definite rates of evaporation from one or more sources, and definite rates of deposition onto one or more substrates, with a high (fraction of a percent) uniformity of film thickness over large surfaces (such as that of an artificial satellite). Some of the laboratory deposition techniques have been carried over to the production stage in new industries. Notable examples of successful industrial processes are the use of thermal evaporation of A1 for aluminization of foils, inert and reactive sputtering of Ta for thin-film

microminiaturized resistors and capacitors with component densities exceeding 10 4 /cm 2 , and chemical deposition of Nb 3 Sn on foils for the winding of superconducting magnets. Several microanalytical techniques have been improved and others invented to determine the composition and microstructure of thin films. Electron diffraction and electron microscopy, which owe their present high level of sophistication to the availability of thin films, are now responsible for much of the understanding of the structure of thin films. Among the various by-products in terms of new analytical techniques, that of the moir6 fringes is significant since it is capable of revealing images of lattices and lattice defects down to a dimension of a few angstroms. Structure and Growth (Chap. 3V). The unique growth stages of a vapor-deposited film consist of a statistical process of nucleation of the vapor atoms, suxface-diffusion-controlled growth of the threedimensional nuclei, and the formation of a network structure and its subsequent filling. The most characteristic stage is that of the liquid-like coalescence of the nuclei to form the network structure. This stage plays a dominant role in the microstructure and epitaxial growth of films, and the introduction and annihilation of structural defects. Studies of these filnvgrowth stages provide an insight into the basic crystal-growth processes. The deposition parameters may be exploited to influence the kinetics of the film-growth stages and thereby obtain films with structures ranging from a completely disordered (amorphous) to a highly ordered (monocrystalline) form. Further, one can prepare films with an atomically smooth surface, or a'rough film with an effective surface area hundreds of times larger than the geometrical area. The control of deposition conditions allows the preparation of stoichiometric films of multicomponent alloys and compounds. The phenomenon of epitaxy may be used to prepare suitably oriented single-crystal films of elements, alloys, and compounds on single-crystal native or foreign substrates. The atomistic growth processes can be controlled to yield in many materials a multitude of metastabilized normal and abnormal structures which are impossible to obtain in the bulk form by any known physical method. Thus, for example, it is possible to prepare thick films of amorphous W and Zr, fee Ta and Hf, cubic Co, amorphous Au-Co alloys, metastable binary alloys over a large solubility range, sphalerite CdS, wurtzite GaAs, cubic GeTe, superperiodic structures in certain binary

alloys, and surface superstructures of a number of metals, insulators, and semiconductors. Thus, thin film techniques have opened a new and exotic dimension in materials research.. Mechanical Effects (Chap. V). Thin films are unusual specimens for the study of mechanical effects in materials in the presence of a high internal structural disorder. Vapor-deposited films are generally under enormous stresses (—109 to 10 10 dynes/cm 2 ) and further contain a high density of lattice defects which, even in the most favorable case of epitaxial films, amount to ~ 1 0 u dislocations/cm2. The level of the intrinstic stress is comparable with the yield strength of many bulk materials and should have a strong influence on the physical and mechanical stability and the properties of films. The high density of defects and the presence of free surfaces make it difficult to generate or move dislocations in a film. This condition results in the enhancement of the tensile strength of films up to 200 times the value in the corresponding bulk material, a value which is considerably larger than can be obtained by the severest cold-work treatment of the bulk material. Thus, thin films provide a medium for the study of the high yield strength and superplastic behavior of materials./' Transport Phenomena (Chaps. VI, VH, and VIII). Studies of the electronic properties of films have been largely stimulated by many attractive microelectronic-device applications. As a result of these studies, thin films of alloys, compounds, and mixtures used as resistors, and thin insulator films used in capacitors now find widespread applications in integrated circuits. The marked sensitivity of the surface conductivity of thin semiconducting films to a transverse electric field has been successfully exploited in the promising thin-film transistor (TFT). Single and multilayer thin piezoelectric films have been employed as hypersonic transducers in the gigacycle range. Because of the surface scattering of carriers in films of thickness comparable with the mean free path (mfp), the electrical and thermal transport properties of metals and semiconductors are modified. This modification, called the "size effect," depends on the ratio of the film thickness to the mfp, which may be varied by changing the film thickness and temperature, and by applying a magnetic field. Such size-effect studies have provided tests for transport theories and have also ( offered convenient and, in some cases, direct methods of determining bulk-transport parameters such as the concentration and mobility of the carriers, the mfp's and their temperature and energy dependence for

