Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Ion implantation patterning of high temperature superconducting thin films and multilayers Wong, Andre Wing Gai 1999

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1999-390055.pdf [ 10.98MB ]
JSON: 831-1.0085566.json
JSON-LD: 831-1.0085566-ld.json
RDF/XML (Pretty): 831-1.0085566-rdf.xml
RDF/JSON: 831-1.0085566-rdf.json
Turtle: 831-1.0085566-turtle.txt
N-Triples: 831-1.0085566-rdf-ntriples.txt
Original Record: 831-1.0085566-source.json
Full Text

Full Text

ION IMPLANTATION PATTERNING OF HIGH T E M P E R A T U R E SUPERCONDUCTING THIN FILMS AND MULTILAYERS B y Andre W i n g Gai Wong B. Sc. (Physics) Dalhousie University, 1992 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES PHYSICS AND ASTRONOMY We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA 1999 © Andre W i n g G a i Wong, 1999 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Physics and Astronomy The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V 6 T 1W5 Date: Abstract This thesis was motivated by the suggestion that selectively implanting Y B a 2 C u 3 0 7 ( Y B C O ) films with a highly reactive ion, such as Si, could pattern without destroying or removing material. If true, this would greatly simplify conventional methods of pattern-ing multilayer structures. This led to systematic studies on the use of ion implantation to pattern Y B C O thin films and multilayers. The samples used were obtained by developing a method referred to as scanning pulsed laser deposition. This technique resulted in the reproducible growth of highly crystalline Y B C O thin films with high transition temperatures and critical currents. A study of Si implantation was first done to elucidate the exact mechanism by which it rendered Y B C O non-superconducting. Measurements of films implanted at various energies and doses revealed that implantation at required doses severely damaged the films crystalline structure, destroying superconductivity. The growth of high quality multilayers require that the underlying patterned film retain its as-grown crystalline quality. A damaged film can regain that quality (to some extent) with high temperature annealing. Measurements of annealed implanted films revealed that crystalline damage at levels >10 displacements per atom (dpa) could not be completely removed at accessible annealing temperatures. However, when the im-plantation damage was kept below 1-2 dpa, an annealing temperature of ~ 9 0 0 ° C was successful in recovering most of the original structure. Unfortunately, at these tempera-tures, the Si-implanted film phase separated, forming islands of a Si-mixed material in a sea of Y B C O , and thus regained its original high transition temperature. To address the original claims that Si implantation would not destroy the film, a ii film was rendered non-superconducting with the appearance that it maintained its crys-tallinity by implanting only near the films surface. In this case, the film was effectively passivated and a low temperature anneal resulted in oxygen leaving the underlying Y B C O structure to the more energetically favored Si02 states in the implanted layer. Replenish-ment of oxygen from the atmosphere was hindered due to the passivating layer capping the film. Patterning films by implanting Si was thus deemed to be unsuitable for multilayer structures. A new technique of substitutional ion implantation patterning was developed using M g ions. M g substitutes for the C u in the C u - 0 planes, drastically reducing transi-tion temperature with very low concentrations. The low required concentrations allowed the use of low implant doses. This, coupled with the relatively low mass of the M g ion, reduced the implantation damage to levels easily removed with high temperature annealing. As well, the high temperature anneal can incorporate M g into the Y B C O matrix, forming the compound, Y B a 2 ( C u i _ x M g x) 307. The process of implanting M g into Y B C O , followed by a high temperature anneal resulted in the formation of a highly crystalline, non-superconducting material at 77K. M g implantation was used to success-fully pattern films with a resolution of 10 ^m. Bilayers with a top Y B C O layer and a bottom YBa 2 (Cui_ x Mg3;)307 were fabricated. Resistivity and x-ray measurements reveal the high quality of both layers. iii Table of Contents Abstract ii List of Tables vii List of Figures viii Acknowledgment xii Preface xiv 1 Introduction 1 1.1 The History 1 1.2 Basic Properties of Y B a 2 C u 3 0 7 4 1.2.1 Oxygen Doping 4 1.2.2 Anisotropy 8 1.3 H T S T h i n Films 10 1.4 H T S T h i n F i l m Patterning 13 1.5 Issues with Multilayer Patterning 15 2 YBCO Film Growth and Processing 17 2.1 F i l m Growth 17 2.1.1 Introduction 17 2.1.2 Control of Substrate Temperature 20 2.1.3 Composition Control 21 iv 2.1.4 Growth 23 2.1.5 Characterization 25 2.1.6 Discussion 30 2.1.7 Defects 33 2.2 Annealing 36 2.2.1 Low Temperature Annealing 37 2.2.2 High Temperature Annealing 38 2.3 Concluding Remarks 39 3 Reactive ion implantation 41 3.1 Original Experiments and Interpretation 42 3.2 Inconsistencies with the Literature 43 3.3 The Experiments 47 3.4 New Interpretation 56 3.5 Titanium Implantation 57 3.6 Recovery of Crystallinity 61 3.7 Concluding Remarks 63 4 Substitutional Ion Implantation 65 4.1 Impurity Doping in Y B C O 65 4.1.1 Y Site Substitution 66 4.1.2 B a Site Substitution 67 4.1.3 C u Site Substitution 67 4.1.4 Addition of Impurities 71 4.1.5 Incorporation via Implantation 71 4.1.6 Choice of Implant Ion 72 4.2 Magnesium Ion Implantation 73 v 4.2.1 Experiment 73 4.2.2 Determination of the M g Concentration 74 4.2.3 Results 75 4.2.4 Discussion and Concluding Remarks 82 5 Structuring with Mg ion implantation 85 5.1 Patterning 85 5.2 Bilayer growth 89 5.2.1 Characterization 89 5.3 Concluding remarks 96 6 Conclusions and Future Improvements 97 Appendices 100 A X-Ray Diffraction Measurements 100 B SQUID Magnetization Measurements 108 B . l Zero Field Cooled(ZFC) Measurements 112 B.2 Field Cooled(FC) Measurement 113 C Diffusion 115 Bibliography 117 vi L is t of Tab les 2.1 Measured transition temperature, rocking curve width and c-axis length of all films grown in a four month period. The residual resistivities (ex-trapolated from the normal state) are given for a few of the films. Critical current measurements were performed on two films of differing quality. The numbers assigned to the samples correspond to the order in which they were grown 31 3.1 Summary of the work in the literature on ion implantation into H T S films and crystals. The threshold damage is the level where there was no de-tectable crystallinity 45 4.1 Summary of doping work with Y B C O 67 vii List of Figures 1.1 Structure of Y B a 2 C u 3 0 7 ( Y B C O ) 5 1.2 Schematic of the phase diagram of Y B C O showing the effect of oxygen doping on the structural and superconducting properties 6 1.3 a, b and c axes lengths as a function of oxygen content 7 1.4 Electrical resistivities (pc, pa and pb) of Y B a 2 C u 3 O x 9 1.5 H - T phase diagrams of a 3D, a 2D and a thin film superconductor . . . . 11 1.6 Illustration of the birth and spread growth scheme 12 1.7 Illustration of the growth of a spiral defect 14 1.8 Schematic o'f a H T S crossover and a via connect 16 2.1 Temperature Calibration of Block Heater 22 2.2 Pulsed Laser Deposition System 24 2.3 X-ray 9/29 scan of Y B C O film 26 2.4 X-ray rocking curve of a Y B C O thin film 27 2.5 Temperature dependence of magnetization and resistivity 28 2.6 Remnant field map 29 2.7 Transition temperature as a function of rocking curve width 32 2.8 S E M micrograph of an as-grown film . . . 34 2.9 Atomic Force Microscope line scan of pinhole 35 2.10 Diagram of annealing furnace 37 2.11 Magnetization measurements of deoxygenated films annealed ex-situ at temperatures ranging from 2 5 0 ° C to 4 0 0 ° C 38 viii 2.12 S E M micrograph of high temperature annealed film 39 3.1 Previous d.c. magnetization curves of silicon implanted thin films . . . . 43 3.2 Previous x-ray diffraction curves of the silicon implanted thin films . . . 44 3.3 TRIM-92 prediction of Si ion and atomic vacancy distributions 48 3.4 Rocking curve width as a function of Si ion implant dose 49 3.5 d.c. magnetization curves of silicon implanted films (120nm) as a function of dose and annealing temperature 50 3.6 d.c. magnetization curves of Si implanted Y B C O films at a dose of 3 x l 0 1 6 c m - 2 52 3.7 X-ray diffraction scans of Si implanted films 53 3.8 Transition temperatures and c-axis lengths vs annealing temperature for Si implanted films 54 3.9 Transition temperatures and c-axis lengths versus ion dose for Si implanted films 55 3.10 Illustration of new deoxygenation interpretation 58 3.11 X-ray diffraction measurements on T i implanted films annealed at 9 0 0 ° C 59 3.12 d.c. magnetization measurements of T i implanted films 60 3.13 Recrystallization from annealing as a function of implantation damage . . 62 4.1 a and b axes lengths as a function of impurity doping into Y B C O . . . . 68 4.2 Transition temperature of YBa 2 (Cui_ : E M a ; )307 ( M = Zn, A l , Co, A l and Ga) as a function of x 70 4.3 TRIM-92 prediction of the M g and damage profiles 74 4.4 SIMS depth profile of Y , C u and M g in a YBa 2Cu307/YBa2(Mgo .oo8Cuo.99 2) 3 07 bilayer 75 4.5 d.c. magnetization curves of M g implanted films 76 4.6 Transition temperatures of implanted films versus annealing temperature 77 ix 4.7 X-ray diffraction ( X R D ) scans of films implanted with M g and annealed at 9 0 0 ° C . . . 79 4.8 X-ray rocking curves of the (005) Bragg peak of films implanted with M g and annealed at 9 0 0 ° C 80 4.9 Temperature dependence of the d.c. resistivity of M g implanted thin films 81 5.1 S E M micrograph of the backscattered electrons of a film patterned using M g ion implantation 86 5.2 Electron Probe Microanalysis(EPMA) of a border between an implanted and an unimplanted region of a patterned film 87 5.3 S E M micrograph of the backscattered electrons of a Y B C O film patterned using M g ion implantation at a dose of l x l O 1 6 c m - 2 88 5.4 Resistivity curves of a YBa 2Cu307/YBa2(Mgo.oo5Cuo.995)3 0 7 bilayer . . . 90 5.5 X R D scan of a YBa2Cu307 /YBa 2 (Mgo.oo5Cuo.995)3 07 bilayer . . . . . . . 91 5.6 X R D scan of (0013) Bilayer Bragg peak 92 5.7 Secondary Ion Mass Spectra of the M g depth profile in a Y B M g C O / Y B C O bilayer with and without annealing 94 5.8 SIMS depth profile of M g implanted bilayers 95 6.1 Potential processing method of a H T S flux transformer structure using ion implantation 98 A.1 Illustration of x-ray diffraction from a line of periodic points 101 A.2 Illustration of 9/29 x-ray diffraction in k-space 103 A.3 Illustration of the Bede x-ray double crystal diffractometer 105 A.4 Illustration of a rocking curve in k-space 106 x A . 5 X-ray rocking curves of a Y B C O film (005) (*), very high purity (vhp) Y B C O crystal grown in Y S Z crucibles (006) (•), ultra high purity (uhp) Y B C O crystal grown in B a Z r 0 3 crucibles (006) (•) and a Si crystal (111) (<) 107 B. l Illustration of the Meissner Effect 109 B.2 Schematic of the S Q U I D Magnetometer detection system I l l B.3 Illustration of a Zero Field Cooled (ZFC) and a Field Cooled (FC) sample 114 xi Acknowledgment I would first like to express my sincerest gratitude to my thesis supervisor, Dr. Walter Hardy. Although his guidance was not in the form of daily supervision, I was always made to feel that I could seek his advice whenever necessary. It was very reassuring to know that someone with his knowledge and wisdom was just an e-mail message, or a door-knock, away. As well, I would like to thank the other members of my Ph .D. thesis committee, Drs Jim Carolan, Tom Tiedje, Birger Bergersen and Nick Jaeger, for providing some useful advice during the course of my thesis work. Extended thanks are given to Nick Jaeger for providing the facilities at the Center for Advanced Technology in Microelectronics for the photolithography and ion implantation. Gratitude is also extended to Dr. Doug Bonn for providing the financial support for myself and this thesis work. Several people were important for the completion of this thesis. Firstly, Dr. Q .Y . M a for providing early guidance and later making it possible for me to continue the implantation work with his resources at Columbia University. Secondly, Dr. Ruixing Liang, who stepped in at an important juncture of my thesis to assist in the growth of the films. His meticulous and devoted approach to solving materials science problems is one that I am striving to emulate. Finally, Dr. Massoud Badaye, who joined the lab during the last two years of my studies, was a source of comfort and companionship which enabled me to see the light at the end of the tunnel. This work was quite multidisciplinary and required the assistance of people in various departments within the university. These include Dr. Al ina Kulpa, Center for Advanced Technology in Microelectronics, Hiroshi Kato, formerly of the Department of Electrical xn Engineering and Dr. Mat i Raudsepp, Department of Earth and Ocean Sciences. O n a more personal note, I met met many friends who have made the past six years at U . B . C . very enjoyable. In the lab, there are Saeid Kamal , Ahmed Hosseini, Pinder Dosanjh, Chris Bidinosti, Patrick Turner, and Richard Harris, and some others who have since left. In addition, I thank Michael Gardner for his measurements on my films. There are a few of us who started graduate school together six years ago, and have been constant companions; Lori Paniak, Michael Saliba, Alex Busch, Nick Fameli, Ermias Gete and Makoto Fujiwara. In particular, I have formed strong friendships with Ermias and Makoto and as our paths lead us in vastly different directions, I hope we can maintain that friendship for many years to come. Finally, the most gratitude must be extended to sweet Rose who has always been there for me. Thank you. xiii Preface In this age of specialized research, where P h D graduates are criticized for knowing more and more about less and less, it was important (at least for the author, and probably even more so for the reader) to think of where this research fits with the larger scope. As these words are being typed in the early months of 1998, the world is experiencing a technological revolution. When I began university, I was excited about the prospect of using a P C with 8 M H z clock speed. Now as I prepare to leave, the Pentium 200 M H z P C is already commercially obsolete. The remarkable aspect is that, now, literally a world of information is accessible to almost anyone. What was once only the privilege of the academic and military elite, is now available at all universities, most public schools and libraries and even in some coffee shops. How did this happen? The bridge from the unique tool to one accessible to all is crossed when the tool can be made well in large quantities. In the computer and information industry it was the ability to reproducibly fabricate devices on a micron scale over an area of 10s of c m 2 that enabled this bridge to be crossed. Crystal growth, patterning and processing had to be achieved with the highest of precision, billions of times each day. That accomplishment was due to advancements in the materials science of semiconductors. When the demand for information truly becomes global (the Internet is a North Amer-ican phenomenon - less than 1% of the world's population has access to a computer), the present material of choice, silicon, will unlikely continue to be. The electronic material of the next generation is undetermined, and a potential candidate are the high temperature superconductors. If that is the case, the same materials science would need to be fine tuned again before this material can become as ubiquitous as its predecessor. xiv This thesis tells a materials science story about the high temperature superconductor, Y B a 2 C u 3 0 7 ( Y B C O ) . It is a difficult material to grow and pattern well. Thus, devices made with this material are expensive and unlikely to be a fixture in everyday commercial products. This thesis confronts this problem by developing a new method of patterning this material that could make patterning much simpler. This method is to use ion implantation to pattern Y B C O . In chapter 1, the reader will be introduced to the materials aspects of Y B C O , Y B C O thin films and H T S multilayers. Present day methods of patterning devices will also be described. The detailed process of growing and characterizing films for this study is given in Chapter 2, as well as some other processing issues. Chapter 3 will provide the details of reactive ion implantation, the early work done in patterning Y B C O thin films using ion implantation including the old interpretation of the mechanisms. This leads to a systematic study to understand reactive ion implantation, resulting in a modification of the old interpretation and a new idea for ion implantation patterning, substitutional ion implantation. Chapter 4 provides the details of substitutional ion implantation, a method better suited for multilayer patterning. This is described along with a background of impurity doping into Y B C O . Finally, chapter 5 demonstrates the successful fabrication of some multilayer structures by substitutional ion implantation. xv Chapter 1 Introduction The motivation of this thesis was to develop a new and simpler method of patterning high temperature superconducting (HTS) multilayers using ion implantation. In this introductory chapter, the original idea, its origin, and where the research for this thesis began will be presented. This is followed by an overview of some of the relevant basic properties of YBa2Cu3C>7_,5(the most studied of the H T S materials and the material used in this thesis), H T S thin films and the patterning of H T S thin films and multilayers. 1.1 The History Wil l the H T S materials become the electronic material of the next generation? It has been nearly twelve years since its discovery, and although H T S materials are making headway in some niche areas (passive microwave filters are most notable) there is effectively no industry comparable to that for other electronic materials. 1 What is the problem? The complete answer to that question cannot be answered simply. However, it is safe to say that what needs to be improved is the materials science of the H T S materials. For microelectronics applications, the issue now is not how can one make a high quality Josephson device, but how can one make 107 stable devices performing at nearly identical specifications. A reliable method of growing high quality, large area, thin films to serve as wafers is required, as well as a scheme for fabricating and patterning devices on these 1The Low Temperature Superconductor(LTS) industry had over 3 billion dollars in revenue[l] in 1995 compared with <100 million in the HTS industry(1997)[2]. This is not to mention its much more imposing competitors in the microelectronics industry. 1 Chapter 1. Introduction 2 wafers with the potential of large scale integration. The research of former research associate at U B C , and now member of the faculty at Columbia University, Dr .Q.Y.Ma, was focused on this issue and initiated the work comprised in this thesis. It is therefore relevant now to discuss the original concept (introduced by Q .Y .Ma in [3]) of using ion implantation for patterning high temperature superconducting thin films and multilayers. From the moment these materials were discovered, numerous attempts have been made to incorporate semiconductor materials with superconductors in hybrid systems. It was this desire that led to the patterning of superconductors with unconventional techniques. In an attempt to grow YBa2Cu307(YBCO) on silicon wafers, it was immediately found that silicon could easily diffuse into Y B C O , degrading its superconducting properties[4]. This led to the idea of Selective Epitaxial Growth (SEG). 2 A pattern of silicon was first deposited onto the substrate, and a Y B C O film was subsequently grown on top. The portions that grew on top of the silicon would be degraded or "poisoned" from the silicon diffusion. Modifications of this technique have included using other predeposition materials (e.g. Ti[7], Si xNj,[8], T i N or W[9]), which frustrate the subsequent growth, instead of diffusing into the Y B C O , improving the resolution of the technique. The clear advantage of SEG was that patterning could be achieved without removing material or exposing it to possible contaminants. However, SEG could not be used beyond the first deposited Y B C O layer since the Si diffused (or growth frustrated), insulating portions of the film were no longer crystalline, and epitaxy would not be maintained with subsequent depositions. Since patterning of only single layers is limited to a small range of devices, the next step was to improve this method of patterning to expand its appli-cability to multilayer structures. Many HTS devices require several patterned epitaxial 2 This was not a completely new idea, and had been previously used to pattern semiconductors [5] and optoelectronic devices[6]. However, this was the first application to HTS materials. Chapter 1. Introduction 3 layers (e.g. Josephson junctions[10, 11, 12], S Q U I D magnetometers[13, 14] and infrared bolometers[15]). Often the resultant device is made with three or four epitaxial layers of superconductor and dielectrics. Because of the short coherence length of H T S materials, and their highly anisotropic nature, the H T S films in multilayer structures must be grown as nearly single crystalline and as flat as possible. Thus, present day multilayer pattern-ing is tedious, requiring a mixed recipe of various lithographic and etching techniques (for example see references [11] and [12]). A reliable fabrication method of multilayer struc-tures will be required for commercialization of any such devices. The incorporation of an impurity for patterning Y B C O would be most advantageous in multilayer patterning, since each layer would remain planar. However, if patterning using an impurity such as Si was to be applied to multilayers, a scheme for incorporating it into the material without destroying its crystallinity was required. Ion implantation has long been used in the semiconductor industry for incorporat-ing dopants, since very fast and large scale processing is routine. As well, a wealth of information[16, 17] exists regarding the modeling of ion transport in materials, and the recrystallization of radiation damaged material. Implanting silicon into Y B C O was con-ceived as an alternative method of poisoning the material without necessarily degrading the crystallinity and was successful in patterning single layer devices[18, 19]. The early results led to optimistic predictions and it was thought that reactive ion implantation (as it became referred to) could easily become a standard H T S multilayer patterning technique. Looking deeper into the problem revealed some fundamental problems, i.e. the interpretation by which the silicon poisoned the superconductivity was hand-wavy - to quote a phrase popular among physicists - and at the very least required further understanding. As well, the limitations of this method were not known. This was the starting point for this thesis. First, we back track and briefly overview some of the basic properties relevant for this thesis. Chapter 1. Introduction 4 1.2 B a s i c P r o p e r t i e s of Y B a 2 C u 3 0 7 The most commonly studied and used material is Y B a 2 C u 3 0 7 ( Y B C O ) . First discovered in 1987 [20], Y B C O has an easily accessible transition temperature and is relatively easy to fabricate. There exists a wealth of information regarding its properties, and the interested reader is directed to the collection of reviews edited by Ginsberg[21]. A select number of relevant aspects are described. 