Sputter deposition of Y-Ba-Cu-O superconductor and SrTiO₃ barrier layer thin films


Material Information

Sputter deposition of Y-Ba-Cu-O superconductor and SrTiO₃ barrier layer thin films
Physical Description:
xii, 259 leaves : ill. ; 29 cm.
Truman, James Kelly, 1960-
Publication Date:


bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1992.
Includes bibliographical references (leaves 247-256).
Statement of Responsibility:
by James Kelly Truman.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001867913
notis - AJU2429
oclc - 28997471
System ID:

Full Text








To my family


I wish to extend boundless gratitude to Dr. Paul

Holloway for being the most patient, challenging,

understanding, supportive, and helpful advisor and friend I

could have imagined. The University of Florida should grant

him a bonus sabbatical year, restricted to long, relaxed

fishing excursions, for having had to work with me for so


I am thankful for the advise, time, and efforts of the

other members of my doctoral committee who include Drs.

Joseph Simmons, Robert DeHoff, and Rolf Hummel of the

Department of Materials Science and Engineering and David

Tanner of the Department of Physics. Their insights and

enlightened critiques have made my education a more enriched


The various staff personnel in the Department of

Materials Science and Engineering who worked with me over

the years were a great help. I would like to acknowledge

Wayne Acree for electron microprobe analysis, Amy Holtzer

for X-ray diffraction work, and Eric Lambers for Auger

analysis and training. I would especially like to thank Joe

Rojo for keeping me connected to UF while I was in Rochester


and Ludie Hampton for being a good secretary and a good


I went through a lot over the years with other graduate

students and made many friends who have unique places in my

memories. I would like to single out my good friends Steve

Wallace and Tom Bussing for being long-time sources of

support, commiseration, entertainment and irreverence.

Further, I would like to thank Carl Mueller for making many

of the sputter targets used in my work and for being my

partner in all aspects of high temperature superconductor

work in our laboratory. I would also like to thank Greg

Lindauer for writing the software which controlled the

resistance versus temperature system and Ed Clausen for

invaluable help when I first delved heavily into personal

computers. Finally, I am grateful for the experience of

having worked and socialized with people from a variety of

cultures, since doing so greatly expanded the views of this

small-town midwesterner.

I have been fortunate that my advisor brought in a

steady stream of visiting scientists and post-doctoral

associates, since I benefited greatly from their presence.

In particular, Dr. Markku Leskela was mainly responsible for

getting me involved in the high temperature

superconductivity field and became a good friend. Also Dr.

Georg Thurner's negative SIMS and Auger analysis work was

invaluable and I really enjoyed our political discussions

and musical shared interests.

I have often found it useful to get on the telephone

and ask questions to those whom I considered to be the best

source of answers. Over the years I especially relied on Dr.

Bill Westwood and Grant Este of Bell-Northern Research and

Mike Wagner of Tektronix. They were always willing to

entertain (and were sometimes entertained by) my questions

on sputter deposition and vacuum equipment.

My summer at IBM Watson provided me a unique

opportunity to work with a true sage in Dr. Norm Braslau.

The technical activities of that summer were rewarding, but

Norm's reflections on being a scientist caused me to

reevaluate my own professional attitudes and the role of the

scientist in today's world.

My employer during the past two years, CVC Products,

has been very supportive while I chipped away at my

dissertation among the trivial activities of work, marriage,

and parenting. My boss, Paul Ballentine, and the company's

president, Christine Whitman, provided a good work

environment and steady encouragement. Also, Dr. Alan Kadin

of the University of Rochester essentially assumed the role

of my local advisor, and his persistent pressure on me to

finish my dissertation was a needed motivation. I am

grateful for all of his assistance and the friendship we


Finally, my family has been unbelievably patient

throughout my seemingly never-ending doctoral program. I owe

my wife Cristina a million back rubs and my daughter Isabel

a house full of kittens for all they have had to sacrifice

for the past few years (although those long holidays on the

beach with Grandma while I was writing could not have been

too bad). Their love and support were invaluable. I am

especially grateful for the encouragement and support my

mother, Suzanne Higgins, has provided throughout my entire

education and particularly while I finished my dissertation.

I never would have done any of this if it were not for her.


ACKNOWLEDGEMENTS ......................................

ABSTRACT ..............................................


1 INTRODUCTION.................................

2 LITERATURE REVIEW...........................

Sputter deposition of YBa2Cui307- Thin
Principles of Sputter Deposition.......
Deviation of the Compositions of
Sputter Target and Sputtered
Processing Variations for YBa2Cu307-,
Thin Films............................
In Situ Growth of YBa2Cu3O7 .............
Explanations for Deviation of Y-Ba-Cu-O
Film and Target Compositions.........
Run-to-Run Irreproducibility of Film
Summary ................................

Barrier Layers for the Growth of YBa2Cu307-,
on Practical Substrates...................

Directions for Research......................

3 EXPERIMENTAL PROCEDURES.....................

Identification of Negative Ions by SIMS.....

AES Study of Long Time Sputtering Effects
on YBa2Cu3O7-x Targets......................

Processing of Y-Ba-Cu-O Thin Films..........

















Sputter Target Fabrication.................. 69
RF Sputter Deposition of Y-Ba-Cu-O and
Barrier Layer Thin Films................... 71
Post-deposition Heat Treatment............... 76
Physical Characterization of Thin Films..... 78
Electrical Characterization of the
Y-Ba-Cu-O Films........................... 81

Y-Ba-Cu-O.................................. 90

SIMS Study of Negative Ions Sputtered
from Y-Ba-Cu-O ............................ 91
Effect of Oxygen Addition.............. 97
Temperature Dependence of 0- Yield..... 98
Discussion of Negative Ion Data........ 100
Summary of SIMS Negative Ion Study..... 118

AES Depth Profiling Study for Ba
Migration................................. 120

5 Y-Ba-Cu-O THIN FILM PROCESSING.............. 124

Towards a 1:2:3 Film Composition by RF
Sputter Deposition from a Single
Y-Ba-Cu-O Target .......................... 125
Experimental............................ 125
Sputter Target Preparation and
Condition............................ 129
Composition of Y-Ba-Cu-O Films......... 132
Discussion of Y-Ba-Cu-O Film
Composition Data..................... 146
Summary of Sputtered Y-Ba-Cu-O Film
Composition Data..................... 159

Towards Superconducting Y-Ba-Cu-O Films..... 160
Heat Treatments to Form Superconducting
Y-Ba-Cu-O ............................ 161
Properties of Superconducting Y-Ba-Cu-O
Films................................. 162
Discussion of Properties of
Superconducting Y-Ba-Cu-O Films...... 174
Summary ................................ 180


SUBSTRATES ................................. 182

Sample Fabrication.......................... 183

Results and Discussion...................... 185
Y-Ba-Cu-O on Si or Sapphire Without a
Barrier Layer........................ 185
SrTiO3 Films on Si and Sapphire
Substrates............................ 188
Y-Ba-Cu-O/SrTiO3/Sapphire.............. 195
Y-Ba-Cu-O/SrTiO3/Si.................... 198

Summary ..................................... 203

7 SUGGESTIONS FOR FUTURE STUDY................ 204

8 SUMMARY AND CONCLUSIONS..................... 210


SYSTEM....................................... 214

General Description......................... 214

Description of Equipment.................... 217
Vacuum ................................... 217
Gas Flow and Control................... 220
Sputter Sources and Power Supplies..... 221
Baffles and Shields..................... 224
Substrate Mounting and Positioning..... 226

Operation................................... 230
Cold Start-Up.......................... 230
Regular Start-Up....................... 231
Target Mounting......................... 232
Sample Loading........................... 233
Pump-Down............................... 234
Presputtering and Deposition........... 235
Sample Removal...... ................... 237

Shut Down ............................. 238

MEASUREMENT SYSTEM......................... 240

Description of System....................... 240

Operating Procedure......................... 244

REFERENCES................................ ............ 247

BIOGRAPHICAL SKETCH.................................. 257

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy





Chairman: Paul H. Holloway
Major Department: Materials Science and Engineering

The commercial application of superconducting YBa2Cu307x

thin films requires the development of deposition methods

which can be used to reproducibly deposit films with good

superconducting properties on insulating and semiconducting

substrates. Sputter deposition is the most popular method to

fabricate Y-Ba-Cu-O superconductor thin films, but when used

in the standard configuration suffers from a deviation

between the compositions of the Y-Ba-Cu-O sputter target and

deposited films, which is thought to be primarily due to

resputtering of the film by negative ions sputtered from the

target. In this study, the negative ions were explicitly

identified and were found to consist of predominantly 0-.

The sputter yield of 0- was found to depend on the Ba

compound used in the fabrication of Y-Ba-Cu-O targets and

was related to the electronegativity difference between the

components. An unreacted mixture of Y203, CuO, and BaF2 was

found to have the lowest O- yield among targets with Y:Ba:Cu

= 1:2:3. The high yield of 0- from YBa2Cu3v07.was found to

depend on the target temperature and be due to the excess

oxygen present. The SIMS negative ion data supported the

composition data for sputter-deposited Y-Ba-Cu-O films.

Targets using BaF2 were found to improve the Ba deficiency,

the run-to-run irreproducibility and the nonuniformity of

the film composition typically found in sputtered Y-Ba-Cu-O

films. Superconducting Y-Ba-Cu-O films were formed on SrTiO3

substrates by post-deposition heat treatment of Y-Ba-Cu-O-F

films in humid oxygen.

The growth of superconducting YBa2Cu3O_7- thin films on

common substrates such as sapphire or silicon requires the

use of a barrier layer to prevent the deleterious

interaction which occurs between Y-Ba-Cu-O films and these

substrates. Barrier layers of SrTiO3 were studied and found

to exhibit textured growth with a preferred (111)

orientation on (100) Si substrates. However, SrTiO3 was

found to be unsuitable as a barrier layer for the growth of

YBa2Cu307-x on Si since Ba reacted with the Si after migrating

through the SrTi03 layer. For sapphire, no textured growth

of SrTiO3 was observed but it was found to be a suitable

barrier layer since it prevented any interaction between

Y-Ba-Cu-O films and sapphire substrates.



The discovery of high critical temperature (Tc)

La-Ba-Cu-O superconductors by Bednorz and Mueller [1] and of

the YBa2Cu3Ov7. compound with a Tc of 90 K by Wu et al. [2]

heralded the onset of an unprecedented amount of research

activity in a single area. High temperature

superconductivity became one of the rare scientific subjects

which attracted the attention of the general public and

garnered coverage by the popular press. The promise of

economical levitating trains and lossless power transmission

seemed to have moved beyond the realm of science fiction.

Even proven superconductor applications such as magnet wire

and magnetic detectors could clearly benefit from the cost

reductions of operation in liquid nitrogen, rather than in

liquid helium as required by conventional superconductors.

Furthermore, the integration of high temperature

superconductors with conventional semiconductor technology

could have important consequences for microelectronics.

Passive superconductor applications such as interconnects

offer the possibility of low-loss signal transmission, while

active devices such as Josephson junctions offer the

possibility of fast switching speeds with low heat

generation [3]. Because of its predominance in the

semiconductor industry, Si is the primary candidate for

these integrated applications with superconductors.

However, the processing of high Tc materials with the

desired chemical, mechanical, and superconducting properties

has proven to be a formidable task, even with most efforts

concentrating on one material, YBa2Cu307_x. The processing of

high quality bulk YBa2Cu30.x has shown progress but the

quality of the material is still not suitable for most

applications. The limitations are due to brittleness, low

critical current density (Jc) values, and instability in

ambient [4]. On the other hand, applications involving

superconducting thin films are in the early stages of

commercialization and the processing of thin films is now

capable of producing extremely high quality material. In

particular, thin films with good microwave properties, Tc 2

90 K and JC values in excess of 106 A/cm2 at 77 K, are now

commonly reported [5]. However, the processing of YBa2Cu3O.7

thin films is still not without its problems and a good deal

of work remains to be'done.

During the first few years after the discovery of high

temperature superconductors, most processes for the

fabrication of superconducting YBa2Cu3_0x thin films

consisted of depositing a film with a Y:Ba:Cu ratio of

1:2:3, heating the film to 800 900 'C in oxygen to obtain

the tetragonal YBa2Cu3O x crystal structure, and slowly

cooling the film in oxygen to increase the oxygen

concentration and form orthorhombic, superconducting

YBa2Cu3O7. [5, 6]. Since 1989 most efforts have concentrated

on growing superconducting YBa2Cu307- films in situ by

depositing at substrate temperatures of 600 800 'C to form

tetragonal YBa2Cu3O7-x, then backfilling the deposition

chamber with oxygen and slowly cooling to form orthorhombic

YBa2Cu307x. This in situ deposition avoids the high

temperatures of a post-deposition heat treatment and

produces films with properties better suited for device

applications [7].

The crystal structure of the superconducting YBa2Cu3O-x

phase is orthorhombic with 0 S x < 0.7, which is illustrated

in Figure 1-la [8]. When the temperature is raised and the

oxygen stoichiometry is less than approximately 6.3 [9], the

material becomes non-superconducting and transforms to the

tetragonal YBa2Cu3O-x phase illustrated in Figure 1-lb [8].

In Figure 1-lb a composition of YBa2Cu3O, is illustrated,

which is completely depleted of excess oxygen. During slow

cooling in oxygen to form the superconducting, orthorhombic

YBa2Cu3O7x phase, oxygen is incorporated into the structure

between the Cu atoms along the b-axis, resulting in the

formation of Cu-O chains as indicated in Figure 1-la. The

incorporation or removal of this excess oxygen occurs

reversibly with thermal cycling [9]. In spite of an

incredible amount of effort, no theory has been developed to

Figure 1-1.








Structure of the two phases in the YBa2Cu3Ov7
family. Orthorhombic YBa2Cu3O07v, illustrated
in a), is superconducting, whereas
tetragonal YBa2CuO36, illustrated in b), is
insulating. After reference [8].

suitably explain high temperature superconductivity in

YBa2Cu30Ox. Regardless, the incorporation of excess oxygen

and the transformation to an orthorhombic crystal structure
are necessary conditions for superconductivity to occur.