electron-phonon and electron-electron scattering, surface scattering coefficient, and Fermi surface topology. A variety of new phenomena of basic importance has emerged from these studies, e.g., the size-dependent specular scattering of electrons in metal films, thickness-dependent' oscillatory variation of the transport parameters in semimetals due to the size quantization of the energy levels, and oscillatory variation of the transverse magnetoresistance of metal films. Extensive theoretical and experimental investigations of conduction in thin insulating films have led to an understanding of the various transport mechanisms and also to the determination of the barrier parameters and barrier profile for an insulator and the insulator-electrode interfaces. By employing quantum-mechanical tunneling through thin insulators a$ a source of hot carriers, their injection, transfer, and collection properties have been studied. The possibility of using tunnel electrons to obtain a relatively temperature-insensitive and highfrequency-response tunnel (hot-electron) triode has been demonstrated. The characteristic dependence of the current through a tunnel junction on the energy-level diagram of the electrodes and, in the case of inelastic tunneling, on the excitable energy levels in the insulator forms the basis of a new, powerful technique of tunneling spectroscopy. Superconductivity in Films (Chap. IX). Films of a thickness comparable with the penetration depth and coherence length of a superconductor are ideal specimens for studying the superconducting behavior (magnetization, critical current, and critical magnetic field) of type I and type II superconductors in the light of the various theories. Such studies have played a major role in the theoretical 'understanding of superconductivity and have made it possible to determine the superconductivity parameters of penetration depth, coherence length, and the Ginzburg-Landau coupling constant. The detection and measurement of quantized flux in a cylindrical film provided the most striking verification of the existence of electron or Cooper pairs in a superconductor. The extension of the spectroscopic tool of quantum-mechanical tunneling through a thin insulator bounded by one or both superconducting electrodes provided the most direct measurement of the energy gap and the density-of-states function and thus an excellent verification of the Bardeen-Cooper-Schrieffer theory of superconductivity. The observation of the theoretically predicted potentialfree, supercurrent, or Josephson tunneling of Cooper pairs through ultrathin insulators represents one of the most celebrated contributions

of thin films to solid-state research. When a dc potential appears across a Josephson junction, microwave photons of definite frequency are emitted. Conversely, the absorption of such microwaves produces the supercurrent. A Josephson tunnel junction therefore acts as an emitter and detector of microwaves. The frequency of these microwaves has been used to determine the fundamental constant k/e with an unprecedented accuracy. The phase coherence obtained in a tunnel junction makes possible the study of electron quantum interference effects. It is easy to suppress superconductivity, but impossible to enhance it appreciably. Observations of anomalously high enhancements of the transition temperature in highly disordered, amorphous-like structures of films of some materials offer much optimism for the discovery of new high-temperature superconductors. The fact that the zero resistance state of a superconductor can be transformed reversibly to the finite resistance state, or switched to another part of the superconducting circuit by means of a suitable current or a magnetic field, forms the basic principle of a variety of promising superconductive devices. The high switching speed of the basic circuit, called a "cryotron," is achieved only by the use of thin films. Because of their simplicity of fabrication, small size, low power requirements (in microwatts), and lack of polarity dependence, with no need for connections, cryotrons are attractive switching, logic, and storage devices. Ferromagnetism in Films (Chap. X). The attractive possibilities of utilizing the extremely fast magnetization-reversal processes in thin ferromagnetic films for computer circuitry have led to extensive research on the magnetic behavior of these films. These studies have contributed new magneto-optical and magnetoelectron-optical (Lorentz-microscopy) techniqu6s, and various new phenomena which form a highly specialized field of micromagnetics. The uniqueness of the thin film ferromagnetic phenomena is due- to the geometry and microstructure of films. Owing to the pinning of an increasing number of spins at the surface of a ferromagnetic film, the saturation magnetization decreases rapidly below about 30 A thickness. The surface pinning makes it possible to excite spin waves across the film thickness, an interesting phenomenon which yields a direct measurement of the ferromagnetic exchange constant.