1.2.1 O x y g e n D o p i n g One of the more interesting and complex issues of the H T S materials, in particular Y B C O , is the dependence of its properties on the atomic configuration of oxygen. Struc-turally, electronically, and magnetically, the material can be drastically altered with slight changes in oxygen content. S t r u c t u r a l The structure of Y B C O is shown in figure 1.1. The unit cell is layered and comprised of alternating layers of yttrium atoms sandwiched by a pair of C u 0 2 planes and two B a O layers sandwiching a CuOi_<5 layer. For simplicity, this layered structure can be thought of as having the following stacking sequence: B a O - C u O i _ 5 - B a O - C u 0 2 -Y - C u 0 2 . The Cu( l ) is often referred to as the chain copper site and the 0,(5) and 0(1) as the chain site oxygens. The 0(4) is called the apical or interstitial oxygen. When S in CuOi_,$ is zero, the material is fully oxygenated, conversely when 6=1, the material is completely deoxygenated. Figure 1.1 shows a fully oxygenated unit cell. The oxygens in the C u O i _ a chain layer are sufficiently weakly bound so that even at temperatures far below the melting temperature, the system will develop a finite oxygen pressure, which is a function of the stoichiometry and temperature. This accounts for the ease at which oxygen can be inserted and removed from the lattice. At low oxygen content (6 ~1), the Chapter 1. Introduction 5 Figure 1.1: Three unit cells of fully oxygenated Y B C O . This image was constructed from one downloaded from the website Chapter 1. Introduction 6 probability of occupying either the 0 ( 1 ) or 0(5) site are roughly equal and the unit cell is, on average, tetragonal with the a and b lattice constants equal. At lower values of S, a phase transition occurs and the probability of occupying the 0(5) site decreases with an increase in probability of occupying the 0 (1 ) site. Eventually, as the oxygen content is increased above 6.4/unit cell, 1 dimensional chains are formed along the 6-axis, resulting in a orthorhombic unit cell. This is illustrated with the phase diagram in figure 1.2. 250 _ 200 a> 2 150 co Q. E a> c 100 o '.^  'to c CO 50 0 Orthorhombic Tetragonal S C 0 0.25 0.5 0.75 8 in YBa^Cup^g 600 500 400 CD 300 1.00 3 CD 0) »—*• c —* CD 200 ^ 100 Figure 1.2: Schematic of the phase diagram of Y B C O showing the effect of oxygen doping on the structural and superconducting properties. A n antiferromagnetic(AF) region of Y B C O exists for 8 > 0.6, and the material is not superconducting. At lower values of 5, the material becomes a superconductor (SC), with T c increasing with decreasing S. The a,b, and c-axes lattice parameters are very sensitive to the oxygen content, as seen in figure 1.3, where the a and b axes lengths are given for bulk powder samples as a Chapter 1. Introduction 7 function of oxygen content. The c axis lengths are given for thin films as well. 3 . The c-axis length is easily measurable in c-axis oriented thin films, and can provide information on the oxygen content of the film. 11.90 11.85 3.89 , 3.88 <^  3.87 x: B) c 3.86 W 3.85 X CO 3.84 JD CO 3.83 3.82 t 11-80 c o co 11.75 x CC o 11.70 11.65 • MBE film A Sputtered film • powder • a A [ . rj • Aa a n • A A ° ° ••• • • i i 0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 8 in Y B a 2 C u 3 0 7 . 5 Figure 1.3: a, b and c axes lengths as a function of oxygen content. This shows the sensitivity of Y B C O ' s structure to oxygen content (from reference [24]). Electronic and Magnetic The effect of changing the oxygen content from 6=0 to 6=1 is even more dramatic with respect to Y B C O ' s electronic properties. Figure 1.2 also shows a schematic representation of the phase diagram. At 6=1, Y B C O is an antiferro-magnetic insulator, and as 6 is decreased past 0.6, superconductivity is induced, reaching a maximum transition temperature of 93K for the optimally doped samples. In the in-sulating region the coppers in the C u 0 2 plane are divalent (Cu 2 + ) with the unpaired electron spin having 3D long range antiferromagnetic order. As more oxygen is added to the chains, the Cu(l) (nearest neighbor to the inserted oxygen) becomes oxidized, altering the value from the previously monovalent Cu to a divalent copper. As the chains 3The a and 6 axes lengths were unavailable for these thin films from x-ray diffraction Chapter 1. Introduction 8 get longer in length, the energetics favor a transfer of electrons from the planes to the chains, doping the planes with holes, and destroying the long range antiferromagnetic order. The addition of enough oxygen creates long chains, transferring a large number of holes to the copper-oxide planes. The charge carriers in the planes are electron holes. At sufficiently high hole doping, the material undergoes a metal-insulator transition (which coincides with the orthorhombic-tetragonal transition), described in [22]. The transition temperature increases monotonically with decreasing 5, but plateaus at around 5 ~0.4, and then continues to increase with further addition of oxygen. The transition temper-ature peaks at a value of 5=0.07, and additional oxygen doping suppresses T c . Oxygen doping levels below the optimal are referred to as the underdoped region and above this is dubbed an overdoped system. R e v e r s i b i l i t y Any of the previously described properties are reversible. The oxygen content in Y B C O can be set by annealing the sample with the appropriate annealing con-ditions. The dependence of S on oxygen pressure and temperature is well documented[23], and the content can be tuned to any value from 6=0 to 5=1 from any other. 1.2.2 A n i s o t r o p y The anisotropic nature of all the H T S materials is one of their more important and interesting properties. Indeed, anisotropy is an important factor for these materials being superconductors at all. It is believed the the high transition temperatures are a result of the two dimensional nature. Y B C O , whose d.c. resistivity plots at various oxygen doping levels are shown in figure 1.4, is one of the least anisotropic, yet still exhibits a c axis resistivity two orders of magnitude larger than that of the a or 6 directions. Other measurements of the anisotropy are listed in reference [2 5]. Consequences of the large anisotropy are that the vortices are Chapter 1. Introduction 9 102 101 6.68 10° - Pc /A™ -y /,6.88 ivity (Q. cm -1 10 2 / 6.93 - . CO 'lO cn 10 * «»••• •••••••• •-rr lO"3 lo"4 10 5 • • • • • • n f p p (6.93) I I I I I I I 0 1 0 0 2 0 0 3 0 0 Temperature (K) Figure 1.4: Electrical resistivities (pc, pa and pb) of YBa2Cu30 x as a function of temper-ature at various values of x. The c-axis resistivity is two orders larger than that of the a or b axes directions. From reference [26]. Chapter 1. Introduction 10 not as rigid as in a 3D superconductor, and are commonly viewed as stacked pancakes, loosely coupled, instead of long rigid tubes. A vortex glass instead of a rigid lattice is formed which melts into a vortex liquid at fields lower than the upper critical field(a vortex liquid is much more difficult to pin than a vortex lattice). For many applications the anisotropy constrains the usefulness of the H T S materials by limiting the operating temperatures to undesirably low values. Anisotropy also requires that the grains within the material be perfectly aligned if the current carrying capabilities of the material are not to be compromised. Control of the orientation during crystal growth is vital to obtain the desired properties. Misalignment of the c-axis, in particular, will result in degraded transport behavior. 1.3 H T S Thin Films The main drive behind the development of H T S thin films is microelectronic applica-tions(SQUIDs, microwave components), and has been pursued for more than a decade. The applications now have expanded to include high current carrying wires 4. There exist many schemes to grow H T S thin films, the more popular ones being sputtering, laser ablation and Metal-Organic Chemical Vapor Deposition ( M O C V D ) . For this work, the chosen method was laser ablation, and a detailed description is given in chapter 2. The typical H T S thin film is c-axis oriented and has a thickness on the order of several microns or less. Its thickness is less than the c-axis penetration depth, and therefore has effectively no lower critical field. Even in the absence of any applied magnetic field, there are vortices within the film, albeit with zero net vorticity. The vortex-vortex interaction is much lower in the 2D case, and vortices can therefore be created and annihilated much more easily. Wi th the application of a field, there will exist a number of vortices 4 A recent breakthrough in the fabrication of HTS wires has been to grow films on buffered rolled Ni substrates, which in principle can be made to arbitrarily large lengths[29] Chapter 1. Introduction 11 aligned in the direction of the applied field, with a number of pairs of vortices with zero net vorticity. The thinner the film, the more the system behaves as a two dimensional superconductor. The H-T phase diagram changes drastically, as shown in figure 1.5. The Meissner state almost disappears and a distinct line exists between the vortex lattice and vortex liquid states. Figure 1.5: H-T phase diagrams of a 3D, a 2D and a thin film superconductor. The 2D superconductor has a phase diagram much different than the typical 3D superconductor. There is essentially no Meissner state, and the vortices are more loosely pinned resulting in a vortex lattice melting line (denoted by T T O 2 D ) occuring much below H c i . The phase diagram of a HTS thin film resembles the 2D superconductor, yet still retains some of the 3D qualities. One of the biggest issues with HTS thin films is the nature and source of the defects. It is clear that all "good" films are highly defected, as indicated by the very high critical currents observed. The reason for this can be understood by considering the growth processes. Growth of HTS thin films can be considered as a sort of "cheating", in that to avoid the inaccessible conditions required for equilibrium growth, non-equilibrium techniques are employed to achieve growth. As well, there is no perfect match for a 3D superconductor 2D superconductor thin film Chapter 1. Introduction 12 substrate , a n d thus l a t t i ce m i s m a t c h o f at least 2% is a lways present. S o m e t h i n g must be c o m p r o m i s e d . W h a t is c o m p r o m i s e d is the g r o w t h mode . In t h i n f i l m g rowth , i t is most desi rable to have a layer by layer g r o w t h m o d e (also referred to as the van der Merwe mode) . In th i s g r o w t h mode , each mono laye r o f the f i l m covers the surface before the subsequent layer begins . If s t r a in exists (say, due to a m i s m a t c h i n l a t t i ce parameters ) , then this g r o w t h m o d e w i l l break d o w n at w h a t is referred to as the t h re sho ld thickness . B e y o n d th is th ickness , the s t r a in can no longer be a c c o m m o d a t e d by the f i l m , a n d i t "cracks" , r e su l t i ng i n defects ex tend ing t h r o u g h o u t the f i l m . If the misma tches are very large between the subs t ra te and f i lm l a t t i ce parameters , the layer by layer g r o w t h cannot be m a i n t a i n e d , a n d an i s l and (or Volmer-Weber ) g r o w t h mode , s chema t i ca l l y shown i n figure 1.6, develops. In th is g r o w t h m o d e , the film is fo rmed by the coalescence of sp read ing discrete i s lands , f o r m i n g an in te rconnec ted ne twork s t ruc ture . T h e spreading rate of the i s lands depend s t rong ly on the subst ra te t empera tu re , the ambien t pressure a n d the energy o f the a toms as they are depos i t ed onto the substra te . A n o t h e r g rowth scheme, d u b b e d the Stranski-Krastanow mode , is a h y b r i d of the two where the i n i t i a l g r o w t h is a m i x t u r e of layer by layer a n d the i s l a n d g r o w t h modes. F i g u r e 1.6: I l l u s t r a t i o n of the b i r t h a n d spread g r o w t h scheme. T h i s g r o w t h scheme results i n a h i g h l y or iented , bu t defected film. The re is consensus tha t the g r o w t h scheme for laser ab la ted films is the i s l a n d g rowth , or 2 D nuc lea t ion growth[30, 31, 32]. T h i s g r o w t h scheme has been d i r e c t l y observed by Chapter 1. Introduction 13 High Resolution Transmission Electron Microscopy (HRTEM)[33], where distinct islands were evident after only what was supposed to be the growth of a couple of layers of unit cells. The island growth scheme has some serious consequences for the microstructure of the film. Problems arise because it is very unlikely that two or several different islands meet exactly, and the point of coalescence becomes a source of a defect which can extend throughout the thickness of the film. (i.e. stacking fault, dislocation core). A striking example of how such a defect can form is the screw dislocation. The mismatch between coalescing islands becomes a nucleation point for the screw dislocation. The growth of Y B C O is very anisotropic and is much quicker in the a-b plane than perpendicular to it. As more material is added during growth, the preference is to grow within the a-b plane, spiraling about the defect. This is illustrated in figure 1.7. For the films used in this work, this example of the screw dislocation was not promi-nent for reasons unclear to the author (neither the Atomic Force nor the Scanning Elec-tron Microscopes revealed this structure on any of the samples). It has been recently demonstrated[34], that the growth of spiral defects are very sensitive to growth conditions, and it is likely that this sort of defect could have been induced with slight modifications of the growth conditions (probably at the cost of the other film properties). A study of the formation of spirals in laser ablated films would be a thesis on its own, requiring extensive use of high resolution microscopy. That was not consistent with the goal of this thesis and was not pursued. However, other consequences of the island growth were quite apparent in as-grown films, described in chapter 2. 1.4 H T S Thin Film Patterning Patterning H T S thin films is done in a way similar to that for other materials, and can be classified into three different categories. The first two methods involve the removal Chapter 1. Introduction 14 Figure 1.7: The birth and spread growth scheme produces many defect points that can result in defects extending throughout the thickness of the film. One striking example are screw dislocations. This defect, which likely results from the poor coalescence of two islands, becomes a nucleation point. As more material is added, it grows in a spiral about the nucleation point. of material by etching. Firstly, patterning can be done on ex-situ grown films, where etching is done before the final annealing step. In this case the film is deposited at room temperature and standard lift-off techniques are done on a non-crystalline film[35, 36]. The patterned film undergoes a final anneal step to obtain a crystalline Y B C O structure. The second class, and that used much more extensively, is to pattern the film after it is completed. A standard photo resist mask is placed on the film and wet etched using H3PO4, HC1, or HN03[37, 38] Alternative methods of etching include ion beam[39], focused ion beam[40], plasma[41] and chemical plasma[42] etching. A technique of laser writing directly onto the film has been used[43], and as well as the use of an STM[44] to Chapter 1. Introduction 15 induce abrasion into the film, patterning lines as narrow as 35 nm. Both techniques have the advantage of removing material without the use of a photoresist mask, but have the disadvantage of being a much slower process so that large scale patterning is unlikely to be practical. The third class includes methods which achieve patterning without the removal of material, and rely on the sensitivity of the properties of Y B C O to impurities and heat or radiation treatment. Methods have included the use of ion[45] or electron[46] beams to locally amorphisize the film, rendering the film insulating. Locquetei.a/.[47] have proposed a technique which can electrochemically oxidize or reduce the material on a local scale. Features were made with widths of 20 /^m. As already alluded to, S E G patterns by frustrating the growth of the film. This technique has been used to pattern very narrow features, such as nanobridges[8] as well as SQUIDs[48], and bolometers[49]. 1.5 Issues with Multilayer Patterning The difficulty with multilayer patterning, as briefly mentioned, is maintaining flat, single crystalline films. The two greatest source of degradation are the insulating crossovers and via connects, which are shown schematically in figure 1.8. Wi th crossovers, the top Y B C O film must be deposited over an underlying step edge. It has been shown that large angle (> few degrees) step edges often result in a large angle grain boundaries in the overlying film[50], which severely degrade the critical current[51]. To maintain high critical currents, the ramp slope can be made very gradual, but this is not an ideal solution since this is more space consuming and will decrease the level of integration. Other alternatives are to grow thicker films so that it is higher than the step height. However larger film thicknesses are more difficult to grow well and often have poor crystallinity and rough surfaces. The critical current density has been observed to decrease with Chapter 1. Introduction 16 increased film thickness/step height ratio[52]. Pianarization of the multilayers is another alternative[53], however, this requires several more processing steps. HTS film Crossover Via connect Figure 1.8: Schematic of a H T S crossover and a via connect. The via-connects electrically link a top layer superconducting film to a bottom one. The critical current of the via is often very much degraded. No ideal method of repro-ducibly making via-connects exist. State-of-the-art H T S multilayer processing used by industry (TRW[11], Northrup Grumman[12]) and large research institutions (NIST-Boulder[28]) use sophisticated recipes to maintain high performance in their multilayer devices, but it is not obvious how these recipes could be scaled to large or very large scale integration. Ideally, for multilayer patterning, one would want a scheme where material is not removed or destroyed. How-ever, presently no such scheme exists. This thesis develops such a scheme, by using the knowledge of the physics of the H T S materials and the tools of semiconductor processing. Chapter 2 Y B C O Film Growth and Processing The first challenge of this research was to obtain sufficiently high quality samples. It was vital that the macroscopic material properties of the samples were reproducibly satisfactory. In this chapter, details of the film optimization and characterization are discussed. This is followed by a description of ex-situ annealing experiments conducted on these films. 2.1 Film Growth 2.1.1 Introduction For all H T S device applications, the ideal material is a single crystal thin film. Despite the mature status of YBa2Cu 3 Or(YBCO) film growth and success in obtaining superior superconducting properties, the crystallinity of the films, as measured by x-ray diffrac-tion ( X R D ) , still does not approach the limit imposed by the substrate. Typically Y B C O thin films, of thicknesses ~ 100 nm, grown on SrTi03(STO) exhibit (005) rocking curve widths (a description of rocking curves is included in Appendix A) in the range 0 .1°-0.3° [54, 55, 56, 57, 58], while the (200) rocking curve widths of high quality S T O sub-strates are less than 0 .03°( fu l l width at half maximum). To our knowledge, the H T S films with the highest crystallinity are Y B C O grown on N d G a 0 3 [58] substrates and N d B a 2 C u 3 0 7 ( N B C O ) on S r T i 0 3 ( S T O ) [59], which exhibit (005) rocking curve widths of about 0 .05° . For Y B C O films of this thickness, grown on S T O , the narrowest published 17 Chapter 2. YBCO Film Growth and Processing 18 (005) rocking curve width was 0 .11° [56]. In this section, the details of the film optimiza-tion is discussed resulting in the reproducible growth, by pulsed laser deposition(PLD),of highly crystalline Y B C O thin films on SrTiOa substrates with an average (005) rocking curve width, Aw = 0 . 0 7 ° ± 0 . 0 2 ° , and a narrowest measured width of 0 .037° . O f the many methods of depositing Y B C O thin films (the interested reader is referred to reference [60]), one of the more popular is Pulsed Laser Deposition (PLD) or Laser Ablation. Laser ablation is popular for H T S thin film growth because of the complex composition of these materials. If we consider the various methods of growing thin films, the most obvious (and common) method is thermal evaporation. The vapor can be deposited onto a substrate placed appropriately near the evaporating source. Methods such as evaporation and Molecular Beam Epitaxy ( M B E ) work on this principle with a separate target for each element in the final film. In these schemes, very large area and very good composition control can be obtained. Wi th the h igh-T c materials, which have a more complicated composition than any other technologically important material, this scheme of thin film deposition is not so easy 1. To grow Y B C O using thermal sources requires several targets, one for each of the metallic elements, Y , B a and C u . A more popular alternative is to work with a single stoichiometric target, but since the volatilities of each of the elements (Y, Ba, C u , O) are so diverse, thermal heating is nearly impossible. A n attractive alternative is P L D . W i t h P L D , an ultra-violet laser pulse of width < 40ns is focussed onto the stoichiometric target (or several) comprised of the material to be deposited. The energy of a single laser pulse is typically several hundred millijoules, resulting in a transfer of power of over 100 M W to the target. This results in the process of ablation, which is the formation of a plasma from the ionized and dissociated atoms in the target. The velocity of the plasma is directed perpendicular to the surface of 1Although, for very large area (>2 inches diameter)films, co-evaporation is becoming a method of choice Chapter 2. YBCO Film Growth and Processing 19 the target, and the placement of a heated substrate at an appropriate distance from the target receives the ablated material. Pulsed laser deposition is a difficult method of film growth to control, as many factors can seriously degrade the quality of the film. To reproducibly produce high quality films, all growth parameters, as well as the the target and substrate quality had to be carefully selected. The substrate, SrTiOa single crystals, were initally obtained from several vendors and characterization had to be done on each and every single sample. This involved measurement of the x-ray rocking curves of the (200) Bragg peak and observation of the surface using the S E M (Atomic Force Microscope was not available at the time). Once a reliable vendor was determined, this characterization was done only on a representative from each batch. Equally important was the determination of a reliable source of a Y B C O target. Characterization of the target included 9/29 X R D measurements, to probe for impurity phases, and determination of the target density. A benchmark for the density was found to be ~ 87% (denser targets were not required to obtain high quality films). Our final choices of vendors were Applied Technology Enterprises2 for the substrates and Superconductive Components3 for the targets. The handling of the substrates and the films required the following precautions. 1) The substrates were shipped in sealed bags with packages of dessicant. After re-moval from the packaging, care was taken to handle the substratea; Teflon tweezers were used to prevent chipping of the edges. 2) The substrates were cleaned in boiling acetone ( H P L C grade) for a few minutes, then in boiling isopropanal(HPLC grade). The substrates were dried with nitrogen gas from liquid nitrogen dewars, for cleaniness. 