Throughout this dissertation, the following convention

will be adopted to distinguish between orthorhombic and

tetragonal YBa2Cu307-: the use of YBa2Cu307- implies the

orthorhombic superconducting phase, whereas the tetragonal,

insulating phase will be designated YBa2Cu306, even though

the tetragonal phase may actually be YBa2Cu306,y, as discussed


The desired superconducting properties of YBa2Cu3O7-x

films are a sharp resistive transition with a Tc near 90 K

and a Jc of 106 A/cm2 or greater. To achieve these

properties, the films must have the correct metallic

stoichiometry Y:Ba:Cu of 1:2:3 and consist of only the

orthorhombic YBa2Cu3O07. structure and no impurity phases.

Further, the critical current in superconducting YBa2Cu307-

is anisotropic and depends on the crystalline quality of the

material. Specifically, the critical current is higher along

the a-b planes than in the direction of the c axis [7]. Thus

for most applications it is desirable to grow thin films

with the c-axis oriented normal to the plane of the film.

The elimination of grain boundaries and other crystalline

defects also increases Jc [4, 10]. Hence a single crystal,

c-axis oriented, epitaxial YBa2Cu3O-x film is desired. This

dictates the lattice parameter of materials to be used as

substrates for YBa2Cu3Ovx films.

Since lattice matching to YBa2Cu307.x is one of the

primary criteria in the choice of a substrate material,

SrTiO3 and LaA103 have been the most popular substrates for

the growth of high quality YBa2Cu307- films [5, 11]. Although

these substrates are useful for certain applications, the

large-scale commercialization of YBa2CuO307 will require

applications utilizing more common, low-cost substrates.

However, the temperatures of post-deposition heating are

sufficiently high to promote harmful interactions between

YBa2Cu307-x and some substrates. Unfortunately this is true

for substrates such as sapphire (a-AO120) and the

semiconductor Si, which react with YBa2Cu3O7- to such an

extent as to destroy superconductivity [5]. To permit the

growth of YBa2Cu307-x on Si or sapphire, an idea which has

received much study is the insertion of an intermediate

barrier layer to prevent the reaction between the film and

the substrate. The lower temperatures used for in situ

processing will limit but not stop the reaction between Si

and YBa2Cu3Ov7,. Furthermore, very little is known about the

effects of the reaction on the semiconducting properties of

Si or on Si devices. Thus a barrier layer will probably be

needed even for in situ YBa2Cu3O7-x film growth on Si or


Many different thin film growth techniques have been

employed in the processing of YBa2Cu3Ov7 thin films with

varying degrees of success in satisfying the demands

discussed above. Sputter deposition has been the most

commonly used technique, but laser ablation, and to a much

lesser extent electron beam evaporation, are also commonly

used. Sputter deposition potentially offers the benefit of

being able to grow stoichiometric films from a single

multicomponent target. Electron beam evaporation requires

the simultaneous control of three different sources and has

been found to be impractical beyond research applications.

Laser ablation can produce the YBa2Cu3x07 films of the same

high quality as sputter deposition, while more easily

obtaining the correct film composition and having

potentially higher deposition rates [12]. However, films

grown by laser ablation suffer from poor thickness

uniformity and poor surface smoothness over large areas.

Comparatively, sputter deposition can produce high quality

films at moderate deposition rates with good thickness

uniformity and excellent surface smoothness on substrates

presently as large as three inches in diameter [13]. Thus

sputtering appears to be the best-suited physical deposition

technique for manufacturing-scale production of YBa2Cu307-

films. However, for reasons to be discussed below,

consistently obtaining the desired Y-Ba-Cu-O film

stoichiometry by sputter deposition is not as simple as is

usually the case for the sputter deposition of other

multicomponent thin films.

The goal of the present work was to address two issues

which could lead to improved processing and

manufacturability of YBa2Cu3O7- thin films grown by sputter

deposition. First, a study was performed to better

understand the mechanics of the sputter deposition of

YBa2Cu3O_7- thin films from single multicomponent oxide

targets. As will be discussed in more detail below, the

reproducible deposition of Y-Ba-Cu-O thin films with a 123

composition has been a major problem encountered in the

sputter deposition of Y-Ba-Cu-O. However, very few

diagnostic studies have been done to explicitly identify the

reasons) for this lack of control. Further definition of

these reasons was an objective of this study. The results

would serve as a guide for improving control of the film

composition. Second, as discussed above, the growth of

YBa2Cu3O7- on practical substrates will require the use of

barrier layers. This is also true for the growth of

YBa2Cu3Ov7x on the most common single crystal oxide substrate,

sapphire. An investigation of one possible barrier layer

material was performed in this study.

In the balance of this dissertation, a review of the

literature is presented in Chapter 2 which emphasizes the

sputter deposition of YBa2Cu307- thin films and the issues

associated with this method. Possible explanations for the

deviation of film and target compositions are discussed as

well as methods to obtain the desired film composition. The

requirements for barrier layers on Si for YBa2Cu307-x are then

presented along with a discussion of various barrier layer

materials reported in the literature.

In Chapter 3 the experimental methods used in this

study are described. Additional information on the sputter

deposition system used in this study and on the resistivity

versus temperature measurement apparatus are presented in

Appendix A and Appendix B, respectively.

Presented in Chapter 4 are the results and discussion

of studies of two possible mechanisms for the deviation of

the composition of Y-Ba-Cu-O thin films and targets observed

in sputter deposition. These mechanisms are the resputtering

of the growing film by negative ions and the migration of

atoms to the target surface during sputtering. First, the

results of a secondary ion mass spectroscopy (SIMS) study

show that the dominant negative ion sputtered from Y-Ba-Cu-O

targets is O-. The proper choice of starting compounds in

the sputter target will be shown to influence the 0- yield.

A discussion on the formation of negative ions is presented.

Secondly, the results of an Auger electron spectroscopy

(AES) study of the surface composition of YBa2Cu3O7- sputter

targets as a function of sputter time and substrate

temperature are presented. The migration of target

components, specifically Ba, to the surface during

sputtering is found not to occur.

Chapter 5 contains a description of efforts to deposit

Y-Ba-Cu-O films with a 123 composition by rf planar

magnetron sputter deposition and to form superconducting

YBa2Cu3,07 by post-deposition heat treatment. Electron

microprobe analysis showed that films sputter deposited from

targets using BaO2 or BaCO3 precursors were Ba deficient and

exhibited very poor run-to-run reproducibility of the film

composition. The use of sputter targets containing unreacted

mixtures of Y203, BaF2, and CuO was found to improve the Ba

deficiency in Y-Ba-Cu-O films, the run-to-run

reproducibility in the film composition, and the uniformity

of the film composition across two inch diameter substrates.

As suggested by the results of the negative SIMS study in

Chapter 4, the use of BaF2 in unreacted targets reduced the

Ba deficiency problem as result of a decreased 0- yield.

Finally, the properties of superconducting YBa2Cu3O.7x films

on SrTiO3 substrates are described.

In Chapter 6 is presented a study of the suitability of

SrTiO3 barrier layers for preventing interaction between

Y-Ba-Cu-O films and substrates of Si or sapphire. The SrTi03

layers were found to exhibit textured growth with a (111)

preferred orientation on the (100) Si substrates. However,

SrTiO3 was not a suitable barrier layer for Si since Ba

still reacted with the Si after migrating through the SrTiO3

layer. For sapphire, SrTiO3 was found to be a suitable

barrier layer since it prevented any interaction between the

Y-Ba-Cu-O film and the sapphire substrate.

Chapter 7 contains suggestions for future work based on

the review of the literature and the questions raised by the

experiments in this effort.


Finally, in Chapter 8 is presented a summary of the

results from the previous chapters.


Presented in this chapter is a review of published

literature relating to the sputter deposition of YBa2Cu3O7_x

thin films and the growth of YBa2Cu3O7x on barrier layers on

practical substrates, particularly Si. In addition, the

reactions between the superconductor thin films and these

substrates are reviewed. The number of publications since

1987 in the area of YBa2Cu30-,_ thin film fabrication and

reactions, including journal articles and conference

proceedings, is easily in excess of one thousand. Thus a

comprehensive review of every paper is simply impossible.

Hence, this review is general in scope, but focuses on

relevant important publications.

2.1 Sputter Deposition of YBa.Cu OQ Thin Films

Sputter deposition is typically the method of choice

when depositing multicomponent thin films since the desired

film composition can usually be obtained from a single

target with the same composition. For targets containing

compounds in which one of the elements is highly volatile,

such as oxygen in oxides, the film is usually deficient in

the volatile element unless special steps are taken.

However, the ratio of metallic species in the film usually

equals that in the target for steady state conditions [14].

Certain exceptions arise, as discussed below, when even the

ratio of metallic species in the film does not equal that in

the target. This problem has been the bane of investigators

studying sputter deposition of YBa2Cu3O07x thin films.

Generally, films deposited from Y-Ba-Cu-O sputter targets

(where Y-Ba-Cu-O represents any combination of Y, Ba, Cu,

and 0) onto unheated substrates are deficient in Ba relative

to the target, with Cu deficiencies sometimes being

reported. Furthermore, the reproducibility of film

compositions between successive depositions, even with all

deposition parameters kept constant, has been marginal at

best unless special precautions are taken.

The Y:Ba:Cu ratio needed in a film to get a

superconducting transition is not strictly 1:2:3, since a

film made of less than 50% YBa2Cu307x can have a continuous

superconducting network due to percolation effects [15].

However, the closer to a uniform 1:2:3 stoichiometry, the

more desirable are the properties of the film, including a

sharper resistive transition and a higher critical current

density. Furthermore, when deposited on an unheated

substrate, sputtered Y-Ba-Cu-0 films are insulating and

require post-deposition heat treatment at 800 900 'C in

oxygen to form the superconducting YBa2Cu307.x phase [6]. Many

reports during the past two years have emphasized the in

situ growth of superconducting YBa2Cu307_x during which the

substrate temperature is elevated during deposition to 600 -

800 "C [7]. The heating of substrates has also been found to

affect the film composition. Hence, much of the work in the

sputter deposition of YBa2Cu307- has centered on controllably

obtaining a 123 film composition in as-deposited thin films.

A brief overview of the basic principles of sputter

deposition will be given. Four general explanations for the

deviation of the sputtered film compositions from that of

the sputter target will be outlined. These will provide the

background necessary for understanding the summary of the

many processing variations presented in the literature in

attempts to grow stoichiometric YBa2Cu307- thin films.

2.1.1 Principles of Sputter Deposition

The basic principles of sputter deposition are best

understood by examining the simplest configuration, the

planar diode, which is illustrated in Figure 2-1 [16].

Sputtering is a process in which energetic particles strike

the surface of a material and eject particles from the

material's surface by a momentum exchange process. For film

deposition by sputtering, the ejected particles M are

collected on a substrate. A heavy inert gas, usually Ar, is

backfilled into a vacuum chamber to a pressure of 1 to 100

mTorr. A power supply maintains a potential between cathode

and anode electrodes in order to sustain an Ar plasma. The

plasma provides the Ar' ions which sputter the target. The




M Ar+ M

target (cathode)

Figure 2-1.

Schematic illustration of planar diode
sputter deposition. Ar+ ions are
accelerated to the negatively charged
target. Target particles M are ejected and
travel to the substrate, thereby forming a
film. After ref. [16].

target, which is the source of material to be deposited, is
the cathode and is negatively biased to accelerate Ar' ions
from the plasma so that its surface is sputtered by the
ions. A conducting target can be biased using a DC
potential, and the target can double as the cathode. A
nonconducting target requires the use of a RF potential and
the target is usually attached to a conducting cathode
electrode. Material sputtered from the target, M in Figure
2-1, collects on a substrate, which is usually positioned
beyond an opening in the anode. The sputtered species are
primarily low energy atoms, although energetic neutral atoms



anode 1

molecules as well as low and high energy positive and

negative ions are also observed [17]. Also created in the

sputtering process are the secondary electrons necessary to

sustain the plasma. The deposition rates from diodes are

generally low. Also, bombardment of the substrate by

electrons and the plasma leads to undesirable heating when a

diode source is used.

Limitations of the planar diode have led to the

development and widespread use of magnetron sputter cathodes

[18]. A magnetron is essentially a diode apparatus with

specially configured magnets placed beneath the cathode

electrode. The resulting magnetic field, combined with the

electric field from the applied potential on the cathode,

confines the electrons in the plasma to a net motion in a

closed loop above the cathode. The most common magnetron

configuration is the circular magnetron, for which the

target is a flat disk. The magnetic field and resulting
-4 -4
E x B electron drift path for a circular magnetron are

illustrated in Figure 2-2 [19]. Since most of the electrons

in the plasma are confined to motion in an annular path, the

probability of collisions with gas atoms is much greater and

the plasma is much denser within this region. This results

in the brightly glowing ring which can be seen in the plasma

above a circular magnetron cathode and also in a greatly

increased Ar' density in this region. Thus an increased flux

of Ar' is incident on the target and an increased sputter



Figure 2-2. Schematic of circular planar magnetron
cathode, illustrating the shape of the
magnetic field and the drift path of the
electrons resulting from the combined
magnetic and electric fields. After
reference [19].

rate results beneath the ring in the plasma. The sputtering
pattern in the magnetron results in the formation of a
circular groove in the target, which is commonly called the
racetrack. Finally, the sputter rate in the central area of
the target can be low enough that a net redeposition from
the racetrack region can occur [20].
Compared to a diode source, the confined motion of
electrons in a magnetron results in the formation of a far
greater number of Ar' ions for a given pressure and
potential. Thus, magnetrons can give an order of magnitude
or more increase in the deposition rate [18]. Also, the
confinement of the electrons greatly reduces the substrate
heating due to electron and plasma bombardment.