The planar geometry of a ferromagnetic film produces a very high demagnetizing field perpendicular to the plane as compared with that in the plane of the film. This anisotropy becomes uniaxial when the film is deposited in the presence of a magnetic field so that the magnetization is aligned with the field direction. This direction of magnetization can be reversed to the energetically equivalent direction in a very short time by means of a small magnetic field. This switching process is the attractive feature for computer applications. The static and dynamic behavior of the magnetization of a ferromagnetic film is dominated by the various magnetic anisotropies and ripple or dispersion in the uniaxial anisotropy in the fonn of quasiperiodic local angular deviations of a few degrees from the average direction. The magnetization reversal involves the rotation of domains and some motion of the domain boundaries or walls. In contrast to bulk ferromagnets, thin ferromagnetic films have domains extending throughout the film thickness. The domain walls are of Neel or cross-tie type instead of the Bloch type formed in bulk materials. This difference again arises from the high magnetostatic energy normal to the film plane so that the magnetization rotates in the film plane. Studies of these various features have contributed knowledge to thin film computer technology, basic feiromagnetism, and the structure of films. Optical Properties (Chap. XI). Since the early use of metal films in mirrors and interferometers, interest in the applications of thin metal and dielectric films in optical devices has dominated research in this field. The ingenious exploitation of the interference phenomenon in thin films has led to the development of sophisticated multilayer systems with nearly ideal reflection, antireflection, polarization, narrow- and wideband reflection and transmission filtering, and absorptance and emittance properties. Such optical coatings now find routine and indispensable industrial applications in optical instruments. The ideally suited, clean, smooth surfaces of vapor-deposited films have been utilized to determine the optical constants of numerous materials. Thin film absorption in the infrared and ultraviolet and photoemission studies have yielded valuable information on the basic electronic parameters of materials. The film geometry makes it possible to excite surface modes in an ionic lattice and affect plasma-radiation emission and absorption in metals. Studies of these absorption phenomena have thrown light on the lattice dynamics of the materials.

Future Developments. With rapid technological advances in the preparation of films with controlled, reproducible, and well-defined structures, thin films are expected to play an increasingly important role in the studies of a variety of solid-state phenomena of basic and practical interest. The properties of a large variety of new and exotic materials obtained by thin film techniques will undoubtedly draw considerable attention in the future. A multitude of thin-film optical, magnetic, electronic, and superconductive devices have been successfully operated. With increasing flexibility and diversity of the application-oriented industry, a natural course of growth and selection of new and promising devices based on complete and careful utilization of the science and technology of thin films will dominate future developments. This remark is best illustrated by the development by Weimer et al. [1] of a television camera with a completely integrated self-scanned solid-state image sensor, employing more than 105 thin-film components obtained by simple vacuum deposition techniques in a single pump-down cycle. Figure 1 shows the

a* _

P •••

; .

1

-

;

'

1

1

'

1

•

- - — — -1!—

I

•

•

>

FIG. 1. A photomicrograph of a 256-stage thin-film transistor decoder connected to a 256-element line-scan sensor. The notations stand for 256 elements each of photoconductors (?), diodes (D), capacitors (C), CdSe thin-film transistors (TFT), and resistors (R). The center-tocenter spacing between elements is about 52 pL (Courtesy of P. K. Weimer et al [1])

photomicrograph of a 256-stage thin-film transistor decoder for television scanning, driven by two 16-stage shift registers, and the associated photoconductors, diodes, capacitors, and resistors. This example represents perhaps the best exploitation of both the technology and science of thin films for an integrated microelectronic circuit of the future. REFERENCE 1. P. K. Weimer, G. Sadasiv, J. E. Meyer, L. Meray-Hovxath, and W. S. Pike, WESCON 67,13/3 August, 1967; Proc. IEEE, 55:1591 (1967).

1 1

TBI N FILM

DEPOSITION TECHNOLOGY

1. INTRODUCTION This chapter describes various methods and the related technology for preparing thin solid films for research, development, and production purposes: Although this subject cannot be dealt with in depth in this book, the summary provided here should help the reader to evaluate various techniques critically. A general description with references to important reviews and representative publications is presented. The deposition techniques for thin films may be broadly classified under three headings: thermal evaporation (physical vapor deposition), cathodic sputtering, and chemical deposition. A comparative summary of the various methods within these categories is presented in Table V at the end of 'Sec. 4.