3) After cleaning, the substrate surface was inspected under the optical microscope 2 P.O. Box 1622, Irmo, South Carolina 29063 31145 Cheasapeake Ave., Columbus OH 43212 Chapter 2. YBCO Film Growth and Processing 20 and re-cleaned if necessary. Clean substrates were stored in standard Fluoroware storage containers. Throughout the handling, special attention was given to avoid any contact with the substrate surface. The target also required some special treatment. After each filmtgrowth, the surface of the target was cleaned using a sharp edge of a clean piece of a used Y S Z (Yttria Stabilized Zirconia) crucible (used for growing Y B C O single crystals). Handling of the targets was always done while wearing protective gloves and any additional contact with the surface was avoided. Optimization of the film growth parameters required an additional year of constant research. Early on, the reproducibility of the films' macroscopic properties was poor. Depending on the day, the transition temperature, crystallinity and surface varied wildly, despite attempts to maintain constant laser energy, substrate temperature, and oxygen pressure. However, with the assistance of research associate, Dr. Ruixing Liang 5 , a breakthrough came with the realization that control of substrate temperature and, in particular, c o m p o s i t i o n , were most vital. What follows is a description of our method of controlling these two parameters. 2.1.2 C o n t r o l of Subs tra te T e m p e r a t u r e The substrate was heated with a US Thin Film Products Inc.6(model US-GUN-II) 2 inch diameter block heater. This is comprised of a heating element embedded into a block of non-corrosive alloy of nickel, iron and molybdenum. A thermocouple was placed into a hole drilled into the block, such that the tip of the thermocouple was directly behind of the center of the face of the block. Temperature control was achieved with a feedback circuit from the output of the thermocouple to the current into the heating element. A 4 w w w .fluoroware. com 5Department of Physics and Astronomy, U.B.C. 6 Chapter 2. YBCO Film Growth and Processing 21 temperature stability of ± 2 ° C was obtained. However, in this temperature measurement, the thermocouple does not directly measure the temperature of the substrate, only the temperature of the heater block. Thus the temperature of the substrate must be properly calibrated with respect to the temperature of the heater. This calibration would not provide an absolute temperature of the substrate. However, this was not vital, since during optimization of the film growth, a large range of temperatures was scanned. What was important was the reproducibility of the temperature, especially when changes in the system(e.p. changing thermocouples in the heater) occured. The calibration was done by placing a type K thermocouple on the surface of a substrate which was attached to the heater with a conducting paste. Data was taken by increasing the temperature of the block heater in units of 20 degrees, maintaining it at that temperature until a steady state was reached. Measurements from the heater thermocouple and the reading from the substrate thermocouple were compared. Since the films are grown in a reduced oxygen pressure, and cooled in one atmosphere of oxygen, the conditions for calibration change in these temperature ranges. Near the growth temperature ( ~ 8 0 0 ° C ) , the pressure in the chamber was increased to 130 mTorr (the growth pressure). At temperatures near the final annealing temperature of 4 3 0 ° C , the growth chamber was filled with oxygen to near atmospheric pressure. A plot of the substrate temperature versus the heater temperature is shown in figure 2.1. 2.1.3 Composition Control Control of composition is essential for the growth of highly crystalline thin films[61, 62]. In the fabrication of bulk Y B C O , one has good control of composition where 5 nines pure starting materials can be accurately weighed to several significant figure precision. For thin film growth of Y B C O by P L D , the control is much poorer since the composition of the film depends on the composition of the plume at the substrate surface. Although Chapter 2. YBCO Film Growth and Processing 22 200 1 • • •— 200 400 600 800 Heater Temperature (°C) Figure 2.1: Substrate temperature as a function of heater temperature at various chamber pressures. one can start with a very pure and stoichiometric target, it has been observed[62] that the composition at the substrate surface can vary drastically with distance from the center of the plume (for example, the center of the plume tends to be slightly copper rich while its edges are copper deficient). In this case one can attempt to make precise adjustments to the deposition system such that the composition at the substrate surface would happen to be the same as the target. This can be achieved by conditioning the beam to obtain a more homogeneous intensity at the target [63] or by placing the substrate at a particular angle relative to the plume[64]. In general, these methods provide poor reproducibility since any slight changes to the system parameters (e.g. slightly defocused laser beam,target location not precisely the same) disrupt the optimal condition. In what follows, we describe a more reproducible method of composition control, based on the Chapter 2. YBCO Film Growth and Processing 23 belief that, except for the initial few laser shots, the composition within the entire volume of the plume is the same as that of the target. In particular, one arranges to average out the spatial compositional inhomogeneity at the substrate by moving the plume relative to the substrate. Over the course of a typical film growth (1500 laser shots), the copper deficient edges of the plume can, for example, compensate for the copper rich center of the plume. If the plume was not moved relative to the substrate, then the center of the film and its edges would remain non-stoichiometric. (We have, in fact, observed that when the plume was not moved, there was a ring about the center of the deposited area with a high T c , while the center and edges had very low T c ' s ) . In our system, the diameter of the plume at the substrate surface was roughly 2 cm; therefore to ensure that all portions of the plume were directed reasonably evenly to all areas of the substrate, the plume must be moved at least 1 cm along the radius of the target (circular with radius of 1.5 cm) in all directions. To achieve this, the laser spot (demagnified by lenses to a slightly defocused circular spot of diameter of 0.3 cm at the target) was circularly scanned about the entire front surface of the target by repeating a path consisting of three concentric circles, the largest circle having a diameter close to that of the target. See the inset in figure 2.2. This was done to cover as much area of the circular target as possible with the laser scan. Recently, W u[65] have described a related type of scanning P L D system that was motivated by the desire to produce large area, uniform thickness, Y B C O films. 2.1.4 Growth The scanning P L D system is shown schematically in figure 2.2. The K r F excimer laser (A=248nm) was followed by an aperture, two lenses and two mirrors, to direct the laser beam into the vacuum chamber. The beam was scanned using a computer controlled mirror having two axes of rotation. The substrate was heated by adhering it, with a Chapter 2. YBCO Film Growth and Processing 24 Excimer Laser Laser beam path Computer controlled mirror to vacuum chamber Focusing lens Figure 2.2: A schematic of the pulsed laser deposition system for the growth of Y B a 2 C u 3 0 7 thin films. The inset shows the scanning pattern of the laser beam on the target. Scanning the beam on the target was done to move the plume relative to the substrate in order to even out the compositional inhomogeneities in the plume. In addition, scanning the beam prevents the-laser beam from repeatedly ablating the same spot on the target. conductive paste, to a resistive heater. Special care was taken to obtain homogeneous thermal contact. For optimization, films were grown at various substrate temperatures, oxygen pressure, laser fluence, substrate-target distances, and scanning schemes. The optimal deposition parameters, for this system, were found to be a substrate temperature of 7 9 0 ° C , oxygen pressure of 130 mTorr, laser fluence of 1.2 J c m - 2 , a substrate-target distance of 4.8 cm and the circular scan previously described. After growth, films were first held at the growth temperature and pressure for 5 minutes, then cooled slowly ( 1 0 ° C / m i n ) in a 200 Torr oxygen atmosphere to 6 0 0 ° C , at which point the chamber was filled with oxygen to a pressure of 750 Tofr. After maintaining the film at this temperature for 30 minutes, it was cooled at an even slower rate ( 5 ° C / m i n ) to 4 3 0 ° C , where it was annealed for 90 minutes, after which the heater was turned off. A l l samples Chapter 2. YBCO Film Growth and Processing 25 used in this study were grown under these optimal conditions. 2.1.5 C h a r a c t e r i z a t i o n After growth, the films (of nominal size 0.5 cm x 0.5 cm x 120-170 nm), were all char-acterized by X-Ray Diffraction ( X R D ) and d.c. magnetization, and selected films by magneto-optic and resistivity measurements. Details of the X-ray diffraction and d.c. magnetization measurements are given in Appendix A and Appendix B , and are briefly described here. The measurements were conducted with a Rigaku rotating anode x-ray generator with two different diffractometers. A Rigaku powder diffractometer was used to verify the alignment of the films, detect impurity phases and determine the c-axis length. After careful alignment of the film with respect to its (00/) Bragg peaks, 8/26 diffraction spectra were obtained by slowly scanning ( 1 ° / min) from 5 to 70 degrees at the highest usable power level. Additional scans at higher angles were also performed to determine the precise location of higher order Bragg peaks (up to (0013)). A full scan of 6/26 is shown in figure 2.3. The 6/29 scans indicate highly c-axis oriented films. The 6/26 peak widths were sufficiently sharp that Bragg peaks from the K a i and K q 2 x-rays are discernible already in the (004) peak. The precise c-axis lengths were obtained by the usual technique of plotting the individual c-axis values (calculated from 26 Bragg angles of the (001) reflections) versus the function, i(6) — (l/(sin((9) + 1/6) * cos2#, and extrapolating to i(9) = 0, described in more detail in Appendix A . The mean c-axis length was 1 . 1 6 8 6 ± . 0 0 0 4 nm, consistent with the c-axis length of very high purity, fully oxygenated Y B C O single crystals, 1.1685 nm. The c-axis lengths of all films are listed in table 1. A Bede double axis diffractometer , which provided a highly collimated and monochromatic x-ray beam, was used to determine the width of the (005) rocking curves and 6/26 peaks. The very narrow angular dispersion was achieved through the use of the Bragg reflections from a channel cut (220) silicon crystal collimator, which was then Chapter 2. YBCO Film Growth and Processing 26 followed by a (111) silicon crystal monochrometer (with an angular resolution of 0 . 0 0 3 ° ) . The narrowest beam width available on the apparatus was used at both the detector and monochrometer output. Rocking curve lineshapes and widths were insensitive to the choice of slit width at the detector. Representative rocking curve scans of the (005) peak of three films are shown in figure 2.4 where symbols represent data points and the lines are simple Lorentzian fits to the data. 10° ' • I 5 10 15 20 25 30 35 40 45 50 55 60 65 70 2theta (degree) Figure 2.3: A x-ray 6/29 scan of sample 14 (see table for sample identification). The inset (with a linear vertical scale) shows that the separation of the K a i and K q 2 peaks is discernable already in the (004) Bragg peak. The films' superconducting properties were determined by measuring the temperature Chapter 2. YBCO Film Growth and Processing 27 -0.2 -0.1 0.0 0.1 0.2 Aco(degree) Figure 2.4: (005) X-ray rocking curves of samples 10 (circles), 17 (diamonds) and 3 (squares). The inset shows the (200) rocking curve typical of good SrTiC>3 substrates. Lines represent Lorentzian fits to the data. dependence of the diamagnetic signal using a commercially obtained S Q U I D magnetome-ter (see Appendix B). Measurements were taken while warming the sample in an applied field of 3 Oe (oriented normal to the film surface), after the film was cooled to 5K in zero applied field. The temperature dependence of the d.c. resistivity, measured in the same cryostat used for the magnetization measurement, was obtained using the standard four probe measurement (with current reversal) at a current density of 0.3 A era ' 2 . Although magnetization measurements were conducted on all films, the temperature dependence of resistivity was measured only on a selection of the films. This was because resistivity mea-surements required the placement of A u contacts, which prevented usage of the samples Chapter 2. YBCO Film Growth and Processing 28 300 03 -15 . . . . I 0 50 100 150 200 250 300 Temperature (K) Figure 2.5: Magnetization (black) and resistivity(grey) (measured in the same cryostat) as a function of temperature of sample 17. The transition temperature of this film was taken to be 90.1 K for other experiments. The magnetization and resistivity as a function of temperature of one sample are both shown in figure 2.5. The transition temperature was chosen to be the temperature where the diamagnetic signal strength was 1% of the the temperature independent signal strength at lower (<85K) temperatures. This value of the transition temperature was typically 0.3-0.4 K lower than the temperature at which zero resistiv-ity was observed. The transition width, measured by magnetization, was roughly 5 K, while the transition width, measured by resistivity was 0.8K. The additional broadening in magnetization can be attributed to the large demagnetization factor of a thin film Chapter 2. YBCO Film Growth and Processing 29 or iented p e r p e n d i c u l a r to the magne t i c f ie ld. T h e c r i t i c a l cur rent a n d superconduc t ive homogene i ty o f the films were o b t a i n e d w i t h a magne to -op t i c i m a g i n g m e t h o d , developed by fel low U B C P h . D student , M i c h a e l G a r d -ner, a n d desc r ibed i n more de t a i l i n [66]. T w o films, one w i t h a h i g h T c a n d na r row r o c k i n g curve (sample 17 i n tab le 1), the o ther w i t h a r e l a t ive ly b r o a d r o c k i n g curve w i d t h a n d lower T c ( sample 3) , were selected for i m a g i n g . F i g u r e 2.6: R e m n a n t field m a p at 80 K of sample 17 o b t a i n e d us ing a magne to -op t i c technique w h i c h provides a l o c a l B 2 m a p , w i t h a r e so lu t ion of 14 mic rons , o f the film. T h e supe rconduc t ive homogene i ty of the f i l m is evident f rom the close resemblence of the image to the p red ic t ed roof top pa t t e rn . F i l m d imens ions are r o u g h l y 0 .4cm x 0 .5cm x 170nm A magne t i c field was a p p l i e d u n t i l comple te pene t r a t i on of the film was observed. T h e field was then removed , a n d the remnant field i n the film i m a g e d w i t h a magne to -op t i c Chapter 2. YBCO Film Growth and Processing 30 sensitive garnet film and a C C D camera. The result is a local Bz map of the film, with a spatial resolution of 14 microns, resulting from the distribution of pinning sites in the film. The remnant field map for sample 17 is shown in 2.6. The superconductive homogeneity of the film is apparent from the fact that the trapped field pattern closely resembles the ideal "rooftop" pattern, which is expected for a homogeneous distribution of pinning sites in a rectangular shaped film. The corresponding currents in the film, which would result in this field shape, can be easily calculated using an iterative algorithm and the relation between field and current. The calculated value for the currents give the lower bound for the critical currents in the film. These lower bounds (which varied slightly from one region of a sample to another) were found to be (at 80K) 4.4 x l 0 6 A c m - 2 (a maximum of 5 .0xl0 6 A c m _ 2 a n d a minimum of 3 .8xl0 6 A c m - 2 ) for sample 17 and 4.2 x l 0 6 A c m - 2 (maximum 5.2xl0 6 A c m - 2 , minimum 3.6xl0 6 A c m - 2 ) for sample 3. Considering that T c is slightly higher for sample 17, the two films of quite different structural quality (Aw = 0 .105° for sample 3 and Au; = 0 .066° for sample 17) exhibit very similar critical current densities. 2.1.6 Discussion Table 2.1 summarizes the characteristics of all films grown during a four month period. The reproducibility of the film growth process is evident, since all films grown exhibit rocking curve widths less than 0 .11° (already lower than any other published value for Y B C O films of this thickness on S T O ) and transition temperatures higher than 89K. In the present scanning scheme, the desire was to reproducibly obtain a uniform composition throughout the surface of the substrate, limiting inhomogeneities to the c-axis direction. This was because it was unreasonable to expect that surface diffusion would average out large scale (e.g. substrate dimension) compositional inhomogeneities. However, on smaller length scales (~ film thickness), diffusion can be relied upon to achieve a uniform Chapter 2. YBCO Film Growth and Processing 31 S a m p l e (005)Aw w i d t h T c c-axis Po J c @ 8 0 K (degree) ( K ) (nm) (fifl cm) ( A c m " 2 ) 1 0.063 89.4 1.16810 2 0.083 89.7 1.16827 3 0.105 88.7 1.16834 10 4.2xl0 6 4 0.079 89.9 1.16830 5 0.097 89.5 1.16844 6 0.081 89.7 1.16878 7 0.093 89.1 1.16879 8 0.054 89.9 1.16877 9 0.079 89.9 1.16917 7 10 0.037 90.3 1.16927 4 11 0.042 90.1 1.16910 12 0.044 90.1 1.16917 13 0.104 89.3 1.16866 12 14 0.066 89.7 1.16845 15 0.079 89.9 1.16850 16 0.086 89.7 1.16830 17 0.066 90.1 1.16851 5 4.4xl0 6 M e a n 0 . 0 7 ± 0 . 0 2 8 9 . 7 ± 0 . 4 1 . 1 6 8 6 ± 0 . 0 0 0 4 Table 2.1: Measured transition temperature, rocking curve width and c-axis length of all films grown in a four month period. The residual resistivities (extrapolated from the normal state) are given for a few of the films. Critical current measurements were per-formed on two films of differing quality. The numbers assigned to the samples correspond to the order in which they were grown. Chapter 2. YBCO Film Growth and Processing 32 compositional distribution. Ideally, a full scan of the target would correspond to the growth of as few layers of Y B C O as possible as this would provide uniformity within each monolayer. In this scanning system, one full scan corresponded roughly to the growth of 5 layers of Y B C O (~ 6 nm). A faster scan was not possible due to limitations imposed by the scanning system. Still, good uniformity in the a-b plane was achieved over 5 unit cells, and it reasonable to expect diffusion in the c-axis to readily occur on this length scale. Figure 2.7: Transition temperature of all films plotted as a function of their (005) rocking curve width. The solid line is only a guide to the eye. The broadening observed in all rocking curves (figure 2.4), were well fitted within experimental error to Lorentzian lineshapes. A . Gauzzi [67] have discussed the type of disorder in ion-beam sputtered Y B C O thin films that would result in Lorentzian Chapter 2. YBCO Film Growth and Processing 33 broadening of the rocking curves. It was suggested that a reduction of long range lat-tice order from homogeneously dispersed defect disorder, resulting in localization, was responsible for the observed Lorentzian broadening and T c suppression. Figure 2.7 is a plot of transition temperature as a function of rocking curve width; a clear trend is evident as T c suppression is correlated with rocking curve broadening. If the film were comprised of a granular structure, then degradation of superconductivity would manifest itself more in a suppression of critical current density, as even slight increases in grain boundary angles can drastically reduce critical current densities[51]. However, two films showed quite different rocking curve widths and yet exhibited very similar critical cur-rent values. As well, scanning electron micrographs of the film surface did not reveal any granularity. These factors suggest that the films have single crystal-like structure. Using the treatment of A. Gauzzi, the lattice correlation length (distance over which the atomic positions are correlated) can be obtained from the rocking curve widths using the following expression, where rc is the correlation length, d the c-axis length, n the order of the Bragg reflection and Au;, the rocking curve width. Thus, for the narrowest rocking curve width, we obtain rc=115 nm, and for the broadest curve, r c=40 nm. 2.1.7 Defects Most of the characterization techniques previously mentioned are blind to the microscopic defects that pervade most Y B C O laser ablated thin films. The large number of defects can be strongly linked to the growth mechanism. In these films, several defect structures were apparent. Chapter 2. YBCO Film Growth and Processing 34 O n e of the more obvious defect s t ructures were p inholes d ispersed u n i f o r m l y t h rough-out the f i l m surface, evident f rom the S E M m i c r o g r a p h s h o w n i n figure 2.8. T h e pinholes were e x a m i n e d more c losely w i t h an A t o m i c Force M i c r o s c o p e ( A F M ) . A profile of a t y p i c a l p inho le is shown i n figure 2.9. T h e p inholes h a d d iamete r s on the scale o f 200nm, a n d depths of 10 n m . A l t h o u g h these features do not degrade ei ther the c r i t i c a l current or the t r a n s i t i o n t empera tu re , they pose serious p rob lems for m u l t i l a y e r g r o w t h or h i g h -re so lu t ion pa t t e rn ing . U s i n g h i g h t empera tu re annea l ing , desc r ibed i n a la ter sect ion, we c o u l d essent ia l ly e l im ina t e a l l p inholes w i t h o u t deg rada t ion o f the f i l m proper t ies . F i g u r e 2.8: S E M m i c r o g r a p h o f an as-grown f i l m . C l e a r l y evident are p inholes (black arrows) a n d ou tg rowths (whi te arrows) d ispersed t h r o u g h o u t the f i l m surface. O u t g r o w t h s , apparent f rom the w h i t e spots i n the S E M m i c r o g r a p h are not as p r o m i -nent as the p inholes . A t the s tar t o f the f i lm g r o w t h o p t i m i z a t i o n , the ou tgrowths were Chapter 2. YBCO Film Growth and Processing 35 Figure 2.9: Atomic Force Microscope line scan of a pinhole in an as-grown film. The diameter of the typical pinhole was 200 nm, and its depth 10 nm. very dense. The outgrowths are believed to result from an off composition at the sub-strate surface. The composition of the outgrowths can be identified, if the density is high enough, by x-ray diffraction. The most prominent x-ray peaks were identified as BaCuO"2 and C u O , which suggested that the early films were Y poor. As the film growth improved, the density of outgrowths decreased. Under the optimal conditions, the films were essentially free of outgrowths. A common problem, especially with laser ablated films, is the occurrence of boulders. These are pieces of the target that were not completely ablated and became embedded in the film. The boulders are distinguished by their irregular shapes and are generally much larger in size than the outgrowths (boulders can be as large as 1 /im). A n example of a boulder can be found in a later section (figure 2.12). During growth under optimal conditions, boulders were not a serious problem, even though no special care was taken to remove them. Chapter 2. YBCO Film Growth and Processing 36 Impurities in the film were determined with Electron Dispersive X-ray (EDX) analysis and the assistance of Dr. M . Raudsepp7 The primary impurities in the film were found to be A l and C. However, since the superconducting and structural properties had reached the required standard, no further investigation was done. It suffices to say that the concentrations were low enough that the electronic and structural properties were not affected. 2.2 Annealing In this section, experiments to determine the effects of low temperature (T<500°C) and high temperature (T>800°C) annealing were conducted. The annealing was conducted in an oxygen flow tube furnace. The typical annealing schedules were comprised of a slow warm up to the desired annealing temperature, a dwell at the annealing temperature, followed by a slow cool down to room temperature. The oxygen was purified by flowing it through a catalytic converter (an oxidizer to convert CO to easily trapped C 0 2 ) and a liquid nitrogen cold trap. This eliminated the two most prominent contaminants to the film, H 2 0 and C 0 2 . Figure 2.10 is a diagram of the annealing furnace setup. Special care had to be taken to protect the film from contamination from the furnace tubes walls. Samples were placed inside an annealing cell, made from two Y B C O pellets. Prior to annealing the film, the cell was baked in flowing oxygen for 12 hours at 950°C to remove any possible contaminants (water, C 0 2 ) . Once cooled to room temperature, the seal to the tube furnace was broken, and the film was quickly placed inside the cell, and the furnace resealed. 