2.1.2 Deviation of the Compositions of Sputter Target and
Sputtered Films

As pointed out above, one of the main advantages of

sputtering is that the composition of a film is usually the

same as that of the target. The reason is the formation of

an altered layer on the target surface which is depleted and

enriched in the high and low sputtering rate constituentss,

respectively [17, 21]. After a stabilization period of

typically a few minutes (i.e. sputter depths of about 100 A)

during which the altered layer is formed, the sputtered flux

usually has the same composition as the target. To explain

this, a few concepts will be briefly discussed. The incident

flux, FAr, is the number of Ar' ions per unit time per unit

area striking the target. (It is assumed that only Ar is

being used as the sputter gas). The total flux of component

i sputtered from the target by the incident Ar* ions

is represented by FAr. The total sputter yield, Y,, is

defined as the total flux of particles sputtered from the

target per incident ion [22]. For a single component target

comprised of element M, the total sputter yield equals the

yield of M, or
Y, = FM (2.1)

For a multicomponent target, it is typically assumed the

sputter yields for each component are equal to those

determined from elemental targets for sputtering with Ar' of

a specified energy and angle of incidence [22]. To account

for different concentrations of each component in a

multicomponent target, the fractional concentration of

component i in the surface region of the target, ci, is

included in the expression for the flux of i sputtered from

a multicomponent target [22],

FAr = FArYArci. (2.2)

To understand the formation of the altered layer during

sputtering, consider a target made of an alloy AB with a

surface composition initially equal to that of the bulk. Let

the sputter yield of component A, Y,, be twice that of

component B, Y,, for sputtering with Ar' of a specified

energy and angle of incidence. The initial sputtered flux

will be rich in component A and the target surface will

become depleted in A. Eventually a steady state condition

will be reached in which the surface concentration of A will

become half the fraction of A in the bulk with a concomitant

rise in the surface concentration of B. In other words,
yAr b
CA B A (2.3)
c yAr Cb
where cb and cb are the bulk concentrations of A and B,

respectively [21]. The region near the surface is the

desired altered layer. Thus after an initial transient

period, the steady state sputtered flux and the deposited

thin film will have the same composition as the target bulk.

In general, the composition of sputter deposited films

equals that of the multicomponent target. However, there are

four main mechanisms, to be discussed in more detail below,

which can prevent the sputtered film composition from being

equal to that of the target. The first mechanism involves

migration of one or more of the target species) to the

target surface during sputtering independent of sputter rate

[21]; this will be discussed in Section The second

mechanism entails differing non-unity sticking coefficients

of the components on the substrate [21]. The sticking

coefficient, Si, is defined as the ratio of the number of

particles of specie i incident on a surface to the number of

particles of specie i which stick to the surface, or as the

probability that a single incident particle will stick to a

surface [23]. The importance of this will be discussed in

Section The third mechanism involves a variation in

the angular distribution of species sputtered from a

multicomponent target [24]. The angular distribution of a

sputtered flux usually follows a cosine distribution.

However, for some multicomponent targets, the sputtered

angular distributions of some species) may not follow a

simple cosine distribution, as will be discussed below in

Section The fourth and final mechanism involves an

unexpectedly high sputter yield of negatively charged ions,

which is the number of negative ions sputtered per incident

Ar* ion [25]. This issue will be discussed in Section Any combination of these mechanisms may be present,

but the exact contribution of each is difficult to predict. Migration of species to the target surface

The typical power dissipation during sputtering is

several hundred watts. As a result, sputter targets are kept

cool by water which flows across the back side of the

cathode. To effectively cool, the target must be in good

thermal contact with water or the conducting cathode on

which it rests. However, the surface of poor thermal

conductors (e.g. most insulators) may still get hot even

though the bottom of the target is adequately cooled. Even

metal targets with high thermal conductivity can reach

temperatures of 200 300 'C when sputtering at a high rate.

During the initial stages of sputtering of a multicomponent

target, the target surface becomes depleted in the component

with the highest sputter yield, as discussed above. The

formation of an altered layer, with the composition

necessary to produce a sputtered flux with the same

composition as the target bulk, would require that the

surface remain depleted in this component. The depletion of

a component at the target surface represents a driving force

for the migration of atoms from deeper in the target to the

surface. In an adequately cooled target, the rate of

migration of the depleted component is low compared to the

sputter rate, and the altered layer maintains the correct

stoichiometry. However, in a hot target, a mobile component

may migrate to the target surface at a rate faster than the

sputter rate can keep.the surface depleted in the component.

This will prevent the formation of an altered layer with the

desired composition and the sputtered flux would not have

the same composition as the bulk. Rather, the sputtered flux

and the resulting film composition would be rich in the

mobile component [21]. Sticking coefficient variations

Non-unity sticking coefficients can result in the

deficiency of a component in the deposited film. A highly

volatile element in a target, such as oxygen, will commonly

be deficient in deposited films. The oxygen which is not

sputtered from the target as an oxide molecule or which does

not rapidly react with a deposited electropositive element

often does not stick to the substrate and is pumped away

[26]. When substrates are at room temperature, it is usually

safe to assume that the sticking coefficient of all

nonvolatile components sputtered from a target, i.e. most

cationic species, is approximately equal to unity [27].

However, the substrate temperature, even for a planar

magnetron apparatus, can easily reach 200 'C due to

bombardment by electrons and reflected excited state neutral

Ar [28]. At this temperature the sticking coefficients for

high vapor pressure components such as Pb may be less than

one, resulting in a film composition not equal to that of

the target. Anaular distribution variations

A variation in the angular distribution of species

sputtered from a multicomponent target can result in

unexpected deviations and non-uniformities in film

composition. For normal incidence ions, the flux of

sputtered atoms from a single component target follows a

cosine distribution centered about the normal to the target

surface [22]. However, for a binary alloy target,

significant differences between the angular distributions of

the components have been reported [24]. The difference in

angular distributions'is more pronounced at lower ion

energies ( < 1 keV ) and for greater differences in the mass

of the components [29]. Specifically, the lower mass

component is preferentially sputtered along the target

surface normal and the heavier component is sputtered along

more oblique angles. For a two component target, there must

exist an angle at which a substrate can be positioned which

will result in a film with the same composition as the

target. However, this is not necessarily the case for an

alloy target with three or more components, particularly

when their masses differ significantly. For most compound

targets, the angular distribution cannot always be predicted

from the mass difference of the components, as was reported

for WSi2.3 [29]. Resputterina by negative ions

The species sputtered from a target by Ar* are mostly

atoms, but can also include high energy excited state

neutral atoms as well as molecular clusters and positive and

negative ions. Also, molecules are often sputtered from

compound targets [26]. Positively charged particles are

attracted to the target because of its large negative

potential. Conversely, negative ions are accelerated away

from the target towards the substrate due to repulsion by

the large negative potential, which is the same potential

which attracts the Ar' ions. The negative ions then enter

the plasma where they may collide with electrons, other

ions, atoms, or excited species and lose their additional

electron, depending on the mean free path for collision.

This mean free path in turn depends on the target potential

and the pressure [30]. The collision with electrons is most

likely to remove the extra electron from a negative ion. The

energetic, now neutral atom has a small collision

cross-section and can travel through the plasma to the

substrate, with the number of energy-reducing collisions

again depending on the mean free path [30]. For mean free

paths on the order of the target-to-substrate distance,

which are possible at low operating pressures or high target

potentials, energetic negative ions can pass through the

plasma without collision and bombard the substrate.

The yield of negative ions during sputtering is usually

less than 1% [26] and their effects are generally not

critical. However, large yields of negative ions have been

observed from certain compounds, particularly those which

are highly ionic and contain a strongly electronegative

element such as F or O bonded to a strongly electropositive

element. In such cases, the negative ions can bombard the

substrate with a flux comparable to the deposition rate of

the film and thus greatly resputter the growing film or even

cause etching of the substrate. Hanak and Pellicane [31]

were the first to identify the cause of such effects during

the sputtering of TbF3. as the negative ion F-. An additional

problem with negative ions is that when the target contains

more than two elements, the resputtering of the growing film

by negative ions can result in the depletion of one (or

more) of the elements. For example, films deposited from

BaPb1_-BixO3 targets were found to be strongly deficient in Ba

[32]. Resputtering by negative ions was conjectured to be

the cause, although no further verification was reported.

The only model predicting the degree of film

resputtering by negative ions was published by Cuomo et al.

[25]. During the sputter deposition of films from targets

consisting of selected compounds of gold and a rare earth

(RE) metal such as Sm, the deposition rate on the substrate

ranged from unexpectedly low to a negative rate, i.e.,

substrate etching had occurred. Analysis by SIMS found that

the Au- negative ion yield from SmAu was four orders of

magnitude greater than that from Au. It was concluded that

the unexpectedly low deposition rates were due to

resputtering of the depositing film by negative ions formed

during sputtering of the Au-RE targets.

Cuomo et al. developed a model which attributed the

formation of Au- to the high electron affinity of Au in

combination with the low ionization potential of a given

rare earth metal. In generalizing the model, the target

material was simplistically treated as a simple ionic

compound AB, in which negative B ions were present in the

solid. Negative ions were suggested to be sputtered from AB

because of the high electron affinity of the B atom, such

that sputtered B retained the negative charge it possessed

in the lattice. During the formation of the solid, the

energy required to remove an electron from the A atom was

assumed to be equal to the ionization potential of A, I,.

Forming the negative ion B- resulted in a gain of energy

equal to the electron affinity of B, EA,. Thus the total

energy difference IA-EAB was suggested to be a measure of

the difficulty in transferring an electron from A to B

during the formation of the ionic solid. Among a series of

compounds, comparing the values of I,-EAB was suggested to

provide an estimate for the ease of charge transfer in AB

and the likelihood of removing negative ions by sputtering.

Conceptually, the explanation of Cuomo et al. suggested

how a negative ion could be created during sputtering of a

target material. However, this model did not provide a

realistic picture since no solid, including Au-RE

compounds, is perfectly ionic with complete localization of

charge. Further, although the model suggested how negative

ions may be sputtered from the target, it was not able to

predict a threshold voltage for the formation of negative

ions. Rather, the model was used to empirically predict the

degree of film resputtering during deposition from a variety

of Au-RE target materials. Specifically, it was found that

if the value of IA-EA, was less than 3.4 eV for Au-RE

compounds, where component A was the rare earth element and

component B was Au, sputter deposition with Ar at an

accelerating voltage of -1000 V would result in substrate

etching by Au- ions. However, the rate of resputtering by

negative ions was found to be low enough in some cases that

substrate etching did not occur but the net deposition rate

decreased. No attempt was made to explain this. Also, for a

given value of IA-EA,, the degree of resputtering was found

to depend strongly on the negative accelerating potential on

the target. Thus the value of I,-EAB alone was not

sufficient to predict when negative ion rates would be

observed or significant. However, this value did provide a

helpful qualitative comparison among similar Au-RE


Kester and Messier [33] compared the predictions of the

model of Cuomo et al. [25] for the formation of negative

ions to qualitative effects of resputtering of a large

number of oxides, including changes in deposition rate and

film morphology. Negative ion resputtering effects were

observed for RF magnetron sputter deposition of BaTiO3 and

other titanate perovskites, zirconate perovskites, niobates

and other oxides. All of the oxide compounds which exhibited

negative ion resputtering had values of IA-EA, greater than

3.4 eV, in contrast to the prediction of the empirical model

of Cuomo et al. Furthermore, for just one material, BaTiO3,

the degree of resputtering varied from none to severe

depending on the RF power, i.e. the accelerating potential

on the target. Thus, in agreement with the results of Cuomo

et al., the value of IA-EAB was not alone sufficient to

predict negative ion effects. However, for the same

deposition conditions and among similar materials (e.g.

zirconates or titanates), the amount of resputtering was

found to increase with a decreasing value of IA-EAB. The

applicability of this binary compound model to predict

sputtering of ternary oxides has been attempted, but the

calculation of a meaningful IA-EAB value remains. It was

assumed that only the I,-EAB value between the element with

the lowest ionization potential and oxygen was important,

whereas negative ions may have also formed between the other

element and oxygen. In summary, it was found that the I,-EA,

values was but one of many factors that influence whether

resputtering by negative ions will occur, and to what


The formation of negative ions has understandably

received considerable attention from the Secondary Ion Mass

Spectroscopy (SIMS) analysis community [22]. In SIMS

analysis, the information sought is the mass-to-charge ratio

versus the intensity of ionic species sputtered from a

material's surface, as will be discussed below in Chapter 3.

In order to apply SIMS in quantitative surface compositional

analyses, much effort has gone into relating the sputtered

ion intensities to the actual surface compositions. However,

most of these efforts are at least partially empirical, and

a first-principles theory relating sputtered ion intensities

to actual surface composition has not been developed.

Semi-quantitative models have been developed to predict the

the sputtered ion yield from metallic materials, for which

the emission of ions is understood to be due to the

tunneling of electrons from the surface of metals to

sputtered particles [22]. On the other hand, only a

qualitative understanding, based on the concept of bond

breaking, exists for the emission of ions during the

sputtering of ionic compounds. In general, it is observed

that the more ionic the character of the bonds in a

compound, the more likely the sputtered species are to

retain their charge from the lattice [22]. An increased

ionic bond character is obtained with an increasing

electronegativity difference between species in the

compound. Thus the more electronegative a specie is in the

compound, the greater.the probability that it will retain

the negative charge in the lattice and be sputtered as a

negative ion [22]. However, this provides no basis for

calculating the fraction of the specie which will be

sputtered as a negative ion.

2.1.3 Processing Variations for YBaCuaQ, Thin Films

As discussed above, the reproducible sputter deposition

of Y-Ba-Cu-O thin films with a 1:2:3 composition from a

YBa2Cu3Ov7 target is very difficult. An additional

complication is that the film composition as well as the

thickness depends on the position above a YBa2Cu3Ov7x sputter

target, as shown in Figure 2-3 [34]. In all such data a

depletion of Ba is seen in the film above the magnetron

racetrack region of the target, which is where the sputter

rate is the highest. Thus the first step in the sputter

deposition of YBa2Cu3O7., is the establishment of processing

conditions for a particular system to obtain a 1:2:3 film

composition. A large variety of sputter configurations and

geometries have been used in the growth of YBa2Cu307- thin

films. Almost every conceivable sputter deposition technique

has been employed and for each of these a large matrix of

deposition parameters have been varied. Also,

unconventional sputtering geometries and different types of


0 1.0

*" 0.5 *








Distance from center of target (cm)

0.0' --k -- -' -
-10 -5 0 5
.Distance from center (cm)

Figure 2-3.

Variation in composition and
Y-Ba-Cu-O film as a function
above sputter target [34].

thickness of
of position

-5 0

targets have been studied. These processing variations will

be briefly reviewed; more detailed discussion is given in a

review by Leskela et al. [35].