2. THERMAL EVAPORATION 2.1 General Considerations Solid materials vaporize when heated to sufficiently high temperatures. The condensation of the vapor onto a cooler substrate yields thin solid films. Thin films of carbon deposited by evaporation inside an electric bulb were probably first observed by Edison. Historically [ 1 ], however, the deposition of a metal film from a wire exploded by a high current density is attributed to Faraday, and by vacuum evaporation, to Nahrwold, and Pohl and Pringsheim. The deposition by the thermalevaporation method is simple and very convenient and is at present the most widely used. Excellent and detailed reviews of the "know-how" of the subject are given by Holland 12, 3 ] . Because of collisions with ambient gas atoms, a fraction of the vapor atoms proportional to exp (— and He 4. A part of the ions are reflected projectiles and the rest axe ejected atoms. A considerable number of secondary electrons are generated because of their kinetic ejection [60] via interaction of the energetic projectile with the valence electrons of the target, potential ejection [61] via neutralization of the ion at the surface, and Auger (excited) emission. The fraction of secondary electrons per Ar+ and He+ ion bombardment of a W surface is ~0.1 and 0.24, respectively, for energies ~200 to 1,000 eV. Let us now say a few words about the theories of sputtering."^The hot-spot evaporation theory, formulated by von Hippel [62] and investigated by Townes [63], has yielded to the overwhelming experimental evidence, due largely to the work of Wehner [51,54], that sputtering is not an energy but rather a momentum-transfer process, as originally suggested by Stark [64]. The early momentum-transfer or collision theory of Kingdon and Langmuir [65] has seen numerous •developments. Henschke's [66] billiard-ball model, valid at low energies, considers sputtering as the result of double or triple collisions of the ion with the lattice atoms followed by its back reflection by the lattice.

More sophisticated theories, such as those of Key well [67] and others [68,69], consider sputtering as essentially a radiation-damage phenomenon. Accordingly, the incident ion displaces a number of atoms (referred to as "knock-ons") during its passage through the material and thus loses its energy, or "cools." Some fraction of the knocked-on atoms will diffuse to the surface and emerge as sputtered atoms. The knocked-on atoms may also have sufficient energy to produce additional displaced atoms which can contribute to the total sputtering yield. The exact details of the interaction of an ion with the target atom, the momentum transfer, and the collision mean free path clearly depend on its energy. With increasing energy (£), the interaction changes from billiard-ball to Coulombic and then to a nucleonic model. Pease's [69] theory gives the following expression for the sputtering yield:

where A is the cross section for imparting an energy greater than E d , the energy required to displace an atom from its lattice site, E is the mean energy of the struck atom, E s is the sublimation energy, and n is the number of atoms per unit volume. Note that in this theory, a threshold is set by the displacement energy Ed. The ejection of atoms along close-packed directions [54] can be understood on the basis of focused collision sequence (focusons) along close-packed directions as postulated by Silsbee [70]. The applicability of a focuson process for low-energy sputtering has, however, been questioned by Harrison et al. [71], and Lehmann and Sigmund [721, who have shown that ejection patterns can result simply from the regular crystal structure at the surface of a solid. 3.2 Glow-discharge Sputtering A cheap and simple source of ions for sputtering is provided by the well-known phenomenon of glow discharge due to an applied electric field between two electrodes in a gas at low pressures. A typical dc current-voltage characteristic of such a diode structure and its visual appearance are shown in Fig. 7(a) and (£>), respectively. The gas breaks down to conduct electricity when a certain minimum voltage is reached.