7Department of Earth and Ocean Sciences (EOS), U.B.C Chapter 2. YBCO Film Growth and Processing 37 catalytic converter Figure 2.10: Annealing furnace used for annealing films 2.2.1 Low Temperature Annealing As described in the previous chapter, the oxygen content of Y B C O affects all of its properties. Experiments were conducted on deoxygenated films (T c =0K) to determine how easily oxygen could be reintroduced into the film. To obtain the deoxygenated samples, the films were annealed at lower base pressures. For this step, the growth chamber was used. The films were attached onto the block heater using the same scheme as outlined earlier in this chapter (i.e. using a conductive paste), and heated to 500°C in a pressure of « 1 0 - 6 Torr and left to dwell for five hours. After measurement, the films were placed in the tube furnace, where they were annealed to temperatures ranging from 250°C to 400°C with a dwell time of five hours. The results are shown in figure 2.11. It was concluded that an annealing temperature of at least 400°C (for an annealing Chapter 2. YBCO Film Growth and Processing 38 time of five hours) was required to reoxygenate a deoxygenated film. Even under these conditions, the films were not as homogeneous as the as-grown films (see figure 2.5). Better uniformity could be obtained with extended annealing times. 3 E CD CN I o c CD O E O -4—' c c n o 0.5 0.0 -0.5 -1.0 -1.5 0 250°C anneal 400°C anneal 20 40 60 80 100 Temperature (K) Figure 2.11: Magnetization measurements of deoxygenated films annealed ex-situ at tem-peratures ranging from 250°C to 400°C. Data was taken with a field of 3 Oe oriented normal to film surface. 2.2.2 High Temperature Annealing Y B C O films have been shown to recrystallize when annealed at temperatures >800°C[80]. At these elevated temperatures, the cations become mobile, whereas at lower annealing temperatures, only oxygen is mobile. The dangers of annealing at too high a temperature are the decomposition of the material and interdiffusion of the Sr and T i in the substrate with the metallic elements in the film. With these samples, high temperature (900°C) Chapter 2. YBCO Film Growth and Processing 39 F i g u r e 2.12: S E M m i c r o g r a p h of a f i lm annea led at 9 0 0 ° C . T h e h igh t empera tu re anneal-i n g was effective i n r e m o v i n g the 2 0 0 n m pinholes i n the as-grown f i l m . T h e wh i t e piece i n the lower left corner was ident i f ied as a b o u l d e r f r o m the target due to i t s large size a n d i r regu la r shape, a n d was used to focus the image . annea l i ng was successful i n r econs t ruc t ing the surface, effectively r e m o v i n g a lmos t a l l p inholes , w i t h o u t any degrada t ion i n the film. T h i s is shown i n the S E M m i c r o g r a p h of a film annea led at 9 0 0 ° C (figure 2.12). M e a s u r e m e n t s o f the r o c k i n g curve a n d t r a n s i t i o n t empera tu re of the annea led f i l m show no d e v i a t i o n f rom the as-grown values. 2.3 Concluding Remarks In s u m m a r y , we have demons t ra t ed the r ep roduc ib l e g r o w t h of h i g h l y c rys t a l l i ne Y B C O t h i n f i lms by pu l sed laser depos i t ion , w i t h r o c k i n g curve w i d t h s a p p r o a c h i n g tha t o f Chapter 2. YBCO Film Growth and Processing 40 the substrate. The Lorentzian line shapes of the rocking curves, the dependence of transition temperature and the independence of critical current density on the x-ray rocking curves widths suggests that the disorder was homogeneously dispersed, indicating a single-crystal like film. The lattice correlation lengths in the films were estimated to be in the range 40-115 nm. Films could be deoxygenated by annealing in low ( 1 0 - 6 Torr) base pressure and ele-vated ( 5 0 0 ° C ) temperatures. To completely replenish the oxygen, restoring the transition temperature to near 90K, required annealing temperatures of 4 0 0 ° C , and five hours of annealing time. High t e m p e r a t u r e ( 9 0 0 ° C ) annealing was effective in removing the pri-mary defect in the films, pinholes. Measurements with X R D and magnetization show no evidence of decomposition nor interdiffusion after high temperature annealing. Wi th the samples in hand, the next task was to confront the open issues connected with reactive ion implantation, alluded to in the introductory chapter. Chapter 3 React ive ion implantat ion As mentioned in chapter 1, implantation of Si ions (and other reactive ions such as A l and B) into Y B C O had been thought to be successful in patterning by rendering select portions of a Y B C O film non-superconducting without destroying its crystallinity. However, these results were inconsistent with the literature. In this chapter, the original experiments and interpretation of reactive ion implantation will be discussed followed by a discussion of the inconsistencies this interpretation had with the literature. A description of the new experiments conducted to resolve the inconsistencies will then be discussed, leading to a new interpretation and a modification of this technique which is more suitable for multilayer patterning. The Si ion implantation for the experiments described in this chapter was done in collaboration with the Center for Advanced Technology in Microelectronics, University of British Columbia. Additional Si implant work was performed by the Center for Elec-trophotonic Materials and Devices, McMaster University 1 . The T i ion implantation, described near the end of the chapter, was done by Implant Sciences2 through a collab-oration with Dr. Q Y . M a at Columbia University. 1 htm 2 www. implantsciences .com 41 Chapter 3. Reactive ion implantation 42 3.1 Original Experiments and Interpretation The original experiments and results are described in detail in [3, 69, 70, 87]. The goal was to develop a method of patterning Y B C O thin films without removing or destroying material. Using ion implantation, patterning without removing material was easily achieved. Silicon ions were implanted into Y B C O thin films at implant parameters chosen such that the Si ion distribution was centered in the middle of the film thickness. Doses were selected based on the belief that a reactive ion like silicon could "strip" oxygen from the Y B C O lattice - effectively deoxygenating it. If superconductivity was lost due to deoxygenation, instead of crystal damage, then the second of the two conditions would also be satisfied. Figure 3.1, taken from reference [3], shows the apparent effect of implanting a high dose of silicon ( 5 x l 0 1 6 c m - 2 at 100 keV) into the thin film, and subsequently annealing the film at low annealing temperatures. After implantation, some superconductivity remains, as shown in 3.1(B). Annealing at 4 0 0 ° C ( C ) and 5 0 0 ° C ( D ) further degraded the film's superconductivity. The magnetization curves alone do not demonstrate the degradation of superconductivity from a deoxygenation effect produced by the implanted silicon. In fact one might expect that degradation after implantation would be due to lattice damage. What was required was a simultaneous measurement of the crystallinity of the films. Figure 3.2 was the most striking of results, and suggested that the crystallinity remained intact despite being implanted at 5 x l 0 1 6 c m - 2 with an accelerating energy of 100 keV. The interpretation of these results was that the silicon destroyed superconductivity by removing oxygen, forming Si02, and not necessarily from crystal damage. The silicon concentration was estimated to be 0.1/unit cell; where and how the Si ions could sit in the lattice without significant disruption of the crystallinity, as measured by x-ray diffraction, was not thoroughly addressed. The deoxygenation argument was based on Chapter 3. Reactive ion implantation 43 Temperature (K) Figure 3.1: The original results, from reference [3] show that the as-implanted film(B) degrades with subsequent annealing at 3 0 0 ° C ( C ) and 4 0 0 ° C ( D ) . This is shown in com-parison to the virgin film (A) the fact that the heat of formation of S i 0 2 (AH o =-210 kcal/mole) was much lower than that for C u O (-38 kcal/mole) (or BaO(-134 kcal/mole) for that matter), and thus would be energetically favored. A low temperature anneal could "activate" the deoxygenation process, where the mobile oxygen in the chains could leave the Y B C O matrix and organize to form S i 0 2 . 3.2 Inconsistencies with the Literature The word "striking" used previously to describe the x-ray scan (figure 3.2) was appropri-ate since it was a stark contrast to all previous results for ion implantation into Y B C O . Ion implantation of Y B C O thin films had been extensively studied, and the range of Chapter 3. Reactive ion implantation 44 to "c 13 .ci * c^o -«—* CO c cu -I—» c ' c xi i l l >> -4—' 'to c a> c 003 LaAlO 005 006 LaAlO 002 001 | I 004 L i / > 007 0 0 20 20 40 60 20 20 40 60 Figure 3.2: X R D scans from the original results, also from reference [3], which suggested that the crystallinity of the film was unaffected by Si implantation (5xl0 1 6 c m - 2 at 100 keV). The top graph is for the original film, while the bottom graph shows the x-ray scan of a silicon implanted film. implant ions extends from the light, H+[81], to the very heavy, Ba+[72] and almost ev-erything in between. From early on, it was clear that with implant parameters similar to those used in [4], the lattice damage was extensive enough to destroy crystallinity and as well superconduc-tivity. To compare the extent of lattice damage from different ion species and conditions, it is important now to introduce a parameter which quantifies the damage resulting from ion implantation which takes into account the ion species and its acceleration energy. Using a popular Monte Carlo program, Transport of Ions in Matter (TRIM) [16], which predicts the number of displaced atoms per incident ion (dpi) in the target, the more commonly used unit, displacements per atom (dpa), can be easily determined. If $ is the Chapter 3. Reactive ion implantation 45 Reference Ion Energy Damage Film thickness Threshold (keV) (dpi) (nm) Damage (dpa) [71] (1987) 0+ 500 1553 500 0.17 [72] (1989) 0+ 80 661 > 10000 [72] (1989) Cu+ 160 2493 > 10000 [72] (1989) F+ 100 920 > 10000 [72] (1989) Ba+ 360 5855 > 10000 [73] (1990) 0+ 25 307 156 < 5.1 [74] (1991) F + 120 1004 200 0.74 < 7.4 [77] (1992) Ar+ 180 2176 200 0.15 < 0.73 [76] (1992) 0+ 25 307 160 [75] (1992) N+ 60 475 120 [78] (1993) 0+ 40 434 200 [79] (1993) P+(into Bi2212) 170 2001 2000 [81](1995) H+ 50 5 200 [80] (1995) Ar+ 100 1369 100 0.18 < 1.8 [82] (1996) Ne+ 200 1524 400 0.25 [83] (1997) 0+ 120 856 > 10000 [84] (1998) 0+ 180 969 250 <0.26 Table 3.1: Summary of the work in the literature on ion implantation into H T S films and crystals. The threshold damage is the level where there was no detectable crystallinity. implant dose, v the unit cell volume, t the film thickness (if damage extends throughout the film), and n the number of atoms per unit cell, then dpi • <f> • v , *° =  ±T^r <3:2> Using this parameter, the damage reported in various implantation studies on H T S sam-ples is summarized in table 3.1. The threshold damage is that level where there was no detectable crystallinity, and at these levels, there is also a complete loss of superconductivity. From table 3.1, the threshold value has been within the range 0 . 1 - 5 dpa. In contrast, a similar calculation of the damage in the Si implanted films, described in ref [4], results in a value of 50 dpa. Chapter 3. Reactive ion implantation 46 Although the literature has shown that the crystal structure of Y B C O was destroyed at even low implant dose (< 1 0 1 5 c m - 2 ) , this damage was not permanent. The loss of crystallinity can be reversed with a subsequent high temperature post-anneal. Under appropriate conditions, almost all of the crystal structure can be recovered. McCal lum [72] investigated the crystallinity of ion implanted Y B C O single crystals using R B S channeling, with only the top 200 nm of the crystal damaged. W i t h a high temperature anneal of 8 5 0 ° C for 2 hours, a substantial amount of the damage can be recovered with-out a change in the stoichiometry. Martinez eta/.[73] implanted 156 nm Y B C O films with 25 keV oxygen ions. Again, using R B S channeling, it was observed that up to 90% of the damaged portion could be recovered with annealing temperatures of 9 9 0 ° C . As well, an activation energy of the regrowth was identified, and estimated to be 0.42 eV. W u, by also implanting 25 keV O ions into 160 nm films, found that high tem-perature ( ~ 9 0 0 ° C ) annealing could recover 80% of the damage. Work done by L i and Pennycook suggest that annealing at high enough temperatures can recover the ion damaged regions of the film. However, it is only at low enough fluences (i.e. low enough damage) and at annealing temperatures > 8 0 0 ° C that complete recovery of a c-axis oriented film possible. In all these cases where there was post-annealing after implantation, the restoration of crystallinity was, w i t h o u t except ion , accompanied by a reappearance of superconductivity. The implant parameters used in the original Si ion implantation work had damage levels of 20-100 times higher than the threshold observed for the ions listed in table 3.1. Yet, according to the claims of the authors, the crystallinity was, to some level, retained. Secondly, at the annealing conditions used, superconductivity was further degraded. However, the literature showed that annealing out damage would, if anything, recover superconductivity. These were the inconsistencies, and the next task was to resolve them. Chapter 3. Reactive ion implantation 47 3.3 The Exper iments W i t h regard to the inconsistencies just discussed, the possibility existed that, for some reason, silicon was unique. Many material aspects of Y B C O were still not fully un-derstood and perhaps it was simply due to ignorance why there was no explanation describing how silicon could be implanted into Y B C O at high dose without imparting the expected damage. This led to a systematic study of silicon implantation into Y B C O thin films. The first objective was to determine the threshold damage level for Si implantation into Y B C O thin films. Silicon ions were implanted into Y B C O films (of thickness 120 nm) at an accelerating energy of 90 keV (close to the previously used value of 100 keV) and at doses ranging from l x l 0 1 3 c m ~ 2 to l x l 0 1 4 c m ~ 2 , and post-annealed at tempera-tures ranging from 5 0 0 ° C to 8 0 0 ° C . Typical ion current used was 2-4 JJLK c m - 2 for these relatively low doses. In later implants, where higher doses were used, the ion current was reduced to 200-500 n A c m - 2 . Figure 3.3 shows the predicted (from TRIM-92[16]) atomic displacement and silicon concentration profile at energies of 30, 60 and 90 keV. The to-tal number of atomic displacements is shown to be roughly proportional to ion energy. T R I M calculations also indicate that the most displaced atoms were oxygen atoms. At an energy of 90 keV, the damage and silicon profiles would extend throughout the film. It should be noted that at these doses, the concentration of Si ions (10~ 4 -10 - 3 /unit cell) is not expected to noticeably deoxygenate (or chemically affect in any way) the properties of the film. The effect of implantation on (005) rocking curve width and intensity are shown in figure 3.4. A loss of crystallinity is observed with increasing ion dose, with almost a complete loss of x-ray intensity at a dose of l x l 0 1 4 c m - 2 . This corresponds to a threshold damage of 0.14 dpa. Magnetization measurements of each of the films after implantation and Chapter 3. Reactive ion implantation 48 0 500 1000 1500 Film Depth (Angstroms) Figure 3.3: a) Atomic displacements and b) Silicon ion distribution resulting from ion implantation, as a function of film depth at energies of 30, 60, and 90 keV. These were obtained from TRIM-92 Monte Carlo calculation. The total number of displacements per ion were 413, 888 and 1240 for 30, 60 and 90 keV respectively. Chapter 3. Reactive ion implantation 49 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 Aco(degree) Figure 3.4: (005) rocking curve of 120 nm film at various doses with an implant energy of 90 keV. At this energy, the lattice damage is predicted to extend throughout the thickness of the film, a) Original film b) dose = l x l 0 1 3 c m ~ 2 c) dose = l x l 0 1 4 c m ~ 2 d) dose = l x l 0 1 4 c m ~ 2 and annealed at 5 0 0 ° C e) dose= l x l 0 1 4 c m - 2 and annealed at 8 0 0 ° C annealing are shown in figure 3.5. There is a degradation of superconductivity with increasing dose with a complete loss at a dose of 6 x l 0 1 3 c m ~ 2 . This agrees with work summarized in table 3.1, in particular with the systematic study with A r + implantation at 100 keV[80], where a dose of 10 1 4 cm~ 2 was also found to be the damage threshold to inhibit a film of similar thickness (100 nm). Implantation studies on other ions, as listed in the table, have threshold damage levels within the range 0.1 -5 dpa. Annealing implanted films restored their superconducting properties and to some ex-tent, their crystalline structure. Annealing at 5 0 0 ° C , where only oxygen was mobile, only Chapter 3. Reactive ion implantation 50 Dose = 1 0 1 3 c m " 2 13 E c cu E o E cu c o E cu 60 80 Temperature (K) Dose = 1 0 u c m " 2 0.0 - - -2.0 c cu E o E -4.0 ~ -6.0 CU c cn o -8.0 unorineoled G—e—e-500°C 800°C 20 40 60 80 Temperature (K) Dose= 6 x 1 0 1 3 c m " 2 • 1 • 1 • 1 III... unannealed 500°C / 8oo°c y o—e—e—e— 20 40 100 80 60 60 80 Temperature (K) Transit ion Temperature vs Dose too 40 y 20 ^ 0 o 800°C 500°C . 1 unannealed • * 100 2 4 6 8 Dose (10 1 3 cm - 2 ) 10 Figure 3.5: d.c. magnetization plots of films(120 nm) implanted at doses ranging from 10 1 3 to 10 1 4 cm~ 2 , and annealed at temperatures of 5 0 0 ° C and 8 0 0 ° C . The accelerating energy was 90 keV. Chapter 3. Reactive ion implantation 51 partially removed the damage, as seen in figure 3.4 but restored the transition tempera-ture to near 90K in the case of the lowest dose and to 77K at a dose of 1 0 1 4 c m - 2 , shown in figure 3.5. This suggested that T c reduction is strongly related to oxygen disorder, which can be regained with a low temperature anneal ( 5 0 0 ° C ) . Removing disorder within the cation sub-lattice required much higher annealing temperatures - an anneal of 8 0 0 ° C was required to regain a rocking curve intensity near its original value. Accompanying the restoration of rocking curve intensity and width was a recovery of the original 90K transition temperature for all doses. These results reveal that inhibition of superconductivity can be achieved even at low doses. The fact that annealing the films restores superconductivity suggests that the degradation is due to lattice damage, in particular oxygen disorder. This is consistent with the work done with A r + implantation [80], and more importantly is inconsistent with the previous silicon work. O n both counts (i.e. crystallinity lost at low dose, restora-tion of superconductivity after annealing) these results could not resolve the questions arising from the original data. It was obvious that the original results were misinterpreted. Why were strong x-ray diffraction peaks observed even after implantation at 5 x l 0 1 6 c m - 2 , since it has now been established that implanting even at low doses ($<10 1 4 cm~ 2 ) would destroy the crystallinity, and hence superconductivity. 3 The only possible explanation is that in the original experiments, the damage d i d not extend through the entire thickness of the film. In the present measurements, care was taken to ensure a precise measurement of the thickness (profilometry and S E M ) of the film as well as an accurate determination of the implant depth. To test this hypothesis, the next step was to implant just the upper half of the film, purposely leaving an "untouched" portion at the bottom. Doses were 3 Destroying the crystallinity is of course not the only way to inhibit superconductivity. In fact there has been work done on introducing point defects (undetectable by x-ray diffraction) by irradiation to drive YBCO through a metal-insulator transition Chapter 3. Reactive ion implantation 52 chosen to be similar to that of the original experiments (~ 10 1 6). Several energies (30, 60, 90 keV) and slightly thicker films (170 nm) were used. From figure 3.3 it can be seen that an "untouched" region of the film should remain, ranging in thickness from « 95 nm in the case of a 30 keV implant to 10 nm, for the 90 keV implant. The doses used were in the range (3xl0 1 5 - 3 x l 0 1 6 c m - 2 ) , leaving a concentration of silicon (0.03-0.3/unit cell) which should noticeably deoxygenate (or chemically affect in some way) the film. 0.5 i 1 1 1 1 1 1 1 1 • i -2 0 1 ' 1 ' 1 ' 1 1 1 1 1 0 20 40 60 80 100 Temperature (K) Figure 3.6: d.c. magnetization curves as a function of temperature of Y B C O films implanted (unannealed) at 3 x l 0 1 6 c m ~ 2 with energies of 30, 60 and 90 keV. Samples were first cooled in zero field then warmed in a field of 3 Oe, oriented normal to the film surface. Figure 3.6 shows the d.c. magnetization plots of 170 nm films implanted at high dose at various energies. TRIM-92 Monte Carlo calculations of the damage and S i + depth profiles (at these energies), predict a bottom layer unaffected by implantation. If the implantation damage had extended throughout the film, these doses would be much more than sufficient to completely destroy superconductivity. The observation of Chapter 3. Reactive ion implantation 53 cn c Z3 _ Q O CO c CD. 90 keV 002 001 1"' 1" •'1 ^  SrTiO, 60 keV 004 J 005 11 1 11 J SrTiO, 007 30 keV 5 10 15 20 25 30 35 40 45 50 55 60 2<9(deg) Figure 3.7: X-ray diffraction scans of Si implanted films. The implantation only extended through the top portion of the film. This is evident since X R D could detect the (00/) peaks from the c-axis oriented crystalline bottom layer. This is despite the fact that the dose (3x l0 1 6 cm- 2 ) was 2 orders higher than that required to destroy crystallinity. For the 30 keV implant, an "untouched" thickness of 95 nm was left at the bottom, 50 nm for the 60 keV implant, and roughly 10 nm for the 90 keV implant. Chapter 3. Reactive ion implantation 54 Figure 3.8: Left d.c. magnetization of the film implanted at 60 keV. A two hour, 3 5 0 ° C anneal resulted in a complete inhibition of superconductivity. However, subsequent an-nealing at higher temperatures restored the original transition temperature. Right Tran-sition temperatures (black squares) and c-axis lengths (open circles) as a function of annealing temperature for all implant energies. A l l films were implanted at 3 x l 0 1 6 c m ~ 2 . superconductivity is therefore consistent with the prediction of an "untouched" bottom layer. It should be noted that the magnetization curve of the 60 keV implant bears a close resemblence to that of the original experiments (see figure 3.1(B)). In figure 3.7, the x-ray diffraction scans are shown for films implanted such that the bottom layer is left "untouched". Since x-ray diffraction probes the entire thickness of the film and even into the substrate, the (00/) peaks corresponding to a c-axis oriented film are evident in all scans. Again, these results appear similar to that of the original Chapter 3. Reactive ion implantation 55 Figure 3.9: Transition temperatures and c-axis lengths, after a two hour, 3 5 0 ° C anneal, as a function of ion dose. The implant energy was 90 keV. experiments. The effect of low temperature annealing on the apparent transition temperature and c-axis length is shown in figure 3.8. The transition temperatures were determined using d.c. squid magnetometry while the c-axis lengths were determined by analyzing the x-ray diffraction scans. The left side of figure 3.8 shows the effect on the magnetization of the film implanted at 60 keV. The right side summarizes the results for all implant energies. It should be noted that the c-axis values must correspond to the bottom, "untouched" portion of the film since, as shown in the previous section, doses of this magnitude reduce the intensity of Bragg peaks to a point where they are barely noticeable. Low temperature annealing strongly affected these observed transition temperatures and c-axis lengths. The unannealed implanted film exhibited a T c and c-axis length comparable to the as-grown film. At an annealing temperature of 3 5 0 ° C , we observed a complete loss Chapter 3. Reactive ion implantation 56 of superconductivity regardless of ion energy. This was accompanied by a lengthening of the c-axis which corresponded to the length of the YBa2Cu3O7_5(r5~0.5-0.6) unit cell[9] - a loss of 0.5-0.6 oxygen atoms per unit cell. This was consistent with a dose of 3 x l 0 1 6 c m ~ 2 of silicon ions (0.3/unit cell of Y B C O ) if it is assumed that each Si ion combines with two oxygen atoms. Higher annealing temperatures resulted in a relaxing of the c-axis to its original value of 1 . 