All types of sputter deposition techniques, including,

but not limited to, DC planar diode [36] and magnetron [37],

unbalanced DC magnetron [38], RF planar diode [39] and

magnetron [40], RF cylindrical magnetron [41], and ion beam

deposition [42] have been successfully employed in the

growth of YBa2Cu3O7.- thin films. The most commonly used

technique has by far been RF planar magnetron sputtering,

and the reasons for its popularity will be discussed. Any RF

sputtering technique is more versatile for depositing

Y-Ba-Cu-O films than the related DC technique (i.e. diode or

magnetron) since either insulating or conducting targets may

be used, and Y-Ba-Cu-0 targets are conducting only when

reacted to form orthorhombic YBaCu3O7x. Further, among RF

methods, the RF magnetron's magnetic field confines

electrons and reduces the substrate heating which occurs

with RF diodes [41]. For a given power, a magnetron source

also allows lower pressure sputtering as compared to a diode

source. Lower pressure sputtering results in line-of-sight

deposition with higher rates, which is important for

patterning of thin films [43]. For ion beam sputtering the

deposition rates are generally quite low [44]. The RF planar

magnetron also offers the convenience of using

easy-to-fabricate flat, round targets, unlike the targets

required for a cylindrical magnetron.

The most common method for obtaining 1:2:3 film

compositions has been to use RF planar magnetron sputter

deposition from a single oxide target enriched in those

components found to be deficient in the deposited film,

while keeping all other deposition parameters constant

[5, 35]. However, choosing the correct target composition is

not simple. This process is iterative and usually requires

several different target compositions until a 1:2:3 film

composition is obtained. Target compositions reported in

the literature are nearly as numerous as the number of

publications. The addition of excess Ba and Cu are common,

but the amounts vary from system to system.

Many groups found the fabrication of oxide sputter

targets with a variety of compositions to be too

painstaking. Attempts to fabricate a metallic target with a

Y:Ba:Cu ratio equal to 1:2:3 were unsuccessful because

neither an alloy or compound with this stoichiometry could

be formed and sintered targets of mixed Y, Ba, and Cu were

highly unstable in air, typically corroding and crumbling in

a short time [45]. On the other hand, the film composition

was controlled in situ by simultaneous or sequential

deposition from multiple sources containing either oxide or

metallic targets. Target combinations used include YBa2Cu307.x

and Ba or Cu [46], Y2Cu20O and BaCuO2 [47], Y and Ba2Cu3 [48],

and YCu and BaCu [49]. Films were deposited with the

deposition rates from'each source set in the correct ratio

to give a 1:2:3 composition. However, the control of film

composition over reasonably large area (= 1 cm2) using

multiple sources proved to be difficult due to the need to

precisely control deposition rates and geometrical factors.

Further, the composition of the films were still found to

deviate from the values predicted from deposition rates.

Also, the atmospheric stability of metal targets containing

Ba, such as Ba and Ba2Cu3, was found to be extremely poor

[45]. Thus the amount of effort in using multiple sputter

targets has greatly decreased during the past two years.

Sputter deposition parameters which have been varied

systematically include sputter gas pressure and percent of

oxygen in the sputter gas [50], power [51], substrate bias

[52], source-to-substrate distance [53], length of

presputter time [54, 34], and substrate temperature [55,

56]. Substrate temperature has also been controlled for the

purpose of in situ superconducting YBa2Cu307- depositions,

which will be discussed below. The effect of varying each

parameter is difficult to isolate since it can depend on

other parameters as well, but general trends can be

observed. Higher Ar pressures [50] and lower accelerating

voltages [51] give film compositions closer to that of the

target. The addition of oxygen to the Ar sputter gas can

lead to severe deficiencies of Ba and lesser deficiencies of

Cu in films [57]. Increasing the substrate temperature

causes an larger depletion of Cu in films sputter deposited

onto substrates positioned directly above the target [55,

56, 58]. In fact, it was found necessary to keep the

substrate constantly cooled to obtain the same Cu/Y ratio in

the film as in the target [58]. Negatively biasing the

substrate had no effect at higher Ar pressures, but at low

Ar pressure a small negative bias improved the film

composition [59]. On the other hand, too large of a negative

substrate bias caused resputtering of the film by Ar' from

the plasma. Systematic trends due to source-to-substrate

distance are difficult to quantify among different

deposition systems, but film compositions nonetheless are

sensitive to this parameter.

Selinder et al. [54] and Kadin et al. [34] found very

long presputtering times of 30 60 hours, depending on the

target, to be necessary to obtain a 1:2:3 film composition

from a YBa2Cu30,7- target. The long sputtering times were

suggested to be necessary due to the slow loss of oxygen

from the target as it converted from orthorhombic YBa2Cu3O_07

to tetragonal YBa2Cu30O, which is completely depleted of

excess oxygen, as discussed in Chapter 1. No test was

performed to actually observe oxygen leaving the target, nor

was the target analyzed to determine that tetragonal

YBa2Cu3O, had formed.

Unconventional, off-axis sputtering geometries have

been successfully implemented to obtain films with a 1:2:3

composition from a slightly compensated target [60, 61, 62,

63]. The substrates were positioned at an angle to the

substrate normal such that they were not directly above the

target. The off-axis angle depended on the geometry of the

particular deposition chamber and reported values vary from

30 to 90 degrees from the target normal. The film

composition was found to depend on the off-axis angle [63].

The substrates required rotation during deposition to obtain

a uniform composition. The main disadvantage of off-axis

substrate positions has been low deposition rates.

Regardless, since 1990 the use of off-axis substrate

positioning has emerged as the most popular variation of

sputter deposition of YBa2Cu3O7- films. The concern for

compositional reproducibility has overshadowed the drawback

of low deposition rates.

2,1.4 In Situ Growth of YBa4Cu2-Q

As pointed out in Chapter 1, in situ growth of

superconducting YBa2Cu3Ov-x films has received much attention

since 1989, with the number of YBa2Cu3Ox films grown by

post-deposition annealing greatly decreasing. The in situ

formation of YBa2Cu3Ov- films is desirable because of the

lower temperature seen by the substrate and a much smoother

film morphology. The smoother film morphology results from

the growth of either strongly textured films with low angle

grains boundaries or epitaxial, nearly single-crystal films

and is particularly important for microwave applications.

Further, most implementations of in situ sputter deposition

of YBa2Cu307- have also utilized off-axis substrate positions

[64, 65, 66].

The temperatures used in the typical post-deposition

heat treatment, 800 900 'C, are undesirably high for two

main reasons. First, such heat treatments can cause

interdiffusion or reaction of the Y-Ba-Cu-O film and the

substrate, thereby destroying superconductivity and/or

damaging the substrate. Second, many potential applications

of YBa2CuO37-. require its growth on silicon substrates. Any

Si devices already present would be damaged by heating to

such temperatures. For in situ growth of YBa2Cu307,

substrate temperatures of 650 800 'C are typically used

during deposition for crystallization of the perovskite-like

structure [67]. Substrate temperatures as low as 450 'C have

also been reported [68], although the quality of the

YBa2Cu307- films was not very high and post-deposition heat

treatment was still required. This is not surprising since

the crystallization temperature for YBa2Cu3O, is reported to

be in the range 500 600 'C [67].

In addition to elevated substrate temperatures, the in

situ formation of YBa2Cu3Ov7 also requires the addition of

oxygen to the Ar sputter gas, with reported values ranging

from 20 to 50% 02 [64, 65, 66]. The deposition is followed

by slow cooling in the chamber in an oxygen background in

order to increase the oxygen content of the film [64, 65,

66]. Very high quality, epitaxial films have been grown on

substrates including SrTiO3, MgO, and LaA103. Fair quality

YBa2Cu30O7x films have been grown directly on Si, but

interface reactions and lattice mismatch have limited Tc and

Jc values [69, 70].

An additional benefit of the in situ growth of

YBa2Cu3Ov07 films which has been observed is an improved

resistance to environmental degradation [5]. This probably

results from the formation of nearly epitaxial, smooth

YBa2Cu307 films with decreased exposed grain boundary area

[71]. Films of YBa2Cu3Ov7. formed by post-deposition heat

treatment have a rougher film surface morphology and a

greater grain boundary area, which makes them more

susceptible to atmospheric degradation [71], as discussed

in Section 2.1.6. Further, in a conventional post-deposition

annealing process, the as-deposited Y-Ba-Cu-O films can

suffer environmental degradation prior to annealing, which

in turn prevents the complete reaction to form YBa2Cu3Ov7

films after annealing [72]. The resulting impurity phases

can limit Tc and Jc.

Control of the film composition for in situ YBa2Cu3Ox

growth has added complications. During sputter deposition,

the substrate temperature and the oxygen in the sputter gas

synergistically affect the film composition [57, 63]. For a

constant percent oxygen, increasing substrate temperature

leads to a Cu deficiency. Increasing the percent of oxygen

for a given substrate temperature causes a shortage of Ba in

the film. These factors must be accounted for in the

iteratively determined composition of a target to compensate

for the deficient elements.

As discussed above, the majority of YBa2Cu3O7- thin

films, whether deposited by sputtering, laser ablation, or

co-evaporation, are now grown by in situ processes. However,

a process in which films are grown without intentional

substrate heating and then heated after deposition to form

YBa2Cu30o7- would be advantageous for coating non-uniformly

shaped objects or large areas, due to the difficulty of

uniformly heating such substrates in situ. One notable

process exists in which post-deposition annealing is used to

form YBa2Cu307- films with surface morphology, Tc and Jc

equivalent to the best in situ grown films. In this process,

Y, Cu, and BaF2 are co-evaporated from electron-beam sources

and then subjected to a post-deposition heat treatment to

form YBa2Cu3Ov7 [73, 74]. The use of BaF2 as the Ba source

adds the complication of a post-deposition heat treatment in

humid oxygen, which is necessary to reduce the BaF2 by the

reaction [75]

BaF2 + H2O(g) -4 2HF g) + BaO. (2.4)

The resulting BaO then reacts with CuO, Y203 and/or ternary

Y-Cu-O oxides to form YBa2Cu3O7-x. A temperature of 850 "C or

above is required for the reduction of BaF2 and the

formation of YBa2Cu30_x with humid oxygen. However, recent

data show that decreased oxygen partial pressure during

post-deposition annealing (e.g. annealing in mixtures of

humidified 02 and Ar) leads to a lower anneal temperature to

form YBa2Cu307- from Y, Cu, and BaF,. In fact the temperature

can be lowered to values equal to those employed for in situ

film growth (e.g. 730 'C) [74]. Thus the additional

difficulty of in situ processing may not ultimately be

necessary to achieve films of the highest quality. However,

a caveat in the use of BaF2 is its greater toxicity than

BaCO3 or BaO2, of which one must be aware when handling the

materials [76].

2.1.5 Explanations for Deviation of Y-Ba-Cu-O Film and
Target Compositions

Deviations of the compositions of sputter deposited

films from Y-Ba-Cu-O target compositions has been discussed

in the literature in terms of all four of the mechanisms

described in Section 2.1.1, namely, 1) migration of one or

more of the target species) to the target surface during

sputtering, 2) differing sticking coefficients of the

components on the substrate, 3) angular distribution

variation of sputtered species, and 4) unexpectedly high

negative ion yields from sputtering. These four reasons will

be discussed here relative to their effects and probability

during magnetron sputter deposition of Y-Ba-Cu-O

superconductor thin films.
 Migration of Ba in Y-Ba-Cu-0 target

Liou et al. [77] suggested that the Ba deficiency in

sputtered Y-Ba-Cu-O films may be due to the migration of Ba

to the surface of the YBa2Cu3Ov_ target during sputtering.

Very little explanation was given as to why the migration of

Ba in the target would occur, but the point bears further

consideration. Since sintered YBa2Cu3Ov-x or

off-stoichiometric Y-Ba-Cu-O targets are not good thermal

conductors, they could reach temperatures as high as 300 'C

during sputtering. As discussed above in Section,

such higher temperatures could accelerate migration of

components to the target surface, prevent the formation of

an altered layer on the surface with the desired

composition, and result in a sputtered flux with a

composition different from that of the target bulk.

Considering Ba migration in a Y-Ba-Cu-O target, the

sputtered flux and film composition would be rich in Ba

compared to the target. However, sputtered Y-Ba-Cu-O films

have usually been reported to be Ba-deficient [57], whereas

films rich in Ba relative to the target have never been

reported. It is conceivable that the length of a typical

deposition (three or more hours) may be greater than the

time it takes to deplete Ba from the target, which could

account for the Ba deficiency in the films. If this were

true, eventually the flux of Ba migrating to the target

surface would not be able to compensate for the removal of

Ba from the surface by sputtering, and the amount of Ba in

the sputtered flux would decrease. Thus a Ba deficiency in

the film composition would be seen. Further, if Ba continued

to be more and more depleted from the target with additional

sputtering, a time-dependent surface composition could

develop. As a result of this, in successive depositions the

film compositions could differ. A study of the composition

of the surface of Y-Ba-Cu-O sputter targets after extended

periods of sputtering has not been reported, which would be

necessary to test the suggestion of Ba migration of Liou et

al. Experiments to test this suggestion are presented in

Section 4.2. Sticking coefficient variations for Cu

The decrease in Cu concentration in Y-Ba-Cu-O films

with increased substrate temperature discussed above has

been suggested to result from a decreased sticking

coefficient for CuO due only to higher substrate temperature

[78]. This hypothesis appeared to be supported by the report

of Argana et al. [58] 'of the need to cool the substrate in

order to obtain the same Cu/Y ratio in RF planar magnetron

sputter-deposited films as in the target. Authors supporting

the lower sticking coefficient argument appeared to

implicitly assume that CuOx was being sputtered, which could

have had a lower coefficient. This assumption ignored the

fact that metallic Cu could have been sputtered, with CuOx

reformed by surface reaction on the film. In addition, the

deficiency of Cu should not be strictly a thermal sticking

coefficient effect, since no such deficiency has been

reported for YBa2Cu30-7. deposited in situ by laser ablation,

even though the substrate temperature has been taken as high

as 800 'C [5, 6, 35]. Thus the decrease in Cu concentration

with increasing substrate temperature must also have been

related to the sputtering process itself. Furthermore, the

authors do not suggest whether CuO should be sputtered as a

molecule or form on the substrate. If metallic Cu was

sputtered, then a thermal sticking coefficient argument

would not be reasonable since the coefficient for Cu would

be expected to unity at temperatures up to 500 *C [79].