The attendant glow discharge maintains itself at a constant voltage and is referred to as "normal glow." The region where both voltage and current increase together is called the "abnormal glow." A luminous layer which covers the cathode partially in the normal glow and entirely in the abnormal glow is known as the "cathode glow." A fairly well defined region of relatively low luminosity known as "Crookes" or "cathode" dark space is adjacent to it. This is followed by a bright "negative glow" region, after which ill-defined regions of the Faraday dark space and the "positive column" can be seen. The cathode dark space is the most important region. Most of the applied voltage is dropped (called "cathode fall") across it. Ions and electrons created at the breakdown are accelerated across this region. The energetic electrons produce more ions by collisions with the gas atoms in the negative glow, and the energetic ions strike the cathode to produce sputtering and emit secondary electrons which are essential for

CATHODE CATHODE ' DARK NEGATIVE LAYERS \ SPACE GLOW

FARADAY DARK SPACE

POSITIVE OOLUMN

FIG. 7. (a) Typical current-voltage characteristics of an electrical discharge through gas at low pressures, (b) A visual representation of the principal regions in a low-pressure dc glow discharge.

sustaining the glow. The thickness d of the cathode dark space is inversely proportional to the pressure p of the gas (Paschen's law) such that for AT gas the product pd= 0.3 Torr-cm. The number of collisions of ike electrons traversing this region is the same at all pressures and is about 8 for Ax gas. Effective sputtering is possible only when both the number of ions and';,their energy axe large and controllable. This is conveniently effected ah the abnormal-glow discharge region. The ion energy is less than or ffiqual to the cathode fall, depending on whether or not it collides with gas'jatoms during its transit. Several factors which influence the operation of glow discharge as a technique for sputter deposition of films will now be considered. (1) Pressure. As the gas pressure is increased, the discharge current increases, the voltage falls, and the cathode dark space decreases. The ^timber of ions increases (approximately proportional to p 2 ), but their energies decrease. Since the yield increases proportionally with the number of ions, but decreases with decreasing ion energy linearly or less pan linearly in the practical range of a few kilovolts, a net increase in the total number of atoms ejected results. There is, however, an upper Emit to it, since with increasing pressure, ejected atoms suffer more collisions and are thus prevented from reaching the anode. For example [62], at 0.1 Torr, less than 10 percent of the ejected atoms may reach fbie anode. The sputtering yield of Ni[73], bombarded by 150-eV Ar ions, is fairly constant ("0.47) for pressures down to 20 mTorr and thereafter shows an apparent drop. Thus, the optimum pressure range for glow-discharge sputtering is between 25 and 75 mTorr (or microns). (2) Deposit Distribution. Because of collisions with the ambient gas atoms at high pressures, the sputtered atoms are diffusely scattered (fairing transit and therefore reach the anode with randomized directions and energies. As a result of the diffuse nature of material transport, the atoms deposit at places not necessarily in the line of sight of the cathode. Also note that, because of collisions, the energetic ions hit the cathode at high oblique angles, which actually is helpful in increasing the yield. At constant pressure and constant applied voltage, the deposition rate S low at large distances from the cathode and shows a decided maximum at the center. As this distance is decreased, a more uniform deposit first results which then becomes annular in nature with a maximum thickness on a circle slightly smaller than the target. The optimum conditions of deposition with uniformity of deposit extending to about half the area

of the target are obtained when the cathode-anode distance is about twice the length of the cathode dark space. As the anode (or any other physical obstruction) approaches the cathode dark space, l i e glow discharge extinguishes and no sputtering occurs. This fact is advantageously utilized in preventing glow discharge between the back side of the cathode or the high-tension lead wires and the neighboring support materials. It is achieved by connecting the support materials and any auxiliary electrode such as Al foil wrapped around the high-tension lead wires to anode potential, and by keeping the distance between the cathode and the nearest anode less than the cathode dark space. Ions and electrons in a plasma do not recombine efficiently because the difference in their masses makes it difficult to conserve momentum. Thus, low-temperature plasmas may diffuse at substantial distances from the cathode. Recombination, however, takes place much more easily at walls and pointed and contaminated regions of the glow-discharge geometry, which act as sinks for a diffusing plasma. The recombination, resulting in a neutral molecule, releases considerable energy as heat. The plasma-sink regions also distort the uniformity of the glow discharge and hence the sputtering rate. (3) Current and Voltage Dependence. The sputtering rate is proportional to the current for a constant voltage which is thus a very convenient control parameter. The voltage dependence is nonlinear (see Fig. 5), but for a certain range of applied voltages, depending on the gas and the target material, the glow-discharge sputtering may be operated in a linear range so that the sputtering rate is proportional to the product of current and voltage. Typical conditions employed for plane cathode sputtering are 1 to 5 kV potential with a current density 1 to 10 mA/cm 2 obtained from a rectified power supply capable of supplying up to 500 mA. A high-wattage current-limiting series resistor is essential to prevent arcing. (4) Cathode. A plane cathode of area about twice that required for a uniform deposit is used. The cathode material may be a plate, foil, or electroplated deposit onto a suitable (normally low-yield materials such as stainless steel) support target material. Because of the bombardment of ions, the cathode gets hot- The temperature increases rapidly to approach an equilibrium value. Both the rate of temperature rise and the maximum temperature attained depend on the power dissipated at the cathode, the thermal characteristics (such as conductivity and emissivity)