1 6 8 6 ± 0 . 0 0 0 3 nm, and a restoration of T c to near its as-grown value of 90K. Figure 3.9 shows the transition temperatures and c-axis lengths of films implanted at different doses and annealed at 3 5 0 ° C . The reduction in T c and the lengthening of the c-axis with increasing silicon concentration suggests that oxygen content in the undamaged bottom layer decreases as a result of its reaction with silicon near the surface of the film. 3.4 N e w Interpretat ion By implanting the surface of the Y B C O film with enough Si, a low temperature (350 °C) anneal could render it non-superconducting. The observation of T c = 0 was accompanied by a lengthening of the c-axis to a length consistent with a deoxygenated Y B C O . However, the deoxygenated film was not stable at higher annealing temperatures. Annealing at temperatures higher than 3 5 0 ° C , restored the transition temperature to near its original 90K value and the c-axis length to its original length of 1.1685 nm. The interpretation of this behavior is the following. At these doses, extensive damage is imparted into the film - to the point of amorphization. By implanting only near the surface, an effective bilayer system was created, where the surface was a heavily damaged Y - B a - C u - O - S i material passivating a bottom Y B a 2 C u 3 0 7 layer. It is worth recalling from chapter 1, that one has a very high oxygen mobility due to the C u - 0 chain structure in the Y B C O matrix. A highly disordered system, as we have at the surface of our bilayer, will not conduct Chapter 3. Reactive ion implantation 57 oxygen as easily. The implanted, unannealed film appears to have a T c of 90K, and a c-axis length of 1.1685nm, which is expected since the bottom layer of the bilayer was left untouched, and both d.c. magnetization and X R D "ignore" the amorphous non-superconducting surface region. Si02 is a more energetically favored phase than CuO, and given a strong enough push, it will form (this has been verified by X-ray Photoemission Spectroscopy (XPS) measurements[86]). Therefore, at low annealing temperatures (too low for oxygen atmosphere to replenish any oxygen deficiencies in the film), Si strips oxygen from the fully oxygenated bottom layer, until there is no unbonded silicon remaining. The plot in figure 3.9 provides strong evidence for this interpretation. Once all the silicon has been used to form Si02, reoxygenation will take place. At higher annealing temperatures, oxygen can diffuse through the capping layer, replenishing the oxygen deficiencies. Thus annealing at 450°C and higher restores the original transition temperature and c-axis length to the bottom layer of the bilayer. Given the geometry of the present samples (0.5 cm x 0.5 cm x 100 nm), it is much more reasonable to expect oxygen to diffuse through the capping layer than the films' edges. For clarity, this interpretation is illustrated schematically in figure 3.10. 3.5 Titanium Implantation There was some concern by the author that the reason for the observed changes in T c and c-axis length was clustering by the silicon atoms. Such clustering would be the result of diffusion of like atoms to each other, forming pockets of, say, 50-100 atoms; silicon has shown some tendency to do this. Clustering could easily explain the restoration of T c with increasing annealing temperature as islands of Si clusters would be embedded in a sea of Y B C O . A n explanation of the other observed changes would have required Chapter 3. Reactive ion implantation 58 1. As implanted Si-Y-Ba-Cu-0 material 2. Low temperature (350°C) anneal Si _,. i V / ^ o o o \ ' \ \ o \ YBa£u.p 7 3. Higher temperature (> 4 0 0 ^ ) anneal oxygen YBa 2 Cu 3 0 7 . 5 5°= ion dose oxygen SIOj SK>2 sio2 Y B a ^ u p ? F i g u r e 3.10: 1. A d isordered S i - Y - B a - C u - 0 m a t e r i a l , w i t h lower oxygen m o b i l i t y , caps the u n d e r l y i n g Y B a 2 C u 3 0 7 layer, a n d the apparent T c of the sample is 9 0 K 2 . A t a low annea l ing t empera tu re of 350°C, h i g h l y m o b i l e oxygen f rom the u n d e r l y i n g layer is a t t r ac ted to the S i a toms i n the c a p p i n g layer since A H / ( S i 0 2 ) < A H ^ ( C u O ) . T h e c a p p i n g layer prevents the oxygen (from a tmosphere) f rom rep len i sh ing the oxygen deficient layer, f o r m i n g YBa 2 Cu307_<5 w i t h a lower T c . T h e extent o f deoxygena t ion depends on the dose of ions i m p l a n t e d in to the top layer. O x y g e n can o n l y replenish f rom the sides of the sample - a m u c h slower process, r e q u i r i n g h igher annea l ing tempera tures . 3 . A t the h igher annea l i ng t empera tu re of 450°C, oxygen (which p r o b a b l y diffuses t h r o u g h the c a p p i n g layer) is capable of r ep len i sh ing oxygen deficiencies, res tor ing T c to 9 0 K . Chapter 3. Reactive ion implantation 59 further investigation. As it turned out, it was unlikely that clustering was the reason, a conclusion derived from experiments with titanium implantation. Ideally, clustering could have been verified by direct observation using sufficiently high resolution microscopy, but none was available. The alternative was to study the implantation of an atom similar to silicon, with the exception that it would not have a high tendency to diffuse or cluster. Titanium, like silicon, is very reactive, and one could expect it to behave as described in the previous section. Unlike silicon, its diffusion in Y B C O is very low, and will not move even at very high annealing temperatures, making clustering unlikely. 24 26 28 30 32 34 36 38 40 42 44 20(degree) Figure 3.11: X R D scans of T i implanted films annealed at 9 0 0 ° C for several hours. W i t h increasing dose, the formation of impurity phases, identified as BaTiOa become apparent. Experiments with titanium implantation, identical to that of the silicon implantation, were carried out with similar results. When the implant parameters were chosen such that the titanium distribution and damage extended throughout the film thickness, no Chapter 3. Reactive ion implantation 60 evidence of superconductivity was detected. This was observed for all doses used (3xl0 1 5 , l x l O 1 6 and 3x l0 1 6 cm~ 2 ) Again, this corresponded to a loss of crystallinity. Annealing the implanted films restored the transition temperature to above 85K, near its original value. The restoration of a high T c coincided with the formation of other impurity phases, apparent from figure 3.11. Implanting only near the surface was successful in deoxygenating the bottom layer. However, the effect was not as drastic as with the silicon implantation. The most ener-getically favored oxide phases are T13O5 and T i 2 0 3 . The deoxygenation would not be as efficient as in the case with Si, since only 1.5 or 1.4 oxygen atoms are stripped for each T i ion. The transition temperature could be lowered to 50K by annealing at 3 0 0 ° C , and higher annealing temperatures restored superconductivity to near 90K. This is shown in figure 3.12. Temperature (K) Figure 3.12: d.c. magnetization plots of films (170nm) T i implanted at doses of 3 x l 0 1 5 c m - 2 and l x l O 1 6 c m - 2 with an energy of 80 keV. This implants T i in only the top half of the film. The circles represent unimplanted films, diamonds implanted films, triangles annealed at 3 0 0 ° C and squares annealed at 4 0 0 ° C . Titanium showed similar behavior to silicon surface implantation, thus there was no Chapter 3. Reactive ion implantation 61 compelling evidence that silicon clustering was responsible for the observed changes in T c and c-axis lengths. 3.6 Recovery of C rys ta l l i n i t y The ability to recover the crystallinity after implantation is vital for multilayer patterning. The underlying film must be able to serve as a substrate for subsequent growth. As discussed in an earlier section, high temperature annealing can be successful in recovering crystallinity. The plot in figure 3.4 shows a restoration of ~70% of the x-ray intensity with an anneal at 8 0 0 ° C of a film implanted at low dose (10 1 4 cm~ 2 ) . A t higher damage levels, the extent of damage recovery is not as successful, or requires higher annealing temperatures. Figure 3.13 is a survey of the amount of crystalline re-covery after implantation and post annealing as a function of the damage incurred by the crystal. The data points were taken from previous measurements documented in the liter-ature and the present ones done with Si and T i implantation. The amount of restoration was quantified by comparing the intensities of the (005) 9/29 peaks, normalizing with respect to the intensity of the substrate peak, which was also done for the curves in figure 3.4. The data is scattered, since it is difficult to compare the more rigorous method of crystalline recovery using R B S channeling, to the relatively crude scheme of comparing x-ray diffraction intensities. The line separates the data for samples annealed at temper-atures below 9 0 0 ° C and those annealed at 9 0 0 ° C and higher. This graph gives an idea of the annealing conditions required to recover the crystalline structure after implantation damage. The trend is clear, and consistent. To regain enough of the crystalline structure for multilayer patterning, the damage must be kept as low as possible, and it appears that careful high temperature annealing would be successful in recovering damage at the level of a few dpa. Beyond the level of roughly 10 dpa, it would be very difficult to recover Chapter 3. Reactive ion implantation 62 Figure 3.13: The amount of recovery from annealing as a function of the damage from implantation. The solid line separates those that were annealed at T < 9 0 0 ° C , and those annealed at T > 9 0 0 ° C (annealing temperatures are indicated beside each point). The diamond represents silicon implantation from this chapter, and the triangles from the T i implantation. Black circles are from reference [80], the squares [72], and the white circles from reference [76]. Chapter 3. Reactive ion implantation 63 enough of the structure suitable for multilayers, or very high temperatures (~1000°C) would be required. Annealing at such a high temperature was not an ideal option for fear of decomposition and interdiffusion. The preferred choice was to anneal at a more accessible annealing temperature (900°C) with longer dwell times. 3.7 Conc lud ing R e m a r k s It is evident that the original experiments have been misinterpreted. A quick glance at the literature would have revealed the inconsistencies. The strategy of this research described in this chapter was to hypothesize what had happened and attempt to prove it by reproducing the results. What has been shown was that in the original experiments, the film thickness had been underestimated, or the implant depth overestimated - in either case the implantation did not extend throughout the film. This accounted for the detection, using X R D , of a highly c-axis oriented crystalline film. The interpretation of the degradation of superconductivity after low temperature annealing was partly correct. Deoxygenation does occur; however, it is not stable. Only under the specific conditions that a surface is somehow passivated, and annealing temperatures kept low can this occur. Once the annealing temperature is raised above ~400°C, superconductivity is restored. The HTS structures and devices patterned by Si implantation were clearly produced by extensive lattice damage, and they would probably be very stable devices. At damage levels of 50 dpa, only very high temperature annealing would restore superconductivity. However, this is clearly unsuitable for multilayer patterning. The experiments described in this chapter show that multilayer patterning using ion implantation requires a post-annealing step at high temperatures. This is because im-planting at a dose greater than 10 1 4 will cause extensive damage. Fortunately, to some Chapter 3. Reactive ion implantation 64 level, this damage can be removed with the appropriate post-annealing treatment. This technique requires two important modifications. Firstly, an implantation ion must be used that is stable in Y B C O . Thus a high temperature anneal can incorporate it into the structure. Secondly, the damage must be kept as low as possible and preferably below a few dpa. These two requirements have led to a modified reactive ion implantation process, which the present author dubs substitutional ion implantation. Chapter 4 Substitutional Ion Implantation Substitutional doping refers to the incorporation of the impurity into the Y B C O ma-trix, by the substitution of an atom in the unit cell. This has been described by many sources[90]. The altered material may be thermodynamically stable, and in this case will be unlikely to change with further processing. This chapter will attempt to convince the reader that substitutional ion implantation is the most suitable method of patterning multilayers using ion implantation, and secondly, the most appropriate choice of implant ion, in this context, is magnesium. We begin by first providing a review of impurity doping into Y B C O , leading to the choice of magnesium. The experimental results of M g ion implantation is thoroughly discussed. The chapter concludes with a summary of the optimal processing parameters for multilayer patterning. The M g ion implantation was done by Implant Sciences, through a collaboration with D r . Q . Y . M a at Columbia University. The Secondary Ion Mass Spectrometry (SIMS) was performed by Charles Evans and Associates1. 4.1 Impurity Doping in YBCO For H T S applications, doping has shown to be useful in improving the stability of films[91, 92], creating normal(N) layers for junctions[93], and in lowering the surface impedance in Y B C O single crystals[94]. Our goal was to use impurity doping to selectively (via implantation) lower the transition temperature of the Y B C O film and thereby achieve 1 65 Chapter 4. Substitutional Ion Implantation 66 patterning of structures. By considering the effects of various dopants in Y B C O , an appropriate choice of implant ion for patterning can be made. Several requirements exist for multilayer patterning using impurity doping. 1. The impurity must incorporate into the Y B C O lattice. 2. The doped material must possess a T c much less than that for Y B C O . 3. The doped material must be (or nearly be) a single phase material. 4. The concentration of the impurity must be as low as possible, since its incorporation will result in the removal of another element from the lattice (most likely, Cu) There is a diverse range of effects associated with atomic substitution in Y B C O , and elements have been substituted for all three metallic elements in Y B C O . For the goal of lowering the transition temperature with as small an impurity concentration as possible, two factors are targeted. First, the rate of T c depression, and second, the solubility limit. Table 4.1 summarizes these parameters taken from the work done on doping both bulk and film samples. 4.1.1 Y Site Substitution Substitution for Y has been mainly done with rare earth elements. The effect on the transition temperature is not dramatic as shown in table 4.1. It should be mentioned that the suppression of T c for Pr doping Y B C O , listed in this table, is more likely due to cation mixing of Pr and Ba. 2 Substitutions with other metallic elements have been done mainly with C a and Na. These enter with lower valence and therefore in maintaining charge balance, force doping of holes into the C u - 0 planes. The effect of Na and C a doping is to overdope the material, hence reducing the transition temperature at a rate of 30-40 and 70-80 K * x respectively. 2In a recent paper[95], PrBa2Cu307 was found to be a superconductor (fabricated under high pres-sure) at temperatures near 100K. Chapter 4. Substitutional Ion Implantation 67 Ion S u b s t i t u t i o n site S o l u b i l i t y l i m i t d T c / d x R e f p r 3 + Y 1.0 50-130 [97] E u 3 + Y 1.0 1-2 [97] E r 3 + Y 1.0 1-2 [97] S c 3 + Y 0.5 3-4 [97] Ca 2 + Y 0.2-0.5 70-80 [98] N a 1 + Y 0.5 30-40(x<0.03),200(0.03<x<0.05) [99] L a 2 + B a 0.7-1.0 — [97] F e 3 + mixed Cu(l) ,Cu(2) 0.8 900(x>0.01) [100] C o 3 + Cu( l ) 1.0 700(x>0.025) [100, 93] Ga 3 + Cu( l ) — 200-250(x>0.01) [101] A l 3 + Cu( l ) 0.4 250-300(0.04<x<0.12) [100] Z n 2 + Cu(2) 0.4 1200(x<0.03),500(x>0.03) [100, 102] N i 2 + Cu(2) 0.3-0.4 300(x<0.05) [100, 102] Table 4.1: Summary of doping work with Y B C O . 4.1.2 B a Site S u b s t i t u t i o n Only the rare earth element, L a , was successfully incorporated into Y B C O on the B a site. It has been difficult to fabricate single phase materials by doping on this site. 4.1.3 C u Site S u b s t i t u t i o n Substitution for the copper site can be divided into two categories. Elements that sub-stitute for the copper chain site, C u ( l ) , and those that substitute for the copper plane site, Cu(2). Fe, Co and A l are known to substitute for the chains and in fact force an orthorhombic to tetragonal (O-T) transition. Zn and Ni , which substitute for the plane site do not induce a structural transition, but drastically affect the films' superconducting and normal state properties. The O - T transition observed with changing oxygen doping, can also be induced by doping into the chain copper site, Cu( l ) . The addition of more oxygen to the YBCO(6 .0 ) Chapter 4. Substitutional Ion Implantation 68 unit cell, induces a preference for the oxygen to occupy only the 0(1) sites, resulting in the 0 - T transition at an oxygen level of ~6.4. N M R measurements have shown that Fe, Co and A l substitute for the Cu( l ) site, and a similar 0 - T transition has been observed. See figure 4.1. The preference for either an orthorhombic or tetragonal phase is determined by the energetics within and between C u - 0 chains. By doping an impurity into the chain, the M - 0 bond strengths change, altering the energetics. W i t h enough doping, a phase transition is induced. to 3.9 3.8 j /-' i i i 9- 3 o = b Fe - = a 3.9 3.8 / i i i e o-Co 1 1 3.9 3.8 i i i Al 3.9 3.8 i i i Zn 1 1 3.9 3.8 * 1 1 Ni 1 1 0 0.2 0.4 0.6 0.8 1.0 x in Y B a 2 C u 3 _ x M x 0 7 Figure 4.1: The O - T transition is apparent from the a and b axes lengths. Fe, Co and A l substitute for Cu( l ) , while Ni and Zn substitute for Cu(2). From ref. [100]. Chapter 4. Substitutional Ion Implantation 69 Superconductivity is widely believed to occur within the C u - 0 planes. Thus any drastic perturbation of these planes should have serious effects on superconductivity. The two most prominent impurities which have been demonstrated to substitute for the copper plane site, Cu(2), are Zn and Ni , with drastic differences in their effects. Zn and Ni both are known to substitute as divalent ions. Z n 2 + is non magnetic with a 3 d 1 0 electronic configuration , while N i 2 + is magnetic with a 3d 8 configuration. The experience with conventional superconductors predicts a strong suppression of superconductivity with magnetic ions. Conversely, non-magnetic impurities would have little effect on the transition temperature. W i t h the high temperature superconductors the opposite is observed. Zn, the non-magnetic impurity, has the most drastic effect on T c , while Ni , the magnetic impurity, has a much less drastic effect. See figure 4.2. The magnetic pair-breaking effect seen in conventional superconductors is not expected in superconductors that do not have s-wave gap symmetry. It is believed that the pairing mechanism, induced by antiferromagnetic spin fluctuations, is fundamentally different from the phonon mediated mechanism of conventional superconductors. The gap symmetry, predicted by this mechanism, is one with line nodes, and the presence of non-magnetic Zn is to suppress the spin fluctuation operator at its substitution site, eventually reducing the transition temperature to zero Kelvin. The microwave surface resistance of H T S materials is technologically important since one of the most promising applications are microwave filters for use in cellular phone and Personal Communication Service (PCS) networks. The very low microwave loss of the H T S materials give it a distinct advantage over more conventional materials. Zn and Ni doping have strong effects on lowering the microwave surface resistance. This is described well in [94], but is briefly overviewed here. The temperature dependence of the surface impedance is determined by the behavior of the quasiparticles, and is thus proportional Chapter 4. Substitutional Ion Implantation 70 x in YBa^Cu^MJiA Figure 4.2: Transition temperature of Y B a 2 ( C u i _ x M x ) 3 0 7 ( M = Zn, A l , Co, A l and Ga) as a function of x. Figure obtained from reference [21]. to the density of quasi-particles, ne and their lifetime, r . Measurements with high quality Y B C O crystals show a broad peak at intermediate temperatures (~40K) resulting from a competition (as temperature is decreased) between the decreasing number of quasi-particles and the increasing lifetime. The addition of Zn or Ni put an upper limit on r , thus suppressing the peak, and lowering the surface resistance. It is believed that it is the presence of this sort of scattering impurities which are responsible for the low surface impedance of commercially available H T S films. For the purpose of patterning, the ideal impurity is one that can drastically reduce the transition temperature with as low a concentration as possible. Armed with the Chapter 4. Substitutional Ion Implantation 71 knowledge of previous impurity doping studies, it would seem the most suitable choice is Zn doping. However, other more practical issues with implantation must first be addressed. 4.1.4 Addition of Impurities In implantation patterning, an impurity is added to stoichiometric Y B C O . If substitution were to occur, that would imply that an excess amount of the substituted atom (most likely copper) would exist in the material. Work done on the addition of impurities to stoichiometric Y B C O has been limited to only a few investigations, listed in [9]. This includes work with T i , W , Cr , Nb and Si addition to Y B C O powder. The results are not impressive. The impurity-added material, in these cases, were multi-phased, and the effect on transition temperature was minor, since regions of Y B C O still existed. Often the material possessed a T c near 90K even with large amounts of the impurity added to the powder. The only success with addition of impurities (for a completely different purpose), was work on the addition of Au[92] and Ag[91] into Y B C O thin films to improve stability by passivating the grain boundaries. Preliminary measurements of the critical current, for example, are promising. 4.1.5 Incorporation via Implantation There are two main issues associated with doping via implantation. The first is the extent of damage imparted to the lattice. However, as shown in the previous chapter, if the damage is sufficiently low, then it can be removed with high temperature annealing. Removal of damage is not the only concern. Immediately after implantation, the im-purity ions are distributed throughout the sample, and would not occupy any particular site, since after all, there is no lattice. During recrystallization, the impurity must be Chapter 4. Substitutional Ion Implantation 72 incorporated into the Y B C O lattice to effectively and permanently alter, the materials superconducting properties. In the fabrication of bulk doped Y B C O , the initial powders are carefully mixed and calcined at temperatures of over 9 0 0 ° C and left at that tempera-ture for several hours. It is reasonable to expect that similar processing conditions would also be required to form a doped film. It is clear that for incorporation via implantation, the post-anneal accomplishes two things. First, to recrystallize the film, and second to incorporate the ion into the lattice. 