An alternative explanation for the decrease in Cu

concentration with increasing substrate temperature for

sputtered Y-Ba-Cu-O thin films is suggested by the optical

emission data of Fledderman [80]. Optical emission

spectroscopy was used to study the excited-state atoms and

molecules created during the ion beam sputtering of

YBa2Cu3Ov-x targets. Species present included Y, Ba, Ba*, Cu,

YO, BaO, and CuO. The intensity of emission was monitored as

a function of sputter ion energy, oxygen content of the

beam, and target temperature. As the YBa2Cu307-x target

temperature was increased above 200 'C, the intensity of the

Cu emission lines increased. Although not acknowledged by

Fledderman, this suggests that the sputter yield of Cu from

YBaCu307- increased as a function of temperature. As

discussed below, a Y-Ba-Cu-O thin film being deposited by

sputter deposition may be subjected to resputtering by

energetic O- ions or neutralized O, whose energy and

impingement rate depend on the plasma conditions. Increasing

the substrate temperature would result in an increase in the

yield of Cu sputtered from the film by the 0- ions or O

atoms. Thus, the decrease in the Cu concentration in

Y-Ba-Cu-O films with increasing substrate temperature could

be due to a temperature-dependent resputter rate of Cu from

the film, rather than due to a decreasing sticking

coefficient. A study of this issue is suggested for future

work, as discussed below in Chapter 7. Angular distribution variation for Y. Ba. and Cu

The significant differences in atomic mass among Y, Ba,

and Cu would make them good candidates for differing

sputtered angular distributions. Wehner et al.[81] measured

the composition of a Y-Ba-Cu-0 film deposited on a curved Ta

foil substrate by DC triode sputter deposition from a planar

YBa2Cu30,7 target. The curved substrate was flattened out and

the composition of the Y-Ba-Cu-O measured as a function of

the angle from substrate normal. These data indicate that Y

and Cu were ejected preferentially in the normal direction,

while the heavier Ba atoms were ejected more obliquely.

Similar results were observed by Burbidge et al. [82] for

films deposited by DC planar magnetron sputtering from a

YBa2Cu3Ov7x target onto 1 cm2 substrates located at various

angular positions. Wehner et al. explained the deviations in

Y-Ba-Cu-O film compositions in terms of angular distribution

variations. However, no effort was made in either study to

account for the effect of negative ions impinging on the

films above the target. As discussed below, the angular

variation in negative ion flux above the target could have

caused the observed Ba deficiency, rather than angular

variations in the sputtered flux of target atoms. Also, the

data of Kageyama and Taga [83] showed sputtered angular

distributions of Y, Ba, and Cu which did not agree with

those of Wehner et al., with the distributions being

strongly dependent on the deposition conditions including RF

power, substrate-target distance, and sputter gas pressure.

This is not surprising since, as discussed above in Section, the angular distribution of components sputtered

from a compound target does not always follow the prediction

of those of heavy mass being ejected more strongly along the

target normal.

Furthermore, an interesting result of Kageyama and

Taga, which has also been observed in studies of off-axis

deposition [63], was that regardless of the angular

distribution patterns of Y, Ba, and Cu, the film

compositions always converged to the target composition at

an angle of 90' to the target normal. Also, the higher the

sputter gas pressure, the closer the film composition to the

target composition [62]. None of these authors offered

explanations why such results were observed. These data

suggest that the Y-Ba-Cu-O film compositions were was not

simply determined by the mass-dependent angular distribution

variations of Y, Ba and Cu. It is suggested below in Chapter

7 that a valuable future study would be to isolate the

composition variation in Y-Ba-Cu-O films due only to the

angular variations in the sputtered flux of target atoms. Negative ion effects for sputtered Y-Ba-Cu-O

Negative ions have been considered to be the major

reason for the difficulty in obtaining 1:2:3 film

compositions by sputter deposition from YBa2Cu3Ov7. targets

[5, 35]. The fact that YBa2Cu307-x contains highly

electropositive Ba and Y bonded to highly electronegative

oxygen led to the suggestion that the dissociation of Ba-O

and Y-O bonds during sputtering of YBa2Cu3O7- led to the

formation of negative ions, namely 0- [57]. Thus negative

ion effects, particularly film resputtering, were thought to

be primarily responsible for the film composition

deviations. Also, the'results of many of the process

variations could be interpreted in terms of their effects)

on the negative ions, as will be discussed below. A few

groups attributed most of the problems to bombardment of the

substrate with electrons [84, 85, 86]. Electron effects do

not seem likely since similar problems were observed for

both magnetron and diode sources, and the electrons would

largely be confined away from the substrate in the magnetron


Rossnagel and Cuomo [57] were the first to explicitly

show that sputtering of YBa2Cu3O07- produced a large yield of

negative ions. The negative ions were accelerated by a grid

placed closely over the target. The particles were then

energy analyzed and counted in a four-grid energy analyzer.

Thus the analysis could only identify a particle's charge;

it could not identify the ionic specie. Ion beam sputtering

of a fresh YBa2Cu3O,07 target with Ar' was found to give a

negative ion yield of 0.3. In other words, for every 10 Ar*

ions which bombarded the target, approximately three

negative ions were created. More oxygen in the sputter gas

resulted in an increased negative ion yield. Also, the

deposition rates of Y-Ba-Cu-O films deposited by RF planar

magnetron sputtering from a YBa2Cu3,Ov7 target were found to

decrease with an increasing percentage of oxygen in the

sputter gas. A 100% oxygen sputter gas resulted in etching

of the substrates rather than film growth. These

phenomenological results led to the conclusion that the

dominant negative ion was 0-. No explicit identification of

0- or any other negative ion was obtained. It was suggested

that BaO or Y203 were responsible for the generation of O-

since they were reported to have high O- yields in SIMS

[87]. A fact not addressed by Rossnagel and Cuomo was that

neither compound was present in the single phase YBa2Cu307-

target. However, Shah and Garcia [85] reported that

resputtering of films occurred when using either a BaO or a

Y203 target but not a CuO target, although they did not

attribute the effect to O-. This suggests that the

dissociation of Ba-O or Y-O bonds during sputtering can lead

to resputtering by negative ions.

That negative ion effects can best explain the

deviation of Y-Ba-Cu-O film and target compositions is

strengthened by the fact that many of the results obtained

from variations in sputter deposition parameters can be

interpreted in terms of their effects) on negative ions.

For example, increasing the sputter gas pressure or

decreasing the RF power would increase the probability of

scattering of the negative ions [26] and reduce the energy

with which they bombarded the substrate, thereby decreasing

the amount of resputtering. Many of the changes in

sputtering parameters were enacted in attempts to reduce the

deleterious effects of negative ions on film composition and

properties. For example, substrates were biased with

negative voltages to in efforts to repel the negative ions

[59]. Although not acknowledged by the authors, the success

of this biasing depended on the sputter gas pressure and the

power applied to the target because scattering of the

negative ions could cause loss of the extra electron and

formation of a neutral particle, which would not be repelled

by the negative bias. Also, off-axis substrate locations

were used to avoid the direct path of negative ions

accelerated away from the target surface by the negative

voltage on the target. Although never stated in the

literature, the effectiveness of off-axis substrate

locations also depended on the sputter gas pressure and

power, because scattering of the negative ions would cause

their trajectory to be diverted away from the target normal.

In summary, attempts to avoid film resputtering by negative

ions have been based on either reducing the energy of the

negative ions striking the substrate or preventing the

negative ions from striking the substrate altogether.

2.1.6 Run-to-Run IrreDroducibilitv of Film Composition

Once an optimum target composition had been obtained

for a given single oxide target sputter deposition process,

even if all other parameters were kept constant, the film

composition could still vary between successive depositions.

This irreproducibility was found to depend on the target

firing temperature and target materials) [77], and was more

of a problem for systems without load-locked chambers [54].

Firing pressed oxide targets at only 500 'C prevented

reaction among the starting compounds, nominally Y203, CuO,

and BaCO3 [88], and was reported to improve the stability of

the targets and the reproducibility of the film composition

[77, 89]. The 500 'C firing resulted in just enough

sintering to allow handling and mounting of the targets.

Thus these targets essentially consisted of mechanical

mixtures of the starting compounds. In fact, an unsintered,

loose powder target was reported to result in easy

fabrication and improved run-to-run composition control [34,


There have been essentially three reasons suggested in

the literature for the run-to-run variation in film

compositions: 1) the atmospheric instability of YBaCu307-,,

2) the time-dependent loss of excess oxygen from YBa2Cu3O7x

during sputtering, and 3) the migration of Ba in YBa2Cu3O,07

targets during sputtering. Each of these suggestions will be

discussed below.

First, a brief discussion of the atmospheric stability

of YBa2Cu3O7- and related Ba-containing compounds is in

order, particularly with regard to understanding the

run-to-run irreproducibility of film compositions. The main

issue is that YBa2CuO307 and related Ba-containing compounds

such as BaO2 are hygroscopic and react with water [77]. The

amount of water present in the air as humidity in a typical

laboratory is sufficient to degrade the surface of a bulk

YBa2Cu307.- sample, resulting in a loss of superconductivity

at the surface. The reaction of YBa2Cu307-x with water results

in an enrichment of Ba at the surface, due primarily to the

formation of Ba(OH)2 at the expense of YBa2Cu3Ovx [90]. Other

atmospheric degradation products on the surface can include

hydrated Ba(OH)2, BaO2, hydrated BaCxO,, and adsorbed water.

Eventually the Ba(OH)2 may react with CO2 in the air and

convert to BaCO3 [72]. Thus the surface of a superconducting

YBa2Cu307_x sample becomes insulating. Also, the atmospheric

degradation of YBa2Cu307-. is often accompanied by cracking

and crumbling of the surface of the samples [40].

For thin films, it has been observed that the

resistance to atmospheric degradation is improved for in

situ deposited YBa2Cu3Ov-x films versus YBa2Cu30,7- films formed

by post-deposition heat treatment. As discussed in Section

2.1.4, this is probably due to the smoother film morphlogy

and decreased grain boundary area of in situ deposited

YBa2Cu307-x films. This is supported by the results of

Zandbergen et al. [71] and Buyukliamanli [90], who suggested

that the atmospheric degradation of YBa2Cu3Ov7x is dominated

by the reaction of water at particle or grain boundaries,

and decreasing the particle or grain boundary area greatly

improves the resistance to environmental degradation. A

single crystal, epitaxial YBa2Cu3O_7 thin film would be much

more stable in air than a typical YBa2Cu3O_7- target, which is

typically sintered from a polycrystalline powder with a

particle size of 5 50 pm and after firing has a density

equal to 50% 80 % of the theoretical maximum, depending on

the firing temperature [88]. The particle surface area in

such a case would be relatively large.

Based on the above discussion of atmospheric

degradation, the improved run-to-run reproducibility

observed with unreacted targets made of Y203, CuO, and BaCO3

is probably due to the improved atmospheric stability of

these compounds relative to YBa2CuaO-x. However, the firing

at 500 'C in air used to sinter the unreacted targets would

convert some of the BaCO3 on the surface to Ba(OH)2 or other

compounds [90]. This must be taken into account when using

such targets. For unreacted targets, the amount of material

on the surface with a composition different from the bulk of

the target would be less, and less sputtering would be

required to remove the degradation products.

The use of BaF2 rather than the usual BaCO3 as the Ba

source in unreacted Y-Ba-Cu-O sputter targets has been found

to improve the run-to-run control of the film composition

[77]. Also, the use of BaF2 as the Ba source greatly

improved the atmospheric stability of the as-deposited

Y-Ba-Cu-O-F films as well as the annealed, superconducting

YBa2Cu307.x films [91]. In the as-deposited films, the use of

BaF2 has been suggested to make them non-hydroscopic, in

contrast to as-deposited films containing other Ba compounds

[75]. However, BaF2 actually does react with the water in

air to form a thin surface layer of Ba(OH)2, which

essentially passivates BaF2 from further reaction [92]. For

superconducting YBa2Cu3_07 films, the improved atmospheric

stability gained by the use of BaF2 has been suggested to be

due to the formation of an oxyflouride overlayer, which

passivates the YBa2Cu30o7 surface [93].

Klein and Yen [94] studied the emission spectra

resulting from ion beam sputtering of YBa2Cu3O,7x. The

emission data provided the identification of hydrogen as it

was sputtered away from the YBa2Cu37-x. The emission spectra

was collected for a YBa2Cu3,07- sample which had been left in

ambient laboratory conditions for several weeks. It was

found that more than 30 minutes of ion beam sputtering at an

accelerating voltage of 800 V and a beam current of 40 mA

was required to bring the intensity of the H peak to near

zero. Although not suggested by the authors, the source of

the H was most likely Ba(OH)2 formed by the reaction of

YBaCu3Ov-x with water. This data illustrates that removal of

the atmospheric degradation products from a YBa2Cu3O.,

target requires substantial presputtering.

The length of time which the target is exposed to

atmosphere and the relative humidity of the ambient air can

both affect the degree to which surface degradation products

are formed [90]. The degradation products must be sputtered

away before the sputtered flux has the same composition as

the target bulk. The amount of presputtering needed for a

given target will depend on many factors including the

density of the target, the firing conditions, the humidity

of the environment in which the target was fired and stored,

and the sputtering conditions. This will be discussed

further below in Section 5.1.4.

Second, as discussed above in Section 2.1.3, Selinder

et al. [54] and Kadin et al. [34] found that very long

presputtering of a YBa2Cu3O7-, sputter target was necessary to

get a 1:2:3 film composition. This was suggested, although

never proven, to be necessary to completely remove the

excess oxygen in YBa2Cu307- in order to form tetragonal

YBa2Cu306. Until the excess oxygen was completely removed,

the Y-Ba-Cu-O film composition would be dependent on the

amount of time spent presputtering prior to deposition.

Selinder et al. [54] thus suggested that a run-to-run

irreproducibility in the film composition would occur as a

result of the time-dependent loss of the excess oxygen from

YBa2Cu3Ovx during sputtering. No suggestion was given as to

how the loss of excess oxygen from the target actually

affected the film composition. This will be discussed

further in Chapter 4 below.