of the cathode, the gas pressure, etc. Typically, at 1 kV and 1 mA/cm2 the temperature of an Au target rises to 200 to 300°C in about 1 min operation. These temperatures do not significantly alter the sputtering yield, but they have other undesirable effects such as that of heating the substrate, or heating the gas resulting in changes in density and discharge conditions. Although a heavy, high-thermal-conductivity cathode may have only tolerable temperature rise, it is generally desirable to cool the cathode with lunning water or some other cooling fluid. Several cooling systems are described in the literature [74j. In addition to plane cathodes, wire [751, cylindrical [76], and concave [2] cathodes have also been studied. The wire geometry which is useful for deposition inside a cylindrical substrate enhances ion bombardment and thus gives deposition rates that axe considerably higher than those obtained by plane cathodes. The concave cathode concentrates bombardment between the boundary and center of the cathode, thus producing a deposit roughly annular in shape with little deposit in the middle. Mention should also be made of the multiplecathode designs [74,77-79] which allow simultaneous or sequential sputtering from various cathodes. (5) Contamination Problem. Even if a leakproof sputtering system is initially pumped down to a high vacuum (~10 - 8 Torr) and then sputtering gas of high purity (the commercially available high-purity Ar generally requires passing through a cold trap to remove oil and water vapors) is admitted, contaminants may still appear from (1) the outgassing as a result of plasma-discharge heating of the walls and contaminated components of the sputtering chamber which are not adequately grounded or shielded, and (2) the decomposition of oil vapors as a result of back streaming from the diffusion pump operated at high pressures. Since only fluid pumps (mechanical and/or diffusion) can be used for pumping a high-pressure sputtering chamber, the use of an optically dense baffle, preferably cooled, and some provision for throttling of the diffusion pump axe highly advisable. Mass-spectrometer analysis [80,81] of the composition of the background gases before, during, and after sputtering shows an immediate decrease in the concentration of reactive gases such as 0 2 , N 2 , and water vapors, but a sharp increase of H 2 occurs during sputtering. This is possibly due to cracking of higher-mass hydrocarbons in the glow discharge. The presence of H 2 influences the sputtering yield considerably [81]. Analysis by flash photolysis (Chap. Ill, Sec. 2.1) of the gas

content of both glow-discharge and ion-beam low-pressure sputtered films of Mo prepared in the author's laboratory confirms the presence of large amounts (>3 percent) of H 2 in both high- and low-pressure (~10" 5 Torr) sputtered films. In addition to chemical sorption of the ambient gases by the film, Winters and Kay [230] have shown that an impact-activated sorption of the sputtering gas occurs which increases rapidly in the case of Ni films sputtered with Ax ions of energy greater than 100 eV. Depending on the deposition conditions, a concentration of 10"1 to 10~4 Ax atoms per Ni atom was found. (6) Deposition Control. One of the chief advantages of the sputtering technique is that the rate of deposition remains constant with time, provided the current density and voltage do not vary, a condition that is easily attained by using an automatic pressure controller and a regulated power supply. One can also control the rate by controlling the discharge-current density electronically. A quartz-crystal oscillator (discussed later) may be used for monitoring and controlling rate by means of a feedback mechanism to control the discharge current. It is important, however, to position the monitor so that it does not disturb the plasma and is well shielded. A small magnet may be used to deflect the ions. 3.3 Sputtering Variants Several systems have been employed for deposition of films by sputtering. These sputtering variants axe shown schematically in Fig. 8. The simplest and most widely used one (thoroughly discussed in Ref. 2, among others) utilizes the glow discharge between two electrodes and is commonly referred to as a diode arrangement (