4.1.6 Choice of Implant Ion From the previous discussion, the most appropriate choice would be Zn doping, since it has the largest d T c / d x suppression and the solubility limit is high enough to completely suppress superconductivity. Unfortunately, due to its relatively large ion mass, the pre-dicted damage would be 1964 dpi. Assuming that all the Zn substitutes into the Cu(2) site, the dose required to lower the T c of a 150 nm Y B C O film to zero Kelvin would result in a damage level of 45 dpa - clearly beyond the level at which this damage can be removed. Implanting this much Zn is not practical, since the resultant material will definitely not be single phased. The excess amount of copper should be identical to the concentration of Zn (~10% to completely suppress T c ) . Whatever is implanted, to regain a highly crystalline, quasi-single phased material the dose must be limited to about 10 1 5 cm~ 2 . A t the same time, one needs an ion for which only a very small amount (< 1%) can reduce T c a sufficient amount. As well, the lower the atomic number the better, since this also reduces the lattice damage imparted to the film. Chapter 4. Substitutional Ion Implantation 73 4.2 Magnesium Ion Implantation In the bulk form, M g has been found to enter into the Y B C O lattice to form the solid solution Y B a 2 ( C u i _ x M g x ) 3 0 7 , with a solubility limit of x=0.01-0.025[106, 107, 108]. The effect of M g doping on T c is dramatic, with a T c suppression of 1400K*x, similar to that of Zn doping (Mg is also believed to substitute for the Cu(2) site in the C u - 0 planes). Thus, it is believed that M g can permanently poison the superconductivity. The relatively low mass of M g limits the extent of implantation damage, making it well suited for implantation patterning. In this section, results are presented showing the incorporation of M g into Y B C O by ion implantation for the purpose of patterning. High temperature post-annealing resulted in the formation of a single phase, highly crystalline, M g doped Y B C O structure, with a finite resistivity at 77K. 4.2.1 Experiment M g + ions were implanted with an E A T O N 3206 ion implanter at doses of 2 x l 0 1 5 , 4 x l 0 1 5 and l x l 0 1 6 c m ~ 2 , corresponding to x=0.008, 0.02 and 0.04 M g respectively. Figure 4.3 shows the predicted M g distribution and damage profile. At an accelerating energy of 80 keV, we expect a distribution of M g + ions peaked at a depth of 80 nm, and a damage level of 1070 atomic displacements/ion (at a dose of 2 x l 0 1 5 c m ~ 2 , this results in in 2 displacements per atom (dpa)). The films were annealed after implantation with an annealing schedule comprised of a ramp of 5 ° C / m i n to the desired annealing temperature, a dwell for nine hours and a cool down of 2 ° C / m i n . Samples were characterized with d.c. magnetization and resistivity, x-ray diffraction ( X R D ) , scanning electron microscopy (SEM) and electron probe microanalysis ( E P M A ) . Chapter 4. Substitutional Ion Implantation 74 0.10 Depth (nm) Figure 4.3: TRIM-92 predictions of the M g and damage profiles resulting from an implant with energy 80 keV. At a dose of 2 x l 0 1 5 c m - 2 , this results in a damage level of 2 dpa. 4.2.2 Determination of the Mg Concentration The implanted depth profile is Gaussian, which is predicted by T R I M and previously verified by Secondary Ion Mass Spectrometry (SIMS) [109]. Thus, the M g concentration, if determined by dividing the total dose by the number of unit cells of Y B C O , would not provide an accurate value, since near the peak, the true concentration would be much higher, and near the surface, much lower. It was first assumed that after post-annealing, the distribution would be homogeneous. Before any further studies were done, it was important to verify this using SIMS. Figure 4.4, is the SIMS measurement of a bilayer of 140 nm Y B C O / 1 4 0 nm Y B M g C O with the concentration of C u , Y and M g measured as Chapter 4. Substitutional Ion Implantation 75 a function of depth, after post-annealing. The bilayer was fabricated by first depositing a film of Y B C O , implanting and then post annealing it. A Y B C O top film was then deposited on top. (A more detailed discussion of bilayers is given in chapter 5.) It can be seen that the M g concentration is essentially uniform throughout the bottom layer. 0.00 70 90 110 130 150 170 190 210 230 250 270 Film depth (nm) Figure 4.4: Secondary Ion Mass Spectra of a YBa 2Cu307/YBa2(Mgo.oo8Cuo.992)3 0 7 ( Y B -M g C O ) bilayer. The top graph shows the depth profile of C u and Y , while the bottom graph shows the interface width between the top Y B C O and bottom Y B M g C O layers as well as the uniform depth profile of M g in the bottom layer 4.2.3 Results The effect of annealing on the magnetization of M g + implanted films, measured by d.c. magnetization, is shown in figure 4.5. Prior to annealing, the implanted films showed no indication of superconductivity at any temperature, and with subsequent annealing, Chapter 4. Substitutional Ion Implantation 76 E CO CM c CD E o E o • CD C O) CO 0.5 0.0 -0.5 -1.0 -1.5 -2.0 1.0 0.0 -1.0 r -2.0 2 x 1 0 1 5 c m 2 (2 dpa) 600°C • / • 4 x 1 0 1 5 cm (4 dpa) 800°C " • ' I l t t i r t u 900°C T • • • • • • ^ as-grown -3.0 0.5 0.0 -0.5 \ -1 .0 -1.5 1 0 x 1 0 1 5 c r r i 2 ( 1 0 d p a ) 800,900°C as-grown «/ • • • • • * 20 40 60 80 Temperature (K) 100 Figure 4.5: d.c. magnetization curves of M g implanted films. The doses, which are labeled with each plot, range from 2xl0 1 5 to l O x l O 1 5 c m - 2 . As well, the effect of annealing is also shown. W i t h increasing annealing temperature, superconductivity is regained (the reappearance of superconductivity in the l O x l O 1 5 graph is not so apparent on this scale). In stark contrast to Si and T i implantation, the maximum T c of the M g implanted films is limited to lower values. This is due to the fact that M g can incorporate into the Y B C O lattice, forming YBa2(MgxCui_a;)307. Chapter 4. Substitutional Ion Implantation 77 100 cu o CU Q_ E CD C o cn c o 80 60 40 - 20 0 200 0 dpa) 400 600 800 Annealing temperature (°C) 1000 Figure 4.6: Transition temperatures of implanted films versus annealing temperature. Each dataset is labelled with the implanted ion and the damage level, measured in dis-placements per atom (dpa) calculated from TRIM-92 Monte Carlo simulations. The inset compares the transition temperatures of the Mg implanted, annealed films with that of the Y B a ( C u 1 _ I M g x) 307 bulk samples, from ref 6(squares) and 7(diamonds). superconductivity was restored. At an annealing temperature of 800°C, T c increases to near 70K for the lowest dose, but remains at zero Kelvin for the highest dose. Annealing at 900°C restores superconductivity to all films, but again the extent varies with the damage level. It is now worth comparing transition temperatures at various damage levels and annealing temperatures. This is shown in figure 4.6. The transition temperatures for Mg implanted films are compared with films implanted with T i + at the same energy (80 keV) and with similar (3xl0 1 5 and l x l O 1 6 c m - 2 ) doses. The implantation damage (measured in units of dpa) is given for each parameter set. If it is thermodynamically Chapter 4. Substitutional Ion Implantation 78 unfavorable for the implanted ion to be integrated into the Y B C O structure, then high temperature (> 8 0 0 ° C ) annealing will reform islands of pure Y B C O , resulting in high transition temperatures. For example, to our knowledge and suggested from figure 3.11, T i is not soluble in Y B C O . Thus the T i + implanted films, despite sustaining much higher damage levels, could be annealed at 9 0 0 ° C to obtain higher transition temperatures (85K for 3x l0 1 5 , 83K for l x l 0 1 6 c m ~ 2 ) than the M g + implanted films, which reached a maximum T c of 77K. This near full restoration of T c , at similar damage levels, was also observed for S i + and A r + implanted films[89, 80]. The transition temperatures of the 9 0 0 ° C annealed, M g + implanted films are consistent with the values measured for the bulk samples doped at the same concentration, shown in the inset of figure 4.6. At concentrations less than the solubility limit of M g in bulk Y B C O , T c suppression in the films and bulk are very similar, suggesting that the suppressed T c ' s observed for the annealed M g implanted films is due to the formation of a M g doped Y B C O film and not to remnant defects from implantation damage. The phase purity of a film implanted at 2 x l 0 1 5 c m - 2 (x=0.008), which is less than the solubility limit, is apparent from figure 4.7. Details of the x-ray measurements are given in Appendix B . A t this dose, a highly c-axis oriented film, with no other phases was observed. For comparison, the x-ray scans of the higher dose implants are also shown. With increasing dose, one can detect the appearance of impurity phases and the peak of randomly oriented Y B C O . At a dose of 4xl0 1 5(x=0.02), the first indication of impurity phases is detected. When the concentration of M g exceeds the solubility limit, the excess M g is accommodated by forming other impurity phases, with the two most prominent ones identified as M g O and BaCu02- Based on the dose at which these impurity phases become observable, it is estimated that the solubility limit of M g into Y B C O , doped via ion implantation, is approximately x=0.02. The crystallinity of the implanted, annealed films is vital for the subsequent growth Chapter 4. Substitutional Ion Implantation 79 2 theta(deg) Figure 4.7: 9 0 0 ° C . The X-ray diffraction ( X R D ) scans of films implanted with M g and annealed at doses are labeled with each scan. Chapter 4. Substitutional Ion Implantation 80 20 15 10 [ 0 -1 .0 15 ^ 10 -4—' cn c 5 CD - Lorentzian Fit -• Gaussian Fit • Datapoints 0.0 Aoj(degree) 4x1015cm"2(x=0.02) Figure 4.8: X-ray rocking curves of the (005) Bragg peak of films M g implanted with increasing doses and annealed at 9 0 0 ° C . The implant doses are included with each plot. The intensity is labeled with respect to the films' original, as-grown value. At the high-est dose the Lorentzian lineshape could not be regained even after high temperature annealing, and resembles more a Gaussian lineshape. Chapter 4. Substitutional Ion Implantation 81 100 200 Temperature (K) 300 Figure 4.9: The temperature dependence of the d.c. resistivity of M g implanted films annealed at 9 0 0 ° C . The implant doses were a) 2 x l 0 1 5 c m ~ 2 (x=0.008) b) 4 x l 0 1 5 c m ~ 2 (x=0.02) and c) l x l 0 1 6 c m - 2 (x=0.04) of top layer films. The quality of the top layer film can, at best, be equivalent to the underlying film. The x-ray rocking curves of the (005) Bragg peaks are shown in figure 4.8 for the implanted films annealed at 9 0 0 ° C . The original unimplanted films exhibited Lorentzian shaped rocking curves with widths < 0 .12° . At an implant dose of 2 x l 0 1 5 c m ~ 2 (damage level of 2 dpa), shown in the bottom graph, the original Lorentzian lineshape was regained after annealing. In comparison, at 5 times the damage level (10 dpa), resulting from a dose of l x l 0 1 6 c m ~ 2 , the lineshape of the rocking curve could not be fitted with the same Lorentzian, and exhibited Gaussian broadening, typical of a granular structure[67]. This suggests that at high enough damage levels, a limit was reached where a single Chapter 4. Substitutional Ion Implantation 82 crystal-like crystallinity could not be recovered even after high temperature annealing. The temperature dependence of the d.c. resistivity, measured at 1.5 A c m - 2 , of M g implanted, 9 0 0 ° C annealed films is shown in figure 4.9. The midpoint of the transitions corresponded well with the transition temperatures measured by d.c. magnetometry (fig-ure 4.6). The lowest dose curve exhibited a metallic normal state with a transition width of 10K. The metallic nature of the normal state degraded with increasing dose, and at the highest dose a semiconducting temperature dependence was observed. This temperature dependence, along with the Gaussian broadening of the rocking curve, suggests that the film is comprised of separate grains. The film, implanted at 2 x l 0 1 5 c m - 2 and annealed at 9 0 0 ° C , had a resistivity of 240 fj,fl cm at 77K with zero resistivity at 66K (shown in the inset). 4.2.4 Discussion and Concluding Remarks In this chapter, an improved method of patterning H T S films for multilayers was de-scribed. Substitutional ion implantation using M g can reduce the transition temperature below 77K, while still maintaining a crystalline structure. High temperature annealing was more successful in restoring crystallinity after implantation because the impurity could incorporate into the Y B C O matrix. The only drawback to M g is its low solubility limit, so that the transition temperature cannot be driven to zero Kelvin. However, even if the solubility limit were much higher, implanting that much M g would not have been practical, due to the damage and the excess copper. W i t h these constraints, the low solubility was not an issue. The concentration of M g (the value of x) was determined by assuming that the post-annealed distribution were uniform. This was confirmed by SIMS measurement. How-ever, for films much thicker than the lateral straggling of the implant, a 9 0 0 ° C , 9 hour anneal was not sufficient to obtain a uniform distribution. This is discussed further in Chapter 4. Substitutional Ion Implantation 83 chapter 5. A n alternative scheme to homogenize the implant profile, is to implant at various ener-gies using lower doses. This is often used in semiconductors[17], and has been previously studied with the H T S samples[109]. A question arises, "What is it about M g that makes it so effective in reducing T c ? " . A quick answer would be that the similarities between M g and Zn doping, imply a similar mechanism. Firstly, although there has yet to be an independent verification, M g is believed [106] to substitute for the Cu(2) site. As well, it also tends to incorporate as a non-magnetic divalent ion, with a 2p 6 electronic configuration. However, to determine the exact magnetic state of M g in the C u - 0 plane is not so easy, since M g 2 + is not an isolated ion, but sits in a lattice with C u and O and the crystal field splitting must be considered. There is a possibility that hybridization between the C u 3d orbitals and the O 2p orbitals in the C u - 0 plane are responsible for conduction. Thus, Mg's lack of 3d orbitals would have a drastic effect. To reach a satisfactory answer to these questions will certainly require more experimental work. In the context of patterning methodologies, it may be sufficient to leave it an open question. For patterning multilayers, the optimal conditions can now be stated. 1. The dose for M g implantation should be such that the concentration is just below the solubility limit, x~0.02, but not exceeding 5 x l 0 1 5 c m ~ 2 , For a Y B C O film of thick-ness 140 nm, this is a dose of 2 x l 0 1 5 c m - 2 . The film thickness must be very accurately determined, otherwise the M g concentration can be easily miscalculated. 2. The energy should be such that the peak of the implant distribution is near the center of the film's thickness. However with the high temperature anneal, diffusion in the c-axis direction is probably sufficient that even slightly lower energies would result in a uniform distribution. 3. Post annealing of the implanted film must be done at least at 9 0 0 ° C (for 9 hours). Chapter 4. Substitutional Ion Implantation The time can be exchanged for higher temperat Chapter 5 Structuring with Mg ion implantation The motivation of this research was not to do a materials study of implanted Y B C O thin films, although up to now that has been the focus. The motivation was to fabricate multilayer structures, which are, broadly speaking, three dimensionally patterned films. This chapter will describe the patterning of Y B C O thin films with M g implantation and the growth and characterization of bilayer structures using this processing technique. 5.1 Patterning The optimal conditions for patterning using M g implantation was outlined in the final section of the previous chapter. Standard photolithographic techniques were used to mask the film prior to implantation. W i t h the assistance of A . K u l p a 1 , the following processing recipe was used. 1. Spin Shipley S1400-27 positive photoresist onto 1 c m 2 sample. The spinner was set at 4500 rpm, which laid a 1.1 / im thick photoresist layer on the film 2. Photoresist (PR) baked for 25 minutes in air at 9 6 ° C , and cooled to room temper-ature. 3. Exposed P R to U V light for 27-29 seconds at 277 Watts (25.3 m W / c m 2 ) 4. Developed P R by dipping in developer for 30 seconds. 5. Post-baked P R for 25 min at 1 1 0 ° C . After implantation, the photoresist was removed by placing the sample in hot acetone 1 Center for Advanced Technology in Microelectronics, Department of Electrical Engineering, U . B . C . 85 Chapter 5. Structuring with Mg ion implantation 86 for 30 seconds. The resultant structure, which contained lines 100 fim wide, is shown in figure 5.1. Figure 5.1: S E M micrograph of the backscattered electrons of a film patterned with standard photolithography and M g ion implantation. The implant dose was 2 x l 0 1 5 c m - 2 , and the accelerating energy, 80 keV. The lateral diffusion of the M g ions during high temperature annealing was determined using Electron Probe Microanalysis ( E P M A ) in the Department of Earth and Ocean Sci-ences, U . B . C . In this technique, an electron beam (diameter « l / i m ) , is used to excite the core electrons to higher energy states. A subsequent de-excitation to lower energy levels emits an x-ray characteristic of the atom. Thus, measurements of the x-ray emis-sions is a useful tool in obtaining information on the composition of the material. The probing volume of the E P M A is several u.m3 and tear-shaped. To investigate diffusion resulting from annealing, the line profile across a boundary between the implanted and unimplanted regions was measured before and after annealing. The profile is shown in figure 5.2. Before annealing, a border width of 1-2 microns is evident, and after anneal-ing, this width broadens slightly to roughly 4 microns. It should be noted that the initial Chapter 5. Structuring with Mg ion implantation 87 unimplanted implanted 60 I I ' ' ' l I • 1 — • — r — — ' — ' i 30 1 1 1 -10 - 8 - 6 - 4 - 2 0 2 4 6 8 10 Distance from border (jum) Figure 5.2: Electron Probe Microanalysis (EPMA) scan of a border between an implanted and an unimplanted region on a patterned film width of this border is limited by the width of the probing beam, which was previously mentioned to be 1-2 microns. The image shown in figure 5.1 was a film processed under the optimal conditions (below the solubility limit), and the surface is as smooth as that of annealed unimplanted films. Higher doses result in a Mg concentration exceeding the solubility limit, and the multi-phased nature of the material also appears on the surface. An S E M micrograph of a film implanted at l x l O 1 6 is shown in figure 5.3. The implanted region of the film is very rough from the precipitation of off-composition outgrowths, and the inability of annealing to completely recover 10 dpa worth of damage. The brightness of the white spots reveal their high average atomic number, and E D X measurements show some indication of the Chapter 5. Structuring with Mg ion implantation spots being Y rich precipitates. 88 Figure 5.3: S E M micrograph of the backscattered electrons of a Y B C O film implanted with M g ions with a dose of l x l O 1 6 c m - 2 and annealed at 9 0 0 ° C . The film was masked such that alternating lines of 100 yum were implanted. In this image, a border between the implanted and unimplanted is shown. The rough surface in the implanted region resulted from off composition and the extensive damage in the film. Chapter 5. Structuring with Mg ion implantation 89 5.2 Bilayer growth The bilayers were fabricated by first depositing a thin film (~200nm) 2 of Y B C O using the scanning pulsed laser deposition system described in chapter 2. The films were im-planted at the optimal conditions determined previously; an implant dose of 2 x l 0 1 5 c m - 2 with an accelerating energy of 80 keV. The post annealing conditions were a slow warm up ( 1 2 0 ° C / h r ) to the set temperature of 9 0 0 ° C , and dwell for 9 hours, and a slow cool down ( < 1 2 0 ° C / h r ) to room temperature. These processing conditions result in a highly crystalline YBa2(Mgo.oo5 (^0.995)307 (hence, Y B M g C O ) film which has a transition tem-perature below 80 K . The Y B M g C O films were loaded back into the deposition chamber for the growth of top layer Y B C O films. The growth of top layer films were modified slightly from that described in chapter 2 in order to reduce the time the Y B M g C O film was kept in vacuum at elevated temperatures. As well, bilayers with the Y B C M g O on top of a Y B C O film were grown using the same processing techniques except only that the order of some of the steps was reversed. 