Third, the run-to-run irreproducibility in the film

composition has been suggested to be due to the possible

migration of Ba to the target surface during sputtering and

the resulting time-dependent concentration of Ba on the

target surface [77], as was discussed above in Section In fact, the run-to-run irreproducibility due to Ba

migration to the target surface can be viewed as a

time-dependent manifestation of the same mechanism suggested

for the deviation between the film and target composition:

If more Ba were depleted from the target during successive

depositions, the degree of Ba deficiency would vary from run

to run.

The improvement in the reproducibility of film

compositions due to the use of unreacted composite targets

of Y203, BaF,, and CuO was suggested to be due to the

minimizing of the segregation of Ba in the target due to its

tighter bonding in BaF2 [77]. However, no data were

presented to support this proposition.

2.1.7 Summary

In summary, for films deposited under typical

conditions, a deviation of the Y-Ba-Cu-O target and

sputtered film compositions usually exists with films

deficient in Ba and also possibly Cu. The exact film

composition is strongly dependent on the sputter deposition

parameters including RF power, target composition, sputter

gas pressure, substrate temperature, percent of oxygen in

the sputter gas, and the angle between the target and

substrate normals. The deviation of the Y-Ba-Cu-O film

composition has been discussed in the literature in terms of

four possible explanations: 1) migration of Ba to the target

surface and the eventual depletion of Ba from the target, 2)

sticking coefficient differences of the components striking

the substrate, particularly at increasing substrate

temperatures, 3) variations in the angular distribution of

sputtered components due to differences in their masses and

4) negative ion effects, particularly resputtering of the

growing film. The first explanation may be viable but has

never been experimentally tested. The second explanation,

as suggested above, is probably incorrect since an

explanation for the decrease in Cu with increasing substrate

temperature must also account for why this only occurs above

a sputtering source. The third explanation could account for

some data presented in the literature, but the results were

not consistent among different researchers. Also, the effect

of variations in angular distributions were never clearly

isolated from those of negative ions, so there magnitudes

are unknown. Finally, the fourth explanation for the

deviation of film and target compositions has been

considered the dominant mechanism, since most experimental

results pointed to resputtering of the film by negative ions

accelerated away from the sputter target. However, no

explicit identification of the deleterious negative ion(s)

has been reported. Further, although much effort has gone

into avoiding the effects of negative ions, little work has

been published on trying to lower the number of negative

ions sputtered from the target in the first place.

2.2 Barrier Layers for YBaCAuQ2_ on Practical Substrates

The growth of very high quality YBa2Cu3Ox thin films on

ceramic oxide substrates such as SrTiO3, MgO, and LaA103 is

well documented [5, 6, 11, 35]. However, from the viewpoint

of applications and cost, it is highly desirable to grow

superconducting YBa2Cu.O7x films on substrates such as Si and

sapphire (A1203), which are more affordable and more

commonly used. The potential importance for high speed

operation by integration of superconducting YBa2Cu30_x thin

films with conventional semiconductor technology is


When a conventional post-deposition heat treatment

process of 800 900 'C is used, films of YBa2Cu307, react to

some degree with all substrates on which it has been

deposited. For successful substrates such as SrTiO3 and MgO,

the reaction does not appear to degrade the superconducting

properties because it is either limited to near the

YBa2Cu307-x/substrate interface or results in substitution

(e.g.) of Sr for Ba in the lattice with only marginal net

consequences on TC and Jc [95]. On the other hand, for heat

treatments in oxygen at 2 900 'C the reaction with other

oxides such as A1203 proceeds as much as 0.5 pm into the

YBa2Cu307- and films thinner than this would no longer be

superconducting [95]. Even for thicker films, the

superconducting cross-section is reduced, lowering the total

current necessary to exceed Jc. The reaction between

YBa2Cu3Ov7 and Si or GaAs is severe and leads to complete

loss of superconductivity [96]. The reaction of a Y-Ba-Cu-O

film and a Si substrate leads to the formation of BaSiO3 at

the Si interface and depletion of Ba from the Y-Ba-Cu-O

[97]. Even if an in situ growth process is used, the lower

temperatures do not eliminate reaction between the YBa2CuO_307

film and Si and the thickness of the film must be greater

than the reaction layer. For example, a 200 A thick

interface reaction layer forms between Y-Ba-Cu-O and Si at a

substrate temperature of 600 'C during the typical four

hours required to deposit a = 0.5 pm thick Y-Ba-Cu-O film

by off-axis sputtering [98]. While the effects of Si on the

superconductor are clear, the effects of Ba, Y, or Cu on any

semiconductor device on the substrate have been less well

studied. It would seem however that a barrier layer would be

equally necessary for avoiding degradation of the


The obvious method of preventing deleterious

interaction between Y-Ba-Cu-O films and substrates has been

the insertion of an intermediate layer between them. The

terms barrier layer and buffer layer have been used

interchangeably in the literature to describe the

intermediate layer. In this work the term barrier layer will

be used. The ideal barrier layer should 1) be chemically

passivating and prevent interaction between YBa2Cu3Ov, and

the substrate, and 2) not harm the YBa2Cu3O-x film or the

substrate, such as creating electrically active defects in

Si. It has become clear over the last two years that with

the strong anisotropy in the critical current, the ideal

barrier layer would also 3) grow very strongly textured or

epitaxially on the substrate to in turn allow epitaxial

growth of YBa2Cu3O7- with a c-axis orientation as needed for

maximizing J,. Data presented over the last two years have

shown that grain boundaries may be sources of resistance

destroying superconductivity. Therefore epitaxial growth,

which eliminates grain boundaries, would also be very

desirable. Finally, the ideal barrier should 4) have a

matched thermal expansion coefficient between that of

YBa2Cu307. and the substrate in order to reduce the stress

generated by thermal cycling and thereby prevent cracking.

A large number of barrier layer/substrate combinations

have been reported in the literature, with the majority of

work focusing on barrier layers on Si or sapphire, as

expected. Much of the early work was very Edisonian in

nature, with groups apparently trying almost every source

material available in their laboratories as barrier layers

[35]. Barrier layer materials reported include fluorides,

nitrides, oxides, and metals (e.g. BaF2,, CaF, BN, WN, VN,

NbN, Ta2Os, Y203, SiO2, BaTiO3, MgO, ITO, TiO,, ZrO2, Cu, Ag,

Au, Pt, Nb, Ni, and Ti) [35]. In most of these studies,

preventing interface reactions was the only objective. No

regard made to growing textured or epitaxial barrier or

superconductor layers. Most of the barrier layers were not

successful, particularly on Si and sapphire. The best

results on Si substrates were obtained with a ZrO2 barrier

layer [99], whereas for sapphire, the best improvements in

the superconducting properties of YBa2Cu3O.x were obtained

using Pt [100] and ZrO2 [101]. Although barrier layers of Ag

on sapphire and other oxide substrates yielded high quality

YBa2Cu3Ox. films, Ag was not stable against interaction with

YBa2Cu30.x [102]. In fact, Ag was found throughout the

YBa2Cu3O07- films. However, silver's incorporation into

YBa2Cu3O,.x did not cause degradation of the superconducting

properties. Rather, Ag apparently improved the

superconducting properties (Ta, J,) by a proposed mechanism

of lowering the grain boundary resistance [102].

Recent barrier layer studies have emphasized the growth

of strongly textured YBa2Cu3.07 films on Si (100) substrates.

The best results reported presented two different barrier

layers which came close to being ideal. First, a bilayer

barrier consisting of an initial layer of MgAl204 on Si

followed by a layer of BaTiO3 facilitated the growth of

strongly textured YBa2Cu3O07- with complete c-axis orientation

[98]. An in situ YBa2Cu3O7-x growth process was used. The

lattice mismatch between the first layer and Si, the first

and second layers, and the second layer and YBa2Cu37-x were

all small enough to permit the YBa2Cu3Ovx to grow with its

c-axis normal to the surface of the (100) oriented Si

substrate. Also, the thermal mismatch among the layers, Si,

and YBa2Cu3O-x was small enough to avoid microcrack formation

in the YBa2Cu3O,7x. A sharp transition at Tc = 87 K and a Jc

at 77 K of 105 A/cm2 were measured. Secondly, by optimizing

the amount of Y203 in the film, barriers of Y203-stabilized

ZrO2 (Y-ZrO2) were grown epitaxially on Si and epitaxial

YBa2Cu30O7. was grown on the barrier [103]. The Y203

concentration was optimized to adjust the lattice parameter

of the Y-ZrO2. The Si surface preparation was found to be

crucial and consisted of an innovative procedure to hydrogen

passivate the freshly-cleaned Si surface prior to

deposition. The clean, ordered Si surface was then exposed

by heating above 500 'C in situ.

2.3 Directions for Research

In order to develop sputter deposition processes for

YBa Cu307.x with improved compositional control, in light of

the literature it is clear that a better understanding of

the mechanisms behind the deviation of the film and target

compositions are needed. In an RF planar magnetron

sputtering system the situation is very complicated, and it

is difficult to sort out which mechanismss, whether one or

more of those presented in the literature or one not yet

suggested, is (are) affecting the film composition. It is

likely that more than one mechanism may be acting

synergistically to influence the Y-Ba-Cu-0 film composition.

Experiments are needed to independently study the effect and

magnitude of each possible mechanism. Also needed are

explanations of the deviation between the target and film

compositions which account for the effects of more than one

mechanism acting simultaneously. It would then be possible

to intelligently design sputter deposition processes with

control over the Y-Ba-Cu-O'film composition, rather than

following the Edisonian approach which has dominated the

literature to date.

In this study it was desired to evaluate two of the

suggested mechanisms for the deviation between the

composition of Y-Ba-Cu-O targets and sputtered films. First,

an explicit identification of the negative ion(s) proposed

to cause film resputtering and to be the primary reason for

the deviation of the film and targets compositions was

clearly needed. This information could then be used to

choose target materials and/or fabrication procedures which

minimized the number of negative ions created during

sputtering. Secondly, it was considered valuable to

experimentally determine if the migration of Ba to the

target surface during sputtering and its subsequent

depletion were in fact occurring. Such information would

offer guidelines on the requirements for target fabrication,

mounting, and cooling.

In this study it was also desired to study barrier

layers for the growth of superconducting YBa2Cu3,07- films on

practical substrates. In the barrier layer study presented

below it was desired to combine the best substrate for

growing YBa2Cu3Ov7x, SrTiO3, with both the most widely used

semiconductor substrate, Si, and the most common single

crystal oxide substrate, sapphire (a-A203). Since

YBa2Cu307-x was known to be relatively stable with SrTiO3, it

was thought that SrTiO3 might make a stable barrier layer

between YBa2Cu3O7- and Si or sapphire. At the start of this

study, chemical stability was the key issue with the growth

of strongly textured YBa2Cu3v07 on a barrier layer on Si or

sapphire being a secondary issue. However, as the study

progressed, the importance of the textured growth of the

barrier layer on Si became important and was addressed as



The general experimental procedures used in the course

of this work are presented in this chapter; more specific

experimental procedures are presented as needed in other

chapters. First, a secondary ion mass spectroscopy (SIMS)

study of the negative ions sputtered from YBa2Cu3v07_ and

related compounds is discussed. Second, an Auger electron

spectroscopy (AES) sputter depth profiling study, performed

to determine if Ba is migrating to the target surface during

sputtering of YBa2Cu30v7-, is presented. Third, the processing

system and steps required for the sputter deposition of

Y-Ba-Cu-O and barrier layer thin films, including the

fabrication of sputter targets, RF planar magnetron sputter

deposition parameters, and post-deposition heat treatment,

are covered. The sputter deposition system is described in

more detail in Appendix A. Finally, the physical and

electrical characterization of the Y-Ba-Cu-O thin films are


3.1 Identification of Negative Ions by SIMS

In order to identify the species and determine the

relative yield of the negative ions being created by the

sputtering of Y-Ba-Cu-O targets, a negative SIMS study was

carried out, the results of which are presented in

Chapter 4. For a complete discussion of the principles of

SIMS characterization, the reader is referred to an

excellent, comprehensive book recently written by

Benninghoven et al. [22]. A brief explanation will suffice

for understanding the data presented here. The information

provided by SIMS is related to the composition of a sample's

surface. An ion beam is used to sputter the surface, and an

analyzer is used to mass separate the secondary ions created

during sputtering. The secondary ions which are ejected come

from the uppermost layers (15 A) of a solid's surface,

thereby giving SIMS an excellent surface sensitivity. The

mass-to-charge ratio (m/e) of the sputtered species versus

the intensity of each specie provide direct information on

the composition of the sputtered area. For accurate

quantitative analysis, correction factors generally must be

applied to the raw data.

A 3M model'610 static SIMS was used in the negative

secondary ion detection mode with a primary ion beam voltage

of 0.5 kV, an Ar pressure of 5 x 10-s Torr, and a constant

beam current of 5 nA, typically. Samples were all mounted

simultaneously on a multiposition carousel and analyzed

under the same vacuum conditions in order to eliminate

artifacts in the data, particularly with regard to beam

alignment. The analysis system contained both a 3M SIMS and

a Perkin Elmer Phi model 545 scanning Auger spectrometer.

The sample normal on the carousel was 60' away from the

Auger electron analyzer axis, in the plane defined by the

analyzer and the UHV manipulator axis. The manipulator was

located on top of the stainless steel chamber and mounted

with the axis vertical. The ion gun was mounted on a

2.75 inch diameter flange on top of the chamber, 15' away

from the analyzer axis. The quadrapole mass analyzer was

mounted on a 4.5 inch diameter flange 15' from the

manipulator axis on the other side on the plane formed by

the manipulator and analyzer axes. The incident angle of of

the ion beam on the sample was about 30' and the take-off

angle to the quadrapole was approximately the same. The

distance from the sample to the quadrapole mass analyzer

was about 5 inches. A modified Bessel box filter was used to

energy select secondary ions and an extraction voltage was

used to attract ions from the sample to the mass analyzer.