5.2.1 Characterization The bilayers were characterized with d.c. resistivity, X R D , and Secondary Ion Mass Spectrometry (SIMS). Resistivity measurements of the top layer films are shown in figure 5.4. In the left graph, where the current flows through both the Y B C O and Y B M g C O layers, a clear double transition is evident, revealing that the bottom Y B C M g C O layer retains its lower transition temperature even after growth of the top layer Y B C O film. The right side of figure 5.4 shows a metallic temperature dependence in the normal state and a sharp transition with zero resistivity observed at 9IK, 1 which is similar to as-grown 2The intended film thickness was 140nm. However after growth and profilometry measurements, the true thickness was closer to 200 nm. This realization came after the implantation, and thus the Mg content, from a 2 x l 0 1 5 c m - 2 dose, would not be x=0.008 (described in the previous chapter), but instead, x=0.005. Chapter 5. Structuring with Mg ion implantation 90 10 8 O D c o "to Z] "to CD * 2 0 •••• 70 80 90 10O-Y B C O YBMgCO 0 1.0 0.5 0.0 88 90 92 94 96 Y B C O YBMgCO 50 100 150 200 250 300 " 0 50 100 150 200 250 300 Temperature (K) Temperature (K) Figure 5.4: The temperature dependence of the d.c. resistivity of Y B C O films (black) grown on top of Y B M g C O films (grey). The left graph shows a double transition corre-sponding to the top Y B C O film ( T c = 90K) and the bottom Y B M g C O film ( T c = 79K). The graph on the right reveals that the resistivity of the top Y B C O film is very similar to that of a high quality as-grown film films. A magneto-optic technique of imaging a remnant field in the film was used to obtain the critical current of the top layer film. Details of this technique are given in [66]. At 84.4K (above the transition temperature of the bottom Y B M g C O film), the average critical current density was calculated to be 2 .5xl0 6 A c m - 2 (maximum of 2 .6xl0 6 A c m - 2 , minimum of 2 .3xl0 6 A c m - 2 ) . In comparison, the critical current (measured using the same magneto-optic technique) of an as-grown film was determined to be over 4 x l 0 6 A c m - 2 at 80K[110]. X-ray diffraction 9/29 scans can provide a macroscopic average of the orientation, the phase purity and the crystallinity of the material. The 9/29 scan of the bilayer is shown in figure 5.5. A highly c-axis oriented bilayer is evident, with only the slightest of indications of a randomly oriented Y B C O peak. The c-axis lattice parameters for Y B C O and Y B M g C O were previously measured to be 1.1685 and 1.1663 nm respectively, and Chapter 5. Structuring with Mg ion implantation 91 5 10 15 20 25 30 35 40 45 50 55 60 65 70 2 t h e t a Figure 5.5: X R D scan of YBa2Cu30 7 /YBa 2 (Mgo .oo5Cu 0 .995)3 07 bilayer. The highly c-axis oriented nature of the bilayer was apparent. The x-ray scan was performed using the highest usable power level and scanned at a rate of l ° / m i n u t e thus the Bragg peaks of the two will overlap at low diffraction orders. W i t h the resolution of the x-ray diffractometer, separation of the 2t9 Bragg peaks will only become apparent in higher order Bragg peaks (i.e. I in(00/)>ll) . In figure 5.6, this separation is clear in the 9/29 scan of the (0013) peak of the bilayer. The diffraction pattern was fit to a function comprised of two components corresponding to the Y B C O (short dashed line) and Y B M g C O (long dashed line) layers and a constant background term. Each of the film layers was assumed to contribute Lorentzian shaped diffraction peaks from the K Q I (A= 0.1540598 nm)= and K q 2 (A = 0.1544418 nm) x-rays. The only other constraints in the fit were the locations of the four peaks (observed from the data and predicted from the x-ray wavelengths). The ratio of the K Q L and Ka2 intensities and the widths Chapter 5. Structuring with Mg ion implantation 92 6000 117 118 119 120 26>(deg) Figure 5.6: X R D scan of the (0013) Bragg peak of the bilayer. The data (dark circles) was fit to a function represented by the solid line. This fit was comprised of a Y B C O (short dashed line) and a Y B M g C O (long dashed line) diffraction component each split by the K Q i and K q 2 x-ray doublet. of the peaks were left as free parameters. The best fit was obtained with F W H M values of 0 .45° for both the Y B C O Kal and K A 2 peaks, and 0.32° for both Y B M g C O peaks. The ratio of the intensities of K Q i and Ka2 were found to be 2.1 ± 0.2 for both the Y B C O and Y B M g C O contributions. This is in good agreement with that expected from the x-ray source (2.0). These results are consistent with that expected for a crystalline Y B C O / Y B M g C O bilayer. Compositional mapping was done with SIMS. In the previous chapter (figure 4.4) the SIMS result of the depth profile of Y , C u , and M g in a bilayer of Y B C O / Y B M g C O was shown. The top graph revealed a uniform Y and C u concentration throughout the Chapter 5. Structuring with Mg ion implantation 93 bilayer. A step function-like M g depth profile(with an interface width of roughly 15 nm) is shown in the bottom graph, suggesting that significant intermixing or diffusion did not occur during the deposition of the top Y B C O film. Post-annealing at 9 0 0 ° C was successful in obtaining a fairly uniform M g distribution throughout the bottom layer. The lateral diffusion of M g was determined using E P M A . If implantation is to be used to pattern several layers in a multilayer structure, then vertical diffusion is also important. The film thickness is typically only a few or several hundred nanometers, and with an annealing temperature of over 9 0 0 ° C , there is the concern that M g would spread throughout the structure. To mimic the actual fabrication conditions, the processing sequence for the bilayers was modified from the previous ones. A film was first grown, and removed from the chamber. Without further processing, an additional film was grown. The top film was implanted with the same conditions (80keV, 2xl0 1 5 ) and the entire structure annealed at 9 0 0 ° C . SIMS analyses were done on a portion of the sample that was not annealed, and one that was. The results are shown in figure 5.7. The M g depth profiles of both the annealed and unannealed samples were well fit to Gaussians. This is in agreement with that predicted by T R I M simulations and general diffusion theory, described in the appendix. The solution to equation C.4 for diffusion from a planar source is 2 C(x,t) =-4=exp(-4r) (5.1) where m is the total amount of the diffusing species. If we compare the Gaussian shaped distribution profiles before and after annealing, then the diffusion coefficient can be obtained from the following: rr2 - rr2 D(T) = ^ m t u d (5.2) where t is the annealing time, and T, the annealing temperature for the diffused profile; Chapter 5. Structuring with Mg ion implantation 94 0.05 Film Depth (nm) Figure 5.7: Secondary Ion Mass Spectra of the M g depth profile in a Y B M g C O / Y B C O bilayer with and without annealing. The solid lines are Gaussian fits to both profiles. a represents the width of the Gaussian profile. From 5.7, we obtain the values of a and o~initial, which are Oinitiai=71nm and cr=210nm. From equation 5.2, and assuming that the depth profiles are determined primarily from annealing at 9 0 0 ° C (i.e. we ignore the ramp up and cool down stage) we obtain a diffusion coefficient of 2.5xl0~ 1 5 cm 2 / s at a temperature of 1173K. For comparison, the diffusion coefficient for oxygen at the same temperature is ~ 1 0 _ 8 c m 2 / s [ l l l ] , and for Zn and most other transition metals, ~10 - 1 0 cm 2 / s [112] . These measurements were made on polycrystalline samples, and are likely dominated by in-plane diffusion. In contrast, the value of 2 . 5 x l 0 - 1 5 c m 2 / s is strictly for c-axis diffusion, and it is expected that diffusion in this direction would be several orders of magnitude lower. Measurements done on oxygen diffusion[113] observed that Chapter 5. Structuring with Mg ion implantation 95 diffusion in the c-axis direction is 1 0 - 4 - 1 0 - 6 smaller than that in the a-b plane. In the above case, the lateral straggling of the M g implantation distribution was much thinner than the bilayer thickness. Thus the post-anneal conditions were not sufficient to obtain a uniform profile. To overcome this, a double implant can be done. A thin film of Y B C O (200 nm) was first grown, M g implanted and post-annealed. A n additional layer was grown on top, implanted and annealed. The resultant bilayer possessed a transition temperature of 79K, measured by d.c. magnetization. The SIMS measurement of this bilayer is shown in figure 5.8, and the uniformity of the M g concentration is apparent. CO c CD 20 60 100 140 180 220 260 300 340 380 Bilayer depth (nm) Figure 5.8: Secondary Ion Mass Spectra of a bilayer with M g implanted into both layers. The uniformity of the M g concentration is apparent. For comparison, the C u and Y concentrations were also mapped. Chapter 5. Structuring with Mg ion implantation 96 5.3 Concluding remarks The goal of this thesis was to develop a multilayer patterning method without removing or destroying material. This chapter demonstrated the ability to pattern, with a resolution of several microns, Y B C O thin films. As well, bilayers have been demonstrated where the bottom film is normal at 80K, while the top film maintains its 90K transition temperature. The quality of the top film is supported by its high critical current at 84K and the appearance of sharp 6/26 x-ray diffraction peaks corresponding to both the top and bottom films. If the concentration of M g is kept below the solubility limit (x~0.01-0.02), then the post-processing surface is still suitable for additional layers. However, at doses corresponding to M g concentrations beyond the solubility limit, the compositional excesses result in outgrowths on the surface. As well, these doses also result in damage levels that cannot be completely recovered. Annealing at 9 0 0 ° C can easily induce diffusion on the order of several hundred nanome-ters. This is useful in the case of a implant into a single layer film, where diffusion can homogenize the concentration. In some types of bi- or multilayers, with typical film thickness on this order, an implantation in one layer will diffuse into adjacent layers. This would not occur, however, when the bottom film is implanted and an additional film (which only undergoes film growth annealing conditions) grown on top. M g implan-tation patterning is limited to only one sub-surface layer. The surface layer would not require any post-annealing since no further films would be grown on top. A double implant was done on a "bilayer" of Y B M g C O / Y B M g C O , with a total thick-ness of 400 nm. A very homogeneous M g distribution was obtained after high temperature post-annealing. Chapter 6 Conclusions and Future Improvements The motivation of this thesis was to develop a multilayer patterning technique that did not remove or destroy material. That was accomplished through the technique of substitutional ion implantation. However, some modifications and compromises had to be made. First, the initial concept of deoxygenating a sample by implanting a reactive ion, such as Si, was shown not to be suitable for multilayer patterning. The doses required to deoxygenate the sample resulted in extensive lattice damage. The only recourse was a high temperature anneal to regain the crystallinity, which restored superconductivity in implanted regions. The method of patterning using substitutional ion implantation was to implant an ion that would be thermodynamically stable in Y B C O . Unfortunately, the non-superconducting portion could not be made insulating; it is normal at 77K and a superconductor at T<65K. Secondly, an additional processing step (the high temperature anneal) was nec-essary after implantation. Unfortunately, these compromises place constraints on the types of multilayer structures that can be patterned with this technique. Substitutional implantation patterning can only be applied to the first layer in the structure. However, this still allows for the growth of planar films in the second and third layers (if the second layer does not require patterning), which would still be a clear advantage over conventional patterning techniques. With this, possibly the most suitable application of this processing technique is in SQUID magnetometers where implantation 97 Chapter 6. Conclusions and Future Improvements 98 can be used i n con junc t i on w i t h e t ch ing techniques. S Q U I D magne tometers require the c o n s t r u c t i o n of a f lux t ransformer w h i c h is pa t t e rned by f ab r i ca t i ng a c o i l i n the first layer, fo l lowed by an i n su l a to r i n the second layer, a n d is comple t ed w i t h another c o i l i n the t h i r d layer , dep ic t ed i n figure 6.1. T h e v ia s to the b o t t o m c o i l w o u l d s t i l l require an e t ch ing step, however, the t op c o i l is free o f the h i l l s a n d val leys tha t were unavo idab le w i t h o ther p a t t e r n i n g schemes. top view cross section F i g u r e 6.1: Ion i m p l a n t a t i o n can be used i n the fo l l owing m u l t i l a y e r p a t t e r n i n g process: a) T h e first layer film is pa t t e rned in to a S Q U I D us ing ion i m p l a n t a t i o n ; b) A d ie lec t r ic layer is depos i t ed on top o f the pa t t e rned film. A v i a can be pa t t e rned us ing t y p i c a l e t ch ing methods ; c) T h e top m u l t i - t u r n supe rconduc t i ng sp i r a l can be g rown on the p l ana r surface, also pa t t e rned us ing t y p i c a l e t ch ing methods . T h e v i a connect is i n d i c a t e d w i t h the ova l . F u r t h e r improvemen t s to s i m p l i f y the process have been ident i f ied: 1) T h e h i g h t empera tu re annea l c a n be replaced , a n d op t ions i nc lude a R a p i d T h e r m a l A n n e a l ( R T A ) or laser annea l ing . B o t h c o u l d speed up the process, a n d w i t h laser Chapter 6. Conclusions and Future Improvements 99 annealing, there also exists the possibility of localized annealing. It is possible that only the surface of the implanted layer would need to be recrystallized for growth of top layer films. 2) Multiple implantations have not been extensively investigated, but were previously done to obtain a more homogeneous implant profile. Using multiple implantations at lower doses may also reduce the damage imparted to the crystal. Five implants at l x l 0 1 5 c m ~ 2 , each followed by a high temperature anneal would likely result in a more crystalline material than one implant at 5x l0 1 5 cm~ 2 , with a post-anneal. 3) At present only impurity doping into Y B C O was investigated, and Mg is believed to be the most suitable choice for implantation patterning. However, light element doping has not been a thoroughly investigated with other HTS materials. There exists the possibility that transition temperatures can be more strongly suppressed with a different HTS material and dopant. Appendix A X-Ray Diffraction Measurements X-Ray Diffraction(XRD) was the standard method of quantifying the crystallinity of the thin films throughout this thesis. Two types of X R D measurements were performed. First, a standard powder scan on a Rigaku1 powder diffractometer using the following recipe: 1. Align sample with respect to the (200) peak of SrTiOs , the substrate of the film. 2. At a power level of 20kW x 10mA, scan this peak at a rate of l ° / m i n . (this intensity was used to normalize the x-ray intensities from day to day) 3. Align the sample with respect to the (005) peak of the Y B C O film. 4. At a power level of 50 k W x 150 m A , scan the (00/) peaks of Y B C O from 1=1 to 1=7, or equivalently from 2 0 = 5 ° to 2 0 = 7 0 ° at a rate of l ° / m i n . 5. The (008), (0010), (0011), and (0013) peaks were scanned at 50 k V and 150mA for a range of 2° about the center of the peak with a rate of 0 . 2 ° / m i n . The typical result would look like what is seen in figure 3.7. To understand what the peaks mean and where they come from, we briefly overview x-ray kinematical diffraction theory 2. Consider a lattice in real space with an incoming x-ray beam of momentum k and a diffracted x-ray beam of momentum k', depicted in figure A.1 . It can be easily seen that 1 2The kinematical theory assumes that the diffraction x-rays do not interact with incoming x-rays or those diffracted from other crystal planes. To consider this, the more involved x-ray dynamical theory is appropriate. The interested reader is referred to reference [114]. 100 Appendix A. X-Ray Diffraction Measurements 101 d • d cos G' d cos 9 Figure A . l : X-rays with momentum k impinge on the line of diffraction points, resulting in diffracted x-rays with momentum k' for constructive interference of the diffracted waves: But since ti(cos 9' + cos 9) = mX d-cos9' = - d • k' d • cos 9 = d • k where k and k' are the unit vectors in the k and k' directions, then d • (k - k') = mX or (A.l) (A.2) (A.3) (A.4) d • (k - k') = 2-nm (A.5) The distance, d, is just the Bravais lattice vector, Ft, in real space, and we can therefore write which is equivalent to writing R • (k - k') = 27rm Di(k'-k)-R (A.6) (A.7) Appendix A. X-Ray Diffraction Measurements 102 The above expression for k-k' is also the definition of the reciprocal lattice vector, and therefore k' - k = K (A.8) where K is the reciprocal lattice vector. Thus, the Laue condition is that the difference between the diffracted and incident wave vectors must be equal to a reciprocal lattice vector for a diffraction peak to appear. The magnitudes of k and k' are equal in elastic diffraction. Therefore from the Laue condition we also see that k = |k' - K | (A.9) and thus, k-K=^K (A.10) The component of k in the direction of the reciprocal lattice vector K must be half the length, K. Another approach to the same result is to consider the crystal as a set of reflect-ing planes. By setting a condition for constructive interference, a Bragg condition for diffraction peaks is easily obtained. Namely, 2dsin0 = mA ( A . l l ) where d is the interplanar distance, and 9 is the angle of incident of the x-ray. For standard powder diffraction, the crystallites are oriented in all possible directions. The 9/29 scan probes reciprocal space by moving in a straight path and will produce peaks at angles where a reciprocal lattice vector connects two points on the circumference of the Ewald Sphere with radius k. Because the crystallites are oriented in every direction, this scan can "pick up" diffraction peaks corresponding to reciprocal lattice vectors in all directions. In the case of c-axis oriented thin films, the grains within the film all Appendix A. X-Ray Diffraction Measurements 103 have a preferred orientation, and since 6/26 x-ray diffraction scans move in a straight and vertical direction, only the (00/) reciprocal lattice vectors will be detected. The appearance of only (00/) peaks indicate the absence of misaligned grains, thereby verifying the highly c-axis orientation of the film. 6 Figure A.2: A n Ewald sphere for the x-rays are drawn within the reciprocal lattice of a c-axis oriented thin film of Y B a 2 C u 3 0 7 illustrating the source of the (00/) diffraction peaks. This shows the geometry for the (005) diffraction peak. The breadth of the peaks is determined by several factors: the crystallite size, the homogeneity, and the thickness of the film. The shape is most commonly fitted to either a Gaussian or Lorentzian line profile, although it is rarely precisely one or the other. Because of this, it is usually not (but can be) used as a measure of crystallinity. The rocking curve (RC), which is described later, is the more common tool for characterizing the crystallinity. A n accurate value of the c-axis length can be determined from the 6/26 scan. Because of the orientation, the interplanar distance, d, in the Bragg equation is, to a first approx-imation, the c-axis length. But several factors (which are dependent on the Bragg angle) Appendix A. X-Ray Diffraction Measurements 104 such as sample eccentricity, x-ray absorption, and divergence of the x-ray beam cause deviations, A c , of the value d from the true c-axis length. A standard extrapolation method is often used to deal with these effects, introduced by Taylor and Sinclair[115] and Nelson and Riley[116], in the 40's. This method is valid if the dominating correction is absorption. The technique is to plot the calculated value of d as a function of / w = c o s 2 e ' ^ + ^ » < A - 1 2 > and extrapolate the line to f(#)=0, since f(#)oc A c / c . The 9/29 provides information on the macroscopic orientation of the film, the presence of impurity phases and the c-axis length. For a more detailed knowledge on the films' crystallinity, a measurement referred to as the x-ray rocking curve (RC) was performed using a Bede3 Double Crystal Diffractometer (shown schematically in figure A.3). The following recipe was used: 1. The sample was aligned such that the x-ray beam was positioned in the center of the sample. 2. The detector and the sample were oriented to their optimal positions with respect to its (005) Bragg peak. 3. The peak was scanned by keeping the detector stationary and rotating the sample about the optimal Bragg position. The x-ray rocking curve is a different probe of k-space than the 9/29 scan. Whereas the 9/29 scan moves vertically in k-space, the rocking curve moves horizontally, hence detecting the angular distribution of the reciprocal lattice vector, K corresponding to that Bragg peak. The angular distribution is sensitive to the quality of the crystal. This is illustrated in figure A.4. It is most common to fit the rocking curves of crystals with a function comprised of Lorentzian and Gaussian components. The width of the Appendix A. X-Ray Diffraction Measurements 105 Si(111) monochrometer Collimator / / / Figure A.3: The Bede diffractometer is constructed such that the C u ^ - a x-rays pass through a collimator(constructed with a channel cut into a Si(220) crystal) and is directed toward a S i ( l l l ) monochrometer, which selects only the Cupcai x-ray(A=0.1540598 nm). The incoming beam to the sample is thus very monochromatic and nicely collimated. The measurement of the x-ray rocking curve is done by holding the detector still and rotating the sample about a Bragg angle (the (005) peak is the most common choice). The Rigaku diffractometer setup, used for the 9/29 powder diffraction described earlier, is very similar with the exceptions that the S i ( l l l ) monochrometer is replaced with a graphite x-ray mirror and there is no collimator. The absence of the monochrometer and collimator results in a much more intense and dispersive beam comprised of Cuj<-Qi, Cuj<-a2 and CUK/3 x-rays. rocking curves decrease with increasing crystallinity. To illustrate, the rocking curves, measured on the same diffractometer, are shown for several crystals which are known to have very high crystallinity: S i ( l l l ) , a very high purity YBa2Cu30"7 crystal grown in a Yt tr ia Stabilized Zirconia(YSZ) crucible, and an ultra high purity Y B a 2 C u 3 0 7 crystal grown in a BaZrO"3 (BZO) crucible. The rocking curves are shown in comparison to that of the best films in this work, illustrated in figure A.5. There is clearly a close correlation between rocking curve width and the quality of the crystal. The shape as well as the width of the rocking curves provide information on the crystalline quality. It had been discussed in chapters 2 and 3 that the Gaussian shaped Appendix A. X-Ray Diffraction Measurements 106 Figure A.4: The source of the rocking curve is most easily seen when viewed in k-space. The incoming x-ray, k and the diffracted x-ray, k', are drawn within the reciprocal lattice of c-axis oriented Y B C O . As the sample is "rocked" by an angle A u , the resultant k-k' vector is rotated about K , the reciprocal lattice vector. Previously, it was shown that a Bragg peak will only result when k-k' = K . At sufficiently large Au, k-k' is no longer equivalent to K , and the intensity of the scattered x-rays disappear. The rocking curve width reveals the angular distribution of the K vector, and hence the crystallinity of the film. rocking curve indicated a granular film. The Lorentzian was more typical of a crystalline, but defected sample. This was only applicable to samples of broader rocking curve widths (lower quality crystals). Wi th the very high quality crystals (for example those in figure A . 5 ) , the rocking curves were fit best with a function comprised mainly of a Gaussian lineshape. In fact the narrower the rocking curve width, the more Gaussian its shape. This is most likely due to the fact that the widths of the samples' rocking curves are approaching the limit of the x-ray beam, whose spot profile is very well approximated by a Gaussian. Appendix A. X-Ray Diffraction Measurements 107 Figure A.5: X-ray rocking curves of (listed in order of decreasing rocking curve width) a Y B C O film (005) (*), very high purity(vhp) Y B C O crystal(006) (•), ultra high pu-rity(uhp) Y B C O crystal (006) (A) and a Si crystal (111) (•). A l l curves were fit to a function comprised of a mix of Gaussian and Lorentzian (or Cauchy) profiles. The nar-rower the rocking curve, the more Gaussian the shape: Y B C O film was 100% Lorentzian, the vhp Y B C O crystal, 60% Lorentzian, the uhp Y B C O crystal, 37% and the silicon crys-tal, 10%. Appendix B SQUID Magnetization Measurements Characterization of the superconducting properties is done by probing the samples' elec-tromagnetic properties. Generally, this is accomplished by applying an external field and observing the response of the sample. Unfortunately, many of the accessible techniques have some unavoidable constraints. For example, to measure the d.c. resistivity of the sample, the placement of metal contacts was necessary, rendering the sample unusable for many subsequent measurements. Microwave loss measurements require very specific substrates onto which the films were grown. These substrates, in particular LaAlC>3, do not result in the most crystalline films, and were not used. For many of the experiments in this work, these constraints were unacceptable, and the most suitable method was d.c. S Q U I D Magnetization. Magnetization measurements probe the most fundamental property of a superconduc-tor, the Meissner Effect. Shown schematically in figure B . l , this is the near complete expulsion of magnetic flux from the sample when the temperature and the field are suffi-ciently low (i.e. T < T C and H < H C ) . If the samples are extremely pure, this is a completely reversible process. To briefly overview, we can write (in Gaussian units), the magnetic induction B in terms of the magnetization of the sample, M , and the applied field, H . B = ( H + 4TTM) (B.l) But in a superconductor, the magnetic flux, B = 0 and thus, M = - — H (B.2) 108 Appendix B. SQUID Magnetization Measurements 109 T > T C T < T C T < T C H>0 Hc>H>0 H=0 B=0 Figure B . l : The Meissner effect is shown in the bottom of the above figure. In contrast, a perfect conductor is shown on the top. In both cases, at a temperature above its transition to the perfect or superconductor state, an external field will penetrate through the sample. As the sample is cooled to below the transition, nothing should occur in the perfect conductor as B remains zero. However, for the superconductor, which has the more stringent constraint that B=0, currents are induced at the surface to oppose the applied field(perfect diamagnetism). When the applied field is removed, the perfect conductor generates persistent currents to maintain the flux in the sample. In the superconductor, the currents die out with the removal of the applied field. Since both of these conductors require the generation of persistent currents to satisfy their field constraints, zero resistivity must be a property of both. However, it is the additional property of perfect diamagnetism which distinguishes the superconductor from all other electronic materials. Appendix B. SQUID Magnetization Measurements 1 1 0 From the definitions of the permeability, fi, and the susceptibility, x> B = fiH (B.3) M = x H (B.4) we obtain the expression for x, Thus, the measurement of the magnetic susceptibility, x, probes superconductivity at its most fundamental level and as well, the measurement of x(T) can provide information on the quality of the superconductor. For the perfect superconductor, x= - 4^  when T < T C and 0 for T>T C . The transition from the superconducting to the normal state is narrow for pure samples and wide for poor ones. Measurement of x(T) is most easily done with the SQUID magnetometer. The mag-netometer used in this thesis was obtained from Quantum Design1. The magnetometer detects the magnetic behavior of a sample by measuring the creation of a voltage signal from the changes in the magnetic field as the sample is moved through the field. The signals are induced in a second derivative coil set (which eliminates drifts in the ap-plied uniform magnetic field) and is coupled into the SQUID sensor via a RFI isolation transformer. A schematic of the SQUID magnetometer is shown in figure B.2. The output is a voltage signal which was a function of the distance the sample was moved through the coils, and is characterized by a tri-peak function illustrated in figure B.2. A fit by the magnetometer software to this function produced a value for the magnetization, and hence the susceptibility. Care was taken to ensure that the shape of the output signal did not deviate from the expected tri-peak function since the relevance of the value obtained for the magnetization was sensitive to the quality of the fit. 1 Appendix B. SQUID Magnetization Measurements 111 Heater Signal Coil Second Derivative Coil Set Figure B . 