Cylindrical targets of 0.25 inch diameter and 0.25 inch

thick were produced using the same process as for the two

inch diameter sputter targets used in film deposition, as

discussed below. Starting powders for the SIMS targets were

YBa2Cu3Ox and 123 mixtures of CuO, Y203, and either BaF2,

BaO2,or BaCO3. A YBa2Cu30, target was formed by reducing a

YBa2Cu3O07 target at 600 'C for six hours in 90% N2/10% H2

forming gas. The orthorhombic YBa2Cu3O,. target was fired at

950 'C in air for 168 hours to maximize formation of the

superconducting phase, as verified by X-ray diffraction. The

1:2:3 mixtures were fired at 500 'C for 12 hours to promote

sintering but to avoid reaction. Quickly after the firing

process, the back side of the targets was sanded down to

reduce the thickness to 0.125 inch, as needed to fit under

the clips on the SIMS' sample carousel. The samples were

immediately loaded into the SIMS' UHV vacuum chamber and

pumpdown was begun.

The UHV chamber was pumped by sorption roughing and ion

pumps down to a base pressure of 5 1 x 10-9 Torr. The poppet

valve to the ion pump was then partially closed to a small

conductance and ultra-high purity (99.9999%) Ar was

backfilled into the chamber to establish the typical Ar

sputtering pressure of 5x10-5 Torr for sputtering. To study

the effect of oxygen on the negative ion yield, 10% 02

(99.9995%) was sometimes added to the sputter gas. As will

be discussed below in Chapter 4, it was desired to

determine if Ba with its low work function on a freshly

sputtered YBa2Cu307- surface could supply electrons to

thermal 02 and form 0-. To test this, the YBa2Cu3O07- target

was sputtered with Ar* and then the chamber was backfilled

with O0 to 1x10-6 Torr with the mass spectrometer operating.

Finally, to simulate the heating of targets during

sputtering deposition, the SIMS targets were sometimes

sputtered while being'heated up to 300 'C by a Ta foil

envelope heater.

To collect SIMS data, the sample alignment was first

checked by observing the location of an electron beam on the

target surface. The sample position was then adjusted using

a cylindrical mirror analyzer to locate the elastically

reflected electrons at the incident beam energy of 2 keV.

The target surface was presputtered for two minutes with an

ion beam voltage of 3 keV to remove surface contamination

and artifacts. An AES survey was then taken to check the

cleanliness of the sample surface. For SIMS analysis, the

ion beam voltage was reduced to 0.5 keV to simulate voltages

closer to those used in sputter deposition. The Ar pressure,

emission current, and beam voltage were kept constant to

keep the ion beam current constant at 5 nA. The raw SIMS

data were collected on an x-y recorder. The mass/charge

range was set at 1 to 50 amu, while the y-axis counts per

second (cps) range was varied to optimize the resolution of

the data plot, but was typically 104 105 cps. The

intensity of the 0- peak was characterized by the maximum

peak intensity. Only relative O- yield data were compared

among samples. The results of these experiments are

presented in Chapter 4.

3.2 AES Study of Long Time Sputter Effects on YBaCuQ10

In order to determine if the devation of film and

target composition was due to the migration of Ba or other

species to the target surface during sputtering, an Auger

electron spectroscopy (AES) sputter depth profiling study of

a heated YBa2Cu307x target was performed. To simulate the

conditions of actual magnetron sputter deposition, the

target was heated to a temperature for deposition (300 'C)

and sputtered with Ar ions for extended periods of time. If

changes in the surface concentration of Ba or other species

were measured with increasing sputter time, it would

indicate that migration of Ba from the bulk to the target

surface was occurring.

A cylindrical 0.25 inch diameter by 0.25 inch thick

YBa2Cu3yO7 target was fabricated as described in Section 3.1.

The YBa2Cu3O7 x target was mounted in a Ta envelope and heated

to 300 'C by passing a controlled DC current through Ta wire

leads spot-welded to the envelope. The temperature was

monitored by a type K thermocouple spot-welded to the

envelope. The analysis was performed in the same UHV vacuum

system as the negative SIMS measurements and similar vacuum

conditions were obtained. The Physical Electronics Model 545

scanning Auger Electron Spectrometer described above was

used for AES analysis while sputtering with Ar ions with a

voltage of 3 keV and current of 5 nA. AES depth profiling

was performed for sputter times from 0.1 to 30 hours. The

electron beam was rastered over an area of = 0.25 mm2 and

the ion beam diameter was approximately 1 mm, versus an

average grain (particle) size of = 50 mu in the target. Thus

the area sputtered and analyzed contained a large number of

grains and grain boundaries, as is the situation in actual

magnetron sputtering. The raw AES first derivative data were

output in real time to a strip chart recorder. The

peak-to-peak heights of the Ba, Cu, and O Auger signals were

monitored as a function of sputter time. The intensity of

the Y peak was too low to accurately measure changes. The

results of this study are presented in Chapter 4.

3.3 Processing of Y-Ba-Cu-0 Thin Films

Thin films of Y-Ba-Cu-0 were grown by RF planar

magnetron sputter deposition from single targets.

Cold-pressed, sintered targets were prepared from Y203, CuO,

BaCO3, BaO2, or BaF2. A BaF, precursor was chosen to

determine if improvements suggested by the results of the

negative SIMS study presented in Chapter 4 could be realized

in actual sputter deposition of Y-Ba-Cu-O films. The various

aspects of Y-Ba-Cu-0 film processing and characterization

are described below. Since the growth and characterization

of barrier layer films are very similar, the barrier layers

are discussed along with the Y-Ba-Cu-0 films.

3.3.1 Sputter Target Fabrication

Powders for Y-Ba-Cu-O sputter targets were combined in

the desired Y:Ba:Cu ratio using Y203, CuO, and BaCO3, BaO2,

or BaF2 powders of 99.5% purity or better from either AESAR,

Aldrich, or Alfa Chemicals. The particle diameter of each

powder was 5 um or less. Acetone was added to the mixtures

and the resulting slurry was shaken by hand in a sealed

plastic jar. The mixtures were transferred to a glass beaker

and carefully heated at 160 'C for three hours in a fume

hood to remove the acetone. A binder solution of polyvinyl

alcohol, deionized water, and glycerine was added to the

powders to improve the strength of cold-pressed targets. The

powders were transferred to a plastic jar along with 0.5 cm

diameter ZrO2 milling media. Complete mixing was obtained by

ball milling for 12 hours. The powder mixture was poured

into a two inch diameter stainless steel die. The powder was

cold pressed at a pressure of 13,000 psi into a two inch

diameter by 0.125 inch thick target.

For unreacted, composite targets, green targets were

heated at 500 'C for 12 hours in air in order to burn off

the binder and to promote sufficient sintering for

mechanical stability. For targets in which reaction of the

starting compounds was desired, for example to form

orthorhombic YBa2Cu3O7x, firing in air was done at

temperatures up to 950 'C for times ranging from 12 to 120

hours. The phases formed in the targets as a function of

temperature were determined by X-ray diffraction, as

reported below in Chapter 5.

After firing, targets were bonded with CERAC Ag epoxy

to a copper backing plate for improved thermal conductivity

to prevent cracking. The total thickness of the target,

epoxy, and backing plate was kept at about 3/16 inch due to

mounting restrictions in the sputter source.

3.3.2 RF Sputter Deposition of Y-Ba-Cu-O and Barrier Layer
Thin Films

All Y-Ba-Cu-O and barrier layer thin films used in this

study were grown in the same home-built multisource sputter

deposition system. The system, described in detail in

Appendix A, contained.two US Gun model I DC and two US Gun

model II RF planar magnetron sputter sources. All films

discussed in this study were deposited using an RF source.

Home-made and/or commercially obtained two inch diameter

sputter targets, bonded to a Cu backing plate as discussed

above, were mounted in the sources. Conductive Ag paste was

applied to both the Cu backing plate and the mounting stage

of the sputter source to insure good thermal contact.

Cooling water flowing beneath the stage of the sputter

source thus kept the targets cool during sputtering.

Prior to loading for deposition, substrates were

cleaned to remove hydrocarbon contaminants, metallic

impurities, and native oxide in the case of Si. Substrates

used in this study included (100) Si, (100) SrTiO3, and

(1120) (A-plane) sapphire, which were obtained from

Virginia Semiconductor, Commercial Crystal Laboratories, and

Sapphikon, respectively. All substrates were given a minimum

degreasing clean of five minutes in ultrasonically agitated

methanol, followed by blow drying with nitrogen or freon.

For sputter rate calibration or films deposited for

compositional analysis by electron probe for microanalysis

(discussed below), a Si substrate was used and no other

cleaning or etching was done. However, for the growth of

superconducting Y-Ba-Cu-O films or of barrier layers, a more

thorough cleaning was usually performed. The exception was

SrTiO3 substrates, which were purchased precleaned and thus

were only given a methanol degreasing. A more complete

degreasing included an additional first two steps of two

minutes in trichloroethane followed by two minutes in

acetone prior to methanol cleaning. The containers of both

solvents were placed in an ultrasonic tank. Sapphire

substrates were then cleaned in a 1:1:10 HF:HC1:H20 mixture

for 10 minutes to remove metallic surface contaminants,

rinsed in flowing deionized water and blow dried with N2.

Silicon substrates, after being degreased and and then

rinsed in deionized water, were etched in a 10:1 deionized

water:buffered oxide solution to remove the native oxide.

This etch was followed by rinsing in flowing deionized


After cleaning, all substrates were immediately loaded

onto the home-built substrate holder described in Appendix

A. For verification of the anticipated film thicknesses and

to establish sputter rates, a portion of a Si substrate was

masked with a cleaved'piece of a Si wafer and held in place

on the substrate holder by a tungsten clip, as described in

Appendix A. The substrates and the holder were blown clean

with N2 filtered to remove particulates > 0.5 pm. The

substrate holder was then attached to the arm of the

substrate rotation assembly, as illustrated in Figure A-5.

The distance between the target and the substrate and the

position of the substrate above the target were set as

desired. Typically, the substrate was 6 cm directly above

the target.

The diffusion-pumped, liquid nitrogen trapped glass

bell jar, with a stainless steel clam shell liner, high

vacuum system was typically pumped to a base pressure of

1x10-6 Torr for depositions. Argon or Ar+O2 sputter gas was

backfilled into the chamber at a controlled flow rate. In

conjunction with a limiting of the pumping speed by a

throttle valve, a constant pressure of 10 mTorr was

typically established. The RF plasma was then ignited and

the reflected RF power minimized with a tuning network. The

desired RF power (typically 50 W) was established and the

target was presputtered with a shutter protecting the

substrate. New Y-Ba-Cu-O targets were sputtered for at least

three hours prior to use. A presputter of at least one hour

was used for all subsequent depositions from a target, with

three hours not being uncommon. For reacted Y-Ba-Cu-O

targets, the amount of presputtering was found to strongly

effect the film composition.

After presputtering, the shutter was opened and

Y-Ba-Cu-O deposited for the estimated length of time needed

to grow a film of the desired thickness (based on the

sputter rate of a deposition with the same parameters).

Frequent manual returning of the RF matching network was used

during deposition to maintain constant deposition power.

Substrates were electrically floating and nominally

unheated, although during deposition a temperature of up to

200 'C was possible due to electron, ion, and neutral

bombardment of the substrate. After deposition, the shutter

was closed and the RF power and the sputter gas flow were

brought to zero. The substrates and target were allowed to

cool under vacuum at a pressure of 5 2 x 10-6 Torr for at

least 15 minutes. The pumping system was valved off and the

chamber backfilled with nitrogen gas to bring it up to

atmospheric pressure. Finally, the substrates coated with

newly grown Y-Ba-Cu-O films were removed.

After a deposition, the film thickness was determined

by removing the Si mask and measuring the height of the

resulting step with a Sloan Dektak Model I profilometer.

Typical film thicknesses were 0.5 to 1.0 pm. The sputter

rate was calculated by dividing the thickness by the time of

the deposition in minutes. Since RF power, gas pressure, and

other deposition parameters were held constant, the sputter

rate was assumed to remain constant throughout a deposition

[20]. Deposition rates for RF power of 50 W were typically

30 50 A/min.

The deposition parameters of a typical Y-Ba-Cu-O

deposition were as follows: 50 W RF power, 10 mTorr Ar, 6 cm

substrate-to-target (S-T) distance, and substrate

electrically floating, nominally unheated, and positioned

face-down directly above target center. The prime emphasis

of this study was to control the film composition by using

these typical deposition parameters and varying the target

composition, the starting Ba compound, and the target firing


Different target compositions were used to iteratively

approach a cationic Y:Ba:Cu ratio of 1:2:3 in the film, as

will be discussed in Chapter 5. Barium precursor compounds

of BaCO3, BaO2, or BaF2 were studied to determine their

effects) on film composition. Target compositions for all

Ba compounds included 1:2:3 and 1:4:4 (indicating the ratio

of Y:Ba:Cu). After BaF2 was found to give the most

consistent results among the Ba precursor compounds, the

target composition using BaF2 was iteratively refined to

1:1.6:2 in order to get a film composition near 1:2:3.

Barrier layer films of SrTiO3 were RF sputter deposited

from commercially obtained stoichiometric oxide targets onto

Si (100) and sapphire (1120) substrates. Initial

depositions used the standard deposition parameters

established for Y-Ba-Cu-O films, since this allowed

presputtering of the Y-Ba-Cu-O target while the barrier

layer was being grown. These included 50 W RF power,

10 mTorr Ar, 6 cm substrate-to-target (S-T) distance, and

substrate electrically floating, nominally unheated, and

positioned face-down directly above target center. However,

when a pure Ar sputter gas was used, SrTiO3 films were found

to be highly stressed in the as-deposited condition and

exhibited cracking and even peeling, as discussed in

Chapter 6. Oxygen was added to the sputter gas to reduce the

stress in the SrTiO3 films. The deposition rate for SrTiO3

was typically 40 50 A/min and the thickness of the barrier

layer films was typically 0.4 0.5 um. After the growth of

the barrier layer, either the substrate was rotated over the

second RF source and Y-Ba-Cu-O was deposited without

breaking vacuum, or the sample was removed from the vacuum

chamber and subjected to heat treatment prior to Y-Ba-Cu-O

growth. The heat treatment was intended to relieve stress in

the barrier layer and to promote crystallization and/or

grain growth.

3.3.3 Post-deposition heat treatment

As-deposited Y-Ba-Cu-O thin films were not

superconducting but rather were insulating and amorphous. In

order to form superconducting YBa2Cu3O7-x, the films required

a post deposition heat treatment. The heat treatments were

done in a three-zone Applied Test Systems tube furnace with

an enclosed 51 mm diameter quartz tube for containing the

desired ambient gas. Samples were loaded onto a quartz boat

and positioned in the center of the quartz tube. A cap was

placed on the end of the tube and gas regulated by a valve

was flowed through the tube and exited through an oil

bubbler. The furnace was powered by three Leeds and Northrup

model 11906-223 zero voltage power packs and was controlled

by an Omega model CN-2010 microprocessor, which regulated

both setpoints and heating and cooling rates.

The heating cycle was similar for all Y-Ba-Cu-O films.

The samples were brought from room temperature to the

desired high temperature setpoint of 850 900 'C in 1 to 2

hours. Then the temperature was held constant for a one hour

soak. After the soak, the temperature was lowered very

slowly ( 53 'C/min.) to room temperature by either simply

shutting off power to the furnace or by having the

controller regulate the cooling. Occasionally, during

cooling an additional 2 to 6 hour soak was performed at

500 'C in an attempt to improve the oxygen stoichiometry of

the YBa2Cu3O7-x film. For films deposited from targets with

BaO2 or BaCO3 precursor compounds, heat treatments were

performed in 99.995% purity oxygen flowing at 10 30 sccm.

For BaF2 containing targets, the films required heating in

humid oxygen in order to reduce the BaF2 to BaO and permit

formation of YBa2Cu307- [75], as discussed in Chapter 2.

Oxygen was humidified by passing it at an accelerated flow

rate of 200 sccm through a 1000 ml flask of deionized water

at room temperature. The humid oxygen was used during

heating from room temperature and during the high

temperature soak, but dry oxygen at a reduced flow rate of

30 sccm was used during cooling.

3.3.4 Physical Characterization of Thin Films

The composition of as-deposited Y-Ba-Cu-O thin films

was determined by Eledtron Probe Microanalysis (EPMA) on a

JEOL Superprobe model 733 EPMA System. All EPMA work was

performed by MAIC staff at the University of Florida. To

assure the surface smoothness which is critical to

quantitative EPMA and to avoid secondary fluorescence of the

elements in the film due to emission from the substrate,

0.5 1 pm thick films were deposited for composition

analysis on two inch diameter Si (100) substrates. The EPMA

analysis was performed using wavelength dispersive

spectroscopy with an 8 keV electron accelerating voltage, a

20 pm beam diameter, and a 40' tilt between the crystals and

the normal to the sample surface. The 8 keV voltage was used

to prevent the electron beam from penetrating beyond the

films into the substrate. Data was collected for 60 seconds

to obtain a quantitative accuracy of 2 atomic percent.

Calibration standards were chosen with Y, Ba, and Cu bonded

to oxygen in order to approximate the bonding condition in

the oxide targets: YAG for Y, CuO or Cu for Cu, and BaSO,

for Ba. Also, CaF2 was used as the calibration standard for

F. Tracor Northern phi-rho-z correction software was applied

to the K values to obtain quantitative composition. Analysis

of a bulk, textured YBa2Cu307- sample obtained from David

Tanner in the Department of Physics at the University of

Florida yielded a 1:2:3 composition, thereby verifying the

accuracy of the measurement. Further, the composition values

determined by EPMA for two films were verified by Rutherfod

Backscattering Spectroscopy. The film thicknesses had to be

at least 0.4 um to prevent fluorescence from the substrate.

To assign a single composition value for a given deposition,

five data points were averaged for a substrate of 1 cm2

positioned 0.5 inches from the center of the substrate

holder, which was in turn centered directly above the

target. For composition profiles across two inch diameter

substrates, an EPMA line scan was performed with 25 data

points spaced 2000 pm apart. Data were plotted as atomic

percent versus position from substrate center.

The crystallinity and structure of thin films was

determined by X-ray diffraction (XRD). Samples were mounted

with double-stick tape onto a 1" by 3" glass slide. All XRD

data collection were performed by MAIC staff at the

University of Florida. A Phillips model APD 3720

diffractometer was used with Cu Ka radiation and generator

settings of 40 kV and 20 mA. The angle 20 was varied from

15' to 90' at a rate of 3'/min. A typical full-scale

intensity of 1000 cps was plotted versus 20. Phase and

structure identification were done by comparing the

d-spacing data to standard values from a JCPDS [104] card

file and from values in the recent Y-Ba-Cu-O literature.

The morphology of the Y-Ba-Cu-O and barrier layer thin

films was studied by scanning electron microscopy (SEM) and

optical microscopy. A JEOL model JSM-35C SEM was used to

obtain images of the film surface at magnifications up to

10,000 times. Both Y-Ba-Cu-O and barrier layer films had to

be coated with 100 A of gold to prevent charging of

nonconductive regions. Optical microscopy provided

complementary information at lower magnifications and

required no sample preparation. In particular, identifying

microcracks in barrier layer and Y-Ba-Cu-O thin films

required both SEM and optical microscopy. A Nikon model

Epiphot-TME optical microscope was used at magnifications up

to 1000X.

The distribution of elements across the surface and

throughout the thickness of the films was determined by

Auger Electron Spectroscopy (AES) surveys, images, and/or

depth profiles. The distribution of Y, Ba, Cu, O and

sometimes F was studied for as-deposited and heat treated

Y-Ba-Cu-O films. The ability of barrier layer films to

prevent interaction of Y-Ba-Cu-O and a given substrate was

also determined by AES sputter profiling. The AES analysis

was done on a Perkin Elmer model 660 Scanning Auger

Microprobe. An electron beam accelerating voltage of 10 keV

and a current of 30 nA was used. The peak energies used for

each element were as follows (values in eV): Y = 1746, Ba =

584, Cu = 920, 0 = 510, F = 650, Si = 1619, Sr = 1649, Ti =

418, and Al =1396. For sputtering, an Ar' beam produced in

an ion gun with a 25 mA emission current and an Ar pressure

of 1.1 x 10-' Torr was accelerated by a voltage of 3 keV

and rastered in a standard 3 mm x 3 mm area. Sputter rates

were estimated from calibrated values on Ta20s known for the

accelerating voltage used. The data collection and reduction

was computer-controlled by a Perkin Elmer personal computer

and software written by Perkin Elmer. The Auger surveys and

depth profiles were plotted on either a Hewlett Packard

laser printer or pen plotter.

3.3.5 Electrical Characterization of the Y-Ba-Cu-0 Films

Resistivity versus temperature characterization is the

most common way to obtain information about the

superconducting transition properties of a thin film or bulk

material [3, 5]. In this approach, the resistivity of a

material is measured as a function of a decreasing

temperature. At a specific temperature value, the

resistivity may begin to drop suddenly, indicating the onset

of superconductivity. As the temperature is lowered a

resistivity of zero may eventually be observed, indicating

complete transition to the superconducting state. The

temperature at which the resistivity equals zero is the

critical transition temperature, Tc. Of course, a

measurement of zero resistivity cannot truly be made. The

"zero" value is typically taken as the lowest resistivity

value measurable by a given system, which can either by

limited by the equipment at hand or by noise. For the common

DC four point probe technique used in this study, discussed

below, the equipment limitations were usually the output

range of the current source and the sensitivity of the

voltmeter available. Since the resistive transition for high

quality superconducting YBa2Cu30O7- samples usually drops

steeply, the exact definition for zero resistivity may not

be critical. However, many authors have reported on

YBa2Cu3Ov-x samples for which the resistive transition was not

sharp and extended over several degrees, as discussed below.

In such cases, a clear statement of what comprises zero

resistance, or zero voltage for DC four contact

measurements, is necessary to be able to compare reported Tc

values. In this study, for reasons discussed below, the

value for zero volatge was 0.9 gV. Unfortunately, in the

literature the definition for zero resitivity or voltage is

rarely given. To avoid the issue of what comprises zero

resistivity, T, values are often reported as the temperature

at the midpoint of the superconductor resistive transition.

This was especially true during the first few years of the

high temperature superconductor field, since the quality of

much of the material was not high and the resistive

transitions were often not sharp. Even today Tc values based

on either the midpoint of the transition or on zero

resistivity are utilized. However, the basis for the Tc

value is often not stated.

Resistivity versus temperature data for a good quality

post-deposition heat treated YBa2Cu3O thin film on SrTiO3

are shown in Figure 3-1 [105]. In this figure, the normal

state resistivity decreases linearly from room temperature,

300 K, to an extrapolated intercept on the resistivity axis

near a value of zero at 0 K, as expected for metallic

conduction. The superconducting transition occurs near 90 K

and the resistivity drops to zero within a narrow

temperature range which is nominally less than 5 K. The

relatively soft shoulder on the transition is typical of

YBa2CuO x.. The room temperature resistivity is about

900 pQcm, which is high considering the relatively sharp

superconducting transition, even for post-deposition heat

treated films. During the last few years, room temperature

resistivity values less than 200 1ucm have been reported

for in situ grown YBa2Cu307- films [5, 13, 65].

At the end of the superconducting transition a small

tail can appear in which the final drop to zero resistivity

may take several degrees. The extent of the tail portion of

the R-T data depends on the definition of zero resistivity,

which depends on the measurement capabilities, as discussed



E /GL 51-A3
U 0.6

> 0.4 ..

0.2 /

I /
0 / ..... L ...a---- ------- |I ------ 1,-----
0 60 120 180 240 300

Figure 3-1. Example of good quality resistance versus
temperature data for post-deposition heat
treated Y-Ba-Cu-O film [105].

above. As a result, the data in the tail portion of the R-T

plot are often ignored in the literature.

Several parameters are commonly used to compare high

temperature superconductor R-T data, including the critical

transition temperature, TC, the transition width, AT, and

the resistivities at room temperature (300 K) and 0 K, pr

and po, respectively. The relationship of these parameters

to physical properties of superconducting Y-Ba-Cu-O thin

films will be discussed in Chapter 5. As pointed out above,

Tc can be defined either as the midpoint of the

superconductive transition, TC,mid, or as where the

resistivity goes to zero, To0. The determination of Tcld

will be discussed below. The value for po is taken as where

the linear extrapolation of the normal state portion of the

curve intersects the resistivity axis. This parameter gives

an indication of the residual normal state resistivity in a


Similar to the case for Tc, the definition of AT is not

consistent among publications. Strictly, AT is defined as

the difference of the temperature at the onset of

superconductivity, Tonst,, which is nominally where the

resistivity begins to drop, and the temperature where the

resistivity goes to zero, Tc,o. The problems with the

definition of zero resistivity were discussed above. The

definitions of Tonset are inconsistent because of the soft

shoulder between the normal state and superconducting

transition portions of the R-T data for YBa2Cu307-x. As a

result, in some samples a Tnset of above 100 K would be

obtained, when in reality the transition to the

superconducting state does not occur until about 92 K. It

then follows that the values of AT could be unexpectedly


To solve the problem for the determination of Tons.t, one

of a few methods is commonly used [106]..The method used in

this study is illustrated in Figure 3-2, using R-T data

3.0 YBCO/SrTiO


o 2.0 p o,


n 1.0
0.5 T T",
T o To
0.0 I
0 20 40 60 80 100120140160180200220240260280300
Temperature (K)

Figure 3-2. Illustration of graphical method used to
calculate Tc and AT from R-T data.

presented in Chapter 5. The linear extrapolation of the

normal state portion of the curve is drawn to determine po,

as discussed above. A line is then drawn through the

superconducting transition region of the curve. Where the

two lines intersect is taken to be Tont. In this way, the

uncertainty due to a soft shoulder on the transition is

removed. Further, once T.nt has been defined, the

resistivity corresponding to the onset, Ponse,, can be

determined. Subsequently, values of Tcmid and AT can be

determined. Another common practice is to take AT be the

difference between the temperatures corresponding to the

resistivity values equal to 90% and 10% pon.t [106]. The

value for Tcmid is taken to be the temperature where the R-T

data crosses the value of 50% of ponet"

Data in the literature are presented either as

resistance versus temperature or resistivity versus

temperature plots. Either type of data presentation allows

the determination of T, and AT. However, recent publications

have more frequently presented resistivity versus

temperature data, since the normal state resistivity values

are also indicative of film quality.

In this study, room temperature resistivity values for

Y-Ba-Cu-O thin films were determined using a home-built four

point probe measurement system. Room temperature

resistivity data were used as a first screening test for the

superconductivity of a film. If a film was insulating or a

very poor conductor, it was unlikely to exhibit a

superconducting transition and further electrical testing

was not performed. An Alessi model CPS four point probe test

fixture and model C4S four point probe head, with collinear

spring-loaded osmium probe tips of 0.005 inch diameter

spaced 0.05 inch apart, were used. A Lake Shore Cryotronics

model 120 current supply provided a constant current to the

outer two of the four contacts. A Keithley model 196

voltmeter was used to measure the resulting voltage across

the inner two contacts. The polarity of the current was then

reversed and the voltage remeasured. The film sheet

resistance was calculated by the relationship [107]

S 4.532 V (a/square). (3.1)
Rs- I
The film resistivity in Ocm was then calculated by

multiplying R, by the film thickness in centimeters.

Superconducting properties of a film were determined by

direct current (DC) resistivity versus temperature (R-T)

data. The home-built, computer-controlled R-T system and the

operating procedure are discussed more fully in Appendix B.

In brief, the R-T system performed the same resistance

measurement as the room temperature four point apparatus but

with a controlled variation in temperature from room

temperature down to 12 K, if necessary. Thin film samples of

Y-Ba-Cu-O were loaded into a holder and contacted by four

collinear rhodium-tipped copper probes. The holder was then

inserted into a cryostat cooled by a closed-cycle He

refrigerator and the temperature automatically brought to

the desired value. A constant DC current was applied to the

outer two contacts and the resultant DC voltage across the

inner contacts was measured. The polarity of the current was

reversed and the voltage remeasured. An average resistance

was calculated from the opposite polarity voltage

measurements as discussed in Appendix B. The temperature was

successively lowered until a zero voltage measurement was

obtained, at which time the measurement was terminated by