2 : The S Q U I D Magnetometer detects the magnetization of the sample by moving it through a second derivative coil set within an applied field. The magnetization of the sample induces current into the circuit and is coupled into the S Q U I D sensor via an R F I isolation transformer. The transformer heater, turned on while the magnet is charging and at the beginning of a sample measurement, is used to eliminate all persistent currents in the S Q U I D input circuit. Appendix B. SQUID Magnetization Measurements 112 The films used in this study were not defect-free, and contained many strong pinning sites. The measured value of x w a s therefore dependent on the measurement history. W i t h this in consideration, two methods of measuring x (T) were possible. B . l Zero Field Cooled(ZFC) Measurements The zero field cooled measurement is done by cooling the sample to a temperature far below its transition temperature, and then applying the field. In this case the field is expelled from the sample in the manner depicted in figure B . l . The recipe used for Z F C measurements was as follows: 1. The magnet was set to apply a field of 3 Oe, oriented normal to the films' surface. 2. Sample was inserted into the cryostat such that the sample was located just above the magnet and left to cool to a temperature of 5 K . 3. Once the temperature of the sample reached 5 K , the sample was slowly inserted into the magnet. 4. The position of the sample in the field was verified and the sequence activated to measure the magnetization of the sample as it warmed to a temperature above its transition. At each temperature setting (typically in steps of 5 K far from the transition narrowing to 0.25 K steps near T c ) , the sample was scanned vertically 2 cm (the region where the field remained homogeneous) 2. The Z F C measurement has the disadvantage that the width of the transition is not entirely an intrinsic property of the material. Because of its thin film geometry, the field near the edges are much more intense (see the diagram in figure B . l ) and have an effective strength much larger than the actual field strength. Thus, near the edges, the field will penetrate the sample, hence broadening 2 For fitting purposes, the sample should be scanned its maximum distance of 5 cm, however the field homogeneity would be compromised for this improved fit. This is more of a concern with samples that exhibit a magnetic hysteresis (like superconductors) than other magnetic samples. Appendix B. SQUID Magnetization Measurements 113 the true transition width. However, the transition temperature as well as sufficient degradation of the samples' superconductivity will be evident in the ZFC measurement. B .2 F i e l d Coo led (FC ) Measurement The field cooled measurement is done by first applying the magnetic field while the sample is in its normal state, then cooling it through the transition temperature. If the sample is extremely pure, then the field would be completely expelled. However, for highly defected samples, such as most thin films, there is pinning of the field shown schematically in figure B.3. However, since the field is not "pinched" at the edges in the manner described in the ZFC measurements, the width of the measured transition is believed to be more accurate than the ZFC measurement. A drawback is that the SQUID signal was dominated by the strong pinning in the samples, which was in addition to the diamagnetic signal in the other regions of the sample. The resultant SQUID signal was a distorted tri-peak function which was fit with weak paramagnetism. Due to the distorted shape of the tri-peak function, this fit was up to interpretation. The FC measurement was therefore not a standard one, and was conducted on only a few samples. Appendix B. SQUID Magnetization Measurements 114 Zero Field Cooled vortices Field Cooled Figure B .3: The Zero Field Cooled (ZFC) sample expels the magnetic field. The result is a strong diamagnetic signal measured by the S Q U I D magnetometer. The highly defected Field Cooled(FC) sample traps much of the field into bundles, or more formally, vortices. In this case, the sample contains a mixture of diamagnetism and the paramagnetic signal from the trapped fields. The result is a weak paramagnetic signal. A p p e n d i x C Di f fus ion If there is only a concentration gradient of the diffusing species, then Fick's first and second laws can be applied to the process. The equations for Fick's first and second laws are: r)C clC f)C dC d2C d2C d2C ~dt ~ V x x d ^ + VyyW + Dzz~d* {L'2) where C is the concentration of the diffusing species, Da are the nth components of the diffusion tensor (of the diffusion coefficients in the ith direction), and J i s the flux of the diffusing material. In the present case, we only have a concentration gradient in the z or c-axis direc-tions. Within any given x-y or a-b plane the concentration of the diffusing species is homogeneous. If we assume that D is independent of position, then the equations can be simplified to: 1: BC J = - D - t (C.3) and: -8T = D ^ ( a 4 ) 1This assumption of D being independent of position is the case for a homogeneous medium. For an inhomogeneous medium, (i.e. D=D(x)), there is a loss of generality when this assumption is made. To treat that problem properly, one must be careful as to where in equation C.4 to insert D. See reference [117] 115 Appendix C. Diffusion 116 for the one dimensional case, which is applicable to the present work. The diffusion coefficient,/), is temperature dependent and in many situations follows the Arrhenius relationship: D = D0exp(^) (C.5) where D0 is a temperature independent constant, E 0 , the activation energy, k# Boltz-mann's constant (1.38 x 1 0 _ 2 3 J / K ) , and T is the temperature. If D is independent of concentration, C and position x, then equation C.4 can be written: (C6) where t'=Dt, and also independent of. temperature. Given the same diffusion process at different temperatures, the time will scale with the Arrhenius factor that appears in the diffusion coefficient. A point to note is that in Y B C O thin films, grain boundaries become a shunt for diffusion of the material since diffusion along a grain boundary can be much faster than in the bulk. Bibliography R . W . Kiernan, Superconductor Industry, Winter 1995, 13 (1995) Estimated by summing the revenues of the companies producing H T S products: Conductus, STI, American Superconductor, Illinois Superconductor, Colorado Su-perconductor, Pirelli Q . Y . M a , P. Dosanjh, J . F . Carolan and W . N . Hardy, Appl. Phys. Lett 63, 3633 (1993) Q . Y . M a , E .S . Yang, G . V . Treyz, C h i n - A n Chang, Appl. Phys. Lett. 55, 896 (1989) B. D . Joyce, J . A . Baldrey, Nature 195, 485 (1962) K . Nagai, M . Ozeki, J. Cryst. Growth 68, 200 (1984) C . Rossel, U . Kaufmann, H . Downar, R. Schulz, Physica C 185-189, 255 (1991) D. P. Kern, K . Y . Lee, R . B . Laibowitz, A . Gupta, J. Vac. Sci. Technol. B 9, 2875 (1991) C . A . J . Damen, Reactive Patterning Techniques for High-Tc Superconductors, P h D thesis, University of Twente, The Netherlands (1997) W . H . Mallison, S.J. Berkowitz, A .S . Hirahara, M . J . Neal and K . C h a r , Appl Phys. Lett 68, 3808 (1996) C L . Pettiette-Hall, J . Murduck, J . F . Burch, M . Sergant, R. Hu, J . Cordrump, M . Luong, and R . K . Ellis, IEEE trans. Appl Supercond. 7, 205 (1997) J . Talvacchio, M . G . Forrester, B . D . Hunt, J . D . McCambridge, and R . M . Young, IEEE trans. Appl Supercond. 7, 2051 (1997) Y . Q . Shen, Z . J . Sun, R. Kromann, T . Hoist, P. Vase and T . Freloft, Appl. Phys. Lett 67, 2081 (1995) F . Ludwig, D. Koelle, E . Danstker, D . T . Nemeth, A . H . Miklich, John Clarke, R . E . Thomson, Appl. Phys. Lett. 66, 373 (1995) S.J. Berkowitz, A .S . Hirahara, K . Char, E . N . Grossman, Appl. Phys. Lett. 69, 2125 (1996) 117 Bibliography 118 [16] J . F . Zeigler, Ion Implantation:Science and Technology, Academic Press, San Diego (1988) [17] S.K. Gandhi, VLSI Fabrication Principles, Wiley, New York (1983) [18] Q Y . M a , S.H. Hong, J .R. Miller, J .Z . Sun, W . J . Gallagher and R . B . Laibowitz, Supercond. Sci. Technol. 9, A92 (1996) [19] D . C . Degroot, D . A . Rudman, K . Zhang, Q Y . M a , H . Kato, N . A . F . Jaeger, Appl. Phys. Lett. 69, 2119 (1996) [20] M . K . W u , J .R. Ashburn, C . J . Torng, P . H . Hor, R . L . Meng, L . Gao, Z . J . Huang, Y . Q . Wang, C . W . Chu, Phys. Rev. Lett. 58, 908 (1987) [21] Physical Propreties of High Temperature Superconductors I-V, D . M . Ginsberg (ed.), World Scientific, Sinapore (1989-1995) [22] V . E . Zubkus, O . E . Parfionov, E . E . Tornau and P.J . Kundrotas, Physica C 198, 141 (1992) [23] Paul Richard Shleger, Thermodynamics of Oxygen Ordering in YBa^Cu^O^x, Ph .D . thesis, University of British Columbia, (1992) [24] Jinhua Ye, Keikichi Nakamura, Phys. Rev. B 48, 7554 (1993) [25] D. Basov, R. Liang, D . A . Bonn, W . N . Hardy, B . Dabrowski, M . Quijada, D . B . Turner, J .P. Rice, D . M . Ginsberg and T . Timusk, Phys. Rev. Lett. 74, 598 (1995); K . Zhang, R. Liang, D . A . Bonn, W . N . Hardy, Phys. Rev. Lett.73, 2484 (1994); A . Kulpa, A . C D . Chaklader, G . Roemer, D .L . I . Williams, W . N . Hardy, Supercond. Sci. Technol. 3, 483 (1990) [26] K . Kitazawa, J . Shimoyama, H . Ikuta, K . Kishio, Proceedings of the 10th Anniver-sary HTS Workshop on Physics, Materials and Applications, eds. B . Batlogg, C . W . Chu, W . K . Chu, D . U . Gribser, K . A . Muller, World Scientific, Singapore, 151 (1996) [27] Y . Huang, K . L . Merkle, L .P . Lee, M . Teepe and K . Char, Appl. Phys. Lett 71, 3703 (1997) [28] H . Q . L i , R . H . Ono, L . R . Vale, D . A . Rudman, and S.H. Liou, Appl. Phys. Lett 69, 2752 (1996) [29] D.P. Norton, A . Goyal, John D . Budai, David K . Christen, Donald M . Kroeger, Eliot D . Specht, Qing He, Bernd Saffian, M . Paranthaman, Charles E . Klabunde, Dominic F . Lee, Brian D. Sales and Fred A . List, Science 274, 755 (1996) Bibliography- lid [30] M . A . Harmer, C . R . Fincher, B . A . Parkinson, J. Appl. Phys. 70, 2760 (1996) [31] R. Sum, H.P. Lang, H . J . Gunterodt, Appl. Surf. Sci. 86, 140 (1995) [32] B . Dam, N . J . Koeman, J . H . Rector, B . Stauble-Pumpin, U . Poppe, R. Griessen, Physica C 260, 1 (1996) [33] S.K. Streiffer, B . M . Lairson, C B . Eom, B . M . Clemens, T . H . Geballe, J . C . Bravman, Phys. Rev. B 43, 13007 (1991) [34] B . Dam, preprint to be published in Physica C, 1998 [35] B . Hauser, M . Diegel, H . Rogalla, Appl. Phys. Lett. 52, 844 (1988) [36] P . M . Mankiewitz, J . H . Scofield, W . J . Skocpol, R . E . Howard, A . H . Dayem, E . Good, Appl. Phys. Lett. 51, 1753 (1987) [37] Y . Yoshizako, M . Tonouchi, T . Kobayashi, Jpn. J. Appl. Phys. pt.2 26, L1533 (1987) [38] I. Shih, C X . Qiu, Appl. Phys. Lett. 52, 1523 (1988) [39] A . Enokihara, H . Higashino, K . Setune, T . Mitsuyu, K . Wasa, Jpn. J. Appl. Phys. pt.2 27, L1521 (1988) [40] S. Matsui, Y . Ochiai, Y . Kojima, H . Tsuge, N . Takado, K . Asakawa, H . Matsutera, J . Fujita, T . Yoshitake, Y . Kubo, J. Vac. Sci. Technol. B 6, 900 (1988) [41] H . Tsuge, S. Matsui, N . Matsukura, Y . Koj ima, Y . Wada, Jpn. J. Appl. Phys. pt.2 27, L2237 (1988) [42] S. Matsui, N . Takado, H . Tsuge, K . Asakawa, Appl. Phys. Lett. 52, 69 (1988); J . W . C . de Vries, B . Dam, M . G . J . Heijmann, G . M . Stollman, M . A . M . Gijs, C . W . Hagen, R.P . Greissen, Appl. Phys. Lett. 52, 1904 (1988) [43] H . Tabei, H . Miyazawa, K . Hotta, S. Yoshizawa, and M . Murakawa, Physica C 2 3 5 -240, 3431 (1994); T h . Schuster, H . Kuhn, A . Raiber, T . Abeln, F . Dausinger, H . Hugel, M . Klaser, and G . Muller-Vogt, Appl. Phys. Lett 68, 2568 (1996) [44] J . Boneberg, M . Bohmisch, M . Ochmann, and P. Leiderer, Appl. Phys. Lett 72, 3805 (1997) [45] R . H . Koch, C P . Umbach, G . J . Clark, P. Chaudari, R . B . Laibowitz, Appl. Phys. Lett. 51, 200 (1987) [46] A . J . Pauza, A . M . Campbell, D . F . Moore, R . E . Somekh, A . N . Broers, IEEE trans. Appl. Supercond. 3, 2405 (1993) Bibliography- nO Jean-Pierre Locquet, Frederic Arrouy, Erich Machler, Michel Despont, Peter Bauer and Erica J . Williams, Appl. Phys. Lett. 68, 1999 (1996) P. Seidel, E . Heintz, F . Schmidl, K . Zach, H - J . Kohler, H . Schneidewind, J . Borck, L . Dorrer, S. Linzen, T . Kohler, W . Michalke, M . Manzel, E . Steinbeiss, H . Bruchlow, E . - B . Kley, H . - J . Fuchs, IEEE trans. Appl. Supercond. 3, 2353 (1993) P. Langlois, D . Robbes, M . Lam Chok sing, C . Gunther, D . Bloyet, J . F . Hamet, R. Desfeux, H . Murray, J. Appl. Phys. 76, 3868 (1994) C . L . Jia, B . Kabiaus, K . Urban, K . Herrman, G . J . Cui , J . Schubert, W . Zander, A.I. Braginski, C . Heiden, Physica C 175, 545 (1991) V . R . Todt, X . F . Zhang, D . J . Miller, M . St. Louis-Webber, and V . P . Dravid, Appl. Phys. Lett. 69, 3746 (1996); G . Friedl, B . Ross, M . Romheld, L . Schultz, W . Jutzi, Appl. Phys. Lett.59, 2751 (1991) S. Kuriki , T . Kamiyama, D . Suzuki, IEEE trans. Appl. Supercond. 3, 2461 (1993) A . P . Marathe, T . Van Duzer, IEEE trans. Appl. Supercond. 7, 3834 (1997) M . L . Lucia, J.Santamaria, F.Sanchez-Quesada, W . Lopera, M . E . Gomez, and P. Prieto, Physica C 260, 149 (1996) C . C . Chin, H.Takahashi and T . Morishita, J. Mater. Res. 8, 951 (1993) M . Kawasaki, J . Gong, M . Nantoh, T . Hasegawa, K . Kitazawa, M . Kitazawa, M . Humagai, K . Hirai, K . Horiguchi, M . Yoshimoto, and H . Koinuma, Jpn. J. Appl. Phys., Part 1 32, 1612 (1993) A . Cassinese, A . D i . Chiara, F . Miletto Granozio, S. Saiello, U . Scotti di Uccio, and M . Valentino, J. Mater. Res. 10, 11 (1995). M . Ece, E . G . Gonzalez, H . - U . Habermeier, and B. Oral , J. Appl. Phys. 77, 1646 (1995) M . Badaye, F . Wang, Y . Kanke, K . Fukushima, and T . Morishita, Appl Phys. Lett. 66, 2131 (1995) There are many sources describing H T S thin film growth methods, but the most complete are the Proceedings of the Applied Superconductivity Conference. For the most recent, see IEEE trans. Appl. Supercond. 7 (1997). P. Schneider, G . Linker, R. Schneider, J . Reiner, and J . Geerk, Physica C 266, 271 (1996) Bibliography 121 [62] Y . Watanabe, Y . Seo, M . Tanamura, H . Asami, and Y . Matsumoto, J. Appl. Phys. 78, 5126 (1995) [63] B . Dam, J . Rector, M . F . Chang, S. Kars, D . G . Degroot, and R. Greissen, Appl. Phys. Lett. 65, 1581 (1994) [64] S .G. Lee, D.S. Hwang, Y . K . Park, and J . C . Park, Appl. Phys. Lett. 65, 764 (1994) [65] K . H . Wu, R . C . Wang, S.P. Chen, H . C . L in , J . Y . Luang, T . M . Uen, and Y . S . Gou, Appl. Phys. Lett. 69, 421 (1996) [66] M . W . Gardner, S.A. Govorkov, Ruixing Liang, D . A . Bonn, J . F . Carolan, W . N . Hardy, Jour. Appl. Phys.83, 3714 (1998) [67] A . Gauzzi and D . Pavuna, Appl. Phys. Lett. 66, 1836 (1995) [68] L i Yijie, Ren Congxin, Chen Guoliang, Chen Jiamin, Yang Jie and Zou Shichang, Solid State Commun. 76, 663 (1990) [69] Q . Y . M a , P. Dosanjh, A . Wong, J . F . Carolan, and W . N . Hardy, Supercond. Sci. Technol. 7, 295 (1994) [70] Q . Y . M a , A . Wong, P.Dosanjh, J . F . Carolan, and W . N . Hardy, Appl. Phys. Lett. 65, 240 (1994) [71] G . J . Clark, A . D . Marwick, R . H . Koch, R . B . Laibowitz, Appl. Phys. Lett 51, 139 (1987); G . J . Clark, F . K . Legoues, A . D . Marwick, R . B . Laibowitz, R. Koch, Appl. Phys. Lett.51, 1462 (1987) [72] J . C . McCal lum, C . W . White, and L . A . Boatner, Nucl. Instr. and Meth. 40/41, 608 (1989) [73] J . A . Martinez, B . Wilkens, N . G . Stoffel, D . Hart, L . Nazar, T . Venkatesan, A . Inam, X . D . Wu, Appl. Phys. Lett..57189 (1990) [74] Yijie L i , Ren Congxin, Chen Guoliang, Chen Lianmin, Yang Jie, Lin Zixin, Chen Zhengxiu, Wang Le, Zou Shichang, J. Mater. Sci. Lett. 10, 171 (1991) [75] L . A . Abroyan Bulletin of the Russian Academy of Sciences 56, 959 (1992) [76] X . D . Wu, L . Luo, R . E . Muenchausen, S.R. Foltyn, J. Mater. Res. 7, 531 (1992) [77] Yijie L i , Guangchen Xiong, Zizhao Gan, Physica C 199, 269 (1992) [78] S.J. Pennycook, R. Feenstra, M . F . Chisholm, D.P. Norton, Nucl. Instr. and Meth. B 7 9 , 641 (1993) Bibliography 122 [79] W . T . Lin , H.P. Kao, Y . F . Chen, Y . K . Fang, J. Mater. Sci. 9, 781 (1993) [80] Yijie L i , S. Linzen, F . Machalett, F . Schmidl, P. Siedel, Physica C, 243, 294 (1995) [81] Y . H . L i , C . Leach, T . J . Tate, M . J . Lee, Yupu L i , A . Kilner, P . G . Quincey, J. Mater. Sci 30, 5927 (1995) [82] Y . H . L i , A . E . Stanton-Bevan, T . J . Tate, M . J . Lee, Yupu L i , J . A . Kilner, R . E . Somekh, J. Mater. Sci 15, 1836 (1996 [83] H Y . Nan, X . Yao, M . Murakami, S. Tanaka, Physica C 282-287, 1043 (1997) [84] M . Kuhn, B . Schey, R. Klarmann, W . Biegel, B . Strizker, J . Eisenmenger, P. Lei-derer, Physica C 294, 1 (1998) [85] Ruixing Liang, Department of Physics and Astronomy, University of British Columbia, private communication [86] Q Y . M a , IEEE trans. Appl. Supercond. 7, 2713 (1997) [87] Q Y . M a , A . Wong, J . F . Carolan, W . N . Hardy, H . Kato, D . Hui , and N . A . F . Jaeger, IEEE trans. Appl. Supercond. 5, 1181 (1996) [88] S.H. Hong, J .R. Miller, Q Y . M a , E.S . Yang and G . M . Luke, Appl. Phys. Lett. 67, 2717 (1995) [89] A . Wong, A . Kulpa, R. Liang, P. Dosanjh, J . F . Carolan, W . N . Hardy, H . Kato, and N . A . F . Jaeger, IEEE trans. Appl. Supercon. 7, 2134 (1997) [90] There are literally hundreds of papers describing this. For an early but good example, see J . M . Tarascon, P. Barboux, P .F . Miceli, L . H . Greene, G . W . Hull , M . Eibschutz, S.A. Sunshine, Phys. Rev. B 37, 7458 (1988) [91] C . Hart, O. Ares, J . L . Pena, P. Bartolo-Perez, V . Sosa, W . Cauich, G . A . Hirata, L . Cota-Araiza, M . H . Farias, Appl. Phys. Lett. 67, (1995) [92] R. Pinto, P.R. Apte, M.S .R . Rao, Ramesh Chandra, C P . D'Souza, S.P. Pai, L . C Gupta, R. Vijayaraghavan, K.I . Gnanasekar, M . Sharon, Appl. Phys. Lett. 68, 1006 (1996) [93] B . Moeckly, K . Char, Physica C 265, 283 (1996) [94] D . A . Bonn, S. Kamal , K . Zhang, Ruixing Liang, D . Barr, E . Klein, W . N . Hardy, Phys. Rev. B 50, 4051 (1994) Bibliography 123 [95] J . Ye, Z. Zou, A . Matsushita, K . Oka, Y . Nishihara, T . Matsumoto, Phys. Rev. B 58, R619 (1998) [96] S.B. Ogale, I. Takeuchi, M . Rajeswari, R . L . Greene, T . Venkatesan, D . D . Choughule, Appl Phys. Lett. 66, (1995) [97] J . T . Markert, Y . Dalichaouch, M . B . Maple, Physical Properties of High Temperature Superconductors I D.M. Ginsberg (ed.), World Scientific, Singapore (1989) [98] Y . Tokura, J . B . Torrance, T . C . Huang, A.I . Nazzal, Phys. Rev. B 38, 7156 (1988); A . Tokiwa, Y . Soyono, M . Kikuchi, R. Suzuki, T . Kajitani, N . Kobayashi, T . Sasaki, O. Nakatsu, Y . Muto, Jpn. J. Appl. Phys. 27, L1009 (1988); M . Kosuge, B . Okai, K . Takahashi, M . Ohta, Jpn. J. Appl. Phys.27, L1022 (1988) [99] Y . Dalichaouch, M.S . Torikachvili, E . A . Early, B . W . Lee, C L . Seaman, K . N . Yang, H . Zhou, M . B . Maple, Solid State Commun. 65, 1001 (1988) [100] J . M . Tarascon, P. Barboux, P .F . Miceli, L . H . Greene, G . W . Hull , M . Eibschutz, S.A. Sunshine, Phys. Rev. B 37, 7458 (1988) [101] Y . Maeno, T . Tomita, M . Kyogoku, S. Awaji, Y . Aoki , K . Hoshino, A . Minami, and T . Fujita, Nature, 328, 512 (1987); G . Xiao, M . Z . Cieplak, A . Gavrin, F . H . Streitz, A . Bakhshai, and C L . Chien, Phys. Rev. Lett. 60, 1446 (1988); M . Hiratami, Y . Ito, K . Miyauchi, T . Kudo, Jpn. Jour. Appl. Phys. 26, L1997 (1987) [102] D . A . Bonn, D . C Morgan, K . Zhang, R. Liang, D . J . Baar, W . N . Hardy, J. Phys. Chem. Solids 54, 1297 (1993); D . A . Bonn, S. Kamal , K . Zhang, R. Liang, D . J . Baar, E . Klein, ^W.N. Hardy, Phys. Rev. B 50 4051 (1994); T . R . Chien, Z. Wang, N.P. Ong, Phys. Rev. Lett 67, 2088 (1991) [103] S.X. Dou, N . Savrides, X Y . Sun, A . J . Bourdillon, C C Sorrell, J . R Zhou, K . E . Easterling, J. Phy.C. 20, L1003 (1987); M . F . Tan, W . W . Rhodes, P .K . Gallagher, J. Appl. Phys. 63, 821 (1988); T . Komatsu, O. Tanaka, K . Matusita, M . Takata, T . Yamashita, Jpn. J. Appl Phys. 27, L1025 (1988); R . J . O Jarvinen, K . J . Niemi, T . A . Mantylam, E.S . Heikkila, P . T . Vuorinen, Physica C 153, 882 (1988) [104] R . J . O Jarvinen, K . J . Niemi, T . A . Mantylam, E.S . Heikkila, P . T . Vuorinen, Physica C 153, 882 (1988); T . Suzuki, T . Yamazaki, A . Koukitu, M . Maeda, H . Seki, K . Takahashi, J. Mater. Sci. Lett.l, 926 (1988) [105] S.X. Dou, N . Savrides, X Y . Sun, A . J . Bourdillon, C C Sorrell, J.P. Zhou, K . E . Easterling, J. Phy.C. 20, L1003 (1987); R . J . O Jarvinen, K . J . Niemi, T . A . Mantylam, E.S. Heikkila, P . T . Vuorinen, Physica C 153, 882 (1988); T . Suzuki, T . Yamazaki, A . Koukitu, M . Maeda, H . Seki, K . Takahashi, J. Mater. Sci. Lett.l, 926 (1988) Bibliography 124 [106] W . M . Tiernan, R . B . Hallock, J . C . W . Chien, and B . M . Gong, Phys. Rev. B 44, 4661(1991) [107] L . Raffo, R. Caciuffo, D . Rinaldi and F . Licci, Supercond. Sci. Technol. 8, 409 (1995) [108] M . F . Yan, W . W . Rhodes, P .K . Gallagher, J. Appl. Phys. 63, 821 (1988) [109] Q . Y . M a , A . Wong, J . F . Carolan, W . N . Hardy, H . Kato, D . Hui, and N . A . F . Jaeger, "Ion implantation in high temperature superconducting films, IEEE trans. Appl. Supercond. 5, 1181 (1995) [110] A . Wong, R. Liang, M . Gardner, and W . N . Hardy, J. Appl. Phys. 82, 3019 (1997) [111] S.J. Rothman, J . L . Routbort, J . E . Baker, Phys. Rev. B. 44, 8852 (1989) [112] J . L . Routbort, S.J. Rothman, N . Chen, J . N . Mundy, J . E . Baker Phys. Rev. B. 43, 5489 (1991) [113] S.J. Rothman, J . L . Routbort, U . Welp, J . E . Baker, Phys. Rev. B. 44, 2326 (1991) [114] B . E . Warren, X-ray Diffraction, (1940) [115] A . Taylor and H . Sinclair, Proc. Phys. Soc. (London), 57, 126 (1945) [116] J . B . Nelson and D.P. Riley, Proc. Phys. Soc. (London), 57, 160 (1945) [117] N . G . van Kampen, Z. Phys. B - Condensed Matter, 68, 135 (1987) 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items